Point-of-care antimicrobial coating protects orthopaedic implants from bacterial challenge

One pot neat synthesis of PAM polymer

We aimed to design a branched block copolymer that could effectively chemisorb to metal surfaces and self-assemble to form a uniform coating on the metal surface. Although linear PEG-PPS polymers were shown to effectively bind to gold surfaces, we found that branched PEG-PPS polymers were less effective at coating titanium and steel surfaces and required pre-silanization²⁸. We reasoned that we could improve chemisorption by using an analog structure to PPS that does not contain a methyl group in every PPS unit, which we thought limited its ability to effectively pack (Supplementary Fig. 1a). To test our reasoning, we synthesized a PAM analog of PPS, which has the same number of carbon units but no methyl side chain (Supplementary Fig. 1b). The synthesis of PAM is based on a “one-pot” thiol-ene copolymerization from a stoichiometric mixture of 1,3 propanedithiol and diallyl sulfide (Fig. 1a). The polymerization was initiated using 365 nm UV light (20 mw/cm²) using a 2,2-Dimethoxy-2-phenylacetophenone (DMPA) photoinitiator producing a linear polymer with thiol and/or allyl functional groups at each end depending on if a slight excess of the dithiol or diallyl is used, enabling further functionalization. The reaction can proceed in the presence of atmospheric oxygen and wet solvent (no special drying procedure was used). Thus, the synthesis of PAM is significantly faster and more straightforward than the anhydrous and inert conditions required for anionic polymerization used for PPS synthesis. The results of the synthesis of poly allyl mercaptan (PAM) from the thiol–ene radical copolymerization of the 1,3 propanedithiol and allyl sulfide are presented in Fig. 1a. Size-exclusive chromatography (SEC) analysis reveals that the PAM product has an Mn of 767 MW (Supplementary Fig. 2), which is equivalent to a degree of polymerization (DP) of 10. Additionally, this DP number is also in agreement with the calculation from ¹H-NMR residue terminal analysis (Supplementary Fig. 1b).

We next examined the ability of PAM to switch from a hydrophobic to a hydrophilic polymer upon oxidation. We used FT-IR to monitor the chemical transformation of thio-ether to sulfone/sulfoxide upon exposure of PAM to oxidation. After oxidation exposure, the PAM FT-IR spectra showed new peaks at 1100, 1300, and 1030 cm⁻¹ representative of sulfone and sulfoxide groups, which provides direct proof of thio-ether oxidation (Fig. 1b). Thus, the PAM polymer is oxidation responsive similar to PPS²⁹, and can switch from hydrophobic to hydrophilic behavior in the presence of reactive oxygen species³⁰.

Although the ability of PEG-PAM to bind gold has no direct relationship to its ability to bind titanium surfaces, PEG-PPS was able to bind to gold and thus we wanted to assess the ability of PAM to bind to gold. Surface plasmon resonance (SPR) was used to confirm the ability of PEG-PAM to chemisorb to gold surfaces. The SPR response was immediate after the deposition of the PEG-PAM copolymer (0.1 mg/ml in methanol) and reached a plateau quickly. Rinsing yielded a second plateau, which was associated with the chemisorbing monolayer on the metal surface (Supplementary Fig. 3). The PEG-PAM dissociation constant (K_D) was calculated to be ~4.35 μM in water.

Tetra-PEG-PAM self assembles and oxidizes

After successful preparation of the hydrophobic PAM block, we aimed to synthesize the branch block copolymer of PEG and PAM by grafting the PAM block to a tetra-PEG-SH core using thiol-ene coupling. One mole of tetra-PEG-SH was used for every four moles of PAM. Our ¹H-NMR results (Fig. 1c) indicated a total disappearance of allyl protons from PAM terminals, which means almost quantitative coupling between tetra-PEG-SH and PAM moieties. Following purification, tetra-PEG-PAM is soluble in water and it forms a hydrogel at 6% w/v with a storage modulus of ~400 Pa and a loss modulus of ~30 Pa (Fig. 1d). To demonstrate that the hydrogel was formed through self-assembly and not through covalent crosslinking during PEG-PAM synthesis, PEG-PAM hydrogels were oxidized using H₂O₂. If crosslinking is through covalent bonds, the modulus of the hydrogel should not change significantly; however, if the crosslinking is through self-assembly, the gel would be completely disrupted when the PAM block changes from hydrophobic to hydrophilic, and the storage modulus would be lower in value than the elastic modulus. Upon exposure to H₂O₂, the mechanical properties of the PEG-PAM hydrogel were similar to that of PEG polymer, indicating that PEG-PAM forms a hydrogel through self-assembly.

In situ tetra-PEG-PAM synthesis effectively coats metal surfaces

Our goal was to use tetra-PEG-PAM mixed with antibiotics to effectively coat orthopaedic implant surfaces and prevent infection. However, the fact that PEG-PAM has high viscosity in both aqueous and organic solvents and forms a hydrogel at concentrations over 5% w/v complicates the coating. Since high viscosity and gelation occur only when tetra-PEG and PAM are conjugated to form an amphiphilic branched co-polymer, we reasoned that we could avoid working with this highly viscous solution and effectively coat the implant by applying a solution of tetra-PEG and PAM precursors prior to in situ conjugation to form tetra-PEG-PAM (Fig. 2a). Titanium wires were submerged in a solution containing a mixture of PAM and tetra-PEG-thiol dissolved in chloroform, resulting in a final concentration of 6% w/v of tetra-PEG-PAM polymer. The titanium wires were subsequently irradiated under UV light (365 nm, 20 mw/cm²) for 5 min to induce the formation of tetra-PEG-PAM directly on the implant surface and the solvent was subsequently evaporated under vacuum. The drying process can also be accomplished through passive evaporation at the bench surface or through flowing nitrogen gas over the implant surface. All of these approaches are expected to be OR and surgery compatible. The resulting coated implant was analyzed using scanning electron microscopy (SEM), elemental analysis, and gross images using a rhodamine-labeled coating to assess the chemical composition and coating microstructure of the surface. SEM and gross images show the coating covering the implant surface changing it from a rough appearance to a smooth appearance (Supplementary Fig. 4a, b). Elemental analysis revealed an increase in sulfur concentration from 0-wt% to 2-wt% and a reduction in chlorine content to 99.6% relative to the coating solution, consistent with the incorporation of PAM and removal of solvent on the implant surface (Supplementary Fig. 4c, 5). To estimate the coating thickness, mock titanium-coated ultra-flat gold slides were used and analyzed with AFM and 3D profilometer (Fig. 2b, c, Supplementary Fig. 4d, e). AFM and 3D profilometer results showed clear differences between coated and uncoated areas in both roughness and thickness. The thickness of the tetra-PEG-PAM layer was found to be 2 µm.

In situ PEG-PAM antibiotic coating as an antimicrobial strategy

Antibiotics are introduced into the coating by mixing the coating precursors, tetra-PEG and PAM, with the desired antibiotic prior to coating the implant (Fig. 2a). Since the “thiol-ene” photoconjugation in solution does not release heat, a wide range of antibiotics can be incorporated into the coating. Utilizing vancomycin (Vanc), one of the most commonly used antibiotics after orthopaedic surgery to prevent infection, we studied the parameters that affect antibiotic loading, such as the role of tetra-PEG-PAM concentration in the coating and sequential coating layers. SEM images of coated pins that contained Vanc displayed similar smooth morphology as those with PEG-PAM only (Fig. 2d). Utilizing PEG-PAM from 2% to 20% w/v in the coating solution resulted in an increased concentration of Vanc loaded with increasing percent of the loaded polymer, ranging from 30–50 µg/mm² (Supplementary Fig. 6). Similarly, undergoing the coating process several times results in an increase in Vanc loading with each new layer, ranging from 100–300 µg/mm² for 1 to 8-layers respectively (Fig. 2e). The thickness of the coating was again estimated using flat titanium-coated glass slides and 3D profilometry and increased with each additional layer added reaching 100 µm coating for 8-layers (Fig. 3f). Vanc release in a PBS bath shows that the concentration of Vanc in solution increased over time and was above the MIC (0.5 μg/mL for Xen 36) for 14 days (Fig. 2g). Pins that were coated with 3 layers of PEG-PAM + Vanc resulted in concentrations above the MIC for 17-days (Fig. 2h).

To measure in vitro antimicrobial activity against S. aureus, we performed zone of inhibition (ZOI) assays with our PEG-PAM coated titanium pins with/without Vanc. Vanc loaded pins with one layer of 6% w/v coating (60 µg/pin) were placed on top of S. aureus (Xen36) agar plates (Fig. 3a). Distinctive dead zones were observed on the plates containing tetra-PEG-PAM/Vanc coated pins, while uncoated pins and tetra-PEG-PAM pins showed no dead zone. Quantification of the ZOI area showed a statistically increased area for the Vanc containing group (Fig. 3a).

The ability to effectively incorporate a variety of antibiotics was confirmed by coating separate implants each with one of 8-different clinically relevant antibiotics including ceftriaxone, cefazolin, cefepime, tobramycin, piperacillin and tazobactam, clindamycin, linezolid, and rifampin. All of these antibiotic-loaded titanium pins were used to verify the ability to inhibit S. Aureus growth in vitro (Fig. 3b, c). The coated pins were incubated in a solution of bioluminescent Staphylococcus aureus Xen36, which contain a bioluminescent lux operon construct integrated into a stable bacterial plasmid that naturally produces a blue-green light emitted only by metabolically active bacteria. Bioluminescence readings were taken at 0, 2, 8, and 24 h after bacteria inoculation. The cefazolin, tobramycin, piperacillin and tazobactam, and rifampin groups had almost no bioluminescent signal after 8 h, which indicated an efficient inhibition of S. aureus growth from antibiotics released from the coated pins (Fig. 3b, c). Quantification revealed a steady decrease in bioluminescence reaching baseline levels by 24 h. In contrast, pins containing only tetra-PEG-PAM resulted in a significant bioluminescence increase by 24 h (Fig. 3b, c). Antibiotic-coated implants were also challenged with other infectious bacterial strains, including Pseudomonas aeruginosa PAO-1 and Escherichia Coli RR1 with ZOI assays. With the appropriate selection of antibiotics in the PEG-PAM coating (tobramycin for PAO-1 and piperacillin and tazobactam for RR1), distinctive dead zones were observed in the plate containing tetra-PEG-PAM/antibiotic coated pins, while tetra-PEG-PAM pins without antibiotic showed no dead zone (Fig. 3d, e).

PEG-PAM coating does not inhibit osseointegration long-term

Implant surfaces are carefully designed for optimal biocompatibility and osseointegration. Thus, any coating technology must demonstrate that osseointegration is not compromised. To evaluate the effect of the PEG-PAM/Vanc coating on osseointegration, push-in-force measurements were used. The distal femur of 24 mice were implanted with an untreated control Ti implant and 24 mice were implanted with a Ti implant coated with PEG-PAM/Vanc. The technique for implantation was the same as that used in the mouse model of knee PJI, except no bacteria was used for this experiment (Fig. 4a–c, Supplementary Fig. 9)⁹. This animal model was developed by practicing orthopaedic surgeons specifically as a small animal screen for the evaluation of translational technologies. Briefly, a sterile titanium pin (6 mm length × 0.8 mm diameter) is placed retrograde, from the knee joint into the femoral canal by first generating a channel using a 25-gauge and then a 21-gauge needle, consistent with the clinical practice of reaming. At 2- and 4-weeks following implantation, 12 mice from each group were euthanized and the femurs were harvested. The mechanical withholding strength was measured by pushing the implant in a retrograde fashion into the femoral canal using a custom-made stainless-steel pushing rod mounted on a 1000 N load cell (Instron, Canton, MA). The axial load on the implant was applied at a crosshead speed of 1 mm/min, and the maximum load (N) to displacement was measured as the implant push-in value (Fig. 4c)³¹.

We found that in both the control and the PEG-PAM/Vanc coated implant group there was a statistical increase in the mean load to displacement from 2–4 weeks (Fig. 4d). In the control group, the mean load to displacement increased from 13.2 N (standard deviation (SD) 4) to 30.2 N (SD 4.6) while in the PEG-PAM/Vanc coated group the mean load to displacement increased from 7 N (SD 5.1) to 27.6 N (SD 5.6). The mean load to displacement was statistically significantly lower in the PEG-PAM/Vanc coated group at 2 weeks, but there was no difference between the groups 4-weeks post-implantation. This data suggests that the process of osseointegration was ongoing in both groups between 2 and 4 weeks and that the PEG-PAM/Vanc coating does not inhibit the establishment of osseointegration by 4 weeks post-implantation.

To assess whether the delay in osseointegration corresponds to the presence of the coating, we assessed the in vivo degradation rate of PEG-PAM coating using the same PJI mouse model. In vivo biodegradability of PEG-PAM coating was assessed using non-invasive imaging through labeling PEG-PAM with Cy5.5. Following implantation, the decay of fluorescent signal was monitored in the mice for 3 weeks. We found a steady fluorescence decrease in signal over time with no detectable level at 3-weeks (Fig. 4e). Given that Cy5.5 last more than 4 weeks in vivo³² we conclude that coating degradation, rather than fluorophore degradation, is responsible for the observed decrease and that our coating is largely resorbed by 3-weeks. These results are consistent with the observed osseointegration and with the notion that the implant surface retains its ability to undergo osseointegration after coating resorption.

PEG-PAM coating can withstand mechanical forces during implantation

Though not all orthopaedic implant surgeries impose mechanical shear to the implant surface, the press-fit technique in the PJI model does and thus, we set to determine to what extent the coating sheared after press-fit implantation. Two distinct methods were used to qualitatively evaluate the mechanical stability of the PEG-PAM/Vanc coating following press-fit implantation. In each case, 6 Ti implants were coated with PEG-PAM/Vanc. The first method employed a synthetic bone mimic (10 pounds per cubic foot (PCF) polyurethane foam block, Sawbones). 10 PCF was selected as it has similar tensile modulus and compressive strength as the trabecular bone in the human distal femur^33,34. We compressed the flat surfaces of two Sawbones blocks together using lag screws. The potential space between the two blocks was then reamed first with a 25- and then a 21-gauge needle as per our implantation protocol and the coated implants were inserted by hand (Fig. 4f, Supplementary Fig. 7). Following insertion, the screws in the blocks were removed, the blocks separated, and the implants carefully lifted from the blocks and imaged via scanning electron microscopy (SEM). Using this method, the force on the coating was only felt during insertion, thus creating a more realistic model of clinical reality in which the bone is first reamed and then the implant is inserted and left in place.

The second set of 6 coated implants were press-fit in a retrograde fashion into freshly harvested cadaver mouse femurs, and then pulled out from the direction in which they had been inserted (Supplementary Fig. 8). Following insertion and removal, the implants were imaged via SEM. The insertion site in the intercondylar notch was sequentially reamed first with a 25- and then a 21-gauge needle prior to implant insertion (Supplementary Fig. 9). Unlike the Sawbones technique described above, in the cadaver femur technique the coating experienced both the force of insertion and removal. The post-implantation and removal SEM images for both the Sawbones and cadaver mouse femur methods showed that the coating was largely intact, with only rare surface abrasions (Fig. 4f, Supplementary Fig. 7, 8).

Antibiotic loaded Tetra-PEG-PAM coating effectively protects the implant surface from bacterial challenge in a PJI model

We next evaluated if our implant coating technology could effectively prevent infection after periprosthetic joint infection using the knee PJI model. Bioluminescent Staphylococcus aureus Xen36 (10³ bacterial light units) was used to inoculate the intra-articular portion of the metal pin in the joint space to induce the formation of a controlled infection that can be non-invasively quantified using the Xenogen in vivo imaging system (IVIS Lumina II, PerkinElmer, Hopkinton, MA). We tested joints treated with implants that had no coating, PEG-PAM or PEG-PAM/Vanc. In the mice that received implants with one layer of 6% w/v tetra-PEG-PAM/Vanc coating the bioluminescent signals were significantly lower (p < 0.05), compared to mice with implants that were not coated or were treated with the tetra-PEG-PAM coating alone (Fig. 5a, b), indicating that the tetra-PEG-PAM polymer alone has no antibacterial properties and cannot protect the implant surface. The Vanc coated implants had a bioluminescence signal that was not above baseline from post-operative day (POD) 1 onward, suggesting eradication of infection below the level of detection by noninvasive imaging. However, bacteria could still be present at low levels at the implant surface and surrounding tissue.

To assess the presence of bacteria at the implant surface and tissue, at POD 21, the implants and surrounding tissue were retrieved, and the number of colony-forming-units (CFUs) was assessed (Fig. 5c). As expected, animals that received infected implants had 100% of the implants and tissue grow out of bacteria. Similarly, animals that received PEG-PAM-only implants had 100% of the implants and tissue grow out of bacteria, again showing that PEG-PAM coating itself is not antimicrobial nor protects the surface from infection. Animals that received implants that were coated with PEG-PAM/Vanc resulted in none (0%) of the implants containing CFUs, showing that PEG-PAM/Vanc coating effectively and completely protects the implant surface from infection.

Periprosthetic osteolysis is a major complication of orthopedic joint infections, where bone loss occurs and the implant loosens due to the local inflammatory response to infection. Animals treated with implants coated with tetra-PEG-PAM alone showed a dramatic degree of periprosthetic osteolysis that became evident by POD 21 (Fig. 5d). In contrast, animals treated with PEG-PAM/Vanc coated implants showed no detectable radiographic periprosthetic osteolysis (Fig. 5d), consistent with the high efficacy of these coatings in preventing bacterial replication and subsequent host immune response around the implants and the surrounding bone and soft tissue. To examine the effect of the PEG-PAM/Vanc antibiotic implant coating on the microscopic anatomy of the distal femur, histologic sections of the tissue were evaluated (Supplementary Fig. 10a). Uncoated pins, as well as pins coated with PEG-PAM without antibiotics, demonstrated morphologic changes in the bone and surrounding joint tissue. These changes included increases in the size of the distal femur and changes to the normal bony architecture, as well as synovial hyperplasia and an inflammatory cell infiltrate. These alterations were less pronounced in distal femoral tissue from PEG-PAM/Vanc coated pins. Taken together, these findings show that the tetra-PEG-PAM/Vanc coating protected the implant surface and surrounding tissue from infection and the resulting inflammatory reaction that can lead to periprosthetic osteolysis.

To challenge the coating further, the implants were infected with orders of magnitude higher numbers of bacteria. We infected the implants with 10⁴, 10⁵, 10⁶ bacterial light units instead of 10³ as previously done. Surprisingly, in 10⁴ and 10⁶ bacterial challenge conditions, one layer of tetra-PEG-PAM/Vanc effectively protected the implant surface from infection with no implants containing CFUs at day 21-post implantation (Fig. 5e, Supplementary Fig. 10b). However, animals receiving 10⁵ bacteria a single animal contained CFUs at both the implant and surrounding tissue (Supplementary Fig. 10c), indicating some variability in the coating or surgical procedure. Nevertheless, completely preventing infection at 10³, 10⁴, and 10⁶ bacterial challenge conditions demonstrates that our PEG-PAM/Vanc implant coating is able to completely protect the implant surface from infection.

Given that it is likely that this technology would be applied in conjunction with the current standard of care, IV antibiotics, we compared IV antibiotic delivery alone or in combination with PEG-PAM/Vanc coating. Utilizing a mouse-equivalent dose of the standard of care IV delivered antibiotic resulted in a higher bioluminescence signal than the PEG-PAM/Vanc group from post-operative day 0 to day 10. We have performed studies directly comparing IV-delivered antibiotics and our PEG-PAM/Vanc coating. Our results demonstrate that implants coated with PEG-PAM/Vanc were statistically more protected from infection than the standard of care. Bioluminescence curves of animals treated with PEG-PAM/Vanc coated implants and the combined treatment of PEG-PAM/Vanc coated implants plus IV antibiotics were indistinguishable from each other (Fig. 5f), suggesting that the coating alone effectively prevents infection.

Loading of multiple antibiotics in PEG-PAM coating

In practice, there is often a need to introduce multiple antibiotic types into a single implant to establish broad-spectrum antimicrobial therapy. Multiple types of bacteria may be present, requiring multiple types of antibiotics for adequate treatment. Multiple antimicrobial agents can also be used to ensure that the bacteria are completely eliminated, reducing the risk of generating antibiotic-resistant bacterial strains. Thus, we investigated the incorporation of two antibiotics into the same implant coating using PEG-PAM. Implants were coated with rifampin or a mixture of rifampin and vancomycin using the same coating procedure as used above and resulted in 37.8 µg/implant for rifampin in rifampin-only coatings and 28.0 µg/implant rifampin and 12.8 µg/implant vancomycin for the implants that contained both antibiotics. Both implants coated with tetra-PEG-PAM/Rifampin and tetra-PEG-PAM/Vancomycin/Rifampin effectively protected the implant surface from infection (Fig. 6a, b), decreased the number of CFUs on and around the implant surface (Fig. 6c), and protected the implant from periprosthetic osteolysis (Fig. 6d).

Loading of vancomycin in PEG-PAM coating on different sized stainless-steel implants

In orthopaedic surgery, implants are usually diversified with different metals of multiple sizes. Thus, we explored the compatibility of our coating technology using different sized stainless-steel implants (0.1, 0.6, and 0.8 mm). All these implants were coated with vancomycin using the same procedure and were examined in our mouse model of knee PJI. Our in vivo results indicated these implants coated with tetra-PEG-PAM/Vanc effectively protected the implant surface from infection (Fig. 6e, f), decreased the number of CFUs on and around the implant surface (Fig. 6g), and protected the implant from periprosthetic osteolysis (Fig. 6h).

Antibiotic loaded PEG-PAM coating effectively protects the implant surface from bacterial challenge in spine implant infection model

To demonstrate the versatility of our coating technology, we explored the possibility of transferring the coating process to a spine implant infection model³⁵. Besides PJI, spine implant infection is another devastating complication that occurs in 2–8% of all elective spine surgeries. Briefly, an L-shaped stainless-steel pin (0.1 mm) is placed through the L4 spinous process and laid cranially along with the posterior spinal elements. Bioluminescent Staphylococcus aureus Xen36 is used to inoculate the bend of the metal pin to induce the formation of a controlled infection that can be non-invasively quantified using the Xenogen in vivo imaging system (IVIS Lumina II). We performed the same PEG-PAM coupling with vancomycin on the 0.1 mm “L” shaped stainless-steel implant. Our in vitro release experiment indicated an elution of vancomycin for 1 week above MIC. However, the coating was ineffective in preventing infection resulting in the same level of infection as PEG-PAM coating alone. (Fig. 7e, f).

To overcome this limitation and deliver more antibiotics in the spine surgery model, we designed and synthesized a more stable polymer network: PEG-PAMDA (polyallyl mercaptan diallyl), which is covalently cross-linked by a thiol-ene click reaction between tetra-PEG-SH and PAMDA. In addition, based on our previous results, we increased the coating layer from 1–3 to maintain 3 times more antibiotic (352 ug) loading compared with the single-layer PEG-PAM coating (140 ug). (Fig. 7c) Animals treated with this improved PAMDA/Vanc coated implants had significantly decreased bacterial infection when compared to PAMDA coating alone (Fig. 7g–h). These results indicate that a relatively simple modification of the coating, that allows for further crosslinking, is more effective for preventing infection in spinal implants.

As mentioned, the off-label application of antibiotic powder to the wound is used at times to prevent infection. This is often the case for spine surgeries. We recently assessed and published on the use of vancomycin powder to prevent infection³⁵. In these studies, we find that vancomcyin powder completely suppresses bioluminescence signal; however, 20% of the animals have persistent infections despite the large amounts of vancomycin powder used of 2, 4, and 8 mg per wound.

Demonstration of PEG-PAM coating on a clinically used implant in operating room

To further demonstrate the potential applicability of our coating technology, we performed the PEG-PAM coating process in the operating room. The preparation of the coating process included the prepolymer solution as described above, antibiotic (rifampin), a human intramedullary hip implant, spray bottle or paintbrush, and a UV light source. Rifampin (Rif) and the prepolymer solution were mixed together and then transferred to the spray bottle. The human intramedullary hip implant was either sprayed or painted by brush with the PEG-PAM/Rif polymer solution and then irradiated under UV light for 2 min to yield a homogeneous PEG-PAM/Rif coating on the surface of the implant (Fig. 8). This process was easily performed under sterile conditions.