Bacterial biofilm formation is considered to play a major role in the pathogenesis of BAI (Costerton et al., 1999; Holmberg et al., 2009; Anderson and Patel, 2013). Biofilm formation is initiated by bacterial cells attaching to the surfaces of medical devices. Subsequently, bacteria replicate and produce extracellular matrix forming complex communities consisting of bacteria, bacterial exopolysaccharides, proteins, extracellular DNA, and host proteins (Costerton et al., 1999). Bacteria in biofilms are considerably more tolerant to antibiotics and less accessible to cells and molecules of the human immune defense system than their planktonic counterparts (Otto, 2009; Chen et al., 2013). This might be due to the extracellular polymeric matrix of the biofilm, making the bacteria less accessible for phagocytes and effector molecules, and to the persister state of the bacteria. Persisters are metabolically-inactive, antibiotic tolerant bacteria that maintain the ability to multiply after antibiotic treatment (Harms et al., 2016), thus explaining the recurrence of BAI (Gerdes and Semsey, 2016; Fisher et al., 2017).

Another important element in the pathogenesis of BAI is bacterial colonization of the tissue surrounding the implant (Boelens et al., 2000a; Ciampolini and Harding, 2000). In vivo studies showed that S. epidermidis applied on the surface of titanium implants, both as adherent cells and as a pregrown biofilm, rapidly relocated from the implants to the surrounding tissue (Riool et al., 2014). Similarly, large numbers of S. aureus were cultured from mouse tissues around infected titanium (Riool et al., 2017a,b) and silicon elastomer implants (de Breij et al., 2016). In a murine model of chronic osteomyelitis, S. aureus was found in osteoblasts and osteocytes, as well as in canaliculi of live cortical bone (de Mesy Bentley et al., 2017).

Bacterial invasion of the peri-implant tissue and subsequent development of infection is facilitated by dysregulation of the local immune response resulting from the presence of a foreign body. The phagocytic and intracellular killing activities of neutrophils and macrophages are reduced due to altered cytokine tissue levels in the presence of a biomaterial (Boelens et al., 2000a,b,c; Broekhuizen et al., 2010; Zimmerli and Sendi, 2011). In agreement, microscopical examination has revealed that many of the bacteria reside within these inflammatory phagocytes (Broekhuizen et al., 2010). Interestingly, studies in mice infected with S. epidermidis as well as in infected peri-catheter tissue biopsies obtained from deceased intensive care unit patients showed that bacteria present in tissue surrounding the implants had incorporated bromodesoxyuridine, demonstrating that the bacteria can replicate in the peri-implant tissue (Broekhuizen et al., 2010). Furthermore, bacteria may adapt to the tissue and intracellular micro-environment by the formation of so-called small colony variants. The presence of such intracellular small colony variants further complicates treatment as they are more resistant to antimicrobial compounds (Tuchscherr et al., 2010; Zaat, 2013).

In addition to the limited activity of antibiotics against biofilm-encased bacteria, persisters, and intracellular bacteria, the emergence of resistance among staphylococci as well as other bacterial species causing BAI constitutes a major challenge to the efficacy of (combinations of) conventional antibiotics. The emergence of multidrug-resistant (resistant to at least one agent in three or more antimicrobial classes), extensively drug-resistant (resistant to at least one agent in all but one or two antimicrobial classes), and pan-drug-resistant (resistant to all agents in all antimicrobial classes) pathogens, is accelerated by the selective pressure exerted by extensive use and abuse of antimicrobials (Magiorakos et al., 2012). Bacteria belonging to the so-called ESKAPE panel (Enterococcus faecium, S. aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) are increasingly prevalent and resistant and thereby a particularly dangerous group of bacteria (Rice, 2008). Currently, the majority of hospital infections in the United States is caused by multidrug-resistant ESKAPE bacterial strains (Boucher et al., 2009). The World Health Organization recently endorsed a global action plan to tackle antibiotic resistance to avoid the dark scenario of a “post-antibiotic era” (Chan, 2015). One of the key objectives of this plan is to develop novel antimicrobial drugs with a mode of action different from those of current antibiotics.

Already in 1987, Gristina suggested that adhering tissue cells and bacteria compete for a spot on the implant's surface, the so-called “race for the surface” concept (Gristina, 1987). In case this race is won by the bacteria, this will result in infection instead of tissue integration. Gristina also realized that colonization of the tissue around implants was another possible mechanism of infection (Gristina, 1987). Bacterial adhesion and subsequent biofilm formation may be prevented by modifying the physicochemical surface properties of biomaterials such as the surface charge, hydrophobicity/hydrophilicity, and surface chemistry. One strategy is to use hydrophilic polymer coatings, e.g., immobilized PEG, as applied on contact lenses, shunts, endotracheal tubes, and urinary catheters (Banerjee et al., 2011; Busscher et al., 2012). Another approach is functionalization of the surface with a dense layer of polymer chains commonly known as polymer brush coatings (Nejadnik et al., 2008; Neoh et al., 2013; Keum et al., 2017). Large exclusion volumes of tethered polymer chains result in surfaces difficult to approach by proteins or bacteria.

Another approach to prevent implant colonization is the immobilization of AMPs on surfaces of medical devices, which can be performed with a variety of chemical techniques. An excellent overview of immobilization strategies has recently been published by Silva et al. (2016). There are several common “rules” for success. The structural characteristics important for the antimicrobial activity of the peptides should not be altered by the immobilization process. Length, flexibility, and kind of spacer connecting the peptide to the surface, orientation of the immobilized peptides, and the AMP surface density are additional important parameters (Costa et al., 2011). Interestingly, even short surface-attached peptides not likely to have a free interaction with the bacterial cytoplasmic membrane, have antimicrobial activity (Hilpert et al., 2009). This is thought to be due to destabilization of the bacterial membrane by displacement of positively charged counter-ions, disrupting the ionic balance, changing bacterial surface electrostatics, and activating autolytic enzymes (Hilpert et al., 2009). An example of a contact-killing surface is the hydrogel network with the covalently attached stabilized inverso-CysHHC10 peptide (Cleophas et al., 2014). This coating demonstrated high in vitro antimicrobial activity against S. aureus, S. epidermidis, and E. coli. Furthermore, brush coating molecules may also possess active functional groups with antimicrobial activity, e.g., by conjugation with the AMPs Tet20 (Gao et al., 2011a) and Tet213 (Gao et al., 2011b). Another example is polyurethane with a brush coating tethered with the AMP E6 for the prevention of catheter-associated infections (Yu et al., 2017). This surface coating reduced bacterial adhesion on the catheter surface in a mouse urinary catheter infection model. A variety of AMPs, like GZ3.27 (De Zoysa and Sarojini, 2017), GL13K (Chen et al., 2014; Zhou et al., 2015), SESB2V (Tan et al., 2014), bacitracin (Nie et al., 2016, 2017), hLF1-11 (Costa et al., 2014; Godoy-Gallardo et al., 2015), LL-37, Melimine, lactoferricin, and Mel-4 (Chen et al., 2016; Dutta et al., 2016) have been covalently coupled onto various surfaces, such as glass, silicon, and titanium, with different degrees of success (summarized in Table 1). Chimeric peptides comprised of both a titanium binding domain and an antimicrobial motif are also used to create contact-killing surfaces (Yucesoy et al., 2015; Liu et al., 2016; Yazici et al., 2016). Due to their titanium-binding domain, the peptides preferentially bind the implant, while the freely exposed antimicrobial domain is available for combatting invading bacteria. Titanium surfaces modified with these chimeric peptides were found to significantly reduce adhesion of different Streptococcus species, S. aureus, S. epidermidis, P. aeruginosa, and E. coli, compared to bare titanium. Immobilization of GL13K onto titanium dental implants even enabled osseointegration when tested in rabbit femurs (Chen et al., 2017). Another promising strategy is the development of multifunctional coatings by combining the well-known RGD cell adhesive sequence with the lactoferrin-derived AMP LF1-11, resulting in in vitro cell integration as well as inhibition of bacterial colonization by S. aureus and Streptococcus sanguinis (Hoyos-Nogués et al., 2017). Recently, others described a self-assembling coating of recombinant spider silk protein fused to the AMP Magainin I for different biomaterials, which reduced numbers of live bacteria on the coated surfaces (Nilebäck et al., 2017). It should be noted that not in all studies described above the absence of unbound peptide within the coating is verified. Thus, in those cases it cannot be excluded that the antimicrobial activity of the coating is caused by a combination of bound and released AMP.

It should be noted that surface attachment of peptides does suffer from some disadvantages. The antimicrobial activity of the surface with immobilized AMPs is critically dependent on the chemical tethering procedure and the orientation of the covalently attached AMPs. The antimicrobial activity of the resulting coating may be strongly reduced compared to the activity of the peptide in free form (Bagheri et al., 2009; Onaizi and Leong, 2011; Dutta et al., 2016). Apart from this reduction of activity due to the tethering process, proteins, blood platelets, and dead bacteria may block the antimicrobial groups on the surface. Moreover, since the antimicrobial activity is restricted to the surface of the implant, there is a lack of antimicrobial impact on bacteria in the tissue surrounding the implant. Contact-killing surfaces will only eradicate bacteria that are in direct contact with the active surface, meaning that clearance of any bacteria further away from the surface will depend on efficient phagocytosis and systemic or local antibiotics. However, as mentioned before, due to the presence of a biomaterial the local host immune response is dysregulated, and therefore phagocytosed bacteria may not be killed and may even persist intracellularly (Boelens et al., 2000a,b).

As described above, the peri-implant tissue is an important niche for bacterial survival. Therefore, antimicrobial-releasing surfaces or coatings from which the antimicrobial agent also reaches this niche are preferred to prevent BAI. Antibiotic-releasing coatings are widely used for medical devices such as sutures and central venous catheters and urinary tract catheters. However, these coatings have two major disadvantages: (i) a patient may be infected with a bacterium resistant to the released antibiotic, and (ii) due to the local release a gradient of the antibiotic will be present near the implant thereby increasing the risk to select for resistant bacteria. Coatings releasing antibiotics for orthopedic devices remain mainly experimental (Lucke et al., 2003; Kälicke et al., 2006; Darouiche, 2007; Moojen et al., 2009; Alt et al., 2011). The first commercially available gentamicin-releasing intramedullary tibia nail has recently shown promising results in a first prospective study (Fuchs et al., 2011; Metsemakers et al., 2015; Alt, 2017). In view of the increasing development of antibiotic resistance among bacteria, the use of antibiotics in medical devices is discouraged by government regulatory agencies like the American Food and Drug Administration (FDA, 2007; Brooks et al., 2013). Obviously, coatings releasing antimicrobial agents that are less likely to induce resistance, such as AMPs, are preferred in view of both managing resistance development and compatibility with use of antibiotics for prophylaxis or treatment. To prevent the spread of bacteria from the implant surface to the surrounding tissue, and to eradicate bacteria contaminating tissue during surgery, a rapid initial release of antimicrobials is required. If this release is delayed, bacteria may “escape” into host cells before effective levels of the antimicrobial agent have been established. Subsequently, prolonged local release of the antimicrobial agent at sufficiently high concentrations will be required to eradicate any residual bacteria (Zilberman and Elsner, 2008; Emanuel et al., 2012).

Application of AMPs in antimicrobial surface coatings is a subject of increasing interest and different types of release-coatings have been described, including hydrogels, nanotubes, microporous calcium phosphate coatings, and polymer coatings (summarized in Table 2). Hydrogels with the AMP Cateslytin strongly adhere to dental implant surfaces. The hydrogels showed potent antimicrobial activities against Porphyromonas gingivalis, an important causative agent of peri-implantitis, without signs of toxicity (Mateescu et al., 2015). Another example is a gelatin-based hydrogel on titanium surfaces allowing for the controlled release of the short cationic AMP HHC36 preventing S. aureus, S. epidermidis, E. coli, and P. aeruginosa biofilm formation (Cheng et al., 2017).

Self-organized and vertically oriented titanium oxide nanotubes loaded with the broad spectrum AMP HHC36 showed in vitro bactericidal activity against S. aureus in liquid surrounding the nanotubular surface and reduced bacterial colonization on the surface ~200-fold (Ma et al., 2012). GL13K-eluting coatings on these titanium oxide nanotubes prevented growth of Fusobacterium nucleatum and P. gingivalis in an in vitro disk-diffusion assay (Li et al., 2017). In vitro release of Tet213 from microporous calcium phosphate coatings applied on titanium showed bactericidal activity against S. aureus and P. aeruginosa (Kazemzadeh-Narbat et al., 2010). In a similar approach, release of PSI 10 from microporous calcium phosphate coated magnesium alloy inhibited S. aureus growth in vitro and promoted in vivo bone repair (Tian et al., 2015). Furthermore, controlled release of Tet213 linked to collagen IV inhibited S. aureus biofilm formation in vitro (Shi et al., 2015). However, these types of coatings have not yet been tested in vivo.

Injection of the LL-37-inspired AMPs OP-145 (de Breij et al., 2016), SAAP-145, and SAAP-276 (Riool et al., 2017a) along subcutaneous implants in mice did not reduce the numbers of S. aureus in the surrounding tissue. This might be because the AMPs did not effectively penetrate the tissue or were not taken up by the host cells and thereby not capable of killing internalized bacteria. However, when these AMPs were released from Polymer-Lipid Encapsulation Matrix (PLEX) coatings, the numbers of viable S. aureus bacteria were reduced in the peri-implant soft tissue in mice (Riool et al., 2017a) and even in bone in a rabbit humerus intramedullary nail infection model (de Breij et al., 2016). This clearly illustrates the benefit of the PLEX coating technology allowing controlled and prolonged release of the AMPs at the implant-tissue-interface. The SAAP-276-PLEX-coated implants were able to significantly, but not completely, reduce the number of doxycycline-resistant S. aureus in the peri-implant tissue, in contrast to the doxycycline-PLEX coated implants which failed to reduce their numbers in the tissue (Riool et al., 2017a). This underlines the potency of SAAP-276-PLEX coatings in the fight against BAI caused by multidrug-resistant staphylococci.

Although the AMPs mentioned above reduced the colonization of the peri-implant tissue in vivo when released from a coating, they might still not be able to act against intracellular bacteria. Apparently, the rapid initial release of the AMPs killed the vast majority of the infecting bacteria, preventing biofilm formation on the implant surface as well as colonization of the tissue, thereby protecting both these sites against colonization. Treatment of infections featuring intracellular bacteria remains difficult, as observed with the conventional antibiotic vancomycin (Broekhuizen et al., 2008), and likely with the novel AMPs as well. A possible way to improve the intracellular entry of AMPs is by adding a specific domain (“tag”) to the peptides as a signal for uptake by the host cells (Splith and Neundorf, 2011; Ye et al., 2016). However, intracellular localization of bacteria does not seem to occur to a large extent when AMPs are used in BAI prevention, as shown for instance with the AMP-PLEX coatings described above. By directly killing the bacteria on the implant-tissue interface the AMPs prevented bacterial invasion into the tissue and internalization by and survival in host cells.

Several novel technologies are arising for manufacturing implants with particular focus on the possibility of personalization. We will briefly address additive manufacturing and electrospinning with regards to the strategies of incorporation of antimicrobial agents and potential for AMPs.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.