**4. Discussion**

Electrospun scaffolds with incorporated silver nano-hydroxyapatite are nontoxic to cells at a concentration that was antimicrobial, making them suitable for use as antimicrobial bone implants. In 2017, Anjaneyulu et al. [25] published a study looking into nanoscale electrospun fibers containing silver-doped hydroxyapatite and its hemocompatibility. Our study took a different direction, producing scaffolds with fibers to allow ingrowth of cells with larger pore sizes having positive impact on osteoblast cell growth [31]. Our study also included an in-depth microbiology study looking at both contact and noncontact tests. Investigating the impact of the scaffolds on bone regeneration via analyzing the toxicity on primary MSCs, for relevant cell toxicity, and 3T3 fibroblasts allowed toxicity comparison between labs.

The rapid manufacturing of nHA pioneered by Wilcock et al. [10] was replicated to produce silver-doped nHA powder. The characterization of nanoscale hydroxyapatite indicated it was comparable to that produced in previous studies. We were able to go one step further by confirming via TEM XRD that the denser (black) areas on the TEM contained silver. This had been hypothesized by the previous paper but not directly confirmed [21]. Under XRD we confirmed that the material was both hydroxyapatite and that it contained silver. TEM analysis showed that the nHA was in a nanoscale particulate form. The lack of an XRD-measurable silver phosphate phase in the 5 mol.% Ag nHA demonstrated that the silver had been incorporated into the nHA, either by substitution of silver into the HA crystal lattice for calcium or by doping of the silver ions onto the HA structure (i.e., not replacing the ions of the HA crystal). This result was expected for our method as our previous reports demonstrated a pure hydroxyapatite phase precipitated at 5 mol.% using this reaction [21].

The silver-doped nHA was successfully electrospun into PCL scaffolds. PCL is a popular material used in bone-related applications as its long-term degradation profile is suited to the slower healing rate of bone compared to soft tissues. Incorporating silver-containing tricalcium phosphate was reported by Schneider et al. [26], who found it effective against *E. coli* but did not test against the more common bone infection-related pathogen *S. aureus*. Jiang et al. [19] assessed antibacterial activity using both of these bacteria with a scaffold formed by in situ foaming, which allowed inclusion of silver/nHA up to 40% of total weight. This technique does not produce scaffolds, however, instead producing a monolithic block that would need to be carved to dimensions required during theater. Anjaneyulu et al. [25] investigated using both bacteria types and manufactured formed scaffolds but did not assess bone cell biocompatibility.

Our sca ffolds degraded in an accelerated degradation study and showed a two-phase silver release system. Initially, a rapid burst of silver was released, decreasing in rate until day 7. During the same initial period of degradation, the polymer had not lost any mass. This rapid silver release is likely to arise from surface- (or near surface-) bound material. After the initial 7 days, the material lost mass and silver was released for the remainder of the experiment. Rothstein et al. described this in a model of an initial surface erosion system, which then transitions into bulk erosion [32]. The combination of the two allows for a rapid release when first implanted to clear immediate microorganisms and then a slower sustained release of the remainder of the implant life to prevent colonization from opportunistic bacteria, which may later attach to the device through a hematogenous route. Liu et al. [15] used electrochemical deposition of silver onto their electrospun fibers to provide an initial burst of silver over two days instead of a continual release. This would help prevent potential buildup of silver toxicity but would not prevent later infections from occurring. A continuous release of calcium and phosphorus was detected using ICP, which has been widely reported to stimulate bone regeneration [33]. An implanted material can act as a desirable location for bacteria to colonize with an infection requiring far fewer bacteria to start than in a normal tissue [34]. A continuous silver release would prevent opportunistic bacteria from infecting the implant at a later date. Due to the limited nature of our degradation method we cannot report on the exact amount of silver released per unit of time in vivo, as physiological conditions would impose a more complicated environment that may impact silver release. Despite not knowing the exact amount released, we showed that it is su fficient to prevent bacteria growth in non-accelerated studies.

Antimicrobial tests using both Gram-positive and Gram-negative bacteria showed that an antimicrobial response was found for 2, 5, and 10 mol.% incorporated silver-doped nHA. In a contact agar di ffusion assay, both the silver-containing materials were found to produce an area of inhibition, whereas hydroxyapatite by itself does not [35]. Looking at the reduction of bacteria in solution, both bacteria were reduced beyond detection within 96 h. In our study we found silver was far more effective against *S. aureus* than against the *E. coli* bacteria. This was unexpected, as the thicker wall of the Gram-positive *S. aureus* is arguably better defended against the uptake of silver particles than thinner Gram-negative cell walls [36]. Resistance to silver can be acquired by bacteria [37], although it is a relatively broad spectrum medicant. This is significant, as *S. aureus* is among the most common pathogens found in bone infections [1,2]. Recent studies have shown that treating the motile *S. aureus* in an infected 3D tissue model is more challenging than treating a monolayer bacteria assay [38]. Therefore, this would be the logical progression for the testing of this material. One side e ffect of silver is that it can also be toxic to mammalian cells at elevated concentrations, which we investigated with our sca ffolds.

The toxicity of the sca ffolds was related to both the silver concentration and on the cell type being tested. Against the standard cell line of mouse fibroblast 3T3 cells there was a small indication of toxicity with sca ffolds containing 10 mol.% silver. When using a primary rat MSCs there was clear toxicity detected with this highest concentration of silver. This shows the importance of not just using a single cell line to determine toxicity of a material but also of the correct choice of relevant cells. For a more in-depth review of the di fferent modes of action of silver nanoparticles on various cell types, see Zhang et al. [39]. Using 3T3 cells is a useful method of comparing relative toxicities between di fferent laboratories as they are well known and utilized in international standards for cytotoxicity (ISO 10993-5). The primary MSCs are more di fficult to standardize between labs and are, therefore, harder to use to compare toxicity. They do give a more realistic insight on what levels of toxicity might be harmful in the environment the medical device is intended for. From the results presented, 5 mol.% silver-doped material at 20 wt. mol.% to PCL had an antimicrobial impact without being toxic to mammalian MSCs.

Using a longer study of 21 days, the 10 mol.% silver nHA sca ffolds maintained a viable and growing cell population on par with the plain PCL material. Hydroxyapatite appeared to boost the metabolic activity of the MSCs by day 21 with the combination of 5 mol.% silver and hydroxyapatite, resulting in even higher levels of cell activity. Increasing the amount of silver to 10 mol.% did not increase the trend of increased proliferation. The initial high release rate from the material was potentially too toxic for the cells early on but, once the release level had fallen to a consistent 50 mol.%, surviving cells would then have been able to proliferate. After this period, the metabolic activity of cell on the 10 mol.% silver nHA scaffold increased at a similar growth rate to other materials. This is highly advantageous in a wound-healing application as it would allow an initially high burst to clear bacteria, which then lowers to allow the mammalian cells to begin the healing process.

Hydroxyapatite is widely reported to increase MSC differentiation into osteoblasts [40–42] so we focused on investigating the impact of silver on this process showing increased ALP production. Huge disparity can be found in the literature on whether silver has a significant impact on differentiation of MSCs. Some authors find that silver nanoparticles have no impact on MSC differentiation [43,44]. Other papers report on an impact of silver nanoparticles on MSC differentiation down the osteogenic lineage [45,46]. Conversely, some authors argue that silver particles can cause adipogenic differentiation [47] of MSCs, whereas others have found no impact at all on adipogenic differentiation [48]. One aspect that varies between these papers is the size of the silver nanoparticles, which might be a reason between the widely different reporting of the impact of silver on differentiation. Powers et al. found impact on neurodevelopment from silver on zebrafish PC12 cells, which was very dependent on both silver composition and silver nanoparticle size [49]. Another variation between papers is the wide range of different MSCs that are tested for differentiation, from both different tissue and animal origins. As even MSCs from different tissue origins vary in behavior, it is not surprising that different conclusions have been reported by multiple studies [50]. We found that 5 mol.% silver had the highest ALP production at day 21, although there was no significant difference to 10 mol.% silver. It is possible that either diminishing returns or an already reached optimum concentration of silver is the reason for no further increase at 10 mol.% silver. Another explanation is that the slightly more toxic impact of 10 mol.% silver offsets any additional osteogenic stimulation. Our study demonstrated that silver-doped nHA incorporated into PCL scaffolds may have a positive effect regarding MSC differentiation along an osteoblastic lineage due to increased levels of ALP production.

Limitations to the study include no mapping of elements within the electrospun membrane containing nHA. However, Figure 2b does show evidence that the material is present within the electrospun membranes but the precise location of the nHA is not known. Another limitation is that, while using bacteria typical of infections in bone, neither of the isolates was taken from a patient bone infection. Another limitation on the microbiology side is the simple nature of the tests used; it should be considered that (along with other disparities) osteomyelitis is associated with a hypoxic environment, which was not reflected by our experiments. Further work is needed to address these issues, including the use of a bone infection model to more accurately reproduce the conditions found in vivo. In a recent study we found that a material with antimicrobial impact on planktonic bacteria and simple biofilm models was then not effective on *P. aeruginosa* in tissue-engineered skin models, believed to be due to the bacteria motility [24]. This is something we intend to investigate in future work.

This research found that silver nHA containing scaffolds has the potential to act as an antimicrobial scaffold while stimulating bone tissue regeneration. The scaffolds degrade slowly and release a continuous amount of silver into the local environment over time. They are nontoxic to mammalian cells at a concentration that is toxic to both Gram-negative and Gram-positive bacterial strains. Investigating the impact of silver on MSC differentiation has suggested that higher values of silver promote MSC differentiation.
