A Niclosamide-releasing hot-melt extruded catheter prevents Staphylococcus aureus experimental biomaterial-associated infection

Biomaterial-associated infections are a major healthcare challenge as they are responsible for high disease burden in critically ill patients. In this study, we have developed drug-eluting antibacterial catheters to prevent catheter-related infections. Niclosamide (NIC), originally an antiparasitic drug, was incorporated into the polymeric matrix of thermoplastic polyurethane (TPU) via solvent casting, and catheters were fabricated using hot-melt extrusion technology. The mechanical and physicochemical properties of TPU polymers loaded with NIC were studied. NIC was released in a sustained manner from the catheters and exhibited in vitro antibacterial activity against Staphylococcus aureus and Staphylococcus epidermidis. Moreover, the antibacterial efficacy of NIC-loaded catheters was validated in an in vivo biomaterial-associated infection model using a methicillin-susceptible and methicillin-resistant strain of S. aureus. The released NIC from the produced catheters reduced bacterial colonization of the catheter as well as of the surrounding tissue. In summary, the NIC-releasing hot-melt extruded catheters prevented implant colonization and reduced the bacterial colonization of peri-catheter tissue by methicillin sensitive as well as resistant S. aureus in a biomaterial-associated infection mouse model and has good prospects for preclinical development.

Material preparation. TPU was loaded with NIC at 2, 5 and 10% (w/w) using the solvent casting approach adopted from the work of Shaqour et al. 29 , as shown in Fig. 1A. First, the required amount of NIC was suspended in chloroform and sonicated for 30 min. Then, a magnetic stirrer was used to homogeneously distribute the NIC particles for 30 min. Subsequently, the TPU was added to the system while stirring and the suspension was left overnight to dissolve the polymer. The ratio between TPU and chloroform was 12.5% (w/v). The polymer solution with suspended NIC was then poured into a petri dish with a diameter of 200 mm to allow the chloroform to evaporate. Finally, the casted films were vacuum dried for 3 days at 25 °C and 50 mbar.
Fibers and catheters were extruded using an in-house single screw extrusion setup attached with an e3d v6 stainless steel nozzle (E3D-Online, United Kingdom) with a diameter of 0.8 mm or a specially in-house designed and 3D-printed coaxial nozzle 30 (Fig. 1B). To extrude the fibers and catheters, first, cast films (with or without NIC) were cut into stripes and directly fed into the extruder. Subsequently, fibers and catheters were extruded at a temperature of 180 °C with a screw extrusion speed of 75 rpm. The extruded fibers were cut into 4 cm segments and used to evaluate the effect of NIC loading on the mechanical properties of the TPU, while the extruded catheters segments (0.5 and 1 cm length) were used for material characterization and for microbiological characterization. Catheter segments for microbiological testing were further sterilized by incubation in ethanol 70% for 1 min followed by air-drying for 30 s 29 . Microscopy. Catheter segments were inspected using an S9i microscope (Leica, Belgium). Catheters from each group were imaged and then analyzed using ImageJ software version 1.52a 31 (https:// imagej. nih. gov/ ij/) by calculating the equivalent diameter (Eq. 1) of the measured area for the inner and outer circles for each catheter (n = 6 for each group) 29 .
Mechanical testing. Tensile tests were performed according to ISO 527-1 standard 32 on an AGS 5 kND machine (Shimadzu, Germany). The tensile rate was set to 20 mm/min and the distance between grips was 20 mm. Extruded fibers (n = 5 per group) were used for this test. Then, the tensile stress at 100% strain was measured.
(1) Equivalent diameter = 4 * Area π Fourier-transform infrared spectroscopy. NIC powder before and after heating to 180 °C was examined using Fourier-transform infrared (FTIR) spectroscopy to study the effect of heat on NIC molecules. Moreover, to study the interactions between NIC and the TPU molecules, FTIR analysis was conducted on the produced catheters. The FTIR spectrometer used (Nicolet 8700, ThermoScientific, USA) was equipped with a diamond attenuated total reflectance (ATR) accessory. All ATR-FTIR spectra were recorded in the 400-4000 cm −1 range at room temperature. The spectral resolution and accuracy were 4 cm −1 and ± 1 cm −1 , respectively.
X-ray diffraction. X-ray diffraction (XRD) analysis was performed to analyze the solid-state characteristics of NIC pure powder, non-loaded TPU, and NIC-loaded TPU catheters. The XRD analyses were run in transmission mode on PANalytical X'Pert Pro X-Ray Diffractometer (PANalytical B.V., The Netherlands) equipped www.nature.com/scientificreports/ with an X'Celerator detector using a standard XRD method. The instrumental parameters used are listed in the supplementary Table 1.
In vitro drug release assay. Catheter segments loaded with NIC (2, 5 and 10% (w/w)) and non-loaded, with a length of 1 cm were weighted using a microbalance (Sartorius, Germany). NIC is a hydrophobic drug, and therefore is poorly soluble in aqueous solutions such as PBS 33

Evaluation of antimicrobial properties of TPU-catheters loaded with NIC.
Catheter segments loaded with 2, 5, and 10% (w/w) NIC and non-loaded catheter segments were analysed for antibacterial activity (release) and prevention of biofilm formation using the MSSA, MRSA, MSSE and MRSE strains. Firstly, a modified Kirby-Bauer disk diffusion assay 35 was performed to determine the zone of inhibition (ZOI) of 0.5 cm catheter segments loaded with NIC and non-loaded catheters. Briefly, 1 to 3 colonies of each strain were incubated in 5 mL of TSB overnight at 37 °C and 120 rpm. Two hundred microliters of an inoculum of 1 × 10 6 CFU/mL was spread on blood agar plates with a cotton swab. The catheter segments were inserted vertically into the blood agar plates, which were then incubated at 37 °C for 24 h. The next day, the catheter segments were transferred to freshly inoculated blood agar plates and this step was repeated for 10 days. Each day the resulting zones of growth inhibition (including catheter diameter) were measured in mm.
Secondly, 1 cm catheter segments were placed in tubes containing suspensions of each of the 4 bacterial strains (approx. 5 × 10 6 CFU/mL) in 1 mL of TSB. Catheter segments were incubated in the bacterial suspensions at 37 °C and 120 rpm for 24 h. After the incubation, 2 measures of bacterial growth were quantified: (1) planktonic bacterial growth in the medium; and (2) biofilm formation on the catheter surfaces.
To assess the antibacterial activity, planktonic cells in the medium were enumerated by quantitative culture, performed as follows: aliquots of the suspensions were taken, tenfold serially diluted, and the dilutions and undiluted suspensions plated on tryptic soy agar (TSA) plates; these plates were incubated at 37 °C for 24 h and CFUs were determined. Similarly, biofilm formation was quantitatively measured by enumerating the viable cells attached to the surface. In brief, catheter segments were removed from the medium and rinsed 3 times in PBS to remove planktonic cells. Immediately thereafter, catheter segments were transferred to 1 mL of PBS, vortexed for 30 s, and sonicated for 15 min using an ultrasonic bath (Branson CPX2800-E, 40 kHz) to detach and disperse adherent-biofilm cells. This procedure does not affect viability of the bacteria. The bacteria recovered from the catheter segments were quantitatively cultured. For each experiment, triplicates were analyzed for each sample (NIC-loaded and non-loaded catheter segments).
Finally, scanning electron microscopy (SEM) was performed to visualize the morphological changes of MSSA bacterial cells adhered to NIC loaded and non-loaded catheter segments. As previously described in this section, bacterial suspensions were incubated with NIC loaded and non-loaded catheter segments until the washing steps with PBS. Prior to SEM imaging, samples were fixed in a solution of 4% (v/v) paraformaldehyde supplemented with 1% (v/v) glutaraldehyde (Merck, USA) overnight at room temperature. Samples were rinsed twice with demineralized water for 10 min and dehydrated in a graded ethanol concentration series from 50 to 100% of ethanol. To reduce the sample surface tension, samples were immersed in hexamethyldisilane (Polysciences Inc., USA) overnight and air-dried. Before imaging, samples were mounted on aluminum SEM stubs and sputtercoated with a 4 nm platinum-palladium layer using a Leica EM ACE600 sputter coater (Microsystems, Germany). Images were acquired at 8 kV using a Zeiss Sigma 300 SEM (Zeiss, Germany) at the Electron Microscopy Center Amsterdam (ECMA; Amsterdam UMC, Amsterdam, The Netherlands). Of each tube, 6-8 fields were inspected and photographed at magnifications of 100× and 500×. Catheter segments loaded with 10% of NIC were not included in the results as the material was partially degraded by the fixation procedure for SEM visualization. www.nature.com/scientificreports/ In vivo evaluation of TPU-catheters loaded with NIC using a murine BAI model. The efficacy of NIC-loaded catheters to prevent infection was evaluated in vivo using a murine BAI model. As S. aureus is one of the most frequent pathogens in medical device-related infection 5,6 , the MSSA and MRSA strains were used for the in vivo evaluation of the NIC loaded catheters. All experiments involving animals were carried out in accordance with the European directive 2010/63/UE governing animal welfare and protection, which is acknowledged by the Italian Legislative Decree no 26/2014, the company policy on the care and use of laboratory animals, and the ARRIVE guidelines. All the studies were revised and agreed by the Aptuit Committee on Animal Research and Ethics (Animal Welfare Body) and approved by Italian Ministry of Health (authorization n. 51/2014-PR). Male CD-1 mice (Charles River, Italy), weighing between 18 and 20 g, were housed in solid bottomed plastic cages with sawdust litter at a temperature of 20-22 °C, a relative humidity range of 45-65%, and lighting from approximately 06:00 to 18:00 h daily. Mice were fed a standard maintenance diet (A. Rieper SpA, Italy) and drinking water was filtered from normal domestic supply. Diet regimen was ad libitum. A period of acclimatization of 5 days was implemented before any experimental procedure. Animals were monitored during the entire period of the studies and clinical signs were recorded.
Briefly, the mice were anesthetized using inhaled isoflurane at 2.5% and the back of the mice was shaved using an electric razor and cleaned using an aqueous solution of benzalkonium chloride 0.175% (v/v). Two small incisions were made in both flanks of the mice and two subcutaneous pockets were created using forceps, a 1 cm catheter segment was inserted into each pocket and the incisions were closed using surgical staples (Autoclip wound closing system, Clay Adams, USA) under sterile conditions. Each mouse received two catheter segments with the same concentration of NIC, either non-loaded, 2 or 5% (w/w) NIC loaded catheters (n = 7 mice per group, so n = 14 catheter segments per group). Three additional mice were added to each experimental group and dedicated to histological analysis on day 3 post-infection, similarly at the same time point, a group comprised of mice implanted with TPU catheters in the absence of infection was also added for comparison in the histological analysis (Fig. 2).
Animals were anesthetized using inhaled isoflurane (2.5%) and the blood collected by terminal cardiac puncture at 1, 3, 7 or 14 days post-infection. Blood, catheter segments and the respective surrounding tissue were collected. Biopsies were aseptically retrieved using a 12 mm diameter biopsy punch and accordingly processed for bacterial load determination: catheter segments were separated from their surrounding tissue and rinsed3 times in 0.9% saline solution to remove non-adherent bacteria. Next, catheter segments were transferred into tubes containing 1 mL of sterile PBS, vortexed for 30 s and sonicated for 15 min using an ultrasonic bath (Branson CPX2800-E, 40 kHz) to detach and disperse biofilm cells from the catheter surfaces. Recovered bacteria from the catheters were then quantitively cultured. Surrounding tissue was homogenized in 1 mL of PBS using a Precelys tissue homogenizer device (Bertin Technologies, France). Tissue homogenates were quantitatively cultured to In vivo efficacy evaluation of non-loaded, 2 and 5% (w/w) NIC loaded TPU catheters using a murine BAI model. Two 1 cm catheters segments were implanted subcutaneously in the mice and a bacterial challenge (10 6 CFU in 100 μL) of either MSSA or MRSA was injected intraluminally into each catheter. Blood, catheter segments, and the surrounding tissue were collected of mice sacrificed on day 1, 3, 7 and 14 after challenge. Bacterial colonization of the catheter and the surrounding tissue, and the NIC concentration in plasma and in the tissue surrounding the catheter were quantified. www.nature.com/scientificreports/ enumerate viable bacterial cells residing in the tissue. In case of bacterial growth on day 14 post-infection in the experiment with MRSA, S. aureus colonies were retrieved from all of the experimental groups and the MIC for NIC was determined for all single retrieved colonies following the CLSI guidelines 36,37 . For histopathology analysis, biopsies were retrieved from histology dedicated mice at day 3 post-infection and preserved in 10% neutral buffered formalin. Following fixation, samples were embedded in paraffin, sectioned at a nominal thickness of 3-5 µm and stained with hematoxylin-eosin for histological analysis.
The NIC in vivo release profile in animals carrying NIC-loaded catheters was determined in plasma and in the tissue surrounding the catheter segments collected on day 1, 3, 7 and 14 post infection. First plasma was collected as follows: 1 mL of blood was placed in K3-EDTA collection tubes (Greiner Bio-One, USA) and centrifuged at 3000 RCF, 4 °C for 10 min. Fifty microliters of supernatant-containing plasma were then retrieved and mixed with 150 µL of HEPES 0.1 N and stored at − 20 °C until analysis.
Second, catheter surrounding tissue was processed as previously described and resulting tissue homogenates were further homogenized by enzymatical digestion: 1 mL of a solution of collagenase type 1 (8 mg/mL in PBS) and incubated for 3 h at 37 °C with agitation. Digested tissue homogenates were kept at − 20 °C until analysis. Prior to the analytical procedure, plasma and tissue homogenate samples were deproteinized with 2 volumes of acetonitrile containing diclofenac (200 ng/mL) as internal standard, vortexed and then centrifuged at 3000 RPM for 10 min. After centrifugation, supernatants were diluted (160 µL ultrapure water + 200 µL of supernatant) using a Hamilton Microlab STARlet small liquid handler (Hamilton Company, Switzerland). Levels of NIC in supernatants were determined by analyzing 2 μL of the supernatants using a Waters ultrahigh performance liquid chromatography system (UPLC; Milford, USA) coupled with an API400 (Applied Biosystems/MDS Sciex, USA) in tandem mass spectrometry mode (LC-MS/MS). Chromatographic separation was performed using a Waters Acquity UPLC BEH C18 30 × 2.1 mm, 1.7 µm analytical column. Mobile Phase A consisted of 0.1% (v/v) formic acid in water; mobile phase B was 0.1% (v/v) formic acid in acetonitrile.

Statistical analysis.
Quantitative data was expressed as the average ± standard deviation, with the number of samples stated in each experiment. The statistical analysis of all the in vitro and in vivo characterization experiments was performed using a one-way analysis of variance (ANOVA) with Dunnett's comparison test to evaluate post-hoc differences between the test groups with respect to the control group. Alpha was set at 0.05 for all analyses. Statistical analysis and data representation were performed using GraphPad Prism version 8.00 (GraphPad Software, San Diego, CA, USA; https:// www. graph pad. com/ scien tific-softw are/ prism/). *Indicates a p-value of 0.01 to 0.05, **indicates a p-value of 0.001 to 0.01, ***p-value of 0.0001 to 0.001, ****indicates a p-value < 0.0001.

Results
TPU films loaded with NIC and catheter/fiber extrusion. The NIC loaded and non-loaded TPU films were successfully produced. The non-loaded TPU films were transparent while the NIC-loaded ones were yellowish and opaque. TPU is highly soluble in chloroform 21 . The level of coloring and opaqueness increased with increasing NIC content. However, the NIC distribution was homogeneous as no agglomerates were spotted in the films.
The stress at 100% strain was used to assess the mechanical properties of the produced fibers and whether the addition of NIC affected TPU's mechanical properties. The extruded fibers from the 0.8 mm nozzle had a diameter of 0.91 ± 0.08 mm. There were no significant differences between the NIC-loaded and non-loaded fibers and tubes in term of diameter dimensions. The generated stress-strain curves for TPU and drug loaded samples are shown in supplementary Fig. 1. The stress at 100% strain was 11.1 ± 0.3, 10.7 ± 0.5, 11.4 ± 0.5 and 9.9 ± 0.4 MPa for the non-loaded and 2, 5 and 10% (w/w) NIC loaded TPU fibers, respectively (Fig. 3A). Only the 10% (w/w) NIC loaded fibers showed a slight but significant decrease in the stress (p = 0.0012) when compared to non-loaded TPU fibers, which may affect the mechanical properties of extruded catheters.
The produced catheters had an average outer and inner diameter of 1.24 ± 0.62 mm and 0.62 ± 0.01 mm, respectively, with no differences between the different types of catheters (Fig. 3B). The results of water contact angle measurements from non-loaded and NIC-loaded TPU catheters are shown in Fig. 3C. All catheters were hydrophobic (i.e. contact angle higher than 90°) 38 , with a contact angle of 106.5 ± 1.9°, 106.8 ± 1.5°, 116.3 ± 5.0° and 118.4 ± 2.5° for non-loaded, 2, 5 and 10% NIC loaded TPU catheters, respectively. The hydrophobicity of the NIC-loaded catheters increased with higher NIC loading, which is expected since NIC is considered a hydrophobic molecule 39 . However, only the 5 and 10% (w/w) NIC-loaded catheters (p < 0.0001) showed a significant increase in the contact angle when compared to non-loaded catheters.
Thermogravimetric analysis (Fig. 3D,E) showed the thermal degradation behavior for NIC powder, nonloaded and NIC-loaded TPU catheter segments. NIC and TPU had an onset degradation temperature (T onset ) of 260 °C and 295 °C, respectively. Measurements of the NIC loaded catheters showed that the T onset for NICloaded TPU was slightly higher than that of non-loaded catheters. T onset for 2, 5 and 10% NIC loaded TPU were 315 °C, 313 °C and 311 °C, respectively. Such increase in T onset was also reported when TPU was loaded with ciprofloxacin 29 or tetracycline hydrochloride 40 , which indicates that, like NIC, the presence of these compounds improved the thermal stability of the polymer.
Fourier-transform infrared spectroscopy and X-ray powder diffraction analysis. The FTIR analysis for the untreated NIC powder and NIC heated at 180 °C revealed no major changes in the spectra due to the heating (supplementary Fig. 2). The FTIR spectra of NIC powder, non-loaded and NIC loaded TPU are shown in Fig. 4A. The NIC spectrum showed characteristic bands at 3576.34 cm −1 , 3088.44 cm −1 and 1679.69 cm −1 corresponding to -OH, -NH, and C=O groups, respectively (Fig. 4B) 41  www.nature.com/scientificreports/ a typical distribution of absorption bands for this type of material 42 . The vibrational band at 3330 cm −1 corresponds to NH stretching vibrations. The peaks in the range from 3000 to 2800 cm −1 relate to CH asymmetrical and symmetrical stretching vibrations 42 . The double peak in the region of 1680 to 1740 cm −1 (C=O stretching vibrations), the bands above 1500 cm −1 (probably N-H bending vibrations and C-N stretching vibrations), and the strong bands in the region of 1300 to 1000 cm −1 (asymmetrical and symmetrical O-C-O stretching vibrations) are associated with the bonds of urethane groups 42 . The FTIR spectrum of NIC-loaded TPU tubes has characteristic peaks similar to those of the spectrum of non-loaded TPU tubes (Fig. 4A). Moreover, also the  www.nature.com/scientificreports/ new specific peaks due to the presence of NIC appeared in the spectra of the loaded with 5% and 10% (w/w) NIC. These peaks characteristic for NIC are practically invisible at the spectrum with the lowest NIC content. It was probably caused by the low intensity of these bands and the fact, that they were overlapped by the TPU peaks. For the higher content of NIC, the intensity of a peak of hydrogen-bonded N-H group in the amorphous region of TPU (3320 cm −1 ) significantly decreased and the band of free N-H group with no H-bond formation observed at 3450 cm −1 became stronger. Probably the drug particles weakened the force between H-bond groups of TPU and hindered the formation of a new H-bond, a similar effect was observed for the thermally processed composite of montmorillonite and polyamide 43 . Additionally, the strong band at 3535 cm −1 was observed, which can be assigned to the hydroxyl groups of NIC. In the comparison to the pure drug, the intensity of this peak increased, due to the amorphization of the drug 44 .
The XRD patterns of the non-loaded and NIC-loaded TPU tubes and pure NIC powder are reported in the Fig. 4C,D. NIC powder exhibited the characteristic peaks corresponding to the crystal form of the molecule 44 . On the contrary, the XRD pattern of TPU indicated its amorphous nature 45 . The TPU tubes loaded with NIC showed an XRD pattern without the sharp NIC diffraction peaks, which suggests a transformation of crystalline NIC to its amorphous form, this phase change was also reported by Jara et al. when hot-melt extruding NIC with the polymer poly(1-vinylpyrrolidone-co-vinyl acetate) (PVP-VA) at 180 °C for the production of enteric formulations with increased bioavailability 46 .

NIC thermostability and in vitro drug release.
Heating of NIC at 180 °C did not affect the antibacterial activity of the drug as judged from the MIC and MBC against MSSA and MRSA strains (Table 1). This indicates that NIC is thermally stable at the processing temperature and is expected not to lose its antimicrobial activity during extrusion at 180 °C.
The estimated amount of NIC (µg) loaded per 1 cm catheter segment was 258 ± 3, 573 ± 5 and 1110 ± 7 µg, for tubes loaded with 2, 5 and 10% NIC, respectively. All NIC loaded catheter segments showed an initial burst release of NIC in the first 24 h, followed by a gradual and sustained release over 27 days (Fig. 5A). Overall, the 2, 5 and 10% NIC-loaded catheters showed similar drug release kinetics over time (Fig. 5A,B). During the first 10 days, about 70% of NIC was released from the catheter segments loaded with 10% of NIC, while around 80% of the drug was released from catheter segments loaded with 2 and 5% of NIC. After 20 days, catheters loaded with 10% of NIC released 80% of the drug whereas a release of 90% was achieved for catheters with 2 and 5% of NIC (Fig. 5A,B).   (Fig. 5C-F). As expected, non-loaded catheter segments did not produce any ZOI. For the first 5 days, catheters segments loaded with 2% of NIC produced a ZOI between 4 and 8 mm depending on the bacterial strain. On subsequent days, the ZOI was gradually reduced reaching in some strains a minimum diameter of 2 mm at 10 days. Catheter segments loaded with 5% of NIC exhibited a larger ZOI, ranging from 8 to 12 mm during the first 5 days. On the last day of experimentation, the ZOI produced for all the strains tested ranged from 6 to 10 mm. Compared to catheters with other loading percentage, catheter segments loaded with 10% of NIC showed the largest inhibition zones after 10 days. These results showed that released NIC from the catheters exhibited in vitro antimicrobial activity overtime.

Evaluation of antimicrobial properties of TPU-catheters loaded with NIC. The in vitro antibacte-
rial activity of NIC-loaded catheters was assessed using MSSA, MRSA, MSSE, MRSE strains ( Fig. 6A-H).
Biofilm formation was significantly reduced on the surface of all NIC-loaded catheter segments, with reductions higher than 2 LogCFU for both tested S. aureus strains (Fig. 6A,B). Planktonic growth around all NICloaded catheters was reduced when compared against the non-loaded group (Fig. 6E,F). However, similar reductions on the bacterial load were observed for all the drug loadings, indicating a possible bacteriostatic effect of released NIC as the LogCFU/catheter values are similar to those of the inoculum. The biofilm formation for both S. epidermidis strains on all NIC-loaded catheters segments was significantly reduced (> 1.6 LogCFU reductions) (Fig. 6C,D) whereas corresponding planktonic bacterial growth was only reduced by catheter segments loaded with NIC at 5 and 10% (Fig. 6G,H). The reduction of the bacterial burden on catheter segments challenged with MSSA was confirmed by visualization with SEM ( Fig. 7A-C). In non-loaded catheter segments, a high number of bacterial cells were attached to the intraluminal surface forming a biofilm structure. In contrast, the catheter segments loaded with 2 and 5% of NIC showed only few bacteria adhered to the surface.
In vivo evaluation of TPU-catheters loaded with NIC in a BAI model. As the 10% NIC may affect the mechanical properties of TPU and to avoid potential toxic effects, in vivo efficacy of 2 and 5% NIC loaded catheters was assessed in a BAI model against the MSSA strain (Fig. 8A,B). On day 1, a significantly lower bacterial load was detected on both NIC loaded catheters (4.29 and 3.40 LogCFU/catheter for 2 and 5% loadings, respectively) with respect to non-loaded catheter segments (6.06 LogCFU/catheter) (p < 0.0001). Additionally, a significant reduction of 1.24 LogCFU in the bacterial counts was also observed in the surrounding tissue of 5% NIC loaded segments (p < 0.01). On day 3 and 7 post infection, bacterial burden of both NIC-loaded catheters and in the respective surrounding tissues was progressively reduced over-time. Finally, the maximum effect was www.nature.com/scientificreports/ observed after 14 days when the bacterial load was greatly reduced (~ 5.7 Log CFU reduction p < 0.0001) on both NIC loaded catheters which was in line with further bacterial reduction of ~ 4.2 Log CFU (p < 0.0001) in corresponding surrounding tissue (Fig. 8A,B). In a subsequent experiment, mice implanted with catheter segments loaded with 2 and 5% of NIC were challenged with the MRSA strain (Fig. 8C,D). On day 1 after infection, the bacterial load on 2 and 5% NIC catheters (4.29 and 3.40 LogCFU/catheter; p < 0.001 and p < 0.0001, respectively) was significantly lower if compared to unloaded catheters (6.06 LogCFU/catheter). Moreover, bacterial counts in the surrounding tissue of 5% NIC catheters was significantly reduced by a value of 1.23 LogCFU (p < 0.01), as similarly observed with the MSSA strain. On the following days, the bacterial burden on both NIC loaded segments and in the surrounding tissue gradually decreased over time until day 14 (Fig. 8C,D). Few colonies of 2 and 5% NIC groups, isolated at day 14, showed increased MIC values for NIC (2 colonies = MIC 1 µg/mL, 1 colony = MIC 2 µg/mL, and 1 colony = MIC 8 µg/mL). The MIC recorded for the original strain was replicated for most of the isolated colonies (i.e. 56 colonies MIC 0.25 µg/mL). Finally, no increase in terms of MIC values of NIC was observed in colonies isolated from non-loaded TPU groups.
Histology of tissue surrounding the catheters was performed on tissue samples from mice infected with MSSA and sacrificed on day 3 (Fig. 8E). In mice implanted with non-loaded catheter segments in absence of infection (control group), the inflammation reaction was characterized by a small number of neutrophils and macrophages, with minimal fibrin deposition restricted to peri-catheter tissue. In addition, no necrosis was observed in this group (Fig. 8E-1). On the other hand, in biopsies from mice with infected non-loaded catheter segments, the inflammation was characterized by the presence of abundant neutrophils and less macrophages which were embedded in abundant edema and fibrin. Extensive necrosis was observed around the catheter segments, variably extending into the surrounding tissues. Moreover, a full-thickness coagulative skin necrosis was observed in this group (Fig. 8E-2). Intralesional bacterial biofilm formation was observed within the necrotic debris that surrounded the catheter segment. Additionally, features such as epidermal hyperplasia, furunculosis and skin ulceration were registered in the tissue. Animals with NIC-loaded catheter segments showed morphological changes similar to those present in the infected mice with non-loaded catheters, however the severity was not as marked. Additionally, in the same biopsies, we observed a moderate number of neutrophils and macrophages with moderate fibrin deposition. Furthermore, a minimal number of necrotic debris was found in the surrounding tissue and no bacterial aggregates were observed, which agrees with the observed quantitative decrease in the bacterial load in surrounding tissue.
The NIC distribution in situ in the surrounding tissue is shown in Fig. 8F. An initial NIC level of 407 ± 389 and 932 ng/g ± 922 of tissue was recorded in mice bearing catheter segments with 2 and 5% NIC, respectively. After 3 days, NIC levels of 781 ± 382 ng/g and 312 ± 206 ng/g (no significant difference between both groups) were recorded in the surrounding tissue of catheter segments loaded with 2 and 5% NIC, respectively. On the following experimental days, the in situ NIC concentration decreased in the surrounding tissue, reaching the lowest values of 65 ± 51 ng/g (NIC 2%) and 135 ± 62 ng/g (NIC 5%) on day 14. The in vivo level of NIC in plasma is shown in Fig. 8G. After 1 day, NIC plasma levels of 20 and 47 ng/mL were measured for catheters with 2 and 5% of NIC, respectively. On the consecutive time points, the NIC plasma levels for both NIC loadings gradually dropped over time, reaching levels at the limit of detection on day 14. www.nature.com/scientificreports/

Discussion and conclusion
Catheters are essential medical devices which are mainly used for diagnosis, prevention, and treatment of many diseases and health conditions 1 . Catheters are mainly fabricated by HME process in which medical grade polymers are fed into an extrusion machine, then heated up to melting and finally extruded through a nozzle 47 . Antiinfective catheters can be produced by loading the active pharmaceutical ingredient (API) within the polymeric matrix of the catheter. This can be done during the HME processes, however, the thermal stability of the loaded API should be taken into consideration to avoid any possible degradation 48 .
We developed a novel NIC eluting catheter adapting the methodology previously described by Shaqour et al. 29 to prevent ventilator-associated pneumonia. We used NIC, as a repurposed antibacterial drug, to minimize the potential development of antibiotic resistance which currently is a rising global health issue 49 . Moreover, the potential antibacterial activity of NIC has been tested against multi-drug resistant bacteria such as methicillinresistant S. aureus and S. epidermidis, two of the most prominent pathogens found in medical device-related infections 13,14 . Furthermore, NIC has not only been proven biologically active in different preclinical models but also not toxic in clinical trials 15,21,[50][51][52][53] . The method used for producing the catheters is a small scale HME production equipment. This system is widely used on a larger scale for mass production of conventional catheters used in medical practice nowadays. We believe that using such a widely used industrial process will ease the move toward mass production of the NIC-loaded catheters. www.nature.com/scientificreports/ We confirmed that NIC can withstand processing temperatures (up to 180 °C) without undergoing degradation 46 and any detrimental changes in its in vitro antibacterial activity. Moreover, NIC has a well-studied safety profile thereby it has been tested to treat different medical indications using preclinical models 24,52,[54][55][56][57][58] . Considered together, these results prove NIC's suitability to fabricate anti-infective medical devices by using not only HME but also different processes involving temperatures at least up to 180 °C such as melt extrusion 3D-printing technologies. After proving the suitability of NIC for HME applications, NIC-loaded TPU catheters were fabricated at loadings of 2, 5 and 10% NIC (w/w). Overall, the addition of NIC at 2 and 5% (w/w) did not alter the mechanical properties of the TPU fibers which would likely be the same for NIC-loaded catheters. The mechanical properties of the TPU are generally controlled by the ratio between the hard segment and the soft segment 59 . The TPU grade used in this study has a soft and hard segment in a ratio of 3:1. In order to overcome the change in the mechanical properties at higher NIC loading, different TPU grades with different ratios may be used. However, in this case more tests should be conducted to study the release profile of NIC from the matrix. The incorporation of NIC into the polymeric matrix of TPU catheters was confirmed by the appearance of new functional groups in the FTIR spectra. These changes in NIC-loaded TPU likely are due to covalent modifications, while the shift of spectral bands characteristic of the TPU or the immobilized NIC may indicate non-covalent interaction 60,61 . Moreover, NIC underwent a phase transition from crystalline to an amorphous form within the NIC-loaded catheters. This drug phase transition is likely caused by the combined effect of solvent casting and the HME process during catheter production 46,62 . The amorphous form of NIC improves the drug's bioavailability, which in turn is pivotal for a successful delivery of the drug within a host and to exert the intended antibacterial effect 62 .
Catheters loaded with NIC presented a steady drug release over 27 days. In similar drug-loaded implants, the drug release is mainly mediated by passive diffusion, which governs how the drug migrates towards the external media 63 . In drug delivery systems governed by diffusion, the release kinetics relies on the solubility of the drug in the polymer, the diffusion coefficient of the drug, the drug loading, and the degradation of the polymer 64 . In fact, we observed that the total amount of released NIC to the media was dependent on the drug loading.
The in vitro antibacterial properties of NIC-loaded catheter segments were demonstrated by both a disc diffusion assay and the quantification of biofilm-forming bacteria on catheter surfaces and planktonic bacterial growth in liquid media 65 . Not only did the NIC prevent biofilm formation, the apparent release of NIC in the medium inhibited the bacteria to grow. The released concentrations of NIC however did not kill the planktonic bacteria, since their numbers did not significantly decrease relative to the inoculum, suggesting a bacteriostatic activity of released NIC. The antibacterial activity of NIC against both S. aureus and S. epidermidis has been described to be dependent on drug concentration 13,14,18,66 . As NIC is partially soluble in aqueous solutions such as broth media 33 , the solubility of released NIC from catheters in TSB might be limited under the experimental conditions. Moreover, the bacteriostatic activity of released NIC among the tested bacterial strains is similar, which might indicate a similar drug exposure among the tested strains. Hence, free-solubilized NIC in TSB media was probably not high enough to exert a bactericidal effect on planktonic bacterial cells.
Based on the in vitro studies, we next investigated whether the protective activity of NIC-loaded catheters with 2 and 5% of NIC would translate to in vivo efficacy using a BAI model. In prior studies, this in vivo model has been proven useful to fast track the development of new antimicrobials and anti-infective medical devices to combat medical device-related infections 67 . In this in vivo model, the NIC-loaded catheters exhibited a similar antibacterial efficacy against both MSSA and MRSA strains for at least 14 days. The in vivo efficacy was reflected in significantly reduced bacterial colonization of both the catheter segments and the peri-catheter tissue.
From the in vivo experiments with MRSA, a limited number of colonies re-cultured from the tissue of mice with 2% or 5% NIC-loaded catheter segments displayed elevated MIC values. As the antibacterial indication is not approved for NIC, there are no data available of resistance concentration breakpoints for clinical susceptibility. The shifts in the MIC of NIC could be due to either emergence of resistance or tolerance, and in case of the later, it could be circumvented by delivering therapeutically attainable NIC concentrations to the infection site. Apparently, the elevated MIC had not resulted in dominance of this trait in the bacteria in the tissue since most of the retrieved colonies had lower MIC values. Resistance development of Gram-positive bacteria to NIC has to the best of our knowledge not been reported until now but should be kept in mind as a possibility of prolonged in vivo exposure to concentrations released, which are decreasing over time. Therefore, we strongly suggest performing NIC susceptibility assays with bacteria re-cultured from in vivo models with NIC as the antimicrobial agent. On the other hand, Gram-negative bacteria possess inherent resistance to NIC, mediated by efflux pumps and nitro-reduction 68 . Interestingly, it has been reported that co-administration of NIC and colistin exerted in vivo synergistic antibacterial efficacy against resistant Pseudomonas aeruginosa in a skin abscess infection model, which is a difficult to treat infection with standard antibiotics alone 68 . Therefore, the development of catheters containing NIC and synergistic antibiotics holds potential to decrease the likelihood of emergence of newly NIC tolerance/resistance while increasing the antibacterial efficacy against Gram-positive and Gram-negative bacteria.
The reduction of bacterial colonization in surrounding tissue was also confirmed by histological analysis, which is relevant since peri-catheter tissue is a niche allowing infecting bacteria to survive in medical devicerelated infections 69,70 . Also, NIC-loaded catheters hold the potential to elicit a complete anti-infective protection against bacteria coming from the skin or emanating from contaminated hubs 5,71,72 . In addition, our data suggest that NIC is rapidly released and retained into peri-catheter tissue, since the local tissue concentration was several folds higher than in plasma. Importantly these NIC plasma levels have been reported to not exert systemic toxicity 21,50,51 . The higher NIC levels in tissue indicate that NIC likely slowly diffuses through the tissue and into the blood vessels. From similar reported in vivo release kinetics, we can hypothesize that once in blood NIC undergoes a rapid systemic clearance which in fact will minimize the systemic concentrations of NIC 73 . In addition, considering the amount of NIC in the loadings of 2 and 5% per 1 cm catheter, and the initial mean www.nature.com/scientificreports/ weight of the mice (~ 20 g) at the beginning of the experiments, we approximately administered to the mice a NIC dose of 12.5 mg/kg and 31.25 mg/kg per a NIC loading of 2 and 5% respectively. Different preclinical studies have reported that daily subcutaneous NIC doses of 1 and 10 mg/kg and intraperitoneal NIC doses of 10, 20, 30 and 50 mg/kg did not exert any toxic effect in mice after several days of treatment 68,[74][75][76][77][78] . Moreover, the acute oral toxicity LD 50 value for mice is 1500 mg/kg while the subcutaneous LD 50 value is > 1000 mg/kg 79 . In its anthelmintic indication NIC is administered orally at a dose of 2000 mg/patient and is well tolerated by human patients 80 . Furthermore, the amount of NIC loaded in the catheters is gradually released into the peri-catheter tissue which modulates the real concentration of bioavailable NIC overtime and therefore minimizes the exposure to potential systemic toxic NIC levels. Although the BAI model does not completely mimic a catheterization in mice, this model offers relevant valid in vivo information on the functionality of NIC-loaded catheters to resist S. aureus infection in potential clinical conditions 67 . Moreover, our results suggest a potential clinical application for NIC-loaded TPU catheters, as NIC was able to combat infection by the MRSA strain tested, and MRSA strains are known to cause most of the life-threatening staphylococcal infections 53,54 . Furthermore, the approach followed in this research work could be applied for developing novel NIC-functionalized medical devices to prevent or treat staphylococcal infections such as those associated with wounds, heart valves, urinary catheters, intra-uterine devices, voice prostheses, endotracheal tubes, and prosthetic implants. In future studies it would be interesting to use combinations of NIC and other antibacterial agents to generate broad-spectrum antibacterial catheters. In conclusion, this research work serves as a proof of concept of the potential use of NIC to develop medical devices using high-temperature manufacturing technologies, for preventing medical device-related infections.

Data availability
The data that support the findings of the study are included in this published article (and its supplementary information files) and are available in the Zenodo repository, https:// doi. org/ 10. 5281/ zenodo. 61283 56.