On the function of biosynthesized cellulose as barrier against bacterial colonization of VAD drivelines

Biosynthesized cellulose fabrication and characterization

The Surface Micro-Engineered BC used in this study was synthesized by a strain of the bacterium Acetobacter xylinum in static culture. The wild type Acetobacter xylinum strain ATCC-700178 (LGC Standards, Wesel, Germany) was used for BC fermentation¹⁷. The bacteria were grown in a medium prepared as reported in Table 1 and sterilized by autoclaving. The resulting BC membranes were made of a three-dimensional network of randomly arranged cellulose nanoribbons, forming a multi-layered hydrogel with high porosity¹⁷. The average pore diameter was < 500 nm.

The BC membranes were harvested after 1 week of incubation at 27 °C. BC purification was performed as reported in Bottan et al.¹⁷. Briefly, at the end of the culturing period a thick (3–4 mm) cellulose layer was formed. BC substrates were then harvested. To remove bacteria from the BC the pellicles were washed in NaOH 1 M for 80 min at 80 °C, and subsequently in deionized (DI) water at room temperature (RT) until neutral pH was reestablished. The BC substrates were fully hydrated upon harvesting from the bacterial fermentation culture. This stage is defined as nondehydrated bacterial cellulose (NDH). The NDH substrates were then washed and dehydrated overnight at RT. Dehydrated cellulose substrates (DH) were then rehydrated with DI water. Finally, the rehydrated substrates (RH) were autoclaved (121 °C, 1.1 bar for 15 min) and stored in PBS at 4 °C. Membranes had a residual endotoxin content < 20 EU/device, as prescribed for fully implantable medical devices¹³. In particular, sterile biosynthetic cellulose sheets measuring 10 × 20 cm were used for the animal experiments.

The surface porosity (Φ) was measured on planar images at high magnification (25,000×) as the ratio of the area of pores divided by the area of the field of view (Supplementary Fig. 1). The area of the pores was calculated using the Analyze Particle tool of ImageJ on a binary mask obtained from the SEM image setting a signal intensity threshold. To obtain a value for the average pore diameter, Φ was divided by the number of individual pores detected from the processed image. This evaluation was performed on multiple fields of view to obtain a standard deviation for the average pore diameter.

After preparation, the BC substrates were characterized using scanning electron microscope (SEM) imaging. To prepare the samples for SEM, the substrates were washed twice in Milli-Q water, and then rinsed for 10 min in increasing concentrations of filtered ethanol (30, 50, 70, 90 and 95%). They were then rinsed twice in 100% ethanol for 15 min each. Ethanol dehydration was followed by gradual replacement with hexamethyldisilazane (Sigma-Aldrich) that was let to evaporate in a fume hood overnight. Samples were finally coated with a 5 nm-thick film of gold/palladium (60/40 wt%). The substrates were imaged using a JEOL JSM-7500FA scanning electron microscope (SEM), equipped with a cold field emission gun (FEG). Initially, low magnification images at 30° tilt angle were acquired, to allow for perspective 3D imaging. The BC nanofibers were imaged at higher magnification and resolution, operating the machine at an acceleration voltage 2 kV in order to minimize any possible beam damage effect. In both cases the SEM imaging was performed by collecting secondary electron (SE) signal.

In vitro permeation assay

A custom-developed setup was exploited for the permeation assay (Fig. 3). A test membrane (i.e. the BC membrane) was mounted to separate the two chambers (Fig. 3A). The upper chamber was then filled with a stabilized suspension of 2 µm microparticles (SPHERO Fluorescent Light Yellow Particles 1%w/v 1.97 µm, Spherotech Inc.), while the lower chamber was filled with an equivalent volume of pure solvent. After 24 h incubation at room temperature, the test membrane was retrieved and the number of beads permeating from the upper to the lower chamber was evaluated at different time points by inspecting the membrane surface by means of bright-field microscopy.

For bacterial colony formation, cubes of agar with lateral size of 1 cm were prepared and inoculated with a solution of S. aureus (Fig. 3C). The agar cubes were then leaned on a BC membrane separating them from an agar plate. Samples were incubated overnight at 37 °C and the colonization of the agar plate surface (i.e. by bacteria crossing the BC membrane) was evaluated. BC membranes in which an array of holes was created by puncturing with a syringe tip were included as internal controls, following the same experimental scheme.

Experimental animals

The animal study was performed on adult non-gravid, horned and dehorned female goats (German improved white goat/Preclinics GmbH) on the basis of the governmental permit G 0099/18 issued by LaGeSo (Berlin, Germany) and according to the guidelines of the EU directive 2010/63, the German animal protection act (TierSchG) and the regulation for the protection of laboratory animals (TierSchVersV). The animal study complies with the ARRIVE guidelines.

Large animals were necessary for this study since the implanted s.d.m. (i.e. the external silicon layer of commercial L-VAD drivelines) measured 6–10 cm in length. Goats represent the most adequate model. They can be easily trained and handled. Using protective vests, self-mutilation of the implants can be prevented even in group housing. For this study, the protective vests for maintenance of proper wound hygiene were fabricated based on the design published by Grosshauser et al.²⁰ (Fig. 1).

A total of eight (n = 8) animals were included in the study. Animals were divided into two groups, a 6-week (animals Nr. 1–4) and a 12-week (animals Nr. 5–8) endpoint group based on the time span between the implantation and explantation of the s.d.m. Each animal (6-week and 12-week group) received four s.d.m. (n = 4) to reduce the total number of animals. On both flanks of the animal, one control s.d.m. without BC coating and one BC covered s.d.m. were implanted following a staggered scheme. In total, each group of animals received sixteen individual s.d.m. (n = 16), i.e. eight control s.d.m. (n = 8) and eight BC covered s.d.m. (n = 8). No animal was excluded during this study.

Housing, husbandry and animal welfare

Animals were allowed to adapt to the new environment, personnel and the protective vests for 2 weeks before the surgery. In addition, all animals were trained to walk the path from the stable to the room where the anesthesia induction was performed and were acclimated to the inhalation masks. Training as well as later induction of anesthesia was performed in groups of two animals. This approach was chosen to reduce stress to individual animals. In particular, induction of anesthesia directly via Isofluran application using an inhalation mask rendered premedication by intramuscular injection unnecessary.

In general, animals were accommodated in stables in groups of 2–4. Each stable was equipped with an automated water dispenser and appropriate enrichment. Animals were fed with hay and feed pellets. Upon study-associated manipulations or health checkups, animals were additionally fed with carrots and apples. The bedding consisted of a combination of sawdust and straw. Pens were cleaned and disinfected once a week. Temperature was kept at 18 ± 2 °C and air humidity at 55 ± 10%. Stables were illuminated by daylight, ensuring a natural circadian rhythm.

All animals were inspected daily by veterinarians and animal care attendants. Each animal was weighted weekly. Post implantation, wounds were inspected regularly at intervals ranging from daily to once a week. The frequency of wound controls and the associated dressing changes were primarily dependent on the condition of the wound dressings. General health and wound condition were documented at every inspection.

Perioperative management for implantation

All animals were fasted for 12 h before anesthesia induction. Anesthesia was induced by inhalation of Isoflurane (3.0–5.0%) using a facemask connected to the respirator. Afterwards, an IV access was established on an ear vein and a distal leg vein. Sedation was intensified by an IV bolus (10–15 mg/kg) of Sodium Thiopental. Following endotracheal intubation, anesthesia was maintained with inhaled Isofluran (0.8–1.5%) and IV Fentanyl (1–5 µg/kg bolus/1–5 µg/kg/h). All animals received a Fentanyl (75 µg/h) skin patch to one foreleg before surgery. Moreover, an oropharyngeal temperature probe and a ruminal tube were placed. All animals received preoperatively a single shot of Ampicillin/Sulbactam (2000 mg/1000 mg) intravenously. Continued antibiotic prophylaxis until healing of the wound was ensured by an intramuscular injection of Amoxicillin (15 mg/kg) before the end of surgery. The injection with Amoxicillin was repeated at 48 and 96 h after surgery. During anesthesia heart rate and respiratory rate were monitored continuously. Proper mechanical ventilation was ensured by capnometry.

Implantation procedure

Commercial s.d.m. of HeartMate3 (HM3; Abbott, Chicago, IL, USA) drivelines were used in this study. The external silicon layer of the drivelines (i.e. the s.d.m.) was divided in 6–10 cm long pieces. Previous studies indicated an increase in the risk of infection by a skin-velour interface compared to a skin-silicone interface, hence the velour portion of the s.d.m. was discarded in this study^21,22. Sterile BC sheets were unfolded on wet sterile gauze and aligned in a manner that the rounded edge was situated at the right top corner. This approach ensures correct application of the cellulose to the s.d.m.

The BC membranes adopted for this study exhibited excellent tensile strength and proved very conformable¹³. These properties allow the material to wrap around the high curvature of the silicone drivelines used for this study. The application of the BC protective layer was performed by tightly rolling the membrane around the driveline silicone surface. Due to the properties of hydrated BC, the membranes were tightly adhering to the s.d.m. after application. The process was performed without damaging or tearing the BC layer. Afterwards, the protruding ends of the cellulose were trimmed to the exact length of the s.d.m. on both sides (Fig. 2A). After implantation of the BC-protected s.d.m. into the subcutaneous tunnel, the surrounding tissue naturally applied sufficient pressure to the implant to maintain the BC membrane in place without unfolding. A single suture at the end of each driveline tied the membrane to the silicone driveline outer layer and further stabilized the implant configuration. The correct configuration was retrieved upon explantation of the test articles for the ensuing analysis.

Implantation was performed for all animals in aseptic technique. Prior to surgery, both flanks of the animals were shaved using hair clippers. Thereafter the skin of both flanks up to the spine was thoroughly disinfected using a 7.5% povidone-iodine-solution. Surgical drapes were used to create a rectangular surgical field measuring approximately 15 × 20 cm on both flanks of the animal.

Two subcutaneous tunnels were created on each side of the animal by incising the skin at four points with a scalpel and subsequent atraumatic preparation in the subcutaneous tissue with surgical scissors. Each tunnel measured 5–9 cm in length. The two tunnels on the same animal side were approximately 4–6 cm apart. Implantation of control s.d.m. and BC covered counterparts was performed by channeling through the subcutaneous tunnels and securing them with a non-absorbable 2–0 Ethilon (Ethicon, USA) purse-string suture at each exit site (Fig. 2B, C). For additional stability, the s.d.m. ends protruding from the subcutaneous tissue were tied to the sutures securing the exit sites (Fig. 2B).

Dressing changes

Dressing changes were performed aseptically either by using a touchless technique or by wearing sterile surgical gloves. Adhesions and contaminants were removed using sterile 0.9% saline solution and sterile compresses. Wounds were covered using sterile compresses and adhesive bandage. Edges of the bandage were further secured using an extra bandage layer. The first wound control and dressing change was performed within the first three postoperative days. Consecutive dressing changes were performed regularly every 1–7 days depending on the integrity of the wound closure material.

Sample collection

Animals were euthanized directly after being under deep anesthesia induced by inhalation of Isoflurane (5.5%) followed by IV administration of 2500 mg Sodium Thiopental, 0.5 mg Fentanyl, 2 mg Pancuronium Bromide and 60 ml of 7.45% potassium chloride solution. The s.d.m. and the adjacent tissue specimens were then gathered directly.

An approximately 3 cm thick (dissected to the muscle fascia) square tissue section (10 × 10 cm) was cut out of the skin utilizing a scalpel. Importantly, the tissue size was selected such that the edges of the sample were at least 2 cm away from the implant at all points. Once explanted, the two tissue specimens (from both flanks of the animal, containing all 4 s.d.m.), were transferred to another sterile instrument table for further dissection and sample collection. Specifically, each sample (s.d.m. and surrounding tissue) was dissected longitudinally into three sections. Two sections contained the exit sites (left and right, where the s.d.m. exits the skin) and the middle section containing the subcutaneous part of the implant. The middle section measured approximately 1 cm in length.

The middle section was cut longitudinally in order to separate the s.d.m. from the adjacent tissue. Control s.d.m. were freed from the tissue and transferred into 15 ml Falcon tubes filled with 3 ml sterile 0.9% NaCl solution. BC-covered s.d.m. were separated prior to transfer into 15 ml Falcon tubes filled with 3 ml sterile 0.9% NaCl solution. Therefore, the BC was dissected and peeled off the corresponding s.d.m. This separation resulted in three distinct groups for microbiologic testing: (1) control bare s.d.m. (B-s.d.m.), (2) BC-coated s.d.m. deprived of BC (BCB-s.d.m.) and (3) BC. Transfer of implant specimens into Falcon tubes was performed under sterile conditions.

Microbiology

Wound swabs collected from the subcutaneous tunnel and corresponding exit sites prior to the s.d.m. implantation were sent to a specialized veterinarian laboratory (LABOKLIN, Germany) for microbiologic testing.

Post-collection specimens were processed in a microbiology laboratory within 6 h. Falcon tubes containing the implants immersed in 0.9% NaCl were vortexed for 30 s (Vortex Genie 2, Scientific Industries, Bohemia NY, USA), followed by sonication in an ultrasound bath at 40 kHz and 0.2 W/cm² (BactoSonic/BANDELIN electronic GmbH & Co. KG) for 1 min and another 30 s vortexing. Finally, 100 µl of the sonication fluid were plated on a brain heart infusion (BHI; BD, Le Pont de Claix, France) agar and incubated for 24 h at 37 °C. Sterile not implanted BC (same batch as the implanted BC) and a sterile driveline mantle of a HeartWare HVAD (HW; Medtronic, Minneapolis, MN, USA) driveline were used as negative controls.

Software

The cartoons in Fig. 2 were created with Biorender.com.