BD-2 and BD-3 increase skin flap survival in a model of ischemia and Pseudomonas aeruginosa infection

The main aim of this work was to study the usefulness of human β-defensins 2 (BD-2) and 3 (BD-3), which are part of the innate immune system, in the treatment of infected ischemic skin flaps. We investigated the effect of transducing rat ischemic skin flaps with lentiviral vectors encoding human BD-2, BD-3, or both BD-2 and BD-3, to increase flap survival in the context of a P. aeruginosa infection associated with a foreign body. The secondary endpoints assessed were: bacterial counts, and biofilm formation on the surface of the foreign body. A local ischemic environment was created by producing arterialized venous flaps in the left epigastric region of rats. Flaps were intentionally infected by placing underneath them two catheters with 105 CFU of P. aeruginosa before the surgical wounds were hermetically closed. Flap biopsies were performed 3 and 7 days post-operatively, and the specimens submitted to immunohistochemical analysis for BD-2 and BD-3, as well as to bacterial quantification. Subsequently, the catheter segments were analyzed with scanning electron microscopy (SEM). Flaps transduced with BD-2 and BD-3 showed expression of these defensins and presented increased flap survival. Rats transduced with BD-3 presented a net reduction in the number of P. aeruginosa on the surface of the foreign body and lesser biofilm formation.


Materials and Methods
In all groups, flaps were intravascularly injected with a 100-µl solution of recombinant rat Vascular Endothelial Growth Factor-A that was left to act for 90 min. In the NaCl group, before closing the surgical wounds, one milliliter of a 0.9% sodium chloride solution was instilled under the flap into the vicinity of the silicone catheter segments. In the PA group, one milliliter of a 0.9% sodium chloride solution containing 10 5 CFU Pseudomonas aeruginosa was instilled under the flap into the vicinity of the silicone catheter segments. In the GFP, BD-2, BD-3 and BD-2 + BD-3 groups, besides the procedure described for the PA group, a solution containing a lentivirus coding for Green fluorescent protein, human β-defensin 2, human β-defensin 3 and human β-defensins 2 and 3 was injected in the flap's vessels, respectively. www.nature.com/scientificreports www.nature.com/scientificreports/ Bacterial growth. All bacteria were inoculated in Luria-Bertani (LB) broth and grown at 37 °C on a rotary shaker at 250 rpm, overnight. Exceptions were performed and are properly indicated.
Plasmids production. Competent Escherichia coli DH5α were transformed by heat-shock with pMD2.G and psPAX vectors, while competent E. coli Stbl3 were transformed with pLenti6.BD-2 and pLenti6.BD-3. Posteriorly, bacteria were grown in LB media, at optimal temperature in ampicillin presence, at 37 and 30 °C, respectively.
Pseudomonas aeruginosa -infection model. P. aeruginosa CECT 110 + strain (also known as P. aeruginosa ATCC ® 10145 ™ , which is widely used for antibiotic testing and for microbiological assays since 1947) was used to infect rats [31][32][33] . This strain was grown until the mid-exponential phase and it was serially diluted in 0.01 M PBS at a OD 600nm of 0.0001 corresponding to 1.0 × 10 5 CFU/ml. At all times, to confirm P. aeruginosa CFUs (colony forming units), spreadings of bacteria were performed in LB agar dishes and colonies were counted after overnight incubation at 37 °C.
Before performing the antimicrobials assays in the animal model of ischemia, tests were conducted to determine the ideal bacterial concentration. The concentrations initially used were: 10 3 ; 10 4 ; 10 5 ; 10 6 and 10 7 CFU/ml. We selected the concentration of 10 5 CFU/ml, since this concentration systematically resulted in clinically significant surgical site infection and partial flap necrosis in our model. Higher concentrations led to complete or nearly complete flap necrosis (10 6 CFU/ml; n = 2) and animal's death (10 7 CFU/ml; n = 3), whereas lower concentrations (10 3 and 10 4 CFU/ml; n = 3 for each concentration) did not produce clinically recognizable infection.
Lentivirus particles were produced according to ViraPower ™ Lentiviral Expression Systems (ThermoFisher ® ) recommendations, using HEK293FT cells (Invitrigen TM , ThermoFisher ® , R700-07). Recombinant BD-2 and BD-3 expression in HEK293 was confirmed by RT-PCR. The total RNA of cells lines was extracted using the removal, the depilatory cream was cleaned from the abdomen with warm saline. After shaving the abdomen and placing the animals on the operation table, body temperature was recorded with a rectal thermometer and the rats were kept on a heating pad (Skaldo, Ardes ™ ), in order to maintain a constant body temperature. A substantial amount of an alcoholic solution (Cutasept F ® , Hartmann ™ ) was sprayed over the operative site. The product was left in the operative site for at least 15 seconds. Application was repeated 3 times. After the last application, the disinfectant solution was left in contact with the operating field for at least 2 min before proceeding with the surgery. All surgical procedures were performed in strict sterility conditions. Surgical Model. Succinctly, a variant of the rat abdominal arterialized venous flap was used as model of ischemic flap (Fig. 2) 41,42 . This flap was transduced with an ex vivo infusion of a 100 µl solution containing approximately 4.7 × 10 9 plaque-forming-units of recombinant lentiviruses coding for Green Fluorescent Protein (GFP) or BD-2 and/or BD-3 43,44 . This model was combined with an adapted version of the rat foreign body infection model (Figs 1 and 2) 45 .
In all animals, under a surgical operating microscope, a 5 cm long and 3 cm wide fasciocutaneous flap was raised on the left side of the rat's abdomen 42 . Cranially, the flap was connected exclusively to the skeletonized lateral thoracic vein 46 . This vein was temporarily clamped. Caudally, the superficial caudal epigastric artery was ligated with an 8/0 Nylon suture. The superficial caudal epigastric vein (SCEV) was cannulated with a 27-gauge ophthalmic cannula (BD Bioscience ™ ). It was then perfused with a 100-µl solution whose composition varied amongst the experimental groups (Figs 1 and 2). In all cases the solution contained 5 µg of recombinant rat Vascular Endothelial Growth Factor A (VEGF; Immunotools ® ) in DMEM, in order to increase transduction efficiency 43 . The composition of the solution injected in the flap's venous system was as follows: NaCl Group (VEGF); PA Group (VEGF); GFP Group (VEGF + GFP coding lentivirus); BD-2 Group (VEGF + BD-2 coding lentivirus); BD-3 Group (VEGF + BD-3 coding lentivirus); BD-2 + BD-3 Group (VEGF + BD-2 + BD-3 coding lentivirus) (Fig. 1). After injection, SCEV was immediately clamped for 90 min, in order to leave the injected solution in contact with the flap's vascular system 43 .
Subsequently, vascular clamps were removed and the SCEV was arterialized by connecting it to the femoral artery. In order to achieve this, the SCEV was cut from the femoral vein with a 1-mm long cuff of adjacent femoral vein tissue. The ostium in the femoral vein was closed with a continuous Nylon 11/0 suture. The same suture line was used to perform a side-to-end arteriovenous anastomosis between the SCEV and the ventral flank of the femoral artery through a 1-mm long ostium previously created in the latter vessel. Interrupted stitches were used for this anastomosis. The lateral thoracic vein was preserved cranially to ensure outflow from this arterialized venous flap 42 .
Two 1-cm segments of sterile 14-gauge silicone catheters (Mediplus ™ ) were placed in the central aspect of the surgical wound. Surgical wounds were closed with a running 5/0 Nylon suture. Just before closing the lateral aspect of the skin wound, 1-ml of a 0.9% sodium chloride solution was instilled under the flap into the vicinity of the silicone catheter segments using an 18-gauge silicone catheter (Mediplus ™ ). In all groups except the NaCl group, the latter solution contained 10 5 CFU of P. aeruginosa. The running suture was closed immediately after injection, in order to avoid any spilling to the area adjacent to the flap (Fig. 2).
No anticoagulants or antibiotics were administered pre, intra or postoperatively. All surgical procedures were performed by the same surgeon (D.C.), in order to minimize inter-surgeon variability. Thermography. One hour after flap reperfusion, with the rat still under anesthesia, the rat's abdomen was submitted to infrared thermography with a FLIR ® E6 camera placed 25 cm above the abdomen. This evaluation intended to confirm the flap's relative ischemia compared to the contralateral side 42 . Rats were placed on their (2019) 9:7854 | https://doi.org/10.1038/s41598-019-44153-y www.nature.com/scientificreports www.nature.com/scientificreports/ backs for 10 min prior to this evaluation. Thermographic measurements were made at a constant room temperature (22 °C) and humidity (50%) 42,47 . Post-operative care and assessment. After thermographic evaluation was completed, a transparent Tegaderm ™ (3 M Deustschland GmbH ® ) transparent film dressing was applied over the flap and adjacent skin.
Following surgery, rats were kept in solitary rat cages and offered rat chow and water ad libitum.
Rats were assessed daily by the same blinded researcher, in order to reduce inter observer bias and variability 48 . The following parameters were evaluated: animal wellbeing, flap viability, and presence of complications.
On the third and seventh days postoperatively, rats were anesthetized as described above. Objective measurement of flap survival was performed on these days based on digital photographs, which were later analyzed by a blinded observer using the free Image J ® software 40,49 . The total flap area and the necrotic flap area were determined for each flap. The necrosis rate for each rat was defined by the proportion of the total flap area that was occupied by the necrotic flap area, being expressed as a percentage of the total flap surface area 50 .
After disinfecting the rat's abdomen as detailed above, on the third postoperative day, a sample measuring approximately 1 cm in length and 0.5 cm in width was surgically excised from the most lateral aspect of the caudal third of the viable flap. These specimens were subjected to histological analysis, as well as bacterial counts by both culture and real-time PCR. The integumentary defect created by the biopsy was submitted to hemostasis and closure with a continuous 5/0 Nylon suture taking part of the integumentary redundancy in this region. A transparent Tegaderm ™ (3 M Deustschland GmbH ® ) transparent film dressing was again applied over the flap and adjacent skin.
On the seventh day after the first surgery, the lateral aspect of the flap was elevated, thus exposing the underlying catheter segments. These were then carefully removed, taking care not to touch the rat's skin and immediately immersed in the fixative solution.
The flap was then removed for histological analysis, as well as bacterial counts by culture and real-time PCR (Supplemental Tables 2 and 3).
Rats were euthanized on the seventh postoperative day by exsanguination under general anesthesia after cutting the common carotid arteries and the external jugular veins 51 . Evaluation of flap transduction by fluorescence microscopy. Four rats were submitted to the procedures described above for the GFP group with the exception that no bacteria were instilled. Seven days after the surgery, the skin flap was harvested, fresh-frozen and observed under the fluorescence microscope. Two rats were submitted to an analogous the procedure, but no lentiviruses were used to transduce the flaps in these animals. Fluorescence microscopy images were obtained in the same manner, using the same image acquisition parameters, by two blinded observers.  Table 2]. Each reaction was performed in triplicate. Thermal cycling conditions were 95 for 20 seconds followed by 55 cycles of 95 for 3 seconds and 60 for 30 seconds. The gene expression was normalized to the endogenous control β-actin, which is known to have significant basal expression 52 . Gene relative expression was calculated using the 2 −ΔCT × 1000 formula, an adaptation of the 2 −ΔΔCt method described by Livak and Schmittgen [53][54][55] . This allowed us to calculate the number of mRNA molecules of the gene of interest per 1000 molecules of the endogenous control (β-actin).
Viable bacterial cell counts. Biopsies were collected in sterility and cut in small pieces (between 0.05 and 0.10 mm of maximum diameter). Next, they were macerated in 0.9%(w/v) NaCl (Merck ® ) at 100 mg/mL concentration for 5 min with a Pellet pestle (Sigma-Aldrich ® ). Serial dilutions were performed from the supernatant in 0.9%(w/v) NaCl (between 10 −1 and 10 −5 ) and were spread on LB dishes in duplicated. The dishes were incubated at 37 °C during 14 h followed by quantification of CFUs. The CFUs counted were verified by running a series of tests, namely, Gram stain (Merck ® ); MacConkey growth (Carl Roth ® ) and Oxidase test (BioMérieux ® ) according to the manufacturers' instructions.
To certify the inoculation dose of P. aeruginosa, the effective average bacterial inoculation dose under the flap was determined by bacterial growth and found to be (9.735 ± 0.120) X 10 4 CFU, which was similar to the initial inoculation concentration of 10 5 CFU/ml. There were no statistically significant differences between the initial bacterial inoculum in the different experimental groups.

P. aeruginosa quantification by real-time PCR.
Biopsies were collected in sterility, were immediately immersed in RNAlater ® (Sigma-Aldrich ® ), and incubated overnight at 4 °C. On the following day, the biopsies were cut in small pieces (between 0.05 and 0.10 mm) and stored at −80 °C.
Forty-five to 50 mg of excised tissue was used to extract genomic DNA by NZY Tissue gDNA Isolation Kit (Nzytech ® ) according to the manufacturer's instructions.
To quantify P. aeruginosa, universal primers were used for 16 S rDNA (Supplementary Table 3) 56 . The real-time PCR was performed using Applied Biosystems 7500 real-time PCR System (Thermo Fisher Scientific ® ).
Thermo-cycling program was 40 cycles of 95 °C for 20 s, 60 °C for 30 s and 72 °C for 40 s with an initial cycle at 50 °C for 2 min and 95 °C for 10 min. Next, a dissociation curve was performed at 95 °C for 15 s and a range between 60 to 95 °C.
Histological processing. Flap biopsies were collected from the lateral aspect of the surgical flap of the rat 3 and 7 days postoperatively, as described above. Specimens were incubated in 10% (v/v) Formalin. After paraffinization (VWR ® ), specimens were cut on a microtome (Leica ® ) as 3 µm thick slices. These slices were stained with Hematoxylin-Eosin (HE) and Masson's Trichrome (MT), according to the manufacturers' indications 57,58 . For immunohistochemical staining with anti-BD-2 and -BD-3, slices were prepared in an analogous fashion, although they were cut 4 µm thick.
Quantification of bacteria and leucocytes on the surface of the foreign body by scanning electron microscopy. Bacteria, leucocytes and phagocytes were identified on the surface of catheters as described by van Gennip et al., Polliack and Saint-Guillain et al., respectively [60][61][62] . Only catheters collected from rats that had survived the entire experiment (7 days) were used for counting purposes. Average bacterial density on the surface of the catheter was based on manual counting bacteria in 20 random SEM fields at a 7500X magnification on each of the two catheter segments in each rat by a blinded researcher. When only one catheter segment could be retrieved, average bacterial density was done analogously using 40 random SEM fields at 7500X magnification. When more than 300 bacteria were found in a given SEM field, bacteria were considered uncountable. Using the same methodology, the disposition of bacteria in each SEM field was recorded as planktonic, biofilm or mixed.
Average leucocyte density on the surface of the catheter was performed in a similar fashion, with the exception that SEM fields used were obtained at 750X magnification. The average phagocyte density on the surface of catheters was determined in the same way. Phagocytes were defined as those leucocytes that showed evidence of pseudopods in direct contact with bacteria and/or biofilm.
Bacteria and leucocyte were included in counts only if the top upper edge of the cell was in the SEM field 63,64 . SEM fields with clots or largely occupied by biofilms were discarded, as bacteria were not accessible to counting using the methodology employed.
Statistical analysis. Qualitative variables were expressed as percentages. Quantitative variables were expressed as means ± standard deviation. The SPSS 21.0 ® software was used for descriptive and inferential www.nature.com/scientificreports www.nature.com/scientificreports/ statistical analysis. The Kolmogorov-Smirnov test was used to assess if variables were normally distributed. ANOVA and t-Student test were used to compare averages in normally distributed data. Kruskal-Wallis and Mann-Whitney tests were used to compare means in non-normally distributed data. Wilcoxon rank-sum test was used to compare ordinal data. Proportions were analyzed with the Chi-square test or Fisher's exact test. Association between numerical variables was assessed using Pearson's correlation coefficient. Relationship between ordinal variables was evaluated with resort to Spearman Rank Correlation Coefficient 65,66 . Differences in survival in the different experimental groups was tested with the log rank test. A two-tailed p < 0.05 was considered to be statistically significant.

Inoculation of arterialized venous flaps results in a model of persistent ischemic flap infection.
In order to obtain a rat model of ischemia and P. aeruginosa infection in the presence of a foreign body, we followed the steps summarized in Fig. 2. Ischemia was obtained by using a variant of the rat abdominal arterialized venous flap, established by our group 42 . Infection was obtained by inoculating P. aeruginosa and two 1-cm segments of sterile 14-gauge silicone catheters were used to simulate foreign body.
By thermography analysis, one hour after surgery, the average difference between the flap surface's temperature and the temperature of the surface of the homologous contralateral region was 2.34 ± 1.06 °C (p < 0.001). Figure 3 shows the typical thermographic appearance of the flap one hour after surgery, illustrating the lower temperature of the flap compared to the contralateral side. Throughout the experiment, the flap's temperature remained inferior to that of the homologous contralateral region.
Overall our data shows that a model of ischemia, infection and foreign body can be successfully obtained by adapting our previous established model of the rat abdominal arterialized venous flap, following inoculation with P. aeruginosa and introduction of silicone catheters. To confirm transduction efficiency, seven days after surgery, flaps from rats transduced with GFP were analyzed by fluorescence microscopy. As shown in Fig. 4, flaps from GFP lentivirus-transduced rats showed GFP expression. The expression was greater in endothelia and in the epidermis (Fig. 4). Quantification of BD-2 mRNA expression by real-time PCR seven days after transduction, revealed 591.12 × 10 6 % ± 894.00 × 10 6 % relative expression of BD-2 in transduced flaps versus 3.53 × 10 6 % ± 11.35 × 10 6 % relative expression in non-transduced flaps (p = 0.014) (Fig. 4). Analogously, the relative BD-3 expression determined in a similar fashion was 253.09 × 10 6 % ± 354.61 × 10 6 % in transduced flaps and 25.94 × 10 6 % ± 87.44 × 10 6 % for non-transduced flaps (p = 0.018) (Fig. 4). Using the same methodology, no statistically significant differences were found in the relative expression of BD-2 in the BD-2 and in the BD-2 + BD-3 groups, nor in the relative expression of BD-3 in the BD-3 and in the BD-2 + BD-3 groups. Figure 5 summarizes the main clinical, optical microscopy and SEM findings in the different experimental groups. On the third postoperative day, the average area of necrotic flap was higher in the P. www.nature.com/scientificreports www.nature.com/scientificreports/ aeruginosa (61.98% ± 18.52%) and GFP (67.39% ± 20.67%) groups than in the groups BD-2 (37.16% ± 9.04%), BD3 (17.79% ± 5.48%) and BD-2 + BD-3 (35.09% ± 9.43%) (p ≤ 0.001). The necrotic area was the smallest in the NaCl (11.07% ± 5.72%) and in the BD-3 groups. There was no statistically significant difference between these two groups (p = 0.836) (Fig. 6).
In summary, our results demonstrate that transduction of flaps with BD-2 or BD-3 improves infected flap survival at day 7 following surgery, by decreasing tissue necrosis.

Evidence of flap ischemia and defensin expression by histology and immunohistochemistry.
Histological analysis of the flaps revealed signs of venous congestion, marked edema, epidermolysis and areas of necrosis in all experimental groups. However, these changes were more pronounced in the P. aeruginosa and GFP groups (Fig. 5). Immunohistochemistry confirmed expression of BD-2 and BD-3 in the transduced rats (Fig. 5). BD-2 expression occurred mainly in endothelia and perivascular tissues, although it was also observed in fibroblasts and in the epidermis. BD-3 expression was strongest in the epidermis and skin appendages (Fig. 5). Using the mentioned semi-quantitative score for immunohistochemistry staining 59 , BD-2 expression was as follows: 0.25 ± 0.45 for the NaCl group; 0.23 ± 0.44 for the PA group; 0.36 ± 0.50 for the GFP group; 2.07 ± 0.59 for the BD-2 group; 0.13 ± 0.35 for the BD-3 group; 1.64 ± 0.50 for the BD2 + BD-3 group. Proceeding in a similar way, BD-3 expression in the different experimental groups was the following: 0.67 ± 0.49 for the NaCl group; 0.69 ± 0.48 for the PA group; 0.79 ± 0.70 for the GFP group; 0.27 ± 0.59 for the BD-2 group; 2.80 ± 0.41 for the BD-3 group; 2.50 ± 0.52 for the BD2 + BD-3 group. BD-3 expression was stronger than BD-2 expression. However, this difference did not reach statistically significance.

BD-2 and BD-3 decrease bacterial numbers and biofilms on the surface of foreign bodies.
We then assessed the number of bacteria on the surface of the foreign body in the different experimental groups. At the end of the experiment, it was possible to retrieve 56 catheters that were distributed as follows: 8 in the NaCl group, 6 in the PA group, 12 in the GFP group, 9 in the BD-2 group, 11 in the BD-3 group, and 10 in the BD-2 + BD-3 group.
In all rats, infection was confirmed by growth in P. aeruginosa MacConkey agar, Gram stain and positive oxidase reaction. Table 1   Values are expressed as average ± standard deviation. CFU, colony forming units; SEM, Scanning electron microscopy As shown in Table 1, significant differences were found in the number of bacteria on the surface of the foreign body (Fig. 7). The number of bacteria on the surface of catheters was lower in the NaCl, BD-3, and BD-2 + BD-3 groups. There were no statistically significant differences between these 3 groups. The BD-3 group presented a smaller number of bacteria on the surface of catheters than the P. aeruginosa (p < 0.042), GFP (p < 0.001) and BD-2 (p = 0.035) groups.
Numerous leucocytes could be observed on the surface of catheters in all groups. (Figs 5Y to D' , 7 and 11). Apart from the average number of leucocytes being superior in the GFP group (4.92 ± 3.68 leucocytes/SEM field) than in the NaCl group (0.75 ± 0.93 leucocytes/SEM field; p = 0.016), no other statistically significant differences were found in leucocytes' and phagocytes' numbers on the surface of catheters in the remaining experimental groups (Fig. 7). www.nature.com/scientificreports www.nature.com/scientificreports/ Bacterial contamination with cocci were observed in 8.92% ± 22.21% of SEM fields (Fig. 9). There were no statistically significant differences between the different experimental groups.
Overall, data shows that in the model used the surface of foreign bodies presents significant numbers of bacteria. Flap transduction with either BD-2 or BD-3, and in particular with the latter, remarkably lowers bacterial numbers on the surface of foreign bodies.  www.nature.com/scientificreports www.nature.com/scientificreports/

Bacterial counts and biofilms on the surface of the foreign body were correlated with flap necrosis.
We then assessed whether bacteria accumulation on the surface of the foreign body was correlated with flap necrosis. Table 2 synthetizes the most relevant correlations in the data obtained in this study.
Flap necrosis rates on the 3 rd and 7 th day postoperatively were highly correlated (Pearson's correlation factor = 0.922; p < 0.001). Flap necrosis was also correlated with several features of catheters' surface. In fact, flap necrosis rate on the third postoperative day was positively correlated with the average number of bacteria per SEM field and the percentage of SEM fields with biofilm (Pearson's correlation factor = 0.609; p < 0.001 and 0.596; p < 0.001, respectively). It was negatively correlated with the following variables: percentage of SEM fields with planktonic bacteria; percentage of SEM fields without bacteria; percentage of SEM fields without biofilm (Pearson's correlation factor = -0.409, p = 0.004; -0.426, p = 0.003 and -0.681; p < 0.001, respectively).
Analogously, flap necrosis on the seventh day postoperatively was positively correlated with the following findings: average number of bacteria per SEM field (Pearson's correlation factor = 0.626; p < 0.001) and the percentage of SEM fields with biofilm (Pearson's correlation factor = 0.563; p = 0.001). Moreover, flap necrosis rate on the seventh day after surgery was significantly negatively correlated with the percentage of SEM fields with planktonic bacteria, without bacteria and without biofilm (Pearson's correlation factor = -0.379, p = 0.03; -0.462, p = 0.007; -0.674, p < 0.001, respectively).
The average number of bacteria per SEM field was also positively correlated with the number of leucocytes (Pearson's correlation factor = 0.276; p = 0.041); and phagocytes per SEM field (Pearson's correlation factor = 0.401; p = 0.002).
P. aeruginosa counts on the 3 rd postoperative day determined by culture of flap biopsies, were correlated with flap necrosis on that day (Pearson's correlation factor = 0.287; p = 0.041) and on the seventh day after surgery (Pearson's correlation factor = 0.395; p = 0.016) (Fig. 12). However, there were no statistically significant correlations between P. aeruginosa counts after flap biopsy on the seventh day after surgery and flap necrosis either on the 3 rd or 7 th postoperative days. (Fig. 12).
There were no statistically significant correlations between bacterial quantification after flap biopsy and culture or real-time PCR, on the one hand, and SEM bacterial counts on the other hand.
In conclusion, our data shows that the bacterial counts on the surface of the foreign body and biofilm formation were correlated with flap necrosis.

Discussion
Our study showed that transducing an ischemic skin flap in the rat with human β-defensins, the BD-2 and BD-3 genes, increased skin flap survival in the context of a P. aeruginosa infection in the presence of a foreign body. In addition, BD-2 and particularly BD-3 transduction reduced P. aeruginosa numbers and biofilm formation on the surface of the foreign body. Other authors had already shown that BD-2 and BD-3 are overexpressed in the presence of P. aeruginosa keratitis in mice and that overexpression was associated with diminished P. aeruginosa numbers in the eyes of affected mice 15,67 .
However, as far as the authors could determine, this is the first description of the transduction of BDs genes to treat a P. aeruginosa infection in vivo 14 . It is also the first time that the transduction of an AMP gene is performed www.nature.com/scientificreports www.nature.com/scientificreports/ with the intent to treat an infection taking place in a poorly perfused region, simulating local ischemia. This scenario is frequent in the clinical setting, occurring after trauma, radiotherapy or in ischemic regions of the body, namely in the lower limbs of atherosclerotic and/or diabetic patients 68,69 . This study describes for the first time the morphometric features of the bacteria on the surface of a foreign body and thoroughly compares them with the clinical evaluation of the overlying skin flap, and with bacterial quantification using microbiological cultures of skin biopsy and real-time PCR. Other authors had already shown the possibility of treating a Staphylococcus aureus infection transducing a normally perfused flap with the AMP cathelicidin LL-37 43 .
The groups treated with BDs presented a smaller percentage of SEM fields with any form of biofilm relatively to the GFP group (p ≤ 0.002; Fig. 10). Arguably, the GFP group seems to be the most appropriate control group, as it involves expression of a foreign protein using a viral vector. This alone has been shown to affect local bacterial clearance 70 . These data suggest that in vivo these AMPs are able to prevent the formation of biofilms, which are one of the most common causes of clinical persistent infection [71][72][73][74][75][76] . This adds to the knowledge of BD physiology, as most papers in this field refer to the action of AMPs in vitro, not addressing the role of these substances in vivo 77 .
Flap necrosis rates on the 3 rd and 7 th day postoperatively were highly correlated (Pearson's correlation factor = 0.922; p < 0.001). This suggests that, in this rodent model, the ongoing P. aeruginosa infection associated with the foreign body led to continuous breakdown of the overlying integument 72 . This had already been amply observed in clinical practice 72,78 . Recently, it has been shown that P. aeruginosa induces the production and release of cytotoxic amyloids with prion features 79,80 . It would be interesting to investigate in future studies whether this mechanism plays a significant role in the model used in the present work, and whether these processes could be curtailed or even halted by BDs transduction.
Interestingly, in this work, flap necrosis was significantly correlated with several features of catheters' surface, namely bacterial numbers and biofilm presence. In opposition, in the present study, flap necrosis rates in the different groups was not correlated with bacterial determination by real-time PCR of flap biopsies. Moreover, only bacterial counts of microbiological cultures of biopsies of the flap performed on the third post-operative day correlated, and even then, only poorly, with flap necrosis rates on the third (Pearson's correlation factor = 0.287; p = 0.041) and seventh (Pearson's correlation factor = 0.395; p = 0.016) days after surgery. This lends support to the notion that in bacteria capable of producing biofilms, such as P. aeruginosa, the protracted release of bacterial cells from the biofilm medium is one of the main determinants of continuous tissue damage to the overlying  www.nature.com/scientificreports www.nature.com/scientificreports/ tissues 45,60,[72][73][74] . In fact, a recent meta-analysis demonstrated that biofilms are present in more than three quarters of chronic wounds, increasing their extension or preventing their closure 73 .
All these data suggest that the model herein used adequately mimics clinically relevant skin ulcers and infections. Noteworthy, a similar model had already been described by Van Wijngaerden et al., who introduced a catheter in the back of the rat to test antibiotic efficacy 45 . Furthermore, other authors, have also placed a catheter under a well perfused flap in the groin region of the rat 43 . Notwithstanding, these authors did not use SEM to evaluate the surface of the catheter used as a foreign body. It was only recently that van Gennip et al. emphasized the potential advantages of studying the morphology and interaction of bacteria and immune cells on the surface of foreign bodies with SEM 60 . However, these latter authors placed the foreign body inside the peritoneal cavity 60 .
We believe that our model presents the additional advantage of simulating an ischemic environment surrounding the foreign body. In fact, as shown in Fig. 5, one hour after surgery, the average temperature difference between the flap and the contralateral side was 2.34 ± 1.06 °C (p < 0.001). These data confirm that the surgical skin flap placed over the foreign body was ischemic, since skin temperature is known to be proportional to integumentary perfusion 81 .
In all groups, a 100 µl solution was injected intravascularly. To ensure that the same number of viral particles was administered in each of the transfected groups, in the BD-2 + BD-3 group, the solution was composed of 50 µl of the solution used in the BD-2 group and 50 µl of the solution used in the BD-3 group. The BD-2 + BD-3 group did not show statistically significant differences relatively to either the BD-2 or the BD-3 groups, regarding flap necrosis on 7 th post-operative day (Fig. 6), in the number of bacteria (Fig. 7) or bacterial distribution on the surface of the foreign body (Fig. 10). Hence, these data suggest that the simultaneous expression of the two defensins does not have a synergistic or an antagonistic effect 82 .
One of the limitations of the present methodology was the impossibility to precisely differentiate between leucocytes types (Fig. 11) 60,[83][84][85][86][87] . Although immunological methods have been proposed to distinguish leucocyte types using SEM, it has been shown that these methods may significantly affect leucocyte adhesion and morphology, making interpretation of data difficult 88 . In future works, it would be interesting to characterize with other techniques leucocyte numbers and cell types both on the surface of the foreign body and in the surrounding tissues. www.nature.com/scientificreports www.nature.com/scientificreports/ Another limitation of this study was the presence of contamination of the wound by bacteria other than P. aeruginosa, namely by cocci, in 8.92 ± 22.21% of SEM fields (Fig. 9). This contamination affected all experimental groups equally. It is no doubt related to the ischemic nature of the skin flap used, leading to partial flap necrosis, surgical wound dehiscence and contamination with saprophytic flora. Hence, studies similar to this one but with normally perfused flaps are, therefore, warranted, in order to minimize contamination with other bacterial species.  Table 2. Synthesis of the most relevant correlations in the data obtained in this study. SEM, scanning electron microscopy. www.nature.com/scientificreports www.nature.com/scientificreports/ The authors must concede that multiple other delivery methods of the BDs genes could have been employed 89,90 . Many of the other available vectors have potential advantages and disadvantages. For example, it has been shown that intradermal delivery of the transgenes might be superior to intravascular perfusion using lentivirus vectors in skin flaps 91 . Consequently, further studies are also necessary in this field, namely to test the superiority of other vectors, as well as the susceptibility of other bacteria species, particularly bacteria resistant to conventional antibiotics. Furthermore, it would be interesting to assess if BDs could be used to reverse established biofilm on the surface of foreign bodies.
Although the potential clinical merits of AMPs have been extoled for the past two decades, their clinical application has been tardy and incipient, due to potential systemic toxicity, susceptibility to proteases, and high cost of peptide production 10,11,13,14,92,93 . However, despite several in vitro studies (Supplemental Figs 1 and 2) many questions remain to be answered regarding the in vivo mechanisms of BD-2 and BD-3 bacterial clearance, biofilm impediment and increased tissue survival. The present work provides little help in answering these questions, making further studies warranted 4,14,20,94 .
Finally, the authors would like to note that controlled local production of a specific AMP by a flap placed over a difficult wound (for example, osteomyelitis in a mangled extremity) may theoretically allow to eradicate a bacterial infection and thus permit wound closure 43 . If the introduction of the AMP sequence was to cause oncogenesis, the flap could be simply removed and replaced by another flap or eventually by a skin graft 43,95 .