Clinical SMA strains were obtained from hospitalized patients with invasive infections that had originated from medical manipulation in the surgical wards and surgical ICUs of six general hospitals in Tianjin, China, between 2006 and 2012 (Table 1). The MICs of SMA to 12 antibiotics commonly used for Gram-negative bacilli were determined by microbroth dilution, analyzed according to the American National Clinical and Laboratory Standards Institute (CLSI) guidelines.29 The following strains were assessed in parallel for quality control: ATCC27853, ATCC25922 and ATCC25923, preserved in the Infectious Disease Institute of the Second Hospital of Tianjin Medical University, China.

Using a methodology previously reported by Ceri27,28 we established an in vitro model of SMA BBF, in a Mueller–Hinton broth (MHB)-silica film system, as previously described.30,31 Cryopreserved SMA was recovered in sheep blood agar plates incubated aerobically overnight. A fresh single colony was transferred to fresh MHB and incubated for 8h at 35°C, from which a 200μL suspension of 0.5 McFarland was prepared and transferred to a 12-well flat-bottom plate, in which sterile silica film (1cm×1cm×1mm, L×W×T) and 1.8mL MHB were co-cultured at 35°C for 12 and 24h. After washing three times with 0.9% sodium chloride to remove planktonic bacteria, the BBF on the silica films was prepared. The culture medium was regarded as the negative control. Morphology was observed by scanning electron microscopy (SEM) and fluorescence microscopy (FSM) as described below.

As previously described,32,33 the biofilm was fixed in 2.5% (v/v) glutaraldehyde in PBS (0.1M, pH 7.4) for 24h, then immersed in saturated calcium chloride solution for 15min, and rinsed with ddH2O between each step. The film was immersed in 5% silver nitrate solution for 15min, immediately stained with 1% hydroquinone for 2min, then rinsed with ddH2O. The film was fixed in 5% sodium thiosulfate solution for 2min, then rinsed in ddH2O and analyzed by FSM.

As previously described,25 the silica biofilm was fixed in 2.5% (v/v) glutaraldehyde in PBS (0.1M, pH 7.4) at 4°C for 2h, fixed again with 1% osmic acid for 1h, then rinsed with PBS, dehydrated through a series of ethanol dilutions, then treated with isoamyl acetate. The specimen was dried in a vacuum, then coated with platinum–palladium and analyzed by SEM at 5–10kV.

MIC values of the following antibiotics were calculated with a broth microdilution assay using the twofold dilution method according to the CLSI guidelines34: ciprofloxacin (CIP), levofloxacin (LEV), piperacillin (PIP), ceftazidime (CAZ), cefoperazone/sulbactam (SCF), erythromycin (ERY), sulfamethoxazole (SXT), and gentamycin (GM) within the range of CIP was 256–0.125mg/L, or 512–0.125mg/L.

After dilution of the fresh SMA to 3×1010CFU/mL, 100μL of the culture was transferred to each well of a flat-bottom 96-well microtiter plate. As previously described,27,28,35,36 BBF were formed by immersing the pegs of a modified polystyrene microtiter lid into this plate, which was then incubated at 35°C for 24h. Peg lids were removed, rinsed in PBS, then placed on flat-bottom microtiter plates containing antibiotics, and incubated at 35°C. After 24h peg lids were removed, rinsed three times in PBS, then transferred onto flat-bottom microtiter plates containing 100μL of MHB per well, and biofilms were transferred from pegs to wells by centrifugation at 805×g for 20min. The optical density was measured at 650nm (OD650) on a microtiter plate colorimeter before and after incubation at 35°C for 6h.

Adequate biofilm growth was defined as a mean OD650 difference (OD650 at 6h minus the OD650 at 0h) of ≥0.05. BICs are defined as the lowest concentration of antibiotic in which the OD650 was 10% or less of the mean of the positive control well readings, representing at least a 1−log10 growth difference.

The susceptibility of biofilms to LEV, SCF and PIP alone and in combination with ERY was determined by the checkerboard method. Five strains of SMA (numbered 0020, 0037, 0040, 0088, 0256) were selected for the following test. Silica films were transferred to a 24-well flat-bottom microtiter plate. Then 2.0mL MHB was transferred to the well (blank control), 1.8mL MHB was transferred to each of the other wells. According to the MIC values of LEV, SCF and PIP against each strain, LEV, SCF and PIP was added to yield a final concentration of 1/2 MIC, 1 MIC and 2 MIC, and ERY was added to yield a final concentration of 1/16 MIC and 1/4 MIC. The plate was incubated at 35°C for 24h, then after washing with sterile physiological saline, the silica films were ultrasonicated at 60W for at 15min. The colony concentration was calculated from the OD650 by the ATCC27853 standard curve (Y=1.26514×10−8X+0.48028, with a linear range between 1.5×108 and 9.6×103CFU/mL, p<0.01). The density of colonies on the silica film was recorded from three wells in parallel.

CFU per centimeter was indicated by mean±standard deviation, and the interaction of combined testing was analyzed by factorial experiment design ANOVA. The impact of antibiotics on biofilm growth was calculated using an ΔE model is based on the Bliss independence theory, described by the equation Ii=(IA+IB)−(IA×IB), where Ii is the predicted growth inhibition caused by the theoretical combination of drugs A and B, and IA and IB represent the growth inhibition caused by each drug individually. Since I=1−E, where E represents growth inhibition, the following equation is derived: Ei=EA×EB. Ei represented the predicted growth inhibition of the theoretical non-interactive combination of the drugs A and B, and EA and EB represent the growth inhibition caused by each drug individually. Interaction was described as the difference (ΔE) between the predicted and measured growth inhibition (ΔE=Epredicted−Emeasured). Statistically significant interactions of <100% were considered weak, those of 100–200% were considered moderate, and those >200% were considered strong.

The antibiotic susceptibility of SMA isolates was assessed and in total, 47 (92.1%) of isolated strains were resistant to three or more antibiotics. The majority of tested strains were resistant to cefotaxime (94.1%) amikacin (90.2%) cefritaxone (88.2%) GM (82.4%) and CAZ (60.8%), and almost half of all tested strains were resistant to cefoperazone (49.0%), PIP (45.1%) and cefepime (45.1%).

The capacity of clinical isolates of SMA strain 0314 to form biofilms on the surface of the silica films incubated in MHB was assessed by fluorescent microscopy. Mature biofilms were formed by 42 of the 51 SMA isolates (82.35%). After 24h, silver staining revealed mature biofilm to be composed of black irregular sheets, scattered dots and black cottony membranes, and fine rod-shaped bacteria were observed at the borders (Fig. 1). No black aggregation was observed on the surface of silica film cultured in MHB in the absence of SMA. These findings were confirmed by SEM (Fig. 2). Biofilm ultrastructure was observed after 12h (Fig. 2C), and mature biofilm was observed at 24h (Fig. 2A and B). SMA was observed to be clustered on the surface of the biofilm, and SMA was mostly short rods of about 1.5μm×1.0μm, partially or entirely encompassed by the extracellular matrix (Fig. 2A and C). Fine bacilli alternated cross lots of extracellular mucus filaments, some cells were visible in split phase or undergoing apoptosis, and the bacterial community present on the biofilm was heterogeneous (Fig. 2B). No bacteria were observed on the surface of silica film cultured in MHB in the absence of SMA (Fig. 2D).

Fig. 1.

In vitro SMA BBF formation was visualized by fluorescence microscopy (×200). After 24h SMA strain 0314 cultured in MHB formed BBF on silica film. Black irregular sheets, scattered black dots, and cotton-like membranes were observed, and fine rod-shaped bacteria were observed at the borders (A). No black aggregation was observed on the surface of silica film cultured in MHB in the absence of SMA (B).

(0.14MB).

Fig. 2.

Ultrastructure of in vitro SMA BBF visualized by scanning electron microscopy. After 12h the biofilm of SMA strain 0314 cultured in MHB on silica film was observed (C), and was mature at 24h (A, B). SMA biofilm ultrastructure included clusters of proliferating bacteria (A, C). SMA was mostly short rods of about 1.5μm×1.0μm, partially or entirely encompassed by the extracellular matrix. A large number of fine bacilli alternated across numerous filamentous strands (B). No bacteria were observed on the surface of silica film cultured in MHB in the absence of SMA (D). B ×20,000, Bar=50μm; A, C, D ×5000; Bar=20μm.

(0.24MB).

The effect of antibiotic agents on biofilm formation was analyzed by comparing MICs with BIC. The formation of biofilms enhanced the resistance of some strains to all tested antibiotics aside from GM and ERY, to which all the tested strains were resistant even in the absence of biofilm, and LEV (Table 2). After biofilm formation, 21 (50.0%) tested isolates were resistant to LEV, 33 (78.57%) to SXT, 33 (78.57%) to PIP, 32 (76.19%) to CAZ, 32 (76.19%) to SCF, and 32 (76.19%) to CIP. The average BIC90 was also much higher than the MIC90 (Table 3).

The effects of combining ERY with other antibiotics was investigated in five SMA isolates which were determined to have moderate susceptibility to LEV, SCF and PIP. The MIC of LEV in the chosen strains (0020, 0037, 0040, 0080 and 0256) was between 0.5 and 64mg/L, the MIC of SCF was between 8 and 32mg/L and the MIC of PIP was between 2 and 64mg/L. Addition of ERY did not significantly enhance the efficacy of SCF (p=0.06), but did significantly enhance LEV and PIP efficacy (all p<0.05). Factorial ANOVA analysis revealed that the efficacy of SCF, LEV, and PIP was significantly improved in the presence of ERY (F=8.460, p=0.000; F=20.825, p=0.000; and F=4.506, p=0.001, respectively; Table 4, and one-way ANOVA, p<0.05). The synergy of antibiotics was assessed by the ΔE model, and the combination of ERY and LEV was found to have a synergistic effect against four of the five strains tested (Table 5).

SMA strain 0020 biofilm formation was also assessed in the presence of SCF, LEV, PIP and/or ERY (Fig. 3). We observed that LEV damaged BBF in the presence or absence of ERY (Fig. 3A and B), but the combination of ERY with LEV reduced the number of bacteria anchored to the BBF, thinned the polysaccharide matrix and altered bacterial morphology. Some spherical, irregularly shaped, or cracked bacteria were observed. In the absence of antibiotics, the BBF was mature and bacterial cells were packaged in a lot of sticky polysaccharide matrix. Bacterial growth was strong and some bacteria were observed to be in the division phase (Fig. 3C). Neither PIP nor SCF, in the presence or absence of ERY, altered the state of the bacteria or biofilm (Fig. 3D).

Fig. 3.

Impact of antibiotic agents on ultrastructure of in vitro BBF visualized by scanning electron microscopy. SMA strain 0020 BBF formed a complete mature structure in which the bacterial were packaged with dense matrix (C). In the presence of erythromycin (B) or erythromycin and levofloxacin (A) the BBF matrix became thin, fewer adhered bacteria were observed and some exhibited spherical or irregular form. The black arrow indicates a bacterial cell undergoing division phase. In the presence of erythromycin and piperacillin the BBF remained similar to that seen in the blank control (D). D ×20,000; Bar=100μm; A, B, C ×5000; Bar=50μm.

(0.23MB).