Sample and phenotype determination
A total of 57 unrelated non- duplicate A. baumannii isolates from different hospitalized patients were selected randomly from the collection at the National Center for Infection Control and Antibiotic Resistance in Israel. The sample was composed of isolates collected from 15 medical centers from Israel and Europe in the years 2008–2019. The specimens were isolated from blood (n = 25), sputum (n = 11), skin (n = 8), rectum (n = 4), tracheal aspirate/bronchoalveolar lavage (n = 4) and other sites (n = 5). The isolates were identified to the species level by VITEK® MS (bioMérieux SA, Marcy l’Etoile, France). Further confirmation of the species was performed by molecular genotyping using methods described by Evans et al. [26]. In brief, typing of A. baumannii isolates was conducted by the alignment of OXA-51-like β-lactamase sequence relationships using OXA-69A and OXA-69B primers. Clonality was determined by sequencing of the OXA-51-like gene to assign isolates to international clonal complexes [27]. Of the 57 isolates, ten belonged to clonal complex 2, 33 isolates belonged to clonal complex 3, and 14 belonged to other clonal complexes (Table 1).
All isolates were categorized by phenotype observation as mucoid or non-mucoid. Phenotype was determined following overnight incubation at 35 ± 2 °C on blood agar (tryptic soy agar supplemented with 5% sheep blood; Hylabs, Rehovot, Israel). Three clinical microbiologists evaluated the phenotype according to several colony morphology parameters, including texture, elevation, margin, size and shape [28]. If the categorization by the three microbiologists was not identical, the isolate was excluded (Table 1).
We also included in the sample three well-characterized specimens as controls: (i) a non-mucoid and thin-capsulated A. baumannii (ATCC 19606); (ii) a mucoid A. baumannii with a significant capsule (CAP-AB). This isolate harbors a capsule locus type KL17 including wza-c genes and three different glycosyltransferases gtr38, 39, 40. (iii) A hypermucoid string test-positive K. pneumoniae (HM-KP).
All isolates were stored at − 80 °C before sub-culturing and analysis.
Principle of the density-dependent gradient method
The method is based on bacterial cells passing through a gradient matrix as a result of centrifugal force. The gradient was composed of multiple phases, each phase with a different concentration (20–80% V/V) of colloidal silica particles of 15–30 nm (Percoll, GE Healthcare, Uppsala, Sweden). Bacterial cells were inoculated to the top of a gradient and, due to centrifugation, migrated to different locations within the density gradient based on capsule size. This method assumes that cell size is similar between A. baumannii strains.
Validation of gradient composition
The validation was first performed on the three control strains and then confirmed on six isolates selected randomly from the sample: four with a mucoid phenotype and two with a non-mucoid phenotype. The nine isolates were prepared by inoculating a single colony in 10 mL BHI (Hylabs, Rehovot, Israel). The cultures were incubated for 18 h at 35 ± 2 °C with shaking (250 rpm), to reach a stationary growth phase (final normalized optical density OD600 of 0.7–0.9). The cultures were centrifuged for 10 min at 3200×g and the bacterial cell pellet was re-suspended with sterile phosphate buffer saline (PBS; Bio-Lab, Jerusalem, Israel) to a final volume of 2 mL.
In order to achieve robust separation and easy visualization of bacteria in the gradient matrix, we determined the optimal silica concentrations of the phases and the minimum number of phases required. To determine the optimal silica concentrations, a panel of seven single concentrations of silica (20, 30, 40, 50, 60, 70, 80% V/V) with a final volume of 500 μL were prepared. An aliquot of 100 μL of each of the nine isolates was placed on the top of each tube and was centrifuged for 10 min at 8000×g. The migration of the bacterial band was visually examined in the seven tubes and concentrations with bands that yielded a clear band within the gradient, without any smears, were selected. We rejected tubes in which bacterial cells could not penetrate the matrix or bacterial bands had settled at the bottom of the tube. The minimum and maximum silica concentrations of the remaining tubes composed the range of work.
Gradient matrix preparation
The matrix was prepared as follows: 1 mL of the lower density silica concentration was placed in an empty round-bottom tube (12/75 mm, Heinemann Labortechnik GmbH, Duderstadt, Germany). Using a disposable syringe (23G needle), 1 mL of the higher density concentration was inserted at the bottom of the tube, allowing the lower density concentration to rise above it.
Sample preparation and analysis
We used two techniques to prepare the samples: (i) bacterial cells from BHI broth as described above in the “validation of gradient composition” section. Following centrifugation, the cell pellet was re-suspended for a final volume of 1 mL. (ii) Bacterial cells directly from plate: a 5 μL loop-load of colonies from an overnight incubation on blood agar (35 ± 2 °C) was suspended in 1 mL PBS. In order to evaluate whether the growth media affected the results, all tests were performed in duplicates from BHI broth and from blood agar plates.
An aliquot (600 μL) of the bacterial cells was inoculated to the top of the gradient matrix. Following centrifugation for 30 min at 3000×g, results were read by measuring the height of the bacterial band from the liquid level down to the middle of the bacterial band. A bacterial band that migrated to the bottom phase of the gradient was classified as a thin or non-capsulated strain, while a bacterial band that concentrated in the top phase was suspected to be a capsulated strain.
Total carbohydrate quantification
This test is based on the reaction of carbohydrates with sulfuric acid, which produces furfural derivatives that develop a detectible color when reacting with phenol [29, 30]. The test quantifies the total polysaccharides present in the bacterial cell, including capsular polysaccharides and LOS. Extraction was performed by centrifuging 1 mL of a culture for 5 min at 14,500×g and washing the cells with 1 mL of 50 mM NaCl five times. During the last washing cycle, the cells were re-suspended in 50 mM ethylenediaminetetraacetic acid to normalize the solution to OD600 2.0. The sample was incubated at 35 ± 2 °C for 1 h in order to release the polysaccharides from the cell. Following centrifugation, 200 μL of the supernatant of each isolate was used for quantification. A range of known sucrose/fructose (1:1 W/W) concentrations (0–240 μg/mL) was used to construct a standard curve. 200 μL of 5% phenol (in water) and 1 mL of 93% sulfuric acid were added to the samples and to the standards. Furfural derivatives were measured by optical absorbance at 490 nm (OD490) after 10 min. The total carbohydrate amount was calculated from the standard curve.
Tem
Six isolates selected randomly from the sample were imaged: one isolate determined to be thin or non-capsulated by the density-dependent gradient test, three isolates suspected to be capsulated by the density-dependent gradient test, and the two A. baumannii controls. The samples were prepared according to a standard protocol [31]. An aliquot of 1 mL of a culture grown overnight in BHI medium at 35 ± 2 °C with shaking was centrifuged and the supernatant was removed manually. The bacterial cells pellet was re-suspended and fixed for 3 h in Karnovsky mixture [31]. To remove fixative residues, the cells were washed three times with 0.1 M sodium cacodylate buffer. The cells were post-fixed in 1% OsO4, 0.5% K2Cr2O7, 0.5% K4[Fe (CN)6] and stained with 2% uranyl-acetate. The samples were washed with distilled water, dehydrated using ethanol and embedded with resin (Epon EMBED 812, EMS, Hatfield, PA) and polymerized at 60 °C for 24 h. Ultrathin sections (90–70 nm) were obtained using an ultra-microtome (EM UC7, Leica Biosystems, Buffalo Grove, IL). Samples were imaged using a TEM at 120kv (G-12 Spirit, FEI, Hillsboro, OR). Capsule thickness was determined by measuring five different sites on each isolate. For each isolate, the mean and standard deviation of the five measurements were calculated.
India ink staining
The samples were prepared according a standard protocol [15]. Briefly, a drop of India ink (Black 17, Pelikan, Hannover, Germany) was placed on a microscope slide and mixed with a colony from an overnight culture on blood agar. The slide was then dried under air at room temperature for 30 min. The slide was saturated with 1% crystal violet (Merck, Rehovot, Israel) for 1 min, rinsed with deionized water and dried under air at room temperature for 2 h. Observation of the slide was performed with a light microscope (Leica LMD7, Leica Microsystems, Buffalo Grove, IL).
Statistical analysis
We used a Student’s t-test to compare total carbohydrates and capsule thickness between the capsulated and thin/non-capsulated isolates.