Introduction

The genome of each microorganism is a source of knowledge that can be applied for strain differentiation, based on bioinformatic tools and available techniques of molecular biology, suitable for epidemiological investigations. Among the species of the genus Acinetobacter, A. baumannii strains manifest the highest pathogenicity (Wong et al. 2017; Skariyachan et al. 2019). They are highly opportunistic microorganisms, responsible for hospital infections related to ability to adapt to different environmental conditions (Antunes et al. 2014). At the beginning of the twenty-first century, no complete genome sequence of Acinetobacter sp. was known. Barbe et al. (2004) published the first sequence of Acinetobacter sp. ADP1 genome, and later Smith et al. (2007) published the complete genome of A. baumannii ATCC 17,978. Subsequently, the first genomic sequence of the multidrug-resistant A. baumannii strain was published by Adams et al. (2008). Presently, complete sequences of the genomic DNA of A. baumannii are known for about 250 strains (http://www.ncbi.nlm.nih.gov, database retrieved on 10 December 2020). As indicated by various research teams, repeated sequences of A. baumannii and other microorganisms have great impact in the process of generating pathogenicity for immunocompromised hosts (Zhou et al. 2014; Shariat and Dudley 2014; Nabil et al. 2015) or adaptation skills to different environmental conditions (Zhou et al. 2014; Shariat and Dudley 2014; Karah et al. 2015).

The presence of tandem DNA repeats in genomes of A. baumannii was confirmed by several groups (Martín-Lozano et al. 2002; Turton et al. 2009; Irfan et al. 2011; Pourcel et al. 2011; Minandri et al. 2012; Ergin et al. 2013; Ahmed and Alp 2015; Villalón et al. 2015). Based on these sequences, different methods of differentiation of A. baumannii strains have been developed; however, they take into account only their diversifying power of evolutionary changes of the Acinetobacter genus (Touchon et al. 2014). Thus, their features responsible for drug resistance or pseudo-immunological bacterial responses, encoded in the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) system, which evolved to protect the cells from exogenous phage and plasmid DNA invasion, are ignored in such analyses. On the other hand, as suggested by Touchon et al. (2014), the next step in the process of strains’ classification should be focused on confrontation of the genetic and phenotypic features related to pathogenicity of bacterial species. To address some of the above features, an optimized method for A. baumannii differential analysis is proposed in this report. It is based on combination of the previously described method based on analysis of repeated sequences (Nowak-Zaleska et al. 2008, 2016) and whole genome alignment.

Materials and methods

Bacterial strains

We used 51 A. baumannii isolates from diagnostic materials of the hospital environments of Antoni Jurasz University Hospital in Bydgoszcz. These isolates were derived from 11 hospital wards (Dermatology, Endocrinology, Geriatrics, General and Endocrine Surgery, General and Vascular Surgery, Intensive Care Units, Neurology, Nephrology, Neurosurgery, Orthopedic, Plastic Surgery), 2 clinics (Orthopedic Outpatient Clinic, Surgical Outpatient Clinic), and Rehabilitation Department. The isolates were collected during the period of 2003–2006 (Table 1). The following strains were isolated from different diagnostic materials: 10 from bronchoalveolar lavages, 8 from bedsores, 2 from blood, 1 from cerebrospinal fluid, 2 from drains, 2 from drain swabs, 1 from needle tip, 1 from pus, 8 from respiratory secretions, 2 from tracheostomy tube swabs, 1 from tube swab, 9 from ulceration wounds, and 4 from urine. Strains were identified based on ID GN phenotypic identification system, including drug sensitivity. This identification was conducted using Kirby–Bauer method, according to CLSI instructions (for details, see Nowak-Zaleska et al. 2008, 2016).

Table 1 Characteristics of multidrug-resistant Acinetobacter baumannii clinical isolates
Full size table

Locus identification with repeated sequences

The isolates of A. baumannii were differentiated on the basis of previously published polymorphisms of repeated sequences located in the CRISPR region (Touchon et al. 2014), variation in the gene encoding the EmrA homologue of E. coli (Nowak-Zaleska et al. 2016), and three newly identified (in this study) polymorphic regions (Tables 2 and 3).

Table 2 The sizes of PCR products for designed pairs of primers calculated for selected Acinetobacter baumannii genomes
Full size table
Table 3 Identification of proteins within amplified genomic regions of Acinetobacter baumannii MDR-TJ strain
Full size table

DNA-technology methods

The genetic material from the isolates was obtained using Genomic Mini Set, purchased from A&A Biotechnology (Gdynia, Poland), following the manufacturer’s instruction. For the DR-PCR/RFLP genotyping method, sequences of primers, the PCR reaction conditions, and enzymatic digestion of PCR products were previously described (Nowak-Zaleska et al. 2008). Briefly, the amplification reactions were conducted according to the following time–temperature profile: 94 °C for 2 min, during the initial denaturation step, 35 cycles consisting of the DNA denaturation at 94 °C for 1 min, hybridization at 68 °C for 1 min, and extension at 72 °C for 2 min. The amplification products were subjected to the restriction fragment length polymorphism (RFLP) analysis using HaeIII and SsiI restriction enzymes. Separation of restriction fragments was performed electrophoretically, in 12% polyacrylamide gels, and results were documented using Versa Doc Imaging System, ver. 1000. The homologous region of the emrA resistance-related gene, containing 6-nt repeats, was analyzed as described previously (Nowak-Zaleska et al. 2016). Identification of three newly discovered polymorphic regions was possible after multiple alignment of nine A. baumannii genomes (see Table 2), using the MAFFT 7.271 software (Katoh et al. 2002). Subsequently, three pairs of primers, shown in Table 3, were used in the PCR analysis. The PCR reactions were conducted in 25 μl reaction mixtures, using the Eppendorf AG 22,331 thermal cycler. The PCR mixtures were as follows: 1.5 U of RUN DNA polymerase (purchased from A&A Biotechnology), PCR reaction buffer containing 10 mM KCl, 10 mM (NH4)2SO4, 0.1% Triton X-100, 20 mM Tris, pH 8.5, 2 mM of Mg2Cl, 2 mM of each deoxynucleoside triphosphates, 25 pM of suitable pairs of primers, and 50 ng/μl of template DNA. Amplified PCR products were separated using 2% agarose gel electrophoresis and standard ethidium bromide staining procedure (Sambrook et al. 1989). Images of the gels were obtained using Versa Doc Imaging System, ver. 1000.

Statistical analysis

Statistical analysis was performed using Epi Info 7.2.3.1 software using two-tailed Fisher exact test analysis. The values “1” and “0” were representing resistant and susceptible strains for different antibiotics used in our study. Similarity matrices of different genotypes and resistance features and phylogenetic trees were constructed using package MVSP ver. 3.22.

Results and discussion

To enhance the currently available methods of differentiation of A. baumannii strains, we were searching for previously unknown PCR-derived fragment length polymorphism variations in randomly identified regions of selected genomic sequences. The theoretical values of PCR fragment lengths of the newly discovered polymorphic regions for nine A. baumannii genomes are presented in Table 2. Among three identified polymorphic regions, only one was characterized by the highest length polymorphism. It was recognized as a gene fragment coding for DNA polymerase III subunit gamma/tau, with the Protein_id = AFI95102.1 in the MDR-TJ A. baumannii genome (GenBank accession no. CP003500.1) (Table 3).

In order to increase variation of analyzed A. baumannii isolates, two other previously described variable regions in the genomes of A. baumannii were included in our study (Nowak-Zaleska et al. 2008, 2016). The combined application of the three genetic aforementioned genotypic methods, DR-PCR/RFLP, different number of P-A dipeptide repeats encoded in the N-terminal part of EmrA-homologue gene, as well as three new variables, namely, Aci7 and Aci8, Aci13 and Aci14, and Aci17 and Aci18 (Table 4), combined with known information about resistance patterns for each isolate (Table 5), allowed for recognition of 31 different clusters shown in Table 1.

Table 4 Set of different genotypes shown as PCR length polymorphisms in nucleotide base pairs for 51 MDR Acinetobacter baumannii isolates
Full size table
Table 5 Set of different antibiotic resistance patterns determined for 51 MDR Acinetobacter baumannii strains
Full size table

Detailed analysis of bacterial isolates and diagnostic material revealed significant differences between A. baumannii isolates from bronchoalveolar lavage (BAL) and other clinical samples (p < 0.0001), as well as significant correlation between resistance pattern II and genotype 8th (p < 0.01), presented in Table 1. In addition, significant correlation (p < 0.05) between the frequency of occurrence of 8th A. baumannii genotype in the first trimester of 2006 year in comparison to other periods of isolation time was also evident. Higher Hunter–Gaston Discriminatory Index (HGDI), presented in Table 1, was determined using the method developed in this study, in comparison to previously published genotyping methods (Nowak-Zaleska et al. 2008, 2016). Furthermore, in the course of statistical data analysis, we observed that strains representing clusters 2 and 4 from combined genetic–phenotypic analysis, shown in Table 1, were isolated in two consecutive years 2005 and 2006 (p < 0.0001). These strains represent the 15th genotype pattern, which was present in 2004 and 2005, but with different resistance patterns II, VI, IX, X, XII, and XIII (p = 0.01). In addition, three strains representing 21st cluster with the resistance pattern X appeared in years 2004 and 2006. Moreover, strains with genotypes 6, 12, and 14 were only present in 2003 (p < 0.0001), in comparison to other genotypes, and what is interesting, the resistance pattern II appeared each year, while patterns V, VI, VII, and VIII appeared only between 2003 and 2005 (p = 0.01).

Combined analysis of similarity matrices, obtained using data from Tables 4 and 5, revealed that out of 19 combinations of genetic and resistance markers, only three were significantly different (p < 0.05) (Table 6), as indicated by χ2 value higher than 4, obtained from two phylogenetic trees presented in Fig. 1. Among significantly different mixed parameters identified, there were (1) 172 bp DNA insertion, located in the CRISPR locus, identified using the SsiI enzyme for genotypes 7 to 11, in combination with resistance to chloramphenicol and gentamycin; (2) 45 and 55 bp DNA insertions in the same locus, identified using the HaeIII enzyme, combined with trimethoprim/sulfamethoxazole resistance or susceptibility patterns; and (3) 184 or 1374 bps DNA length polymorphisms in the second genomic region (see tree new PCR region, Table 4), identified in our study for genotypes 1 to 6 and 7 to 16, in combination with imipenem resistance, characteristic for pattern I or susceptibility features, characteristic for other patterns (Table 6).

Table 6 Set of two joined-similarity matrices obtained for 19 different genotypes indicated by underlined values, and for 19 different antibiotic resistance patterns. All values are from the range between 1 and 100%. Abbreviations "_s" and "_r" indicate intermediate resistance patterns considered two times as susceptible or resistant, respectively. The "0" value was replaced by "1E-06" for diagonal correlation calculation purposes. Significant (p < 0.05) combinations of genetic and resistance/susceptibility features are highlighted in black
Full size table
Fig. 1
figure 1

Phylogenetic trees for different pairs of genetic polymorphisms and resistance/susceptibility features. Branches order obtained based on nearest neighbor method and length–distance calculation based on χ2 method. Significant (p < 0.05) differences indicated in black boxes were identified based on cut-off χ2 value = 4

Full size image

The presence of A. baumannii genotypes over a period of 4 years in the hospital wards (Table 7), and location of determined genotypes over a 4-year period in hospital wards (Table 8), was also assessed. This analysis provides a possibility to identify the presence of specific isolates in various wards over the period of several years. We suggest that such analyses may be useful in epidemiological studies on the origin and migration of particular bacterial strains between different wards of an investigated hospital. It also gives the possibility to analyze the strains regardless of the time period in which they were collected.

Table 7 Presence of A. baumannii genotypes over a period of 4 years
Full size table
Table 8 Location of determined genotypes over a 4-year period in hospital wards
Full size table

Conclusions

In conclusion, 16 different genotypes out of 51 MDR A. baumannii clinical isolates were identified in our study. Based on combined comparative analysis of genetic and resistance patterns, two significantly different patterns of DNA polymorphisms in the CRISPR coding region, resistance to chloramphenicol and gentamycin features, and resistance or susceptibility to trimethoprim/sulfamethoxazole, specific groups of isolates were identified. Out of 19 genetic markers and antibiotic resistance features, three of them were shown to be statistically significantly different using two statistical tools (Table 6, Fig. 1). In addition, 184 or 1374 bp DNA length polymorphisms in genomic region no. 2, located upstream of the GTP cyclohydrolase I gene, with the Locus_tag = "ABTJ_01152", associated in 94% with susceptibility to imipenem, was identified. Finally, the highest genetic diversity, determined within the DNA polymerase III subunit gamma/tau gene, can be recommended for future genotyping of multidrug-resistant A. baumannii strains. We suggest that the optimized methods, proposed in this report and based on combination of Repeated Sequences and Whole Genome Alignment Differential Analysis (RS&WGADA), can be useful in epidemiological studies concerning specific strains of pathogenic bacteria present in investigated hospitals.