Multidrug-Resistant and Extensively Drug-Resistant Acinetobacter baumannii Causing Nosocomial Meningitis in the Neurological Intensive Care Unit

Acinetobacter baumannii is one of the significant healthcare-associated meningitis agents characterized by multidrug resistance and a high mortality risk. Thirty-seven A. baumannii strains were isolated from thirty-seven patients of Moscow neuro-ICU with meningitis in 2013–2020. The death rate was 37.8%. Strain susceptibility to antimicrobials was determined on the Vitek-2 instrument. Whole-genome sequencing was conducted using Illumina technology; the sequence types (ST), capsular types (KL), lipooligosaccharide outer core locus (OCL), antimicrobial resistance genes, and virulence genes were identified. The prevalent ST was ST2, belonging to the international clone IC2, and rarer, ST1, ST19, ST45, ST78, ST106, and ST400, with prevalence of KL9 and OCL1. Twenty-nine strains belonged to multidrug-resistant (MDR) and eight extensively drug-resistant (XDR) categories. Genes conferring resistance to beta-lactams (blaPER, blaGES, blaADC, blaCARB, blaCTX-M, blaTEM, and blaOXA-types), aminoglycosides (aac, aad, ant, aph, and arm), tetracyclines (tet), macrolides (msr and mph), phenicols (cml, cat, and flo), sulfonamides (dfr and sul), rifampin (arr), and antiseptics (qac) were identified. Virulence genes of nine groups (Adherence, Biofilm formation, Enzymes, Immune evasion, Iron uptake, Regulation, Serum resistance, Stress adaptation, and Antiphagocytosis) were detected. The study highlights the heterogeneity in genetic clones, antimicrobial resistance, and virulence genes variability among the agents of A. baumannii meningitis, with the prevalence of the dominant international clone IC2.


Introduction
Carbapenem-resistant Acinetobacter baumannii (CRAB) has recently been labeled as a "critical" pathogen, i.e., constituting a significant global human health risk, in the World Health Organization's list of antimicrobial-resistant bacteria intended to guide the research and development of new effective antibiotic treatments [1]. The recent COVID-19 pandemic was associated with an increasing risk of hospital-acquired ventilator-associated secondary infections, for which A. baumannii was a leading cause [2]. A. baumannii has been frequently Microorganisms 2023, 11, 2020 3 of 21 results are significant for expanding our knowledge about the molecular epidemiology of the nosocomial A. baumannii in critical care units in Russia. The resistome and virulome analysis provides the data that can be used for the development of control strategies against A. baumannii infections and novel antimicrobial and anti-virulence targets for detection systems and treatment.

Research, Bioethical Requirements, and Patients
This study was a retrospective cohort observation analytical study. Under the requirements of the Russian Federation Bioethical Committee, each patient signed informed voluntary consent to treatment and laboratory examination. The study did not contain the personal data of patients, such as the name, date of birth, address, and disease history. The study was a retrospective observational study in the neuro-ICU department in a specialized Neurosurgical Hospital in Moscow, Russia, with 300 beds that care for approximately 8000 patients per year, 95% of whom undergo surgery [20].
Minimal inhibitory concentrations (MICs) of two antiseptics-chlorhexidine digluconate (Sigma-Aldrich, Saint Louis, MO, USA) and benzalkonium chloride (Sigma-Aldrich, Saint Louis, MO, USA)-were determined via the broth microdilution assay on the 96-well plates. Bacterial growth in the wells was detected via the xMARK microplate spectrophotometer (Bio-Rad Laboratories, Inc., Watford, UK). MIC values were compared with antiseptic concentrations recommended for use in ICU: 500 mg/L chlorhexidine digluconate; 10,000 mg/L benzalkonium chloride [22].

Biofilm Formation
The biofilm formation ability of A. baumannii isolates was determined via polystyrene tube assay based on the crystal violet staining method [23]. Briefly, the wells of a 96-well plate containing 0.1 mL of Luria broth (Thermo Fisher Scientific, Waltham, MA, USA) were inoculated with 100 µL of overnight bacterial culture in concentration 1 × 10 7 CFU/mL and incubated at 37 • C for~20 h. The liquid media was discarded, and the adherent cells were washed twice with 0.2 mL of phosphate-buffered saline (PBS) and stained with 0.1% solution of crystal violet for 10 min. The stain was discarded, and the cells were washed three times with 0.2 mL of PBS. The stain was eluted from the adherent cells via 0.2 mL of 96% ethanol for 5 min and transferred in. The absorbance of the eluted stain was measured at 600 nm using the xMark microplate spectrophotometer (Bio-Rad, Hercules, CA, USA). The results were divided into four categories according to the sample optical densities (ODs) in comparison with the control optical densities (ODc): strong biofilm producer (4 × ODc < ODs); medium biofilm producer (2 × ODc < ODs ≤ 4 × ODc); weak biofilm producer (ODc < ODs ≤ 2 × ODc); and non-biofilm producer (ODs ≤ ODc [24].

Whole-Genome Sequencing, Assembly, and Annotation
DNA isolation was performed via the CTAB method [25]. WGS was carried out using Nextera DNA Library Preparation Kit (

Patients and Clinical Data
During the period from January 2013 to Sepember 2020, 37 cases of A. baumannii meningitis (MG) were detected in a Moscow neurosurgery ICU in the patients after neurosurgery operations. Among them were 25 males of 12-75 years (median 46.7) and 12 females of 2-65 years (median 44.8). The diagnoses of the patients were oncological (n = 25), traumatic brain injury (n = 8), and disorders of the cerebral circulation including ruptured aneurysms (n = 4). Meningitis was detected in 37 patients; moreover, 19 respiratory tract infections (RTIs), 7 surgical site infections (SSIs), 5 gastrointestinal dysfunctions (GD), 5 urinary tract infections (UTIs), and 1 blood infection (BI) were detected in combinations from one (MG) to four (MG + RTI + SSI + GD or MG + RTI + UTI + GD) infections. Infections were located in the brain and other body sites simultaneously. The stays in the ICU were 8-229 days (median 67.6 days). The death rate was estimated as 37.8% (14/37 patients) ( Table 1).
Nine strains were resistant to six antimicrobial groups, seven to seven groups, fifteen to eight groups, four to nine groups, and two to ten antimicrobial groups simultaneously. The phenotypic diversity of antibiotic resistance among the strains was very high-23 variants of AMR phenotype were revealed. Carbapenem-resistant strains represented~68% of strains (Table 2).

Antimicrobial Susceptibility
All A. baumannii strains in this study were attributed to two antimicrobial resistance patterns: MDR (n = 29) and XDR (n = 8), according to Magiorakos et al.'s criterium [21]. Major strains were resistant to penicillin and cephalosporins (n = 37), fluoroquinolones (n = 36), aminoglycosides (n = 32), penicillin/beta-lactam inhibitors (n = 28), sulfonamides (n = 26), carbapenems (n = 25), and chloramphenicol (n = 20). At the same time, dominant strains were susceptible to polymyxins (n = 34) and tetracyclines (n = 26) ( Figure 1). Nine strains were resistant to six antimicrobial groups, seven to seven groups, fifteen to eight groups, four to nine groups, and two to ten antimicrobial groups simultaneously. The phenotypic diversity of antibiotic resistance among the strains was very high-23 variants of AMR phenotype were revealed. Carbapenem-resistant strains represented ~68% of strains (Table 2). It was shown that planktonic cells of all A. baumannii strains were sensitive to chlorhexidine digluconate and benzalkonium chloride, with MICs 8-64 mg/L and 4-8 mg/L, respectively. Such levels of susceptibility to antiseptics are within their concentrations recommended for use in ICUs (http://dezreestr.ru/, access date 30 March 2023): 500 mg/L for chlorhexidine digluconate; 10,000 mg/L for benzalkonium chloride (Table 3).

Biofilm Formation
A. baumannii strains were divided into four groups according to their ability for biofilm (BF) formation: strong BF (n = 2); moderate BF (n = 8); weak BF (n = 17); and no BF (n = 10). Notably, strong BF formation was detected for the strains of ST2 only, and moderate BF formation for the strains of ST2, ST1, and ST400 ( Figure 2).

Biofilm Formation
A. baumannii strains were divided into four groups according to their ability for biofilm (BF) formation: strong BF (n = 2); moderate BF (n = 8); weak BF (n = 17); and no BF (n = 10). Notably, strong BF formation was detected for the strains of ST2 only, and moderate BF formation for the strains of ST2, ST1, and ST400 ( Figure 2).
The tet gene of tetracycline efflux MFS-family transporter was detected in 1 strain of ST1 and 18 strains of ST2. The msrE and mphE genes coding ribosomal protection protein and macrolide 2 -phosphotransferase, respectively, were simultaneously identified in the strains of ST2 (n = 11) and ST78 (n = 1). Regarding phenicol resistance markers, the cat (n = 9) genes encoded subunits of chloramphenicol acetyltransferase, and the cmlA (n = 6) and floR (n = 2) genes encoded chloramphenicol/florfenicol efflux MFS-family transporters. AGRs providing resistance to sulfonamides were detected in the strains of all STs: trimethoprim-resistant dihydrofolate reductase dfrA17 gene (n = 9) and the sulfonamide-resistant dihydropteroate synthase genes sul1 (n = 25) and sul2 (n = 11). Three A. baumannii strains of ST2 carried the gene arr-2, encoding the NAD(+)-rifampin ADPribosyltransferase associated with resistance to rifampin. Moreover, the genetic determinant for the antiseptic-resistant qacE gene coding the quaternary ammonium compound efflux SMR-family transporter was detected in the genomes of 25 strains (Figure 3).
A total of 20 types of putative virulence factor genes (VFGs) were predicted in 37 A.
A total of 20 types of putative virulence factor genes (VFGs) were predicted in 37 A. baumannii genomes. The VFG open reading frames (ORFs) associated with Adherence (n = 39), Biofilm formation (n = 537), Enzyme (n = 139), Immune evasion (n = 964), Iron uptake (n = 989), Regulation (n = 124), Serum resistance (n = 32), Stress adaptation (n = 130), and Antiphagocytosis (n = 7) were identified in the genomes (Figure 4).  The VFGs identified in all 37 A. baumannii genomes were the following: ompA, ade, csu, pga, plc, plcD, Capsula locus, LPS locus, Acb locus, two-component system genes, pbpG, and kat. At the same time, the ptmA gene encoding legionaminic and biosynthesis protein A was detected in two strains of ST2 and ST78 only. The bap gene encoding biofilm-associated protein was detected in all strains except two strains of ST1 and four strains of ST19 and ST106. The HU locus was obtained in all strains except one of ST106. The abaI inducer and abaR receptor genes associated with QS were detected in 30 strains. The VFGs identified in all 37 A. baumannii genomes were the following: ompA, ade, csu, pga, plc, plcD, Capsula locus, LPS locus, Acb locus, two-component system genes, pbpG, and kat. At the same time, the ptmA gene encoding legionaminic and biosynthesis protein A was detected in two strains of ST2 and ST78 only. The bap gene encoding biofilm-associated protein was detected in all strains except two strains of ST1 and four strains of ST19 and ST106. The HU locus was obtained in all strains except one of ST106. The abaI inducer and abaR receptor genes associated with QS were detected in 30 strains. The rmlD gene encoding dTDP-6-deoxy-L-lyxo-4-hexulose reductase provides serum resistance, as well as the rfbC gene coding dTDP-4-dehydrorhamnose 3,5-epimerase providing antiphagocytosis, were detected in only one strain of ST106. The wecB and wecC genes encoding UDP-Nacetylglucosamine 2-epimerase (non-hydrolyzing) and UDP-N-acetyl-D-mannosamine dehydrogenase, respectively, providing antiphagocytosis, were identified in the strains of ST19 only. Despite the vast variability of VFGs in the strains under study, the correlation between disease outcomes-patient death or survival-was not obtained.
Upon analyzing Acinetobacter genomes through the VFDB online resource, it was determined that the "katA Catalase (Neisseria)" gene is present in all strains belonging to ST2. This data prompted further examination of kat genes in all studied Acinetobacter genomes. These genes include katA, which encodes a small-subunit mono-functional catalase; katE, the large-subunit mono-functional catalase; katG, the catalase-peroxidase; and katX, a small protein with a catalase-domain. The reference genome of the A. baumannii strain AB5075-UW (CP008706.1) isolated from the tibia/osteomyelitis of the diabetes patient in the USA in 2008 was obtained from GenBank (Table 5).  Note: ST, sequence type; Ident., identity to the reference gene sequence obtained from GenBank; SNP, single nucleotide polymorphism; SAV, single amino acid variation; "-", no match; pseudo, pseudogene.
The katA gene with 100% identity compared to the reference (GB accession number AKA32231.1) was detected in three strains of ST1 and two strains of ST19. All strains of ST2 and ST45 carried the katA gene with 25-26 single nucleotide polymorphisms (SNPs), and only one SNP was significant, resulting in the single amino acid variation (SAV), g367c/A123P. The strains of ST78 and ST400 carried the katA gene with 21 and 23 SNPs, respectively, which caused two SAVs, a211t/T71S and g367c/A123P, in the ST78 strains, as well as g367c/A123P and g649t/V217F in the ST400 strain. The katA gene was not detected in strain B-9401 of ST19, while the pseudogene was identified in strain B-9406 of ST400. It should be noted that SAV g367c/A123P was detected in all variants of the katA gene in this study.
The sequence of the katE gene was matched to the reference (GB accession number AKA32163.1). The 100%-identical genes were detected in stains of ST1 and two strains of ST19. All strains attributed to the ST2 and ST45 carried 35 SNPs in this gene, providing three SAVs: a876c/Q292H; c1304t/A435V; and a1610g/N537S. The strains of ST78 carried the katE gene with 30 SNPs compared to the reference, including two nonsynonymous substitutions, g1303a and c1304t, resulting in one SAV A435I. The strains B-9404 of ST106 and B-9406 of ST400 carried 25 and 24 SNPs, respectively, realizing in one (a1820g/N607S) and two (c1304t/A435V and t1939g/L647V) SAVs, respectively. The katE gene was missing in strain B-9401 of ST19. Interestingly, that SAV in position 435 was common for katE genes in this study, two variants of SAVs were detected: A435V and A435I.

Phylogenetic Analysis
The phylogenetic relationship between 37 A. baumannii genomes determined on the base of the SNPs analysis showed the presence of five main clusters. The first cluster included 24 genome sequences of the strains belonging to ST2 and two strains of ST45; the second cluster included the genomes belonging to ST1 and ST19; while the genomes of the strains belonging to ST78, ST106, and ST400 formed individual clusters each. The phylogenetic analysis also showed that the genomes belonging to ST2 were divided into five sub-clusters, while the genomes of the other STs were combined in a single cluster for a specific type ( Figure 5).

Discussion
The present study characterized 37 Acinetobacter baumannii strains isolated from 37 patients of Moscow neurosurgery ICU with meningitis in 2013-2020, which caused 37.8% of deaths. The overall incidence of nosocomial meningitis in this period was 9.5%; the average proportion of A. baumannii among all nosocomial meningitis causative agents was 22.1%. [31]. Such a high level of mortality is in agreement with reports about A. baumannii as a problematic pathogen to human health due to its high level of drug resistance, increased virulence, and difficulty in treatment, with a high mortality rate (15-73%) because of its multiple drug resistance [32,33].

Discussion
The present study characterized 37 Acinetobacter baumannii strains isolated from 37 patients of Moscow neurosurgery ICU with meningitis in 2013-2020, which caused 37.8% of deaths. The overall incidence of nosocomial meningitis in this period was 9.5%; the average proportion of A. baumannii among all nosocomial meningitis causative agents was 22.1%. [31]. Such a high level of mortality is in agreement with reports about A. bau-mannii as a problematic pathogen to human health due to its high level of drug resistance, increased virulence, and difficulty in treatment, with a high mortality rate (15-73%) because of its multiple drug resistance [32,33].
Seven A. baumannii STs were revealed in the study: ST1, ST2, ST19, ST45, ST78, ST106, and ST400. The prevalent ST2 belonging to the pandemic lineage, named "international clone IC2", was associated in our study with 12 of 14 deaths. ST2 is a well-known dominant type (59%) among the available complete and draft genomes in the GenBank database. This is also consistent with a large number of previous publications that report outbreaks due to IC2s accounting for the bulk of carbapenem-resistant A. baumannii (CRAB) outbreaks, whilst IC2 strains do not seem to be the dominant type in South American countries, Tunisia, Tanzania, Poland, and Japan [34]. Other A. baumannii strains associated with the death of the patients were ST45, belonging to IC2; and ST400, attributed to the distant phylogenetic group. The last ST was previously rare and described for a nosocomial A. baumannii isolate in Germany [35].
Thirteen KLs were identified in our study; the most prevalent was KL9. In A. baumannii, the capsular polysaccharide is a major virulence factor that is important for survival in vitro and in vivo. Specific capsule types are proven to overwhelm mammalian defenses in vivo, including K2 [36]. To date, more than 128KL gene clusters (KL types) have been identified at the KL in A. baumannii genomes [37]. The major A. baumannii outer core locus (OCL) in our study was OCL1, which includes genes for the synthesis of the variable outer core region. This is in agreement with the data of the Kaptive tool that OCL1 is the most common (~74%) amongst isolates attributed to ST1, ST2, ST3, and ST78 [38].
In this study, a majority of the strains were multidrug-resistant MDR, (78%); the rest were extensively drug-resistant XDR, resistant to 6-10 functional groups of antimicrobials. For comparison, the prevalence of XDR A. baumannii isolates in Bulgarian university hospitals during the period from 2014 to 2016 was reported as 12.4%; during the period between 2018 and 2019, it increased to 78.1% [39,40]. It was reported recently that intracranial infections caused by drug-resistant Gram-negative bacilli, including MDR, XDR, or even pandrug-resistant (PDR) A. baumannii, have very serious consequences [41]. The phenotypic diversity of antibiotic resistance among the strains in our study was very high-23 variants of AMR phenotype were revealed. Carbapenem-resistant strains represented 68%. It is known that CRAB infections are generally nosocomial and frequently occur in the intensive care unit (ICU) for up to 31% around the world with high mortality-greater than 50% in some cases [42]. The carbapenem resistance of the strains in this study was supported by eight bla OXA -type carbapenemase genes with the prevalence of bla OXA-23 and bla OXA-80 genes detected in the strains belonging to ST2 only. This finding is in agreement with reports about bla OXA -type carbapenemase genes associated with members of GC1 and GC2 [34]. Moreover, the genes of broad-spectrum beta-lactamases and ESBLs (bla TEM-1 , bla CARB-14 , bla CTX-M-124 , bla PER-1 , bla PER-7 , bla GES-11 , bla GES-12 , and bla ADC-25 ) were detected in our study. Similar sets of beta-lactamase genes were identified in the studies of other authors [43,44].
Additionally, a total of 47 types of antimicrobial resistance genes (ARGs) providing resistance to aminoglycosides, tetracyclines, macrolides, phenicols, sulfonamides, rifamycin, and antiseptics were identified. Aminoglycoside resistance was associated with aac, aad, ant, aph, and arm genes. Recently, many reports have appeared showing that carbapenem resistance in A. baumannii tends to occur in conjunction with aminoglycosides resistance. The co-production of ArmA 16S rRNA methylase and OXA-23 carbapenemase was determined as the most prevalent mechanism conferring MDR in clinical A. baumannii isolates in many countries worldwide, such as Italy, Greece, China, South Korea, India, Yemen, etc. [45]. Among the ARGs providing resistance to other groups of antimicrobials, the tet genes were detected in 19/37 strains in this study, which was approximately the same level as the study in Thailand (56.1%); a similar level of prevalence was also obtained for the arr-2 gene encoding the resistance to rifampin-3/37 strains in our study (9.1%)-in Thailand. Regarding phenicol resistance markers-cat, cmlA, and floR genes-17/37 strains carried these determinants in our study, which was higher than in the Thailand study (8.1%). As well as the prevalence of AGRs providing resistance to sulfonamides, sul1 and sul2 genes were detected in 25/37 strains and 11/37 strains, respectively, in our study, compared to 46.1% and 13.6% in Thailand. In contrast, the msrE and mphE genes providing the macrolide resistance were less common in our study (12/37 strains) than in Thailand (79.6%) [46]. Importantly, three colistin-resistant strains were identified among A. baumannii causing meningitis, all of which were attributed to ST2 as in the previously described colistin-resistant isolates collected from the urinary tract, respiratory tract, skin, soft tissue, and blood of patients in some Russian regions in 2013-2014 [47].
Moreover, the genetic determinant for the antiseptic-resistant qacE gene was detected in the genomes of 25/37 strains in our study, which was higher than the previously published prevalence of this gene in A. baumannii isolates collected from urine, pus, blood, and endotracheal fluid/aspirate in Malaysia (28%). However, the phenotypic resistance of strains in our study was lower (MIC = 4-8 mg/L of benzalkonium chloride; and MIC = 8-64 of chlorhexidine gluconate) than in the Malaysia study (from 12.5 to >50 mg/L and from 25 to >50 mg/L, respectively) [48].
Nine groups of virulence factor genes (VFGs) were identified in A. baumannii in this study, including a total of 20 types of VFGs containing the specific numbers of the open reading frames (ORFs): adherence (n=39); biofilm formation (n = 537); enzyme (n = 139); immune evasion (n = 964); iron uptake (n = 989); regulation (n = 124); serum resistance (n = 32); stress adaptation (n = 130); and antiphagocytosis (n = 7). The adherence group was presented by the ompA gene detected in all strains. A significantly increasing expression of the ompA gene in persisters was recently shown, which proposes ompA as a potential target for drug development against A. baumannii persisters [49]. Additionally, the ptmA gene was detected in two strains of ST2 and ST7; this is in agreement with the opinion that this virulence factor is quite rare [50]. Several genes involved in the biofilm formation of A. baumannii were identified in the strains in this study, such as adeFGH efflux pump genes, bap gene, csu locus, and pga locus. However, the capacity to form biofilms varied among the strains, with the majority of the isolates forming either moderate (8/37) or weak (17/37) biofilms and only a minority (2/37) forming strong biofilms. This is in agreement with other studies aimed at the evaluation of the distribution of VFGs involved in biofilm formation in multi-drug-resistant A. baumannii [51]. The plc and plcD genes contributing to pathogenesis by aiding in the lysis of host cells to facilitate bacterial invasion were also detected in all studied A. baumannii strains. The major differences between the strains were observed in the capsular export region, where the number of genes varied from 11 to 23. These findings were in agreement with recently published data [52]. A number of putative genes associated with acinetobactin biosynthesis were detected in all A. baumannii strains in our study and varied from 16 to 21. In contrast, the heme utilization locus was obtained in all strains except one of ST106. Surprisingly, only this strain of ST106 contained the additional gene rmlD involved in serum resistance, as well as the rfbC gene providing antiphagocytosis activity [53]. Surprisingly, the VFDB online resource revealed "katA Catalase (Neisseria)" in all A. baumannii strains belonging to ST2. A further comprehensive examination of kat genes showed that the strains carried the katA (35/37), katE (36/37), katG (37/37), and katX (23/37) genes. This finding is consistent with current knowledge of the significance of katE and katG genes in H 2 O 2 resistance, as well as the lesser influence of katA and katX genes on peroxide resistance [54]. It should be noted that putative KatE was different in the strains of ST2 and ST45 (Q292H, A435V, and N537S), ST78 (A435I), ST106 (N607S), and ST400 (A435V and L647V) compared to the KatE of ST1 strains. At the same time, putative KatG varied in the strains of ST2 and ST45 (P270S, T422A, P456S, and A457T), ST78 and ST106 (P270S, T422A, P456S, and K505Q), and ST400 (P270S and T422A) compared to ST1 and ST19. This observation demonstrates the ongoing evolution of these virulence factor determinants. The wide variety of the virulence factors identified in the Acinetobacter strains that caused meningitis in neuro-intensive care patients indicates the plasticity of the pathogen genome. This is consistent with the previously described diverse interaction between different pathogenicity factors, such as iron acquisition, cell adherence, cell motility, biofilm formation, and antibiotic susceptibility [55].
In conclusion, this study provides a detailed account of the genetic determinants associated with multidrug resistance and virulence in clinically significant A. baumannii isolates from the patients of a Moscow neurosurgery ICU with meningitis. The sequence type ST2 was the prevalent genetic line in this cohort. Numerous variants of antimicrobial resistance genes and virulence gene sets were revealed in the genomes of meningitis agents. At the same time, the correlation between STs and their molecular-genetic characteristics was obtained. This observation could form the basis for the future design of novel diagnostic and treatment strategies. We propose to continue this study by focusing on the further investigation of antimicrobial resistance islands, plasmid compositions, the localization of the genetic mobile elements, etc.