Carbapenem Resistance in Acinetobacter baumannii and Other Acinetobacter spp. Causing Neonatal Sepsis: Focus on NDM-1 and Its Linkage to ISAba125

Carbapenem-resistant determinants and their surrounding genetic structure were studied in Acinetobacter spp. from neonatal sepsis cases collected over 7 years at a tertiary care hospital. Acinetobacter spp. (n = 68) were identified by ARDRA followed by susceptibility tests. Oxacillinases, metallo-β-lactamases (MBLs), extended-spectrum β-lactamases and AmpCs, were detected phenotypically and/or by PCR followed by DNA sequencing. Transconjugants possessing the blaNDM−1(New Delhi metallo-β-lactamase) underwent further analysis for plasmids, integrons and associated genes. Genetic environment of the carbapenemases were studied by PCR mapping and DNA sequencing. Multivariate logistic regression was used to identify risk factors for sepsis caused by NDM-1-harboring organisms. A. baumannii (72%) was the predominant species followed by A. calcoaceticus (10%), A. lwoffii (6%), A. nosocomialis (3%), A. junni (3%), A. variabilis (3%), A. haemolyticus (2%), and 14TU (2%). Fifty six percent of the isolates were meropenem-resistant. Oxacillinases present were OXA-23-like, OXA-58-like and OXA-51-like, predominately in A. baumannii. NDM-1 was the dominant MBL (22%) across different Acinetobacter spp. Isolates harboring NDM-1 also possessed bla(VIM−2, PER−1, VEB−2, CTX−M−15), armA, aac(6′)Ib, aac(6′)Ib-cr genes. blaNDM−1was organized in a composite transposon between two copies of ISAba125 in the isolates irrespective of the species. Further, OXA-23-like gene and OXA-58-like genes were linked with ISAba1 and ISAba3 respectively. Isolates were clonally diverse. Integrons were variable in sequence but not associated with carbapenem resistance. Most commonly found genes in the 5′ and 3′conserved segment were aminoglycoside resistance genes (aadB, aadA2, aac4′), non-enzymatic chloramphenicol resistance gene (cmlA1g) and ADP-ribosylation genes (arr2, arr3). Outborn neonates had a significantly higher incidence of sepsis due to NDM-1 harboring isolates than their inborn counterparts. This study demonstrates the significance of both A. baumannii and other species of Acinetobacter in cases of neonatal sepsis over an extended period. Oxacillinases and blaNDM−1 are the major contributors to carbapenem resistance. The dissemination of the blaNDM−1 is likely linked to Tn125 in diverse clones of the isolates.


INTRODUCTION
Acinetobacter, a non-fermenting Gram-negative coccobacilli, has been noted as a pathogen in neonatal units in several parts of the globe and outbreaks caused by this organism have also been reported (Huang et al., 2002;Chan et al., 2007;Hammerum et al., 2015). Though A. baumannii has been predominantly implicated in neonatal infections or sepsis, "other species" within the genus have also been reported to have caused sepsis in neonates (Beaufort et al., 1999;Kilic et al., 2008).The ability of the species to acquire antibiotic resistance has now resulted in the acquisition of the potent carbapenem resistant gene, bla NDM−1 (Bonnin et al., 2012;Boulanger et al., 2012;Chuanfu et al., 2013). This enzyme, New Delhi Metallo-β-lactamase-1 (NDM-1), is an addition to the wide array of antimicrobial resistance mechanisms that have been described for Acinetobacter spp.
There have been reports of NDM-1 in Acinetobacter from different countries . Variants of this enzyme from Acinetobacter have also been reported (Espinal et al., 2011). A study has shown that bla NDM−1 is possibly a chimera that was constructed in A. baumannii (Dortet et al., 2014). Though most studies report the presence of this enzyme in A. baumannii, NDM-1 has also been detected in other species such as A. lwoffii and A. pittii (Hu et al., 2012;Pasteran et al., 2014). Apart from NDM-1, carbapenem resistance in Acinetobacter is mediated predominantly by the carbapenemhydrolyzing-oxacillinases belonging to molecular class D (OXAtype carbapenemases) and also by other metallo-β-lactamases (MBLs) belonging to the class B enzymes such as the IMP-and VIM-types.
In addition to the antibiotic resistance, the identification within this genus remains difficult (Vaneechoutte et al., 1995). Acinetobacter genus consists of more than 30 species (Turton et al., 2011) and identification requires molecular methods which are not available in the clinical laboratories in developing countries. Thus, these organisms are not identified to the species level using appropriate methods. Until this is done, the association of the various species to nosocomial cases can never be achieved.
Most investigations of Acinetobacter in neonatal units have been ad hoc studies that were triggered by an outbreak. There is a dearth of studies that focus on neonatal infections caused by Acinetobacter for an extended period of time. In this study we investigate (i) the prevalence of Acinetobacter in cases of neonatal sepsis for a period of about 7 years, (ii) the carbapenemresistant determinants and the genetic context of bla NDM−1 in Acinetobacter, (iii) the transmission of bla NDM−1 and (iv) association of clinical factors with episodes of sepsis caused due to NDM-1-harboring isolates. The emergence of NDM-1 in this unit makes it relevant to carry out such a study. The use of molecular methods in this study to discriminate the different species of Acinetobacter also gives an insight about the different species of Acinetobacter as a causative agent of neonatal sepsis.

Setting and Patients
The specimens were collected from the neonatal intensive care unit of the IPGMER and SSKM Hospital, Kolkata, India from 2007(January) to 2014(June). Due to some unavoidable circumstances specimens could not be collected from January 2012 to June 2012. This unit has a 16 bed Level III unit, 26 bed Level II unit and 8 bed neonatal surgical units. This unit had 1185 admissions per year (departmental census 2013), including both intramural and extramural births. Blood cultures of the neonates having sepsis were processed by previously described methods (Roy et al., 2013). In brief, 1 ml of blood for culture was drawn with aseptic precautions from a peripheral vein into Peds Plus vials. Blood culture was performed with a BACTEC9050 system (Becton Dickinson, Sparks, MD, USA) from 2008 onwards and manually prior to that. For any culture that flagged positive, Gram-staining was performed and subculture was done on appropriate media based on Gram's stain: MacConkey agar (Difco Laboratories, Detroit, MI, USA) and 5% sheep blood agar (Difco Laboratories) for Gram-negative and Gram-positive organisms, respectively. Clinical data were collected from the hospital registers.

Identification of Acinetobacter
The identity of the Acinetobacter spp. was confirmed by VITEK 2 compact system (BioMérieux, Marcy l'Etoile, France) from 2012 onwards and prior to that by the Mini API system (BioMérieux). Molecular identification of all Acinetobacter spp. was carried out by the ARDRA (amplified ribosomal DNA restriction analysis) method (Dijkshoorn et al., 1998).

Antimicrobial Susceptibility and MIC
Antimicrobial susceptibility testing was done by the VITEK 2 compact system (BioMérieux) and results were interpreted according to the custom set parameters (CLSI, 2014, FDA guidelines and natural resistance interpretation).
The MIC values (mg/L) of meropenem and ceftazidime were also determined using Etest method (AB Biodisk, Solna, Sweden) and were interpreted according to CLSI (CLSI, 2014). The MIC 50 and MIC 90 for meropenem and ceftazidime were calculated as the MIC at which 50 and 90% of the isolates were inhibited.

Detection of β-Lactamase and Carbapenemase Phenotypes
The production of ESBL, MBL, AmpC was detected by the commercial diagnostic tablets (Rosco Diagnostica A/S, Taastrup, Denmark) following the manufacturer's instructions. The results were interpreted as follows: an increase in ≥5 mm of inhibition zone diameter around boronic acid and dipicoloinic acid, in comparison to the diameter with meropenem alone was considered a positive result for AmpC and MBL respectively, while an increase in ≥5 mm in diameter around clavulanic acid and cloxacillin against cefotaxime alone suggests ESBL and AmpC production respectively (Giske et al., 2010).  (Ellington et al., 2007;Roy et al., 2011a;Datta et al., 2014) extendedspectrum-β-lactamase genes (bla CTX−M,VEB,PER ) Saladin et al., 2002;Colom et al., 2003;Woodford et al., 2006) and bla AmpC genes (Perilli et al., 2007) (Table S1).

Conjugation and Electro-Transformation of bla NDM-1
Conjugal transfer of bla NDM−1 to the sodium azide-resistant E. coli J53 recipient (Pasteran et al., 2014) was attempted by a solid mating assay . Transconjugants were selected with varying concentrations of meropenem (0.1-2 mg/L) and 100 mg/L of sodium azide. Electro-transformation into commercially available MAX Efficiency R DH10B TM Competent Cells (Life Technologies, Carlsbad, CA, USA) was carried out for non-conjugative strains using 0.1 mg/L meropenem.
Susceptibility test for ceftazidime and meropenem was performed for the bla NDM−1 carrying transconjugants and transformants by Etest method and was followed by PCRs.

Plasmid Characterization
Plasmids harbored by the donor (bla NDM−1 -harboring isolates), transconjugants or transformants were purified by Kado and Liu method, 1981 and were subsequently assessed for the number and sizes (approx) using E. coli V517 and S. flexneri 2a YSH6000 (Roy et al., 2011b) as mega-plasmid marker (Kado and Liu, 1981).

Sequencing of the Amplified Product
Amplified products of bla NDM , bla VIM , bla VEB , bla PER , bla AmpC , bla CTX−M were sequenced directly on both strands using Big dye Terminator v3.1 Cycle Sequencing Kit and analyzed with an automated sequencer (ABI 3730 DNA Analyzer, Perkin Elmer, USA). The bla NDM in some isolates could not be directly sequenced and were cloned into a TOPO TA cloning vector (Life technologies) and sequenced with the M13 primers.

Characterization of Integrons
Integrons were amplified and sequenced with primers derived from the 5 ′ and 3 ′ conserved segments (Novais et al., 2006) and submitted to the INTEGRALL database (http://integrall.bio.ua. pt) for nomenclature. Some of the large products (>1500 bp) were sequenced by primer walking technique (Sverdlov and Tatyana, 2005).

Genetic Environment of the Carbapenem-Resistant Genes
The genetic structures surrounding the bla NDM−1 , were studied by PCR mapping and DNA sequencing using primers based on previously reported structures Bonnin et al., 2012; Table S1). In the NDM-1-possessing transconjugants and transformants, the proximity of the IS elements to genes were determined using combinations of the respective IS primers and the OXA-23-like and OXA-58-like primers (Lin et al., 2010).

Pulsed Field Gel Electrophoresis (PFGE)
PFGE was carried out for all isolates in a CHEF-DR III apparatus (Bio-Rad Laboratories, Hercules and CA) by ApaI enzyme (Abbo et al., 2005) (New England Biolab, Massachusetts).
When indistinguishable isolates of Acinetobacter spp. were cultured from two or more neonates treated in the NICU during an interval of up to 15 days, they were considered as a single episode of cross-transmission (Roy et al., 2010).

Statistical Analysis
The data for the 68 neonates along with all their recorded variables (clinical factors, results of tests, organisms found, antibiotic resistance patterns etc.) were entered in IBM SPSS Statistics Version 20.0 and analyzed systematically using established statistical procedures. Characteristics were described as median and standard deviation (SD) for continuous variables and as frequencies and percentages for categorical variables.
Features like sex, gestational age, birth weight, mode of delivery, use of mechanical ventilator and place of birth (intramural or extramural) that are known to be associated with sepsis were analyzed for association with episodes of sepsis caused due to NDM-1-harboring isolates. Multivariate logistic regression was used to identify the risk factors for infection by NDM-1-harboring organisms. All available clinical factors were entered into the regression simultaneously and a backward selection process was used to identify the risk factors. All relationships were evaluated at a 95% significance level.
Chi-square test of independence was carried out to determine the association of sepsis due to NDM-1 with outcome (death or discharge) and onset of sepsis (early or late). Finally, a simple t-test of proportions was used to test if the incidence of infection via Acinetobacter baumannii or other species was significantly different for each type of antibiotic resistance. All these tests were evaluated at a 95% significance level.

Demographics of the Neonates
Twenty out of 68 (29.4%) neonates from whom Acinetobacter spp. was isolated were female while 46 were male (gender information was missing for 2 neonates). The median gestational age for the neonates was 32.5 weeks (SD 3.4 weeks) and median birth weight was 1363 grams (SD 828 grams). Eighty two percent (56 of 68) of the neonates were pre-term while 77.9% (53 of 68) of the neonates were of low birth weight. Sixty five percent (44 of 68) of the neonates were delivered by cesarean section while the rest had a normal vaginal delivery. Most of the neonates in this unit were born at the facility (inborn)-46 of 68 (67.6%) and the rest were born at other facilities and later transferred to this unit (out born). Sixty two percent of the neonates had early onset sepsis while 31% had late onset of sepsis (time of onset of sepsis could not be recorded with certainty for 5 neonates).

Identification of Acinetobacter spp. by ARDRA
During this period, 68 non-duplicate Acinetobacter spp. were isolated from the blood specimens of 68 septicaemic neonates. All isolates were analyzed in this study. The year-wise breakup of the isolates is depicted in Figure S1 indicating the higher prevalence of Acinetobacter calcoaceticus-baumannii complex (ACB complex) comprising of A. baumannii, A. calcoaceticus, A. nosocomialis, and A. pittii.
The identity of the isolates using the ARDRA method was compared with Mini API or VITEK 2 results. Twenty six of the 38 strains were identified as the same species when results of the mini API and ARDRA (Table S2) were compared. Similarly, 19 of the 30 isolates were assigned as the same species by VITEK 2 and ARDRA (Table S2).
Antimicrobial Susceptibility Tested by VITEK 2 Table 1 summarizes the antibiotic resistance pattern found in Acinetobacter spp. during this period. More than 75% isolates were resistant to extended spectrum cephalosporins (79% isolates were resistant to ceftazidime, 75% isolates were resistant to cefepime), 85% isolates were resistant to aztreonam, 50% were resistant to doripenem, 51% were resistant to both imipenem and meropenem. More than 60% isolates were resistant to at least one aminoglycosides (60% resistant to amikacin, 71% resistant to gentamicin). Eighty four percent isolates were resistant to ciprofloxacin. But all isolates showed high susceptibility to minocycline (99%) and tigecycline (100%).
All the Acinetobacter isolates were categorized into two groups, the ACB complex (A. baumannii, A. calcoaceticus, A. nosocomialis, and A. pittii) and other species (A. lwoffii, A. junni, A. haemolyticus, A. variabilis, and 14TU). Several analyses in this     Frontiers in Microbiology | www.frontiersin.org study were done where comparison between these two groups were made. While analyzing resistance to different groups of antibiotics against these two categories, differences were noted for doripenem ( Table 1).The percentage of doripenem-resistant isolates was significantly higher in ACB complex (55%) than in other species (20%) (p-value 0.0404). However, since the number of strains in the category "other species" is small (n = 10), these differences should not be overemphasized.

Distribution of MIC Values of Different Groups of Antibiotics
There was a marked increase in resistance to carbapenems among Acinetobacter spp. from 25% in 2007 to 87% in 2013 ( Figure S1).The range of MIC values for meropenem and ceftazidime over this period were 0.025− ≥32 mg/L and 1− ≥256 mg/L respectively. For individual years, the range of MIC values is depicted in Table S3.
Over this period the MIC 50 and MIC 90 values for meropenem were 32 mg/L and for ceftazidime were 256 mg/L. Detailed year-wise distribution of MIC 50 and MIC 90 values (mg/L) for meropenem and ceftazidime is shown in Table S3. This result depicted a significant increase in MIC value of meropenem since 2011 whereas for ceftazidime high resistance persisted throughout this period (except for 2010) ( Table S3).

Detection of Different β-Lactamases by Phenotypic Test
Of the 68 isolates, 19, 24, and 28% showed phenotypic evidence of MBLs, ESBLs and AmpCs respectively. Among MBL-producers (n = 13), 50% isolates showed ESBL and AmpC phenotype both and 64% showed only ESBL phenotype. Whereas, among non-MBL (n = 55) phenotype, 18% were ESBL producers, 27% were AmpC producers and 11% produced both of the enzymes. Presence of MBL sometimes complicates the detection of ESBLs and AmpCs. Therefore, the presence of ESBLs and AmpCs in these isolates was confirmed by PCR subsequently.
Detailed distribution of these genes in the different species has been presented in Table 1. bla NDM−1 was present in 21% of the ACB complex (A. calcoaceticus, A. baumannii, A. nosocomialis) and 30% of the other species (A. haemolyticus, A. lwoffii). In contrast, most of the oxacillinases were confined within ACB complex except 2 isolates of A. lwoffii carried either bla OXA−23−like or bla OXA−58−like gene.
While correlating the phenotypic and genotypic test results of all the strains, it was observed that one strain that could not be detected as MBL producers by phenotypic tests harbored bla NDM−1 (A_134) as detected by PCR. Similarly, there were 8 strains which did not show the presence of ESBL by the phenotypic tests but harbored bla PER−1 or bla CTXM−15 . Of these 8 strains, 3 showed the presence of NDM-1. Similarly, 17 of the AmpC non-producers (as tested by the phenotypic tests) harbored bla AmpC gene and 5 of these strains also possessed NDM-1. This showed the failure to detect ESBLs and AmpCs by phenotypic tests particularly in MBL-producing strains.

Diversity of the Acinetobacter Isolates
The analyses of the pulsotypes were done separately for A. baumannii, A. calcoaceticus, and A. lwoffii. Strains of A. baumannii were primarily diverse and dissemination of any particular clone was not evident (Figure 1). One sporadic case of cross-transmission was seen: A_151 (DOB 10/01/2013) and A_153 (DOB 24/01/2013). A few pairs of strains were found to be indistinguishable but could not be considered as episodes of cross-transmission as they were isolated from neonates who were not treated in the NICU during an interval of 15 days (Figure 1). In A. calcoaceticus and A. lwoffii group no cross-transmission was noted and all the isolates were distinct.
The number of isolates in other groups (A. junni, A. nosocomialis, 14TU, A. variabilis, A. haemolyticus) was small for cluster analysis and were visually analyzed. All were diverse.

Characterization of the bla NDM-1 -Harboring Isolates
Twenty two percent (n = 15) of the isolates harbored NDM-1 across all species of Acinetobacter. The presence of NDM-1 was not significantly different between the ACB complex and the other species (p-value: 0.7640). Most of the isolates (n = 11) harboring NDM-1 had an MIC value ≥32 mg/L for meropenem except for a few that had low MIC values (0.125, 0.25, 0.5, 1 mg/L) ( Table 2). One of the isolates, A_117 (0.5 mg/L) had at least three carbapenem resistant determinants bla NDM−1 , bla OXA−23−like and bla VIM−2 gene. All NDM-1-harboring isolates were resistant to a series of antibiotics except tigecycline and colistin.
PFGE of the A. baumannii and A. calcoaceticus isolates possessing NDM-1 revealed that diverse clones of these species harbored NDM-1 (Figure 1). No cases of cross-transmission were noted in the NDM-1 isolates.,
A single plasmid was detected in all of the bla NDM−1 -harboring transconjugants and the sizes of these plasmids ranged between 86 and 213 kb ( Table 3).
Investigation of the immediate genetic environment of bla NDM−1 gene revealed the presence of a conserved structure that always associates with the complete transposon, Tn125. This structure starts with one copy of the ISAba125 at the 5 ′ end of bla NDM−1 gene and ble MBL gene followed by trpF, tat, cutA1, groES, groEL, insE, and ISAba125 at the 3 ′ end. All strains showed similar genetic structure surrounding the bla NDM−1 . As the primer walking and PCR results were consistent for all transconjugants, sequences were thoroughly analyzed for 4 isolates (Figure 2).
Analysis of the integrons revealed different combinations of resistance genes bracketed between the 5 ′ and 3 ′ conserved region and each of these different resistance gene cassettes are presented in Figure 3. Most commonly found genes in this conserved segment were aminoglycoside resistance (aadB, aadA2, aac4 ′ ), non-enzymatic chloramphenicol resistance gene (cmlA1g) and ADP-ribosylation gene (arr2, arr3).

Association of Clinical Factors and Mortality of the Neonates with the Presence of NDM-1-Harboring Acinetobacter
The results of a comparison of neonates with sepsis caused due to NDM-1-harboring Acinetobacter with those without NDM-1 is shown in Table 4. Multivariate logistic regression established that neonates who were outborn had a significantly higher incidence of NDM-1 harboring isolates than their inborn counterparts [Odds Ratio (OR) 2.39; p-value <0.001]. None of the other factors were found to be statistically significant.

DISCUSSION
The data show that the potential impact of Acinetobacter sepsis on neonates in a tertiary care center is significant. There has been an increase of sepsis due to lactose non-fermenting Gram-negative bacteria in recent years and Acinetobacter is an important contributor to this burden (Viswanathan et al., 2014). This study also shows that though A. baumannii is still by far the most common of the Acinetobacter species causing sepsis in this unit, other species are also emerging in the clinical settings. Though infections due to other species of Acinetobacter has been reported (Turton et al., 2011;Tien et al., 2012), infection in the neonatal population caused by species of Acinetobacter other than A. baumannii are rare. The use of the ARDRA method in this study has helped in the elucidation of the species providing an insight about the presence of the other species of Acinetobacter in cases of neonatal septicaemia.
Another issue that emerges from the study is drug resistance. Carbapenem resistance is high and the clinician is left with very little choice of antibiotics. The only antibiotic that still retains activity is tigecycline, an observation that has also been made by other authors (Golanbar et al., 2011;Hsieh et al., 2014). In this unit, piperacillin/tazobactum and amikacin are being used as a pre-emptive (first line) therapy for clinically suspected cases of sepsis. As a second line of therapy ofloxacin and amikacin are used in the unit. In rare cases where the above therapies are ineffective and the antibiogram results are still awaited, meropenem is used.
Further, this study also shows that the phenotypic detection of ESBLs and AmpCs in presence of MBLs is challenging. The failure to detect the ESBLs in the MBL-producing clinical isolates may lead to the hidden spread of such β-lactamases complicating the situation even further. The authors have recently devised a method to circumvent this problem (Datta et al., 2015).
Resistance to carbapenems and cephalosporins could be correlated to the diverse repertoire of genetic determinants that were identified. The most prevalent were the oxacillinases. Among the transmissible oxacillinases, OXA-23-like was the most common and this result is confirmatory of other studies (Woodford et al., 2006;Golanbar et al., 2011). The presence of the oxacillinases was significantly higher in the ACB complex compared to the other Acinetobacter species. The presence of ESBLs like PER-1, VEB-2 and CTXM-15 were also noted along with the carbapenemases in some isolates but compared to the carbapenemases their numbers were lower (Ben et al., 2011). Of the class B β-lactamases, only NDM-1 and VIM-2 were identified. Most earlier reports of NDM-1 or a variant of the gene in clinical settings have been in A. baumannii (Wang et al., 2007;Boulanger et al., 2012;Revathi et al., 2013) but the presence of NDM-1 in this study was noted across other species of Acinetobacter along with A. baumannii.
PFGE revealed that NDM-1 harboring isolates were diverse. Acinetobacter spp. possessing NDM-1 was isolated predominantly from neonates delivered at other (extramural) centers reflecting the dissemination of this gene but not any clone in particular. A study with A. baumannii isolates collected across Europe also reported diversity of clones in Acinetobacter infection (Bonnin et al., 2012). Studies that have reported the presence of NDM-1 in other genuses have also described similar diversity of clones (Pasteran et al., 2014).
The bla NDM−1 gene was organized in a composite transposon bracketed between two copies of ISAba125 in all the isolates irrespective of the species. This arrangement of the gene in Acinetobacter has been reported earlier (Pfeifer et al., 2011;Bonnin et al., 2012;Fu et al., 2012). The similar organization of the bla NDM−1 in the diverse clones of A. baumannii and also the other species lends credence to the hypothesis that the current dissemination of the gene in Acinetobacter is likely linked to Tn125 and not plasmids. The linkage of bla OXA−23−like and bla OXA−58−like to ISAba1 and ISAba3 respectively as observed here, has also been reported in several earlier studies. This shows that the genetic context of these genes in Acinetobacter remain similar even in geographically distant places Lin et al., 2010). Analysis of the class I integron showed that though there was variability in the genetic determinants present between the 5 ′ and 3 ′ CS segments, most of the genes were not associated with carbapenem or cephalosporin resistance (except carb-2/A_157).
A few isolates harboring NDM-1 and oxacillinases had low MIC values (even as low as 0.125 mg/L in one isolate). Similar observation was also reported from earlier studies (Datta et al., 2014;Zmarlicka et al., 2015). Further work is needed to understand the reason for such differences which can occur due to alterations in the expression of the enzymes or other changes in the mechanism of resistance toward β-lactamases.
Statistical analysis indicated that a neonate born in other facilities (outborn) had a significantly higher incidence of NDM-1 harboring isolates than neonates born at this center. This indicates the possibility of the neonates having acquired NDM-1-harboring isolates from other hospitals where they were born. This is supported by the fact that most had a late onset of sepsis, the initial infection probably having acquired from the other hospitals. The clonal diversity of the isolates also lends support to this possibility.
This study documents the significance of both Acinetobacter baumannii and the other species, in cases of neonatal sepsis in a unit over a period of 7 years. We have been able to gain a unique assessment of the diverse genetic determinants responsible for carbapenem resistance in these isolates. The genetic context of the carbapenem-resistant genes also provides insight about the importance of the ISAba125 in the spread of NDM-1 in Acinetobacter. The emergence of NDM-1 among an already existing repertoire of oxacillinases is a challenge for clinicians and microbiologists alike. As new resistance mechanisms constantly evolve, both laboratory detection systems and infection control measures need to be enhanced.

ETHICAL APPROVAL
The study protocol was carefully reviewed and approved by the Institutional Ethics Committee of the National Institute of Cholera and Enteric Diseases (Indian Council of Medical Research) (No. A-1/2013-IEC, dated 10th January 2014). Individual informed consent was waived because this study used currently existing sample collected during the course of routine diagnosis of sepsis and did not pose any additional risks to the patients. The patient records/information was anonymized and de-identified prior to analysis.

FUNDING
The study was supported by Indian Council of Medical Research (ICMR) intramural fund. SC and SD were supported by senior research fellowships from ICMR and SR received a postdoctoral fellowship from ICMR.

ACKNOWLEDGMENTS
We extend our thanks to Dr. Arun K. Singh and the staff of the Department of Neonatology for their support during the study. We also thank the technical staff of central facility of Bose Institute, Kolkata for sequencing data. Our sincere thanks to Dr. Simona Pollini, Dr. George A. Jacoby, Dr. Anne Marie Queenan, Dr. Olivier Moquet, and Dr. Kyungwon Lee for providing us the control strains for PCRs.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2016.01126