Evaluation of bacteriophage cocktail on septicaemia caused by colistin-resistant Acinetobacter baumannii in immunocompromised mice model

Acinetobacter baumannii (AB) is a member of the ESKAPE group of bacteria (i.e., Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, AB, Pseudomonas aeruginosa, and Enterobacter species) notoriously known for the antibiotic resistance crisis1. AB are opportunistic pathogens that cause a variety of infections involving the respiratory tract, urinary tract, skin and blood (bacteremia) and ventilator-associated pneumonia apart from local wound infections2. AB isolates have been reported to acquire resistance to almost all the antibiotics, including carbapenems as well as colistin12. AB is a part of commensal flora of human body, however, it may also be present in the hospital environment due to its’ capability to resist desiccation and disinfectants2. The infections due to multidrug-resistant (MDR)-AB and extensively drug-resistant (XDR)-AB are reported to be causing high morbidity, and mortality in patients admitted to intensive care units (ICUs). Infections caused by the drug-resistant AB compel for the administration of costly and toxic antibiotics and longer stay in the hospitals, leading to high treatment costs and occasional deaths2. Since many of the strains of this bacterium have achieved superbug state3, it has become an exigency to look for alternatives to antibiotics. Bacteriophage (phage) therapy may be one such alternative for severe infections caused by MDR- and XDR-AB strains4. Phage therapy is highly specific and effective in lysing targeted drug-resistant bacteria4. The pre-clinical evaluation of phage therapy in animal models is, however, an essential step before venturing into clinical trials. Artificial burns, pneumonia, pulmonary infections with various microbes in rodent models have successfully been treated with phages567. Although several AB specific phages have been isolated and characterized in terms of their potential therapeutic applications8910, only a few in vivo studies in animal models are available in literature and majority of these are restricted to local infections1112. One study in animal model regarding systemic use of AB specific bacteriophages showed the use of single phage simultaneously with the bacterial challenge13. On similar lines, Wang et al14 demonstrated good efficacy with simultaneous injection of a single phage in their study. These studies have, however, not simulated the real clinical conditions. In clinical settings, patients with AB septicaemia reporting to ICU may be of variable duration (i.e., ranging from recent infection to as old as many hours/days). A recent report regarding systemic/local use of AB specific in a clinical case with disseminated infection is one of the most highlighted ones in the field of bacteriophage therapy15. However, in this case report also the doses of bacteriophage cocktail were used randomly. While working with the bacteriophage therapy for P. aeruginosa infection in murine burn model during 2008-2011, it was observed that there were deaths in certain situations when fixed dosage of bacterial challenge and phage cocktail were given at different time points of the disease16. It could be speculated that a sudden lysis of bacteria in the systemic circulation might have resulted in an endotoxic crisis due to the optimum concentration of phage versus bacterial load. The other issue is the evolution of resistant mutants during phage therapy when a single bacteriophage is being administered. The use of a single phage has been found to be unsuccessful in solving the above issue17. Also, one cannot be sure regarding the dose of the phage cocktail if the patient reports to the ICU with variable bacterial load because of variations in the reporting time. Currently, no data are available regarding doses and dose schedule in such a situation. Therefore, this study was planned to evaluate the different doses of the cocktail of lytic bacteriophages on septicaemia of variable duration caused by AB in an immunocompromised mouse model.

Material & Methods

Sample collection: A total of 140 AB isolates comprising of 70 each from clinical and environmental origin, were screened for antibiotic susceptibility. The clinical strains were isolated from non-duplicate specimens of blood, pus, urine, sputum, nasal aspirate, tips of the endotracheal tubes, pericardial and pleural fluids of patients admitted to ICU of a tertiary care hospital of Banaras Hindu University, Varanasi, India. The study was carried out in the department of Microbiology, Institute of Medical Sciences, Banaras Hindu University between December 2014 to July 2017 after seeking clearance from the Institutional Animal Ethics Committee. The 70 environmental AB isolates were from water specimens of sewer, ponds (3 different sites of sewer of University hospital and 3 sites from Assi nala and 6 ponds) and river Ganga (6 different ghats of river Ganga at different times) in Varanasi. The isolation media used was Acinetobacter selective medium (MDR Leeds Acinetobacter medium containing cefsulodin, cephradine and vancomycin as selective components, Himedia, Mumbai).

Bacterial isolates and their identification: All the clinical and environmental isolates were identified as AB by using the standard biochemical and molecular method21. In multiplex PCR, two target regions of the chromosome of AB were selected i.e. recA gene of Acinetobacter species yielding 425 bp which was genus specific and 16S-23S ribosomal DNA intergenic spacer region yielding 208 bp amplicon for AB specific identification following the methods described earlier18. Further study was conducted only on AB confirmed isolates where amplification for both genus-specific and species-specific amplification could be seen.

The minimum inhibitory concentration of imipenem, meropenem, levofloxacin, tigecycline, polymyxin-B and polymyxin-E (colistin) against AB was carried out by broth dilution method following the recommendation of Clinical and Laboratory Standards Institute (2015)19. All the antibiotic powders were procured from HiMedia Pvt. Ltd., Mumbai. Known reference strain of AB ATCC (19606) was used as a positive control. Since minimum inhibitory concentration (MIC) breakpoint for tigecycline against AB was not mentioned in the guideline, the breakpoint mentioned for P. aeruginosa was considered.

Phage isolation and purification: Isolation of bacteriophages was done from different water sources (river, ponds and sewer) by using double agar overlay method with slight modification as described earlier20. In brief, for isolation of bacteriophages, the AB was plated as lawn culture (10⁸ cfu/ml) on Muller-Hinton agar (MHA) (HiMedia). Water specimens from different water bodies were treated with one per cent chloroform (v/v) for 20 min and centrifuged for 15 min at 10,778×g. The supernatant in the volume of one ml was flooded on the five hour old lawn culture growth (log phage) of the AB (isolated strains from different hosts) on 90 mm nutrient agar plate and incubated overnight at 37°C. Next day, the lawn was washed with three ml TMG (Tris-HCl, magnesium sulphate, gelatin pH 7.4) buffer and centrifuged at 10,778×g for 15 min. To the supernatant (1 ml) one drop of chloroform was added and mixed well by vortexing or by inversion for 15 min. Centrifugation was done at 10,778×g for 10 min. The lawn culture in log phase of the host was prepared again, and 100 μl of the supernatant collected as mentioned above was inoculated at 10-12 places to screen for lysis. The surface with clear plaque was cut and collected in one ml of the TMG buffer and propagated further and plaque counting was done by soft agar overlay method21. The single isolated plaque was picked up for further processing. The number of phage particle was increased by the soft agar overlay method. After bulk production, the bacteria were killed with one per cent chloroform and centrifuged. The clear supernatant was preserved at 4°C for further use. For purification (toxin-free) and concentration of phages, the harvested fluid was subjected to membrane dialysis-135 against polyethylene glycol (PEG 6000, HiMedia) (20% in 2.5M NaCl) for overnight and then washed thrice with phosphate buffered saline (PBS) at 4°C. The endotoxin estimation was done using enzyme-linked immunosorbent assay kit for detection (Thermo Scientific™ Pierce™ LAL Chromogenic Endotoxin Quantitation Kit, MA, USA).

Results

The MIC determination of the recommended antibiotic for clinical use revealed that Ab12 was resistant to all the antibiotics tested, which include imipenem, meropenem, levofloxacin, polymyxin and colistin. Ab12 was used for further experiments.

There were three types of plaques seen on the lawn during the primary isolation. The plaques were either circular with a well-defined margin or having an irregular margin. There was variation in the plaque size also. Overall, 22 bacteriophages were isolated from different water sources against 22 clinical MDR-AB and designated as ɸAb1 to ɸAb22.

Bacteriophage host range determination: Table II shows the percentage of the bacterial isolates lysed after spotting the 22 bacteriophages. When compared with the environmental isolates, the AB from clinical origin were observed to be more susceptible. Furthermore, with the exception ɸAb9 that lyzed A. lwoffii, none of the other phages lysed any other bacterial species tested.

When a dendrogram was prepared based on the susceptibility pattern of 20 randomly selected AB isolates of clinical as well as environmental origin, 22 bacteriophages showed two major clusters I and II. Interestingly, only two pairs of bacteriophage (11 & 18 and 15 & 19) were found to have identical lytic patterns. Rest of the viruses had completely distinct lytic patterns and differed from each other (data not shown).

Genotyping of bacteriophages by ERIC-PCR: ERIC PCR-based whole-genome analysis of 22 phages produced an average of 1-9 bands per phage ranging between 100-4500 bp. When dendrogram was prepared based on the presence and absence of banding pattern in the gel, it was found that only two pairs (1 & 11 and 13 & 15) had an identical banding pattern (data not shown). Rest of the viruses were observed to have unique sets of bands. There were two well-defined clusters I and II where branching occurred at the similarity level of 25 per cent, and the cluster II practically had quite unrelated bacteriophages with a similarity level of 25 per cent. Interestingly, three phages observed to be in the II^nd cluster were common by both the methods of differentiation (ERIC-PCR and bacteriophage host range determination) of phages i.e. ɸ2, ɸ4, and ɸ5 (data not shown).

It was further observed that the six most virulent phages (ɸAb4, ɸAb5, ɸAb7, ɸAb8, ɸAb9 and ɸAb14) of the clinical AB were dissimilar to each other by both the methods, i.e., lytic activity as well as genotyping. Therefore, for pre-clinical evaluation randomly three phages ɸAb4, ɸAb7, ɸAb14 out of the above six phages were selected for the cocktail preparation.

Animal model: The three most active phages lysing 48.6, 52.9 and 47.1 per cent of the clinical isolates respectively, were selected for the in vivo evaluation of their efficacy in treating septicaemia caused by AB in the immunocompromised mouse model. The LD100 could be observed to be 1×10⁸ cfu/ml of Ab12. Bacteriophage cocktail was administered at a constant dose of 100 µl containing 10⁹ pfu/ml.

In group A1 where simultaneous administration of bacteriophage and AB challenge, no mortality occurred and maximum level of illness went up to grade 3, i.e., moderate illness, severe lethargy, ruffled fur and hunched back in three mice after 24 h. These signs improved by 48 h and all the five recovered by 72 h (Table I and Figure). In case of group A2 where bacteriophage cocktail was administered six hours before the bacterial challenge, severe sickness was observed at 12 h and by 24 h post cocktail administration slight illness, lethargy, ruffled fur remained. By 72 h here too all mice recovered (Table I and Figure).

Close

Figure

The outcome of the phage therapy on Acinetobacter baumannii septicaemia at different time points and dosage. Details of study groups: A1- bacterial challenge and phage cocktail given simultaneously, A2-phage cocktail given six hours before bacterial challenge, A3 phage cocktail given six hours after bacterial challenge, A4-phage cocktail given 12 h after bacterial challenge; A5-phage cocktail given 24 h after bacterial challenge with 1x10⁸ cfu/ml, A5.1, A5.2 and A5.3 all with bacterial challenge dose 10⁸ cfu/ml and phage cocktail administered consisted of 10⁷, 10⁶, and 10⁵ cfu/ml.

Export to PPT

In group A3 where bacteriophage cocktail was administered six hours after the bacterial challenge, at 24 h, all the five mice were quite ill and mice no. 3 had exudative accumulation around eyes along with, severe lethargy, ruffled fur and hunched back. This mouse died when examined at 48 h, however, at 72 h the remaining four mice were normal (Table I and Figure). In group A4, where bacteriophage cocktail was administered 12 h after the bacterial challenge, the severity of sickness was quite high, with the average grade point of 2.6 at six hours after the start of therapy. The severity of illness was seen increasing between 12-48 h from the start of the therapy with one death at 48 h. Later at 72 h the rest of the mice showed improvement and fully recovered by 96 h (Table I and Figure).

In group A5, where bacteriophage cocktail was 24 h after administered the bacterial challenge, though the severity of illness was less at six hours after the start of therapy, but it increased with time, and at 48 h three mice were found dead. However, the remaining mice fully recovered by 96 h (Table I and Figure).

The three different dosages i.e. 10⁷, 10⁶, 10⁵ pfu/ml were given to A5.1, A5.2 and A5.3 groups of mice respectively, and it was observed that after six hours of administration of phage cocktail, all the five mice of the three groups were sick uniformly having moderate illness, severe lethargy, ruffled fur and hunched back.

In group A5.1, where a dose of 100 μl having 10⁷ pfu/ml, bacteriophage cocktail was given, three of the five mice had increased severity and developed the additional sign of exudative accumulation around the eyes. Also mouse no. 5 was found dead at 48 h. In the rest of the mice, severe illness persisted at 72 h, however, they recovered by 96 h (Table I and Figure). In case of group A5.2 which received the bacteriophage cocktail dose of 100 μl having 10⁶ pfu/ml, severe illness persisted up to 72 h with the peak at 24 h. In this group also severe illness continued up to 72 h. However, all of them recovered by 96 h (Table I and Figure).

The group A5.3 received 10⁵ pfu/ml in 100 μl of the cocktail. All the mice had all the signs of severe infection at 24 h after the start of the therapy. The sickness persisted up to 72 h with complete recovery at 96 h (Table I and Figure).

The quantification of bacteriophage and AB bacteria carried out in the dead mouse after therapy showed only phage, not the bacteria. The peritoneal fluid, liver, spleen, heart blood were found to have only phages. None of the organs of the dead mice treated with phage cocktail yielded AB bacteria.

Discussion

This study was aimed to look at the efficacy of phage therapy as an alternative to antibiotics in case of septicaemia caused MDR AB in an immunocompromised mouse model. This work was carried out after considering the safety of the phage concentration employed based on a previous study form our laboratory which determined the optimum dose (volume and concentration) of the phage cocktail in the treatment of septicaemia caused by P. aeruginosa16. In that study, a dose of 100 µl of the bacteriophage cocktail in the concentration of 10¹² pfu/ml was employed empirically at different time points after bacterial challenge and mortality was observed only when the phage cocktail was given six hours before and after the bacterial challenge. The mortality, however, decreased to nil with the decreasing concentration of the phage cocktail to 10⁹ pfu/ml. By simulating the actual clinical scenario in ICU using the in vivo mouse model, 100 µl of the phage cocktail at the concentration of 10⁹ pfu/ml was administered simultaneously, and also six, 12 and 24 h after the bacterial challenge. This concentration of the phage caused mortality in 60 per cent of the mice receiving the phage cocktail 24 h after the bacterial challenge. It could be speculated that this concentration of phage cocktail reached the optimum level for sudden lysis of the bacteria similar to the zone phenomenon. It has been reported that when antibiotics inhibiting cell-wall synthesis (penicillins, cephalosporins and carbapenems) are administered as a bolus, sudden lysis occurs in cases of gram-negative bacterial septicaemia2425. Bacteriophages are also known to disrupt the cell wall of bacteria26. In this study, when the concentration of phage cocktail was decreased to 10⁶ and 10⁵ pfu/ml, no mortality could be observed in the groups of mice challenged with bacteria 24 h before.

A report by Schooley et al15 in 2017 demonstrated the potentials of personalized bacteriophage-based therapeutic cocktail in an elderly patient with disseminated MDR AB having the inexorable downhill clinical course for three months duration. This report discussed the different aspects of phage therapy. Amongst many, there is special mention regarding the possibility of a sudden massive release of endotoxin during therapy. Notably, in a clinical case, the treating physicians have good access to monitor the vitals and decide the dosage of phage cocktail accordingly.

There are many reports regarding the role of bacteriophages against AB mostly limited to in vitro evaluation27282930. However, there are studies on the evaluation of AB specific phage therapy in rodent models such as on wounds29 and in mouse pneumonia model30 administering the phages by the nasal route14. There is a report showing the efficacy of bacteriophages in BALB/c septicaemia model as well30. In this study, the phage therapy was initiated two hours after the bacterial challenge and showed the significantly reduced mortality30. The limitation of this study was, however, that they used a single experimental setup.

Based on the results of the present study alongwith our previous findings studies, it can be deduced that in septicaemia treatment with phages, the therapy should be started with a low dose of the phage cocktail. However, there are many questions which still need to be addressed including: (i) whether further lower doses need to be considered, if yes then how low; (ii) whether lower but multiple doses of the phage cocktail can be administered and; (iii) whether low dosage in the form of slow intravenous infusion are safe?

In conclusion, this study supports the systemic use of bacteriophages as therapeutic agents to combat MDR, XDR and PDR AB infections in patients. Further pre-clinical studies are, however, needed to determine the safety of the dosing schedule.