The role of doxycycline in the therapy of multidrug-resistant E. coli – an in vitro study

This study assessed the in vitro antibacterial activity of combinations of amikacin and doxycycline or tigecycline against multidrug-resistant E. coli isolates. Twenty-four different pulsotypes, including 10 extended-spectrum β-lactamase (ESBL)-, 10 carbapenem-resistant, 2 New Delhi Metallo-beta-lactamase (NDM)- and 2 Klebsiella pneumoniae carbapenemase (KPC)-E. coli isolates were collected. All 24 isolates were susceptible to amikacin and tigecycline. Only 30% of ESBL and 50% of carbapenem-resistant E. coli were susceptible to doxycycline. Both of the NDM-E. coli had a MIC of 64 μg/ml. The checkerboard method showed that for the ESBL- and carbapenem-resistant E. coli, the synergistic effects of amikacin/doxycycline were 80% and 90%, respectively. For the two KPC- and two NDM-E. coli, the FIC index of amikacin/doxycycline were 0.5/0.375 and 0.5/0.281, respectively. For the ESBL- and carbapenem-resistant E. coli isolates, the combinations of amikacin and doxycycline exhibited synergistic activities against 80%, and 80% and 10% vs 60%, and 80% and 10% of the isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. The synergistic effect seems to be similar for doxycycline and tigecycline based combinations with amikacin. In conclusion, the antibacterial activity of doxycycline can be enhanced by the addition of amikacin and is observed against most multidrug-resistant E. coli isolates.


The in vitro antibacterial activity of antibiotic combinations assessed by the broth method.
The in vitro determination of the inhibitory effect of combination regimens followed the time-killing method was defined by the CLSI 17 . In brief, bacterial suspensions were diluted to concentrations 5.0× 10 5 colony-forming units (CFU)/mL in fresh Mueller-Hinton broth. Drug concentrations of amikacin, tigecycline and doxycycline were adjusted to those of 1xMIC, 1/2xMIC, and 1/4xMIC. Each drug alone and the combination of amikacin and tigecycline or doxycycline were tested. Bacterial counts were measured at 24 h by enumerating the colonies in 10-fold serially diluted specimens of 100 μ L aliquots plated on the nutrient agar (Difco Laboratories, Sparks, MD) at 37 °C.
Definitions Synergy was defined as a ≥ 2-log 10 decrease in the CFU/ml between the combination and its most active constituent after 24 h and the number of surviving organisms in the presence of the combination must be ≥2 log 10 CFU/ml below the starting inoculum. Bacteriostatic activities were defined as the presence of ≥ 2 log 10 , but < 3 log 10 reductions, and bactericidal activities were defined as the presence of ≥ 3 log 10 reductions in the CFU/mL at 24 h, relative to the initial inoculum 17 . All experiments were performed in duplicate.
The in vitro antibacterial activity of antibiotic combinations assessed by the checkerboard method. To evaluate the effect of the combinations, the fractional inhibitory concentration (FIC) was calculated for each combination by the broth microdilution technique as recommended by the CLSI and as previously described 14,18,19 . Briefly, the 96-well microdilution plates were inoculated with each test organism to yield the appropriate density (10 5 CFU/ml) in 100 μ l of Mueller-Hinton broth (MHB) and incubated at 35 °C in ambient air for 24 h. One well with no antibiotic was used as a positive growth control on each plate. The plates were read for visual turbidity, and the results were recorded at 35 °C in ambient air using a magnifying mirror reader after 24 h of incubation, as turbidity in the wells indicated the growth of the microorganism. The MIC was determined as the well in the microtiter plate with the lowest drug concentration at which there was no visible growth. The following formulas were used to calculate the FIC index: FIC of drug A = MIC of drug A in combination/MIC of drug A alone, FIC of drug B = MIC of drug B in combination/MIC of drug B alone, and FIC index = FIC of drug A + FIC of drug B. Synergy was defined as a FIC index ≤ 0.5, indifference was defined as a FIC index > 0.5 but ≤ 4, and antagonism was defined as a FIC index > 4 20 . All drug combinations were performed repeatedly to validate the data.
The detection of β-Lactamase genes. Plasmid DNA was extracted as templates and polymerase chain reaction (PCR) was used to amplify CTX-M, TEM, IMI, IMP, VIM, KPC, OXA and NDM using specific primers as previously published [21][22][23] . For AmpC genes, the following primers were used: (a) CMY-2-forward (TTT TCA AGA ATG CGC CAG GC), CMY-2-reverse (CTG CTG CTG ACA GCC TCT TT); and (b) DHA-1-forward (CTG ATG AAA AAA TCG TTA TC) and DHA-1-reverse (ATT CCA GTG CAC TCA AAA TA). For SHV genes, the following primers were used: (a) SHV-forward (GAT CCA CTA TCG CCA GCA GG) and SHV-reverse (ACC ACA ATG CGC TCT GC TTT G); and (b) SHV-12-forward (ATG CGT TAT ATT CGC CTG TG) and SHV-12-reverse (TTAGCGTTGCCAGTGCTCG). Amplicons were purified with PCR clean-up kits (Roche Diagnostics, GmbH, Penzberg, Germany) and were sequenced on an ABI PRISM3730 sequencer analyzer (Applied Biosystems, Foster City, CA, USA). Pulsed-field gel electrophoresis. PFGE was performed as described previously 24 with a CHEF DR II apparatus (Bio-Rad Laboratories, Hercules, Calif.). In brief, the DNA in the plugs was digested with XbaI, and electrophoresis was performed in a 1% agarose gel (in 0.5x TBE [Tris-borate-EDTA] buffer). The electrophoretic conditions used were as follows: initial switch time, 2.0 s; final switch time, 35.0 s; run time, 21 h; gradient, 6 V/cm; angle, 120°; and temperature, 14 °C. The bacteriophage lambda ladder pulsed-field grade (PFG) and low-range PFG molecular weight markers were loaded onto all gels. The PFGE patterns were visually examined and interpreted according to the criteria of Tenover et al. 25 . The similarities of the PFGE profiles of each strain were compared using a Dice coefficient at 1.0% of tolerance and 0.8% of optimization. Figure 1 shows the PFGE profile of the enrolled 10 ESBL-, 10 carbapenem-resistant-, two KPC-2 producing and 2 NDM-E. coli isolates (one was NDM-1, and the other was NDM-5), and all of them had different PFGE profiles. Table 1 shows their MIC values and the susceptible rates of amikacin, doxycycline, and tigecycline. All of the 24 E. coli isolates were susceptible to amikacin and tigecycline. However, only 30% of ESBL E. coli and 50% of carbapenem-resistant E. coli were susceptible to doxycycline. For doxycycline, both of the NDM positive E. coli had MICs of 64 μ g/ml, and in contrast, both KPC-producing E. coli had MIC values ≤ 2 μ g/ml.

Results
The ESBL and carbapenemase genes detected among the clinical isolates are presented in Table 2. For ten ESBL E. coli isolates, genes encoding CTX-M were detected for all isolates. Additionally, genes encoding TEM and CMY were detected for three and two isolates, respectively. For ten carbapenem-resistant E. coli isolates, genes encoding CMY were detected for all isolates. However, genes encoding CTX-M and TEM were detected for four and two isolates, respectively. For two KPC-producing E. coli isolates, genes encoding CMT, TEM, and CTX-M were detected for one isolate. For two NDM positive E. coli isolates, both had the gene encoding CMT and TEM, and one had the KPC-2 gene.
The results of the checkerboard methods are shown in Table 3. For the ESBL E. coli, the FIC 50/90 of doxycycline and the tigecycline-based combination were 0.375/0.563 and 0.5/0.563, respectively. The synergistic effects of amikacin/doxycycline and amikacin/tigecycline were 80% and 60%, respectively. For carbapenem-resistant E. coli, the FIC 50/90 of doxycycline and the tigecycline-based combination were 0.375/0.5, and 0.5/0.563, respectively. The synergistic effects of amikacin/doxycycline and amikacin/tigecycline were 90% and 80%, respectively. For both, there was no antagonism among the two combinations. For the two KPC E. coli and the two NDM E. coli, the FIC index values of amikacin/doxycycline were 0.5/0.375 and 0.5/0.281, respectively, and the FIC index values of amikacin/tigecycline were 0.375/0.5 and 0.265/0.312, respectively.
The in vitro activities of the combination of amikacin and doxycycline at the drug concentrations of 1xMIC, 1/2xMIC and 1/4x MIC against each isolate are shown in Table 4. For ESBL E. coli, the reduction of CFU at 24 hours ranged from 2.99 to 4.2, 0.05-4.2, and 0.29-4.08 log 10 , at concentrations of 1x, 1/2x and 1/4xMIC, respectively. The combinations of amikacin and doxycycline exhibited bactericidal effects against 90%, 70%, and 10% of the tested isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. These combinations were synergistic against 80%, 80%, and 10% of the isolates at the concentrations of 1x, 1/2x and 1/4xMIC, respectively. For carbapenem-resistant E. coli isolates, the reduction of CFU at 24 hours ranged from 0. 28    and 0.73-2.00 log 10, at concentrations of 1x, 1/2x and 1/4xMIC, respectively. The combinations of amikacin and doxycycline exhibited bactericidal effects against 90%, 50%, and 10% of the tested isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. These combinations were synergistic against 60%, 80%, and 10% of the isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. For KPC E. coli, at the concentration of the 1x MIC combination, one of two strains had a synergistic effect, and the reduction of the CFU at 24 hours compared to the initial inoculum was 3.79 log 10 and was − 3.94 compared to most active antibiotic. At the concentration of 1/2x MIC, both strains had synergistic effects, and the reduction of the CFU at 24 hours compared to the initial inoculum was 3.79/2.10 log 10 and was 6.45/4.49 compared to the most active antibiotic. Both strains had synergistic effects.
The NDM strain combinations of amikacin and doxycycline were not performed because the MIC of doxycycline was too high.
The in vitro activities of combinations of amikacin and tigecycline at the drug concentrations of 1xMIC, 1/2xMIC and 1/4x MIC against each isolate are also shown in Table 4. For ESBL-E. coli, the reduction of CFU at 24 hours ranged from 2.18 to 3.72, 1.77-3.72 and 0.87-3.68 log 10, at concentrations of 1x, 1/2x and 1/4xMIC, respectively. The combinations of amikacin and tigecycline exhibited bactericidal effects against 90%, 90%, and 20% of the tested isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. These combinations were synergistic against 50%, 100%, and 20% of the isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. For carbapenem-resistant E. coli, the reduction of CFU at 24 hours ranged from 1.12 to 3.82, 2.00-3.82 and 0.30-1.56 log 10, at concentrations of 1x, 1/2x and 1/4xMIC, respectively. The combinations of amikacin and tigecycline exhibited bactericidal effects against 90%, 70%, and 0% of the tested isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. These combinations were synergistic against 30%, 100%, and 0% of the isolates at concentrations of 1x, 1/2x and 1/4xMIC, respectively. For KPC E. coli, at the combined concentrations of 1x MIC and 1/4 x MIC, both two strains have no synergistic effect. At the combined concentration of 1/2x MIC, both strains had synergistic effects and with a value of 2.56/6.15 log 10 compared to most active antibiotic. The reduction of CFU at 24 hours compared to the initial inoculum was 3.73/3.53 log 10 , exhibiting a bactericidal effect. One of the two NDM strains at the combined concentration of 1x MIC had a synergistic effect, and the reduction of CFU at 24 hours compared to the initial inoculum was 4.00 log 10 was − 2.38/0 log 10 compares to the most active antibiotic. At the combined concentration of 1/2x MIC, both strains had a synergistic effect, and the reduction of CFU at 24 hours compared to the initial inoculum was 4.00/3.68 log 10 and was 6.45/6.58 log 10 compared to the most active antibiotic. However, no synergistic effect was noted at the combined 1/4 x MIC.

Discussion
Antibiotic combination therapy has become the possible resolution for the treatment of severe multidrug resistant organism infections, and various antibiotic combination regimens for treating multidrug resistant E. coli have been recommended based on in vitro and in vivo studies. However, research investigating the in vitro antibacterial activity of the combinations of an aminoglycoside (amikacin) and tigecycline or doxycycline against multidrug-resistant E. coli isolates is scarce. This is the first study to assess this type of combined antibiotic regimen against multidrug-resistant E. coli, including ESBL-, carbapenem-resistant, NDM-and KPC-producing E. coli isolates. Based on this in vitro study, we have several significant findings. Most important, although tigecycline and amikacin displayed greater in vitro activities against multidrug-resistant E. coli than doxycycline, the synergistic effect seems to be similar between the combination of doxycycline and amikacin and the combination of tigecycline plus amikacin. As doxycycline is safe, inexpensive, and almost universally availability, further large in vitro and in vivo studies are warranted to clarify its role as a new adjunctive therapy to improve the outcomes of multidrug-resistant E. coli infections.
Although doxycycline is a cheap antimicrobial agent, it exhibits a broad spectrum of activity against different pathogens, including Gram-negative bacteria, and remains as a useful or even drug of choice in the treatment of many infectious diseases 26,27 . Even in this era of the increasing prevalence of multidrug-resistant organism infections, doxycycline is efficacious against multidrug-resistant A. baumannii 28 , Pseudomonas aeruginosa 29 , and Stenotrophomonas maltophilia 30 . In this first study investigating the in vitro activity of doxycycline against multidrug-resistant E. coli, we found that most clinical isolates, including seven (70%) ESBL-, five (50%) carbapenem-resistant and two (100%) NDM -E. coli, were not susceptible to doxycycline. However, even sub-inhibitory concentrations of an aminoglycoside combined with doxycycline can exhibit synergistic activities against more than 80% of tested isolates. For this combination, using 1/2xMIC of doxycycline (2 μ g/mL, which is achievable in serum) produces the best synergism 7,31 . Therefore, our findings indicate the potential role of doxycycline-containing combinations in the management of multidrug-resistant E. coli infections.
Tigecycline, the first glycylcycline, exhibits potent activity against a wide range of clinically significant gram-positive and gram-negative bacteria, including multidrug-resistant strains (e.g., oxacillin-resistant Staphylococcus aureus, vancomycin-resistant enterococci, and ESBL-producing Enterobacteriaceae), and  Table 3. Continued anaerobes (e.g., Bacteroides spp) 32 . Like several previous studies [33][34][35] , we found that the MIC values of tigecycline against all tested isolates remained low (≤ 1 μ g/mL), and all of the tested isolates were susceptible to tigecycline. However, in vitro activity does not equate to an in vivo response, and the current suggested dosage of tigecycline for adults only achieves low serum concentrations; therefore, tigecycline cannot be recommend for the treatment of bloodstream infections, even those caused by so called "tigecycline-susceptible" isolates. To overcome this barrier to the treatment of critical conditions and the emergent tigecycline-resistant strains, tigecycline-containing combinations have been proposed as possible solutions. In this study, we found that using 1/2xMIC of tigecycline (0.5 μ g/mL) in combination with sub-inhibitory concentrations of an aminoglycoside, synergism can be achieved for all of the 24 tested isolates. However, if we use 1/4xMICs of tigecycline (0.25 μ g/mL) in combination with 1/4xMICs of an aminoglycoside, synergism was found for only two of the tested isolates. A previous study showed that the serum attainable concentration of tigecycline was only 0.38 and 0.93 μ g/mL after a single dose injection of 50 mg and 100 mg tigecycline, respectively 36 . Therefore, if we formulate tigecycline-containing combination regimens based on the recommended dosages (100 mg loading, followed by 50 mg every 12 h), we can obtain the synergistic effect with tigecycline and amikacin despite low serum levels of tigecycline (< 1 μ g/mL).

The results of the checkerboard method of amikacin-based combinations with doxycycline and tigecycline against 10 extended-spectrum β-lactamase (ESBL)-, 10 carbapenem-resistant-(CRE), 2 New Delhi Metallo-beta-lactamase (NDM)-and 2 Klebsiella pneumoniae carbapenemase (KPC)-producing
In this study, we found an unusual association between NDM-1 and KPC-2 in one E. coli isolates, and it is the first detection of this combination in Taiwan. As previously reported 37 , this isolate should be multi-drug resistant against most antibiotics, excluding tigecycline. Previous studies only found this double carbapenemase-producer in K. pneumoniae, E. cloacae, Citrobacter freundii and Enterobacter hormaechei isolates from Brazil, Pakistan, China, and India [38][39][40][41][42] . However, we did not find the mutation of outer membrane porin (Omp) in KPC or NDM-producing isolates. As previous reports [43][44][45] , we found that the mutation of OmpA, OmpC, or OmpF was only presented in carbapenem-resistant strains. Overall, all of these findings indicate the worldwide emergence of double, or even multiple, carbapenemase-producing bacteria among Enterobacteriacae, including in Taiwan.
Finally, recent studies 46,47 showed that the different resistance mechanisms of multidrug-resistant organisms may influence the synergistic effects of combination therapy. For carbapenem-resistant K. pneumoniae, Laishram et al. 46 found that isolates producing NDM carbapenemase alone showed significantly more synergy than isolates producing OXA-48-like carbapenemase. Furthermore, Hong et al. 47 found that clinical isolates of KPC-producing K. pneumoniae with high porin expression were more responsive to a combination of colistin-doripenm-ertapenem than isolates with low expression (100% [8/8] vs 0% [0/4]; p = 0.002). In this study of limited clinical isolates, we did not assess whether the MDR E. coli with different resistant mechanisms had different responses to antibiotic combination therapy. However, further investigations are warranted to clarify this issue.
In conclusion, despite the lower susceptible rate of doxycycline, the antibacterial activity of such an ancient antimicrobial agent can be enhanced by the addition of amikacin. The synergistic effect of such combinations seems to be as effective as the tigecycline/amikacin combination against most multidrug-resistant E. coli isolates, and warrants further in vivo investigation to confirm their therapeutic efficacy.  Table 4. The log change (log 10 CFU/ml) from the starting inoculum and the most active single agent after 24 h of incubation with different concentrations of antibiotics combinations including 1x, 1/2x and 1/4x MICs of amikacin, doxycycline and tigecycline for 10 extended-spectrum β-lactamase (ESBL)-, 10 carbapenem-resistant-, 2 New Delhi Metallo-beta-lactamase (NDM)-and 2 Klebsiella pneumoniae carbapenemase (KPC)-producing E. coli isolates. -cidal refers to the bactericidal effect and -static refers to the bacteriostatic effect. ND refers to not done. a -cidal refers to bactericidal effect and -static refers to bacteriostatic effect. ND refers to not done.