| BackgroundAntibiotics resistance promoted by the misuse of antibiotics in both animal husbandry and public health has become a global healthcare issue.In recent years,multidrug-resistant Gram-negative bacteria(MDRGNB)have emerged developing resistance to all aminoglycosides,fluoroquinolones andβ-lactams.In such a case,the colistin appears to be the last treatment option.Colistin(polymyxin E)was discovered in the 1940s and is active against many gram-negative bacillusin.It was once discarded in the early 1980s due to serious neurotoxicity and nephrotoxicity,which seems to be the reason why colistin resistance in gram-negative bacteria is relatively rare now.It was assumed that colistin resistance was chromosomally mediated and transmitted vertically.However,a plasmid-mediated colistin resistance gene in Escherichia coli,mcr-1,was firstly reported in 2015,indicating that it was transmissible between bacteria of the same species or of the different species.Thus,acquisitions of mcr-1 by MDRGNB have the potential to result in untreatable infections,which would be a serious issue for human health.Therefore,it is of major importance to improve the detection of colistin-resistant isolates to prevent any spread of resistance.Furthermore,the patients that suffer from MDRGNB infections such as Acinetobacter baumannii and Pseudomonas aeruginosa are always in critical life-threatening situations.A fast and accuracy detection of colistin resistance would undoubtedly be helpful to optimize the treatment and therefore improve the patient outcomes.Broth microdilution method(BMD)is the standard reference one recommended by the European Committee on Antimicrobial Susceptibility Testing(EUCAST)and the Clinical and Laboratory Standards Institute(CLSI)for determining susceptibility to colistin.However,it is quite laborious and requires 16-24 hours to perform after bacteria isolation and identification.The disk diffusion test and E-test also need16-24 hours and show high rates of false susceptibility due to the slow diffusion of colistin molecules in agar.Among all the automatic susceptibility testing systems,Vitek 2 could provide antibiograms in a shorter time,4-10 hours,but it displays unreliable results for detecting colistin resistance with a low sensitivity of 42%.Raman spectroscopy has already proved to be a powerful tool for fast,robust and reliable characterization and detection of bacteria.The Raman spectrum originated from the molecular vibrations can be considered as a fingerprint of the biochemical composition of microbial cell.This vibrational spectrum could provide valuable information on the interaction between antibiotics and bacterial cells at the molecular level.It leads to its potential application in performing the antibiotic susceptibility testing.Base on the above background,in this study we propose a new method of screening colistin resistance in E.coli,A.baumannii and P.aeruginosa by Raman spectroscopy.Objectives(1)To analyze the Raman spectral variations between that of bacteria exposed to and unexposed to colistin,which can provide a theoretical basis for the new colistin susceptibility test.(2)To explore the spectral differences between Raman spectra recorded from different areas or different layers of the sample,which can further indicate the repeatability for the detection of the colistin resistance of the new colistin susceptibility test.(3)To determine the incubation concentrations of colistin for E.coli,A.baumannii and P.aeruginosa respectively,which will be utilized for the screening of colistin-resistant isolates.(4)To build a novel,fast,convenient screening method of colistin-resistant isolates for the three species of bacteria.Methods(1)Principal component analysis(PCA)was performed to observe and detect the spectral variations induced by the colistin treatment;Bacterial suspensions of 0 McF,0.2McF,0.4 McF and 0.8 McF were prepared in 0.3 McF of the CAMHB and the spectra of these samples were obtained to indicate the source of the Raman signal;Concentrations of proteins and nucleic acids and the number of bacterial cells of the bacterial suspension were measured to indicate the substance differences between the bacteria with and without colistin;The same amount of normal growth bacterial cells and cells treated with colistin were prepared in the CAMHB and their Raman spectra were obtained to explore the influence of the lost of cellular content on the Raman spectrum.(2)The advantage of the confocal Raman to sampling with high axial resolution were utilized to record Raman spectra from different layers of the sample(10μm,20μm,30μm and 40μm away from the bottom of the sample)and from different areas at the sample surface(100μm,200μm,300μm and 400μm away from the center of the sample surface);The common logarithms of the intensity ratio of the 746 cm-1 band to the 481 cm-1 band were calculated to describe the homogeneity of the mixture sample and therefore further indicate the high detection repeatability of the new method.(3)Isolates of the three species with colistin MICs of 0.5μg/ml,1μg/ml,2μg/ml and4μg/ml were obtained and incubated with a serial concentrations of colistin to find out the minimal concentration of colistin needed to separate the Raman spectra of bacteria with and without colistin into two clusters by HCA;The incubation concentrations of colistin were determined for the three species according to the cut-off value of CLSI and EUCAST for colistin resistant isolates.(4)Totally 123 clinical isolates including 42 isolates of E.coli(30 susceptible and 12resistant isolates),41 isolates of P.aeruginosa(30 susceptible and 11 resistant isolates),and40 isolates of A.baumannii(30 susceptible and 10 resistant isolates)were tested to evaluate the new method.Results(1)Results showed that there were obvious differences between the Raman spectra of the bacteria exposed to colistin and those unexposed to.The PC1 loading plot indicated that the protein and nucleic acid content in the bacteria exposed to colistin was significantly lower than that of unexposed to while the starch substance content was obviously higher;The Raman specta of samples with different proportion of CAMHB and bacterial cell changed significantly;The intensity of the 746 cm-1 band(assigned to nucleic acid)increased as the proportion of bacterial cell rose;In addition,the intensity of the 481 cm-1band(assigned to starch substance)increased as the proportion of the CAMHB rose;The concentrations of protein and nucleic acid in the bacterial suspension treated with colistin were quite higher than that of without colistin while the number of bacterial cells in the bacterial suspension treated with colistin was significantly less than that of without colistin;Raman spectra of the same amount of bacteria with and without drug displayed different pattern,which indicated that proteins and nucleic acids loss from bacterial cells would induced relevant and detectable modifications of the spectral features.(2)The mean of log R values calculated from Raman spectra recorded from different surface areas of the bacterial sample was 0.200±0.0419 and that from different layers of the bacterial sample was 0.187±0.037;The small fluctuation of this log R value demonstrated a slight spectral variation between the Raman spectra while sampling randomly.(3)For the three species,the minimal concentration of the colistin needed to separate the treated and untreated bacteria rose as the colistin MIC increased;This minimal colistin concentration was remarkably higher for A.baumannii than for E.coli and P.aeruginosa with the same colistin susceptibility;The incubation concentrations of colistin in our method were determined to be 1μg/ml for E.coli,1μg/ml for P.aeruginosa and 3μg/ml for A.baumannii.(4)Method evaluation with clinical isolates:The total detection sensitivity was90.9%and specificity was 91.1%;The detection sensitivity and specificity for E.coli was 91.7%and 93.3%respectively;90.9%and 86.7%for A.baumannii and P.aeruginosa,90.0%and 93.3%.Conclusions(1)The studied samples in our new colistin susceptibility test were mixtures of the deposit of the CAMHB(starch substance)and bacterial cells;The proportion of the bacterial cells turned down in the samples exposed to the colistin due to losing cellular substance(proteins and nucleic acids)and a decrease in the number of cells;As a result,the intensity of characteristic bands of bacterial cells decreased.Therefore,the Raman spectral variations between the bacteria with and without drug could be recognized by HCA.(2)The log R value of spectra recorded from different layers or from different surface areas of the mixture sample changed slightly,which demonstrated that the mixture was homogeneous and therefore our Raman measurement provided a new spectroscopical method with high repeatability for the detection of the colistin resistance and a random sampling at different points of the bacterial cells mixture was enough to get a relevant screening of the colistin resistance in a bacterial strain.(3)For the three species with the same colistin susceptibility,the minimal concentration of colistin needed to separate the treated and untreated bacteria was remarkably higher for A.baumannii than for E.coli and P.aeruginosa,which might be due to the larger amount of cell surface negative charges for A.baumannii than for E.coli and P.aeruginosa.Any other factors such as the shape or the volume of the bacteria might be relevant,which needed further study.(4)The results of the method evaluation with 123 clinical isolates showed that it performed a most accurate colistin resistance detection for E.coli while for A.baumannii the detection accuracy was relatively lower.However the total detection sensitivity of90.9%and specificity of 91.1%and the rapid detection in 1.5 h demonstrated that our new method could be effective for the screening of the colistin resistance in Gram-negative bacteria.This suggested that it hold great potential to be an effective method for preventing the spread of colistin resistance. |