Subscriber access provided by University of Newcastle, Australia
Communication
Protein-adsorbed Magnetic-Nanoparticle Mediated Assay for Rapid Detection of Bacterial Antibiotic Resistance Taku Cowger, Yaping Yang, David Rink, Trever Todd, Hongmin Chen, Ye Shen, Yajun Yan, and Jin Xie Bioconjugate Chem., Just Accepted Manuscript • DOI: 10.1021/acs.bioconjchem.7b00016 • Publication Date (Web): 13 Feb 2017 Downloaded from http://pubs.acs.org on February 15, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Bioconjugate Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Bioconjugate Chemistry
Protein-adsorbed Magnetic-Nanoparticle Mediated Assay for Rapid Detection of Bacterial Antibiotic Resistance
Taku A. Cowger,†,# Yaping Yang,‡,# David E. Rink,† Trever Todd,† Hongmin Chen,†,* Ye Shen,¶ Yajun Yan, ‡,* Jin Xie†,*
†
Department of Chemistry and Bio-Imaging Research Center, University of Georgia, Athens, Georgia, 30602, USA
‡
College of Engineering, University of Georgia, Athens, Georgia, 30602, USA
¶
Department of Epidemiology and Biostatistics, University of Georgia, Athens, Georgia, 30602, USA
Corresponding Author: * E-mail:
[email protected] (H.C.);
[email protected] (Y. Y.);
[email protected] (J.X.)
#
These authors contribute equally to this work.
1 ACS Paragon Plus Environment
Bioconjugate Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 22
Abstract: Antibiotic susceptibility tests have been used for years as a crucial diagnostic tool against antibiotic resistant bacteria. However, due to lack of biomarkers specific to resistant types, these approaches are often time consuming, inaccurate, and inflexible in drug selections. Here, we present a novel susceptibility test method named protein-adsorbed nanoparticle mediated matrix-assisted laser desorption/ionization mass spectrometry, or PANMS. Briefly, we adsorb five different proteins (β-casein, α-lactalbumin, human serum albumin, fibrinogen, and avidin) onto the surface of Fe3O4. Upon interaction with bacteria surface, proteins were displaced from the nanoparticle surface, the amounts of which were quantified by MALDIMS. We find that the protein displacement profile was different distinctive among different bacteria strains and, in particular, between wild-type and drug-resistant strains.
More
excitingly, we observe bacteria resistant to drugs of the same mechanisms share similar displacement profiles on a linear discriminant analysis (LDA) map. This suggests the possibility of using PANMS to identify the type of mechanism behind antibiotic resistance, which was confirmed in a blind test. Given that PANMS is free of drug incubation and the whole procedure takes less than 50 minutes, it holds great potential as a high throughput, low cost, and accurate drug susceptibility test in the clinic.
Keywords: Iron oxide nanoparticles, bacteria, protein adsorbed nanoparticles, detection of resistance
2 ACS Paragon Plus Environment
Page 3 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Bioconjugate Chemistry
Antibiotics were discovered over 70 years ago and its use has greatly reduced mortality caused by infectious diseases.1-4 However, the misuse of antibiotics has led to evolution in bacteria to acquire resistance mechanisms.2,5,6 Each year in the United States, at least 2 million people become infected with bacteria that are resistant to antibiotics, causing 23,000 deaths annually.1,4,7 Due to uncontrolled sale and regulation of antibiotics, developing countries are even more affected by the resistance issue.3,10,11 While tremendous efforts are devoted to finding alternative antibiotics,12,13 it is equally important to develop diagnostic tools that can identify bacterial resistance at early time points of therapy so that promote change to more appropriate treatments can be made. However, there is a lack of biomarkers that are specific for resistance phenotypes. Current drug susceptibility tests are mostly conducted in a trial-and-error manner by incubating bacteria with different antibiotics and accessing proliferation. These include traditional broth dilution tests,12 gradient methods,13 and disk diffusion tests,14 which often take 24 hrs or longer.12,15-18 Recently developed automated systems can reduce the diagnosis time to 3.5 to 16 hrs;19-21 however, their use has been limited by the high cost, moderate accuracy, and inflexible drug selections.
In the present study, we exploit bacterium surface phenotype changes as a biomarker for antibiotic resistance. It is well documented that the bacterium surface properties are changed when they mutate to become drug resistant. For instance, bacteria may undergo mutations to its surface proteins that are involved in membrane constitution or even intracellular proteins that are involved in the biosynthesis of the components such as lipids of cell membrane to change its membrane permeability to antibiotics. In some cases, the mutations may occur to the multidrug efflux pumps that are unusually embedded into the cell membrane to enable them to have new or enhanced activities towards specific antibiotics. Such surface phenotype changes involve subtle or substantial changes of multiple surface components that are either directly or indirectly involved in drug resistance. These contain fingerprint information that is 3 ACS Paragon Plus Environment
Bioconjugate Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
distinctive among mutates and may be utilized to identify types of drug resistance. There has been progress by the Rotello group on investigating mammalian cell surface phenotype changes when they are treated with therapeutics.22 To the best knowledge, however, there has been no attempt of developing an antibiotic susceptibility test based on surface phenotype attributes of bacteria.
Scheme 1. Schematic illustration of protein-adsorbed magnetic-nanoparticle-mediated MALDI mass spectroscopy (PANMS).
Specifically, we have developed a novel susceptibility test called protein-adsorbed nanoparticle facilitated matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS) method, or PANMS. The procedure of PANMS is illustrated in Scheme 1. Briefly, we adsorb five proteins of different isoelectric points (pI) and molecule weights onto the surface of Fe3O4 nanoparticles; the resulting five nano-conjugates are used as sensors to be incubated with bacteria analytes for 10 minutes. Due to interaction with bacteria, proteins 4 ACS Paragon Plus Environment
Page 4 of 22
Page 5 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Bioconjugate Chemistry
are displaced from the nanoparticle surface, with the amplitude of the displacement varied among bacterium strains (Scheme 1). This is followed by magnetic separation and highthroughput MALDI-TOF MS analysis, which quantifies the amounts of the proteins displaced. Magnetic separation technologies have played a critical role in a variety of biomedical applications,23 which offers major advantages in terms of throughput and cost.24,25 For each bacterial strain, we generated a 5-channel protein displacement profile based on the MALDI quantification. It is found that the displacement profile is distinctive among different bacteria strains and, in particular, between wild-type and drug-resistant strains. More excitingly, we find that bacteria resistant to drugs of the same mechanisms share similar displacement profiles on a linear discriminant analysis (LDA) map. Notably, PAMNS is drug free and the whole procedure takes less than 50 minutes. This sets a big contrast from conventional approaches, which often need repeated and lengthy proliferation assays. These advantages suggest the great potential of PANMS as a new means to identify resistant strains of bacteria.
We first synthesized 15 nm iron oxide nanoparticles (Fe3O4 NPs) through a thermal decomposition method (Scheme S1).26 The particle size was confirmed by both transmission electron microscopy (TEM) and dynamic light scattering (DLS). The as-synthesized NPs were coated with a layer of oleic acid and were hydrophobic. The NPs were then incubated with glucosamine (GA, 50 mg/ml) in an aqueous solution with heating (90 °C) for ligand exchange. A similar method has been used by us and others for surface modification.26,27
Successful tethering of glucosamine to Fe3O4 NP surface was confirmed by zeta potential analysis and DLS (Figure S1). The ligand exchange did not affect the morphology of the nanoparticles (Figure S1a), and the overall nanoparticle size remained almost unchanged (Figure S1b). The resulting, glucosamine coated Fe3O4 NPs, or GA-Fe3O4 NPs, were rendered positively charged and were able to be dispersed in water (Figures S1c&S2). 5 ACS Paragon Plus Environment
Bioconjugate Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
However, due to large surface energy and magnetic interaction, these nanoparticles tend to agglomerate in aqueous solutions (data not shown).
Such a semi-stable status makes GA-Fe3O4 NPs susceptible to protein adsorption, which decreases surface energy. This attribute was utilized by us and others to modify magnetic nanoparticles.28,29 In the present study, we adsorbed five proteins, which are β-casein (CAS), α-lactalbumin (LAC), human serum albumin (HSA), fibrinogen (FIB), and avidin (AVI), to the surface of GA-Fe3O4. The five proteins cover a wide range of pI (3.6, 4.3, 4.2, 6.3, and 9.8 for CAS, LAC, HSA, FIB and AVI, respectively, Fig. S3a) and have different molecular weights (4-5 kDa, 14 kDa, 66 kDa, 350 kDa, and 66 kDa, respectively). According to Coomassie Blue and ICP assays, it was estimated that the protein-to-Fe3O4 nanoparticle ratios were 55.2 ± 0.6, 34.3 ± 1.2, 34.8 ± 0.6, 20.0 ± 2.2, and 37.4 ± 1.3 for CAS, LAC, HSA, FIB and AVI, respectively. For all of the five proteins, the particles became very stable in aqueous solutions after adsorption (Figure S2). The hydrodynamic size was increased to 25.7 ± 3.0, 25.36 ± 3.9, 34.30 ± 4.2, 35.38 ± 4.1, 27.78 ± 3.1 respectively, for CAS-, LAC-, HAS-, FIBand AVI-coated Fe3O4 NPs (Figure S1b).
We incubated the five protein-adsorbed Fe3O4 NPs (50 µg Fe3O4/mL) with three representative bacteria strains, which are E. coli BW25113/F+, E. coli BW25113/F-, and B. subtilis (109 cells/mL, O.D.600 = 1). E. coli BW25113 is a derivative of Escherichia coli, which is a common gram-negative bacterium; F+ and F- indicate whether there is presence of the F’ sex factor in the bacteria. B. subtilis is a gram-positive strain that is also widely studied.30,31 After 10 minutes of incubation, a magnetic bar was applied to the bottom of the vessel to remove nanoparticles; the suspension, which contained the displaced proteins, was collected and filtered through a 0.22 µm filter to remove bacteria.
6 ACS Paragon Plus Environment
Page 6 of 22
Page 7 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Bioconjugate Chemistry
For fast and high-throughput analysis of protein contents, we used MALDI-TOF MS based quantification.32-34 For all of the five proteins, we observed nearly perfect linear response between 0.1 to 10 µg/ml (Figure S4). By comparing to the pre-determined calibration curves, the amounts of proteins in the suspensions were quantified; the proportions of displaced proteins relative to their initial amounts were calculated. It was found that E. coli BW25113/F+ and E. coli BW25113/F- showed similar displacement among proteins, with βcasein displaced the most (24.3 ± 4.5 %, 21.55 ± 6.8 %, respectively) and fibrinogen the least (1.0 ± 0.3 %, 2.3 ± 2.0 %, respectively) (Figure 1a). As a comparison, B. subtilis showed a different displacement pattern, with the highest displaced protein being CAS (32.6 ± 6.8 %) and the lowest being FIB (21.1 ± 4.5 %) (Figure 1a). These results suggest that the subtle difference in surface properties can be discerned by PANMS.
Figure 1. a) Protein displacement from untreated E. coli BW25113/F+, E. coli BW25113/F-, and B. subtilis strains upon nanoparticle addition. b) Protein displacement from Ampicilintreated strains and c) Protein displacement from Kanamycin-treated strains. Data shows significant changes in protein displacement suggesting phenotypic changes occurring in bacteria.
7 ACS Paragon Plus Environment
Bioconjugate Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Drug treatment may lead to altered surface properties and hence different readings by PANMS. This was investigated by treating E. coli BW25113/F+, E. coli BW25113/F-, and B. subtilis with low levels of ampicillin and kanamycin (50 µg/ml, 12 hrs), and then analyzing the bacteria by PANMS. For all three strains, we observed altered protein displacement profiles after drug treatments. Taking CAS for example, on E. coli BW25113/F+, the displacement ratio was 24.3 ± 4.5% for the wide-type, and was slightly increased to 30.0 ± 3.3% and 34.6 ± 2.1% for the ampicillin and kanamycin treated strains, respectively; similar results were observed with E. coli BW25113/F-. For B. subtilis, on the other hand, the displacement for CAS was 32.6 ± 6.4% for the wide-type, but increased further to 53.4 ± 2.2% and 43.2 ± 4.1% after ampicillin and kanamycin treatment, respectively. Notably, while ampicillin is a cell wall production inhibitor, kanamycin is an aminoglycoside antibiotic and mainly targets intracellular proteins.35 It is reasoned that kanamycin treatment is likely to induce the changes in cell membrane permeability to prevent the intracellular proteins targets from exposure to the antibiotic, which is reflected as surface phenotype changes.
We then assessed whether PANMS can discern drug resistance strains from those that are not. To this end, we established mutants of E. coli BW25113/F+, E. coli BW25113/F-, and B. subtilis that were resistant to ampicillin, amoxicillin, tetracycline hydrochloride, terramycin, kanamycin and neomycin, respectively. The mutations were induced through a UV-irradiation method that was reported previously.36 Resistance against these antibiotics is common in the clinic, and the mechanisms generally fall into four categories. These are: 1) outer membrane impermeability (commonly seen in strains resistant to ampicillin and amoxicillin),37 2) 30S subunit target site alteration (tetracycline hydrochloride and terramycin),38 3) aminoglycosidemodifying enzyme production (kanamycin and neomycin),39 and 4) DNA target alteration (nalidixic acid and moxifloxacin).40,41 Notably, the acquired resistance was drug-specific. For instance, a penicillin derivative (ampicillin, amoxicillin) resistant E. coli BW25113/F+ strain 8 ACS Paragon Plus Environment
Page 8 of 22
Page 9 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Bioconjugate Chemistry
was not resistant to kanamycin or antibiotics associated with different mechanisms of action, which was confirmed by broth dilution tests (Table S1).
For analysis, all eight resistant strains, along with the three parental strains, were subject to PANMS. For each strain, we obtained the five protein displacement percentages based on the MALDI quantification (Figure 1a & Table S1). For ease of comparison, for each resistant strain, we computed the ratios between the resistant and parental strains (Table S1), and generated a 5-channel color intensity map (Figure 2). It is clear that each strain has a unique set of protein displacement profile. The profile was significantly changed when bacteria mutated to acquire resistance, but was different among strains with different resistance mechanisms (Table S1).
Figure 2. Color intensity map showing the changes in protein displacement observed with resistant strains of each bacteria along five protein channels. Each row depicts similar results, which is consistent with its resistance mutation. The PANMS is able to distinguish between resistance mutations. 9 ACS Paragon Plus Environment
Bioconjugate Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
We then examined whether strains with similar resistance mechanisms share similar protein displacement profiles. For the purpose, we used linear discriminant analysis (LDA) to process multiple sets of PANMS data with the eight mutates (n=24). LDA is a common classification method used to characterize or separate two or more classes of objects, contingent on a classifier built from the linear combination of multiple variables. Recently it has been implemented in biological fields as a detector for morphology and the identification of proteins, cells, and bacteria.42,43 It can be easily visualized that our results from the eight mutated strains were clustered into four areas on a LDA map (Figure 3). Notably, data dots in the same cluster were from either the same bacteria strain or one that shares a similar resistant mechanism (for instance, data from ampicillin and amoxicillin resistant strains were closely clustered). The LDA results confirm the high precision and reproducibility of the PANMS test, and the capacity of the assay on differentiating strains with different resistant mechanisms. More excitingly, it suggests the feasibility of identifying the resistant mechanism of an unknown bacterium with PANMS without actually incubating with a panel of antibiotics. Leave-one-out cross validation (LOOCR) was performed to check the reliability of the LDA approach in classification of resistant strains. The process was repeated with each case left out in turn. Final results from the validation (Table S1) clearly demonstrated that the discriminant function successfully classified all cases with no error.
10 ACS Paragon Plus Environment
Page 10 of 22
Page 11 of 22
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Bioconjugate Chemistry
Figure 3. Linear discriminant analysis of PANMS categorizes each resistance mutation into a specific region and allows for more qualitative results using 3 different bacteria strains; (a) E. coli BW25113/F+, (b) E. coli BW25113/F-, and (c) B. subtilis.
Table 1. Blind test results based on phenotypic responses of two antibiotics (PRMC, SMC) known to induce a resistant mutation known from PANMS dataset using E. coli BW25113/F+, and two other antibiotics (LMC, VMC) inducing resistant mutation not yet identified by PANMS.
Mechanism of action
P value
PRMC
2
0.30
SMC
2
0.09
LMC VMC
N/A N/A