Analytical Letters

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/lanl20

Characterization of Tigecycline-Sensitive and

Tigecycline-Resistant Escherichia coli by Surface-
Enhanced Raman Spectroscopy (SERS) and
Chemometrics

Saba Bashir, Saqib Ali, Haq Nawaz, Muhammad Irfan Majeed, Mashkoor
Mohsin, Ali Nawaz, Nosheen Rashid, Fatima Tahir, Anwar ul Haq, Mudassar
Saleem, Muhammad Zamman Nawaz & Kashif Shahzad

To cite this article: Saba Bashir, Saqib Ali, Haq Nawaz, Muhammad Irfan Majeed, Mashkoor
Mohsin, Ali Nawaz, Nosheen Rashid, Fatima Tahir, Anwar ul Haq, Mudassar Saleem, Muhammad
Zamman Nawaz & Kashif Shahzad (2022): Characterization of Tigecycline-Sensitive and
Tigecycline-Resistant Escherichia�coli by Surface-Enhanced Raman Spectroscopy (SERS) and
Chemometrics, Analytical Letters, DOI: 10.1080/00032719.2022.2030349

To link to this article: https://doi.org/10.1080/00032719.2022.2030349

View supplementary material Published online: 04 Feb 2022.

Submit your article to this journal Article views: 72

View related articles View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=lanl20
ANALYTICAL LETTERS
https://doi.org/10.1080/00032719.2022.2030349

RAMAN

Characterization of Tigecycline-Sensitive and Tigecycline-

Resistant Escherichia coli by Surface-Enhanced Raman
Spectroscopy (SERS) and Chemometrics
Saba Bashira†, Saqib Alia†, Haq Nawaza, Muhammad Irfan Majeeda,
Mashkoor Mohsinb , Ali Nawazb, Nosheen Rashidc, Fatima Tahira, Anwar ul Haqa,
Mudassar Saleema, Muhammad Zamman Nawaza, and Kashif Shahzada
a
Department of Chemistry, University of Agriculture Faisalabad, Faisalabad, Pakistan; bInstitute of
Microbiology, University of Agriculture Faisalabad, Faisalabad, Pakistan; cDepartment of Chemistry,
University of Education, Faisalabad Campus, Faisalabad, Pakistan

ABSTRACT ARTICLE HISTORY

Tigecycline has been considered the last treatment option for vari- Received 18 October 2021
ous difficult-to-treat diseases caused by multidrug-resistant patho- Accepted 13 January 2022
gens. The recent emergence of resistance in bacteria has become a
KEYWORDS
severe worldwide public health concern. In this study, surface-
Tigecycline resistance;
enhanced Raman spectroscopy (SERS) combined with chemometrics surface-enhanced Raman
was used for the rapid discrimination of tigecycline-sensitive E. coli spectroscopy (SERS);
(TSEC) and tigecycline-resistant E. coli (TREC) strains. For this purpose, principal component
the SERS spectra of TSEC and TREC strains were acquired with and analysis (PCA); hierarchal
without tigecycline exposure and their mean spectra were compared cluster analysis (HCA);
to separate the spectral features associated with the development of partial least squares –
resistance. To analyze the SERS spectra, principal component analysis discriminant analysis
(PCA) and hierarchal cluster analysis (HCA) were employed for the (PLS-DA)
clustering of TSEC and TREC strains with and without tigecycline
exposure based on their biochemical differences. Additionally, partial
least squares – discriminant analysis (PLS-DA) successfully distin-
guished exposed and unexposed strains with 99% sensitivity, 99.8%
specificity, and 100% accuracy.

Introduction
Escherichia coli is one of the primary pathogens causing urinary tract, skin, and skin
structure and intra-abdominal infections. Limited antibiotic treatment options against
carbapenemase and extended-spectrum b-lactamase-producing E. coli and the discovery
of plasmid-mediated colistin resistance gene have left the tigecycline as the final treat-
ment option because of its wide antibacterial activity (Hoban et al. 2005; Nordmann,
Naas, and Poirel 2011; Elnahriry et al. 2016; Moet et al. 2007). During phase 3 clinical
trials, different strains of E. coli were reported with decreased susceptibility to

CONTACT Haq Nawaz haqchemist@yahoo.com Department of Chemistry, University of Agriculture Faisalabad,

Faisalabad 38000, Pakistan; Muhammad Irfan Majeed irfan.majeed@uaf.edu.pk Department of Chemistry,
University of Agriculture Faisalabad, Faisalabad 38000, Pakistan; Mashkoor Mohsisn mashkoormohsin@uaf.edu.pk
Institute of Microbiology, University of Agriculture Faisalabad, Faisalabad 38040, Pakistan.
†
These authors have equal contributions.
Supplemental data for this article is available online at https://doi.org/10.1080/00032719.2022.2030349.
ß 2022 Taylor & Francis Group, LLC
2 S. BASHIR ET AL.

tigecycline due to overexpression of AcrAB efflux pump (Keeney et al. 2008). In 2019,
the presence of transferable plasmid-mediated genes tet(X3/X4) was reported in tigecyc-
line-resistant isolates of E. coli and Acinetobacter baumannii from humans and animals
(He et al. 2019). Among pathogenic bacteria, a rapid prevalence of tet(X) genes is a
leading public health issue that requires reliable and sensitive detection to prevent those
infections. SERS is promising for the rapid discrimination of pathogens at the strain
level as it requires no sample preparation, provides fast results, and is highly sensitive
(Zeiri et al. 2004). Previously, SERS has been employed for the identification of carbape-
nem, colistin, quinolone, ampicillin, and polymyxin B resistant E. coli (Lin et al. 2019;
Kim et al. 2019; Zhou et al. 2015; Bashir et al. 2021).
Recently, the identification of TSEC and TREC pellets using SERS has been reported by
us (Bashir et al. 2021). In the current study, instead of using pellets, the supernatants of E.
coli strains were used and exposed to tigecycline. The goal is to differentiate the TSEC and
TREC strains using SERS in conjunction with chemometric techniques based on the pres-
ence and absence of Tet(X)/flavin-dependent monooxygenase that is released into superna-
tants during centrifugation in TREC strains. Tigecycline exposure selectively hydroxylated it
into 11a-hydroxytigecycline (Moore, Hughes, and Wright 2005). For this purpose, four E.
coli strains were used and their SERS spectra were recorded before and after tigecycline
exposure to identify the changes associated with their biochemical composition.
Multivariate data analyses including PCA, HCA, and PLS-DA were employed to classify/
discriminate the spectra of the exposed and unexposed TSEC and TREC strains. According
to the best of our knowledge, there is no previous report published which uses SERS for
the antibiotic susceptibility testing of the TSEC and TREC strains exposed to tigecycline.

Materials and methods

Synthesis of silver nanoparticles (AgNPs)
The AgNPs, a SERS substrate, were synthesized by using a precursor, capping agent, and a
reducing agent by chemical reduction. The precursor was silver nitrate (AgNO3) and the
reducing agent was trisodium citrate (Na3C6H5O7) that also acted as a capping agent. To
prepare 1 mM AgNO3, 33.72 mg of AgNO3 were added to a beaker containing 200 ml of
deionized water followed by the addition of 8 ml of 1% Na3C6H5O7 that was heated to boil-
ing with continuous magnetic stirring for one hour at 600 rpm. The solution was cooled to
room temperature and gray Ag NPs were produced (Kashif et al. 2020).

Bacterial sample preparation

Tigecycline-unexposed E. coli strains
Two TSEC strains (PKNR 5090 and PKNR 5073) and two TREC strains (PK 2248 and
PK 4114) were obtained from the Institute of Microbiology, University of Agriculture
Faisalabad, Pakistan. All four strains were inoculated in 5 ml of Brain Heart Infusion
(BHI) broth contained in culture dishes and incubated for 3 hours at 37  C followed by
centrifugation at 10,000 rpm for 15 min.
 ANALYTICAL LETTERS 3

Tigecycline-exposed E. coli strains

A volume of 5 mL of tigecycline solution (10 mg/mL) was mixed with 495 mL distilled
water and inoculated with 10 mL of each of the TSEC (PKNR 5090/5073) and TREC
(PK 2248/4114) strain’s stock solution separately in Eppendorf tubes. Each tube was
vortexed for 1 min and placed in an incubator in the dark for 3 h at 37  C followed by
centrifugation at 10,000 rpm for 10 min. The supernatant was used for subsequent SERS
analysis of tigecycline-exposed E. coli strains.

SERS measurements
Fifty microliters of AgNPs were mixed with an equal volume of each supernatant and incu-
bated at room temperature as illustrated in Figure 1. Seven SERS spectra were acquired for
each tigecycline exposed and unexposed E. coli strain using the conditions in Table S1.

Data preprocessing and analysis

The raw spectra were imported into Matlab 7.8 (MathWorks) and the wavenumbers
were truncated from 500 to 1800 cm 1. The spectrum of AgNPs on an aluminum slide
was subtracted from the sample spectra. The preprocessing including smoothing, base-
line correction, and normalization by Savitzky-Golay filtering (13-point window and
third order), rubberband algorithm, and vector normalization. The data were further
analyzed by PCA, HCA, and PLS-DA in Matlab and MetaboAnalyst 5.0 (https://www.
metaboanalyst.ca) to effectively detect hidden correlation among variables.

Results
Mean SERS spectra
The SERS spectra of unexposed TSEC and TREC strains were compared to identify the
features related to tigecycline resistance and strain-specific features. The peak assign-
ments of all E. coli strains and the pure tigecycline are provided in Table S2. The com-
parison of mean spectra of all unexposed strains is illustrated in Figure 2 in which each
TSEC strain is plotted against each TREC strain to observe the features related to their
biochemical composition. The spectral changes that remained consistent in all compari-
sons at 657, 1010, 1095, 1331, 1557, and 1587 cm 1 are considered to be associated with
the tigecycline resistance development in E. coli. The band at 1557 cm 1 is solely pre-
sent in unexposed TREC strains, and the features at 1095 and 1587 cm 1 are only pre-
sent in unexposed TSEC strains. The remaining features discriminate them based on
their peak intensity ratios. The intensities of the bands at 1010 and 1331 cm 1 are rela-
tively high and the intensity at 657 cm 1 is lower in unexposed TREC strains. Several
features were strain-specific, as the band at 627 cm 1 is only present in PKNR 5073.
This strain has a relatively high intensity 733, 962, and 1455 cm 1 bands and low inten-
sity at 1246 cm 1 compared to the others. The features at 1010, 1127, 1213 cm 1, and
1029 cm 1 are absent in PKNR 5073, unlike the others. The bands at 1678 and
1730 cm 1 only appeared in PK 2248. The intensity at 733 cm 1 is relatively low and at
1010 cm 1 is relatively high in PK 2248. The bands at 1172 and 1373 cm 1 were present
4 S. BASHIR ET AL.

Figure 1. Schematic of the synthesis of silver nanoparticles and SERS spectra of E. coli strains exposed
and unexposed to tigecycline.

in all E. coli strains. The comparison of present and absent bands in all unexposed E.
coli strains is provided in Table S3.
Figure 3 illustrates the comparison of the mean spectra of unexposed and exposed E.
coli strains. The SERS spectrum of pure tigecycline with characteristic features at 618,
743, 1029, 1112, 1246, 1286, 1365, 1418, 1455, and 1563 cm 1 is plotted above all pairs
to identify which features of the tigecycline appeared in the exposed strains. The mean
spectra of unexposed versus exposed PKNR 5090 are plotted in Figure 3a. The features
of tigecycline that appeared in exposed PKNR 5090 are at 618, 1112, 1246, 1286, and
1563 cm 1. The characteristic features of unexposed PKNR 5090 at 657, 733, 962, 1095,
1127, 1172, 1213, 1331, 1455, and 1587 cm 1 disappeared with exposure to tigecycline
except for at 1010, 1246, and 1373 cm 1. The features at 695, 1313, 1438, and
1623 cm 1 only appeared in the exposed PKNR 5090.
Figure 3b shows exposed and unexposed PKNR 5073 spectra in which the features of
unexposed PKNR 5073 at 627, 657, 733, 962, 1029, 1095, 1172, 1331, 1455, and 1587 cm 1
disappeared with exposure to tigecycline. The bands at 1246 and 1373 cm 1 are present in
both. The peaks at 686, 1010, 1313, 1353, 1430, and 1623 cm 1 are only associated with
exposed PKNR 5073. The features of tigecycline in exposed PKNR 5073 are similar to those
in exposed PKNR 5090. The relatively high intensity of the band at 1246 cm 1 in exposed
PKNR 5090 andv5073 is due to the presence of this band in their unexposed forms and
pure tigecycline. The comparisons of unexposed versus exposed PK 4114 and unexposed
versus exposed PK 2248 spectra are illustrated in Figure 3c and d. With the exposure to
tigecycline, all bands of unexposed PK 4114 and 2248 disappeared. The features at 624,
695, 924, 1150, 1313, and 1438 cm 1 are observed in exposed PK 4114. The peaks at 624,
 ANALYTICAL LETTERS 5

Figure 2. Comparison of mean SERS spectra of each unexposed tigecycline-sensitive with each unex-
posed tigecycline-resistant E. coli strain: (a) unexposed PKNR 5090 versus unexposed PK 4114, (b)
unexposed PKNR 5090 versus unexposed PK 2248, (c) unexposed PKNR 5073 versus unexposed PK
4114, and (d) unexposed PKNR 5073 versus unexposed PK 2248.

695, 780, 924, 1150, 1313, and 1445 cm 1 are only in exposed PK 2248. Both of the
exposed TREC strains only have one band at 1365 cm 1, similar to tigecycline.
The SERS spectra of all exposed TSEC and TREC strains are compared in Fig. 4 and
Table S4. The prominent spectral changes to discriminate the exposed TSEC from
exposed TREC strains are at 618, 624, 924, 1010, 1112, 1150, 1286, 1365, 1373, 1563,
and 1623 cm 1. The bands at 618, 1010, 1112, 1286, 1373, 1563, and 1623 cm 1 are
only present in exposed TSEC strains. The bands at 618, 1112, 1286, and 1563 cm 1 are
associated with tigecycline. The prominent bands at 924, 1150, and 1365 cm 1 only
appeared in the exposed TREC strains, and a band at 618 is shifted to 624 cm 1 with
an increase in intensity in all exposed TREC strains.

Principal component analysis

The ability of SERS to differentiate E. coli strains was evaluated by PCA. Based on high
explained variance, the significant PCs were calculated. A Scree plot showing explained
variance against PCs is in Figure 5a and their relevant score plot is in Figure S1 to
determine which PCs are significant to explain the maximum variance in the data. The
optimal number of PCs was six explaining 95% of the total variance in data. The max-
imum variance of 67.6% is explained by PC–1. Figure 5b is shows a combined PCA
6 S. BASHIR ET AL.

Figure 3. Mean SERS spectra of exposed and unexposed E. coli strains plotted with pure tigecycline
spectrum (cyano): (a) unexposed PKNR 5090 versus exposed PKNR 5090, (b) unexposed PKNR 5073
versus exposed PKNR 5073, (c) unexposed PK 4114 versus exposed PK 4114, and (d) unexposed PK
2248 versus exposed PK 2248.
 ANALYTICAL LETTERS 7

Figure 4. Comparison of mean SERS spectra of all exposed tigecycline-sensitive and tigecycline-resist-
ant E. coli strains.

scatter plot for the exposed and unexposed strains. PC–1 distinguished the exposed and
unexposed strains, while PC–2 separated the TSEC and TREC strains. The exposed
TSEC and TREC strains were clustered on the positive axis and the unexposed TSEC
and TREC strains are clustered on the negative axis of PC–1. Moreover, PC–2 has clus-
tered the unexposed TREC and exposed TSEC strains on its positive axis and unex-
posed TSEC and exposed TREC strains on its negative axis.
To identify the SERS features linked to each class, pairwise PCA was performed and
the scores and loadings of each analysis are depicted in Figure 6. The score plot for
pairwise PCA for unexposed TSEC and TREC strains is depicted in Figure 6a(i). The
unexposed TREC strains were on the positive axis and the unexposed TSEC strains on
the negative axis of PC–1. The loadings shown in Figure 6a(ii) on the negative axis at
657, 733, 1095, 1455, and 1587 cm 1 are associated with unexposed TSEC strains. Those
on the positive axis at 1010, 1127, 1331, 1557, 1678, and 1730 cm 1 are associated with
unexposed TREC strains. Figure 6b(i) demonstrates a PCA score plot for exposed and
unexposed TREC strains. The exposed TREC strains are clustered on the positive axis
and unexposed TREC strains are clustered on the negative axis of PC–1. Fig. 6b(ii)
shows the loadings on the negative axis at 657, 733, 962, 1010, 1029, 1331 and
1557 cm 1 are due to unexposed TREC strains. The loadings on the positive axis at 624,
695, 780, 924, 1150, 1313, 1365, and 1438 cm 1 are caused by exposed TREC strains.
The scores of pairwise PCA analysis for exposed and unexposed TSEC in Fig. 6c(i)
show good differentiation of both categories by PC–1. The exposed TSEC strains are
clustered on the positive axis and unexposed TSEC strains are clustered on the negative
axis of PC–1. Their loadings are in Fig. 6c(ii). The positive loadings of PC–1 at 618,
695, 1010, 1112, 1286, 1313, 1353, 1430, 1563, and 1623 cm 1 are associated with
exposed TSEC strains. The negative loadings at 657, 733, 962, 1095, 1331, and
1587 cm 1 are associated with unexposed TSEC strains. The scores of pairwise PCA for
exposed TSEC and TREC strains are in Fig. 6d(i) clustering exposed TREC strains on
the positive axis and exposed TSEC strains on the negative axis of PC–1. Fig. 6d(ii)
8 S. BASHIR ET AL.

Figure 5. (a) Scree plot for combined principal component analysis showing the percentage of explained
variance and eigenvalues for top ten PCs. (b) Score plot for all exposed and unexposed E. coli strains.

shows the PC–1 loadings in which the features on the positive axis of the PC–1 at 624,
924, 1150, and 1365 are related to exposed TREC strains. The features on the negative
axis of PC–1 at 1010, 1112, 1250, 1286, 1313, 1563, and 1623 cm 1 are associated with
exposed TSEC strains.

Hierarchal cluster analysis

HCA was performed using Ward’s clustering algorithm based on the distance of the
squared Euclidean to cluster and sub-cluster the data based on the similarity in their
SERS features. Figure 7 depicts the first cluster into eight categories, each containing
seven spectra of individual strains. The second cluster contains four categories including
unexposed TSEC and TREC and exposed TSEC and TREC strains on the HCA
 ANALYTICAL LETTERS 9

Figure 6. Pairwise principal component analysis: a(i) score plot of unexposed tigecycline-sensitive E. coli
(TSEC) versus unexposed tigecycline-resistant E. coli (TREC) strains, a(ii) loading plot of unexposed TSEC
versus unexposed TREC, b(i) score plot of unexposed TREC versus exposed TREC strains, b(ii) loading plot
of unexposed TREC versus exposed TREC strains, c(i) score plot of unexposed TSEC versus exposed TSEC
strains, c(ii) loading plot of unexposed TSEC versus exposed TSEC strains, d(i) score plot of exposed TSEC
versus exposed TREC strains, and d(ii) loading plot of exposed TSEC versus exposed TREC strains.

dendrogram. The E. coli strains with or without tigecycline are partitioned into two
well-separated clusters in the third clustering. The classification accuracy was deter-
mined to be 94.6% by dividing the correctly classified spectra by total spectra.
10 S. BASHIR ET AL.

Figure 7. Dendrogram of hierarchal cluster analysis showing discrimination among tigecycline sensi-
tive and resistant strains exposed and unexposed to tigecycline.

Partial least squares – discriminant analysis

To build a PLS-DA model, all data were randomized and split into 60% and 40% training
and test data sets, respectively. The model is trained on the training data set with leave-one-
out cross-validation to choose the optimal number of latent variables (LVs). This value was 7
based on the percentage of variance and the scoring map of these components (Figure S2).
The optimal LVs were used to test the model on the test data-set that was not involved in its
construction. The predicted class against each spectrum of exposed and unexposed E. coli
strains for training and test data sets are presented in Figure 8a and Fig. and demonstrate a
correct prediction. The sensitivity, specificity, and accuracy were 99%, 99.8%, and 100%.
To further evaluate and validate the model’s performance, the receiver operating
characteristic (ROC) approach was used, as illustrated in Figure 8c. The ROC curve was
plotted between the true positive rate/sensitivity and the false positive rate/1-specificity.
The area under the ROC curve was 0.95 demonstrating the excellent performance of
PLS-DA that discriminated between exposed and unexposed E. coli strains.

Discussion
The SERS signatures in bacteria are due to nucleic acids, lipids, phospholipids, polysac-
charides, amino acids, and proteins. Different biochemical compositions of E. coli
 ANALYTICAL LETTERS 11

Figure 8. (a) Predicted class for each spectrum of exposed and unexposed E. coli strains of training
data set. (b) Predicted class for each spectrum of test data set. (c) Receiver operating characteristic
curve between true-positive and false-positive rate with the optimal operating point (green circle).

strains lead to their unique SERS spectra, allowing the discrimination of TSEC and
TREC on the strain level. Furthermore, the exposure of tigecycline to TSEC and TREC
strains is associated with different mechanisms resulting in spectral changes to discrim-
inate the exposed TSEC from TREC strains. To identify the SERS bands associated with
the development of tigecycline resistance in TREC strains, each unexposed TSEC strain
was compared with each unexposed TREC strain. Those SERS bands were mainly asso-
ciated with proteins and carbohydrates indicating the presence of Tet(X). When the
TSEC and TREC strains were exposed to tigecycline, several common spectral features
were present in the spectra of exposed TSEC strains and pure tigecycline (Figure 3a and
b). The absence of Tet(X) in these strains indicated no conversion of tigecycline to any
other form. On the other hand, the TREC strains have plasmid-mediated tet(X3/X4)
genes encoding for Tet(X) that decrease the antibacterial activity of tigecycline by select-
ively hydroxylating it into 11a-hydroxytigecycline (Moore, Hughes, and Wright 2005).
Hence, most spectral features of pure tigecycline are not observed in exposed
TREC strains.
The combined PCA depicts a clear differentiation of each E. coli strain (inter-strain
variability) and indicates different biochemical composition for each exposed and unex-
posed TSEC and TREC strain. In pairwise PCA, the data points are much closer
together in score plots for exposed and unexposed TREC and exposed and unexposed
TSEC [Figure 6b(i) and c(i)] compared to score plots of unexposed TSEC and TREC
and exposed TSEC and TREC. These results indicate greater variability in strains before
and after exposure to tigecycline compared to the variability between exposed/
12 S. BASHIR ET AL.

unexposed TSEC and TREC strains. These results explain why the PC–1 of combined
PCA has differentiated exposed and unexposed strains while PC–2 was used for the
clustering of exposed/unexposed TSEC and TREC strains. The information extracted
from loadings by pairwise PCA is in agreement with the mean SERS spectra (Figure 3).
Moreover, in the HCA dendrogram, all unexposed TSEC and TREC strains are
together on the upper cyano cluster and all exposed TSEC and TREC strains are in the
lower green cluster of the longitudinal coordinate depicting the highest variability in
their SERS features. Only three spectral data points of unexposed PK 2248 strain are
misclassified into unexposed PK 4114 strain, indicated by a red asterisk. This occurred
because both are TREC strains with similar features associated with tigecycline resist-
ance. The sensitivity, specificity, and accuracy of the PLS-DA model also indicate the
good performance of the model. The results obtained from all chemometric analyses
(PCA, HCA, and PLS-DA) are compatible, showing the internal consistency and sup-
porting the findings of the mean SERS spectra.

Conclusions
SERS was employed for the rapid discrimination of TSEC and TREC strains based on
their exposure to tigecycline. The spectra of each E. coli strain were clustered separately
by PCA showing their unique spectral features. HCA also clustered SERS spectra of E.
coli strains into distinct clusters in a dendrogram. The PLS-DA classification algorithm
has distinguished the exposed and unexposed E. coli strains into two categories. The
sensitivity, specificity, and accuracy were 99%, 99.8%, and 100%. The results suggest
that SERS coupled with multivariate data analysis discriminated between TSEC and
TREC strains exposed to tigecycline.

Disclosure statement
The authors have no known competing financial interests or personal relationships that influ-
enced the results reported in this paper.

Funding
This work was supported by the Higher Education Commission (HEC) of Pakistan under Grants
6400/NRPU and 8494/NRPU.

ORCID
Mashkoor Mohsin http://orcid.org/0000-0002-5460-5780

References
Bashir, S., H. Nawaz, M. I. Majeed, M. Mohsin, A. Nawaz, N. Rashid, F. Batool, S. Akbar, M.
Abubakar, S. Ahmad, et al. 2021. Surface-enhanced Raman spectroscopy for the identification
of tigecycline-resistant E. coli strains. Spectrochimica Acta. Part A, Molecular and Biomolecular
Spectroscopy 258:119831. doi:10.1016/j.saa.2021.119831.
 ANALYTICAL LETTERS 13

Bashir, S., H. Nawaz, M. I. Majeed, M. Mohsin, S. Abdullah, S. Ali, N. Rashid, M. Kashif, F.

Batool, M. Abubakar, et al. 2021. Rapid and sensitive discrimination among carbapenem resist-
ant and susceptible E. coli strains using Surface Enhanced Raman Spectroscopy combined with
chemometric tools. Photodiagnosis and Photodynamic Therapy 34:102280. doi:10.1016/j.pdpdt.
2021.102280.
Elnahriry, S. S., H. O. Khalifa, A. M. Soliman, A. M. Ahmed, A. M. Hussein, T. Shimamoto, and
T. Shimamoto. 2016. Emergence of plasmid-mediated colistin resistance gene mcr-1 in a clin-
ical Escherichia coli isolate from Egypt. Antimicrobial Agents and Chemotherapy 60 (5):
3249–50. doi:10.1128/AAC.00269-16.
He, T., R. Wang, D. Liu, T. R. Walsh, R. Zhang, Y. Lv, Y. Ke, Q. Ji, R. Wei, Z. Liu, et al. 2019.
Emergence of plasmid-mediated high-level tigecycline resistance genes in animals and humans.
Nature Microbiology 4 (9):1450–6. doi:10.1038/s41564-019-0445-2.
Hoban, D. J., S. K. Bouchillon, B. M. Johnson, J. L. Johnson, and M. J. Dowzicky. 2005. In vitro
activity of tigecycline against 6792 Gram-negative and Gram-positive clinical isolates from the
global Tigecycline Evaluation and Surveillance Trial (TEST Program, 2004). Diagnostic
Microbiology and Infectious Disease 52 (3):215–27. doi:10.1016/j.diagmicrobio.2005.06.001.
Kashif, M., M. I. Majeed, M. A. Hanif, and A. Ur Rehman. 2020. Surface enhanced Raman spec-
troscopy of the serum samples for the diagnosis of Hepatitis C and prediction of the viral
loads. Spectrochimica Acta. Part A, Molecular and Biomolecular Spectroscopy 242:118729. doi:
10.1016/j.saa.2020.118729.
Keeney, D., A. Ruzin, F. McAleese, E. Murphy, and P. A. Bradford. 2008. MarA-mediated overex-
pression of the AcrAB efflux pump results in decreased susceptibility to tigecycline in
Escherichia coli. The Journal of Antimicrobial Chemotherapy 61 (1):46–53. doi:10.1093/jac/
dkm397.
Kim, S., S. H. Lee, Y. J. Kim, H. J. Lee, and S. Choi. 2019. A rapid tag-free identification of
Escherichia coli antibiotic-resistant isolates using Raman scattering. Analytical Methods 11 (42):
5381–7. doi:10.1039/C9AY01713E.
Lin, Z., X. Zhao, J. Huang, W. Liu, Y. Zheng, X. Yang, Y. Zhang, M. Lamy de la Chapelle, and
W. Fu. 2019. Rapid screening of colistin-resistant Escherichia coli, Acinetobacter baumannii
and Pseudomonas aeruginosa by the use of Raman spectroscopy and hierarchical cluster ana-
lysis. The Analyst 144 (8):2803–10.
Moet, G. J., R. N. Jones, D. J. Biedenbach, M. G. Stilwell, and T. R. Fritsche. 2007. Contemporary
causes of skin and soft tissue infections in North America, Latin America, and Europe: Report
from the SENTRY Antimicrobial Surveillance Program (1998–2004). Diagnostic Microbiology
and Infectious Disease 57 (1):7–13.
Moore, I. F., D. W. Hughes, and G. D. Wright. 2005. Tigecycline is modified by the flavin-
dependent monooxygenase TetX. Biochemistry 44 (35):11829–35. doi:10.1021/bi0506066.
Nordmann, P., T. Naas, and L. Poirel. 2011. Global spread of carbapenemase-producing
Enterobacteriaceae. Emerging Infectious Diseases 17 (10):1791–8. doi:10.3201/eid1710.110655.
Zeiri, L., B. V. Bronk, Y. Shabtai, J. Eichler, and S. Efrima. 2004. Surface-enhanced Raman spec-
troscopy as a tool for probing specific biochemical components in bacteria. Applied
Spectroscopy 58 (1):33–40. doi:10.1366/000370204322729441.
Zhou, H., D. Yang, N. P. Ivleva, N. E. Mircescu, S. Schubert, R. Niessner, A. Wieser, and C.
Haisch. 2015. Label-free in situ discrimination of live and dead bacteria by surface-enhanced
Raman scattering. Analytical Chemistry 87 (13):6553–61. doi:10.1021/acs.analchem.5b01271.