**Robust SERS Platforms Based on Annealed Gold Nanostructures Formed on Ultrafine Glass Substrates for Various (Bio)Applications**

**Lan Zhou 1, Simone Poggesi 1,2, Giuliocesare Casari Bariani 1,2, Rakesh Mittapalli 1, Pierre-Michel Adam 1, Marisa Manzano <sup>2</sup> and Rodica Elena Ionescu 1,\***


Received: 14 February 2019; Accepted: 29 March 2019; Published: 10 April 2019

**Abstract:** In this study, stable gold nanoparticles (AuNPs) are fabricated for the first time on commercial ultrafine glass coverslips coated with gold thin layers (2 nm, 4 nm, 6 nm, and 8 nm) at 25 ◦C and annealed at high temperatures (350 ◦C, 450 ◦C, and 550 ◦C) on a hot plate for different periods of time. Such gold nanostructured coverslips were systematically tested via surface enhanced Raman spectroscopy (SERS) to identify their spectral performances in the presence of different concentrations of a model molecule, namely 1,2-bis-(4-pyridyl)-ethene (BPE). By using these SERS platforms, it is possible to detect BPE traces (10−<sup>12</sup> M) in aqueous solutions in 120 s. The stability of SERS spectra over five weeks of thiol-DNA probe (2 μL) deposited on gold nano-structured coverslip is also reported.

**Keywords:** SERS on ultrafine solid supports; glass coverslips; BPE; thiol-DNA probe; annealed gold nanostructures

#### **1. Introduction**

In recent decades, the use of gold nanoparticles (AuNPs) in the field of light–matter interactions has attracted considerable interest for their potential applications in various sciences, such as biomedical, agricultural, environmental, and forensic investigations, because of their unique optical and chemical properties. AuNPs serve as miniaturized platforms, ideal for the development of ultrasensitive bioassays [1,2]. In fact, a considerable number of protocols have been developed for the preparation of AuNPs, which can be classified into three main groups: (i) the top-down approach based on physical manipulation, for example using an ultrasonic field, electron beam lithography [3], or laser irradiation [4]; (ii) the bottom-up method based on the chemical reduction of chloroauric acid to AuNPs in the presence of reducing and stabilizing agents [5], and (iii) the "on solid supports" approach, using an annealed microscope glass slide coated with thin gold film [6–10].

Among the optical methods, surface enhanced Raman spectroscopy (SERS) has attracted scientists' attention, being used in the identification of unknown substances in analytical chemistry [11], electrochemistry [12], physical chemistry [13], solid state physics, biochemistry, biophysics, and even medicine [14–16]. Nowadays, SERS effects on metal (Au)-coated surfaces are explained using electromagnetic and chemical mechanisms [17,18]. The SERS electromagnetic mechanism is caused by the interactions between the laser excitation on the metal-labeled surface and the scattered Raman

field. The chemical mechanism is caused by the inelastic tunneling of ballistic electrons to the lowest unoccupied molecular orbital of the chemisorbed molecule. The return of the electron to its initial state in the metal—i.e., the recombination of the electron and the hole—emits a Raman-shifted photon [19–21]. Usually, SERS substrates are fabricated either by immobilizing a colloidal silver nanoparticle (AgNP) on 3-aminopropyltriethoxysilane-coated glass coverslips [22] or by dropping tiny volumes of colloidal AgNPs onto microscope glass coverslips [23–25]. Despite the simplicity, such SERS substrates are not stable, and AgNPs are easily displayed by water streams. On the contrary, for biological applications, naked AgNPs must be strongly attached to transparent and biocompatible solid supports for further use in different chemical and biomolecule functionalization steps without any nanoparticle displacements. To solve the inconvenience of the stability of nanoparticles on solid supports, a solution was reported in 2013, which consisted of heating the gold-coated microscope glass pieces at a high annealing temperature in an oven for8h[6]. However, these substrates require training to carefully cut the microscope slide into small pieces to avoid scratching that may affect the homogeneity in the AgNP formation.

A second solution is proposed in the present work and consists of replacing microscope glass pieces with ultrafine coverslips, thus eliminating the cutting step. It should be noted that glass coverslips are typically used for running conventional biological assays and have never been used for robust SERS (bio)applications. The aim of this work is therefore to validate the use of ultrafine glass coverslips as easy-to-handle and inexpensive SERS supports after a high annealing treatment on a hot plate for several hours. A wide variety of chemicals and biomolecules can be detected with these new SERS platforms. To prove the concept, a Raman model molecule, 1,2-bis-(4-pyridyl)-ethene (BPE) [26,27], was selected to study the SERS spectroscopic performances of annealed gold nanostructures on ultrafine glass coverslips.

#### **2. Materials and Methods**

#### *2.1. Materials*

The coverslips were cleaned using Decon 90 (Decon LaboratoriesTM Decon 90TM) liquid detergent (Fisher Scientific, Göteborg, Sweden) and ultrapure water (18.2 MΩ cm) produced by a Millipore Milli-Q water purification system (Molsheim, France). The same water was used for all rinsing steps.

For SERS investigations, a BPE (1,2-bis-(4-pyridyl)-ethene) molecule was purchased from Sigma-Aldrich (Schnelldorf, Germany). Several BPE solutions were prepared from 97% concentrated stock solution, to form six concentrations that were subsequently tested in ultrapure water: 10<sup>−</sup>3, 10−5, 10−7, 10−9, 10<sup>−</sup>12, and 10−<sup>15</sup> M, respectively. For SERS stability studies, a fragment of DNA modified in 5 position with C6 thiol group (TGTTTGAGCGTCATTTCCTTCTCACTATTTAGTGGTTATGAGATTACACGAGG, 53 pb), provided by Eurofins Genomics (Eberseberg, Germany) and here called thiol-DNA probe (10 ng/μL), was suspended in 1xSSPE buffer containing 3 M sodium chloride, 0.23 M sodium phosphate dibasic, 25 mM ethylenediaminetetraacetic acid, pH 7.4. The thiol-DNA was designed to detect *Brettanomyces bruxellensis* spoilage yeast. All the reagents required for the preparation of the SSPE buffer were provided by Sigma-Aldrich.

#### *2.2. Instruments for the Characterization of Gold Nanoparticles Annealed on Coverslips*

Metal evaporation was performed with Plassys MEB 400 (Plassys, Bestek, France). A hot plate (Thermo Fisher Scientific, Waltham, MA, USA) was used for annealing under clean room conditions.

Nanostructured coverslips were characterized with a scanning electron microscope (SEM) (FEG-SU8030, Tokyo, Japan) and an atomic force microscope (AFM) (Bruker ICON, Billeric, MA, USA) with cantilever ScanAsyst-Air in silicon nitride with a tip height of 2.5–8.0 mm. A spring constant of 4 N/m and a reflective aluminum coating on the back side in standard ScanAsyst-Air mode were used to characterize the morphology of AuNPs (data not shown).

SERS spectra were recorded with backscattering geometry using a modified Jobin-Yvon LabRAM (Horiba scientific, Longjumeau, France) and an excitation wavelength of 632.8 nm (11 mW) from the He–Ne laser source, and all the spectra were recorded with a 10× objective Olympus MPlanFl with a 5.2 μm<sup>2</sup> laser spot area. The acquisition time varied from 10 to 120 s, and all the spectra were recorded 3 times with a D filter range between 0 and 0.3.

For sterilization, a Tuttnauer Autoclave Steam Sterilizer 2540ML (Tuttnauer, Villenoy France) was used. The samples were dried in an oven provided by VWR company (DRY-Line drying oven DL 53), and all operations were made under a biological hood provided by Thermo-scientific MSC 1,2 ADV (Illkirch Cedex, France).

#### *2.3. Sample Preparation: Cleaning, Gold Evaporation, and Annealing of Coverslips*

Glass coverslips (Carl Roth GmbH + Co., KG, Karlsruhe, Germany) were degreased with Millipore distilled water and a detergent solution (Decon 90) (ratio 2:8, *v*/*v*) in an ultrasonic distilled water bath (Elmasonic S30H model, Elma Schmidbauer GmbH, Singen, Germany) at 50 ◦C for 15 min according to the procedure used by Jia et al. [6]. In addition, an ultrasonic bath was made with distilled water at 50 ◦C for 5 min. The next step was to carefully rinse each coverslip with distilled water, dry them under a stream of nitrogen, and deposit them on a hot plate at 100 ◦C for 10 min. Further, the coverslips were labelled with a scotch band on an external side for correct handling, fixed on a circular evaporation plate (200 mm diameter), and finally exposed to gold vapors in the evaporator. Different gold thicknesses (2 nm, 4 nm, 6 nm, and 8 nm, respectively) were evaporated on squared glass coverslips at <sup>1</sup> <sup>×</sup> <sup>10</sup>−<sup>5</sup> Torr pressure at 25 ◦C using an evaporation rate of 0.03 nm/s. The resulting gold-coated glasses (4 sets of 12 coverslips/set) were systematically heated on a hot plate preheated to three different temperatures (350 ◦C, 450 ◦C, and 550 ◦C) for different time periods (1, 3, 6, and 9 h, respectively) (Figure 1).

**Figure 1.** Square glass coverslips coated with gold thin films (2 nm, 4 nm, 6 nm, and 8 nm) after 3 h at three different temperatures (350 ◦C, 450 ◦C, and 550 ◦C).

After the annealing procedure, the coverslips underwent an additional cleaning process according to the procedure describe by Jia et al. [6], which involves washing with 70% ethanol in an ultrasonic bath at 30 ◦C for 20 min, rinsing with sterile water, and further washing in an ultrasonic bath with sterile water for 10 min at 30 ◦C. Then, the coverslips were allowed to dry in oven at 50 ◦C for 20 min. After that, the annealed glass coverslips were biofunctionalized by adding 10 μL of thiol-DNA at 10 ng/μL in 1xSSPE at 4 ◦C overnight. The coverslips were subsequently washed with 1.5 mL of sterile water and dried over the biohood. The thiol-DNA was previously treated with a buffer solution containing 10 mM Tris(2-CarboxyEthyl)Phosphine hydrochloride) (TCEP) and 3 M sodium acetate in order to release the thiol group.

#### *2.4. SERS Measurements on Coverslips*

Different BPE concentrations were tested (10<sup>−</sup>3, 10−5, 10−7, 10−9, and 10−<sup>12</sup> M) by deposing tiny drops of 2 μL on gold coverslips. In order to increase the sensitivity of the SERS experiments, different combinations of spectral acquisition time and laser filtering were used: for 10−<sup>3</sup> M and 10−<sup>5</sup> M, the acquisition time was 10 s using the D0 filter, whereas for the lower concentrations, a D0.3 filter was used to avoid the background noise due to the longer excitation time necessary for comparable spectra acquisition. An acquisition time of 30–120 s was studied. The stability of SERS spectra over five weeks for gold nanostrutured coverslips modified with thiol-DNA probe with an acquisition time of 10–30 s and using a D0.3 filter is also reported.

#### **3. Results and Discussion**

#### *3.1. SEM Characterization*

It is well known that the SERS properties of gold nanostructures are strongly influenced by the size, distribution, and spacing between particles [8]. In the preparation of the AuNP process, including cleaning, evaporation, and annealing protocol, the morphology of the substrate can be modulated by controlling different experimental conditions, such as the Millipore and Decon 90 distilled water ratio, gold film thickness, evaporation pressure, evaporation rate, annealing time, and annealing temperature. In our experiments, three parameters—the thickness of the evaporated gold film, the annealing temperature, and the annealing time—played an important role in the morphology of the gold nanoparticles and SERS properties. Thus, SEM studies were conducted for four different gold film thicknesses (2 nm, 4 nm, 6 nm, and 8 nm, respectively) at three different annealing temperatures (350 ◦C, 450 ◦C, and 550 ◦C) and for four different annealing times (1, 3, 6, and 9 h) (Figure 2).

#### 3.1.1. Influence of the Annealing Temperature on the Formation of Gold Nanoparticles

Evaporated gold films of 2 nm, 4 nm, 6 nm, and 8 nm on coverslips showed different colors, from light blue (2 nm Au) to blue (4 nm Au), to light green (6 nm Au), or to darker green (8 nm Au). These colors changed significantly for each gold thickness after 3 h of annealing at different temperatures. The highest temperature produced a violet color for the 2 nm gold film, whereas for the 4 nm, 6 nm, and 8 nm films, the color appeared from light violet to dark purple, respectively (Figure 1). SEM images of the evaporated samples and the annealed samples are shown in Figure 2. The size of the gold nanoparticles increased with the increase of the thickness of the film (2 nm, 4 nm, 6 nm, and 8 nm), which corresponded to the color variation before and after annealing at different temperatures.

For glass coverslips coated with 2 nm Au, the interparticle distances, or proportion of background (PB), increased when temperatures rose from 350 ◦C to 550 ◦C (Figure 3). On the contrary, for samples coated with 6 nm and 8 nm Au, the PB values at 550 ◦C ranged from 60.94% to 63.63%, while at 450 ◦C, the PB values were 61.39% and 65.53%, respectively. Interestingly, the PB values for the 4 nm Au sample showed no great variation when annealed at 350 ◦C (60.29%), 450 ◦C (60.07%), or 550 ◦C (60.41%).

In conclusion, the temperature definitively influenced the sizes and shapes of the gold nanoparticles and the interparticle distances, with respect to the gold thickness evaporated on the coverslips.

**Figure 2.** SEM images of square glass coverslips gold coated (2 nm, 4 nm, 6 nm, and 8 nm) after 3 h at different temperatures (350 ◦C, 450 ◦C, and 550 ◦C). AuNPs: gold nanoparticles.

**Figure 3.** The proportion of background for the annealed square glass coverslips gold coated (2 nm, 4 nm, 6 nm, and 8 nm, respectively) after exposure at three different temperatures (350 ◦C, 450 ◦C, and 550 ◦C). These numbers are also used in Figure 2 to indicate the SEM image for every substrate.

#### 3.1.2. Influence of Gold Thickness on Coverslips on Nanoparticle Distribution

As robust and stable SERS platforms, square coverslips coated with gold of 2 nm, 4 nm, 6 nm, and 8 nm were proposed and annealed at 550 ◦C for 3 h. For these samples, the particle size distribution, and the proportion of background are reported in Figure 4A,B.

**Figure 4.** The size distribution of AuNPs on coverslips (**A**) and the proportion of background for different gold thicknesses (2 nm, 4 nm, 6 nm, and 8 nm) after annealing at 550 ◦C for 3 h (**B**).

Figure 4A shows that by increasing the thickness of gold, the size of the AuNP nanoparticles and their distribution percentage increase. Similarly, the size of nanoparticles affects the interdistance between particles. Thus, after the annealing of 2 nm Au film on the glass, the AuNPs ranged mainly from 6 (size distribution 31.8%) to 8 nm (29.9%), whereas for 4 nm Au, the nanoparticles ranged from 10 (37.9%) to 15 nm (38.8%). For 6 nm Au, the AuNPs ranged from 20 (39.7%) to 30 nm (35.9%), and finally, for 8 nm Au, the AuNPs ranged from 20 (38.1%) to 40 nm (24.8%).

On the other hand, it was found that the proportion of background for the coverslips coated with four different gold thicknesses (Figure 4B) was the smallest for 4 nm Au (60.41%) and the highest for 2 nm Au (74.77%). Additionally, for coverslips coated with 6 nm and 8 nm Au, the background was 60.94% and 63.63%, respectively.

In the current SERS studies, the 4 nm Au coated coverslips after 550 ◦C showed the largest nanoparticle surface coverage and the lowest interparticle distances compared with the other tested thicknesses (2 nm, 6 nm, and 8 nm, respectively).

3.1.3. Influence of Annealing Time on the Nanostructuration of Coverslips

In order to prepare large-scale gold nanoparticles with stable optical characteristics, the effect of 4 nm Au evaporated and annealed at 550 ◦C at different annealing times (1, 3, 6, and 9 h) is illustrated in Figure 5. On other hand, the nanoparticle size distribution and the proportion of background for SEM images (Figure 2) are analyzed using the public domain ImageJ software platform, developed at National Institutes of Health (Figure 6).

Experimentally, the coverslips heated for 1 h at 550 ◦C formed nanoparticles in the range of 15–20 nm (36.8–23%). Similar sample evolution was obtained for glasses after 6 h at the same temperature when the AuNPs ranged from 10 (23.6%) to 15 nm (38.8%), while the AuNP size after 3 h displayed a uniform distribution from 10 (37.9%) to 15 nm (38.8%) compared with the others, corresponding to the SEM image (Figure 5).

**Figure 5.** SEM images of AuNPs on square glass coverslips coated with 4 nm and annealed for different time periods (**A**) 1 h, (**B**) 3 h, (**C**) 6 h, and (**D**) 9 h at 550 ◦C.

The coverslips annealed for 9 h showed a high distribution at 5–10 nm (45.8–19.0%) (see Figure 6A). However, as shown in Figure 6B, the proportion of background increased following the evaporated gold film thickness, becoming thicker over time: 9 h (70.72%) > 6 h (64.87%) > 3 h (61.19%) >1 h (56.51%). In detail, even though the sample annealed for 9 h had a very high distribution of 45.8% at 5 nm, the largest proportion of the background of the sample was 70.72%, which corresponded to the coverage of the smallest area of the 9 h sample. On the other hand, the samples annealed for 1 h and 3 h had lower proportions compared with the samples annealed for 6 h and 9 h. The lower proportion of background of the larger surface coverage was obtained, and considering the uniform distribution of the nanoparticles and better surface coverage, we used the samples annealed at 550 ◦C for 3 h.

**Figure 6.** Analysis of AuNPs based on SEM images showing the size distribution of gold nanoparticles on annealed coverslips (**A**) and the proportion of background after annealing the 4 nm gold coated coverslips for different time periods (1, 3, 6, and 9 h) at 550 ◦C (**B**).

#### *3.2. SERS Characterization*

The SERS tests were initially performed on classical microscope glass slide supports modified with 4 nm gold film and annealed at 550 ◦C in the oven for 8 h according to the procedure described by Jia et al. [6]. These substrates proved to be inappropriate for SERS measurements, because the glass slide produced strong fluorescence interferences that abnormally altered the optical signals. Therefore, several solid supports are here proposed for SERS investigations: plastic petri dishes, glass coverslips, plastic pipettes, Eppendorf tubes, plastic cuvettes, and quartz crystals microbalance (QCM) (Supplementary Materials, Figure S1). The SERS measurements show that the best solid supports were the ultrafine glass coverslips for further gold nanostructuration due to the absence of fluorescence interferences. Then, the SEM morphology also confirmed the evolution of SERS signals for annealed

gold coated coverslips at 350 ◦C, 450 ◦C, and 550 ◦C for 3 h (Figure S2A–C). Among the different spectroscopic investigations (Figure S3), it was found that the best SERS substrate is the ultrafine square glass coverslip coated with 4 nm Au (Figure S3B) annealed at 550 ◦C for 3 h (Figure S2C), due to the absence of background SERS peaks that could mask the presence of specific SERS peaks produced by (bio)molecules once immobilized on nanoparticles.

3.2.1. SERS Spectrum of BPE Molecule on 4 nm Gold-Annealed Coverslip

The SERS tests were carried out in the presence of a model molecule, 1,2-bis-(4-pyridyl)-ethene, which has interesting bonds and atoms giving good SERS spectra when deposited on annealed gold coated coverslips, as demonstrated in Figure 7.

**Figure 7.** Surface enhanced Raman spectroscopy (SERS) spectrum of the 1,2-bis-(4-pyridyl)-ethene (BPE) (1 mM) on annealed gold nanostructured coverslip (4 nm Au, 550 ◦C for 3 h on a hot plate), after three times of acquisition of 10 s and using a D0 filter.

As reported, the main peaks at 1601 and 1630 cm−<sup>1</sup> correspond to the C–N stretching mode in the pyridyl ring and the BPE vinyl group vibration, respectively [9], while the peaks at 1193 and 1235 cm−<sup>1</sup> refer to the ring breathing mode of pyridine and the vibrational movement of the nitrogen atom in pyridyl, respectively. In the present work, the peak at 1012 cm−<sup>1</sup> can be attributed to the chemical absorption of BPE molecules onto AuNPs on coverslips.

#### 3.2.2. SERS Spectra of Different BPE Concentrations

SERS signals were recorded and compared for different BPE concentrations deposited on gold-annealed coverslips. These SERS measurements confirmed that the best conditions for the detection of very low BPE concentrations are as follows: 4 nm Au on glass heated at 550 ◦C for 3 h. As is shown in Figure 8, gold nanostructured coverslips made possible the SERS detection of the lower concentration of the BPE molecule at 10−<sup>12</sup> M.

**Figure 8.** SERS spectra of BPE molecules of different concentrations (10<sup>−</sup>3, 10−5, 10−7, 10−9, and 10−<sup>12</sup> M) using 4 nm gold-coated coverslips annealed at 550 ◦C for 3 h on a hot plate. Inset-photo of a coverslip after the deposition of five different BPE concentrations.

Figure 9 depicts the intensity variation of the three main SERS peaks (1193 cm<sup>−</sup>1, 1630 cm−1, and 1601 cm<sup>−</sup>1) versus the decimal logarithmic of four BPE concentrations. Whatever the wavenumber, the intensity variation exhibited an autonomous decay when the concentration of BPE decreased. A linear fit was used to model the experimental SERS measurements. A pronounced linear behavior was observed for the 1630 wavenumber, for which a linear regression coefficient of 0.9976 was calculated. The values of the modeled slopes that represent the sensitivity were of the same order of magnitude.

**Figure 9.** SERS intensity of three wavenumber main peaks (1193 cm<sup>−</sup>1, 1630 cm−1, and 1601 cm−1) as a function of BPE concentration.

The enhancement factor (EF) was calculated using the equation EF = (ISERS/IR) × (NR/NSERS) and was found to be equal to 2.71 <sup>×</sup> <sup>10</sup>7. ISERS represents the intensity of the 1630 cm−<sup>1</sup> BPE band, measured for 10−<sup>5</sup> M concentration, while IR represents the intensity of the Raman band for 10−<sup>3</sup> M on reference glass. NSERS and NR represent the number of molecules formed as a layer of 10 nm thickness under the laser spot and the number of the BPE molecules in the focal volume. The values of IR and NR are the same as those reported in [28]. Gold nanoparticles covered 40% of the surface under the spot. By using the same calculation method, NSERS was found to be equal to 125 molecules for a surface spot laser of 5.2 μm2. In the case of 10−<sup>5</sup> M BPE content, an ISERS value of 9000 was measured.

3.2.3. SERS Signal Stability of Thiol-DNA Deposited on Gold-Annealed Coverslip Substrate

A thiol-DNA biofunctionalized gold-annealed coverslip was tested over five weeks to evaluate the substrate's SERS signal stability (Figure 10). Interestingly, the intensity of SERS increased after two weeks and decreased after four weeks. This confirms that the nanostructuration of the coverslip was stable for more than a month. The stability tests were stopped after five weeks, because the thiol-DNA probe on AuNPs presented a strong attenuation of the SERS intensity.

**Figure 10.** Evolution of SERS intensity of the thiol-DNA probe on annealed gold-coated coverslip over five weeks and obtained after three acquisition times, showing a 10 s spectra with a D0.3 filter (**A**). Photo of the sample after five weeks (**B**).

#### **4. Conclusions**

Large-scale, annealed, gold nanostructures were fabricated for the first time on ultrafine glass coverslips. Several parameters have been optimized to conclude that 4 nm gold-coated coverslip heated at 550 ◦C on a hot plate for 3 h had the greater sensitivity of the SERS spectrum to different BPE concentrations. By using the newly SERS annealed coverslips platforms it was possible to detect a BPE concentration of 10−<sup>12</sup> M. Moreover, the stability of SERS spectra intensity over five weeks of a thiol-DNA probe (10 ng/μL) was also monitored. On the basis of these results, annealed gold coverslips can be considered as ideal substrates in the construction of ultrasensitive SERS nanobiosensors.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-6374/9/2/53/s1, Figure S1: SERS spectra of various naked solid supports: plastic petri dish, glass coverslip, plastic pipette, Eppendorf tube, plastic cuvette and quartz QCM crystal, Figure S2: SERS signals of naked annealed gold films (2, 4, 6 and 8 nm) on coverslips after 3 h, Figure S3: SERS signals of naked annealed gold films on coverslips at 550 ◦C for different periods.

**Author Contributions:** Conceptualization and methodology, R.E.I.; software, S.P., L.Z. and G.C.B.; validation, R.E.I., L.Z., and S.P.; formal analysis, R.E.I. and P.-M.A.; investigation, R.M.; resources and data curation, R.E.I. and L.Z.; writing—original draft preparation, R.E.I.; writing—review and editing, L.Z., M.M. and R.E.I.; visualization, P.-M.A.; supervision, project administration and funding acquisition, R.E.I.

**Funding:** This research was funded by EIPHI Graduate School grant number ANR17-EURE-0002.

**Acknowledgments:** R.E.I., G.C.B. and R.M. are thankful for the financier support of EIPHI Graduate School (contract "ANR17-EURE-0002"). The authors thank the NANOMAT Champagne-Ardenne Regional Platform for their financial support. L.Z. kindly thanks the Chinese Scholarship Council (CSC) for funding her PhD scholarship in France (October 2017–April 2021). S.P. thanks the Master thesis research program at the University of Udine for the financial support in France (September–December 2018).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **The Application of a Nanomaterial Optical Fiber Biosensor Assay for Identification of** *Brucella* **Nomenspecies**

#### **Kelly McCutcheon 1, Aloka B. Bandara 2, Ziwei Zuo 1, James R. Heflin <sup>1</sup> and Thomas J. Inzana 2,3,\***


Received: 22 March 2019; Accepted: 14 May 2019; Published: 21 May 2019

**Abstract:** Bacteria in the genus *Brucella* are the cause of brucellosis in humans and many domestic and wild animals. A rapid and culture-free detection assay to detect *Brucella* in clinical samples would be highly valuable. Nanomaterial optical fiber biosensors (NOFS) are capable of recognizing DNA hybridization events or other analyte interactions with high specificity and sensitivity. Therefore, a NOFS assay was developed to detect *Brucella* DNA from cultures and in tissue samples from infected mice. An ionic self-assembled multilayer (ISAM) film was coupled to a long-period grating optical fiber, and a nucleotide probe complementary to the *Brucella* IS*711* region and modified with biotin was bound to the ISAM by covalent conjugation. When the ISAM/probe duplex was exposed to lysate containing ≥100 killed cells of *Brucella*, or liver or spleen tissue extracts from *Brucella-*infected mice, substantial attenuation of light transmission occurred, whereas exposure of the complexed fiber to non-*Brucella* gram-negative bacteria or control tissue samples resulted in negligible attenuation of light transmission. Oligonucleotide probes specific for *B. abortus*, *B. melitensis*, and *B. suis* could also be used to detect and differentiate these three nomenspecies. In summary, the NOFS biosensor assay detected three nomenspecies of *Brucella* without the use of polymerase chain reaction within 30 min and could specifically detect low numbers of this bacterium in clinical samples.

**Keywords:** *Brucella abortus*; *Brucella melitensis*; *Brucella suis*; optical fiber; biosensor; nucleotide probe; light transmission; diagnosis

#### **1. Introduction**

Brucellae are bacterial pathogens responsible for brucellosis of domestic and wild animals and are zoonotic pathogens for humans. *Brucella* spp. are small gram-negative, nonmotile, aerobic, and slow-growing coccobacilli. Despite the recognition of brucellae as a single genospecies based on DNA-DNA hybridization studies, they are systematically classified based on host specificity. The main terrestrial nomenspecies are *B. abortus* (cattle), *B. melitensis* (goats and sheep), *B. suis* (pigs), *B. canis* (dogs), *B. ovis* (sheep), and *B. neotomae* (woodrats) [1,2]. In addition, *Brucella* spp. can also be isolated from marine mammals [3,4]. Human infections are acquired by consuming unpasteurized milk and dairy products or by direct exposure to animals and their carcasses. Human brucellosis resulting from direct exposure is primarily a disease of farmers, shepherds, veterinarians, microbiologists, butchers, and slaughterhouse workers [5,6].

Wild animals play an important role in the epidemiology of *Brucella* infections. *Brucella* spp. remain enzootic in wild elk and bison in the Greater Yellowstone region that includes areas of Montana,

Idaho, and Wyoming. As a result, these animals are a reservoir for *B. abortus* in the United States [7]. Transmission of *Brucella* spp. to susceptible cattle normally occurs by ingestion or oral contact with infected fetuses that have been aborted, fetal fluids and membranes, or uterine discharges [8]. Elk that congregate on feeding grounds from November through April overlap with the peak time period when *Brucella* is transmitted to other animals (February through June) [9]. Maichak et al. reported that as many as 12% of the elk attending feeding grounds come into contact with non-infectious elk fetuses placed on these sites [10]. Bison normally congregate in large numbers, which increases the likelihood they will come into contact with *Brucella-*infected fetuses and discharges. Such congregation increases the possibility that infected bison could transmit *Brucella* to cattle herds in the area [7]. As a result, farmers may unnecessarily kill elk or bison that wander out of the park and onto private farmlands.

Development of reliable and cost-effective diagnostic tests for use in elk and other wildlife is a high research priority. Reliable and portable diagnostic assays that can be carried out in the field by non-specialist personnel are urgently needed to minimize the spread of the disease among wildlife and its transmission to domestic animals and humans.

Biosensors combine biological molecules with a physicochemical transducer. Biological components incorporated into biosensors may include nucleic acids, enzymes, antibodies, etc., and the transducer may be optical, electrochemical, thermometric, or piezoelectric. Regardless, the detection of the target biological material results in a measurable signal. The advantages of optical fibers (light, inexpensive, and low interference) have established them as essential instruments of sensor technology [11]. Biosensors that consist of optical fibers transmit light based on total internal reflection through their transduction elements. The sensor produces a signal that can be analyzed and is in proportion to the concentration of the molecule that binds to the biological element on the sensor. Grating devices in the optical fiber induce a periodic variation in the refractive index of the optical fiber's core. As a result, there is a significant drop in the amount of light transmitted through the fiber at a specific wavelength. The specific wavelength can be changed to account for temperature, pressure, or the type of binding event [12].

Layer-by-layer films, also known as ionic self-assembled multilayer (ISAM) films, are a novel type of materials that enable the user to modify the structure and thickness of the thin film at nanometer levels. The assembly of such films is simple and inexpensive [13,14]. As a result, optical fibers containing these nanoscale overlays substantially enhance, through direct light transmission, the detection of antigen binding to antibody or DNA hybridizing to complementary DNA. Furthermore, these sensors can be organized into a device that is rugged and portable [15,16]. For the detection of infectious agents, these fiber-optic biosensors can be used as rapid diagnostic or screening tests prior to culture, serology, or other means of diagnosis. A variety of fiber grating-based biosensor platforms have recently been developed [17–20]. For the work described here, a nanomaterial optical fiber biosensor (NOFS) assay was successfully used to detect *Brucella* DNA in culture lysates and in infected animal tissues.

#### **2. Materials and Methods**

#### *2.1. Oligonucleotide Primers and Probes*

The oligonucleotide probes and primers (Table 1) were designed manually based on the DNA sequences of the respective genes/regions in GenBank and were purchased from Integrated DNA Technologies, Collinsville, IL, USA. The IS*711* DNA region is present in all known nomenspecies of *Brucella*, but not in other bacteria. Therefore, the primers IS*711*-For and IS*711*-Rev and probes IS*711*-BIO and IS*711*-DIG from the IS*711* region (Table 1) were used for detection of all *Brucella* nomenspecies and to distinguish them from other bacterial species. In order to identify and differentiate the three major *Brucella* nomenspecies (*Brucella abortus*, *B. melitensis*, and *B. suis*), the oligonucleotide probes BruAb2\_0168 (GenBank accession AE017224.1), Melitensis\_0466 (GenBank accession AE008918.1), and Suis\_TraJ (GenBank accession CP024421.1) (Table 1) were used. A search using the NCBI BLAST

program confirmed the specificity of the DNA regions used for identifying and distinguishing the respective *Brucella* species.


**Table 1.** Oligonucleotide probes and primers used for detecting major *Brucella* nomenspecies.

#### *2.2. Bacterial Strains and Culture Conditions Used*

The *Brucella* nomenspecies and other bacterial species used as controls in this study are listed in Table 2. The bacteria were grown to mid-log phase in brain heart infusion (BHI) broth. Bacteria were harvested by centrifugation, washed, resuspended in phosphate buffered saline, pH 7.2 (PBS), and killed by boiling for 20 min (confirmed by viable plate count). Serial dilutions of killed cell suspensions were made in PBS, and genomic DNA was harvested using the DNAeasy Blood and Tissue kit (Qiagen, Valencia, CA, USA).

**Table 2.** Bacterial species and strains used.


#### *2.3. PCR*

PCR, when used, was performed in 25 μL volumes and included 10 pmol each of the primers Primer-IS*711*-For and Primer-IS*711*-Rev (Table 1), 1 mM MgCl2, 200 μM dNTPS, 1X concentration of One*Taq* Standard Reaction Buffer (New England Biolabs, Ipswich, MA, USA), 1.25 units of One*Taq* DNA Polymerase (New England Biolabs), and template DNA. Template DNA included either 26 ng of genomic DNA or 1 <sup>μ</sup>L of heat-killed, cell lysate from 1 <sup>×</sup> 105 cells/mL to 3 <sup>×</sup> 1010 cells/mL. Reaction conditions were an initial denaturation temperature of 95 ◦C for 5 min, followed by 30 cycles of 95 ◦C/1 min, 57 ◦C/1 min, 72 ◦C/1 min, and a final extension at 72 ◦C/10 min.

#### *2.4. Enzyme-Linked Immunosorbent Assay (ELISA)*

An ELISA was designed using Magnalink Streptavidin Magnetic beads (Solulink Inc., San Diego, CA, USA). The protocol was as described by the manufacturer (Solulink, Inc.) with modification. Briefly, 60 pmol of the biotinylated probe (Probe-IS*711*-BIO) (Table 1) in 250 μL of nucleic acid binding and wash buffer (50 mM Tris-HCl, 150 mM NaCl, 0.05% Tween 20, pH 8.0) was incubated for 30 min at room temperature with the beads in 1.5 mL microcentrifuge tubes. The heat-denatured PCR products or genomic DNA were incubated with the beads coupled to the biotinylated probe in hybridization buffer (3× SSC, 0.05% Tween 20) for 2 h at 45 ◦C. A digoxigenin-labeled probe (Probe-IS*711*-DIG) (Table 1) in hybridization buffer was then incubated with the bead/probe/DNA triplex for 2 h at 45 ◦C. The DIG Detection Starter Kit from Roche (Sigma-Aldrich, St. Louis, MO, USA) was used to determine binding of the probe to the triplex complex. The ELISA was designed solely to confirm that the designed probe would hybridize with the *Brucella* genomic DNA, and not as a diagnostic assay itself. Therefore, quantitative data were not obtained.

#### *2.5. Fabrication of the ISAM Film*

ISAM films were fabricated using procedures described by the authors [21]. The ISAM method simply involves the alternate dipping of a charged substrate (optical fiber) into an aqueous solution of a polycation and an aqueous solution of a polyanion at room temperature. The optical fiber was immersed in an aqueous 10 mM poly-allylamine hydrochloride (PAH) (pH 7.0) solution for 2 min then rinsed three times in distilled water. The fiber was then immersed in a similar aqueous solution of 10 mM poly-1-[p-(3 -carboxy-4 -hydroxyphenylazo) benzenesulfonamido]-1,2-ethanediyl (PCBS) (pH 7.0) for 2 min and rinsed again. The final layer was always the negatively-charged PCBS. These two steps were repeated until the optimal number of bilayers was obtained, which, for this assay, was four layers (Figure 1).

**Figure 1.** Assembly of the ionic self-assembled multilayer (ISAM) film. Polycationic and polyanionic solutions were alternately deposited on the optical fiber to form the ISAM film.

#### *2.6. Coupling the Probe to the ISAM Film*

The ISAM film was incubated with 0.6 mL of 0.17 M freshly prepared *N-*(3-dimethylaminodipropyl)-*N'-*ethylcarbodiimide (EDC), 0.17 M *N-*hydroxysulfosuccinimide (NHS), and 60 pmol of biotinylated oligonucleotide probe in PBS, pH 7.0, at room temperature for 30 min.

#### *2.7. Conjugation of Streptavidin to the ISAM Film*

An alternative method to couple the probe onto the ISAM film involved using a streptavidin intermediate. Four bilayers were deposited onto the optical fiber, leaving PCBS with negatively-charged carboxyl groups exposed. Then, 40 μL of streptavidin (1 mg/mL in PBS, pH 7.0) was mixed with 0.6 mL of cross-linker solution (0.17 M EDC and 0.17 M NHSS in PBS, pH 7.0). The mixture was added to the fiber and incubated for 8 h, with mixing every 15 min. The fiber was then rinsed and the biotinylated probe was added for spontaneous coupling to streptavidin.

#### *2.8. NOFS Assay*

The NOFS assay consists of turnaround point long-period gratings (TAP-LPGs) with ISAM films adsorbed on fiber cladding. The TAP-LPGs are TrueWave RSTM (OFS) single-mode optical fibers with a grating period of 116 μm written by a 248 nm excimer laser through a chrome-plated mask. The grating couples to the LP0,14 cladding mode of the fiber. White light model SLD-11OESL003 (FiberLabs, Inc. Fugimino-Shi, Saitama, Japan) was coupled to the optical fiber, and the spectra were measured by an optical spectrum analyzer (ANDO AQ6317) following the deposition of materials onto the TAP-LPG.

Bacteria grown in broth medium were harvested and washed in PBS. Serials dilutions of cultures were inoculated to agar medium to determine the colony forming units (CFU)/mL. *Brucella* cultures were lysed by boiling for 30 min. Loss of viability was confirmed by viable plate count before removing the bacteria from the biosafety level-3 laboratory. Prior to beginning the assay, preparations of genomic DNA, PCR products, and lysates of bacterial cells were boiled for 5 min. The film/probe duplex was incubated with the heat-denatured sample (genomic DNA, DNA regions amplified by PCR, amplified DNA from killed cells, dilutions of lysed cells, or dilutions of extracts of tissues from infected animals) for 50 min to allow hybridization between the probe and sample DNA to occur [21]. The TAP-LPG was tuned beyond the turnaround point such that the two narrow-band peaks merged into a single broadband peak that changed attenuation strength as the coupling between the core and cladding mode was modified by the addition of material to the cladding surface. As light in the range of 1400–1700 nm was transmitted through the ISAM fiber, an optical analyzer was used to record the attenuation in light transmission at 1550 nm. This attenuation in light transmission occurred due to the increase in coupling of light out of the core of the optical fiber due to sample DNA hybridizing to the DNA probe. An example of the series of spectra, as material binds to the cladding surface, is shown in Figure 2. The thickness of the ISAM films used is determined in order to set the attenuation at approximately half of the maximum attenuation that occurs before peak split into two separate narrowband peaks.

**Figure 2.** Transmission spectra after different steps of the assay. Adding the probe to the ISAM-coated fiber caused a large increase in attenuation. The attenuation was further increased after exposing the sensor to the positive control. However, further exposure of the fiber to the negative control did not result in any further change in attenuation, as was expected. As a result, the negative control spectrum overlaps and is indistinguishable from that of the positive control.

#### *2.9. Detection of Brucella DNA in Tissues of Infected Mice*

Groups of two female BALB/c mice each were inoculated intraperitoneally with 6 <sup>×</sup> <sup>10</sup><sup>4</sup> CFU/mouse of *B. abortus* strain 2308, *B. melitensis* strain 16 M, or *B. suis* strain 1330. Two mice were injected with PBS as controls. One week after inoculation the mice were euthanized, and 0.1 g of spleen and liver samples were collected. The tissues were ground with 1 mL of PBS. Half of the volume of the extracts of the ground tissues were used in viable plate count determination to determine the number of bacteria/g of tissue. The remaining half (500 μL of heat-denatured cell-free extract corresponding to 0.05 g of tissue) was used per each run of the NOFS assay, as previously described [21]. DNA in these samples were not amplified by PCR prior to NOFS testing. Serial dilutions of the extract were also cultured onto BHI agar, and bacterial colony counts were determined as CFU after 72 h of incubation at 37 ◦C with 5% CO2.

#### *2.10. Statistical Analyses*

The standard deviations of the means were calculated from assays repeated at least three times. The online calculator (http://www.danielsoper.com/statcalc3/calc.aspx?id=43) was used to determine the analysis of variance, which was used to compare the transmission attenuation between different samples. The online calculator (http://www.socscistatistics.com/tests/studentttest/Default2.aspx) was used to calculate *p*-values from the Student *t-*test and to compare the attenuation of light transmission recorded for infected versus control tissue extracts. Student *t-*tests were also used for analysis of the attenuation of light transmission after exposure of the probe to two different PCR products. Results with calculated *p-*values of less than 0.05 was considered significant. The cutoff value in percent attenuation of light transmission that was used to differentiate negative from positive samples was calculated by multiplying the standard deviation of the true negative isolates tested by 3. This cutoff value could change depending on the optical fiber used and varied from 0.6% light attenuation to 3.2% light attenuation. Larger cutoff values were due to larger standard deviations of the negative controls.

#### **3. Results**

#### *3.1. Specificity of the DNA Probes*

DNA amplification of *Brucella* and heterologous species using oligonucleotide primers to the IS*711* region (Table 1) confirmed the specificity of the IS*711* region for *Brucella* nomenspecies. An approximately 300 bp-size amplicon was obtained when 50 ng of genomic DNA or lysates containing at least 5 <sup>×</sup> 103 killed cells of each *Brucella* nomenspecies was used in PCR reactions. However, visible amplicons were not seen in agarose gels when lysates representing 8 <sup>×</sup> <sup>10</sup><sup>2</sup> or fewer *Brucella* cells were used in PCR reactions. PCR amplicons were also not seen when lysates containing up to 3 <sup>×</sup> <sup>10</sup><sup>7</sup> killed cells of *Escherichia coli*, *Pseudomonas aeruginosa*, or *Salmonella* Typhimurium (negative controls) were used (Figure 3).

#### *3.2. Validation of Target DNA for Hybridization to the DNA Probe*

An ELISA was used to validate that target DNA hybridized to probes of the IS*711* region. After the DNA and initial bead-bound probe were allowed to hybridize, a second DIG-labeled oligonucleotide IS*711* probe to a different region of the DNA was added. Only if the sample DNA bound to the first probe would the second DIG-labelled probe bind and specifically detect *Brucella* DNA. The use of genomic DNA or lysate of killed cells in the absence of PCR did not produce a colorimetric change, indicating there was inadequate complementary DNA sequence from the genomic DNA to be detected in this assay. However, following DNA amplification of the test sample (genomic DNA or lysate containing 8 <sup>×</sup> 103 cells of *Brucella*), a positive reaction was obtained (Figure 4), but not if lysate representing 8 <sup>×</sup> <sup>10</sup><sup>2</sup> or fewer *Brucella* cells were tested (not shown). These results were consistent with results obtained by gel electrophoresis of PCR products and confirmed that the probe successfully bound to amplified DNA from the IS*711* region and was valid for use in the NOFS assay.

**Figure 3.** PCR amplicons from *Brucella* and control strains. Lanes and lysates representing the number of cells from nomenspecies used for PCR: 1 and 21, molecular size standards; 5 and 11, blank wells; 2–4, 3 <sup>×</sup> 107, 3 <sup>×</sup> 105, and 3 <sup>×</sup> 10<sup>3</sup> cells of *Pseudomonas aeruginosa* (the same negative results for the same number of cells are not shown for *Escherichia coli* and *Salmonella* Typhimurium); 6–10, 8 <sup>×</sup> 106, 8 <sup>×</sup> 105, <sup>8</sup> <sup>×</sup> 104, 8 <sup>×</sup> <sup>10</sup>3, and 8 <sup>×</sup> <sup>10</sup><sup>2</sup> cells of *B. abortus*; 12–16, 6 <sup>×</sup> 106, 6 <sup>×</sup> <sup>10</sup>5, 6 <sup>×</sup> <sup>10</sup>4, 6 <sup>×</sup> 103, and 6 <sup>×</sup> <sup>10</sup><sup>2</sup> cells of *B. suis*; 17–20, 5 <sup>×</sup> <sup>10</sup>6, 5 <sup>×</sup> 105, 5 <sup>×</sup> 104, and 5 <sup>×</sup> 103 cells of *B. melitensis*; 22, positive control (26 ng of *B. abortus* genomic DNA; 23, negative control.

**Figure 4.** Magnetic bead ELISA. All tubes contained the beads, biotinylated probe of IS*711* gene, and digoxigenin-labelled probe to a distinct IS*711* DNA region. Tube 1 contained the PCR amplicon from a lysate of 7 <sup>×</sup> <sup>10</sup><sup>6</sup> cells of *Brucella abortus*. Tube 2 contained all the reaction components of tube 1, but the genomic DNA was not amplified by PCR.

#### *3.3. Identification of Brucella Nomenspecies by NOFS Assay*

Reaction of the ISAM/probe (IS*711*) duplex with the entire 25 μL of PCR amplicons from a lysate representing 10<sup>4</sup> cells of *B. abortus* strain 2308, *B. melitensis* strain 16 M, or *B. suis* strain 1330 in 500 μL of PBS resulted in 18.7%, 18.6% and 20.11% attenuation of light transmission, respectively, with positive results being above 0.6% light attenuation. When lysate from 10<sup>2</sup> cells of these same nomenspecies were tested, 8.8%, 14.2% and 13.6% attenuation of light transmission was obtained, respectively (Figure 5). These results indicated that the NOFS assay was capable of detecting PCR products with at least 102 cells of *Brucella*, which is much lower than the number of cells that could be detected by gel electrophoresis or ELISA. In contrast, when lysate from 10<sup>4</sup> cells of *P. aeruginosa*, *E. coli*, or *S.* Typhimurium were tested by the NOFS assay following PCR, less than 0.2% attenuation of transmission was obtained for any of the non-*Brucella* species tested (Figure 5).

Reaction of the ISAM/IS*<sup>711</sup>* probe duplex containing streptavidin with lysate representing 4 <sup>×</sup> 102 or 4 <sup>×</sup> <sup>10</sup><sup>4</sup> cells of heat-killed *B. abortus* without the use of PCR resulted in 4.3% and 14.5% transmission attenuation, respectively. Reaction of the same duplex lysate representing 5 <sup>×</sup> <sup>10</sup><sup>4</sup> cells of heat-killed *E. coli* failed to produce a positive transmission attenuation (Figure 6). These results confirmed that the assay could specifically detect low numbers of *Brucella* without the use of PCR.

**Figure 5.** Detection of *Brucella* DNA amplified by PCR from *B. suis*, *B. abortus*, and *B. melitensis* by nanomaterial optical fiber biosensors (NOFS) assay. Each experiment consisted of 3 sequential steps: The biosensor was first tested with sample amplified by PCR from a lysate containing 10<sup>4</sup> cells of a negative control strain (*E. coli*, *Salmonella*, or *P. aeruginosa*), followed by an amplified sample of lysate from 10<sup>4</sup> *Brucella* cells, then lysate from an amplified *Brucella* culture containing 102 cells.

**Figure 6.** Detection of *B. abortus* DNA from lysates of killed cells by NOFS assay without PCR amplification. This experiment consisted of 3 sequential steps: The biosensor was first tested with lysate representing 5 <sup>×</sup> 10<sup>4</sup> cells of a negative control strain (*E. coli*), followed by lysate containing <sup>4</sup> <sup>×</sup> 104 cells of *B. abortus*, followed by lysate containing 4 <sup>×</sup> 102 cells of *B. abortus*.

When 10 replicates of each of the *Brucella* nomenspecies above were tested with lysates containing 10<sup>2</sup> cells/reaction by NOFS with streptavidin and without PCR, all were positive for attenuation of light transmission and significantly greater in comparison to three different non-*Brucella* species tested in duplicate as negative controls (*p* ≤ 0.0004, pooled averages). The average light attenuation for *B. abortus* was 3.81% ± 0.92%, for *B. suis* was 3.50% ± 1.15%, and for *B. melitensis* was 5.15% ± 1.63% (all above the respective cutoff value for a positive result). Light attenuation results using lysates containing 10<sup>4</sup> cells/reaction of the negative control species *E. coli*, *P. aeruginosa,* and *Salmonella* were 0.41% ± 1.28%, 0.93% ± 1.68%, and 1.04% ± 0.89%. These results could be obtained in 30 min and confirmed that the NOFS assay was a highly sensitive, specific, and rapid assay for the detection of *Brucella* DNA.

#### *3.4. NOFS Assay to Detect and Distinguish Di*ff*erent Brucella Nomenspecies and to Distingusih Brucella from Non-Brucella Bacterial Types*

When the ISAM/probe BruAb2\_0168 complex (specific for *B. abortus* but not for other *Brucella* nomenspecies) was reacted directly with lysate containing 105 heat-killed cells of *B. abortus* strain 2308 (without PCR), light transmission was attenuated by 5.4%. However, when the same ISAM/probe complex was reacted with a similar number of *B. melitensis* 16 M cells, transmission was attenuated only by 0.2%. In a separate assay, when the ISAM/probe Suis\_*TraJ* complex (specific for *B. suis*) was reacted with lysate containing 105 cells of *B. suis* strain 1330, light transmission was attenuated by 3.8%. However, when the same ISAM/probe complex was reacted with lysate containing 105 cells of *B. abortus*, no positive attenuation of transmission was observed. When the ISAM/probe IS*711* complex was reacted with lysate representing 105 cells of 15 non-*Brucella* bacterial samples (Table 3), less than 2.2% light transmission attenuation was observed (all below the respective positive cutoff value). Thus, the NOFS assay was specific for *Brucella* and could detect and distinguish different *Brucella* nomenspecies.

**Table 3.** Percent NOFS light attenuation from non-*Brucella* bacterial types.


*3.5. Identification of Brucella in Tissues from Infected Mice by NOFS Assay*

When the ISAM/probe BruAb2\_0168 complex (specific for *B. abortus*) was reacted with 2 spleen or 2 liver extracts from *B. abortus-*infected mice, light transmission was attenuated by 6.79% ± 0.34% and 3.38% ± 0.78%, respectively (positive values were above 3.2% light attenuation for these assays). The average bacterial loads in the spleen and liver extracts used in the assays were 3.8 <sup>×</sup> <sup>10</sup><sup>4</sup> and 4 <sup>×</sup> 103 cells, respectively. However, when the same ISAM/probe complex was reacted with 2 spleen or 2 liver extracts from control mice inoculated with PBS, there was no positive attenuation of transmission for any of the samples (mean attenuation was −1.47% and −1.78%, respectively). Therefore, the NOFS assay could detect *B. abortus* in infected mouse spleen and liver. When the ISAM/probe Melitensis\_0466 (specific for *B. melitensis*) was reacted with 2 spleen or 2 liver extracts from *B. melitensis-*infected mice, light transmission was attenuated by 7.6% and 6.1%, respectively. However, when the same ISAM/probe complex was reacted with 2 spleen or 2 liver extracts from PBS-injected mice, positive attenuation of transmission was not seen. Similarly, when the ISAM/probe Suis\_TraJ complex (specific for *B. suis*) was reacted with 2 spleen or 2 liver extracts from *B. suis-*infected mice, light transmission was attenuated by 6.9% and 5.1%, respectively, but light transmission was less than 1% when the probe complex was reacted with 2 spleen or 2 liver extracts from PBS-injected mice (Table 4).



<sup>a</sup> All samples included streptavidin as a linker in the NOFS assay. <sup>b</sup> The positive cutoff value for this assay was 3.2%.

#### **4. Discussion**

Biosensors are becoming established diagnostic modalities and, when combined with PCR, have been used for detection of DNA using impedance spectroscopy [22,23] and piezoelectric gold electrode [24]. However, such biosensors are very expensive (i.e., may cost over \$10,000 apiece), require high maintenance by experienced personnel, and have the additional PCR step. Therefore, these assays may also not be practical for many laboratories. Optical transduction methods such as surface plasmon resonance (SPR) are rapid and sensitive devices that have been developed for detection of bacterial agents [25]. However, these assays require the use of LED and spectroscopy to generate excited light and receive a signal. SPR sensors are also expensive and require highly trained personnel. Unlike other published biosensors, the NOFS assay described here utilizes nanometer-thick layers that can include a variety of materials, such as DNA, antibodies, and antigens. TAP-LPGs that are coupled to a DNA probe specific to the bacterium and supplemented with additional layers of biotin-streptavidin further enhance the limit of detection of the assay. As a result, PCR was not required for adequate detection of DNA with this NOFS assay. When the target DNA binds to the complementary oligonucleotide probe, thus altering the thickness of the film, the refractive index is also changed. As a result, the transmission characteristics of the fiber are modified, resulting in attenuation of the percent light transmitted. Due to the high specificity of the compatible DNA probe and target, specific DNA can be detected.

DNA probe assays have previously been used to identify the common nomenspecies of *Brucella* [26]. Specific primers have been used with DNA amplification to successfully differentiate *B. abortus* biovars 1, 2, and 4, *B. melitensis*, *B. suis* biovar 1, and *B. ovis*. [27,28]. Several investigators have shown that by targeting highly conserved genes (i.e., 16S rRNA [29], 16S-23S intergenic spacer regions [30], *bcsp*31 or IS*711* for all *Brucella* species [31,32], *alk*B for *B. abortus*, and BMEI1162 for *B. melitensis*[27,33]), probes and primers can be developed for direct detection of these agents. In this communication, DNA amplification was used to confirm that the IS*711* DNA region was specific for all three nomenspecies of *Brucella*, and an ELISA was used to demonstrate that the oligonucleotide probe specifically binds to the IS*711* region of *Brucella.* The NOFS assays, which included a biotin-streptavidin linker, were used to detect as few as 100 cells of *Brucella* with a high degree of sensitivity and specificity in the set of samples studied here, even without prior PCR amplification. The NOFS assay was also capable of detecting *Brucella* in the tissues of infected mice. The probes to BruAb2\_0168, Melitensis\_0466, and Suis\_TraJ DNA regions were specific for *B. abortus*, *B. melitensis*, and *B. suis*, respectively, as determined by NOFS. Therefore,

these designed oligonucleotide probes could be used to distinguish each of the respective *Brucella* nomenspecies from each other or heterologous bacteria. Major advantages of the NOFS assay were that it could be completed in less than 1 h, did not require particular expertise to perform, and did not require a large amount of bench space.

The detection of antibodies to the lipopolysaccharide (LPS) O-antigen by serological methods is the accepted diagnostic method for brucellosis in all hosts. However, specificity can be problematic due to the structural similarity of the O-antigen side chain of *Brucella* with that of other bacteria, particularly *Yersinia enterocolitica* O:9, *Vibrio cholerae*, and *E. coli* 0:157. At this time, no other antigens have been identified that can successfully replace the LPS O-antigen in diagnostic assays. Molecular diagnostic tests are now important methods in clinical microbiology, although they remain restricted to larger laboratories that have the funds, expertise, and equipment to utilize this technology. Real-time (q)PCR assays can detect the DNA of infectious disease agents with same day results [34–36]. However, qPCR technology is restrictive due to the large cost of equipment and expertise needed to carry out these assays. Therefore, qPCR is normally not available in medical settings that have or utilize small laboratories, particularly in rural communities where infections due to *Brucella* may be more prevalent. Furthermore, *Brucella* can be exceptionally difficult to detect in blood, and although isolation from animal tissues may be more productive, as we show here by detection of *Brucella* by NOFS from mouse tissues, such isolation is normally not practical with human tissues. Nonetheless, the NOFS assay can be modified to detect antibodies to *Brucella* by coupling the antigen to the fiber, rather than a DNA probe, or alternatively coupling antibodies to the fiber to detect a specific antigen. We have described such an assay using antibody-coupled sensors to detect methicillin-resistant *Staphylococcus aureus* [21] and *Francisella tularensis* [37].

The most prominent reservoir of *B. abortus*in the United States in in bison and elk in the Yellowstone National Park area [7]. Cattle farmers are particularly concerned that bison or elk that wander onto their farmlands may be infected with *Brucella* and transmit the agent to their cattle, resulting in their loss of *Brucella*-free status. The NOFS assay has the advantage that samples collected from anesthetized animals or aborted fetuses could be used in a small regional facility to rapidly detect the presence of *B. abortus*. In addition, *B. suis* is the most prevalent *Brucella* nomenspecies in the United States and is present in feral hogs in 14 U.S. states [38]. *B. suis* can be transmitted to humans through hunting (field dressing and butchering) or other close contact [39]. Although *Brucella* diagnosis in humans can be difficult due to non-specific flu-like symptoms, detection of the agent in the animal's tissues can strongly support the diagnosis. Therefore, with the correct primers and probes, this NOFS assay can be adapted to detect any of the *Brucella* nomenspecies.

The NOFS assay described here is at the proof-of-principle stage. Further work will be required to develop a diagnostic test ready for regulatory approval. Such work will require a large number of *Brucella* strains of the different nomenspecies, as well as additional negative control strains, to adequately determine the sensitivity and specificity of the assay. Although the limit of detection of the assay has been determined (about 100 cells/mL), due to the low number of strains of *Brucella* available to us, sensitivity (defined as the true positive rate divided by false positives) was not able to be accurately calculated. Although the specificity of the assay appeared to be 100%, additional control strains encompassing a wide variety of species would need to be tested to confirm this. In addition, the NOFS assay can easily be modified to detect antigens, antibodies, or DNA from a wide variety of infectious agents, including viruses, fungi, and parasites, as well as other bacteria. In addition, the assay could be used to detect DNA encoding for antibiotic resistance genes to aid in screening patients that may be colonized with bacteria carrying specific antibiotic resistance genes. Such an assay would be highly beneficial in hospital infection control situations, particularly with an assay that can be completed in a short period of time for reasonable cost, and without the need for highly trained personnel.

**Author Contributions:** Project administration: T.J.I. and J.R.H.; investigation, A.B.B., Z.Z., and K.M.; formal analysis: A.B.B., Z.Z., and K.M.; writing—original draft preparation, A.B.B. and K.M.; writing—review and editing, T.J.I., J.R.H., A.B.B., and K.M.

**Funding:** This study was funded, in part, by USDA-NIFA award 2010-37610-30877.

**Acknowledgments:** We thank Ben Fox for excellent technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Biosensors* Editorial Office E-mail: biosensors@mdpi.com www.mdpi.com/journal/biosensors

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18