Fluorescent Nanoparticle-Based Indirect Immunofluorescence Microscopy for Detection of Mycobacterium tuberculosis

A method of fluorescent nanoparticle-based indirect immunofluorescence microscopy (FNP-IIFM) was developed for the rapid detection of Mycobacterium tuberculosis. An anti-Mycobacterium tuberculosis antibody was used as primary antibody to recognize Mycobacterium tuberculosis, and then an antibody binding protein (Protein A) labeled with Tris(2,2-bipyridyl)dichlororuthenium(II) hexahydrate (RuBpy)-doped silica nanoparticles was used to generate fluorescent signal for microscopic examination. Prior to the detection, Protein A was immobilized on RuBpy-doped silica nanoparticles with a coverage of ∼5.1×102 molecules/nanoparticle. With this method, Mycobacterium tuberculosis in bacterial mixture as well as in spiked sputum was detected. The use of the fluorescent nanoparticles reveals amplified signal intensity and higher photostability than the direct use of conventional fluorescent dye as label. Our preliminary studies have demonstrated the potential application of the FNP-IIFM method for rapid detection of Mycobacterium tuberculosis in clinical samples.


INTRODUCTION
Tuberculosis (TB) is a global public health emergency, fueled by the spread of human immunodeficiency virus (HIV)/Acquired Immune Deficiency Syndrome (AIDS) and the emergence of drug-resistant stains of Mycobacterium tuberculosis (M. tuberculosis). Approximately 2 billion peopleone third of the human population-are currently infected with TB, with one new infection occurring every second. Each year there are more than 8.8 million cases and close to 2 million deaths attributed to TB worldwide. Experts at the World Health Organization (WHO) predicted these numbers would escalate in coming decades, nearly 1 billion people would become newly infected, over 150 million would become sick, and 36 million would die worldwide between now and 2020-if control was not further strengthened [1]. Rapid and accurate diagnosis of tuberculosis is a critical step in the management and control of TB. For decades, diagnosis has largely relied on acid-fast staining and culture of bacilli. However, the sensitivity of acid-fast staining is poor, and culture is a relatively time-consuming process. Many ef-forts have been directed toward developing techniques for rapid diagnosis of tuberculosis with higher sensitivity and reliability [2], including methods based on molecular biology (molecular diagnosis techniques) [3], such as nucleic acid amplification tests (NAA tests) [4,5], DNA probes [6,7]; and methods based on immunology (serodiagnosis techniques) [8], such as enzyme-linked immunosorbent assay (ELISA) [9,10], immunochromatographic assay [11], latex agglutination assay [12]. Recently, more simple, direct, and visually detectable assays have been developed for rapid diagnosis of TB with Au nanoparticles [13,14]. These approaches have contributed much on the improvement of sensitivity and accuracy of the detection but still exhibit deficiencies in some extent [15]. NAA tests have been the subject of a number of investigations. Many commercial kits are available including the Amplicor and MTD tests which are currently US FDA approved. The NAA tests have high specificity and work better to rule-in TB. However, sensitivity of NAA tests is lower and it is less good to rule-out TB. Serological tests for the diagnosis of tuberculosis have been attempted for decades. Dozens of commercial kits are available, most of which are focused on antibody detection. However, assays based on antibodies detection are hard to distinguish active TB from BCG vaccination and past infection. Therefore, more studies are needed to develop and improve the detection methods for tuberculosis.
Dye-doped silica nanoparticles [16,17], exhibiting such important advantages as high luminescence and photostability compared to conventional fluorescent dyes, have been widely applied in biological imaging and ultrasensitive bioanalyses, including cell staining [18], DNA detection [19,20], cell surface receptor targeting [21][22][23][24], and ultrasensitive detection of Escherichia coli O157:H7 [25]. Owing to the dye-encapsulated structure, thousands of dye molecules embedded in one nanoparticle contribute to the luminescence of one particle, causing significant signal amplification. In this paper, we establish a rapid immunological method for detection of M. tuberculosis by combining highly luminescent RuBpy-doped nanoparticles with indirect immunofluorescence microscopy. Since direct anchoring of antibodies onto solid supports via covalence methods is always faced with the loss of activity of the antibodies, Protein A was applied as an affinitive adsorber. In order to obtain full antibody activity, M. tuberculosis was first recognized with the specific antibody in solution then signaled by Protein A functionalized fluorescent nanoparticles. This method was used to detect M. tuberculosis in mixed bacterial samples and spiked sputum samples. Meanwhile, signal intensity and photostability of the method were compared with conventional fluorescent dye fluorescein isothiocyanate labeling method.

Bacteria
The H37Ra strain of M. tuberculosis was obtained from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). M. tuberculosis was cultured by Dr. Songlin Yi (Hunan Tuberculosis Hospital, Hunan, China) on modified Lowenstein-Jenson medium at 37 • C for 3-4 weeks to obtain pure bacterial culture for use in establishing detection method. M. tuberculosis was harvested in pH 7.4, 0.01 M phosphate buffered saline (PBS) to form predominantly single-cell suspension using previously described method [26]. E. coli strain DH5α (Microbial Culture Collection Center of Guangdong Institute of Microbiology, Guangdong, China) was grown overnight in Luria-Bertani broth at 37 • C. The bacterial suspensions were counted in a Petroff-Hausser chamber, and the concentrations of bacteria were adjusted for use in experiments.

Instrumentation
The morphology and uniformity of RuBpy-doped silica nanoparticles were measured with an atomic force microscope (AFM) SPI3800N-SPA400 (Seiko). Size distribution analysis of RuBpy-doped silica nanoparticles was determined at 25 • C by dynamic light scattering (DLS) using Zetasizer 3000HS A (Malvern). The volume-weighted average diameter obtained by the manufacturer's software was used for the calculation of the average nanoparticle volume. A refractive index of 1.47 was used for nanoparticles (the refractive index of silica). Viscosity was determined at 30 • C using a cone plate digital viscometer LVDV-III+CP (Brookfield). Determination of protein concentration according to the Bradford method was done with a UV-Vis spectrophotometer DU-800 (Beckman) [28].

Biological modification of the RuBpy-doped silica nanoparticles
RuBpy-doped silica nanoparticles were prepared using the water-in-oil (W/O) microemulsion method that had been described before [21]. In order to immobilize Protein A onto the nanoparticles, the surface of the RuBpy-doped silica nanoparticles was first activated with CNBr. Nanoparticles (11.2 mg) were suspended in 2 ml of 2 M sodium carbonate solution by ultrasonication. A solution of CNBr in acetonitrile (0.78 g of CNBr dissolved in 2 ml of acetonitrile) was then added dropwise to the particle suspension under stirring at room temperature for 5 minutes. After the activation reaction, the particles were washed twice with ice-cold water and twice with pH 7.

Indirect immunofluorescence detection of M. tuberculosis with bioconjugated nanoparticles
Rabbit anti-M. tuberculosis antibody was added to a 500 μl suspension of M. tuberculosis in PBS (antibody final concentration: 5 μg/ml) and incubated at 37 • C for 1 hour. The suspension was subsequently washed with PBS twice. Nanoparticle-Protein A conjugates (0.1 mg/ml) were then added, and the mixture was incubated at 37 • C for 1 hour.
To remove the free nanoparticle-Protein A conjugates that did not bind to the bacteria, the mixture was centrifuged at 8000 rpm for 2 minutes, and then the supernatant was discard. The pellet was washed twice again. Smear slide was made by spreading the pellet on glass slide and observed with fluorescence microscopy or confocal microscopy. For controls, the rabbit anti-p53 antibody or PBS only was substituted for the primary antibody. Another bacterium E. coli DH5α was treated with the same strategy to test the crossreaction with bioconjugated nanoparticles. For immunofluorescence detection of M. tuberculosis with FITC-labeled antibody, the FITC-conjugated rabbit anti-M. tuberculosis antibody was added to a 500 μl suspension of M. tuberculosis in PBS (antibody final concentration: 25 μg/ml) and the mixture was incubated at 37 • C for 1 hour. The suspension was subsequently washed with PBS for three times and then spread on glass slide for microscopic examination.

Preparation of mixed bacterial sample
The mixed bacterial sample was prepared by mixing FITClabeled E. coli and unlabeled M. tuberculosis. The FITClabeled E. coli was first obtained according to the following method. E. coli was incubated at a concentration of 10 9 cells/ml with 0.5 mg of FITC in 0.1 M Na 2 CO 3 -NaHCO 3 buffer (pH 9.2) at 37 • C for 2 hours in the dark. The E. coli was then washed for three times with PBS to remove free FITC and resuspended in PBS. A 500 μl of mixed bacterial sample was prepared by easily mixing 1.8×10 6 cells/ml FITClabeled E. coli and 3.6×10 5 cells/ml unlabeled M. tuberculosis. The mixture was detected with the FNP-IIFM method.

Preparation of spiked sputum sample
Sputum (2 ml) from healthy individual was collected and equally divided into two portions. One portion was spiked with M. tuberculosis, whereas the other portion was used as the unspiked sample. Then samples were liquefied with the NALC-NaOH method. In brief, the samples were mixed with equal volumes of NALC-NaOH solution (2% NaOH, 1.45% Na-citrate, and 0.5% NALC), shaken vigorously for digestion, and the mixtures were allowed to stand for 15 minutes at room temperature. Then the samples were diluted with 8 ml of water. To remove big agglomerates in the sputum, the mixtures were centrifuged at 1000 rpm for 2 minutes. The precipitates were disposed and the supernatants were centrifuged at 4000 g for 15 minutes. After the supernatant fluids were carefully decanted, the sediments were resuspended in 10 ml of PBS and centrifuged again at 4000 g for 15 minutes. The supernatants were discarded. The resulting pellets were suspended in 500 μl of PBS and detected with the FNP-IIFM method.

Microscopy imaging
An inverted fluorescence microscope ECLIPSE TE300 (Nikon) equipped with a 100 W mercury lamp, a filter block (consisting of a 450-490 nm bandpass excitation and a 515 nm longpass emission filter), and a color CCD (Digital Camera DXM1200, Nikon) was used for common smear microscopic examination. Confocal microscopy was performed on an inverted Olympus IX70 microscope with an argon/krypton laser emitting at 488 nm to excite both RuBpy-doped nanoparticles and FITC fluorescence. We used a dichroic beam splitter (DCB) around 560 nm, together with either a longpass (LP) 560 nm filter for RuBpy-doped nanoparticles signal or an LP 505 nm filter for FITC signal. The RuBpy-doped nanoparticles signal was displayed in the pseudocolor red and the FITC signal in green. To study the differentiation between M. tuberculosis and E. coli in mixed bacterial samples with the FNP-IIFM method, the smears were scanned by sequential excitation mode. In brief, an argon/krypton laser emitting at 488 nm and a helium/neon laser emitting at 543 nm were used to excite FITC and RuBpy-doped silica nanoparticles fluorescence, respectively. We used a DCB around 560 nm, together with the following emission filter: either a bandpass (BP) 505-525 nm when the argon/krypton laser (FITC signal) was used or an LP 560 nm when the helium/neon laser (RuBpy-doped silica nanoparticles signal) was used. A ×60 objective (Olympus PlanApo NA 1.4 oil) was used for routine studies. Pixel format was 512 × 512.

Highly luminescent and photostable fluorescent nanoparticles
We used an easy and efficient water-in-oil microemulsion method to synthesize RuBpy-doped silica nanoparticles. The obtained nanoparticles were uniform and well dispersed as shown in the AFM image (Figure 1(a)). Dynamic light scattering (DLS) measurements for the nanoparticles showed that the size distribution of RuBpy-doped nanoparticles was narrow and the volume-weighted mean hydrodynamic diameter determined was 63.8 nm (Figure 1(b)). For the structure of dye-doped silica nanoparticles, dye molecules are trapped inside the silica matrix, which endows the nanoparticles with two important merits. For one thing, the fluorescence emitted by one nanoparticle is contributed by thousands of dye molecules embedded in the silica matrix. So it is easy to see that one dye-doped nanoparticle is much more luminescent than one dye molecule, which is called the significant signal amplification effect. This attribute makes the dye-doped nanoparticles be advantageous  in improving detection sensitivity in many aspects and very suitable for detection of bacteria with higher sensitivity. As another advantage, due to the protective function of the silica matrix, the nanoparticles are much more photostable than ordinary dye molecules. As shown in Figure 1(c), after continuous intensive illumination with a laser source for 80 seconds, the fluorescence intensities of both RuBpy and FITC dyes were decreased to below 20%, while the fluorescence intensiy of RuBpy-doped nanoparticles remained above 80%.

Covalent immobilization of Protein A on nanoparticles
Covalent attachment of antibodies directly to solid supports via glutaraldehyde, carbodiimide, succinimide ester, and so forth is always found with the loss of biological activity of the antibodies. One of the main reasons for such reduction is attributed to the random orientation of the asymmetric macromolecules on support surface [30]. Several approaches for achieving oriented antibody coupling for good steric accessibilities of active binding sites and increased stability have been developed, including the use of Protein A or Protein G [31], chemical or enzymatic oxidation of the immunoglobulin (IgG) carbohydrate moiety [32], and the use of biotinavidin or streptavidin techniques [33]. Protein A, a highly stable 42 kDa coat protein extracted from Staphylococcus aureus, is capable of binding to the Fc portion of immunoglobulins, especially IgGs, from a large number of species [34]. In our scheme, Protein A was used as an affinitive adsorber to avoid direct attachment of antibody to nanoparticles. For immobilization of Protein A on the RuBpy-doped silica nanoparticles, the CNBr method was used to activate the surface of silica nanoparticles and then couple the Protein A. The surface coverage of Protein A on the nanoparticles was quantified by the Bradford method, and the average mass of one particle was determined through the viscosity/light scattering method, then the number of Protein A molecules attached to one particle could be calculated. The amount of Protein A immobilized on nanoparticles was calculated approximately as [29]: Dilan Qin et al. where q is the amount of Protein A immobilized onto a unit mass of the nanoparticles (mg/mg); C i and C t are the concentrations of the Protein A in the initial solution and in the supernatant after the immobilization reaction, respectively (mg/ml); V is the volume of the aqueous phase (ml); and m is the mass of the nanoparticles (mg). C i and C t were determined by the Bradford method [28]. The amount of Protein A immobilized on nanoparticles calculated according to (1) in our experiment was ∼0.41 mg/mg. The average mass of one particle was then determined and calculated as where m i is the average mass of one nanoparticle (mg); C is the concentration of the nanoparticle suspension (mg/ml); N is the number of nanoparticles in a unit volume of suspension liquid (particles/ml), which was calculated through the viscosity/light scattering method [35] as where 4/3π(d/2) 3 is the average volume of a nanoparticle; d is the volume-weighted diameter determined by light scattering; and φ is the volume fraction of the particles determined by viscosity and calculated as where h is the viscosity of the nanoparticle suspension; h 0 is the viscosity of the solvent without nanoparticles. According to (2)-(4), the average mass of one nanoparticle calculated was ∼8.8 × 10 −17 g. So there were ∼3.6 × 10 −14 mg Protein A on one particle, that is, ∼5.2 × 10 2 Protein A molecules on one particle. It provided a foundation for optimal binding of the nanoparticle-Protein A conjugates with the antibody in the later process.

Detection of M. tuberculosis in pure culture
A method of fluorescent nanoparticle-based indirect immunofluorescence microscopy (FNP-IIFM) was developed for the rapid detection of Mycobacterium tuberculosis. The principle for this method was illustrated in Figure 2. In this scheme, M. tuberculosis was first recognized by a rabbit anti-M. tuberculosis antibody and then the nanoparticle-Protein A conjugates were used to generate fluorescent signal. To examine the binding of bioconjugated nanoparticles to bacteria, the incubated bacteria were imaged using either fluorescence microscopy or confocal microscopy. Pure M. tuberculosis suspension was first immunodetected with the FNP-IIFM method and the resulting confocal images were shown in Figures 3(a), 3(f). The bacteria displayed a bright fluorescence. This indicated that large quantities of nanoparticles had bound to the M. tuberculosis cells. In order to demonstrate whether the binding of bioconjugated nanoparticles to M. tuberculosis was solely through the antigen-specific targeting pattern or there were other nonspecific interactions between the Protein A-nanoparticle conjugates and other surface molecules of the bacteria, two controls were set in which the primary antibody was substituted with the following: (1) PBS only; (2) a rabbit anti-p53 antibody. No fluorescence was observed to associate with the M. tuberculosis in both controls as shown in Figures 3(b), 3(g) and 3(c), 3(h), suggesting that there was little nonspecific interaction between the Protein A-nanoparticle conjugates and the M. tuberculosis cell wall. These results identify that the bioconjugated nanoparticles bind to M. tuberculosis through the antibody-mediated antigen binding pattern. Another bacterium E. coli DH5α was also tested with the FNP-IIFM method. No labeling of the bacteria with the nanoparticle bioconjugates was observed as shown in Figures 3(d), 3(i). The result shows that the anti-M. tuberculosis antibody does not cross-react with E. coli DH5α, and the nanoparticle bioconjugates do not attach to E. coli DH5α nonspecifically, which indicates that the FNP-IIFM method can be used to detect Mycobacterium tuberculosis in pure culture. The fluorescence enhancement capability of the bioconjugated nanoparticles label in the FNP-IIFM method has also been investigated. The detection of M. tuberculosis with bioconjugated RuBpy-doped nanoparticles was compared with the commercial FITC conjugated rabbit anti-M. tuberculosis antibody. The final antibody concentration used in the FITC method was 25 μg/ml. It was 5-fold higher than that used in the FNP-IIFM method. We used higher concentration of antibody in the FITC method because the induced fluorescence signal was too low when the antibody concentration was 5 μg/ml. Figures 3(e), 3(j) showed the confocal images of M. tuberculosis recognized by the FITC method. The fluorescence signal from the bacteria recognized with the FITC method ( Figure 3(j)) was much weaker than the signal with the FNP-IIFM method (Figure 3(f)). Although the primary antibody used in the FNP-IIFM method was only one fifth of that used in the FITC method, the average fluorescence intensity of M. tuberculosis recognized with the FNP-IIFM method was determined to be above five times of that with the FITC method. The experiment reveals the signal advantage that the fluorescent nanoparticles possess over conventional fluorescent dye.

Detection of M. tuberculosis in mixed bacterial samples
To evaluate the detection capability of the FNP-IIFM method in complex samples, artificial complex samples consisting of M. tuberculosis and E. coli were used for test. In order to estimate the accuracy of the detection with the FNP-IIFM method in bacterial mixture, E. coli was labeled with FITC to distinguish from M. tuberculosis prior to the detection. Then the FITC-labeled E. coli was mixed with unlabeled M. tuberculosis to constitute the mixed bacterial samples and detected with the FNP-IIFM method. The results obtained with confocal microscopy were shown in Figure 4(a). The image in Figure 4(a)-A showed the FITC fluorescence associated with E. coli in the mixture (pseudocolor green, emission filter: BP 505-525 nm). Figure 4(a)-B showed the fluorescence of the bioconjugated RuBpy-doped nanoparticles which had bound to bacteria (pseudocolor red, emission filter: LP 560 nm). If the nanoparticles also attached to E. coli, the fluorescence would appear yellow in the overlay image (the combination of green plus red). The overlay image in Figure 4(a)-C showed no colocalization of the red fluorescent nanoparticles with E. coli, so the bioconjugated nanoparticles only bound to the M. tuberculosis. Besides, the detection was also observed with the less-expensive fluorescence microscopy. As shown in Figure 4 good. These results indicate that the FNP-IIFM method can be used to detect M. tuberculosis in mixed bacterial samples.
Meanwhile, the photostability of the fluorescent label in the FNP-IIFM method was also investigated. We compared the photostability of RuBpy-doped nanoparticles bound on M. tuberculosis and FITC dyes labeled on E. coli. The fluorescence of FITC was dim after being continuously irradiated for 2 minutes while that of the nanoparticles was still bright, as shown in Figure 5. It is demonstrated that the bioconjugated RuBpy-doped silica nanoparticles used in the FNP-IIFM method possess much better photostability in comparison with the FITC dye label.

Detection of M. tuberculosis in spiked sputum
In order to demonstrate the usefulness of our method for M. tuberculosis detection under clinical condition, M. tuberculosis was spiked into sputum and detected with the FNP-IIFM method. The result was compared with unspiked sputum control to make certain whether the M. tuberculosis could be detected in the sputum. Sputum from healthy individual was collected and equally divided into two portions. One portion was spiked with M. tuberculosis, whereas the other portion was used as the unspiked sample. The spiked sample and unspiked sample were parallelly pretreated and detected by the FNP-IIFM method. For sample pretreatment, we used the NALC-NaOH method to liquefy the sputum. After liquefaction for 15 minutes, the viscosity of the sputum was greatly decreased. However, there were some visible big agglomerates in both the spiked and unspiked sputum which could neither be liquefied nor be brokenup by vigorously vortexing. These big agglomerates caused poor smear quality such as uneven thickness, and had better been removed before immuno-reaction. To remove the big agglomerates, we centrifuged the liquefied sputum samples at low centrifugal speed (1000 rpm, 2 minutes) and disposed the precipitates. The supernatants were detected with the FNP-IIFM method. As we expected, the sputum samples were much complex mixtures containing a great deal of bacteria and impurities shown in Figure 6. In the unspiked sputum sample, no fluorescent bacterium was found as shown in Figure 6(b). It indicates that the bioconjugated nanoparticles have little nonspecific interaction with the sputum components and the oral bacteria. In the spiked sputum sample, we found highly luminescent bacteria in many microscopic fields as shown in Figure 6(a) (the luminescent bacterium indicated by the arrow). By comparing with the unspiked sample, the luminescent bacteria were considered to be M. tuberculosis recognized by the bioconjugated nanoparticles. The high intensity of fluorescence associated with the recognized M. tuberculosis well distinguished the object bacteria from the complex background. The time needed to finish detecting M. Tuberculosis with the FNP-IIFM method in sputum is <4 hours after the receipt of specimen (sample pretreatment: <1 hour, immunoassay and smear examination: <3 hours). This result demonstrates that our FNP-IIFM method is useful for rapid detection of M. tuberculosis in sputum.

CONCLUSIONS
We have developed a new method for the detection of M. tuberculosis using fluorescent nanoparticle-based indirect immunofluorescence microscopy. With this method, M. tuberculosis can be detected in both mixed bacterial samples and sputum samples. Total assay time including sample pretreatment is within 4 hours. Comparing with conventional fluorescent dyes, the use of fluorescent nanoparticles as label in immunofluorescence microscopy offers advantages of higher luminescence and higher photostability. This method can integrate with epifluorescent filter techniques to further shorten the time needed for detection. In addition, by substituting the antibody to suit to other bacteria, this technique has the potential to develop to a universal method for detecting a wide variety of bacteria in biomedical and biotechnological areas.