Rapid onsite detection of bacterial spores of biothreat ...

Appl Microbiol Biotechnol DOI 10.1007/s00253-015-7151-7

METHODS AND PROTOCOLS

Rapid onsite detection of bacterial spores of biothreat importance by paper-based colorimetric method using erbium–pyrocatechol violet complex M. S. Shivakiran 1 & M. Venkataramana 2 & P. V. Lakshmana Rao 2

Received: 31 July 2015 / Revised: 14 October 2015 / Accepted: 5 November 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract Dipicolinic acid (DPA) is an important chemical marker for the detection of bacterial spores. In this study, complexes of lanthanide series elements such as erbium, europium, neodymium, and terbium were prepared with pyrocatechol violet and effectively immobilized the pyrocatechol violet (PV)–metal complex on a filter paper using polyvinyl alcohol. These filter paper strips were employed for the onsite detection of bacterial spores. The test filter papers were evaluated quantitatively with different concentrations of DPA and spores of various bacteria. Among the four lanthanide ions, erbium displayed better sensitivity than the other ions. The limit of detection of this test for DPA was 60 μM and 5 × 106 spores. The effect of other non-spore-forming bacteria and interfering chemicals on the test strips was also evaluated. The non-spore-forming bacteria did not have considerable effect on the test strip whereas chemicals such as EDTA had significant effects on the test results. The present test is rapid and robust, capable of providing timely results for better judgement to save resources on unnecessary decontamination procedures during false alarms.

Keywords Dipicolinic acid . Pyrocatechol violet . Dye displacement . Bacterial spores . Point-of-care detection . Paper-based colorimetric detection

* P. V. Lakshmana Rao [email protected] 1

Center of Excellence, Department of Biotechnology, Vignan’s University, Vadlamudi, Guntur Dist., Andhra Pradesh 522 213, India

2

DRDO-BU Center for Life Sciences, Bharathiar University, Coimbatore, Tamil Nadu 641046, India

Introduction Aftermath anthrax spore attacks in the USA in 2001, enormous research has been focused on development of rapid and accurate detection systems for biothreat agents. During the 2001 anthrax attacks in the USA, the perpetrators had used Bacillus anthracis spores as the threat weapon. Since then, B. anthracis has re-emerged as a major bioterrorism agent because of its high stability, aerosolization, and notable virulence capacity. Prior to that, numerous times, B. anthracis spores had been used as a biological weapon (Mock and Fouet 2001; Riedel 2004). These bacterial spores can be easily generated in bulk with a fair knowledge of microbiology and minimal laboratory equipments. Early detection of the threat agents greatly helps in decision framing and mobilizing suitable response during biological emergencies. Toward this achievement, several methods such as ELISA, PCR, reverse transcription (RT)-PCR, immuno-capture PCR, LAMP, and advance tools such as sensors and fingerprinting techniques have been developed for the rapid detection of spores (Edwards et al. 2006). These tools are definitive but are limited to referral labs as these methods demand some or all of the following needs: expensive reagents, trained personnel, well-equipped laboratory settings, and extensive sample processing prior to the assay. The primary responders who first encounter the pathogen at the field cannot use these methods. Moreover, they may not be technically well qualified to handle diagnostic equipments. Hence, simple, easy-to-interpret, and disposable methods of detecting the spores of biothreat agents are the need of hour. Bacterial spores are created by Bacillus and Clostridium group of bacteria during the time of unfavorable conditions such as lack of nutrition, stress, or artificial spore induction using certain salts. These spores have thick spore coat which protects the inner components from

Appl Microbiol Biotechnol

environmental factors such as high temperature, chemical disinfectants, UV radiation, and high pressure. One of the key metabolites of the spores is calcium dipicolinate (Ca-DPA)/dipicolinic acid (DPA), chemically known as 2, 6-pyridinedicarboxylic acid. DPA comprises 5–15 % of total spore dry weight. It stabilizes the bacterial chromosome and contributes to the spore resistance and its germination. DPA forms tridentate fluorescent complex with terbium (Tb3+), europium (Eu3+), and other metal ions. It is well documented that DPA transfers energy to the bound metal ion which results in long-lived emission in the visible spectrum (Barnes et al. 2011). This formed a basis for developing detection methods for bacterial spores (Rosen et al. 1997; Pellegrino et al. 1998). In later years, Tb3+ became the most commonly used reagent for spore detection. Further, colorimetric and luminometric based detection systems were developed by dye displacement method employing Tb3+ ions (Clear et al. 2013) and Eu3+ ions (Goncalves et al. 2013) for detection of DPA. They first prepared a complex of metal ions with pyrocatechol and betanin, respectively, and then DPA was allowed to replace these ligands competitively. The released dye was read for the results. Although these methods are fairly simple to perform, these have some setbacks, viz., the need to be performed in the laboratory and the interpretation of results through plate readers or spectrophotometers. Moreover, the dye–metal complex has to be prepared fresh for better results. The first responders will not be able to perform these preparations at the site of the test. Hence, simpler and easy-to-do detection systems are needed. In this perspective, the present study deals with the preparation and evaluation of a simple, instant, reagent-free, disposable paper-based colorimetric detection system for presumptive identification of bacterial spores of biothreat relevance with B. anthracis in particular.

Materials and methods Chemicals, reagents, and materials The following chemicals and materials were used in this study. These materials were purchased from Sigma, India: 2, 6-pyridine dicarboxylic acid (DPA), europium chloride (EuCl3), terbium chloride (TbCl3), erbium chloride hexahydrate (ErCl 3 ·6H 2 O), neodymium chloride hexahydrate (NdCl3·6H2O), and pyrocatechol violet (PV). And, the following chemicals were purchased from Himedia Labs, India: HEPES, polyvinyl alcohol (PVA, MW-140,000), filter papers (FPs), nutrient broth, magnesium sulphate, manganese chloride, ferrous sulphate, and calcium nitrate. The bacterial cultures were purchased from National Collection for Industrial Microorganisms (NCIM), Pune, India.

Preparation of metal complexes, reagents, and media The stock solutions of DPA, lanthanide salts, and the dye were prepared in 50 mM HEPES (pH 7.4). The stocks were prepared at 10-mM solutions and stored at room temperature. To prepare the PV–metal (PV-M) complex, we first mixed TbCl3, EuCl3, ErCl3·6H2O, and NdCl3·6H2O solutions with PV at 200, 100, and 50 μM in equimolar ratio. Polyvinyl alcohol was prepared in 50 mM HEPES at concentrations of 0.5, 1, 2, and 4 %. The sporulation media was prepared from nutrient broth and certain salts as mentioned here: The sporulation medium was prepared by adding 3.0 g tryptone, 6.0 g bacteriological peptone, 3.0 g yeast extract, and salts such as 1 mL of Ca (NO3)2 (1 M), MnCl2 (10 mM), and FeSO4 (1 mM) to 1 L of distilled water, and the pH of the media was adjusted between 7.2 and 7.4. The media was autoclaved at 121 °C for 20 min and stored at 4 °C till use. Preparation of bacterial spores The spores were prepared as described elsewhere with suitable modifications to improve the spore yield (Redmond et al. 2004). The bacteria were inoculated in nutrient broth and incubated overnight at 37 °C in shaker. The next day, 250 μL of overnight culture was added to 25 mL of sporulation medium and incubated at 37 °C for 72 h in shaker with 300 rpm. After the incubation, the culture was collected by centrifugation at 10,000 rpm for 10 min at room temperature (RT). The culture pellet was resuspended in 25 mL of chilled sterile distilled water and stored at 4 °C for 30 min. The resuspended spores were centrifuged as above. This washing procedure was repeated ten times. The spores were resuspended in 25 mL of cold sterile water and stored at 4 °C overnight. The next day, another round of wash was given before microscopy. A small volume of spores was smeared on the microscopic slide and observed under phase contrast objectives to determine the purity of spores and efficiency of spore preparation. One milliliter of spore suspension was serially diluted in sterile water for spore count analysis. Stoichiometry studies The stoichiometric relation of Er3+, Eu3+, Nd3+, and Tb3+ with PV was determined using Job’s method of continuous variation (Hirose 2001). To determine the equilibrium maximum of the metal–PV complex, the metal solution and PV solution were mixed in molar ratios (M/PV + M) varying from 1 to 0. Then, the absorbance of the complex at 625 nm was plotted against the metal–PV ratio. Further, the metal–PV complexes were titrated with various concentrations of DPA to assess the displacement of PV from the metal–PV complex. These complexation reactions were prepared in 96-well microtiter plates,


and the absorbance readings were taken in Biotek Hybrid H1 multimode reader. Preparation of paper based colorimetric detection system Whatman no.1 filter paper (FP) discs were soaked in various molar concentrations of metal–PV complex prepared in 50 mM HEPES with 15 % ethylene glycol for 60 min. Various methods were used to fix the complex on to the FP: (a) UVexposure for 5 min at 900 W followed by drying at 50 ° C in hot air oven, (b) hot air drying at 80 °C for 20 min, and (c) the FP discs were dipped in 2 % PVA for 30 min followed by drying at 50 °C in hot air oven. Once the FP discs were dried, they were cut into smaller squares of 1 cm2 and stored in dry and dark place in pouches until further use. Assessment of the sensitivity of the assay The sensitivity of the FP discs prepared with different metal– PV complexes was determined by serial dilution of DPA. DPA was serially diluted in twofold ranging from 500 to 0.7 μM, and 20 μL was added to previously cut FP squares. The change in the color of the FP squares was observed in comparison to the control. The sensitivity of the assay was further evaluated on bacterial spores. For this, we employed the spores of Bacillus subtilis NCIM 2117. The spores were resuspended in sterile distilled water at a concentration of 5 × 1010 mL−1. These spores were serially diluted tenfold in sterile water and autoclaved in pressure cooker for 30 min. After cooling, 20 μL from each sample was added onto FP squares. For control samples, 100 μM DPA was used as positive and water as negative control. The quantitative assessment of the reaction was performed using ImageJ software. Immediately after the assay, the FP strips were photographed using digital camera (Sony, Model-DSC WX300, Japan), and the images were stored in .PNG format. The images were split into RGB colors and the color intensity of red was measured. The intensity measurement was repeated three times for each reaction at three different fields, and the average was considered. The change in the red intensity (ΔR = Rsample − Rblank) was compared across the reactions. Validation of paper based assay The spores were prepared as mentioned above from various bacteria procured from NCIM (Table 1). The spores were tenfold diluted in distilled water serially ranging from 5 × 1010 to 5 × 101 spore mL−1. The non-spore-forming bacteria such as Escherichia coli and lactic acid bacteria were also included in this test to investigate their reactivity with the test strips. Twenty microliters of each diluted sample was dropped on to the FP strips. About 30–60 s of time was given for color development. The intensity of the color was compared with

Table 1 No.

List of bacteria used in this study Name

Source

01

Bacillus cereus NCIM 2155

NCIM, Pune

02

Bacillus cereus NCIM 2158

NCIM, Pune

03 04

Bacillus thuringiensis NCIM 2977 Bacillus thuringiensis NCIM 2978

NCIM, Pune NCIM, Pune

05 06

Bacillus subtilis NCIM 2117 Bacillus brevis NCIM 2532


07

Clostridium perfringens MTCC 450

ATCC, USA

08 09

Clostridium sporogenes NCIM 2559 Clostridium difficile isolate


10 11

E. coli ATCC 10536 Lactic acid bacteria isolate

ATCC, USA curds

known standard, viz., 100 μM DPA. Since the essential principle of this assay is metal complex formation with PV, some chelating compounds can interfere with the reaction assay. Hence, we examined the effects of some routinely used buffers and chemicals in the laboratory such as 1× phosphate-buffered saline (PBS), 50 mM EDTA, 0.9 % NaCl solution, 50 mM carbonate buffer, and 100 mM Tris buffer.

Results Stoichiometric studies of metal–PV complex The absorption spectra of titration of metal solutions against PV showed that the ligand, i.e., PV, undergoes shift in absorption from 440 to 600 nm as the metal ions form complex with the ligand. Among the four metal ions, Er3+ appeared to display the lesser binding affinity as revealed when metal ion molar concentration was plotted against A600/A440 and the highest being Eu3+ ions (Fig. 2a). The complexation studies of lanthanide metal ions with PV by using Job’s continuous variation method revealed varying stoichiometries among the metal ions with PV (Fig. 2b). The displacement experiments were further restricted to erbium–PV complex as Er-PV complex alone displayed 1:1 stoichiometric ratio. The displacement pattern of PV from Er-PV complex is dependent on molar concentration of DPA added to the complex as depicted in the Fig. 2c. As DPA was added to 50 μM Er-PV complex at concentrations ranging from 0 to 100 μM, A600/A440 was reduced exponentially indicating the displacement of PV by DPA from M-PV complex. Preparation of paper based platform Filter papers were selected as the platform for the reaction. The preparation of FP strips is summarized in Figs. 1 and 2.

Appl Microbiol Biotechnol Fig. 1 Schematic representation of preparation of paper based colorimetric detection for spores

PV–metal complexes of various concentrations of 50, 100, and 200 μM were added to FPs. Interestingly, some filter papers such as Whatman Cat. No. 1001-917 and Whatman Cat. No. 1001-325 displayed blue color development immediately after adding PV solution probably due to some transition metal ions present in the filter paper (Fig. 3a). However, the Whatman no.1 filter paper (Cat. No. 1001 090) did not show any such color transformation; hence, further experiments were conducted with this filter paper. After soaking in the PV–metal complex, various methods were used to fix the complex on to the FP. First, we tried two methods: (a) UV exposure for 5 min at 900 W and (b) hot air drying at 80 °C for 20 min. Both of these methods did not retain the PV–metal complex on the FP. The metal–PV complexes did not deposit homogenously onto the FPs resulting in irregular color patches after drying. Therefore, in alternative approach, the

FPs were soaked in various concentrations of polyvinyl alcohol (PVA) after PV–metal complex adsorption on FP (Fig. 3b). It was found that PVA effectively controlled the leaking of the complex from the test filter paper and assisted in homogenous deposition of complex. Treating the FP with PVA before application was assumed to decrease the retention of the reagent. In our study, we found that 2 % of the PVA was found optimum as it was sufficient to prevent the color leach and did not interfere with diffusion of the test sample that is added on to the filter paper. High concentrations of PVA, i.e., 3 % and above, although displayed effectiveness in preventing the color bleach, they proved to be impediment during testing of the suspect sample. Once the FPs were ready for testing, various concentrations of DPA were added to test strips. The displacement of PV from metal–PV complex was not immediate and non-homogenous

Fig. 2 Stoichiometric studies of metal–PV and DPA. a Representative titration points, and computer curve fitting to a 1:1 binding model, for addition of metal to a solution of PV (50 μM in 50 mM HEPES buffer, pH 7.4). b Continuous variation plot showing change in M-PV complex

absorbance at 625 nm. The experiment was repeated three independent times and the average ± S.D. is presented here. c Representative titration points, and computer curve fitting to a 1:1 binding model, for addition of DPA to a solution of M-PV (50 μM in 50 mM HEPES buffer, pH 7.4)


Fig. 3 Treatment of filter papers with PVand M-PV complex by different methods. a The FP strips were dipped in 50 μM PV and then rinsed in 50 mM HEPES. (i) Untreated filter paper, (ii) Whatman Cat. No.-1001 917, (iii) Whatman Cat. No. 1001 325, (iv) Whatman Cat. No. 1001 090.

b The FP discs were soaked in M-PV complex for 60 min and then (i) UV exposure for 5 min at 900 W followed by drying at 50 °C in hot air oven, (ii) hot air drying at 80 °C for 20 min, and (iii) soaking in 2 % PVA for 30 min followed by drying at 50 °C in hot air oven

on filter paper platform. When DPA is loaded onto the FP strips, the DPA diffused gradually with decreasing concentration toward the periphery of the spot and therefore the color displacement was maximum at the center of the spot and weaker at the periphery unlike in the solutions where the sample is diffused uniformly. Even then, the color change in FP strips was definitely evident to the experimenter. However, to quantify the results, each image was analyzed in the ImageJ software at three different spots of each test, and the average is shown in the graphs.

Sensitivity evaluation of different metal–PV complexes with DPA

Fig. 4 Effects of different concentrations of DPA on various metal–dye complex coated FPs. a Strips of FP embedded with 50 mM PV–metal complex were cut into 1-cm2 strips. Twenty microliters of twofold diluted DPA was added to each strip. The change in color intensity was recorded by photography. b Quantitative analysis of change in color intensity (ΔR) of the dye displacement reaction. The images were split into RGB channels using ImageJ software, and the intensity of the image in R channel was measured. The values shown here are the mean ± SD of the R value in triplicates in three independent fields

After the immobilization of the PV–metal complex in the FP, the sensitivity of the test strips was evaluated with various concentrations of the DPA. As Fig. 4a, b shows, all the four lanthanide ions considered in this study produced different color intensity on the FP. Among them, Er-PV complex displayed maximum change in color intensity in comparison to the other ions followed by Tb-PV and the least was Eu-PV


complex. The clarity in the differences of the color intensity was better in Er-PV complex up to 3 μM DPA followed by Tb-PV complex up to 15 μM DPA. Although the quantitative analysis of ΔR reveals that the Er-PV, Nd-PV, and Tb-PV strips have >40 units compared to Eu-PV strips which have 120). It is also reasonable that only the bacterial spores can be effectively employed rather than any other non-sporulating bacteria for biological attacks because spores can survive extreme conditions, need no moisture or nutrients, and easily aerosolized. Moreover, biological weapon agents such as B. anthracis and Clostridium perfringens are spore formers. Therefore, it is unlikely that one encounters non-sporulating bacteria in suspected samples. The reaction principle is based on co-ordination complex formation of metal ions with chelating agents. Any chemical reagent capable of complexing with lanthanide ions, par with DPA or better, can easily interfere with the detection of the analyte in the sample. Since the current report deals with complex colorimetric reactions, there are several possibilities that certain chemicals interfere with the chemical complex. For instance, reports showed that the colored complex formed by DPA with Tb3+ ions can easily be destroyed by other chelating agents which competitively bind to Tb3+ such as EDTA, EGTA, and BAPTA (Barnes et al. 2011). Similarly, we too found that EDTA and PBS displayed colorimetric reaction on

Comparison of limit of detection of lanthanide series element based detection of bacterial spores

No.

Method

Target

LOD

Reference

1

Tb3+-based photoluminescence

B. subtilis

4.4 × 105 CFU mL−1

Rosen et al. (1997)

2

Terbium–pyrocatechol-based dye displacement Eu–betanin-based dye displacement

DPA

∼108 Bacillus spores mL−1

Clear et al. (2013)

B. anthracis and B. cereus

(1.1 ± 0.3) × 106 spores mL−1

Goncalves et al. (2013)

3

3+

5

−1

4

Tb -based luminescence

B. globigii

5

Tb(EDTA)-based luminescence

B. anthracis

6

Delayed gate fluorescence detection using Tb3+–DPA complex Ancillary ligand enhanced anion analyte binding lanthanide-based sensors Erbium-based dye displacement on paper platform

B. subtilis

104 spores mL−1

Hindle and Hall (1999)

DPA

103–104 spores mL−1

Cable et al. (2013)

B. subtilis

106 spores mL−1

This study

7 8

1.21 × 10 spores mL

Pellegrino et al. (1998) Barnes et al. (2011)


FP as they displaced the PV from the colored complex (Fig. 7). To reduce such false negatives, a novel method was described wherein ancillary ligands such as DO2A when bound to lanthanide ions greatly enhance the analyte binding to those metal ions by 1–2 magnitudes and also mitigate interference of environmental or physiological contaminants (Barnes et al. 2011; Cable et al. 2013). Similarly, complexation of PV to metal ion may lead to a change in electron density making the binding site for DPA more electropositive and therefore attracts the anionic DPA molecules. But, in the present assay, the anion should also effectively displace PV which is already bound to metal ion. For this, the metal ion should possess more affinity toward DPA than PV. Figures 2 and 4 show that the PV displacement was greater in Er-PVand Tb-PV complexes. It may be hypothesized that gadolinium break, PV-M stoichiometry, and polarizabilities of metal ions converge to enhance affinity for DPA in Er-PV and Tb-PV complexes and concurrently release PV. Although the LOD of the present assay was one million spores, this test is essentially a preliminary investigation to eliminate unnecessary false or hoax situations during biological emergencies. However, such preliminary tests must be followed by confirmatory laboratory investigations such as PCR (Makam et al. 2013), RT-PCR (Ryu et al. 2003), ELISA (Sastry et al. 2003), conventional microbiological tests, MALDI-TOF (Dybwad et al. 2013), or combination of these tests to identify the threat agent. During biological threat scenario, it will be difficult to investigate a large number of samples by sophisticated confirmatory tests. Therefore, simple tests can be handier for more meaningful investigations in the field conditions. Moreover, the paper-based assay presented here further supports the fabrication of low-cost, reagent-free onsite detection platforms for biodefense applications especially for the resource-poor settings. The present test system can definitely give leads for further improvisation of its sensitivity and to eliminate the effects of interfering agents. Acknowledgments The authors are thankful to Dr. K. Kadirvelu, JointDirector, DRDO-BU Center for Lifesciences for his kind support. Funding This work was completely carried out with the internal funds of the institute. Compliance with ethical standards Conflict of interest All the authors declare that there is no conflict of interest. Ethical statement This work presented here did not involve any human volunteers or animals for any experimentation by any of the authors.

References Barnes LS, Kaneshige KR, Strong JS, Tan K, von Bremen HF, Mogul R (2011) Effects of terbium chelate structure on dipicolinate ligation and the detection of Bacillus spores. J Inorg Biochem 105:1580–1588 Cable ML, Kirby JP, Gray HB, Ponce A (2013) Enhancement of anion binding in lanthanide optical sensors. Acc Chem Res 46(11):2576– 2584 Clear KJ, Stroud S, Smith BD (2013) Dual colorimetric and luminescent assay for dipicolinate, a biomarker of bacterial spores. Analyst 138: 7079–7082 Curiel D, Sanchez G, Mas-Montoya M, Tárragaa A, Molinaa P (2012) Rational design of a fluorescent receptor for the recognition of anthrax dipicolinate. Analyst 137:5499–5501 Dybwad M, Laaken AL, Blatny JM, Paauw A (2013) Rapid identification of bacillus anthracis spores in suspicious powder samples by using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS). Appl Environ Microbiol 79(17): 5372–5383 Edwards KA, Clancy HA, Baeumner AJ (2006) Bacillus anthracis: toxicology, epidemiology and current rapid-detection methods. Anal Bioanal Chem 384:73–84 Goncalves LCP, Da Silva SM, DeRose PC, Ando RA, Bastos EL (2013) Beetroot-pigment-derived colorimetric sensor for detection of calcium dipicolinate in bacterial spores. PLoS One 8(9):e73701 Hindle AA, Hall EAH (1999) Dipicolinic acid (DPA) assay revisited and appraised for spore detection. Analyst 124:1599–1604 Hirose K (2001) A practical guide for the determination of binding constants. J Incl Phenom Macrocycl Chem 39:193–209 Janssen FW, Lund AJ, Anderson LE (1958) Colorimetric assay for dipicolinic acid in bacterial spores. Science 127:26–27 Makam SS, Majumder S, Kingston JJ, Urs RM, Tuteja U, Sripathi MH, Batra HV (2013) Immuno capture PCR for rapid and sensitive identification of pathogenic Bacillus anthracis. World J Microbiol Biotechnol 29(12):2379–2388 Mock M, Fouet A (2001) Anthrax. Ann Rev Microbiol 55:647–671 Pellegrino PM, Fell NF Jr, Rosen DL, Gillespie JB (1998) Bacterial endospore detection using terbium dipicolinate photoluminescence in the presence of chemical and biological materials. Anal Chem 70: 1755–1760 Redmond C, Baillie LWJ, Hibbs S, Moir AJG, Moir A (2004) Identification of proteins in the exosporium of Bacillus anthracis. Microbiology 150:355–363 Riedel S (2004) Biological warfare and bioterrorism: a historical review. BUMC Proc 17:400–406 Rosen DL, Sharpless C, McGown LB (1997) Bacterial spore detection and determination by use of terbium dipicolinate photoluminescence. Anal Chem 69:1082–1085 Ryu C, Lee K, Yoo C, Seong WK, Oh HB (2003) Sensitive and rapid quantitative detection of anthrax spores isolated from soil samples by real-time PCR. Microbiol Immunol 47(10):693–699 Sastry KSR, Tuteja U, Santhosh PK, Lalitha MK, Batra HV (2003) Identification of Bacillus anthracis by a simple protective antigenspecific mAb dot- ELISA. J Med Microbiol 52:47–49 Yamada K, Takaki S, Komuro N, Suzuki K, Citterio D (2014) An antibody-free microfluidic paper-based analytical device for the determination of tear fluid lactoferrin by fluorescence sensitization of Tb3+. Analyst 139:1637–1643