PHOTON-UPCONVERSION NANOPARTICLES FOR SINGLE-MOLECULE IMMUNOSENSING OF CANCER BIOMARKERS AND BACTERIA

The recent progress in the field of immunoassays has been driven by introduction of various kinds of nanomaterials. In particular, photon-upconversion nanoparticles (UCNPs) proved to be excellent immunoassay labels due to their ability to emit light of shorter wavelengths under near-infrared excitation (antiStokes emission), which prevents autofluorescence, minimizes light scattering, and thus reduces the optical background interference. These unique photoluminescent properties allow counting of individual biomolecules labeled with UCNPs by conventional wide-field epiluminescence microscopy and enable the development of single-molecule (digital) immunoassays. We have introduced a novel label based on UCNPs conjugated with streptavidin via poly(ethylene glycol) and applied it in a digital upconversion-linked immunosorbent assay (ULISA) for the detection of a cancer biomarker prostate specific antigen (PSA). The digital readout based on counting of individual immunocomplexes improved the sensitivity 16× compared to conventional analog readout and allowed to reach a limit of detection (LOD) of 23 fg·mL−1 (800 aM). Human serum samples were successfully analyzed achieving an excellent correlation with electrochemiluminescence reference method. The conjugates of UCNPs with streptavidin are also suitable for the detection of pathogenic bacterium Melissococcus plutonius, the causative agent of honeybee disease European foulbrood. The ULISA assay provided an LOD of 340 CFU·mL−1 and successfully analyzed real samples of bees, larvae and bottom hive debris. Due to the high reliability and relatively simple detection scheme, the digital ULISA can pave the way for a new generation of digital immunoassays with a strong potential for commercialization.


INTRODUCTION
Due to their high specificity, immunochemical assays are widely useful for detection of various analytes within complex matrices. The most commonly used immunoanalytical method -enzyme-linked immunosorbent assay (ELISA) -relies on signal generation by enzymes, typically horseradish peroxidase [1]. The enzymes provide highly sensitive detection, because single enzyme molecule can transform numerous molecules of substrate. However, enzymes also suffer several disadvantages, most notably low stability [2]. The recent progress in nanotechnology has provided various nanomaterials, which can be used as labels in immunoassays [3]. The detection schemes can be based on: (i) transformation of substrate, similarly to the use of enzymes, or (ii) on direct detection of the label. For the detection based on substrate transformation, various kinds of catalytic nanoparticles (nanozymes) can be employed, typically mimicking the peroxidase activity [4]. The direct detection of the label often exploits fluorescent nanoparticles, including quantum dots [5] and carbon dots [6]. The convenient readout of fluorescence signal is, however, accompanied by a serious disadvantage, the presence of background signal, which is especially important in case of complex biological matrices.
The background signals can be avoided by the use of materials exhibiting anti-Stokes emission, such as photon-upconversion nanoparticles (UCNPs). The typical UCNPs consist of a hexagonal NaYF4 host crystal doped with well-defined amounts of lanthanides, in particular Yb 3+ and Er 3+ [7]. UCNPs strongly reduce autofluorescence and light scattering because they can be excited by near-infrared (NIR) light and emit light of shorter wavelengths [8]. Further advantages of UCNPs include a high photostability and multiple emission bands that can be tuned individually for the multiplexed assay formats [9]. The preparation of bioconjugates of UCNPs, especially with antibodies or streptavidin, allowed the development of upconversion-linked immunosorbent assays (ULISAs). ULISA assays were previously reported for detection of different analytes, including α-fetoprotein [10], cardiac troponin I [11], and sulfaquinoxaline [12]. Recently, we have developed a microscopic approach for detection of single UCNPs and applied it for the detection of cancer biomarker prostate specific antigen (PSA) [13]. The measurement was based on readout of individual sandwich immunocomplexes, consisting of anti-PSA antibody immobilized on the surface of microtiter plate, captured analyte PSA, and conjugate of anti-PSA antibody with UCNP.
With the increasing sensitivity of the readout methods, the minimization of non-specific binding is getting ever higher importance. In this work, we have focused on the preparation of conjugates of UCNPs with streptavidin, based on PEG and coating by bovine serum albumin (BSA), with the focus on high sensitivity and the lowest possible level of non-specific binding. The conjugates were employed for the detection of PSA and bee pathogen Melissococcus plutonius.

Preparation of UCNP-PEG-streptavidin labels
The NaYF4:Yb 3+ ,Er 3+ and NaYF4:Yb 3+ ,Tm 3+ -based UCNPs were synthesized by high-temperature coprecipitation [14]. To prepare alkyne-PEG-neridronate, sodium neridronate was mixed with alkyne-PEG-NHS, the product was dialyzed (MWCO 3.5 kDa) against water and freeze-dried. The surface of the UCNPs was modified by mixing an aqueous UCNP dispersion with alkyne-PEG-neridronate. After stirring for 24 h at room temperature, dialysis (MWCO 14 kDa) was performed. The streptavidin-azide (7 Bioscience) was coupled to the surface of UCNPs via copper-catalyzed click chemistry. The alkyne-PEG-UCNPs were mixed with streptavidin-azide and sodium L-ascorbate and purged with argon to remove oxygen. Afterwards, CuSO4 was added to generate the Cu(I) catalyst in situ. The prepared conjugates were sonicated, dialyzed (MWCO 100 kDa) and stored at 4 °C [15].
For the conventional analog readout, dried microtiter plate was scanned using custom-built upconversion microtiter plate reader (Chameleon, Hidex). Before performing the digital readout, glycerol was added to the dry wells to minimize local heating effects during NIR excitation. A modified epifluorescence microscope (Eclipse Ti-E, Nikon) [13] was used to record 9 wide-field images per well. The total number of UCNPs was counted using software NIS elements (Nikon). For both readout modes, the average and standard deviation were calculated from three wells and analyzed using a four-parameter logistic function [15].

Preparation of UCNP-BSA-streptavidin labels
The NaYF4:Yb 3+ ,Er 3+ UCNPs were first modified by carboxylated silica layer using water-in-oil microemulsion method [16]. Afterwards, streptavidin was conjugated to the UCNPs via BSA layer using copper-free click chemistry. The BSA-alkyne was conjugated to the UCNPs to prepare a click-reactive product. The carboxyl groups of UCNPs were activated by mixture of EDC and NHS. The dispersion was sonicated for 10 min, followed by a short centrifugation and redispersion in solution containing BSA-alkyne. After 90 min of mixing, the UCNP-BSA-alkyne conjugate was purified by centrifugation (1700 g, 5 times).
Fluorescent click-reactive streptavidin-azide was prepared by coupling 5-azidopentanoic acid (APA) and 5(6)carboxyfluorescein (CF) with the streptavidin. Free carboxyl groups of APA and CF were activated using EDC/sulfo-NHS and incubated for 1 h. Then, the APA and CF solutions were mixed with streptavidin and incubated for 4 h. The click-reactive streptavidin-azide was purified by dialysis (MWCO 12.4 kDa).

ULISA for M. plutonius detection
To allow specific detection of M. plutonius, rabbit polyclonal antibody was prepared [18]. The rabbits were immunized by M. plutonius cell wall fraction, followed by a booster antigen injection after 35 days. After 10 days, rabbit serum was collected and immunoglobulin G fraction was purified on a protein G column. To prepare the detection conjugate, antibody was modified by biotin. Biotinamidohexanoic acid Nhydroxysuccinimide ester (NHS-LC-biotin) was added in 15 times molar excess to the anti-M. plutonius antibody. The Ab-biotin conjugate was purified using Amicon Ultra 0.5 mL centrifugal filters 100K (MWCO 100 kDa) and stored at 4 °C.
For the ULISA assay, a high-binding microtiter plate (Microlon, Greiner Bio-One) was coated with anti-M. plutonius antibody in PBS at 4 °C overnight. All following steps were carried out at room temperature, after each incubation, the plate was washed 4 times with washing buffer. The plate was blocked using 5 % powdered milk in PBS, followed by incubation either with standard dilutions of bacteria in assay buffer or spiked in bee, larvae or bottom hive debris extracts diluted in assay buffer. Afterwards, the microtiter plate was incubated with biotin-conjugated anti-M. plutonius antibody and UCNP-BSA-streptavidin conjugate. After drying, the microtiter plate was scanned using up conversion microplate reader equipped with a continuous 980 nm laser (Laserland) and CCD spectroscope QE65 Pro (Ocean Optics) [17].

Digital detection of PSA using UCNP-PEG-streptavidin label
The PEG-based conjugates were employed in the sandwich ULISA for the detection of PSA (Figure 1). After optimization of the assay parameters, the two readout modes were compared. The digital mode provided LOD of 23 fg·mL −1 (800 aM), which is approximately 20 times lower than the LOD of the analog mode (410 fg·mL −1 ). Compared to our previous work on PSA detection using UCNP-SiO2-antibody conjugates (LOD 1.2 pg·mL −1 ) [13], improvement of sensitivity of 50 times was achieved. This can be attributed to two factors: (1) The use of biotinylated antibody, which acts as a flexible linker facilitating the binding of the nanoparticle label. Furthermore, the high affinity of biotin-streptavidin interaction allows decreasing of UCNP label concentration, which has a positive effect on reduction of the non-specific binding. (2) The level of non-specific binding was further reduced by the PEG coating on the surface of the UCNP [15].
The optimized assay was used for the determination of PSA levels in clinical human serum samples. Random serum samples were collected in a hospital and analysed by an electrochemiluminescence immunoassay (Roche Elecsys) as a standard method. The digital ULISA provided recovery rates between 74 and 102 %. The achieved results, especially the low LOD value, demonstrate the potential of digital ULISA for monitoring of patients after radical prostatectomy, where monitoring of low PSA concentrations is required [19].

Detection of M. plutonius using UCNP-BSA-streptavidin label
In order to detect M. plutonius, the causative agent of European foulbrood, rabbit polyclonal antibody was prepared. The antibodies were tested by ELISA assay (LOD 1.4×10 5 CFU·mL −1 ), which later served also as a standard method for comparison of ULISA performance. The silica-coated UCNPs (Figure 2A) were modified by BSA layer with click-reactive BCN, which served for conjugation of azide-modified streptavidin. The successful bioconjugation was confirmed by dynamic light scattering and gel electrophoresis. Afterward, the UCNP-BSA-streptavidin was used as a label in the sandwich assay for M. plutonius. LOD of 340 CFU·mL −1 with a wide working range up to 10 9 CFU·mL −1 was achieved, which is 400 times better than the LOD of the conventional ELISA [17].
The practical impact of the assay was demonstrated on the analysis of real samples of spiked bees, larvae and bottom hive debris. The 10 times diluted real samples provided LODs in the range between 540 CFU·mL −1 and 8.5×10 3 CFU mL −1 (Figure 2B). Since the amount of M. plutonius in infected apiaries with clinical symptoms is typically around 10 5 CFU·mL −1 [20], the ULISA assay is a suitable method for early-stage diagnosis of European foulbrood infections.

CONCLUSIONS
The streptavidin-conjugated UCNPs can serve as highly sensitive labels in immunoassays based on biotinylated antibodies. We have developed a digital ULISA for the detection of PSA, which provided subfemtomolar LOD of 23 fg·mL −1 (800 aM). The assay was successfully employed to analysis of clinical samples and provided excellent correlation with standard electrochemiluminescence method. Furthermore, we have developed ULISA assay for detection of M. plutonius, the causative agent of European foulbrood, with LOD of 340 CFU·mL −1 , and demonstrated its practical applicability on the analysis of real samples of spiked bees, larvae and bottom hive debris.