A Compact and Sensitive Time-Resolved-Optical-Reader for Bioassay Using Low-Energy Excitable and Long-Lived-Fluorescence Nanolabels

Abstract

Time-resolved fluorescence measurement, a technique utilized to improve sensitivity for fluorescence detection, is becoming more accessible due to the recent development of bright nanolabels with long fluorescence lifetimes and low-energy excitation. Currently, instruments taking advantage of this technique for broader applications (in-field or point-of-care testing) are lagging behind in their development. In this work, a low-cost, compact, and sensitive time-resolved-optical-reader was developed aiming to advance such a technique for wider applications.

I. INTRODUCTION

Time-resolved fluorescence measurement, utilizing long-lived-fluorescence nanolabels, is a sensitive biosensing method. In this method, after pulsed excitation, sample autofluorescence will fade out within tens of nanoseconds; however, these nanolabels will continue emitting fluorescence for a longer time for signal recording [1]. This method inherently possesses a high signal/background ratio. In spite of such a merit, its applications in point-of-care or in-field testing are still limited. The reason lies in many current time-resolved nanolabels needing high-energy or high-power excitation, or not offering lifetimes at milli-second levels [2–3]. As a result, sophisticated/expensive instruments are usually needed for the use of these nanolabels [4]. Recently, bright nanolabels with low-energy (visible light) excitation and lifetimes of milli-seconds are in rapid development, and such optical features can significantly simplify the design of time-resolved instruments (e.g., adopting non-UV grade optics or low-speed electronics), lower instrument complexity/cost, and facilitate their wide applications.

In some recent work, on the basis of newly developed nanolabels using Mn:AgZnInS/ZnS nanocrystals (NCs) (excitable at 405 nm, ~ 1.3 ms lifetime, > 40% quantum yield), a low-cost and small instrument for time-resolved measurement was developed, but this instrument was designed for a cuvette with a large sample volume (5 mL) [5]. Moreover, the laser-diode (LD) driver in this device had strict requirements on the operation voltage/current of LDs, making it less flexible to use LDs with different operation-voltages/currents or output laser power levels. Although an optical path was later designed to match with widely-used microplates for small sample volumes, this instrument still needs manual tuning before experiments and its limit of detection (LOD) on nanolabels is high at ~ 350 pg/mL [6]. Further still, the effect of the laser pulse width on the LOD of the instruments was not studied there. In this work, we designed a new time-resolved optical (TRO) reader overcoming the shortcomings of the earlier instruments. The TRO reader adopted a new optical path allowing the reader to match with microplates without manual optical tuning in experiments, and also used a new LD driver. This new LD driver can rapidly switch a constant current to turn LD on/off, is more adaptive for using different LDs (e.g., with different operation conditions or output laser power levels), and can render the reader to easily be modified when different LDs are needed in testing. Through further optimization of the laser pulse width and other timing parameters, the TRO reader achieved a LOD at ~ 21 pg/mL on Mn:AgZnInS/ZnS-NC-based nanolabels. The TRO reader is fully packaged for portability/mobility and should be suitable for many applications.

II. EXPERIMENTAL

A. Materials and Apparatus

Photomultiplier tubes (PMTs) H11706-20 and H10721-20 were from Hamamatsu. LD (L405P20), pre-amplifier (TIA60), aspherical lens (C230TMD-A), and various optic components were from Thorlabs. Excitation filter ZET405/20x, emission filter ET620/60m, and dichroic mirror ZT491rdc-UF1 were from Chroma. NXP LPC4370 MCU and other electronics were from Mouser. Bovine serum albumin (BSA) Fraction V was from MP Biomedicals. Avidin and PBS buffer were from Thermo Fisher. Non-binding microplates were from Corning, and biotin-coated microplates were from G-Biosciences. A stock solution of water-soluble Mn:AgZnInS/ZnS-NC-based nanolabels was prepared using a previous method [7–8]. Oscilloscope Tektronix TDS-2012B 100MHZ and function generator Gwinstek GFG-3015 were used. Non-time-resolved tests were performed using a Perkin Elmer 2030 microplate reader.

B. Testing of the TRO Reader

For the calibration of the TRO reader, nanolabels in a wide concentration range were prepared through series dilution using 1% BSA (1% BSA in DI water). Triplicates of 100 μL nanolabels in each concentration were added to microwells of a nonspecific microplate and read by the TRO reader. For the biotin-avidin bioassay, nanolabels were conjugated with avidin through the EDC-NHS cross linking chemistry [6]. Dilutions of avidin-conjugated nanolabels were prepared, and triplicates of 100 μL each dilution were placed into biotin-coated microwells for 30-min biotin-avidin-binding incubation. After incubation, all reaction wells were washed using PBS buffer and read by both the TRO reader and Perkin Elmer 2030.

III. RESULTS AND DISCUSSION

A. Overview of the Developed TRO Reader

Fig. 1(A) and (B) illustrate the system schematic and the operational timing of the developed TRO reader, respectively. The laser beam from the laser diode (LD) is collimated into a ~ 2-mm wide ellipse across its major axis through an aspherical lens. This parallel beam is then directed by a dichroic mirror into a microwell of a microplate. Nanolabels in the microwell will be excited to emit fluorescence, which will be collected into the PMT to generate a current. This current is further converted into a voltage by the pre-amplifier, which is then sampled by the ADC for data processing. Under operation, the LD is repetitively pulsed by the LD driver under the control of the MCU. Within every repetition cycle, after the LD is turned off, the PMT and ADC are sequentially turned on with certain time delays to collect the decayed fluorescence from nanolabels. Fig. 1(C) presents a full view of the developed TRO reader outlining its major optoelectronic and electronic components. The TRO reader is compact and easy to fit into a suitcase for portability/mobility. The optoelectronics side includes a microplate (sample holder), optical filters, a gated-PMT, and a LD. The electronics side is shown in Fig. 1(D), which consists of a low-noise pre-amplifier and a stack of prototype boards containing the LD driver, a power management board, and an MCU board with a 12-bit ADC and multiple GPIOs.

Fig. 1.

(A) System schematic of the compact time-resolved optical reader. (B) Timing diagram for the operation of the time-resolved optical reader. The decayed fluorescence from nanolabels after laser-off is for signal recording. (C) Full view of the compact time-resolved optical reader which integrates an optical path, a laser-diode (LD), a gated-PMT, and an electronics case. (D) Internal view of the electronics in the electronics case.

B. Laser-Diode Driver and Laser Performance

Fig. 2(A) shows the basic schematic of the LD driver. Considering that the LD is a current-driven device and needs to turn on/off fast, a constant current (40 mA) generated by LM317T and a precise resistor (30.9 Ω, 5 ppm/°C) is switched between the laser-diode leg and the Zener-diode leg through on/off control of MOSFETs in the two branches. N-MOS IRFD220 is used for the current swap control due to its fast switching time (< 22 ns) and low drain-source on-resistance (~ 0.8 Ω). Controlled by the MCU GPIOs, only one MOSFET in the two branches is on at a time. When the MOSFET in the LD leg is on, the constant current is pulled into the LD. During this state, the adjust-terminal voltage of LM317T is dependent on the LD and changes to match its typical operation voltage (5 V). When the MOSFET in the Zener-diode leg is on, the constant current is pushed to the Zener diode. The adjust-terminal voltage is set at the Zener diode breakdown voltage (5.6 V). Such a breakdown voltage is beneath the maximum rated voltage (5.8 V) of the LD and offers overvoltage protection during the current switching. The circuit structure of this driver is adaptive to LDs with different operation currents/voltages simply by changing the precise resistor and Zener diode accordingly.

Fig. 2.

(A) Schematic of the LD driver using a current push-pull approach. (B) System configuration for the laser stability testing. C) Plot of the rising time of the laser-diode under the LD driver. (D) Plot of the falling time of the laser-diode under the LD driver. (E) Laser stability over three hours for different laser pulsing conditions. (F) Photo-stability of 100 ng/mL nanolabels in 1% BSA under continuous laser pulsing.

Fig. 2(B) depicts the testing configuration using a non-gated PMT (H10721-20) to test the switching dynamics of LD and the LD stability over continuous pulsing under the LD driver. Fig. 2(C) and (D) illustrate the oscilloscope outputs of the rising edge and the falling edge of a laser pulse at 0.5 kHz with 80% duty. The rising time and the rising delay time are 215 ns and 290 ns, respectively. The falling time and the falling delay time are 65 ns and 135 ns, respectively. Such time parameters under other pulsing conditions (0.5 kHz with 10% or 40% duty, or 0.33 kHz with 80% duty) were tested and are close to the ones tested under 0.5 kHz using 80% duty. These dynamic time parameters indicate that the LD can be quickly turned on/off for the pulsed excitation to nanolabels. The laser stability in the reader operation is also critical, otherwise measurement errors could arise from the laser instability. To test the laser stability, the LD was pulsed continuously for three hours under different pulsing conditions and its light intensity was measured every thirty minutes. Because each sensing measurement on an individual sample using the TRO reader takes several seconds, the laser stability over three hours of continuous use should be a sufficient amount of time to reliably encompass any experiments in this work. Fig. 2(E) shows the normalized laser stability over three hours under different pulsing conditions (0.5 kHz with 10% or 40% or 80% duty, or 0.33 kHz with 80% duty). Each data point in Fig. 2(E) has a relative standard deviation less than 0.4%. Under each testing condition, the variation of the signals over time is no more than 0.7% relative to the mean of the signals. The laser stability is thus certified. In this TRO reader, 20 mW laser power is focused in a spot of ~ 1.44 mm², which produces a laser power density at ~ 14 mW/mm². Such a high density could photobleach nanolabels causing measurement errors. Triplicate 100 μL of 100 ng/mL nanolabels in 1% BSA were loaded in microwells and subjected to continuous laser pulsing for five minutes in order to test the possible photobleaching effect on nanolabels. Fig 2(F) shows the photostability of nanolabels read every minute under different conditions of continuous laser pulsing. Each data point in this figure has a relative standard deviation in the range from 1% to 7%. Under each testing condition, the variation of the signals over time is no more than 4% relative to the mean of the signals. No noticeable photobleaching was observed for all tested laser pulsing conditions.

C. Time-Resolved Reading and Timing Optimization

Fig. 3(A–C) present the PMT response for DI water, 1% BSA and 1 μg/mL nanolabels in 1% BSA, respectively. In these tests, the PMT gate was reprogrammed to include the laser excitation pulse, so that the fluorescence signal of nanolabel before and after laser-off can be monitored. In Fig. 3(A) and (B), both backgrounds were observed during the laser pulse. This could be the result of autofluorescence from the microwell, BSA or their synergetic effect; however, after laser-off, these backgrounds drop close to zero immediately due to the short lifetime of autofluorescence. Fig 3(C) shows the fluorescence signal of 1 μg/mL nanolabels in 1% BSA before and after laser-off. It is clear that the fluorescence decay of nanolabels was observed after laser-off. Fig. 3(D) illustrates the time-resolved measurements of 0 μg/mL and 1 μg/mL nanolabel in 1% BSA, where the PMT gate was turned on after laser-off. In Fig. 3(D), the signal-to-background ratio (SBR, the ratio of the signal from 1 μg/mL nanolabel in 1% BSA to the background from 1% BSA) for an average of the twenty datapoints after laser-off was calculated to be ~ 21. Using the data in Fig. 3(B) and (C), the SBR for an average of the twenty datapoints before laser-off was calculated to be ~ 1.4. This confirms that the time-resolved measurement has a higher SBR or is more sensitive than the measurement under laser excitation, due to the reduction of background in time-resolved measurement.

Fig. 3.

(A) The response of PMT for DI Water in microwell. (B) The response of PMT for 1% BSA in microwell. (C) The response of PMT for 1 μg/mL nanolabels with 1% BSA in microwell. (D) The response of PMT under a gate signal only turning on PMT during the fluorescence decay of 1 μg/mL nanolabels with 1% BSA in microwell. The inset figure shows the background for 0 μg/mL nanolabels.

For the operation of the TRO reader, four timing parameters should be optimized to achieve a lower LOD. Fig. 4(A) illustrates these four timing parameters: the laser pulse width, the PMT delay (the time between laser-off and PMT-gate-on), the sampling delay (the time between PMT-gate-on and ADC-on), and the sampling window (the time between ADC-on and PMT/ADC-off). In the timing optimization, only one parameter was varied while all other parameters were fixed at their optimized values. To investigate how the varied parameter affects the LOD on nanolabels, calibration measurements (on triplicate series dilutions of 200, 100, 50, 10, 5, 1, 0.5, 0.1, 0.05, 0.01, and 0 ng/mL of nanolabels in 1% BSA) were conducted using the TRO reader and the LOD was calculated from the calibration curve using the 3σ method for further comparison. The optimization is an iterative procedure until all parameters are in their optimal values. Table 1 presents LODs under different timing parameters. For instance, in the testing of laser pulse width, the PMT delay, the sampling delay and the sampling window were kept at their optimal values (50 μs, 5 μs, and 150 μs), respectively, but only the laser pulse width was changed. From Table 1, it can be seen that a longer laser pulse under 0.5 kHz produces a lower LOD. The reason is shown in Fig. 4(B) where a longer laser pulse charges the fluorescence of nanolabels to a higher level and the time-resolved signal is consequently higher. Therefore, a longer pulse would achieve a lower LOD. A longer pulse under 0.33 kHz with 80% duty was also tested, yet it was found that its LOD is very close to the one under 0.5 kHz with 80% duty. The reason also can be found out in Fig. 4(B) – the fluorescence of nanolabels is relatively saturated under these two long pulses and their time-resolved signals right after laser-off are very close to each other. Through the tests on the laser pulse width, 0.5 kHz with 80% duty was selected as the optimal laser pulsing condition. The optimization of other timing parameters was performed in a similar approach. The final set of the optimal timing parameters is 0.5 kHz with 80% duty for laser pulse width, 50 μs for the PMT delay, 5 μs for the sampling delay, and 150 μs for the sampling window. With these optimal parameters, the calibration curves of nanolabels measured using both the TRO reader and Perkin Elmer 2030 are presented in Fig. 4(C). Fig. 4(D) presents the calibration curves using both readers at a lower nanolabel concentration to distinguish the detection capability of two readers. In the testing, both readers used light at 405-nm wavelength for excitation and used the same filter for emission collection. The LOD of the TRO reader on nanolabels is ~ 21 pg/mL, which is more than three orders lower than that of Perkin Elmer 2030 (~ 104 ng/mL).

Fig. 4.

(A) Operation timing of the TRO reader. (B) Effect of the laser-pulse-width on the fluorescence intensity of nanolabels before and after laser-off. (C) Calibration curves of nanolabels using the TRO reader with optimal timing parameters and a standard Perkin-Elmer microplate reader (non-time-resolved) – both curves were normalized from 1 μg/mL. (D) Calibration curves of nanolabels normalized from 10 ng/mL.

Table 1.

Timing-parameter optimization for a lower LOD

Laser Pulse Width	0.5 kHZ 10%	0.5 kHZ 40%	0.5 kHZ 80%	0.33 kHZ 80%
LOD (pg/mL)	~ 779	~ 250	~ 21	~ 22
PMT Delay	50 μs	100 μs	200 μs	-
LOD (pg/ml)	~ 21	~ 142	~ 279	-
Sampling Delay	5 μs	15 μs	25 μs	-
LOD (pg/ml)	~ 21	~ 61	~ 296	-
Sampling Window	150 μs	200 μs	250 μs	-
LOD (pg/ml)	~ 21	~ 371	~ 500	-

D. Bioassay Application of the TRO Reader

A biotin-avidin bioassay was performed to examine the effectiveness of the TRO reader in sensitive detection. Different dilutions of a stock avidin-nanolabel conjugates were applied to bind with biotin molecules coated in microwells. As a reference, non-conjugated nanolabels in the same dilutions were applied to biotin-coated microwells as well. After wash, the microwells were read using both the TRO reader and Perkin Elmer 2030. Fig. 5(A) and (B) present the assay data using two different readers, respectively. Under non-time-resolved reading, no matter which dilution was applied in the assay, the signal of nanolabels bound in microwells cannot be distinguished from the background according to the 3σ method. Comparatively, with the TRO reader, the signals from 10×, 100×, and 1000× dilutions can be distinguished from the background. Although Perkin Elmer 2030 and the TRO reader have different electronics and optics in their setups, the TRO reader as a standalone and compact instrument presents better detection capability in bioassay.

Fig. 5.

(A) Non-time-resolved measurement (Perkin Elmer 2030) on the biotin-avidin assay in a microplate. (B) Time-resolved measurement (the TRO reader) on the same biotin-avidin assay. Assay using non-conjugated nanolabels was performed in the same way as a reference.

IV. CONCLUSION

A fully packaged and compact TRO reader was developed and optimized. It can be coupled with widely-used microplates for measurement without optical tuning before measurements, and it is adaptive to LDs with different operation currents/voltages or output laser-power levels. This reader presents a low LOD (~ 21 pg/mL) in detecting Mn:AgZnInS/ZnS-NC-based nanolabels under optimal timing parameters. Its effectiveness in sensitive detection was presented through comparing its detection performance with that of a standard non-time-resolved microplate reader on a biotin-avidin assay.