Monitoring Enzymatic Reactions with In Situ Sensors

Please verify that (1) all pages are present, (2) all figures are acceptable, (3) all fonts and special characters are correct, and (4) all text and figures fit within the margin lines shown on this review document. Return to your MySPIE ToDo list and approve or disapprove this submission.

Monitoring Enzymatic Reactions with In Situ Sensors I.T. Young*a, V. Iordanovc, A. Kroonb , H. Dietrichb , R. Moermanb , L.R. van den Doela, G.W.K. van Dedemb , A. Bosschec, B.L. Grayc, L. Sarroc, P.W. Verbeeka, L.J. van Vlieta a

Pattern Recognition Group, Faculty of Applied Sciences, Delft University of Technology, Lorentzweg 1, NL-2628 CJ Delft, The Netherlands b Kluyver Laboratory for Biotechnology, Faculty of Applied Sciences, Delft University of Technology, Julianalaan 67, NL-2628 BC Delft, The Netherlands c Electronic Instrumentation Laboratory, Faculty of Information Technology Systems and DIMES, Delft University of Technology, Mekelweg 4, NL-2628 CD Delft, The Netherlands ABSTRACT

In previous publications and presentations we have described our construction of a laboratory-on-a-chip based on nanoliter capacity wells etched in silicon. We have described methods for dispensing reagents as well as samples, for preventing evaporation, for embedding electronics in each well to measure fluid volume per well in real-time, and for monitoring the production or consumption of NADH in enzyme-catalyzed reactions such as those found in the glycolytic pathway of yeast. In this paper we describe the use of light sensors (photodiodes) in each well to measure both fluorescence (such as that evidenced in NADH) as well as bioluminescence (such as evidenced in ATP assays). We show that our detection limit for NADH fluorescence in 100 µM and for ATP/luciferase bioluminescence is 2.4 µM. Keywords: microarray, quantitative microscopy, photodiodes, fluorescence measurements, bioluminescence measurements, embedded instrumentation

1

INTRODUCTION

The objective of our research program has been to design and build an intelligent analytical system that measures different molecular analytes simultaneously using specific molecular interactions in wells on specially-constructed chips. This system can be used as an aid for decision making in professional and technical environments. The technology we propose is generic and useful in a variety of applications such as quality management in the biotechnology, pharmacology, and food industry, medical diagnostics, and environmental monitoring. The advantage of the proposed technology over existing methods is that large numbers of different chemical and biochemical analyses can be performed simultaneously in a very short time, using minimal amounts of reagents and sample. The large amounts of quantitative data from the measurement system can be mixed with qualitative information from experts in the field and historical data for making decisions in complex environments thereby increasing speed, consistency and accuracy, and decreasing costs. We have recently evaluated the performance of our laboratory-on-a-chip system in the measurement of NADH fluorescence associated with enzyme-catalyzed reactions in the glycolytic pathway of yeast and the bioluminescence associated with luciferase-catalyzed reactions involving ATP.

2

BACKGROUND

We have developed an electro-spray mechanism for dispensing reagents into wells whose volumes range from 60 pl to 500 nl (Figure 1a)1. We have shown that this mechanism is fast, reliable, reproducible, and suitable for commercial exploitation. Further, we have devised techniques for dispensing samples (analytes) in such a way that each well is filled with the same volume of fluid, gas bubbles are eliminated, and evaporation is prevented. Patent applications have been filed for the techniques that we have developed2,3.

*

[email protected]; phone +31-1-15-278-1416; fax +31-15-278-6740; ph.tn.tudelft.nl

SPIE USE, V. 3 4966-15 (p.1 of 7) / Color: No / Format: A4/ AF: A4 / Date: 2003-01-06 02:08:00


Figure 1a: Electro-spray deposition of reagent in a well that is 200 µm ¥ 200 µm and 6 µm deep for a volume of 240 pl. The spray is from above; the shadow of the nozzle is below.

Figure 1b: Filling and covering mechanism for analytes in wells. This procedure guards against cross-contamination and evaporation and eliminates air bubbles.

Because it is important to know the amount of fluid that has been deposited in each and every well, we have developed two independent methods for assessing the fluid volume per well, one based upon the use of in situ electrodes and one based upon the interference fringes observed through a digital imaging microscope system. The former system4,5 is “realtime”; we can follow fluid evaporation and filling (see Figure 2). The latter6 is absolute in that it provides an absolute measurement of volume that is accurate in the axial (z) direction to 20 nm, that is, about 70 water molecules. The slower interference-based (optical) technique can be used as the calibration produce for the significantly faster electrical technique7.

2.60

Voltage (V)

2.40 2.20 2.00 1.80 1.60 1.40 1.20 1.00 160

210

260

310

360

410

460

510

Volume (pl)

Figure 2a: Construction of an in situ liquid volume sensor using aluminum electrodes

Figure 2b: Measured output voltage as a function of the liquid volume in a 540 pl well during well filling

The fluorescence signals have been primarily acquired using a digital imaging microscope system that has been modified to collect all the light from a given well on as few pixels of the digital camera as possible8. In this sense, a photomultiplier system would work as well but our configuration gives us considerably more flexibility. This modification is show in Figure 3. We have monitored the reaction rates derived from cell-free extracts for various enzymatic reactions that are found in the glycolytic pathway of yeast (Saccharomyces cerevisiae). The enzymes are Glucose-6-phosphate dehydrogenase (G6PDH), Lactate dehydrogenase (LDH), and Alcohol dehydrogenase (ADH). These reactions are: G6P-pyruvate + NAD ´ Gluconolactone + NADH

(G6P-DH)

Pyruvate + NADH ´ Lactate + NAD

(LDH)

Ethanol + NAD ´ Acetaldehyde + NADH

(ADH)



Figure 3a: A Zeiss Axioskop using an ordinary 2 0 ¥ J/J0.75 FLUAR objective, an inverted 5 x objective, and a Princeton Versarray CCD camera

Figure 3b: The trinocular tube has been removed and replaced by an inverted Zeiss 5 ¥ J/J0.25 Zeiss FLUAR lens whose working distance is 9.3 mm.

For all three enzymes the measurement is based on the conversion of NAD(P) to NAD(P)H or vice versa and the fluorescence of NADH with an absorption maximum at l = 340 nm and an emission maximum at l = 450 nm. Our detection limit for NAD(P)H is about 5 µM, which corresponds to 3 x 109 molecules in a 1 nl well. Data are automatically incorporated into an analysis module o f our software environment GENLAB (http://www.genlab.tudelft.nl/)9 which runs under MatLab. Figure 4a shows one image of an array of 5 ¥ 5 wells taken at one point in time. Figure 4b shows the analysis of a sequence of images taken over 2000 s to produce a profile per well of the consumption of NADH in the LDH-catalyzed reaction. The rate constant for each well can be determined and used to assess enzyme kinetics. The data shown below are based upon cell-free extracts from Saccharomyces cerevisiae.

Figure 4a: GEN L AB display of measurements in one time frame of the NADH fluorescence from the LDH reaction in 5 ¥ 5 microarray wells

Figure 4b: Results of the LDH reaction in 25 wells. The enzyme concentration LDH is 0.1 U/ml. The data follow a first-order exponential decay reaction model. The italic number in each graph is the decay time t.

The measurements shown above were made using Princeton Versarray 512B CCD camera. This instrument provides 16 bits of dynamic range and variable readout rates. A detailed analysis of the issues involved in camera choice and usage can be found in8, 10. As mentioned earlier, spatial sampling of the region within a well is not necessary. Further, signalto-noise considerations show that it is not desirable.



A more appropriate sensing technology would involve the use of one light-sensing device per well and this option is possible by embedding photodiodes in the “floor” of each well. The problem associated with this approach, however, is that the sensor is now “looking into“ the excitation source, a step backward from Ploem’s invention of epiillumination fluorescence microscopy in the 1960’s11. If we are to actually achieve this goal then filtering will be required between the liquid sample and the surface of the photodiode. This is achieved through the use of a polycrystalline silicon layer.

3

THIN-FILM POLY-SI FILTER

An array of photodiodes can be fabricated in such a way that it is aligned with our wells and that there is a thin film (≈75 µm) of re-crystallized silicon in between the photodiode and the liquid sample. The photodiode current can then be sampled and digitized to measure the photon flux at the sampling time. The crystalline structure, with a grain size of about 2 µm, is capable of blocking excitation wavelengths below 400 nm and thus permitting only the emission wavelengths to pass12. A microarray chip incorporating these photodiodes and a schematic of the structures are shown in Figure 5a,b.

Transmittance [dB]

0

-10

-20

poly-Si on glass (simulated) poly-Si on glass (measured)

-30

-40 350

400

450

500

550

600

650

700

Wavelength [nm]

Figure 5a: Micro-array chip with in situ photodiodes and thin-film Si filters

Figure 5b: Schematic representation of a well including the in situ filter and photo-diode

Figure 5c: Comparison of the simulated t o measured filter characteristics for poly-Si thin film filter deposited on glass

With this technique a suppression of about 35 dB has been achieved in practice (see Figure 5c) and we have reason to believe that this can be improved to about 50 dB. The results for measuring NADH fluorescence are shown in Figure 6. 3

without poly-Si filter (right scale)

80 70 60

2

50 40

with poly-Si filter (left scale)

30

1

20 10 0 0

200

400

600

Photodiode Current [µA]

Photodiode Current [nA]

90

Figure 6: The poly-Si filter clearly suppresses sufficient excitation light to make fluorescence measurements possible. Note the use of two different vertical scales. The photodiode current has been sampled at time intervals of 210 ms.

0 800

NADH Concentration [µM]

This technique currently achieves a sensitivity in the range of 100 µM. This is not as good as the sensitivity achieved through the use of the CCD camera, 5 µM. Our poly-Si filter deposition technique will have to be improved in order to reach the sensitivity necessary to achieve the results shown in Figure 4b, results obtained with a scientific CCD camera. That this problem is primarily associated with the short wavelength blocking effects of the poly-Si filter can be seen in our experiments with bioluminescence where no excitation source is used.



4 BIOLUMINESCENCE MEASUREMENTS We have used a Roche ATP Bioluminescence HS II Assay Kit13 to test our filter-covered photodiode arrays as well as to determine the suitability of our technology for those applications where bioluminescence is more appropriate than fluorescence. The specific reaction is: ATP + D-luciferin + O2 Æ oxyluciferin + PPi +AMP + CO2 + light

(LUCIFERASE)

where the emission spectrum has a maximum at l = 562 nm. We covered a photodiode with 500 nl of solution and measured the luciferase-catalyzed reaction over an interval of 10 minutes with sampling intervals of 210 ms using varying concentrations of ATP: 0, 10, 20, and 40 µM corresponding to 0, 5, 10, and 20 pmol. We measured the signals from an “empty” photodiode and from the reaction where the ATP concentration was zero as controls. The first four minutes of the results are shown in Figure 7.

Photodiode Current [pA]

Photodiode Response to Bioluminescence 15 12 9

40 µM 20 µM 10 µM blank empty

6 3

Figure 7: Photocurrent of the HS II ATP bioluminescence kit on a single photodiode. The signals were recorded without any liquid (empty), filled with HS II reagent and an ATP concentration of zero (blank), and in the presence of different, non-zero concentrations of ATP.

0 0

50

100

150

200

250

Time [s]

The empty photodiode produced a signal with an average value of 3.24 pA and a standard deviation of 0.07 pA while the blank photodiode (ATP concentration = 0) produced a signal with an average value of 4.32 pA and a standard deviation of 0.08 pA = 80 fA. This indicates that while the blank well might have an increased average value, the contribution to noise in the measurement process is determined by the photodiode itself; the blank liquid does not contribute to the noise. These statistical measurements were made in the interval 50 s to 200 s, when the electronic transient behavior had stabilized. When we subtract the blank average value from the measurements where ATP was added, we obtain the result shown in Figure 8. Normalized Photodiode Response Photodiode Current [pA]

5 4 3

40 µM 20 µM 10 µM blank

2 1

Figure 8: The average value of the blank signal has been subtracted from the photodiode response. The resulting data is displayed in the range 50 ≤ t ≤ 200 s.

0 -1

0

50

100

150

200

250

Time [s]

When we look at the average values of each of these curves versus the ATP concentration, we find that the sensitivity is about 0.1 pA/µM. The standard deviation (s) of the noise in the blank, 80 fA, then translates into 800 pM. This is very close to the levels that would make a bioluminescence assay practical. The data are not strictly linear in the ATP concentration. This can be explained by the fact that we are operating outside of the linear range of the HS1II kit as specified by the manufacturer, which is 1 pM to 1 µM13.



5

SUMMARY AND CONCLUSIONS

The photodiodes are capable of producing a measurement of light flux that is both sensitive and quantitative. The sensitivity can be improved in several ways all of which are being investigated at this time: 1) Instead of sampling the photodiode current at intervals of 210 ms, we can integrate the current on capacitors that are embedded in the chip in a manner similar to CCD camera sensors and then read out the total charge after a sufficiently long integration period; 2 ) We can improve the thin-film poly-Si filters to produce greater wavelength selectivity for fluorescence measurements so that a rejection ratio of some 50 dB instead of 35 dB can be achieved; 3 ) For bioluminescence, we can use photodiodes that are not covered by the poly-Si filters. Around the wavelength of interest, l = 562 nm, the poly-Si filter has a –10 dB suppression meaning that we are losing 90% of the incoming photons. As there are no excitation photons to contend with, this means that we are losing emission signal. In our next series of experiments we will be using filter-free photodiodes. In an ideal situation we would be able to measure the ATP activity in a single (yeast) cell. At the present time the sensitivity we have achieved of 800 pM means that at the 3s level of 2.4 µM we are measuring the ATP associated with more than 106 cells. This assumes that a single yeast cell, on average, contains 6.5 µgr per gram (dry weight) cell14. We are clearly a long way from a single-cell metabolic assay. The use of a silicon substrate means that a variety of options are available to us for in situ instrumentation. In this paper we have shown that through the use of poly-Si thin-film filters and embedded photodiodes, we can measure fluorescence and bioluminescence phenomena. We will continue to improve this development and to explore other, novel possibilities for embedded instrumentation.

ACKNOWLEDGMENTS This work was partially supported by the Physics for Technology program of the Foundation for Fundamental Research in Matter (FOM) and the Delft Inter-Faculty Research Center Intelligent Molecular Diagnostic Systems (DIOC-IMDS).

REFERENCES 1. 2. 3. 4. 5. 6.

7.

8.

9.

R. Moerman, J. Frank, J.C.M. Marijnissen et al., “Picoliter dispensing in wells of a micro-array by means of electrospraying,” Journal of Aerosol Science, 30:1, pp. 551-552, 1999. R. Moerman, J. Frank, and J.C.M. Marijnissen, "Method of the dosed application of a liquid onto a surface", Patent #WO 0035590, 2001. R. Moerman, "Device for carrying out a reaction, and a method of carrying out a reaction in the device", Patent #PCT/NL02/00095, 2002. K.T. Hjelt, L.R. van den Doel, P. Szczaurski et al., “Monitoring of Liquid Evaporation in Sub-Nanoliter Reactors,” presented at the Eurosensors XIII, The Hague, Netherlands, pp. 695-698, 1999. K.T. Hjelt, L.R. van den Doel, W. Lubking et al., “High-resolution liquid volume detection in sub-nanoliter reactors,” Sensors and Actuators A, 83, pp. 61-66, 2000. L.R. van den Doel and L.J. van Vliet, “Temporal phase-unwrapping algorithm for dynamic interference pattern analysis in interference-contrast microscopy,” Applied Optics - Optical Technology and Biomedical Optics, 40:25, pp. 4487-4500, 2001. I.T. Young, K.T. Hjelt, L.R. van den Doel et al., “Measuring Liquid Volumes in Subnanoliter Wells,” presented at the Biomedical Instrumentation Based on Micro- and Nanotechnology II, San Jose, California, R.P. Mariella and D.V. Nicolau ed., Vol. 4265, pp. 75-80, Publisher SPIE, 2001. L.R. van den Doel, R. Moerman, G. van Dedem et al., “Monitoring Enzyme-catalyzed Reactions in Micromachined Nanoliter Wells using a Conventional Microscope based Microarray Reader,” presented at the Novel Micro- and Nanotechnologies for Bioengineering Applications, Photonics West, BIOS 2002, San Jose, California, R.P. Mariella and C.J. Murphy ed., Vol. 4626, pp. 366-377, Publisher SPIE, 2002. L.F.A. Wessels, E.P. van Someren, and M.J.T. Reinders, “Information Processing for Intelligent Molecular Diagnosis,” Pattern Recognition Letters, 20:11-13, pp. 1457-1465, 1999.



10. J.C. Mullikin, L.J. Van Vliet, H. Netten et al., “Methods for CCD Camera Characterization,” presented at the SPIE Conference on Image Acquisition and Scientific Imaging Systems, San Jose, California, Vol. TC-2173, pp. 73-84, Publisher SPIE, 1994. 11. J.S. Ploem, “The use of a vertical illuminator with interchangeable dichroic mirrors for fluorescence microscopy with incident light,” Zeitschrift für wissenschaftliche Mikroscopie, 68, pp. 129-142, 1967. 12. V.P. Iordanov, W. Lubking, R. Ishihara et al., “Silicon thin-film UV filter for NADH fluorescence analysis,” Sensors and Actuators A, 97-98, pp. 161-166, 2002. 13. Roche Molecular Biochemicals, “ATP Bioluminescence Array Kit CLS II,” Roche Diagnostics GmbH, Mannheim, 1999. 14. Personal Communication: J. van Dam to A. Kroon and V.P. Iordanov, 2002.
