A new sub-nanosecond LED at 280 nm: application to protein fluorescence

C D McGuinness¹, K Sagoo², D McLoskey² and D J S Birch^1,2

¹Photophysics Group, Department of Physics, University of Strathclyde, Glasgow G4 0NG, UK
²Horiba Jobin Yvon IBH Ltd, Skypark 5, 45 Finnieston Street, Glasgow G3 8JU, UK

Email: djs.birch@strath.ac.uk

Received 9 August 2004, in final form 14 September 2004
Published 11 October 2004

Keywords: LED, light emitting diode, protein fluorescence, pulsed optical sources, sensors, fluorescence decay time, fluorescence lifetime

Introduction

Molecular fluorescence has spawned across the disciplines many new techniques in spectroscopy, sensing and imaging. Progress in such techniques is often limited by the available technology. Here we report an advance in technology, which we believe will be of significance to a wide variety of research, and in particular in the biological and medical as well as chemical sciences.

In recent years new light sources have emerged in the form of nitride semiconductor light emitting diodes (LEDs) and laser diodes at wavelengths which span the visible and visible-ultraviolet boundary, and at considerably lower cost and higher reliability than conventional sources. Progress in these sources has attracted global attention and has been extensively reviewed [1, 2]. Of relevance to our purpose here, such diodes can be pulsed on the sub-nanosecond timescale, making them ideal for fluorescence decay measurement. Fluorescence decay studies of molecules are now accepted as having expanded the field of fluorescence from spectroscopy and microscopy into kinetics and sensing. The well-known virtues of fluorescence include high sensitivity (down to the single-molecule level), specificity, non-destructive nature, in situ sample compatibility and easily accessible spectral range. Fluorescence decay measurements bring additional benefits by offering independence from changes in fluorophore concentration including the effect of photo-bleaching, routine measurement, time discrimination, kinetic rates and unambiguous calibration.

Previously reported fluorescence decay measurements using LEDs, which typically have µW average powers and mW peak powers, include a device of 450 nm peak wavelength giving 3 ns pulses at 10 kHz repetition rate [3], a LED of 525 nm peak wavelength giving 25 ns pulses at 11 kHz repetition rate [4] and LEDs of 525 nm, 560 nm and 590 nm peak wavelengths pulsed at up to 10 MHz to give 1.9 ns, 3 ns and 3.6 ns pulse durations respectively [5]. Samples studied include quinine sulfate [3], feldspars and quartz [4] and rhodamine dyes in solution and as sensors of metal ions [5]. A pulsed output of 380 nm from a LED of 450 nm peak wavelength when operated at currents >50 mA has also been reported [6]. Commercial pulsed LED sources are also available. Throughout all this work the shortest wavelength available with LEDs has hitherto been 370 nm, precluding the study of one of the most important intrinsically fluorescent samples, namely the amino acid fluorescence of proteins, which absorb below 300 nm.

Proteins are fundamental to life and the mechanisms by which each protein folds to a unique conformation still remain one of the much researched, yet unsolved problems of our time. Along with NMR, infrared spectroscopy, circular dichroism and other techniques, protein intrinsic fluorescence decay and its associated observation of fluorescence anisotropy, collisional quenching and resonance energy transfer are key tools available to the protein researcher [7].

Measuring the fluorescence decay allows the dynamics of protein structural changes to be observed directly [8]. Moreover, excitation of the native fluorescence of proteins emanating from the amino acids tyrosine and tryptophan eliminates the possibility of perturbation of the local environment when using extrinsic fluorescent probes. Previous pulsed optical sources for exciting intrinsic protein fluorescence include synchrotron radiation [9], mode locked lasers [10] and flashlamps [11]. The latter has hitherto been the only low-cost and simple-to-use option. However, when compared to a LED, a flashlamp is much less convenient as its electrodes need regular cleaning and its gas needs replenishing, it has lower repetition rate (typically ~40 kHz), has increased potential for radio frequency distortion of decays due to higher voltage switching and poorer pulse-to-pulse temporal reproducibility, although LED and flashlamp pulse intensities are often comparable [5].

In fluorescence measurements shown here on human serum albumin (HSA) in a sol-gel sensor matrix we demonstrate what, to the best of our knowledge, may well be the first report of protein intrinsic fluorescence excited using a LED. We believe this approach will be of interest to protein and indeed other researchers who study fluorescence excited below 300 nm, but who wish to obtain data faster than with flashlamps, more easily than using mode-locked lasers and more conveniently than with synchrotrons.

Techniques, results and discussion

We have used the time-correlated single-photon counting (TCSPC) technique [12] to record and IBH reconvolution software to analyse the fluorescence decays. The data accumulation rate in TCSPC is proportional to the source repetition rate, but is limited to ~2% of the source repetition rate if pile-up effects are to be avoided [12]. We have recently reported switching circuitry which allows LED pulse rates up to 10 MHz [5], but in reality a source repetition rate of ~1 MHz is sufficient to accumulate the fluorescence decay of most samples in only a few minutes. Only in single molecule and imaging applications might higher repetition rates prove routinely beneficial. Here we have used the IBH NanoLED drive circuitry at 1 MHz to achieve LED pulsing and TCSPC synchronization.

The LED we have used is based on AlGaN fabrication technology [13]. Figure 1 shows a typical LED spectral profile (actual peak at 282 nm) recorded with an IBH f/3 monochromator at 2 nm bandwidth and incorporating a holographic grating in a Seya Namioka geometry. Figure 2 shows the full LED spectral profile, where a longer wavelength emission peak can be seen at 430 nm. The latter peak cannot be used for exciting other fluorophores as it has a long decay time of ~500 µs. Hence care is needed when using cut-off filters to select fluorescence excited with the 280 nm band if the 430 nm band is not to be detected along with the Stokes shifted fluorescence. For this reason we found it necessary to use a monochromator to pre-filter the LED excitation at 280 nm rather than simply exciting with the diode directly. Similarly, although selection of fluorescence using a cut-off filter provides higher sensitivity, further discrimination against detecting the 430 nm band is achieved by using a monochromator to select fluorescence. Figure 3 shows a typical instrumental pulse recorded using an IBH TBX-04 detector under TCSPC conditions at a time calibration of 27 ps per channel, as used for all the time-domain measurements. The ~600 ps full width at half maximum (fwhm) is significantly faster than previous reports with visible LEDs [3-6] and compares very favourably with a typical hydrogen flashlamp pulse also shown. Figure 4 shows the emission spectrum of HSA in an alcohol-free hydrated sol-gel matrix of tetramethylorthosilicate (TMOS) prepared at pH 7.5 using standard hydrolysis and condensation reactions [14] and obtained with the 280 nm LED as the excitation source in an IBH Model 5000U fluorimeter. The spectrum shows the well-known features [15] without artefacts.

Protein retains enzyme activity in silica sol-gels [16] which have useful features such as biocompatibility and a nanometre pore size, facilitating immunoassay of analytes such as metal ions, glucose etc by preventing protein aggregation, transport of analytes of interest and exclusion of high molecular weight interferents such as extraneous protein [17]. Figure 5 shows the fluorescence decay of HSA in the hydrated sol-gel using 280 nm LED excitation and recorded with an IBH Model 5000U fluorometer equipped with excitation and emission monochromators, the latter tuned to transmit 335 nm to select the protein fluorescence. The log scale clearly shows the sharp LED pulses free from afterglow or after-pulsing. The triple decay parameters of 0.53 ± 0.05 ns, 2.43 ± 0.15 ns, 6.07 ± 0.05 ns (errors all 3 std. dev.) and relative intensities 8%, 38%, 54% respectively are consistent with work in other laboratories, e.g. using a mode-locked laser [18], and also our own measurements with a hydrogen flashlamp of 0.59 ± 0.06 ns, 2.47 ± 0.19 ns, 6.04 ± 0.05 ns and 9%, 36% and 55% relative intensities. The triple decay parameters can be attributed to the now widely accepted tryptophan conformer model [19]. The quality of the goodness of fit (LED chi-sq 1.00, flashlamp 1.08) shows the data to be free from the effects of scattered excitation or scattered fluorescence as might be expected for a porous medium. This bodes well for the downstream design of integrated fluorescence lifetime sensors using protein immunoassays and LED excitation. In preliminary measurements, a range of 280 nm LEDs at 1 MHz gave up to a factor of ~12 higher protein fluorescence count than the hydrogen flashlamp at 40 kHz. No doubt this will be improved upon further with better collection optics, higher power LEDs and indeed laser diodes, but it is still significant as it stands.

HSA contains a single tryptophan residue that simplifies the kinetics, but as well as direct tryptophan excitation at 280 nm energy transfer from tyrosine to tryptophan also occurs. Indeed the 280 nm LED is ideal for tyrosine excitation [20], but if tryptophan is to be preferentially excited, 295 nm is closer to its absorption peak [7-9] and gives ~3× more fluorescence. A wavelength of 295 nm can in principle be achieved with the same LED fabrication technology as we have used here [13]. Indeed, we have performed fluorescence decay measurements with a 340 nm LED, and LED wavelengths down to 250 nm have also recently become available. As well as proteins 280 nm is also ideal for exciting other fluorophores such as naphthalene, stilbene, etc.

Taken together this means that all of the vast range of aromatic molecules is now accessible using LED excitation at extremely low cost. The exciting opportunities created by these developments will undoubtedly also impact well outside areas of fluorescence spectroscopy.

Conclusion

We have demonstrated protein fluorescence decay excited with an ultraviolet LED. We expect increasing use of this approach. However, this may not necessarily lead to the nanosecond flashlamp becoming obsolete since its broad-band spectrum is very convenient and indeed essential for the measurement of time-resolved excitation spectra. Similarly the ultrafast pulses of mode locked lasers still offer the ultimate in time resolution. Nevertheless, we envisage the low cost, ease of use, convenience and compactness of 280 nm to 300 nm nanosecond LEDs will make fluorescence techniques more widely used by protein researchers. This includes not only decay kinetics as shown here but also emission spectroscopy, microscopy, imaging and sensing using steady state, modulated and pulsed modes of operation.

Acknowledgments

DJSB and CDM would like to thank EPSRC for research grants and a research studentship held by CDM.

[1] 

Nakamura S 1999 InGaN-based violet laser diodes Semicond. Sci. Technol. 14 R27-40 

[2] 

Akasaki I 2002 Nitride semiconductors J. Cryst. Growth 237-239 905-11 

[3] 

Araki T and Misawa H 1995 Light emitting diode-based nanosecond ultraviolet light source for fluorescence lifetime measurements Rev. Sci. Instrum. 66 5469-72 

[4] 

Chithambo M L and Galloway R B 2000 A pulsed light-emitting-diode system for stimulation of luminescence Meas. Sci. Technol. 11 418-24 

[5] 

O'Hagan W J, McKenna M, Sherrington D C, Rolinski O J and Birch D J S 2002 MHz LED source for nanosecond fluorescence sensing Meas. Sci. Technol. 13 84-91 

[6] 

Araki T, Fujisawa Y and Hashimoto M 1997 An ultraviolet nanosecond light pulse generator using a light emitting diode for test of photodetectors Rev. Sci. Instrum. 68 1365-8 

[7] 

Valuer B 2002 Molecular Fluorescence (Weinheim: Wiley-VCH)  

[8] 

Eftink M R 1994 The use of fluorescence methods to monitor unfolding transitions in proteins Biophys. J. 66 482-502 

[9] 

Munro I, Pecht I and Stryer L 1979 Subnanosecond motions of tryptophan residues in proteins Proc. Natl Acad. Sci. USA 76 56-60 

[10] 

Small E W 1991 Laser sources and microchannel plate detectors for pulse fluorometry Topics in Fluorescence Spectroscopy vol 1: Techniques ed J R Lakowicz (New York: Plenum) pp 97-182 

[11] 

Birch D J S and Imhof R E 1981 Coaxial nanosecond flashlamp Rev. Sci. Instrum. 52 1206-12 

[12] 

Birch D J S and Imhof R E 1991 Time domain fluorescence spectroscopy using time-correlated single-photon counting Topics in Fluorescence Spectroscopy vol 1: Techniques ed J R Lakowicz (New York: Plenum) pp 1-95 

[13] 

Whitaker T 2004 SUVOS pushes UV LEDs and lasers to shorter wavelengths  http://Optics.org June

[14] 

Brinkler C J and Scherer G W 1990 Sol-Gel Science: The Physics and Chemistry of Sol-Gel Processing (New York: Academic)  

[15] 

de Lauder W B and Wahl Ph 1971 Fluorescence studies of human serum albumin Biochem. Biophys. Res. Commun. 42 398-404 

[16] 

Avnir D, Braun S, Lev O and Ottolenghi M 1994 Enzymes and other proteins entrapped in sol-gel materials Chem. Mater. 6 1605-14 

[17] 

McCartney L J, Pickup J C, Rolinski O J and Birch D J S 2001 Near-infrared fluorescence lifetime assay for serum glucose based on allophycocyanin-labelled concanavalin A Anal. Biochem. 292 216-21 

[18] 

Hirshfield K M, Toptygin D, Grandhige G, Kim H, Packard B Z and Brand L 1996 Steady-state and time-resolved fluorescence measurements for studying molecular interactions: interaction of a calcium-binding probe with proteins Biophys. Chem. 62 25-38 

[19] 

Szabo A G and Rayner D M 1980 Fluorescence decay of tryptophan conformers in aqueous solution J. Am. Chem. Soc. 102 554-63 

[20] 

Kabuyama Y, Homma M K, Kurosaki T and Homma Y 2002 Early signaling events induced by 280 nm UV irradiation Eur. J. Biochem. 269 664-70