Fred K. Friedman, Laboratory of Metabolism, Center for Cancer Research, NCI, NIH
Allen Markowitz, Division of Bioengineering and Physical Science, NIH
Laser flash photolysis is a useful technique for disruption of photolabile chemical bonds. The kinetics of the subsequent re-formation of such bonds can then be monitored with a suitable detection system. This technique has proven useful for following the reassociaton kinetics of hemeproteins with carbon monoxide. We describe the construction of this apparatus along with microcomputer based signal acquisition and processing.
The essential components of the system are:
The ideal laser light source delivers a pulse with sufficient energy to break the photolabile bond, and whose duration is small relative to the reassociation time. A model DL2100A dye laser, model PDL-11 electronics console, and model 2K dye circulator (Lumenx, Durham NH)) are used. This system yields an output energy of 1 joule with a pulse width of 0.4 µs. A high voltage of 18 kV was routinely used to generate a laser pulse. Rhodamine 110 in ethanol was used as the laser dye, since it lases at 570 nm, a light wavelength which photodissociates the cytochrome P450-CO complex.
The sample holder, a 1 cm square quartz spectrofluorometer cell, was placed in a water- cooled aluminum block such that the detection light was perpendicular to the laser pulse. In association with an external circulator water bath, it provides temperature regulation in the range of 0 to 50 degrees.
The detection light source is a 45 watt tungsten halogen lamp (General Electric model Q6.6A/T2½/CL). Its power supply (Lambda model LES-F-01-OV) supplys variable voltage to the lamp. Light from the lamp is focused with a lens and passed through a 450 nm bandpass filter (Kodak); this wavelength exhibits the greatest difference in transmittance upon formation of the cytochrome P450-CO complex.
The Wratten number 50 filter (Kodak) between the sample and the PMT reduces the intensity of the scattered laser light which impinges on the PMT; this prolongs PMT lifetime and reduces PMT hysteresis. A PMT ( Hamamatsu model R106) is used in conjunction with a high voltage power supply (Hamamatsu model C956-04). The latter is connected to an external +15 volt power supply with a variable resistor that controls the high voltage level.
A major noise source is the PMT. The S/N ratio of the PMT is given by:
S/N = Is/sqrt((2q(Is+Id)Gf))
Is = signal current
Id = dark current
G= current amplification factor of PMT
f= bandwidth of the system in Hz
q= electronic charge = 1.6x10^19 coul
Examination of this equation reveals that the S/N is increased by lowering G and increasing Is. The former is accomplished by decreasing the PMT high voltage. Although Is is lowered by this change, Is can then be readjusted by raising the light intensity. It should also be noted that decreasing the PMT high voltage also reduces the maximum linear response of Is to light intensity, and care must thus be taken to ensure that Is remains in the linear response region.
ELECTRONIC SIGNAL PROCESSING AND CONTROL
In order to view the time dependence of the transmittance change, the PMT output current was converted to voltage, amplified and filtered. The PMT current is first input to a preamplifier (Hamamatsu model C1053-3) which is a current to voltage converter with a conversion factor of 0.3 V/µa.
According to the Nyquist sampling theorem, the sampling rate must be greater than twice the signal frequency in order to eliminate frequency aliasing. If this condition is not met, high frequency components of the signal appear as low frequency components which cannot be subsequently corrected by software filtering. To prevent aliasing, the signal is applied to a 8-pole low-pass Bessel variable filter (Frequency Devices model 902) with a range of 0 to 29.9 kHz. This device also introduces a variable offset and gain (0-20 db) in order to subsequently utilize the full scale of the A/D converter, and allows for improved resolution of the signal. A/D conversion and data storage are performed using an IBM 386 microcomputer with an internal A/D converter board (Data Translation model DT2821-G8DI). This 12 bit board has a 150 kHz throughput and variable gain.
Data collection is initiated by triggering the A/D converter with a start switch. After an adjustable delay time during which baseline signal data was acquired, photolysis is initiated by electronically triggering the laser pulse.
Data acquisition is performed using Labtech Notebook software (Laboratory Technologies, Wilmington MA), which produces data in ASCII format. The S/N ratio is then improved by the following signal processing techniques.
Data from repetitive experiments are first averaged to improve S/N. The data are then subjected to digital filtering using a Fortran program, which employs a FFT subroutine, to remove high frequency noise. However, an FFT requires periodic input data. Since data for the reassociation reaction is a single and nonrepetitive decaying signal, the algorithm thus imposes periodicity via introducing a transient to connect the initial and final data points. To eliminate the resulting frequency artifacts, detrending with curve inversion was used prior to performing the FFT: a linear function was added to the data to yield a transformed data set with zero end values, followed by curve inversion around the origin. The user then provides a cutoff frequency for removing frequency components above the range of interest, and these components were rejected using a first-order low- pass filter. An inverse FFT is then performed, followed by reversing the above detrending and curve inversion operations.
In a typical application with rat liver microsomes, the high voltage and lamp voltage were experimentally determined to minimize the S/N ratio due to the PMT, and were 800 V and 4.0 V, respectively. Optimum settings correspond to the maximum light intensity consistent with a linear PMT response, which avoided photo-degradation of the sample. To ensure signal fidelity in accordance with the Nyquist criterion, we use sampling rates of 10-50 kHz with the Bessel filter set at half this frequency. The amplifier gain and offset voltage were set to 20 db and 2.1 V, respectively, in order to utilize full scale resolution of the A/D converter with an input range of ±10 volts. Approximately 130 baseline readings are taken before the laser pulse, in order to determine the equilibrium value.
Software enhancements of the data can be performed with three techniques, which differ in their relative merits. 1) Repetitive signal averaging is the simplest approach and does not distort the data. However, this approach can be limited owing to photo-degradation of the sample after multiple laser flashes. In addition, the marginal efficacy decreases as the repetitive number increases because S/N varies with the square root of this number. 2) A moving point average distorts data and has the additional disadvantage that this operation excludes a series of initial and final points; this tends to conceal any initial rapid reaction components. 3) The FFT approach for low pass filtering is potentially the most powerful. However, one must carefully examine the frequency domain data to ensure selection of an appropriate cutoff frequency which is higher than that of the reaction components. Selection of a lower frequency would distort the data owing to exclusion of relevant frequency components above the cutoff frequency.
Return to Home Page