- Email: [email protected]
A high speed optical multichannel analyzer
J. Biochem. Biophys. Methods 35 (1997) 161–174
A high speed optical multichannel analyzer a, b a 1 ,c J.W. Cole *, R.W. Hendler , P.D. Smith , H.A. Fredrickson , T.J. Pohida c , W.S. Friauf 2 ,a a National Center for Research Resources, National Institutes of Health, Bethesda MD 20892, USA National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda MD 20892, USA c Division of Computer Research and Technology, National Institutes of Health, Bethesda MD 20892, USA b
Received 17 July 1997; received in revised form 14 August 1997; accepted 15 August 1997
Abstract An optical multichannel analyzer capable of recording spectra at sampling rates up to 100 kHz is described. The instrument, designed to gather data on the kinetic reaction mechanisms of biological preparations such as cytochrome oxidase and bacteriorhodopsin, features a massively parallel approach in which each photosensing element of the detector array has a dedicated amplifier, integrator, analog to digital converter, and sample buffer. The design has 92 such elements divided in two separate arrays, each of which sits at the focal plane of a 1 / 4 m Ebert spectrometer. The spectrometers may be tuned to cover independent, 130 nm wide, regions of the spectrum from 350 nm to 900 nm with a dispersion of 2.8 nm per element. Each detection channel has 12-bit resolution with an electronic dark count of 1 count and may be sampled 1024 times during a single experiment with dynamically variable sampling intervals from 10 ms to several seconds. Time averaging of up to thousands of consecutive laser-initiated kinetic cycles allows analyses of spectral changes , 0.001 optical density units. A personal computer with custom software provides a number of features: entry of experiment parameters; transfer of data from temporary buffers to permanent files; real time display; multiple spectrum averaging; and control and synchronization of associated system hardware. Optical fibers or lenses provide coupling from a parabolic reflector Xenon arc monitoring light source, through the sample chamber, to the entry slit of the monochromator. The instrument has been used for extensive studies on the rapid kinetics and definition of reaction sequences of the energy-transducing enzymes cytochrome oxidase and bacteriorhodopsin. Some results from these studies are discussed. 1997 Elsevier Science B.V. *Corresponding author. Address for correspondence: 9400 Wooden Bridge Rd, Potomac, MD 20854, USA. Present address: P.O. Box 68, Ashton, MD 20861, USA. 2 Present address: 5616 Oakmont Ave, Bethesda, MD 20817, USA. 1
0165-022X / 97 / $17.00 1997 Elsevier Science B.V. All rights reserved. PII S0165-022X( 97 )00037-7
162
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Keywords: Optical multichannel analyzer; Reaction mechanism; Cytochrome oxidase; Bacteriorhodopsin
1. Introduction Spectrophotometry is an important investigational technique used for studies of chemical and biological reactions. In optical spectroscopy, the absorption of light, at specific wavelengths through a reaction mixture, provides information about intermediates, products, and reaction pathways. One particular application of this technique is in the study of the rapid kinetics of chemical transformations. Usually, several chromophoric substances are involved in these transformations. As the reaction proceeds, spectral changes are produced by the transformations of these chromophores, each of which contributes to the absorbance changes at the monitored wavelength(s). To date, most instruments have been limited to the continuous monitoring of rapid kinetic reactions at only one or two wavelengths. With limited spectral information, deconvolution of the total observed absorbance change into the relative contribution of each chromophoric component present is virtually impossible. A few laboratories have tried to solve the problem by using a two flash technique. First, a rapid reaction is initiated by a laser flash. Secondly, a monitoring flash (usually Xenon) is used to obtain a spectrum at a single time point. In this procedure, each time point requires a separate enzyme sample in a separate experiment. It must be assumed that each time point has a completely reproducible zero time point. The instrument described here collects up 1024 sequential spectra on a single sample. The ability to obtain a timed sequence of complete spectra has become particularly important in recent years because of advances in the analysis of spectral deconvolution, which utilize all of the information in multichannel spectra and permit the isolation of the contribution(s) of each component [1,2]. This analysis was first applied in equilibrium studies where spectra are collected slowly. With the realization that the same approaches would be valuable in the analysis of rapid kinetic data, the need for an instrument to collect a sequence of entire spectra rapidly was apparent. The instrument described in this paper was designed and built to address this need. Many commercial spectrophotometer systems are currently available based on a variety of light transducer technologies including photomultiplier tubes (pmt), photodiodes, and charge-coupled arrays. These transducers produce an output signal that is proportional to the intensity of the incident dispersed light, which is either mechanically scanned across a pmt or exposed directly on to a series of adjacent transducers each of which correspond to an incident wavelength. The output signals are multiplexed into ‘sample and hold’ devices to obtain a time sequence of data points corresponding to the light intensity on specific transducer elements. Current commercial instruments are limited by the speed of the multiplexing and data readout electronics. A multiplexed ‘sample and hold’ approach reduces the time resolution of the instrument because of the excessive ‘wait time’ between successive readings of one specific transducer. The instrument described in this paper overcomes this problem by using parallel processing
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
163
techniques. In place of multiplexing, each individual transducer has a dedicated amplifier, integrator, analog to digital converter (ADC), and sample buffer. This design approach permits the instrument to simultaneously sample, using 12-bit ADCs, all 92 channels at a maximum rate of 100 000 samples per second. All parameters, such as the sampling rate and the number of samples at each specific rate, are programmable via a personal computer (PC). The 92 transducers are provided in two 46 element diode arrays, each placed at the focal plane of a separate spectrograph, which enables independent ranges of wavelength to be observed (130 nm as described herein). In contrast, existing commercial instrumentation, capable of comparable sampling rates, is limited by a fixed sampling rate, a lower number of samples per channel, and a smaller detector element size requiring greater than 100-fold the amount of light to be incident on the reaction mixture.
2. General system description The instrument consists of ten subsystems, shown in Fig. 1, which are described below.
2.1. Light sources This subsystem consists of two different light sources: one to monitor the progress of the reaction; and one to initiate the reaction (i.e. actinic). Each is directed into the
Fig. 1. System diagram. Each of the sub-systems is described in detail in the text. The dotted lines indicate optical coupling.
164
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
reaction chamber. The monitoring light source is a 75 watt continuous Xenon lamp (Model: LPS-220, Photon Technology International, South Brunswick, NJ). The intensity of this source is held constant by a light feedback regulator. The actinic light source is a pulsed laser (Surelite-10 ND/ YAG, Continuum Lasers, Santa Clara, CA) that emits a 5 ns pulse with a maximum energy of 150 mJ at 530 nm. It can be delivered either as a single shot or in a repetitive mode at a maximum rate of 8.3 Hz. The actinic light impinges on the reaction cell at a 908 angle relative to the monitoring light path.
2.2. The reaction chamber For rapid kinetic studies, it is essential that the reaction be started instantly and synchronized with the data acquisition. In some cases, a single reactant is energized to undergo a chemical change by a laser pulse (for example, the bacteriorhodopsin photocycle). In other cases, two reactants must be rapidly mixed together either for spontaneous reaction or for subsequent induction to react by a laser pulse (for example, the oxidation of carbon monoxide-liganded reduced cytochrome oxidase by O 2 ). In the latter case, the rapid mixing of the reactants requires the use of a stopped-flow apparatus. For this purpose, a Kintek Instruments (University Park, PA) stopped-flow apparatus was used.
2.3. Optics The optical subsystem consists of a series of lenses and / or optical fibers arranged to couple the monitoring and actinic light sources through the reaction chamber to the spectrographs. The monitoring light, after passing through the reaction chamber is split, by either a beam splitter for the bacteriorhodopsin experiments or a bifurcated optical fiber bundle for the stopped flow / cytochrome oxidase experiments The exiting light was focused on the input slits of two spectrographs. Each spectrograph disperses the white light received from the reaction chamber into 130 nm bands selected by spectrograph settings. The 1 / 4 meter spectrographs used in this application (Model 77200, Oriel Corporation, Stratford, CT) contain 1200 line / mm gratings (Model: 77232). A 46element photodiode array (described below) is mounted at the output port of each spectrograph providing a dispersion of approximately 2.8 nm per element.
2.4. Front end electronics The front end electronics consist of 92 independent circuits, each comprised of a photodiode, a current amplifier, and an integrator (Fig. 2). The photodiode current is not integrated directly because the total charge during the shorter intervals is not great enough relative to switching transients and other noise sources to provide an adequate signal-to-noise ratio. Also, the integrating capacitor needed to provide the desired output voltage would not be large enough relative to parasitic capacitances. Circuit component values were chosen to produce a final analog output signal voltage between 0 and 10 volts, corresponding to the input voltage range of the ADCs (conversion electronics). The photodiode arrays (Model: S4112-46) are produced by Hammamatsu Photonics
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
165
Fig. 2. Front end electronics and conversion electronics. The photocurrent from each element of the photodiode array is connected to an individual amplification and integration circuit. The integrated photodiode signal is digitized by a dedicated 12-bit ADC and stored by a 1024 deep Sample FIFO buffer.
(Hammamatsu City, Japan). Each element has a relatively large active area of 4.4 mm 3 0.9 mm which maximizes the critical optical coupling of the light onto the detector array. They also possess a broad spectral response curve, ranging from 200 to 1050 nm. The first stage operational amplifier (Model: AD549, Analog Devices, Norwood, MA), selected because its very low input bias current, is ideally suited for measuring small current values and fluctuations. A 2 ms-wide pulse signal with a programmable repetition period of 10 to 70 ms causes a temporary closure of the ‘capacitor clear’ switch which resets the integrator output to zero volts. Fig. 3 shows the relationship between the integrator output signal, the ADC convert signal, and the ‘capacitor clear’ signal. In this particular example, the ‘capacitor clear’ signal is set at a 10 ms repetition interval. The front end electronics are constructed on four identical 23-channel copper-clad boards. Each board receives 23 photodiode currents via a ribbon cable from the photodiode arrays. Two boards are required per photodiode array and are housed with the photodiode array and spectrograph. In summary, there are two front end assemblies, each containing one spectrograph, one photodiode array, and two front end boards.
2.5. Computer control and interface A PC with plug-in boards is the interface between the user and the custom electronic hardware sub-systems. A custom software package interprets user commands which are then relayed to the hardware through two Input / Output (I / 0) driver cards located within the PC. The first of these cards is a five channel counter timer board (CTM-05) and the second a 24-bit parallel digital I / 0 interface board (PI0-12), both manufactured by
166
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Fig. 3. Timing diagram indicates the temporal relationship between the front end integrator output, the capacitor clear signal and the ADC convert signal.
Keithley / Metrabyte (Tauton, MA). Each card sends and receives a number of signals to and from the hardware (Figs. 4 and 5). The CTM-05 outputs are: SHIFT IN Writes control word into Control FIFO (first-in, first-out memory) —see below. START Starts experiment. BUS CONTROL Dictates data flow direction on bus (16-bit control word in or 12-bit data word out). INIT Initializes interface with Sample FIFO —see below. RESET Resets all electronic hardware. CLEAR RATE Programs (3-bits) capacitor clear pulse repetition period (10–70 ms). The CTM-05 inputs are: INPUT READY Control FIFO ready for Control word. CONTROL FIFO EMPTY Control FIFO is empty. SAMPLE FIFO EMPTY Sample FIFO is empty. The P10-12 outputs are: CONTROL WORD Data word pairs used by timing electronics (16-bits).
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
167
Fig. 4. PC control interface. The bus control line from the PC CTM- 05 board dictates whether the PC will be sending 16-bit control words or receiving data from the Sample FIFOs.
Fig. 5. Experiment synchronization electronics. Adjustable delays allow the A / D convert signal to be synchronized with the other experiment signals (laser and stopped flow instrumentation).
168
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
ADDRESS LINES Select one channel for sample read (7-bits). READ Read strobe for Sample FIFOs. The P10–12 inputs are: DATA 12-bit sample words from Sample FIFOs. It is important to note that the PI0–12 functions in a dual capacity: it sends 16-bit timing word pairs (discussed below) and receives the 12-bit measurement data words. In order to perform these functions using the same I / 0 port, a bus arbitration interface was developed. A dedicated output from the CTM-05 dictates the direction of data transfer on the bus. A ‘0’ indicates that the 16-bit word pairs will be written to the measurement hardware while a ‘1’ indicates that the 12-bit data words will be read from the hardware.
2.6. Experiment synchronization electronics The experiment synchronization electronics allow the user to control various timing parameters such as those associated with the laser, stopped-flow apparatus, and measurement systems. A block diagram of the experiment synchronization electronics is shown in Fig. 5. The electronics have been designed to allow a variety of different experiments to be performed by simply changing switch settings. Either single experiments or experiments that average many spectra may be performed. During operation, the laser requires an 8.3 Hz TTL pulse train to activate the flashlamp. A single 100 ms TTL pulse, which is synchronized to the flashlamp pulse train, is required to fire the laser. All other signals required for the experiment, including those which activate the stopped-flow apparatus and initiate data collection, are provided prior to the initiation of the laser pulse.
2.7. Timing electronics The data sampling control electronics sequentially interpret control word pairs previously sent from the computer and produce a user-defined data acquisition sequence. The word pairs encode both the sampling rate and the number of samples desired at that rate, which the user enters via a PC. Fig. 6 schematically illustrates the timing electronics. The words written by the PC are 16-bits wide and are sent in pairs. The word pairs are stored until the beginning of an experiment in a 16-bit, 64 word FIFO (Model: 74 AC2708, Integrated Device Technologies, Santa Clara, CA). To initiate an experiment, a start signal is sent to the timing electronics and is synchronized with the master clock. The first word pair is shifted out of the Control FIFO and interpreted to produce the desired sampling sequence. Upon completion of the first set of instruction words, the next set of words is immediately shifted out and interpreted. This process continues until all the word pairs have been shifted out of the Control FIFO. The computer senses when the Control FIFO is empty and then reads back the sample data stored in the 92 individual channel Sample FIFOs (see Fig. 2). The timing electronics operate on a 10 ms clock and the actual period between ADC samples is a constant 10 ms. The effective sampling rate (in multiples of 10 ms) is obtained by storing the ADC
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
169
Fig. 6. Experiment data sampling control electronics. Control words from the PC are deciphered to define the sampling rate and the corresponding number of samples.
output when desired. This eliminates potential transient problems that may arise if the ADC sampling frequency is changed during an experiment. Sampling periods from 10 ms to 20.97 s can be selected and the number of samples set from 1 to 512. The selected sampling period is encoded by 21-bits (bits 1–15 of the first and bits 1–6 of the second words) and the number of samples, for each sampling rate, is specified by 9-bits (bits 7–15 of the second word). Additional word pairs are used to allow collection of up to 1024 samples in a highly flexible timing schedule. A series of registers, counters, and comparators are the main components used in this sub-system. The sampling control electronics also produce the ‘capacitor clear’ pulses that reset the integrator capacitors within the front end electronics (see above). The time interval between each ADC convert and ‘capacitor clear’ pulse is precise, repeatable, and consistent throughout an experiment.
2.8. Conversion electronics The conversion electronics perform the tasks of sampling the analog output voltage signals produced by the front end integrator circuits and storing the sample data. Each integrator output has a dedicated 12-bit ADC whose output is selectively stored in a 1024 word FIFO. Fig. 2 illustrates the major components of the conversion electronics and the connection to the front end electronics. There are 92 of these circuits. An ADC convert signal is sent simultaneously to each ADC to initiate conversion. The ADC store signal is sent simultaneously to each Sample FIFO to write the corresponding ADC output samples into memory. The time interval between ADC store pulses is user-
170
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
specified with the PC. Note that the ADC convert signal, as shown in Fig. 3, occurs 0.2 ms prior to the ‘capacitor clear’ signal. The 12 bit ADCs (Model: AD1678, Analog Devices, Norwood, MA) can operate at a maximum frequency of 200 kHz. The minimum sampling period of this instrument is 10 ms. The Sample FIFOs (Model: 7202, Integrated Device Technologies Inc., Santa Clara, CA) are 9 bits wide by 1024 words deep. Two of these devices operate in parallel to provide 12 bits of storage. These devices contain tristate buffer outputs permitting the outputs from each channel to be connected to a common data bus. Physically, the conversion electronics are located on twelve identical PC cards. Each card contains 8 conversion circuits for a total of 96 channels of which 92 are used and 4 are spares.
2.9. Readout electronics After each experiment the Sample FIFOs contain a user specified number of sample data words. The readout electronics enable the PC to read the 12 bit data words from the FIFOs. Data word retrieval is accomplished with a multiplex scheme that correlates a coded 7-bit binary word with each of the 92 individual channels. The coded word is produced by the computer and specifies which channel will provide the data. A read pulse is sent from the computer to the desired channel causing a data word to be placed on the data bus. This process is repeated until the FIFO raises an empty flag. The computer senses the empty flag and modifies the code word to read the next channel. This process is repeated for each channel. In the computer, the data words are placed into a data array which is used for subsequent analysis and processing.
2.10. Software The custom software developed for controlling and monitoring of the entire system was written in Power Basic (Spectral Publishing Co., Sunnyvale, CA). The software interfaces the user, through the plug-in cards, to the instrument. A set-up screen is used to enter the desired timing parameters for an experiment. A maximum of 32 entries may be used. The first number defines the number of samples, while the second number specifies the rate at which these samples are to be taken. One complete entry creates one word pair which is then written to the Control FIFO located in the timing electronics. After the experiment’s parameters have been entered, the user initiates the experiment. The instrument then returns an output screen which shows the transmittance measurements made by all 92 channels (i.e. at all wavelengths) at one time sample. The user may scroll the spectrum transmittance measurements through each of the separate time samples. However, it is frequently more useful to view a single channel output over the entire experiment to show the kinetics of the reaction at a particular wavelength.
2.11. Calibration Since each channel has a slightly different gain, offset, and dark current, provision for calibration is essential. Use of nearly 200 trimming potentiometers is impractical for several reasons, including the impossibility of maintaining the monitoring light source
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
171
sufficiently stable for the protracted time needed to adjust all the potentiometers. Accordingly, the computer is calibrated every day in the following way. Each diode is calibrated independently using two neutral density filters of 0.1 and 1.0 A units. The absorption spectra for these two filters over the entire wavelength range of interest is first obtained using a separate calibrated spectrophotometer. Each of the spectra is then fitted to parameters of an empirical equation, which is included in a calibration module of the software for the spectrophotometer. In this way, the exact absorbance of each of the two calibration filters is known at each wavelength for each of the photodiodes. With this information, the raw signals in the computer at the two known absorbance values are used to compute a straight line from which the dark current value and the factors to convert computer values to a corresponding absorbance for each signal are available. This technique is used instead of using total darkness to obtain the dark current value because total darkness would result, on some channels, in slightly negative outputs which cannot be read by the unipolar ADCs.
3. Performance The system as described is sensitive in the visible portion of the spectrum and is limited at the ultra-violet end to 380 nm by the detector sensitivity and at the infra-red end to 900 nm by the grating sensitivity. Two independent 130 nm regions may be selected by each monochromator with a dispersion of 2.8 nm per detector element. Full scale (4096 counts) readings, at the 10 ms sampling interval, are obtained at 650 nm with 121 mW of light incident on each element. By accounting for the grating and mirror reflectivities, this corresponds to 4.5 mW incident on the monochromator slit for each 130 nm band. The dark noise of the complete system, i.e. photodiode through the computer, evaluated with a comprehensive set of sampling intervals, is 2 counts peak to peak. The system was first used to study the kinetics of transfer of four electrons from cytochrome c to the two heme and two copper centers of oxidized mammalian cytochrome oxidase [3]. The first electron is bound with a time constant (t ) of 5 ms and is shared equally between heme a and CuA . The subsequent binding and internal transfer events involving the other three electrons and the heme a3 and Cu B centers are controlled by the level of O 2 present in the system. In a separate study, the kinetics of the bacteriorhodopsin photocycle were examined [4]. It was found that there are two separate decay routes for the M photocycle intermediate. A fast form of M with a t of | 2 ms decays through the O intermediate, whereas a slower form with a t of 6 ms decays directly back to the ground state. Fig. 7 shows difference spectra obtained for the decay of the fast (top panel) and slow (bottom panel) forms of the M intermediate obtained from spectra acquired with the new spectrometer and analyzed by the SVD deconvolution procedure as described previously [2]. Both panels show the disappearance of an M intermediate (trough at | 412 nm). The top panel shows the simultaneous appearance of the O intermediate (peak at | 640 nm), whereas the bottom panel shows that the decay of the slow M intermediate is concomitant with the appearance of bacteriorhodopsin (peak at | 570 nm). Also shown in the bottom panel, is the decay of O formed in the
172
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Fig. 7. Difference spectra for the decays of the fast (top panel) and slow (bottom panel) forms of the M intermediate in the bacteriorhodopsin photocycle. The vertical dotted lines are drawn at 412, 570, and 640 nm. These spectra were generated from 520 raw spectra acquired with the instrument described in this paper and analyzed by the SVD procedure previously described [2]. Note that the disappearance of the fast M intermediate ( | 412 nm) is compensated by the appearance the O intermediate ( | 640 nm) whereas the disappearance of the slow M intermediate is compensated by the appearance of the ground state, bacteriorhodopsin ( | 570 nm). The bottom panel also records the decay of O intermediate formed in the earlier kinetic phase.
earlier kinetic phase. Quantification of the difference spectra reveal that fast M accounted for 54%, and slow M for 46% of the total M in the single turnover experiment shown. Subsequent studies with the bacteriorhodopsin system [5] showed that the ability of actinic light to control the operation and kinetics of the photocycle depend on the interaction of particular membrane lipids with the bacteriorhodopsin protein. The instrument was also used to define the kinetics and reaction pathways in the oxidation of fully reduced cytochrome oxidase by O 2 [6]. Four kinetic events were resolved with t ’s of | 0.01, | 0.09, | 1.2, and | 30 ms. It was further found that rather than a linear sequence of electron transfer events, as commonly believed, there is a branching after
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
173
the first electron transfer event, followed by the final oxidation of each of the intermediates in the two arms of the branches leading to the formation of the fully oxidized product.
4. Conclusions A rapid scanning, 92 channel spectrophotometer system 3 has been designed and built to obtain time-resolved absorption spectra of various chemical substances undergoing rapid changes. The system uses two independently adjustable 46 element photodiode arrays allowing simultaneous measurements over two wavelength regions. Each photodiode output is independently amplified, integrated, measured via an ADC, and digitally stored. The system offers 12-bit resolution and has a maximum sample rate of 100 000 samples per second. The instrument has successfully collected data for analysis of the fast kinetic changes occurring in two biochemical energy-transducing enzyme systems.
4.1. Simplified description of the method and ( future) applications The instrument described in this paper has been used successfully to examine the rapid kinetics of electron transfer from cytochrome c to the heme and copper redox centers of cytochrome aa3 [3] and then the movement of electrons from the reduced metal centers to O 2 [6]. It has also been used to examine the kinetics and pathways of the bacteriorhodopsin photocycle [4] and the role of membrane lipids in controlling the cycle [5]. There are several possible modifications that may improve the performance and sensitivity of the instrument. First, the thermal leakage current produced by the photodiode elements could be reduced with the addition of a photodiode cooling system. Product data sheets show that the thermal leakage current approximately halves for every 8 to 10 degrees Celsius temperature decrease. Peltier cooling devices are currently being tested as a means for decreasing the background noise of the photodiodes. With a sufficiently lowered background noise it is reasonable to substantially increase the dynamic range of the instrument by adding the capabilities of offset and gain to each channel and / or the use of 16-bit A / D converters. A possible method for increasing the sensitivity of the system is the use of an image intensifier mounted on the face of the photodiode array. Photons impinging onto the intensifier are amplified by creating several output photons for each input photon. The increased number of output photons are then directed onto the photodiode array creating a reasonable output signal for a very small incident light signal. This offers the advantage that fewer photons are required to study a particular sample which in many cases may respond differently to different light intensities.
3
Detailed electronic circuits are available upon request to T.J. Pohida, DCRT, NIH, Building 12, Room 2033, 12 South Drive, Bethesda, MD 20892.
174
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Acknowledgements The authors gratefully acknowledge the contributions of J. Fessler and R. Solomon for electrical fabrication and K. Greenway and J. Sullivan for mechanical fabrication.
References [1] Shrager RI, Hendler RW. Titration of individual components in a mixture with resolution of difference spectra, pKs, and redox transitions. Anal Chem 1982;54:1147–52. [2] Hendler RW, Shrager RI. Deconvolutions based on singular value decomposition and the pseudoinverse: a guide for beginners. J Biochem Biophys Methods 1984;28:1–33. [3] Hendler RW, Bose S, Shrager RI. Multiwavelength analysis of the kinetics of reduction of cytochrome aa3 by cytochrome c. Biophys J 1993;65:1307–17. [4] Hendler RW, Dancshazy Zs, Bose S, Shrager RI, Tokaji Zs. Influence of excitation energy on the bacteriorhodopsin photocycle. Biochemistry 1994;33:4604–10. [5] Mukhopadhyay AK, Dracheva S, Bose S, Hendler RW. Control of the integral membrane proton pump, bacteriorhodopsin by purple membrane lipids of Halobacterium halobium. Biochemistry 1996;35:9245– 52. [6] Bose S, Hendler RW, Shrager RI, Chan SI, Smith PD. Multichannel analysis of single-turnover kinetics of cytochrome aa3 reduction by O 2 . Biochemistry 1997;36:2439–24497.
A high speed optical multichannel analyzer a, b a 1 ,c J.W. Cole *, R.W. Hendler , P.D. Smith , H.A. Fredrickson , T.J. Pohida c , W.S. Friauf 2 ,a a National Center for Research Resources, National Institutes of Health, Bethesda MD 20892, USA National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda MD 20892, USA c Division of Computer Research and Technology, National Institutes of Health, Bethesda MD 20892, USA b
Received 17 July 1997; received in revised form 14 August 1997; accepted 15 August 1997
Abstract An optical multichannel analyzer capable of recording spectra at sampling rates up to 100 kHz is described. The instrument, designed to gather data on the kinetic reaction mechanisms of biological preparations such as cytochrome oxidase and bacteriorhodopsin, features a massively parallel approach in which each photosensing element of the detector array has a dedicated amplifier, integrator, analog to digital converter, and sample buffer. The design has 92 such elements divided in two separate arrays, each of which sits at the focal plane of a 1 / 4 m Ebert spectrometer. The spectrometers may be tuned to cover independent, 130 nm wide, regions of the spectrum from 350 nm to 900 nm with a dispersion of 2.8 nm per element. Each detection channel has 12-bit resolution with an electronic dark count of 1 count and may be sampled 1024 times during a single experiment with dynamically variable sampling intervals from 10 ms to several seconds. Time averaging of up to thousands of consecutive laser-initiated kinetic cycles allows analyses of spectral changes , 0.001 optical density units. A personal computer with custom software provides a number of features: entry of experiment parameters; transfer of data from temporary buffers to permanent files; real time display; multiple spectrum averaging; and control and synchronization of associated system hardware. Optical fibers or lenses provide coupling from a parabolic reflector Xenon arc monitoring light source, through the sample chamber, to the entry slit of the monochromator. The instrument has been used for extensive studies on the rapid kinetics and definition of reaction sequences of the energy-transducing enzymes cytochrome oxidase and bacteriorhodopsin. Some results from these studies are discussed. 1997 Elsevier Science B.V. *Corresponding author. Address for correspondence: 9400 Wooden Bridge Rd, Potomac, MD 20854, USA. Present address: P.O. Box 68, Ashton, MD 20861, USA. 2 Present address: 5616 Oakmont Ave, Bethesda, MD 20817, USA. 1
0165-022X / 97 / $17.00 1997 Elsevier Science B.V. All rights reserved. PII S0165-022X( 97 )00037-7
162
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Keywords: Optical multichannel analyzer; Reaction mechanism; Cytochrome oxidase; Bacteriorhodopsin
1. Introduction Spectrophotometry is an important investigational technique used for studies of chemical and biological reactions. In optical spectroscopy, the absorption of light, at specific wavelengths through a reaction mixture, provides information about intermediates, products, and reaction pathways. One particular application of this technique is in the study of the rapid kinetics of chemical transformations. Usually, several chromophoric substances are involved in these transformations. As the reaction proceeds, spectral changes are produced by the transformations of these chromophores, each of which contributes to the absorbance changes at the monitored wavelength(s). To date, most instruments have been limited to the continuous monitoring of rapid kinetic reactions at only one or two wavelengths. With limited spectral information, deconvolution of the total observed absorbance change into the relative contribution of each chromophoric component present is virtually impossible. A few laboratories have tried to solve the problem by using a two flash technique. First, a rapid reaction is initiated by a laser flash. Secondly, a monitoring flash (usually Xenon) is used to obtain a spectrum at a single time point. In this procedure, each time point requires a separate enzyme sample in a separate experiment. It must be assumed that each time point has a completely reproducible zero time point. The instrument described here collects up 1024 sequential spectra on a single sample. The ability to obtain a timed sequence of complete spectra has become particularly important in recent years because of advances in the analysis of spectral deconvolution, which utilize all of the information in multichannel spectra and permit the isolation of the contribution(s) of each component [1,2]. This analysis was first applied in equilibrium studies where spectra are collected slowly. With the realization that the same approaches would be valuable in the analysis of rapid kinetic data, the need for an instrument to collect a sequence of entire spectra rapidly was apparent. The instrument described in this paper was designed and built to address this need. Many commercial spectrophotometer systems are currently available based on a variety of light transducer technologies including photomultiplier tubes (pmt), photodiodes, and charge-coupled arrays. These transducers produce an output signal that is proportional to the intensity of the incident dispersed light, which is either mechanically scanned across a pmt or exposed directly on to a series of adjacent transducers each of which correspond to an incident wavelength. The output signals are multiplexed into ‘sample and hold’ devices to obtain a time sequence of data points corresponding to the light intensity on specific transducer elements. Current commercial instruments are limited by the speed of the multiplexing and data readout electronics. A multiplexed ‘sample and hold’ approach reduces the time resolution of the instrument because of the excessive ‘wait time’ between successive readings of one specific transducer. The instrument described in this paper overcomes this problem by using parallel processing
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
163
techniques. In place of multiplexing, each individual transducer has a dedicated amplifier, integrator, analog to digital converter (ADC), and sample buffer. This design approach permits the instrument to simultaneously sample, using 12-bit ADCs, all 92 channels at a maximum rate of 100 000 samples per second. All parameters, such as the sampling rate and the number of samples at each specific rate, are programmable via a personal computer (PC). The 92 transducers are provided in two 46 element diode arrays, each placed at the focal plane of a separate spectrograph, which enables independent ranges of wavelength to be observed (130 nm as described herein). In contrast, existing commercial instrumentation, capable of comparable sampling rates, is limited by a fixed sampling rate, a lower number of samples per channel, and a smaller detector element size requiring greater than 100-fold the amount of light to be incident on the reaction mixture.
2. General system description The instrument consists of ten subsystems, shown in Fig. 1, which are described below.
2.1. Light sources This subsystem consists of two different light sources: one to monitor the progress of the reaction; and one to initiate the reaction (i.e. actinic). Each is directed into the
Fig. 1. System diagram. Each of the sub-systems is described in detail in the text. The dotted lines indicate optical coupling.
164
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
reaction chamber. The monitoring light source is a 75 watt continuous Xenon lamp (Model: LPS-220, Photon Technology International, South Brunswick, NJ). The intensity of this source is held constant by a light feedback regulator. The actinic light source is a pulsed laser (Surelite-10 ND/ YAG, Continuum Lasers, Santa Clara, CA) that emits a 5 ns pulse with a maximum energy of 150 mJ at 530 nm. It can be delivered either as a single shot or in a repetitive mode at a maximum rate of 8.3 Hz. The actinic light impinges on the reaction cell at a 908 angle relative to the monitoring light path.
2.2. The reaction chamber For rapid kinetic studies, it is essential that the reaction be started instantly and synchronized with the data acquisition. In some cases, a single reactant is energized to undergo a chemical change by a laser pulse (for example, the bacteriorhodopsin photocycle). In other cases, two reactants must be rapidly mixed together either for spontaneous reaction or for subsequent induction to react by a laser pulse (for example, the oxidation of carbon monoxide-liganded reduced cytochrome oxidase by O 2 ). In the latter case, the rapid mixing of the reactants requires the use of a stopped-flow apparatus. For this purpose, a Kintek Instruments (University Park, PA) stopped-flow apparatus was used.
2.3. Optics The optical subsystem consists of a series of lenses and / or optical fibers arranged to couple the monitoring and actinic light sources through the reaction chamber to the spectrographs. The monitoring light, after passing through the reaction chamber is split, by either a beam splitter for the bacteriorhodopsin experiments or a bifurcated optical fiber bundle for the stopped flow / cytochrome oxidase experiments The exiting light was focused on the input slits of two spectrographs. Each spectrograph disperses the white light received from the reaction chamber into 130 nm bands selected by spectrograph settings. The 1 / 4 meter spectrographs used in this application (Model 77200, Oriel Corporation, Stratford, CT) contain 1200 line / mm gratings (Model: 77232). A 46element photodiode array (described below) is mounted at the output port of each spectrograph providing a dispersion of approximately 2.8 nm per element.
2.4. Front end electronics The front end electronics consist of 92 independent circuits, each comprised of a photodiode, a current amplifier, and an integrator (Fig. 2). The photodiode current is not integrated directly because the total charge during the shorter intervals is not great enough relative to switching transients and other noise sources to provide an adequate signal-to-noise ratio. Also, the integrating capacitor needed to provide the desired output voltage would not be large enough relative to parasitic capacitances. Circuit component values were chosen to produce a final analog output signal voltage between 0 and 10 volts, corresponding to the input voltage range of the ADCs (conversion electronics). The photodiode arrays (Model: S4112-46) are produced by Hammamatsu Photonics
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
165
Fig. 2. Front end electronics and conversion electronics. The photocurrent from each element of the photodiode array is connected to an individual amplification and integration circuit. The integrated photodiode signal is digitized by a dedicated 12-bit ADC and stored by a 1024 deep Sample FIFO buffer.
(Hammamatsu City, Japan). Each element has a relatively large active area of 4.4 mm 3 0.9 mm which maximizes the critical optical coupling of the light onto the detector array. They also possess a broad spectral response curve, ranging from 200 to 1050 nm. The first stage operational amplifier (Model: AD549, Analog Devices, Norwood, MA), selected because its very low input bias current, is ideally suited for measuring small current values and fluctuations. A 2 ms-wide pulse signal with a programmable repetition period of 10 to 70 ms causes a temporary closure of the ‘capacitor clear’ switch which resets the integrator output to zero volts. Fig. 3 shows the relationship between the integrator output signal, the ADC convert signal, and the ‘capacitor clear’ signal. In this particular example, the ‘capacitor clear’ signal is set at a 10 ms repetition interval. The front end electronics are constructed on four identical 23-channel copper-clad boards. Each board receives 23 photodiode currents via a ribbon cable from the photodiode arrays. Two boards are required per photodiode array and are housed with the photodiode array and spectrograph. In summary, there are two front end assemblies, each containing one spectrograph, one photodiode array, and two front end boards.
2.5. Computer control and interface A PC with plug-in boards is the interface between the user and the custom electronic hardware sub-systems. A custom software package interprets user commands which are then relayed to the hardware through two Input / Output (I / 0) driver cards located within the PC. The first of these cards is a five channel counter timer board (CTM-05) and the second a 24-bit parallel digital I / 0 interface board (PI0-12), both manufactured by
166
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Fig. 3. Timing diagram indicates the temporal relationship between the front end integrator output, the capacitor clear signal and the ADC convert signal.
Keithley / Metrabyte (Tauton, MA). Each card sends and receives a number of signals to and from the hardware (Figs. 4 and 5). The CTM-05 outputs are: SHIFT IN Writes control word into Control FIFO (first-in, first-out memory) —see below. START Starts experiment. BUS CONTROL Dictates data flow direction on bus (16-bit control word in or 12-bit data word out). INIT Initializes interface with Sample FIFO —see below. RESET Resets all electronic hardware. CLEAR RATE Programs (3-bits) capacitor clear pulse repetition period (10–70 ms). The CTM-05 inputs are: INPUT READY Control FIFO ready for Control word. CONTROL FIFO EMPTY Control FIFO is empty. SAMPLE FIFO EMPTY Sample FIFO is empty. The P10-12 outputs are: CONTROL WORD Data word pairs used by timing electronics (16-bits).
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
167
Fig. 4. PC control interface. The bus control line from the PC CTM- 05 board dictates whether the PC will be sending 16-bit control words or receiving data from the Sample FIFOs.
Fig. 5. Experiment synchronization electronics. Adjustable delays allow the A / D convert signal to be synchronized with the other experiment signals (laser and stopped flow instrumentation).
168
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
ADDRESS LINES Select one channel for sample read (7-bits). READ Read strobe for Sample FIFOs. The P10–12 inputs are: DATA 12-bit sample words from Sample FIFOs. It is important to note that the PI0–12 functions in a dual capacity: it sends 16-bit timing word pairs (discussed below) and receives the 12-bit measurement data words. In order to perform these functions using the same I / 0 port, a bus arbitration interface was developed. A dedicated output from the CTM-05 dictates the direction of data transfer on the bus. A ‘0’ indicates that the 16-bit word pairs will be written to the measurement hardware while a ‘1’ indicates that the 12-bit data words will be read from the hardware.
2.6. Experiment synchronization electronics The experiment synchronization electronics allow the user to control various timing parameters such as those associated with the laser, stopped-flow apparatus, and measurement systems. A block diagram of the experiment synchronization electronics is shown in Fig. 5. The electronics have been designed to allow a variety of different experiments to be performed by simply changing switch settings. Either single experiments or experiments that average many spectra may be performed. During operation, the laser requires an 8.3 Hz TTL pulse train to activate the flashlamp. A single 100 ms TTL pulse, which is synchronized to the flashlamp pulse train, is required to fire the laser. All other signals required for the experiment, including those which activate the stopped-flow apparatus and initiate data collection, are provided prior to the initiation of the laser pulse.
2.7. Timing electronics The data sampling control electronics sequentially interpret control word pairs previously sent from the computer and produce a user-defined data acquisition sequence. The word pairs encode both the sampling rate and the number of samples desired at that rate, which the user enters via a PC. Fig. 6 schematically illustrates the timing electronics. The words written by the PC are 16-bits wide and are sent in pairs. The word pairs are stored until the beginning of an experiment in a 16-bit, 64 word FIFO (Model: 74 AC2708, Integrated Device Technologies, Santa Clara, CA). To initiate an experiment, a start signal is sent to the timing electronics and is synchronized with the master clock. The first word pair is shifted out of the Control FIFO and interpreted to produce the desired sampling sequence. Upon completion of the first set of instruction words, the next set of words is immediately shifted out and interpreted. This process continues until all the word pairs have been shifted out of the Control FIFO. The computer senses when the Control FIFO is empty and then reads back the sample data stored in the 92 individual channel Sample FIFOs (see Fig. 2). The timing electronics operate on a 10 ms clock and the actual period between ADC samples is a constant 10 ms. The effective sampling rate (in multiples of 10 ms) is obtained by storing the ADC
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
169
Fig. 6. Experiment data sampling control electronics. Control words from the PC are deciphered to define the sampling rate and the corresponding number of samples.
output when desired. This eliminates potential transient problems that may arise if the ADC sampling frequency is changed during an experiment. Sampling periods from 10 ms to 20.97 s can be selected and the number of samples set from 1 to 512. The selected sampling period is encoded by 21-bits (bits 1–15 of the first and bits 1–6 of the second words) and the number of samples, for each sampling rate, is specified by 9-bits (bits 7–15 of the second word). Additional word pairs are used to allow collection of up to 1024 samples in a highly flexible timing schedule. A series of registers, counters, and comparators are the main components used in this sub-system. The sampling control electronics also produce the ‘capacitor clear’ pulses that reset the integrator capacitors within the front end electronics (see above). The time interval between each ADC convert and ‘capacitor clear’ pulse is precise, repeatable, and consistent throughout an experiment.
2.8. Conversion electronics The conversion electronics perform the tasks of sampling the analog output voltage signals produced by the front end integrator circuits and storing the sample data. Each integrator output has a dedicated 12-bit ADC whose output is selectively stored in a 1024 word FIFO. Fig. 2 illustrates the major components of the conversion electronics and the connection to the front end electronics. There are 92 of these circuits. An ADC convert signal is sent simultaneously to each ADC to initiate conversion. The ADC store signal is sent simultaneously to each Sample FIFO to write the corresponding ADC output samples into memory. The time interval between ADC store pulses is user-
170
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
specified with the PC. Note that the ADC convert signal, as shown in Fig. 3, occurs 0.2 ms prior to the ‘capacitor clear’ signal. The 12 bit ADCs (Model: AD1678, Analog Devices, Norwood, MA) can operate at a maximum frequency of 200 kHz. The minimum sampling period of this instrument is 10 ms. The Sample FIFOs (Model: 7202, Integrated Device Technologies Inc., Santa Clara, CA) are 9 bits wide by 1024 words deep. Two of these devices operate in parallel to provide 12 bits of storage. These devices contain tristate buffer outputs permitting the outputs from each channel to be connected to a common data bus. Physically, the conversion electronics are located on twelve identical PC cards. Each card contains 8 conversion circuits for a total of 96 channels of which 92 are used and 4 are spares.
2.9. Readout electronics After each experiment the Sample FIFOs contain a user specified number of sample data words. The readout electronics enable the PC to read the 12 bit data words from the FIFOs. Data word retrieval is accomplished with a multiplex scheme that correlates a coded 7-bit binary word with each of the 92 individual channels. The coded word is produced by the computer and specifies which channel will provide the data. A read pulse is sent from the computer to the desired channel causing a data word to be placed on the data bus. This process is repeated until the FIFO raises an empty flag. The computer senses the empty flag and modifies the code word to read the next channel. This process is repeated for each channel. In the computer, the data words are placed into a data array which is used for subsequent analysis and processing.
2.10. Software The custom software developed for controlling and monitoring of the entire system was written in Power Basic (Spectral Publishing Co., Sunnyvale, CA). The software interfaces the user, through the plug-in cards, to the instrument. A set-up screen is used to enter the desired timing parameters for an experiment. A maximum of 32 entries may be used. The first number defines the number of samples, while the second number specifies the rate at which these samples are to be taken. One complete entry creates one word pair which is then written to the Control FIFO located in the timing electronics. After the experiment’s parameters have been entered, the user initiates the experiment. The instrument then returns an output screen which shows the transmittance measurements made by all 92 channels (i.e. at all wavelengths) at one time sample. The user may scroll the spectrum transmittance measurements through each of the separate time samples. However, it is frequently more useful to view a single channel output over the entire experiment to show the kinetics of the reaction at a particular wavelength.
2.11. Calibration Since each channel has a slightly different gain, offset, and dark current, provision for calibration is essential. Use of nearly 200 trimming potentiometers is impractical for several reasons, including the impossibility of maintaining the monitoring light source
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
171
sufficiently stable for the protracted time needed to adjust all the potentiometers. Accordingly, the computer is calibrated every day in the following way. Each diode is calibrated independently using two neutral density filters of 0.1 and 1.0 A units. The absorption spectra for these two filters over the entire wavelength range of interest is first obtained using a separate calibrated spectrophotometer. Each of the spectra is then fitted to parameters of an empirical equation, which is included in a calibration module of the software for the spectrophotometer. In this way, the exact absorbance of each of the two calibration filters is known at each wavelength for each of the photodiodes. With this information, the raw signals in the computer at the two known absorbance values are used to compute a straight line from which the dark current value and the factors to convert computer values to a corresponding absorbance for each signal are available. This technique is used instead of using total darkness to obtain the dark current value because total darkness would result, on some channels, in slightly negative outputs which cannot be read by the unipolar ADCs.
3. Performance The system as described is sensitive in the visible portion of the spectrum and is limited at the ultra-violet end to 380 nm by the detector sensitivity and at the infra-red end to 900 nm by the grating sensitivity. Two independent 130 nm regions may be selected by each monochromator with a dispersion of 2.8 nm per detector element. Full scale (4096 counts) readings, at the 10 ms sampling interval, are obtained at 650 nm with 121 mW of light incident on each element. By accounting for the grating and mirror reflectivities, this corresponds to 4.5 mW incident on the monochromator slit for each 130 nm band. The dark noise of the complete system, i.e. photodiode through the computer, evaluated with a comprehensive set of sampling intervals, is 2 counts peak to peak. The system was first used to study the kinetics of transfer of four electrons from cytochrome c to the two heme and two copper centers of oxidized mammalian cytochrome oxidase [3]. The first electron is bound with a time constant (t ) of 5 ms and is shared equally between heme a and CuA . The subsequent binding and internal transfer events involving the other three electrons and the heme a3 and Cu B centers are controlled by the level of O 2 present in the system. In a separate study, the kinetics of the bacteriorhodopsin photocycle were examined [4]. It was found that there are two separate decay routes for the M photocycle intermediate. A fast form of M with a t of | 2 ms decays through the O intermediate, whereas a slower form with a t of 6 ms decays directly back to the ground state. Fig. 7 shows difference spectra obtained for the decay of the fast (top panel) and slow (bottom panel) forms of the M intermediate obtained from spectra acquired with the new spectrometer and analyzed by the SVD deconvolution procedure as described previously [2]. Both panels show the disappearance of an M intermediate (trough at | 412 nm). The top panel shows the simultaneous appearance of the O intermediate (peak at | 640 nm), whereas the bottom panel shows that the decay of the slow M intermediate is concomitant with the appearance of bacteriorhodopsin (peak at | 570 nm). Also shown in the bottom panel, is the decay of O formed in the
172
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Fig. 7. Difference spectra for the decays of the fast (top panel) and slow (bottom panel) forms of the M intermediate in the bacteriorhodopsin photocycle. The vertical dotted lines are drawn at 412, 570, and 640 nm. These spectra were generated from 520 raw spectra acquired with the instrument described in this paper and analyzed by the SVD procedure previously described [2]. Note that the disappearance of the fast M intermediate ( | 412 nm) is compensated by the appearance the O intermediate ( | 640 nm) whereas the disappearance of the slow M intermediate is compensated by the appearance of the ground state, bacteriorhodopsin ( | 570 nm). The bottom panel also records the decay of O intermediate formed in the earlier kinetic phase.
earlier kinetic phase. Quantification of the difference spectra reveal that fast M accounted for 54%, and slow M for 46% of the total M in the single turnover experiment shown. Subsequent studies with the bacteriorhodopsin system [5] showed that the ability of actinic light to control the operation and kinetics of the photocycle depend on the interaction of particular membrane lipids with the bacteriorhodopsin protein. The instrument was also used to define the kinetics and reaction pathways in the oxidation of fully reduced cytochrome oxidase by O 2 [6]. Four kinetic events were resolved with t ’s of | 0.01, | 0.09, | 1.2, and | 30 ms. It was further found that rather than a linear sequence of electron transfer events, as commonly believed, there is a branching after
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
173
the first electron transfer event, followed by the final oxidation of each of the intermediates in the two arms of the branches leading to the formation of the fully oxidized product.
4. Conclusions A rapid scanning, 92 channel spectrophotometer system 3 has been designed and built to obtain time-resolved absorption spectra of various chemical substances undergoing rapid changes. The system uses two independently adjustable 46 element photodiode arrays allowing simultaneous measurements over two wavelength regions. Each photodiode output is independently amplified, integrated, measured via an ADC, and digitally stored. The system offers 12-bit resolution and has a maximum sample rate of 100 000 samples per second. The instrument has successfully collected data for analysis of the fast kinetic changes occurring in two biochemical energy-transducing enzyme systems.
4.1. Simplified description of the method and ( future) applications The instrument described in this paper has been used successfully to examine the rapid kinetics of electron transfer from cytochrome c to the heme and copper redox centers of cytochrome aa3 [3] and then the movement of electrons from the reduced metal centers to O 2 [6]. It has also been used to examine the kinetics and pathways of the bacteriorhodopsin photocycle [4] and the role of membrane lipids in controlling the cycle [5]. There are several possible modifications that may improve the performance and sensitivity of the instrument. First, the thermal leakage current produced by the photodiode elements could be reduced with the addition of a photodiode cooling system. Product data sheets show that the thermal leakage current approximately halves for every 8 to 10 degrees Celsius temperature decrease. Peltier cooling devices are currently being tested as a means for decreasing the background noise of the photodiodes. With a sufficiently lowered background noise it is reasonable to substantially increase the dynamic range of the instrument by adding the capabilities of offset and gain to each channel and / or the use of 16-bit A / D converters. A possible method for increasing the sensitivity of the system is the use of an image intensifier mounted on the face of the photodiode array. Photons impinging onto the intensifier are amplified by creating several output photons for each input photon. The increased number of output photons are then directed onto the photodiode array creating a reasonable output signal for a very small incident light signal. This offers the advantage that fewer photons are required to study a particular sample which in many cases may respond differently to different light intensities.
3
Detailed electronic circuits are available upon request to T.J. Pohida, DCRT, NIH, Building 12, Room 2033, 12 South Drive, Bethesda, MD 20892.
174
J.W. Cole et al. / J. Biochem. Biophys. Methods 35 (1997) 161 – 174
Acknowledgements The authors gratefully acknowledge the contributions of J. Fessler and R. Solomon for electrical fabrication and K. Greenway and J. Sullivan for mechanical fabrication.
References [1] Shrager RI, Hendler RW. Titration of individual components in a mixture with resolution of difference spectra, pKs, and redox transitions. Anal Chem 1982;54:1147–52. [2] Hendler RW, Shrager RI. Deconvolutions based on singular value decomposition and the pseudoinverse: a guide for beginners. J Biochem Biophys Methods 1984;28:1–33. [3] Hendler RW, Bose S, Shrager RI. Multiwavelength analysis of the kinetics of reduction of cytochrome aa3 by cytochrome c. Biophys J 1993;65:1307–17. [4] Hendler RW, Dancshazy Zs, Bose S, Shrager RI, Tokaji Zs. Influence of excitation energy on the bacteriorhodopsin photocycle. Biochemistry 1994;33:4604–10. [5] Mukhopadhyay AK, Dracheva S, Bose S, Hendler RW. Control of the integral membrane proton pump, bacteriorhodopsin by purple membrane lipids of Halobacterium halobium. Biochemistry 1996;35:9245– 52. [6] Bose S, Hendler RW, Shrager RI, Chan SI, Smith PD. Multichannel analysis of single-turnover kinetics of cytochrome aa3 reduction by O 2 . Biochemistry 1997;36:2439–24497.