1632
Anal. Chem. 1908, 60. 1632-1635
As the final step of the procedure an addition of hydroxylamine hydrochloride, NH,OH.HCl (in a solution of EDTA) was included in order to minimize any oxidation of diaminonaphthalene (DAN), which served as the complexing agent for Se(IV) in the subsequent fluorometric determination. Although the addition of NHzOH has also been recommended by other workers who use the DAN-fluorometric method (2, 3), it appears to be rather doubtful advice, as this agent is known to be an effective reductant for Se(IV), resulting in the formation of elemental selenium. As such, it has been commonly used in gravimetric methods for selenium and its reduction properties on Se(1V) have been discussed elsewhere ( 4 ) . It is obvious that if NHzOH would reduce any Se(1V) to the elemental state, this fraction would become unavailable for further complex formation with DAN or for determination by other spectrometric techniques as well. In order to elucidate the potential problem, it was decided to examine the effect of hydroxylamine in the present procedure in some detail. The results of the study are presented in this paper. EXPERIMENTAL SECTION For details of the digestion procedure ref 1should be consulted. Experiment 1. Wheat corn (0.200 g) plus 0.150 g of selenourea, (NH,),CSe ( E N Pharmaceuticals, Inc.), was taken through the given procedure, resulting in the predicted formation of Mn04-, indicating complete oxidation. After the addition of 6 M HC1 and the subsequent heating and cooling, 8 mL of the 0.04 M EDTA/10% NH,OH-HCl solution was added, and the development of color due to the formation of elemental selenium was observed (see "Results"). Experiment 2. Two samples of wheat corn, each of 0.300 g, were dissolved according to the recommended procedure, and after final cooling, 1.6 pg of Se(1V) (in 200 pL) was added to each solution. To one of the solutions the recommended 8 mL of EDTA/NH20Hsolution was added immediately after the cooling, whereas for the other solution, this addition was done after the elapse of 1 h. Both solutions were thereafter (1h) transferred to 100-mLvolumetric flasks, which were filled to volume. The solutions were immediately analyzed for selenium by hydride generation atomic absorption spectrometry using an automatic hydride generator (P.S. Analytical, U.K.) and an atomic absorption spectrometer (Perkin-Elmer Model 300). R E S U L T S A N D DISCUSSION Experiment 1. The large amount of selenium present allowed a visual observation of elemental selenium. After the addition of EDTA/NH,OH, the solution became yellow, orange, and at last (within 10 min) definitely red owing to the formation of elemental (red) selenium. This clearly demonstrated that the suspected reaction had taken place. Experiment 2. The absorbance signal for the solution to which hydroxylamine was added just before measuring was
taken as 100%. The signals for the solution to which the hydroxylamine was added 1h before measuring were recorded as 80% of the former signal. This also indicated that the unwanted reaction had, to some extent, taken place. General Discussion. The reduction of Se(IV) to elemental selenium by hydroxylamine is not a very fast reaction, as indicated in experiment 1. Thus, the success of the method seems to depend on an immediate addition of the complexing agent (DAN) after the addition of hydroxylamine. By that means the selenium will probably be completely complexed. This may also explain the lack of interference from hydroxylamine with the DAN-selenite reaction reported earlier (1, 2). Moreover, acidity may also influence the sample solution, since both the precipitation rate and the rate of the complex formation with DAN are pH-dependent. The rate of precipitation of elemental selenium is also probably even slower when trace concentrations of selenium are concerned, as indicated in experiment 2. However, it is recommended that the addition of hydroxylamine should be omitted, thus making the procedure even more attractive. With fluorometric methods the claimed purpose with the addition was to minimize oxidation of DAN by residual HN03. But the amount of H N 0 3 in the final solution is probably very minute and the DAN is always added in quite large excess, so this precaution is perhaps be unnecessary. Any excess of H N 0 3 may also be destroyed by addition of formic acid (3),as also mentioned by the authors. For other determination techniques, such as atomic absorption spectrometry and inductively coupled plasma, the digestion method should be equally well suited; moreover, the presence of hydroxylamine will not serve any purpose with these techniques and can certainly be omitted. The digestion method itself appeared to be very convenient. The use Mn(I1) as a redox indicator seems to be a nice solution to a problem common with many digestion procedures, i.e. to decide when the oxidation of the sample is completed. Registry No. Se, 7782-49-2; "OB, 7697-37-2; H3P04,766438-2; HzOz,7722-84-1;",OH, 7803-49-8; seleno urea, 630-10-4. LITERATURE C I T E D (1) Dong, A.; Rendig, V. V.; Burau, R. G.; Besga, G. S. Anal. Chem. 1887, 59, 2730-2732. (2) Cukor, P.; LOR, P. F. J . Phys. Chem. lQ65, 69, 3232-3239. (3) Reamer, D. C.; Veillon, C. Anal. Chem. lQ83, 55. 1605-1606. (4) Bye, R. Talanfa IQS3, 3 0 , 993-996.
R a g n a r Bye Department of Chemistry Agricultural University of Norway 1432 Aas-NLH, Norway
RECEIVED for review December 29,1987. Accepted March 25, 1988.
Characterization of Unseparated Nucleic Acid Restriction Enzyme Fragments by Electric Birefringence Frequency Dispersion Sir: Electric birefringence of nucleic acids in gels is a common tool for investigating the dynamics of internal motions and the mechanism of electrophoretic transport of the polymers (1-3). Classically, interest has centered on large polymers, but recently several investigators have reported the behavior of small (50-25000 bp (base pair)) monodisperse fragments (4-6). Monodisperse fragments oriented in an electric field usually obey an exponential relaxation relation,
which can be related to the electrophoretic mobility (6, 7). Electric birefringence, and less commonly, relaxation of fluorescence polarization (8-11), are convenient methods for testing such theories. Electric birefringence measurements are most commonly made in the time domain. The sample is oriented in a pulsed electric field, and relaxation constants are evaluated from the time course of the birefringence decay. A few workers have
0003-2700/88/0360-1632$01.50/00 1988 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 60, NO. 15, AUGUST 1, 1988
employed the frequency dependence of electric birefringence to extract the same information (12,13). Birefringence decay is due to randomization of chain orientation when an aligning field is removed. For monodispersed fragments in a agarose gel the decay is approximately exponential. Two theories relate birefringence decay to chain length. Equation 1 describes (10) the relaxation of a semirigid rod: 7
= 6rqL2P/kT
I Laser
(1)
= 6*qL3/kT
Lens
Polarizer
Analyzing Polarizer
II
where q is solvent viscosity, L is chain length, P is persistence length, Iz is Boltzmann's constant, and T is temperature. Reptation theory (14) predicts a stronger dependence on chain length, as shown in eq 2. Stellwagen has observed an 7
1633
, I
Detector
Cell
I
Signal Power Lock-in Generator Amplifier Amplifier Flgure 1. Apparatus for swept frequency birefringence measurements.
(2)
L2.8dependence for fragments in the range 600-3000 bp, in approximate agreement with eq 2 (5). If a gel contains a sufficiently dilute solution of DNA fragments of different sizes, fragment orientation and relaxation in an electric field will proceed independently. Under these conditions, the observed relaxation will be a simple sum of the exponential relaxations of the individual fragments. In the frequency domain, relaxation dispersion will have several characteristic frequencies, each corresponding to a different fragment. It follows that measurement of the multiple relaxation times or frequencies provides the s e e mobility and, therefore, size information as agarose gel electrophoresis. However, because no physical separation is required, measurement of relaxation times or frequencies of a mixture of fragments should be inherently faster than electrophoresis of the same mixture. T o the extent that relaxation is a n exponential decay, its governing equations are formally the same as those of a lowpass resistance-capacitance (RC) filter (15).Although orientation depends on the direction of the aligning field, the measured parameter, birefringence, depends only on the magnitude. If a sinusoidal electric field drives the orientation, the resulting birefringence will follow the magnitude of the field and will appear the same on both the positive and negative half-cycles. This behavior is analogous to passage through a full-wave rectifier. In particular, the birefringence will have a component at the second harmonic of the driving frequency (12,13). A further complication is that birefringence is proportional to the square of electric field strength. The rectified low-pass filter response must be squared to describe this effect. Using these concepts, we can describe the birefringence response to a cosinusoidal electric field, V = cos ( u t ) ,by eq 3. AN=
1 (1 + w272)1/2XlK[
5(
2 cos 2(ot - $9)
1-
3
-
2 cos 4(wt - $9) 15 cp
= arctan
(-UT)
(34
In eq 3 and 3a is the birefringence, w is the frequency, X is the wavelength, 1 is optical path length, K is the Kerr constant, V is the alternating current (AC) voltage, t is time, and 9 is the phase angle of the response. If the frequency of the driving cosine wave is swept measurement of birefringence with a lock-in amplifier referenced to the second harmonic of the driving field and with a 45' phase angle will produce the waveform of eq 4.
K = -8XlKV 9T
,
Equation 4 maximizes at w = 0.9217. On a log frequency scale the full width at half-maximum (fwhm) is 0.7 decade. In practice, the swept frequency technique is limited to the range of operating frequencies of the lock-in amplifier. This range is usually about 5 decades and commonly extends from about 1 Hz to about 100 kHz. Heterodyning or other special techniques are required to extend the operating range (16). In principle, measurements can be made by orienting the molecules in any external field, although an electric field will usually be the most practical choice. Electric birefringence is a convenient experimental probe, but electric dichroism, fluorescence polarization, or other orientation-sensitive phenomena could also be employed.
EXPERIMENTAL SECTION Phage lambda Hind I11 restriction fragments (Bethesda Research Laboratories) were dissolved in 0.25-2 % agarose gels (SeaPlaque, FMC Marine Colloids) and buffered with 0.04 M Tris acetate/O.OOl M EDTA (TAE) to pH 8. Samples were prepared by mixing solutions of buffered DNA fragments and agarose gels at 50 O C . Prior to use, the buffered fragments were heated in a water bath at 65 "C for 15 min to dissociate complexes between fragments and render all fragments available for analysis. The sample/agarose mixture was placed in a section of 1 or 2 mm square quartz capillary tube and allowed to gel in a refrigerator. The apparatus for swept frequency measurements is shown as Figure 1. The Kerr cell consisted of the capillary tube containing the sample gel and Teflon buffer reservoirs at either end. Each reservoir contained a platinum electrode,with a total l-cm spacing. Birefringence measurements were made as the increase in transmission through crossed linear polarizers. A 2-mW polarized He-Ne laser, with auxiliary Glan prism, functioned as the light source. A focusing lens was included for use with 1 mm cross section cells but omitted for 2-mm cells. Sheet Polaroid was used as the analyzing polarizer. A Hamamatsu 1P28 photomultiplier tube served as the detector. The oscillating electric field was generated by a voltage-controlled signal generator (Wavetek 21) and amplified through a power operational amplifier (Apex PA84) capable of providing *IO0 V at &40 mA. Electrical contact was made to the buffer reservoirs of the Kerr cell. The lock-in amplifier (EG&G, Princeton Applied Research Corp. Model 5209) was referenced to the signal generator. The experiment was controlled by a personal computer (Zenith Z-158), which provided the sweep voltage for the signal generator and collected and stored lock-in amplifier readings.
ANALYTICAL CHEMISTRY, VOL. 60, NO. 15, AUGUST 1, 1988
1634
1
23,130
/\
2’
-
1.5
Bp
-1
i
3,416 82
3
I
-.
0 1
1 -1
-
I
I
-
1
3
5
Log Frequency, Hertz
Flgure 2. Swept frequency birefringence measurement of Hind I I I fragments. The response 45’ out of phase with the driving signal is shown
The time domain apparatus was similar, except that the signal generator/power amplifier was replaced by a 0-1000 V direct current power supply and a field effect transistor switch and the Kerr cell electrodes were 2 mm apart. After orientation, birefringence relaxation was measured directly by using a fast 12-bit analog-digital converter (MetraByte DAS-16F). The polarity of the pulse was reversed between each measurement to prevent net migration.
RESULTS AND DISCUSSION Figure 2 shows a typical swept frequency measurement of Hind 111 fragments in a 1.0% agarose gel (SeaPlaque). Peaks for the seven fragments between 23.1 kbp and 564 bp are observed. The relaxation frequency of the 125-bp fragment is beyond the range of our present equipment. The relaxation frequencies are proportional to IPS,as previously observed for individual fragments (5). The fwhm of the completely resolved 23- and 4.3-kbp bands are the predicted 0.7 decade. The incompletely resolved bands are narrower. This effect is still incompletely understood. Rigorously, rotation of the plane of polarization of light by molecules does depend on orientation. There may be partial cancellation of signals when the laser is passed through a gel containing a mixture of two or more fragments whose orientations are slightly different. A rigorous multicomponent response must be derived for amplitudes, not magnitudes. Equation 4 predicts a band shape that is not quite symmetric about l / r . However, the peaks have a longer tail than predicted on the high-frequency side. This deviation from the idealized shape for exponential decay is evidence that the relaxations are not rigorously described by a single exponential. Because a linear frequency sweep containing 3200 points was used, the time required to carry out the measurement was 11min. The signal intensity is proportional to 2\rz, indicating (1-3) a constant birefringence contribution per base pair. Vertical scale changes are used in Figure 2 to make all signals clearly visible. For fragments smaller than about 2 kbp it is useful to employ a gel of good optical clarity to minimize noise caused by light scattering. By varying the agarose concentration, the relaxation frequency can be changed. If the fragment size is similar to the pore size, the relaxation time is concentration dependent ( 5 ) . For example, as the agarose concentration increases from 0.25% to 0.8% the relaxation frequency of the 9.4-kbp fragment decreases by 40%. Similar behavior is observed for other fragments. This effect allows some manipulation of relaxation frequencies, extending the range of observable fragment size with an instrument of fixed scan range.
Reproducibility of peak frequency measurements on the same sample taken a t intervals over 2 days is f1.0%. Reproducibility between samples is controlled by gel concentration reproducibility. For high-accuracy work, standards could be added to a sample to mark the high and low ends of the expected molecular weight range. It is possible to make measurements at the frequency of the driving signal. However, the signals are about 10% as strong as at the harmonic. A rectified cosine wave should have no component at the fundamental frequency. The existence of this signal is further evidence that the relaxations are not rigorously exponential decays. One important practical consequence of the weak signal at the fundamental is that the lock-in amplifier need not have extraordinary subharmonic rejection for this application. We have obtained essentially the same results with the PARC 5209 lock-in amplifier and the less sophisticated PARC 5101. The agarose gel itself is electrically birefringent, but with a very short time constant well beyond the range of our instrument. We do not see a signal from a gel containing no nucleic acid or fragments. In the time domain, multifragment measurements are made directly by extracting the exponential time constants from the observed birefringence relaxation. The technique is applicable to both rapid (microsecond) and slow (second) decays. The time required to acquire the relaxation of a multicomponent sample is the relaxation time of the largest fragment. For a 25-kbp fragment the relaxation constant is about 2 s. However, the success of the multiexponential deconvolution procedure depends on the assumption that the observed individual decays are exponential. Any deviation from the assumed functional form leads to systematic errors in the derived constants. A simplex fit technique (17) easily extracts accurate decay constants from artificial data sets that are the sum of four exponential decays. In practice, we have been able to extract only four of the expected eight relaxation times in a Hind I11 fragment sample. The time constants do not agree well with the values from relaxation frequencies measured by frequency dispersion. The errors are probably caused by deviation from exponential behavior at early times in each component’s decay. This effect has been observed in single-fragment measurements (5). Furthermore, the larger fragments do not behave as rigid rods. Unlike the simplex fitting technique, the frequency dispersion measurement is purely empirical. It requires only that orientation occur a t a finite rate, so that the driving signal and response are 45’ out of phase a t some measurable frequency. The direct decay measurements, however, assume exact adherence to exponential decay. We conclude that time domain measurements are best used to extend the working range of a swept frequency instrument to frequencies beyond the range of the lock-in amplifier employed. In this case derived lifetimes, even if systematically in error, may be adequate to identify particular fragments. It is possible to impove the signal/noise ratio of the experiment by redesign of the Kerr cell so that the signal path is 5-10 mm long and the electrode spacing is 0.5-1 mm. Similarly, a logarithmic frequency sweep would provide a more efficient use of time than the linear sweep now employed. With these refinements, the sensitivity of the experiment will be increased by a factor of 10, while the time is reduced to 2-5 min. From eq 4, it should be possible to resolve 5.5 bands per decade with a 10% intensity decrease between the maxima. The narrow widths of observed overlapping bands suggest that the ultimate resolution should be 10-15 bands per decade. At the present state of development, we can resolve 4 or 5 fragments per decade. Further improvement is possible.
Anal. Chem. lW8, 6 0 , 1635-1637
In DNA fragaent analysis by electric birefringence there is no actual separation of the fragments. They are not automatically available for subsequent experiments. If individual fragments are required, then an electrophoresis or other separation must be performed. Arguably, the separation can be faster if the mobility of each fragment is known in advance so that an optimized voltage program can be used. Electric birefringence relaxation requires no gel staining. In preliminary experiments we have successfully used birefringence imaging subsequent to conventional agarose electrophoresis as a stainless form of densitometry. This technique appears especially useful for high molecular weight nucleic acids, for which the specific birefringence is quite large. Frequency dispersion effects should be observable in fluorescence polarization as well as in electric birefringence. Swept frequency experiments may provide an efficient means of analyzing the fluorescence-tagged fragment extensions prepared by dideoxy procedures, for example. Because there is no dilution of the sample as in a true electrophoresis, design of efficient illumination and signal collection systems should be straightforward. This possibility is under investigation in our laboratories.
LITERATURE CITED (1) Fredericq. E.; Housler, C. Elecfric Dichroism and Electric Birefringence; Clarendon: Oxford, 1973. (2) Molecular Electro-Optics; Krause, S., Ed.; Plenum: New York, 1981. (3) Porschke, D. Annu. Rev. Phys. Chem. 1985,35, 159-178. (4) Stellwagen, N. C. Siopo!~mers1981,1 0 , 399-433.
1635
(5) Stellwagen, N. C. J . Siomol. Strucf. Dyn. 1985,3 , 299-314. (6) Wigmenga, S. J.; Maxwell, A. Biopolymers 1988,25, 2173-2186. (7) Stellwagen, N. C. S/opo&mers 1985,24, 2243-2256. (8) Hurley, I. Slopo&mers 1986. 25, 539-554. (9) Lumpkln, 0. J.; DeJardin, P.; Zlmm, B. H. Biopolymers 1985, 24, 1573-1 593. (10) Odjik, T. Macromolecules 1983, 76, 1340-1344. (11) Noolandl, J.; Rousseau, J.; Slater, G. W.; Turmel, C.; Lalande. M. phvs. Rev. Left. 1987,58, 2428-2431. (12) Thurston, G. 6.; Bowling, D. I. J . Collo/d Interface Scl. 1969, 3 0 , 34-45. (13) Jennings, 6. R.; Brown, B. L. f u r . Polym. J . 1971, 7 , 805-826. (14) deGennes, P. G. J . Chem. Phys. 1971,55, 572. (15) Horowkz, P.; HIII, W. The Art of Electronics; Cambrldge University Press: London, 1980. (16) Meade, M. L. Lock-in AmplIfiefS : Principles and Applicaflons; Peregrinus: London, 1983. (17) Danlels, R. W. An Introduction to Numerical Methods and Opflmizafion Technlques; North-Holland: Amsterdam, 1978.
'
Present address: Department of Chemistry, Ohio Wesleyan University, Newark, OH 43105
Stephen J. Parus* Reed A. Shick Martin Matsumura' Michael D. Morris* Department of Chemistry University of Michigan Ann Arbor, Michigan 48109
RECEIVED for review January 12, 1988.
Accepted April 15, 1988. This work was supported in part by PHS Grant GM37006 and in part by a grant from the AMAX Foundation.
TECHNICAL NOTES Reduction of Slgnal Reflections for Fast-Pulse Recording with Microchannel Plate Detectors L. Q. Huang, R. J. Conzemius,* G . E. Holland, and R. S. Houk
Ames Laboratory-U.S. Department of Energy and Department of Chemistry, Iowa State University, Ames, Iowa 50011 Microchannel plate detectors (MCPD) have been widely used in nuclear sciences, physics, and military applications (1-4). In recent years MCPDs have also gained popularity in mass spectrometry (5,6),particularly in time-of-flight mass spectrometry (TOF-MS). TOF-MS requires a detector yielding an output signal with rise times approaching 1 ns, high gain, low noise, and a large ion collection area (e.g. several square centimeters). The ion collection area should be a flat surface perpendicular to the path of the arriving ions. An MCPD meets these needs, and proper installation can yield measurement resolutions better than 100 ps (7,8). In the development of a TOF-MS (9),it is useful to monitor the ion signal with the detector placed in different locations along the ion flight path. A common problem in pulse transmission is the presence of reflections that distort signal pulses. Signal reflections in transmission lines may be caused by an improper impedance match between the MCPD anode and the high-frequency vacuum feedthrough even though proper cabling and impedance matching is used outside of the housing. Electrical pulses in a transmission line can be viewed as similar to the constructive and destructive interference in
* Author to whom correspondence should be addressed.
a optical wave within a cavity. Impedance matching of the transmission line is analogous to the matching of indexes of refraction at the end of the cavity, and the length of the electrical transmission line is analogous to the length of the cavity. The velocity of propagation of the electronic pulse can be estimated from the transmission line characteristics. The impedance match throughout the transmission line is critical, particularly for accurate transmission of a fast pulse (1-2 ns) obtained from an MCPD. Signal reflections can be minimized by keeping transmission lines as short as possible, but the detector must then be close to a feedthrough flange. The techniques required for reduction of unwanted reflections in high frequency signal transmission are standard practice in experienced laboratories (IO). However, descriptions of these techniques do not appear in a readily appreciated form and are given here for those less experienced in the measurement of fast pulses, especially when a homemade, movable MCPD assembly within a vacuum chamber is used.
EXPERIMENTAL SECTION Two schematic views of the MCPD mechanical arrangement are shown in Figure 1. The characteristics of the microchannel (MCP) as well as pertinent experimental parameters are given in Table I.
0003-2700/8S/0360-1635$01.50/0 0 1988 American Chemical Society