Figure 7.3. Block diagram of a near-IR time-correlated single-photon counting (TCSPC) detector for CGE. The laser is focused onto a capillary column with the emission collected using a 40x microscope objective (N.A. = 0.85). The fluorescence is imaged onto a slit and then spectrally filtered and focused onto the SPAD. L, laser singlet focusing lens; C, capillary; BD, beam dump; MO1, collecting microscope objective; MO2, focusing microscope objective; SPAD, single-photon avalanche diode; AMP, amplifier; CFD, constant fraction discriminator; TAC, time-to-amplitude converter; ADC, analog-to-digital converter; and MCS, multichannel scaler (Legendre et al., 1996).

selected for this instrument was a single-photon avalanche diode (SPAD) possessing an active area of 150 fim offering a high single-photon detection efficiency (>60% above 700 nm). The counting electronics were situated on a single TCSPC board, which was plugged directly into a PC-bus exhibiting a dead time of <260 ns, allowing efficient processing of single-photon events at counting rates exceeding 2 x 106 counts/s. This set of electronics allowed for the collection of 128 sequential decay profiles with a timing resolution of 9.77 ps per channel. The instrument possessed a response function of approximately 275 ps (FWHM), adequate for measuring fluorescence lifetimes in the subnanosecond regime.

One of the most important aspects associated with lifetime measurements in sequencing applications is the processing or calculation algorithm used to extract the lifetime value from the resulting decay. The accuracy of the base call depends directly on the lifetime differences between fluorophores in the dye set and the relative precision in the measurement. Algorithms used in this application require not only the calculation of the lifetimes precisely, even under the situation of poor photon statistics, but also the ability to perform on-line measurements during the electrophoresis. The typical calculation algorithm for lifetime determinations, nonlinear least squares (NLLS), can deconvolve the IRF from the overall decay and provide a more accurate lifetime value. Unfortunately, this algorithm is calculation intensive and it produces large errors in cases where photon statistics are poor. Moreover, it is more suitable for static measurements rather than dynamic (on-line). Two other simple algorithms for on-the-fly fluorescence lifetime determinations that have been evaluated are the maximum likelihood estimator (MLE) and the rapid lifetime determination method (RLD) (Soper and Legendre, 1994).

MLE calculates the lifetime via the following relation (Hall and Selinger, 1981):

1 + (eT/Tf - 1) - m (emT/Tf - 1)-1 = N-1J2 iN (7.1)

where m is the number of time bins within the decay profile, Nt is the number of photocounts in the decay spectrum, Ni is the number of photocounts in time bin i, and T is the width of each time bin. A table of values using the left-hand side (LHS) of the equation is calculated by setting m and T to the experimental values and using lifetime values (tf) ranging over the anticipated values. The right-hand side (RHS) of the equation is constructed from actual decay data over the appropriate time range. The fluorescent lifetime is determined by matching the value of the RHS obtained from the data with the table entry from the LHS. The relative standard deviations in the MLE can be determined using N-1/2.

Fluorescence lifetimes are calculated using the RLD method by integrating the number of counts within the decay profile over a specified time interval and using the following relationship (Ballew and Demas, 1989):

where At is the time range over which the counts are integrated, D0 is the integrated counts in the early time interval of the decay spectrum, and D1 represents the integrated number of counts in the later time interval. Both the MLE and RLD methods can extract only a single lifetime value from the decay, which in the case of multiexponential profiles would represent a weighted average of the various components comprising the decay.

Wolfrum and co-workers have developed a special pattern recognition technique for calling bases using lifetime discrimination methods (Koellner et al., 1996). Basically, the method involves comparing a simulated decay pattern to the measured decay and searches for the pattern that best fits the measurement. This algorithm is equivalent to the minimization of a log-likelihood ratio, where fluorescence decay profiles serve as the pattern. Since the pattern recognition algorithm uses the full amount of information present in the data, it potentially has the lowest error or misclassification probability.

Soper et al. (1995) demonstrated the feasibility of performing on-line lifetime determinations during capillary gel electrophoresis (CGE) separation of DNA sequencing ladders. C-terminated fragments produced from Sanger chain-termination protocols and labeled with a near-IR fluorophore at the 5' end of the sequencing primer were electrophoresed and the lifetimes of various components within the electropherogram were determined. The average lifetimes determined using the MLE method was found to be 581 ps with a standard deviation of ±9 ps (RSD = 1.9%). This result indicated that MLE could produce high precision, even for ultra-dilute conditions. The favorable accuracy and precision was aided by the use of near-IR fluorescence detection, which minimized scatter contributions into the decay as well as background fluorescence. Four-Lifetime/One-Lane DNA Sequencing

In the four-lifetime/one-lane method, four DNA ladders of differently labeled fragments generated by Sanger chain termination reactions are separated in one lane, typically in a single capillary gel column, and the base calling is done with lifetime discrimination as opposed to spectral discrimination. Obviously, the dye sets suitable for color discrimination are not necessarily appropriate for the use in lifetime discriminations. New dye sets must be developed that suit this identification method. In lifetime discrimination, it is not necessary to use dyes with discrete emission maxima; therefore, structural variations in the dye set can be relaxed. For instance, Flanagan et al. (1998) developed a dye set that consisted of a series of structurally similar near-IR tricarbocyanines that possessed identical absorption (765 nm) and emission (794 nm) maxima. The lifetimes of the dyes were varied by incorporating a single halide (I, Br, Cl, or F) into the molecular structure. This dye set, with lifetimes ranging from 947 ps to 843 ps when measured in a polyacrylamide gel, have been suggested for use in a four-lifetime/one-lane sequencing experiment. Sequencing primers labeled with this dye set demonstrated uniform mobilities in gel electrophoresis applications, irrespective of the linker structures (see Table 7.2). However, since these were tricarbocyanine dyes, the lifetimes were found to be <1.0 ns and the lifetime differences among the dye set was somewhat small (Atf = 70 ps, ~8% relative difference). Lassiter et al. (2002) optimized the experimental conditions for using these dyes for DNA sequencing by lifetime discrimination.

Wolfrum and co-workers demonstrated the use of a four-lifetime/one-lane approach for DNA sequencing using CGE (Lieberwirth et al., 1998). In their work, four dye labels selected from the rhodamine, cyanine, and oxazine families were covalently tethered to the 5' end of oligonucleotide primers. The dyes exhibited similar absorption and emission maxima and were excited efficiently with a 630-nm pulsed diode laser operated at a repetition rate of 22 MHz. The labels exhibited fluorescence lifetime values of 1.6, 2.4, 2.9, and 3.7 ns, with the difference between the dyes adequate for efficient identification in sequencing applications (see Figure 7.4a). This dye set allowed for the use of a simple detection system that was equipped with a single laser and a single avalanche photodiode. The instrumental response function of the entire system was measured to be ~600 ps (FWHM). The time-resolved data were managed using the TCSPC technique. Using appropriate linker structures, dye-dependent mobility shifts were minimized, eliminating the need for post-electrophoresis corrections. The dye-labeled sequencing fragments were identified by both MLE and pattern recognition algorithms, with the latter method providing higher overall base-calling accuracy. Using an M13 template, Wolfrum and co-workers were able to demonstrate a read length of 660 bases with a probability of correct classification >90% (see Figure 7.4a-d). Two-Lifetime/Two-Lane

To show that fluorescence lifetimes could also be obtained from multiple electrophoretic lanes, a scanning system for measuring fluorescence lifetimes from multi-lane slab gels has been reported by Lassiter et al. (2000). In that report, a modified microscope head was inserted into an automated slab gel sequencer, which consisted of a near-IR time-resolved scanning imager and implemented a two-lifetime/two-lane sequencing approach. Two dyes in each lane were identified by their lifetimes on-line during gel electrophoresis. Two commercially available cyanine-based near-IR dyes, IRD700 and Cy5.5 were chosen as fluorescence reporters and were labeled at the 5' end of a sequencing primer. A-terminated bases were labeled with IRD700, and T-terminated bases labeled with Cy5.5 were elec-trophoresed in one lane, while C (IRD700) and G (Cy5.5) tracts occupied an adjacent lane. The similar absorption and emission properties of these two dyes allowed efficient processing of the emission on a single detection channel and excitation with a single source. The lifetimes for these two dyes were calculated by the MLE algorithm and determined to be 718 ± 5 ps and 983 ± 13 ps for IRD700 and Cy5.5, respectively. Figure 7.5 shows an


0 0

Post a comment