Victoria L. Singer and Iain D. Johnson
Molecular Probes, Inc., 4849 Pitchford Avenue, Eugene, OR 97402 USA.

× Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø

The field of DNA typing is rapidly shifting towards the use of fluorescence-based detection methods in a wide variety of applications. The most important reason for this shift is the availability of numerous spectrally distinct fluorescent labels, which facilitate multiplex analysis. However, the broad dynamic range of quantitation possible with fluorescence and the ease of automation of fluorescence-based assays also play a role. How does the testing lab select the best fluorescent label for a particular bioanalytical application from among the multitude of available choices? Optimal dye selection requires consideration of the spectral properties of fluorescent labels in relation to the characteristics of the instrument used for detection. Instruments currently in widespread use include fluorescence microplate readers, ultraviolet light transilluminators and Polaroid^® or CCD cameras, automated sequencers, laser scanners, and capillary electrophoresis systems. Instrument parameters to be considered include the excitation source type, the optical filters used for signal discrimination and the sensitivity and spectral response of the detector. The most important dye characteristics in optimizing compatibility are the excitation spectrum, the fluorescence Stokes shift and the emission spectral bandwidth. Other factors, including the extinction coefficient, fluorescence quantum yield, photostability and environmental sensitivity, determine the fluorescence output ("brightness") of the labeled biomolecules. In addition, the effects of dye physical characteristics on labeling efficiency and biological structure/function properties are of major importance. These effects produce assay performance manifestations such as modified electrophoretic behavior, hybridization efficiency or enzymatic turnover.

Fluorescence-based assays are now in widespread use in forensics laboratories for quantitating DNA yields in solution or in gels and for DNA typing. Fluorescence provides useful detection tools because it has the capacity to allow simultaneous multiparameter analysis of complex samples. In addition, fluorescence detection techniques have intrinsically higher sensitivity than comparable colorimetric detection methods because the optical background due to incident (excitation) light can, in principle, be completely eliminated. The dynamic range for quantitation is also very broad. As the forensics community moves towards the use of more and more of these tools, it is useful to examine some of the characteristics of the fluorophores in common use, in order to determine the optimal dyes to choose in developing new applications and in order to understand which of these properties might play a role in generating artifactual assay results.

Fluorescence is a phenomenon that results when a molecule known as a fluorophore or fluorescent dye absorbs incident light and in response emits light at a different wavelength (Figure 1). The absorbed photon creates an excited electronic singlet state in the fluorophore, which exists for a finite time (typically 1-10 nanoseconds) that is characteristic of that fluorophore. During this time the molecule can undergo conformational changes and can interact with its environment, partially dissipating the energy of the excited state. Fluorescence emission is only one of a number of competing excitated state deactivation processes. Therefore, the number of fluorescence photons emitted per photon absorbed (quantitatively expressed by the fluorescence quantum yield, see below) is almost always less than one. The fluorophore then emits light at a longer wavelength than that of the absorbed photon — this is simply because the energy of the emitted photon is lower than that of the absorbed photon, due to energy dissipation during the excited state lifetime.

Because of the wide variety of bioanalytical applications utilizing fluorescence-based detection methods, it is impossible to designate one particular fluorophore as universally "better" than another. Instead, a more reasonable approach is to determine which dye or dyes are best suited to a particular application. For example, the performance of any given fluorophore in short tandem repeat (STR) analysis is directly related to the chemical properties of that fluorophore and of its nucleic acid conjugates. Thus a useful question to ask is: What are the desired characteristics of fluorophores designed for use in STR analysis?

The major factors that affect fluorophore performance in actual applications are:

The molar extinction coefficient is a direct measure of a dye's ability to absorb light. This capability is clearly important in determining the amount of light a molecule can generate via fluorescence emission. Most fluorophores in common use have molar extinction coefficients at their wavelength of maximal absorption ranging between 5000 and 200,000 cm^-1M^-1 (Haugland, 1996). The fluorophore labels most commonly used in STR analysis have molar extinction coefficients between 70,000 and 90,000 (Table 1). Note that the rhodamines have particularly large extinction coefficients. The Stokes shift is the difference in wavelength between the fluorescence excitation maximum and the fluorescence emission maximum. Interestingly, in general, dyes with large Stokes shifts tend to have relatively small extinction coefficients, whereas dyes with small Stokes shifts tend to have relatively large extinction coefficients.

The extinction coefficient of a dye can be measured at places other than its absorption maximum. The effect of exciting a fluorophore at a wavelength other than its maximum is illustrated in Figure 2. The emission spectrum shape is unchanged, regardless of where excitation takes place. However, the intensity of the emission is directly dependent on the amount of light the dye has absorbed. The amount of light absorbed is dependent on the relationship between the absorption spectrum and the excitation source wavelength(s). Thus matching excitation lasers and/or filters to the absorption spectra of dyes is extremely important in obtaining optimal signals.

The fluorescence quantum yield is a measure of the efficiency with which the excited molecule is able to convert absorbed light to emitted light. It is defined as the fraction of absorbed photons that are converted to fluorescence emission. Typical quantum yields for commonly used fluorophores range from 0.05 to 1.0. For the fluorophore labels commonly used for STR, typical quantum yields are between 0.5 and 0.95 (Table 2). Most dyes exhibit a decrease in quantum yield upon coupling to proteins and nucleic acids. The quantum yield is very sensitive to the environment, pH, temperature, and the structure of the fluorophore itself. Note that although TAMRA (tetramethylrhodamine) has a very large extinction coefficient (Table 1), its quantum yield is not particularly high (Table 2). This is mainly due to the flexibility inherent in the structure of TAMRA, as compared, for example, with ROX (carboxy-X-rhodamine) (Figure 3). A lack of structural rigidity is the main reason why the commonly used pH indicator, phenolphthalein, is not fluorescent, while the structurally similar fluorescein is an excellent fluorophore (Figure 4). The excited state of phenolphthalein is deactivated by internal rotation about the central portion of the molecule, dissipating the absorbed energy as heat. This deactivation mechanism is not feasible for fluorescein, so absorbed energy is instead emitted in the form of fluorescence. Similarly, TAMRA has two positions able to rotate to a greater degree than ROX and thus has a lower fluorescence quantum yield. This molecular flexibility of TAMRA is responsible for the sensitivity of its fluorescence emission intensity to temperature (Figure 5). The practical concern for the forensics laboratory is that if an STR gel undergoes excessive heating during the course of electrophoresis (because of running the gel at too high current or power, or because of the buffer capacity being exceeded during the electrophoretic run), the TAMRA labeled fragments may appear less fluorescent relative to other labeled fragments.

Dye environment has a strong effect on some fluorophores and minimal effects on others. The strongest environmental factors that affect fluorescence yield are pH, choice of solvent, the presence of quenchers, and binding to DNA or proteins. Fluorescein, as indicated earlier, is much like phenolphthalein in its structure. It is also a pH indicator. Fluorescein (FAM) is progressively less fluorescent as the pH decreases below pH 9.0 (Figure 6). Oregon Green ^® 488 dye, another fluorinated fluorescein, has a different pK_a and thus has fairly constant signals above pH 6.0. Rhodamine Green^® dye, on the other hand, is relatively pH insensitive. For DNA typing, this means that the pH of the buffer in the gel can be extremely important for detecting FAM signals. Since Tris buffer pH is affected by temperature (the higher the temperature the lower the pH), the result of running a gel too fast (too hot) can be a drop out of FAM labeled bands simply due to a decrease in buffer pH in the gel itself.

For most DNA typing applications, solvent effects and the presence of many fluorescence quenchers are not important, since most reactions and detection take place in aqueous solution and materials that are potential quenchers are generally removed during the DNA purification process. One exception to this statement is the quenching effect due to hemoglobin on DNA quantitation. For example, high concentrations of hemoglobin appear to quench the fluorescence of bound PicoGreen^® reagent under some assay conditions.

The effect of binding to nucleic acids is very important for certain dyes. Nucleic acid stains exhibit large increases in fluorescence upon binding noncovalently to nucleic acids. The reason for this increase, or fluorescence enhancement, is that the relatively flexible dye structures are held in more rigid conformations when bound by nucleic acids. This is true whether the dyes intercalate between the bases, bind the DNA grooves, or interact in other ways. In addition, water interactions with the dyes are usually excluded during DNA binding (a kind of solvent effect). Although all nucleic acid stains act on these same general principles, the difference in their ability to detect DNA is due to a combination of factors, including fluorescence enhancement, affinity, extinction coefficient and quantum yield of the nucleic acid complex (Table 3). All of these factors working together combine to give PicoGreen^® reagent a higher sensitivity for double-stranded DNA detection than ethidium bromide or Hoechst 33258.

Spectral bandwidth, dye purity and electrophoretic mobility distortion are also important factors affecting STR analysis. The importance of spectral bandwidth, especially in the emission spectrum, lies in the capability of instruments to distinguish one dye from another. For example, although fluorescein (FAM), tetramethyl-rhodamine (TAMRA) and Texas Red^® dyes can be spectrally distinguished from one another using appropriate optical filters, the BODIPY^® dyes exhibit narrower emission spectra, making it easier to design filters that exclude signals from the other two fluorophores (Figure 7). Signals that "bleedthrough" from one channel or filter to another make data interpretation extremely difficult.

Impure dye or the presence of unwanted isomers in reactive dyes or conjugates can give rise to shadow bands or unexpected shoulders on peaks, which could be misinterpreted as mixed alleles or mixed samples.

The electrophoretic mobility of DNA fragments is affected by the addition of most fluorophores. This is both due to the physical size of the fluorophore — which affects the physical size and shape of the conjugate — and to the ionic charge present on the dye — which affects mobility by altering the charge:base ratio of the resulting nucleic acid conjugate. A well-known example of this phenomenon is the mobility retardation caused by binding to noncovalent nucleic acid stains. Because most such stains are cationic, they impede electrophoretic mobility in direct proportion to the number of dye molecules bound per base pair. The same phenomenon occurs when fluorescent dyes are covalently coupled to STR primers. Thus, in selecting a set of fluorophores to use in multiple fluorophore analysis, it is critical to determine what, if any, the effects of those particular fluorophores are on electrophoretic mobility, and then to correct for observed changes using software or other methods.

A wide variety of factors affect the performance of fluorophore labels and nucleic acid stains. Taking these factors into consideration allow the laboratory that is developing new assays and the laboratory that is using existing assays to design better experiments and to better interpret experimental results.

Table 1. Molar extinction coefficients were determined for free dyes in solution.

Table 3. Comparison of Double-Stranded (ds)DNA Quantitation Reagents.

Extinction coefficients were determined for free dye in aqueous solution.

Figure 1. Jablonski diagram, showing a schematic view of the fluorescence process. During stage 1 a photon of energy hu_EX is absorbed by the fluorophore, creating the S₁' singlet electronic excited state. The excited state lifetime is stage 2, during which the molecule undergoes conformational changes and interacts with its environment. This results in the relaxed singlet excitation state, S₁. During stage 3 a photon of energy hu_EM is emitted.

Figure 2. Effect of exciting a fluorophore at a place in its spectrum other than the absorbance maximum. A representative fluorophore is excited at three different wavelengths (EX 1, EX 2 and EX 3) and the resulting emission spectra are shown (EM 1, EM 2 and EM 3, respectively).

Figure 3. Comparison of the structures of ROX and TAMRA. TAMRA, shown on the left, is very similar to ROX, on the right. However, the ring structures present in the ROX molecule prevent some of the intramolecular rotational motion to which TAMRA is subject in the presence of heat.

Figure 4. Comparison of the structures of phenolthalein and fluorescein. Phenolphthalein and fluorescein differ structurally only due the presence of a single oxygen atom and the presence of bonds formed with that oxygen. The phenolphthalein molecule can rotate much more freely than the fluorescein molecule, because of the absence of that oxygen.

Figure 5. Temperature-dependence of the fluorescence intensity of FAM, JOE, TAMRA and ROX dyes. The fluorescence emission intensity of each dye was measured as a function of temperature, using a minifluorometer.

Figure 6. Comparison of pH sensitivity several fluorophores. The fluorescence intensity of Oregon Green^® 488 dye (solid circles), FAM (open circles) and Rhodamine Green^® (open squares) was measured as a function of pH.

Figure 7. Comparison of the fluorescence emission spectra of several dyes. Normalized fluorescence emxission spectra are shown for goat anti-mouse IgG conjugates of fluorescein (FL), tetramethylrhodamine (TMR) and Texas Red^® dye (TR) (dashed lines), and BODIPY FL, BODIPY TMR and BODIPY TR conjugates (solid lines).