It has long been recognized that, compared with linear oligonucleotide probes, stem–loop hairpin probes can have better specificity in gene detection (37
). Molecular beacons are not only stem–loop hairpin probes but also have the ability to ‘switch on’ fluorescence upon binding to a complementary target, and are therefore becoming a powerful tool for many applications (38
). However, as with any biochemical sensor/transducer system, it is necessary to understand the structure–function relationships of molecular beacons to fully exploit their potential. In many applications, the choices of the sequence of a probe are limited by target-specific considerations, such as the sequence context surrounding a SNP of interest. Probe features that can be adjusted independent of target choice include probe length and stem length. To directly test these aspects of probe design, we systematically studied the relative effects of varying probe and stem lengths on target affinity, mismatch discrimination and hybridization kinetics.
Insight into the structure–function relationship of molecular beacons can be gained by considering the free energy differences between unbound molecular beacons and molecular beacon–target duplexes. The formation of the stem– loop structure between self-complementary sequences at the ends of a molecular beacon is driven by a favorable free energy difference ΔGs
that depends on the stem length Ls
, stem sequence, ionic conditions, and the temperature θ (8
). A longer Ls
leads to a larger ΔGs
and results in a more stable stem–loop structure against thermal fluctuations. Likewise, the free energy difference ΔGp
due to the binding of the probe to its complementary target depends on the probe length, probe sequence and temperature. Since ΔGp
is larger than ΔGs
for a typical molecular beacon design, the stem–loop opens upon molecular beacon–target binding. Competition between ΔGp
largely determines the conformational state of a molecular beacon at a given temperature with a given ionic environment and thus dictates the stability, specificity and hybridization kinetic rates. Although these general trends can be understood qualitatively from an energy point of view, it is still necessary to quantify the structure–function relationship of molecular beacons.
Here we have performed a systematic study of molecular beacons with Lp = 17, 18, 19 and 20 bases and Ls = 4, 5 and 6 bases and show how these structural differences affect function. It was found that molecular beacons with longer (more stable) stem lengths and shorter probe lengths have an improved ability to discriminate between targets over a wider range of temperatures; however, this comes at the cost of reduced molecular beacon–target duplex stability and a decreased rate of hybridization. Molecular beacons with longer stems also generated melting curves with a broader transition between unbound molecular beacons in the stem– loop conformation and molecular beacon–target duplexes. Consequently, there was a reduction in the optimal ratio of molecular beacons bound to wild-type targets to those bound to mutant targets. These findings can be explained by the effect of molecular beacon structure on ΔGp and ΔGs. As the probe length is decreased (lower ΔGp) or as the stem length is increased (higher ΔGs) the difference between ΔGp and ΔGs becomes smaller and the preference of target hybridization becomes less favorable. Any further reduction in this free energy difference, possibly due to a point mutation in the target, will subsequently have an amplified effect on the binding of molecular beacons to these targets.
by lengthening the stem of a molecular beacon also lowered the rate of hybridization due to the larger energy barrier that must be overcome for target binding to occur. It is interesting to note, however, that even the kinetic rate constants for the dual-labeled linear probes were slightly lower than previously reported for unmodified oligonucleotides (18
). Although it is not clear what caused the lower rates of hybridization it could be a consequence of the experimental set-up and the buffer conditions. Another possibility is that there was some interaction between the fluorophore and the quencher, or between the end-labels and the single-stranded target that is not present with unmodified oligonucleotides (47
In determining the binding specificity of a molecular beacon–target duplex, we used melting curves to establish the difference in the fraction of molecular beacon bound to wild-type target and that to mutant target. Therefore, the average width of each curve shown in Figures B, and is an indication of the range of temperatures at which the molecular beacon can differentiate between wild-type and mutant targets and the height reflects how well the molecular beacon can differentiate between targets. The melting curves, however, are dependent on the initial concentrations of molecular beacons and targets. This implies that in order to realize the maximum specificity in living cells the concentration of molecular beacons must be tailored such that the maximum differential between the fraction of molecular beacon bound to wild-type targets and that to mutant targets occurs at physiological temperature. This is very critical in applications in which detection of point mutations of the target is desired.
Using van’t Hoff plots, the changes in enthalpy ΔH12
and entropy ΔS12
associated with beacon/target binding were obtained for molecular beacons with different probe and stem lengths hybridized to wild-type and mutant targets. In general, the measured changes in ΔH12
owing to structural variations and probe/target mismatches are in good agreement with those reported in the literature (19
). For example, for a molecular beacon with a five-base stem and an 18-base probe hybridized to mutant target B, the changes of ΔH12
due to a single-base mismatch (G·A) were, respectively, 58.8 kcal/mol and 161 cal/mol K (entropy unit, eu), while the reported values are in the range of 50.5– 75.2 kcal/mol for ΔH12
and 138.4–215.0 eu for ΔS12
). Further, for a molecular beacon with a five-base stem, increasing the probe length Lp
from 18 to 19 bases resulted in a change in ΔH12
by 8.64 kcal/mol and in ΔS12
by 23.5 eu, respectively, which are comparable to the average values of the unified NN parameters ΔH0
= –8.36 kcal/mol and ΔS0
= –22.4 eu (27
). Unexpectedly, however, when the probe length Lp
of the molecular beacon changed from 17 to 18 bases, both ΔH12
changed about 3.7 times the average values of ΔH0
. One possible reason is that, due to the competition between ΔGp
, the effect of increasing Lp
from 17 to 18 bases is ‘amplified’, i.e. the dissociation constant K12
at θ = θm
is more sensitive to certain changes in probe length. Additional experiments and analysis need to be carried out to study the underlying reasons of this intriguing difference.
The thermodynamic studies present here were conducted in a buffer containing 10 mM monovalent cation and 5 mM divalent cation. For in vivo applications of molecular beacons, the ionic milieu will change, which should affect both melting temperature and kinetic rates. Preliminary studies indicated that kinetic on-rate constants in phosphate buffered saline are significantly slower than the on-rate constants reported here using a buffer that has a relatively high Mg++ ion concentration (data not shown).
In calculating the rate constants for the formation of molecular beacon–target duplexes, we assumed that hairpin unfolding and probe–target binding are linked events and therefore can be analyzed using a two-state approach (25
). However, this assumption might not apply to all scenarios and a multi-step reaction might need to be considered.
The basic features and structure–function relationships revealed in this study have important implications for the design of molecular beacons. This study has demonstrated clearly that in designing molecular beacons for a specific application, both stem and probe lengths must be carefully chosen. For example, when high probe specificity is required, as in the case of detecting point mutations or polymorphisms, molecular beacons will offer improved discrimination when relatively more stable (longer) stems are matched with shorter probe domains. Conversely, when studying RNA expression in living cells in real-time, it may be more important to have fast hybridization kinetics. In this case, molecular beacons with less stable (shorter) stems and longer probe domains would be preferred. Mismatch discrimination is further improved if the mutation is positioned centrally within the probe domain. Finally, since molecular beacons with a four-base stem length may have high background fluorescence, the use of longer stems would be preferred in most applications. In summary, the quantitative studies of structure–function relationships of molecular beacons can provide guidance to the design of molecular beacons to achieve an optimal balance among specificity, S:B ratio and kinetic rates desirable for a specific application.