Detection of tmRNA molecules on microarrays at low temperatures using helper oligonucleotides

Over the last decade microarrays have quickly found applications in microbial diagnostics, for detecting different pathogenic viruses, bacteria and other microbes [1] or for analyzing species composition in environmental and medical samples [2]. Also, many different biosensor technologies based on nucleic acid hybridization have been developed and proposed for quick and cost effective "in-the-field" detection and identification of diseases, pathogens or contaminants [3, 4].

The most common target molecule for diagnostic and phylogenetic studies is 16S rRNA (or corresponding gene). It was used in the 1970s [5] and continues to be the most widely-used marker for discriminating bacterial species [2, 6]. The advantages of ribosomal small subunit RNA are its presence in all species in high copy numbers and the different evolutionary rates of different regions of 16S rRNA, making various taxonomic studies possible [7, 8]. Nevertheless, alternative marker molecules [9–12] have to be considered in case 16S rRNA is not suitable for precise detection and distinguishing between closely related species [13].

One interesting novel marker that has shown great potential in molecular diagnostics is the tmRNA transcript of bacterial ssrA gene. In living cells, tmRNA is present in relatively high copy numbers (around 1000 molecules per cell) [14, 15] and is responsible for assisting ribosomes during translation when protein synthesis stalls. tmRNA molecules contain regions of species-specific sequence heterogeneity and can therefore be successfully used as markers for bacterial diagnostics [16, 17].

Nucleic acid hybridizations in microbiology and molecular diagnostics have been performed at various temperatures ranging from 4°C and RT to around 40°C or even higher (50°C and above). It is suggested in previous studies, that the hybridization of complex target molecules is hindered below 42°C, leading e.g. to low signal intensities and bad probe specificity [18]. Low temperature hybridization is of great interest for emerging technologies, such as membrane biosensors, where the denaturation of membranes and proteins have to be avoided and "laboratory-on-chip" and embedded solutions, where maintaining different compartments with varying temperature can be complicated and costly. Modern oligonucleotide design tools allow the hybridization affinity and specificity of local regions to be estimated quite precisely at different temperatures. One of our main goals was to develop a hybridization method that would be suitable for use below 37°C.

Several difficulties can arise in the detection of bacterial RNA by hybridization. Target RNA degradation has to be prevented and nonspecific hybridizations with wrong targets avoided. The latter is rather difficult on highly conserved RNA molecules [19]. Strong secondary structures can block the hybridization sites inside the molecule and thus prevent hybridization almost completely or retard it significantly [20, 21]. The secondary structure of RNA is much stronger than that of the corresponding DNA [22] and the detection of RNA is more difficult [23]. It is suggested that secondary structure may be the main reason why hybridization-based detection fails at room temperature [18, 21] and it has to be disrupted, or (in the case of synthetic molecules) its formation has to be minimized, to gain access to the target regions of the RNA molecule [21, 24]. The latter is especially crucial in the case of rRNA and tmRNA molecules as they both fold into complex secondary and tertiary structures.

For certain applications it is also important to be able to estimate the relative or absolute abundances of different bacterial strains or species quantitatively. Quantification of hybridization poses additional challenges, especially if the process is too slow to reach equilibrium.

Several approaches to improve the efficiency of hybridization have been described. The hybridization temperature can be increased. It is predicted that while about 70% of a 70-mer cDNA molecule is inaccessible at 42°C, only 30% of DNA and 50% of RNA remains inaccessible at 65°C [20]. Some authors have suggested that a higher temperature increases hybridization specificity, but other authors have found no such effect [23]. Designing probes for specific exposed areas of the molecule also increases the hybridization efficiency [20]. Measuring or predicting the effect of secondary structure is difficult [25, 26], especially as the parts of molecule that do not form double-helical stems can themselves be blocked by higher-level structures [20]. Cleaving target molecules to smaller fragments is one widely-used option; it can expose most hybridization sites that are normally blocked [20, 25].

Alternatively, specific helper oligonucleotides (chaperones) can be added to the hybridization solution to increase hybridization efficiency [18, 23, 27]. These molecules bind to target molecules and block specific or nonspecific intramolecular interactions that cause secondary structure formation. Chaperones are specific to certain target molecules and they also increase the specificity of hybridization. Chaperones can also be marked with fluorophores or other detectable markers, solving the problem of detecting hybridized intact RNA [23, 27]. It is reported that chaperones immediately side-to-side with hybridization probes are most effective in increasing the effectiveness of hybridization at low temperature [28]. This has been explained by the prevention of hairpin structure formation and by the effect of base stacking between capture probe and chaperone [27].

In this study we evaluated the effect of short DNA helper oligonucleotides (chaperones) on the hybridization of synthetic Streptococcus pneumoniae tmRNA molecules to DNA microarray probes. The practical objective was to find an improved protocol for detecting bacterial species by tmRNA hybridization. The theoretical objective was to elucidate the effect of the complex structure of a longer RNA target molecule on hybridization kinetics. In addition, a practical use for chaperones as a interesting novel tool in secondary structure analysis was demonstrated.

Ideally, hybridization can be viewed as an equilibrium between free and bound target molecules. The relative abundances of molecules in both states are determined by the difference between rates of duplex formation and dissociation. These rates in turn are determined by the free energy changes in the corresponding processes. Signal intensity is directly proportional to the number of molecules hybridized and is thus determined by the difference in free energy change between hybridization and dissociation. Equilibrium is usually preferred for hybridization because the signal strength is greatest, since the maximum numbers of target molecules are bound to the probes. Also, as hybridization rates may differ among molecules, only the equilibrium state guarantees that actual signal intensities correlate with the concentrations and hybridization affinities of the target molecules. If equilibrium is not reached, a rapidly but weakly hybridizing target may give a stronger signal than a slowly but strongly hybridizing one, so its relative concentration is overestimated.

At low temperature hybridization experiments, two important factors can influence the signal intensities. Both of them are strongly influenced by secondary structures of the target molecule. First, the equilibrium can change and the probe cannot compete with the secondary structure any more [21, 26]. Second, the concentration of accessible target configurations is lower at low temperature due to the increased stability of tightly packed secondary structures. The hybridization rate is proportional to the concentration of accessible target molecules and is therefore much slower at low temperatures [26].

If the actual rate of hybridization is low, we may not reach equilibrium during the experiment. In that case the signals are initially low but increase over time. As we detected such behavior in our experiment, we concluded that the low signal intensities at low temperature are at least partially caused by slow hybridization.

By hybridizing with the target molecule, chaperones block the formation of at least some secondary or tertiary structure variants and thus increase the rate of hybridization at low temperatures [21, 24]. Although it is shown, that designing chaperone specifically to the the region that forms intramolecular bonds increases hybridization signals significantly [28], this is not always possible.

We were unable to establish any correlation between the pairing probability of individual nucleotides in tmRNA, calculated with sFold, and the relative amplification of the signals in the presence of chaperones. This can be explained by the effect of tmRNA tertiary structure and the non-equilibrium state. The hybridization probability calculated by sFold only takes direct intramolecular pairings between nucleotides into account; it does not consider the blocking of potential hybridization sites by the globular structure of the molecule. Also, the probabilities of different conformations are calculated by absolute free energy levels, not taking into account the kinetics of secondary structure formation. Some conformations with low free energy may form very slowly, simply because they have to cross an unfavorable intermediate state.

When probe hybridization is blocked by the formation of a hairpin-like structure, the chaperone for the immediate neighborhood should work best because it does not allow the hairpin to form. It has been demonstrated that designing chaperones for the immediate neighborhood should work best if the accessibility of the region is hindered by the formation of secondary structure elements such as hairpins [28].

We were able to see such an effect with chaperone F, where the hybridization profiles of all probes suggest the presence of a hairpin. The chaperone binds to region 248-261 and the signals in the immediately preceding region (230-245) are strongly amplified in its presence. Thus, one can infer that without treatment, these regions are probably hybridized, forming a hairpin. Indeed, the ΔG plot generated by mFold suggests that there is a local energetic minimum between those regions. Nevertheless, it is important to notice that according to mFold this hairpin was not present in the most stable molecule conformation, as determined by the global energy minimum. Also, many other places with similar local energy profiles did not show a similar hybridization pattern. Thus, a hairpin, even if energetically favorable, may often not be the prevailing secondary structure pattern.

A high-level (tertiary) structure of RNA may also form by intramolecular hybridization between distant parts of a longer sequence. Thus the sites that affect the accessibility of a certain part of the molecule may be spatially separated from it. In that case, designing a chaperone for the immediate proximity of the probe may not work, because it either cannot hybridize because of blockage by the three-dimensional structure, or fails to make the neighboring site accessible because it is blocked by some other region. In that case, the best results should be obtained by a chaperone hybridizing specifically to the tertiary structure-forming region, so the tertiary structure cannot form and the conformation of the molecule is loosened. Such chaperones can potentially increase the probe hybridization rate in many regions of the molecule.

In the current experiments, the effect of tertiary structure is suggested by the fact that a single chaperone was able to amplify signals in different regions of the tmRNA molecule (chaperones A and D) and different chaperones amplified signals in the same region (chaperones A, D and E). Such an effect was also suggested by the generic increase of signal intensities of all non-overlapping probes, irrespective of location, in the presence of chaperones.

As expected, chaperones almost completely block the signals of the overlapping probes. Nevertheless, this effect is not absolute, as seen with chaperone E. The reason is still unclear. One possibility is weak chaperone binding, so the higher affinity capture probes outcompete it during hybridization. However, the chaperone has to have a sufficiently high affinity to outcompete the secondary structure of tmRNA successfully. This contradiction may indicate that at least in some cases, the factor limiting hybridization is not secondary but tertiary structure. Although intramolecular double-stranded regions are energetically weak, they fold the tmRNA molecule in such a way that some parts of it are not easily accessible. In such a case, the energetically much stronger target-probe hybridization is slow because the probability that a probe will hit its target area is very low, especially as the probes are immobilized on a surface. Chaperones were applied at a higher temperature at which there was no tertiary structure. Also, the relative concentration of chaperone molecules was much higher, both because they were applied in abundance and because they were free in solution. Thus, the chaperones could hybridize to any region, and if the temperature was lowered, the tertiary structures either did not form or were much weaker.

We thus conclude that while the hybridization of tmRNA can sometimes be relatively slow at low temperatures, it can be significantly increased by using specific helper oligonucleotides (chaperones). The exact effect of certain helper nucleotides on the strength of the signal of certain capture probes depends on many factors, including probably the three-dimensional structure of the target molecule. The effect is not always local, meaning that a chaperone in the immediate proximity of the capture probe may not increase the signal strength, while one at a distant location might. As the structures of denatured nucleotide sequences cannot be precisely predicted at present, experimental verification of chaperones is necessary.