Jean-Louis H. A.
Duprey
a,
Dario M.
Bassani
b,
Eva I.
Hyde
c,
Gediminas
Jonusauskas
d,
Christian
Ludwig
e,
Alison
Rodger
f,
Neil
Spencer
a,
Joseph S.
Vyle
g,
John
Wilkie
a,
Zheng-Yun
Zhao
a and
James H. R.
Tucker
*a
aSchool of Chemistry, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. E-mail: j.tucker@bham.ac.uk
bInstitut des Sciences Moléculaires, CNRS UMR 5255, Université Bordeaux, 351 Cours de la Libération, Talence 33405, France
cSchool of Biosciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
dLaboratoire Ondes et Matière d'Aquitaine, UMR CNRS 5798, Université Bordeaux, Talence 33405, France
eHenry Wellcome Building for Biomolecular NMR Spectroscopy, Institute of Cancer & Genomic Sciences, College of Medical & Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
fDepartment of Molecular Sciences, Faculty of Science and Engineering, Macquarie University, North Ryde, NSW 2109, Australia
gSchool of Chemistry and Chemical Engineering, Queen's University Belfast, David Keir Building, Stranmillis Road, Belfast, BT9 5AG, UK
First published on 15th August 2018
The labelling of DNA oligonucleotides with signalling groups that give a unique response to duplex formation depending on the target sequence is a highly effective strategy in the design of DNA-based hybridisation sensors. A key challenge in the design of these so-called base discriminating probes (BDPs) is to understand how the local environment of the signalling group affects the sensing response. The work herein describes a comprehensive study involving a variety of photophysical techniques, NMR studies and molecular dynamics simulations, on anthracene-tagged oligonucleotide probes that can sense single base changes (point variants) in target DNA strands. A detailed analysis of the fluorescence sensing mechanism is provided, with a particular focus on rationalising the high dependence of this process on not only the linker stereochemistry but also the site of nucleobase variation within the target strand. The work highlights the various factors and techniques used to respectively underpin and rationalise the BDP approach to point variant sensing, which relies on different responses to duplex formation rather than different duplex binding strengths.
An alternative approach to point variant sensing is to generate different and distinct read-outs for the two possible duplexes formed with the probe strand, as illustrated for this work in Fig. 1. From a sensor design point of view, such an approach is attractive since it removes the need to establish a temperature window in which one target strand binds and the other does not. This means that sensing can be carried out at any desired temperature, so long as each target can form a stable duplex. Although commercial assays of these so-called base discriminating probes§ (BDPs) have yet to emerge, there are plenty of examples from the research literature over the last few years,6–19 with read-out methods other than fluorescence also recently established.20
Our main contribution to this area has been the development of anthracene-containing probes in which the tag is attached to the DNA backbone via the non-nucleosidic linkers serinol17 or threoninol.11,15,16 These linkers are particularly attractive due their relative ease of preparation and, in the case of threoninol, the ready availability of both the D- and L-isomers.21 While others have developed similar strands containing pyrene12,18 and thiazole orange10,13,19 tags, anthracene is attractive due to its well documented intercalative30 and photochromic22,38 properties. Our previous work established that the formation of duplexes with fully matched base pairs on either side of the anthracene tag brought about a decrease in emission intensity, whereas the formation of those containing a base-pair mismatch on the 5′ side with respect to the tag brought about an increase (Fig. 1).16,17 Although these observations have proven to be broadly applicable to different sequences,16 making this sensing method attractive for probing almost any nucleobase variation, there has been no in-depth rationalisation of this intriguing OFF/ON behaviour, without which it cannot be deemed a robust transferable approach. Herein, through a combination of spectroscopic techniques and molecular dynamics simulations, we provide an explanation for the sensing mechanism. In particular, we show how both the choice of the linker group stereochemistry as well as the position of the mismatched base-pair with respect to the fluorophore are essential for generating an effective sensing response. This study is important for highlighting the various factors that underpin the design of a successful base discriminating probe, which in turn should inform the design of the most effective probes of this type for commercial or clinical applications.
Oligonucleotide name | Sequence |
---|---|
a B = A, C, G, T or Ab (Ab denotes an abasic nucleoside with no nucleobase, prepared for the 3′-GBG strand only). | |
5′-CTC | 5′-TGGACT-CTC-TCAATG-3′ |
Probe 5′-CXC | 5′-TGGACT-CXC-TCAATG-3′ |
Target 3′-GBG | 3′-ACCTGA-GBG-AGTTAC-5′ |
Target 3′-GAB | 3′-ACCTGA-GAB-AGTTAC-5′ |
Target 3′-BAG | 3′-ACCTGA-BAG-AGTTAC-5′ |
CLC-trimer | 5′-CLC-3′ |
Notwithstanding that the sensing behaviour operates for other flanking bases,16 for the sake of comparison with most of our previous work, it was decided to retain two cytosines either side of the anthracene tag for this detailed study, giving the 15-mer probes 5′-CLC- and 5′-CDC (Table 1). In addition, three control compounds were made: the unmodified strand 5′-CTC, the 3-mer CLC-trimer and the monomer L-Phos. So that we could investigate and rationalise the effect of variations in the target DNA sequence on both the thermal stabilities and the photophysical properties of various duplexes, a series of 15-mer target sequences (3′-GBG, 3′-GAB, 3′-BAG where B = G, A, C, T) were also prepared. In particular, this would allow us to examine the effect of introducing changes in the adjacent 5′- (upstream) or 3′- (downstream) positions as well as those directly opposite the tag. Purification was performed via preparative RP-HPLC on the 5′-DMT protected strands, which were then detritylated and de-salted. The purity and composition of each oligonucleotide was confirmed using analytical HPLC and mass spectrometry respectively, as detailed in the ESI.†
Sequence | τ 1 wt (ns) (%) | τ 2 wt (ns) (%) | τ 3 wt (ns) (%) | χ 2 | |||
---|---|---|---|---|---|---|---|
a Relative contribution of each decay constant to the total emission decay. | |||||||
5′-CLC | 1.05 | (38) | 4.28 | (36) | 12.4 | (26) | 1.04 |
3′-GAG | |||||||
5′-CDC | 1.50 | (33) | 3.68 | (41) | 9.08 | (25) | 0.99 |
3′-GAG | |||||||
5′-CLC | 0.94 | (32) | 4.08 | (40) | 12.9 | (28) | 1.01 |
3′-GAA | |||||||
5′-CLC | 0.88 | (2) | 4.6 | (21) | 14.4 | (77) | 0.97 |
3′-AAG | |||||||
5′-CLC | — | 5.1 | (20) | 13.3 | (80) | 1.02 | |
3′-CAG | |||||||
5′-CLC | — | 5.9 | (29) | 13.5 | (71) | 0.97 | |
3′-TAG |
The effect on the lifetime distribution of the mismatched duplexes was also investigated. Those formed with the 5′-CDC probe were again all tri-exponential with comparable decay profiles. However, whereas duplexes for the 5′-CLC system with downstream mismatches (e.g. 3′-GAA) also gave tri-exponential decays, those for upstream mismatches changed from a tri-exponential decay profile to essentially a bi-exponential one, with the fastest quenching pathway absent (or almost completely suppressed, only 2%, in the case of 3′-AAG).
Target | ||||
---|---|---|---|---|
a Average of at least 3 runs and determined by taking the maximum of the first derivative of the melting curve. | ||||
Probe | 3′-GAG | 3′-GAA | 3′-AAG | 3′-GAbG |
5′-CTC | 55.0 | 42.5 | 42.0 | 40.0 |
5′-CLC | 52.5 | 44.5 | 46.0 | 53.5 |
5′-CDC | 48.0 | 37.5 | 35.0 | 48.0 |
Fig. 3 CD spectra of duplexes 5′-CTC/3′-GAG, 5′-CLC/3′-GAG and 5′-CDC/3′-GAG in 10 mM pH 7 phosphate buffer, 100 mM NaCl. Main: [duplex] = 5 μM; insert: S0 → S1 region, [duplex] = 500 μM. |
The imino regions in each NMR spectrum, denoting the H-bonding base pairs (N–H⋯N resonances, where the proton donor group is from either a T or G base), are shown in Fig. 4. The unmodified duplex (Fig. 4a) shows a cluster of peaks in the A-T region and six peaks further upfield in the G-C region, indicating the loss of one terminal GC base pair signal out of the possible seven due to fraying of the duplex ends. The spectra for the anthracene-modified duplexes show interesting trends. For the two matching duplexes, distinct upfield shifts are observed for two of the G-C signals, which are assigned to NH protons on each G base lying either side of the modification site. In the case of the L-threoninol system, these signals are distinct from one another at 11.20 ppm and 12.15 ppm. For the mismatched duplex, and as expected from these assignments, one upfield peak is no longer observed (due to the introduction of a G-A mismatch) with one residual peak now at 11.9 ppm.
Fig. 5 Images from AMBER molecular dynamics models of 5′-CLC/3′-GAG duplex (left) and 5′-CDC/3′-GAG duplex (right) after 10 ns. The carbon atoms of the anthracene monomer unit are coloured yellow. |
If we first consider the single stranded probes, the quantum yield for the D-threoninol linker system is more than twice as high as for its L-threoninol counterpart, with a value fairly close to that of the anthracene monomer alone (L-Phos). This indicates that with a D stereochemistry, the tag is positioned such that its excited state is much less sensitive to quenching processes attributed to vicinal bases. However, upon duplex formation, the anthracene necessarily comes into contact with more bases, which for both stereochemistries and for almost all sequence variations, with the one noted exception, decreases the emissiveness of the anthracene tag. This effect is not surprising, given that DNA bases are known to quench the excited states of various organic fluorophores.25 However, it is worth noting that the anthracene quenching is uniform when varying the base opposite the tag for both the L-system and D-systems (i.e. Probe 5′-CLC with targets 3′-GBG, where B = G, A, C and T, see ESI†) and this trend continues for an abasic site at this position. This gives a strong indication that the base directly opposite the anthracene tag plays no direct role in quenching its excited state. This contrasts with probes with longer tether lengths between the anthracene and the L- or D-threoninol backbone, for which either base variations15 or base modifications (i.e. epigenetic changes)11,15 may be sensed by changes in fluorescence emission intensity.
As noted above, the one notable exception to the general trend of anthracene emission intensity decreasing upon duplex formation is for the L-system, but only when duplexes are formed that contain a base-pair mismatch immediately upstream of the tag. Furthermore the effect is observed for both transversions and transitions (i.e. an increase in emission is observed for all mismatched duplexes 5′-CLC/3′-BAG, where B = A, C, T). Our spectroscopic measurements and molecular simulations indicate that the origin of this effect lies in the extent to which a given stereochemistry enables the anthracene tag to experience a local environment that is less quenching upon the introduction of a mismatching base pair. For example, the data from the variable temperature UV/vis studies (Table 3) indicates that of the two isomers, the L stereochemistry confers substantially more stabilisation than the D in the matching systems, suggesting a greater interaction between the tag and the duplex. For the mismatched duplexes, there is an expected drop in the duplex melting temperature for the unmodified and the modified systems with either stereochemistry. However, whereas the location of the mismatch (upstream or downstream) does not appear to substantially affect this change in stability, the smaller drop in Tm values for the L-isomer systems indicates that this stereochemistry allows the anthracene to insert further into the mismatched duplexes, providing greater hydrophobic stacking interactions that hold them together more strongly in the absence of local hydrogen bonding.
CD spectroscopy also reveals an interesting stereochemical effect (Fig. 3). The characteristic B-DNA structure is observed for the matching duplexes, with additional anthracene-dependent signals that are attributed to be a result of induced CD (ICD).39 These signals arise due to the transfer of chirality to the chromophore and are expressed as a positive Cotton effect at ca. 260 nm (S0 → S2). The ICD signals are most likely due to excitonic coupling between the anthracene and proximate bases and indicate different electronic environments experienced by the tag in various duplexes. The less intense peak for the D-isomer system would again be consistent with this stereochemistry leading to a weaker interaction with the base pair stack. However, the orientation of the anthracene is also an important consideration. The position in which the long axis is at 45° to that of the adjacent base pairs is expected to give the strongest ICD signal, compared to an angle closer to 0 or 90°. The spectrum of concentrated samples reveals an additional CD band in the 330–420 nm region (S0 → S1) which is stronger for the L-isomer duplex, again indicating a strong interaction with the base pair stack. This band has previously been observed in untethered anthracene groups that intercalate into DNA24,40 and indicates a binding mode where the anthracenyl short axis can align relative to the base pair stack.41,42 The S0 → S2 anthracene band (at ∼ 260 nm) in spectra of the mismatched 5′-CLC duplexes (see ESI†) is clearly still prominent, which again can be explained by significant interactions with the bases. However its broadening suggests that the tag can now adopt a greater number of conformations in the more loosely held duplex.
The 1H NMR spectra of the duplexes (Fig. 4) also support the existence of different local environments for the anthracene tag that affect their sensing properties. For both matching duplexes (5′-CLC/3′-GAG and 5′-CDC/3′-GAG), the two signals for the imino protons associated with the GC base pairs directly adjacent to the anthracene tag are shifted upfield compared to the unmodified control duplex, with the largest change observed in the L-linker system. While this can be explained by weaker H-bonding interactions and a greater exchange with the solvent, intercalation and stacking can also induce upfield shifts due to the ring current effect of the chromophore.43 Based on the molecular modelling studies showing a better orbital overlap of the anthracene with the upstream base (with respect to the tag) and a shorter distance of approach (vide infra), this more upfield-shifted signal is ascribed to the 3′-AG imino proton. For the mismatching 5′-CLC/3′-AAG system, it is interesting to note that the remaining H-bonded imino G signal (i.e. 3′-AA) is shifted further upfield than the equivalent base in the 3′GAG duplex, which is consistent with a greater degree of duplex insertion for the tag in the mismatched system, in support of the proposed model.
The lifetime data in Table 2 provide access to quenching rates of the excited anthracene tag, which are important for elucidating the sensing mechanism. The presence of multiple lifetimes in oligonucleotide conjugated fluorophores is not unusual and has been studied previously using 2-aminopurine (2AP) nucleobase,44 or nucleobase-appended BODIPY45 and PNA conjugated thiazole orange.27 The multiple decay parameters present in our data indicate a dynamic profile for the anthracene in which it partitions between different environments in its interaction with the duplex. These different environments are not interconverting within the lifetime of the excited states. As would be expected, the monomer L-Phos emits with a single lifetime (5.1 ns) that is somewhat shorter than that of the chromophore in a non-aqueous environment, due to quenching by water. Based on this, we attribute the 4 ns decay component (τ2) observed in all the ss and dsDNA species to the anthracene chromophore located in a predominantly aqueous environment. In the case of the L-isomer, this would correspond to the anthracene adopting an extra-helical conformation, while for the D-isomer, a minor groove conformation is also possible, according to the molecular modeling simulations (Fig. 5).
The fluorescence lifetime measurements and the time-resolved absorption data respectively indicate the presence of additional fast (ca. 1 ns) and ultrafast (ps) quenching processes in the matching 5′-CLC/3′-GAG system. Furthermore it is clear that the point variant sensing mechanism, triggered by the conversion of a matching to a mismatching base pair immediately upstream of the tag, is accompanied by the disappearance of both of these processes. In assigning these processes, we note the absence of any transient species on the nanosecond timescale ascribable to oxidized or reduced anthracene, or to the population of the anthracene triplet state. This effectively rules out the 1 ns decay process (assigned to τ1 in Table 2) being due to photoinduced electron transfer (PET) that forms the corresponding anthracene radical ion (or population of the triplet state through charge recombination following PET). Instead, it most likely corresponds to fast (8.5 × 108 s−1) internal conversion (IC) of the S1 state as a result of collisional quenching by the close proximity of neighbouring bases. It is worth noting that the lifetime measurements reveal that at least one of these fast decay pathways assigned to the τ1 decay component is also present in both ssDNA probes as well as the CLC-trimer (see ESI†). This suggests that quenching is not solely dependent on the presence of bases in the target strand, but rather on the tightness of the immediate cavity of adjacent bases in the same strand that frame the tag.
An ultrafast (ps) decay process, such as that observed for the 5′-CLC/3′-GAG duplex, could provide an explanation for the similarity in fluorescence lifetimes between the D and L systems (both unbound and bound as fully matched complexes) despite having very different quantum yields. However the question arises as to whether this picosecond decay is also due to collisional interactions or alternatively a PET process. There is indeed evidence for the formation of an anthracene radical ion whose transient decay is detected at 690 nm (see ESI†). However the increase in signal intensity is weak and its rate of formation (k ≈ 2.5 × 1012 S−1) is in fact faster than that of the ultrafast decay component of the S1 state (270 ps). Therefore it would appear that collisional quenching by proximate bases is the most likely explanation for both the ultrafast and fast decay processes of the S1 state, which at least in the case of the 5′-CLC/3′-GAG system, start in the picosecond domain and continue into the low nanosecond domain. However the fact that this additional ultrafast electron transfer process is also absent in the mismatched 5′-CLC/3′-AAG system might suggest that the guanine base in the upstream position from the anthracene tag in the matched target (i.e. 3′-AG) is involved in this process through its oxidation.26 This would thereby result in an increase in fluorescence emission intensity by its removal (i.e. the formation of a mismatch). However, such a reliance on the presence or absence of guanine in the target strand for the observed base variant sensing behaviour would be at odds with our previous studies16 that have demonstrated the same mismatch-induced sensing effects when targeting C-to-A and T-to-A variations (i.e. targets not involving pairing with a guanine base in the matching target strand).
The molecular dynamics models are useful for visualising the environment around the anthracene in terms of how it might effect possible decay pathways for the excited state. For example, the cavity around the anthracene widens to some extent in the mismatched system (Fig. 6), which could account for the absence of the τ1 pathway involving collisional quenching by proximate base pairs. Another important consideration in this respect is the third and longest decay component, τ3 (range 9–14 ns). The wider cavity would give more room for the tag to insert itself further into the duplex, placing it into a more hydrophobic environment, where it is protected from quenching water molecules. In fact, this decay parameter is certainly similar to the lifetime of 9-alkoxyanthracene derivatives in non-polar organic solvents (7 ns).46 The absence of τ3 in the CLC trimer (ESI†) and guanine tagged anthracenes47 is in agreement with this hypothesis, as these conjugates have more open structures that would preclude conformations in which quenching processes from water and proximate bases could be avoided.
The question as to why only a change to the upstream position (with respect to the tag, i.e. 3′-AG) gives rise to the sensing effect can most likely be explained in terms of the anthracene-base distance and/or overlap. Once again, the molecular dynamics simulations are informative as they consistently show a better orbital overlap of anthracene with this base and a shorter distance of approach.32 They also shed considerable light on the importance of the threoninol linker stereochemistry. In the case of the matching duplex 5′-CLC/3′-GAG, the stereochemistry of the linker allows the anthracene to intercalate into the centre of the duplex via the major groove (Fig. 5), to the extent that the tag can locate itself in approximately the same position as a natural nucleobase. As such, it benefits from considerable overlap with the adjacent nucleobases and the resulting stabilisation through hydrophobic pi-stacking interactions.48 At the same time it does not clash with the adenine opposite and allows a natural, un-kinked structure to be maintained by the duplex. Hence, the melting point for this duplex is almost as high as that of unmodified DNA (Table 3). In contrast for the matching duplex with the D-threoninol isomer, the stereochemistry is such that duplex intercalation is much less favoured (Fig. 5, see also ESI†), with the anthracene instead orientating itself towards the minor grove, leading to a lower melting point. For this duplex, the models indicate that the adenine directly opposite the tag follows suit by flipping out of the duplex, presumably to maximise the stacking interactions between the residual H-bonding base pairs either side of the modification site.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ob01710g |
‡ In large scale genotyping to identify many variants within a single genome, high throughput sequencing technologies are typically used. |
§ With other read-out methods emerging, the term base discriminating probe (BDP) is suggested as a more convenient descriptor than the hitherto more commonly used base-discriminating fluorophore (BDF). BDFs have in any case often referred only to probes that contain a fluorophore tag directly attached to a nucleoside (e.g. see ref. 6), as opposed to a non-nucleosidic linker, as is the case with this work. |
This journal is © The Royal Society of Chemistry 2018 |