Streamlining effects of extra telomeric repeat on telomere folding revealed by fluorescence-force spectroscopy

Tandem repeats of guanine rich sequences are ubiquitous in the eukaryotic genome. For example, in the human cells, telomeres at the chromosomal ends comprise of kilobases repeats of T2AG3. Four such repeats can form G-quadruplexes (GQs). Biophysical studies have shown that GQs formed from four consecutive repeats possess high diversity both in their structure and in their response to tension. In principle, a GQ can form from any four repeats that may not even be consecutive. In order to investigate the dynamics of GQ possessing such positional multiplicity, we studied five and six repeats human telomeric sequence using single molecule FRET as well as its combination with optical tweezers. Our results suggest preferential formation of GQs at the 3’ end both in K+ and Na+ solutions although minority populations with a 5’ GQ or long-loop GQs were also observed. Using a vectorial folding assay which mimics the directional nature of telomere extension, we found that the 3’ preference holds even when folding is allowed to begin from the 5’ side. Interestingly, the unassociated T2AG3 segment has a streamlining effect in that one or two mechanically distinct species was observed at a single position instead of six or more observed without an unassociated repeat. Location of GQ on a long G-rich telomeric overhang and reduction in diversity of GQ conformations and mechanical responses through adjacent sequences have important implications in processes such as telomerase inhibition, alternative lengthening of telomeres, T-loop formation, telomere end protection and replication.


Introduction
As early as 1962, guanosine moieties were known to self-assemble into tetrameric structures via Hoogsteen hydrogen bonds (1). Planar association between four guanines by eight hydrogen bonds in the presence of a central coordinating monovalent cation (e.g. Na + and K + ) generates a G-quartet or a G-tetrad (2,3). Stacking of two or more G-quartets creates a very stable DNA secondary structure called the G-Quadruplex (GQ) (4). Complete sequencing of human genome has identified about 300,000 putative sequences that can fold into GQs (5). Expansion of GQforming motifs have been implicated in pathogenicity associated with several human neurological disorders (6). Additionally, GQ-forming sequences have been reported in various viral genomes including human immuno-deficiency virus, Epstein-Barr virus and human papillomavirus (7).
Despite an apparently simple, repeating G-rich sequence, structural studies with circular dichroism (CD), nuclear magnetic resonance (NMR) and X-ray crystallography have revealed extreme polymorphism in human telomeric GQs composed of four T2AG3 repeats (18)(19)(20)(21)(22)(23)(24). Such diversity suggests co-existence of different conformations in the telomere with implications on rational designing of GQ-targeting drugs. GQ polymorphism and conformational dynamics in longer telomeric DNA containing five or more T2AG3 repeats, which should better mimic a telomere overhang, have been examined only in a few studies (25)(26)(27)(28)(29).
GQ formation requires a minimum of four G-rich segments (30). As human telomeric DNA contains multiple hexanucleotide (T2AG3) repeats and stretches for several kilobases, it has been assumed that GQs can form anywhere along the G-rich strand (31). Hence, under physiological conditions, a long telomeric DNA with more than four T2AG3 repeats can harbor different forms of terminal structures depending on the position of GQ(s). GQ locations on telomeric overhangs can modulate telomerase extension or mechanism of alternative lengthening of telomere (ALT) (14). An exonuclease hydrolysis assay suggested preferential GQ formation via association of four consecutive hexanucleotide repeats at the 3' end of long telomeric DNA, leaving out a T2AG3 repeat at the 5' end (26). Another study employing site-specific labeling of guanine residues suggested the presence of hybrid GQ structures, incorporating one or more hexanucleotide repeat within a single loop to connect adjacent Gs in a G-quartet (29). Ensemble average techniques used in the above studies are limited by time resolution and hence cannot univocally deconvolute individual species from a heterogeneous population. Dynamic exchange between GQ conformations in telomeric DNA spanning five to seven T2AG3 repeats have been observed using single molecule fluorescence resonance energy transfer (smFRET) (32) and single molecule optical tweezers studies of human telomeric DNA of four to seven repeats indicated that GQs more frequently form using four consecutive repeats (25). In a recent study we demonstrated extreme mechanical diversity heterogeneity of GQs in 22 nt long human telomeric DNA (four telomeric repeats, where at least six different mechanically distinct species were observed (30)). However, diversity in mechanical responses of GQs formed in longer telomeric DNA is yet to be explored.
In this study, we employed smFRET to probe GQ formation of human telomeric DNA, spanning five and six T2AG3 repeats. Using guanine to thymine (G to T) mutation at the central G of a T2AG3 repeat, which is known to restrict the participation of the repeat in GQ formation (24,30), we created variants without positional multiplicity to help interpret the data. Using smFRET and its combination with optical tweezers, which we refer to as fluorescence-force spectroscopy, we found evidence for preferential formation of GQs at the 3' end of five-repeat telomeric DNA, with GQs formed at the 5' end or at an internal position representing minority populations. Such GQs formed at the 3' end unravel non-cooperatively via strand slippage, contrary to cooperative unfolding of canonical GQs formed from four telomeric repeats (30). Furthermore, using a superhelicase-based vectorial folding assay, we show that a GQ form preferentially at the 3' end even when it is allowed to fold starting from the 5' side. Although constituting a minor population, GQs formed at the 5' end of a five-repeat telomeric DNA show extreme mechanical stability and cannot be perturbed by forces of ~ 28 pN. GQs preferentially formed at the 3' end of six telomeric repeats and manifested similar mechanical properties. Overall, we found that an unassociated T2AG3 repeat has a streamlining effect on GQ conformations in that one or two mechanically distinct conformations are observed at a single location instead of six or more observed for GQs without an unassociated repeat.

DNA constructs
All DNA oligonucleotides were purchased from Integrated DNA Technologies. The five and six telomeric repeat sequences referred to as hTel28 and hTel34 respectively are as follows:

Fluorescence-force spectroscopy
An integrated single molecule fluorescence-optical trap instrument was recently developed in our lab to probe conformational changes of biomolecular systems under tension (34,35). In short, an optical trap was formed by an infrared laser (1064 nm, 800 mW (maximum average power), EXLSR-1064-800-CDRH, Spectra-Physics) through the back port of the microscope (Olympus) on the sample plane with a 100X immersion objective (Olympus). Tension was applied on the sample tethers by translating the piezo-electric stage that holds the microscope and the applied force was read out via position detection of the tethered beads using a quadrant photodiode (UDT/SPOT/9DMI). The instrument was calibrated as described previously (34,35). A confocal excitation laser (532 nm, 30 mW (maximum average power), World StarTech) was used to scan the sample via a piezo-controlled steering mirror (S-334K.2SL, Physik Instrument). Two avalanche photodiodes were used to record the fluorescence emission, filtered from infrared laser by a band pass filter (HQ580/60 m, Chroma) and excitation by a dichroic mirror (HQ680/60 m, Chroma.

Data acquisition
In the absence of force, a prism-type total internal reflection microscope, with 532 nm laser excitation and back-illuminated electron-multiplying charge-coupled device camera (iXON, Andor Technology, South Windsor, CT) was used to detect smFRET signals (33). The donor and acceptor intensities, i.e. ID and IA respectively were corrected for background signals and crosstalk and used to estimate smFRET efficiency E using IA/(IA+ID). E histograms were constructed by averaging the first ten data points of each molecule's time trace.
A detailed data acquisition procedure for integrated single molecule force-fluorescence spectroscopy has been described by Hohng et al (34). Briefly, a tethered bead was trapped and its origin was determined by stretching the tether along opposite directions along x and y-axes. The These derived parameters (x ‡ , u(0), and G ‡ ) were then used to reconstruct the force profiles (36,37).

Circular Dichroism (CD) Spectroscopy
CD spectra of the human telomeric oligonucleotides were recorded on an Aviv-420 spectropolarimeter (Lakewood, NJ, USA), using a quartz cell of 1 mm optical path length. The oligonucleotides were diluted to 20 M in a buffer containing 20 mM Tris pH 8 and appropriate concentration of K + /Na + ions. Before each measurement, the oligonucleotides were heated to 90 o C for ~ 5 min and slowly cooled to room temperature, to avoid formation of intermolecular structures. An average of three scans was recorded between 220 and 320 nm at room temperature. The spectra were corrected for baseline and signal contributions from the buffer.

Vectorial GQ folding using a superhelicase Rep-X
In order to eliminate end-effects of the sequence on Rep-X unwinding and GQ folding, we Hua et al (38).

SmFRET analysis of five repeats human telomeric DNA
Our hTel28 construct consists of a central 28 nt long human telomeric DNA sequence spanning five hexanucleotide repeats, (GGG(TTAGGG)4T), flanked by two duplex handles at the 5' and 3' ends. CD spectrum of hTel28 in 100 mM K + showed a positive peak at ~ 289 nm with shoulders at 265 and 255 nm and a negative peak at 235 nm, suggestive of hybrid GQs (Figs. 1a, S1a) (39). The construct was immobilized on to the PEG-passivated surface via the duplex stem at the 5' end whereas a 3'end partial duplex served as a tether to the -DNA for use in fluorescence-force spectroscopy. The donor (Cy3) and acceptor (Cy5) fluorophores were placed adjacent to the ends of the telomeric sequence such that FRET efficiency, E, reflects the end-toend distance of the telomeric sequence (Fig. 1b).
Formation of secondary structures was first examined through smFRET as a function of K + concentration (Fig. S1b). The peak centered at E = 0 represents molecules with a missing or inactive acceptor, and hence can be ignored. In the absence of K + , we observed a peak centered at E ~ 0.16. Transition into secondary structures commenced between 2 and 10 mM K + , as  extension to the 28 nt long human telomeric repeat GGG(TTAGGG)4T was annealed to a 18 nt long biotinylated strand and immobilized on a PEG-passivated quartz surface through biotin-neutravidin interaction. The 5' end of the biotinylated strand is labeled with Cy5 (acceptor). FRET was measured between Cy3 and Cy5. The main hTel28 strand is labeled with Cy3 (donor) at the 3' end of the telomeric repeat and is followed by dT17 and an 18 nt extension that is annealed to 30 nt long -bridge. (c) E histograms of hTel28, hTel28.1, hTel28.3 and hTel28.5 in 100 mM K + concentration. (d) Schematic representation of GQ formation after site-specific G to T mutation of hTel28.
In a five-repeat telomeric sequence, GQs can potentially form at the 3' or 5' ends by association of four consecutive T2AG3 repeats (3' and 5' GQs respectively), or with an inner long loop bearing the second, third or fourth T2AG3 segment (long-loop GQs). Therefore, at least five different GQ structures can be conceived depending on which repeat is excluded (Fig. S2). In order to shed light on the conformational landscape, mutations can be introduced in biomolecules to selectively depopulate certain species from a heterogeneous population (24,29,40). Previous studies have also shown that guanine to thymine substitution of the central G of a hexanucleotide repeat T2AG3 precludes its participation in GQ formation in 100 mM K + or Na + (24,30) S1c).

Vectorial folding of hTel28, mimicking telomerase action
Our smFRET studies under refolding conditions suggest predominant GQ formation at the 3' end of hTel28s. However, because a telomerase adds one repeat at a time to the 3' end of a telomeric DNA under a physiological situation of telomere elongation, a GQ may first form on the 5' side during telomere elongation. To address this point, we employed a highly processive superhelicase Rep-X to mimic the vectorial nature of telomerase action (38,41,42). In a previous study, we used this 5' to 3' DNA helicase to unwind RNA/DNA heteroduplex and to reveal the RNA strand in the direction of transcription (5' to 3') at the speed of transcription to mimic cotranscriptional RNA folding (38). Here, we annealed hTel28 to its complementary C-rich DNA and Rep-X was loaded through the 3' overhang in the absence of MgCl2 and ATP (Fig. 2a).
Duplex unwinding was then initiated by addition of MgCl2 and ATP. The helicase translocates in the 3' to 5' direction along the C-rich DNA strand, unwinding the duplex and releasing the G-rich DNA in the 5' to 3' direction, which is the direction of telomere extension. The disengaged G-rich hTel28 can fold into GQ as it is being revealed starting from the 5' side (Fig. 2a). GQ folding in nascent telomeric DNA synthesized during telomerase extension has been directly observed by Jansson et al (43).
In the fully annealed state, hTel28-duplex showed a FRET peak at E ~ 0.07 (Fig. 2b). Rep-X mediated unwinding reaction was quenched after a minute by washing away ATP and the resulting smFRET distribution showed that ~ 47 % of molecules fold into a GQ with a major peak at E ~ 0.73 (Fig. 2b). The rest stayed at E ~ 0.07 probably because these molecules did not get unwound. A control experiment performed with a slowly hydrolysable ATP analog, AMP-PNP, showed only ~ 5 % folded population (Fig. S3d). Under identical unwinding buffer conditions, major peaks centered at E ~ 0.72 and 0.62 were observed in hTel28.1 and hTel28.5, respectively (Fig. S5a). Hence, the major state observed at E ~ 0.73 was attributed to 3' GQs.

Dynamics of hTel28 under tension
For fluorescence-force spectroscopy, the hTel28 construct was annealed with -DNA via the bridge prior to its immobilization on the PEG-passivated surface. The other end of the -DNA was tethered to an optical trapped bead (Fig. 3a).  Average E vs force response of hTel28.1 (e), hTel28.5 (g) and hTel28.3 (h). All single molecule trajectories were collected at a time resolution of 20 ms. Error bars represent standard errors.
The most common was Type I partial unfolding, defined by stable E values of ~ 0.75 up to ~ 2 pN, followed by a gradual decrease to ~ 0.25 at ~ 28 pN (Fig. 3b, cycles 2 and 3). On relaxation, the unraveling pathway was retraced without any significant hysteresis (Fig. 3c). Type II partial unfolding also showed reversible E changes but with a different range, from ~ 0.68 to ~ 0.37 ( Fig. 2g).

Interestingly, a single molecule of hTel28 under tension can interchange between complete and
Type I partial unfolding. For example, Fig. 3b shows complete and partial unfolding in consecutive pulling cycles 1 and 2 and vice versa in the cycles 3 and 4. Overall, switching from complete to Type I partial unfolding in consecutive cycles and vice versa were observed in ~ 33 % (19 out of 58 events) and ~ 16 % (24 out of 150 events) pulling cycles, respectively. However, Type II partial unfolding events rarely switched to other types of unfolding in the subsequent cycle (4 events, ~ 7 %).

Conformational dynamics of hTel28 variants under tension
We next examined the hTel28 variants under tension in order to gain mechanistic insight into GQ formation in five-repeat telomeric DNA. In 100 mM K + , hTel28. We next studied the conformational dynamics of hTel28.3 as a representative of long-loop GQs.
Because the hTel28.3 harbors a G to T mutation in the innermost T2AG3 segment, a GQ would form preferentially with the four outer repeats with an inner 9 nt long loop (Fig. S5a) (Fig. 3h).
In all, mechanical behavior of hTel28 was recapitulated by the hTel28 variants studied. We can attribute complete unfolding to long-loop GQs, and attribute Type II partial unfolding to 5' GQs.
Type I partial unfolding was observed in both long-loop GQs and 3' GQs. Interestingly, among the variants, mechanical heterogeneity could be noted only in long-loop GQs (hTel28.3).
Combining these results with zero-force smFRET studies, we propose that in 100 mM K + , the 3' GQs constitute the preferred conformation of hTel28.

Effect of an unassociated G-rich segment at the 3' end of GQ
We previously classified diverse mechanical behavior of hTel22 that contain four telomeric repeats into (1) abrupt and complete unfolding, (2) partial unfolding and (3) an ultrastable population that could not be perturbed by forces up to ~ 28 pN (30). GQs formed at the 3' or 5' ends of hTel28 would have a 6 nt overhang to an hTel22-like GQ core. Therefore, an ultrastable GQ with a pendant 6 nt tail would show gradual E decreases as the ssDNA tail is stretched. Thus, the reversible E changes between ~ 0.68 and ~ 0.39 under applied forces shown by 5' GQs and classified as Type II partial unfolding of hTel28 may be due to an ultrastable GQ population. On the other hand, Type I partial unfolding which entails a greater degree of unraveling (Fig. S5c) can be due to partial unfolding via mutual slippage of the G-rich strands as proposed to occur in hTel22 (30). Interestingly, Type I partial unfolding, although dominant in hTel28, was observed only in ~ 11 % of hTel22 molecules (30). Hence, in order to investigate the effect of a short addendum at the 5' end of GQ, we substituted the 6 nt T2AG3 segment at the 5' end with (dT)6 (hTel28.1T). Thus, hTel28.1T serves as a 22 nt long GQ forming sequence with a pendant (dT)6 at its 5' end.
hTel28.1T folded in a K + concentration-dependent manner (Fig. S6a) Type II partial unfolding, supporting our assignment of hTel28.5 to an ultrastable GQ. In the second and third pulling cycles of Fig. 4b, we observed a gradual decrease in E from ~ 0.77 due to ssDNA stretching, followed by abrupt unfolding (Fig. 4b). Refolding also occurred abruptly (Figs. 4b and d). This second type of behavior showed a wide range of unfolding forces (~ 1 and 28 pN) (Figs. 4c and S6b), similar to cooperative unfolding observed for hTel22 (30). The funfold values (131 events) can be grouped into three force clusters centered at ~ 3.5, 10 and 20 pN (Fig.   4c). We also observed a third type of unraveling behavior, characterized by gradual and reversible changes from E ~ 0.77 to ~ 0.25 (24 events, ~ 10 %) and reminiscent of Type I partial unfolding of hTel28 (Fig. S6d) (30). In all, we could recapitulate conformational dynamics of hTel22 with hTel28.1T. The stark contrast in mechanical behavior of hTel28.1 and hTel28.1T suggests that GQ conformations are affected by the unassociated G-rich strand at the 5' end in a way that a (dT)6 cannot mimic.

Formation of GQs in six-repeat human telomeric DNA
We next investigated the formation of GQs in six-repeat human telomeric DNA, hTel34 (Fig.   5a). Like hTel28, CD spectra of hTel34 in 100 mM K + shows signatures of hybrid GQs (Fig. 5b) (39). Secondary structure formation commenced at < 10 mM K + (Fig. S7a), and in 100 mM K + , we observed two non-interconvertible (within our observation time) major populations, centered at E ~ 0.54 and 0.71, that we designate as E1 and E2 states (Fig. 5c, Fig. S7b). A minor population at E ~ 0.33 (E0) is similar in mechanical response to (dT)34, an unstructured ssDNA of the same length ( Fig. 5i and 5j), and therefore is likely an unfolded population. and S7d). Switch from E2 to E1 after a pulling cycle accounted for ~ 29 % (14 out of 48 events, Fig. 5d (cycle 4)) of the events originating at E2 whereas a E1 to E2 switch was observed in ~ 13 % (12 out of 89 events, Fig. 5d (cycle 2)) of the events initiating from E1.
With two additional T2AG3 repeats, hTel34 can form 15 GQ structures that differ in which four of the six repeats are chosen (Fig. S2). In order to gain information on its internal architecture we

Conformational dynamics of hTel28 and hTel34 in 100 mM Na +
We next investigated if the preferential GQ formation at the 3' end of more than four telomeric repeats is dependent the cations present. Instead of hybrid GQ signatures in 100 mM K + , CD spectra of hTel28 in 100 mM Na + showed signatures of antiparallel GQs (Fig. S9a) (45). In the absence of force, we observed a major population at E ~ 0.72 (Fig. S9b). Under tension, unfolding predominantly occurred via an abrupt change in E from ~ 0.72 to ~ 0.1 at forces peaked at ~ 8 pN (Figs. S10a, b and d) and refolding occurred at lower forces of ~ 2.5 pN (Fig.   S10e). Additionally, Type II partial unfolding was observed in ~ 30 % (31 out of 104 events) of pulling cycles (Figs. S10a (cycle 3) and b). We did not observe Type I partial unfolding.
Among the hTel28 variants, stable GQs were formed only in hTel28.1, suggesting that GQ forms preferentially in the 3' end also in Na + solution (Fig. S9b). Other variants did not form stable GQs in Na + likely because of poorer coordination by smaller Na + compared to K + (46).
Under tension, hTel28.1 predominantly showed Type II partial unfolding in 100 mM Na + (87 out of 114 events, ~ 76 %) (Figs. S10f and g) with the rest showing complete unfolding with two funfold clusters centered at ~ 5 and 19 pN (Figs. S10f-i). We next investigated hTel28.1T to test if an unpaired T2AG3 has any additional effect in Na + that (dT)6 does not have. hTel28.1T showed a major population centered at E ~ 0.75 in 100 mM Na + (S11a). Under tension, it predominantly showed abrupt, complete unfolding with two funfold clusters centered at ~ 7 and 15 pN (Fig. S11b) with refolding occurring at lower forces (Fig. S11c). The rest (~ 20 %, 22 out of 110 events) showed Type II partial unfolding. Overall, hTel28.1T was very similar to hTel28, both in its zero-force E value and its mechanical response. Thus, the GQ-core in Na + remains unaffected by an additional T2AG3 overhang at the 3' end. hTel34 adopted anti-parallel GQ-like conformation in 100 mM Na + and showed a single population centered at E ~ 0.5 (Figs. S12a and b) (45). The mechanical response of hTel34 was gradual and reversible similar to E1 in 100 mM K + (Figs. S12c and d). However, this does not necessarily report on the dynamics of GQ within hTel34 because stretching of a single stranded overhang, if present, can overwhelm the response of the underlying GQ. To test the above possibility, we next interrogated hTel34' behavior in 100 mM Na + because it can report on the shorter, 22 nt, segment located in the 5' end of hTel34. The major FRET population at E ~ 0.72

Discussion
Despite its simple repetitive sequence, telomeric GQ possess extremely heterogeneous populations. Structural studies have revealed at least six different GQ conformations (e.g. parallel, antiparallel, hybrid etc.) based on strand symmetry, strand orientation and glycosidic conformation (18)(19)(20)(21)(22)(23)(24). Recently, we identified six mechanically different, interconvertible GQ species in a 22 nt long human telomeric DNA using single molecule fluorescence-force spectroscopy (30). In the context of longer telomeric sequences harboring more than four T2AG3 repeats, as we studied here using five repeats and six repeats using fluorescence-force spectroscopy, GQ polymorphism in tandem with their positional multiplicity along the DNA add to the complexity. Indeed, previous studies of four to seven repeats using smFRET only (hence at zero force) (32) or optical tweezers only (25) found evidence of multiple states. Using a series of mutants designed to mimic positionally defined GQs, we were able to separate positional multiplicity and GQ polymorphism.

Five telomeric repeats
Here, we found that the mechanical response of GQ in a long telomeric DNA can depend on its location (Fig. 6a). For example, although both the 3' GQ and the 5' GQ in hTel28, mimicked by hTel28.1 and hTel28.5, respectively, utilize four consecutive T2AG3 segments, on application of force, the 3' GQ former unravels gradually whereas the 5' GQ withstands forces as high as 28 pN (Fig. S4c). This may stem from differences in the underlying GQ conformations as evidenced    (30).
In all, our study highlights the regulatory role of an unassociated G-rich strand on the GQ core of a five-repeat telomeric sequence. A T2AG3 overhang at the 5' end biases the molecule toward the partial unfolding pathway whereas when present at the 3' end it renders GQ ultrastable. On the other hand, an inner loop harboring a T2AG3 segment manifests cooperative unfolding similar to canonical hTel22-GQs (30). We hypothesize position-dependent transient interactions between the unpaired T2AG3 segment with the underlying GQ core, which directly affects the conformational distribution by streamlining a mechanically diverse population to a single conformation (Fig. 6). Interactions between unpaired G-nucleotides and the GQ core was also observed in the GQ-forming c-Myc promoter (49). However, detailed structural studies are required to understand the exact nature of such interactions. Nonetheless, at applied forces of up to ~ 28 pN, because the 3' GQs in hTel28 are only partially unraveled, they rarely convert into long-loop GQs (~ 16 %) upon relaxation. In contrast, a long-loop GQ, when unfolded cooperatively, loses memory of its initial state and can relax into a mechanically different longloop GQ or even a 3' GQ (~ 33 %).

Six telomeric repeats
From hTel34 which contains six human telomeric repeats, we observed two major populations E1 and E2 at zero force. To reduce the compounding effect of stretching of ssDNA tail(s), we also designed the internally labeled construct hTel34' whose FRET response is sensitive to the conformation of the four 5' most repeats. By comparative analysis of responses of hTel28 and hTel34, we assigned the major population, E1 (or E1') of hTel34 (or hTel34') to 3'GQ with a 12 nt ssDNA overhang harboring two T2AG3 segments. The minor populations, E2 (hTel34) and E2' (hTel34') show abrupt, cooperative unfolding under tension similar to long-loop GQs in hTel28.
The funfold of E2 reveals two force clusters at ~ 8 and 16 pN with underlying x ‡ s of ~ 3.7 and 4 nm respectively (Fig. S13b). Although similar in x ‡ , the cluster at funfold=16 pN had significantly higher u(0) and G ‡ (10990 vs 59.83 s and 11.6 vs 5.7 kBT), indicating presence of long and short-lived species that are structurally similar (24). The funfold clusters in E2' correspond to x ‡ s of ~ 3.6 and 3.7 nm respectively, similar to E2. Hence, we suggest that E2 of hTel34 are E2' of hTel34' represent the same conformation (Fig. S13b).
GQs formed at the 5' end of hTel28 were ultrastable. Analogous behavior might be overshadowed by ssDNA stretching in hTel34, but should be reflected as stable high FRET over the complete pulling cycle in hTel34'. Although E2' showed a no-force FRET value of ~ 0.88, expected for a 5' GQ, it did not show stable high FRET under tension. Rather, E2' showed abrupt unfolding, reminiscent of long-loop GQs in hTel28. So we suggest that E2 of hTel34 is unlikely due to a 5' GQ. Can it be due to a 3' GQ? A 3' GQ would leave a 12 nt ssDNA region in hTel34' which can be stretched under tension, giving rise to initial gradual FRET decrease. Because we did not observe any FRET decrease until abrupt unfolding, we can also rule out the possibility that E2' and E2 correspond to a 3' GQ. Overall, our data suggest that E2 represents long-loop populations although we cannot tell where there is a single long loop of 15 nt in length or two separate 9 nt long loops. The difference in transition distances between hTel34 and hTel28 may be attributed to structural differences stemming from the size and arrangement of the intervening loop(s).

Dynamics in Na + solution
In 100 mM Na + , hTel28 showed the abrupt unfolding response in majority of events, reminiscent of hTel22. We estimated x ‡ , u(0) and G ‡ as ~ 4.3 nm, ~ 50 s and 6.5 kBT, respectively.
Among the variants of hTel28, only hTel28.1 showed stable folding in Na + , suggesting that 3' GQ formation is favored also in Na + . Upon substituting the hexanucleotide G-rich segment at the 5' end with (dT)6, we observed two funfold clusters corresponding to long-lived (u(0) ~ 1190 s, funfold ~ 14.8 pN) and short-lived (u(0) ~ 68 s, funfold ~ 6.8 pN) isostructural species with x ‡ of ~ 4 nm, which is close to a previously reported value for hTel22 ( Fig. S15) (30). Thus, GQ conformation is not significantly affected by the presence of a T2AG3 overhang at the 5' end in 100 mM Na + . We estimated similar x ‡ s in 3' GQ forming hTel28 and hTel34' (Fig. S14). This substantiates our hypothesis that GQ forms preferentially at the 3' end of hTel34 also in Na + .
Taken together, our study suggests a complex, sequence dependent interplay between GQ and the unpaired Gs in five and six-repeat telomeric DNA. We note that, the predominant Type I partial unfolding observed in hTel28 and hTel34 is exclusive to 100 mM K + . Thus, although GQ formation is polarized at the 3' end in Na + and K + , the streamlining effect of an associated telomeric repeat on GQ folding appears unique to K+ conditions. and lengthening by ALT, respectively (14,(50)(51)(52). Thus GQ formation at the 3' end of the telomeric DNA is likely to inhibit the aforementioned processes. On the other hand, although a population minority, the long-loop GQ species might be more susceptible to helicase-catalyzed by providing an internal loading site for helicases (53). Telomeres are assumed to be protected by T-loop formation, which involves invasion into the double-stranded telomeric region by the G-rich telomeric overhang and subsequent annealing with the C-rich strand (54). Stable GQs at the 3' end might interfere with T-loop formation as well. In the cellular milieu, processes such as transcription and replication generate tension. However, our study shows that long telomeric repeats cannot be fully unwound at average stall forces of motor proteins and polymerases (55,56). Thus, GQs can act as potential kinetic traps and inhibit replication and transcription.