• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Structure-based simulations complemented by conventional all-atom simulations to provide new insights into the folding dynamics of human telomeric G-quadruplex*

    2021-07-30 07:43:40YunQiangBian邊運(yùn)強(qiáng)FengSong宋峰ZanXiaCao曹贊霞JiaFengYu于家峰andJiHuaWang王吉華
    Chinese Physics B 2021年7期

    Yun-Qiang Bian(邊運(yùn)強(qiáng)), Feng Song(宋峰), Zan-Xia Cao(曹贊霞),Jia-Feng Yu(于家峰), and Ji-Hua Wang(王吉華)

    Shandong Key Laboratory of Biophysics,Institute of Biophysics,Dezhou University,Dezhou 253023,China

    Keywords: molecular dynamics simulation,structure-based model,all-atom model,DNA G-quadruplex

    1. Introduction

    Functions of proteins and nucleic acids are supported by their folded structures. Therefore, correct folding is a crucial step for them to achieve their physiological significance.In recent years, the folding problem of DNA G-quadruplex(GQ) structures,[1-3]a family of noncanonical four-stranded helices built by G-quartets (Fig. 1(a)), has attracted much attention.[4-25]One of the most controversial systems is the human telomeric (Htel) GQ, which plays relevant roles in genome stability and anticancer therapies.[26-28]It is well known that the Htel GQ is extremely polymorphic.[29]Influenced by the kind of monovalent cations and other factors, the Htel sequence can fold into a parallel, antiparallel or hybrid GQ topology (Fig. S1 in the Supplementary Material).[30-42]This outcome indicates that the folding landscape contains several well-separated deep basins,and multiple different folds may coexist at the thermodynamic equilibrium.As a result, Htel GQ folding is expected to follow a typical kinetic partitioning or multiple-pathway model, according to which only a fraction of the molecules fold to the native structure rapidly and the other fractions become trapped in various competing basins of attraction.[43]

    For Htel GQ, the identification of competing folds is difficult in experiments, since their measurable signals may mutually overlap. However, several studies have indeed detected the kinetic partitioning folding behavior involving different folds.[44-46]In a study conducted by Marchand and coworkers on the folding of various Htel and c-myc sequences,using electrospray mass spectrometry and CD experiments,they suggested a competition between four three-quartet GQs and two two-quartet GQs.[44]In another study reported by Bessiet al., the sequence Htel24 (dTTGGG[TTAGGG]3A)was straightly demonstrated by real-time NMR spectroscopy to fold via a kinetic partitioning between the hybrid-1 (H1)and hybrid-2(H2)GQs in the presence of K+(Fig.1(b)).[46]Of particular interest is that the H2 conformation was unveiled to be kinetically more accessible. Despite of these explicit experimental demonstrations, understanding of the detailed kinetic partitioning folding process is still largely incomplete in experiments due to the temporal and spatial limits of techniques. Molecular dynamics(MD)simulations are deemed to be powerful methods to provide complementary information for experiments. Accumulated studies have demonstrated that such simulations indeed provide basic insights into the structural dynamics by focusing on specific regions of the conformational space.[47-51]For instance, by conducting extensive atomistic simulations, Stadlbaueret al.have systematically investigated the structural stability and lifetimes of various potential folding intermediates of GQs, such as G-hairpins,[47]G-triplexes[48]and late-stage intermediates.[49]However, because the kinetic partitioning folding landscape is considerably rugged, contemporary all-atom simulations are still not sufficient to sample the full folding landscape. Interestingly,in recent studies, coarse-grained models have been validated to be effective in investigation of Htel GQ folding.[52-54]For example,using a high-resolution coarse-grained DNA model,Cragnoliniet al.provided a direct view of the interconversion between different GQ folds.[53]Another example is the hybrid atomistic structure-based model,a variant of the structurebased model or Gˉo-like model which is widely used to study the folding of proteins and nucleic acids.[55-58]We have validated the effectiveness of the hybrid atomistic structure-based model in the investigation of GQ folding.[59,60]This model is more appropriate to study the folding of DNA GQs than the traditional structure-based model. It is known that the reorientation of glycosidic configurations is an important aspect to complicate the folding of DNA GQs. Traditional structurebased models cannot correctly characterize this conformational transition because local conformations are strongly biased to the native conformations.

    Fig. 1. Schematic illustration for the G-quartet and the two GQ topologies formed by human telomeric sequence Htel24.(a)The structure of a G-quartet stabilized through eight Hoogsten bonds between four guanine bases. Also shown are the syn and anti glycosidic configurations of a guanine base depending on the χ dihedral. (b) The sequence and two GQ folds of Htel24.The red and blue squares denote the guanine nucleotides with syn and anti configurations,respectively.

    In contrast,the hybrid atomistic structure-based model removes the favor of local native conformations as it borrows descriptions of local interactions from classical force fields,and thus it is capable of describing the reorientation of glycosidic configurations more accurately. By developing a doublebasin hybrid atomistic structure-based model, we studied the competition between H1 and H2 conformations in the folding of Htel24 and successfully reproduced the experimentally observed kinetic preference of H2.[60]

    As mentioned above,exploring the full folding landscape of Htel GQ is still unaffordable for conventional atomistic simulations. Although coarse-grained models can improve the sampling efficiency, they are suffered from the problem of low accuracy in comparison with conventional atomistic models. To alleviate this problem, employing a multiscale framework that merges the high accuracy of all-atom models and the high efficiency of coarse-grained models becomes a reasonable way. Developing such multiscale methods has also been an attractive issue in the field of biophysics. Available examples are the resolution-exchange MD simulations[61,62]and the atomic interaction-based coarse-grained(AICG)model.[63]In the resolution-exchange method,multiple replicas with different levels of coarse-graining are employed and they are allowed to exchange with each other through a metropolis criterion. This strategy is able to largely improve the sampling efficiency of all-atom models because the conformation space sampled by the low-resolution replica can also be visited by the all-atom replica during the simulation. Similar methods include the multigraining MD[65]and the multiscale essential sampling.[66,67]Compared with the strategy of coupling coarse-grained and atomistic models directly in a simulation as in the resolution-exchange method,the AICG model adopts a different strategy that refines the coarse-grained potential function by performing a set of all-atom simulations. For example, one can define new coarse-grained angle and dihedral potentials to characterize the flexibility of protein chains more accurately by extracting structural information from all-atom simulations on corresponding fragments.[64]Several other methods, such as the Bolzmann inversion[68]and the force-match,[69,70]also employ this refine framework. Other multiscale simulation strategies can be found elsewhere.[71-75]In this study,we developed a multiscale simulation framework to couple the hybrid atomistic structure-based model and the all-atom model by two steps. First, a large number of conformations were acquired by performing structure-based simulations. Then,a set of conformations were selected as seeds to conduct large-scale conventional all-atom simulations after reconstructing the structures to atomistic models. It is worth noting that the reconstruction is easy because only hydrogen atoms are not considered in the atomistic structure-based model. Afterwards, the folding thermodynamics and kinetics were analyzed by applying a combination of the Markov state model (MSM)[76]and the transition path theory (TPT).[77]Gromacs-4.6.2[78]was used to conduct all the simulations,and PyEmma[79]was employed to build the MSM and to perform the TPT analysis. For simplicity, four G-repeats in the sequence were labeled as GGG1-GGG4in the 5′-3′direction(Fig.1(b)). Consequently,the substructure established by GGG1and GGG2was simplified to GGG12,and the other substructures were similarly named. The results show that the allatom simulations can complement the findings of the hybrid atomistic structure-based model and can provide more atomistic insights into the folding landscape.

    2. Theoretical method

    2.1. Hybrid atomistic structure-based model

    In this section, we introduce the hybrid atomistic structure-based model briefly. Details have been presented elsewhere.[59,60]All the heavy atoms of the system are retained in this model,and the solvent is implicitly modeled by the Langevin dynamics. Since modeling ions explicitly is still challenging in coarse-grained models,the current model does not consider the roles of ions. The Hamiltonian includes two parts:Vhybrid=Vff+Vsb,(1)

    whereVffis the potential of local interactions andVsbis the potential of long-range structure-based interactions.Vffincludes the bonded potential, the angular potential, the proper and improper dihedral potential and the 1-4 short-range potential. In this study,descriptions of local interactions were borrowed from the AMBER DNA force field OL15.[80-82]This force field was chosen because it merges all the newest modifications of DNA dihedral potential which guarantees us to acquire more accurate results. To describe the coexistence of two different folds of Htel24, which are characterized as the H1 and H2 GQ conformations,Vsbmerges the information of these two structures:

    ULJis a 12-6 Lennard-Jones attraction potential for the atom pairs forming native contacts(NC),andUexlis a generic hardcore repulsion potential for the nonnative (NN) atom pairs.It should be noted that the native contacts were defined as those atom pairs within a cutoff of 0.4 nm in the native structure. The native contacts shared by the H1 and H2 conformations were named“core”contacts,while those formed only in one of the two conformations were called specific H1 or H2 contacts. Here,εcore,εH1,εH2,εexl,r0i jandσare the user parameters. Determination of these parameters can be found in our previous study.[60]It is worth noting that various models have been developed to study multi-state biological processes.[83-88]We built the double-basin Hamiltonian by employing a mix-contact map strategy,[83-86,88]which has been validated to be successful in investigation of protein folding and allostery.In this framework,native contacts are classified into different libraries depending on whether they form in both reference structures(core native contacts)or merely form in one reference structure (“specific” native contacts). These information are then integrated into the long-range interaction term with different potential strengths. This is different from the other widely used double-basin model developed by Okazakiet al.[87]In that model, two structure-based singlebasin potentials are built asV(R|R1)andV(R|R2)+ΔV.Ris the coordinates of protein structure andR1(R2)represents the coordinates of the reference structure. ΔVis a parameter modulating the relative stability of the reference structures. Then,a smoothed double-based potentialVMBis obtained by solving a secular equation.

    To build the double-basin potential, the reference structure of H1 conformation was directly taken from the Protein Data Bank(PDB code:2GKU),while the structure of H2 conformation was modified from a similar topology formed by sequence Htel25 (5′-TAGGG[TTAGGG]3TT-3′, PDB code:2JSL). Using the Discovery Studio, the 3′-thymine in Htel25 was truncated, and the adjacent thymine was mutated to an adenine nucleotide. Likewise,the 5′-adenine was mutated to a thymine nucleotide. Lastly,the modified structure was equilibrated under 300 K with a 100 ns NPT all-atom MD simulation.

    Using the above model, multiple unbiased simulations around the folding temperature of H1 conformation (330 K)were performed starting from an unfolded structure. The staring structure was randomly selected from an ensemble of unfolded structures obtained from a 300 ns conventional all-atom simulation under 600 K.The time step was chosen as 2 fs,and the inverse friction constant in Langevin dynamics was 1 ps.Each trajectory was accumulated to 1010steps.

    2.2. The Markov state model (MSM) and the transition path theory(TPT)

    MSM is powerful to identify metastable states during the conformational changes of biomolecules. In an MSM,by choosing reasonable features to characterize the conformations and reducing the dimension of conformational space with dimension reduction methods, the space is first transformed into a set of discrete microstates. Then, after selecting a lag time under which jumps between these microstates can be regarded as a Markov chain, a few macrostates or metastable states are identified with the usage of Perron cluster analysis(PCCA+)[89]or other methods. In addition,the transition time between two metastable states is acquired by estimating the mean first passage time(mfpt).

    On the basis of metastable states obtained from the MSM,folding pathways can be identified using a TPT analysis. By defining the reactant state and the product state,the probability flux tween microstatesiandj,fi j,is defined as

    2.3. Seeding conventional all-atom simulations

    In the hybrid atomistic structure-based model, only hydrogen atoms are ignored. Therefore,conformations obtained from structure-based simulations are easily converted to allatom structures. We borrowed a procedure in the construction of an MSM to obtain a set of broadly distributed seeds. When building the MSM, three metrics were used as input features to characterize the conformational changes,including the rootmean-square deviation(RMSD)with respect to the H1 conformation, the RMSD with respect to the H2 conformation and the twelveχdihedral of each guanine nucleotide. In the estimation of RMSD, atoms P, C4′and N1 were used for purine bases,and the atoms P,C4′and N3 were used for pyrimidine bases. Using the time-lagged independent component analysis (TICA)[90]with a lag time of 1000 simulation steps, the space was reduced by retaining 90% of the total kinetic variance. After that,500 microstates were obtained by clustering the conformations using thek-means algorithm. From each of the 500 microstates, ten conformations were randomly selected as starting conformations to conduct conventional allatom simulations.

    The DNA force field AMBER OL15 was used in the allatom simulations. Each starting conformation was solvated within a periodic cubic box,and the water molecules were explicitly modeled by the TIP3P model. K+and Cl-ions were added to neutralize the system and to maintain a salt concentration of 100 mM. The electrostatic interaction was treated using the particle mesh Ewald(PME)method with a cutoff of 1.0 nm. The cutoff of the van der Waals interactions was also selected as 1.0 nm. All bonds were constrained using the linear constrain solver algorithm,and the MD time step was set to 2 fs. The Berendsen algorithm was used for both temperature and pressure coupling. The whole system was first subjected to a minimization of 1000 steps, followed by an equilibrium run with an NPT ensemble at 1 atm and 330 K for 2 ns. The last conformation was used for an additional 20 ns NPT simulation at 330 K.Overall,5000 trajectories with a total simulation time of 100 μs were collected to construct the MSM.

    3. Results

    3.1. Efficiency of the hybrid atomistic structure-based simulations

    Examination of structure-based simulations shows that the system hopped among three ensembles, which were determined to be the unfolded state, the H1 state and the H2 state (Fig. 2). This is an indication that the hybrid atomistic structure-based model enabled us to sample a broad range of the folding landscape of Htel24. Consequently, conventional all-atom simulations seeded from these structure-based simulations could also cover a large enough conformational space to investigate the folding mechanism and could be used to estimate the real time scale of the folding dynamics.

    Fig.2.A typical trajectory of the hybrid atomistic structure-based model with a double-basin potential. The RMSDs of all the heavy atoms with respect to the H1 and H2 conformations are colored red and green,respectively.

    3.2. MSM

    From the conventional all-atom simulations, we built a high-resolution MSM.The RMSD with respect to the H1 conformation,the RMSD with respect to the H2 conformation and the twelveχdihedrals were used as input features. The definition of RMSD can be found in the section of theoretical methods. Using the TICA method with a lag time of 2 ps,the conformational space was projected onto an 8-dimension space by retaining 90%of the total kinetic variance.After that,the conformations were clustered into 300 microstates with the usage ofk-means clustering algorithm. Because the implied time scales reached plateaus after about 3 ns (Fig. 3(a)), this lag time was used to build the MSM.Moreover,since there exists a gap between the zero-based seventh and eighth implied time scales(Fig.3(b)),300 microstates were lumped into nine macrostates using the hidden MSM method,[91]which were labelled as states 1-9.

    Fig.3. Construction of the MSM.(a)Implied time-scales as a function of lag time. (b)The sorted eigenvalue spectrum estimated from a lag time of 3 ns. The predicted folding time,which is indicated by the slowest time-scale,is approximately 6.53 ms.

    Fig.4. MSM and TPT analysis. (a)A randomly selected representative structure for each metastable state. (b)Distributions of the glycosidic configurations for states 5,6 and 7. Indicated by red arrows are the guanine nucleotides adopting nonnative configurations. The syn and anti glycosidic configurations are colored by red and blue as in Fig.1(b). (c)The free energy profile projected onto the 1st and 2nd slowest TICA components. Also shown are the centers of the nine metastable states. (d) Transition rates from the unfolded state to the H1 and H2 GQ topologies estimated from the TPT analysis.

    Figure 4(a) and Figs. S2 and S3 in the Supplementary Material demonstrate the representative structure of each metastable state. State 1 is shown to be highly heterogeneous in terms of structure (Fig. S2). Since it grouped various random coils together,this state was denoted as the unfolded state.States 2 and 3 are partially folded as implied by the structures and can be named as the hairpin state and triplex state, respectively. In state 2 the two terminal G-repeats GGG1and GGG4built a stable hairpin GGG14; in state 3, the G-repeat GGG3docked upon the hairpin and formed a triplex structure GGG134. All the remaining states adopted well-formed fourstranded topologies. It is clear that state 4 is a two-quartet GQ state, which has been suggested to be an important intermediate participating in the formation of three-quartet DNA GQs. Structures of state 5 through state 8 are equivalent to the topology of H1 conformation. However, deep analysis shows that state 5 through state 7 can be looked at as misfolded states(misfolded-1 to misfolded-3),since their distributions of syn/anti glycosidic configurations were incorrect in comparison with the native H1 structure. As shown in Fig. 4(b) and Fig.S4, guanine nucleotide G10 in state 5 took a syn configuration,which should be anti in the native state. Similarly,in states 6 and 7,G11 and G23 adopted the nonnative syn configurations,respectively(Fig.4(b),Figs.S5 and S6). In comparison with the three misfolded states, all guanine nucleotides adopted the native configurations in state 8 (Fig. S7). Therefore, state 8 is characterized as the native state of H1. The competition of the misfolded H1 states and native H1 state is also clearly shown in the free energy profile (Fig. 4(c)). For the last state, i.e., state 9, the representative structure resembles the topology of H2 conformation,and the distribution of glycosidic configurations is demonstrated to be the same as that of the native state of H2(Fig.S8). As a result, state 9 is denoted as the native H2 state.

    After identifying the metastable states,the real time-scale transition rate from the unfolded state to the H1 state or the H2 state was obtained by estimating the mftp. In good agreement with the experimental demonstration,the H2 conformation was shown to be kinetically more favorable. It is interesting to note that the ratio ofk1andk2(k1/k2)estimated in this study is on the same magnitude of order with that calculated by the previous simulations (0.20 vs 0.25).[60]Therefore, the all-atom simulations reproduced the folding dynamics discovered by the hybrid atomistic structure-based model.

    3.3. Formation pathways of H1 and H2 conformations

    The TPT analysis identified multiple formation pathways for the H1 state and one formation pathway for the H2 state(Fig.5). Four pathways,which have probabilities greater than 10%,are discussed for the formation of H1 state. In the dominant pathway(pathway-1),the DNA folded into the H1 state by successively traversing through the ensembles of hairpin GGG14,triplex GGG134and two-quartet GQ state without visiting the misfolded states.

    Fig. 5. Formation pathways of the H1 and H2 states. The four major pathways in which H1 forms are colored by violet,red,blue and brown,respectively. Also shown are the probabilities of the pathways. Indicated by arrows are the guanine nucleotides adopting nonnative configurations. The syn and anti glycosidic configurations are colored by red and blue as in Fig.1(b). The gradient color indicates a fluctuating configuration between syn and anti.

    In contrast, in each of the other three competing pathways, the system was observed to be trapped in a misfolded state before forming the native structure. For example, in pathway 2, the formation of the H1 state proceeded through the hairpin ensemble GGG14into the triplex state GGG134followed by the formation of the misfolded-2 state in which G11 adopted an incorrect configuration. At the end, the process continued through the nonnative glycosidic configuration reoriented to find the correct pattern. Pathways 3 and 4 are similar to pathway 2 except that different misfolded states were visited. In comparison with the formation of H1 state,the H2 state is demonstrated to form with a higher cooperativity. As shown in the formation pathway of H2 state,the DNA directly folded to the native H2 conformation after going through the GGG14hairpin ensemble.This indicates that the substructures GGG12,GGG23and GGG34can form cooperatively during the formation of H2 state, which has also been observed in the structure-based simulations. It is thus another evidence that the all-atom simulations recovered the folding dynamics revealed by the hybrid atomistic structure-based simulations.

    4. Discussion and conclusion

    The combined power of the conventional all-atom model,MSM and TPT allows us to complement the folding mechanism of Htel24 revealed by the double-basin hybrid atomistic structure-based model. The relevance of current results to the findings of previous studies is discussed in the following.

    The set of all-atom simulations conducted in this study provided an estimation of the real time-scale for the folding dynamics revealed by the hybrid atomistic structure-based model. As shown in Fig.3(b),the folding time of Htel24 under 330 K was estimated to be 6.53 ms. In the absence of direct experimental measurement for the folding time under this temperature, concrete conclusions cannot be made regarding this issue. However, on the one hand, due to the structurebased model favors the H1 and H2 topologies,some potential kinetic traps (such as alternative GQ folds) were eliminated during the simulations. Consequently, the folding time scale estimated here may represent a lower limit and gives a hint of how fast Htel24 may fold if potential competing folds can be avoided by a careful design of experiments. On the other hand,experiments showed that the folding time of Htel24 decreases as the temperature increasing. Since the experimental folding time under 316 K was demonstrated to be on the order of seconds,[45]the time-scale predicted here may be reasonable. This argument can be further verified by an experiment that measures the folding rate.

    In comparison with the results of hybrid atomistic structure-based simulations,[60]the all-atom simulations identified more metastable states, including a hairpin state, a triplex state and two misfolded states (misfolded-2 and misfolded-3). Of interest is the triplex state, which was not stable enough to be characterized as a folding intermediate previously. During the structure-based simulations, the folding temperature was chosen with the aim to sample sufficient folding and unfolding events.However,native contacts,which are the most important interactions stabilizing the structures in the structure-based model, could be broken easily under this temperature. As a result,the lifetime of a triplex structure was largely shortened. In contrast, in the conventional all-atom simulations,more factors increasing the stability of the structures can be considered,such as the nonnative interactions and the metal ions. With the formation of these additional important interactions,the stability and lifetime of a triplex state can be significantly increased. The participation of triplex state in the formation of Htel GQ is consistent with several previous studies.[19,21,39,92]However, whether the G-triplex structure can be characterized as a stable folding intermediate of GQs still remains controversial. While many experimental studies support the involvement of G-triplex structures as intermediates in the folding of GQs,computational studies have demonstrated that such structures are unstable and thus cannot be characterized as intermediates. For example,Liet al. detected the existence of a G-triplex state by investigating the folding kinetics of Htel GQ using a magnetic tweezer.[19]However,by performing large-scale atomistic MD simulations, Stadlbaueret al.reported that the G-triplex structures have lifetimes below 1 μs,which is an indication that these structures are thermodynamically unstable.[49]Therefore,the structural stability of a G-triplex structure requires further investigation to reconcile the contradictions between previous studies.

    The all-atom simulations also provided another possible explanation for the kinetic preference of H2 conformation. In the structure-based simulations, the H2 conformation was revealed to fold by two steps, while the H1 conformation was suggested to fold by three steps. The higher cooperative folding manner of H2 conformation was believed to be a reason of its kinetic preference. Here, the all-atom simulations also showed the similar high cooperative formation of H2. Moreover, it was revealed that misfolded states adopting inappropriate combinations of syn/anti guanine nucleotides prefer to participate in the formation of H1 conformation. This preference may also be a potential factor inducing the slower kinetic accessibility of H1,since such misfolded structures have been suggested to profoundly influence the folding processes of DNA GQs.[22,49]However, why guanine nucleotides prefer to adopt nonnative syn/anti configurations during the formation of H1 state remains an interesting issue required further investigation. It is the major reason leading to the multipathway formation mechanism of H1 state and one-pathway formation mechanism of H2 state. In addition, two-quartet GQ structures were suggested to merely participate in the formation of H1 state, which is another possible illustration for the different formation mechanisms of H1 and H2 states. In future, we will systematically explore different preference of misfolded states and two-G-quartet GQs during the formation of H1 and H2 states by performing large-scale molecular simulations.

    It is worth noting that the role of metal ions was not discussed. The influence of metal ions during the folding of biomolecules is an important issue.[94]Especially, for Htel DNA GQ,monovalent metal ions(typically K+and Na+)can determine the final type of folded structure.[2]However, the hybrid atomistic structure-based model is essentially a highresolution coarse-grained model,in which explicitly modeling of ions still remains a challenge. To our knowledge, there is not any coarse-grained model that can address the question of nucleic acids folding with ions explicitly modeling at this stage. Therefore,we did not consider the influence of ions in the structure-based model and validated our results by comparing with known experimental observations. As discussed above,the structure-based model enabled us to reproduce the experimental observed two-native-state folding thermodynamics and kinetics of Htel24. This is a validation that simulations based on the structure-based model have captured most of the folding properties. It should be noted that, although we performed conventional all-atom simulations with ions explicitly modeled in this study to complement the results of structurebased model, the simulation time is very limit which is only 20 ns for each trajectory. Particularly,the metal ions may not enter to the center of the adjacent G-quartets in such a short time scale. Consequently, using present all-atom simulations to describe the influence of ions is inaccuracy.To alleviate this problem, we are performing additional long time-scale simulations to explore the important role of monovalent ions and the results will be discussed in our future work.

    In conclusion, performing large-scale conventional allatom simulations,we have complemented the results of the hybrid atomistic structure-based model for the two-native-state kinetic partition folding of Htel24. The real folding time-scale has been estimated and more details of the folding landscape are acquired. Moreover,the all-atom simulations also provide a possible illustration for the experimentally observed different kinetic accessibility of H1 and H2 conformations. These findings suggest that the multiscale computational framework coupling the multiple-basin hybrid atomistic structure-based model and the conventional all-atom model is a potential useful method to study the complex kinetic partitioning folding process of Htel GQ and the other DNA GQs.

    男女下面插进去视频免费观看| 免费在线观看日本一区| 又黄又粗又硬又大视频| 午夜精品国产一区二区电影| 亚洲第一av免费看| av有码第一页| 国产精品偷伦视频观看了| 少妇人妻久久综合中文| 人人妻人人添人人爽欧美一区卜| 99热全是精品| 丝袜喷水一区| 美女高潮到喷水免费观看| 精品福利永久在线观看| 99久久精品国产亚洲精品| 国产精品 欧美亚洲| 91精品国产国语对白视频| 欧美精品亚洲一区二区| 男人舔女人的私密视频| 捣出白浆h1v1| 国产成人影院久久av| 免费不卡黄色视频| 国精品久久久久久国模美| 一区二区三区乱码不卡18| 建设人人有责人人尽责人人享有的| 亚洲少妇的诱惑av| 国产精品.久久久| 男女免费视频国产| 七月丁香在线播放| 热re99久久国产66热| 国产视频一区二区在线看| 精品亚洲成国产av| 国产高清不卡午夜福利| 男人操女人黄网站| 欧美久久黑人一区二区| 久久国产精品影院| 久热这里只有精品99| 99精国产麻豆久久婷婷| 成人黄色视频免费在线看| 国产男人的电影天堂91| 中文乱码字字幕精品一区二区三区| 国产高清视频在线播放一区 | 亚洲少妇的诱惑av| 亚洲成色77777| 丝袜人妻中文字幕| 久久久久精品国产欧美久久久 | 亚洲欧美中文字幕日韩二区| 无遮挡黄片免费观看| 中文字幕人妻丝袜一区二区| 免费人妻精品一区二区三区视频| 亚洲精品国产一区二区精华液| 真人做人爱边吃奶动态| 女人被躁到高潮嗷嗷叫费观| 在线精品无人区一区二区三| 亚洲av国产av综合av卡| 国产精品三级大全| 下体分泌物呈黄色| 日韩 欧美 亚洲 中文字幕| 看十八女毛片水多多多| 亚洲欧美成人综合另类久久久| 国产视频首页在线观看| 国产精品一区二区精品视频观看| 欧美在线黄色| 久久人妻熟女aⅴ| 日本色播在线视频| 亚洲欧美日韩高清在线视频 | 色视频在线一区二区三区| 老司机亚洲免费影院| 亚洲国产成人一精品久久久| 亚洲国产日韩一区二区| 国产熟女午夜一区二区三区| 一级毛片女人18水好多 | 青草久久国产| av天堂久久9| 亚洲精品国产av成人精品| 啦啦啦啦在线视频资源| 欧美老熟妇乱子伦牲交| 亚洲一区二区三区欧美精品| 成年女人毛片免费观看观看9 | 久热这里只有精品99| 中文字幕人妻丝袜制服| 在线av久久热| 女人爽到高潮嗷嗷叫在线视频| 成人国产一区最新在线观看 | 久久精品国产综合久久久| 亚洲精品国产区一区二| 在线天堂中文资源库| 亚洲美女黄色视频免费看| 伦理电影免费视频| 亚洲熟女毛片儿| av欧美777| 中文字幕另类日韩欧美亚洲嫩草| 菩萨蛮人人尽说江南好唐韦庄| 亚洲,一卡二卡三卡| 国语对白做爰xxxⅹ性视频网站| 欧美 日韩 精品 国产| 尾随美女入室| 国产一区二区三区av在线| 欧美少妇被猛烈插入视频| 欧美 亚洲 国产 日韩一| 精品亚洲成国产av| 国产福利在线免费观看视频| 久久久精品94久久精品| 久久久久久免费高清国产稀缺| 久久毛片免费看一区二区三区| 欧美xxⅹ黑人| 爱豆传媒免费全集在线观看| 亚洲精品日本国产第一区| 久久久国产欧美日韩av| 久久99一区二区三区| 可以免费在线观看a视频的电影网站| 国产成人a∨麻豆精品| 美女视频免费永久观看网站| 国产欧美亚洲国产| 日韩一卡2卡3卡4卡2021年| 一区二区三区四区激情视频| 成年av动漫网址| 日韩,欧美,国产一区二区三区| 亚洲人成电影观看| 久久人人爽av亚洲精品天堂| 久久人人97超碰香蕉20202| 国产在线视频一区二区| 国产一区二区在线观看av| 亚洲,一卡二卡三卡| 悠悠久久av| 欧美黑人欧美精品刺激| 美女主播在线视频| 丰满少妇做爰视频| 免费少妇av软件| 色94色欧美一区二区| 日本wwww免费看| 久久国产亚洲av麻豆专区| 飞空精品影院首页| 日韩 欧美 亚洲 中文字幕| 国产免费现黄频在线看| 成年人黄色毛片网站| 少妇的丰满在线观看| av电影中文网址| 97在线人人人人妻| 日本a在线网址| 一本色道久久久久久精品综合| 丰满少妇做爰视频| 国产淫语在线视频| 亚洲精品久久久久久婷婷小说| 久久午夜综合久久蜜桃| 久久人人97超碰香蕉20202| 久久久国产精品麻豆| 亚洲av男天堂| 精品少妇一区二区三区视频日本电影| 国产三级黄色录像| 成人三级做爰电影| 欧美亚洲日本最大视频资源| 免费女性裸体啪啪无遮挡网站| 亚洲精品国产一区二区精华液| 亚洲五月色婷婷综合| 欧美变态另类bdsm刘玥| 别揉我奶头~嗯~啊~动态视频 | 久久久久久久大尺度免费视频| 90打野战视频偷拍视频| svipshipincom国产片| 国产无遮挡羞羞视频在线观看| 九草在线视频观看| 日韩中文字幕视频在线看片| 久久久久久免费高清国产稀缺| av有码第一页| 国产精品免费大片| 午夜视频精品福利| 天天躁夜夜躁狠狠久久av| 制服人妻中文乱码| av在线app专区| 久久人妻福利社区极品人妻图片 | 国产欧美日韩精品亚洲av| 99热国产这里只有精品6| 久久精品国产综合久久久| 国产免费又黄又爽又色| 波多野结衣一区麻豆| 国产免费视频播放在线视频| 美女中出高潮动态图| 一级,二级,三级黄色视频| 9色porny在线观看| 亚洲伊人色综图| 少妇裸体淫交视频免费看高清 | svipshipincom国产片| 欧美日韩黄片免| 国产亚洲欧美在线一区二区| 欧美国产精品一级二级三级| 成年av动漫网址| 少妇裸体淫交视频免费看高清 | 亚洲精品久久午夜乱码| 免费观看a级毛片全部| 欧美日韩亚洲国产一区二区在线观看 | 无遮挡黄片免费观看| 精品福利永久在线观看| 国产亚洲午夜精品一区二区久久| 亚洲精品一卡2卡三卡4卡5卡 | 99久久人妻综合| 欧美日韩综合久久久久久| 日本猛色少妇xxxxx猛交久久| 久久久久久免费高清国产稀缺| 看十八女毛片水多多多| 成人18禁高潮啪啪吃奶动态图| 三上悠亚av全集在线观看| 99国产综合亚洲精品| 中文精品一卡2卡3卡4更新| 一边摸一边做爽爽视频免费| 国产有黄有色有爽视频| 亚洲欧美日韩高清在线视频 | 免费在线观看影片大全网站 | 久久精品久久精品一区二区三区| 美女中出高潮动态图| 欧美精品一区二区免费开放| 在现免费观看毛片| 美女高潮到喷水免费观看| 国产精品人妻久久久影院| 一级黄色大片毛片| 色94色欧美一区二区| 男女边吃奶边做爰视频| 亚洲图色成人| 又大又爽又粗| 丝袜美足系列| 日韩欧美一区视频在线观看| 久久这里只有精品19| 一级黄色大片毛片| 久久精品人人爽人人爽视色| 青春草亚洲视频在线观看| 国产深夜福利视频在线观看| cao死你这个sao货| av国产久精品久网站免费入址| 91麻豆精品激情在线观看国产 | 美女福利国产在线| 午夜免费男女啪啪视频观看| 视频区图区小说| 五月天丁香电影| 尾随美女入室| 一边亲一边摸免费视频| 精品人妻1区二区| 久久精品成人免费网站| 一本久久精品| 欧美人与善性xxx| 日本vs欧美在线观看视频| 黄色视频在线播放观看不卡| 人体艺术视频欧美日本| 国产成人av激情在线播放| 一个人免费看片子| 在线观看免费午夜福利视频| 日韩一区二区三区影片| 免费女性裸体啪啪无遮挡网站| 国产精品一区二区免费欧美 | 丝袜美腿诱惑在线| 精品一区在线观看国产| 亚洲伊人色综图| 汤姆久久久久久久影院中文字幕| 亚洲国产最新在线播放| 久久亚洲国产成人精品v| 久久久久久亚洲精品国产蜜桃av| 午夜两性在线视频| 交换朋友夫妻互换小说| 亚洲中文日韩欧美视频| 蜜桃在线观看..| 日韩制服骚丝袜av| 黄网站色视频无遮挡免费观看| 夫妻性生交免费视频一级片| 久久午夜综合久久蜜桃| 国产精品麻豆人妻色哟哟久久| 欧美亚洲 丝袜 人妻 在线| 99九九在线精品视频| 亚洲第一青青草原| 久久久国产一区二区| 999久久久国产精品视频| 天天操日日干夜夜撸| 真人做人爱边吃奶动态| 国产在线视频一区二区| av天堂久久9| 少妇精品久久久久久久| 18禁黄网站禁片午夜丰满| 97精品久久久久久久久久精品| 人人妻人人澡人人爽人人夜夜| 多毛熟女@视频| 久久人人97超碰香蕉20202| 亚洲精品自拍成人| 国产在视频线精品| 波野结衣二区三区在线| 高潮久久久久久久久久久不卡| 欧美精品高潮呻吟av久久| 欧美日韩视频精品一区| 水蜜桃什么品种好| 免费女性裸体啪啪无遮挡网站| 免费在线观看日本一区| 国产伦理片在线播放av一区| 夫妻午夜视频| 国产免费又黄又爽又色| 99久久精品国产亚洲精品| 欧美日韩黄片免| 精品福利永久在线观看| 99国产精品一区二区三区| 岛国毛片在线播放| 亚洲情色 制服丝袜| 亚洲欧美色中文字幕在线| 国产极品粉嫩免费观看在线| 999精品在线视频| 老鸭窝网址在线观看| 一级毛片我不卡| 中文字幕人妻丝袜一区二区| 久久久久久久久免费视频了| 国产淫语在线视频| 国产一区亚洲一区在线观看| 国产亚洲精品第一综合不卡| 亚洲国产av新网站| 女人被躁到高潮嗷嗷叫费观| 欧美成狂野欧美在线观看| 老司机在亚洲福利影院| 国产福利在线免费观看视频| 国产精品久久久久成人av| 美女高潮到喷水免费观看| 久久影院123| 国产又爽黄色视频| 首页视频小说图片口味搜索 | 亚洲伊人久久精品综合| 少妇粗大呻吟视频| 亚洲天堂av无毛| 亚洲精品一卡2卡三卡4卡5卡 | 国产精品国产三级专区第一集| 欧美人与性动交α欧美软件| 国产高清不卡午夜福利| 亚洲成人手机| 久久精品国产亚洲av涩爱| 国产精品三级大全| 欧美97在线视频| 又大又黄又爽视频免费| 另类精品久久| 亚洲精品自拍成人| a 毛片基地| 国产高清不卡午夜福利| 美女福利国产在线| 在线天堂中文资源库| 久久午夜综合久久蜜桃| 1024香蕉在线观看| 亚洲精品国产色婷婷电影| 天天影视国产精品| 精品国产乱码久久久久久男人| 免费在线观看日本一区| 国产在线免费精品| 亚洲,一卡二卡三卡| 午夜福利,免费看| 欧美另类一区| 精品一区二区三卡| 自拍欧美九色日韩亚洲蝌蚪91| 亚洲少妇的诱惑av| 操美女的视频在线观看| 日本五十路高清| 日韩伦理黄色片| 另类精品久久| 欧美成狂野欧美在线观看| 亚洲五月色婷婷综合| 天天操日日干夜夜撸| 五月天丁香电影| 免费观看人在逋| 久久精品久久精品一区二区三区| 精品少妇黑人巨大在线播放| xxxhd国产人妻xxx| 国产三级黄色录像| 91精品伊人久久大香线蕉| 国产一区二区三区综合在线观看| 脱女人内裤的视频| 亚洲中文av在线| 亚洲精品久久午夜乱码| 日本黄色日本黄色录像| 日日夜夜操网爽| 亚洲成av片中文字幕在线观看| av一本久久久久| 天天躁狠狠躁夜夜躁狠狠躁| 十分钟在线观看高清视频www| 国产深夜福利视频在线观看| 精品久久久久久电影网| tube8黄色片| 精品国产一区二区久久| 亚洲图色成人| 久久久精品国产亚洲av高清涩受| 少妇粗大呻吟视频| 国产极品粉嫩免费观看在线| 亚洲国产精品国产精品| 中文字幕色久视频| av欧美777| 亚洲欧洲日产国产| 赤兔流量卡办理| 国产片内射在线| 久热这里只有精品99| 黑人猛操日本美女一级片| 国产午夜精品一二区理论片| 亚洲情色 制服丝袜| 国产精品二区激情视频| 久久久久久久精品精品| 国产免费现黄频在线看| 亚洲欧美清纯卡通| 国产免费福利视频在线观看| 超碰成人久久| 欧美精品人与动牲交sv欧美| 国产精品国产三级国产专区5o| 亚洲五月婷婷丁香| 国产成人欧美| 精品一区二区三区四区五区乱码 | a级毛片在线看网站| 欧美日韩av久久| 满18在线观看网站| 女性被躁到高潮视频| av网站免费在线观看视频| 夜夜骑夜夜射夜夜干| 大话2 男鬼变身卡| 欧美性长视频在线观看| 国产精品久久久久久人妻精品电影 | 在线观看免费午夜福利视频| 亚洲专区中文字幕在线| 免费不卡黄色视频| 男人添女人高潮全过程视频| 国产亚洲午夜精品一区二区久久| 免费看不卡的av| 欧美大码av| 国产午夜精品一二区理论片| 制服人妻中文乱码| 精品人妻熟女毛片av久久网站| 高清不卡的av网站| 久久久国产精品麻豆| 午夜日韩欧美国产| 99精国产麻豆久久婷婷| 亚洲国产精品成人久久小说| 亚洲av国产av综合av卡| 亚洲精品国产区一区二| 亚洲欧洲国产日韩| 在线观看人妻少妇| 日韩中文字幕视频在线看片| 日韩精品免费视频一区二区三区| 国产片内射在线| 熟女av电影| 久久热在线av| 国产欧美亚洲国产| 王馨瑶露胸无遮挡在线观看| 狠狠婷婷综合久久久久久88av| 一本综合久久免费| 免费女性裸体啪啪无遮挡网站| 老司机亚洲免费影院| 精品福利永久在线观看| 国产精品国产三级国产专区5o| 久久九九热精品免费| 中文字幕亚洲精品专区| 黄色片一级片一级黄色片| 嫩草影视91久久| 99久久精品国产亚洲精品| 女人高潮潮喷娇喘18禁视频| 色综合欧美亚洲国产小说| 久久亚洲精品不卡| 久久久国产一区二区| 丝瓜视频免费看黄片| 国产成人av教育| 久久精品国产a三级三级三级| bbb黄色大片| 99久久人妻综合| 女人精品久久久久毛片| 亚洲国产毛片av蜜桃av| 亚洲综合色网址| 精品一品国产午夜福利视频| 欧美国产精品一级二级三级| 久久亚洲国产成人精品v| 精品一区二区三区av网在线观看 | 黄网站色视频无遮挡免费观看| 久久性视频一级片| 中文字幕亚洲精品专区| 免费不卡黄色视频| 欧美在线黄色| 亚洲欧洲精品一区二区精品久久久| 久久毛片免费看一区二区三区| 亚洲av电影在线观看一区二区三区| 欧美人与善性xxx| 男男h啪啪无遮挡| 欧美日韩综合久久久久久| www日本在线高清视频| 母亲3免费完整高清在线观看| 亚洲中文字幕日韩| 视频区图区小说| 在线观看人妻少妇| 美女扒开内裤让男人捅视频| 国产午夜精品一二区理论片| 午夜免费鲁丝| 美女高潮到喷水免费观看| 久久精品国产综合久久久| 国产伦人伦偷精品视频| 亚洲欧美日韩另类电影网站| 成人黄色视频免费在线看| 成人国产一区最新在线观看 | 日韩欧美一区视频在线观看| 两个人看的免费小视频| 国产精品.久久久| 老司机深夜福利视频在线观看 | 免费久久久久久久精品成人欧美视频| 国产成人一区二区在线| 啦啦啦在线免费观看视频4| 岛国毛片在线播放| 婷婷色综合www| 精品福利永久在线观看| 性色av一级| 国产极品粉嫩免费观看在线| 午夜福利视频在线观看免费| 色综合欧美亚洲国产小说| 另类亚洲欧美激情| av在线播放精品| 国产高清不卡午夜福利| 曰老女人黄片| 两性夫妻黄色片| 嫩草影视91久久| 狂野欧美激情性xxxx| 国产av一区二区精品久久| 精品一区二区三区av网在线观看 | 18禁观看日本| 国产精品亚洲av一区麻豆| 美女中出高潮动态图| 久久影院123| 国产一区二区激情短视频 | 亚洲国产精品国产精品| 2018国产大陆天天弄谢| 欧美黑人欧美精品刺激| 亚洲 欧美一区二区三区| 亚洲av电影在线进入| 狂野欧美激情性bbbbbb| 一区二区av电影网| 亚洲精品久久成人aⅴ小说| 欧美精品av麻豆av| 国产在线视频一区二区| 午夜福利影视在线免费观看| 天天躁夜夜躁狠狠久久av| 婷婷色av中文字幕| 少妇人妻久久综合中文| 午夜福利乱码中文字幕| 91老司机精品| 人人妻人人爽人人添夜夜欢视频| 国产一区二区 视频在线| 久久鲁丝午夜福利片| 在线观看免费高清a一片| 国产熟女欧美一区二区| 少妇 在线观看| 国产精品一国产av| 999精品在线视频| 香蕉国产在线看| 国产成人欧美| 精品久久蜜臀av无| 日韩精品免费视频一区二区三区| 自拍欧美九色日韩亚洲蝌蚪91| h视频一区二区三区| 精品第一国产精品| 天天添夜夜摸| 老司机亚洲免费影院| 亚洲激情五月婷婷啪啪| 久久久久久久国产电影| 精品国产一区二区久久| 国产精品香港三级国产av潘金莲 | 久久久国产精品麻豆| www日本在线高清视频| 久久精品亚洲av国产电影网| 嫩草影视91久久| 熟女少妇亚洲综合色aaa.| 中文字幕人妻丝袜一区二区| av电影中文网址| 中国美女看黄片| 亚洲伊人色综图| 国产精品 欧美亚洲| 啦啦啦 在线观看视频| 欧美性长视频在线观看| 中文欧美无线码| av天堂在线播放| 欧美另类一区| 成人免费观看视频高清| 国产精品久久久久久精品电影小说| 制服诱惑二区| av在线老鸭窝| 精品第一国产精品| 人体艺术视频欧美日本| 老司机在亚洲福利影院| 男人操女人黄网站| 男人添女人高潮全过程视频| 高清不卡的av网站| 麻豆av在线久日| 欧美成人精品欧美一级黄| 一级毛片 在线播放| av视频免费观看在线观看| 国产1区2区3区精品| 国产成人精品久久二区二区91| 一区二区三区激情视频| 久久久久视频综合| 国产在线视频一区二区| 色婷婷久久久亚洲欧美| 亚洲天堂av无毛| 国产淫语在线视频| av在线播放精品| 电影成人av| 乱人伦中国视频| 久久久久久久久免费视频了| 一级毛片黄色毛片免费观看视频| 久久精品久久久久久久性| 久久久久久亚洲精品国产蜜桃av| 亚洲视频免费观看视频| 色播在线永久视频| 伊人亚洲综合成人网| 亚洲专区中文字幕在线| av一本久久久久| 欧美日韩亚洲综合一区二区三区_| 婷婷成人精品国产| 婷婷丁香在线五月| 啦啦啦 在线观看视频| 一边摸一边抽搐一进一出视频| 99精品久久久久人妻精品| 精品国产一区二区久久| a 毛片基地| 欧美日韩视频精品一区| 久久精品aⅴ一区二区三区四区| 亚洲成人手机| 一区二区三区精品91| 亚洲图色成人| 国产精品一区二区在线不卡|