100% found this document useful (1 vote)
119 views

Lecture 12' - Structural Transitions in Nucleic Acids II

This document provides an outline for a lecture on modeling the helix-coil transition in DNA. It discusses DNA structure, the DNA melting process, and experimental monitoring of DNA melting curves. It introduces an equilibrium model for predicting DNA melting curves based on statistical thermodynamics. The model treats DNA as an ensemble of conformations, assigns statistical weights to each conformation based on free energy, and uses partition functions to calculate values like the equilibrium constant KD and the fraction of stacked base pairs ΘB. The goal is to fit model melting curves to experimental data to understand DNA thermodynamic parameters.

Uploaded by

curlicue
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
119 views

Lecture 12' - Structural Transitions in Nucleic Acids II

This document provides an outline for a lecture on modeling the helix-coil transition in DNA. It discusses DNA structure, the DNA melting process, and experimental monitoring of DNA melting curves. It introduces an equilibrium model for predicting DNA melting curves based on statistical thermodynamics. The model treats DNA as an ensemble of conformations, assigns statistical weights to each conformation based on free energy, and uses partition functions to calculate values like the equilibrium constant KD and the fraction of stacked base pairs ΘB. The goal is to fit model melting curves to experimental data to understand DNA thermodynamic parameters.

Uploaded by

curlicue
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 39

Lecture 12 – Structural Transitions in

Nucleic Acids II
Outline
 Introduction
 The DNA Helix-Coil Transition
 Very quick review of basic DNA structure:
 Focus: base-pair stacking.
 DNA melting and melting curves:
 Thermal denaturation: breaking the stacks.
 Experimental monitoring of base-pair stacking %...

 Modeling DNA Melting:


 Idea: Generalize our Aligned Zipper model (Lecture 12).
 To treat concentration-dependence, shifted structures, loops…
 Focus is still: Equilibrium Treatment…
1. Weighting conformations (both stacks and loops);
2. Thermodynamic parameter sets;
3. Models of duplex formation;

 Comparisons with experiment.


Our Focus: the Helix-Coil
Transition in DNA
 In particular, we focus on two
related processes:
 dsDNA melting
 B helix to two coils.
 dsDNA annealing
 two coils to a B helix.
 Note: single dsDNA species.
 Understanding these:
 aids in modeling more complicated
transitions.
 e.g., many competing species.
 Ultimate focus: complex annealing.
Single-Stranded DNA (the ‘Coil’)
 An unbranched, polynucleotide chain:
 monomers units = nucleotides.
 each contains three components:
a negatively charged phosphate (PO4-);
a 2’-deoxyribose sugar;
one of 4 heterocyclic bases (A,T,G,C);
 pairs linked by phosphodiester bonds.
 Nitrogenous bases are of two types:
 Purines (2 rings): Adenine (A), Guanine (G).
 Pyrimidines (1 ring): Thymine (T), Cytosine (C).
 ssDNA has a 5’ to 3’ polarity.
 5’ and 3’ ends are chemically distinct.
 by convention, DNA sequence is written 5’ to 3’.
The B Helix
 Natural dsDNAs in solution…
 adopt a double-stranded, helical structure.
 strand orientations are antiparallel.
 under physiological conditions: B helix.
 Helix characterized by Watson-Crick
base pairing:
 A pairs with T (2 H+-bonds).
 G pairs with C (3 H+-bonds).
 At right, we show the dsDNA formed
by annealing of:
 5’-CTAGTCGTGGTTC-3’
 5’-GAACCACGACTAG-3’
Stabilizing Interactions

 DNA B-Helix structure stabilized by:


 Hydrogen bonding between bases (minor);
 Stacking between H+-bonded base-pairs (primary):
1. Van der Waals interaction:
 Optimal stacking of adjacent rings;
2. Induced dipole-induced dipole interaction:
 Favorable alignment of the moments of adjacent rings.
 Results in the characteristic helix shape.
3. Hydrophobic interaction:
 Favorable sequestering of hydrophobic rings.
 In DNA melting…the helix destabilized
 generally implemented by increasing temperature, T.
 destabilizes the stacks…unwinding the helix.
 unwound helix separates into free ssDNAs (‘coils’).
Monitoring the Helix-Coil
Transition
 Degree of stacking is experimentally observable:
 Let ΘB = mean fraction of stacked base pairs.
 Ultraviolet absorbance at 260 nm (A260)…
is inversely proportional to ΘB;
Also called the ‘hypochromicity’.
 DNA melting accompanied by  40% increase in A260.
 Plot of A260 vs. T yields ΘB vs. T
 The DNA ‘melting curve’.
DNA Melting Curves
 ΘB decreases monotonically from 1 to 0 (for fully matched strands).
 sigmoidal shape indicates DNA melting is cooperative.
 One sigmoid = cooperative melting of entire DNA;
 The DNA melting temperature (Tm):
 For fully matching strands: temp. at which ΘB = ½;
 Width (∆T) is non-zero (e.g., for 10-mers, ∆T  10 oC).

 Melting curves of longer DNAs show more structure:


 several independently melting regions (AT’s less stable).
Why focus on DNA Melting?
 Fitting of model curves with experiment:
 facilitates investigation of DNA thermodynamic parameters.
 describe the fundamental properties of DNA interaction.
 Helps to understand more general DNA mixtures.

 Modeling provides a demonstration of general techniques:


 Equilibrium chemistry;
 Statistical Thermodynamic weighting;

 Although parameter values vary by polymer, transition,


 general principles apply to modeling other biopolymers:
 protein folding and structural transitions.
 RNA folding, etc.

 We will use an Equilibrium approach…


 based on Statistical Thermodynamics.
The Aligned-Zipper Model
 In L. 11, we adopted a simplified model…
 Three non-zipper assumptions:
 Homoduplex-melting:1 kind of base-pair.
 Strands perfectly-aligned: no shifting.
 Strand-separation neglected: no [strand] effects.

 This allowed an aligned Zipper model:


 Annealing: forward transition (coil to helix).
 Melting: reverse transition (helix to coil).
 from a fully-helical state, H = …hhhhhh….
 to a fully-melted state, C = …cccccc….

 Model Application proceeded by:


1. Defining model parameters:
 The nucleation parameter, σ.
 The propagation parameter, s.
 Applying our Zipper-expression for <Ph>:
 Result: DNA Melting Curve.

 However, our model a bit too simple!


Need for a Better Model
 Most dsDNAs of interest are not homo-polymers:
 Generally contain all 4 bases (A, T, G, C).
 At least 10 propagation parameters, si required.

 Strand-separation is also significant:


 Results in a dependence on strand-concentration.
 Particularly for oligonucleotides.

 Annealing, Melting also much more complicated:


 Shifted alignments, looped structures can be significant.
Melting Curve Prediction:
 We adopt an equilibrium model.
 Our simple equilibrium of interest:
 Quantities of Interest:
 Fraction of stacked base pairs (bps): ΘB = ΘextΘint;
 fraction of associated strands: Θext = 2CAB/Ctot;
 mean fraction of stacked bps per dsDNA conformation: Θint ;

 Let’s begin with an estimation of Θext


 First, we need some simple equations:
 Mass Action: KD = CACB /CAB = 1/Kassoc.
 ssDNA Strand Conservation:
 CAo = CA + CAB and CBo= CB + CAB
Melting Curve Prediction (cont):

 Continuing our estimation of Θext


 Analysis is system-dependent:
 Mass Action: KD = CACB /CAB = 1/Kassoc.
 ssDNA Strand Conservation:
 CAo = CA + CAB and CBo= CB + CAB

 Idea: Combine to yield a quadratic eq. ( solve for Θext…)


 Usual is an ‘Equivalent co-polymer’ treatment: assume A = B;
 Good for long polymers; not so good for oligonucleotides.

 Result: Θext = [(Ctot/KD+1) - (1+2Ctot/KD)1/2] / (Ctot/KD)


Statistical Thermodynamics
 Computing ΘB still requires estimates of KD and Θint.
 Tool…Statistical Thermodynamics.
 assumption: system always (nearly) at equilibrium.
 note a limitation…no rate information.
 Consider an Ensemble of Systems:
 large number of instances of our system…O(1023).
each prepared identically.
 members distributed over all accessible conformations:
single-stranded states (unstacked ssDNAs, hairpins).
distinct double-stranded states.
 Stat-thermo addresses:
 equilibrium probabilities of state occupancy.
 changes in system variables which accompany equilibrium
transitions.
Ising Model of Stacking
 Assumption: stack-formation is all-or-none.
 each base has either single-stranded or stacked ‘character’.
 big simplification…

 Each dsDNA conformation is then specified by:


 alignment between the interaction strands.
 stacking pattern.
 no worrying about “partial” stacks.

 Conformation specified by location of helical and ss regions.


The Gibbs Factor
 So, how do we estimate relative occupancies?

 As before, each conformation, i gets a statistical weight, ωi.


 related to its standard free energy of formation, ∆Gio:
 ωi = exp[-∆Gio/RT]; R = molar gas constant.
 the ‘Gibbs Factor’.

 Relative probability of observing states i and j:


 estimated by the ratio of weights:
P(i)/P(j) = ωi/ωj = exp[-δ(∆G)o/RT]…
 at equilibrium, more stable states much more likely.

 What about the absolute probabilities?


 we need to normalize by dividing by the Partition Function…
The Partition Function, Z
 Z = the sum of the statistical weights of all states:

Z = Σi ωi = Σi exp[-∆Gio/RT]

 (we called this ‘Q’, earlier…)


 equal to the product of external and internal Z’s:
Z = ZextZint.

 As before, the absolute probability of observing any state, i:


P(i) = ωi/Z.

 All thermodynamic quantities derivable from Z.


 macroscopic observables correspond to ensemble averages.
 ensemble average <X> of observable X:
<X> = Σi Xi ωi / Z;
Xi = X value characteristic of state i.
Estimating KD
 KD = equilibrium constant of dissociation.
 estimated by the partition functions of products and reactants.
 ‘reactants’ = all double-strands (dsDNAs).
 ‘products’ = all fully melted single strands (ssDNAs).

 For dsDNA melting: KD = Z(ss)2 /Z(ds)=1/(βZc).


 Zc = ratio of internal partition functions = Zint(ds).
 β = ratio of external partition functions = Zext(ds)/Zext2(ss).
= the ‘strand association parameter’.

 Note: If we like, we can also model hairpin melting:


KD = Z(ss)/Z(hp) = 1/Zc.

 Also note that KD = 1/Kassoc.


Estimating Θint
 Recall that ΘB = Θint Θext.
 KD allows us to model Θext.
 however, Θint must also be estimated…
The mean fraction of stacked bps/duplex.
 Let fi denote the fraction of stacked base pairs in
conformation i.
 then, Θint is just the ensemble average of fi…
 denoted, <fi>.
 Θint can be estimated from the partition function:

Θint = <fi> = Σi fiωi/Zc

 Here, Zc = Zint is the conformal partition function;


 Only the weight of associated conformations included.

.
Estimating Statistical Weights
 Now we know how to compute Θint and Θext.
 We also have: Kassoc = βZc = β Σi ωi…
 How do we estimate the statistical weight, ωi of each
conformation?
 by decomposition.
 We model a given dsDNA conformation X…
 a ‘linear’ chain, consisting of simpler structures:
 base-pair doublets, hairpin loops, internal loops, bulges…
 each weighted independently…
 weights form a set of thermodynamic parameters.
 product of weights = overall weight.
Example
 Decomposition of a conformation into subunits…

 Overall Statistical Weight


 product of a set of smaller weights…
 which are determined by the identities the subunits.
Statistical Weight (cont.)

 Overall weight of conformation X denoted ωX.


 ωx estimated by a product of smaller weights.

 Distinct weight for each type of structure…


 si - each stacked base pair doublet of type i.
 σ1/2 - each junction between stacked/unstacked pairs.
 f(m) - each internal loop of m broken base pairs.
 sbulge - each bulge loop.
 F(n) - each terminal (hairpin) loop of n bases.
 send- each dangling end.
 we must also assign a weight for dsDNA chain association…β.
 We now address each, in turn.
Statistical Weight of a Stacked
Base-pair Doublet, s
 Nearest-neighbor model:
 Enthalpy (∆Ho) and Entropy (∆So) of doublet stack formation.
 depend only on the identity of the base pair doublet.
 10 types of Watson-Crick base pairs = 20 distinct parameters.
 Statistical weight of a stacked base-pair:
 snn(i) = exp[-∆Gio/RT].
 ∆Gio = ∆Hio–T∆Sio = Gibbs free energy of stacking.
 Many Nearest-neighbor parameter sets:
 10 Watson-Crick pairs (SantaLucia, et al., 1998).
 Various singlet-mismatches (Allawi, et al., 1997, etc.).
 Example:
Sequence-Dependence of s
 Stacking ∆Go’s depend on GC content…
 And will vary with specific doublet identity:
 i.e., adjacent pairs of base-pairs.
 We will expect the size of our propagation
parameter:

s = exp[-∆Gcho/RT],
 to increase with GC content;

 Duplexes with higher GC-content:


 should form more easily…
 and be more resistant to melting.
Modeling End Unraveling:
The Cooperativity Parameter (σ)
 Unraveling at a duplex end:
 generally modeled by σ1/2.
 σ accounts for the cooperativity of DNA melting.
 formation of an isolated base much more difficult.
 Consensus value: σ = 4.5 x 10-5 (0.1 M [Na+]).
 Some care is required:
 σ always included in chain association parameter, β.
 σ1/2 sometimes included in terminal loop weight, F(m).
 1 factor of σ1/2 sometimes ‘normalized’ into the zero free
energy state (Benight, et al., 1988).
 σ is also salt-dependent (S. Kozyavkin, 1987).
Statistical Weight of an Internal Loop
 Internal loop of m unbonded base pairs:
 Statistical weight = σfn(m):

 End unraveling:
 accounted for by 2 factors of σ1/2.
 Normalized probability of loop closure:
 Jacobson-Stockmayer: f(n) = 1/n1.5; unrestricted loop, n links.
 Purely entropic in origin (no T-dependence).
 Empirical form (R. Wartell, 1977)
 f(m) = 1/[(1-1.38-0.1m)(m+1)1.7], m > 3.
 Accounts for volume exclusion and chain stiffness.
 Note: Due to the large penalty…
 Looped conformations usually discarded for oligos.
Statistical Weight of a Bulge Loop
 Bulge Loop
 only one strand has unpaired bases.
 Example:

 Perturbation to intact helix, δ∆Go.


 statistical weight, sbulge = exp[-δ∆Go/RT]
 1-base bulges well-studied (Zhu and Wartell, 1999):
 statistical penalty roughly σ1/2 ; sequence-dependent.
 Larger bulges less well-studied:
 parameters for RNA bulges (Freier, et al., 1986).
 statistical penalty > σ, increases with size.
Statistical Weight of a Terminal Loop
 A terminal loop of n unpaired bases:
 “hairpin” loop.
 Statistical weight = σ1/2Fend(n).
 Example:

 Strand unraveling:
 modeled by σ1/2.
 Normalized probability of loop closure:
 Fend(n) = M(n)/(n+1)1.5 (Benight, et al., 1988).
 M(n) accounts for steric hindrance, chain stiffness.
Statistical Weight of a Dangling End
 Dangling ends (overhanging, unpaired bases):
 stacking of first dangling base against duplex core.
 often as stabilizing as an extra stack.

 Energetics sequence dependent.


 Nearest-neighbor model, δ∆Go (Bommarito, et al., 2000).
 Energies depend, to 1st order only on:
 Identity of dangling base + duplex core bases;
 Statistical weight, sdangle = exp[-δ∆Go/RT].
 Values for all dangles reported.
Bimolecular Helix Initiation
 Strand Association Parameter: β.

 β Accounts for both nucleation and end unraveling.


 includes a factor of σ (one σ1/2 for each duplex end).
 Length, temperature dependent (W. Hillen, 1981).
 β = KNa+b[1-Θ(int)]; K = 5000, a = -2.8, b = -3.2 ([Na+] = 0.1 M).
 Nearest-neighbor model (SantaLucia, et al., 1998):
 Simpler, approximate treatment of β (deviations < 20%).
 length-independent “initiation” free energy, ∆Gonuc.
 β = exp[-∆Gonuc/RT].
Impact of Strand Anchorage

 For duplex conformations formed on microchips


 well known to be much less stable.

 Impact may be treated as a multiplicative correction:


 A. Fotin, et al., 1998.
 length, nature of linker – no substantial effect.
 δ∆Ho = 24 +/- 4 kcal/mol
 δ∆So = 70 +/- 12 cal/(mol K)
 Then, δ∆Go = δ∆Ho – T δ∆So
 sanchorage = exp[-δ∆Go/RT].
Example 1 – Simple DNA Duplex

 Kd for formation from isolated strands:


 one factor of β for helix initiation.
 an appropriate factor of s for each stacked base pair.
 recall internal partition function for each isolated strand = 1.
 Kd = 1/βsCC/GG2sCA/GTsAA/TT2 = 1/Ka.
 Approx. form for a conformation of this type:
 let s = mean weight of doublet stacking.
 Kd = 1/βsl-1.
Example 2 – Simple DNA Hairpin

 Kd for formation from unfolded single-strand:


 one factor of ‘σ1/2F(7)’, for the terminal loop.
 one factor of σ1/2, for unraveling at the free end.
 an appropriate factor of s for each stacked base-pair.
 recall internal partition function for the isolated strand = 1.
 Ka = σsCC/GG2sCA/GTsAA/TT2 F(7); then Kd = 1/Ka.
 Approx. form for a conformation of this type:
 let s = mean weight of doublet stacking.
 Ka = σsl-1M(n)/(n+1)1.5.
Models of Duplex Formation

 General Model = all conformations considered.


 Problem: Exponential number of statistical weights.
computationally too intensive.
 Solution: choose a simpler model of duplex formation.
various models have been investigated…
Models of Duplex Formation (cont.)
 All-or-None Model (2 states):
 full-length duplex and dissociated ssDNA pair.
 good approximation for KD (complementary strands);
 Appropriate for short oligos (< 15 bps);
 Fully Aligned Model (no shifted alignments).
 still an exponential number of states, but…
 Z can be computed in TIME O(N2)…Poland’s Algorithm.
 appropriate for DNA Melting (all lengths).
 Staggered Zipper Model (1 duplex/conformation):
 includes all shifted alignments; no internal loops larger than 1.

 Z computable in TIME = O(N2).


 With singlet bulges, TIME = O(N3).
 appropriate for annealing/renaturation of short DNA.
limit: N < 100-150 bps.
Simplest Application: Short Oligos
 For short oligos, a 2-state model often used;
 Only 2 conformations:
 un-melted (2 ssDNAs) + fully-aligned dsDNA;
 Applied model can be statistical (melting curves), or van’t Hoff.
 Generally, focus is Tm value…
 Van’t Hoff assumption: @ Tm, Θ = Θ(ext) = ½.
 Example: Length 9 bp mis-matched oligo:

 Result: All-or-none model gives good agreement for short oligos:


 Both ∆Go and Tm predictions are acceptable (SantaLucia, et al.);
 However: for oligos longer than 15 bps, a shifted Zipper model required...
Limitations of the 2-state Model
 Is a 2-state model good for long oligonucleotides?
 Study: 100 dsDNAs of length 23 bps (A. Suyama, et al).
 Experimental vs. Calculated Tm values:

 Result:
 correlation pretty good, but...calculated values too low!
 Unpredicted stabilization probably due to melting intermediates.
 Adequate for predicting gross behavior/trends;
 Inadequate for accurate or detailed prediction.
Longer DNAs
 Statistical Zipper Model (1 duplex/conformation):
 very successful for predicting DNA melting, up to 150 bps;
 e.g.: Differential melting curves for 4 lac DNA fragments:
 Watson-Crick SZM predictions shown;
 Note add’l structure: 2 melting regions (two peaks) in (d)…
 For polymers > 150, a general, aligned model usually required.

a: 80 bps,
b: 101 bps,
c: 188 bps,
d: 219 bps
experimental.
---- theoretical.
Conclusion
In this Lecture, we have:
 Discussed the Helix-Coil transition of biopolymers.
in the context of DNA melting and renaturation.
 Described physical methods necessary for modeling:
Equilibrium Chemistry and Statistical Thermodynamics.
Note: also apply to protein and polysaccharide modeling.
 Investigated the generalization of the model:
 DNA strand design:
 Stat-thermo modeling of error/efficiency.
 Quantitative design for minimized error.

Next come Real-world applications:


 Low-error Tag-Antitag system design.

You might also like