Lighting up biological tissue with rotaxanes

2 05 2012

Water-soluble, deep-red fluorescent squaraine rotaxanes

Erin L. Cole ,  Easwaran Arunkumar ,  Shuzhang Xiao ,  Bryan A. Smith & Bradley D. Smith

Org. Biomol. Chem., 2012, Advance Article; DOI: 10.1039/c2ob06783h

By Loruhama M. Delgado

Squaraines are fluorescent dye molecules that emit in the red or near infrared region. Although, these dyes have been extensively studied, their use for biological imaging applications are limited by their susceptibility to nucleophilic attack and their propensity for hydrolysis in water. For this reason Smith and coworkers encapsulated squaraine dyes in macrocyclic tetralactams to produce squaraine rotaxanes. The encapsulation stabilizes the squaraines and prevents fluorescence quenching by self-aggregation.

They synthesized several rotaxanes by using a Leigh-type clipping reaction to trap the squaraines inside macrocyclic tetralactams. After that, they used click chemistry to attach protected polar groups to the ends of the squaraine, followed by a deprotection to produce eight different water-soluble rotaxanes.

They measured the stability of their compounds by monitoring their absorbance in water, and in 10% fetal bovine serum (FBS).  These studies showed that rotaxanes containing two or three stoppers groups attached to the ends of the squaraines were less stable than those containing four groups. Out of all the rotaxanes, only two that contained four stoppers groups were kept for future screening. These two compounds showed great stability in a pH range of 6-10. They also observed that these two compounds didn’t show any affinity to the protein albumin, suggesting that these compounds could be used as non-targeted tracers.

For this purpose, they performed in vivo studies with mice to compare two squaraine rotaxanes with Indocyanine Green (ICG), which is another non-targeted tracer. They took optical images of mice organs, which showed that the two rotaxanes signals were mostly in the bladder, while the one from ICG was located in the intestines. They then proceeded to compare the two rotaxanes by ex vivo fluorescence using excised organs. In these studies they found out that one of the rotaxanes tracer showed more pixel intensity values than the other one, but that both of these compounds showed very low retention in the tissues of the organs in mice.

I think that overall this communication was interesting. The narrative was engaging, the authors did a good job explaining their findings, and the paper was of appropriate length. Although they were able to get water-soluble and stable rotaxanes that exhibit some desirable characteristics, I don’t think these rotaxanes (the ones reported) could be used in broad biomedical applications. They show little retention in the tissues, so I find this application really hard, because they don’t monitor their dyes after two hours of being injected, and you can’t really see what happened before that or how they were distributed in the organs. On the other hand, these compounds do seem to have low toxicities, so maybe future derivatives will in fact be very useful as bioimaging tools.


Fancy soap to harvest light

2 05 2012

Artificial Light-Harvesting System Based on Multifunctional Surface-Cross-Linked Micelles

Hui-Qing Peng, Yu-Zhe Chen, Yan Zhao, Qing-Zheng Yang, Li-Zhu Wu, Chen-Ho Tung, Li- Ping Zhang, & Qing-Xiao Tong

Angew. Chem., Int. Ed., 2012, 51, 1-6. DOI: 10.1002/anie.201107723

By Luis R. Rivera-Ríos (Berti)

Energetic crisis has made scientists examine the efficient chemical processes used by plants for doing work. In particular, into the details of how chlorophyll aggregates harvest Sunlight. In 1990, A. Scherz and colleagues reported (Scherz et al. PNAS, 1990, 5430) the use of the Triton X-100 detergent to encapsulate photosynthetic pigments making in the process the first model system using self-assembled micelles. The approached used by A. Scherz to construct the light-harvesting system (LHS) was from a biological perspective. In contrast, the article I’m summarizing this week, although also using micelles, reports the construction of an antenna from a more “chemical” perspective. Here Hui-Qing and coworkers used a cross-linked micelle (SCMs) as a template to first covalently modify the exterior with a donor (DPA) chromophore and, second, to attach an anionic acceptor (EY) molecule using electrostatic interactions. This model system used the “bullet proof” copper catalyzed dipolar cyclo-addition to first make the cross-link micelles and second to modify them with one equivalent of DPA per micelle monomer. The synthesis of the antenna is a one-pot reaction at room temperature and the purification is a simple product filtration after it precipitates out from the reaction mixture. 1H NMR was used for monomer characterization; IR spectroscopy (alkyne stretch bands) was used to monitor the attachment of the DPA; and dynamic light scattering was used for the characterization of the micelles before and after donor attachment. The resulting active micelles range from 15 nm to 25 nm in size. If higher efficiencies are desired for the energy transfer process, the composition of the system could be easily modified by attaching a different alkynyl donor and by adding a different anionic acceptor. In this example the resulted LHS presented no self-quenching with a donor quantum yield of 80% and with no excimer formation. DPA–SCMs have a λext at 330–420 nm and a λem band at 390–520 nm, this emission is complementary to EY λext at 530 nm. By using this strategy the dipole–dipole coupling-mediated energy transfer was not lost and self-quenching was prevented. All of the luminescence experiments (and the corresponding controls) show that the donor contributes directly to the acceptor emission. The Förster radius was estimated to be 3.6 nm for donor-acceptor and 2.5 nm for donor-donnor. DPA and EY effective concentrations were of [DPA-SCMs] = 23 mm and [EY] = 1.34 mm, this gives us an idea of the low concentrations needed for making an efficient LHS. The authors decided to investigate how the chromophores communicate between them after realizing that one acceptor quenched multiple donors. They concluded that there were two mechanisms of energy transfer: 1) direct quenching, and 2) the “energy-migration pathway”, which is an effect of the antenna architecture. Finally, a model of the LHS was built in which each micelle was divided into harvesting units composed of ~48 DPAs surrounding each acceptor. This work can’t be compared to that of Scherz in 1990, but I still think that the authors should’ve cited that article. This work represents a very promising strategy for the building of harvesting systems due to the simplicity of its chemistry and the efficiency of the product. Although, a better system will consist of an assembly that can self-correct errors within the Förster radius, prior to any needed covalent fixation step.

Commented previously in:

Honing in on hydrophobicity’s autograph

24 04 2012

Signature of hydrophobic hydration in a single polymer

Isaac T. S. Li & Gilbert C. Walker

Proc. Natl. Acad. Sci. USA 2011, 108, 16527. DOI: 10.1073/pnas.1105450108

By Luis M. Negrón Ríos

There is still a lot of progress to be made in theoretical studies (molecular simulations) regarding the understanding of the hydrophobic effect from small solutes to large proteins (protein folding).  However, correlation between theoretical and experimental studies of hydrophobic hydration (ΔGhyd) data was still missing until the publication of this article of Li and Walker.  According with theoretical studies, the ΔGhyd scale with the solvent accessible surface area (SASA) of molecules follows macroscopic interfacial thermodynamics (water surface tension).  However, at the microscopic scale, macroscopic interfacial thermodynamics does not follow a correlation between ΔGhyd and SASA.

In this article, they explain this behavior by measuring the “hydrophobic signal” which is the temperature dependence of ΔGhyd. Using Atomic Force Microscopy (AFM), they performed single chain elastic stretching on hydrophobic polymers that form globules in presence of water.  These hydrophobic polymers (polystyrene (PS), poly(4-tert-butylstyrene) (PtBS), and poly(4-vinylbiphenyl) (PVBP)) were deposited on a silicon surface.  The force required to pull a single polymer to unravel the formed globule provides information of ΔGhyd of the exposed monomer in water.  To obtain the temperature dependence of ΔGhyd by this procedure, they repeated this experiments thousands of times at different temperatures (25–80 °C).  They divided their results, in three main findings: temperature, size and polymer/monomer dependence.

First, for each polymer ΔGhyd is strongly dependent on temperature and does not correlate with interfacial thermodynamics.  To understand this explanation we have to separate the macroscopic (above 1 nm) and microscopic scale (below 1 nm).  In the case of macroscopic scale, the temperature dependence of ΔGhyd follows a correlation described by interfacial thermodynamics, which means that ΔGhyd decreases with increasing temperature having a similar trend of water surface tension by increasing temperatures.  For that reason, in the macroscopic scale, this tendency is dominated by “surface area” and is “enthalpy driven” as the number of disrupted hydrogen bonds in the large hydrophobic particle scale with the surface area.  However, for small hydrophobic particles measuring less than 1 nm (microscopic scale), the ΔGhyd increases at low temperature by reaching a maximum and then decrease at high temperature.  This behavior is predicted to scale with volume instead of surface area.  According to the article, this particular increase of ΔGhyd at low temperatures results from the lowered entropy of the surrounding water molecules.

Second, results of temperature dependence of ΔGhyd at different sizes (backbone + side chain) are showed for the studies with PS (7.2 Å), PtBS (9.5 Å) and PVBP (11.4 Å).  In these, they point out that for PS, the ΔGhyd increases “monotonically” with increasing temperature.  In contrast, for PtBS and PVBP the ΔGhyd varies parabolically with temperature showing maxima at 55.1 °C and 47.8 °C, respectively.  They state that this behavior is characteristic for hydrophobic solutes whose sizes are in a “crossover regime” between ΔGhyd of small (<7 Å) and large (>20 Å) solutes.

The third finding is the relationship of ΔGhyd in free monomer versus the monomer on a polymer.  They found that the cost of hydrating free monomers is higher than hydrating monomers that are part of the polymer.  This is because the monomers of the polymeric chain are stabilized by hydrophobic interactions with other monomers that are part of the chain.  The high impact of all the presented results is a validation of theoretical predictions of temperature dependency of ΔGhyd and experimental evidence that the crossover length scale is in the order of 1 nm.

About the article, I have to agree with Garde and Patel’s comment that this study fills a gap between theoretical studies (molecular simulations) and experimental models.  This article helps to see the hydrophobic effect in a scale aspect and the presented trends between ΔGhyd at different temperatures.   There is a lot of work that have to be done to understand the hydrophobic effect in natural systems like folding in proteins.  I understand that polymers are good candidates for these kinds of studies, but to address the challenge of hydrophobicity we need materials that can be good analogues to biological systems such as proteins.  I like the article, not only because of the methodology used, but also because the implications of the reported findings makes a strong contribution to understand other aspects of water that are still elusive in terms of the hydrophobic effect.  We can dedicate many years to the story of the hydrophobic effect, but like characters that appear in stories, there are a lot of parameters, anomalies and phenomena that will continue appearing in the story of this phenomenon.

Fluoroaromatics stacking direct protein assembly

24 04 2012

Stacked Fluoroaromatics as Supramolecular Synthons for Programming Protein Dimerization Specificity

Christopher J. Pace, Hong Zheng, Ruben Mylvaganam, Diane Kim, &  Jianmin Gao*

Angew. Chem. Int. Ed. 2011, 51, 103. DOI: 10.1002/anie.201105857

By Ana Victoria Morales

Because little is known about aromatic interactions for protein design, Gao and colleagues decided to study them. Previous work by this group showed that a stacked phenyl and perfluorophenyl pair controls the dimerization specificity of a protein. That led them to the studies done in this article that involved the analysis of the aromatic stacking energetics by incorporating several stacked aromatic pairs into the protein α2D.

Fluorinated analogues of phenylalanine in the residues Phe10 and Phe29 were incorporated into α2D. All mutant proteins were able to fold into homodimeric complexes. A van’t Hoff analysis was used to study the thermodynamic parameters of the dimerization. The melting temperatures for the α2D homodimers varied from 29-78 ºC, a difference of almost 50 ºC between the wild type and the (Z,Z) which is indicative of the greater hydrophobicity of fluorocarbon compounds. The folding free energies (∆Gf) ranged from -5.9 to -12.8 kcal mol-1. The plot of the folding free energy against LogP of the fluoroaromatic side chains showed a poor correlation as well as the plot of the folding free energy against the surface area. Because of this, other factors than hydrophobicity must be contributing to the stability of the homodimers. Interestingly, the folding stability agreed with the magnitude of the dipole moments, this is indicative that dipole-dipole coupling may be stabilizing the homodimers. For example, (F345F345) showed the most favorable folding free energy between all variants and it had the greatest dipole moment of all the amino acids. Then, another plot was done of the free energy against the combination of the LogP and the dipole moment. It had a much better correlation (R2=93), indicating the contribution of both hydrophobicity and dipole moments of the aromatic rings.

Dipole-induced-dipole interactions were studied on α2D single mutants. The stability is supposed to reveal the best fluorinated phenylalanine analogues for targeting a native phenylalanine. Surprisingly, the (F,Zo) mutant gave the most stable homodimer. This was surprising because Z is more hydrophobic and it was expected to have a more favorable quadrupole with the phenylalanine. This is indicative of the combination of the hydrophobicity and dipole moment for strong stacking interactions with the native aromatic residues. It was then hypothesized that aromatic interactions can direct orthogonal molecular assembly or self-sorting behavior of peptides. Thermodynamic equilibrium through a disulfide cross-linking experiment of a three component system was studied with mutants (F, F), (F345F, F345F), and (Z, Z). Theoretically speaking, random dimerization should have given six species. In the LC-MS experiment only two significant peaks were observed, those of the heterodimer of mutants (F,F) and (Z, Z) which is stabilized by quadrupole interaction and the other (F345F, F345F) homodimer stabilized by the dipole-dipole coupling.

I enjoyed reading this article because I learned a lot, and assume that everybody learns something because as the authors say that this one of the first investigation of the energetics of aromatic stacking in proteins. The article put into perspective how much fluorine could change the energetics of the aromatic residues. There are small but very significant details to be considered such as the amount of fluorines incorporated and their position. Fluorinated amino acids are useful because of the benefit of fluorination in NMR analysis, PET imaging and protein stabilization. The information obtained is very useful because it provides insights for the energetic considerations of incorporating fluorinated aromatic amino acids into target proteins. It also highlights the importance, not only of the hydrophobic contributions, but the dipole contributions to aromatic stacking. Zo was the most efficient in targeting native Phe residues through aromatic stacking, this finding helps in the design of enzyme inhibitors. It was also proven that self-sorting does occur with these peptides by the use of stacked aromatics as supramolecular synthons. Nevertheless, for this last study I would have liked to see more experiments with several combinations of the fluorine analogues to see if the amount of fluorines in the analogue and their position affect the self-sorting. It would have been nice to see if there were other types of assemblies, but that may need another article. Overall, it is a nice article that has to be read carefully to understand the next step and why the experiments were done.

Asymmetric catalysis using a chiral supramolecular box

18 04 2012

Self-Assembly of a Confined Rhodium Catalyst for Asymmetric Hydroformylation of Unfunctionalized Internal Alkenes

Tendai Gadzikwa, Rosalba Bellini, Henk L. Dekker & Joost N. H. Reek*

J. Am. Chem. Soc., 2012, 134 (6), 2860; DOI: 10.1021/ja211455j

A synopsis by Yazmary Melendez

In this article, J. Reek et al. continue their work developing encapsulated catalysts for the selective hydroformylation of unfunctionalized internal alkenes. They created a supramolecular box using 3 ([ZnII(salphen)]) as the template and 2 (a chiral phosphonyl-based ligand) as the dipyridyl pillars. The choice for 2 was made because, once the box self-assembled, the conformation would bring two phosphorus atoms in close proximity. Once the active catalyst was formed, the cavity within the box would be sterically crowded enough to diminish the conformational freedom of the alkenes and thus enhance the stereoselectivity.

They confirmed the assembly with proton NMR and confirmed it was a discrete structure via a titration, which was monitored by circular dichroism. They also used a diffusion NMR experiment to determine the amount of each compound in the structure and concluded the results were consistent with that of a 2+2 structure, which was also consistent with the computational models.

After obtaining the desired supramolecule, they tested its use in Rh-catalyzed asymmetric hydroformylation of cis- and trans-2-octene. They used two monodentate ligands (1 and 2) to compare this function in both, and observed in both cases a preference for aldehyde a. Although, when 2 was used in box 1, they observed the regioselectivity had reversed, causing a mild shift in favor of aldehyde b.  They also observed that the use of ligand 1 in the reaction increased the ratio of the R enantiomer of aldehyde b from 61 to 86%. Afterwards, they tested the substrate scope for 1 by doing hydroformylation reactions on a series of cis- and trans-2-alkenes. Using ligand 1, they observed the systems favored the formation of the innermost aldehyde with the trans-2-olefins yielding 80% of the major enantiomer the cis- isomers yielded 90%.

They performed further experiments to test out the overall performance as well as the stability of the ligand. In order to do this, they studied its performance at high temperature, undergoing the catalysis at 70 °C. The results suggested that the system still exhibited high enantiomeric ratios for the innermost aldehyde while similar experiments with ligand 2 produced primarily racemic mixtures.

After reading this paper, I thought it was well constructed in the sense that all the experiments I assumed they needed to do were reported. They confirmed the type of assembly and underwent a lot of studies to confirm their original assumptions regarding the enhanced selectivity afforded by the smaller the box’s cavity. I can’t really criticize their experimental design or their results. I think, however, the figures were lacking a little bit. Specifically, since they were all grayscale, when you take a look at Figure 1, for example, you can’t really appreciate the Spartan PM3 structure of the supramolecule. Furthermore, I think that the numbering was somewhat confusing, since they used numerals to identify everything. I think it would’ve been nicer if they used 1, 2 and 3 for the supramolecules, the template and the pillars (respectively), like in the paper, and then when they were to talk about the ligands and such they could’ve used roman numbers or Greek letters or something else. That way, when they talk about 1, for example, it would’ve been unambiguous that they were referring to the supramolecule and not the first ligand. I just found this issue confusing when I read it the first time.

Exploring supramolecular space

15 02 2012
Chemical space is a framework used to describe the vast amounts of known molecules of natural or synthetic origin. The main premise is that similar structures have similar properties (e.g., SAR), and thus, can be clustered in groups (‘galaxies’) using such properties as criteria (parameters) for their classification. The concept is primarily used in the field of drug discovery and to a lesser degree in materials development.
Supramolecules are not explicitly excluded from the working definition of chemical space, nor are they implicitly included. Thus, the term of supramolecular space in meant to specifically target the study of how intrinsic parameters (i.e., spatial covalent connectivity) determine the properties of the resulting supramolecules. It can be distinguished from the standard definition of chemical space by the explicit inclusion of extrinsic parameters (e.g., media, co-solutes, temperature, pH), which very often dictate the very existence of the supramolecule in the first place.
Since supramolecular space is very vast, and just like our outer universe, continues to expanding at an accelerated rate. So, this blog will focus primarily on selected “galaxies” (topics) that are most related to our research interests. The following is a partial list of topics that will be discussed in this site:
  1. Assemblies of molecules: Not simple aggregates, but groups of molecules interacting via specific non-covalent interactions (NCIs) (although some interesting cases are at the edge of this definition) with atomic/molecular precision.
  2. NCIs and reversible covalent interactions: Reversibility and timescale are of the essence here
  3. Complex molecular systems (e.g., hierarchical, dynamic libraries): With emphasis on synthetic systems, but with with specific interest in combinations of synthetic and natural systems
  4. Molecular information transfer
  5. Design and synthesis of molecular machinery
If you think we should expand this list of topics leave your suggestions in the comments.
This week we will discuss the first two articles related to supramolecular space. In one of them Jean discusses a recent article by the groups of Wheeler, Houk and Kool dealing with a quantitative assessment on the nature of hydrogen bonds within the context of base-pairing in DNA. Hydrogen bonding interactions are arguably the most studied of all NCIs, yet there are still important details to be found out as Jean explains in his post. Andrew, on the other hand, discusses a recent article by the groups of Rastrelli and Prins where they report the advantages of using 13C labeling strategies for the NMR analysis of a complex supramolecular dynamic library. I’m looking forward to a lively discussion both in the comments and on Friday’s GM.

Analysis of a dynamic library enabled by 13C-labeling strategies

15 02 2012

Marta Dal Molin,  Giulio Gasparini,  Paolo Scrimin,  Federico Rastrelli and Leonard J. Prins
Chem. Commun., 2011, 47, 12476-12478; DOI: 10.1039/C1CC15295E

By Andrew James Surman

As we saw a few months ago, discussing the Würthner group’s review on self-sorting in non-covalent assemblies [Chem. Rev., 2011, 111, 5784-5814], development and use of increasingly complex supramolecular systems is significantly limited by a lack of tools to characterise and quantify them. HPLC is wonderful for well-behaved covalently bound systems, but where real-time measurements are required or interactions are too weak to withstand chromatography, this option isn’t available to us. While 1H NMR spectroscopy is vital during synthesis, and common for studying interactions in simple supramolecular systems, increasing complexity quickly leads to the overlap of resonances, hampering quantification of species and structural analysis. As a result, its use in complex synthetic systems relies on serendipitous resolution, or very careful (and rather limiting) design of systems to facilitate resolution. In our group we know that even mixing two components (a deoxyguanosine derivative and a salt) can lead to horribly complex forest of 1H resonances, where reasonable assignment requires a considerable amount of extra NMR experiments (time-consuming, inconvenient) and where reliable quantification of species can be impractical or impossible.

Despite these limitations, the authors of this week’s article believe that NMR is the answer: previously that have observed that NMR spectroscopy “is one of the few techniques that potentially allows direct identification and quantification of all species present in a solution as a function of time” [Chem. Comm., 2008, 3034-3036] and here add that “the limited use of NMR spectroscopy as an analytical tool within the context of systems chemistry is quite remarkable, considering that NMR is widely accepted as the tool for studying complex biomolecular structures.” They propose we learn from that field, and use isotopic enrichment of what would otherwise be low-abundance nuclei (13C, 15N) as a means to simplify spectra, and increase sensitivity for nuclei whose resonances are usually better-resolved than 1H. As a demonstration of the approach they apply it to an imine-type dynamic covalent library, closely related to several previously studied in the Scrimin & Prins groups (working towards studying mechanistic mimics of serine protease), setting themselves the task of characterising the contribution of spectator groups to the stability of a compound in the minimum number of experiments.

The complete library was formed from four separate aldehyde ‘scaffolds’ (P1-4), which are able to react with amines (A and A+) or hydrazides (H and H+) to form a range imines (PxA or PxA+) or hydrazides (PxH or PxH+). The aldehydes all bore negatively charged phosphonate groups, a transition state analogue for ester hydrolysis, except the control P1. The amines and hydrazines either bore a pendant quaternary amine (A+, H+), the postitive charge of which was previously shown to interact with phosphonates and bias product distribution, or phenyl groups (A, H) as a control. 13C labels were introduced during aldehyde formation by quenching of a lithiated (o-directing) phenol with carbonyl-13C-N,N-dimethylformamide.

Mixing all the components (P1-4, A and A+, H and H+) in MeOH resulted in a 1H spectrum in which many of the iminic resonances were overlapped in the 8 – 9.5 ppm range. In contrast, as a result of the 13C labelling, each of the 26 components of library (including hydrazone E/Z isomers) was manifested by a single (iminic carbon) resonance in a 1H-decoupled DEPT-90 spectrum (improved sensitivity over standard 13C), and all were resolved, allowing quantification down to 0.1 mM. These resonances were assigned through stepwise addition of components to build up the library, most could be quantified in a reasonably short experiment (the 18 best-resolved could be quantified in a 20 min, all 26 required 12h), and concentration down to 0.1 mM could be detected.

Despite the success of the labelling approach in allowing quantification of all 26 components, it was not possible to obtained thermodynamic stabilities of all the species from analysis of the mixture. Since the imines are far less stable, only stoichiometric amounts of hydrazides were used to allow coexistence of all 26 species (and show off the labelling approach), so ‘cross-talk’ between equilibria was possible (the authors have published on the limitations of dealing with these equilibria before) [Chem. Comm., 2007, 1, 1340-1342]. Instead, the hydrazone and imine libraries were analysed separately, and the energy scales correlates by competition experiments with P1-4A and P1-4H. Findings followed the main trends expected (hydrazones more stable than imines; electrostatic interactions of charged amines with phosphonates lending stability), however significant contributions from ‘spectator groups’ on the aldehydes were less predictable, and the authors are still working on a fuller treatment (treating hydrazone isomers separately).

This communication neatly proves a point which people in many other areas of chemistry might consider obvious (that isotopic labelling helps resolve spectra of complex systems), but which synthetic supramolecular chemists seem rarely to use. The spectra (Fig 2) speak for themselves, and it is seems a marked improvement over the 1H-13C HSQC method they proposed in previous work on these systems; I imagine that it would become more impressive as unlabelled spectra get more crowded. Nonetheless, I will be interested to see if they include an estimate of errors in the follow-up: while they note that DEPT-90 is fairly insensitive to variations in 1JCH values, I would be interested to see some data.

Dissecting hydrogen bonds in DNA

15 02 2012

Measurement and Theory of Hydrogen Bonding Contribution to Isosteric DNA Base Pairs

J. Am. Chem. Soc., 2012, 134 (6), pp 3154–3163 DOI: 10.1021/ja210475a

By Jean Carlos Rivera

There is conflicting evidence on the extent of the interactions of a Thymine (T) like compound, 2,4-difluorotoluene (F), when it is substituted in place of T in the sequence of a DNA duplex. In F the carbonyl groups that function as hydrogen bond acceptors in T are replaced by fluorine, resulting in a less polarized structure with extremely similar size and shape. Some crystal structures of F substituted DNA duplexes reveal that the distance between Adenosine (A) and F is relatively short, implying the possibility of hydrogen bond formation between F and A, while others show a larger distance implying little to no interaction. It has also been shown that the substitution of F for T was unfavorable to the helix, a result that was contrasted by experiments that showed that DNA polymerase handled F efficiently. The larger chloro version of F, 2,4-dichlorotoluene (L), was also handled extremely well by the polymerase, a result that was puzzling as well. So, in order to completely understand the effects of these substitutions, the groups of Wheeler, Houk and Kool undertook the impressive set of experiments and simulations described in this article.

To understand how base pairing (and mismatches) affect the stability of the DNA duplex, they performed extensive UV-Vis melting experiments. They measured the melting point of a wide variety of sequences, 12 and 14 nucleotides long, introducing one and two mismatches respectively. The data obtained from these experiments allowed them to determine the thermodynamic parameters of the melting/hybridization of each sequence, allowing them to assess the overall contribution of each mismatch to the stability of the DNA duplex. They obtained similar results using Isothermal Titration Calorimetry on selected sequences.

The stabilization that each base pair confers to a DNA sequence can be dissected into π-π stacking interactions with adjacent base pairs and hydrogen bond formation between complementary nucleobases. The extent of these interactions can be appropriately measured by terminal base pairing (TBP) measurements.

Their results show that F destabilizes the DNA duplex, and looses any selectivity towards A. This is because hydrogen bond formation between F and A is weaker than the interactions of A with the solvent. Formation of the duplex implies placing F in close proximity of A stripping it of solvent interactions, an action that results in an overall destabilization of the duplex. This destabilizing interaction between F and A is also highlighted by the TBP experiments. Finally, they used computational methods to determine the energetics of each base pair they formed during the melting experiments. The measured pairing energy was stripped down to its components using smaller molecules with similar motifs, allowing them to determine the major energetic contributions of the recognition motifs in each base pair.

In the last part of the article, they proceeded to explain why there are different crystal structures for similar base pairs. One possibility they suggest is that the F-A base pair can effectively “breathe” which causes the formation of crystals with different distances between the nucleobases. They also explain that the apparent recognition of F and L by the DNA polymerase is due to the protein’s ability to recognize the shape of the nucleobase instead of hydrogen bond complementarity.

With regards to the narrative, I find that the article was very well written. The only flaw I can point out is that it becomes a bit repetitive towards the end. This is mainly because they state their findings in the introduction, the results section and in the discussion. The figures were nice, the graphs really helped understand the data presented in the tables. In my opinion a graphical representation of the TBP results would have been better than the table they presented. Overall I think this is a really nice article, although a bit too fundamental for my taste.