Ojima I. (ed.) Fluorine in Medicinal Chemistry and Chemical Biology

Подождите немного. Документ загружается.

Fluorinated Amino Acids and

Biomolecules in Protein Design and

Chemical Biology

He Meng , Ginevra A. Clark , and Krishna Kumar

16.1 Introduction

While materials scientists have long appreciated the usefulness of ﬂ uorinated compounds,

chemical biologists and protein scientists have just begun to uncover their value. Fluorine

has unique and fascinating properties. The judicious replacement of hydrogen with ﬂ uorine

can provide tools for unraveling the workings of biological systems. Rational design using

ﬂ uorinated amino acids has been successful in deciphering collagen stability, in mapping

ligand – receptor interactions, in unveiling protein folding and dynamics, and in creating

extra - biological structures. In addition, the third - phase properties of perﬂ uorocarbons have

been utilized in biomolecule enrichment, artiﬁ cial membranes, and small molecule micro-

arrays. Unusual properties of ﬂ uorinated molecules will continue to proffer novel ways to

perturb and observe biological systems.

16.1.1 Fluorine in the Context of Biological Systems

Noncovalent interactions are central to many key biological functions. The self - assembly

of the plasma and organelle membranes, association between receptors and ligands, and

folding of RNA and protein molecules are all controlled by noncovalent interactions.

Noncovalent forces in the design of proteins have proved useful in gaining insight into

Fluorine in Medicinal Chemistry and Chemical Biology Edited by Iwao Ojima

412 Fluorine in Medicinal Chemistry and Chemical Biology

sequence/structure and function relationships. We and others have employed ﬂ uorocarbons

as building blocks to construct biomolecule analogues and exploited their unique charac-

teristics in biological systems. Organic ﬂ uorine exhibits unusual properties. At the level

of substitution of hydrogen by a single ﬂ uorine atom or a few ﬂ uorine atoms (e.g., H to

F, CH

to CF

), the effects are mainly manifested in electronic, steric, and hydrophobic

properties of the resulting compounds. On the other hand, for perﬂ uorocarbon chains,

third - phase properties, as distinct from water and hydrocarbons, become more relevant.

We ﬁ rst review here the enrichment and self - assembly of biomolecules based on ﬂ uorous

phase separation. We then highlight some recent uses of ﬂ uorinated amino acids in protein

design and chemical biology.

16.1.2 Unique Properties of Fluorine

Fluorine is the most electronegative element. While ﬂ uorine forms the strongest hydrogen

bonds in the ionic state, it does not readily participate in hydrogen bonding once covalently

bound to carbon [1] . This observation was at ﬁ rst surprising, but careful scrutiny of crystal

structures has revealed few organic molecules displaying F

…

H intermolecular distances

required for hydrogen bonding. These interactions only occur in the absence of better

hydrogen - bond acceptors. This property illustrates an essential feature of ﬂ uorine: it holds

its unshared electrons tightly, evident in its low polarizability ( α = 0.557 × 10

− 24

c m

− 3

= +0.13 kcal/mol) [2, 3] . The C – F bond (485 kJ/mol) [4] is the strongest formed by

carbon bonded singly to any element. Perﬂ uorocarbons are chemically inert and thermally

stable. The C – F bond has a reversed and large dipole moment ( µ = 1.85 D) [5] compared

with that of a C – H bond ( µ = 0.4 D) [2] . Despite electron localization on ﬂ uorine, perﬂ uo-

rocarbons are minimally polarizable, resulting in weak intermolecular interactions and

high vapor pressures.

16.1.3 Phase Separation Properties of Fluorocarbons

The most striking feature of perﬂ uorocarbons is their insolubility. They phase separate

from nonpolar organic solvents and water at room temperature. This property was exploited

by Horv á th in 1994 with the introduction of ﬂ uorous biphasic catalysis (FBC) [6] . This

new paradigm for separation and puriﬁ cation of catalysts and organic molecules has been

used widely [7 – 9] . Gladysz and Curran [10] have formalized the deﬁ nition of the term

“ ﬂ uorous ” in analogy to “ aqueous ” as “ of, relating to, or having characteristics of highly

ﬂ uorinated saturated organic materials, molecules or molecular fragments. Or, more simply

(but less precisely), ‘ highly ﬂ uorinated ’ or ‘ rich in ﬂ uorines ’ and based upon sp

- hybridized

carbon. ”

The Hilderbrand – Scatchard solubility parameter δ (Equation 16.1 ) can be used to

estimate the miscibility of ﬂ uorocarbons with organic solvents [11 – 13] .

δν=

()

∆E

(16.1)

Fluorinated Amino Acids and Biomolecules in Protein Design and Chemical Biology 413

∆ E

is the vaporization energy of the pure component (cal/mol) and v is the molar volume

(cm

/mol) at temperature T . Perﬂ uorocarbons have relatively small δ values due to large

molar volumes and exceedingly low propensities for intermolecular interactions [11] .

In contrast, water has a δ value [14] of

23 5

12 32

. cal cm

−

. The mutual solubility of two

nonpolar liquids (1 and 2) can be estimated primarily on the difference between δ values

| δ

− δ

|. Solubility also depends on the temperature ( T ) and molar volumes (v

and v

)

(Equation 16.2 ).

νν δδ

12 12

()

⋅−

()

< RT

(16.2)

In general, when δ

= δ

, there is no heat of mixing and two liquids are miscible in all

proportions. When | δ

− δ

| is less than

12 32

. cal cm

−

for an average molar volume

of 100 mL, the liquids are still completely miscible at room temperature [12] . When the

differences in δ become substantial, phase separation occurs.

16.2 Fluorous - based Methods in Chemical Biology

16.2.1 Fluorous Enrichment of Peptides

Fluorous afﬁ nity separation was originally used to remove catalysts from complex reaction

mixtures [6] . A perﬂ uoroalkyl moiety (generally no shorter than − C

) is appended to a

compound of interest. Tagged molecules are then rapidly separated from other components

in the mixture by either liquid – liquid extraction or liquid – solid - phase extraction. Fluorous

afﬁ nity - based separation has recently been used in biomolecule puriﬁ cation, proteomics,

and microarray experiments [15 – 20] .

Solid - phase peptide synthesis (SPPS) [21] has greatly facilitated the preparation of

these important biomolecules. The ﬁ nal products often reside in a mixture containing

similar compounds, due to incomplete coupling of certain residues. Puriﬁ cation of the

desired product from the crude mixture is tedious, costly, and environmentally unfriendly.

Our laboratory has developed a new ﬂ uorous capping reagent, a trivalent iodonium salt,

to simplify the puriﬁ cation of polypeptides synthesized on solid support (Figure 16.1 a)

[15, 16] . In analogy to routine capping steps using acetic anhydride (Ac

O)/diisopropyl

ethylamine (DIEA), this reagent reacts aggressively with free amines of α - amino acids to

deliver secondary amines with ﬂ uoroalkyl chains. The newly formed bonds between ﬂ uo-

roalkyl tags and peptides are stable enough to survive subsequent coupling, deprotection,

and cleavage conditions in both t - Boc and Fmoc chemistry. The capped peptides can be

removed either by centrifugation or by ﬂ uorous solid - phase extraction (FSPE). Removal

of deletion products greatly simpliﬁ es the puriﬁ cation of desired peptides on reversed -

phase high performance liquid chromatography (RP - HPLC).

Complementary to our capping approach, Boom and Overkleeft have pursued a dif-

ferent strategy using the tagging method (Figure 16.1 b) [22, 23] . This method is suitable

for SPPS using Fmoc chemistry. Peptides are assembled with routine capping steps

(Ac

O/DIEA). Upon the completion of the peptide chain, benzyloxycarbonyl - (or

414 Fluorine in Medicinal Chemistry and Chemical Biology

Figure 16.1 Fluorous puriﬁ cation of peptides synthesized on solid phase. (a) Capping method and trivalent iodonium salt reagent

used in both t - Boc and Fmoc chemistry. (b) Tagging method and reagents for Fmoc chemistry.

Fluorinated Amino Acids and Biomolecules in Protein Design and Chemical Biology 415

methylsulfonylethoxycarbonyl - ) based ﬂ uorous ponytails are appended to the full - length

products. The cleavage step frees all peptides from the resin. Tagged full - length peptides

are easily puriﬁ ed on Fluophase ™ columns or by FSPE. An additional required detagging

step then delivers the ﬁ nal products.

In addition, ﬂ uorous tags have been developed for puriﬁ cation of oligonucleotides

[18] and oilgosaccahrides [19] synthesized on solid support.

Peters and colleagues have implemented a similar strategy in proteomics [24] . Pro-

teins of interest were ﬁ rst subjected to enzymatic digestion. Peptide fragments were then

selectively labeled with ﬂ uorous tags via cysteine residues or by β - elimination/Michael

addition to phosphorylated peptides. Tagged peptides were signiﬁ cantly enriched and

isolated from complex mixtures by FSPE. These samples can be directly analyzed by

MALDI - MS or ESI - MS. Fluoroalkyl tags do not undergo fragmentation, providing a

distinct advantage over biotin - based afﬁ nity reagents. This ﬂ uorous derivatization and

enrichment strategy has been extended to analyze small molecules and peptides using

desorption/ionization on silicon mass spectrometry (DIOS - MS) [25] .

16.2.2 Fluorous Small - Molecule Microarrays

Microarrays have tremendously accelerated gene sequencing and biological sample screen-

ing [26, 27] . The distinct advantages are the high - throughput nature and the minuscule

amounts of sample required. In contrast, small molecules of interest in pharmacology

and biology are discovered by tedious and expensive procedures. Small - molecule microar-

rays (SMMs) offer an attractive alternative for this discovery process. One challenge

confronting SMMs is the immobilization of small molecules onto glass slides. This step

conventionally relies on covalent modiﬁ cations of small molecules, often requiring

multiple chemical steps. Fluorous SMMs appear to be promising for alleviating this

difﬁ culty.

Pohl and co - workers have developed ﬂ uorous - based carbohydrate microarrays [28] .

The noncovalent interactions between ﬂ uorous tagged carbohydrates and ﬂ uorous func-

tionalized slides are the focus here in the array fabrication. Sugars (mannose, galactose,

N - acetylglucosamine, and fucose) were selected as initial targets. The C

tails were

appended to protected trichloroacetimides of these monosaccharides to produce ﬂ uorous

tagged sugars. Next, ﬂ uorous sugars were spotted onto a commercially available glass

slide coated with a Teﬂ on/epoxy mixture. The ﬂ uorous - based microarrays were then inter-

rogated using ﬂ uorescein isothiocyanate - labeled jack bean lectin concanavalin A (FITC -

ConA). FITC - ConA bound exclusively to mannose as expected. This experiment

demonstrated that the C

groups anchored ﬂ uorous sugars onto the glass slide, and

that this type of microarray could be used for investigating carbohydrate – protein

interactions.

Spring and co - workers have applied the same principle for fabricating ﬂ uorous - based

SMMs to illustrate the recognition of small - molecule ligands by proteins [29] . The

prototype binding pair of biotin – avidin was employed, in which biotin was conjugated

to ﬂ uorous tags and “ printed ” onto ﬂ uoroalkyl - coated glass slides. Avidin interacted with

biotin on the glass surface as judged by the ﬂ uorescence emanating from dyes linked

to avidin.

416 Fluorine in Medicinal Chemistry and Chemical Biology

Schreiber and co - workers have also developed ﬂ uorous - based SMMs (Figure 16.2 ) for

the discovery of histone deacetylase (HDAC) inhibitors [30] . Twenty small molecules com-

prised of putative active and inactive HDAC inhibitors were appended to C

- based

anchors. Among them was included suberoylanilide hydroxamic acid (SAHA), a known

inhibitor of multiple members of the HDAC family. Once the array was produced as above,

puriﬁ ed His - tag fused HDAC2, HDAC3/NCoR2, and HDAC8 were allowed to interact

with the array. Detection by Alexa - 647 - labeled anti - His antibody revealed several binders

for each enzyme. The nonﬂ uorous tagged compounds were examined to check their inhibi-

tory capacity against their targets in a ﬂ uorescence - based biochemical activity assay.

Further, these compounds were also studied by surface plasmon resonance (SPR) for

binding to HDAC3/NCoR2. It is important to note that the results obtained from these three

different techniques are in good agreement, validating the ﬂ uorous - based SMM method.

Figure 16.2 Fluorous small - molecule microarrays. Small - molecule histone deacetylase

(HDAC) binders are noncovalently immobilized onto a glass slide coated with ﬂ uorocarbon

compounds. An antibody labeled with a ﬂ uorescent dye recognizes HDAC proteins. See color

plate 16.2.

( Source: Vegas, A. J., Brander, J. E., Tang, W. et al. , Fluorous - based small - molecule microarrays

for the discovery of histone deacetylase inhibitors, Angew. Chem. Int. Ed. (2007), 46 , 7960 –

7964. Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.)

Fluorinated Amino Acids and Biomolecules in Protein Design and Chemical Biology 417

16.2.3 Fluorinated Lipids

Phospholipids are ubiquitous components of biological membranes and play key roles in

conﬁ ning cellular components, regulating cell – cell interactions, and supporting membrane

protein and lipid functions. They are therefore commonly used to fabricate model mem-

branes such as vesicles and supported lipid bilayers for studies in cell biology, drug

delivery, and materials science. Clustered display of ligands and receptors has been pos-

tulated to be mediated by phase - separated microdomains (or “ lipid rafts ” ). Using ﬂ uoro-

carbons to control lipid self - assembly and present clustered biological ligands would be

very useful for probing such mechanisms.

Bendas, Schmidt, and Tanaka have designed ﬂ uorinated glycolipids with a glycerol

diether core. The glycolipid sLexFL88 (sialyl Lewis

hexasaccharide with a n - perﬂ uorooc-

tanyl tail) self - associates into atypical clusters in 1,2 - distearoyl - sn - glycero - 3 - phosphocho-

line (DSPC) lipids [31] . With 1 mol% of sLexFL88 in DSPC, the domains average ∼ 300 nm

in diameter. The size of these clusters increases with increasing concentrations of sLexFL88.

CHO (Chinese hamster ovary) cells expressing E - selectins were used to explore leukocyte

rolling on these model membranes. The size of sLexFL88 clusters and the distance

between clusters dramatically inﬂ uenced cell binding and rolling [32] . Clustering of

sLexFL88 enhanced the cell - binding behavior; however, cells migrated more slowly on

membranes composed of sLexFL88 and LacFL88 than on those made from sLexEO6 and

DSPC [32] .

Our group has fabricated supported lipid bilayers using 2 - dipalmitoyl - sn - glycero - 3 -

phosphocholine (DPPC) and its ﬂ uorinated derivative [33] . The ﬂ uorinated derivative was

synthesized by replacing the terminal n - hexyl groups of acyl chains with n - perﬂ uorohexyl

groups. Such membranes contain intricate composition - dependent structures, which are

stripes of ∼ 50 – 100 nm interspersed between ∼ 1 µ m sized domains. Furthermore, variable -

temperature atomic force microscopy (AFM) revealed that domains and stripes are intrin-

sic features of two gel phases of DPPC and its ﬂ uorinated counterpart at low temperatures.

These results suggest that ﬂ uorinated lipids self - assemble into segregated microdomains

in biologically relevant lipid environments. Clustering derived from ﬂ uorination could be

used for investigating polyvalent interactions and producing biomaterials with speciﬁ c

surface functionalities.

16.3 Fluorinated Amino Acids in Protein Design

16.3.1 Unique Properties of Fluorinated Amino Acids

There are myriad ways by which ﬂ uorine can be exploited to perturb the properties of

biomolecules. Fluorine is scarce in living systems and the use of

F NMR for studying

protein structure and dynamics is extremely useful. Fluorine is massively electron with-

drawing and signiﬁ cantly perturbs acidity constants of amino acids [34] . For example, the

p K

values for pentaﬂ uorophenylalanine (2.2 and 8.3) [35] and hexaﬂ uorovaline (3.17 and

6.30) [36] are greatly shifted from their hydrocarbon counterparts (2.16 and 9.1 for Phe;

2.61 and 9.71 for Val) [35] . It can also inﬂ uence hydrogen - bonding properties of the

418 Fluorine in Medicinal Chemistry and Chemical Biology

backbone. The inductive properties reverse the quadrupole of aromatic rings, altering

magnitudes of π – π interactions and π – cation interactions [37] . Fluorine substitution also

inﬂ uences the conformational preferences of proline, resulting in a change in the trans/cis

ratio of the peptide bond [38] . Finally, substitution of hydrogen by ﬂ uorine inﬂ uences the

α - helical propensities of amino acids [39] .

While a C – F substitution dramatically changes the electronic properties of a C – H

bond, it exerts only a minor steric inﬂ uence [40] . The van der Waals radius of ﬂ uorine

(1.47 Å ) is only slightly larger than that of hydrogen (1.2 Å ) [41] . In general, a C– F group

is nearly isosteric with a C – H group. A CH

to CF

substitution, on the other hand, signiﬁ -

cantly increases the bulk. The volume of the van der Waals hemisphere changes from

16.8 Å

for CH

to 42.6 Å

for CF

[4] . The van der Waals volume of CF

is comparable

to an ethyl group and is only slightly smaller than that of an isopropyl group [42] .

Another important consideration is how substitution of CH

with CF

inﬂ uences

hydrophobicity. The driving force for folding of globular proteins, and many ligand

binding events, is hydrophobic interactions [43, 44] . The strength of the hydrophobic effect

is related to the size and the molecular nature of the hydrophobic group. Whitesides and

co - workers have demonstrated that the interaction of a hydrophobic ligand with carbonic

anhydrase is directly proportional to the surface area of the hydrophobic group [45] .

Fluorocarbons were more hydrophobic than hydrocarbons as the surface area is larger for

ﬂ uorocarbons. (Hansch parameters for CF

( Π = 1.07) and CH

( Π = 0.5) groups) [46] .

The Hansch parameters for side - chains of hexaﬂ uoroleucine and Leu are 1.87 and 1.58

[47] , pointing to the superior hydrophobicity of the ﬂ uorinated amino acid. As such, the

substitution of CH

with CF

should result in an increase in the conformational stability

of a protein if the CF

group is shielded from exposure to solvent in the folded

structure.

16.3.2 Incorporation of Fluorinated Amino Acids into Proteins

In order to use ﬂ uorinated amino acids to study biological systems, they need to be syn-

thesized and incorporated. Despite the challenges in both steps, there are several methods

available. For instance, enantiomerically pure ﬂ uorinated amino acids may be prepared by

asymmetric synthesis or by stereochemical resolution using enzymatic methods [48] .

Fluorinated amino acids can be introduced into proteins biosynthetically, or chemically

by SPPS. Several reviews that detail the synthesis of enantiomerically pure ﬂ uorinated

amino acids and incorporation methods into proteins are available [48 – 51] .

16.3.3 Fluorine NMR for Structure Determination

Nuclear magnetic resonance (NMR) methods have emerged as an important complement

to X - ray crystallography for determining protein structure. Structural details can be

obtained for proteins that are not readily crystallized, such as membrane proteins or molten

globules. However, even with NMR methods, it is inherently difﬁ cult to obtain structural

information on dynamic protein states. One major problem is that

H NMR resonances

tend to overlap, making interpretation of spectra difﬁ cult. Because

F has a large magne-

Fluorinated Amino Acids and Biomolecules in Protein Design and Chemical Biology 419

togyric ratio, is present in 100% natural abundance, and displays a broad dispersion in

chemical shifts, it represents an invaluable probe for exploring protein dynamics and

protein – protein interactions [52] .

Rat intestinal fatty acid binding protein (IFABP) binds and transfers fatty acids to

their metabolic destination. Structures of apo - and holo - IFABP have been solved by NMR

and X - ray crystallography. IFABP consists of two short α - helices and 10 antiparallel β -

sheets arranged to form a ligand - binding cavity (Figure 16.3 ). The side - chains of several

aromatic amino acids in IFABP play an important role in both structure and ligand binding.

Knowledge of the details of the conformation and dynamics of these side - chains would

allow a deeper understanding of the protein – ligand interaction and protein folding. To

extract such details, Frieden and co - workers replaced eight phenylalanine (Phe) residues

with 4 - F - Phe [53] . One - dimensional (1D)

F resonance assignments were based on the

Figure 16.3 The structure of apo - IFABP (PDB code: 1IFB). Eight Phe residues (shown in

stick representation), the D

– E and I – J regions, and location of G121 are indicated by labels.

The structure was generated using MacPyMOL (DeLano Scientiﬁ c LLC, Palo Alto, CA, U.S.A.).

See color plate 16.3.

420 Fluorine in Medicinal Chemistry and Chemical Biology

chemical shift of individual proteins containing one or two 4 - F - Phe residues. The ligand -

bound (oleic acid) and ligand - free forms were then analyzed by 2D

F –

F NOE (nuclear

Overhauser effect) spectra and linewidth measurements. Upon ligand binding, the NOE

spectrum revealed more exchange cross - peaks, and the 1D

F spectrum indicated that

most of the peaks were broadened. These results suggested that aromatic side - chains in

the binding cavity surprisingly become more ﬂ exible upon ligand binding [54] , even

though the backbone is more rigid in the ligand - bound form [55] .

Fluorine labels have also been used to explore the folding process of IFABP. Protein

folding involves interactions among side - chains and backbone hydrogen bonds [56] . One

challenge in elucidating this process is the characterization of “ intermediates ” between

folded and unfolded states. The “ acid state ” of IFABP may well resemble such an inter-

mediate – also referred to as a “ molten globule. ” Pulsed - ﬁ eld gradient (PFG) NMR indi-

cated that apparent hydrodynamic radii of the acid state IFABPs were larger than that of

the native state, but smaller than that of the fully denatured state. One - dimensional

spectra showed that IFABP was structured at pH 2.8, whereas it was mostly unfolded at

pH 2.3. It is of note that even at pH 2.3, there was still a small portion of the protein in

its native - like structure. More importantly, at pHs lower than 4.8, signiﬁ cant changes in

chemical shift and linewidths were observed for Phe128, 17, 68, and 93, suggesting that

the D – E turn and I – J regions are involved in the early stages of unfolding. The overall

structure of the hydrophobic core remained intact at pH 2.8, as indicated by

F –

F NOE

between Phe68 and Phe93 as well as circular dichroism (CD) measurements. Collectively,

changes in side - chain orientations occur before structural changes in the backbone upon

lowering the pH, even though these side - chains are buried in the hydrophobic core. Upon

oleic acid binding, the spectrum at pH 2.3 resembles the spectrum at higher pH, suggesting

that the ligand binding may shift the unfolded protein to a native - like structure [57] . The

folding kinetics of IFABP were also monitored by

F NMR on a slow - folding mutant in

which Val replaced Gly121 [58] . This single - site replacement is postulated to disrupt a

normal nucleation site in the I – J region. The G121V mutant folds more slowly, is less

stable, and exhibits the same structure and dynamics as wild type, allowing for a closer

examination of its folding behavior by stop - ﬂ ow

F NMR and CD. In the process of

refolding, the secondary structures formed twice as quickly as the stabilized conformation

of side - chains. A local, nonnative - like structure involving Phe62, Phe68, and Phe93

appeared within milliseconds, followed by arrangement of Phe2 and Phe17, and ﬁ nally

Phe47 into their conformations. This is followed by an overall rearrangement. Without

NMR, it would have been a serious challenge to shed light on both side - chain dynamics

and intermediate states during folding of IFABP [58] .

Mehl and co - workers have incorporated 4 - CF

- Phe into proteins to elucidate struc-

tural changes in nitroreductase upon addition of the cofactor ﬂ avin mononucleotide. When

4 - CF

- Phe was introduced in the active site (Phe124), the

F signal shifted upon ligand

addition. If it was introduced at a distant site (Phe36), the change in chemical shift was

quite subtle. In addition, substitution of 4 - CF

- Phe did not alter the activity of the enzyme

[59, 60] . These results suggest that 4 - CF

- Phe could be a sensitive probe for probing ligand

binding to proteins.

Membrane proteins are good pharmaceutical targets, since they are displayed on the

cell surface where drugs can bind without permeating lipid bilayers. Furthermore, they are

involved in cell - signaling interactions, viral entry, and a host of other processes important