Document [original]

Deciphering the Fluorine CodeThe Many Hats Fluorine Wears in a

Protein Environment

Published as part of the Accounts of Chemical Research special issue “Chemical Biology of Peptides”.

Allison Ann Berger,

†

Jan-Stefan Voller,

‡

Nediljko Budisa,

‡

and Beate Koksch*

,†

†

Institute of Chemistry and Biochemistry −Organic Chemistry, Freie Universita

t Berlin, Takustrasse 3, 14195 Berlin, Germany

‡

Institute of Chemistry, Technische Universita

t Berlin, Muller-Breslau-Str. 10, 10623 Berlin Germany

CONSPECTUS: Deciphering the ﬂuorine code is how we

describe not only the focus of this Account, but also the

systematic approach to studying the impact of ﬂuorine’s

incorporation on the properties of peptides and proteins used

by our groups and others. The introduction of ﬂuorine has

been shown to impart favorable, but seldom predictable,

properties to peptides and proteins, but up until about two

decades ago the outcomes of ﬂuorine modiﬁcation of peptides

and proteins were largely left to chance. Driven by the

motivation to extend the application of the unique properties

of the element ﬂuorine from medicinal and agro chemistry to

peptide and protein engineering we have established extensive

research programs that enable the systematic investigation of

eﬀects that accompany the introduction of ﬂuorine into this class of biopolymers. The introduction of ﬂuorine into amino acids

oﬀers a universe of options for modiﬁcations with regard to number and position of ﬂuorine substituents in the amino acid side

chain. Moreover, it is important to emphasize that the consequences of incorporating the C−F bond into a biopolymer can be

attributed to two distinct yet related phenomena: (i) the ﬂuorine substituent can directly engage in intermolecular interactions

with its environment and/or (ii) the other functional groups present in the molecule can be inﬂuenced by the electron

withdrawing nature of this element (intramolecular) and in turn interact diﬀerently with their immediate environment

(intermolecular). Based on our studies, we have shown that a change in number and/or position of as subtle as one single

ﬂuorine substituent has the power to considerably modify key properties of amino acids such as hydrophobicity, polarity, and

secondary structure propensity. These properties are crucial factors in peptide and protein engineering, and thus, ﬂuorinated

amino acids can be applied to ﬁne-tune properties such as protein folding, proteolytic stability, and protein−protein interactions

provided we understand and become able to predict the outcome of a ﬂuorine substitution in this context. With this Account, we

attempt to analyze information we gained from our recent projects on how the nature of the ﬂuorine atom and C−F bond

inﬂuence four key properties of peptides and proteins: peptide folding, protein−protein interactions, ribosomal translation, and

protease stability. These results impressively show why the introduction of ﬂuorine creates a new class of amino acids with a

repertoire of functionalities that is unique to the world of proteins and in some cases orthogonal to the set of canonical and

natural amino acids. Our concluding statements aim to oﬀer a few conserved design principles that have emerged from systematic

studies over the last two decades; in this way, we hope to advance the ﬁeld of peptide and protein engineering based on the

judicious introduction of ﬂuorinated building blocks.

■INTRODUCTION

Of all the halogens, ﬂuorine has the greatest abundance in the

earth’scrust,andeventhoughinorganicﬂuorine canalso befound

in signiﬁcant concentrations in marine and terrestrial organisms,

no organoﬂuorine compounds have ever been isolated from

within the animal kingdom. From a few tropical plants and

actinobacteria have been isolated ﬂuoroacetate, ﬂuoro fatty acids,

ﬂuoroacetone, ﬂuorocitrate, a ﬂuorinated nucleoside component

of the antibiotic nucleosidin, and ﬂuorothreonine.

As reviewed

by Gouverneur et al. in 2008

and by Liu et al. in 2016,

ﬂuorine

has proven to be highly beneﬁcial in the pharmaceutical and

agrochemical industries, with over 150 ﬂuorinated small

molecules reaching the market since the 1950s and an increase,

since 2010, from 20% to 30% of administered drugs containing

ﬂuorine atoms or ﬂuoroalkyl groups. In addition, ﬂuorine-18

and

ﬂuorine-19

are extraordinarily useful tools in medical imaging by

means of positron emission tomography and magnetic resonance

imaging, respectively.

In light of ﬂuorine’s impact on medicinal chemistry and drug

development, it became obvious to introduce this unique element

into biopolymers with the intention of improving or tuning their

Received: May 5, 2017

Published: August 12, 2017

Article

pubs.acs.org/accounts

Acc. Chem. Res. 2017, 50, 2093−2103

This is an open access article published under an ACS AuthorChoice License, which permits

copying and redistribution of the article or any adaptations for non-commercial purposes.

Downloaded via TU BERLIN on December 27, 2021 at 11:28:55 (UTC).

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

properties and facilitating structural studies.

Meanwhile, ﬂuorine

has been studied in the context of nucleic acids,

carbohydrates,

lipids,

and proteins, and the latter is the focus of this Account.

Fluorine has been shown to impart often favorable but seldom

predictableproperties topeptides and proteins, andtheoutcomes

of such studies have been reviewed comprehensively else-

where.

10,11

Recent reviews have also been written about the

synthesis of ﬂuorinated building blocks,

and strategies for their

chemical or ribosomal incorporation.

Here we attempt to

summarizehowthenatureoftheﬂuorineatomandtheC−Fbond

inﬂuence four key properties of peptides and proteins.

■WHAT WE HAVE LEARNED OVER THE LAST TWO

DECADES ABOUT FLUORINATED ANALOGUES OF

NONPOLAR ALIPHATIC AMINO ACIDS AND

PROLINE

In the early stages of the research programs of the Koksch and

Budisa groups about 20 years ago, we were struck by how the

introduction of ﬂuorine into peptides and proteins and its impact

on various properties of the resulting unnatural biopolymers

could lead to such unpredictable outcomes. At that point in time,

this unique element was far from being a suitable tool for rational

design approaches. Thus, each of our working groups aims at

understanding the molecular interactions that underly the impact

of ﬂuorination in the context of peptide and protein environ-

ments.

Table 1. Structures, Names, and “Three Letter Codes”for All Amino Acids Referred to in This Account

IUPAC names and common synonyms. Abu: (2S)-2-aminobutanoic acid, homoalanine; MfeGly: (2S)-2-amino-4-monoﬂuorobutanoic acid,

monoﬂuoroethylglycine; DfeGly: (2S)-2-amino-4,4-diﬂuorobutanoic acid, diﬂuoroethylglycine; TfeGly: (2S)-2-amino-4,4,4-triﬂuorobutanoic acid,

triﬂuoroethylglycine. Val: (2S)-2-amino-3-methylbutanoic acid, L-valine; (2S,3S)-TfVal: (2S,3S)-2-amino-4,4,4-triﬂuorobutanoic acid, (2S,3S)-

triﬂuorovaline, TfV (diastereomeric mixtures); (2S,3R)-TfVal: (2S,3R)-2-amino-4,4,4-triﬂuorobutanoic acid, (2S,3R)-triﬂuorovaline, TfV

(diastereomeric mixtures); HfVal: (2S)-2-amino-3-(triﬂuoromethyl)-4,4,4-triﬂuorobutanoic acid, (2S)-4,4,4,4′,4′,4′-hexaﬂuorovaline, 43,4′3-F6Val,

HfV; Ile: (2S,3S)-2-amino-3-methylpentanoic acid, L-isoleucine; DfpGly: (2S)-2-amino-4,4-diﬂuoropentanoic acid, diﬂuoropropylglycine; (2S,3S)-4′-

TfIle: (2S,3S)-2-amino-3-(triﬂuoromethyl)pentanoic acid, 4′3-F3Ile; (2S,3S)-5-TfIle: (2S,3S)-5,5,5-triﬂuoroisoleucine, 53-F3Ile; Leu: (2S)-2-amino-4-

methylpentanoic acid, L-leucine; (2S,4S)-TfLeu: (2S,4S)-2-amino-5,5,5-triﬂuoro-4-methylpentanoic acid, (4S)-triﬂuoroleucine, (4S)-53-F3Leu, TfL

(diastereomeric mixtures); (2S,4R)-TfLeu: (2S,4R)-2-amino-5,5,5-triﬂuoro-4-methylpentanoic acid, (4R)-triﬂuoroleucine, (4R)-53-F3Leu, TfL

(diastereomeric mixtures); HfLeu: (2S)-2-amino-5,5,5-triﬂuoro-4-triﬂuoromethyl pentanoic acid, (2S)-5,5,5,5′,5′,5′-hexaﬂuoroleucine, 53,5′3-F6Leu,

HfL; Pro: (2S)-pyrrolidine-2-carboxylic acid, L-proline; (4S)-FPro: (2S,4S)-4-ﬂuoropyrrolidine-2-carboxylic acid, 4-S-Flp; (4R)-FPro: (2S,4R)-4-

ﬂuoropyrrolidine-2-carboxylic acid, 4-R-Flp

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2094

The Koksch laboratory has systematically investigated

ﬂuorinated analogues of Abu (Table 1) in the context of peptide

model systems that are designed to reconstitute natural protein

environments. We have made the following observations: (1)

small numbers of ﬂuorine atoms introduce polar character into

otherwise hydrophobic amino acid side chains, representing a

departure from earlier studies that had demonstrated the

hydrophobicity/lipophilicity of heavily ﬂuorinated side chains;

(2) ﬂuorine dramatically changes the secondary structure

propensity of aliphatic amino acids and, thus, the folding

properties of accordingly modiﬁed peptides and proteins; (3)

ﬂuorine can considerably inﬂuence the proteolytic stability of

peptides; and (4) the impact of ﬂuorine not only depends on the

nature of the ﬂuorinated side chain but also on the immediate

environment with which it interacts.

The Budisa group has focused on reprogramming protein

translation in living cells for the eﬃcient use of ﬂuorinated amino

acids as building blocks for the biological expression of proteins.

Our particular aim is to understand the chemical reactivity and

stereochemistry of the ﬂuorinated building block in translation

and its impact on the kinetics of protein folding.

Most recently,

we have used ﬂuorinated Pro (Table 1) as an excellent molecular

model for this purpose, and it enabled us to make the following

observations: (1) ﬂuorinated Pro analogues dramatically increase

biological expression of mussel proteins;

(2) the folding rate of

globular proteins is aﬀected not only by cis/trans isomerization

but also the puckering of the proline ring;

(3) speciﬁcﬂuorine-

tyrosine polar contacts can increase the stability of protein self-

assembly;

and (4) there is a chiral bias with respect to

ﬂuorinated substrates for ribosomal synthesis, resulting in

dramatic diﬀerences in the rates of peptide bond formation.

We will show that the introduction of ﬂuorine creates a

fascinating class of amino acids with properties that are unique in

the protein world.

■THE NATURE OF THE C−F BOND AND HOW IT

IMPACTS SOME PROPERTIES OF AMINO ACIDS

Fluorine isthe mostelectronegative element, which means ithas a

large inductive eﬀect and that the C−F bond is polarized with

ionic character and a large dipole moment, facilitating electro-

static/dipolar interactions with its immediate environment. The

three lone pairs of ﬂuorine are tightly held, imparting the C−F

bond with low polarizability, for example, making it a poor H-

bond acceptor. The C−F bond possesses a low lying σ*C−F

antibonding orbital as a consequence of the great extent of bond

polarization; stereoelectronically aligned electron-rich bonds like

C−H can thus donate electron density into this orbital to stabilize

certain conformations via hyperconjugative interactions.

How do the physicochemical properties of the C−F bond

impact the measurable properties of amino acids, properties that

in turn inﬂuence the characteristics of the peptides and proteins

that contain them?

Hydrophobicity

From studies on the Abu, Val, Ile, and Leu scaﬀolds, we have

learned that the volume and hydrophobicity of the side chain is

inﬂuenced signiﬁcantly by ﬂuorination (Figure 1). In particular,

MfeGly, DfeGly, and DfpGly are markedly polar due to the

presence of highly polarized geminal and vicinal C−H bonds, as

well as a strong ﬂuorine-induced dipole moment within the side

chain. Consider DfpGly, for which the van der Waals volume of

the side chain is close to that of Leu but its hydrophobicity is

lowered to a level slightly below that of Val.

Secondary Structure Propensity

The intrinsic propensity of an amino acid for a certain secondary

structure is an important property that must be considered in

protein engineering.

Inspired by initial studies from the Cheng

group

that we have expanded upon over time, we have learned

that the number of ﬂuorine atoms within an aliphatic side chain

variesindirectlywiththepropensityforthepeptidecontainingthe

given building block to adopt an α-helical secondary structure

(Table 2). Only (2S,3S)-5-TfIle, with the highest helix propensity

of all the triﬂuoroalkyl-bearing amino acids, bucks this trend.

Ring Puckering

Pro is unique among the 20 proteinogenic amino acids in that it

possesses a pyrrolidine ring that spans the α-carbon (Cα) and

nitrogen of the backbone and restricts the conformational space

to distinct cis or trans states, for which there are characteristic

torsion angles between Pro and the preceding residue. More than

90% of Pro in protein structures adopt the trans conformation,

and the cis/trans conversion, requiring a 180°rotation about the

peptide bond, is associated with a relatively high energetic barrier

(∼20 kcal/mol).

Wennemers and co-workers have extensively

studied the conformational eﬀects of avariety of ring substituents,

Figure 1. Relationship between van der Waals volume of the side chain

and retention time in a reversed-phase HPLC assay. 2-Aha: (2S)-2-

aminoheptanoic acid. The black oval highlights the ﬂuorine-induced

increase in polarity of DfpGly.

Table 2. Helix Propensities for Selected Natural and

Fluorinated Amino Acids

amino acid helix propensity and literature reference

Ala 1.46 ±0.01

Abu 1.22 ±0.14

Leu 0.994 ±0.093

MfeGly 0.873 ±0.068

Ile 0.53 ±0.05

DfeGly 0.497 ±0.060

Val 0.41 ±0.04

(2S,3S)-5-TfIle 0.26 ±0.03

HfLeu 0.128 ±0.023

TfeGly 0.057 ±0.022

(2S,3S)-TfVal 0

(2S,3R)-TfVal 0

(2S,3S)-4′-TfIle 0

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2095

including aminoproline, which provides a pH-triggered switch.

The introduction of ﬂuorine at C-4 can be summarized as follows:

(4R)-FPro strongly prefers the Cγ-exo ring pucker due to the

gauche eﬀect; the presence of ﬂuorine reduces the energy barrier

to isomerization by aﬀecting the pyramidalization of the

pyrrolidine nitrogen; thus, (4R)-FPro stabilizes the trans acyl-

FPro conformation, while (4S)-FPro favors the cis conforma-

tion

(Figure 2). This demonstrates how targeted ﬂuorination of

the Pro scaﬀold contributes to the elucidation of its roles in

chemical or biological systems and oﬀers new avenues for peptide

and protein engineering.

■PEPTIDE/PROTEIN FOLDING

α-Helical Coiled Coils

Coiled coils typically consist of two to seven right-handed α-

helices that are wound around each other forming a left-handed

superhelix. The primary structure is amphiphilic and can be

characterized by a so-called heptad repeat (abcdefg)n. Positions a

and dare typically occupied by apolar residues (Leu, Ile, Val and

Met) that form a hydrophobic core at the interface of the helices.

By contrast, the positions eand gare frequently occupied by

charged amino acids (most commonly Glu, Lys and Arg) that

form interhelical electrostatic interactions. The remaining heptad

repeat positions b,c, and fare exposed to the solvent and can in

principle be occupied by any hydrophilic residue. Several prior

studies had demonstrated that global substitution of hydrophobic

core with highly ﬂuorinated analogues of hydrophobic amino

acids leads to increased thermal stability.

In contrast, work from

our group revealed that single substitutions of minimally

ﬂuorinated building blocks within the hydrophobic core can

lead to thermal destabilization. The degree of destabilization

depends on how eﬃciently the ﬂuorinated residue packs against

neighboring side chains.

29−33

Of particular interest was the

ﬁnding that the polarized β-methylene group of DfpGly

inﬂuences the thermal stability of coiled coil dimers in a

position-dependent way (Figure 3).

To address the question of whether single substitutions with

highly ﬂuorinated side chains, thus borrowing from both extreme

regimes described above, can restore or enhance the thermal

stabilityof suchstructures,we studiedthe substitutionof Leu with

HfLeu,and ofVal withTfVal inthe heterodimericparallel peptide

model system referred to as VPE/VPK (Figure 4).

Whereas a

higher thermal stability is observed for VPK containing HfLeu at

position d19 (74.4 °C) in complex with VPE, compared to the

parent helix bundle (70.7 °C), the melting points of VPK with

(2S,3R)-TfVal or (2S,3S)-TfVal at position a16 were similar or

lower, 70.8 and 67.5 °C, respectively. In this model system the

central hydrophobic positions a′16 (Val), d′19 (Leu), and a′23

(Val) of VPE directly interact with the substituted position d19 of

VPK. Therefore, the enhancement in the thermal stability for the

HfLeu containing analogue is likely a result of greater hydro-

phobicity and eﬃcient packing with the hydrocarbon side chains

in the hydrophobic core, as had been previously demonstrated by

our group by means of phage display experiments (Figure 5).

On the other hand, position a16 has d′12 (Leu), a′16 (Val), and d′19

(Leu) as its nearest neighbors and the shorter side chain of TfVal

appears to contribute less to hydrophobic core formation.

Section Summary. The combination of hydrophobicity,

polarization of neighboring groups, and the properties of the

immediate environment determines the outcome of ﬂuorine

substitution within a hydrophobic protein environment.

β-Sheets

In collaboration with Czekelius

23,24

(2S,3S)-5-TfIle, (2S,3S)-

TfVal, (2S,3R)-TfVal, and (2S,3S)-4′-TfIle were synthesized in

their enantiomerically pure forms, incorporated into peptides,

Figure 2. Schematic representation of the eﬀects of ﬂuorinating Pro at

position 4 of its pyrrolidine ring. Complete discrimination between

electronicand stericeﬀectsisimpossible;thus,theterm“stereoelectronic

eﬀect”

is used.

Figure 3. Impact of ﬂuorine depends on its immediate packing

environment. Adapted from ref 30 with permission from John Wiley

and Sons.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2096

and found to show extremely low α-helix propensities in all cases

except (2S,3S)-5-TfIle (Table 2). Obviously, ﬂuorination of

isoleucine’s methyl group at the β-branching point abolishes α-

helix propensity, while ﬂuorination of the δ-position rescues α-

helix propensity to some extent.

Wecarriedoutasystematicstudyontheeﬀectofﬂuorineonthe

rate of amyloid formation that took size, hydrophobicity, ﬂuorine

content, and secondary structure propensity of the building

blocks into consideration.

We exploited a de novo designed

model peptide (VW18) that adopts an α-helical coiled coil

conformation upon dissolution in aqueous media and undergoes

Figure 4. Parallel heterodimeric model system VPE/VPK as (A) helical wheel representation indicating Val16 and Leu19

and (B) ribbon and stick

model.

Reproduced from refs 34 and 36 with permission from Elsevier and John Wiley and Sons, respectively.

Figure 5. Cartoon showing packing interactions at positions (A) a16 and (B) d19 of VPE/VPK within the context of the phage display experiment.

Figure 6. Eﬀect of single substitution with ﬂuorinated amino acids on the structural transition of VW18.

(A) internal ﬁbril architecture and (B, C) ThT

ﬂuorescence at 485 nm versus time (h). Adapted from ref 22 with permission from The Royal Society of Chemistry.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2097

a time-dependent spontaneous structural transition into β-sheet

rich amyloid ﬁbrils.

Three judiciously placed valine residues,

one at each an f(Val3), a b(Val13), and a c(Val14) solvent-

exposed position of the α-helical coiled coil are responsible for

this structural transition. Val13 was systematically replaced by

MfeGly, DfeGly, TfeGly, and Leu, and Val14 was substituted by

these residues in addition to (2S,3S)-TfVal and (2S,3R)-TfVal.

The rate of amyloid formation, as determined in a standard ThT

ﬂuorescence assay, varies directly with the number of ﬂuorine

atoms at both positions 13 and 14 (Figure 6B and C).

Remarkably, both the Leu and MfeGly variants exhibit reduced

amyloid formation rates compared to the more highly ﬂuorinated

species.

Section Summary. Secondary structure propensity is an

important factor in amyloid formation kinetics, as is hydro-

phobicity. This observable is likely determined by the extent to

which the initial helical coiled coil structure of the peptide is

stabilized (Leu and MfeGly) or destabilized (TfeGly and TfVal).

Finally, DfeGly exhibits the greatest position dependence, which

may reﬂect the diﬀerence in polarity of the ﬁbril environments of

positions 13 and 14 (Figures 6A and 7).

(4S)-FPro and (4R)-FPro in Enhanced Green Fluorescent

Protein

Large globular protein folding requires escape from misfolded

states caused by proline isomerization. The high barrier for

isomerization results in characteristic refolding times of 10−1000

s at room temperature. For that reason, we were motivated to

study and understand how Pro ring ﬂuorination would inﬂuence

these dynamics. To this end we solved the high-resolution crystal

structure of enhanced green ﬂuorescent protein (EGFP) with 10

Pro-residues globally replaced by (4S)-FPro or (4R)-FPro

(Figure 8A).

Remarkably, EGFP containing (4R)-FPro was

not expressed to a detectable level, whereas global substitution

with (4S)-FPro yielded a soluble protein with faster refolding

kinetics (Figure 8B).

Section Summary. Analysis of the structures revealed the

following:(i)the majorityof Proresidues inEGFP areinvolved in

trans peptide bonds but anotable exception is Pro89 (Figure8D);

(ii) all Pros display Cγ-endo puckered pyrrolidine rings apart

from Pro56, which adopts a Cγ-exo conﬁguration; (iii) the

ﬂuorine atom in (4S)-FPro promotes endo puckering and thus

preorganizes the pyrrolidine rings of 9 out of the 10 Pros into an

optimal spatial arrangement, which is likely also responsible for

the improved refolding properties. All of these eﬀects result in 12

new ﬂuorine-induced stabilizing interactions. Doubtless, the

puckering of the pyrrolidine rings in the structure, which is

dramatically inﬂuenced by ﬂuorination, is the basic driving force

behind the observed phenomena, also recently observed by

Raines and co-workers in ribonuclease A.

■PROTEIN−PROTEIN INTERACTIONS

Inaneﬀort toaddress thequestionof ﬂuorine’seﬀectonprotein−

protein interactions in an otherwise well-characterized natural

system, we substituted Lys15 of the bovine pancreatic trypsin

inhibitor (BPTI) (Figure 9A), the residue that is part of the

“interaction loop”and binds at the catalytic subsite S1 of the

digestive serine protease trypsin (Figure 9B), with Abu, DfeGly,

or TfeGly by solid-phase peptide synthesis and native chemical

ligation.

Thermal denaturation studies under acidic conditions

in the presence of either 6 M guanidinium chloride (GdmCl) or 8

M urea revealed that the Lys15Abu variant is signiﬁcantly less

stable than the native or ﬂuorinated analogues (Figure 9C).

Remarkably, the ﬂuorinated side chains have a stabilizing eﬀect

that is greater even than that of the wild type in the presence of

GdmCl,buttheoppositeistrueinthepresenceofurea;thispoints

to the fact that the positive formal charge of the solvent-exposed

Lys contributes to the thermal stability of the natural inhibitor.

Thus, DfeGly and TfeGly do not behave like canonical

hydrophobic groups, which previous studies had found to be

destabilizing at position 15.

Rather, the introduction of ﬂuorine

into the gamma methyl group facilitates electrostatic interactions

with an aqueous environment due to the highly polarized nature

of the C−F bond.

BPTI strongly inhibits trypsin by forming numerous contacts

to the S1 subsite of the enzyme, and a water-mediated H-bond

between the ammonium group of BPTI-Lys15 and Asp189 of β-

trypsinplaysan especiallyimportantrole. Weconductedstandard

inhibition assays with bovine β-trypsin

and found that

substitution of Lys with Abu at P1 dramatically reduces inhibitor

activity, whereas replacement with either DfeGly or TfeGly

results in inhibitors that are as active as wild-type BPTI.

We determined high-resolution crystal structures of all BPTI

variants in complex with bovine β-trypsin (note that we refer to

the designated Protein Data Bank atom identiﬁer labels in the

following ﬁgures and text, for example, 3EGFAC describes one of

the gamma ﬂuorine atoms (FAC) of residue TfeGly (3EG)). All

complexes were superimposable, except at the P1 side chain and

the water molecules within the S1 pocket. Analysis of the data

revealedevidencefor a“ﬂuorophilicenvironment”forDfeGlyand

TfeGly within the S1 subsite of trypsin (Figure 10) based on

multipolar interactions, as previously described by Diederich and

co-workers for ﬂuorinated small molecule drugs in complexes

with enzymes.

Interestingly, structural water molecule C is closer to ﬂuorine

atom 3EGFAC in the TfeGly structure (3.4 Å) than it is to

hydrogen atoms ABAHG3 in the Abu structure (3.6 Å) or OBFHG

intheDfeGly structure(3.7 Å)(Figure11). Althoughit cannotbe

ruled out that this is a consequence of the slight shift in

conformation within the side chain that occurs due to the

electrostatic repulsion between 3EGFAC and 3EGN, it may also

indicate that a weak OH···FC H-bond exists between the TfeGly

side chain and water C. Further investigation of this hypothesis

Figure 7. Schematic of the relationship between helix propensity, size/

hydrophobicity, and rates of amyloid formation for VW18 with

ﬂuorinated aliphatic amino acids at position 13 or 14.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2098

would require neutron diﬀraction studies.

B-factor analysis for

“new”waters B and C in the various structures provided further

evidence to support this conclusion, namely that these values are

signiﬁcantly lower in the DfeGly and TfeGly complexes

compared to Abu.

Section Summary

In contrast to nonpolar hydrocarbon amino acid side chains,

ﬂuorinated aliphatic groups can participate in multipolar and

water-mediated interactions within the binding pockets of

proteins and enzymes.

■RIBOSOMAL TRANSLATION

Protein translation machinery naturally discriminates between

unnatural/noncanonical amino acids and canonical building

blocks via three key quality control mechanisms: (i) amino-

acylation and editing of tRNAs by aminoacyl-tRNA synthetases

Figure 8. Role of ﬂuorine-modulated ring puckering in the refolding kinetics of EGFP. (A) EGFP structure (PDB ID: 2Q6P) indicating Pro residues. (B)

Refolding features of EGFP containing all L-Pros or all (4S)-FPros. Structural analysis enabled the mapping of novel ﬂuorine interactions, exempliﬁed by

residues (C) 211 and (D) 89.

Figure 9. (A) BPTI protein (PDB IDs: 3OTJ

and 2FTL

). (B) BPTI-trypsin complex with inhibitor in orange and enzyme in blue, black sphere

represents position of Lys15. (C) Melting temperatures of BPTI variants.

Figure 10. Distance analysis andtopview of theunnatural side chain atposition 15 of BPTI(mainchain orange) within theS1 binding pocket ofβ-trypsin

(main chain blue):

(a) Abu; (b) DfeGly; and (c) TfeGly. Reproduced from ref 39 with permission from The Royal Society of Chemistry.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2099

(aaRS), (ii) formation of ternary complexes with EF-Tu-GTP,

and (iii) peptide-bond formation mediated by the ribosome.

These mechanisms evolved in the presence of the natural

canonical amino acid pool; however, because virtually all

ﬂuorinated amino acids are of chemical synthetic origin and

were consequently absent during the evolution of these

proofreading mechanisms, it has generally been assumed that

translational quality control is not eﬀective against these

substrates. Whereas aminoacylation studies with ﬂuorinated

Figure 11. Water-mediated H-bond network around the unnatural side chain at position 15 of BPTI within the S1 binding pocket of β-trypsin.

Reproduced from ref 39 with permission from The Royal Society of Chemistry.

Figure 13. Ribosomal peptide bond formation with ﬂuorinated Pro analogs reveals a chiral bias

toward (4R)-FPro. Note that ring pucker (directly

inﬂuenced by ﬂuorination, as shown in Figure 2) determines the rate of peptide bond formation.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2100

amino acids are decades old,

there has been a dearth of detailed

mechanistic studies regarding their editing.

TfeGly

We reported recently that the editing domain of isoleucyl-tRNA

synthetase (IleRS) is unable to discriminate between tRNAIle

charged with the cognate substrate Ile or the ﬂuorinated amino

acid TfeGly with respect to the hydrolysis rate (Figure 12).

Surprisingly, the competing reaction, the dissociation of TfeGly-

tRNAIle from the enzyme into solution for subsequent translation

via the ribosome is also exceptionally slow, which had never been

observedwithaminoacidsthatdonotcontainﬂuorine.Asaresult,

hydrolysis outcompetes dissociation and the bioincorporation of

TfeGly is not supported by the wild-type translation system.

However, the ﬁrst reported ribosomal translation of TfeGly was

achieved by mutation of the enzyme’s editing domain.

Section Summary. It is important to note that the kinetics of

the individual processes comprising editing against TfeGly diﬀer

signiﬁcantly from those found in editing against nonﬂuorinated

substrates, which are characterized by much faster reaction rates

(Figure 12). Obviously, the CF3group can engage in speciﬁc

interactions with the synthetase.

FPro

Another factor that can potentially limit the production of

ﬂuorinated polypeptides is the rate of peptide bond formation

within the ribosome. For example, peptide bond formation with

Pro is exceptionally slow compared to all other proteinogenic

aminoacids,leadingtoribosomestallingwhenapeptidesequence

contains multiple such residues. A recent report showed that the

introduction of a single ﬂuorine atom into Pro to yield (4R)-FPro

abolishes ribosome stalling by elevating the peptide bond

formation rate to a level seen in the other proteinogenic amino

acids (Figure 13).

Interestingly, its diastereomer (4S)-FPro

produced the opposite eﬀect.

Section Summary. Fluorine-induced eﬀects on the stereo-

electronics of the Pro pyrrolidine ring are fully reﬂected in the

reaction rates that characterize ribosome-mediated peptide bond

formation.

■PROTEASE STABILITY

Peptide and protein-based therapeutics are characterized by their

excellentselectivity comparedtosmallmolecules,butalsobytheir

low metabolic stability. Numerous reports have demonstrated

thatﬂuorinatedbuildingblockscanimprovethisproperty,though

this is not always the case.

46,47

To systematically investigate the inﬂuence of ﬂuorinated

analogues of Abu on stability toward protease degradation, the

peptide sequence Abz-KAAFAAAAK (FA), was designed as a

substrate with a central Phe residue to satisfy the known substrate

speciﬁcities of serine and aspartic endopeptidases. The central

portion of the FA sequence, -A-F-A-A-, can be written -P2-P1-

P1′-P2′-. DfeGly, TfeGly, and Abu were substituted for alanine at

positions P2, P1′, and P2′, and the variant sequences named

accordingly. All peptide substrates were subjected to degradation

bytreatmentwithhuman bloodplasma,elastase, α-chymotrypsin,

and pepsin.

48,49

In the context of the blood plasma component

elastase, TfeGly substitution immediately adjacent to the central

Phe results in protease protection. For the digestive enzymes α-

chymotrypsin and pepsin, the following ﬂuorination patterns

result in dramatic reductions in turnover: substitution at P2′with

either DfeGly or TfeGly in the pepsin substrate, and the presence

of DfeGly at P2 in the α-chymotrypsin substrate (Figure 14).

Section Summary

In about 25% of all cases studied here, peptide substrates were

protected due to the incorporation of a ﬂuorinated amino acid.

■GENERALIZATIONS THAT CAN OR CANNOT BE

MADE, FUTURE DIRECTIONS, AND POTENTIAL

APPLICATIONS

In a league of its own, ﬂuorine enables protein engineering to

achieve highly desirable outcomes, and we have come a long way

inbeingabletopredictthem.Forexample,MfeGlyhasthehighest

helix propensity of the ﬂuorinated amino acids described so far;

thus, we feel conﬁdent in stating that it would support helical

folding in a native protein environment. In contrast, adding one

moreﬂuorine substituent (DfeGly) may insteadfacilitate amyloid

formation. Furthermore, the stereochemistry of a single ﬂuorine

substituent in a single Pro ring can dictate whether an unnatural

protein product is expressed.

As has been known within the ﬁeld of inorganic ﬂuorine

chemistry for many decades, where ﬂuorine’s reactivity and

toxicity are major challenges, this element is a “diva”that must be

handled with care. For the peptide community, the challenge in

Figure 14. Digestion of peptide substrates by α-chymotrypsin or pepsin.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2101

using ﬂuorine as a tool lies in our ability to juggle the interplay

between the speciﬁc properties of the ﬂuorinated building block

and its responsiveness to the environment it is exposed to. It is

important to keep in mind that here are numerous entirely

untapped aspects of the C−F bond that are waiting to be explored

in a protein environment, including charge−dipole interactions

and metal cation coordination.

Finally, from our vantage point, the next hurdle will be

determining the way in which not just biomolecules in the

laboratory, but whole living organisms accommodate ﬂuorine.

O’Hagan

andco-workers’groundbreakingpublicationaboutthe

ﬂuorinase enzyme provided a major impulse for the detailed

investigation of the eﬀects of ﬂuorine’s incorporation into

peptides and proteins. At this juncture, almost two decades

later,thepeptidecommunityisreadytousherinanewerathatwill

focus on the uptake, toxicity, and metabolism of organoﬂuorine

building blocks and that will be inspired by thekinds of systematic

and interdisciplinary studies that we have reviewed here.

■AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected].

ORCID

Nediljko Budisa: 0000-0001-8437-7304

Beate Koksch: 0000-0002-9747-0740

Notes

The authors declare no competing ﬁnancial interest.

Biographies

Allison Ann Berger received a B.A. degree from Reed College and a

Ph.D. degree from TSRI La Jolla, both in Chemistry. After completing a

two-year postdoctoral stay at the Max-Planck-Institute for Infection

Biology as an Alexander von Humboldt fellow (2005−2007), she joined

the group of Beate Koksch as staﬀscientist.

Jan-Stefan Voller studied chemistry at TU Berlin and received his Ph.D.

degree in 2016 at Freie Universita

t Berlin in the group of Beate Koksch,

working on a collaborative project with the group of Nediljko Budisa at

TU Berlin. Currently, he is pursuing postdoctoral research in the Budisa

lab on the ribosomal translation of novel ﬂuorinated and other

noncanonical amino acids.

Nediljko Budisa has been Professor of Biocatalysis at TU Berlin since

2010. He received his Ph.D. degree in 1997 and has done pioneering

workingenetic-codeengineeringandmostrecentlyinchemicalsynthetic

biology (xenobiology). His research in the context of ﬂuorine

biochemistry focuses primarily on the development of in vivo methods

for introducing genetically encoded protein modiﬁcations in individual

proteins, complex protein structures, and whole proteomes.

Beate Koksch received a Ph.D. degree from University Leipzig and

pursued postdoctoral studies at TSRI La Jolla and postodoctoral lecture

qualiﬁcation at University Leipzig under Klaus Burger. She has been

Professor of Chemistry at Freie Universita

t Berlin since 2004. Her group

investigates ﬂuorinated amino acids in the context of peptides and

proteins, studies complex folding mechanisms in neurodegenerative

diseases and develops new multivalent scaﬀolds.

■ACKNOWLEDGMENTS

We thank the many members of our respective research groups

fortheiroutstandingcontributions totheworkreviewedhere,and

Susanne Huhmann for excellent assistance with the artwork. This

work was funded by Grant RTG-1582 from the Deutsche

Forschungsgemeinschaft.

■DEDICATION

Dedicated to the memory of Klaus Burger.

■REFERENCES

(1) Harper, D. B.; O’Hagan, D. The fluorinated natural products. Nat.

Prod. Rep. 1994,11, 123−133.

(2) Purser, S.; Moore, P. R.; Swallow, S.; Gouverneur, V. Fluorine in

medicinal chemistry. Chem. Soc. Rev. 2008,37, 320−330.

(3) Zhou, Y.; Wang, J.; Gu, Z.; Wang, S.; Zhu, W.; Acena, J. L.;

Soloshonok, V. A.; Izawa, K.; Liu, H. Next generation of fluorine-

containingpharmaceuticals,compounds currentlyinphaseII−IIIclinical

trials of major pharmaceutical companies: new structural trends and

therapeutic areas. Chem. Rev. 2016,116, 422−518.

(4) Brooks, A. F.; Topczewski, J. J.; Ichiishi, N.; Sanford, M. S.; Scott, P.

J. H. Late-stage [18F]fluorination: new solutions to old problems. Chem.

Sci. 2014,5, 4545−4553.

(5) Tirotta, I.; Dichiarante, V.; Pigliacelli, C.; Cavallo, G.; Terraneo, G.;

Bombelli, F. B.; Metrangolo, P.; Resnati, G. 19F magnetic resonance

imaging (MRI): from design of materials to clinical applications. Chem.

Rev. 2015,115, 1106−1129.

(6) Ojima, I., Ed. Fluorine in Medicinal Chemistry and Chemical Biology;

Wiley-Blackwell, UK, 2009.

(7) Egli, M. The Steric Hypothesis for DNA Replication and Fluorine

Hydrogen Bonding Revisited in Light of Structural Data. Acc. Chem. Res.

2012,45, 1237−1246.

(8) Oberbillig, T.; Mersch, C.; Wagner, S.; Hoffmann-Roder, A.

Antibody recognition of fluorinated MUC1 glycopeptide antigens.

Chem. Commun. 2012,48, 1487−1489.

(9) Bucher, C.; Gilmour, R. A Modular Synthesis of Fluorinated, Chiral

Polar Lipids. Synthesis 2011,2011, 549−552.

(10)Salwiczek,M.; Nyakatura,E.K.; Gerling,U.I. M.;Ye,S.; Koksch,B.

Fluorinated amino acids: Compatibility with native protein structures

and effects on protein-protein interactions. Chem. Soc. Rev. 2012,41,

2135−2171.

(11) Marsh, E. N. G. Chapter Twelve −Designing Fluorinated

Proteins. Methods Enzymol. 2016,580 (2016), 251−278.

(12) Smits, R.; Cadicamo, C. D.; Burger, K.; Koksch, B. Synthetic

strategies to alpha-trifluoromethyl and alpha-difluoromethyl substituted

alpha-amino acids. Chem. Soc. Rev. 2008,37, 1727−1739.

(13) Budisa, N.; Voller, J.-S.; Koksch, B.; Acevedo-Rocha, C. G.;

Kubyshkin, V.; Agostini, F. Biocatalysis with unnatural amino acids:

Enzymology meets Xenobiology. Angew. Chem., Int. Ed. 2017,

DOI: 10.1002/anie.201610129.

(14) Merkel, L.; Budisa, N. Organic fluorine as a polypeptide building

element: in vivo expression of fluorinated peptides, proteins and

proteomes. Org. Biomol. Chem. 2012,10, 7241−7261.

(15) Larregola, M.; Moore, S.; Budisa, N. Congeneric bio-adhesive

mussel foot proteins designed by modified prolines revealed a chiral bias

inunnaturaltranslation.Biochem. Biophys. Res. Commun. 2012,421,646−

50.

(16) Steiner, T.; Hess, P.; Bae, J. H.; Wiltschi, B.; Moroder, L.; Budisa,

N. Synthetic biology of proteins: tuning GFP’s folding and stability with

fluoroproline. PLoS One 2008,3, e1680.

(17) Dietz, D.; Kubyshkin, V.; Budisa, N. Applying γ-substituted

prolines in the foldon peptide: polarity contradicts preorganization.

ChemBioChem 2015,16, 403−406.

(18) Doerfel, L. K.; Wohlgemuth, I.; Kubyshkin, V.; Starosta, A. L.;

Wilson, D. N.; Budisa, N.; Rodnina, M. V. Entropic contribution of

elongation factor P to proline positioning at the catalytic center of the

ribosome. J. Am. Chem. Soc. 2015,137, 12997−13006.

(19) O’Hagan, D. Understanding organofluorine chemistry. An

introduction to the C−F bond. Chem. Soc. Rev. 2008,37, 308−319.

(20) Chou, P. Y.; Fasman, G. D. Conformational parameters for amino

acids in helical, β-sheet, and random coil regions calculated from

proteins. Biochemistry 1974,13, 211−222.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2102

(21) Chiu, H.-P.; Suzuki, Y.; Gullickson, D.; Ahmad, R.; Kokona, B.;

Fairman, R.; Cheng, R. P. Helix propensity of highly fluorinated amino

acids. J. Am. Chem. Soc. 2006,128, 15556−15557.

(22) Gerling, U. I. M.; Salwiczek, M.; Cadicamo, C. D.; Erdbrink, H.;

Grage, S.; Wadhwani, P.; Ulrich, A.; Behrends, M.; Haufe, G.; Czekelius,

C.; Koksch, B. Partial side chain fluorination influencing amyloid

formation: a symphony of size, hydrophobicity, and α-helix propensity.

Chem. Sci. 2014,5, 819−830.

(23) Erdbrink, H.; Peuser, I.; Gerling, U. I. M.; Lentz, D.; Koksch, B.;

Czekelius, C. Conjugate hydrotrifluoromethylation of α,β-unsaturated

acyl-oxazolidinones: synthesis of chiral fluorinated amino acids. Org.

Biomol. Chem. 2012,10, 8583−8586.

(24) Erdbrink, H.; Nyakatura, E. K.; Huhmann, S.; Gerling, U. I. M.;

Lentz, D.; Koksch, B.; Czekelius, C. Synthesis of enantiomerically pure

(2S,3S)-5,5,5-trifluoroisoleucine and (2R,3S)-5,5,5-trifluoro-allo-isoleu-

cine. Beilstein J. Org. Chem. 2013,9, 2009−2014.

(25) Fischer, G. Peptidyl-prolyl cis/trans isomerases and their effectors.

Angew. Chem., Int. Ed. Engl. 1994,33, 1415−1436.

(26) Siebler, C.; Erdmann, R. S.; Wennemers, H. Switchable proline

derivatives: tuning the conformational stability ofthe collagen triple helix

by pH changes. Angew. Chem., Int. Ed. 2014,53, 10340−10344.

(27) Renner, C.; Alefelder, S.; Bae, J. H.; Budisa, N.; Huber, R.;

Moroder, L. Fluoroprolines as tools for protein design and engineering.

Angew. Chem., Int. Ed. 2001,40, 923−925.

(28) Shoulders, M. D.; Raines, R. T. Collagen structure and stability.

Annu. Rev. Biochem. 2009,78, 929−958.

(29) Ja

ckel, C.; Salwiczek, M.; Koksch, B. Fluorine in a native protein

environment - how the spatial demand and polarity of fluoroalkyl groups

affect protein folding. Angew. Chem., Int. Ed. 2006,45, 4198−4203.

(30) Salwiczek, M.; Samsonov, S.; Vagt, T.; Nyakatura, E. K.; Fleige, E.;

Numata, J.; Colfen; Pisabarro, M. T.; Koksch, B. Position-dependent

effects of fluorinated amino acids on the hydrophobic core formation of a

heterodimeric coiled coil. Chem. - Eur. J. 2009,15, 7628−7636.

(31) Salwiczek, M.; Koksch, B. Effects of fluorination on the folding

kinetics of a heterodimeric coiled coil. ChemBioChem 2009,10, 2867−

2870.

(32) Woolfson, D. N. The design of coiled-coil structures and

assemblies. Adv. Protein Chem. 2005,70,79−112.

(33) Monera, O. D.; Zhou,N. E.; Kay, C.M.; Hodges, R. S.Comparison

of antiparallel and parallel two-stranded alpha-helical coiled-coils.

Design, synthesis, and characterization. J. Biol. Chem. 1993,268,

19218−19227.

(34) Huhmann, S.; Nyakatura, E. K.; Erdbrink, H.; Gerling, U. I. M.;

Czekelius, C.; Koksch, B. Effects of single substitutions with

hexafluoroleucineand trifluorovalineon thehydrophobiccoreformation

of a heterodimeric coiled coil. J. Fluorine Chem. 2015,175,32−35.

(35) Nyakatura, E. K.; Reimann, O.; Vagt, T.;Salwiczek, M.; Koksch, B.

Accommodating fluorinated amino acids in helical peptide environ-

ments. RSC Adv. 2013,3, 6319−6322.

(36) Ja

ckel, C.; Koksch, B. Fluorine in peptide design and protein

engineering. Eur. J. Org. Chem. 2005,2005, 4483−4503.

(37)Pagel,K.;Wagner, S-C.;Araghi,R.R.; vonBerlepsch,H.;Bottcher,

C.; Koksch, B. Intramolecular Charge Interactions as a Tool to Control

the Coiled-Coil-to-Amyloid Transformation. Chem. - Eur. J. 2008,14,

11442−11451.

(38) Arnold, U.; Raines, R. T. Replacing a single atom accelerates the

folding of a protein and increases its thermostability. Org. Biomol. Chem.

2016,14, 6780−6785.

(39) Ye, S.; Loll, B.; Berger, A. A.; Mulow, U.; Alings, C.; Wahl, M. C.;

Koksch, B. Fluorine teams up with water to restore inhibitor activity to

mutant BPTI. Chem. Sci. 2015,6, 5246−5254.

(40) Krowarsch, D.; Otlewski, J. Amino-acid substitutions at the fully

exposed P1 site of bovine pancreatic trypsin inhibitor affect its stability.

Protein Sci. 2001,10, 715−724.

(41) Kawamura, K.; Yamada, T.; Kurihara, K.; Tamada, T.; Kuroki, R.;

Tanaka, I.; Takahashi, H.; Niimura, N. X-ray and neutron protein

crystallographic analysis of the trypsin-BPTI complex. Acta Crystallogr.,

Sect. D: Biol. Crystallogr. 2011,67, 140−148.

(42) Hanson, W. M.; Domek, G. J.; Horvath, M. P.; Goldenberg, D. P.

Rigidification of a flexible protease inhibitor variant upon binding to

trypsin. J. Mol. Biol. 2007,366, 230−243.

(43) Muller, K.; Faeh, C.; Diederich, F. Fluorine in pharmaceuticals:

looking beyond intuition. Science 2007,317, 1881−1886.

(44) Budisa, N. Prolegomena to future efforts on genetic code

engineeringbyexpandingitsamino acidrepertoire.Angew. Chem., Int. Ed.

2004,43, 3387−3428.

(45) Voller, J.-S.; Dulic, M.; Gerling-Driessen, U. I. M.; Biava, H.;

Baumann, T.; Budisa, N.; Gruic-Sovulj, I.; Koksch, B. Discovery and

investigation of natural editing function against artificial amino acids in

protein translation. ACS Cent. Sci. 2017,3,73−80.

(46) Budisa, N.; Wenger, W.; Wiltschi, B. Residue-specific global

fluorination of Candida antarctica lipase B in Pichia pastoris.Mol. BioSyst.

2010,6, 1630−1639.

(47) Meng, H.; Krishnaji, S. T.; Beinborn, M.; Kumar, K. Influence of

selective fluorination on the biological activity and proteolytic stability of

glucagon-like peptide-1. J. Med. Chem. 2008,51, 7303−7307.

(48) Asante, V.; Mortier, J.; Schluter, H.; Koksch, B. Impact of

fluorination on proteolytic stability of peptides in human blood plasma.

Bioorg. Med. Chem. 2013,21, 3542−3546.

(49) Asante, V.; Mortier, J.; Wolber, G.; Koksch, B. Impact of

fluorination on proteolytic stability of peptides: a case study with α-

chymotrypsin and pepsin. Amino Acids 2014,46, 2733−2744.

(50) O’Hagan, D.; Schaffrath, C.; Cobb, S. L.; Hamilton, J. T. G.;

Murphy,C.D.Biochemistry:Biosynthesisofanorganofluorine

molecule. Nature 2002,416, 279.

■NOTE ADDED AFTER ASAP PUBLICATION

This paper published ASAP on 8/12/2017. Figure 13 was

replaced and the revised version reposted on 8/16/2017.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.7b00226

Acc. Chem. Res. 2017, 50, 2093−2103

2103