Revisiting secondary structures in NCA polymerization: Influences on the analysis of protected polylysines

Article abstract of DOI:10.1021/ma5000392

Two series (degree of polymerization: 20-200) of polylysines with Z and TFA protecting groups were synthesized, and their behavior in a range of analytical methods was investigated. Gel permeation chromatography of the smaller polypeptides reveals a bimodal distribution, which is lost in larger polymers. With the help of GPC, NMR, circular dichroism (CD), and MALDI-TOF, it was demonstrated that the bimodal distribution is not due to terminated chains or other side reactions. Our results indicate that the bimodality is caused by a change in secondary structure of the growing peptide chain that occurs around a degree of polymerization of about 15. This change in secondary structure interferes strongly with the most used analysis method for polymers - GPC - by producing a bimodal distribution as an artifact. After deprotection, the polypeptides were found to exhibit exclusively random coil conformation, and thus a monomodal GPC elugram was obtained. The effect can be explained by a 1.6-fold increase in the hydrodynamic volume at the coil-helix transition. This work demostrates that secondary structures need to be carefully considered when performing standard analysis on polypeptidic systems.

Full text of DOI:10.1021/ma5000392

Article
pubs.acs.org/Macromolecules
Revisiting Secondary Structures in NCA Polymerization: Influences
on the Analysis of Protected Polylysines
†
†
†
rk, Hans Joachim Rad
er,‡
‡
David Huesmann, Alexander Birke, Kristina Klinker, Stephan Tu
̈
̈
,†
and Matthias Barz*
†
Institute of Organic Chemistry, Johannes Gutenberg-Universitat
̈
Mainz, Duesbergweg 10-14, 55128 Mainz, Germany
Max Planck Institute for Polymer Research, Ackermannweg 10, 55128 Mainz, Germany
S Supporting Information
‡
*
ABSTRACT: Two series (degree of polymerization: 20−200)
of polylysines with Z and TFA protecting groups were
synthesized, and their behavior in a range of analytical
methods was investigated. Gel permeation chromatography
of the smaller polypeptides reveals a bimodal distribution,
which is lost in larger polymers. With the help of GPC, NMR,
circular dichroism (CD), and MALDI-TOF, it was demon-
strated that the bimodal distribution is not due to terminated
chains or other side reactions. Our results indicate that the
bimodality is caused by a change in secondary structure of the
growing peptide chain that occurs around a degree of
polymerization of about 15. This change in secondary
structure interferes strongly with the most used analysis method for polymersGPCby producing a bimodal distribution
as an artifact. After deprotection, the polypeptides were found to exhibit exclusively random coil conformation, and thus a
monomodal GPC elugram was obtained. The effect can be explained by a 1.6-fold increase in the hydrodynamic volume at the
coil−helix transition. This work demostrates that secondary structures need to be carefully considered when performing standard
analysis on polypeptidic systems.
INTRODUCTION
at a certain chain lengthfor example, a coil-to-helix transition.
In the case of polypeptides three major secondary structures are
observed in nature: α-helix, β-sheet, and random coil. Since β-
sheets often precipitate when formed during the NCA
polymerization, polymer chemists usually deal with helical
and random coil structures in solution.
■
The synthesis of polypeptides based on the ring-opening
polymerization of α-amino acid-N-carboxyanhydrides (NCAs)
has attracted synthetic chemists since the early 20th century,
when Leuchs discovered this class of monomers.
end of the 20th and the beginning of the 21st century, various
techniques to control the polymerization have emerged.
Even nowadays, the synthesis of polypeptides under controlled
or even living conditions seems demanding, which is usually
attributed due to the presence of side reactions. While it
appears reasonable that at higher degrees of polymerization side
reactions like the activated monomer mechanism (AMM) etc.
occur, it is interesting that also at low molecular weight polymer
1
−3
Since the
4
−11
A number of methods can be used to characterize the
different secondary structures. X-ray crystallography is the most
used method when it comes to the study of protein secondary
16
12
and tertiary structures in the solid state. However, high-
quality single crystals are required for this technique. In
solution, NMR techniques are also often used to determine
protein structures and the chemical shift alone can be used to
1
7,18
1
3,14
identify secondary structures.
Further, circular dichroism
side reactions are a matter of debate.
Side reactions due to
1
9,20
21,22
(
CD)
and IR
spectroscopy are useful tools for the
the cleavage of relatively labile protecting groups, such as
benzyl esters or trifluoroacetic acid amide, might occur under
certain conditions due to a nucleophilic attack of the amino
terminus at the carbonyl carbon. But in the case of robust
protecting groups such as benzyl groups, or even trifluoroacetic
acid amide at ambient or low temperatures, these side reactions
are very unlikely to occur. On the other hand, also
conformational effects are known to influence the propagation
speed of the NCA polymerization and therefore influence the
determination of secondary structures.
It is known since several decades that there is a change in
secondary structure as the polypeptide chain grows. Oligo(γ-
ethyl glutamate), for example, was found by CD spectroscopy
and NMR in trifluoroethanol to change from a random coil
conformation to an α-helix at a degree of polymerization
15
23
̈
around seven. Kricheldorf and Muller also investigated the
2
,5
molecular weight distribution.
In all cases precise sample characterization becomes rather
demanding when polymers undergo a change in superstructure
Received: January 6, 2014
Revised: January 9, 2014
Published: January 22, 2014
©
2014 American Chemical Society
928
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
secondary structure of polypeptides in the solid state with the
standards. In all cases the elution diagram was evaluated with PSS
WinGPC from Polymer Standard Service Mainz.
CD spectroscopy was performed on a Jasco J-815 spectrometer in a
1
3
24,25
help of IR spectroscopy and C NMR CP/MAS.
The different techniques that have been developed to control
4,8
cell with a path length of 1 mm. The temperature was kept constant at
NCA polymerization can lead to different end groups, which
2
0 °C. Spectra were recorded at concentrations of 0.5, 0.3, and 0.2 mg
1 +
may also have an effect on the secondary structures. In this
−
−1
−
mL for PLys(Z) (in HFIP, 3 g L K TFA ), PLys(TFA) (in HFIP,
3
work we have used primary amines as well as their HBF salts
−1
+
−
4
g L K TFA ), and deprotected Lys (in Milli-Q water),
to initiate the polymerization of lysine NCAs giving us an
alkylamide at the C-terminus and a free amine at the N-
respectively. Each spectrum represents the average of two scans. The
mean residual ellipticity θMR was calculated from the observed
ellipticity θ (see Supporting Information).
For the spectra of the fractions Z1_S and Z1_B, which were
collected from the GPC outlet, the concentration was unknown, and
2
6
terminus. Thus, the effects reported in this work may not be
present in all controlled NCA polymerization techniques,
although it appears likely that these effects are much more
related to the individual monomer than to a specific end group.
In this work we report the synthesis of poly(N-ε-
benzyloxycarbonyl-L-lysine) (PLys(Z)) and poly(N-ε-trifluor-
oacetyl-L-lysine) (PLys(TFA)) with a degree of polymerization
varying from 20 to 200. Polylysine has attracted much attention
and is used in a variety of applications like coatings for
3
0 scans were recorded and averaged.
The mass measurements were carried out with a REFLEX MALDI-
TOF mass spectrometer (Bruker, Bremen, Germany), equipped with a
37 nm nitrogen laser. 1,8,9-Trihydroxyanthracene (dithranol) was
3
used as a MALDI matrix. For sample preparation 50 μL of the matrix
dissolved in THF (5 mg/mL) was mixed with 10 μL of the
fractionated polymer solution and crystallized on the stainless steel
target immediately before measurement. The mass spectrometer was
calibrated externally with a C /C fullerene mixture. The instrument
2
7
improved cell attachment or as polycations for gene
2
8
delivery. Its use in these and other important industrial
applications, e.g. detergent formulations, water treatment, or
coatings, makes polylysine an industrially relevant polypeptide,
and a good insight into its polymerization behavior is
indispensable. The ring-opening polymerizations (ROP) were
initiated by primary amines as well as their tetrafluoroborane
60
70
was operated in linear and reflection mode, both at an acceleration
voltage of 20 kV. In linear mode the Bruker HIMAS detector was used,
providing high sensitivity and low resolution and in reflection mode
dual channel plates were used, providing high resolution but lower
sensitivity. The mass spectra were smoothed and baseline corrected
with the XMASS data processing program (Bruker). mMass was used
9
,25
ammonium salts. By combining different methods like GPC,
CD spectroscopy, NMR, and MALDI-TOF, we show that in
the case of the protected polylysines PLys(Z) and PLys(TFA)
termination reactions are absent under the applied conditions.
Most importantly, we report on a change in secondary structure
around a degree of polymerization of 15 that interferes with
standard GPC analysis. Finally, we would like to provide a
model that explains the observed effects by taking changes in
the hydrodynamic volume during the coil-to-helix transition
into account.
3
0−32
to evaluate the spectra.
Initiator Synthesis. To 5 mL (5.59 g, 36.74 mmol) of HBF ·Et O,
4
2
4
.31 mL (3.20 g, 36.74 mmol) of neopentylamine was slowly added.
The mixture was cooled using an ice bath. The addition resulted in the
precipitation of a slightly brown solid. The ether was removed in vacuo,
and the solid was recrystallized two times from ethyl acetate and
washed with cyclohexane. The product was dried in vacuo yielding 2.72
g (15.59 mmol, 42%) of neopentylammonium tetrafluoroborate as a
1
colorless solid. H NMR (300 MHz, DMSO-d ) δ [ppm] = 7.58 (s,
6
+
13
3
H, NH ), 2.63 (s, 2H, CH ), 0.93 (s, 9H, C(CH ) ). C NMR (75
3 2 3 3
+
MHz, DMSO-d ) δ [ppm] = 49.94 (CH NH ), 30.21 (C(CH ) ),
6
2
3
3 3
−1
+
2
2
6.78 (3H, C(CH ) ). FT-IR: (neat) v [cm ] = 3248 br w (NH ),
963 w (CH), 1617 w, 1508 m, 1482 w, 990 br s (BF ), 849 w, 520
3
3
3
−
EXPERIMENTAL SECTION
■
4
m. Anal. Calcd for C H NBF : C, 34.43; H, 8.06; N, 8.00. Found: C,
Materials and Methods. DMF was dried by stirring over
molecular sieve (3 Å) and BaO. It was then distilled in vacuo at low
to ambient temperature onto molecular sieve to remove dimethyl-
amine impurities, was protected from light, and was stored at −80 °C.
Ethyl acetate, THF, hexane, and cyclohexane were distilled from Na/
K; other solvents were used as received unless otherwise stated.
Neopentylamine was dried over CaH2 and distilled before use.
Protected L-lysines were purchased from Orpegen.
5
14
4
3
4.35; H, 7.99; N, 8.07.
Monomer Synthesis. N-ε-Benzyloxycarbonyl-L-lysine-N-carbox-
yanhydride. In a three-necked flask equipped with a reflux condenser,
dropping funnel, and septum, 14.00 g (50 mmol) of Z-protected lysine
was suspended in 100 mL of THF, and the suspension was heated to
7
0 °C. 6.6 mL (55 mmol) of diphosgene was added over 30 min. The
solution was heated until all solid disappeared (30−60 min). Dry
nitrogen was then bubbled through the solution for 2−3 h to remove
excess HCl and phosgene. The solution was concentrated in vacuo, and
dry cyclohexane was added to precipitate the NCA. The suspension
was stored in the fridge for 1 h. The solid was collected by filtration in
an inert atmosphere and washed with cyclohexane. It was then
dissolved in the smallest possible volume of THF, and again
cyclohexane was added to precipitate the product. The suspension
was left in the fridge overnight. The solid was then collected by
filtration in an inert atmosphere and dried in a stream of dry nitrogen.
1
13
H and C NMR spectra were recorded on a Bruker AC 300 or AV
4
00 at room temperature. The spectra were calibrated using the
29
solvent signals.
Infrared spectroscopy was measured on a Jasco FT/IR-4100 with an
ATR sampling accessory (MIRacle, Pike Technologies). IR spectra
were analyzed using Spectra Manager 2.0 (Jasco).
Gel permeation chromatography (GPC) was performed with DMF
containing 0.25 g/L LiBr as eluent at 50 °C. The column was packed
with HEMA 300/100/40. A refractive index detector (G 1362A RID)
was used to detect the polymer. GPC in HFIP was performed with 3
The recrystallization was repeated and AgNO (0.1 M in water) was
3
+
−
g/L K TFA at 40 °C. The columns were packed with modified silica
added to the filtrate, to confirm the absence of chloride ions (AgCl
would immediately precipitate as a colorless solid). The Lys(Z)NCA, a
colorless solid (13.23 g, 43.2 mmol, 86%), was transferred to a Schlenk
(
PFG columns, particle size: 7 μm, porosity: 100 and 1000 Å). A
refractive index detector (G 1362A RID, Jasco) was used to detect the
polymer. Molecular weights were calculated using a calibration
performed with PMMA standards (Polymer Standards Services
GmbH). GPC in water was performed with buffered aqueous solution
1
tube and stored at −80 °C; mp 99.1 °C. H NMR (400 MHz, DMSO-
d
) δ [ppm] = 9.09 (s, 1H, CONH−C ), 7.40−7.25 (m, 6H, Ar H,
6
α
NH(Z)), 5.01 (s, 2H, NHCH
CH NH), 1.74−1.63 (m, 2H, CH−CH
CH
(C COOCNH), 156.06 (COOBn), 151.94 (C NHCOO), 137.21
Ph), 4.43 (t, 1H, C
), 1.44−1.31 (m, 4H, CH
). C NMR (75 MHz, DMSO-d ) δ [ppm] = 171.61
H), 2.99 (q, 2H,
2
α
(
50 mM sodium phosphate, 150 mM sodium chloride, pH 7). The
2
2
2
−
1
3
following parts were used: Jasco pump (pU-2086 Plus series), a Jasco
UV/vis detector (UV-2077 Plus), a Jasco RI-detector (Jasco RI 2031
2
6
α
α
−1
Plus series). The flow rate was set to 0.4 mL min . A Superose6 10/
(Ar), 128.29 (Ar), 127.67 (Ar), 65.09 (PhCH ), 56.97 (C ), 39.88
2 α
300 GL column was used. Calibration was done using protein
(C ), 30.58 (C ), 28.72 (C ), 21.55 (C ).
ε β δ γ
9
29
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
N-ε-Trifluoroacetyl-L-lysine-N-caboxyanhydride. 10.15 g (41.9
mmol) of Lys(TFA) was dissolved in 150 mL of ethyl acetate, and
the suspension was heated to 70 °C. 6 mL (50.3 mmol) of diphosgene
was added via a syringe over 3 h. After an additional 1 h, in which the
solution remained slightly cloudy, a nitrogen stream was passed
through the solution for 1 h, and solution was concentrated in vacuo.
GPC at Different Time Points. GPC data at different time points
of the polymerization were obtained by taking samples after the
appropriate times and precipitation in ethyl ether, followed by washing
and lyophilization as mentioned above.
Fractionation of the Polymer via GPC. 20 mg of Z1 (see Table
−1
+
−
1) was dissolved in 0.5 mL of HFIP containing 3 g L K TFA . 100
35 mL of AcOEt was added, and the cloudy solution was filtered. The
product was crystallized by slowly adding 80 mL of hexane to the
filtrate over 2.5 h. The product was collected by filtration and
recrystallized two times. The product was dried in vacuo for 1 h,
Table 1. PLys(Z) Polymers and Their Analytical Data from
NMR, HFIP GPC, and DMF GPC
yielding 7.12 g (26.6 mmol; 63%) of colorless crystals. The product
Mn
M (HFIP
Đ
(HFIP
GPC)
M (DMF
Đ
(DMF
GPC)
n
n
1
a
a
was stored at −80 °C; mp 101 °C. H NMR (400 MHz, DMSO-d , δ)
X
n
(NMR)
GPC)
GPC)
6
polymer (NMR) [kg/mol] [kg/mol]
[kg/mol]
9
.40 (t, 1H; CONH−CF ), 9.09 (s, 1H; CONH−CH), 4.43 (t, 1H;
3
CH−CH ), 3.17 (q, 2H; CH −NH), 1.86−1.58 (m, 2H; CH−CH ),
1
CH −NH).
Z1
Z2
Z3
Z4
24
57
6.2
15.0
22.1
51.3
6.7
11.5
14.9
20.2
1.38
1.29
1.19
1.15
4.0
7.5
1.70
1.52
1.20
1.15
2
2
2
.50 (p, 2H; CH −CH −CH −NH), 1.44−1.21 (m, 2H; CH −CH −
2
2
2
2
2
2
84
13.4
16.6
Polymer Synthesis. Polymerizations were carried out in Schlenk
196
tubes in DMF. The NCAs were dissolved in in DMF at a
a
−3
Relative to PMMA standards.
concentration of 100 mg cm . The initiator (neopentylammonium
tetrafluoroborate or neopentylamine) was added to initiate the
polymerization. The solution was stirred at 20 °C, and the reaction
μL of this solution was injected using the autosampler. Fractions of 0.4
mL were collected and combined to Z1_S (elution volume: ca. 19.5−
flask was opened to the Schlenk line to allow the escape of CO . The
2
2
1 mL) and Z1_B (elution volume: ca. 16.5−19 mL) for CD
progress of the polymerization was monitored by IR. Once all NCA
bands disappeared the mixture was precipitated in diethyl ether and
washed with ether twice. The polymer was then freeze-dried from
dioxane yielding a colorless fluffy powder (typical yields 80%).
Poly(N-ε-benzyloxycarbonyl-L-lysine) by Neopentylammonium
Tetrafluoroborate. 694 mg of Lys(Z)NCA (2.27 mmol) was
transferred under nitrogen counter flow into a predried Schlenk-tube
equipped with a stir bar and again dried in high vacuum for 1 h. Then
the NCA was dissolved in 6.6 mL of dry DMF. To prepare a stock
solution of the initiator (always used on the day of preparation) 104
mg of neopentylammonium tetrafluoroborate was weighed in and
measurement. Afterward, the solvent was removed under reduced
pressure for MALDI-TOF analysis.
RESULTS AND DISCUSSION
■
In this work, we have synthesized various PLys(Z) and
PLys(TFA) homopolymers with different degrees of polymer-
ization (between 20 and 200) and investigated their behavior in
a variety of standard analytical methods like NMR, GPC,
MALDI-TOF, and CD spectroscopy. The polymers were
synthesized by ring-opening polymerization of the correspond-
ing NCAs. All NCAs were synthesized using the Fuchs−
−1
dissolved in 10 mL of dry DMF (c = 0.59 mol L ). 38 mL (0.023
mmol) of the initiator solution was added to the monomer solution via
syringe. The solution was stirred at 40 °C and kept at a constant
pressure of 1.25 bar of dry nitrogen via the Schlenk line to prevent
33
Farthing method (Scheme 1). Lys(Z) and Lys(TFA) NCAs
impurities from entering the reaction vessel while allowing CO to
2
Scheme 1. Synthesis of Lys(Z) and Lys(TFA) NCAs
escape. Completion of the reaction was confirmed by IR spectroscopy
−1
(
disappearance of the NCA peaks (1853 and 1786 cm )). Directly
after completion of the reaction the polymer was precipitated in cold
ether and centrifuged (4500 rpm at 4 °C for 15 min). After discarding
the liquid fraction, new ether was added and the polymer was
resuspended using sonication. The suspension was centrifuged again
and the procedure was repeated. The polymer was then dissolved in
dioxane and lyophilized to obtain a fluffy powder (595 mg, yield 83%).
1
For different chain lengths yields ranged from 72 to 91%. H NMR
(
400 MHz, DMSO-d ): δ [ppm] = 8.80−7.80 (64H (1n), br, −NH−
6
CO−CH−), 7.54−6.98 (358H (5n), br, −C H ), 5.30−4.65 (115H
(
were purified by recrystallization. Yields and melting points
(Lys(Z)-NCA: mp = 99.1 °C; Lys(TFA)-NCA: mp = 101 °C)
6
5
2n), br, −O−CH −C H ), 4.35−3.60 (43H (1n), br, −CH−CH ),
2
6
5
2
2
3
.12−2.68 (115H (2n), br, CH−CH ), 2.70−1.70 (332H (4n), br,
2
were in agreement with published literature. The silver nitrate
CH −CH −CH −NH), 0.82 (9H (initiator), br, C(CH ) ).
2
2
2
3
3
test was negative for all purified NCAs. All synthesized NCAs
have been stored at −80 °C over months without any
detectable degradation or oligomerization.
The synthesis of poly(N-ε-benzyloxycarbonyl-L-lysine) initiated by
neopentylamine was carried out in the same way at 20 °C. Since
neopentylamine is a liquid, it was added using an Eppendorf pipet.
The synthesis of PLys(TFA) was analogous to the synthesis of
PLys(Z). Exact details can be found in the Supporting Information.
Deprotection of Poly(N-ε-benzyloxycarbonyl-L-lysine). 143
mg of PLys(Z) was dissolved in 4.3 mL of acetic acid (c = 0.033 g/
mL), and 0.6 mL of HBr in acetic acid (33% w/w; 4-fold excess in
regard to (Z)-group) was added. After the addition of the HBr a
precipitate formed. The mixture was stirred vigorously for 90 min at
room temperature. Then 1 mL of water was added which dissolved
part of the precipitate. After an additional 30 min another mL of water
was added which dissolved the rest of the precipitate. The solution was
stirred for 1 h and extracted two times with ether to remove acetic
The ring-opening polymerization of NCAs was performed in
absolute dimethylformamide (DMF) using neopentylammo-
10
nium tetrafluoroborate and neopentylamine as initiators
(Scheme 2). The polymerizations were allowed to proceed
under the exclusion of light until full conversion of the NCAs
was observed by FT-IR (i.e., when the NCA associated
−1
carbonyl peaks at 1853 and 1786 cm are no longer
detectable). Afterward, the polymers were purified by
precipitation in ethyl ether, washed with diethyl ether, and
freeze-dried from dioxane.
1
Tables 1 and 2 display the analytical data of the synthesized
acid. The aqueous phase was freeze-dried, yielding 86 mg of PLys. H
1
polymers. Precise H NMR analytics were enabledeven at
NMR (300 MHz, D O) δ [ppm] = 4.30 (t, 37H (1n), C H), 3.00 (t,
2
α
7
1
9H (2n), CH−CH ), 1.90−1.60 (m, 146H (4n), CH −CH ), 1.59−
high degrees of polymerizationby using neopentylamine as
2
2
2
.27 (m, 82H (2n), CH −NH ), 0.85 (s, 9H, (CH ) ).
initiator, allowing integration of the 9 initiator protons (see
2
2
3 3
9
30
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
Scheme 2. Polymerization of Lys(Z) and Lys(TFA) Using
Neopentylamine
Table 2. PLys(TFA) Polymers and Their Analytical Data
from NMR (DMSO-d ), HFIP GPC, and DMF GPC
6
Mn
M (HFIP
Đ
(HFIP
GPC)
M (DMF
Đ
(DMF
GPC)
n
n
a
a
Xn
(NMR)
GPC)
GPC)
polymer (NMR) [kg/mol] [kg/mol]
[kg/mol]
Figure 2. CD spectrum of PLys(Z) in HFIP, showing different
secondary structures at different length.
T1
T2
T3
T4
20
65
4.5
14.6
20.2
32.1
3.4
9.5
1.59
1.10
1.11
1.13
6.2
21.9
27.9
44.9
1.95
1.21
1.21
1.16
90
11.9
19.0
distributions can be seen in HFIP GPC (Figure 1), DMF GPC
143
(
Figure 2), and DOSY NMR (Figure S9). The effect is
strongest for Z1 (DP = 24), less pronounced in the polymer Z2
DP = 57), and almost absent in polymer Z3 (DP = 84). The
a
Relative to PMMA standards.
(
Figure 4). It was ensured that the initiator itself is well soluble
at the applied precipitation conditions for the polymer. Thus,
initiator peaks are polymer associated, which was additionally
confirmed by DOSY (diffusion ordered spectroscopy) NMR
experiments (Figure S9).
PLys(TFA) polymers show the same trend, as can be seen in
Figure S1.
These bimodal distributions can emerge due to two different
reasons, namely termination reactions or faster propagation
than initiation of polymer chains as well as effects caused by
secondary structure formation in solution. Thus, chemical as
well as secondary structure associated effects have to be
analyzed to attribute the observed effects to one of them.
Chemically, the termination of a portion of the chains might
stop their growth, while other chains continue to grow, leading
to a bimodal distribution. It appears, however, that in the case
of PLys(Z) and PLys(TFA) the presence of a bimodal
distribution vanishes for larger polymers, which is counter-
intuitive since side reactions such as activated monomer
mechanism as well as urea formation should be more
pronounced in higher molecular weight polymers due to
elongated reaction times. In fact, when samples were taken and
analyzed during different time points of the polymerization
(Figure S2), the distribution undergoes a transition from
monomodal to bimodal and back to monomodal. This effect
was present in neopentylammonium tetrafluoroborate as well as
neopentylamine initiated polymerizations. Thus, side reactions
are very unlikely responsible for the bimodal distributions.
As can be seen by comparing the molecular weights from
NMR and GPC in Tables 1 and 2, molecular weights are
underestimated when a correction regarding weight per bond is
not taken into account. After correcting the molecular weight
by a factor of 1.75 (Table S1)correcting for the differences in
molecular weight per bond in the repeating unit of the
polymerthe molecular weight is overestimated by GPC. This
effect gets less pronounced for large polymers as already
3
4,35
reported by Flory and co-workers
and even drops to an
underestimation of a factor of 0.69 in Z4. We would like to
emphasize the difference in molecular weights obtained by
HFIP and DMF GPC. We are well aware that mentioning these
findings seems trivial but underlines the problems of molecular
weight termination by GPC of polylysine or polypeptides in
general.
It can be seen in Tables 1 and 2 that the shorter polymers
have a very broad distribution thatas can be seen in Figures 1
and 2stems from a bimodal distribution. These bimodal
Figure 1. GPC of PLys(Z) at different length in HFIP (a) and DMF (b). Short polymers show bimodal distributions.
9
31
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
Interestingly, this effect is much less pronounced in the
polymerization of Glu(OBn) NCA.
polymers remained soluble at all times during the polymer-
ization.
10
However, it has been argued that PLys synthesized by NCA
polymerization has dead chain ends (e.g., 62% at an M/I ratio
The finding, that secondary structures play a strong role in
PLys analytics, was further supported by NMR spectra in
DMSO-d and HFIP-d , which displayed two peaks for the α-
1
3
of 20, due to reaction with DMF) so MALDI-TOF
spectrometry was used to determine the integrity of the
polypeptides. In this case, termination of propagating species
due to side reactions can be analyzed. Because of the mass
6
2
proton as well as the initiator (Figure 3, Figures S6 and S7). In
3
6,37
discrimination,
it was, however, not possible to obtain
MALDI-TOF spectra for the complete bimodal molecular
weight distribution of PLys(Z) polymers. Thus, our smallest
PLys(Z) (Z1) was separated into 12 fractions by HFIP GPC.
The individual fractions could be analyzed individually (Figures
S3 and S4). The fractions did not display side products, and the
peaks nicely correspond to the individual PLys(Z) chains with
neopentylamine initiator and amine end group (Tables S2 and
S3). Within the detection limits of MALDI-TOF we could not
verify dead chain ends even at higher molecular weights;
however, MALDI-TOF analysis is limited in the case of
PLys(Z) to a molecular weight of around 13 000 g/mol, which
corresponds to a degree of polymerization of 50−60. Signal
intensity for higher molecular weight polymers vanishes and
thus limits the use of this methodology for PLys(Z) polymers.
In addition, it limits the comparison of intensities between
different fractions. In summary, MALDI-TOF analysis of the
low molecular weight fraction displayed the absence of
termination products within the accuracy of the analytical
method. All species in the spectra are PLys(Z) polymers
initiated by neopentylamine. These findings exclude termi-
nation, initiation by dimethylamine, or the presence of activated
monomer mechanism as reasons for the bimodal distribution of
PLys(Z). These findings were independent from the polymer-
ization methodology used.
1
Figure 3. H NMR spectra of pLys(Z) with different chain length in
DMSO-d . Inset a shows the α-protons and inset b the initiator.
6
Figure 3, two α-proton peaks of PLys(Z) can be seen at about
.2 and 3.8 ppm (insert a). These chemical shifts are similar to
4
the shifts of a lysine α-proton in a random coil or α-helical
structure, respectively. The shifts cannot be attributed to β-
sheet formation since these structures would cause a signal of
the α-proton at around 4.7 ppm. Furthermore, a similar trend
can also be observed for the initiator protons (inset b, Figure 3)
whose chemical shift also depends on the secondary structure.
As in the GPC traces, comparison to PLys(TFA) shows a
similar trend with two distinct α and initiator protons for the
short polypeptides and only one signal for the longer ones
(Figure S6).
18
Physically, different secondary structures can strongly
influence the hydrodynamic radius, leading in turn to a
difference in elution volume, even if the polymer chains have
similar length. CD spectroscopy is one of the most convenient
1
9,20
ways to estimate the secondary structures of proteins.
We
used this technique to gain insight into the variation in
secondary structure with polymer length. In contrast to DMF
the use of HFIP as solvent enables the correlation of GPC and
CD analysis. As in GPC analysis, we observed a trend from
smaller to larger polymers. As can be seen in Figure 2, the
circular dichroism is stronger in larger polymers, further
supporting the assumption of different secondary structures at
different length. This trend is again also present in PLys(TFA)
Using size exclusion chromatography, we were able to
separate the two secondary structures of the short polymer (Z1
in Table 1) based on their different hydrodynamic volumes.
Selected fractions are shown in Figure 4. The green fraction is
especially interesting since it falls between the two GPC peaks.
It can be seen that this spectrum is basically a mixture of the
two adjacent fractions and is rather broad compared to the
other fractions, fitting well into the hypothesis that the
difference in chain length in this fraction is much larger than
the difference in hydrodynamic volume. However, it has to be
kept in mind that mass discrimination might distort the
MALDI-TOF spectra.
(Figure S5).
The CD spectra were measured at 20 °C to prevent solvent
loss during measurement. However, it was made sure that the
temperature has no influence on the secondary structure
(Figure S10).
All fractions belonging to one peak were combined (Z1_S,
ca. 19.5−21 mL elution volume, smaller polymer; Z1_B, ca.
16.5−19 mL elution volume, bigger polymer). The MALDI-
TOF analysis of the two different fractions showed that Z1_S
However, it has to be noted that, although it looks very
similar, the CD spectrum does not fully match the α-helix
spectrum. This might be due to the protecting groups which
both contain amide bonds. This itself might lead to a different
CD spectrum for an α-helix but might also induce a different
kind of secondary structure that does not occur in natural
peptides. Additionally, it is most likely that the polypeptide
does not form a single helix but an assembly of helical units
with random coil like defects as reported by Flory and co-
was composed of polymers of an X between 5 and 15, while
n
the polymers in Z1_B had up to 60 repeating units (Figure S8).
The CD spectra (Figure 5) of both peaks clearly show that
only the peak of the fraction containing the longer polymers
(Z1_B) shows strong CD activity. The Z1_S fraction shows
close to no CD peaks since the CD of a random coil is weaker
than that of an α-helix, and further the chain is very short,
generating only a weak signal. Also, the concentration in Z1_S
is slightly lower than in Z1_B.
3
4
workers.
β-Sheet formation, as reported by Kricheldorf (by C NMR
13
25
in the solid state), is not very likely to occur in our case, since
9
32
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
Scheme 3. Deprotection of PLys(Z) with HBr
indeed responsible for bimodal distributions in short, protected
polylysines.
In summary, we have demonstrated that the observed
bimodal molecular weight distributions observed by DMF as
well as HFIP GPC of PLys(Z) or PLys(TFA) polypeptides are
not due to side reactions (termination, initiation by impurities,
or activated monomer mechanism). It is much more based on
differences in the hydrodynamic radii of both secondary
structures. In line with our argumentation the bimodal
distribution of a sample vanishes after its deprotection, when
all polymers are in a random coil conformation. We would also
like to mention that the above observed effects are less
prominent in poly(glutamic acid) (PGA) and can also be
hidden by a second block, e.g. poly(ethylene glycol) or
polysarcosine, in block copolymers. When the nonpeptidic
block has a degree of polymerization above 150, a bimodal
distribution is hardly detectable.
Figure 4. MALDI-TOF spectra of different fractions of Z1.
10
Estimates of the Size Difference of α-Helix and
Random Coil. In the last part of this work we would like to
discuss a possible explanation for the bimodal character of the
GPC elugrams in detail. The theoretical model presented here
is based on assumptions, which are most likely but cannot be
proven by the authors. Thus, the presented model can only
serve for a qualitative discussion of the size change that is
difficult or even impossible to determine for polydispers sample
experimentally. However, we feel that this model provides a
deeper understanding of the reported effects.
We have already demonstrated that PLys(Z) and PLys(TFA)
exist in different superstructures (random coil and α-helix)
depending on the degree of polymerization. For P(Lys) in a
random coil conformation the worm-like chain (WLC) model
is a suitable model. To estimate the hydrodynamic radius of a
random coil of PLys(Z) with a DP of 15, the model by
Figure 5. CD spectra of the two GPC peaks of Z1 indicating that only
the peak corresponding to the larger polymer fraction shows
significant CD activity.
Since it is well-known that unprotected PLys shows a
40
38
Yamakawa and Fujii as corrected by Dorfman et al. was
random coil conformation at neutral pH, we deprotected
41
selected.
Z1the polymer showing the strongest bimodal distribution
39
The length of the chain is estimated to be L = 5.6 nm (15 ×
using HBr in acetic acid (see Scheme 3). We expected that
the deprotection should remove any effect coming from
different secondary structures, thus leading to a monomodal
distribution.
3
2
74 pm) with a Kuhn length of l = 2 nm and a diameter of d =
k
.94 nm. This leads to an approximate radius, which is
calculated (see Supporting Information) to be
As shown by NMR in Figure 6, the deprotection was
quantitative (in the experimental limits of NMR) and the
signals from the α-helix disappeared.
Further, GPC measurements in buffer showed only a
monomodal peak in the GPC elugram corresponding to a
monomodal molecular weight distribution (Figure 7a) for the
deprotected polylysine. CD spectroscopy finally revealed that
the only secondary structure present was a random coil (Figure
R hW LC = 1.7 nm
If we assume an α-helical conformation, with a pitch of 5.4 Å
incorporating 3.6 residues and a DP of 15, we have a helix
with a length of 2.25 nm spanning 4.2 turns.
The approximate length of the extended side chain can be
calculated by assuming a median bond length of l = 150 pm and
an angle between bonds of Θ = 110°. The effective length of
the side chain (n = 12 bonds) is thus
42
7b), conclusively proving that different secondary structures are
9
33
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
Figure 6. NMR spectrum of PLys before (Z1, green, HFIP-d ) and after (L1, red, D O) deprotection, showing that only one secondary structure is
2
2
present after deprotection.
Figure 7. (a) GPC trace of L1 (PLys ) after deprotection, showing monomodal distribution. (b) CD spectrum of PLys after deprotection in water,
25
showing random coil conformation.
Figure 8. Hydrodynamic radii of Lys(Z) with DP = 15 in random coil and α-helical conformation.
9
34
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
leff = nl sin(θ/2) = 1.47 nm
ASSOCIATED CONTENT
Supporting Information
GPC, CD, and NMR and experimental data of PLys(TFA),
additional MALDI-TOF spectra, an NMR spectrum in HFIP,
DOSY spectrum, GPC traces at different time points of the
polymerization, and corrected molecular weights of PLys(Z) as
■
*
S
Since the diameter of an α-helix without side chains is
approximately 0.6 nm, the diameter of the whole helix can be
approximated to be 3.54 nm (see Figure 8).
This leads to a cylinder with a length of 2a = 2.25 nm and a
diameter of 2b = 3.54 nm (see Figure 8). The equivalent radius
well as equations for the R estimations. This material is
h
(
by
radius of a sphere with the same volume) of a cylinder is given
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: Barz@uni-mainz.de (M.B.).
Notes
■
2
3
4
⎛ a ⎞
⎜ ⎟
cy
eq
3
R
=
⎠ a = 3.48 nm
⎝
b
leading to a hydrodynamic radius of
The authors declare no competing financial interest.
f
f0
Rh = R ec qy = 2.0 nm
cy
ACKNOWLEDGMENTS
■
The authors are indebted to Prof. M. Schmidt for very fruitful
discussions about the dimensions of different superstructures in
solution. M.B. acknowledges financial support by the SFB 1066
and the NMFZ Mainz. D.H. acknowledges the support by the
“Verband der Chemischen Industrie” (VCI) and the “Max
Planck Graduate Center with the Johannes Gutenberg-
where f is the translational friction coefficient of the cylinder
and f is the translational friction coefficient of a sphere having
the same volume. An equation for f/f was modeled by Ortega
et al. (see Supporting Information). Since the volume is
proportional to R , the volume ratios would be 1:1.6 for worm-
like chain and cylinder, respectively.
0
0
4
3
3
h
̈
Universitat Mainz” (MPGC).
To obtain a random coil with a hydrodynamic radius of 2.0
ABBREVIATIONS
■
nm (the R of an α-helix with DP = 15), the PLys(Z) chain
h
AMM, activated monomer mechanism; CD, circular dichroism;
DMF, dimethylformamide; DMSO, dimethyl sulfoxide; DOSY,
diffusion ordered spectroscopy; DP, degree of polymerization;
GPC, gel permeation chromatography; HFIP, hexafluoroiso-
propanol; NCA, α-amino acid-N-carboxyanhydride; PEG,
poly(ethylene glycol); PMMA, poly(methyl methacrylate);
PLys, poly(L-lysine); PLys(TFA), poly(N-ε-trifluoracetamide-
L-lysine); PLys(Z), poly(N-ε-benzyloxycarbonyl-L-lysine);
ROP, ring-opening polymerization; THF, tetrahydrofuran;
WLC, worm-like chain.
would need to have a degree of polymerization of 30. These
rough estimates show that the transition from coil to helix
appears in the GPC as if the molecular weight of the polymer
would suddenly double. This model can qualitativly explain the
bimodal GPC elugram by the presence of two superstructures
that differ significantly in the individual hydrodynamic volumes.
CONCLUSIONS
■
Surprisingly, a change in secondary structurealthough well-
knownis seldom discussed in context with standard analytics
of polypeptides made by NCA polymerization. The most
common method for determining the molecular weight of
polymers is size exclusion chromatography, which is certainly
simple to perform and gives good estimates of polymer
definition and molecular weights. It has to be kept in mind that
molecular weight determination by GPC depends on the
correlation of hydrodynamic volumes between sample and
standard, which is greatly influenced by secondary structure.
Because of their ability to form secondary structures via
hydrogen bonds, polypeptides are often not in a random coil
conformation or even one distinct conformation.
REFERENCES
■
(
(
1) Leuchs, H. Ber. Dtsch. Chem. Ges. 1906, 39, 857−861.
2) Kricheldorf, H. R. α-Amino Acid-N-Carboxyanhydrides and Related
Heterocycles: Syntheses, Properties, Peptide Synthesis, Polymerization;
Springer: Berlin, 1987.
(
(
3) Kricheldorf, H. R. Angew. Chem., Int. Ed. 2006, 45, 5752−5784.
4) Deming, T. J. Nature 1997, 390, 386−389.
(5) Aliferis, T.; Iatrou, H.; Hadjichristidis, N. Biomacromolecules
2004, 5, 1653−1656.
(6) Dimitrov, I.; Schlaad, H. Chem. Commun. 2003, 2944−2945.
(
́
7) Vayaboury, W.; Giani, O.; Cottet, H.; Deratani, A.; Schue, F.
Macromol. Rapid Commun. 2004, 25, 1221−1224.
(
(
2
(
8) Lu, H.; Cheng, J. J. Am. Chem. Soc. 2007, 129, 14114−14115.
In this work we have demonstrated that secondary structure
formation needs to be carefully considered when performing
analysis on polypeptides. It has been shown that for PLys(Z)
the helix−coil transition occurs around a degree of polymer-
ization of 15 but does not seem to be sharp. More importantly,
standard GPC analysis tends to overestimate polymer dispersity
tremendously whenever the sample consists of polymers
differing in conformation, like random coil or helical (rod-
like) structure.
It should also be noted that the difference in secondary
structures is in no way problematic for the application of
polypeptides. When going to larger polypeptides, an all-α-
helical structure is observed, and even for smaller polypeptides
the difference in secondary structures is lost after deprotection.
9) Peng, H.; Ling, J.; Shen, Z. J. Polym. Sci., Part A: Polym. Chem.
012, 50, 1076−1085.
10) Conejos-Sanchez, I.; Duro-Castano, A.; Birke, A.; Barz, M.;
́
Vicent, M. J. Polym. Chem. 2013, 4, 3182−3186.
(11) Zou, J.; Fan, J.; He, X.; Zhang, S.; Wang, H.; Wooley, K. L.
Macromolecules 2013, 46, 4223−4226.
(
12) Hadjichristidis, N.; Iatrou, H.; Pitsikalis, M.; Sakellariou, G.
Chem. Rev. 2009, 109, 5528−5578.
13) Vayaboury, W.; Giani, O.; Cottet, H.; Bonaric, S.; Schue,
Macromol. Chem. Phys. 2008, 209, 1628−1637.
14) Habraken, G. J. M.; Wilsens, K. H. R. M.; Koning, C. E.; Heise,
A. Polym. Chem. 2011, 2, 1322−1330.
́
15) Vayaboury, W.; Giani, O.; Cottet, H.; Deratani, A.; Schue, F.
(
́
F.
(
(
Macromol. Rapid Commun. 2004, 25, 1221−1224.
(16) Doyle, D. A. Science 1998, 280, 69−77.
9
35
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936

Macromolecules
Article
(
(
17) Wu
̈
thrich, K. Nat. Struct. Mol. Biol. 2001, 8, 923−925.
18) Wishart, D. S. Prog. Nucl. Magn. Reson. Spectrosc. 2011, 58, 62−
87.
(19) Pelton, J. T.; McLean, L. R. Anal. Biochem. 2000, 277, 167−176.
(20) Whitmore, L.; Wallace, B. A. Biopolymers 2008, 89, 392−400.
(21) Surewicz, W. K.; Mantsch, H. H.; Chapman, D. Biochemistry
1
(
9
(
993, 32, 389−394.
22) Jackson, M.; Mantsch, H. H. Crit. Rev. Biochem. Mol. 1995, 30,
5−120.
23) Goodman, M.; Verdini, A. S.; Toniolo, C.; Phillips, W. D.;
Bovey, F. A. Proc. Natl. Acad. Sci. U. S. A. 1969, 64, 444−450.
(
(
(
24) Mu
25) Kricheldorf, H.; Mu
26) Vicent Docon, M. J.; Barz, M.; Canal, F.; Conejos Sanchez, I.;
̈
ller, D.; Kricheldorf, H. Polym. Bull. 1981, 6, 101−108.
̈
ller, D. Macromolecules 1983, 16, 615−623.
Duro Castano, A. Sintesis Controlada De Poliglutamatos Con Baja
Polidispersidad Y Arquitecturas Versatiles, P201131713, 2011.
27) Mazia, D. J. Cell Biol. 1975, 66, 198−200.
28) Christie, R. J.; Nishiyama, N.; Kataoka, K. Endocrinology 2010,
́
(
(
1
51, 466−473.
29) Gottlieb, H. E.; Kotlyar, V.; Nudelman, A. J. Org. Chem. 1997,
2, 7512−7515.
30) Strohalm, M.; Hassman, M.; Kosata, B.; Kodícek, M. Rapid
Commun. Mass Spectrom. 2008, 22, 905−908.
31) Strohalm, M.; Kavan, D.; Novak, P.; Volny,
(
6
(
(
́
́
M.; Havlícek, V.
Anal. Chem. 2010, 82, 4648−4651.
(
(
32) Niedermeyer, T. H. J.; Strohalm, M. PLoS One 2012, 7, e44913.
33) Oya, M.; Katakai, R.; Nakai, H.; Iwakura, Y. Chem. Lett. 1973,
1
(
(
143−1144.
34) Miller, W. G.; Flory, P. J. J. Mol. Biol. 1966, 15, 298−314.
35) Miller, W. G.; Brant, D. A.; Flory, P. J. J. Mol. Biol. 1967, 23,
6
(
7−80.
36) Schriemer, D. C.; Li, L. Anal. Chem. 1997, 69, 4176−4183.
37) Schriemer, D. C.; Li, L. Anal. Chem. 1997, 69, 4169−4175.
38) Tiffany, M. L.; Krimm, S. Biopolymers 1969, 8, 347−359.
39) Ben-Ishai, D.; Berger, A. J. Org. Chem. 1968, 17, 1564−1570.
40) Yamakawa, H.; Fujii, M. Macromolecules 1973, 6, 407−415.
41) Tree, D. R.; Muralidhar, A.; Doyle, P. S.; Dorfman, K. D.
(
(
(
(
(
Macromolecules 2013, 46, 8369−8382.
42) Pauling, L.; Corey, R. B.; Branson, H. R. Proc. Natl. Acad. Sci. U.
S. A. 1951, 37, 205−211.
43) Ortega, A.; García de la Torre, J. J. Chem. Phys. 2003, 119,
914−9919.
(
(
9
9
36
dx.doi.org/10.1021/ma5000392 | Macromolecules 2014, 47, 928−936