Chain length dependence of apomyoglobin folding: structural evolution from misfolded sheets to native helices.

Abstract

Very little is known about how protein structure evolves during the polypeptide chain elongation that accompanies cotranslational protein folding. This in vitro model study is aimed at probing how conformational space evolves for purified N-terminal polypeptides of increasing length. These peptides are derived from the sequence of an all-alpha-helical single domain protein, Sperm whale apomyoglobin (apoMb). Even at short chain lengths, ordered structure is found. The nature of this structure is strongly chain length dependent. At relatively short lengths, a predominantly non-native beta-sheet conformation is present, and self-associated amyloid-like species are generated. As chain length increases, alpha-helix progressively takes over, and it replaces the beta-strand. The observed trends correlate with the specific fraction of solvent-accessible nonpolar surface area present at different chain lengths. The C-terminal portion of the chain plays an important role by promoting a large and cooperative overall increase in helical content and by consolidating the monomeric association state of the full-length protein. Thus, a native-like energy landscape develops late during apoMb chain elongation. This effect may provide an important driving force for chain expulsion from the ribosome and promote nearly-posttranslational folding of single domain proteins in the cell. Nature has been able to overcome the above intrinsic misfolding trends by modulating the composition of the intracellular environment. An imbalance or improper functioning by the above modulating factors during translation may play a role in misfolding-driven intracellular disorders.

Full text

Page 1

Chain Length Dependence of Apomyoglobin Folding: Structural Evolution from
Misfolded Sheets to Native Helices†
Clement C. Chow,§Charles Chow,§Vinodhkumar Raghunathan,‡Theodore J. Huppert,§Erin B. Kimball,§and
Silvia Cavagnero*,§
Department of Chemistry, UniVersity of Wisconsin-Madison, 1101 UniVersity AVenue, Madison, Wisconsin 53706, and
Department of Chemistry, UniVersity of Washington, Seattle, Washington 98195
ReceiVed December 5, 2002; ReVised Manuscript ReceiVed April 12, 2003
ABSTRACT: Very little is known about how protein structure evolves during the polypeptide chain elongation
that accompanies cotranslational protein folding. This in vitro model study is aimed at probing how
conformational space evolves for purified N-terminal polypeptides of increasing length. These peptides
are derived from the sequence of an all-R-helical single domain protein, Sperm whale apomyoglobin
(apoMb). Even at short chain lengths, ordered structure is found. The nature of this structure is strongly
chain length dependent. At relatively short lengths, a predominantly non-native ?-sheet conformation is
present, and self-associated amyloid-like species are generated. As chain length increases, R-helix
progressively takes over, and it replaces the ?-strand. The observed trends correlate with the specific
fraction of solvent-accessible nonpolar surface area present at different chain lengths. The C-terminal
portion of the chain plays an important role by promoting a large and cooperative overall increase in
helical content and by consolidating the monomeric association state of the full-length protein. Thus, a
native-like energy landscape develops late during apoMb chain elongation. This effect may provide an
important driving force for chain expulsion from the ribosome and promote nearly-posttranslational folding
of single domain proteins in the cell. Nature has been able to overcome the above intrinsic misfolding
trends by modulating the composition of the intracellular environment. An imbalance or improper
functioning by the above modulating factors during translation may play a role in misfolding-driven
intracellular disorders.
Significant progress has been made in recent years in
understanding the physical principles and the mechanisms
governing the folding of single domain proteins (1-7). Since
Anfinsen’s key experiments (8), most of the work in this
area has focused on the mechanisms by which sequence
encodes structure. Folding is usually envisaged as the
convergence of an ensemble of disordered conformations
(i.e., the unfolded state) toward a lower energy spatially
ordered compact structure. The initial conditions of these
experiments involve a chemically or photochemically gener-
ated denatured state. The process is then followed as a
function of time, once the solution has rapidly been switched
to conditions thermodynamically favoring the folded con-
formation. The physical forces acting on the polypeptide
chain on its way to the native conformation are then probed,
and experimental information is gained on the energy
landscapes of the folded protein. These in vitro experiments
assume that the pertinent species to be studied is the full-
length polypeptide.
A poorly explored aspect in protein folding is how
precisely structure develops as a function of chain length,
starting from the N-terminus and proceeding toward the
C-terminus. This is a biologically relevant question since,
within the ribosomal machinery of the cell, proteins are
vectorially synthesized from the N-terminus toward the
C-terminus. Most importantly, translation rates in both
prokaryotic and eukaryotic systems are far slower than the
typical folding rates of single domain proteins at physiologi-
cally relevant temperatures (Table 1). The above kinetic
argument suggests that there is ample chance for conforma-
tional equilibration to take place cotranslationally, before
synthesis of the full length chain has been completed.
While it is clear that other complicating factors (e.g.,
presence of chaperones, molecular crowding) may play a role
during cotranslational folding/misfolding events in the in-
tracellular environment, this work takes a model system
approach and investigates how polypeptide conformation and
energy landscapes evolve with chain elongation under native-
like conditions (i.e., in the absence of any denaturing agents).
It is important to emphasize upfront that this strategy is not
aimed at representing the complex conditions found in the
cell’s natural milieu (e.g., high viscosity and molecular
crowding, ribosome, dynamic equilibrium with cotransla-
tionally active chaperones), but rather, at establishing physi-
cal principles and expected trends of the polymer chain
encoding a protein sequence as it gets elongated. These
†This work was supported by the American Heart Association (Grant
0160492Z), the National Science Foundation (Grant 0215368), by the
Research Corporation (Research Innovation Award), and by the
Milwaukee Foundation (Shaw Scientist Award).
* To whom correspondence should be addressed. Phone: (608) 262-
5430. Fax: (608) 262-5430. E-mail: cavagnero@chem.wisc.edu.
‡University of Washington.
§University of Wisconsin.
7090
Biochemistry 2003, 42, 7090-7099
10.1021/bi0273056 CCC: $25.00© 2003 American Chemical Society
Published on Web 05/20/2003

Page 2

principles can then be exploited to devise tests that more
rigorously take the cell’s environment into account.
The chain length dependence of homopolymer and block-
copolymer model polypeptides has been studied before.
Scheraga (9, 10), Blout (11), and Loh (12) investigated the
secondary structure evolution of R-helical, polyproline II
helical, and ?-sheet polypeptides. Hodges and Litowski (13),
and Toniolo et al. (14) studied the chain length dependence
of coiled coil and 3-10 helix formation, respectively. A
different approach has been adopted by Wright and Dyson
(15-18), and Carey and Tasayco (19-21), who have
analyzed the folding of protein segments corresponding to
individual secondary structure modules of globular proteins
with the goal of learning about locally versus globally driven
secondary structure formation in folding. Fersht and co-
workers examined the behavior of polypeptide fragments of
increasing length derived from the chymotrypsin inhibitor
II (CI-2) (22, 23) and barnase (24) sequences.
The target protein of this work is apomyoglobin (apoMb),1
a well-characterized all R-helical single domain protein
(Figure 1a) whose full length chain in vitro folding pathways
have been well-studied (25-29). The A, G, and H helices
form first as part of a molten globule intermediate (Figure
2a). The additional secondary and tertiary structure modules
are subsequently formed within a cooperative step character-
ized by single exponential kinetics (25). We have examined
36, 77, and 119 amino acid long N-terminal fragments.
Lengths were designed to match ends of individual helices
in the native state (Figure 1b). All species were compared
to the 153 amino acid full length protein. Figure 2 (panels b
and c) illustrates two possible limiting models by which a
polypeptide may fold as its chain elongates from the N- to
the C-terminus. The first model (panel b) postulates that the
elongating polypeptide chain is largely dynamic and structur-
ally disordered until most of the polypeptide chain has been
synthesized. According to the alternative limiting model
(panel c), both secondary and tertiary interactions develop
in concert and are extremely close to the corresponding
structures found in the full length protein. The present study
aims at testing the above models and the existence of any
alternative folding/misfolding motifs.
MATERIALS AND METHODS
N-Terminal Fragment Preparation. The wild type apoMb
gene was obtained from Steve Sligar (University of Illinois
at Urbana Champaign) and subsequently subcloned into a
pET blue-1 vector (Novagen, Madison, WI). Briefly, the
apoMb gene insert was obtained by PCR, blunted by T4
DNA polymerase, and finally subcloned into the pETblue-1
vector at the EcoR V restriction site. The genes for the 1-77
and 1-119 apoMb N-terminal fragments were produced by
engineering stop codon point mutations into appropriate
regions of the parent plasmid (QuickChange Site-Directed
Mutagenesis Kit; Stratagene, La Jolla, CA). The gene for
the 1-36 fragment was generated by introducing a Met point
mutation in the parent gene at a position corresponding to
amino acid 36. After expression and reverse phase HPLC
1Abbreviations: apoMb, apomyoglobin; GnHCl, guanidine hydro-
chloride; CD, circular dichroism; TFA, trifluoroacetic acid; HFIP,
hexafluoro2-propanol; far-UV CD, circular dichroism in the far
ultraviolet region; FT-IR, Fourier transform infrared.
Table 1: Kinetic Parameters for the Observed Translation Rates of Prokaryotic and Eukaryotic Species Both in Vivo and in Cell-Free Systems
Translation Rates
system
translation rate
(amino acids/s)
15-20
3-4
2.5-3.5
3-5
translation time for
a 200 residue protein (s)
10-13
50-65
55-80
40-65
T (°C)
37
37
37
37
ref
prokaryotic, in vivo
eukaryotic, in vivo
prokaryotic, cell-free system
eukaryotic, cell-free system
(rabbit reticulocyte)
eukaryotic, cell-free system
(wheat germ)
67, 68
67
69
69
1-1.5130-20022-27
69
In Vitro Protein Folding Rates
folding rate (s-1)
1-106
0.1-150
3-10 × 10-5
protein type
half-life (s)
7 × 10-7-1
5 × 10-3-7
1-3 × 104
T (°C)
37-42
20-25
10
ref
single domain, R helix
single domain, ? sheet
multidomain
(tail-spike protein, luciferase)
1
5
70, 71
FIGURE 1: (a) Three-dimensional structure of Sperm whale myo-
globin (apoMb). The letters denote corresponding helices found in
the native structure. Coordinates were derived from the X-ray
structure of carbonmonoxymyoglobin (72). The image was created
using the MOLSCRIPT (73) and RASTER3D (74) softwares. (b)
Schematic representation of the N-terminal apoMb fragments
studied in this work. The length and color coding of the different
segments matches that of the helices found in myoglobin’s native
state (see panel a).
Chain Length Dependence of Protein Folding/Misfolding
Biochemistry, Vol. 42, No. 23, 2003 7091

Page 3

purification, cyanogen bromide cleavage was performed.
Expression was done in LB medium at 37 °C. Protein
expression was induced by addition of isopropyl-?-D-
thiogalactopyranoside (IPTG, 1 mM, at OD6000.5). Growth
was then continued for at least 5 hrs until saturation. Plasmids
were transformed in Escherichia coli Tuner (DE3) pLac1
competent cells (Novagen). Cell lysis and purification were
performed according to known protocols for wild-type full-
length protein (27, 30). Identity of the desired fragments was
confirmed by high-resolution time-of-flight electrospray mass
spectrometry. All masses were within (1 atomic mass unit
from the theoretical value.
Cyanogen Bromide CleaVage. Each milligram of protein
was treated with 0.1 mL of 0.1 M HCl and 2.5 mg of
cyanogen bromide. The mixture was stirred at room tem-
perature for about 14 h and then lyophilized. The dried
reaction product was dissolved in 0.5 mL of hexafluoro-2-
propanol (HFIP) and then brought to 40% v/v HFIP/0.1%
trifluoroacetic acid (TFA) water. Reverse phase HPLC
purification was then performed. Pure fractions were char-
acterized by electrospray mass spectrometry. Four different
species were generated. Briefly, the C-terminal homoserine
resulting from the cleavage at the Met site was found in both
the lactone and the unesterified open form. Each of these
species was found to either contain or lack the N-terminal
Met. Only the pure N-terminal Met-containing lactone form
was used.
Sample Preparation for Spectroscopic Measurements.
Lyophilized powders containing pure peptides/proteins were
dissolved in 10 mM sodium acetate buffer (pH 6.0), and the
solution pH was then readjusted. Samples were centrifuged
for 30 s on a tabletop centrifuge to remove traces of dust
and any other solid particles. Supernatants were then
equilibrated overnight at 4 °C and used for all the spectro-
scopic measurements below.
UV-Vis Absorption Measurements and Determination of
Peptide Concentrations. Extinction coefficients in 5 M
GnHCl at 280 nm were calculated for all species from amino
acid sequence (31). Known amounts of these samples were
then diluted in excess refolding buffer, and absorbances were
measured. These values were then utilized to determine
extinction coefficients in 10 mM sodium acetate at pH 6.0.
Light scattering components that fit a λ-4analytical expres-
sion were subtracted out, and the residual absorbances (at
280 nm) were utilized to determine sample concentration.
Measurements were done on an HP 8452A diode array
spectrophotometer.
Size Exclusion Chromatography. Size exclusion chroma-
tography was performed at 25 °C on a LCC-500 FPLC
system from Pharmacia (Piscataway, NJ) equipped with a
LKB Uvicord SII detector, operating at 226 nm. A Superdex
75 HR 10/30 column was used. The column was calibrated
with the low molecular weight calibration kit from Pharma-
cia. In addition, Aprotinin (6.5 kDa; USB, OH) was also
used as a standard. Samples were dissolved in 10 mM sodium
acetate buffer (pH 6.0) and eluted in 10 mM sodium acetate,
100 mM NaCl (pH 6.0). Concentrations of injected samples
were 34 µM (1-36 apoMb), 42 µM (1-77 apoMb), 123
µM (1-119 apoMb), and 59 µM (full length apoMb).
FIGURE 2: Schematic illustration of apoMb folding models based either on experimental evidence or on hypothetical limiting mechanisms.
(a) Experimentally determined folding pathways for full-length apomyoglobin starting from a urea unfolded state (17, 25). (b) A hypothetical
limiting mechanism supporting no defined secondary and tertiary structures formation until nearly the entire chain has been synthesized. (c)
An additional limiting folding mechanism illustrating that structure may evolve upon increasing chain length just as in the full-length
protein. (d) Proposed chain length dependence of apoMb folding/misfolding based on the experimental evidence collected in this work.
7092 Biochemistry, Vol. 42, No. 23, 2003
Chow et al.

Page 4

Circular Dichroism. CD spectra were recorded at 25.0 °C
on an Aviv spectropolarimeter (Model 202SF) in 10 mM
sodium acetate buffer (pH 6.0). Either 0.1 or 0.2 cm path
length cuvettes were used. Data were collected as single
scans with a 1 nm bandwidth, 1 nm step size, and a 100 ms
time constant. Data averaging time was 20 s (60 s was used
for <15 µM samples). Sample concentrations ranged from
9 to 12 µM.
Fluorescence. Trp fluorescence emission data were ac-
quired on a Hitachi F-4500 spectrofluorometer. Samples (5
µM) were prepared in 10 mM sodium acetate buffer at pH
6.0. Emission spectra were recorded between 295 and 450
nm (λex: 280 nm). Slit widths for both excitation and
emission were 5 nm. Data were recorded at 25 °C.
Fluorescence-detected GnHCl titrations were performed on
solutions containing different denaturant concentrations.
Accurate GnHCl concentrations were determined by refrac-
tometry (32). Fluorescence emission titration data were
reported as total peak area between 295 and 450 mn. Curve
fitting of the fluorescence-detected urea titration data was
done according to an equation for an equilibrium three-state
unfolding model. A complete derivation of this equation is
provided as Supporting Information.
Infrared Spectroscopy. Fourier transform infrared (FT-IR)
spectra were recorded on a Nicolet 740 FT-IR spectropho-
tometer equipped with a TGS detector and dry air purging.
256 scans were acquired for each sample with 2 cm-1
resolution. A CaF2cell with a 50 µm Teflon spacer was used.
The presence of traces of trifluoroacetic acid is known to
severely interfere with FT-IR data collection in the amide I
region (33, 34). Therefore, the samples were dissolved in
10 mM HCl and lyophilized to remove traces of residual
TFA from HPLC purification, following published protocols
(35, 36). The powders were then dissolved in D2O and
incubated overnight at 4 °C followed by lyophilization.
Samples were then dissolved in D2O again, and pH*
(uncorrected electrode reading) was then adjusted to 5.6 prior
to data collection. Data were recorded at 25 °C.
ThioflaVin T Assay. Samples were prepared by mixing 1.92
mL of peptide solution at a 10 µM concentration with 80
µL of 100 µM Thioflavin T in 10 mM sodium acetate buffer
adjusted at the appropriate pH (37). Emission spectra were
recorded at room temperature on the spectrofluorometer by
Molecular Kinetics (equipped with MOS-250 optics and data
acquisition interface) over the 450-520 nm wavelength
range. Excitation wavelength was set to 450 nm. Excitation
and emission slit widths were set to 5 and 20 nm, respec-
tively.
Hydropathy and Nonpolar Surface Area Calculations.
Cumulative hydropathies were calculated according to Kyte
and Doolittle (38). The SURFACE RACER program (39)
was used for the non polar surface area calculations of Figure
10b. In all cases, each chain length was generated in the
extended chain form by the Biopolymer module of Insight
II 2000 (Accelrys).
RESULTS
Far-UV Circular Dichroism Spectroscopy. Far-UV circular
dichroism (CD) analysis of the N-terminal fragments (Figure
3) reveals that ordered structure is present in all species and
random coil population is very low at all chain lengths even
for the shortest fragments. The full length protein contains
predominantly helical structure. However, as chain length
decreases the doublet typical of R-helix is progressively
replaced by a singlet centered at 217 nm, indicating a
complete change in secondary structure content. The CD data
deconvolution program CONTINLL (40, 41) has been used
to gain insights into the evolution of secondary structure as
a function of chain length (Table 2). This program cannot
be relied upon for quantitative data analysis because of the
presence of putative ?-strand and(or) turn, which are poorly
represented in the experimental data sets used by the
algorithm. Nonetheless, CONTINLL is a very valuable tool
for qualitative data assessment. The program output shows
that the fraction of random coil disordered conformation is
low at all chain lengths, and as polypeptide length increases,
?-strand (or turn) content decreases to be progressively
replaced by helical conformation. It is known that CD
singlets centered at around 217 nm may be due to either
?-turn or ?-strand.
As seen in Figure 3 and Table 2, a significant fraction of
the helicity found in full-length apomyoglobin is acquired
upon transition from the 1-119 fragment to the full length
protein. This suggests that the last few amino acids play an
important role in consolidating native-like structure and in
reshaping the polypeptide energy landscape so that it assumes
FIGURE 3: Far-UV circular dichroism (CD) spectra of N-terminal
apoMb fragments. The full length protein displays a doublet
characteristic of R-helices (minima at 208 and 222 nm, respectively).
The doublet progressively turns into a singlet centered at 217 nm
as chain length decreases.
Table 2: Deconvolution of Far-UV CD Data by the program
CONTINLLa
Deconvolution of Far-UV CD Data
R
?
0.040.40
0.160.29
0.300.18
0.500.04
apoMb N-terminal fragment
(1-36) apoMb
(1-77) apoMb
(1-119) apoMb
full-length (1-153) apoMb
? + t
0.62
0.51
0.40
0.28
rc
0.34
0.33
0.30
0.21
aThe symbols R, ?, ? + t, and rc denote R-helix, ?-strand, ?-strand
+ turn, and random coil, respectively.
Chain Length Dependence of Protein Folding/Misfolding
Biochemistry, Vol. 42, No. 23, 2003 7093

Page 5

native-like features. To further investigate this evidence, we
have replotted the far-UV CD data of the 1-119 fragment
and full-length apoMb in Figure 4. This figure contains a
comparison between the CD spectra of three species: full-
length apoMb, 1-119 apoMb, and the calculated spectrum
resulting from addition of the far-UV CD data for 1-119
apoMb and H helix peptide (17). The H helix peptide is about
25% intrinsically helical, and it comprises only the residues
belonging to the apoMb H helix. Chain lengths have been
readjusted to properly calculate molar ellipticity values on a
per residue basis. The [1-119apoMb-H helix] CD spectrum
is representative of an imaginary polypeptide containing most
of the apoMb amino acids but whose nature is such that the
H helix is not allowed to interact with the remaining
C-terminal portion of the protein. Comparison of this
spectrum and that of the full length protein shows that nearly
half of the overall helicity of full length apoMb is gained as
a result of the interaction between the last portion of the
chain and the preexisting fragment. This indicates that the
C-terminal portion of the amino acid chain plays an important
role in promoting cooperativity in secondary structure
formation.
Fourier Transform Infrared Spectroscopy. To discriminate
whether the nonfull length polypeptides contain either turn
or sheet-like secondary structure (see CD data above), we
have performed additional studies by Fourier transform
infrared (FT-IR) spectroscopy in the amide I region. The
data are presented in Figure 5. As chain length decreases,
FT-IR reveals, in full agreement with the CD data, that the
helical conformation of the full length species gets converted
into a different kind of secondary structure. The predomi-
nantly ?-strand nature of the secondary structure found at
short chain lengths is established by wavelength maxima of
the observed spectral features, particularly the peaks centered
at about 1621 and 1683 cm-1. The second derivative data
provided as Supporting Information confirm these results.
The peak centered at 1621 cm-1is also strongly suggestive
of extended ?-sheet content (42). In summary, the data show
that secondary structure progressively evolves from ?-sheet
to R-helix as chain length increases. The ?-sheet secondary
structure acquired by the polypeptide at shorter chain lengths
is to be regarded as non-native and misfolded since no
?-strand is present in the native full length protein. As further
discussed in the next section, these results are in interesting
contrast with both the full length apoMb folding pathways
and the chain length-dependent secondary structure acquired
by barnase and chymotrypsin inhibitor-2 (CI2).
Fluorescence Emission Spectroscopy. The Tryptophan
fluorescence emission spectra of the N-terminal fragments
and full length apoMb are shown in Figure 6. All samples
contain the same number of fluorophores (i.e., Trp 7 and
Trp 14). These are the only two Tryptophans present in wild-
type apoMb. There are large variations in emission intensity
for the different species, with a notable hyperfluorescence
exhibited by the 1-119 fragment. In all cases, experiments
have been performed in duplicate, and concentrations have
been carefully measured. The apoMb fluorescence is known
to result from the delicate balance between the degree of
solvent exposure of both fluorophores and the proximity of
Trp 7 to Lys 79. This positively charged amino acid exerts
some intramolecular quenching. The fragments and full
FIGURE 4: Superposition of far-UV CD spectra of full length apoMb
(1-153 apoMb), its N-terminal fragment lacking amino acids
corresponding to the last C-terminal helix (1-119 apoMb), and
the peptide corresponding to the apoMb C-terminal H helix (17).
The figure also reports the spectrum of an imaginary polypeptide
resulting from the summation of the spectra of the C-terminal helix
peptide and (1-119) apoMb. The mean residue ellipticities for this
hypothetical species have been recalculated to take the increased
chain length into account.
FIGURE 5: Fourier transform infrared (FTIR) absorbance spectra
of N-terminal apoMb fragments. The data show a progressive
maximum intensity red shift (from 1648 to 1621 cm-1) as chain
length decreases. Sample concentrations range from 0.8 to 7.4 mg
mL-1. Data have been concentration-corrected to reflect individual
absorption intensities at an identical molar concentration. Solution
buffer is 10 mM sodium acetate at pH 6.0.
FIGURE 6: Trp fluorescence emission spectra of N-terminal apoMb
fragments. (λex) 280 nm).
7094 Biochemistry, Vol. 42, No. 23, 2003
Chow et al.