Biochimica et Biophysica Acta 1824 (2012) 1144–1150

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

journal homepage: www.elsevier.com/locate/bbapap

Analysis of amino acid contributions to protein solubility using short peptide tags
fused to a simplified BPTI variant☆
Mohammad Monirul Islam 1, Monsur A. Khan, Yutaka Kuroda ⁎
Dept. of Biotechnology and Life Science, Tokyo University of Agriculture and Technology, 2-24-16 Nakamachi, Koganei-shi, Tokyo 184-8588, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Protein solubility is usually characterized in terms of a hydrophobicity scale, which refers to the free energy of
Received 9 March 2012 transfer of a molecule from an aqueous to a nonpolar solution and is not a “solubility propensity scale” per se.
Received in revised form 25 May 2012 Using a “host–guest” approach, we measured the effects of short poly-amino-acid tags (guests) on the solubility
Accepted 13 June 2012
of a host protein, a simplified bovine pancreatic trypsin inhibitor (BPTI), to which they were fused at the
Available online 20 June 2012
C-terminus. We analyzed 10 amino acid types, representing the full range of biophysical properties (acidic,
Keywords:
basic, polar, and hydrophobic). As anticipated, positively charged residues significantly increased the solubility
Protein aggregation of the model protein, at both pH 4.7 and 7.7, whereas very hydrophobic poly-Ile markedly reduced the solubility
Solubility propensity scale of BPTI. Poly-Asp and poly-Glu barely affected BPTI solubility at pH 4.7, but induced an eight to ten-fold increase
Thermodynamics at pH 7.7, attributable to the ionization of their side chains. Although Pro is the most soluble amino acid, poly-Pro
Protein stability did not affect the protein's solubility. The effects of the other tags on BPTI solubility ranged from none to an eight-
Polypeptide fold increase. To ensure that the measured solubility values were context independent and could provide a “sol-
ubility propensity scale”, we confirmed that the tags remained independent of the structure, thermal stability,
and biochemical activity of the host protein. These observations suggest that this approach is valuable for mea-
suring the solubility propensities of amino acids, which could eventually allow the calculation of a polypeptide's
relative solubility from its amino acid sequence.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction on the amino acid sequence of the protein, more accurately than
using the present hydrophobicity-derived scales.
Protein solubility is an important, but often overlooked, property. The “host–guest” approach, as applied to peptides and proteins,
It is associated with several human diseases [1], and in vitro protein provides an experimental method for measuring the tendencies of
solubility is becoming an important issue in several areas of biotech- amino acids to adopt certain states or conformations. It was used by
nology, including the production of protein pharmaceuticals [2,3]. Scheraga's group in the 1960s [9], and later by Baldwin's group [10],
Solubility is usually estimated in terms of a hydrophobicity-derived to determine the helix propensities of amino acids. It was also used
scale. However, hydrophobicity is not an actual measure of solubility to probe the structural/biochemical context dependencies of helix
propensity, because it refers to the transfer of amino acids from an [11], or β-sheet formation within a protein [12]. However, a host–
aqueous to a nonpolar solution [4,5]. Biophysical attempts to rational- guest-like approach has rarely been used to analyze the contributions
ize the effects of amino acid mutations on protein solubility have of amino acids to protein solubility, and when it has been applied, the
been limited by various intrinsic and extrinsic factors that influence results were confounded by effects arising from the surrounding
protein solubility [6–8]. A genuine solubility propensity scale for structural/biochemical environments of the host protein [13].
amino acids might allow the prediction of protein solubility based The solubilities of amino acids range widely, from a few score mi-
cromolar to several hundred millimolar, which presents a technical
hurdle, among many others, for the construction of a solubility pro-
Abbreviations: Simplified BPTI, A single-disulfide-bonded bovine pancreatic trypsin pensity scale, because such a wide range of values is difficult to mea-
inhibitor variant whose sequence was simplified by multiple alanine replacement;
sure accurately with a single protocol. Peptide sequences consisting
BPTI-19, simplified BPTI containing 19 alanines out of 58 residues; Hydropathy, Kyte
Doolittle hydropathy scale; Hydrophobicity, dislike of water in a general sense of a single amino acid type (poly-amino-acid peptides) can be useful
☆ Data deposition: The coordinates and structure factors of BPTI-19 and BPTI-19-C5H in partly overcoming this problem, because they tend to amplify the
have been deposited in the Protein Data Bank under the PDB IDs: 3AUB and 3AUE adhesive, aggregation, polymerization, and solubility properties of
respectively. the amino acids.
⁎ Corresponding author. Tel./fax: + 81 42 388 7794.
E-mail address: [email protected] (Y. Kuroda).
Recently, we used the biochemical properties of charged amino acids
1
Present address: Dept. of Biochemistry and Molecular Biology, University of to improve protein solubility by fusing short poly-Lys or poly-Arg tags to
Chittagong, Chittagong-4331, Bangladesh. a bovine pancreatic trypsin inhibitor (BPTI) variant, which increased its

1570-9639/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbapap.2012.06.005
 M.M. Islam et al. / Biochimica et Biophysica Acta 1824 (2012) 1144–1150 1145

a 1 10 20 30 40 50

Wild type RPDFCLEPPY TGPGKARIIR YFYNAKAGLC QTFVYGGCRA KRNNFKSAED CMRTCGGA

BPTI-19-C5X RPAFCLEPPY AGPGKARIIR YFYNAAAGAA QAFVYGGVRA KRNNFASAAD ALAACAAA-GGXXXXX

b c

Fig. 1. Sequence and structure of BPTI variants. (a) Sequence of the wild type BPTI and the poly-amino-acid tagged BPTI-19 variants. Alanine is in green, positively and negatively charged
residues are in blue and red, respectively. The poly-amino-acid tag residues are underlined. Two Gly were added as spacer residues, followed by five amino acid (X) of the same type.
Ribbon and surface representation of (b) BPTI-19 and (c) BPTI-19-C5H, a poly His tagged variant. The poly-His tag was partially visible, but fully extended pointing outward the BPTI-
19's molecular surface. The electron densities of the other poly-amino-acid tags were mostly invisible, indicating their highly flexible conformation.

solubility by over six-fold without altering its NMR spectrum or activity those anticipated from hydrophobicity and hydrophilicity scales. How-
[14]. We therefore reasoned that poly-amino-acid peptide tags, which ever, we also observed clear discrepancies, which suggest that the de-
can significantly modify a protein's solubility, could be used to determine velopment of a genuine solubility propensity scale should provide
the contribution of individual amino acids to protein solubility. Here, we new insight into the biophysics of protein solubility.
report a host–guest-like approach in which we used poly-amino-acid
tags to investigate the effects of amino acid mutations/additions on pro-
tein solubility. We chose 10 amino acid types representing the full range 2. Results and discussion
of biophysical properties (acidic, basic, polar, and hydrophobic). To this
end, we fused short poly-amino-acid tags (guests), consisting of one of 2.1. Design of poly-amino-acid-tagged BPTI variants
the selected amino acids, to the C-terminus of our host protein, a simpli-
fied BPTI variant [15,16], in which most of the surface residues were ala- In this study, we used a simplified BPTI in which most of the sur-
nines, which we expected would minimize the interactions between the face residues were alanines [15,16], as a host protein. This is because
poly-amino-acid tag and the host protein. We ensured that the measured alanine is regarded as a biochemically and structurally neutral residue
contributions of the amino acids to the protein's solubility were context [17,18], and because alanine on the protein's surface is thus anticipat-
independent by confirming that the peptide tag did not affect the struc- ed to minimize interactions between the host protein and the tag
tural, functional, or thermodynamic properties of the host protein. The peptide (seven of the ten positive residues and two of the four nega-
solubilization effects of amino acids were generally consistent with tive residues in the wild type are left in BPTI-19; Fig. 1).

Table 1
Solubility of the poly-amino-acid tagged BPTI variants.

Mutants Protein solubilitya(mg/ml) StabilitybTm pIc Activityd Mutants Protein solubilitya (mg/ml) Stabilityb pIc Activityd
(°C) (%) Tm (°C) (%)
pH4.7 pH7.7 pH4.7 pH7.7

BPTI-19 1.14 ± 0.05 1.43 ± 0.15 54.04 9.7 100 C5P 1.69 ± 0.71 1.95 ± 0.38 52.44 9.7 100
C2G 1.21 ± 0.25 1.39 ± 0.16 54.85 9.7 100 C5If 0.11 ± 0.11 0.07 ± 0.03 ND 9.7 100
C5S 3.81 ± 0.52 4.57 ± 0.29 53.74 9.7 100 C5D 3.84 ± 0.52 14.20 ± 1.21 54.29 6.6 100
C5N 9.56 ± 1.57 15.37 ± 1.04e 52.44 9.7 100 C5E 1.54 ± 0.21 11.02 ± 0.80 52.51 6.6 100
C5Q 4.48 ± 0.84 4.71 ± 0.74 54.22 9.7 100 C5K 15.67 ± 0.71e 16.98 ± 2.10e 50.74 10.2 100
C5H 7.90 ± 0.57 0.94 ± 0.33 51.51 9.7 100 C5R 8.07 ± 0.84 8.92 ± 1.17 53.59 11.2 100

The melting temperature of C5I could not be measured as C5I precipitated in the presence of 1.3 M ammonium sulfate. ND: stands for not determined.
a
Solubility was measured in the presence of 1.3 M ammonium sulfate in 50 mM acetate (pH4.7) and 50 mM Tris–HCl (pH7.7), and with an initial protein concentration of 20 mg/ml.
All measurements were replicated more than three times.
b
Thermal stabilities were determined in 20 mM Acetate Buffer of pH4.7 by circular dichroism (CD) in the presence of 1.0 M ammonium sulfate. Tm stands for melting temperature.
c
pI (iso-electric Point) was calculated using PROTEIN CALCULATOR v3.3 (http://www.scripps.edu/~cdputnam/protcalc.html).
d
Trypsin inhibitory activity was measured by monitoring the hydrolysis of N-benzoyl-D,L-arginine-p-nitroanilide (BAPA) at equimolar concentrations of BPTIs and trypsin (280 nM).
BPTI variants and trypsin were incubated in buffer for 5 min before addition of BAPA, and the hydrolysis of BAPA was monitored by measuring absorbance changes over 5 min at 405 nm.
e
Equilibrium might not have been reached, and this value might increase when measured with initial protein concentrations higher than 20 mg/ml.
f
The concentration of C5I was determined by measuring the amount of protein that precipitated, because the concentration in the supernatant was too low for direct measurement.
1146 M.M. Islam et al. / Biochimica et Biophysica Acta 1824 (2012) 1144–1150

BPTI-19 was tagged with different poly-amino-acid peptides com-

posed of a single amino acid type (Table 1). We designed 10 poly-
amino-acid-tagged variants, which we named according to the number
a and type of amino acid included [14]. For example, C5R denotes a BPTI-
18
19 variant with a five-Arg-residue tag added to its C-terminus. The poly-
amino-acid peptide tags could be added to either terminals, as the
15 solubilization effect was largely independent on the site of addition [14].
Protein Solubility (mg/ml)

The 10 amino acid types were chosen to include the full range of
amino acids' biochemical properties because of their occurrence in
12
natural and synthetic poly-amino-acid sequences. Polar (N, Q, S), hy-
drophobic (I), positively charged (R, K), and negatively charged (D, E)
9 amino acids, together with Pro (P), were chosen as representative
amino acids (Table 1). We also included His because His tags are com-
6 monly used in molecular biology, as well as Gln because of its relation-
ship to Huntington's disease [19]. We also added two Gly residues to
all of the variants as a spacer between the poly-amino-acid tag and
3 BPTI to ensure the flexibility of the poly-amino-acid tag and to reduce
any putative interaction between the tag and the molecular surface of
0 BPTI. A BPTI variant with two Gly residues added to its C-terminus
C2G C5S C5N C5Q C5H C5P C5I C5D C5E C5K C5R (C2G)was used as the reference molecule.

b 18 2.2. Effects of poly-amino-acid tags on protein solubility

Protein Solubility (mg/ml) at pH7.7

We measured protein solubility as the maximum protein concen-

15 tration in the supernatant after centrifugation at 20,000 × g for 20 min
C5N
C5D in 50 mM acetate buffer (pH 4.7) or 50 mM Tris–HCl buffer (pH 7.7)
12 in the presence of 1.3 or 1.5 M ammonium sulfate. Poly-Lys increased
protein solubility twelve to thirteen times (Fig. 2), similar to our pre-
C5E vious observation [14]. The solubility increase caused by poly-Arg
9
was more modest (six-fold). In contrast, the negatively charged
poly-Asp and poly-Glu tags barely affected protein solubility at pH
6 4.7 (Table 1; Fig. 2a). The polar (N, Q, S) and poly-His tags increased
protein solubility by three to eight-fold. As expected, the addition of
3 five Ile residues significantly reduced the solubility of BPTI. However,
C5H the addition of the very soluble Pro altered the solubility of BPTI only
slightly (Table 1 and Fig. 2a–b).
0 The solubilization effects of most poly-amino-acid tags were pH
0 3 6 9 12 15 18
independent (Fig. 2c), and those of the poly-Ile, -Pro, -Gln, -Arg, and
Protein Solubility (mg/ml) at pH4.7
-Ser tags were essentially the same at pH 4.7 and pH 7.7. However,
c 18
the solubilization effects of poly-Asp and poly-Glu increased dramat-
ically, by four to ten-fold, at pHs higher than 7.1 (Fig. 2). The pH de-
pendence of the effects of the Asp and Glu tags on protein solubility
15 is attributed to the tendency of proteins to aggregate at pHs close to
Protein Solubility (mg/ml)

their isoelectric points (calculated pI of BPTI, 6.62), whereas at high

pH, the ionized Asp and Glu side chains provide a net negative charge,
12
which contributes to an increase in the overall protein solubility
(Table 1; Fig. 2b). Conversely, the poly-His-tag dramatically reduced
9 the protein's solubility when the pH was increased from pH 4.7 to
pH 7.7 (Fig. 2a and b). This is because the typical residue pKa of free
His is approximately 7.0, at which both the acidic and basic forms co-
6
exist; a change in the pH will affect protonation state of the imidazole
ring, and thereby the solubility of the protein. The poly-Asn tag in-
3 creased the solubility of the protein when the pH was increased, but
no such effect was observed for poly-Gln, whose solubilization effect
0 was pH independent.
pH 4.7 pH 6.2 pH 7.1 pH 7.5 pH 7.7 Our observations were generally consistent with the solubility
changes anticipated from their hydrophobicities [4,5], and the solu-
Fig. 2. Effect of poly-amino-acid tags on protein solubility. (a) Protein solubility in bilities of the individual amino acids [20]. For example, peptide tags
50 mM acetate, pH4.7 (□) and in 50 mM TrisHCl, pH7.7 (■) in the presence of 1.3 M
composed of charged or hydrophilic residues increased the solubility
ammonium sulfate at 25 °C. (b) Correlation plot of protein solubility as function of
pH (pH4.7 vs pH7.7). (c) pH dependency of protein solubility. Measurements were of the protein, as anticipated (Fig. 3). However, there were several
conducted in 50 mM acetate (pH4.7 and 6.2), phosphate (pH6.2) and TrisHCl (pH7.1, striking discrepancies that cannot be overlooked when predicting
7.5 and 7.7) buffers (■: BPTI-19; ▽: C5S; ▲: C5N; ○: C5Q; : C5D; ♦: C5R). The pH the solubility of a polypeptide from its amino acid sequence. First, in
values refer to that of the buffer (in the presence of amonium sulfate). All measure- quantitative terms, the solubilization effects of the tags composed of
ments were started with the same initial protein concentration. Values reported here
were measured independently from those reported in Table 1, and the largest discrep-
charged or polar residues did not correlate well with either their hy-
ancy was observed for C5D at pH 4.7, 20 mg/ml initial concentration (3.84 mg/ml vs dropathy [4] and/or hydrophilicity [21] scales or the solubilities of the
3.48 mg/ml). individual amino acids [20] (Fig. 3b), as demonstrated by the near-
 M.M. Islam et al. / Biochimica et Biophysica Acta 1824 (2012) 1144–1150 1147

a 10

Protein Solubility (mg/ml)

6

0
0 10 20 30 40
Initial Protein Concentration (mg/ml)

b 18 300 12
10

Hydropathy and Hydrophilicity

Amino Acid Solubility (g/kg)
15 250
Protein Solubility (mg/ml)

8
6
12 200
4
2
9 150
0
-2
6 100
-4
3 50 -6
-8
0 0 -10
C2G C5S C5N C5Q C5H C5P C5I C5D C5E C5K C5R

Fig. 3. Dependence of protein solubility on initial protein concentration and correlation with biophysical properties of individual amino acids. (a) Dependency of protein solubility on the
initial protein concentration, at pH4.7 in the presence of 1.3 M ammonium sulfate. In most cases, the final protein solubility in the supernatant became stable (consistent and independent
from the initial concentration) at an initial concentration in the 10–40 mg/ml range (■: BPTI-19; □: C2G; ▽: C5S; ●: C5H; ♦: C5R; *: C5P; ○: C5Q). Values reported here were measured
independently from those reported in Table 1, and discrepancies up to 10% were observed. The largest difference was for C5H at 20 mg/ml initial concentration (7.90 mg/ml vs 7.19 mg/
ml). (b) Comparative view of amino acid hydropathy (Δ), [4] hydrophilicity (□) [21], individual amino acid solubility (●) [20] and protein solubility (pH4.7: □; pH7.7: ■).

zero correlation coefficients between these scales (nor did the other and solid phases, with high reproducibility. Ammonium sulfate also al-
scales correlate with one another). Furthermore, hydropathy (as well leviated a peculiar phenomenon whereby the final protein solubility
as hydrophobicity) is usually a pH-independent scale, but the data for depended on the initial amount of dissolved protein [13,22]. Our mea-
Asp, Glu, His, and to a lesser extent Asn clearly indicate or confirm surements in the absence of salt or at low salt concentrations confirmed
that pH is an important factor influencing protein solubility. The Pro this observation. However, the solubility limit became almost indepen-
tag barely affected protein solubility (Fig. 2a), which is consistent with dent from the initial amount of protein when ammonium sulfate was
its low hydrophobicity, but is in sharp contrast to its solubility, which added to the solution, and the solubility limit was reached consistently
is reported to be as high as 1600 g/L [20]. These observations demon- (Fig. 2a; Supplemental information SI-1c). We speculate that the solu-
strate that the actual contribution made by an amino acid to protein sol- bility limit dependence on the initial protein concentration at low salt
ubility can differ from the effects anticipated based on either concentrations might be related to extremely slow kinetics of aggrega-
hydrophobicity or the solubility of individual amino acids. tion or to residual water and/or buffer molecules contained in the ly-
ophilized protein powder. In either case, these effects were minimized
2.3. Ammonium salts and reliability and reproducibility of the measurements by the addition of ammonium sulfate, ensuring highly consistent and
reproducible measurements. All the solubility measurements reported
Although our protocol allows us to measure, in principle, protein in this study were repeated 3–5 times using different initial protein con-
solubility in a pure aqueous solution, the addition of ammonium sul- centrations, and the final solubility limits of all the variants were within
fate was essential to circumvent technical difficulties associated with 10% of one another (Figs. 2 and 3; Supplemental Information SI-1).
determining protein solubility at very high protein concentrations
(Supplemental information, SI-1). For some tagged BPTI variants, 2.4. Protein stability and independence of the tag peptide from the host
measuring their solubilities in the absence of salt required extremely protein
high protein concentrations, which not only necessitated a large
amount of purified protein, but more importantly, also caused poor Protein solubility dramatically decreases when the protein unfolds,
phase separation or even gel formation. The addition of ammonium and it is therefore important to ensure that the host protein remains
sulfate helped to achieve a well-defined separation of the aqueous fully folded during the solubility measurements. Under all conditions
1148 M.M. Islam et al. / Biochimica et Biophysica Acta 1824 (2012) 1144–1150

(0.0–1.5 M ammonium sulfate at pH 4.7–7.7), all the variants exhibited

a two-state reversible thermal denaturation curve when probed with
circular dichroism (Supplemental information SI-2).
a 1.00
The stability of all variants showed a small increase with increasing
ammonium sulfate concentrations, attributable to a salting-in/salting-
out effect or to small pH changes upon addition of ammonium sulfate
Fraction unfolded

0.75
salt. The melting temperatures (Tm) of the poly-amino-acid-tagged var-
iants were identical to that of the reference C2G variant, both in the
0.50 presence and absence of ammonium sulfate, within an experimental
error of ±0.5–1 °C (Fig. 4a–d). Moreover, in the presence of 1.5 M am-
monium sulfate, the thermal folding/unfolding of the tagged variants
0.25 was almost perfectly reversible (Supplemental information SI-2).
These observations suggest that the residues of the poly-amino-acid
tag did not interact with the surface of the BPTI variant, and that all
0.00 the variants were fully folded at 25 °C, the temperature at which solu-
20 30 40 50 60 70
bility was measured.
Temperature (degC)
2.5. Structural and functional independence
b 1.00
The crystal structures of BPTI tagged with the poly-amino-acid tags
provided a reliable and direct assessment of the minimal interactions
that occur between the poly-amino-acid tags and the molecular surface
Fraction unfolded

0.75
of the host BPTI protein (Fig. 1). In all the poly-amino-acid-tagged vari-
ants, both the backbone and side-chain structures of the host protein
0.50
were almost perfectly retained, with an average backbone deviation of
b0.3 Å (BPTI-19, C5S, C5N, C5H, C5P, C3E, C3R, and C3K; manuscript in
preparation).
0.25 The trypsin inhibitory activities of all of the tagged variants were
completely identical to that of the untagged BPTI (Table 1). Because
the tag is located close to the BPTI–trypsin binding site, the unaltered
0.00 trypsin inhibitory activity confirms that the poly-amino-acid tags did
not interact with the molecular surface of the host protein.
20 30 40 50 60 70
Temperature (degC)
2.6. Insight into polypeptide solubility
c 1.00 A major difficulty in predicting protein solubility is that the effects of
several factors are intertwined. Moreover, technical hurdles originating
from high protein concentrations must be overcome. Our strategy for
Fraction unfolded

0.75
circumventing these factors was to take advantage of the near-neutral
biochemical characteristics of our alanine-simplified BPTI, and the ten-
dency of the poly-amino-acid tags to amplify the effects of the particular
0.50
amino acid on protein solubility. Another advantage of this strategy is
that the poly-amino-acid tags are fully independent of the host BPTI
0.25 protein, and we can therefore expect to measure context-independent
solubility (i.e., independent of the surrounding protein environment).
Although high concentrations of ammonium sulfate are well known
0.00 to reduce protein solubility, we consider that the relative amino acid sol-
ubilities were not strongly affected by its presence. Although our present
20 30 40 50 60 70
results were obtained in the presence of ammonium sulfate, they never-
Temperature (degC)
theless provide some useful insights into how amino acid composition
influences the relative solubility of a polypeptide. In particular, the be-
d 60 havior of the Asp and Glu tags provides an explanation of a protein's ten-
dency to aggregate at pHs close to its isoelectric point (pI), which is not
explained by a pH-independent hydrophobicity scale. According to our
Melting teperature (degC)

measurements, the solubility of acidic proteins, which contain many

50

Fig. 4. Thermal stability of poly-amino-acid tagged variants in the presence of ammonium

sulfate in 50 mM TrisHCl (pH7.7) using circular dichroism. (a) Untagged BPTI,
(b) poly-Asp and (c) poly-Lys variants are shown. All measurements were conducted
40 at a protein concentration of 5 μM with 0.00 M (□), 0.50 M (○), 0.75 M (△), and
1.00 M (▽) ammonium sulfate. Symbols represent the raw data while continuous
lines represent the fitted data. Dotted lines represent thermal denaturation curves in the
absence of ammonium sulfate. Similar patterns of protein stability were observed at
pH 4.7 in 50 mM acetate buffer (Supplemental information, SI-2b). (d) Melting tempera-
30 tures calculated from the above denaturation curves (Ammonium sulfate concentration:
BPTI-19C2G C5S C5N C5Q C5H C5P C5D C5E C5K C5R ■: 0.0 M; ■: 0.5 M and □: 1 M).
 M.M. Islam et al. / Biochimica et Biophysica Acta 1824 (2012) 1144–1150 1149

Glu, Asp, and Asn residues, should increase at high pH, but the solubility described [14,15]. Thermal denaturation experiments were conducted
of basic proteins would be little affected in the pH range 4.7–7.7. Accu- at a scan rate of 1 °C/min in the temperature range of 5–75 °C. The re-
rate predictions of solubility will require the determination of the rela- versibility of the thermal denaturation was confirmed by cooling the
tive solubilities at various temperatures and under different conditions, sample to 5 °C followed by a re-heating to the highest temperature.
including low salt concentrations. Furthermore, the contribution of
cross terms, which originate from mixing two types of amino acids in a 4.4. Crystallization and structure determination
tag, must be assessed. Such corrections could, in principle, be readily ac-
commodated in a solubility prediction scheme. Crystals of BPTI variants were grown at 20 °C using the hanging drop
vapor diffusion technique. The X-ray diffraction data were recorded
3. Conclusions from single crystals using a synchrotron beam line at the Photon Factory
(PF, Tsukuba, Japan). The coordinates and structure factors of BPTI-19
Although solubility is an important property of proteins, few, if any, and its His-tagged variant (BPTI-19-C5H) have been deposited in the
systematic studies have been conducted into how each of the 20 natural Protein Data Bank under the PDB IDs 3AUB and 3AUE, respectively.
amino acids modulates protein solubility. Several factors may have ham- Supplementary data to this article can be found online at http://
pered the construction of a proper “solubility propensity scale”: the dx.doi.org/10.1016/j.bbapap.2012.06.005.
strongly intertwined nature of the intrinsic and extrinsic factors that in-
fluence protein solubility, the wide range of solubilities among amino
Acknowledgements
acids, and technical difficulties associated with reliably measuring high
protein concentrations. However, this study indicates that poly-amino-
We thank Dr. Atsushi Kato, Mr. Yuki Ohta, and members of the
acid peptide tags fused to a simplified host protein can provide a valuable
Kuroda's lab for discussion, help with plasmid preparation, protein ex-
method for measuring the effects of amino acids on protein solubility, in a
pression and preliminary characterization. This work was supported
context-independent manner. Such measurements could yield a genuine
by a JSPS grant-in-aid for scientific research (21300110), a special prior-
solubility propensity scale, which could eventually allow the calculation
ity research area grant (21107505), a Kyowa-Hakko research grant, and
of the relative solubility of a polypeptide from its amino acid sequence.
a Noda foundation research grant to YK. MMI was supported by a JSPS
post-doctoral fellowship. X-ray diffraction data were recorded at the
4. Materials and methods
Photon Factory (PF, Tsukuba, Japan).

4.1. Mutant design, expression and purification

References
All BPTI variants were constructed in pMMHa vectors as previously [1] C.A. Ross, M.A. Poirier, Protein aggregation and neurodegenerative disease, Nat.
described [15]. DNA sequences corresponding to the poly-amino-acid Med. 10 (2004) S10–S17 (Suppl.).
tags were added to the C-terminus of the host BPTI using QuikChange [2] S.B. Fowler, S. Poon, R. Muff, F. Chiti, C.M. Dobson, J. Zurdo, Rational design of
aggregation-resistant bioactive peptides: reengineering human calcitonin, Proc.
site directed mutagenesis (Stratagene) and the plasmid sequences Natl. Acad. Sci. U. S. A. 102 (29) (2005) 10105–10110.
were confirmed by DNA sequencing. All poly-amino-acid tagged variants [3] M.S. Ricci, D.N. Brems, Common structural stability properties of 4-helical bundle
were over-expressed in Escherichia coli JM109(DE3)pLysS cells, and puri- cytokines: possible physiological and pharmaceutical consequences, Curr. Pharm.
Des. 10 (31) (2004) 3901–3911.
fied by reverse phase HPLC, as previously described [23]. Protein identi- [4] J. Kyte, R.F. Dolittle, A simple method for displaying the hydropathic character of a
ties were confirmed by ESI-TOF mass spectroscopy, and purified proteins protein, J. Mol. Biol. 157 (1) (1982) 105–132.
were preserved at −30 °C as lyophilized powder. [5] The solubility of amino acids and two glycine peptides in aqueous ethanol and di-
oxane solutions. Establishment of a hydrophobicity scale. Nozaki Y, Tanford C, J.
Biol. Chem. 246 (7) (1971) 2211–2217.
4.2. Protein solubility measurement [6] C.N. Pace, G.R. Grimsley, J.M. Scholtz, Protein ionizable groups: pK values and
their contribution to protein stability and solubility, J. Biol. Chem. 284 (20)
(2009) 13285–13289.
Protein solubility was determined as the maximum protein con-
[7] P.W. Howe, A straight-forward method of optimising protein solubility for NMR,
centration in the supernatant of super-saturated protein solutions, as J. Biomol. NMR 30 (3) (2004) 283–286.
described previously [14]. In short, solubility measurements were per- [8] T. Arakawa, Y. Kita, A.H. Koyama, Solubility enhancement of gluten and organic
compounds by arginine, Int. J. Pharm. 355 (1–2) (2008) 220–223.
formed at 25 °C in 50 mM buffers at pH4.7 and pH7.7 in the presence
[9] H.A. Scheraga, J.A. Vila, D.R. Ripoll, Helix-coil transitions re-visited, Biophys.
of 1.3 and 1.5 M ammonium sulfate, except when indicated. The sam- Chem. (2002) 101–102 (255–265).
ples with different initial concentrations were mixed thoroughly and [10] A. Chakrabartty, T. Kortemme, R.L. Baldwin, Helix propensities of the amino acids
allowed to equilibrate for 20 min at 25 °C. After equilibration, the sam- measured in alanine-based peptides without helix-stabilizing side-chain interac-
tions, Protein Sci. 3 (5) (1994) 843–852.
ples were centrifuged at 20,000 ×g for 20 min at 25 °C and protein con- [11] M. Blaber, X.J. Zhang, B.W. Matthews, Structural basis of amino acid alpha helix
centrations of the supernatants were determined by UV analysis using a propensity, Science 260 (5114) (1993) 1637–1640.
Hitachi UV-spectrophotometer, U-1800. The solubility of C5I was mea- [12] D.L. Minor Jr., P.S. Kim, Measurement of the beta-sheet-forming propensities of
amino acids, Nature 367 (6464) (1994) 660–663.
sured from the amount of precipitated protein. In order to ensure reli- [13] S.R. Trevino, J.M. Scholtz, C.N. Pace, Amino acid contribution to protein solubility:
ability and reproducibility of the final solubility data, all protein Asp, Glu, and Ser contribute more favorably than the other hydrophilic amino
solubility values were measured at least three times, in different work- acids in RNase Sa, J. Mol. Biol. 366 (2) (2007) 449–460.
[14] A. Kato, K. Maki, T. Ebina, K. Kuwajima, K. Soda, Y. Kuroda, Mutational analysis of
ing days, and with different sets of reagents. protein solubility enhancement using short peptide tags, Biopolymers 85 (1)
(2007) 12–18.
4.3. Thermal stability measurement [15] Y. Kuroda, P.S. Kim, Folding of bovine pancreatic trypsin inhibitor (BPTI) variants
in which almost half the residues are alanine, J. Mol. Biol. 298 (3) (2000)
493–501.
Thermal stabilities of all but the C5I variants were monitored using [16] M.M. Islam, S. Sohya, K. Noguchi, M. Yohda, Y. Kuroda, Crystal structure of an ex-
the Circular Dichroism (CD) signal at 220 nm. Samples for CD measure- tensively simplified variant of bovine pancreatic trypsin inhibitor in which over
one-third of the residues are alanines, Proc. Natl. Acad. Sci. U. S. A. 105 (40)
ments were prepared by dissolving lyophilized proteins at 5–10 μM
(2008) 15334–15339.
concentrations in 20 mM sodium acetate (pH4.7) and 20 mM Tris [17] B.C. Cunningham, J.A. Wells, High-resolution epitope mapping of hGH-receptor
buffer (pH7.7) in the presence of 0.0, 0.5, 0.75, 1.0 and 1.5 M ammoni- interactions by alanine-scanning mutagenesis, Science 244 (4908) (1989)
um sulfate. Small pH variation occurred upon addition of Ammonium 1081–1085.
[18] M.H. Yu, J.S. Weissman, P.S. Kim, Contribution of individual side-chains to the sta-
sulfate at high concentration. Measurements were carried out using a bility of BPTI examined by alanine-scanning mutagenesis, J. Mol. Biol. 249 (2)
1 cm cuvette with a Jasco J-820 spectropolarimeter, as previously (1995) 388–397.
1150 M.M. Islam et al. / Biochimica et Biophysica Acta 1824 (2012) 1144–1150

[19] S. Chen, F.A. Ferrone, R. Wetzel, Huntington's disease age-of-onset linked to poly- predicted surface residues with antigenicity and X-ray-derived accessible sites, Bio-
glutamine aggregation nucleation, Proc. Natl. Acad. Sci. U. S. A. 99 (18) (2002) chemistry 25 (19) (1986) 5425–5432.
11884–11889. [22] S.R. Trevino, J.M. Scholtz, C.N. Pace, Measuring and increasing protein solubility, J.
[20] CRC Handbook of Chemistry and Physics. Editor: David R. Lide. 90th edition:Section Pharm. Sci. 97 (10) (2008) 4155–4166.
7–1, 2009. [23] M.M. Islam, S. Sohya, K. Noguchi, S. Kidokoro, M. Yohda, Y. Kuroda, Thermody-
[21] J.M. Parker, D. Guo, R.S. Hodges, New hydrophilicity scale derived from namic and structural analysis of highly stabilized BPTIs by single and double mu-
high-performance liquid chromatography peptide retention data: correlation of tations, Proteins 77 (4) (2009) 962–970.