Basic protein structure and stability VII: Determinants of protein stability and structure

1
Basic protein structure and stability
VIIDeterminants of protein stability and
structure

• Biochem 565, Fall 2008
• 09/12/08
• Cordes

2
Obvious interactions in native protein structures
hydrophobic interactions
polar interactions (hydrogen bond/salt bridge)
disulfide crosslinks
3
Contributions to protein stability

• type of interaction total contribution
• hydrophobic group burial 200 kcal/mol
• hydrogen bonding small??
• ion pairs/salt bridges lt15 kcal/mol
• disulfide bonds 4 kcal/mol per link
• for globular protein of 150 residues
• Hydrophobic burial is the chief interaction
  favoring protein stability, but this is balanced
  by a huge loss of conformational entropy that
  opposes folding.
• Consequently typical net protein stabilities are
  5-20 kcal/mol--gt so even minor interactions can
  make a difference!

4
Alanine scanning
A way of assessing the importance of amino-acid
side chains for structure/stability etc. Remove
each residue one by one by replacement with
Ala many Ala mutations have no effect on
stability--gt about half! A large group also cause
significant effects--gtseveral kcal/mol Occasional
a mutant will stabilize the protein--gt natural
proteins not maximally stable!
5
The interior is more important for stability than
the exterior
side chains with stability-neutral Ala mutations
side chains with destabilizing Ala mutations
6
Side-chain packing in the hydrophobic core
Protein interiors have a jigsaw
puzzle-like aspect. Their packing densities are
similar to those of crystals of organic
molecules. This dense packing can have
importance both in maintaining stability and in
maintaining a precise three-dimensional structure
which is optimized for activity. One issue with
the cores of proteins is simply volume. Given a
particular backbone configuration, there is a
certain amount of space that has to be filled,
and over or underfilling it can be detrimental to
stability Another constraint is sterics--not
all cores with equal volume are equally stable.
For instance, a simple switch of two residues in
Gene V protein, Leu35/Val47--gtVal35/Leu47,
results in a 4 kcal/mol reduction in stability
(Sandberg Terwilliger, 1991) For a good review
of packing, see Richards Lim, Q Rev Biophys 26,
423 (1993).
7
Effect of nondisruptive hydrophobic core mutations
nondisruptive means not causing any steric
clashes or uncompensated buried charges,
H-bonds etc
mutation to alanine
leucine in core
difference in water-octanol transfer free
energies of leucine and alanine is 2 kcal/mol.
Effects of Leu--gtAla mutations are typically
larger than this, however. Why? Not all buried
Leu--gtAla mutations give the same
destabilization. Why?
8
The cost of cavity formation in protein cores
A Leu--gtAla core mutation leaves a cavity in the
hydrophobic core. In addition to the 2 kcal/mol
transfer free energy difference between Leu and
Ala, there is a penalty of 20 cal-mol/Å2 for
forming this cavity. This is due to loss of van
der Waals interactions with the mutated side
chain. This increases the vertical (i.e.
assuming the structure of the mutant is the same)
cost of a Leu--gtAla mutation from 2 to about 5
kcal/mol! Since proteins are only stable toward
unfolding to the extent of 5 to 15 kcal/mol, such
mutations are potentially devastating, and this
suggests that having good packing in terms of not
having cavities is important to stability.
9
The plasticity of protein cores
Matthews and coworkers solved the crystal
structures of a number of T4 lysozyme
core mutants and found that the protein structure
often adjusts to reduce the cavity size, and that
this reduces the energetic penalty, restoring
some stability...note that this doesnt happen in
all cases
slope is penalty for cavity formation
y-intercept is just the Leu/Ala transfer
free energy difference!
Eriksson et al. Science 255, 178 (1992)
10
Disruptive mutations in hydrophobic cores
Three kinds steric mutants change in shape, not
volume extreme volume mutants increase core
volume polar mutants put polar/charged residue
in core
Polar/charged core mutants are almost invariably
very destabilizing, for obvious reasons.
Charged groups or groups that can form hydrogen
bonds that are isolated within a protein interior
are bad for stability Energetic effects of
increased volume are often hard to
predict--subtle backbone shifts often occur to
accommodate the extra volume
11
The V111I mutant of lysozyme at right illustrates
a typical backbone shift to accommodate
increased volume. Lambda repressor
V36L/M40L/ V47I is more stable than
wild-type despite a 50 Å3 increase in core
volume. A crystal structure shows that the
backbone adjusts to accommodate the mutations.
However, the mutant does not bind target DNA as
well as wild type. Thus, despite increased
stability and despite none of the residues being
directly involved in function, the mutation is
not tolerated. Thus, side chain packing is not
only a determinant of stability, It can also be a
key determinant of the precise structure of the
protein Lim et al. PNAS 91, 423 (1994)
12
Mutations of surface (solvent-exposed) residues

• Although the surface of proteins are very polar
  overall, individual surface positions can usually
  be replaced by many other residues including
  hydrophobics (though there are definitely
  exceptions) without much effect on stability.
• average effect on stability of surface mutations
  is small
• little stability penalty for change of
  individual surface polars to hydrophobics
• However
• too many hydrophobics on a proteins surface will
  reduce solubility and
• promote aggregation
• at least two studies have shown that surface
  polar-to-hydrophobic mutations can reduce
  structural specificity by favoring alternative
  conformations in which the introduced hydrophobic
  side chain becomes buried. This is another type
  of effect which may impact function.

Cordes et al., Protein Sci 8, 318 (1999) Schwehm
et al. Biochem 37, 6939 (1998) Cordes et al. Nat
Struct Biol 7, 1129 (2000). Hill DeGrado Struct
Fold Des. (2000) Pakula Sauer Nature 344, 363
(1990).
13
Sickle-cell hemoglobin a surface
polar-to-hydrophobic mutation that lowers
solubility
Glu b6--gtVal mutation causes self-association and
polymerization
Phe 85
mutation leads to hydrophobic interaction between
hemoglobin tetramers
Val 6
Leu 88
fibril formation at high concentration
picture of sickle-cell hemoglobin fibrils
spilling out of a distorted, ruptured erythrocyte
source Biochemistry by Voet Voet.
14
Energetics of hydrogen bonding in proteins
The relevant situation for protein folding is
arrow 3 or 5, depending upon how solvent-exposed
the hydrogen bond is in the native state. Buried
hydrogen bonds (5) can actually destabilize
proteins, while solvent-exposed ones (3) may be
slightly stabilizing. The same is true of ion
pairs/salt bridges.
unfolded protein
exposed H-bond in folded protein
buried H-bond in folded protein
15
Hydrogen bond inventory
Although hydrogen bonds probably do not stabilize
proteins per se, it is nonetheless important that
all potential hydrogen bond donors and acceptors
be hydrogen bonded to something, be it solvent,
protein backbone, or protein side chains. Alan
Fersht has called this concept hydrogen bond
inventory. This is important when trying to
understand the effect of mutations that impact
hydrogen bonding, because removal of one partner
of a hydrogen bonded pair can be quite
destabilizing if the remaining partner is not
able to satisfy its hydrogen bond potential by
interacting with solvent. Essentially this
same logic is also applicable to ion pairing/salt
bridge interactions. Even though ion pairs dont
contribute much to stability, charged groups
which are neither paired with oppositely charged
groups nor solvated by water can be very
destabilizing! In fact, one observes very few
uncompensated buried polar or charged groups in
proteins, and mutation of one partner of a salt
bridge or hydrogen bond is usually very
destabilizing.
16
Role of solvent-exposed salt bridges
Typical mutations of surface salt bridges are
destabilizing by less than 1 kcal/mol, but there
are cases where larger effects are observed.
(His 31-Asp 70 in lysozyme is an example).
0.0 kcal/mol
Surface salt bridges are thus not large
contributors to protein stability. However, some
salt bridges may be important at the level of
specifying a particular precise structure, much
in the way that hydrophobic packing interactions
are.
Strop Mayo, Biochemistry 39, 1251 (2001)
1.5 kcal/mol
P. furiosis rubredoxin
17
Structural role of buried salt bridges
wild-type Arc R31-E36-R40
Arc-MYL M31-Y36-L40
Substitution of of Arg31, Glu36 or Arg40 by Ala
destabilizes Arc repressor by 3 to 6 kcal/mol.
Mutation of all three by the MYL triad,
however, stabilizes the protein by 4 kcal/mol!!
Waldburger et al. Nature Struct Biol 2, 122
(1995)
Buried salt bridges (and buried
polar interactions in general) not important for
stability per se, but removal of individual
partners can be hugely destabilizing.
It has been hypothesized that buried polar
interactions serve more to impart specificity to
the structure rather than stability, due to the
strict requirement for satisfaction of H-bond
potential (H-bond inventory) and compensation of
charge. This has been directly shown to be true
for some proteins e.g. Lumb Kim, Biochemistry,
34, 8642 (1995)
18
Buried polar residues/interactions in thioredoxin
2TRX.pdb
D26 water-mediated H-bond to C32 carbonyl
T66-G74 sch-mch hydrogen bond
water

C32-C35 disulfide
T77-D9 sch-sch hydrogen bond
19
Effects of mutating buried polar residues in
thioredoxin
IAALV means D26I/C32A/C35A/T66L/T77V AALV means
C32A/C35A/T66L/T77V
IAALV found to have less specific native
state-- cant remove all buried polar residues
Bolon D Mayo SC Biochemistry 40, 10047 (2001).
20
H-bonding motifs N-termini of alpha helices
Many helices have side-chain to
main-chain hydrogen bonds at their N-termini.
Mutations to alanine of side chains involved in
such interactions have effects ranging from
0.5 to 2.0 kcal/mol These residues
usually occupy the position immediately before
the helix starts.
N-cap side-chain to main-chain hydrogen bond
Ser, Thr, Asp and Asn are most stabilizing here.
(small side chains that can act as acceptors) Asp
better than Asn, possibly because of
helix-dipole effects. Gly is also OK at N-cap.
Why?
serine side chain
solvent-exposed amide hydrogens
21
Relative stability of helix N-cap variants of
barnase

• amino acid DDGu (kcal/mol)
• Asp 2.02
• Thr 2.05
• Ser 1.64
• Asn 0.86
• Gly 0.69
• Gln 0.42
• Glu 0.25
• His 0.16
• Ala 0.00
• Val -0.15
• Pro -0.87

from Fersht AR, Structure and Mechanism... Chapt
er 17, p. 527.
The numbers represent the average of two
positions in the protein. The N-cap is defined
as the first residue the carbonyl of which
makes an i,i4 hydrogen bond to an amide. These
are relative free energies of unfolding, so a
higher number means greater stability.
22
Residues with unusual backbone conformation prefer
ences glycines at alpha-helix C-termini
these carbonyls hydrogen bond to solvent
left-handed (aL) conformation here leads to
capping of carbonyls here while terminating the
helix and causing a change in chain direction
Many helices terminate this way, and glycine is
favored at the left-handed position because of
its backbone flexibility and because large side
chains here would point upward and interfere with
solvation of carbonyls.
About one-third of all helices terminate in
glycine!
Schellman motif
23
Relative stability of mutants at C-terminal
ends of helices in barnase
from Fersht Structure and Mechanism..in Protein
Science, Ch. 17, p. 526

• amino acid DDGu (kcal/mol)
• Gly 2.23
• His 0.67
• Asn 0.47
• Arg 0.47
• Lys 0.01
• Ala 0.00
• Ser -0.16
• Asp -0.27

Gly--gt Ala mutations have ranged from 1 to
3 kcal/mol in a number of proteins. The 2.2
kcal/mol number observed here is typical
The residue being mutated is the left-handed (aL)
residue at the C-terminal end of the a-helix.
Since these are relative free energies of
unfolding, a higher number means higher
stability. Glycines can contribute to stability
(at certain positions, relative to other
residues) because of their unique backbone
conformation characteristics. Would the average
glycine be stabilizing, though?
24
Intrinsic secondary structure propensities and
stability

• amino acid DDGu (kcal/mol), alpha-helix DDGu
  (kcal/mol), beta-sheet
• Ala 0.00 0.00
• Arg -0.17 0.40
• Leu -0.17 0.45
• Met -0.19 0.90
• Lys -0.31 0.35
• Trp -0.31 1.04
• Gln -0.33 0.38
• Ser -0.44 0.87
• Ile -0.43 1.25
• Phe -0.47 1.08
• Cys -0.54 0.78
• Glu -0.56 0.28
• Tyr -0.56 1.63
• Asn -0.61 0.52
• Thr -0.61 1.36
• Val -0.63 0.94
• His -0.65-0.88 (0 or charge) 0.37
• Asp -0.68 -0.85

All effects listed relative to Ala
based on effects of surface mutations in helices
in a variety of proteins. Some residues like Ala
consistently stabilize proteins relative to other
residues, when they occur in helices.
based on effects of surface mutations at Thr 53
in beta-sheet of B1 domain of protein G. These
effects depend very strongly upon context, e.g.
what side chains interact with the mutated
position on the same face of the beta-sheet. It
could be argued, therefore, that there is no such
thing as intrinsic beta-sheet propensity.
from Fersht book Chapter 17, p. 528
in helices, effect of average substitution is
very small. In beta-sheets can be larger but
depends upon the sequence/structure context.
25
Stability-activity trade-offs?
It has been shown for many proteins that it is
possible to engineer higher stability by
introducing mutations. In many cases, this does
not appear to impair activity in in vivo and/or
in vitro assays. Moreover, comparable proteins
from thermophilic organisms have higher stability
than those from mesophilic counterparts. This
shows that proteins have not evolved to maximize
stability. Rather, it is likely that they
generally evolve to preserve adequate stability.
However, sometimes stability and activity are
directly at odds with one another, and one is
selected at the expense of the other. Many
thermophilic proteins have low activities at
lower temperatures. Some mutations in the active
sites of enzymes (barnase, T4 lysozyme) have been
shown to give more stable but less active
proteins. For instance, the active site of
barnase is highly positively charged because it
has to bind a negatively charged pentacoordinate
phosphate at the transition state. When
substrate is not bound, the positively charged
side chains repel each other, reducing stability.
source Chapter 17 of Fersht. Structure and
Mechanism in Protein Science
26
Mutation of catalytic residues in T4 lysozyme
Asp 20
active site cleft
protein DTm DDGf, activity , C kcal mol-1
(relative) wild-type 0 0 1 E11F 4.3 1.7 lt0.0001
E11M 4.1 1.6 lt0.0001 E11A 2.6 1.1 lt0.0001 E11H 0.1
0.1 lt0.0001 E11N -0.6 -0.1 lt0.0001 D20N 3.1 1.3 lt
0.0001 D20T 2.2 0.9 lt0.0001 D20S 1.6 0.7 lt0.0001 D
20A -0.8 -0.3 0.0005
Glu 11
Replacement with some amino acids increases
stability but strongly diminishes activity. This
same phenomenon was found to occur for residues
involved in substrate binding. Glu11 and Asp20
are examples of what has been referred to as
electrostatic strain in enzyme active sites.
However, not all mutations which remove the
charge stabilize the protein, emphasizing that
the situation is complex.
Shoichet et al. PNAS 92, 452 (1995)
27
Intrinsically disordered/Natively denatured
proteins
protein sequences with high net charge, low
hydrophobicity tend not to be stable
Not all natural proteins have stable folded
structures! In your average organism 10-20 do
not, by various estimates! Folding sometimes
depends upon binding activity.
Oldfield CJ et al Biochemistry 44, 1989
(2005). review Wright PE, Dyson JH J Mol Biol
293, 321 (1999)