Structure, Vol. 12, 2025–2036, November, 2004, 2004 Elsevier Ltd. All rights reserved.DOI 10.1016/j.str.2004.08.011
Crystal Structures of Ral-GppNHp
and Ral-GDP Reveal Two Binding Sites
that Are Also Present in Ras and Rap
tute one of the links between calcium signaling and
sive cell biology experiments proving this link are still
needed (Feig, 2003).
Ras and Rap have several effectors, which are recog-
that is representative of most Ras subfamily members
(Bos, 1998; Feig, 2003). Ral is unique in the subfamily
in that it has a different switch I sequence, YEPTKAD,
and a unique set of effectors. The changes from I to K
and E to A together reverse the overall electrostatic
character of the effector binding site in Ral relative to
other proteins in the Ras subfamily. A sequence space
analysis (Casari et al., 1995) aimed at determining func-
tionally important amino acid residues in Ral revealed
14 residues (3 in switch I and 11 elsewhere) that are
conserved in 8 Ral variants but not in other members
of the subfamily (Bauer et al., 1999). Mutational experi-
ments and binding studies showed that the two charge
reversal residues were sufficient to render Ras specific
for Ral target recognition and vice versa (Bauer et al.,
1999). The crystal structure of Ral in complex with the
Ral minimum binding domain of one of its effectors,
Sec5, elucidated the structural features of switch I re-
sponsible for binding specificity in Ral (Fukai et al.,
In a study aimed to identify functionally important
residues, 100 GTPases in the Ras superfamily were ex-
pressed in their constitutively active forms (mutants
equivalent to Q61L in Ras) and classified according to
resulting changes in cell morphology (Heo and Meyer,
2003). The constitutively active form of Ras resulted in
polar cell morphology, while Ral, Rap, and R-Ras in-
duced what the authors called the eyelash morphology.
The double mutation L23F/K101L in Ras was sufficient
to change the cellular phenotype from polar to eyelash
morphology. The reverse mutations in Ral, Rap, and
R-Ras resulted in cells with polar morphology. Interest-
ingly, both of these functionally important residues are
located outside of the switch regions. The Ral tree-
in Figure 1.
We present the crystal structures of RalA bound to
the GTP analog GppNHp and to GDP. These structures
reveal an overall catalytic domain that is very similar
to those of its close family members, but with unique
features in the switch regions, including the presence
of a second Mg2?ion bound to GppNHp. The Ral tree-
determinant residues and the two switch-of-function
residues converge to two putative binding sites that are
present in at least three members of the Ras subfamily,
each with distinct features.
Nathan I. Nicely, Justin Kosak, Vesna de Serrano,
and Carla Mattos*
Department of Molecular and Structural
128 Polk Hall-CB 7622
North Carolina State University
Raleigh, North Carolina 27695
RalA is a GTPase with effectors such as Sec5 and
Exo84 in the exocyst complex and RalBP1, a GAP for
Rho proteins. We report the crystal structures of Ral-
GppNHp and Ral-GDP. Disordered switch I and switch
II, located away from crystal contacts, are observed in
one of themolecules in the asymmetric unitof the Ral-
GppNHp structure. In the other molecule in the asym-
metric unit, asecond Mg2?ion is boundto the GppNHp
?-phosphate in an environment in which switch I is
pulled away from the nucleotide and switch II is found
in a tight ? turn. Clustering of conserved residues on
the surface of Ral-GppNHp identifies two putative
sites for protein-protein interaction. One site is adja-
cent to switch I. The other is modulated by switch II
and is obstructed in Ral-GDP. The Ral structures are
discussed in the context of the published structures
of the Ral/Sec5 complex, Ras, and Rap.
Ral is a member of theRas subfamily of GTPases, which
also includes Ras, Rap, R-Ras, TC21, and M-Ras (Bos,
1998; Feig, 2003). These proteins are involved in diverse
signaling pathways (Vojtek and Der, 1998) and function
through conformational changes in the switch I and
switch II regions modulated by the presence of GTP
(“on” state) or GDP (“off” state) (Milburn et al., 1990).
Ral exists in two main isoforms: RalA and RalB, each
consisting of 206 amino acid residues. The functional
roles known for Ral have recently been reviewed (Feig,
2003).Ral modulatestheactivityof varioustranscription
factors affecting a variety of cellular functions, including
the potentiation of cell proliferation controlled by the
Raf-Mek-Erk kinase cascade (Urano et al., 1996). Ral
proteins are involved in vesicle sorting, with roles in
both exocytosis (through interaction with the exocyst
complex) (Moskalenko et al., 2002) and endocytosis
(through its constitutive interaction with phospholipase
D and its GTP-dependent interaction with the effector
RalBP1) (Feig, 2003). RalBP1 contains a GAP domain
for the Rho subfamily GTPases Rac1 and CDC42, sug-
gesting the involvement of Ral in the modulation of actin
cytoskeleton and cell morphology (Cantor et al., 1995;
Jullien-Flores et al., 1995). Ral has also been shown to
bind calmodulin in a calcium-dependent manner (Wang
Results and Discussion
The catalytic domain in the GTPases consists of a six-
stranded ? sheet (?1–?6), five ? helices (?1–?5), and ten
Figure 1. Sequence Alignments of Ral, Ras, and Rap
Secondary structural elements are indicated by shading. Switch regions are boxed. The single amino acid difference between simian and
human RalA (147) is in gray. Residues are numbered and labeled above their respective sequences. The 14 tree-determinant residues (I18,
M35, Y36, E44, K47, A48, K54, L57, I64, I78, N81, A103, T104, M172) are in dark blocks, and the two switch-of-function residues (F34, L112)
are in white boxes. Black brackets under the Rap sequence indicate where the backbone of Rap is shifted relative to Ral and Ras.
connecting loops (L1–L10) (Milburn et al., 1990). Switch
I includes L2 residues 30–38 in Ras and is the effector
recognition site. Switch II consists of L4 and part of ?2,
spanning residues 60–70. L1 is the phosphate binding
loop (P loop) and in Ras consists of residues 10–17. The
P loop interacts substantially with the ?-phosphate of
GTP, but it does not undergo major conformational
changes upon hydrolysis (Wittinghofer and Nassar,
1996). In the GTP bound form, a Mg2?ion bridges the
?- and ?-phosphates of the nucleotide and is coordi-
nated by a Ser residue in the P loop (S17 in Ras), a Thr
(Wittinghofer and Nassar, 1996).
Ral has an additional 11 amino acid residues at the
N terminus relative to Ras and Rap (Figure 1), and these
residues interact constitutively with phospholipase D
(Jiang et al., 1995). The C-terminal region in GTPases
extends 20–30 residues beyond the catalytic domain,
serves as an anchor to the membrane, and plays a key
role in localization (Hancock, 2003). Unfortunately, the
C-terminal region interferes with expression in E. coli
and can impair crystallization. Therefore, our crystal
structures of RalA were obtained with a truncated C
terminus (residues 1–178), in analogy to the crystal
Rap2A (1–167) (Cherfils et al., 1997). These truncated
versions of the GTPases will be referred to as Ral, Ras,
and Rap throughout this paper. Due to the additional
11 residues at the N terminus of Ral, one must subtract
11 from a residue number in Ral to obtain the analogous
residue in Ras or Rap up to residue 122, which is an
insertion in Ral and Rap relative to Ras. Beyond residue
122, it is necessary to subtract 12 from the Ral number-
ing and 1 from the Rap numbering in order to obtain the
in Ral (residue 178) corresponds to residue 166 in Ras
and 167 in Rap (Figure 1). The Ral construct used for
the complex with the Ral binding domain of Sec5 was
slightly different and consisted of residues 9–183 (Fukai
et al., 2003). The catalytic activities of Ras and Rap
GTPases have been shown not to be affected by the
C-terminal truncation (Cherfils et al., 1997; John et al.,
1989; Nassar et al., 1995), and the C-terminal truncation
of RalA does not affect its Kd for Sec5 (Fukai et al.,
2003). Even though the N terminus was present in the
constructs of Ral used for the present study, the first
ten residues are disordered and are not observed in the
electron density maps for any of the structures.
The Crystal Structure of Ral Bound to GppNHp
Ral-GppNHp crystallizes with the symmetry of the te-
tragonal space group P4222 and unit cell parameters
shown in Table 1. A full data set was collected to a
beamline, APS (Argonne, IL). Data collection and refine-
ment statistics are also shown in Table 1. There are two
molecules of Ral in the asymmetric unit, designated as
A and B, respectively, and each presents the switch
regions in very different environments. In Molecule A,
both switches are relatively free of crystal contacts. As
a result, switch I is partially disordered, with virtually no
electron density for residues 46–49. Switch II is com-
pletely disordered from residue 72 to 83. In Molecule B,
both switches are in contact with other Ral molecules
in the crystal and are very well ordered.
The switch I residues in Molecule B do not interact
with the nucleotide in the way observed in Ras-GppNHp
(PDB code 1CTQ), Rap-GTP (PDB code 3RAP), and in
Ral-GppNHp/Sec5 (PDB code 1UAD), although the
GppNHp molecule itself is in a position that superim-
poses well with the nucleotide analog in these struc-
tures. An unusual switch I conformation is stabilized
primarily through a series of contacts across the inter-
face between Molecules A and B in the asymmetric unit
(Figure 2). In this conformation, Y43 (32 in Ras number-
ing) is buried between the P45 ring and the molecular
interface. The O? atom of T46 (35 in Ras numbering) is
also involved in contacts across the molecular interface
and is about 10 A˚from the canonical Mg2?ion. There
is a water molecule coordinated to the Mg2?ion in Ral
where the O? of T35 is in Ras.
A further unusual feature in this region is the presence
of a second Mg2?ion coordinated to the ?-phosphate
of the nucleotide (Figure 3). The first ion is in a position
found in other GTPase structures, including Ras and
Rap (Pai et al., 1990). The second Mg2?ion is unique
to the Ral-GppNHp structure presented here and may
occupy a weak binding site, facilitated by the conforma-
Ral GTPase Binding Sites
Table 1. Data Collection and Refinement Statistics for Ral-GppNHp, Ral-GDP Crystal Form 1, and Ral-GDP Crystal Form 2
Ral-GppNHpRal-GDP (1) Ral-GDP (2)
Unit cell (A˚, ?)
a ? 77.3
b ? 77.3
c ? 116.1
? ? 90
? ? 90
? ? 90
a ? 53.6
b ? 62.1
c ? 112.7
? ? 90
? ? 90
? ? 90
a ? 52.6
b ? 61.5
c ? 112.9
? ? 90
? ? 90
? ? 90
Temperature of data collection
Number of reflections
Rms bond length deviation
from ideal geometry (A˚)
Rms bond angle deviation
from ideal geometry (?)
In favored regions
In allowed regions
Number of protein atoms
Number of nucleotide atoms
Number of magnesium ions
Number of water molecules
The numbers in parentheses describe the relevant value for the highest resolution shell. Rsym? ?|Ii? ?I?|/?I, where Iiis the intensity of the
ith term observed and ?I? is the mean intensity of the reflections. Rwork? ?||Fobs| ? |Fcalc||/?|Fobs| crystallographic R factor calculated by using
90% of the reflections against which the model was refined. Rfree? ?||Fobs| ? |Fcalc||/?|Fobs| calculated by using the test set consisting of 10%
of the total reflections, randomly selected from the original data set.
tion of switch I in this structure and by the high magne-
sium formate concentration found in the crystallization
conditions (200 mM rather than the 5 mM MgCl2usually
included in the GTPase buffer solutions). This Mg2?ion
is present only in Molecule B, where it is coordinated to
theterminal oxygenatomofthe nucleotide?-phosphate
and to five water molecules (Figure 3), away from crystal
contacts. This cluster of atoms is part of an H bonding
network that connects the main chain atoms of residues
71–73 in switch II to the side chain of Glu 44 in switch
I. It also serves to link switch II to main chain atoms in
switch I through the ?-phosphate of the nucleotide and
the canonical Mg2?ion.
The switch II residues 70–79 in Molecule B form two
? strands connected by a type I hairpin turn at residues
D74 and Y75 (Figure 4). G71 is at the beginning of the
? strand leading into the turn. Its backbone N atom
makes a good H bond with one of the O atoms of the
?-phosphate in GppNHp (2.7 A˚), as does G60 in Ras and
Rap bound to GTP or its analogs. Thus, the initial part
of Ral, as expected (Wittinghofer and Nassar, 1996).
This interaction is also seen in Molecule A, even though
residues 72–83 are disordered. Q72 (61 in Ras number-
ing) is turned away from the nucleotide and makes a
good H bond with the carbonyl O atom of A70 (2.9 A˚).
provides a link between Q72 and the ?-phosphate of
GppNHp. E73 is the first of the four residues in the type
I turn (residue i), with its carbonyl O atom H bonding to
Figure 2. The Asymmetric Unit in the Crystal
Structure of Ral-GppNHP
GppNHp is in cyan, Mg2? is in yellow, and
the coordinating water molecules are in red.
Figure 3. Close View of the Mg2?Ions Bound to GppNHp
(A) The model in the absence of an electron density map. Ser28 of the P loop and Gly71 in switch II are shown in magenta.
(B) Same model as in (A), superimposed on an anneal omit Fo? Fcelectron density map contoured at the 3? level. The two magnesium ions,
the coordinated water molecules, and the nucleotide were deleted from Molecule B in the model.
(C) Same model as in (A), superimposed on the final 2Fo– Fcelectron density map contoured at the 1? level.
N of A76 (residue i?3) in a typical manner observed for
two-residue ? turns (Wilmot and Thornton, 1988). D74
and Y75 constitute turn residues i?1 and i?2, the two
central residues of the ? hairpin. While the side chain
of D74 is exposed to solvent, interacting with bound
water molecules, Y75 is completely buried in a hy-
drophobic pocket formed by crystal contacts. The sec-
posed of switch II ? turns from two molecules related
to each other by a 2-fold symmetry axis in the crystal
(Figure 4). The side chain of the last residue on the ?
strand (R79) crosses over to interact with the carbonyl
O atoms of D74 (turn residue i?1) and A76 (turn residue
i?3) of the symmetry-related molecule, helping to stabi-
lize switch II in a type I ? hairpin conformation (Mattos
et al., 1994). The remainder of switch II is in a helical
conformation and represents the beginning of helix ?2,
as it does in Ras and Rap.
The Crystal Structure of Ral Bound to GDP
Ral-GDP crystallizes with the symmetry of the ortho-
rhombic space group P212121with two molecules in the
asymmetric unit. Two slightly different crystal forms
wereobtained inthis spacegroup. Thefirst (crystalform
1) is represented by crystals that grow within 2 weeks,
are abundant, and can be routinely reproduced. A data
set usingthis first crystalform was collected toa resolu-
tion of 1.5 A˚at the SER-CAT beamline at APS (Argonne,
IL). Asingle crystalrepresenting thesecond crystalform
Figure 4. Switch II in Ral-GppNHp
(A) The ? turn conformation in switch II of Molecule B in the Ral-GppNHp structure. Switch II in a symmetry-related molecule contributes to
a four-stranded ? sheet that forms across the interface.
(B) A schematic drawing of the two ? strands that come together in the crystal contact.
Ral GTPase Binding Sites
Figure 5. Ral-GDP Structure
The asymmetric unit in the crystal structure
of Ral-GDP in crystal form 2.
(crystal form 2) was found in a crystallization plate that
in a crystallization drop that had previously contained
crystals of crystal form 1. A room temperature data set
was collected at our home laboratory to a resolution of
2.0 A˚. Table 1 shows that although the unit cell parame-
ters are very similar in the two crystal forms, there is a
shrinking of the unit cell in form 2 relative to 1. The main
difference between the two is that switch II is partially
disordered in crystal form 1 but highly ordered in crystal
form 2. No electron density is observed for residues
72–74 in molecule A and for residues 71–74 in Molecule
B of crystal from 1. Superposition of Molecule A in the
two crystal forms reveals that the ordering of switch II
in form 2 is due to a shift in the relative positions of the
molecules in the asymmetric unit, such that switch II is
exposed to solvent in form 1 while it participates in
extensive crystal contacts in form 2, where the mole-
cules become more closely packed. In all four indepen-
dent Ral-GDP models, only one Mg2?ion is observed
interacting with the nucleotide as expected (Milburn et
The switch I region for Ral-GDP is very similar in all
four molecules. In addition to participating in extensive
interactions across the molecular interface in the asym-
metric unit, the switch I residues are involved in several
intramolecular interactions (Figure 5). Y43 H bonds to
Y51 (2.8 A˚) and to K47 (3.1 A˚), anchoring the two ends
of the switch. L32 is at the base of this site, in van der
Waal’s contact with both tyrosine rings (3.8 A˚in each
case). T46 is turned away from the nucleotide, and its
O? atom interacts with the N?2 atom of Q72 in switch
II (3.0 A˚), while the main chain O atoms of Y43, E44,
P45, and K47 all make H bonds to water molecules that
coordinate the nucleotide bound Mg2?ion.
The conformation for the Ral-GDP switch II is de-
scribed for the model derived from crystal form 2. In the
absence of the ?-phosphate, G71 does not interact with
II moves away from the nucleotide. Residues 74–77 are
involved in crystal contacts with the symmetry-related
molecule. Y75 is at the beginning of helix ?2, which
extends to residue G86, beyond the end of switch II.
There is no indication of the ? turn conformation ob-
served in the Ral-GppNHp structure.
Comparison between Ral-GDP, Ral-GppNHp,
Other than in the switch regions, the overall structure
the complex with Sec5. The root mean square deviation
and 0.47 A˚when the two switch regions are excluded
switch II is partially disordered, but switch I interacts
extensively with Sec5 and has well-defined electron
density (Fukai et al., 2003). This interaction is composed
of several important contacts that include the Ral tree-
determinant residues K47 and A48. Molecule A in our
uncomplexed Ral-GppNHp structure shows a disor-
dered switch I, demonstrating, in conjunction with the
structure of the complex, that there is a disorder-to-
order transition in switch I upon complex formation in
Ral. Analysis of the switch conformations in GTPases
has typically been compounded by the fact that these
sites of protein-protein interactions are often found in
crystal contacts. NMR experiments on Ras-GppNHp,
however, have determined that both switch regions ex-
hibit polysterism, with a small number of discrete well-
experiments suggest that each binding partner for Ras
selects its favorite conformer from a set of pre-ordered
conformations present in the uncomplexed protein. We
propose that in Ral there is also selection of particular
conformers by binding partners, although it is not possi-
ble to determine whether the lack of electron density
for the switch regions in Molecule A of the Ral-GppNHp
structure is a consequence of polysterism (with more
than two or three distinct conformers) or of complete
disorder. In either case, since Ral interacts with effector
proteins only in the GTP bound state, the presence of
this nucleotide must facilitate the selection of certain
conformations complementary to target proteins that
are not favored in the presence of GDP.
The conformational differences in switch I between
Ral-GppNHp, Ral-GDP, and Ral-GppNHp/Sec5 are
shown in Figure 6. In Molecule B of the Ral-GppNHp
structure, switch I exists in a conformation that is very
different from that found in the complex, and which
would cause the most severe clashes with Sec5. At the
beginning of switch I, the backbones of Ral-GppNHp
Figure 6. Comparison of Switch I in Three Forms of Ral
Switch I of Ral-GppNHp (blue), Ral-GDP (green), and Ral-GppNHp/Sec5 (brick red). The box shows an enlarged view of the area where Sec5
interacts most closely with switch I. The backbone of Sec5 is shown in gold.
and Ral-GppNHp/Sec5 superimpose well, but these
structures diverge beyond D42. Thus, for the majority of
switch I, Ral-GppNHp/Sec5 and Ral-GDP cluster much
more closely to each other than to the uncomplexed
SwitchII canbestbecompared betweenourRal-GDP
structure from crystal form 2 and Molecule B of Ral-
GppNHp. There are two major differences between the
switch II conformation in Ral-GDP and that in the Ral-
GppNHp structure (Figure 7). In the Ral-GDP structure,
G71 does not H bond to the nucleotide, as it does in
the Ral-GppNHp and Ral-GppNHp/Sec5 where the
?-phosphate is present. In the second part of switch II,
residues 75–79 change from an ?-helical to a ? turn
conformation in going from the GDP to the GppNHp
bound form. From residue 79 to the end of the switch
at residue 83, the Ral-GppNHp structure resembles the
Ral-GDP helical conformation much more closely (Fig-
ure 7). In the Ral-GppNHp/Sec5 structure, the first part
of switch II is disordered, but it becomes ordered in the
presence of crystal contacts from residue 77 onward
(66 in Ras numbering), adopting a helical conformation
similar to that found in the other two structures.
Comparison between Ral-GppNHp,
Ras-GppNHp, and Rap-GTP
The backbone conformations of Ral,Ras, and Rap differ
from one another primarily at the switch regions. The
switch I backbone atoms superimpose well in Ras-
GppNHp and Rap-GTP (Cherfils et al., 1997). Switch I
in uncomplexed Ral-GppNHp is disordered in Molecule
A and adopts an unusual conformation in Molecule B.
In the complex with Sec5, it is in a similar conformation
to that found for Ras and Rap (Fukai et al., 2003). Al-
though it is clear that the conformation of switch I in
our Ral-GppNHp structure is stabilized by crystal con-
tacts, the fact is that it is an accessible conformation.
Whether or not it is of biological relevance in complexes
with binding partners other than Sec5 remains to be
The switch II region differs significantly in all three
proteins. The conformation of switch II observed in Mol-
ecule B of our Ral-GppNHp structure results in the
unique positioning of two Ral residues relative to the
analogous residues in Ras and Rap. In Ras, R68 is
tucked into the protein, making several H bonds with
carbonyl groups of switch II residues (Buhrman et al.,
2003; Pai et al., 1990). Y71 stacks against the R68 resi-
bic core composed of residues V7, V9, T58, F78, and
Y96. The general features of this structure in Rap are
similar to those of Ras, although the details vary. In Ral,
R79 (68 in Ras numbering) is facing outward, making
good H bonding interactions with backbone atoms of
the symmetry-related switch II type I turn (Figure 4).
Interestingly, Y82 (71 in Ras numbering) is also rotated
outward in this structure, retaining its stacking interac-
tion with R79. The flip of these two side chains from
facing theinterior ofthe proteinin Rasand Rap tofacing
the exteriorsurface in Ralis accompaniedby a flipin the
opposite direction for residue 78 (67 in Ras numbering),
which in Ras and Rap is a methionine that protrudes
Figure 7. Comparison of Switch II in Two Forms of Ral
Switch II of Ral-GppNHp (blue) and Ral-GDP (green). Switch II is
disordered in the Ral-GPPNHp/Sec5 structure.
Ral GTPase Binding Sites
from the protein surface, but in Ral is an isoleucine
turned toward the protein core. F83 in Ral is also an
important component of this core, making good van der
Waals’ contact with I78 (3.5 A˚) and with I18 (4.4 A˚). Both
I18 and I78 are Ral tree-determinant residues (Figure 1).
from random mutagenesis studies that identify residues
43 and 157 (54 and 169 in Ral numbering) as important
in the interaction between Ras and cRaf-1 (Winkler et
al., 1997) and from the fact that this is only one of two
regions outside of the switches where there is a back-
bone shift in Rap relative to Ral and Ras (Figure 1). The
shifted residues are those that lead to and from Loop
L3, which was previously observed to undergo small
conformational changes between Rap-GTP and Rap-
GDP (Cherfils et al., 1997). These changes would affect
residues in the proposed binding site in Rap, perhaps
linking optimal interaction with the molecular switch
mechanism of the protein.
The second site in which Ral tree-determinant resi-
dues cluster is modulated by switch II. It is located
between switch II and helix ?3 in a cleft lined by I78 and
F83 in Ral or by R68 and Y71 in Ras and Rap. A ribbon
diagram of Ral in an orientation that shows the pocket
is presented in Figure 9A, and the electrostatic surfaces
of Ral, Ras, and Rap in the same orientation are shown
in Figures 9B, 9C, and 9D, respectively. At one end of
the pocket, residue 103 (92 in Ras) makes an interaction
with the P loop residue 22 (11 in Ras), which in turn is
near the initial part of switch II. Residue 92 is Asp in
Ras and Rap. Ala 103 in Ral was identified as a Ral tree-
determinant residue. Its presence causes variations in
the pocket that could help modulate interactions with
binding partners specific to Ral. At the other end of
switch II residue D80 (D69 in Ras) and helix ?3 residue
N110 (99 in Ras). The adjacent residue 81 (70 in Ras
and Rap) has been identified as a Ral tree-determinant
residue and faces the solvent in all three GTPases,
where it could interact with binding partners occupying
the site. The cleft that forms between switch II and helix
forms the base of this pocket. In Ral, the cleft is deep
and lined by the Ral tree-determinant residue I78, by
F83, and by F107, forming the bottom of the pocket.
The positions of these three residues are modulated by
the presence of a second layer of hydrophobic residues
as V20, F89, and I111. In addition to being modulated
to those observed in Ras and Rap, this pocket can also
be distinguished in Ral by a relatively neutral surface
compared to the positively charged character due to
the presence of R68 that lines the pocket in the other
two GTPases (Figure 9). Interestingly, this region also
exhibits a shift in the Rap structure relative to Ral and
Ras. The shift includes about half of helix ?3 and all
of Loop 7, spanning residues 94–109 (105–120 in Ral
numbering) (Figure 1). It seems to be a consequence of
a Pro at residue 95 in the middle of helix ?3 causing a
slight kink and the backbone shift in Rap relative to the
other two GTPases.
The Ral tree-determinant residue T104 (I93 in Ras and
Rap) and the switch-of-function residue L112 (K101 in
Ras, I101 in Rap) are also part of helix ?3, but they face
helix ?4 on the opposite side from the pocket (Figure
9A). There is a double and complementary mutation in
Ral relative to Ras and Rap resulting in an H bond be-
Two Novel Binding Sites of Protein-Protein
Interaction on the Surfaces of Ral,
Ras, and Rap
The 11 Ral tree-determinant residues outside of switch
I cluster in two distinct areas on the surface of Ral-
Ral with this cluster of residues shown explicitly. Figure
8B shows an electrostatic potential surface of Ral in the
same orientation, with switch I forming a ridge at the
left of the figure. Following the surface to the right is a
narrow groove, and immediately after that, there is a
ridge delineated by two Ral tree-determinant residues,
M35 and Y36, which together form the left edge to an
extensive, but somewhat shallow, groove that we pro-
pose to be a site of protein-protein interaction. These
two residues would interact prominently with a binding
partner that simultaneously occupied the switch I ef-
fector binding site to the left of this ridge in Figure 8B
and the groove to the right. M35 and Y36 are in fact
observed to interact with Sec5 in the Ral-GppNHp/Sec5
structure, even though the truncated Sec5 protein binds
only at switch I and does not reach into the putative
secondary binding pocket. This pocket is lined by six
hydrophobic residues: F34, V55, L57, I64, F169, and
M172. Residue 34 is one of the two switch-of-function
residues important in determining cell morphology, while
residues 57, 64, and 172 were identified as Ral tree-deter-
and, in particular, the Ral tree-determinant residue K54
is at the edge of the proposed pocket. The pocket is
shown in Figures 8C and 8D for Ras and Rap, respec-
tively, with the analogous residues labeled on the sur-
face, but with the correct numbering for each protein.
The pocket in Ras has a deep cavity at its center that
is lined by L23 (Figure 8C). This is the switch-of-function
residue, which in Ral and Rap is a Phe that essentially
fills part of the groove, resulting in a shallower pocket.
In Rap, the pocket is even smaller than in Ral due to a
shift of residues 42–46 into the cavity. The obstruction of
thecentralcavity inRalandRapby the largerPhe residue
is likely to change the nature of the interaction in the
proposed site, perhaps leading to the eye lash morphol-
ogy observed experimentally (Heo and Meyer, 2003). In
Ral, L57 is at the edge of the groove, making room for
whereas the Ras V160 and Rap V161 side chains are
buried and not directly exposed. F156, which has been
shown to be functionally important and conserved in all
members of the Ras superfamily (Quilliam et al., 1995),
interacts directly with L23 in Ras, F23 in Rap, or F34
in Ral, helping to shape the pocket. In addition to the
convergence of Ral tree-determinant residues to this
site and the established functional significance of some
of the residues already mentioned, further support for
the presence of a binding site at this location comes
Figure 8. The Proposed Binding Site Near the Switch I Effector Region
(A) A ribbon diagram of Ral-GppNHp with the secondary structural elements labeled. Ral tree-determinant residues are in blue, and the switch-
of-function residue F34 is in cyan. Other selected residues that contribute to the pocket are shown in violet.
(B) The electrostatic surface of Ral-GppNHp in the same orientation as that shown in (A). The location of switch I is indicated in green.
Residues shown in (A) are identified and labeled with their respective three-letter codes. The Ral tree-determinant residues are indicated by
an asterisk, and the switch-of-function residue is indicated by a plus sign.
(C) The electrostatic surface of Ras-GppNHp.
(D) The electrostatic surface of Rap-GTP.
The orientation and labels in (C) and (D) are as described for (B). The numbering of the amino acid residues is relative to each protein as
shown in Figure 1. All electrostatic surface calculations shown in this figure and in Figure 9 were generated by using the program GRASP
(Nicholls et al., 1991).
tween the Ral tree-determinant residue T104 and R145
in place of van der Waals’ interactions between two
hydrophobic residues. This H bond may affect the be-
havior of helix ?3 in Ral relative to the other family mem-
bers, thus altering the properties of the pocket between
helix ?3 and switch II. Similarly, there is a direct venue
for changes in residue 112 (101 in Ras and Rap) to be
111 in Ral is involved in modulating the conformation
of the hydrophobic residues that line the groove adja-
cent to switch II. An alternative possibility that needs to
be explored is the presence of a binding site between
helices ?3 and ?4.
In Ral-GDP (crystal form 2), the ? turn structure of
switch II is not observed and I78 is turned toward the
solvent. R79 faces into the pocket and forms a salt bridge
with D80. Arg 113 from Loop 7 also moves to interact
closely with D80. These three residues essentially form a
Ral GTPase Binding Sites
Figure 9. The Proposed Binding Site Modulated by Switch II
(A–D) (A)–(D) are as described in Figure 8, but with the appropriate orientation. In (B)–(D), relevant secondary structural elements are indicated
in green. The many hydrophobic residues that line the pocket are discussed in the text but are not labeled in the figure in order to avoid
cluttering. Ser22 and Arg145 are indicated in violet.
lid over the pocket, making it inaccessible in the GDP
bound form of the GTPase. In Ras-GDP, switch II itself
is closed over the pocket, again completely obstructing
action in the proposed site would be sensitive to the
“on”/ “off” state of the GTPases.
The ability to truncate GTPase binding partners is
essential in obtaining crystals of Ras proteins in com-
plexes, including the complex between Ral and Sec5
to visualize the full range of interactions between the
GTPase and its binding partners. As a result, we have
a good understanding of interactions at switch I, but
little or no structural information on the location and
extent of secondary binding sites. Raf, for example, has
a cysteine-rich region that has been shown through mu-
tational analysis to interact with Ras at a secondary
binding site involving switch II (Drugan et al., 1996). It
is also known that Ras and Rap both interact with Raf
through nearly identical switch I sequences, but that
these interactions lead to diverse biological outcomes
(Bos, 1998). Distinct secondary binding sites in Ras and
Rap could serve to convey functional specificity.
We have identified two putative binding sites on Ral,
Ras, and Rap by mapping the convergence of several
independently determined biochemical features onto
grooves near the switch regions on the surface of Ral-
GppNHp. In addition, the two proposed sites also coin-
cide with areas where the backbone of Rap deviates
from those of Ral and Ras. It is unlikely that the conver-
gence of so many factors would occur coincidentally
without any functional significance. The two sites to-
gether account for all 11 Ral tree-determinant residues
found outside of switch I. The binding site adjacent to
switch I contains the switch-of-function residue 34 and
six Ral tree-determinant residues: 35, 36, 54, 57, 64, and
172 (Figure 8). The site modulated by switch II and ?3
includes the second switch-of-function residue 112 and
the remaining five Ral tree-determinant residues: 18, 78,
81, 103, and 104 (Figure 9). Our proposed picture is one
in which Ral is able to bind unique effectors not only
through its differently charged switch I region, but also
through unique conformational features of switch I and
switch II that drastically change the properties of the
new proposed binding sites from one GTPase to the
next. Furthermore, the accessibility of the site adjacent
to switch II is very different between the GTP and GDP
bound forms in the crystal structures of Ral and Ras,
linking interactions at this site to the state of the bound
compared to a 10-fold decrease in RasQ61L (Frech et
on the intrinsic GTPase activities of Ral and Ras are
consistent with different conformational properties of
switch II. It appears that RalQ72L is constitutively active
to RalGAP (Emkey et al., 1991). This view is supported
for Ras by recent computational studies showing that
Q61 in Ras contributes to the preorganization of the
catalytic conformation of the active site in the Ras/Ras-
tion state in the GTP hydrolysis reaction (Shurki and
Warshel, 2004). The simulations rule out a direct partici-
or through direct electrostatic or steric interaction with
the transition state. Instead, they suggest an indirect or
allosteric role for Q61 (Shurki and Warshel, 2004). We
propose that this would leave more room for variations
in the coevolution of GAP proteins and residue 61 than if
Q61 were required as adirect participant in the reaction.
than Gln as residue 61 in Rap proteins (Figure 1) and is
consistent with the idea that each GTPase/GAP pair has
a unique relationship to each other. The fact that Q72
H bonds to the carbonyl group of residue 70 provides
a link to G71, which in turn H bonds to the ?-phosphate
of the nucleotide. This network of interactions could
conceivably provide a venue for the allosteric involve-
With switch I pulled away from the nucleotide and
Q72 tucked into the ? turn of switch II, there is a fair
amount of space around the ?-phosphate group of
completing its coordination sphere occupies most of
that space, bridging the nucleotide to residues in switch
II. Although one can justifiably argue that the conforma-
tions of the switches stabilized by protein-protein inter-
actions within the crystal facilitate the binding of Mg2?,
which is present in high concentrations, it is also possi-
ble that the increased Mg2?concentration found in the
crystallization conditions could serve to facilitate the
unusual structure of switch I. This conformation would
not allow binding to Sec5, but it could possibly be com-
plementary to another binding partner, such as Exo84.
This would provide a mechanism for a second level of
regulation of protein-protein interactions between Ral
and similarly localized effectors in the cell. Transiently
elevated concentrations of Mg2?could mediate exqui-
wise compete for the same binding site. Regulation of
cellular events by transient fluctuations in Mg2?levels
is not unprecedented (Rijkers et al., 1993). This type of
or a polysteric switch I in the uncomplexed form of Ral-
GppNHp. The crystal structures of Ral in complex with
binding partners other than Sec5 (e.g., Exo84, RalBP1,
and RalGAP) will help elucidate the biologically relevant
range of conformations adopted by switch regions in
Ral and the relevance of the conformations we observe
near crystal contacts in the Ral-GppNHp structure.
Meanwhile, the idea that magnesium is a modulator of
specificity in Ral can be tested, since the affinity of Sec5
Are the Switch Conformations in the Crystal
One of the exciting outcomes of structure is that it often
opens new venues for further research. Our structures
of Ral are consistent with intriguing hypotheses. We
present them here in the form of speculative ideas that
may drive research in new directions.
The first idea is that the ? turn conformation observed
? sheet described above could easily be a mimic of the
way switch II interacts in complexes with other proteins,
since the interface involves primarily backbone interac-
tions. R79 stabilizes the type I conformation of the ?
turn also through backbone interactions. Furthermore,
the turn residue Y75 is buried in the molecular interface
in analogy toY64 in Ras, which has been shown to be
buried in the interface between Ras and Sos and to
contribute significantly to binding affinity (Boriack-
Sjodin et al., 1998). Most interestingly, the ? turn confor-
mation brings the Ral tree-determinant residue I78 into
proximity to other Ral tree-determinant residues in the
site identified between switch II and ?3, making it part
of the hydrophobic core that lines this pocket and com-
pletely changing the electrostatic character of the
to the switch II conformation found in the Ral-GppNHp/
Sec5 structure could be explained by the fact that the
truncated Sec5 binds primarily at switch I and therefore
does not affect the conformation of switch II, which is
partially disordered in that structure (Fukai et al., 2003).
The relationship between Q72 (Q61 in Ras numbering)
and the nucleotide in the ? turn switch II structure is
very different than it is in Ras. Rather than facing the
nucleotide as it does in Ras-GppNHp, Q72 in our hypo-
thetical switch II complex mimic is turned away from
the nucleotide and therefore is excluded from playing
tentwith Q72affectingthehydrolysis reaction?Interest-
ingly, while RalG23V shows a 10-fold decrease in intrinsic
GTPase activity, as does RasG12V, the mutant RalQ72L
shows only a 2-fold decrease in intrinsic GTPase activity
Ral GTPase Binding Sites
the initial data sets, and noncrystallographic symmetry was used
in the beginning stages of refinement.
tion data set collected at 100 K on the Mar345 detector mounted
on a GX-13 X-ray generator in our laboratory. The search model
used with the Ral-GppNHp data set truncated at a resolution of
3.0 A˚was the 1.26 A˚structure of Ras-GppNHp (PDB code 1CTQ).
A Matthews coefficient of 4.5, 2.2, and 1.5 for 1, 2, and 3 molecules,
respectively, indicated the presence of two molecules of Ral-
GppNHp in the asymmetric unit. The first peak search identified the
peak search to identify Molecule B in the asymmetric unit. The Ras
residues were changed to those of Ral, and after a round of rigid
body refinement, simulated annealing was carried out with CNS by
using the mlf target function (Adams et al., 1997). The model was
improved by iterative cycles of model building in O, followed by
positional and individual restrained B factor refinement. This model
was further refined against the 1.5 A˚ data set collected at APS.
Similar molecular replacement strategies were used to solve the
structures for crystal forms 1 and 2 of Ral-GDP, by using the Ral-
GppNHp structure as the search model. The final R factors are given
in Table 1.
or Exo84 for Ral-GppNHp should be dependent on the
magnesium concentration. In any case, in order to test
the various aspects of the hypotheses presented here,
it will be necessary to complement the structures with
additional experiments that reveal the behavior of Ral
both in solution and in the cell.
Cloning, Expression, and Purification
The gene for full-length simian RalA was provided by Larry Feig as
a GST fusion in the pGEX2T vector. Simian RalA is identical in
sequence to the human protein, except for one conservative muta-
tion: D147 (simian) to E147 (human). The construct corresponding
to the truncated version of the protein (residues 1–178) was gener-
ated by PCR and was cloned into the pET21a(?) vector (Novagen).
The new vector was then transformed into E. coli BL21 Rosetta cells
(Novagen). The cells were grown in LB broth at 37?C with shaking
at 225 rpm to an OD600 of approximately 0.7–0.8 and were then
induced with 0.15 mM isopropyl-?-D-thiogalactopyranoside (IPTG)
for 5 hr at 32?C. Cells were harvested by centrifugation for 5 min at
4,000 rpm at 4?C and were stored at ?80?C. Frozen cell pellets from
a 6 L culture were thawed on ice and resuspended in 50 ml 20 mM
HEPES (pH 7.6), 50 mM NaCl, 5 mM MgCl2,1 mM DTT, 5% glycerol,
and 10 ?M GDP (Buffer A) with protease inhibitors (5 mM benzami-
dine, 1 mM pefabloc, 2 ?g/ml antipain, 1 ?g/ml leupeptin, 1 ?g/ml
pepstatin A). The cells were sonicated and centrifuged for 20 min
at 20,000 rpm at 4?C. Purification of Ral from the cell lysate was
done as described previously for Ras (Campbell-Burk and Carpen-
ter, 1995), with the following modifications: HEPES at pH 7.6 was
used instead of Tris at pH 8.0; Ral is eluted off the column with a
200 ml gradient of 0%–40% Buffer B in Buffer A (above), where
chromatography, the fractions containing the protein were pooled
and applied to a Q Sepharose HP 5ml column (Pharmacia) at a rate
of 1 ml/min and eluted with a 110 ml gradient of 0%–11% Buffer B
in Buffer A. The yield from a 6 L culture was 50–80 mg protein. The
truncated Ral-GDP was then exchanged into 10 mM HEPES (pH 7.50),
10 mM NaCl, 5 mM MgCl2,1 mM DTE, and 1 ?M GDP and concen-
trated to 25 mg/ml for crystallization. Ral-GppNHp was obtained by
nucleotide exchange by using procedures published for Ras (John
et al., 1990) and concentrated to 25 mg/ml for crystallization.
Larry Feig donated the clone for the full-length simian RalA. C.M.
isgrateful forhis supportduring thebeginning stagesof thisproject.
Team (SER-CAT) 22-ID beamline at the Advance Photon Source,
Argonne National Laboratory. Use of the Advanced Photon Source
was supported by the U.S. Department of Energy, Office of Science,
Office of Basic Energy Sciences, under Contract No. W-31-109-
Eng-38. We are grateful to the staff at the SER-CAT beamline for
their guidance during data collection and to Greg Buhrman for help
during the initial stages of the Ral-GppNHp structure refinement.
This research is supported by a grant from the National Institutes
of Health (1 R01 CA096867-01A1).
Received: June 21, 2004
Revised: August 26, 2004
Accepted: August 28, 2004
Published: November 9, 2004
Crystallization and Data Collection
CrystalsofRal-GppNHp andofRal-GDPwereobtained bythevapor
diffusion method byusing 4 ?l drops containinghalf protein solution
and half reservoir over a 0.5 ml reservoir at 18?C. The reservoir for
Ral-GppNHp crystallization contained 0.2 M magnesium formate,
20% PEG 3350, and that for Ral-GDP crystallization contained 100
mM sodium acetate (pH 4.6), 8% PEG 4000. A 40% glycerol solution
was added to the drop until the crystals were in about 15% glycerol
before transfer to liquid nitrogen. Diffraction data for Ral-GppNHp
and for crystal form 1 of Ral-GDP were collected at 100 K at the
SER-CAT ID-22 beamline at APS (Argonne, IL), on a Mar 135 CCD
detector. The data were processed with HKL2000 (Otwinowski and
Minor, 1997), and the statistics are shown in Table 1. Data for Ral-
GDP crystal form 2 were collected at 4?C with a Mar 345 phospho-
at 40 kV/50 mA. Data were processed with HKL, and the statistics
are shown in Table 1.
Adams, P.D., Pannu, N.S., Read, R.J., and Brunger, A.T. (1997).
Cross-validated maximum likelihood enhances crystallographic
Bauer, B., Mirey, G., Vetter, I.R., Garcia-Ranea, J.A., Valencia, A.,
Wittinghofer, A., Camonis, J.H., and Cool, R.H. (1999). Effector rec-
ognition by the small GTP-binding proteins Ras and Ral. J. Biol.
Chem. 274, 17763–17770.
Boriack-Sjodin, P.A., Margarit, S.M., Bar-Sagi, D., and Kuriyan, J.
(1998). The structural basis of the activation of Ras by Sos. Nature
ing interconnectivity of Ras, Rap1 and Ral. EMBO J. 17, 6776–6782.
Brunger, A.T. (1997). Free R value: cross-validation in crystallogra-
phy. Methods Enzymol. 277, 366–396.
Brunger, A.T., Adams, P.D., Clore, G., DeLano, W., Gros, P., Grosse-
Kunstleve, R., Jiang, J., Kuszewski, J., Nilges, M., Panni, N., et al.
(1998). Crystallography and NMR System: a new software suite for
macromolecular structure determination. Acta Crystallogr. A47,
Buhrman,G.,de Serrano,V.,andMattos,C. (2003).Organicsolvents
order the dynamic switch II in Ras crystals. Structure (Camb). 11,
fication of Ras proteins. Methods Enzymol. 255, 3–13.
Cantor, S., Urano, T., and Feig, L. (1995). Identification and charac-
Phasing and Refinement
Phases for all structures were obtained by using the molecular re-
placement method with peak searches performed with AMoRe (Na-
vaza, 1994) in the CCP4-4.2 software package (CCP4, 1994). The
program Crystallography and NMR System (CNS) (Brunger et al.,
1998) was used for all reciprocal space refinement, with a randomly
selected 10% of the unique reflections reserved for the calculation
of Rfree(Brunger, 1997). The program O (Jones et al., 1991) was used
for manual rebuilding of the models with visualization of Fo? Fcand
2Fo? Fcelectron density maps. The self-rotation was calculated for
terization of Ral-binding protein 1, a potential downstream target
of Ras GTPases. Mol. Cell. Biol. 15, 4578–4584.
Casari, G., Sander, C., and Valencia, A. (1995). A method to predict
functional residues in proteins. Nat. Struct. Biol. 2, 171–178.
CCP4 (Collaborative Computational Project, Number 4) (1994). The
CCP4 suite: programs for protein crystallography. Acta Crystallogr.
D Biol. Crystallogr. 50, 760–763.
Cherfils, J., Menetrey,J., Le Bras, G., Janoueix-Lerosey,I., de Gunz-
burg, J., Garel, J.R., and Auzat, I. (1997). Crystal structures of the
small G protein Rap2A in complex with its substrate GTP, with GDP
and with GTP?S. EMBO J. 16, 5582–5591.
Drugan, J.K., Khosravi-Far, R., White, M.A., Der, C.J., Sung, Y.J.,
Hwang, Y.W., and Campbell, S.L. (1996). Ras interaction with two
distinct binding domains in Raf-1 may be required for Ras transfor-
mation. J. Biol. Chem. 271, 233–237.
Emkey, R., Freedman, S., and Feig, L. (1991). Characterization of a
GTPase-activating protein for the Ras-related Ral protein. J. Biol.
Chem. 266, 9703–9706.
Feig, L.A. (2003). Ral-GTPases: approaching their 15 minutes of
fame. Trends Cell Biol. 13, 419–425.
Frech, M., Schlichting, I., Wittinghofer, A., and Chardin, P. (1990).
Guanine nucleotide binding properties of the mammalian RalA pro-
tein produced in Escherichia coli. J. Biol. Chem. 265, 6353–6359.
Fukai, S., Matern, H.T., Jagath, J.R., Scheller, R.H., and Brunger,
A.T. (2003). Structural basis of the interaction between RalA and
Sec5, a subunit of the se6/8 complex. EMBO J. 22, 3267–3278.
Hancock, J.F. (2003). Ras proteins: different signals from different
locations. Nat. Rev. Mol. Cell Biol. 4, 373–384.
Heo, W.D., and Meyer, T. (2003). Switch-of-function mutants based
on morphology classification of Ras superfamily small GTPases.
Cell 113, 315–328.
Ito, Y., Yamasaki, K., Iwahara, J., Terada, T., Kamiya, A., Shirouzu,
M., Muto, Y., Kawai, G., Yokoyama, S., Laue, E.D., et al. (1997).
Regional polysterism in the GTP-bound form of the human c-Ha-
Ras protein. Biochemistry 36, 9109–9119.
Jiang, H., Luo, J.-Q., Urano, T., Frankel, P., Lu, Z., Foster, D., and
Feig, L.A. (1995). Involvement of Ral GTPase in v-Src-induced phos-
pholipase D activation. Nature 378, 409–412.
John, J., Schlichting, I., Schiltz, E., Rosch, P., and Wittinghofer, A.
(1989). C-terminal truncation of p21H preserves crucial kinetic and
structural properties. J. Biol. Chem. 264, 13086–13092.
John, J., Sohmen, R., Feuerstein, J., Linke, R., Wittinghofer, A.,
and Goody, R.S. (1990). Kinetics of interaction of nucleotides with
nucleotide-free H-ras p21. Biochemistry 29, 6058–6065.
Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991).
Improved methods for building protein models in electron density
maps and the location of errors in these models. Acta Crystallogr.
A 47, 110–119.
Jullien-Flores, V., Dorseuil, O., Romero, F., Letournaeur, F., Sara-
gosti, S., Berger, R., Tavitian, A., Gacon, G., and Camonis, J.H.
with CDC42/Rac GTPase-activating protein activity. J. Biol. Chem.
Mattos, C., Petsko, G.A., and Karplus, M. (1994). Analysis of two-
residue turns in proteins. J. Mol. Biol. 238, 733–747.
mura, S., and Kim, S.-H. (1990). Structural differences between ac-
tive and inactiveforms of protooncogenic rasproteins. Science 247,
Moskalenko, S., Henry, D.O., Rosse, C., Mirey, G., Camonis, J.H.,
and White, M.A. (2002). The exocyst is a Ral effector complex. Nat.
Cell Biol. 4, 66–72.
Nassar, N., Horn, G., Herrmann, C., Scherer, A., McCormick, F., and
and a GTP analogue. Nature 375, 554–560.
Navaza, J. (1994). AMoRe: an automated package for molecular
replacement. Acta Crystallogr. A 50, 157–163.
Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and
association: insights from the interfacial and thermodynamic prop-
erties of hydrocarbons. Proteins 11, 281–296.
Otwinowski,Z., andMinor, W.(1997). Processingof X-raydiffraction
data collected in oscillation mode. Methods Enzymol. 276, 307–326.
Pai, E., Krengel, U., Petsko, G.A., Goody, R., Kabsch, W., and Witting-
hofer,A. (1990). Refined crystalstructure of the triphosphate confor-
mation of H-ras p21 at 1.35 A˚resolution: implications for the mecha-
nism of GTP hydrolysis. EMBO J. 9, 2351–2359.
Quilliam, L.A., Zhong, S., Rabun, K.M., Carpenter, J.W., South, T.L.,
Der, C.J., and Campbell-Burk, S. (1995). Biological and structural
characterization of a Ras transforming mutation at the phenylala-
nine-156 residue, which is conserved in all members of the Ras
superfamily. Proc. Natl. Acad. Sci. USA 92, 1272–1276.
Rijkers, G.T., Henriquez, N., and Griffioen, A.W. (1993). Intracellular
magnesium movements and lymphocyte activation. Magnes. Res.
Shurki, A., and Warshel, A. (2004). Why does the Ras switch “break”
by oncogenic mutations? Proteins 55, 1–10.
Urano, T., Emkey, R., and Feig, L.A. (1996). Ral-GTPases mediate
a distinct downstream signaling pathway from Ras that facilitates
cellular transformation. EMBO J. 15, 810–816.
Vojtek, A.B., and Der, C.J. (1998). Increasing complexity of the Ras
signaling pathway. J. Biol. Chem. 273, 19925–19928.
Wang, K.L., and Roufogalis, B.D. (1999). Ca2?/calmodulin stimu-
lates GTP binding to the ras-related protein ral-A. J. Biol. Chem.
Wang, K.L., Khan, M.T., and Roufogalis, B.D. (1997). Identification
and characterization of a calmodulin-binding domain in Ral-A, a
Ras-related GTP-binding protein purified from human erythrocyte
membrane. J. Biol. Chem. 272, 16002–16009.
Wilmot, C.M., and Thornton, J.M. (1988). Analysis and prediction of
Winkler, D.G., Johnson, J.C., Cooper, J.A., and Vojtek, A.B. (1997).
Identification and characterization of mutations in Ha-Ras that se-
lectively decrease binding to cRaf-1. J. Biol. Chem. 272, 24402–
Wittinghofer, A., and Nassar, N. (1996). How Ras-related proteins
talk to their effectors. Trends Biochem. Sci. 21, 488–491.
The coordinates of Ral-GppNHp have been deposited in the Protein
Data Bank with the entry code 1U8Y. Coordinates for Ral-GDP have
entry codes 1U8Z and 1U90, corresponding to crystal forms 1 and