Molecular Cell, Vol. 10, 857–869, October, 2002, Copyright 2002 by Cell Press
Transcription Corepressor CtBP Is
an NAD?-Regulated Dehydrogenase
was initially identified as a partner of adenovirus E1A
protein and derives its name from its ability to bind a
PXDLS sequence at the E1A C terminus (Boyd et al.,
1993; Schaeper et al., 1995). The binding of CtBP to
E1A results in the loss of CR1-mediated transactivation,
behavior identical to that of a transcriptional repressor.
Genetic support for the ability of CtBP to function in
vivo as a transcription corepressor was provided by
experiments carried out in Drosophila. Drosophila CtBP
(dCtBP) is maternally expressed and uniformly distrib-
uted throughout the early embryo. Mutations in dCtBP
cause severe segmentation and patterning defects that
have been attributed to the combined loss of repression
activities of Knirps, Kru ¨ppel, and Snail; factors critical
for early development and repression of genes such as
even skipped, rhomboid, and fushi tarazu (Nibu et al.,
1998a, 1998b; Poortinga et al., 1998). All three of these
sequence-specific repressors contain PXDLS-related
motifs that have been shown in vitro and in vivo to
be important in recruiting dCtBP. CtBP1 and CtBP2, a
second highly related factor in vertebrates, have been
linked toa host of disparatetranscription factors includ-
ing several that are important in cellular proliferation,
homeostasis, and development by a conserved PXDLS-
like motif (Chinnadurai, 2002).
In contrast to other corepressors, such as N-CoR/
SMRT and mSin3, which have no intrinsic enzymatic
activity but instead recruit enzymatic components such
by its unexpected sequence similarity to D2-hydroxy-
acid dehydrogenases (D2-HDHs) (Schaeper et al., 1995;
Turner andCrossley, 2001).Remarkably, mostof human
nus) can be aligned with this subfamily of NAD?-depen-
dent dehydrogenases, which includes D-glycerate de-
hydrogenase (D-GDH) and D-lactate dehydrogenase
(D-LDH) among others (Goldberg et al., 1994; Stoll et al.,
1996). This unusual similarity to D2-HDHs has prompted
speculation that CtBP may also bind NAD?and possess
dehydrogenase activity—relevant to its corepression
function. However, initial attempts to define NAD?bind-
ing and dehydrogenase activity in CtBP have been un-
successful (Schaeper et al., 1995). The lack of a known
D2-hydroxyacid substrate,and recent reportsthat CtBP
can induce fission of Golgi membrane by acetylating
lysophosphatidic acid (Spanfo ` et al., 1999; Weigert et
al., 1999), and that RIBEYE, a component of ribbon syn-
apses, is a splice variant of CtBP2 (Schmitz et al., 2000),
have led to further uncertainty regarding the actions of
In this manuscript, we show that CtBP is a bone fide
D2-hydroxyacid dehydrogenase, with a characteristic
dumbbell shape and a deep, narrow cleft for both NAD?
and substrate binding. We further demonstrate that the
dehydrogenase domain is necessary and sufficient to
mediate repression and identify the NAD?and putative
substrate interactions based on the structure of the de-
hydrogenase domain determined in the presence of
NAD?. Furthermore, we show that E1A binding to CtBP
Vivek Kumar,1,2,6Justin E. Carlson,4,6Kenneth A. Ohgi,2
Thomas A. Edwards,4David W. Rose,3
Carlos R. Escalante,4Michael G. Rosenfeld,2,5
and Aneel K. Aggarwal4,5
1Department of Biology Graduate Program
2Howard Hughes Medical Institute
School and Department of Medicine
3Department of Medicine
University of California, San Diego
La Jolla, California 92093
4Structural Biology Program
Department of Physiology and Biophysics
Mount Sinai School of Medicine
New York, New York 10029
Transcriptional repression is based on the selective
actions of recruited corepressor complexes, including
those with enzymatic activities. One well-characterized
developmentally important corepressor is the C-ter-
minal binding protein (CtBP). Although intriguingly re-
lated in sequence to D2 hydroxyacid dehydrogenases,
the mechanism by which CtBP functions remains un-
clear. We report here biochemical and crystallographic
studies which reveal that CtBP is a functional dehy-
drogenase. In addition, both a cofactor-dependent
conformational change, with NAD?and NADH being
equivalently effective, and the active site residues are
tion motif on repressors, such as E1A and RIP140.
Together, our data suggest that CtBP is an NAD?-
repression events in cells.
Transcriptional repression is mediated by a wide variety
of DNA binding transcription factors, serving critical
roles in development and homeostasis. In turn, these
repressors often appear to require association with co-
repressors to mediate inhibition of gene transcription
and Kadonaga, 1999). These corepressors include closely
related factors present in multiple complexes, such as
histone deacetylases (HDACs), that link repression to
chromatin structure and protein modification. Many en-
zymatic activities are linked to recruited cofactor com-
plexes: these include acetylation and phosphorylation,
ADP ribosylation, and methylation (Cheung et al., 2000;
Jenuwein and Allis, 2001).
The discovery of the C-terminal binding protein
(CtBP), because of its sequence homology to D2-
hydroxyacid dehydrogenases, potentially presents an-
5Correspondence: firstname.lastname@example.org (A.K.A), mrosenfeld@
6These authors contributed equally to this work.
is NAD?dependent, requiring active site residues and a
for RIP140-dependent repression of retinoic acid recep-
ical results suggest an allosteric mechanism for recruit-
ment of CtBP to its consensus recognition motif for
specific repression events in cells.
this interaction is specific to the PXDLS motif of E1A,
as a peptide containing this motif is able to effectively
compete with this interaction. Thus, the dehydrogenase
and mediate functional repression.
Structure of CtBP
Giventhatthe dehydrogenasedomainofCtBP isafunc-
tional dehydrogenase and that it alone mediates repres-
and purified the 28–353 residue minimal domain from
bacteria in the presence of NAD?, and crystals of the
complex were obtained from solutions containing so-
dium formate and magnesium acetate. The structure
was solved by multiwavelength anomalous diffraction
(MAD) method (Hendrickson, 1991) using selenomethio-
nine-labeled protein expressed in a strain of E.coli that
is methionine auxotrophic (Table 1). The current model,
refined to 1.95 A˚resolution, includes residues 28–352
with good stereochemistry. The crystallographic asym-
metric unit contains a CtBP monomer that forms exten-
CtBP is thus a dimer, where each monomer is divided
into large and small domains separated by a flexible
hinge region (Figure 2). A cleft at the confluence of the
two domains provides binding sites for NAD?as well
as a putativesubstrate. Because all ofthe NAD?binding
residues stem from the large domain, it has been called
the NAD?or coenzyme binding domain in previous D2-
HDH structures. The small domain will be referred to as
(?32 ? 25 ? 29 A˚) and composed of residues from both
the N (aa 27–121) and C terminus (aa 327–352). The NAD?
elongated in shape (?53 ? 33 ? 39 A˚), and mediates
most of the dimerization contacts (Figure 2). The dimer-
accessible surface area per monomer (Figure 2B). The
hinge between the NAD?and substrate binding domains
is composed of two segments, amino acids 122–124
The NAD?binding domain is the most similar to that
of other D2-HDHs (Dengler et al., 1997; Goldberg et al.,
1996), composed of a parallel ? sheet (?A-?G) flanked
on both sides by ? helices (?A-?H). The connectivity
resembles a Rossmann or a dinucleotide binding fold
(Rao and Rossmann, 1973). This similarity in connectiv-
binding as well as the conservation of a GXGXXG(17X)D
signature motif. In comparing CtBP against other D2-
HDH structures, the most pronounced deviation is the
omission of a 15 residue insert found between amino
acids 265 and 281 in D-LDH and amino acids 263 and
279 in D-2-hydroxyisocaproate dehydrogenase (D2-
HicDH) (see for instance, Dengler et al., 1997; Stoll et
al., 1996). The exclusion of this segment in CtBP may
be important in allowing the binding of the PXDLS se-
quence (discussed below). Otherwise, the CtBP NAD?
binding domain superimposes extremely well with other
D2-HDHs, with rmsds ranging from 1.1 to 1.2 A˚. The
CtBP Is a Functional Dehydrogenase
We first asked whether CtBP is a functional dehydroge-
tating change of NADH to NAD?or vise versa, which
can be monitored by loss or gain of absorbance at 340
nm (Adams et al., 1973). Most dehydrogenases have a
strict substrate specificity; however, they will catalyze
nonoptimal substrates at slower rates. We used an
assay that coupled the reduction of pyruvate to lactic
acid with the oxidation of NADH to NAD?. For these
SF9 cells, yielding essentially a homogenous protein
(Figure 1C). CtBP was able to catalyze this reaction in
a dose-dependent manner (Figure 1A). Even though the
reaction was inefficient, it is specific to CtBP, as both an
equivalent volume of uninfected SF9 cell extract (Figure
1B), and a catalytic site mutant of CtBP (Figure 6D)
were not able to catalyze the reaction. In addition, the
ting that the substrate is not CtBP itself, anything in the
SF9 extract, or buffer (Figure 1B). Thus using a nonopti-
mal substrate, we were able to demonstrate that CtBP
is a functional dehydrogenase.
The Dehydrogenase Domain Alone Is a Repressor
All but the last 90 aa of 440 residue hCtBP1 has signifi-
cant homology to D2-HDHs. Having shown that CtBP
is a functional dehydrogenase, we sought to determine
what role the dehydrogenase domain alone has in re-
pression and recruitment of CtBP to a PXDLS motif. We
used two independent assays to test whether the CtBP
dehydrogenase domain is a repressor. When tethered
to DNA by a Gal4 DNA binding domain (DBD), the dehy-
drogenase domain of CtBP represses as well as full-
length CtBP (Figure 1D, compare lanes 2 and 3), and
the last 90 aa of CtBP1 (C?) has no significant repressive
we used CtBP to repress Gal4DBD/E1A-mediated acti-
vation from a UAS/tk luciferase. We also used the C?
(last 78 aa) of E1A, which was used to clone CtBP in a
two-hybrid screen and is not a repressor until CtBP is
overexpressed. The dehydrogenase domain only was
able to repress Gal4/E1A C?-mediated transcription as
well as full-length CtBP1 or CtBP2 (Figure 1E, compare
lane 4 with lanes 2 and 3).
Because the dehydrogenase domain alone is capable
act with E1A. We used GST E1A C? (the same region
and translated (TnT) CtBP full-length or dehydrogenase
domain to test direct binding. The dehydrogenase do-
main alone is sufficient to interact with E1A (Figure 1F)
as robustly as full-length CtBP1 or CtBP2. Furthermore,
CtBP Is an NAD?-Regulated Dehydrogenase
Figure 1. CtBP Is a Functional Dehydrogenase
(A) Dehydrogenase assays were performed to measure the conversion of pyruvate to lactic acid and coupled oxidation of NADH to NAD?as
described in Experimental Procedures. Varying amounts of full-length CtBP purified from baculovirus-infected SF9 cells were incubated with
20 mM pyruvate and the change in absorbance at 340 nm was measured over time.
(B) As control an equivalent volume of SF9 extract from uninfected cells was used as well as a reaction without any substrate.
(C) Coomassie stain of the CtBP is shown.
(D–F) The dehydrogenase domain of CtBP is sufficient to bind PXDLS motif and mediate repression. (D) When fused to the Gal4/DBD full-
length CtBP and dehydrogenase domain (DD) alone can repress transcription (lane 2 and 3, respectively) from a UAS/tk luciferase reporter.
The C terminus of CtBP1 is not capable of being a repressor as a gal fusion (lane3). (E) The dehydrogenase domain can also repress
transcription when recruited by gal4/E1A C?. The dehydrogenase domain can effectively repress transcription as well as full-length CtBP or
CtBP2 (compare lane 3 with 1 and 2). (F) The dehydrogenase domain binds to GST/E1A C? equivalently as full-length CtBP1 and CtBP2. This
interaction can be disrupted by the addition of a peptide containing the PXDLS motif (lane 3). Results are shown as mean ? standard deviation
and are representative of three independent experiments.
substrate binding domain is more variable with rmsds
ranging from 1.3 to 1.65 A˚. This variability is further
enhanced by loops extending into the active site cleft,
consistent with different substrate specificities.
adenine moiety is oriented toward the entrance to the
cleft while the nicotinamide ring is more deeply buried
toward the substrate binding pocket (Figure 2A). Resi-
the adenine ribose, the pyrophosphate group, and the
carboxyamide group of the nicotinamide, respectively
(Figure 3A). In addition to side chain atoms, main chain
carbonyl (Cys237 and Thr264) and amide (Arg184, Val185,
and Trp318) groups also contribute to NAD?binding (Fig-
Active Site Cleft
NAD?binds in the active site cleft in the characteristic
dent dehydrogenases (Eklund and Branden, 1987). The
Table 1. Data Collection, Phasing, and Refinement Statistics
Number of reflections
Data coverage (%)
MAD Phasing Statistics
Number of sites
FoM (centric/acentric) 3.2 A˚c
FoM (DM) 2.25 A˚d
Resolution range (A˚)
Reflections, F ? 2? (F)
Average B factor (A˚2)
aValues for outermost shell are given in parentheses.
bRmerge? ? |I ? ?I?|/?I, where I is the integrated intensity of a given reflection.
cFoM ? mean figure of merit computed to 3.2 A˚.
dFoM ? overall mean figure of merit at 2.25 A˚after density modification.
eRcryst? ? ||Fo| ? |Fc||/? |Fo|.
fRfreewas calculated using 10% of data excluded from refinement.
ure 3A). The structure reveals why glycines at positions
181and 183(GXGXXG(17X)D)arecritical forNAD?bind-
ing, since substitution by any other amino acids at these
positions would cause steric clashes with the bridging
pyrophosphate bridge. The adenine is nestled in a hy-
drophobic cavity defined by residues Pro205 and Tyr206,
with its N7 atom accepting a hydrogen bond from Asn240
A His/Glu(Asp)/Arg triad is conserved in all D2-HDHs
and implicated as the center for substrate binding and
dues in CtBP are His315, Glu295, and Arg266. The histi-
dine is postulated to be the acid/base catalyst, with
the glutamate/aspartate helping to lower the pKa of the
is proposed to polarize the 2-hydroxyl of the substrate
for catalysis. While a biologically relevant substrate for
CtBP remains to be identified, the structure provides
valuable insights into the nature of a putative substrate.
proate in the CtBP active site (Figure 3B) (Dengler et al.,
1997). The nonaliphatic portion of 2-oxoisocaproate is
found to be accommodated remarkably well in the CtBP
active site, and a plausible hydrogen bonding scheme
can be derived in which Arg97 and Ser100 form hydro-
gen bonds with the terminal carboxylate group, and
Arg266 is in a position to form a bond with the 2-hydroxyl.
Interestingly, Arg97 and Ser100 are primarily hydropho-
bic residues in other D2-HDHs. In D2-HicDH, the termi-
nal carboxylate is instead recognized by Tyr100, which
is an alanine (Ala123) in CtBP (Figure 3B). Thus, CtBP
may use a different set of amino acids to fix the orienta-
tion of a D-2-hydroxyacid substrate in its active site
as compared to other D2-HDHs. His315 and Arg266
are ?4 A˚ from the 2-hydroxyl group of the modeled
substrate. The aliphatic portion of modeled 2-oxoisoca-
NAD?binding domain and His77 that extends from the
substrate binding domain. Both of these residues are
unique to CtBP and probably contribute to its substrate
specificity (Figure 3B). The steric clashes with these
acidsubstrate thatissomewhat smallerinthe “R”group
than D-2-hydroxyisocaproate. Also, the lack of basic
to Arg60 in phosphoglycerate dehydrogenase (PGDH)
to a substrate lacking a phosphate group.
Curiously, we observe an acetate ion in the CtBP sub-
CtBP Is an NAD?-Regulated Dehydrogenase
Figure 2. CtBP Structure
(A) The dehydrogenase domain contains a substrate binding domain linked via a flexible hinge to the NAD?binding domain. NAD?(yellow)
binds in the active site cleft in a bent L-shaped configuration.
(B) The van der Waals (vdw) surface of a CtBP dimer. One monomer is drawn in gray and the other in green. The gray monomer is related to
the monomer in (A) by rotation of ?81?, roughly along the vertical axis of the paper, to give a view of the CtBP dimer down the 2-fold axis.
The NAD?molecules are hidden from view by the vdw surface; their approximate positions are indicated by dashed ovals. Note, the NAD?
binding domain makes the majority of dimeric interactions.
strate binding pocket (Figure 3B), hydrogen bonded to
His315 (2.8 A˚), Arg266 (2.8 and 2.9 A˚), and Arg97 (2.8 A˚).
This prompted us to test whether CtBP could catalyti-
cally modify an acetyllysine residue on a histone H3-
derived peptide, but we found no evidence. An acetate
ion is similar to a 2-hydroxyacid substrate in containing
a terminal carboxylate group, but it lacks the 2-hydroxyl
group. The acetate ion, present in the crystallization
mix and critical for CtBP crystallization, is bound in an
orientation different from what we expect for a real sub-
tion anticipated for oxidation/reduction of the 2-hydroxyl
ity to 2-hydroxyacids appear to permit tight binding in
the CtBP active site.
corresponding to a 7.5? rotation around the hinge region
(Lamzin et al., 1994). Similar rigid body motion has been
described for other NAD?-dependent dehydrogenases,
where the apo form has been termed the “open” confor-
mation and the NAD?-bound form as the “closed” con-
formation (Grau et al., 1981; Lamzin et al., 1994). In
comparing our structure against other NAD?-bound D2-
HDHs, the active site cleft is particularly narrow. For
example, the average distance across the cleft (mea-
sured between residues 101 and 266, and 78 and 294)
is ?10 A˚, as compared to ?16 A˚in D-LDH and ?13 A˚
in D-HicDH. Only FDH has a narrower active site cleft
(?8 A˚), which may reflect the small size of its formate
ion substrate (Lamzin et al., 1994).
A comparison with apo-D-GDH structure—based on
the alignment of NAD?binding domains—suggests that
the CtBP substrate binding domain moves toward the
closed conformation by a rotation of ?5? (Figure 4A).
To evaluate the potential role of NAD?-induced confor-
uated the effects of adding NAD?to binding of E1A to
CtBP. We found that in the presence of NAD?, CtBP
bound to E1A much more efficiently (Figure 5A). This
NAD?dependency was observed both with the full-
length CtBP,as well aswith theisolated dehydrogenase
E1A Is Recruited through a NAD?-Dependent
den, 1987). In the majority of cases, NAD?binding
causes a narrowing of the active site cleft due to a
movement of the substrate binding domain toward the
cleft. The structures of holo- and apo-FDH, for instance,
differ in the location of the substrate binding domain,
Figure 3. CtBP Interactions
(A) Interactions between bound NAD?(yellow) and CtBP residues. Both side chain (Arg184, Asp204, Asn240, and Asp290) and main chain
(Arg184, Val185, Cys237, Thr264, and Trp318) atoms make hydrogen bonds (dotted lines) with NAD?.
(B) The 2-oxoisocaproate substrate (green) bound to D-HicDH (left panel) is positioned in CtBP (middle panel) based on superposition of the
two enzymes. The His315(295)/Glu295(Asp263)/Arg266(234) catalytic triad is conserved between the enzymes, but the other residues lining
the substrate binding pocket are different. The nicotinamide ring of NAD?is included in the panels to aid orientation with respect to (A). The
CtBP structure reveals a bound acetate ion (right panel). The carboxylate group of the acetate ion is oriented differently from the carboxylate
of the modeled 2-oxoisocaproate molecule (middle panel).
sary for binding of the PXDLS motif.
Recently it has been reported that the ratio of NAD?
and NADH can regulate the function of DNA binding
transcription factors (Rutter et al., 2001). In addition,
Zhang et al. (2002) have reported that CtBP binds E1A
two to three orders of magnitude better in the presence
of NADH than NAD?. Given these intriguing findings,
we tested whether E1A binding to CtBP is regulated
differently by NADH versus NAD?. We quantitated the
interaction between TnT CtBP and GST-E1A under dif-
ferent concentrations of NAD?, NADH, and various
NAD?-likemolecules. BothNAD?andNADH werefound
to be equally effective in stimulating E1A-CtBP interac-
ulation in the 1–10 ?M range (Figures 5E and 5F). ADP-
ribose, an NAD?/NADH analog lacking the nicotinamide
ring, also increased E1A-CtBP interaction, though the
concentration requiredfor half-maximalstimulation was
?10? higher than that for NAD?or NADH (Figure 5E).
Titration with ?-nicotinamide mononucleotide (NMN) or
the nicotinamide ring alone, however, was not sufficient
to stimulate E1A-CtBP interaction. Thus, in our assay,
NAD?and NADH appear to be equally effective in trig-
gering the conformational switch in CtBP for E1A bind-
ing. Furthermore, compounds such as ADP-ribose that
are capable of causing conformational change, from
open to closed state, in dehydrogenases can also medi-
ate E1A binding.
To appraise this NAD?/NADH-dependent conforma-
CtBP Is an NAD?-Regulated Dehydrogenase
Figure 4. Conformational Change and PXDLS Binding
(A) The CtBP/NAD?complex (blue) is compared to apo-D-GDH (red) solved in the absence of NAD?(Goldberg et al., 1994). The superposition,
based on NAD?binding domains, reveals a relative displacement in the substrate binding domains, marking “open” and “closed” states. The
N and C termini of CtBP are labeled. NAD?bound to CtBP is drawn in yellow.
(B) Location of residues mutated in Cleftmut(blue) and Catmut(red). Cleftmutmutations of residues F102, I107, and K108, lining a cavity at one
the active site cleft, do not affect E1A binding or repression function. Catmutmutations of the catalytic triad residues (H315, E295, and R266)
and D290 disrupt both E1A binding and repression activity. The PXDLS recognition motif is hypothesized to bind close to these catalytic
residues, aided by interactions from a loop (green) from the 2-fold related subunit. The bound NAD?is drawn in yellow.
tional change further, we undertook limited proteolysis
of TnT CtBP in the absence and presence of varying
concentrations of NAD?and NADH. The digestion was
carried out with the nonspecific protease papain that
to papain digestion, producing a very stable ?40 kDa
fragment, which we used as qualitative assay for NAD(H)-
dependent conformational change (Figure 5G). Using
this assay, we determined the concentration of NAD?
and NADH at which the resistant 40 kDa fragment was
produced. Consistent with our E1A binding data, the
protease digestion showed a conformational switch be-
tween 1 and 10 ?M NAD?and NADH, with both com-
pounds behaving identically (Figure 5G). Thus using two
independent assays, we could not detect any difference
in NAD?- and NADH-induced conformational change of
CtBP translated in an in vitro mammalian system.
dimerization function (Dimmut), and a cavity at one end of
G181V, and G183V for NADmut, and H315A, E295A,
R266A, and D290A for Catmut. Because of the extensive
dimer interface two sets of mutations were generated:
R141A, R142A, R163A, and R171A to disrupt critical
salt links and hydrogen bonding across the interface
(Dim1mut), and C134Y, N138R, R141E, and L150W to in-
troduce steric and electrostatic repulsion across the
interface (Dim2mut). Cleftmutincluded K108A, F102A, and
I107A mutations to disrupt residues lining a cavity near
the entrance to the active site cleft that we initially spec-
ulated might interact with the PXDLS motif (Figures 2B
and 4B). As expected, NADmutstrongly inhibited CtBP’s
NADH dependency observed above with the native en-
zyme. Unexpectedly, however, Catmutand Dimmutalso
severely compromised the ability of CtBP to bind E1A,
while Cleftmutdid not affect interaction with E1A (Figure
5A). Similar results wereobtained with CtBP holoprotein
and the isolated dehydrogenase domain. To test whether
the PXDLS motif actuallycompetes with substrate bind-
ing, we measured CtBP’s dehydrogenase activity in the
presence of a PXDLS peptide and observed little effect.
Taken together, the mutagenesis data suggest that the
PXDLS motif is accommodated outside of the substrate
binding pocket but sufficiently close to it to interact with
The PXDLS Motif Interacts Directly with the Active
In order to further test our hypothesis that the PXDLS
motif makes direct contacts with the CtBP dehydroge-
nase domain, we introduced point mutations, based on
the crystal structure, to disrupt NAD?binding (NADmut),
as well as the substrate/catalytic pocket (Catmut), the
Figure 5. NAD?and NADH Enhance Binding of CtBP to E1A, and This Binding and Functional Repression Requires Residues in the Catalytic,
NAD?Binding, and Dimerization Domains of CtBP
(A) GST E1A C? was used to the test the binding of various mutants of CtBP in the presence or absence of 1 mM NAD?/NADH. Addition of
NAD?/NADH consistently causes a marked enhancement of binding of wild-type (WT) CtBP holoprotein or the dehydrogenase domain only.
Mutations that disrupt the dimerization, catalysis, or NAD?binding abolished the NAD?. Panel on the right shows Coomassie staining of GST/
E1A C? as control for equal loading.
(B) Functional assay testing whether the mutants can repress Gal4/E1A C-based activation of UAS/tk luciferase. Both wild-type holoprotein
and the dehydrogenase domain above were able to repress transcription; however the mutants impaired in PXDLS binding were unable to
repress transcription. Expression levels of the various mutants were equivalent as seen by Western blot (Western blot panel below).
(C) CtBP inhibits ligand-dependent RAR activation via RIP140. Single-cell nuclear microinjection experiments were performed in Rat-1 cells.
Addition of retinoic acid (RA 10?8M) resulted in activation which was repressed by CMV/CtBP expression vector. This repression could be
reversed by the addition of purified ?RIP140 IgG and further restored by addition of CMV/RIP140 expression plasmid.
(D) Anti-CtBP IgG significantly reverses RIP-140-dependent suppression of RA-dependent gene activation, using the single-cell nuclear
(E) GST E1A C? binding to TnT CtBP at various concentrations of NAD?, NADH, ADP-Ribose, ?NMN, and Nicotinamide.
(F) Phosphoimager quantitation of the NAD?and NADH lanes in (E) demonstrating that NAD?and NADH are equally effective in stimulating
(G) Protease digestion of TnT CtBP at various concentrations of NAD?and NADH. Upon cofactor binding and conformational change, a
protease resistant 40 kDa band appears (arrow). Both NAD?and NADH are equivalently effective, and the band appears between 1 and 10
?M NAD(H). Results are shown as mean ? standard deviation and are representative of three independent experiments.
the active site residues. His315, Glu295, and Arg266 lie
at the confluence of the NAD?and substrate binding
domains and are in positions to interact with both a
buried substrate and a PXDLS peptide segment at the
rim of the active site cleft (Figures 3B and 4B). Interest-
and Arg266 is bordered by a loop from the 2-fold related
CtBP subunit that could make additional contacts with
the PXDLS motif (Figure 4B). The loss of these interac-
tions, as in Dimmut, could explain the importance of di-
merization in E1A binding. To determine whether the
CtBP mutants that were impaired in E1A binding were
also defective inrepression, we measured CtBP-depen-
dent repression of Gal4/E1A on a UAS-tk-dependent
the ability of expressed CtBP to repress Gal4/E1A, con-
when the mutant CtBP proteins were expressed at very
high levels, they were able to repress Gal4/E1A, al-
though not as efficiently as wild-type CtBP, arguing that
bound to E1A.
CtBP Is an NAD?-Regulated Dehydrogenase
CtBP Mediates RIP140-Dependent
Repression In Vivo
Nuclear receptors bind p140 and p160 factors off and
bound to LXXLL motifs (Cavailles et al., 1995; Heery et
al., 1997; Torchia et al., 1997). However, RIP140, at best,
is a weak activator (Cavailles et al., 1995). The observa-
tion that CtBP could bind to a variant PXDLS motif in
RIP140 (Vo et al., 2001), and this interaction might be
regulated by acetylation of an adjacent lysine residue
(Zhanget al.,2000), promptedus totest therequirement
for the catalytic residues of CtBP during repression by
nuclear receptor. We found that increasing the levels
of CtBP causes a complete block of ligand-dependent
activation by retinoic acid receptor (Figure 5C). How-
ever, mutations of the catalytic residues (Catmut) abol-
with a role for this domain in recruitment of CtBP to the
RAR activation complex. To determine whether this was
dependent upon RIP140, the ability of a specific ?RIP140
IgG to block CtBP-dependent repression was evaluated
using single-cell microinjection assay. We found that in-
jecting ?RIP140 antibody fully restored the ligand-
dependent activation and that the ?RIP140 antibody
had no effect on unliganded retinoic acid receptor tran-
scription unit(Figure 5D). Thus, CtBPdomains exhibited
similar behavior for both E1A and RIP140-dependent
repression events. Our results provide evidence that
CtBP is a functional dehydrogenase and that it utilizes
its dehydrogenase domain to bind to its recognition
mentation coefficientfor aCtBP monomeras calculated
with the hydrodynamic modeling program HYDROPRO
(Garcia de la Torre et al., 2000), using the coordinates
for a CtBP monomer. In contrast, the predominant peak
withCatmuthasasedimentation coefficientvalue of4.1S,
corresponding to a dimer. Together, these biophysical
measurements show (i) that Catmutand Dim2mutare prop-
erly folded and (ii) that Dim2mutis compromised in its
ability to dimerize.
Unlike Catmut and Dim2mut, NADmut and Dim1mut ex-
pressed in inclusion bodies in E. coli that limited their
biophysical analysis. As another measure of structure,
TnT mutant proteins in the apo form (Figure 6B). All the
mutants, including NADmutand Dim1mut, showed similar
pattern of papain digestion as wild-type CtBP. We next
showed that the Catmutwas indeed defective in catalysis
using the same pyruvate to lactic acid, coupled with
NADH to NAD?assay (Figure 6C). We also carried out
(and Dim2mut) that is consistent with a lack of ability to
dimerize (Figure 6D). We used the proteolysis assay to
show that the NADmutdoes not bind NAD?(Figure 6E).
As mentioned earlier, upon NAD?binding, there is a
protease-resistant 40 kDa fragment of CtBP that is pro-
duced (Figure 6E, compare left panel, lanes 1 and 2);
of NAD?(Figure 6E, compare right panel, lanes 1 and
2). In all, these biophysical and biochemical control ex-
perimentsestablish thestructural integrityof themutant
proteins in assessing E1A binding in vitro and in vivo.
Discussion Biophysical and Biochemical Characterization
of the Mutant Proteins
The mutations above were designed on the basis of the
crystal structure to lie outside of the hydrophobic core,
so as to preserve structure. The dimerization mutations
were similarly designed to disrupt the dimer interface
troscopy, which monitors secondary structure. The far
UV CD spectra of the dehydrogenase domains of wild-
type CtBP, Catmut, and Dim2mutare similar, with the min-
ima at 208 and 222 nm, characteristic of their ?-helical
content (Figure 6A). The depth of the 222 nm minimum
is in approximate agreement with the helicity calculated
from the crystal structure (?41% helicity). We tested
dimerization capability by subjecting the dehydroge-
nase domains of Catmutand Dim2mutto analytical ultra-
centrifugation (AU). Equilibrium sedimentation data for
Catmutprovided a molecular mass (?74 kDa) that is in
drogenase domain dimer with an N-terminal his-tag
(?76 kDa). Dim2mut, however, could not be reliably ana-
lyzed by equilibrium sedimentation because the protein
had a small tendency to precipitate over the long time
period (13 hr) of data collection. Instead, we compared
Dim2mutand Catmutby sedimentation velocity measure-
ments with scans taken every 1 min over a period of 3
by the time derivative method (Philo, 2000), reveals a
predominant peak with a sedimentation coefficient of
2.6S. This matches almost exactly the expected sedi-
The switch between transcriptional repression and tran-
scriptional activation has been the subject of intensive
investigation over the past 7 years, and one of the most
of enzymatic activities that underlie these events
(Berger, 2001; Cheung et al., 2000; Kadonaga, 1998).
sor, as genetically dissected in Drosophila (Nibu et al.,
1998b), and the observation that it exhibits sequence
homology to known D2-HDHs (Schaeper et al., 1995),
has prompted the speculation that CtBP might contrib-
ute another important enzymatic activity to corepressor
evidence that CtBP is indeed a functional dehydroge-
site cleft for NAD?and substrate binding. The dehydro-
genase domain alone is sufficient to mediate repression
and can bind the PXDLS recognition sequence motif
of E1A in an NAD?-dependent “closed” conformation.
While a true substrate for CtBP remains to be identified,
steric features of the substrate binding pocket suggest
a D2-hydroxyacid with a small “R” group and the lack
of a phosphate group.
CtBP has also been identified as brefeldin A ribosyla-
tion substrate (BARS50), whose LPA acyltransferase
function is essential for Golgi maintenance (Spanfo ` et
al., 1999; Weigert et al., 1999). The CtBP structure is
inconsistent with a proposed acyltransferase reaction
Figure 6. Biophysicaland BiochemicalChar-
acterization of CtBP Mutants
(A) Far-UV CD spectra of wild-type CtBP
(solid), Catmut(dashed), and Dim2mut(dotted).
The mutant proteins have roughly equal sec-
ondary structure to wild-type CtBP.
(B) Partial protease digestion of the apo form
of mutant CtBPs. TnT mutant proteins di-
gested with papain have the same pattern as
rial wild-type and Catmutwere prepared and
used to convert pyruvate to lactic acid. The
reaction was monitored by change in ab-
sorbance at 340 nM.
(D) Dim1mutand Dim2mutdo not interact with
with wild-type TnT CtBP but not the Dim mu-
tants (compare lane 1 with 2 and 3).
(E) NADmutcannot bind NAD?. Wild-type (WT)
CtBP upon NAD?binding produces a prote-
ase-resistant 40 kDa fragment (left panel,
compare lane 1 and 2, arrow indicates the
40 kDa fragment). NADmutprotease digestion
pattern is the same in the presence and ab-
sence of NAD?, indicating that it cannot bind
acid. It is not easy to see from our structure how both
lysophosphatidic acid and acyl-CoA can be accommo-
dated within the CtBP dehydrogenase active site. More-
over, the structure shows little relationship to that of
2001), which catalyzes an acyl-transfer reaction similar
to that proposed for CtBP/BARS in the Golgi. However
CtBP could carry out an NAD?-dependent oxidation/
reduction reaction in the Golgi that is more consistent
with our structure. This could be true for both BARS50
as well as RIBEYE, which is a splice variation of CtBP2
consisting of an N terminus extension, found in syn-
apses but whose function is unknown.
We show here that the dehydrogenase domain alone
is sufficient to bind the PXDLS motif. This domain is
highly conserved within CtBP family members from C.
elegans to vertebrates. However the C terminus exten-
sion (C?) is highly variable with no predicted secondary
structure. It is possible that C? is a regulatory region,
likely to mediate CtBP function after recruitment to a
PXDLS motif. In Drosophila, there are three splice vari-
ants of CtBP differing only in the C?, and the shortest
of these splice variants is essentially only composed of
the dehydrogenase domain (Poortinga et al., 1998). We
initially speculated, based on the crystal structure, that
thePXDLS motifmightbind inacavitynear theentrance
to the active site cleft. However, mutations in this cavity
do not disrupt E1A interactions in vitro or the repression
function in vivo. Unexpectedly, mutation of the active
site residues do affect E1A binding, suggesting that
the PXDLS motif interacts with these residues at the
periphery of the active site cleft. The cleft is walled off
by a loop extending from the 2-fold related subunit that
may provide additional interactions with the PXDLS se-
quence. Indeed, CtBP may be a simple dehydrogenase
that has evolved or gained an extra ability to bind a
PXDLS recognition motif.
The p140 is highly recruited to ligand receptors on
cognate DNA sites (Cavailles et al., 1995). An interaction
between CtBP and RIP140 has also been reported (Vo
et al., 2001), which is intriguing because RIP140 is re-
CtBP Is an NAD?-Regulated Dehydrogenase
cruited to nuclear receptors in response to ligand based
on RIP140 LXXLL motifs; it competes with other coacti-
vators (Heery et al., 1997). While at ambient levels of
CtBP, liganded retinoic acid receptor induces recruit-
and Rosenfeld, 2000); we show here that increased ex-
pression of CtBP completely blocks RAR activation, an
effect entirely dependent on RIP140. Again, this effect
requires specific CtBP catalytic residues, consistent
with the expected role for these residues in stabilizing
binding to interacting cofactors. This also implies that,
in the presence of ligand, regulation of CtBP can be
a key component to the nature of the transcriptional
alterations in NAD?levels might modulate the binding
of CtBP to specific repressor complexes, as well as
regulating its own enzymatic activity. Alterations in
NAD?level has been documented in response DNA
damage, and the reported associations between CtBP
and p130/Rb complex (Dahiya et al., 2001; Dick et al.,
2000; Fusco et al., 1998; Meloni et al., 1999), BRCA1 (Li
et al., 1999, 2000; Wong et al., 1998; Yu and Baer, 2000;
Yu et al., 1998), and KU70 (Schaeper et al., 1998) may be
critical regulatory components of cellular homeostasis.
The ratio of NAD?/NADH can vary in response to activa-
ods of food intake and starvation, and rhythmic cycles
in the cellular redox state have been shown to regulate
DNA binding of Clock and NPAS2 heterodimeric tran-
scription factors (Rutter et al., 2001).
Using CtBP prepared and expressed in a mammalian
system, we investigated whether E1A-CtBP interaction
is regulated differently by NADH versus NAD?but found
no evidence for it. Intriguingly, these results differ from
those reported recently by Zhang et al. (2002) for a bac-
terially expressed CtBP, where NADH is two to three
orders of magnitude more effective than NAD?in stimu-
lating CtBP-E1A interaction (Zhang et al., 2002). This
discrepancy may reflect different sources of CtBP; as
we used in vitro transcribed and translated (TnT) CtBP,
while Zhang et al. (2002) used bacterially expressed
CtBP. The rabbit reticulocyte lysate used for the TnT
reactionis knowntoposttranslationally modifyproteins,
including phosphorylation, acetylation, and isoprenyla-
tion. It is possible that one or more of these modifica-
rial CtBP. In all, it is not easy to see from the structure
how NADH could be up to three orders of magnitude
action, considering that the two cofactors differ chemi-
cally by only a hydrogen atom on the nicotinamide ring.
This may be further clarified by comparing our structure
to a complex of CtBP with NADH.
CtBP is not the only transcription corepressor to be
now shown to bind NAD?. The Sir2 family of transcrip-
tional corepressors also binds NAD?as a cofactor for
histone deacetylation reactions (Finnin et al., 2001; Min
et al., 2001; Moazed, 2001; Shore, 2000), and further-
more, there is direct evidence that activity of NAD?-
dependent Sir2 repressors can regulate life span in C.
elegans (Tissenbaum and Guarente, 2001). Whether lev-
tion, or transcriptional silencing remains to be deter-
mined, but it marks an intriguing new direction of future
Protein Interaction Studies
GST fusion proteins (pGEX AHK-E1AC? aa222-end; Amersham-
Pharmacia) were purified and GST pull-down assays were per-
formed according to previously described techniques (Horlein et al.,
1995). GST pull-down assays were performed in PPI250 (20 mM
HEPES, pH 7.9, 250 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.05% NP40,
buffer 1 hr, bound to TnT proteins for 1 hr at 4?C, and followed by
washes (5 ? 10 min each) with PPI250. For experiments using NAD?
(Sigma N6522), and NADH (Sigma N8129), ADP-Ribose (Sigma
A0752), ?NMN (Sigma N3501), and nicotinamide (Sigma N5535), all
compounds were added to PPI250 buffer at the blocking step and
tition, 10 ?g of PXDLS peptide (EQTVPLDLSCKRPR), from E1A, was
added during the binding step.
Assays were conducted in 0.2 M TrisCl, pH 7.3, with 20 mM NaPyru-
vate and 0.132 mM NADH at 25?C with the appropriate amount of
baculovirus CtBP (Adams et al., 1973). The absorbance was mea-
sured at 340 nM in a UVKON XL spectrophotometer. Baculovirus
CtBP was prepared using the BactoBac system from GIBCO-BRL.
Protease Digestion Protocol
35S-labeled TnT CtBP was digested with limited amount of Papain
(2 ?g/ml and 0.6 ?g/ml), for partial digestion, and higher levels (20
?g/ml) for complete digestion, in reaction buffer (100 mM Tris, pH
needed, the reaction buffer was complimented with the appropriate
amount of NAD(H). The digested material was electrophoresed on
a 15% SDSPAGE gel, dried, and exposed to film.
Transfections and Single-Cell Microinjection
All transfections were performed using HEK293 cells in 12-well
plates. 0.3 ?g of UAStk-Luciferase, 0.3 ?g of Gal/E1A, and 50 ng
of CMV/CtBP wereused per well. Cells weretransfected with CaPO4
and harvested 24 hr later. Transfections were normalized using
?-actin lacZ plasmid. Microinjections into Rat-1 cells were per-
formed as previously reported (Torchia et al., 1997).
The minimal CtBP dehydrogenase domain (aa 28–353) was sub-
cloned into the T7 expression vector pET15b (Novagen), and then
following expression in E. coli BL-21(DE3) pLysS cells, the protein
methionine (Semet)-substituted CtBP was purified similarily from E.
coli B834, a methionine auxotrophic strain. Hexagonal crystals of
the CtBP/NAD?complex were obtained from solutions containing
140 mM sodium formate, 70 mM magnesium acetate, and 100 mM
Hepes (pH 7.0). The crystals belong to space group P6422 with unit
cell dimension of a ? b ? 89.1 A˚, c ? 164 A˚, ? ? ? ? 90?, ? ?120?.
ric unit with 48% solvent in the crystal.
Data Collection, Structure Determination, and Refinement
ratory (BNL, beamline X25), extending to 1.95 A˚resolution (Table
1). MAD data were collected at the Advanced Photon Source (APS,
beamline ID32) at 3 wavelengths, corresponding to the edge, peak,
and a high-energy remote point of the selenium K-edge absorption
profile (Table 1). CNS (Brunger et al., 1998) was used to generate
data. The phases were extended to the 1.95 A˚resolution limit of
the native data using solvent flattening, which yielded a readily
interpretable electron density map. The initial model built had an R
factor of47.1% (Rfreeof 47.0%).After around of simulatedannealing,
dropped to 34.8% (Rfree37.4%). NAD?was built into well-defined
density in a Fo-Fc map. Iterative rounds of model building and
refinement lowered the Rfreeto 30.0%, at which point waters were
added. The final model contains a methionine from pET15b, CtBP
residues from Pro28 to Asp352, NAD?, and 414 waters.
structure of a ternary complex of D-2-hydroxyisocaproate dehydro-
genase from Lactobacillus casei, NAD? and 2-oxoisocaproate at
1.9 A resolution. J. Mol. Biol. 267, 640–660.
Dick, F.A., Sailhamer, E., and Dyson, N.J. (2000). Mutagenesis of
the pRBpocket reveals thatcell cycle arrest functionsare separable
from binding to viral oncoproteins. Mol. Cell. Biol. 20, 3715–3727.
Eklund, H., and Branden, C.I. (1987). Crystal structure, coenzyme
conformations and protein interactions. In Pyridine Nucleotide Co-
enzymes, D.N.Y. Dolphin, ed. (New York: Wiley), pp. 51–98.
Finnin, M.S., Donigian, J.R., and Pavletich, N.P. (2001). Structure of
the histone deacetylase SIRT2. Nat. Struct. Biol. 8, 621–625.
Fusco, C., Reymond, A., and Zervos, A.S. (1998). Molecular cloning
and characterization of a novel retinoblastoma-binding protein. Ge-
nomics 51, 351–358.
Garcia de la Torre, J., Huertos, M.L., and Carrasco, B. (2000). Calcu-
lation of hydrodynamic properties of globular proteins from their
atomic-level structure. Biophys. J. 78, 719–730.
Glass, C.K., and Rosenfeld, M.G. (2000). The coregulator exchange
in transcriptional functions of nuclear receptors. Genes Dev. 14,
Goldberg, J.D., Yoshida, T., and Brick, P. (1994). Crystal structure
J. Mol. Biol. 236, 1123–1140.
Grau, U.M., Trommer, W.E., and Rossmann, M.G. (1981). Structure
of the active ternary complex of pig heart lactate dehydrogenase
with S-lac-NAD at 2.7 A˚resolution. J. Mol. Biol. 151, 289–307.
Gray, S., and Levine, M. (1996). Transcriptional repression in devel-
opment. Curr. Opin. Cell Biol. 8, 358–364.
Heery, D.M., Kalkhoven, E., Hoare, S., and Parker, M.G. (1997). A
signature motif in transcriptional co-activators mediates binding to
nuclear receptors. Nature 387, 733–736.
Hendrickson, W.A. (1991). Determination of macromolecular struc-
tures from anomalous diffraction of synchrotron radiation. Science
Horlein, A.J., Naar, A.M., Heinzel, T., Torchia, J., Gloss, B., Kuro-
kawa, R., Ryan, A., Kamei, Y., Soderstrom, M., Glass, C.K., et al.
(1995). Ligand-independent repression by the thyroid hormone re-
ceptor mediated by a nuclear receptor co-repressor. Nature 377,
Jenuwein, T., and Allis, C.D. (2001). Translating the histone code.
Science 293, 1074–1080.
Kadonaga, J.T. (1998). Eukaryotic transcription: an interlaced net-
work of transcription factors and chromatin-modifying machines.
Cell 92, 307–313.
drogenase. J. Mol. Biol. 236, 759–785.
Li, S., Chen, P.L., Subramanian, T., Chinnadurai, G., Tomlinson, G.,
Osborne, C.K., Sharp, Z.D., and Lee, W.H. (1999). Binding of CtIP
to the BRCT repeats of BRCA1 involved in the transcription regula-
tion of p21 is disrupted upon DNA damage. J. Biol. Chem. 274,
Li, S., Ting, N.S., Zheng, L., Chen, P.L., Ziv, Y., Shiloh, Y., Lee,
E.Y., and Lee, W.H. (2000). Functional link of BRCA1 and ataxia
telangiectasia gene product in DNA damage response. Nature 406,
Meloni, A.R., Smith, E.J., and Nevins, J.R. (1999). A mechanism for
Rb/p130-mediated transcription repression involving recruitment of
the CtBP corepressor. Proc. Natl. Acad. Sci. USA 96, 9574–9579.
Min, J., Landry, J., Sternglanz, R., and Xu, R.M. (2001). Crystal struc-
ture of a SIR2 homolog-NAD complex. Cell 105, 269–279.
ing. Curr. Opin. Cell Biol. 13, 232–238.
Nibu, Y., Zhang, H., Bajor, E., Barolo, S., Small, S., and Levine,
M. (1998a). dCtBP mediates transcriptional repression by Knirps,
Nibu, Y., Zhang, H., and Levine, M. (1998b). Interaction of short-
Far UV CD spectra were collected on the bacterially expressed
dehydrogenase domains of wild-type CtBP, Catmut, and Dim2muton
a Jasco-810 spectropolarimeter. The concentrations of wild-type
CtBP, Catmut, and Dim2mutwere 5.7, 9.3, and 4.9 ?M, respectively.
Relative ellipticity was converted to molar ellipticity.
Sedimentation equilibrium data on Catmutwere measured on a Beck-
data were fitted using the WinNONLR package (http://Spin6.mcb.
uconn.edu) and the molecular mass calculated by SEDNTRP (http://
www.bbri.org/rasmb/rasmb.html). Sedimentation velocity data on
Dim2mutand Catmutwererecorded (AT-60rotor,42,000rpm) bytaking
scans every 1 min for 3 hr, in both the absorbance and interference
modes. The data were interpreted using the time derivative of the
concentration profile g(s*) with the program dcdt?(Philo, 2000).
The hydrodynamic modeling was done with programs HYDROPRO
(Garcia de la Torre et al., 2000) using the CtBP crystal coordinates.
We are grateful to K. D’Amico for synchrotron time at the APS. We
thank L. Shapiro and T. Boggon for help with data collection. We
thank L. van Grunsven, C. Svensson, and A. Sundqvist for hCtBP1
and CtBP2 expression plasmids, respectively. We thank C. Nelson
for tissue culture assistance, P. Meyer for illustrations, and M.
Fischer for help with the manuscript. We also thank O. Hermanson
and other members of the Rosenfeld Lab for support and encour-
agement. This research was supported by NIH grants to M.G.R and
A.K.A. M.G.R is an investigator of the HHMI.
Received: December 5, 2001
Revised: August 13, 2002
Adams, M.J., Buehner, M., Chandrasekhar, K., Ford, G.C., Hackert,
M.L., Liljas, A., Rossmann, M.G., Smiley, I.E., Allison, W.S., Everse,
J., et al. (1973). Structure-function relationships in lactate dehydro-
genase. Proc. Natl. Acad. Sci. USA 70, 1968–1972.
Berger, S.L. (2001). An embarrassment of niches: the many covalent
modifications of histones in transcriptional regulation. Oncogene
Boyd, J.M., Subramanian, T., Schaeper, U., La Regina, M., Bayley,
rus 2/5 E1a protein is required for association with a cellular phos-
phoprotein and important for the negative modulation of T24-ras
mediated transformation, tumorigenesis and metastasis. EMBO J.
Brunger, A.T., Adams, P.D., Clore, G.M., Delano, W.L., Gros, P.,
Grosse-Kunstleve, R., Jiang, W., Kuszewski, J., Nilges, M., Pannu,
N.S., et al. (1998). Crystallography & NMR system: a software suite
for macromolecular structure determination. Acta Crystallogr. D 54,
P.J., and Parker,M.G. (1995). Nuclear factorRIP140 modulates tran-
scriptional activation by the estrogen receptor. EMBO J. 14, 3741–
Cheung, P., Allis, C.D., and Sassone-Corsi, P. (2000). Signaling to
chromatin through histone modifications. Cell 103, 263–271.
Chinnadurai, G. (2002). CtBP, an unconventional transcriptional co-
repressor in development and oncogenesis. Mol. Cell 9, 213–224.
Dahiya, A., Wong, S., Gonzalo, S., Gavin, M., and Dean, D.C. (2001).
Linking the Rb and polycomb pathways. Mol. Cell 8, 557–569.
CtBP Is an NAD?-Regulated Dehydrogenase Download full-text
range repressors with Drosophila CtBP in the embryo. Science 280,
Pazin, M.J., and Kadonaga, J.T. (1997). What’s up and down with
histone deacetylation and transcription? Cell 89, 325–328.
Philo, J.S. (2000). A method for directly fitting the time derivative of
ing sedimentation coefficient distribution functions. Anal. Biochem.
Poortinga, G., Watanabe, M., and Parkhurst, S.M. (1998). Drosophila
CtBP: a Hairy-interacting protein required for embryonic segmenta-
tion and hairy-mediated transcriptional repression. EMBO J. 17,
Rao, S.T., and Rossmann, M.G. (1973). Comparison of super-sec-
ondary structures in proteins. J. Mol. Biol. 76, 241–256.
Rutter, J., Reick, M., Wu,L.C., and McKnight, S.L. (2001). Regulation
of clock and NPAS2 DNA binding by the redox state of NAD cofac-
tors. Science 293, 510–514.
Schaeper, U., Boyd, J.M., Verma, S., Uhlmann, E., Subramanian, T.,
and Chinnadurai, G. (1995). Molecular cloning and characterization
of a cellular phosphoprotein that interacts with a conserved C-ter-
minal domain of adenovirus E1A involved in negative modulation of
oncogenic transformation. Proc. Natl. Acad. Sci. USA 92, 10467–
G. (1998). Interaction between a cellular protein that binds to the
C-terminal region of adenovirus E1A (CtBP) and a novel cellular
protein is disrupted by E1A through a conserved PLDLS motif. J.
Biol. Chem. 273, 8549–8552.
Schmitz, F., Ko ¨nigstorfer, A., and Su ¨dhof, T.C. (2000). RIBEYE, a
component of synaptic ribbons: a protein’s journey through evolu-
tion provides insight into synaptic ribbon function. Neuron 28,
Schuller, D.J., Grant, G.A., and Banaszak, L.J. (1995). The allosteric
ligand site in the Vmax-type cooperative enzyme phosphoglycerate
dehydrogenase. Nat. Struct. Biol. 2, 69–76.
Shore, D. (2000). The Sir2 protein family: a novel deacetylase for
gene silencing and more. Proc. Natl. Acad. Sci. USA 97, 14030–
Spanfo `, S., Silletta, M.G., Colanzi, A., Alberti, S., Fiucci, G., Valente,
C., Fusella, A., Salmona, M., Mironov, A., Luini, A., et al. (1999).
Molecular cloning and functional characterization of brefeldin
nance of the Golgi structure. J. Biol. Chem. 274, 17705–17710.
Stoll, V.S., Kimber, M.S., and Pai, E.F. (1996). Insights into substrate
binding by D-2-ketoacid dehydrogenases from the structure of Lac-
Tissenbaum, H.A., and Guarente, L. (2001). Increased dosage of a
sir-2 gene extends lifespan in Caenorhabditis elegans. Nature 410,
Torchia, J., Rose, D.W., Inostroza, J., Kamei, Y., Westin, S., Glass,
C.K., and Rosenfeld, M.G. (1997). The transcriptional co-activator
p/CIP binds CBP and mediates nuclear-receptor function. Nature
Turnbull, A.P., Rafferty, J.B., Sedelnikova, S.E., Slabas, A.R., Simon,
J.W., Fawcett, T., Nishida, I., Murata, N., and Rice, D.W. (2001).
Analysis of the structure, substrate specificity, and mechanism
of squash glycerol-3-phosphate (1)-acyltransferase. Structure 9,
Turner, J., and Crossley, M. (2001). The CtBP family: enigmatic and
enzymatic transcriptional co-repressors. Bioessays 23, 683–690.
Tyler, J.K., and Kadonaga, J.T. (1999). The “dark side” of chromatin
remodeling: repressive effects on transcription. Cell 99, 443–446.
Vo, N., Fjeld, C., and Goodman, R.H. (2001). Acetylation of nuclear
hormone receptor-interacting protein RIP140 regulates binding of
the transcriptional corepressor CtBP. Mol. Cell. Biol. 21, 6181–6188.
Weigert, R., Silletta, M.G., Spano `, S., Turacchio, G., Cericola, C.,
Colanzi, A., Senatore, S., Mancini, R., Polishchuk, E.V., Salmona,
M., et al. (1999). CtBP/BARS induces fission of Golgi membranes
by acylating lysophosphatidic acid. Nature 402, 429–433.
Wong, A.K., Ormonde, P.A., Pero, R., Chen, Y., Lian, L., Salada,
G., Berry, S., Lawrence, Q., Dayananth, P., Ha, P., et al. (1998).
Characterization of a carboxy-terminal BRCA1 interacting protein.
Oncogene 17, 2279–2285.
Yu, X., and Baer, R. (2000). Nuclear localization and cell cycle-
tumor suppressor. J. Biol. Chem. 275, 18541–18549.
Yu, X., Wu, L.C., Bowcock, A.M., Aronheim, A., and Baer, R. (1998).
The C-terminal (BRCT) domains of BRCA1 interact in vivo with CtIP,
a protein implicated in the CtBP pathway of transcriptional repres-
sion. J. Biol. Chem. 273, 25388–25392.
Zhang, Q., Yao, H., Vo, N., and Goodman, R.H. (2000). Acetylation of
adenovirus E1A regulates binding of the transcriptional corepressor
CtBP. Proc. Natl. Acad. Sci. USA 97, 14323–14328.
Zhang, Q., Piston, D.W., and Goodman, R.H. (2002). Regulation of
corepressor function by nuclear NADH. Science 295, 1895–1897.