ABSTRACT
The yeast two-hybrid system is a powerful experimental approach for the characterization
of protein/protein interactions. A unique strength of the yeast two-hybrid system is the provision for genetic selection techniques that
enable the identification of specific protein/protein interactions. We now
report the development of a modified yeast two-hybrid system which enables genetic selection against a specific
protein/protein interaction. This reverse two-hybrid system utilizes a yeast strain which is resistant to cycloheximide
due to the presence of a mutant
cyh2
gene. This strain also contains the wild-type
CYH2
allele under the transcriptional control of the Gal1 promoter. Expression of
the wild-type Gal4 protein is sufficient to restore growth sensitivity to
cycloheximide. Growth sensitivity towards cycloheximide is also restored by the
coexpression of the avian c-Rel protein and its I
[kappa]
B
[alpha]
counterpart, p40, as Gal4 fusion proteins. Restoration of growth sensitivity
towards cycloheximide requires the association of c-Rel and p40 at the Gal1 promoter and correlates with the ability of the c-Rel/p40 interaction to activate expression from the Gal1 promoter. A
genetic selection scheme against specific protein/protein interactions may be a
valuable tool for the analysis of protein/protein interactions.
The c-Rel protein is sequestered in the cytoplasm through association with the
inhibitor protein, I[kappa]B[alpha] (
1
,
2
). Cytokine-induced phosphorylation of I[kappa]B[alpha] signals ubiquitin-mediated degradation of I[kappa]B[alpha] and subsequent nuclear import of c-Rel (
3
-
10
). The nuclear c-Rel protein binds DNA and activates expression of multiple target genes,
including the gene encoding I[kappa]B[alpha] (
11
-
15
). Newly synthesized I[kappa]B[alpha] is able to enter the nucleus (
16
-
18
), and nuclear I[kappa]B[alpha] has been suggested to facilitate the displacement of c-Rel from DNA and mediate export of the nuclear c-Rel/I[kappa]B[alpha] complex to the cytoplasm (
16
,
18
,
19
). The ability of I[kappa]B[alpha] to associate with c-Rel in both the nucleus and the cytoplasm suggests that the
nuclear and cytoplasmic c-Rel/I[kappa]B[alpha] complexes may have distinct functional and structural
properties.
The existence of two c-Rel/I[kappa]B[alpha] complexes with distinct functional and structural
properties is supported by several lines of evidence. First, the ankyrin repeat
domain of I[kappa]B[alpha] is sufficient to inhibit nuclear import of c-Rel, while both the ankyrin repeat domain and the PEST
domain of I[kappa]B[alpha] are required for inhibition of DNA-binding by c-Rel (
1
,
2
). Second, while the ankyrin repeat domain of I[kappa]B[alpha] is sufficient for association with the wild-type c-Rel protein, both the ankyrin repeat domain and the
PEST domain of I[kappa]B[alpha] are required for association with mutant c-Rel proteins that are resistant to inhibition of DNA-binding by I[kappa]B[alpha] (
19
,
20
). Therefore, it can be envisioned that the ankyrin repeat domain is sufficient
for the integrity of the cytoplasmic c-Rel/I[kappa]B[alpha] complex, while both the ankyrin repeat domain and the PEST
domain are required for the integrity of the nuclear c-Rel/I[kappa]B[alpha] complex. However, little is known regarding the specific
amino acid contacts that define each c-Rel/I[kappa]B[alpha] complex.
We have previously utilized the yeast two-hybrid system to identify residues that are required for the formation of
c-Rel/I[kappa]B[alpha] complexes containing either wild-type or mutant proteins (
19
). The yeast two-hybrid system is a potent tool for studying protein/protein interactions
in vivo
(
21
-
24
). In a commonly utilized version of the
Saccharomyces cerevisiae
two-hybrid system (
21
), one of the proteins (A) is bound to the Gal4 enhancer region of the Gal1
promoter via its fusion to the Gal4 binding domain (Gal4BD), while the other
protein (B) is fused to the Gal4 activation domain (Gal4AD, Fig.
1
A). Association of the two fusion proteins (Gal4BD:A and Gal4AD:B) brings the
activation domain of Gal4 to the Gal1 promoter. Localization of the Gal4
activation domain to the Gal1 promoter can be used to activate the expression
of either a reporter gene such as
lacZ
, or a gene necessary for cell viability such as
HIS3
(
21
,
23
).
The use of the
lacZ
reporter gene in the two-hybrid system has allowed for the development of assays for the
identification of interacting proteins and for the identification of mutant
proteins that are unable to interact with their partner protein (
21
-
24
). However, biochemical assays for
lacZ
expression are relatively insensitive and can yield a high frequency of false
positive clones. To circumvent the high frequency of false positives when using
the
lacZ
gene as a single reporter system, two-hybrid systems in which the use of genes that are required for cell
viability such as
HIS3
were developed (
22
). Such genetic selection schemes have proven to be very effective in the
identification of interacting proteins. However, these genetic selection
schemes are not able to identify mutant proteins that have lost the ability to
associate with their partner protein. Therefore, we asked if the standard two-hybrid system system could be reversed. That is, can a genetic system be
developed in which there is a selection against the association of two
proteins? A genetic system which is capable of selecting against the
association of two proteins might provide a simple yet powerful approach towards screening pools of randomly generated mutant proteins. Furthermore, such a genetic selection scheme, in
conjunction with suitable yeast expression libraries, might facilitate the
identification of genes which encode proteins that interfere with a particular
protein/protein interaction. This has led to the development of the reverse two-hybrid system.
To construct pCL12, the Gal1 promoter was subcloned as a
Sal
I-
Sac
I
fragment from pRY171 (
27
) into pRS304 (
28
) which contains the
TRP1
selectable marker. The PCR primers 5'-GCGCGCGCTTGTATGCCTTCCAGATTCACTAAGAC-3' and 5'-GCGCGC
To construct the plasmid encoding Gal4AD:p40-TA, the PCR primers 5'-TCC
All other plasmids have been previously described (
19
). All proteins were expressed from the yeast ADH promoter on episomal plasmids.
The
HIS3
marker was used to select for the presence of the plasmids expressing the
Gal4BD and all the wild-type and mutant Gal4BD:c-Rel proteins and the
LEU2
marker was used to select for the presence of the plasmids expressing the full-length Gal4, Gal4AD, and all wild-type and mutant Gal4AD:p40 proteins.
The
S.cerevisiae
yeast strain JC981 ([alpha]
gal4 gal80 his3 trp1-901 ura3-52 URA3::Gal1
->
lacZ leu2-3,112 cyh2
) was the parental strain from which CL9 was derived. Standard techniques were
used for the selection of
cyh2
yeast strains and transformation procedures (
25
,
26
). To select for cycloheximide (CHX)-resistant yeast strains, ~5 * 10
8
JC981 cells were plated on YPD (yeast extract, peptone, dextrose) plates
containing 10 [mu]g/ml CHX and incubated at 30oC. CHX-resistant colonies were obtained after 5 days. To ensure that the
yeast strains were CHX resistant due to a mutation at the
CYH2
locus (
cyh2
), the CHX-resistant yeast strains were transformed with pRS318 (
25
) and were plated onto leucine-free synthetic dropout (SD) plates with and without 10 [mu]g/ml CHX. pRS318 contains the wild type
CYH2
gene driven by its endogenous promoter and restores CHX sensitivity when
expressed in mutant
cyh2
yeast (
25
).
The pCL25 plasmid, linearized at the unique
Hin
dIII site within
TRP1
to facilitate integration of the plasmid into the
trp1-901
locus, was transformed into the
cyh2
yeast strains. The transformants were plated onto TRP-free SD plates. To ensure the stable integration of pCL25, the strains
were transferred to YPD plates and replica plated onto SD plates lacking
tryptophan (TRP). The surviving colonies were replica plated back onto YPD
plates and then to TRP-free SD plates. The surviving colonies were also replica plated onto SD
plates lacking histidine (HIS), lacking leucine (LEU), lacking uracil (URA),
and containing 10 mg/ml CHX to ensure that the correct phenotype had been
maintained throughout the manipulation of these strains.
Association between c-Rel and p40 proteins was measured using the standard two-hybrid system essentially as previously described (
19
) with the following exceptions. The
cyh2
yeast strains were transformed with the indicated plasmids and 3% of the transformation mixture was plated onto HIS-free, LEU-free and TRP-free SD plates that lacked or contained 25 [mu]g/ml CHX. For analysis of [beta]-galactosidase activity, the transformed yeast
were picked from plates lacking CHX into 50 [mu]l of 100 mM potassium phosphate containing 0.2% Triton X-100, to which was added 0.5 [mu]l of 1 M dithiothreitol (DTT), 2.5 [mu]l of 0.1% sodium dodecyl sulfate (SDS) and 7.5 [mu]l of chloroform. The yeast cells were lysed by vigorous
vortexing for 10 s. The chemiluminescent assay used to measure the relative
light units produced due to the association of Rel and p40 proteins has been
previously described (
19
). Pictures of the plates were taken 9 days following transformation.
The protein encoded by the
CYH2
gene is responsible for the sensitivity of the yeast to cycloheximide (CHX).
Mutation of
CYH2
can produce a CHX-resistant
cyh2
yeast strain. In the reverse two-hybrid system described in this report, a
cyh2
yeast strain containing
CYH2
under the control of the Gal1 promoter is utilized. Because the
CYH2
allele is dominant over the
cyh2
allele, expression of the
CYH2
gene product will confer CHX sensitivity to a CHX-resistant
cyh2
yeast strain (
25
,
30
). Thus, localization of Gal4AD to the Gal1 promoter through association of the
fusion proteins will activate transcription of
CYH2
and confer sensitivity to CHX. In Figure
1
B, the fusion proteins Gal4BD:c-Rel and Gal4AD:p40 are used as an example. Since the only way for the
yeast cell to survive when grown in the presence of CHX is through the absence
of
CYH2
expression, the reverse two-hybrid system could provide a selection scheme to identify mutant proteins
that are unable to associate.
Any combination of two-hybrid fusion proteins expressed in CL9 resulted in equivalent numbers and
sizes of colonies in the absence of CHX (Figs
2
,
3
,
4
and
5
). To determine if the presence of CHX affected colony growth, CL9 was
cotransformed with a plasmid expressing Gal4BD and a plasmid expressing Gal4AD.
Equivalent aliquots of the transformed yeast were plated in the presence and in
the absence of CHX. The presence of CHX did not affect the number of
transformants that were obtained, but the overall sizes of the colonies were
smaller (Figs
2
,
3
,
4
and
5
).
Three domains within p40 have been defined: (i) an N-terminal domain that is responsible for cytokine-dependent phosphorylation and degradation of p40 (
3
-
10
); (ii) a central ankyrin repeat domain that is required for association with
and cytoplasmic retention of c-Rel (
1
,
2
,
19
); and (iii) a C-terminal PEST domain that is specifically required for inhibition of DNA-binding (
1
,
2
). To determine which domains of Gal4AD:p40 were necessary to confer complete
sensitivity to CHX when coexpressed with Gal4BD:c-Rel, the CL9 yeast strain was cotransformed with a plasmid that expressed full-length Gal4BD:c-Rel and with plasmids that expressed various Gal4AD:p40 mutants. Expression of
Gal4BD:c-Rel with either a Gal4AD:p40 fusion protein containing a 70 amino acid
(aa) N-terminal truncation (Gal4AD:p40-[Delta]N70, Fig.
6
) or with a Gal4AD:p40 fusion protein containing an 18 aa C-terminal truncation (Gal4AD:p40-300, Fig.
6
) resulted in complete sensitivity to CHX (data not shown).
To determine which domains of Gal4BD:c-Rel were necessary to confer complete sensitivity to CHX when coexpressed
with Gal4AD:p40, truncated Gal4BD:c-Rel fusion proteins were expressed with Gal4AD:p40 in strain CL9.
Gal4AD:p40 was expressed with Gal4BD:c-Rel-540 and with c-Rel truncated at residue 355 (Gal4BD:c-Rel-355; Fig.
6
) in the CL9 yeast strain. Figure
3
shows that expression of Gal4BD:c-Rel-540 with Gal4AD:p40 was not inhibitory towards growth of CL9 on
plates containing CHX and that further truncation of the Gal4BD:c-Rel protein to residue 355 slightly increased the size of the viable
colonies. Thus, the C-terminal transactivation domain of c-Rel is required for complete growth inhibition in the reverse two-hybrid system.
Truncated Gal4BD:c-Rel proteins expressed with Gal4AD:p40 were assayed for [beta]-galactosidase activity as shown in Figure
3
. Truncation of Gal4BD:c-Rel to amino acid 540 resulted in a decrease in [beta]-galactosidase activity to ~3.9% of the [beta]-galactosidase activity detected when full-length Gal4BD:c-Rel was expressed with
Gal4AD:p40. Truncation of Gal4BD:c-Rel to amino acid 355 resulted in a reduction of [beta]-galactosidase activity to ~0.35%. Therefore, the level of [beta]-galactosidase activity correlated with the
extent of sensitivity to CHX when truncated Gal4BD:c-Rel proteins were expressed with Gal4AD:p40.
Although expression of full-length Gal4BD:c-Rel with Gal4AD:p40 resulted in complete growth inhibition in the
presence of CHX, expression of full-length Gal4BD:c-Rel with Gal4AD resulted in only partial inhibition of growth in the
presence of CHX. As expression of Gal4BD:c-Rel-540 with Gal4AD did not confer significant sensitivity to CHX
(presumably due to removal of the potent C-terminal transactivation domain of c-Rel), we desired to manipulate the reverse two-hybrid system such that Gal4BD:c-Rel-540 could be used as the `bait' protein. However,
expression of Gal4BD:c-Rel-540 with Gal4AD:p40 was not sufficient to confer complete CHX
sensitivity. To enhance the manifestation of the CHX-sensitive phenotype when Gal4BD:c-Rel-540 was expressed with Gal4AD:p40, the C-terminal transactivation domain of c-Rel was fused to Gal4AD:p40. The C-terminal transactivation domain of full-length c-Rel was fused to Gal4AD:p40
either at the C-terminus of p40 (Gal4AD:p40-TA; Fig.
6
) or at the N terminus of p40 (Gal4AD:TA-p40; Fig.
6
) in the expectation that the addition of the c-Rel transactivation domain to the Gal4AD:p40 fusion proteins would
increase transcription of
CYH2
upon association with Gal4BD:c-Rel-540.
Coexpression of Gal4BD with either the Gal4AD:p40-TA or the Gal4AD:TA-p40 fusion protein had no effect on either CHX sensitivity or on [beta]-galactosidase activity (data not shown). Coexpression of
Gal4BD:c-Rel-540 with Gal4AD:p40-TA did not markedly increase growth inhibition in the presence
of CHX as compared with coexpression of Gal4BD:c-Rel-540 with Gal4AD:p40 (Fig.
5
). Expression of Gal4BD:c-Rel-540 with Gal4AD:p40-TA resulted in levels of [beta]-galactosidase activity that were ~3-fold greater than the levels of [beta]-galactosidase activity that
resulted from the expression of Gal4BD:c-Rel-540 with Gal4AD:p40. However, [beta]-galactosidase activity obtained from coexpression of
Gal4BD:c-Rel-540 with Gal4AD:p40-TA was still approximately one-tenth that of [beta]-galactosidase activity obtained from
coexpression of Gal4BD:c-Rel with Gal4AD:p40. Therefore, fusion of the C-terminal transactivation domain of Gal4BD:c-Rel to Gal4AD:p40 at the C-terminus of p40 was not sufficient to enhance the
manifestation of the CHX-sensitive phenotype and Gal1 promoter-dependent gene expression.
Coexpression of Gal4BD:c-Rel-540 with Gal4AD:TA-p40 resulted in nearly complete sensitivity to CHX as compared
with coexpression of Gal4BD:c-Rel-540 with Gal4AD:p40 (Fig.
5
). Furthermore, coexpression of Gal4BD:c-Rel-540 with Gal4AD:TA-p40 resulted in an ~12-fold increase in [beta]-galactosidase activity as compared with
coexpression of Gal4BD:c-Rel-540 with Gal4AD:p40 (Fig.
5
). The level of [beta]-galactosidase activity obtained from coexpression of Gal4BD:c-Rel-540 with Gal4AD:TA-p40 was approximately half that obtained from
coexpression of full-length Gal4BD:c-Rel with Gal4AD:p40 (Fig.
5
). Therefore, the C-terminal transactivation domain of Gal4BD:c-Rel fused to the N-terminus of p40 enhances the manifestation of the CHX-sensitive phenotype in the reverse two-hybrid system, presumably due to increased
transcription of
CYH2
from the Gal1 promoter.
We have developed the reverse two-hybrid system as a genetic scheme to select against protein/protein
interactions. The reverse two-hybrid system takes advantage of the ability of the
S.cerevisiae
wild-type
CYH2
gene product to confer sensitivity to cycloheximide in a dominant manner over
the mutant
cyh2
gene product which confers resistance to cycloheximide (
30
). The Gal1 promoter was utilized to drive expression of the
CYH2
gene, such that transcriptional activation of the Gal1 promoter by the full-length Gal4 protein resulted in restoration of complete cycloheximide
sensitivity. Complete CHX sensitivity in a CHX-resistant
cyh2
yeast strain was also conferred when the wild-type Gal4BD:c-Rel fusion protein was coexpressed with the wild-type Gal4AD:p40 fusion protein. Expression of a mutant p40
protein in the reverse two-hybrid system that was unable to associate with c-Rel did not confer sensitivity to CHX. Therefore, the reverse two-hybrid system could provide an effective selection system
against specific protein/protein interactions in yeast.
Sensitivity to cycloheximide in the reverse two-hybrid system is presumably due to the formation of a protein/protein
complex at the Gal1 promoter. Since the yeast strain that we constructed also
contains the
lacZ
gene driven by the Gal1 promoter, we were able to determine whether CHX
sensitivity correlated with levels of [beta]-galactosidase activity. The degree of CHX sensitivity correlated
with the levels of [beta]-galactosidase activity for each combination of proteins expressed.
Although we did not directly measure
CYH2
mRNA expression directly, it is firmly established that association of Gal4
fusion proteins results in the transcriptional transactivation of a gene driven
by the Gal1 promoter (
23
). Thus, it is likely that the cycloheximide sensitivity in the reverse two-hybrid system is a direct consequence of the association of Gal4 fusion
proteins at the Gal1 promoter with consequent transcriptional activation of the
CYH2
gene.
Several aspects of our results suggest that a threshold level of transcriptional
activation of the
CYH2
gene may be required for complete sensitivity to cycloheximide. First, even
though the c-Rel protein contains a potent C-terminal transcriptional activation domain (
31
-
34
), expression of full-length Gal4BD:c-Rel with Gal4AD was only able to confer partial sensitivity to CHX.
Instead, complete sensitivity to CHX required the coexpression of Gal4AD:p40
fusion proteins that are able to strongly associate with the Gal4BD:c-Rel fusion protein. Most strikingly, while coexpression of Gal4BD:c-Rel-540 with Gal4AD:p40 did not confer complete sensitivity to
cycloheximide, coexpression of Gal4BD:c-Rel-540 with Gal4AD:TA-p40 gave nearly complete sensitivity to CHX. Thus,
localization of both the C-terminal transactivation domain of c-Rel and the Gal4 activation domain to the Gal1 promoter was
necessary for protein/protein interactions to confer complete CHX sensitivity.
Taken together, these results suggest that a threshold level of
CYH2
expression is necessary to confer complete sensitivity to cycloheximide. The
threshold level of
CYH2
expression necessary to confer complete sensitivity to cycloheximide may be
affected by a number of factors, including the strength of a particular
protein/protein interaction, growth temperature, and the concentration of
cycloheximide in the plates on which the transformants are grown.
As the two-hybrid system has been used to study a wide variety of
in vivo
protein/protein interactions, the reverse two-hybrid system could likewise be suitable to study a wide variety of
protein/protein interactions. In the reverse two hybrid system, we demonstrated
that the lack of association between two fusion proteins results in the
complete lack of cell growth in the presence of CHX. Therefore, the reverse two-hybrid system may be suitable for the rapid screening of randomly
generated mutant proteins that are no longer able to associate with their
partner protein. Experiments are currently underway to identify mutant c-Rel proteins which can no longer associate with p40. Additionally, the
reverse two-hybrid system could potentially be used to identify other proteins which
can disrupt a particular protein/protein complex. For example, the reverse two-hybrid system may be particularly suitable for screening expression
libraries to identify regulators of protein/protein interactions. Finally, the
reverse two hybrid system could be used to screen for drugs that abolish a
specific protein/protein interaction. Therefore, the reverse two-hybrid system will likely become a potent tool in the study of
protein/protein interactions.
We wish to thank Phil Simmons for technical assistance, Elizabeth Rottjakob and
John Cannon for technical advice on the yeast two-hybrid system, and members of the Hannink laboratory for stimulating
conversations. This work was supported by PHS grant CA-55027 and by the University of Missouri Molecular Biology Program.
REFERENCES
Return