(Received for publication, March 31, 1997)
From the Fakultät für Biologie, Universität Konstanz, Postfach 5560, 78434 Konstanz, Germany
The eucaryotic protein L7, which associates with
the large subunit of ribosomes, has been shown to be a major
autoantigen in systemic autoimmune arthritis. The N terminus carries a
sequence motif that is similar to the leucine zipper domain of
eucaryotic transcription factors. This domain promotes the
homodimerization of protein L7 through -helical coiled-coil
formation and binds to distinct mRNAs, thereby inhibiting their
cell-free translation. Using a yeast two-hybrid selection, we have
identified from a Jurkat T lymphoma cDNA library ribosomal protein
S7 and the multi-zinc finger protein ZNF7 as proteins that interact
with protein L7. A fragment of L7 carrying the leucine zipper-like
domain is fully sufficient to mediate these interactions. Their
potential biological significance is indicated by low apparent
dissociation constants of S7-L7 (15 × 10
9
M) and, respectively, ZNF7-L7 (2 × 10
9
M) complexes and co-immunoprecipitation of proteins S7,
ZNF7, and L7 from a cell lysate with an anti-L7 antibody. We also show that ZNF7-like L7 and S7 can exist in a ribosome-bound form. This study
provides further evidence suggesting that L7 is involved in
translational regulation through interactions with components of the
translational apparatus.
Eucaryotic protein L7 associates in the cytoplasm with the large ribosomal subunit (1, 2). The N-terminal region of human and rodent protein L7 carries a sequence motif that is similar to the basic region leucine zipper (BZIP)1 domain characteristic of some eucaryotic transcription factors (3, 4). The BZIP-like domain mediates the formation of L7 homodimers (3, 5), which interact with cognate sites on mRNA (3, 4), thereby inhibiting their cell-free translation (6). Constitutive expression of human protein L7 in Jurkat T lymphoma cells suppresses the synthesis of at least two nuclear proteins, arrests the cell cycle in G1, and induces apoptosis (7). Thus, protein L7 is one of the growing number of ribosomal proteins that seem to have extraribosomal functions (8). Like some other riboproteins, protein L7 is a major autoantigen in rheumatoid autoimmune diseases, such as systemic lupus erythematosus and others (9-11). The autoimmune anti-L7 response develops high titers and switches from the oligoclonal high affinity recognition of one major epitope to the polyclonal low affinity recognition of various minor epitopes ("epitope spreading") during the active phase of systemic lupus erythematosus. This suggests that autoimmunogenic L7 is released during such phases and stimulates normally silent B-lymphocyte clones expressing low affinity antigen receptors (12). In this context, we study the functions of protein L7 with the intention to elucidate potentially autoimmunogenic mechanisms (13, 14).
To understand the biological function of protein L7 we employed yeast two-hybrid selection (15) to identify ribosomal protein S7 and the multi-zinc finger protein ZNF7 as proteins that specifically interact with L7. Complexes of L7 with S7 and, respectively, ZNF7 apparently exist in lymphoid cells, as demonstrated by co-immunoprecipitation. These and previous findings (3-7) suggest a function of protein L7 in translational regulation.
Manipulations of Escherichia coli, Saccharomyces cerevisiae, nucleic acids, and proteins were performed as described (16).
Yeast Two-hybrid SystemThe yeast two-hybrid system used in this study was kindly donated by E. A. Golemis (Fox Chase Cancer Center, Philadelphia, PA) and has been described previously (15, 17). We employed yeast strain S. cerevisiae EGY48, which contained for selection the reporter plasmid pJk103 (15)
Construction of Bait Plasmids for the Yeast Two-hybrid ScreenpEG202 (15) was used as an expression vector for
constructing LexA-fused bait proteins. pLexA-L7 expresses LexA fused to the full-length coding sequence of human ribosomal protein L7. pLexA-L71-124 and pLexA-L7125-248 express the
124 N-terminal amino acids or, respectively, the 124 C-terminal amino acids of L7, fused to LexA. PCR was used to prepare DNA fragments that
contained 5 EcoRI and 3
XhoI sites. These
fragments were treated with EcoRI and XhoI and
were ligated into pEG202. Recombinant plasmids were amplified in
E. coli, and the integrity and orientation of their cDNA
inserts were verified by nucleotide sequencing.
500
µg of total RNA were isolated from 1 × 108 Jurkat
cells (RNeasy, QIAGEN, Hilden, FRG). 7 µg of poly(A)+ RNA
were obtained from total RNA (Oligotex, QIAGEN), and cDNA was
prepared (cDNA Synthesis kit, Stratagene, Heidelberg, FRG) according to the instructions of the suppliers. The resulting cDNA
was size-fractionated using a Sephacryl S-500 spin column. Fractions
containing cDNA molecules larger than 700 base pairs were pooled,
unidirectionally ligated into pJG4-5 (15), cut with EcoRI
and XhoI, and used to transform E. coli XL2-blue
MRF (Stratagene, Heidelberg, FRG). Ampicillin-resistant colonies were pooled, resulting in a cDNA library comprising 2 × 106 independent clones. Inserts ranged from 700 to 3500 base pairs with an average length of 1200 base pairs.
S. cerevisiae EGY48, containing pJK103 and
pLexL7, was transformed with the library. Approximately 2 × 107 transformants were plated on synthetic medium lacking
uracil, histidine, tryptophan, and leucine (17). After 3 days, colonies appeared. These were tested in a filter assay (18) for activation of
the -galactosidase reporter gene (lacZ). From colonies
that activated both the leu2 reporter gene (required in the
biosynthetic pathway for leucine) and the lacZ reporter
gene, library plasmids were rescued (17). Inserts were amplified by
PCR, followed by restriction mapping of the product with enzymes
AluI and HaeIII, and assigned to classes
according to their restriction map pattern. At least two cDNAs from
each class were partially sequenced.
Rescued library plasmids of S7 and ZNF791-687 were retransformed into EGY48/pJK103, containing as baits either pRFHM1 (15), which encodes the homeodomain of bicoid, or pLexA-myc (19), which encodes the carboxyl-terminal 176 amino acids of human c-Myc. Transformants were assayed for leu2 and lacZ activity.
DNA and Protein Sequence AnalysisDNA sequences were
determined by the dideoxy-mediated chain termination method using
[-35S]dATP and Sequenase DNA polymerase version 2.0 (U.S. Biochemical Corp.). Sequence analysis and homology searches were
performed using the GCG program package (Genetics Computer Group,
Madison, WI). Secondary structure predictions of proteins were
performed using the PSA algorithm (20). Predictions of probabilities
for putative coiled coil regions were performed with the program Coils (21). The width of the chosen window was 14 amino acids.
Full-length HisL7 was prepared as described (5) and
further purified by HPLC on a C4 reversed-phase column (Vydac,
Hesperia, CA). To produce HisS7, the full-length human S7 cDNA was
amplified by PCR using primers designed to provide BamHI and
XhoI sites for subcloning
(5-GCCTTGGATCCGCCATGTTCAGTTCGAGC-3
for sense primer and
5
-GGCGGCTCGAGTTACAATTGAAACTCTGG-3
for antisense primer). To produce HisZNF791-687, the human ZNF7 cDNA isolated
in the screen was amplified by PCR using primers again providing BamHI and XhoI sites for subcloning
(5
-AAGGATCCACCATGGAGCAGGCCTGTGAGGACA-3
for sense primer
and 5
-CGCCTCGAGCTATCCCATGTGAATTTTTTG-3
for antisense
primer). PCR products were digested with BamHI and
XhoI and ligated into pRSETA (Invitrogen, Heidelberg, FRG).
Recombinant plasmids pHisS7 and pHisZNF791-687 were
amplified in E. coli, and the integrity and orientation of
their cDNA inserts were verified by nucleotide sequencing. HisS7
and HisZNF791-687 were purified from bacterial lysates as
described (5). HisS7 was further purified by HPLC on a C4
reversed-phase column. Protein concentrations were determined
photometrically (22). Proteins were analyzed by SDS-PAGE and Coomassie
Brilliant Blue staining to determine their purity. The purity of the
preparations of HisL7 and HisS7 was >95%, and in the case of
HisZNF791-687 it was >80%.
Apparent dissociation constants (Kd) of L7-ZNF7 and L7-S7 dimers were determined essentially as described previously (5) by measuring the binding of [35S]cysteine-labeled ZNF791-687 to immobilized HisL7 or by measuring the binding of [35S]methionine-labeled L7 to immobilized HisS7. 35S-Labeled proteins were produced using the TNT coupled transcription/translation kit (Promega, Madison, WI) according to the instructions of the supplier. Plasmids used here were phumL7-14 (3), for radiolabeled L7, and pSK-ZNF7791-687. To produce pSK-ZNF7791-687, the BamHI and XhoI fragment from pHisZNF791-687 was subcloned into vector pSK II (Stratagene, Heidelberg).
Polyclonal AntibodiesAntibodies against human proteins were generated using standard immunization protocols (23). Anti-HisL7 antibodies were raised in chickens, and the antibodies were purified from egg yolk (24). In immunoblot analyses, this antiserum recognized a protein band that displayed the electrophoretic mobility of L7. Antibodies against HisZNF791-687 and against the ovalbumin-coupled peptide sequence CGRIHTGEKPYKG (anti-ZNF7PEP) that represents the conserved spacer sequence between zinc finger domains in ZNF7 and many other human zinc finger proteins were raised in rabbits. Both antisera recognized on immunoblots of cell lysates a band with the predicted electrophoretic mobility of ZNF7 (78 kDa) and two additional bands of 55 and 57 kDa, respectively. Polyclonal rabbit anti-S7 serum (25) was a gift from J. Stahl (Max-Delbrück-Center, Berlin). Polyclonal rabbit anti-LexA antibodies were a gift from E. A. Golemis. Polyclonal mouse anti HisL7 antibodies were a gift from A. Hohlbaum (Universität Konstanz, Konstanz, FRG).
ImmunoprecipitationsImmunoprecipitations were performed as described (16). In brief, 5 × 107 Molt-4 T lymphoma cells were lysed in TSA buffer (10 mM Tris-HCl, pH 8.0, 140 mM NaCl, 0.025% NaN3) containing 0.5% Triton X-100. Debris was pelleted, and the lysate was precleared overnight with 50 µl of activated, quenched Sepharose 4B (Pharmacia LKB, Uppsala, Sweden). The precleared extracts were then precipitated with 300 µl of anti-HisL7 serum, or with 300 µl of preimmune serum, both sera immobilized on Sepharose 4B. After 2 h of incubation at 4 °C, beads were washed twice with TSA buffer, containing 0.5% Triton X-100, and once with TSA buffer. The beads were eluted with nonreducing SDS-PAGE sample buffer, fractionated by SDS-PAGE, and transferred to nitrocellulose, and the immunoblots were incubated with the various antibodies employing standard procedures. Super Signal chemiluminescence reagents (Pierce) were used for the detection of signals.
Immunoblotting of RibosomesRibosomes from 5 × 106 Jurkat T lymphoma cells were prepared, fractionated on SDS-PAGE, and transferred to nitrocellulose (membranes a kind gift from J. Horwath, Universität Konstanz, Konstanz, FRG), which were treated as described above.
To
identify proteins, which potentially interact with protein L7 in
vivo, we performed interaction screening using the yeast two-hybrid system. The yeast reporter strain EGY48 containing the
lacZ reporter plasmid pJK103 was transformed with the
plasmid pLexA-L7, which encodes full-length protein L7 fused to the
LexA DNA-binding domain. Fig. 1 verifies
the correct expression of fusion proteins containing the LexA
DNA-binding domain fused to protein L7 and fragments thereof.
Immunoblots of yeast homogenates transformed with the indicated
plasmids and probed with anti-L7 and anti-LexA antibodies showed bands
of the expected size (52 kDa for LexA-L7, 38 kDa for
LexA-L71-124, and 37 kDa for LexA-L7125-248).
None of the bait plasmids alone activated the leu2 or the
lacZ reporter gene, respectively (data not shown). To
confirm that the LexA-L7 fusion proteins were synthesized in yeast and
that the LexA domains are fully functional to bind to LexA operator
sequences, we performed a repression assay (17) (data not shown).
To screen the Jurkat T lymphoma cDNA library, the
EGY48/pJK103/pLexA-L7 cells were transformed with the library plasmids. Approximately 2 × 107 yeast transformants were
screened. 173 colonies grew on leucine-free galactose plates and
displayed -galactosidase activity. Of these, 61 showed strong
galactose-inducible activity of both reporter genes. Library plasmid
DNA was prepared from these positive colonies. Their inserts were
amplified by PCR followed by restriction mapping of the product with
endonucleases AluI and HaeIII (not shown). Inserts isolated independently at least twice were assigned to seven
groups according to the size of the insert and its restriction site
pattern. The inserts of at least two cDNAs from each group were
then partially sequenced. The two most abundant cDNAs could be
matched to known sequences in the data base; 12 independently isolated
cDNAs (restriction group 1) coded for the multi-zinc finger protein
ZNF7 (26), with the longest cDNA encoding amino acids 91-687,
comprising the C-terminal domain of 15 subsequent zinc finger domains.
The full-length cDNA coding for ribosomal protein S7 (27) was
independently isolated seven times (restriction group 2). Restriction
group 3 (six isolated cDNAs) turned out to be the cDNA for
ribosomal protein S5 in a wrong reading frame. The remaining four
restriction groups, consisting of 2-4 cDNAs, coded for as yet
unidentified C2H2 zinc finger proteins, which could not be assigned to any sequence in the data base. This study focuses on the characterization of the abundant cDNAs of proteins S7 and ZNF7 as potential L7 interactors.
To confirm the specificity of the interactions of S7 and ZNF7 with protein L7, we retransformed the isolated library plasmids into EGY/pJK103, containing bait plasmids that encoded LexA fused to protein L7, to the homeodomain of drosophila bicoid, to the carboxyl-terminal domain of the human c-Myc protein. We then assayed for the activation of both reporter genes. The interaction of S7 and ZNF791-687 with protein L7 turned out to be specific, in that only cotransformation of the L7 bait lead to the activation of both reporter genes (Table I).
|
To determine the
region of L7 mediating the interaction of L7 with S7 or ZNF7, we
transformed the library plasmids of the latter proteins into EGY/pJK103
containing bait plasmids coding for LexA fused to full-length protein
L7 (pLexA-L7), the N-terminal half of protein L7 (comprising residues
1-124 (pLexA-L71-124)), or the C-terminal half of L7
(comprising residues 125-248 (pLexA-L7125-248)). As shown
in Table I, the N-terminal region of protein L7 is fully sufficient to
activate both the leu2 reporter gene and -galactosidase activity. No activation of the reporter genes was observed with the
pLexA-L7125-248 bait. The interface on protein L7 that interacts with S7 and ZNF7 thus resides between residues 1 and 124.
The interaction of S7 and
ZNF7 with L7 was further characterized by determination of the
dissociation constants of the S7-L7 and the ZNF7-L7 complexes.
Dissociation constants were determined utilizing a binding assay
employing oligohistidine-tagged proteins immobilized on a
Ni2+-chelate column at different concentrations.
Immobilized HisL7 was incubated with 35S-labeled
ZNF791-687 at a starting concentration of 3.5 × 105 M. Immobilized HisS7 was incubated with
35S-labeled L7 at the same starting concentration. For both
immobilized proteins, we utilized radiolabeled luciferase as a control
ligand. The columns were washed with a total of 15 batch volumes of
binding buffer, and the fractions of column-bound and eluted ligand
were determined for increasing dilutions of HisL7 and HisS7 (Fig.
2). For both proteins a lowest
concentration of approximately 25 × Kd was
found that provided nearly complete retention of liquid phase
radiolabeled ligand; hence, Kd values of 2 nM for the L7-ZNF7 interaction (Fig. 2B) and of
15 nM for the L7-S7 interaction (Fig. 2A) were
estimated. No retention of radiolabeled luciferase as an irrelevant
protein was observed on the HisL7 or HisS7 column.
Binding of S7 and ZNF7 to L7 in Vivo
Having demonstrated the
interaction of ribosomal protein S7 and ZNF7 with ribosomal protein
in vitro as well as in the yeast two-hybrid system, we
further examined the association of these proteins in a lymphoid cell
line. We immunoprecipitated the total cell lysate of MOLT-4 cells with
an immobilized polyclonal chicken anti-L7 antibody. The precipitate was
fractionated by SDS-PAGE and analyzed by immunoblotting using for
detection mouse anti-L7 antibodies, rabbit anti-S7 antibodies, and
rabbit anti-ZNF7 antibodies and an unrelated antibody as a control. As
shown in Fig. 3A, protein bands with the predicted electrophoretic mobility of proteins L7
(lane 1), S7 (lane 2), and ZNF7 (lanes
3 and 4) could be detected in the precipitate, thus
indicating co-precipitation with ribosomal protein L7. No signal was
found either with the unrelated control antibody (lane 5) or
in a control precipitation with immobilized preimmune serum
(lanes 6-10).
Binding of ZNF7 to Ribosomes
The association of ZNF7 with ribosomes was examined by immunoblotting of Jurkat T lymphoma ribosomes and visualization of ribosome-bound ZNF7 by anti-ZNF7 antibodies (Fig. 3B). Anti-L7 (lane 1) and anti-S7 antibodies (lane 2) served as a control for the preparation of ribosomal fractions. Lanes 3 and 4 show that a protein band with the predicted molecular mass of ZNF7 could be clearly detected by the two available anti-ZNF7 antibodies in preparations of ribosomes. The anti-ZNF7 antibody (lane 3), raised against recombinant ZNF7, detected an additional cross-reactive protein band with a molecular mass of 25 kDa, which was not detected by the anti-ZNFPEP antibody (lane 4). ZNF7 thus can exist in a ribosome-associated form.
In this study, we have used the yeast two-hybrid system to identify from a Jurkat T lymphoma cDNA library ribosomal protein S7 and multi-zinc finger protein ZNF7 as proteins that interact with ribosomal protein L7. Co-immunoprecipitation of both proteins with an anti-L7 antibody from the lysate of the MOLT-4 cell-line and dissociation constants of 15 nM for the S7-L7 interaction and 2 nM for the ZNF7-L7 interaction confirmed the potential in vivo interaction of these proteins. Our results show that the N-terminal half of L7, comprising residues 1-124 is fully sufficient to mediate the interaction with both S7 and ZNF7, respectively.
Protein S7 consists of 194 amino acids, has a predicted molecular mass
of 22 kDa, and is, like protein L7, very basic (pI of S7: 10.6) (27,
28). Protein S7 has been shown to be located at the back lobe of the
small 40 S subunit of eucaryotic ribosomes (24). Together with
ribosomal proteins S3, S3a, S14, and S15, it plays an important role in
the biogenesis of the 40 S small ribosomal subunit (29). Ribosomal
protein L7 is located at the surface of the large 60 S ribosomal
subunit (30, 31), to which it associates in the cytoplasm (32). The 50 C-terminal residues of protein L7 carry a RNA-binding domain that binds
with high affinity to 28 S rRNA (4). The N-terminal BZIP-like domain of
L7 is thought to be exposed to the cytoplasm due to its ability to bind
mRNAs with high affinity (3, 4) and because of cross-linking and
labeling experiments (31, 33). This domain also carries the
immunodominant autoepitope of protein L7 (12). These findings and the
data presented in this study suggest that the interaction of protein S7
with the N-terminal half of protein L7 is important for the interaction
of the small 40 S and the large 60 S ribosomal subunit during the
translation of mRNAs. This is supported by the finding that
ribosomal protein L7 could be in situ chemically cross-linked to both the 28 S rRNA of the large ribosomal subunit and
to the 18 S rRNA of the small ribosomal subunit (34). Secondary structure predictions performed on the amino acid sequence of protein
S7 suggest that it has, like protein L7, mainly an -helical conformation (Fig. 4A). A
prediction of putative coiled-coil regions in S7 shows a high
probability (>90%) for such a domain between amino acids 29 and 50 of
protein S7 (Fig. 4B). This region shows also a high
probability for an
-helical conformation, which is required for a
coiled-coil structure. Within this predicted amphipathic helix are two
leucine repeats in the D position of the helix (residues 36 and 43) and
hydrophobic residues in the A position of the helix (residues 40 and
47). In a previous study, we have shown that a putative coiled-coil
region, located in the first N-terminal
-helix of protein L7, is
capable of forming leucine zipper-like homodimers (5). We therefore
suggest that the interaction of proteins L7 and S7 is mediated by the
BZIP-like domain of L7 and the putative leucine zipper domain of
protein S7.
ZNF7 is a transcription factor-like multi-zinc finger protein consisting of 687 amino acids and has a predicted molecular mass of 78 kDa. It is widely expressed in various cell lines. The protein consists of 15 C2H2 zinc finger domains that comprise nearly 70% of the protein (26). The finger domains share 70% sequence similarity and conform almost exactly to the consensus sequence of DNA- and RNA-binding zinc fingers (35). In addition to the nucleic acid binding properties of zinc finger domains, zinc finger-mediated protein protein interactions have been reported (36), including the interaction of a zinc finger domain with a leucine zipper motif (37). However, the interfaces responsible for the interaction of L7 and ZNF7 remain to be specified. The cellular localization of ZNF7 is unknown, although a nuclear localization has been suggested (26). However, a localization in the cytoplasm has not been excluded, and we show in this study that ZNF7 co-purifies with eucaryotic ribosomes. This suggests a ribosomal localization of ZNF7, where it may act as translational regulator. In addition, ZNF7 has been recently shown to be a substrate of mitogen-activated protein kinase in vitro (38), suggesting a role in signal transduction pathways. However, it is unknown whether mitogen-activated protein kinase-mediated phosphorylation of ZNF7 influences the binding of ZNF7 to L7 in vivo.
In summary, we have shown in this and previous studies that protein L7 can interact with various components of the translational apparatus (i.e. mRNA, 28 S rRNA, ribosomal protein S7, and ZNF7). This hints at a complex role of L7 in translational regulation. It is clear, however, that the functional significance and the molecular details of these interactions remain to be analyzed.
This paper is dedicated to Professor Rudolf Pichlmayr.
We thank Erica A. Golemis for providing plasmids and strains and for valuable help in setting up the yeast two-hybrid system. We are also grateful to Joachim Stahl for providing antibodies, to Andreas Hohlbaum and Jörg Horwath for providing reagents and helpful discussions, to Rikiro Fukunaga for communicating data prior to publication, and to Katja Aviszus and Elli Neu for comments on the manuscript.