ABSTRACT
Watermelon stomach is characterized by prominent stripes of ectatic vascular
tissue in the stomach similar to stripes on a watermelon; in patients with this
disorder chronic gastrointestinal bleeding occurs and approximately half of
these patients have associated autoimmune disorders. In the serum of one
patient, an antinucleolar antibody titer of 1:25 600 was found; the antibodies
specifically recognized a
~
100 kDa nucleolar protein, which we referred to as the `Gu' protein. Its cDNA
was cloned and sequenced. The Gu protein is a member of a new subgroup of RNA
helicases, the DEXD box family. Gu protein fused with glutathione S-transferase contains ATP-dependent RNA helicase activity which preferably translocates in the
5
' ->
3
'
direction. Its RNA folding activity, RNA-dependent ATPase and dATPase activities, and its translocation direction
are similar to those of RNA helicase II [Flores-Rozas,H. and Hurwitz,J. (1993)
J. Biol. Chem
. 268, 21372-21383]. Sequencing of 209 amino acids of RNA helicase II peptides showed
96.7% identity with the cDNA-derived amino acid sequence of the Gu protein. The precise biological
roles of this RNA helicase in the biogenesis of ribosomal RNA and the
pathogenesis of watermelon disease and autoimmune disorder require further
study.
Watermelon stomach is another term for gastric antral vascular ectasia (GAVE)
disease which was first described in 1984. This disease is characterized by
prominent stripes of ectatic vascular tissue radiating outward from the pylorus
in a form resembling the stripes on a watermelon (
1
). This disease commonly occurs in older women and may be associated with
gastric hemorrhage, iron-deficiency anemia and autoimmune disorders in ~60% of the patients (
2
,
3
). The pathogenesis of this disorder and its relationships to autoimmune disease
are being studied.
In patient `Gu', a 1:25 600 titer of antinucleolar antibodies was found. To
understand the association of these antibodies to this disease, the cDNA that
codes for the autoantigen was cloned. Immunofluorescent staining showed that
antibodies produced prominent nucleolar fluorescence. Western blot analysis
using this serum stained an ~100 kDa protein, referred to as the Gu antigen, derived from the name of
the patient.
Cloning and sequencing of the cDNA for Gu protein is reported in this paper.
Based on the cDNA-derived amino acid sequence and the RNA helicase activity of the expressed
protein, the Gu protein belongs to a new DEXD box family of RNA helicases (
4
). These RNA helicases have been implicated in pre-mRNA splicing, translation, ribosomal processing, cell growth and
development (
5
). The cDNA and the derived amino acid sequence of the Gu protein may aid in
understanding the relationship of the pathogenesis of this autoimmune disorder
associated with watermelon stomach as well as the role of RNA helicases in the
biogenesis of ribosomal RNA.
HeLa cells were grown at 37oC in Dulbecco's modified Eagle's medium with 10% newborn calf serum and 5%
CO
2
.
The serum, which was obtained from a patient with watermelon stomach disease,
was used in various experiments either as such or as IgG after purification on
a protein A affinity column according to the supplier (Pierce).
Purified immunoglobulins were labeled with Bolton-Hunter Reagent [
125
I] (ICN) according to the supplier, and the labeled IgGs were used to screen a [lambda]gt11 HeLa cDNA expression library (Clontech). A 2 kb insert from
positive clone 1A1 was subcloned into Bluescript II KS (Stratagene) and
sequenced by the dideoxynucleotide sequencing method using a Sequenase kit (US
Biochemical). To find a longer cDNA clone, a 450 bp fragment from the 5'-end of clone 1A1 was
32
P-labeled and used to screen the same cDNA library using QuikHyb solution
(Stratagene). A positive clone (7A) with a 3.3 kb insert was selected. This [lambda]gt11 clone was digested with
Eco
RI and the resulting four insert fragments were subcloned into pTZ18R (Pharmacia
Biotech) and both strands were sequenced. The orientation of each fragment was
determined by sequencing PCR-amplified products.
Additional 5' sequence of Gu cDNA was obtained using HeLa poly(A)
+
mRNA and 5' RACE system (Life Technologies). The PCR product obtained using this kit
was subcloned at the
Sal
I-
Bam
HI sites of pTZ18R and sequenced as described above.
Poly(A)
+
mRNA was isolated from HeLa cells using an mRNA purification kit (Pharmacia
Biotech). Northern blots were done as described (
8
) using
32
P-labeled PCR-amplified cDNA fragments and QuikHyb solution (Stratagene).
The coding sequence for Gu protein was amplified by PCR and subcloned into the
Bam
HI-
Xho
I sites of pGEX 4T-3 vector (Pharmacia) in frame with glutathione
S
-transferase (GST) gene. The fusion protein was expressed in DH5[alpha]
Escherichia coli
cells (Life Technologies, Inc.) and purified from 3 l Luria broth medium using
glutathione Sepharose 4B resin as described (
9
). The protein was eluted from the resin with 50 mM Tris-HCl, pH 8.0, containing 5 mM reduced glutathione and dialyzed overnight
against Buffer A (20 mM HEPES-KOH pH 7.6, 0.1 mM EDTA, 1 mM DTT, 10% glycerol) with 0.1 M KCl. The
dialyzed fraction was loaded onto a pre-equilibrated 5 ml HiTrap heparin sepharose column (Pharmacia) which was
then washed with 50 ml 0.1 M KCl in buffer A. Bound proteins were eluted with a
linear gradient of 0.1-1.0 M KCl in buffer A. The GST-Gu fusion protein eluted at ~0.6 M KCl and was visualized by silver staining (BioRad kit).
Fractions containing the GST-Gu fusion protein were pooled, concentrated to 0.5 ml using Centricon 30
(Amicon) and desalted by washing two times with buffer A (containing 0.1 M KCl)
followed by further concentration using Nanosep 30 (Pall Filtron). The final
fraction (80 [mu]l) was aliquoted into 20 [mu]l fractions (1 [mu]g/[mu]l protein) and stored at -80oC until used. Protein concentrations were determined
using the BioRad protein assay.
RNA helicase activity was determined using 5'- and 3'-tailed substrates synthesized using Ambion's
MAXIscript kit. The 5'-tailed substrate was prepared by transcribing with T3 or T7 RNA
polymerase the Bluescript II KS plasmid (Stratagene) cut with
Bam
HI or
Hin
dIII, respectively. The lower strand T7 transcript was labeled with [[alpha]-
32
P]GTP. The preparation of the 3'-tailed substrate, gel purification of all the substrates, RNA
helicase assay and ATPase assay were done as described (
11
) with the exception that 10% polyacrylamide-SDS gels were used for RNA helicase assays.
A total of 25 [mu]g of the Cibacron blue-agarose fraction containing RNA helicase II (
11
) was subjected to electrophoresis on a 10% polyacrylamide-SDS gel and then electroblotted onto a nitrocellulose filter and detected
by reversibly staining with 0.2% solution of Ponceau-S (Sigma). The band containing the protein was excised. The nitrocellulose
immobilized protein was subjected to trypsin digestion as described (
12
). The resulting peptides were purified through a microbore reverse-phase high performance liquid chromatography (
13
) and subjected to N-terminal micro-sequencing by Edman degradation (
14
).
Nucleic acid and amino acid sequence analyses were done using the Wisconsin
Sequence Analysis Package GCG Version 8.
Indirect immunofluorescence of fixed HeLa cells using serum from a patient with
watermelon stomach disease revealed the nucleolar localization of Gu protein in
HeLa cells (Fig.
1
A and B). Staining of MCF7, COS-7, CHO-K1 and NIH 3T3 cells showed similar nucleolar localization (data not
shown). Western blot analysis showed that the autoimmune antibodies bound to a
specific nucleolar protein with a molecular mass of ~100 kDa (Fig.
1
C). These results indicate that the Gu autoantigen in this patient is an
immunologically conserved 100 kDa nucleolar protein.
To identify the antigen recognized by the watermelon disease autoimmune serum, a
HeLa cDNA expression library in [lambda]gt11 was immunoscreened using purified
125
I-labeled IgG. A clone with 2 kb insert (clone 1A1 in Fig.
2
) was selected. Sequencing showed that the insert contained a poly(A)
+
tail at its 3'-end and 1 kb open reading frame from its 5'-end. Subcloning of this cDNA into a bacterial
expression vector and subsequent expression produced a 40 kDa peptide which
immunoreacted with the patient's serum (data not shown). To isolate the 5' portion of the cDNA, the same library was screened with a
32
P-labeled probe corresponding to the 5'-end of clone 1A1 (Fig.
2
). A clone with 3.3 kb insert was obtained, the 3'-end of which contained the sequences found in clone 1A1. An
additional 5' sequence of 12 nucleotides was obtained using a 5' RACE (rapid amplification of cDNA ends) kit. The sequence for the
Gu protein cDNA was deposited in GenBank with accession number U41387.
To show that the isolated cDNA clones code for the Gu protein, clone 1A1 was
expressed as a GST fusion protein and blotted onto nitrocellulose paper.
Antibodies from the patient serum were bound to the fusion protein and then
eluted. The eluted antibodies recognized a ~100 kDa protein localized in the nucleolus (Fig.
3
A and B).
Northern blot analysis of poly(A)
+
mRNA isolated from HeLa cells showed that the Gu mRNA is ~4.5 kb long (Fig.
3
C). The same sized mRNA was recognized in two independent experiments using
different cDNA probes shown in Figure
2
. Based on this result and the sequence of the longest available cDNA clone, Gu-mRNA presumably contains 0.8 kb 5' untranslated region, an ~2.7 kb coding region, and a 1 kb 3' untranslated region.
Figure
4
shows the cDNA-derived amino acid sequence of Gu protein. The first four amino acid
residues were obtained from a clone derived using a 5' RACE kit and the remaining sequences were derived from clones 1A1 and 7A
(Fig.
2
). The calculated molecular mass of the available cDNA-derived amino acid sequence is 89 kDa.
Gu protein is a basic protein with combined lysine and arginine residues of 17%
and a computed pI of 10.1. The N-terminal region is particularly rich in lysine residues. Groups of lysine
residues (underlined in Fig.
4
) are putative bipartite nuclear localization signals (
15
). The C-terminal region is rich in arginine and glycine residues, a domain
reported to be essential for efficient binding of nucleolar protein C23 to RNA
(
16
). An RNA binding activity of Gu clone 1A1 (C-terminal region) was observed (data not shown). This region of Gu protein
also contains three FRGQR repeats and one PRGQR (double-underlined in Fig.
4
) of unknown function.
The middle portion of the molecule contains regions highly conserved in RNA
helicases (boxed in Fig.
4
). Comparison of the Gu protein sequence with the Peptide Sequence Databases
shows the presence of nine motifs conserved in RNA helicases in species ranging
from virus to human (
4
,
17
,
18
). Two families of RNA helicases have been reported; the DEAD box family and the
DEAH/DEXH box family (
4
). The Gu protein belongs to the DEXD box family, a variation of the DEAD box
family. This variation is similar to DEXH of the DEAH group.
RNA helicase II has been purified from nuclear extracts of HeLa cells (
11
). Peptides of RNA helicase II were sequenced and compared to the cDNA-derived amino acid sequence of the Gu protein. Of these, 202 amino acid
sequences (209 residues compared) were identical to those derived from the cDNA
protein sequence (Fig.
4
). These identities include highly conserved helicase domains and regions that
vary widely among different helicases which are believed to be responsible for
the specific functions intrinsic to the individual protein (
4
).
Determination of the RNA helicase activity of the GST-Gu fusion protein showed that 5 ng of the fusion protein could unwind 50
fmol double-strand RNA with 5' overhangs under the assay conditions and it was 20% efficient in
unwinding 50 fmol double-strand RNA with 3' overhangs (Fig.
5
A, lanes 3 and 8). With 2.5 ng of the fusion protein 20% displacement was
observed using the 5' substrate but no activity was observed when 3' substrate was used (data not shown). No RNA helicase activity was
observed in the absence of ATP (Fig.
5
A, lanes 5 and 9). Comparison of lanes 2 and 3 (Fig.
5
A) showed that the RNA helicase product migrated more slowly than the single-strand RNA on 10% polyacrylamide-0.1% SDS gels. Boiling this product prior to loading onto the gel
resulted in a migration identical to the single strand RNA (lane 4). These
results imply that the Gu enzyme unwinds the double-strand RNAs and, in addition, catalyzes the folding of the radioactive
strand. These results are similar to those observed with RNA helicase II, which
exhibited unwinding and folding activities (
11
).
This work was supported by the DeBakey and Busch Funds (to H.B.) and the
American Heart Association Grant-in-Aid (to J.H.).
REFERENCES
Return