Article

Hu, Z. et al. A novel nuclear protein, 5qNCA (LOC51780) is a candidate for the myeloid leukemia tumor suppressor gene on chromosome 5 band q31. Oncogene 20, 6946-6954

Section of Hematology/Oncology, Department of Medicine, University of Illinois at Chicago, Chicago, Illinois 60607-7170, USA.
Oncogene (Impact Factor: 8.46). 11/2001; 20(47):6946-54. DOI: 10.1038/sj.onc.1204850
Source: PubMed
ABSTRACT
Interstitial deletion or loss of chromosome 5, del(5q) or -5, is a frequent finding in myeloid leukemias and myelodysplasias, suggesting the presence of a tumor suppressor gene within the deleted region. In our search for this gene, we identified a candidate, 5qNCA (LOC51780), which lies within a consistently-deleted segment of 5q31. 5qNCA expresses a 7.2-kb transcript with a 5286-bp open reading frame which is present at high levels in heart, skeletal muscle, kidney, placenta, and liver as well as CD34+ cells and AML cell lines. 5qNCA encodes a 191-kD nuclear protein which contains a highly-conserved C-terminus containing a zinc finger with the unique spacing Cys-X2-Cys-X7-His-X2-Cys-X2-Cys-X4-Cys-X2-Cys and a jmjC domain, which is often found in proteins that regulate chromatin remodeling. Expression of 5qNCA in a del(5q) cell line results in suppression of clonogenic growth. Preliminary sequence results in AML and MDS samples and cell lines has revealed a possible mutation in the KG-1 cell line resulting in a THR to ALA substitution that has not been found in over 100 normal alleles to date. We propose 5qNCA is a good candidate for the del(5q) tumor suppressor gene based on its predicted function and growth suppressive activities, and suggest that further mutational and functional study of this interesting gene is warranted.

Full-text

Available from: Stephen Horrigan
A novel nuclear protein, 5qNCA (LOC51780) is a candidate for the myeloid
leukemia tumor suppressor gene on chromosome 5 band q31
Zhenbo Hu
1
, Ignatius Gomes
1
, Stephen K Horrigan
2
, Jele na Kravarusic
1
, Brenton Mar
1
,
Zarema Arbieva
1
, Brent Chyna
1
, Noreen Fulton
1
, Seby Edassery
1
, Azra Raza
3
and
Carol A Westbrook*
,1
1
Section of Hematology/Oncology, Department of Medicine, University of Illinois at Chicago, Chicago, Illinois, USA;
2
Avalon
Pharmaceuticals, Gaithersberg, Maryland, USA;
3
Rush Cancer Institute, Chicago, Illinois, USA
Interstitial deletion or loss of chromosome 5, del(5q) or
75, is a frequent ®nding in myeloid leukemias and
myelodysplasias, suggesting the presence of a tumor
suppressor gene within the deleted region. In our search
for this gene, we identi®ed a candidate, 5qNCA
(LOC51780), which lies within a consisten tly-deleted
segment of 5q31. 5qNCA expresses a 7.2-kb transcript
with a 5286-bp open reading frame which is presen t at
high levels in heart, skeletal muscle, kidney, placenta, and
liver as well as CD34+ cells and AML cell lines. 5qNCA
encodes a 191-kD nuclear protein which contains a
highly-conserved C-terminus con taining a zinc ®nger with
the unique spacing Cys -X2-Cys-X7-His-X2-Cys-X2-Cys-
X4-Cys-X2-Cys and a jmjC domain, which is often found
in proteins that regulate chromatin remodeling. Expres-
sion of 5qNCA in a del(5q) cell line results in suppression
of clonogenic growth. Preliminary sequence results in
AML and MDS samples and cell lines has revealed a
possible mutation in the KG -1 cell line resulting in a THR
to ALA substitution that has not been found in over 100
normal alleles to date. We propose 5qNCA is a good
candidate for the del(5q) tumor suppressor gene based on
its predicted function and grow th suppressive activities,
and suggest that further mutational and functional study
of this interesting gene is warranted. Oncogene (2001)
20, 6946 ± 6954.
Keywords: tumor suppress or; nuclear protein; 5qNCA;
leukemia; zinc ®nger; jmjC domain
Introduction
Cytogenetic deletions involving chromosome 5 are
frequently seen in malignant myeloid disorders, includ-
ing myelodysplasia (MDS) and acute myeloid leukemia
(AML). These abnormalities include complete loss of
the entire chromosome [75], loss of the long arm of
chromosome 5[5q-], or interstitial deletions of the long
arm [del(5q)] which usually involve multiple chromo-
some bands (Pedersen-Bjegaard et al., 1990; Le Beau et
al., 1993). Almost a third of MDS cases demon strate
chromosome 5 loss, including over half of those cases
that occur as a late sequel of cancer chemotherapy
(Pedersen-Bjegaard et al., 1990; Le Beau et al., 1993;
Rowley et al., 1998; Westbrook and Le Beau, 1993;
Feanaux et al., 1996). In AML, deletions are much less
frequent, but occur predominantly in high-risk pa-
tients, including the elderly, AML arising after MDS,
and AML arising after chemotherapy; they are
invariably associated with poor response to chemother-
apy, rapid progression, and short survival (Samuels et
al., 1988; Schier et al., 1989; Pedersen-Bjegaard,
1993). Identifying the genes responsible for these
syndromes will help to improve the understanding
and treatment of some of the most deadly forms of
AML and MDS.
The association of a chromosomal loss with a
clinical malignancy suggests the presence of a tumor
suppressor gene, whose mutational loss or inactivation
leads to transformation. Speci®cation of a minimal
deletion is the ®rst step for identi®cation of such a
gene; the demonstration of the gene's inactivation by
deletion and mutation is an important second step.
Early studies de®ned a 4 Mb interval within 5q31 that
was deleted in all patients with aggressive MDS,
AML, or secondary MDS (Le Beau et al., 1993;
Bartoloni et al., 1998; Horrigan et al., 1999). Attempts
to further re® ne the 5q31 deletion using FISH and
loss of heterozygosity by ourselves (Horrigan et al.,
1996, 2000) and others (Zhao et al., 1997; Frazer et
al., 1997) has speci®ed two relatively small intervals,
each of about 1 Mb in extent, that are immediately
adjacent and located on either side of the marker
D5S500. Given the uncertainties of FISH and allelic
loss mapping, it is uncertain whether these intervals
represent two distinct tumor suppressor genes or point
to the same gene. Although a number of candidate
genes have been localized to this region of 5q31, and
many have compelling biology (Le Beau et al., 1993;
Horrigan et al., 1999; Xie et al., 2000), none of them
has been shown to meet the required criteria for a
classical tumor suppressor gene: evidence for muta-
tional inactivation in clinical samples with del(5q).
Oncogene (2001) 20, 6946 ± 6954
ã
2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00
www.nature.com/onc
*Correspondence: CA Westbrook, University of Illinois at Chicago,
Departments of Medicine and Genetics, M/C 734, 900 S. Ashland,
Chicago, IL 60607 ± 7170, USA; E-mail: cwcw@uic.edu
Received 10 May 2001; revised 9 July 2001; accepted 17 July 2001
Page 1
The recent release of genomic sequence for 5q31
allowed us to identify additional candidate genes and
to supplement those that we have previously reported
(Horrigan et al., 1999). These resources were applied
to the isolation and characterization of candidate
genes.
Here we report the identi®cation and preliminary
characterization of one such candidate gene, which we
name `5q Nuclear Co-Activator', or `5qNCA'. We
present the c omplete cDNA and protei n sequence of
this gene, demonstrate its nuclear localization and
growth suppressive activity, and the identi®cation of a
mutation in a del(5q) cell line. We propose that
5qNCA is a promising candidate for the tumor
suppressor genes on 5q whose loss or inactivation
may lead to MDS and AML.
Results
Identification of 5qNCA
Genomic sequence near the D5S500 marker was
identi®ed (http://www.genome.wustl.edu/gsc/index.shtml)
and searched for candidate genes. AC004821, which is
known to contain the candidate genes EGR1 and
SGC32445 (Horrigan et al., 2000), is just centromeric
to D5S500 and was selected for further analysis. The
sequence was searched for additional ESTs by BLAST
search (http://www.ncbi.nlm.nih.gov) of the human
EST database (dbEST, http://www.ncbi.nlm.nih.gov/
blast/). Three ESTs were identi®ed (stSG48489,
N24929, stSG12690), all of which mapped to Unigene
cluster #Hs.24125. The longest open reading frame in
Hs.24125, named KIAA1082, was 4.7 kb in size
(Figure 1a), considerably shorter than that predicted
by Northern analysis (see below). To obtain the
complete sequencefor the gene represented by KIAA1082,
GeneScan (http://genes.mit.edu/GENSCAN.html) was
used to predict additional upstream exons. This was
accomplished by sequence scanning AC004821, and by
comparing with genes which showed sequence homol-
ogy; in particular, KIAA0742 (now known to encode
the human homologs of TSGA, Testis-Speci®c Gene
A). The predicted ®nal transcript size of this composite
cDNA, which we have named 5qNCA, is 6.8 kb, and it
contains several ATG start sites and a large open
reading frame of 5286 kb (Figure 1a) (GenBank
#AF338242); the predicted protein is given in Figure
1b.
Intron-exon boundaries were localized by alignment
of the cDNA with the genomic sequence. The genomic
arrangement, and its relation to other nearby genes
and markers, is shown in Figure 2. 5qNCA contains 24
exons and 23 introns and spans over 85 kb. The intron
between exons 1 and 2 is the longest, being approxi-
mately 20 kb in length, while exon 8 is the longest
exon, being 1.25 kb in length. All intron-exon
boundary sequences contain consensus GT/AG se-
quences at donor and acceptor sites.
Expression of 5qNCA
An exon-connection strategy was used to con®rm that
all predicted exons in the composite 5qNCA cDNA
were expressed in a contiguous message. RT ± PCR was
performed using primer pairs designed from the
predicted sequence on normal marrow-derived cDNA.
All predicted exons were con®rmed by this strategy.
a
b
GVCLDC
CDVCETTLFN IHWVCRKCGF
Figure 1 (a) Full length cDNA of 5qNCA compared to related
cDNA sequences in GenBank. The cDNAs in the diagram are
shown approximately to scale as numbered above in basepairs.
5qNCA is depicted as a horizontal bar immediately below the
scale. The alternative ATG start codons are depicted above the
diagram, while the location of sense (S) and antisense (AS)
primers for exon connection are shown below. The additional
GenBank submissions related to this gene are depicted as
horizontal bars below 5qNCA, along with their GenBank
accession number. For each clone, the ATG translation-initiation
site and stop codon are shown as vertical lines numbered from the
start of the clone; the vertical line labeled `A 355' above
AF251039 indicates the single nucleotide insertion at position
355. Also shown is the size of the predicted open reading frame
(ORF), in base pairs. For each clone, the hatched area depicts the
open reading frame, the clear area the untranslated region, and
the dark area represents the 96-bp sequence that is present only in
5qNCA (see text). (b) Predicted protein sequence of 5qNCA. The
5qNCA protein consists of 1761 amino acids with a molecular
weight of 191 kD. The protein contains a zinc ®nger domain
(highlighted), a jmjC domain (boxed) and (a) nuclear localization
domain (underlined)
Oncogene
5qNCA, a candidate for the chromosome 5 tumor suppressor gene
ZHuet al
6947
Page 2
The sequence of the 5' predicted exons was further
con®rmed by sequencing the 1.5 kb RT ± PCR product,
¯anking exon 1 to exon 8, that was generated by
primer pair S1+AS1 (Figure 1a); as expected, all the
cDNA sequences matched those predicted, con®rming
that the expected transcript was expressed in its
entirety.
While this work was in progress, additional cDNA
sequences were deposited in the GenBank that are
related to 5qNCA, as shown in Figu re 1a. KIAA1082,
the ®rst clone isolated for 5qNCA, is much shorter, and
predicts the use of the most 3' ATG at position 2358,
resulting in an open reading frame of 3129 bp.
LOC51708 is a 6322-bp sequence that is almos t identical
to 5qNCA but is 390 bp shorter at the 5' end, and is also
missing 96 nuc leotides from exon 7, resulting in a shorter
open reading frame of 4254 bp, since it utilizes the
downstream ATG that corresponds to nucleotide 1135.
AF25103 is identical to LOC51708, but contains an
additional nucleotide at position 355, resulting in a
frameshift relative to 5qNCA. AF25103 was recently
reported as a gene, C5ORF7, that encodes a potential
nuclear protein on 5q31 (Lai et al., 2000). Alignment of
the four cDNA sequences (Figure 1a) suggests that they
are probably derived from the same gene, with the
dierences noted above. Note that the 96 additional
nucleotides within exon 7 of 5qNCA are not present in
AF251039 and LOC51780 and result in an additional
exon boundary in AF251039 and LOC51780. The
presence of these 96 nucleotides was con®rmed in all
sequenced RT ± PCR products from cell lines KG-1 and
HL-60 to bone marrow mononuclear cells and the
cDNA from a placenta library (data not shown), as well
as all genomic products of 5qNCA to date (data not
shown). We have not observed a 6.3 kb mRNA
transcript on Northern blots, as would be expected for
both AF251039 and LOC51780. We believe that
AF251039 and 5qNCA are both derived from
LOC51780, but that AF251039 may be a rare splice
variant. Alternatively, the clone from which AF251039
was derived may contain a sequencing error or a
mutation resul ting in a truncated protein and loss of
expression of the 5' sequences of 5qNCA that are
homologous to TSGA. The physiologic basis of this
smaller transcript form remains uncertain.
5qNCA expression in normal and malignant tissues
A multiple tissue blot (Clontech, Palo Alto, CA, USA)
was hybridized to a probe derived from a non-
conserved region of 5qNCA (nucleotides 5120 ± 6480).
As shown in Figure 3, the gene is expressed in many
tissues as a transcript of approximately 7.2 kb, with the
highest levels present in the placenta, skeletal muscle,
kidney, heart and liver. Two smaller transcripts are
also present in placental RNA at 3 and 1.35 kb,
respectively. The 1.35-kb transcript appears highly-
abundant in the placenta. 5qNCA is also highly
expressed in human CD34
+
bone marrow cells
(unpublished data). Figure 3b shows a Northern blot
from three leukemia cell lines and three cancer cell
Figure 2 Genomic structure of 5qNCA relative to the sequence
contig AC004821. Structures are oriented telomeric to left and
centromeric to right. The location of exons of 5qNCA are
indicated as vertical lines or boxes on the clone. The location of
reference markers WI-8265, SHGC-11945 and WI-4406 are
indicated on AC004821 by vertical lines
Figure 3 (a) Northern blot analysis of 5qNCA in human multiple
tissues. The human multiple tissue blot (Clontech, Palo Alto, CA,
USA) contains mRNA (2 mg/lane) from the human tissues
indicated above each column. A transcript of approximately
7.2 kb (arrow) is apparent in heart, skeletal muscle, kidney and
placenta; and with lower expression in lung and liver. Three
transcripts of about 7.2, 3.0 and 1.35 kb are detected in placental
tissue. (b) Northern blot analysis of 5qNCA in human leukemia
and solid tumor cell lines. Each lane of the blot contains 3 mg
mRNA isolated from myeloid leukemia cell lines and solid tumor
cell lines, as indicated above each column. A transcript of 7.2 kb is
observed in all six cell lines, as indicated by the arrow in the right
margin. The sizes of RNA markers are indicated at the left in kb
5qNCA, a candidate for the chromosome 5 tumor suppressor gene
ZHuet al
6948
Oncogene
Page 3
lines, con®rming a fairly high level of a single
transcript of 5qNCA. A similar study of 16 additional
myeloid and lymphoid leukemia cell lines revealed that
5qNCA is wide ly expressed in all of the cells examined
(data not shown), with a single transcript of about
7.2 kb. The transcript size suggests that there may be
some additional 5' sequence of 5qNCA that has not yet
been identi®ed.
Sequence homologies and functional motifs
As seen in Figure 4, 5qNCA shares sequence
homologies across the plant and an imal kingdom.
The 3' half of the ORF is highly homologous to
TSGA, testis-speci®c gene A (Hoog et al., 1991) in rat,
mouse, and human; TRIP8, a thyroid-hormone binding
protein (Lee et al., 1995), the human and murine
hairless gene (Panteleyev et al., 1998; Cichon et al.,
1998), Drosophila CG8165, and arabidopsis/corn
ENBP1. Hairless is involved in the dierentiation and
apoptosis of hair follicles, but it has not been well
characterized. ENBP1 is involved in plant nodule
development. TRIP8 has been predicted to be a
thyroid-hormone recept or co-activator. There are no
known functions for TSGA or Drosophila CG8165.
The three conserved motifs for all of these genes
include a zinc ®nger domain (Figure 1a, Table 1), a
jmjC domain, and an LXXLL nuclear localization
signal (Figure 1b). The zinc ®nger demonstrates a
unique C-spacing, Cys-X2-Cys-X7-His- X2-Cys-X2-
Cys-X4-Cys-X2-Cys, which resembles, but is not
identical to, the LIM and PHD ®ngers; this spacing
is shared with TSGA, TRIP8 and ENBP1, but varies
slightly from hairless. 5qNCA contains a region which
is 84.7% identical to the consensus jmjC domain
(Clissold and Ponting, 2001). This domain was ®rst
described in the protein Jumonji and has since been
found in over 145 proteins. Although its exact function
is unclear, a common theme in jmjC containing
proteins is the regu lation of chromatin modeling.
The protein product of 5qNCA
To evaluate the size and cellular localization of the
5qNCA protein product, an expression vector contain-
ing the complete cDNA open reading frame was
constructed in pTracer, a plasmid vector which
contains the GFP marker, Zeocin resistance, and a
histidine tag in-frame with the inserted cDNA. The
recombinant plasmid was introduced into COS-1 cells,
and positive clones were selected with Zeocin and
expanded for further analysis. Protein lysate prepared
from cells containing the 5qNCA express ion vector was
examined by Western blot using anti-His antibody,
which demonstrated the presence of a 191-kD protein
as the major species in three subclones examined
(Figure 5). This protein size correlates with that
expected from the largest predicted open reading frame
of 5.26 kb. To detect the cellular localization of
5qNCA, COS-1 cells expressing the gene construct
were cultured on slides, ®xed, and immunostained with
anti-His antibody. Fluorescent microscopy revealed
that the 5qNCA-His fusion protein localized primarily
to the nucleus, though a very low level of ¯uorescence
was observed in the cytoplasm as well (Figure 6).
Growth suppressive properties of 5qNCA
To study the eects of 5qNCA on leukemic growth,
the gene was introduced into the MUTZ-1 cell line,
Figure 4 Schematic representation of 5qNCA showing predicted
protein domains and homologies with other genes. The upper bar
depicts 5qNCA and the numbers (1 ± 24) indicate the 5qNCA
exons, approximately to scale, while amino acid numbers are
indicated with vertical lines above. Sequence homologies with
other genes are shown as bars below 5qNCA; clone or protein
names and species names are indicated above each bar. Known
protein domains are shown as gray boxes on the top of the ®gure;
`C2H5C2' is the zinc ®nger, `LXXLL' is the nuclear localization
motif, and `jmjC' indicates the jmjC domain
Table 1 Conserved Zinc-®nger domain
Gene name GenBank ID Protein sequence
5qNCA AF338242 GVREMC DVC ETTLFNIH WVC RKC GFGVC LDC YRLR K S R P
TSGA AB018285 GVREMC DVC DTTIFNLH WVC PRC GFGVC VDC YRMK ± ± ± ±
FLYCG8165 CG8165 GTREIC DVC DTTLFNYH WTC RKC GFGVC LDC VKDR K E G L
TRIP8 AB037801 GVREMC DAC EATLFNIH WVC QKC GFVVC LDC YKAK ± ± ± ±
Hr-mouse Q61645 GIPHCCSRCHHGLFNTH WRC SHC SHRLC VAC GRIA G A G ±
Hr-rat P97609 GIPHCCSRCHHGLFNTH WRC SHC SHRLC VAC GRIA G A G ±
Hr-human AF155648 GIPRCCSRCHHGLFNTH WRC PRC SHRLC VAC GRVA G T G ±
Arab EST AAB96846 DERVYC DHC ATSIVDLH RSC PKC SYELC LKC CQEI R E G ±
ENBP1-arab AC002131 EERVFC NHC ATSIVDLH RSC PKC SYELC LNC CQEI R ± ± ±
Arab EST AAF13079 DERLYC DIC KTSIYDLH RNC KSC SFDIC LSC CLEI R N G K
The conserved Cysteine or Histidine residues that comprise the Zinc-®nger motif are highlighted. Hr=hairless; Arab=Arabidopsis
Oncogene
5qNCA, a candidate for the chromosome 5 tumor suppressor gene
ZHuet al
6949
Page 4
which was established from a patient with MDS
(Steube et al., 1997). MUTZ-1 cells have a deletion
of 5q, but express a moderate level of normal 5qNCA
mRNA (see Figure 3b). The expression vector pTracer-
5qNCA, or the empty vector control pTracer, were
introduced into MUTZ-1 cells by electroporation.
Individual Zeocin-resistant clones demonstrating GFP
¯uorescence were picked, expanded, and further tested
for clonogenic eciency, comparing them to cells
transfected with empty vector. As shown in Figure 7,
constitutive expression of 5qNCA in MUTZ-1 cells in
this manner signi ®cantly inhibited clonogenic recovery
in the transfectants. Every MUTZ-1 subclone exam-
ined showe d a signi ®cant decrease in colony formation
varying from 30 to 80%, depending on the clone. Thus,
5qNCA has a growth suppressive eect, consistent with
what might be expected of a tumor suppressor gene.
Mutation analysis of 5qNCA by genomic sequencing of
AML and MDS samples
5qNCA was investigated for mutations by sequencing
the complete coding region of all 24 exons in genomic
DNA; three AML cell lines with chromosome 5
deletion, KG-1, HL-60, and MUTZ-1, (Fur ley et al.,
1986; Gallagher et al., 1979; Steube et al., 1997) along
with 11 clinical leukemia samples with 5q abnormal-
ities, were studied. Primer design was based on
intronic sequences and mutations were detected by
comparing the resulting sequence to the wild-type
5qNCA sequence. As shown in Table 2, four
conservative nucleotide dierences were found that
did not result in a change in the protein coding
sequence, and represent single nucleotide polymorph-
isms. However, an additional seq uence dierence was
found in the cell line KG-1, which resulted in an amino
acid substitution of threonine to alanine in codon 256
(ACG to GCG), which lies in exon 6, immediately
adjacent to the 5' region that is highly conserved
between 5qNCA and TSGA. This amino acid substitu-
tion was not observed in an additional 104 alleles from
normal individuals, suggesting this is a very rare
polymorphism or, more likely, a somatic mutation.
The presence of this sequence change adjacent to a
Figure 6 Subcellular localization of 5qNCA-His fusion protein.
COS-1 cells containing pTracer-5qNCA were immunostained with
an anti-His antibody, His-probe, and examined under a
¯uorescent microscope. (a) shows the morphology of the cells
under visible light, while (b) shows the same cells under UV with
wavelength of 630 nm
Figure 5 Expression of 5qNCA protein in COS-1 cells as
detected by Western blot. Each lane contains 100 mg of protein
isolated from the indicated COS-1 clones. After gel electrophor-
esis, the proteins were transferred to a ®lter, immunostained with
the anti-His antibody (His-Probe) and visualized by ECL. The
clones represented are COS-1 cells (lane 1), COS-1 cells with
empty vector (p-Tracer) (lane 2), COS-1 subline #8 containing p-
Tracer-5qNCA (lane 3), COS-1 subline #15 containing p-Tracer-
5qNCA (lane 4), and COS-1 subline #18 containing p-Tracer-
5qNCA (lane 5)
5qNCA, a candidate for the chromosome 5 tumor suppressor gene
ZHuet al
6950
Oncogene
Page 5
highly-conserved region suggests that it might impact
on the function of the protein; whether it represents a
mutational event contributing to cellular transforma-
tion will require further investigation. Overall, inacti-
vating mutations of 5qNCA appear to be unco mmon
in myeloid leukemias with chromosome 5 deletions.
Discussion
Here we report the complete sequence and preliminary
characterization of a novel gene, 5qNCA (LOC51780),
as a candidate for the 5q31 tumor suppress or gene.
5qNCA encodes a 191-kD protein that is localized
primarily to the nucleus, but its molecular function
remains unknown. The protein motifs that it contains
suggest an interesting biology. Almost half of the C-
terminal portion of the gene is highly conserved among
the plant and animal kingdoms, sharing a zinc ®nger,
nuclear localization signal, nuclear coactivator motif,
and a jmjC domain. All three domains are shared with
hairless, ENBP1, TRIP8, TSGA, and Drosophila
CG815; however, little is known about the molecular
biology of any of these proteins that might shed light
on the function of 5qNCA. Hairless is involved in the
dierentiation and apoptosis of hair follicles (Pante-
leyev et al., 1998; Cichon et al., 1998; Thompson,
1996). ENBP1 is involved in plant nodule development
(Christiansen et al., 1996). TRIP8 is a suspected
thyroid-hormone receptor co-activato r (Lee et al.,
1995). There are no known functions for TSGA or
Drosophila CG8165, though TSGA protein contains a
zinc ®nger that previously has been shown to mediate
binding to nucleic acids (Hoog et al., 1991). Recently,
characterization of the jmjC domain revealed that it is
shared by over 145 proteins, including many that are
transcription factors or are otherwise associated with
chromatin (Clissold and Ponting, 2001). The domain is
found in metalloenzymes that adopt the cupin fold and
regulate chromatin remodel ing. As 5qNCA appears to
be primarily a nuclear protein with a single zinc ®nger,
one might speculate that it functions as a nuclear co-
activator, binding other proteins that regulate cell
growth or dierentiation. This gene, like the others
with which it shares homology, represents an interest-
ing and relatively unexplored category of nuclear
receptor co-activators.
The expression of a full-length clone of 5qNCA in
COS-1 cells con®rmed that the protein is expressed
primarily from the 5.2-kb open reading frame encoding
a 191-kD protein, rather than the predicted 154-kD
product of AF25103 (Lai et al., 2000). The initial
report of AF25103 may contain a sequence error,
resulting in a frame shift that would lead to a smaller
protein, leaving almost 1000 nucleotides untranslated,
including 5' sequences that are homologous with
TSGA, as well as the KG-1 mutation in exon 6. Our
report of the complete cDNA and protein coding of
5qNCA underscores the limitations of the existing
databases, particularly with regard to sequence errors,
and reinforces the necessity of obtaining and con®rm-
ing cDNA sequence from biological sources .
The relatively high level of expression of 5qNCA in
leukemia cell lines and in normal bone marrow cells
suggests that it may play an important role in the
normal function of this tissue; thus, its disruption may
lead to cellular transformation. To test this possibility,
5qNCA was expressed in a leukemic cell line with
del(5q), resulting in signi®cantly reduced clonogenic
recovery of the cells. These results suggest that
constitutive expression of the protein, or correction of
loss of one allele, ha s growth suppressive activity; thus,
the activity is consistent with that of a tumor
suppressor gene, and further investigation is warranted
Figure 7 Eect of 5qNCA-expression on clonogenic eciency of
MUTZ-1 cells. Subclone name refers to the expression vector
introduced, followed by the number of a randomly-selected isolate
indicated by #; MUTZ1-pTracer refers to the transfectants with
empty vector. Each subclone was plated in a 96-well plate in
quadruplicate and each well contains 200 cells in 100 m l of 0.8%
methylcellulose growth medium with 10% FCS. Each experiment
was repeated twice. The numbers of clonogenic formation in
MUTZ1-pTracer, MUTZ1-5qNCA #4, #6, #15 and #18 are
78.8+3.5, 14.4+2.6, 14.6+2.5, 27.5+2.4 and 49.5+2.7, respec-
tively. Compared to the MUTZ-1 cells transfected with the empty
pTracer vector (100% clonogenic recovery), the clonogenic cell
recovery in MUTZ1-5qNCA transfectants #4, #6, #15 and #18
were reduced by 79.6, 79.4, 62.2 and 30.0%, respectively
Table 2 Polymorphic sequence variants of 5qNCA in three cell lines
and 11 clinical samples with 75 or del(5q)
Codon Amino DNA
a
Cell lines
b
Patient
number acid sequence MUTZ-1 KG-1 HL-60 samples
c
88 Ile ATT X X 7/11
Ile AT
C X 4/11
238 Glu GA
G X X X 7/11
Glu GA
A 4/11
952 Val GT
G 2/11
Val GT
A X X X 9/11
1163 Gly GG
T X ND 3/10
Gly GG
C ND X 7/10
Summary of polymorphic sequence variants of 5qNCA in three AML
cell lines and in 11 clinical leukemia samples with 5q abnormalities.
a
Alternative DNA sequence at the indicated codon.
b
`X' indicates
which sequence is detected in the indicated cell line.
c
Indicates the
number of clinical samples with the indicated sequence variant, out
of the total number sequenced. ND=not done
Oncogene
5qNCA, a candidate for the chromosome 5 tumor suppressor gene
ZHuet al
6951
Page 6
to de termine if the gene is mutated or inactivated in
leukemic cell lines.
In a limit ed screen for 5qNCA mutations, a sequence
change near a highly-conserved region was found in
the KG-1 cell line resulting in an amino acid
substitution. The signi®cance of this missense mutation
is unknown. Because KG-1 cells have already lost one
chromosome 5 allele, the presence of a sequence
change in the residual allele may contribute to
transformation, if it results in a change in protein
function; studies are underway to evaluate the
functional eect of this codon change. The survey
revealed many sequence polymorphisms in the other
cases examined, but no nonsense or missense mutations
other than that found in KG-1. We believe that it is
worthwhile to search for additional mutations in more
clinical AML and MDS samples, to determine if
5qNCA is a frequent target of mutational inactivation
and contributes to the initiation or progression of
malignant myeloid disorders.
The heterogeneity of leukemia deletion intervals
along 5q suggests that there may be multiple tumor
suppressor genes on this chromosom e, of which
5qNCA may be one. The gene is located within the
deletion interval reported by Zhao et al. (1997), and is
very close to the marker D5S500, at the boundary of
an adjacent interval reported by Horrigan et al. (1999).
There are other cancer-related deletions on the long
arm of chromosome 5 which might contain tumor
suppressor genes; these include an interval at 5q32
which is thought to contain a gene for the more
indolent 5q- MDS syndrome (Westbrook and Le Beau,
1993; Nagarajan, 1995; Fairman et al., 1995); a
segment at 5q31.1 containing IRF-1 (Willman et al.,
1993); a site of frequent LOH in myeloid leukemias,
hairy cell leukemia at 5q13 (Fairman et al., 1996) and
the APC colon cancer gene located on 5q21-22 (Mao et
al., 1998). Thus, it is possible that many clinical
samples with deletions target a varie ty of loci on 5q31,
so that only a minority of cases will show mutations in
a particular gene.
In summary, 5qNCA is a good candidate for a 5q
tumor suppressor gene based on its predicted function,
its suppressive activity, its chromosomal location, and
the ®nding of a mutant allele in the KG-1 cell line.
Detailed functional studies of 5qNCA, additional
sequencing in clinical sample, and the eects of the
sequence change found in KG-1, will be required to
determine if this gene plays a role as a tumor
suppressor gene or as a partner of a known tumor
suppressor gene.
Materials and methods
Cell lines and clinical specimens
Cell lines used in this study were obtained from ATCC
(Rockville, MD, USA) and from DSMZ (Brunschweig,
Germany). Cells were cultured using recommended condi-
tions or in RPMI-1640 medium (GIBCO BRL, Grand Island,
NY, USA) supplemented with 10% heat-inactivated fetal calf
serum (Sigma, St. Louis, USA) at 378C in an atmosphere of
5% CO
2
in air. CD34
+
cells were isolated from bone marrow
obtained from normal donors with informed consent under
the guidelines provided by the University of Illinois. The
CD34
+
cells were isolated and puri®ed as previously
described (Brandt et al., 1999).
Clinical specimens from MDS and AML patients contain-
ing del(5q) or 75 were obtained from patients during routine
clinical treatment according to institutional IRB guidelines;
all samples used were analysed for chromosome 5 loss of
heterozygosity, and have been reported previously (Horrigan
et al., 1996; Westbrook et al., 2000).
PCR and reverse-transcriptase PCR (RT ± PCR)
DNA was extracted from cells using standard procedures and
PCR was performed by using intron-based primers for direct
genomic sequencing, as described previously (Xie et al.,
2000). RT ± PCR was used for exon connection of 5qNCA
and for cDNA cloning. RNA used for RT ± PCR was isolated
from human bone marrow CD34
+
cells and from leukemia
cell lines using TRIzol Reagent (Gibco Technologies,
Gaithersburg, MD, USA) as described previously (Xie et
al., 2000). Exon-connection RT ± PCR was performed using
the following primers: S1 (5'-aggcgggccccggcgatggcg-3')+AS2
(5'-ttggactggctccttccacttcc-3'), S2 (5'-atggtgttctagccacagagaac-
3')+AS2 (5'-tctatcccttctgctagggagg-3'), S3 (5 '-gacctcaaaatac-
tacctggcc-3')+AS3 (5'-ggaccctgccagtttggattc-3'), and S4 (5'-
gcggctgctgctgctgttcg-3')+AS4(5'-atgggctgagtgaggctgctggtg-3').
Northern blot analysis
Twenty mg of total cellular RNA or 3 mg of mRNA from each
cell line was separated on a 1.0% formaldehyde agarose gel and
transferred to Hybond-N nylon ®lters (Amersham, Arlington
Heights, IL, USA), which were subsequently cross-linked with
ultraviolet light. After prehybridization, the blot was hybri-
dized to a-
32
P-dCTP-labeled cDNA probe, and washed
subsequently with 26SSC+0.1% SDS twice, followed by
16SSC+0.1% SDS twice, and 0.16SSC+0.1% SDS twice.
The blots were exposed to X-ray ®lms with an intensifying
screen at 7808C. A multiple tissue Northern blot, MTN H
(Clontech, Palo Alto, CA, USA), was used to detect tissue-
speci®c expression of 5qNCA. The hybridization probe was
prepared from a 1360 bp fragment of human 5qNCA, which
was ampli®ed by RT ± PCR using primers ¯anking nucleotides
5120 ± 6480, and the probe was labeled with a-
32
P-dCTP by
random primer labeling using Multiprime DNA Labeling
System (Amersham, Arlington Heights, IL, USA).
DNA sequencing
The DNA sequencing reaction was performed using the ABI
Prism BigDye Terminator Sequencing kit (Perkin-Elmer). In
brief, PCR products ampli®ed from each exon were puri®ed
using Qiagen PCR puri®cation kit (Qiagen, Valencia, CA,
USA). A typical 20 ml reaction containing 50 ± 90 ng DNA
template from puri®ed PCR products, 0.32 m
M of forward or
reverse primer, and 8 ml of BigDye terminate mixture was
reampli®ed as follows: 24 cycles at 968C for 30 s, annealing at
50 ± 588C (optimized for each primer pair) for 15 s, and 608C
for 4 min. The reactions were puri®ed and then sequenced
using an ABI prism 377 DNA sequencer (Perkin-Elmer,
Norwalk, CT, USA). Analysis of the sequences was
performed using SeqMan, a DNASTAR software program
(DNASTAR Inc. Madison, WI, USA).
5qNCA, a candidate for the chromosome 5 tumor suppressor gene
ZHuet al
6952
Oncogene
Page 7
Expression of 5qNCA
A cDNA clone of 5qNCA containing the entire 5.2-kb open
reading frame was prepared by RT ± PCR from a human
placenta cDNA library (Clontech, Palo Alto, CA, USA),
con®rmed by sequencing and inserted into the expression
vector pTracer (Invitrogen, Carlsbad, CA, USA). cDNA was
inserted into the EcoRV+NotI sites in frame with a 66His-
Tag. To transfer the expression vector into mammalian cells,
10 mg of puri®ed pTracer-5qNCA plasmid was transfected in
to COS-1 cells by Lipofectin (GIBCO, Gaithersburg MD,
USA). Forty-eight hours after transfection, cells were plated
in 32-mm dishes with growth medium containing Zeocin for
clone selection. Resistant clones became apparent after 2 ± 3
weeks and were examined under a ¯uorescent microscope.
The colonies expressing GFP were then plated in 32-mm
dishes containing Zeocin for subcloning. The subclones were
further con®rmed for GFP expression by ¯uorescence
microscopy, and positive clones were further expanded for
additional study.
To evaluate whether 5qNCA has growth suppressive
properties, the expression vector pTracer-5qNCA was
introduced into MUTZ-1 cells, an MDS cell line with 5q
deletion, by electroporation at 250 V and 950 mFD. 24 h
after electroporation, the transfectants were selected in
800 mg/ml of Zeocin for clone selection, followed by isolation
of individual positive clones in 0.8% methylcellulose. Zeocin-
resistant clones, which contain GFP ¯uorescence were picked,
expanded, and analysed by ¯ow ¯uorescence-activated ¯ow
cytometry (FACS) and by Western blot to con®rm the
expression of 5qNCA-His protein. The selected positive
clones were further tested for clonogenic eciency by plating
the cells in 0.8% methylcellulose culture.
Western blot
For protein extraction, 5610
6
COS-1 cells were suspended in
lysis buer for 1 h at 48C with freshly added protease
inhibitors. The nuclei were removed by centrifugation. One
hundred mg of protein from each sample was solubilized
with SDS-polyacrylamide gel electrophoresis (SDS ± PAGE)
sample buer, and electrophoresed through 8% SDS-
polyacrylamide gel for about 2 h at room temperature. For
immunobloting, the proteins separated by SDS ± PAGE were
electrotransferred onto nitrocellulose ®lters (Amersham,
Arlington Heights, IL, USA). The ®lters with the proteins
were incubated with primary antibody (His-Probe) against
histine (Santa Cruz, CA, USA) for 2 h and were then
incubated with HRP-labeled secondary antibodies for an-
other hour and subsequently developed using ECL detection
systems (Amersham, Arlington Heights, IL, USA).
Subcellular localization of 5qNCA-His fusion protein
COS-1 clones expressing GFP were grown on coverslides in
tissue culture plates for 24 h. The coverslides were subse-
quently washed three times with PBS, followed by ®xation in
4% formaldehyde in PBS for 20 min and then in pre-cold
acetone for 30 s. After blocking by 1% BSA in PBS for
20 min, the slides were incubated with primary antibody His-
probe (Santa Cruz, CA, USA) for 3 h, followed by
incubation with secondary antibody labeled with TRITC
(Sigma, St. Louis, USA) for 1 h. The slides were washed with
PBS, and the cellular localization of 5qNCA-His fusion
protein was evaluated under a ¯uorescent microscope with a
630 nm wavelength.
Clonogenic Assay
The clonogenic assay was performed to evaluate the eect of
5qNCA on clonogenic recovery of leukemia cells as
previously described (Hu et al., 2000). Brie¯y, clonogenic
cells were detected by their capacity to give rise to colonies
when plated in 0.8% methylcellulose with RPMI-1640 plus
10% FCS (growth medium) in a micro-well plate (Linbro,
Flow Lab, Mclean, VA, USA), after plating 200 cells/well.
Colonies containing more than 20 cells were counted after 5 ±
7 days using an inverted microscope.
Acknowledgments
This work has been s upported by PHS grants RO1-CA-
72593 and PO1-CA-75606.
References
Bartoloni L, Horrigan SK, Viles KD, Gilchrist JM, Stajich
JM, Vance JM, Yamaoka LH, Pericak-Vance MA,
Westbrook CA and Speer MC. (1998). Genomics, 54,
250 ± 255.
Brandt JE, B artholomew AM, Fortman JD, Ne lson MC,
Bruno E, Chen LM, Turian JV, D avis TA, Chute JP and
Homan R. (1999). Blood, 94, 106 ± 113.
Christiansen H, Hansen AC, Vijn I, Pallisgaard N, Larsen K,
Yang WC, Bisseling T, Marcker KA and Jensen EO.
(1996). Plant Mol. Biol., 32, 809 ± 8 21.
Cichon S, Anker M, Vogt IR, Rohleder H, Putzstuck M,
Hillmer A, Farooq SA, Al-Dhafri KS, Ahmad M, Haque
S, Rietschel M, Propping P, Kruse R a nd Nothen MM.
(1998). Human Mol. Genetics, 7, 1671 ± 1679.
Clissold PM and Ponting CP. (2001). JmjC: cupin metal-
loenzyme-like domains in jumonji, hairless and phospho-
lipase A2beta. Trends Biochem. Sci., 26, 7±9.
Fairman J, Chumakov I, Chinault AC, Nowell PC and
Nagarajan L. (1995). Pro c. Natl. Acad. Sci. USA, 92,
7406 ± 7410.
Fairman J, Wang RY, Liang H, Zhao L, Saltman D, Liang
JC and Nagarajan L. (1996). Blood, 88 , 22 59 ± 2266.
Feanaux P, More l P and Lai JL. (1996). Semin. Hematol., 33,
127 ± 138.
FrazerKA,UedaY,ZhuY,GiordVR,GarofaloMR,
Mohandas N, Ma rtin CH, Palazz olo MJ, C heng JF and
Rubin EM. (1 997). Genome Res., 7, 495 ± 512.
Furley AJ, Reeves BR, Mizutani S, Altass LJ, Watt SM,
JacobMC,vandenElsenP,TerhorstCandGreavesMF.
(1986). Blood, 68, 1101 ± 1 107.
Gallagher R, Collins S, Trujillo J, McCredie K, Ahearn M,
Tsai S, Metzgar R, Aulakh G, Ting R, Ruscetti F and
Gallo R. (1979). Blood, 54, 713 ± 733.
Hoog C, Schalling M, Grunder-Brundell E and Daneholt B.
(1991). J. Reproduc. Dev., 30, 173 ± 18 1.
Horrigan SK, Bartoloni L, Speer MC, Fulton N, Kravarusic
J, Ramesar R, Vance J M, Yamaoka LH and Westbrook
CA. (1999). Genomics, 57, 24 ± 35.
Horrigan SK, Westbroo k CA, Kim AH , Baner jee M, Stock
W and