Am J Physiol Gastrointest Liver Physiol
280: G58–G67, 2001.
GATA family transcription factors activate lactase
gene promoter in intestinal Caco-2 cells
RIXUN FANG, LYNNE C. OLDS, NILDA A. SANTIAGO, AND ERIC SIBLEY
Department of Pediatrics, Stanford University Medical Center, Stanford, California 94305
Received 27 December 1999; accepted in final form 12 August 2000
GATA-4; GATA-5; GATA-6; enterocyte
is comprised of four principal
cell types that are derived from a proliferating stem
cell population (see Ref. 10 for review). The stem cells,
located in crypts near the intestinal villus base, undergo terminal differentiation as they migrate from the
crypt to the villus tip, where they ultimately die and
are sloughed off into the gut lumen. The four principal
intestinal epithelial cell types have highly specialized
functions. The absorptive enterocytes are the predominant intestinal cell type, and they function to digest
and absorb luminal nutrients across the apical brushborder membrane. Goblet cells secrete mucus that provides a protective barrier lining the gut. The enteroendocrine cells secrete intestinal hormones involved in
signaling gut motility. Paneth cells terminally differentiate during migration toward the base of the crypt
and are lysozyme- and defensin-producing cells. The
THE INTESTINAL EPITHELIUM
Address for reprint requests and other correspondence: Eric Sibley,
Dept. of Pediatrics, Stanford Univ. Medical Center, 750 Welch Rd.,
Suite 116, Palo Alto, CA 94304 (E-mail: [email protected]).
G58
molecular mechanisms regulating terminal differentiation of the intestinal cell types have not been defined.
The GATA family of zinc finger transcription factors
are important regulators of cell lineage differentiation
during vertebrate development. Six GATA family
members have been identified in vertebrate species.
Each vertebrate GATA factor contains two conserved
zinc fingers that bind to a consensus sequence (WGATAR) that is present in the transcriptional regulatory
region of multiple lineage-specific genes (13, 17). The
GATA-1, -2, and -3 subfamily proteins control critical
steps in erythroid and lymphoid development (8, 22,
23, 29). GATA-4, -5, and -6 subfamily proteins are
expressed in overlapping patterns in the developing
heart and endoderm-derived organs of the gastrointestinal tract including the stomach, intestine, liver, and
pancreas (1, 5, 16, 19, 20). The functional role of
GATA-4, -5, and -6 proteins in regulating heart and gut
development is largely unknown. Homozygous disruption of GATA-4 expression during embryogenesis results in the lack of a primitive heart tube and foregut in
mice (15). Targeted mutagenesis of the GATA-4 gene in
embryonic stem cells disrupts visceral endoderm differentiation (25). GATA binding sites have been identified in transcriptional regulatory elements of several
cardiac-specific target genes. Overexpression of
GATA-4 protein has been shown to transactivate these
cardiac-specific cis elements in cell culture (11, 12, 18).
With respect to a role in regulating enterocyte differentiation, distinct patterns of expression for
GATA-4, -5, and -6 were recently described by Gao et
al. (9). The expression patterns of the GATA factors
differ along the proximal-distal villus axis in the
chicken intestine and during differentiation in culture.
In situ transcript levels for GATA-6 are highest in the
proliferating cell region of the crypts, whereas levels
for GATA-4 and GATA-5 increase toward the villus tip.
On stimulation to differentiate, intestinal HT-29 cells
begin to express GATA-5 whereas GATA-6 transcript
levels decline. Few intestinal target genes, however,
have been identified for the GATA-4/5/6 factor subfamily. In the same report by Gao et al. (9), the gene
encoding Xenopus intestinal fatty acid binding protein
(IFABP) was identified as an in vitro target for the
The costs of publication of this article were defrayed in part by the
payment of page charges. The article must therefore be hereby
marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
0193-1857/01 $5.00 Copyright © 2001 the American Physiological Society
http://www.ajpgi.org
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
Fang, Rixun, Lynne C. Olds, Nilda A. Santiago, and
Eric Sibley. GATA family transcription factors activate lactase gene promoter in intestinal Caco-2 cells. Am J Physiol
Gastrointest Liver Physiol 280: G58–G67, 2001.—The GATA
family of transcription factors regulate tissue-specific patterns of gene expression during development. We have characterized the interaction between GATA proteins and the
lactase gene promoter. Nuclear protein bound to the lactase
gene GATA region cis element (⫺97 to ⫺73) was analyzed by
electrophoretic mobility shift assays (EMSA) and supershift
assays with GATA antibodies. Lactase promoter activities
were assayed in Caco-2 cells transfected with wild-type and
mutated luciferase promoter-reporter constructs and GATA4/5/6 expression constructs. EMSA with the GATA region
probe yields a specific DNA-protein complex that requires the
GATA factor binding site WGATAR. The complex is recognized by GATA-4- and GATA-6-specific antibodies. GATA-4/
5/6 expression constructs are able to activate transcription
driven by the wild-type promoter, but not by a promoter in
which the GATA binding site is mutated, in Caco-2 and
nonintestinal QT6 cells. GATA factor binding to the lactase
cis element correlates with functional promoter activation.
We conclude that each of the GATA family zinc finger proteins expressed in the intestine, GATA-4, -5, and -6, can
interact with the lactase promoter GATA element and can
function to activate the promoter in Caco-2 cells.
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
ward oligonucleotides corresponding to nt ⫺100 to ⫺80, nt
⫺73 to ⫺53, or nt ⫺58 to ⫺38 (each with an added 5⬘ Bgl II
site) and GL primer 2 (Promega), the reverse oligonucleotide
proximal to the luciferase gene in pGL3-Basic. The Bgl IIHind III fragments for each of the PCR products were then
cloned into pGL3-Basic to generate the corresponding ⫺100,
⫺73, and ⫺58 bp deletion promoter-reporter constructs.
To clone mutant lactase promoter-reporter constructs, a
fragment of the wild-type pgLac construct was amplified by
PCR using a forward mutant GATA region oligonucleotide
(Mut-A, Mut-B, Mut-C, or Mut-D, sequences shown in Fig. 1)
and GL primer 2. The mutant PCR product and a forward
lactase gene oligonucleotide corresponding to nt ⫺265 to
⫺240 were used in a second PCR reaction to amplify a 0.3-kb
segment of the pgLac template. The internal 0.2-kb Bgl
II-Xho I fragment of the second PCR product was cloned into
pGL3-Basic to generate pMutA, pMutB, pMutC, and pMutD.
Incorporation of the correct deletion and mutated base pairs
was confirmed by sequencing.
Electrophoretic mobility shift assay and supershift assay.
Nuclear extracts were prepared from Caco-2 or QT6 cells
according to a modification of a previously described procedure (2). Cells were harvested by scraping and resuspended
in 5 vol of phosphate-buffered saline followed by centrifugation (400 g for 10 min at 4°C). The cells were resuspended in
five packed cell volumes of cold buffer A [in mM: 10 HEPES
(pH 7.9), 10 KCl, 1.5 MgCl2, 0.5 dithiothreitol, and 1 phenylmethylsulfonyl fluoride (PMSF)] and incubated for 10 min on
ice followed by centrifugation as above. Cells were resuspended in two packed cell volumes of buffer A, lysed by
Dounce homogenization, and then centrifuged (400 g for 20
min at 4°C). The nuclei pellet was resuspended in 3 ml of cold
buffer B [20 mM HEPES (pH 7.9), 10% glycerol, 0.42 M NaCl,
1.5 mM MgCl2, 0.5 mM dithiothreitol, 0.2 mM EDTA, and 1
mM PMSF] followed by Dounce homogenization and stirred
gently for 30 min at 4°C. After centrifugation (100,000 g for
20 min), the supernatant was precipitated with ammonium
sulfate (final concentration 0.33 g/ml) and then centrifuged
(25,000 g for 20 min). The pellet was resuspended in buffer C
[20 mM HEPES (pH 7.9), 10% glycerol, 0.5 mM dithiothreitol, 0.2 mM EDTA, and 1 mM PMSF], dialyzed against buffer
C for 12–18 h at 4°C, and stored at ⫺80°C.
Nuclear extracts were incubated with a radiolabeled
GATA region binding site probe for gel shift assay. The
MATERIALS AND METHODS
Materials and reagents. Restriction endonucleases and
modifying enzymes were purchased from New England Biolabs (Beverly, MA). Radioisotopes were purchased from DuPont NEN. Oligonucleotides were synthesized in the Protein
and Nucleic Acid facility of the Stanford University Beckman
Center.
Subcloning of deletion and mutant lactase promoter-reporter constructs. The lactase promoter deletion and mutant
reporter constructs were generated from pgLac (6), a previously described plasmid in which a 200-bp 5⬘-flanking region
of the rat lactase gene (nt ⫺200 to ⫹13) was cloned upstream
of the firefly luciferase reporter gene in the vector pGL3Basic (Promega). Deletion promoter-reporter constructs were
generated by initial PCR amplification of pgLac using for-
Fig. 1. Sequence of wild-type and mutated GATA region cis element
of the rat lactase gene promoter used in this study. The sequence of
the wild-type GATA region cis element of the rat lactase promoter is
shown at top. Double-stranded mutations corresponding to bases in
boldface were incorporated into synthetic oligonucleotides for gel
shift analysis (see Fig. 3, MutA, -B, -C, and -D) and into site-directed
mutagenesis reporter constructs (see Figs. 5 and 6, pMutA, pMutB,
pMutC, and pMutD). The GATA protein consensus DNA binding site
is underlined.
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
GATA-4/5/6 factors in intestinal cell culture. In addition, Fitzgerald et al. (7) recently reported that
GATA-6 stimulates promoter activity of the human
lactase gene promoter. We report here that GATA-4
and GATA-5, in addition to GATA-6, can activate the
rat lactase promoter in intestinal cell culture. These
studies provide compelling evidence for the potential of
each member of the GATA-4/5/6 factor subfamily to
regulate expression of multiple enterocyte-specific
genes.
Intestinal lactase-phlorizin hydrolase (LPH, lactase)
is the absorptive enterocyte membrane glycoprotein
essential for digestive hydrolysis of lactose in milk.
Lactase is present predominantly along the brushborder membrane of differentiated enterocytes lining
the villi of the small intestine. Expression of the lactase
gene is spatially restricted along both vertical and
longitudinal axes in the gut (24). Along the vertical
axis, immature enterocytes derived from stem cells in
the crypts migrate from the crypt to the villus tip,
differentiate, and begin to express the enzymatic activity of lactase as well as other digestive hydrolases.
Along the longitudinal axis, the lactase gene is expressed maximally in the proximal small intestine and
declines significantly in the distal segments of the
intestine. Lactase gene expression is also temporally
restricted in the gut during intestinal maturation. Enzyme activity is maximal in the small intestine of
preweaned mammals and declines markedly during
maturation (3, 14, 21). The mechanisms regulating the
spatial restriction and maturational decline in lactase
activity have not been fully defined. In vitro binding
studies showed that the lactase gene promoter interacts with specific nuclear proteins from intestinal cells
(28). The homeodomain protein Cdx-2 binds to a distinct cis element, CE-LPH1, of the lactase 5⬘-flanking
region and is capable of activating transcription of the
lactase promoter (6, 27). Regulation of intestine-specific differentiation is likely to involve additional transcription factors that control expression of terminal
differentiation genes such as LPH. Here we identify a
GATA binding site element in the promoter region of
the lactase gene, identify the GATA proteins interacting with the element, and characterize transcriptional
activation mediated by GATA-4, -5, and -6 proteins in
intestinal Caco-2 cell culture.
G59
G60
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
tions were then carried out with synthetic oligonucleotide
primers corresponding to exon sequences for the human
GATA-4 (Ref. 30; GenBank accession no. D78260), GATA-6
(Ref. 26; accession no. NM005257), and glyceraldehyde-3phosphate dehydrogenase (GAPDH) (internal standard; Ref.
4; accession no. J04038) genes: GATA-4F 5⬘-CATCAAGACGGAGCCTGGCC-3⬘ and GATA-4R 5⬘-TGACTGTCGGCCAAGACCAG-3⬘ (218-bp product); GATA-6F 5⬘-CCATGACTCCAACTTCCACC-3⬘ and GATA-6R 5⬘-ACGGAGGACGTGACTTCGGC-3⬘ (213-bp product); and GAPDH-F
5⬘-GGGTCATCATCTCTGCCCCCTCTG-3⬘ and GAPDH-R
5⬘-CCATCCACAGTCTTCTGGTGGCA-3⬘ (208-bp product).
To prevent nonspecific amplification hot-start PCR reactions
(50 ␮l) were carried out with 5 ␮l of the reverse-transcribed
cDNA, 0.4 ␮M primers, 0.20 mM dNTPs, 20 mM Tris 䡠 HCl
(pH 8.4), 50 mM KCl, 1.5 mM MgCl2, and 2.5 units of Taq
polymerase (GIBCO BRL) preincubated with 1:28 with
TaqStart Antibody (Clontech). Control amplifications confirmed that the reactions were entirely dependent on the RT
reaction. Amplification conditions were 94°C for 45 s, 55°C
for 45 s, and 75°C for 45 s performed for 15–40 cycles to
optimize for PCR products in the linear range of exponential
amplification. Simultaneous PCR reactions were performed
with first-strand cDNA generated from RNA of different
human tissues (MTC; Clontech). PCR products were analyzed after electrophoresis on 2% agarose gels using a Molecular Analyst densitometer (Bio-Rad) and sequenced to confirm their identity.
RESULTS
Regulatory regions of lactase gene 5⬘-flanking DNA
mapped by deletional analysis of reporter constructs.
Promoter activity was previously mapped to within
⫺200 bp upstream from the transcription start site of
the rat lactase gene (6). To identify regions of the
lactase gene capable of mediating regulation of gene
transcription, deletion fragments of this promoter region were cloned upstream of the firefly luciferase
cDNA in the reporter plasmid pGL3-Basic and transfected into Caco-2 cells. The ⫺100, ⫺73, and ⫺58 bp
lactase reporter constructs, shown in Fig. 2A, were
transiently transfected into Caco-2 cells, a human adenocarcinoma-derived cell line that mimics a small
intestinal enterocyte phenotype with respect to expression of several digestive hydrolases including lactase
and sucrase-isomaltase. Caco-2 cell extracts were assayed for relative luciferase activity 48 h after transfection.
The ⫺100 bp flanking region reporter construct directed maximal transcriptional activity with 12-fold
higher relative luciferase expression compared with
the promoterless pGL3-Basic. The transcriptional activity driven by the ⫺73 and ⫺58 bp promoter fragments was significantly reduced compared with the
⫺100 bp fragment construct. A positive cis element
maps to the region between ⫺73 and ⫺100 bp of the
lactase gene promoter. To determine whether the activity mediated by the lactase promoter fragment is cell
type specific, the nonintestinal quail fibroblast QT6 cell
line was similarly transfected with the lactase promoter-reporter constructs. QT6 cells possess no lactase
activity, and, as expected, there was no increase in
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
GATA region probe consisted of a double-stranded 25-nt
oligonucleotide (5⬘-TATCCTAGATAACCCAGTTAAATA-3⬘)
from ⫺96 to ⫺73 relative to the start site of transcription of
the rat lactase gene (21). The Sp1 binding site probe consisted of a double-stranded 28-nt oligonucleotide (5⬘GATCGGGGCGGGGCGGGGCGGGGCGATC-3⬘). The sequence of the mutant oligonucleotides (Mut-A, Mut-B,
Mut-C, and Mut-D) is shown in Fig. 1. The Klenow fragmentlabeled probes were generated by annealing single-stranded
oligonucleotides to yield a four-base 5⬘ overhang on each end
followed by fill-in incorporation with nucleotide triphosphates, including [32P]dTTP (3,000 Ci/mM). Unincorporated
nucleotides were removed by purification with a Bio-Spin
P30 column (Bio-Rad). Nuclear extract (15 ␮g) was incubated
in the presence of 40,000 cpm labeled probe (105 cpm/ng) with
0.5 ␮g poly dI:dC with or without 100-fold excess unlabeled
probe in 5 mM HEPES (pH 7.9), 10% glycerol, 25 mM KCl,
0.05 mM EDTA, and 0.125 mM PMSF for 20 min at room
temperature. DNA-protein complexes were resolved on a 6%
Tris-glycine polyacrylamide gel that was then dried and
exposed to X-ray film for autoradiograph detection. Gel supershifts were carried out with the Caco-2 or QT6 nuclear
extracts and the gel shift method detailed above with the
exception that the reaction mixture was incubated for 30 min
at room temperature in the presence of 1 ␮l of a polyclonal
GATA-4 or GATA-6 antibody (Santa Cruz Biotechnology).
Transient transfection assays. Caco-2 cells or QT6 cells
were cultured in DMEM with 10% fetal bovine serum. Fortyeight to sixty hours before transfection, the cells were split
and 35-mm dishes were seeded with 2 ⫻ 105 cells. For each
reporter construct, a DNA transfection mixture was prepared
consisting of 0.6 ␮M of the construct, 0.5 ␮g of pRL-CMV
(Promega) as an internal control, and pBluescript KS⫹ II to
adjust to 3.5 ␮g of total DNA. The individual DNA mixtures
were transfected into cells (50–80% confluent) with Lipofectamine reagent (BRL) according to the protocol of the
manufacturer. For cotransfection experiments, 0.3 pmol of
pxGATA-4, -5, or -6 expression constructs (9), in which Xenopus GATA cDNAs were cloned downstream of a cytomegalovirus (CMV) promoter in pCDNA3, was transfected along
with 0.6 pmol of the luciferase reporter constructs. Cells were
harvested 48 h after transfection (70–90% confluent), and
luciferase activity was measured by the Dual-Luciferase reporter assay system (Promega) as described by the manufacturer, in a Monolight 3010 luminometer. Transfection with
the dual reporters (firefly luciferase for the lactase promoterreporter plasmids and renilla luciferase for the pRL-CMV
control) allowed for simultaneous expression and measurement of both reporter enzymes. The lactase promoter-reporter activity correlated with the effect of the promoter
sequence or GATA expression, and the activity of the cotransfected pRL-CMV provided an internal control. Experimental
lactase promoter-reporter activities were normalized to the
activity of the pRL-CMV internal control and expressed as
relative luciferase activity (means ⫾ SD, n ⫽ 3), thereby
minimizing experimental variability caused by differences in
cell viability or transfection efficiency.
RNA analysis by RT-PCR. Total RNA was extracted from
proliferating Caco-2 cells using Tri reagent (Molecular Research Center) according to the protocol of the manufacturer.
RNA concentrations were determined by optical densitometry at 260 nm, and the absence of RNA degradation was
confirmed by agarose gel electrophoresis. For RT-PCR, cDNA
was initially synthesized from 1.0 ␮g of Caco-2 cell total RNA
using avian myeloblastosis virus RT and the Advantage
RT-for-PCR kit (Clontech) according to the protocol of the
supplier and brought to a final volume of 100 ␮l. PCR reac-
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
G61
relative luciferase activity for any of the reporter constructs compared with pGL3-Basic (Fig. 2B).
Specific nuclear protein(s) from Caco-2 cells interacts
with a GATA region element. Inspection of the sequence for the ⫺73 to ⫺100 bp positive cis element
region of the lactase promoter reveals a consensus
binding site for the GATA family of zinc finger proteins. To identify interactions between this GATA region cis element and nuclear proteins in Caco-2 cells,
we used the electrophoretic mobility shift assay
(EMSA) or the gel shift assay. The wild-type GATA
region oligonucleotide was radiolabeled, incubated in
the presence of Caco-2 cell nuclear extract, and then
migrated through a 6% nondenaturing acrylamide gel.
The autoradiograph in Fig. 3 reveals the position of the
rapidly migrating unbound probe at the base of the gel
and a DNA-protein complex of slower mobility that is
formed after incubation with Caco-2 nuclear extract.
The DNA-protein complex is not competed away by
100-fold excess unrelated unlabeled SP1 oligonucleotide but is competed for by 100-fold excess unlabeled
wild-type GATA region oligonucleotide. A slightly
faster migrating nonspecific band is not displaced with
either competitor. The nuclear protein bound to the
Fig. 3. Electrophoretic mobility shift assay with Caco-2 cell nuclear
extract and GATA region radiolabeled probe. Caco-2 cell nuclear
extract protein (10 ␮g) was incubated with 32P-labeled GATA region
double-stranded oligonucleotide probe alone (⫺) or in the presence of
100⫻ unlabeled specific wild-type (WT), nonspecific (Sp1), or mutant
(MutA, MutB, MutC, MutD; sequences shown in Fig. 1) oligonucleotides. Samples were loaded on a 6% acrylamide gel. NS, nonspecific.
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
Fig. 2. Relative luciferase activity of intestinal cells transfected with
lactase promoter deletions cloned upstream of a reporter gene. Proliferating Caco-2 cells (A) or QT6 cells (B) were transfected with
promoterless pGL3-Basic and with lactase gene-luciferase reporter
constructs that contain the ⫺100 kb, ⫺73 kb, and ⫺58 bp fragments
of the lactase 5⬘-flanking region cloned upstream of a firefly luciferase reporter gene in pGL3-Basic (inset in A). Transfection efficiencies were normalized to renilla luciferase expression of a cotransfected pRL-CMV vector and expressed as relative luciferase activity
(means ⫾ SD: n ⫽ 3).
GATA region probe in Caco-2 cells therefore represents
a specific DNA-protein interaction.
To further define specificity of binding, we assayed
the abilities of four mutant oligonucleotides to compete
for nuclear protein binding to the probe. The mutant
oligonucleotides differ from the wild-type GATA region
element at clusters of three base pairs that span the
length of the cis element and are shown in Fig. 1.
Mutant oligonucleotides Mut-A, Mut-C, and Mut-D
provided in 100-fold excess were able to compete for
protein binding, whereas Mut-B was not (Fig. 3). In
synthesizing Mut-B, the wild-type sequence of GATA
at nucleotides ⫺90 to ⫺87 was replaced with CGGT.
The inability of the Mut-B oligonucleotide to compete
for binding suggests that the GATA sequence is essential for nuclear protein binding. The GATA sequence
comprises the consensus binding site for the GATA
family of zinc finger proteins. A similar interaction
between Caco-2 cell nuclear proteins and an upstream
GATA cis element of the human lactase promoter was
reported by Fitzgerald et al. (7). The factors interacting
with that cis element were not identified. We therefore
proceeded to determine the identity of the specific
Caco-2 nuclear proteins interacting with the GATA
region element of the rat lactase promoter.
GATA family proteins interact with the lactase promoter element. Members of the GATA family of transcription factors (GATA-4, -5, and -6) are expressed in
intestine and are capable of binding to the GATA cis
element in the IFABP promoter (9). To determine
whether GATA family proteins in Caco-2 cells bind to
the lactase gene-positive cis element identified above,
we incubated the radiolabeled GATA region probe and
nuclear extract from Caco-2 cells in the absence or
presence of a polyclonal GATA-4 and GATA-6 antibody
G62
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
(Fig. 4). GATA-5 binding could not be assayed because
human-reactive GATA-5-specific antibody was not
available. The complexes were analyzed by electrophoresis on a nondenaturing gel and analyzed for specific supershift of the previously identified complex.
Both the GATA-4 and GATA-6 antibodies recognized
the nuclear protein bound to the GATA region probe,
resulting in a complex of reduced gel mobility or a
supershift (Fig. 4). There is residual bound DNA-protein complex that is not supershifted by the GATA-4 or
GATA-6 antibody in each reaction, consistent with the
presence of both proteins interacting with the GATA
region probe. Although GATA-5 binding could not be
assayed directly, residual complex remained even in
reactions incubated in the presence of both GATA-4
and-6 antibodies (Fig. 4). As expected, no supershift
was observed for the binding reaction carried out in the
presence of nonimmune serum (Fig. 4).
GATA-4 and GATA-6 mRNA is present in proliferating Caco-2 cells. To confirm GATA-4 and GATA-6 gene
expression in Caco-2 cells, the presence of GATA-specific mRNA was assayed (Fig. 5). Total RNA isolated
Fig. 5. GATA-4 and GATA-6 mRNA detection in Caco-2 cells. One
microgram of total RNA extracted from proliferating Caco-2 was
reverse-transcribed (⫹) or not (⫺) and PCR-amplified with GATA-4-,
GATA-6-, lactase-, or glyceraldehyde-3-phosphate dehydrogenase
(GAPDH)-specific primers (left). Similar PCR reactions were performed with RT cDNA from adult human thymus, intestine, liver,
and spleen tissue (right). A single PCR-amplified product of the exact
predicted size was visualized for each primer set after ethidium
bromide staining of a 2% agarose gel. The negative gel images are
shown.
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
Fig. 4. Gel supershift with GATA region probe and anti-GATA-4 or
-6 or preimmune serum. Gel retardation was carried out with nuclear extract (10 ␮g) from Caco-2 cells and 32P-radiolabeled GATA
region probe. The reaction mixture was incubated for 30 min at room
temperature alone (⫺) or with 1.0 ␮l of polyclonal GATA-4 or
GATA-6 antibody or preimmune serum.
from proliferating Caco-2 cells was reverse-transcribed
and then PCR-amplified with GATA-4 and GATA-6
gene-specific primers. A single PCR product of the
exact predicted size was amplified for both the GATA-4
and GATA-6 primer sets in Caco-2 cells. There was no
detectable amplification of the RT reaction performed
in the absence of RT as a control for genomic DNA
contamination. In addition, GATA-4 and GATA-6
mRNA was detected in adult human small intestine
and liver. GATA-6 mRNA was also detected in human
spleen. Expression of the GAPDH mRNA was detected
as a positive control in all of the samples. GATA-4
mRNA was undetectable in control HeLa cells (data
not shown).
GATA-4, -5, and -6 can activate lactase gene transcription. To determine whether the intestinal GATA
proteins are capable of either activating or repressing
lactase gene transcription, Caco-2 and QT6 cells were
cotransfected with wild-type or mutated promoter-reporter constructs and with the expression constructs
for GATA-4, -5, and -6. In cloning the constructs, Xenopus GATA cDNA inserts were ligated downstream of
a recombinant CMV promoter in the expression vector
pCDNA3. Cotransfection with each of the GATA proteins results in a three- to fourfold transcriptional
activation of the wild-type lactase promoter-reporter
construct pgLac (Fig. 6A, compare pxGATA-4, -5, and
-6 vs. pCDNA3). Gao et al. (9) reported a similar
activation of the Xenopus IFABP gene promoter using
the identical GATA-4/5/6 expression constructs. To determine whether GATA factor overexpression was capable of activating the lactase promoter in nonintestinal cells, we assayed QT6 cells cotransfected with the
GATA expression constructs. Overexpression of the
GATA proteins in the nonintestinal QT6 cells resulted
in comparable transactivation of the lactase promoterreporter construct (Fig. 6B).
Mutant reporter constructs (pMutA, pMutB, pMutC,
pMutD) were generated by site-directed mutagenesis,
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
G63
Fig. 6. GATA-4, -5, -6 activation of lactase promoter reporter constructs. Subconfluent Caco-2 cells (A) or QT6 cells (B) were cotransfected with the wild-type lactase promoter-luciferase reporter construct pgLac and with the GATA protein expression plasmids
pxGATA-4, pxGATA-5, or pxGATA-6 or the empty expression vector
pcDNA-3. Relative luciferase activity was assayed 48 h after transfection. Transfection efficiencies were normalized to renilla luciferase expression of a cotransfected pRL-CMV vector and expressed
as relative luciferase activity (means ⫾ SD; n ⫽ 3).
and they differ from the wild-type promoter at the base
pairs corresponding to the respective Mut-A, -B, -C,
and -D oligonucleotides (see Fig. 1). Cotransfection of
the pMutA and pMutC reporter constructs with the
GATA-4 expression constructs results in a 2.5-fold or
greater transcriptional activation, comparable to that
of the wild-type promoter construct (Fig. 7A). However,
cotransfection of the pMutB reporter construct with
the GATA-4 expression construct results in a significant disruption of GATA factor activation. Cotransfection of nonintestinal QT6 cells with pMutA and pMutC
and the GATA-4 construct results in a similar lactase
promoter transactivation (Fig. 7B). Again, cotransfection with the pMut B reporter construct results in a
disruption of GATA-4 transactivation in QT6 cells. The
GATA consensus binding sequence WGATAR is mu-
Fig. 7. GATA activation of wild-type and mutant lactase promoterreporter constructs in Caco-2 cells. Proliferating Caco-2 cells (A) or
nonintestinal QT6 cells (B) were transfected with wild-type pgLac
(WT) or mutant lactase gene-luciferase reporter constructs (pMutA,
pMutB, pMutC, and pMutD) in the presence of the GATA-4 expression plasmid or the empty vector pCDNA3. Transfection efficiencies
were normalized to renilla luciferase expression of a cotransfected
pRL-CMV vector and expressed as activation relative to cotransfection with pCDNA3 (means ⫾ SD; n ⫽ 3).
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
tated in the pMutB construct. As predicted from the
results of the gel shift analysis, the GATA proteins
cannot bind to the mutant lactase promoter in pMutB
and therefore cannot activate transcription. GATA protein activation of the pMutD construct was also disrupted, suggesting that bases mutated downstream of
the GATA element may be involved in lactase transcriptional activation. In QT6 cells, GATA protein activation of the pMutC construct was significantly
greater than that of the wild-type promoter construct.
The mutated MutC sequence may disrupt a repressive
element or create an enhancing site capable of interacting with QT6 cell nuclear factors.
GATA factor binding to DNA correlates with functional promoter activation. To correlate binding of the
GATA factors to the lactase cis element DNA with the
functional promoter activation data, we performed gel
G64
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
DISCUSSION
The GATA-4, -5, and -6 subfamily of zinc finger
proteins are expressed in overlapping patterns in the
developing heart and endoderm-derived organs of the
gastrointestinal tract including the stomach, intestine,
liver, and pancreas (1, 5, 16, 19, 20). GATA binding
sites have been identified in transcriptional regulatory
elements of several cardiac-specific target genes. However, few intestinal target genes have been identified
for the GATA-4/5/6 factor subfamily. Fitzgerald et al.
(7) demonstrated that GATA-6 can activate the human
lactase promoter in cell culture. Subsequently, the
gene encoding IFABP was identified by Gao et al. (9) as
an vitro target for the GATA-4/5/6 factors in intestinal
cell culture. Both genes are expressed in absorptive
enterocytes. To identify regions of the intestinal lactase promoter involved in regulating lactase transcription, we have characterized the promoter activity of
various lactase reporter-promoter constructs. Fragments of the 5⬘-flanking region linked to the luciferase
reporter gene and transfected into Caco-2 cells were
assayed for transcriptional activity. A ⫺100 bp promoter-reporter construct possessed maximal activity
whereas the ⫺73 bp deletion construct was essentially
inactive compared with the promoterless pGL3-Basic
vector (Fig. 2A). As expected, the promoter constructs
were inactive when transfected into nonintestinal QT6
cells, which do not express the lactase gene (Fig. 2B).
Close inspection of the sequence in this positive element mapped between nucleotides ⫺73 and ⫺100 of
the lactase promoter revealed a consensus binding
sequence for the GATA family of zinc finger transcription factors. Fitzgerald et al. (7) mapped a positive
GATA cis element to a similar region in the human
lactase gene. The 28-bp GATA region sequence is
highly conserved in the rat, pig, and human lactase
genes, providing support for an important functional
role in regulating intestinal lactase transcription.
The nuclear proteins interacting with the GATA
region element were assayed using nuclear extract
prepared from Caco-2 cells. The predominant DNAprotein complex seen on gel shift analysis was determined to be a specific interaction that was competed for
by excess GATA region probe but not by an unrelated
oligonucleotide (Fig. 3). To further define specificity of
binding, we assayed for the ability of four mutant
GATA region oligonucleotides to compete for nuclear
protein binding to the probe. Only Mut-B, in which the
wild-type sequence of GATA at nucleotides ⫺90 to ⫺87
was replaced with CGGT, was unable to compete for
binding to the wild-type probe. The inability of the
Mut-B oligonucleotide to compete for binding suggests
that the GATA sequence is essential for nuclear protein binding. The sequences surrounding the GATA
region comprise the consensus binding site, WGATAR,
for the GATA family of zinc finger transcription factors. Fitzgerald et al. (7) reported a similar interaction
between Caco-2 cell nuclear proteins and the upstream
GATA cis element of the human lactase promoter. We
have identified the proteins interacting with the GATA
region element. Specific antibodies against GATA-4
and GATA-6 recognized the nuclear protein bound to
the GATA region probe, resulting in a supershifted
EMSA complex (Fig. 4). Both GATA-4 and GATA-6
proteins therefore are expressed in Caco-2 cells and
appear to interact with the GATA element of the lactase promoter. GATA-5 binding could not be directly
assayed because human-reactive GATA-5 antibody
was not available. However, we speculate that the
residual EMSA complex not supershifted in the presence of both GATA-4 and -6 antibodies may represent
GATA-5 binding.
GATA-4/5/6 are each expressed in developing gastrointestinal and gut-derived tissues (5). Fitzgerald et al.
(7) demonstrated that GATA-6 RNA is present in
Caco-2 cells but were unable to detect GATA-4 protein
or RNA in Caco-2 cells by gel supershift or Northern
Fig. 8. Electrophoretic mobility shift assays with nuclear extract of QT6 cells cotransfected with GATA expression
constructs. A: nuclear extract protein (10 ␮g) isolated from QT6 cells cotransfected with the wild-type lactase
promoter-luciferase reporter construct pgLac and with the GATA protein expression plasmids pxGATA-4,
pxGATA-5, or pxGATA-6 or the empty expression vector pcDNA-3 was incubated with 32P-labeled GATA region
double-stranded oligonucleotide probe alone (⫺) or in the presence of 100⫻ unlabeled specific wild-type (WT) or
mutant (MutB) oligonucleotides. B: gel supershifts were carried out with the QT6/GATA-4 nuclear extract
incubated for 30 min at room temperature alone (⫺) or with 1.0 ␮l of polyclonal GATA-4 or preimmune serum.
Samples were loaded on a 6% acrylamide gel.
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
shift and supershift analysis with nuclear extract isolated from QT6 cells transfected with the GATA-4/5/6
expression constructs as described above. A GATAspecific complex, a slower migrating complex A, a faster
migrating prominent complex B, and a nonspecific complex are detected in QT6 cells cotransfected with the
GATA-4, -5, and -6 constructs (Fig. 8A). The relative
migration of the GATA-specific complex is slightly different for each factor, with the GATA-6 reactions consistently resulting in a less abundant complex. The
GATA-specific complex is not detected in the QT6 cells
cotransfected with the empty pcDNA3 expression vectors. The GATA-specific complex is competed by 100fold excess unlabeled probe but not by the MutB probe,
in which the GATA sequence is mutated. In addition,
the GATA-4-specific complex is supershifted with the
GATA-4 antibody (Fig. 8B). The faint GATA-6-specific
complex is similarly supershifted with GATA-6 antibody (data not shown), and the GATA-5-specific complex could not be assayed because human-reactive
GATA-5 antibody was not available. These results indicate that the lactase promoter construct is inactive in
QT6 cells (Fig. 2B) but that overexpression of GATA
factors can result in factor binding to the lactase GATA
region cis element and can function to stimulate promoter activity.
G65
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
G66
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
that when bound, GATA-4, -5, and -6 are capable of
activating transcription of the lactase promoter.
To correlate binding of the GATA factors to lactase
cis element DNA with the functional promoter activation data, we performed gel shift and supershift analyses with nuclear extract isolated from the cotransfected QT6 cells and the GATA region probe. QT6 cells
cotransfected with the GATA expression constructs
form a GATA-specific protein-DNA complex that is
supershifted with GATA-specific antibody (Fig. 8).
GATA factor binding to the GATA region cis element of
the lactase promoter therefore is correlated with functional activation of the lactase promoter. Of note, a
faster-migrating prominent complex B band was
formed in gel shifts with QT6 cell nuclear extract and
the GATA region probe and was not supershifted with
GATA antibody. Whether other factors, possibly those
comprising complex B, are involved in repressing lactase transcription in the nonintestinal QT6 cells is yet
to be determined.
Fitzgerald et al. (7) demonstrated that GATA-6 can
stimulate activation of the lactase promoter. We have
demonstrated that each of the members of the zinc
finger GATA binding protein family expressed in intestine (GATA-4 and GATA-5 in addition to GATA-6) can
recognize a GATA consensus binding sequence within
a positive cis element of the lactase promoter and can
activate transcription. Gao et al. (9) demonstrated that
the GATA-4/5/6 proteins are also capable of binding to
a GATA cis element in the promoter of the IFABP gene
and can activate transcription of the IFABP promoter
in enterocyte cell culture. These findings for two intestine-specific genes (lactase and IFABP) support a role
for the GATA family of transcription factors in regulating intestinal gene expression. Gao et al. (9) described differences in relative abundance of specific
GATA proteins along the small intestine crypt-villus
axis and in differentiating intestinal cells in culture.
Relative levels of each GATA factor within enterocytes
may therefore regulate transcription during cell differentiation. Other gene regulatory nuclear proteins may
also interact with the GATA proteins to mediate their
ability to activate or repress transcription during enterocyte differentiation. It will be of interest to determine whether the complex interactions between multiple GATA transcription factors function to regulate
gut differentiation and development.
We thank Dr. Todd Evans for providing the pxGATA-4, -5, and -6
expression vectors. We thank Drs. D. H. Alpers, G. M. Gray, and
M. P. Scott for helpful discussions and suggestions. We thank Libra
White for expert secretarial assistance.
E. Sibley was supported by National Institute of Diabetes and
Digestive and Kidney Diseases Grant DK-02552.
REFERENCES
1. Arceci RJ, King AA, Simon MC, Orkin SH, and Wilson DB.
Mouse GATA-4: a retinoic acid-inducible GATA-binding transcription factor expressed in endodermally derived tissues and
heart. Mol Cell Biol 13: 2235–2246, 1993.
2. Ausubel FM, BR, Kingston RE, Moore DD, Seidman JG,
Smith JA, and Struhl K. Current Protocols in Molecular Biology. New York: Wiley, 1997.
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
blot hybridization. To assay for the presence of GATA-4
and -6 mRNA, we performed RT-PCR amplification
using total RNA isolated from Caco-2 cells (Fig. 5).
Specific RT-PCR products corresponding to both
GATA-4 and GATA-6 mRNA species were detected.
GATA-4 mRNA was undetectable in control HeLa cells
(data not shown). These gel shift and RT-PCR results
indicate that GATA-4, in addition to GATA-6, interacts
with the GATA binding element (Fig. 4) and that both
mRNA species are expressed in Caco-2 cells (Fig. 5).
The discrepancies may be attributable to differences in
GATA-specific antibody affinities and the enhanced
sensitivity of RT-PCR for mRNA detection. Variations
or shifts in the phenotypic expression pattern of Caco-2
cells cultured by different investigators over time may
also account for the discrepancies. In addition, as expected (26), mRNA for both GATA-4 and GATA-6 was
detected in both liver and small intestine, but only
GATA-6 mRNA was detected in spleen, and no expression was detected in thymus (Fig. 5).
GATA transcription factor function has been implicated in the regulation of intestinal epithelial cell differentiation (9). To determine whether GATA family
proteins are capable of either activating or repressing
lactase gene transcription, Caco-2 cells were transfected with wild-type or mutated promoter-reporter
constructs and with GATA-4, -5, and -6 expression
constructs. Each of the GATA expression constructs
was capable of activating the luciferase reporter driven
by the wild-type lactase promoter in Caco-2 cells (Fig.
6A). Fitzgerald et al. (7) initially reported a similar
transactivation function for GATA-6. Our results indicate that the GATA-4 and GATA-5 proteins, in addition to GATA-6, are capable of functioning to activate
lactase promoter activity in intestinal cells. In addition, there was transactivation of the lactase promoterreporter construct by the GATA-4/5/6 expression constructs in cotransfected QT6 fibroblasts (Fig. 6B).
GATA nuclear protein overexpression is therefore sufficient to activate the lactase promoter in a nonintestinal cell type that does not express lactase.
The mutated pMutB lactase promoter-reporter construct, in which the WGATAR binding site was disrupted, was not significantly transactivated by GATA
factor overexpression (Fig. 7). As predicted from the
results of the gel shift analysis, the GATA proteins
cannot bind to the mutant lactase promoter in pMutB
and therefore cannot activate transcription. GATA protein activation of the pMutD construct was also disrupted, suggesting that bases mutated just downstream of the GATA element are also involved in
lactase transcriptional activation. The nuclear proteins
interacting with such downstream sequence would not
have resulted in an additional gel shift complex if the
entire recognition sequence was not contained within
the GATA region probe. It is not unexpected that multiple transcription factors may interact with the sequence throughout the 100-bp lactase promoter region.
The cotransfection results, combined with the gel shift
analysis, suggest that the GATA region of the lactase
promoter is essential for GATA protein binding and
GATA-4, -5, AND -6 ACTIVATE LACTASE TRANSCRIPTION
18. Molkentin JD, Kalvakolanu DV, and Markham BE. Transcription factor GATA-4 regulates cardiac muscle-specific expression of the ␣-myosin heavy-chain gene. Mol Cell Biol 14:
4947–4957, 1994.
19. Morrisey EE, Ip HS, Lu MM, and Parmacek MS. GATA-6: a
zinc finger transcription factor that is expressed in multiple cell
lineages derived from lateral mesoderm. Dev Biol 177: 309–322,
1996.
20. Morrisey EE, Ip HS, Tang Z, Lu MM, and Parmacek MS.
GATA-5: a transcriptional activator expressed in a novel temporally and spatially-restricted pattern during embryonic development. Dev Biol 183: 21–36, 1997.
21. Nudell DM, Santiago NA, Zhu JS, Cohen ML, Majuk Z, and
Gray GM. Intestinal lactase: maturational excess expression of
mRNA over enzyme protein. Am J Physiol Gastrointest Liver
Physiol 265: G1108–G1115, 1993.
22. Orkin SH. GATA-binding transcription factors in hematopoietic
cells. Blood 80: 575–581, 1992.
23. Pevny L, Simon MC, Robertson E, Klein WH, Tsai SF,
D’Agati V, Orkin SH, and Costantini F. Erythroid differentiation in chimaeric mice blocked by a targeted mutation in the
gene for transcription factor GATA-1. Nature 349: 257–260,
1991.
24. Rings EH, de Boer PA, Moorman AF, van Beers EH, Dekker J, Montgomery RK, Grand RJ, and Buller HA. Lactase
gene expression during early development of rat small intestine.
Gastroenterology 103: 1154–1161, 1992.
25. Soudais C, Bielinska M, Heikinheimo M, MacArthur CA,
Narita N, Saffitz JE, Simon MC, Leiden JM, and Wilson
DB. Targeted mutagenesis of the transcription factor GATA-4
gene in mouse embryonic stem cells disrupts visceral endoderm
differentiation in vitro. Development 121: 3877–3888, 1995.
26. Suzuki E, Evans T, Lowry J, Truong L, Bell DW, Testa JR,
and Walsh K. The human GATA-6 gene: structure, chromosomal location, and regulation of expression by tissue-specific
and mitogen-responsive signals. Genomics 38: 283–290, 1996.
27. Troelsen JT, Mitchelmore C, Spodsberg N, Jensen AM,
Noren O, and Sjostrom H. Regulation of lactase-phlorizin
hydrolase gene expression by the caudal-related homeodomain
protein Cdx-2. Biochem J 322: 833–838, 1997.
28. Troelsen JT, Olsen J, Noren O, and Sjostrom H. A novel
intestinal trans-factor (NF-LPH1) interacts with the lactasephlorizin hydrolase promoter and co-varies with the enzymatic
activity. J Biol Chem 267: 20407–20411, 1992.
29. Tsai FY, Keller G, Kuo FC, Weiss M, Chen J, Rosenblatt M,
Alt FW, and Orkin SH. An early haematopoietic defect in mice
lacking the transcription factor GATA-2. Nature 371: 221–226,
1994.
30. Yamagata T, Nishida J, Sakai R, Tanaka T, Honda H,
Hirano N, Mano H, Yazaki Y, and Hirai H. Of the GATAbinding proteins, only GATA-4 selectively regulates the human
interleukin-5 gene promoter in interleukin-5-producing cells
which express multiple GATA-binding proteins. Mol Cell Biol 15:
3830–3839, 1995.
Downloaded from http://ajpgi.physiology.org/ by 10.220.32.247 on June 15, 2017
3. Buller HA, Kothe MJ, Goldman DA, Grubman SA, Sasak
WV, Matsudaira PT, Montgomery RK, and Grand RJ. Coordinate expression of lactase-phlorizin hydrolase mRNA and
enzyme levels in rat intestine during development J Biol Chem
265: 6978–83, 1990. [Erratum. J Biol Chem 265: August 1990, p.
13410.]
4. Ercolani L, Florence B, Denaro M, and Alexander M. Isolation and complete sequence of a functional human glyceraldehyde-3-phosphate dehydrogenase gene. J Biol Chem 263: 15335–
15341, 1988.
5. Evans T. Regulation of cardiac gene expression by GATA-4/5/6.
Trends Cardiovasc Med 7: 75–83, 1997.
6. Fang R, Santiago NA, Olds LO, and Sibley E. The homeodomain protein Cdx2 regulates lactase gene promoter activity
during enterocyte differentiation. Gastroenterology 118: 115–
127, 2000.
7. Fitzgerald K, Bazar L, and Avigan MI. GATA-6 stimulates a
cell line-specific activation element in the human lactase promoter. Am J Physiol Gastrointest Liver Physiol 274: G314–G324,
1998.
8. Fujiwara Y, Browne CP, Cunniff K, Goff SC, and Orkin
SH. Arrested development of embryonic red cell precursors in
mouse embryos lacking transcription factor GATA-1. Proc Natl
Acad Sci USA 93: 12355–12358, 1996.
9. Gao X, Sedgwick T, Shi YB, and Evans T. Distinct functions
are implicated for the GATA-4, -5, and -6 transcription factors in
the regulation of intestine epithelial cell differentiation. Mol Cell
Biol 18: 2901–2911, 1998.
10. Gordon JI and Hermiston ML. Differentiation and self-renewal in the mouse gastrointestinal epithelium. Curr Opin Cell
Biol 6: 795–803, 1994.
11. Grepin C, Dagnino L, Robitaille L, Haberstroh L, Antakly
T, and Nemer M. A hormone-encoding gene identifies a pathway for cardiac but not skeletal muscle gene transcription. Mol
Cell Biol 14: 3115–3129, 1994.
12. Ip HS, Wilson DB, Heikinheimo M, Tang Z, Ting CN, Simon MC, Leiden JM, and Parmacek MS. The GATA-4 transcription factor transactivates the cardiac muscle-specific troponin C promoter-enhancer in nonmuscle cells. Mol Cell Biol 14:
7517–7526, 1994.
13. Ko LJ and Engel JD. DNA-binding specificities of the GATA
transcription factor family. Mol Cell Biol 13: 4011–4022, 1993.
14. Krasinski SD, Estrada G, Yeh KY, Yeh M, Traber PG,
Rings EH, Buller HA, Verhave M, Montgomery RK, and
Grand RJ. Transcriptional regulation of intestinal hydrolase
biosynthesis during postnatal development in rats. Am J Physiol
Gastrointest Liver Physiol 267: G584–G594, 1994.
15. Kuo CT, Morrisey EE, Anandappa R, Sigrist K, Lu MM,
Parmacek MS, Soudais C, and Leiden JM. GATA4 transcription factor is required for ventral morphogenesis and heart tube
formation. Genes Dev 11: 1048–1060, 1997.
16. Laverriere AC, MacNeill C, Mueller C, Poelmann RE,
Burch JB, and Evans T. GATA-4/5/6, a subfamily of three
transcription factors transcribed in developing heart and gut.
J Biol Chem 269: 23177–23184, 1994.
17. Merika M and Orkin SH. DNA-binding specificity of GATA
family transcription factors. Mol Cell Biol 13: 3999–4010, 1993.
G67