|
|
|
|||||||
|
|||||||||
(Received for publication, April 18, 1995; and in revised form, June
15, 1995) From the
Nuclear factor I (NFI) was suggested to be involved in the
expression of the human
Expression of NFI was originally isolated from HeLa
cells as a host protein required for the efficient replication of
adenovirus 2/5 DNA in vitro and in
vivo(10, 11) . NFI specifically recognizes the
DNA consensus sequence
5`-TGG(N In this context it is of interest
whether NFI could contribute tissue specificity in
Figure 1:
A, slot blot
analysis of RNA of HeLa and K562 cells. An autoradiograph of a
representative experiment is shown. The values on the top of
the bars indicate relative amount of expression of
Figure 2:
In vivo footprinting (A and B) and genomic sequencing (C) of the
Figure 3:
Summary of the in vivo footprinting data of the
The in vivo footprinting patterns obtained with both cell lines is
clearly different from that of the protein-free DNA methylated in
vitro. This indicates that, in vivo, several proteins
occupy the
In
contrast, binding of the NFI site is clearly evident in HeLa cells,
where the first two guanosines of the first half of the NFI site are
consistently found to be protected from in vivo methylation
(underlined in GGG(N All conclusions in the last two sections are based on the analysis
of the sense strand. In spite of the application of various
experimental conditions for the PCR (use of formamide, deoxynucleotide
analogues, dimethyl sulfoxide, different temperatures, and
concentrations of compounds), we never obtained interpretable signals
from the antisense strand. This is most probably due to the even higher
GC content of the DNA upstream of the NFI site, which we believe is
responsible for the attenuation of the signals beyond the displayed
region also of the sense strand (not shown). Ambiguous annealing of the
PCR primers to this region may impede the analysis of the antisense
strand.
Figure 4:
FACS analysis of NFI in K562, K562
hemin-induced and HeLa cells. A nonimmune serum served as a control.
The broken line serves as a comparison of differences in
fluorescence. For details see ``Materials and
Methods.''
Figure 5:
A, binding and competition analysis of
NFI
The specificity of the polypeptides
in the shifted bands as NFI site-recognizing species and their relative
DNA binding affinity was checked by competition assays with 100-fold
molar excess of unlabeled DNA (Fig. 5A). Homologous
competition decreased the intensity of the control and of all shifted
bands of HeLa and of K562 cells (Fig. 5A, lanes
3, 7, and 11, respectively). Competitor L1/2 with a higher affinity NFI site (41) competes for binding
again with all polypeptides of the control and both cell lines (Fig. 5A, lanes 4, 8, and 12, respectively). In contrast, a mutant oligonucleotide
(designated It has been
reported that some cells contain a protein that binds exactly to the
NFI site but is not NFI(56) . NFI proteins share their greatest
sequence homology in the N-terminal DNA binding domain of the
polypeptide(57) . Therefore, in order to ascertain that the
complexes investigated here were due to NFI or NFI-like species, we
determined the immunological characteristics of the electrophoretic
mobility shift assay complexes by reaction with a polyclonal antiserum
directed against the recombinant NFI bearing the DNA binding domain (42) (Fig. 5B). This serum caused a strong
reduction of the DNA complex of the recombinant NFI (control) and
induced the formation of a supershifted NFI NFI has been implicated in several aspects of DNA
transcription and replication. Functions of NFI include the potential
to act as transcriptional
activator(28, 32, 58) , repressor (16,
59-62, and (5) and (49) -52 and references
therein), antirepressor(63) , place-holder in chromatin
organization as a kind of ``antirepressor of
replication''(64) , and replication protein in viral
systems(12) . Several studies suggested that NFI also
stimulates Although both HeLa cells
and the majority of K562 cells do not express We do not know the reason
why most of the cells cannot be induced by hemin to activated
The in vivo footprints reveal absence of NFI binding at the NFI site of the
promoter in K562 cells. Although it is not possible to assay from the
present data for a function of NFI in active In contrast
to K562 cells, HeLa cells which uniformly do not express Binding of NFI, or one NFI isoform, to the
May we expect to see in vivo footprints of the
Volume 270,
Number 33,
Issue of August 18, pp. 19643-19650, 1995
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
-Globin Promoter and Possible Role of Nuclear Factor I in
an
-Globin-inducible and in a Noninducible Cell Line (*)
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
-globin gene. Two established cell lines,
which express -globin differentially, were therefore compared for
differences in binding of NFI at the -globin promoter in
vivo. HeLa cells, in which -globin is repressed, show a high
density promoter occupation with several proteins associated with
structurally distorted DNA. Cell line K562, which is inducible for
-globin, surprisingly was found to be heterogeneous consisting
mainly of cells (95%) unable to express
-globin. However, the
promoter of the nonexpressing K562 cells was clearly different from
that of HeLa cells, being occupied only at basal transcriptional
elements. Therefore, the -globin gene in these K562 cells may not
be truly repressed, but in an intermediate state between repression and
active transcription. The NFI site of the -globin promoter
appeared occupied in HeLa but free of proteins in K562 cells. All cells
of both cell lines produce NFI, but the composition and DNA binding
affinity of NFI species differ significantly between the two cell
lines. Therefore, distinct forms of NFI may repress -globin
transcription in HeLa cells. However, NFI is apparently not involved in
establishing the latent transcriptional state of the majority of K562
cells.
-globin is regulated in vivo by the
interplay of the locus control region at -40 kilobases and
diverse promoter elements(1, 2) . Activation of a
particular gene in the -globin cluster is supposed to be achieved
by interaction of factors binding to the locus control region and
factors binding to promoter and enhancer elements, thereby keeping the
chromatin free of histones(3) . However, this histone-free
state which also correlates with DNase I hypersensitivity (4) bestows upon the globin genes only transcriptional
competence. Additional events or factors that bind at the regulatory
elements are required for a particular gene to be actively transcribed.
In the erythroid lineage, the major specific transcription factor for
globin gene expression is GATA-1(5, 6, 7) .
However, the promoter of the human -globin gene contains no GATA-1
site but instead basal transcription elements, a possible SP1/-IRP
site(8) , and a binding site for nuclear factor I (NFI; (9) ). (
)
)GCCAA-3`(12, 13, 14) .
NFI binding sites are found in many viral and cellular promoters and
enhancers (see (3, 4, 5, 6) and
7-12 in (9) ) suggesting a role of NFI as transcriptional
regulator. Most of these genes display tissue specificity in their
expression(15, 16, 17, 18, 19, 20, 21, 22) .
However, NFI is a ubiquitous factor, and it is not known whether it can
influence transcription in a tissue-specific manner. NFI may be
involved in transcription as a ubiquitous factor with specificity
provided through association with other, cell-specific factors.
Precedents for this mode include, for example, the association of the
ubiquitous Jun and Fos with the lymphoid-specific NF-AT
factor(23) . Alternatively, NFI may act as a
cell-specific transcriptional regulator in spite of its ubiquitous
expression. The latter view is supported by the observation of the
presence of different forms of NFI in different cell types (24, 25, 26) . These forms can arise by
expression of different NFI genes (24, 27) , by
differential splicing(28) , by diverse covalent
post-translational modifications(29, 30) , or by
heterodimerization (31) .-globin gene
expression. NFI has been, in fact, implicated in the multistep process
of transcriptional activation of the human -globin gene by in
vitro(32) and in vivo transient assays with
reporter plasmids(28, 33) . These assays revealed a
weak but clear stimulation of -globin transcription following
binding of NFI to the promoter sequence. Originally it was thought that
stimulation occurs by binding to the general positive cis-acting CCAAT genetic element of this
promoter(34) . It was this assumption which led to the
definition of NFI as ``CTF'' (=CCAAT-box transcription
factor) implying a role for NFI as a general transcription
factor(32) . However, we demonstrated that specific and fairly
strong binding of NFI actually occurs at an adjacent previously
unrecognized NFI site within the -globin promoter(9) .
Furthermore, in vivo analysis with reporter plasmids suffers
from copy number effects and does not account for the influence of
chromatin which is known to play an important role in gene
transcription via the presence of specific histones, nucleosomes, and
higher order structures, such as the 30-nm-diameter chromatin filament,
locus boundary elements, and the nuclear matrix or
scaffold(3) . Chromatin structure is particularly important for
NFI binding and function; for example, Lee and Archer (35) demonstrated recently that NFI can bind and activate the
murine mammary tumor virus promoter from transiently transfected,
``naked'' plasmid templates, whereas the chromatin version of
the same sequence in the same cell is refractory to NFI action. For
these reasons, we wondered whether occupation of the NFI site of the
-globin promoter in the chromosomal context could be
correlated with a particular transcriptional state of the -globin
gene. This would clarify a possible importance of this site in vivo and provide clues for an implication of NFI in -globin gene
expression and regulation in situ. To start approaching this
question, we compared the in vivo footprints of an inducible
(K562) and a noninducible (HeLa) cell line. Our present results suggest
that the transcriptional state of K562 cells does not correlate with
NFI binding to the -globin promoter in vivo. Our data are
also compatible with the hypothesis that NFI species found in HeLa
cells could act as repressors of -globin transcription.
RNA Isolation and Slot Blots
K562 cells were
purchased from ATCC (CCL 243); HeLa cells were a laboratory stock.
Logarithmically growing K562 or HeLa cells suspended in RPMI 1640
medium supplemented with 10% fetal calf serum were harvested by
centrifugation (1000 rpm, at 4 °C), washed in phosphate-buffered
saline, and RNA was isolated according to a standard
protocol(36, 37) . Hemin induction of K562 cells was
done with a final concentration of 50 µg/ml for the times indicated
in Fig. 1A. Slot blots were performed on a Schleicher
& Schuell apparatus (SRC 07210 Minifold II). 30 µg of total RNA
in 100 µl of TE were added to 300 µl of 6.15 M formaldehyde, 10 SSC, denatured for 15 min at 65 °C,
and transferred on a nitrocellulose filter. The filter was baked at 80
°C for 2 h. Specific probes for the
-globin mRNA were
oligonucleotides C and E of the first primer set also used in the
ligation-mediated PCR. Glyceraldehyde-3-phosphate dehydrogenase
specific probes were oligonucleotides 5`-CCAGTGAGCTTCCCGTTCAGCTC-3` and
5`-CCACCACCCTGTTGCTGTAGCC-3`. They were radioactively labeled at the 5`
ends as described(9) . The sarcosyl technique was used for
hybridization(38) .
-globin mRNA. This amount was arbitrarily set 1 for the K562 cells
prior to induction with hemin (0 h). Values were normalized by using
the glyceraldehyde-3-phosphate dehydrogenase signals as a standard. B, FACS analysis of -globin in K562, K562 hemin-induced,
and HeLa cells. A nonimmune serum served as a control. The broken
line serves as a comparison of differences in fluorescence. For
details see ``Materials and
Methods.''
Protein Extracts
Purification of
baculovirus-expressed NFI (amino acids 1-257) from infected Spodoptera frugiperda (Sf9) cells is described in (9) . HeLa and K562 whole nuclear extracts were prepared as
described (39) with the exception that 1 mM phenylmethylsulfonyl fluoride was added to all solutions. Protein
concentration was determined by a standard method (40) and was
between 2.5 and 4 µg/µl. Whole cell extracts from uninfected
Sf9 cells were kindly provided by M. Stanglmaier.Electrophoretic Mobility Shift Assay and Antibody
Supershifting
Oligonucleotides were synthesized by G. Arnold
(Laboratory for Molecular Biology-Genzentrum, Martinsried).
Oligonucleotide ``-G wt'' contains the NFI site
in the context of the -globin promoter(9) .
Oligonucleotide L1/2 with a higher affinity NFI site has been
described in (41) . Oligonucleotide ``-G
mut.'' in which the NFI site has been inactivated is the same
as oligonucleotide ``k'' in (9) . Purification and
radioactive labeling of the oligonucleotides and the conditions for
protein-DNA incubation are described(9) . The amount of the
labeled double-stranded oligonucleotide -G wt was usually
5 fmol, whole nuclear extracts were 2-4 µl; cold competitors
were added before the binding reaction in 100-fold molar excess. In
supershift experiments, the binding reaction was on ice, then 1 µl
of nonimmune or anti-NFI-antiserum (described in (42) ) was
added, and incubation was continued for 15 min. Native polyacrylamide
gel electrophoresis (acrylamide:bisacrylamide = 30:0.8) and
autoradiography were as in (43) .Genomic Sequencing and in Vivo
Footprinting
Conditions for base-specific modification in
vitro and piperidine cleavage were as described(44) . For in vivo footprinting with dimethyl sulfate (DMS), K562 or HeLa
cells were grown as described above. -Globin induction of the K562
cells was with 50 µg/ml hemin for 24 h. Treatment with
-amanitin was at a final concentration of 10 µg/ml for 1 h.
The cells were washed with an isotonic phosphate buffer and incubated
at 3 10
cells per ml in RPMI containing 0.2% DMS
(Merck). The reaction was at room temperature for 2 min and was stopped
by adding to the cells 40 volumes of cold phosphate buffer with 2%
-mercaptoethanol and subsequently by removing the medium by
centrifugation. Cells remain viable after this treatment as controlled
by trypan blue exclusion. The DNA was extracted by a standard protocol
and cleaved at modified residues with piperidine. To visualize the DNA
sequence, the ligation-mediated PCR method was used essentially as
described(45) . For the extension step with the radioactive
primer E, four PCR cycles were performed. The primer set used for
analyzing the sense-strand was: primer C,
5`-CAGGAGACAGCACCATGGTGGGTTC-3`; primer D,
5`-GGTGGGTTCTCTCTGAGTCTGTGGG-3`; and primer E,
5`-AGTCTGTGGGGACCAGAAGAGTGCC-3`. The in vivo footprinting
experiments were repeated several times; the results are reproducible
in the sense that the patterns are exactly the same whenever the same in vivo methylated DNA batch was analyzed and similar, but
never contradictory, in the variation of band intensities, when a
different DNA batch or a different set of primers was used.Flow Cytometry
Intracellular expression levels of
-globin and NFI were determined by the use of specific antibodies
and flow cytometry (FACS) as described previously(46) .
Briefly, K562 cells and HeLa cells were washed twice in
phosphate-buffered saline, fixed with 2.5% paraformaldehyde (10 min, 4
°C), and permeabilized for 4 min at room temperature with 0.0025%
digitonin (Sigma) in order to allow intracellular antibody binding.
Cells were adjusted to 2 10
per ml in staining
buffer (phosphate-buffered saline, 4% fetal calf serum, 5 mM EDTA, 0.1% NaN
) and were incubated for 20 min at 12
°C with a rabbit anti--globin immune serum (Sigma; final
dilution: 1:3000) or with protein A-purified rabbit anti-NFI polyclonal
antibodies (100 µg/ml)(42) . After 2 washes in staining
buffer, cells were incubated with F(ab`)
fluorescein
isothiocyanate-conjugated anti-rabbit IgG and IgM antibodies (1:160,
Dianova, Hamburg, FRG). Cells were washed twice, counterstained with
propidium iodine (5 µg/ml in 4 mM sodium citrate, 0.1%
Triton X-100, pH 7.0) and analyzed by flow cytometry (Becton Dickinson,
Heidelberg, FRG).
The
In order to define a system for the study of
-Globin-inducible Cell Line, K562, Is
Heterogeneous with Respect to the -Globin Gene
Expression-globin expression, we used K562 cells, a commonly used
erythroleukemic cell line, for the investigation of expression of the
globin genes (cf. for example, (47) ). K562 cells can
be stimulated to actively transcribe the -globin gene after hemin
induction for different periods of time (Fig. 1A).
There is a clear difference in the amount of -globin mRNA in K562
cells compared to the non--globin expressing HeLa control cells
(at least 12.5-fold more -globin mRNA at 72 h). However, in the
K562 cell population, -globin mRNA already displays a high
uninduced level (0 h) and is induced by a comparatively very low factor
of only about 2.5-fold after 72 h. We wondered whether this reflects a
uniformly low level of induction of all or most of the K562 cells, or
rather a heterogeneous composition of this cell line, with some cells
expressing high levels of -globin and others expressing low levels
or no -globin. We therefore determined -globin expression in
K562 cells by FACS analysis using specific anti--globin
antibodies; we indeed detected different subpopulations (Fig. 1B). In the uninduced state, the great majority
of cells essentially does not express -globin, whereas a few cells
show comparatively high expression; the latter may be the reason for
-globin mRNA being already detectable before induction (see Fig. 1A). After hemin stimulation, the fraction of
cells expressing -globin increases, but does not exceed 5% of
the overall population. This means that a small number of cells are
actually responsible for virtually all of the
-globin expression
in this cell line. As expected, no HeLa cell expresses -globin
detectably (Fig. 1B), which parallels the RNA analysis
(see above). Accepting that protein levels directly mirror ongoing
-globin mRNA synthesis, these results mean that most of the K562
cells do not actively transcribe the -globin gene, i.e. their state of -globin expression is equivalent to that of
HeLa cells.The General
Since HeLa cells could not
be induced to express -Globin Promoter Configuration in Vivo
Differs between K562 and HeLa Cells-globin (data not shown), the condition, in
which promoter elements of the -globin gene in these cells are,
may reflect the dormant, or fully repressed state. On the other hand,
the few K562 cells, which are transcribing -globin and can even be
stimulated (e.g. by hemin; this work), define the active state
of the -globin gene. We wondered whether the -globin
promoter, in the great majority of K562 cells which are silent, is also
in a repressed state as in HeLa cells. To compare the promoter
structure of -globin in the two cell lines, we performed in
vivo dimethyl sulfate (DMS) footprinting analysis. Typical results
are shown in Fig. 2, A and B; all data are
summarized in Fig. 3. ()To properly interpret the
results, we consider band intensities to be altered in vivo only when they are flanked by any two guanosines, the intensity of
which is not altered compared to the in vitro signals.
-globin promoter. Genomic DNA from HeLa or K562 cells was modified in vitro or in vivo as indicated, and specific
regions were detected with the ligation-mediated PCR as described under
``Materials and Methods.'' Amplification products were
analyzed by sequencing gel electrophoresis and autoradiography. The
localization of the NFI, CCAAT, -IRP, and ATA sites is indicated. Lines with open dots denote decreases, arrows denote
increases, and rectangles are examples for bands with equal
intensity of in vivo and in vitro modified DNA. Lane 5 is a longer exposure of lane 4, which was
performed to visualize the signals of the top region of the gel. The dotted lines indicate regions of lanes 4 and 5, which were not evaluated.
-globin promoter in K562 and HeLa
cells. Symbols are as in Fig. 2. For comparison, a summary of
NFI footprints is also displayed, which was obtained by methylation
interference analysis of this region in
vitro(9) .
-globin promoter in both cell lines (Fig. 2, lanes 1 and 4 or 5 versus lanes 2 and 3). However, the in vivo footprinting pattern is
distinct for each cell line (Fig. 2, lane 1 versus lane 4 or 5): HeLa cells generally display a high density
occupation of the promoter with proteins, revealed by the DMS-protected
guanosines (denoted by lines with open dots in Fig. 2, lane 1, and in Fig. 3), whereas K562 cells show only a
minimal occupation of the promoter (Fig. 2, lane 4 or 5, and Fig. 3). Additionally, the promoter in HeLa
cells shows a particular region of about 25 nucleotides with many
DMS-hypersensitive purine residues (denoted by arrows in Fig. 2, lane 1, and in Fig. 3). Also, two
neighboring cytosine residues at two sites (within the -IRP site
and 3` of it; compare Fig. 3) become methylated to some extent
in this region, presumably at the N-3 positions which can occur only
after strong distortion of the double strand state of the DNA (cf., for example, ``Discussion'' in (48) ,
and references therein). For these reasons, we suggest that the DMS
hypersensitivity unequivocally indicates a dramatic alteration of the
secondary structure at this region of the DNA in HeLa cells. In
contrast, unusual reactivity of the DNA bases was not observed in K562
cells. Interestingly, protein binding in K562 cells seems to happen
exclusively next to basal transcription elements, such as CCAAT-box,
ATA-box and cap-site, as opposed to HeLa cells, in which at least the
last two sites appear essentially protein-free (Fig. 2, lane
4 or 5 versus lane 1; see summary in Fig. 3). In
summary, despite the equivalence in the expression pattern, the in
vivo footprinting data indicate that K562 cells possess an
-globin promoter structure clearly different from HeLa cells. In
the latter, the promoter is packed tightly with proteins and the DNA
structure is pronouncedly distorted. In contrast, in K562 cells, the
promoter shows an ``open'' chromatin configuration with
proteins bound only at distinct sequence elements. Therefore, we
suggest that the -globin gene in the analyzed K562 cells may not
be truly repressed, as in HeLa cells, but in an intermediate state
between repression and active transcription.Occupation of the NFI Site in the
The performed in
vivo footprint analysis provides the opportunity to examine
protein interactions at the NFI site of the -Globin Promoter
in Vivo Differs between K562 and HeLa Cells-globin promoter in
the chromosomal context of the two cell lines (cf.
introduction). In vivo footprints with K562 cells revealed no
stable protein occupation of the NFI site (Fig. 2, lane
5), in spite of the fact that the DNA region does not seem to be
particularly inaccessible due to tight protein packaging (cf.
instead the corresponding region in HeLa cells; previous section).
Since clear protection footprints are obtained only if a sufficiently
high portion of the DNA site in question is stably occupied, one reason
for the lack of NFI footprints in K562 cells could be a low rate or a
merely transient protein binding to its site. -Amanitin has been
used to visualize binding of RNA polymerase II in vivo by
trapping the enzyme at the promoter(49) . In an attempt to
enhance a hypothetical insufficient factor binding at the -globin
promoter by the same rationale, we therefore performed in vivo footprints after treatment of induced K562 cells with
-amanitin. Again, no occupation of the NFI site was detectable
(data not shown). Thus, by the methods used, the NFI site within the
-globin promoter does not become bound in K562 cells.)GCCAG; see Fig. 2, lane
1, and summary in Fig. 3). However, this protection pattern
deviates from DMS footprints of NFI made in vitro ( (9) and (14) ; see ``Discussion''). ()Therefore, it is not possible to diagnose unambiguously
whether protection of the NFI site of the -globin promoter in HeLa
cells is due to NFI binding. Nevertheless, it is clear that the NFI
site present at the -globin promoter is utilized differentially in
each cell line, and this may point toward a distinct, hitherto
unrecognized genetic function of this site in the chromosomal context. All CpG Dinucleotides in the
Binding of several eukaryotic
proteins to DNA is known to depend on the methylation state of
cytosines. Most proteins are inhibited from binding when cytosines are
methylated (at so called CpG or HTF islands, (50) ), but some
require methylated cytosines in order to
bind(51, 52) . Therefore, the distinct footprinting
pattern of the -Globin Promoter in
Both Cell Lines Are Unmethylated-globin promoter could be, at least in part, due to
cell line specific methylation of this region of the genomic DNA. In
particular, differential CpG methylation might have been a cause for
lack of binding of NFI to the NFI site in K562 cells. The HTF islands
of the 1- and 2-globin gene loci have been investigated in
several cell types and tissues(53, 54, 55) .
In fact, they have been shown to be unmethylated in K562
cells(55) , whereas the same loci appear heavily methylated in
HeLa cells(53, 54) . However, in HeLa cells also,
unmethylated sites are apparent at the promoter region, just in front
of the -globin genes(53, 54) . These
investigations which are based on the use of methylation-sensitive
restriction enzymes only detect part of the CpG dinucleotides.
Therefore, to investigate a possible effect of DNA methylation on the
protein binding at this region, we performed direct genomic sequencing
of the -globin promoter of both cell lines. We found that all
cytosines of K562 DNA, and hence all 12 CpG pairs from -90 to
-1, are not methylated (Fig. 2C, lane
7). (Again, these results mirror the methylation state of the majority of the K562 population; cf. above.) The same
cytosine residues are also not methylated in HeLa cells (Fig. 2C, lane 8). Thus, both
non--globin-expressing cell populations have equivalent cytosine
methylation patterns. From these results we conclude, that, firstly,
cytosine methylation cannot be the reason for differential protein
associations of the -globin promoter of the two cell lines.
Secondly, methylation-free CpG islands may be necessary but not
sufficient for activation of the -globin promoter. Thirdly, these
observations exclude inhibition by methylation as a possible reason for
lack of NFI binding to the -globin promoter in K562 cells.NFI Is Present and Evenly Distributed in All HeLa and
K562 Cells
NFI was described as ubiquitous protein in mammalian
cells (25 and references therein). Previous Northern blot analysis of
HeLa and K562 cells has shown in particular that both cell lines
contain at least two prominent NFI mRNAs with the same length (8.6 and
4.5 kilobases) and same relative ratio(28, 56) .
However, those types of analysis are not able to trace NFI in
individual cells. Therefore, we wondered whether the absence of an NFI
footprint at the -globin promoter of K562 cells could be
attributed to a heterogeneous distribution of NFI proteins and thus to
their absence from the majority of the cells, as was detected with
-globin (see above). We approached this question by FACS analysis
using antibodies raised against recombinant, baculovirus-expressed NFI,
consisting of amino acids 1-257(42) . The results in Fig. 4demonstrate that NFI is present and expressed in all K562
and HeLa cells. Therefore, mechanisms other than heterogeneous
expression seem to be responsible for the lack of NFI binding in
vivo at this site in K562 cells.
K562 and HeLa Cells Contain Different Forms of
NFI
NFI is expressed by several mechanisms as a family of
homologous polypeptides (cf. introduction) which may exert
distinct functions in the cell(24, 25, 26) .
Thus, differences in the protein binding pattern at the NFI site in
vivo might be attributable to differences in the NFI populations.
In order to assay different NFI forms between the two cell lines, we
performed electrophoretic mobility shift assays of whole nuclear
extracts of K562 and HeLa cells (Fig. 5A). Comparable
protein amounts of these extracts were allowed to bind the target -G wt DNA (provided in excess) which contains the NFI
site in the context of the -globin promoter. A recombinant NFI
fragment with an apparent molecular mass of 35 kDa (cf.
above and (9) ) encompassing the DNA binding domain was used in
a control reaction (
)(Fig. 5A, lane
2). Both cell extracts gave rise to protein
DNA complexes (Fig. 5A, lanes 6 and 9). However,
the two cell types differ in the number of the DNA binding species:
HeLa cells contain several (6 or 7) distinct species that form diverse
complexes with DNA (Fig. 5A, lane 6). This is
in agreement with earlier reports by us (9) and
others(24, 25, 32) , describing multiple DNA
binding forms of NFI in HeLa cells. In contrast, extracts from K562
cells form only 2-3 distinct bands in this assay, a strong one
and one or two rather faint bands, with lower and higher mobilities (Fig. 5A, lane 10). Moreover, no two distinct
bands of the HeLa and of the K562 extract accurately comigrate with
each other. We may infer, therefore, that the -G wt-DNA
binding activities in these two cell types also differ in molecular
shape and/or charge. (
)Differences in conformation, or
subunit composition, were further substantiated by proteolytic clipping
with trypsin (data not shown).
DNA complexes. 20 ng of purified recombinant NFI protein or 10
µg of whole nuclear extracts each of HeLa and K562 cells were
incubated with 5 fmol of radioactively labeled -G wt oligonucleotide. Competition of complex formation was done with
the indicated oligonucleotides (100-fold molar excess over the labeled
probe) prior to the addition of proteins and analyzed by native gel
electrophoresis and autoradiography. B, supershift analysis of
NFIDNA complexes by anti-NFI antiserum. Incubation and analysis
was performed as above. - (lanes 1, 4, 7, and 10) denotes no antiserum, 0 (lanes 2, 5, 8, and 11) denotes nonimmune serum, and
+ (lanes 3, 6, 9, and 12)
denotes anti-NFI antiserum added to the incubation reactions prior to
analysis.
-G mut.), in which the NFI site of -G wt had been inactivated(9) , has no
significant competition effect on the interaction of the control, as
well as of any polypeptide of either cell line with the NFI binding
site (Fig. 5A, lanes 5, 9, and 13, respectively). Since in this assay all shifted bands from
both cell lines show essentially the same qualitative behavior as the
control NFI protein, we conclude that all bands represent NFI protein
species capable of DNA binding. However, the bands of the K562 cell
extracts appear to be more resistant to homologous competition than
HeLa-derived complexes (Fig. 5A, cf. lanes 11 and 12 versus lanes 7 and 8); this striking
behavior points toward a possible difference in the inherent DNA
binding affinities of the NFI species in the two cell types.
Conclusively, the structural differences in the NFI populations between
the two cell lines correlate with a different DNA binding mode of these
factors to the NFI site of the -globin promoter.DNA complex running a
short distance into the gel (Fig. 5B, lane 6).
The complexes formed by HeLa and K562 nuclear extracts with -G
wt all reacted similarly with the anti-NFI antiserum in that the
intensities of all bands were reduced to the same extent as that
observed with the recombinant NFI (Fig. 5B, lanes 9 and 12). Supershifted complexes with the nuclear extracts
were not apparent here; since all NFIDNA complexes in the
extracts (Fig. 5B, lanes 7 and 10)
appear larger than the DNA complex of the recombinant NFI fragment (Fig. 5B, lane 4), we presume that the
additional increase in size caused by the antibody association
precluded entirely entering into the gel matrix. However, we consider
the impairment of the DNA binding by the anti-NFI antiserum to
sufficiently prove that the bound proteins were in fact NFI or, to a
large extent, NFI-like proteins.
-globin
transcription(28, 32, 33) . However, the
involvement of NFI in -globin expression in situ has
never been tested. To define a differential state of transcription in
order to check NFI involvement, we first compared the -globin
expression of two different human cell lines, K562 and HeLa, which are
known to be inducible or dormant for this gene, respectively. However
and to our surprise, the great majority of K562 cells, which is an
inducible cell line commonly used in globin gene research, was silent
for -globin expression and could not be induced by hemin. To our
knowledge, the heterogeneity found in the K562 cell population is
demonstrated here for the first time. This heterogeneity, with only a
minor cell fraction actively expressing -globin, is consistent
with the relatively weak RNA synthesis and inducibility found here and
by others(65, 66) , with the low level of -globin
proteins (67) and with the reduced DNase I hypersensitivity in
the -globin promoter compared to other genes in the -globin
gene cluster(68) . In the latter case, the investigators
already mentioned heterogeneity in the K562 cell population as a
speculative explanation for their findings.-globin detectably,
they differ profoundly in their respective -globin promoter
configurations in vivo. In the inactive HeLa cells, it is
highly packed with various proteins. In contrast, in the majority of
K562 cells it is scarcely bound with proteins but shows distinct
occupation at basal transcription elements. What could be the cause for
the apparent DMS hypersensitivity of the HeLa promoter between the
-IRP and ATA-box? This hypersensitivity suggests distortion of DNA
conformation which may be due to a positioned nucleosome at this locus:
x-ray analysis showed that the DNA does not wind smoothly around a
nucleosome core, but is rather bent fairly sharply or kinked at several
locations(69) . These kinks are not in direct contact with
histones and display departure from good stacking spread over several
base pairs(69) . These features would permit purine
methylation, also enhanced methylation and methylation at functional
groups that are normally not accessible in dsDNA. Indeed, DMS
modification of nucleosomal DNA in vitro reveals no apparent
periodic modulation of reactivity corresponding to the twist of the DNA
and no sites of purine protection in either DNA groove but rather sites
of enhanced reactivity(70) ; these sites were inferred to be
located within and next to a sharp bend(69) . Thus, the
methylation pattern of this particular region of the -globin
promoter in HeLa cells is entirely consistent with a kinked region of
the DNA wrapped around a nucleosome. The signals of decreased intensity
may be due to proteins other than nucleosomes, which may help to
establish the repressed state of HeLa. These characteristics are in
accordance with the transcriptional silence of this promoter in tightly
packed chromatin and may reflect the truly repressed state. In
contrast, the promoter in K562 cells very significantly lacks the DMS
hypersensitive region of that in HeLa cells, suggesting absence of
nucleosomes. This parallels earlier investigations which revealed DNase
I hypersensitivity at the human -globin promoter of K562 cells (68) indicating an open, i.e. nucleosome-free and/or
loosely packed chromatin state. Our findings are also in agreement with
the hypothesis that the locus control region keeps the globin gene
region free of nucleosomes in erythroid cells(3) . Thus, the
-globin promoter of the majority of K562 cells may define a
distinct stage of -globin expression in a cascade of activation
events, which is different from the repressed state of HeLa and of the
active state of the few expressing K562 cells. A latent transcriptional
state in the course of transcriptional activation, known as
``transcriptional competence'' from other globin (see
references cited in (4) ), but also non-globin genes (e.g. the heat shock protein 70 gene; (71) ), is associated with
increased in vivo sensitivity of chromatin to DNase I
(originating from binding of transcription factors; (4) ),
absence (72) or rephasing (73) of nucleosomes, and
relaxation of higher structural levels of chromatin, probably by
partial depletion of histone H1(74) . Therefore, from the data
quoted above and presented here, we suggest that the latent
-globin promoter in most K562 cells may be in the state of
transcriptional competence, or, alternatively, of an even earlier stage
of transition of an extinct promoter to one undergoing maximal gene
expression, because it cannot be induced.-globin expression. Since K562 is an undifferentiated
erythroleukemia cell line(75) , maybe most of the cells do not
possess all required specific DNA binding activators for -globin
gene induction. Alternatively, transcriptional
``coactivators,'' a hemin receptor, or a connecting link in a
signal transduction pathway other than a DNA binding transcription
factor may be missing, thus rendering the cells apparently
unresponsive. In the latter case, the chromatin configuration
of their -globin promoter might be equivalent to that of the few
cells that can be induced. We are currently attempting to clone
individual K562 cells with the aim to generate a homogeneous
-globin-inducible population. However, degeneration to a
heterogeneous population with respect to -globin expression may
occur as a stochastic process in cell culture.-globin transcription in vivo, we may conclude that NFI is apparently not implicated
in the preactivation phase of cellular -globin expression detected
here. In this instance, it is interesting to note that in an
investigation with the -globin promoter as a model, one NFI member
(CTF-1) has been discussed as a factor needed for antirepression in
vitro(76, 77) , counteracting the binding and
action of histone H1, rather than for subsequent direct transcriptional
activation(63) . NFI was thus suggested to act at early but not
later stages of transcriptional activation. Our system provides a
platform on which this proposal may be tested in vivo.
However, our results argue against an involvement of NFI. Apparently,
NFI does not belong to the factors which prepare the -globin
promoter for active transcription and, thus, does not contribute to
cell-specific expression at this stage in vivo.-globin
reproducibly show a clear occupation of the NFI site (contacts at the
underlined Gs in GGG(N)GCCAG). However, it is difficult to
diagnose whether this site is occupied in fact by NFI, because in
vitro the pattern of methylation protection of the consensus
sequence is different (underlined in TGG(N)GCCAA; (14) ). When methylated in vitro, the
underlined residues in GGG(N)GCCAG interfere with binding
in the particular NFI site of the -globin promoter (see Fig. 3and (9) ). To our knowledge, no other DMS
footprint of NFI in vivo and no DMS protection footprints of
the -globin NFI site in vitro have been ever reported
which could be compared directly with our results. However, the
following in vitro data point toward a possible involvement of
NFI. The first residue in the consensus is also protected by NFI (14, 43) and the third residue (G) is less affected by
methylation(9) ; also, the first G of the second half is
considered by some authors not to be necessary for interaction with NFI (e.g.(27) and (59) ). There are also
examples of other proteins that give similar but different footprint
patterns in vivo than in vitro(78) . For
these reasons, the in vivo pattern indeed may have been caused
by NFI protein species.-globin promoter in HeLa cells would be consistent with a
repressor function in this system. This hypothesis is in agreement with
the results of Treisman et al.(79) , who reported a
(weak) expression of -globin after transient transfection of
reporter plasmids into HeLa cells, which normally keep the endogenous
gene silent. In this study, -globin (as opposed to -globin)
expression was shown to be extremely sensitive to the copy number of
plasmids. This deregulated expression would be consistent with a
repressor function of NFI, since titrating out repressing NFI molecules
would allow escape synthesis of -globin. NFI has been repeatedly
discussed as a repressor of transcription also in other systems (16,
59, 60, 62 and references cited in (5) and (49) -52). Particularly, from analyses of other globin
genes it has been speculated that displacement of positive factors by
NFI is a general globin gene regulatory mechanism (80 and references
therein; see also (22) ). Regardless of the identity of the
protein which interacts with the NFI site of the -globin promoter,
occupation in the repressed state indicates that this site might be a
negative cis-acting element in the -globin expression. -globin
NFI site at all? Making some plausible assumptions, ()we
estimated that, theoretically, NFI in both cell lines should give
almost complete coverage of all sites, provided it were freely
diffusible and the DNA fully accessible. Therefore, we think that NFI
should have a real chance to bind to the -globin promoter in the
nucleus yielding visible in vivo footprints. However, our
results suggest that the NFI site, at least in K562 cells, is not
bound. The presence of a factor in nuclear extracts contrasted by the
lack of in vivo binding to regulatory regions has been
reported for several genes, including the HLA-B7 transgene, the
endogenous H-2K
gene, the tyrosine aminotransferase gene,
MHC class II genes, and the muscle creatine kinase gene (83 and
references cited therein). In these examples, the reasons for
transcription factors being precluded from interaction with their
respective sites in the nucleus are not clear. Why should the NFI site
of the -globin promoter not be recognized by NFI in vivo?
We can think of several reasons; for example, occupation of the site by
other factors or a specific orientation of the site toward a positioned
nucleosome (cf. (35) ), which renders it inaccesible
for NFI; lack of additional factors, which promote or stabilize NFI
binding to -globin promoter; cytosine methylation inhibiting NFI
from binding(84, 85) ; elusive binding of NFI;
heterogeneous cellular distribution of NFI; active or passive
sequestration of NFI to particular regions of the nucleoplasm
(heterogeneous nuclear distribution; see (86) and (87) ); or modified binding characteristics of the
polypeptides. We could preclude heterogeneous cellular distribution of
NFI, C methylation of the NFI site, and possibly elusive
binding of NFI as reasons for the lack of in vivo footprints
in K562 cells. Additionally, from the in vivo analysis we have
no indication of an inhibition of binding due to other factors or
nucleosomes. However, we do get preliminary evidence suggesting that
NFI polypeptides from the two cell lines, which differ in their
structure and composition, might also have different DNA binding
affinities. Binding experiments in vitro revealed that NFI
from K562 is apparently more insensitive toward homologous competition
than HeLa NFI (Fig. 5A). This unexpected binding
behavior may be explained by the assumption that large amounts of NFI
in K562 cells possess lower binding affinity compared to that of HeLa
cells, thus not forming visible protein
DNA complexes under the
applied conditions. Adding more DNA would then mobilize this
supplementary protein compensating apparently for homologous
competition. Studies are currently in progress to validate this
hypothesis.
))-globin in either case.))-globin promoter (9). It was
expressed from a baculovirus-based vector in Sf9 cells; it consists of
amino acids 1-257 (about 35 kDa) and is thus a truncated version
of a full-length polypeptide (reported molecular masses of functional
NFI species 52-66 kDa, see (32) ). This recombinant NFI
encompasses the DNA binding and dimerization domain, which is highly
homologous between different NFI species, and which confers all DNA
binding characteristics. In in vitro binding experiments and
footprints, it interacts with the NFI binding site in an
indistinguishable manner when compared to full-length polypeptides (9).)) to 10 (own
calculations based on quantitative electrophoretic mobility shift assay
and footprinting experiments), and the number of all canonical NFI
sites per diploid nucleus was estimated to be approximately 1 out of
every 100 kilobases (81), i.e. in total about 10 per genome. This corresponds to 100 accessible sites, assuming
that about 1% thereof resides in open chromatin (4). Allowing for
mutations at the two positions of the NFI site found in the
-globin promoter reading 5`-GGG(N)GCCAG-3`, which
differ from the canonical sequence, there may be approximately 1600
NFI-like sites in the same nucleus, with which these NFI molecules may
interact with comparable affinity. Thus, there is numerically
sufficient NFI that in principle could interact with them to give
extensive coverage and thus clear protections (protein to DNA ratio
= 6:1 to 60:1). To which extent this happens depends on the NFI
affinity to these sites and on the actual concentrations of the
reaction partners. The affinity of NFI for the site in the -globin
promoter was previously determined quantitatively (41) and was found to
be 2 10
M, i.e.
50-fold weaker than that of the consensus. Thus, NFI
interacts with lower affinity with the NFI site of the
-globin
promoter than with the canonical NFI site. However, its affinity to the
-globin site approximates, for example, that of the trp repressor (82), a specifically DNA-binding protein. By using the
above estimates and equilibrium binding constant, we obtain then for an
average nucleus of 10
µl volume, a
concentration of NFI-like sites of 27.2 nM and an NFI
concentration of 0.17 µM, yielding 96.6% coverage of all
sites, or an NFI concentration of 1.7 µM, yielding 99.7%
coverage of all sites.
Oligonucleotides were synthesized by G. Arnold
(Laboratory for Molecular Biology-Genzentrum, Martinsried). Whole cell
extracts from uninfected Sf9 cells were kindly provided by M.
Stanglmaier (Institute for Biochemistry, Martinsried). We are indebted
to H. Ibelgaufts and M. Müller for critical reading
of the manuscript and valuable comments and suggestions.
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Sawczenko, O. Azooz, J. Paraszczuk, M. Idestrom, N. M. Croft, M. O. Savage, A. B. Ballinger, and I. R. Sanderson Intestinal inflammation-induced growth retardation acts through IL-6 in rats and depends on the -174 IL-6 G/C polymorphism in children PNAS, September 13, 2005; 102(37): 13260 - 13265. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. S. Madsen, J. C. Hershey, M. B. Hautmann, S. L. White, and G. K. Owens Expression of the Smooth Muscle Myosin Heavy Chain Gene Is Regulated by a Negative-acting GC-rich Element Located between Two Positive-acting Serum Response Factor-binding Elements J. Biol. Chem., March 7, 1997; 272(10): 6332 - 6340. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. E.M. Furlong, N. K. Keon, F. D. Thornton, T. Rein, and F. Martin Expression of a 74-kDa Nuclear Factor 1(NF1) Protein Is Induced in Mouse Mammary Gland Involution. INVOLUTION-ENHANCED OCCUPATION OF A TWIN NF1 BINDING ELEMENT IN THE TESTOSTERONE-REPRESSED PROSTATE MESSAGE-2/CLUSTERIN PROMOTER J. Biol. Chem., November 22, 1996; 271(47): 29688 - 29697. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Molecular and Cellular Proteomics |
| Journal of Lipid Research | ASBMB Today |