Báo cáo khoa học: A human-specific TNF-responsive promoter for Goodpasture antigen-binding protein potx

A human-specific TNF-responsive promoter for
Goodpasture antigen-binding protein
´
´
´
Froilan Granero*, Fernando Revert, Francisco Revert-Ros, Sergio Lainez, Pilar Martınez-Martınez
and Juan Saus
´
´
Centro de Investigacion Prıncipe Felipe, Valencia, Spain

Keywords
Goodpasture antigen-binding protein; tumor
necrosis factor; Goodpasture disease;
bidirectional promoter; DNA polymerase j
Correspondence
´
´
J. Saus, Centro de Investigacion Prıncipe
Felipe, Avenida de la Autopista del Saler
16 C ⁄ En Proyecto, 3 (Camino de las
Moreras), 46013 Valencia, Spain
Fax: +34 96 3289701
Tel: +34 96 3289680
E-mail: jsaus@ochoa.fib.es
Present address
*Center for Matrix Biology Vanderbilt
University Medical Center, Nashville, TN,
USA
(Received 8 July 2005, accepted 19 August
2005)

doi:10.1111/j.1742-4658.2005.04925.x

The Goodpasture antigen-binding protein, GPBP, is a serine ⁄ threonine
kinase whose relative expression increases in autoimmune processes. Tumor
necrosis factor (TNF) is a pro-inflammatory cytokine implicated in autoimmune pathogenesis. Here we show that COL4A3BP, the gene encoding
GPBP, maps head-to-head with POLK, the gene encoding for DNA
polymerase kappa (pol j), and shares with it a 140-bp promoter containing
a Sp1 site, a TATA-like element, and a nuclear factor kappa B (NFjB)-like
site. These three elements cooperate in the assembly of a bidirectional transcription complex containing abundant Sp1 and little NFjB that is more
efficient in the POLK direction. Tumour necrosis factor cell induction is
associated with Sp1 release, NFjB recruitment and assembly of a complex
comparatively more efficient in the COL4A3BP direction. This is accomplished by competitive binding of Sp1 and NFjB to a DNA element
encompassing a NFjB-like site that is pivotal for the 140-bp promoter to
function. Consistently, a murine homologous DNA region, which contains
the Sp1 site and the TATA-like element but is devoid of the NFjB-like
site, does not show transcriptional activity in transient gene expression
assays. Our findings identify a human-specific TNF-responsive transcriptional unit that locates GPBP in the signalling cascade of TNF and substantiates previous observations, which independently related TNF and
GPBP with human autoimmunity.

Goodpasture disease and other more common natural
autoimmune disorders have been described only in
humans. Goodpasture antigen-binding protein (GPBP)
is a serine ⁄ threonine kinase that specifically targets a
motif in the NC1 domain of the human a3 chain of
type IV collagen (the Goodpasture autoantigen), which
is highly divergent and does not exist in other species
or human homologous domains [1,2]. GPBP shows
preferential expression in cells and in association with
tissue structures that undergo common autoimmune
attacks – including the glomerular basement membrane

targeted in Goodpasture disease [2]. Moreover, an augmented expression of GPBP with respect to its alternative spliced variant GPBPD26 has been associated
with a number of autoimmune responses [2], all which
suggests that GPBP is relevant in human autoimmune
pathogenesis.
COL4A3BP, the gene encoding GPBP, is headto-head with POLK, the gene encoding DNA polymerase kappa (pol j; GenBank accession number
AB036934). Pol j and its prokaryote counterpart,
pol IV, display low fidelity, moderate processivity, and

Abbreviations
EMSA, electrophoretic mobility shift assay; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GPBP and GPBPD26, Goodpasture
antigen-binding protein and its alternative splicing variant devoid of the exon-encoded 26-residue motif; NFjB, nuclear factor kappa B;
NR, nonrelevant; pol j, DNA polymerase kappa; siRNA, small interfering RNA; TAF, TBP-associated factor; TAF, TBP-associated factors;
TBP, TATA-binding protein; TNF, tumor necrosis factor.

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

5291


TNF transcriptional induction of GPBP

extend mispaired DNA by misaligning primer–template
to generate )1 or )2 frameshift products [3–7]. Furthermore, pol j can bypass DNA lesions in both
error-prone and error-free manners [7–10]. All of this
suggests that pol j is a DNA polymerase that not
only has a role in the cellular response to DNA
damage, but also plays a part in spontaneous mutagenesis by facilitating base pairing at aberrant replication forks.
Tumour necrosis factor (TNF)-a and -b are two different cytokines with similar biological effects that are
primarily secreted by macrophages and Th1 cells in

response to various inflammatory stimuli (reviewed in
[11]). Of all the known biological effects of TNF, the
cytotoxicity seems to be the most prominent, and thus
TNF is the mediator whereby cytolytic immune cells
induce fatal injury to their targets. However, overproduction of TNF in concert with other cytokines often
has negative consequences for the host, for example in
various autoimmune and immune-mediated rheumatologic disorders.
Here we show that TNF increases the mRNA levels
of GPBP and induces COL4A3BP expression from
a promoter that in its constitutive mode transcribes
POLK more efficiently. This is accomplished, at least in
part, by Sp1 release from – and NFjB recruitment
to ) a bidirectional complex that binds to several DNA
elements including one novel element that encompasses
a nuclear factor kappa B (NFjB)-like site and supports
competitive binding of the two transcription factors.
Moreover, we show that the mouse DNA counterpart
is devoid of this versatile element, thereby uncovering
major differences in GPBP expression between human
and mouse that may be important in understanding
human-specific autoimmune pathogenesis.

Results
Tumour necrosis factor increases the mRNA
levels of GPBP in human cells
Human cultured HEK293 and hTERT-RPE1 cells
were induced with human recombinant TNF-a, and
the levels of GPBP mRNA were estimated by a reverse
transcriptase-real time PCR approach (DDCt) (Fig. 1).
We observed an overall increase in the mRNA levels

of GPBP in response to TNF in both cell types. However, the mRNA levels increased more promptly in
hTERT-RPE1 reaching three- to fourfold after 1 h of
induction whereas 3 h of induction were required to
reach 2- to 2.5-fold levels in HEK293 cells. Our findings suggest that TNF induces GPBP transcription in
human cells.
5292

F. Granero et al.

Fig. 1. TNF increases the mRNA levels of GPBP in cultured human
cells. HEK293 (293) or hTERT-RPE1 (RPE1) cells were cultured in
the presence or in the absence of TNFa during the indicated periods of time, disrupted, and total mRNA extracted and subjected to
reverse transcriptase-coupled real-time PCR to estimate the levels
of mRNA by the DDCt method. The values represent fold of
induced vs. noninduced cells after normalization with the corresponding values for GAPDH. We represent the mean of three independent experiments carried out in duplicate ± SD. The mRNA
levels of GAPDH were not affected by cytokine induction.

Identification of a 140-bp bidirectional promoter
in the intergene region of COL4A3BP and POLK
Using cDNA mapping we identified a 140-bp fragment
in the intergene region of COL4A3BP and POLK
(Fig. 2A). To investigate the presence of a bidirectional
promoter in this 140-bp DNA region, we cloned each
of the two orientations of a 772-bp fragment encompassing the region of interest in pFGH (LpromGPBP
and LpromPolj) and assessed its capacity to drive
heterologous gene expression in mammalian cells
(Fig. 2B). The 772-bp fragment efficiently promoted
transcription in each direction, 21-fold for COL4A3BP
and 25-fold for POLK. When we assessed the transcriptional activity of the 140-bp region (SpromGPBP
and SpromPolj), we observed a reduction in the activity that was more evident for COL4A3BP than for the

POLK direction, a 45% reduction vs. 18%. This indicates that although the 140-bp region contains the core
of a bidirectional transcriptional unit and the structural requirements for divergent transcription, there
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS


F. Granero et al.

TNF transcriptional induction of GPBP

A

B

C

Fig. 2. Identification of a bidirectional promoter for human POLK and COL4A3BP and transcription start sites mapping. (A) The one-strand
sequence between ON-GPBP-18m and ON-GPBP-5c is written in upper case. Indicated are the position and sequence of oligonucleotides
used for DNA isolation and the restriction sites used to generate constructs for experimental procedures in (B) and (C). Open and closed
boxes contain DNA sequences present in the indicated pol j (GenBank accession number AF163570) and GPBP (GenBank accession
number AF136450) cDNA clones of which the 5¢-ends and transcription direction are denoted by arrows. Highlighted in grey is the 140-bp
fragment present in Sprom constructs. Bent arrows indicate the transcription start sites mapped in (C). (B) Cells were transfected with either
Lprom (L) or Sprom (S) constructs along with a b-galactosidase-expressing vector. Results are expressed as the quotient (fold) of the reporter gene expression of the promoter constructs vs. empty vector previously normalized with the corresponding b-galactosidase expression
values. Values represent the mean of two independent experiments carried out in duplicate ± SD. (C) Antisense ribonucleotide probes
(2 · 105 cpm) representing the indicated DNA regions were mock digested (1) or digested with RNase in the presence of yeast (2) or human
(3) total RNA extracted from hTERT-RPE1 cells (COL4A3BP) or from HEK293 (POLK). The amounts of RNA used were 40 lg (COL4A3BP)
or 80 lg (POLK). After digestion of the single-strand RNA the mixtures were analysed by denaturing PAGE and autoradiographed. The
numbers and bars indicate the size in nucleotides and position of radiolabelled ribonucleotide markers.

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

5293


TNF transcriptional induction of GPBP

are regulatory elements in the flanking regions that
modulate both transcriptional activity and direction.

F. Granero et al.

A

Mapping of transcriptional start sites for
COL4A3BP and POLK
We used RNA-protection assays to map the transcriptional start sites for each gene (Fig. 2C). When
32
P-labelled RNA probes representing the antisense
strand of COL4A3BP or POLK between the ApaI and
EclXI sites (Fig. 2A) were hybridized with human
RNA, three and two fragments, respectively, were protected from RNase digestion (Fig. 2C), thereby revealing the existence of multiple mRNA species for GPBP
and pol j. To further map the 5¢ end of the individual
mRNAs, we similarly assayed antisense RNA probes
representing DNA sequences which extend into the
corresponding gene (BamHI ⁄ ON-GPBP-5c and ONpolk-2m ⁄ EclXI). These resulted in larger protected
fragments displaying sizes that were consistent with the
extended sequences (Fig. 2C) and thus mapping the 5¢
end of the different mRNA species at the positions
indicated in Fig. 2A.
All of these findings suggest that the DNA region
identified by genomic cDNA mapping is at the intergene region and contains the structural requirements

for bidirectional transcription.

B

C

Identification and characterization of transcriptional DNA elements in the 140-bp promoter
Using matinspector V2.2 from the transfac 4.0 program [12], we have identified three transcriptional elements in the DNA region of interest. Two elements
commonly found associated with transcriptional initiation in TATA-less promoters, a canonical Sp1 binding
site and a TATA-like element [13–15], and a NFjB-like
binding site for transcriptional regulation (Fig. 3A).
The contribution of individual DNA elements in
transcription was first assessed in 140-bp promoter
constructs in which elements expected to be associated
with transcriptional initiation were deleted (Fig. 3B).
Deletion of each individual element had negative consequences on transcription although impacted differently each orientation. Deletion of the Sp1 site (DSp1)
resulted in major transcription impairment in both
directions ( 60–70%) whereas TATA-like element
deletion (DTATA) significantly reduced transcription
in the COL4A3BP direction ( 60%) but had a lesser
negative effect on the POLK direction ( 30%).
Finally, double deletion (DSp1 ⁄ TATA) reported additional transcriptional activity reduction only in the
COL4A3BP direction, in which case the promoter dis5294

Fig. 3. Identification of transcriptional DNA elements in 140-bp promoter. (A) The one-single strand sequence of the 140-bp promoter
is written in upper case and underlined are the DNA elements identified using the MatInspector V2.2 from the TRANSFAC 4.0 program. (B,C) Cells were transfected with SpromGPBP or SpromPolj
(WT) or with mutants thereof in which the NFjB-like (NFjB) and ⁄ or
Sp1 (Sp1) binding sites and ⁄ or TATA-like element (TATA) were
deleted (D). Transcriptional activities were normalized as in Fig. 2B
and results are expressed as percent activity with respect to WT

sequence, which was set at 100%, and are the mean ± SD of
three experiments carried out in duplicate.

played a transcriptional activity ( 20%), which was
only slightly above the values observed for the empty
vector (7–12%).
This suggests that although the two elements are
used in the two directions, COL4A3BP transcription
depends more on TATA-like element than POLK,
which depends mainly on the Sp1 site. The evidence
also suggests that the TATA-like element and the Sp1
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS


F. Granero et al.

site are coordinated when transcribing COL4A3BP
as the relative contribution of each element to total
activity is very similar and too high to be fully independent.
To assess the role of the NFjB-like site in transcription, the multiple promoter constructs were further
deleted and similarly assayed (Fig. 3C). Deletion of the
NFjB-like site (DNFjB) greatly reduced transcription
in the COL4A3BP direction (80%) and little in the
POLK direction ( 30%). Additional deletion of the
TATA-like element (DNFjB ⁄ TATA) did not reduce
transcriptional activity, while additional deletion of the
Sp1 site (DNFjB ⁄ Sp1) rendered the promoter virtually
inactive in the COL4A3BP direction and with an activity in the POLK direction slightly above empty vector.
Triple deletion mutants were not significantly different
from DNFjB ⁄ Sp1 mutants.

Collectively these results suggest that NFjB- and
TATA-like sites are coordinated elements that account
for  80% of transcription in the COL4A3BP direction whereas the Sp1 site accounts for the remaining
 20%. In contrast the Sp1 site accounts for  70% of
transcriptional activity in the POLK direction and the
remaining  30% depends on NFjB- and TATA-like
sites.
Of special interest was the observation that the
double deletion DNFjB ⁄ TATA rendered a promoter
with transcriptional activities in the POLK direction
that were similar and even higher than the wild-type
promoter, while single deletions mutants (DNFjB or
DTATA) rendered promoters with less activity ( 30%
reduction), suggesting that Sp1 site and coordinated
NFjB ⁄ TATA-like elements compete for transcription
factors.
Sp1- and NFjB-based transcriptional complexes
coexist and compete for DNA elements in the
140-bp promoter
Collectively our data suggest that there are two bidirectional transcriptional complexes, one binding to
the Sp1 site which is more efficient in the POLK direction and another operating through NFjB-like and
TATA-like sites (NFjB-TATA-dependent) comparatively more efficient in the COL4A3BP direction. Both,
however, coexist and compete for the promoter suggesting that the two complexes bind to all three sites.
To test this hypothesis we performed electrophoretic
mobility shift assay (EMSA) using nuclear extracts
and double-stranded synthetic oligonucleotides representing the DNA elements and flanking regions of
interest (NFjB-L, SP1 and TATA-L) (Fig. 4). Individual 32P-labelled DNAs were bound to nuclear proteins
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

TNF transcriptional induction of GPBP

Fig. 4. Mapping transcription complex binding sites in the 140-bp
promoter using EMSA. On the left in the composite and in upper
case is the two-strand nucleotide sequence of the 32P-labelled
probe used in EMSA as shown on the right. Each sequence represents the indicated element (parallel lines) and flanking regions in
the 140-bp promoter (NFjB-L, SP1, and TATA-L). The 32P-labelled
DNA fragments were incubated in the presence (+) or in the
absence (–) of nuclear extracts (NE) and the indicated competitors
(Comp). NR is a nonrelevant DNA. The resulting complexes were
analysed by nondenaturing PAGE and autoradiographed. With
arrows we indicate the position of the retarded autoradiographic
bands referred to in the text. In the composite, the individual samples from the same study and exposure have been placed in a different order than in the original gel for illustrative purposes.

to generate complexes whose assembly was efficiently
inhibited by the corresponding unlabelled DNA but
not by similar concentrations of a nonrelevant DNA
(NR), suggesting that nuclear proteins form complexes
with DNA in a sequence-dependent manner. Interestingly, individual unlabelled DNA elements showed
capacity to inhibit or alter the binding of nuclear proteins to the other 32P-labelled elements. In this regard,
NFjB-L was the most efficient as it completely inhibited the formation of one [32P]SP1 ⁄ protein complex
(top arrow), strongly inhibited another (bottom
5295


TNF transcriptional induction of GPBP

arrow), and fully inhibited two of the three complexes
associated with [32P]TATA-L (top and bottom arrows).
In contrast, SP1 showed the least effect on protein
binding to other elements and inhibited the formation

of one of the complexes associated to [32P]NFjB-L
(top arrow) without significantly impacting the other
(bottom arrow), and only caused an alteration in the
relative abundance of individual complexes associated
with [32P]TATA-L (top and bottom arrows). Finally,
TATA-L efficiently inhibited formation of one complex associated with [32P]NFjB-L (bottom arrow)
without significantly impacting the other (top arrow)
and partially inhibited formation of the two [32P]SP1
complexes (top and bottom arrows).
Of special interest was to find that [32P]NFjB-L
associates with nuclear proteins to form two types of
independent complexes, one that can (bottom) and
another that cannot (top) be assembled in the presence

F. Granero et al.

of SP1, suggesting that Sp1 and NFjB transcription
factors compete for NFjB-L. The evidence also suggests that there are two complexes that bind to all
three DNA elements, although in different manners.
Sp1-based complex preferentially binds to SP1 and displays lower affinity for NFjB-L and still much lower
for TATA-L. The NFjB-based complex preferentially
binds to NFjB-L and displays lower but similar
affinity for SP1 and for TATA-L.
Tumour necrosis factor induces asymmetric transcription from the 140-bp promoter
To assess TNF induction of the 140-bp promoter, we
performed transient gene expression assays (Fig. 5A).
Tumour necrosis factor cell stimulation resulted in
promoter induction, but this was more evident in the
COL4A3 BP (three- to fourfold) than in the POLK

A

B
Fig. 5. TNF induces transcription from the
140-bp promoter. (A) Cells were transfected
with SpromGPBP or SpromPolj along with a
b-galactosidase-expressing vector and were
induced or not with TNF. Results are
expressed as the quotient (fold) of the reporter gene expression of the induced vs. noninduced cells after value normalization with
the corresponding b-galactosidase expression values. We represent the mean of four
independent experiments carried out in duplicate ± SD. (B) Cells were transfected with
SpromGPBP or SpromPolj (WT) or with
mutants thereof, in which the NFjB-like
(NFjB) and ⁄ or Sp1 (Sp1) binding sites and ⁄ or
TATA-like element (TATA) were deleted (D),
were induced with TNF (+) or not induced (–),
and transcriptional activities estimated and
expressed as in Fig. 3. We represent the
mean of three experiments carried out in
duplicate ± SD.

5296

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS


F. Granero et al.

(one- to twofold) direction. Similar results were
obtained when promoter transactivation was assessed

in NIH3T3 (shown) or in HEK293 cells using a different reporter gene expression system (data not shown).
To determine the contribution of individual DNA
elements in transcriptional induction, we performed
similar assays on induced and noninduced cells using
specific deletion mutants (Fig. 5B). We found no major
differences in the overall relative contribution of each
of the two main transcriptional activities (NFjB
dependent and Sp1 dependent) between constitutive
and induced modes of expression in the POLK direction. In contrast, deletion of NFjB-like (DNFjB) or
Sp1 (DSp1) sites had significantly less negative impact
on induced than on constitutive expression in the
COL4A3BP direction, although double deletion
(DNFjB ⁄ Sp1) was still very efficient at reducing transcription to values slightly above background. Additional
deletion
of
the
TATA-like
element
(DNFjB ⁄ Sp1 ⁄ TATA) did not further reduce transcriptional activity in the COL4A3BP direction, but double
and triple mutants had an unusual induction in the
POLK direction that was not observed in single
mutants.
These results suggest that TNF induces transcription
from the 140-bp promoter preferentially in the
COL4A3BP direction involving the same DNA elements than when transcribing in a constitutive mode.
TNF induction of the 140-bp promoter involves
Sp1 release and NFjB recruitment from a
transcriptional complex that binds to all three
elements
To further characterize the molecular mechanism of

TNF induction, the transcriptional complex associated
with the 140-bp promoter in each expression mode
was isolated and analysed by western blot using specific antibodies (Fig. 6A). The transcriptional complex
representing constitutive expression contained abundant Sp1 and traces of NFjB which were only detectable after longer developing times (not shown),
whereas the transcriptional complex representing
induced expression contained relatively less Sp1 and
more NFjB (140-bp), thereby suggesting that TNF
induction is mediated by Sp1 release from – and NFjB
recruitment to ) the transcriptional complex. We
obtained similar conclusions when transcription complexes were purified from nuclear extracts of NIH3T3
cells (data not shown).
Major transcriptional induction occurs in the
COL4A3BP direction which depends mainly on a
NFjB-like site. To explore the mechanism of induction
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

TNF transcriptional induction of GPBP

A

B

Fig. 6. Characterization of transcription complex associated to the
140-bp promoter isolated by affinity chromatography procedures.
(A) At the top we represent the 140-, 76- or 26-bp promoters with
horizontal bars and recognized transcriptional elements therein in
black. At the bottom, nuclear extracts from noninduced (–) or TNFinduced (+) cells were subjected to affinity chromatography on columns containing the indicated promoter fragments and the bound
proteins eluted and analysed by western blot using antibodies specific for the indicated transcription factors. (B) The indicated
amounts of nuclear extracts (NE) from noninduced (–) or TNFinduced (+) cells or proteins thereof bound to the indicated DNA
fragments were analysed by western blot using antibodies specific

for the indicated transcription factors.

further, we purified and similarly analysed transcription complexes bound to different promoter fragments
and to the corresponding fragments devoid of the
NFjB-like site (Fig. 6A). Deletion of the NFjB-like
site yields a promoter that binds transcription factors
in a similar fashion to the 140-bp promoter (140bpDNFjB). A 76-bp promoter devoid of the TATA-like
element (76-bp) displayed a relatively higher binding
capacity for Sp1 than the 140-bp promoter, but induc5297


TNF transcriptional induction of GPBP

tion was still associated with a net Sp1 release and
NFjB recruitment. When the NFjB-like element was
deleted in the 76-bp promoter (76-bpDNFjB) the Sp1
binding pattern remained essentially unchanged but
NFjB recruitment associated with induction was significantly impaired. In contrast, a promoter only containing the NFjB-like site (26-bp) showed a reduced
Sp1 binding capacity, and the recruitment of NFjB
upon induction was not associated with Sp1 release.
Finally, specific deletion of the NFjB-like binding site
from the 26-bp promoter resulted in a DNA fragment
without capacity to bind either Sp1 or NFjB (26bpDNFjB).
Subsequently, the contribution of individual DNA
elements in transcriptional induction was also investigated (Fig. 6B). We first determined the consequences
of induction on the nuclear levels of specific transcription factors and found that TNF increased the nuclear
levels of all members of the NFjB family and slightly
decreased the presence of Sp1 in the nucleus, while the
levels of TATA-binding protein (TBP), a general transcription factor critical for transcription preinitiation
complex assembly, remained virtually unchanged.

Transcriptional complexes associated with individual
DNA elements from noninduced nuclear extracts contained abundant Sp1 and only traces of individual
NFjB members. Furthermore, the transcriptional complex associated with NFjB-L contained traces of TBP
whereas SP1 and TATA-L transcriptional complexes
contained abundant TBP. Cell induction resulted in
recruitment of virtually all NFjB transcription factor
members into each individual transcriptional complex
without evident reduction in the amount of Sp1 and
TBP. The minor differences in the content of NFjB
members among individual transcription complexes
probably reflect different affinities of a common transcriptional complex for individual elements along with
different protein–protein interactions, rather than different transcription complexes being assembled in the
140-bp promoter.
Collectively our data reveal the existence of at least
two functionally different transcriptional complexes
that bind to all three DNA elements, one containing
Sp1 and little NFjB which represents the constitutive
mode of expression, and the other containing abundant NFjB and comparatively less Sp1 which represents the induced mode. Our findings also suggest that
the Sp1 binding site is most critical for Sp1 recruitment
whereas the NFjB-like binding site is the critical element for NFjB recruitment. Consequently, net loss of
Sp1 and gain of NFjB during induction requires both
NFjB-like and Sp1 sites. Our data also provide further
evidence for the NFjB-like site to be a regulatory ele5298

F. Granero et al.

ment bearing binding sites for NFjB and Sp1, and for
the Sp1 site and TATA-like element to contain transcriptional initiators.
NFjB-like binding site is critical for adequate
positioning of Sp1 in both constitutive and

induced transcription complexes
Our studies also reveal the existence of two major
binding sites for Sp1 located at NFjB-like and Sp1
sites (vide supra). To further explore the role of each
site in Sp1 binding and assembly, we used Sp1 antibodies (aSp1) and deletion mutants of the 76-bp
promoter in EMSA (Fig. 7). Nuclear proteins

Fig. 7. NFjB-like site is critical for adequate Sp1 assembly in the
transcription complex. 32P-labelled 76-bp promoter or mutants
thereof in which the NFjB-like (NFjB) and ⁄ or Sp1 (Sp1) binding
sites were deleted (D), were incubated in the presence (+) or in the
absence (–) of nuclear extracts (NE) from TNF-induced (+) or noninduced (–) cells and Sp1 specific antibodies (aSp1). The resulting
complexes were analysed by nondenaturing PAGE and autoradiographed for 16 h (top row) or for 40 h (bottom row). We denote by
arrows the supershifted or inhibited autoradiographic band associated with antibody presence.

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS


F. Granero et al.

from either induced or noninduced cells were bound
to 32P-labelled 76-bp promoter retarding its migration in PAGE (gel shift) in a similar fashion. The
presence of aSp1 in the binding mixtures resulted in
a gel supershift that was more evident in noninduced
nuclear mixtures (arrow in 76-bp). Deletion of
NFjB-like binding site (76-bpDNFjB) resulted in a
promoter that still bound nuclear proteins; however,
aSp1 inhibited protein–DNA complex formation (gel
shift inhibition), rather than inducing a gel supershift
(arrow), suggesting that transcription factor assembly

is largely defective. In contrast, deletion of the Sp1
site yielded a promoter (76-bpDSp1) that bound nuclear proteins and aSp1 in a similar fashion as the
76-bp promoter (arrow). Although in this case, the
promoter appears to bind Sp1 less efficiently (longer
exposure times were required to visualize antibodydependent supershift bands) and the relative abundance of supershift bands was similar between
noninduced and induced nuclear extracts. Finally,
double deletion resulted in a DNA that still formed
complexes with nuclear proteins, but these were
not reactive with the specific antibodies (76bpDNFjB ⁄ DSp1).
These results suggest that in the transcription of the
140-bp promoter the Sp1 site is critical for effective
Sp1 recruitment and release whereas the NFjB-like
site is critical for adequate positioning of Sp1 in the
transcriptional complex.
Sp1 and NFjB are components of the transcriptional complex associated with the 140-bp
promoter ex vivo
To investigate the physiological relevance of our
findings mRNA silencing (expression of siRNA),
western blot and ChIP assays were combined to
investigate ex vivo binding of Sp1 and NFjB into
the 140-bp promoter (Fig. 8). Relevant antibodies
(aSp1, ap65) specifically precipitated the 140-bp promoter from noninduced or induced cells as PCR
procedures efficiently amplified the DNA region of
interest from the corresponding immunoprecipitates
but not from precipitates obtained with irrelevant
antibodies (aNR). However, aSp1 was relatively
more efficient in precipitating the 140-bp promoter
from the chromatin of noninduced cells whereas
ap65 was relatively more efficient at precipitating it
from the chromatin of induced cells. As expected,

expression of individual siRNAs resulted in substantial reduction in the cellular levels of the corresponding
transcription
factors
(Immunoblot);
and
consequently, in the capacity of the corresponding
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

TNF transcriptional induction of GPBP

Fig. 8. Ex vivo binding of Sp1 and NFjB to the 140-bp promoter.
HEK293 cells were subjected to transfection in the absence (–) or
in the presence of the indicated small interfering RNA (siRNA) and
either TNF-induced (+) or not induced (–) and used for either western blot (Immunoblot) or ChIP assays using the indicated specific
antibodies (aSp1 or ap65). Similar amounts of individual immunoprecipitates were subjected to PCR and resulting mixtures were
analysed by agarose gel electrophoresis and ethidium bromide
staining (ChlP). For control purposes, a nonrelevant antibody was
included in the ChIP assays and the corresponding immunoprecipitates (aNR) and supernatants (DNA input) PCR-amplified and analysed in parallel.

specific antibodies to precipitate the 140-bp promoter
(ChIP). This suggests that reduction of cellular levels
of individual transcription factors efficiently reduces
transcription factor–promoter binding. Interestingly,
reduction in the expression of one transcription factor affected promoter precipitation by the other transcription factor-specific antibodies, although in a
different manner depending on expression mode. In
noninduced cells, reduction of cellular levels of one
transcription factor was associated with increased
promoter precipitation by the other transcription factor-specific antibodies. This effect was more evident
when reducing the expression of Sp1 than when
reducing the expression of p65 in consonance with

the prevalence of Sp1 vs. p65 in the constitutive
mode of transcription. In contrast, in TNF-induced
cells, reduction in the level of expression of one individual transcription factor was associated with a
major reduction in promoter precipitation by the
other transcription factor-specific antibodies. This
effect was more evident when reducing the expression of p65 than when reducing the expression of
Sp1 in consonance with the prevalence of p65 vs.
Sp1 in the induced mode of transcription.
These findings reveal that the two transcription
factors are part of the transcriptional machinery
associated with the 140-bp promoter ex vivo,
although Sp1 and p65 are more prominent in constitutive and induced modes of expression, respectively.
5299


TNF transcriptional induction of GPBP

This supports the idea that induction associates with
Sp1 release from – and NFjB recruitment to ) the
transcriptional complex. However, there are major
differences between constitutive and induced transcriptional complexes other than the relative abundance of one or the other transcription factor. Thus
binding of Sp1 and NFjB appears to be independent
and competitive in noninduced cells and mutually
dependent and coordinated in induced cells.
A

B

Fig. 9. A DNA element encompassing NFjB-like binding site is critical for transcriptional activity and TNF responsiveness of the 140bp promoter. (A) the alignment of nucleotides 582–900 of GenBank
accession number AF315603 and nucleotides 4557244–4557527 of

GenBank accession number NT_039590.1, respectively, containing
the human and mouse POLK ⁄ COL4A3BP intergene regions are
shown in upper case. Boxed are the human 140-bp promoter and
the corresponding aligned mouse sequence. Highlighted are the
nucleotides identical between mouse and human DNA and
underlined are human DNA elements. The head-to-head arrows are
written over the 7-bp palindrome flanking the triplet-in-tandem
sequence (GGAGGA). With a vertical bar and an arrow, we indicate
one end of the 127-bp promoter and the direction towards the
other end, respectively. (B) Cells were transfected with the pFGHbased constructs containing the promoter in the indicated transcriptional orientation, and either induced with TNF or not. Results are
expressed as counts per minute after value normalization with the
corresponding b-galactosidase activity. Shown are the results of
one experiment carried out in duplicate ± SD. Similar results were
obtained in three independent experiments.

5300

F. Granero et al.

A DNA element encompassing the NFjB-like site
is missing in the mouse DNA counterpart that
shows no transcriptional activity
Alignment of mouse and human COL4A3BP ⁄ POLK
intergene regions reveals high degree of identity and
the presence of several gaps in mouse genome including one major gap of 27-bp encompassing the
NFjB-like site and its flanking regions (Fig. 9A). To
assess whether sequence differences had consequences
in transcriptional activity, the mouse DNA region representing the 140-bp human unit (105-bp) was cloned
in the two orientations in pFGH and its capacity
to drive heterologous gene expression was assessed

(Fig. 9B). The mouse 105-bp DNA fragment had no
significant transcriptional activity in either direction
or mode of expression in experimental conditions in
which human 140-bp was fully active. Taking a closer
look at the human-exclusive sequence, we found that
the NFjB-like site is the core of a 20-bp sequence
consisting of a 7-bp palindrome centred by a tripletin-tandem repeat (head-to-head arrows in Fig. 9A)
containing a recognition site for BamHI endonuclease.
We took advantage of such a restriction site and generated pFGH-based constructs in which half of the
20-bp sequence was deleted (127-bp, Fig. 9A). Interestingly, the 127-bp construct showed no activity in
the COL4A3BP direction and showed only residual
activity when transcribing in the POLK direction
(Fig. 9B).
Collectively our data suggest that bidirectional transcription from the human 140-bp DNA region depends
mostly on a DNA element containing binding sites
for Sp1 and NFjB which is critical for coordinated
Sp1 release and NFjB recruitment mediating TNF
induction.

Discussion
In eukaryotes, TBP and TAFs (TBP-associated factors) are general transcription factors that constitute
TFIID, a multisubunit complex required for RNA polymerase II positioning at either TATA box-containing
or TATA-less promoters [15,16]. The evidence supports that the canonical TATA element (TATAAA)
mediates TFIID recruitment through TBP binding. In
contrast, in TATA-less promoters TFIID recruitment
occurs through TAF binding to DNA sequences
encompassing a transcription start site (initiator). In
the latter, effective transcription requires the intervention of other nongeneral transcription factors which, in
turn, reflect the level of transcription and cell transcription factor repertoire [17–19]. Many housekeeping
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

F. Granero et al.

transcriptional units, including bidirectional, often are
devoid of a canonical TATA element and are transcribed from multiple start sites [14,20,21], suggesting
that regulatory nongeneral factors are determinant to
define transcription start site usage in these genes. We
have mapped three major transcriptional start sites for
COL4A3BP and two for POLK; however, cDNA cloning approaches identify multiple mRNA species initiating between these sites (GenBank accession numbers
AF163570 for POLK and BI826689, BI831357,
BM827753, BI911338, BI599774 for COL4A3BP).
Interestingly, BI826689 and BI831357 initiate in Sp1
flanking regions and BM827753, BI911338 and
BI599774 initiate in the TATA-like element and flanking regions. Among the remaining cDNAs species
identified (GenBank accession numbers BI860449,
BQ899288,
BQ707890,
BG828192,
BG827803,
BF310812,
BF306225,
CN351006,
BI760500,
CA397804,
BM171788,
BI093981,
BU855806,
BE734230,
BE019579,

BE272872,
BE732391,
BE278882,
AA158617,
AA121104,
BE278407,
BQ068905, BE734230), all of which derive from
COL4A3BP, site b appears to be the most commonly
used for initiation.
These observations support that the 140-bp promoter represents one of various alternative ways in
which the COL4A3BP ⁄ POLK intergene region can
be used for divergent transcription and for regulating
GPBP and pol j expression. The latter is accomplished from the 140-bp promoter by competitive
assembly of Sp1 and NFjB into a bidirectional transcriptional complex that binds to three DNA elements (NFjB-like, Sp1 and TATA-like). As the
NFjB-like site renders the 140-bp region competent
for transcription and is the only element supporting
direct Sp1 and NFjB binding – we found that unlabelled validated consensus sequences for NFjB efficiently inhibited nuclear protein binding to [32P]
NFjB-L, but not to [32P]SP1 or [32P]TATA-L in
EMSA – it seems that bidirectional transcription
from the 140-bp region depends mainly on competitive Sp1 and NFjB binding to the NFjB-like site. It
remains to be determined, however, whether the two
transcriptional factors compete for assembly in a single transcription complex or they exist in separate
complexes which compete for the promoter. Tumour
necrosis factor induces COL4A3 BP expression by
increasing the nuclear levels of NFjB but also by
promoting the structural linkage of the two transcription factors, suggesting that additional signals
other than the relative nuclear abundance of Sp1
and NFjB are needed for regulating transcriptional
orientation.
FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

TNF transcriptional induction of GPBP

A number of natural autoimmune responses mediating low or high frequency disorders (e.g. Goodpasture disease or multiple sclerosis) have been
described only in humans, and counterparts in the
animal world have not yet been reported. This
observation supported initial studies directed to identify human-specific biological features in the context
of Goodpasture disease, which resulted in the identification of a unique phosphorylatable region in the
human autoantigen [22,23] and subsequent discovery
of GPBP, a nonconventional protein kinase that specifically targets the autoantigen unique region [1,2].
Now, through the characterization of a DNA element without counterparts in mouse, chicken or rat
genomes (GenBank accession numbers NW_060733
and NW_047617), we identified a TNF-responsive
human-specific transcriptional unit and provided new
insights on GPBP expression that may be of relevance to understanding human-specific autoimmune
pathogenesis.
During maturation from basal to peripheral strata,
keratinocytes undergo an apoptosis-dependent differentiation process that is controlled, at least in part,
by Sp1 and NFjB-related factors [24–26]. Keratinocytes from patients suffering skin autoimmune processes show an increased sensitivity to UV-induced
apoptosis and a premature apoptosis at the basal
keratinocytes [27–29]. Goodpasture antigen-binding
protein is highly expressed in peripheral keratinocytes
in normal epidermis but is overexpressed in keratinocytes expanding from basal to peripheral strata in
autoimmune processes [2]. Finally, it has been reported that altered autoantigens, including phosphorylated versions thereof, are released from keratinocytes
undergoing apoptosis [28,30]. All of this suggests
that in keratinocytes NFjB-induced GPBP expression
is part of a desired cell death program that operates
illegitimately during autoimmune pathogenesis, in
which aberrant counterparts of critical cell components are generated.
The head-to-head arrangement of COL4A3BP and

POLK through a bidirectional promoter suggests that
the products encoded by these genes are partners in
specific cell programs. Consistently, we have found a
coordinated augmented expression of COL4A3BP and
POLK in skin undergoing an autoimmune attack, and
we have cloned a previously unrecognized alternatively
spliced isoform of 76-kDa, pol j76 (GenBank accession number AF315602), whose relative expression
with respect to pol j was found to be increased in
keratinocytes and in skin autoimmune processes
(unpublished observations). Pol j76 may represent the
version of pol j in a somatic mutation-based strategy
5301


TNF transcriptional induction of GPBP

to generate aberrant counterparts of critical cell components as part of a common cell death program that
is relevant in both keratinocyte maturation and autoimmune pathogenesis.

Experimental procedures
Synthetic oligonucleotides
ON-GPBP-6c, 5¢-CTCGCTCGCCCAGGGAAGGAAAA
GGGAAAAGAAGGGA-3¢; ON-GPBP-18 m, 5¢-GGCAT
GGTTAACGTGGTTCTC-3¢; ON-XbaG ⁄ Bpro1m, 5¢-GA
CTCTAGAGGGTTCGGGAGGAGGATCCCG-3¢; ONXbaG ⁄ Bpro1c, 5¢-GACTCTAGACTGGCCCACTATTTA
CCCTCC-3¢; ON-GPBP-5c, 5¢-CTCGATGCCAATTTCAA
ATAGGGAA-3¢; ON-polj-2m, 5¢-GACAAGCCGCCCTG
GAAAGCAGGCCC-3¢; ON-polj-3m, 5¢-GCAGCACAGC
TGCATCCCTACCCCGCCCTCTC-3¢; ON-SP1Del, 5¢-CG
CCGGGAGGGGGACGTAGTGGGGGAGAAT-3¢; ONNFjBmut, 5¢-TCTAGAGGGTTCGGGAGAAGGCTCGG

CGTGTCG-3¢; ON-TATADel, 5¢-CAGGGGAGGGGAG
GGGTGGGCCAGTCTAGA-3¢; ON-SP1-3m, 5¢-GGGGG
ACGGGGCGGGGAGTAGTGG-3¢; ON-SP1-3c, 5¢-CCA
CTACTCCCCGCCCCGTCCCCC-3¢; ON-TATA-1m, 5¢-GA
GGGGAGGGTAAATAGTGGGCCAG-3¢; ON-TATA-1c,
5¢-GTGGCCCACTATTTACCCTCCCCTC-3¢;
ONNFjBmut-2c, 5¢-GCCTTCTCCCGAACCC-3¢; ON-NFjBmut-2m, 5¢-GGGTTCGGGAGAAGGC-3¢; ON-NFjB-1m,
5¢-GGGTTCGGGAGGAGGATCCCGAAGGC-3¢;
ONNFjB-1c, 5¢-GCCTTCGGGATCCTCCTCCCGAACCC-3¢;
ON-SP1del-c, 5¢-CCCCCACTACGTCCCCCTCCC-3¢; ONPromMouse-F, 5¢-GACTCTAGAGGGAGCGTCGCGAG
CCGCCGGGAG-3¢; ON-PromMouse-R, 5¢-GACTCTAG
ACGGGTCCACTATTTACCCTCCCTC-3¢;
ON-EcoRIprom-1m, 5¢-GAGAATTCGGGTTCGGGAGGAGGAT
CC-3¢; ON-EcoRI-prom-2c, 5¢-GAGAATTCCTGGCCCA
CTATTTACCCTCC-3¢; ON-hGAPDH-F2, 5¢-TGGGCTA
CACTGAGCACCAG-3¢; ON-hGAPDH-R2, 5¢-GGGTGT
CGCTGTTGAAGTCA-3¢; ON-GPBP-F1, 5¢-CTGAATCC
AGCTTGCGTCG-3¢; ON-GPBP-R1, 5¢-GCAGAGTAGC
CACTTGCTCC-3¢.

Antibodies
All antibodies were from Santa Cruz Biotechnology Inc.
(Santa Cruz, CA, USA).

Plasmid construction
To obtain LpromGPBP and LpromPolj, a 772-bp DNA
fragment generated by XbaI and EclXI digestion of a 955-bp
fragment obtained by PCR from human DNA using ONGPBP-18 m and ON-GPBP-6c (GenBank accession number
AF315603) was filled-in and cloned into the HincII site of

5302

F. Granero et al.

pFGH (Nichols Institute, Corning Nichols Institute, San
Juan Capistrano, CA, USA) in the orientation of
COL4A3 BP or POLK, respectively. To generate
SpromGPBP and SpromPolj, the 140-bp DNA fragment,
generated by PCR using LpromGPBP, ON-XbaG ⁄ Bpro1m
and ON-XbaG ⁄ Bpro1c, was digested with XbaI and cloned
in pFGH in each of the two orientations. Subsequently,
SpromGPBP was used to obtain constructs in which NFjBlike and ⁄ or Sp1 binding sites and ⁄ or TATA-like element
were selectively deleted (DNFjB and ⁄ or DSp1 and ⁄ or
DTATA, respectively). This was accomplished using ONNFjBmut and ⁄ or ON-SP1Del and ⁄ or ON-TATADel and a
site-directed mutagenesis approach (Clontech, Mountain
View, CA, USA). To obtain the corresponding mutants for
transcription in the POLK direction, the reverse orientation
was cloned by XbaI digestion and re-ligation.
To generate human 127-bp pFGH-based constructs,
SpromGPBP was digested with BamHI and the resulting
DNA fragment cloned into the BamHI site of pFGH in
each of the two orientations. To obtain the 140-bp mouse
homologue constructs, a DNA fragment generated by PCR
using ON-PromMouse-F and ON-PromMouse-R and
mouse DNA was digested and cloned into the XbaI site of
pFGH in each of the two orientations.
For ribonuclease protection assays, LpromGPBP was
digested with ApaI and EclXI and the resulting DNA 503-bp
fragment made blunt-ended and cloned into the HincII site
of Bluescribe M13+ (Stratagene, La Jolla, CA, USA) to generate BS-ApaI ⁄ EclXI. For similar purposes, human DNA

was PCR-amplified using ON-polj-2m or ON-polj-3m
and ON-GPBP-5c, and the resulting 615- and 337-bp fragments cloned similarly to generate BS-polj2m ⁄ GPBP5c and
BS-polj3m ⁄ GPBP5c, respectively.
All DNA constructs were characterized by nucleotide
sequencing.

RNA purification
Total RNA was prepared from cultured cells using TRIREAGENT (Sigma, St Louis, MO, USA) or RNeasy
(Qiagen, Hilden, Germany) following manufacturer’s
instructions.

Ribonuclease protection assays
BS-ApaI ⁄ EclXI was linearized by EcoRI or HindIII digestion at the polylinker region, and BS-polj3m ⁄ GPBP5c and
BS-polj2m ⁄ GPBP5c by digesting the insert with BamHI
and EclXI, respectively.
T3 or T7 ribonucleotide probes representing the antisense
of GPBP or pol j mRNAs were obtained using MAXIscriptTM T7 ⁄ T3 in vitro transcription kit (Ambion, Austin,
TX, USA). Individual ribonucleotide probes were subject to
ribonuclease protection assays using RPAIIITM (Ambion)

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS


F. Granero et al.

and total RNA from human cultured hTERT-RPE1 (Clontech) or HEK 293 cells (ATCC # CRL-1573).

Promoter fragments
These were obtained by PCR using specific oligonucleotides
and SpromGPBP or derived mutants as templates. Thus,

the 140- and 76-bp promoters were obtained using
SpromGPBP, ON-NFjB-1m, and ON-TATA-1c or
ON-SP1-3c, respectively. The140-bpDNFjB and 76-bpDNFjB
fragments were obtained using DNFjB mutant, ONNFjBmut, and ON-TATA-1c or ON-SP1-3c, respectively.
To obtain 76-bpDSp1 we used ON-NFjB-1m, ON-SP1del-c
and DSp1 mutant; for 76-bpDNFjB ⁄ DSp1 we used ONNFjBmut, ON-SP1del-c and DNFjB ⁄ Sp1 double mutant.
To obtain the 26-bp and 26-bpDNFjB promoter fragments,
we annealed ON-NFjB-1m and ON-NFjB-1c or ONNFjBmut-2m and ON-NFjBmut-2c, respectively.
When the DNA was used for protein binding or purification studies, one oligonucleotide was 5¢-end labelled with
32
P or with biotin, respectively.

PCR
For PCR we used Pfu DNA polymerase (Stratagene) or the
Expand Long Template PCR System (Roche, Indianapolis,
IN, USA).

TNF transcriptional induction of GPBP

For transient gene expression assays, NIH3T3 cells
(1.4 · 105) were seeded in 9.5-cm2 plates, cultured for
14–16 h and transfected for 16–18 h with 2 lg each
individual pFGH-derived plasmid and 1 lg pSV-b-galactosidase expression vector (Promega, Madison, WI, USA)
using the calcium phosphate precipitation method of the
Profection Mammalian Transfection System (Promega).
After transfection, cells were rinsed with NaCl ⁄ Pi, fresh
medium was added and the levels of human growth hormone in the media and b-galactosidase activity in cell
lysates was determined after 48 h following the supplier’s
recommendations.
For some purposes, after transfection the cells were cultured in low serum (0.5%, v ⁄ v) medium for 24 h. Fresh

low serum medium containing or not containing human
recombinant TNFa (10 ngỈmL)1) or TNFb (50 ngỈmL)1)
(Sigma) was added and after 48 h, levels of human growth
hormone and b-galactosidase were similarly determined.

Nuclear extract preparation
HEK293 cells were grown approximately to 70–80% confluence, culture medium was removed, and fresh culture
medium containing 0.5% (v ⁄ v) serum was added and culture continued. After 24 h, fresh low serum medium containing or not containing TNF (see above) was added and
cultured for another 12–14 h and used for nuclear extract
preparation essentially as described previously [31].

Real-Time PCR
Total RNA (5 lg) from human cultured cells was reversetranscribed with random hexamers using the Ready-To-Go
system (GE Healthcare, Chalfont St Giles, UK) following
manufacturer’s recommendations, and the mixtures were
used for mRNA determinations performing real-time PCR
analysis with an SDS 5700 (Applied Biosystems, Foster
City, CA, USA). For these purposes, 4 lL of either 1 : 10
dilution of the mixtures for GPBP or 1 : 500 for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were analysed
by the relative quantitation method (DDCt) according to
the manufacturer’s recommendations. GAPDH was used as
an endogenous control to normalize quantification. The
pair of oligonucleotides used were ON-GPBP-F1 and
ON-GPBP-R1 for GPBP; and ON-hGAPDH-F2 and
ON-hGAPDH-R2 for GAPDH.

EMSA and purification of DNA-binding proteins
32

P-labelled or biotin 5¢-modified oligonucleotides were used

either to anneal with the corresponding unlabelled complementary synthetic oligonucleotide or to perform PCR with
a specific unlabelled oligonucleotide to produce doublestranded labelled DNA, which was further gel purified and
used for EMSA (32P) or for nuclear extract protein purification (biotin labelled) following procedures described previously [31]. Competitors for EMSA were prepared by
annealing ON-NFjB-1m and NFjB-1c (NFjB-L);
ON-SP1-3 m and ON-SP1-3c (SP1); ON-TATA-1m and
ON-TATA-1c
(TATA-L);
ON-NFjBmut-2m
and
ON-NFjBmut-2c (26-bpDNFjB used as nonrelevant DNA).

Cell culture and transient gene expression assays

Transfection of small interfering RNA (siRNA),
chromatin immunoprecipitation (ChIP) and western blot coupled assays

Cells were grown in Dulbecco’s modified Eagle’s medium
(NIH3T3 and HEK293) or Dulbecco’s modified Eagle’s
Ham’s
F-12
medium
(hTERT-RPE1)
containing
100 mL)1 penicillin and 100 lgỈmL)1 streptomycin, and
supplemented with 10% (v ⁄ v) of calf (NIH3T3 cells) or fetal
bovine (hTERT-RPE1 and HEK 293) serum.

HEK293 cells were transfected in the absence or in the
presence of 50 nm Sp1 or p65 siRNA [32,33]. After 24 h,
transfection medium was replaced by fresh medium and

culture maintained for 2 days. Subsequently, cells were cultured for an additional day with low serum (0.5%) and
either noninduced or induced with TNFa (10 ngỈmL)1) for

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

5303


TNF transcriptional induction of GPBP

1 h, and used for western blot or ChIP assays. For western
blot purposes, cells were lysed in 10 mm Tris ⁄ HCl (pH 8),
100 mm NaCl, 1 mm EDTA, 1% (v ⁄ v) SDS, 2% (v ⁄ v)
Triton X-100, 1 mm phenylmethylsulfonyl fluoride and
10 lgỈmL)1, centrifuged at 20 000 g for 45 min and supernatants were analysed to estimate Sp1 or p65 content using
specific antibodies. For ChIP purposes, cells were treated
essentially as previously described [34,35]. Pellets containing
DNA immunoprecipitates were dissolved in 60 lL H2O and
2 lL were subjected to PCR using ON-EcoRI-prom-1m
and ON-EcoRI-prom-2c (30 cycles each consisting of 95 °C
for 45 s; 60 °C for 30 s; 72 °C for 45 s with initial heating
at 95 °C for 1 min and a final 5 min extension step at
72 °C) to verify the presence of the 140-bp promoter. The
DNA in the supernatants of nonrelevant immunoprecipitations was ethanol precipitated, similarly hydrated and further diluted 1 : 50 in H2O prior to being used (2 lL) in
PCR procedures as DNA input samples.

Acknowledgements
This work was supported by grants SAF97 ⁄ 0065,
SAF2000 ⁄ 0047, SAF2001 ⁄ 0453 and SAF2003-09772 of
the Plan Nacional de I + D, and grant 98 ⁄ 102-00 of

´
Fundacion ‘La Caixa’ (Spain) (to J.S), and the fellowship program of the F.V.I.B. in collaboration with
BioStratum Inc. and Iberdiagnosis S.A.

References
1 Raya A, Revert F, Navarro S & Saus J (1999) Characterization of a novel type of serine ⁄ threonine kinase that
specifically phosphorylates the human Goodpasture
antigen. J Biol Chem 274, 12642–12649.
2 Raya A, Revert-Ros F, Martinez-Martinez P, Navarro S,
´
Rosello E, Vieites B et al. (2000) Goodpasture antigenbinding protein, the kinase that phosphorylates the
Goodpasture antigen, is an alternatively spliced variant
implicated in autoimmune pathogenesis. J Biol Chem
275, 40392–40399.
3 Wagner J, Gruz P, Kim S-R, Yamada M, Matsui K,
Fuchs RPP & Nohmi T (1999) The dinB gene encodes a
novel E. coli DNA polymerase, DNA Pol IV, involved
in mutagenesis. Mol Cell 4, 281–286.
4 Gerlach VL, Aravind L, Gotway G, Schultz RA,
Koonin EV & Friedberg EC (1999) Human and mouse
homologs of E. coli DinB (DNA polymerase IV),
members of the UmuC ⁄ DinB superfamily. Proc Natl
Acad Sci USA 96, 11922–11927.
5 Johnson RE, Prakash S & Prakash L (2000) The human
DINB1 gene encodes the DNA polymerase polh. Proc
Natl Acad Sci USA 97, 3838–3843.
6 Tang M, Pham P, Shen X, Taylor J-S, O’Donnell M,
Woodgate R & Goodman MF (2000) Roles of the

5304

F. Granero et al.

7

8

9

10

11

12

13

14

15

16

17

18

19

E. coli DNA polymerases IV and V in lesion-targeted

and untargeted SOS mutagenesis. Nature 404,
1014–1018.
Ohashi E, Bebenek K, Matsuda T, Feaver WJ, Gerlach
VL, Friedberg EC et al. (2000) Fidelity and processivity
of DNA synthesis by DNA polymerase j, the product
of the human DINB1 gene. J Biol Chem 275, 39678–
39684.
Zhang Y, Yuan F, Wu X, Wang M, Rechkoblit O,
Taylor J-S, Geacintov NE & Wang Z (2000) Error-free
and error-prone lesion bypass by human DNA polymerase j in vitro. Nucleic Acids Res 28, 4138–4146.
Gerlach VL, Feaver WJ, Fischhaber PL & Friedberg
EC (2001) Purification and characterization of pol j, a
DNA polymerase encoded by the human DINB1 gene.
J Biol Chem 276, 92–98.
Ogi T, Shinkai Y, Tanaka K & Ohmori H (2002) Pol j
protects mammalian cells against the lethal and mutagenic effects of benzo[a]pyrene. Proc Natl Acad Sci
USA 99, 15548–15553.
Aggarwal BB, Samanta A & Feldmann M (2001)
TNFa. In Cytokine Reference. A Compedium of
Cytokines and Other Mediators of Host Defense, Vol. I.
(Oppenheim JJ & Feldmann M, eds), pp. 413–447.
Academic Press Ltd, London.
Quandt K, Frech K, Karas H, Wingender E & Werner
T (1995) MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data. Nucleic Acids Res 23, 4878–4884.
Naar AM, Ryu S & Tjian R (1998) Cofactor requirements for transcriptional activation by Sp1. Cold Spring
Harbor Symp Quant Biol 63, 189–199.
Azizkhan JC, Jensen DE, Pierce AJ & Wade M (1993)
Transcription from TATA-less promoters: dihydrofolate
reductase as a model. Crit Rev Eukaryot Gene Expr 3,
229–254.

Hampsey M (1998) Molecular genetics of the RNA
polymerase II general transcriptional machinery. Microbiol Mol Biol Rev 62, 465–503.
Veenstra GJC & Wolffe AP (2001) Gene-selective
developmental roles of general transcription factors.
Trends Biochem Sci 26, 665–671.
Li X-Y, Bhaumik SR, Zhu X, Li L, Shen W-C, Dixit
BL & Green MR (2002) Selective recruitment of TAFs
by yeast upstream activating sequences. Implications for
eukaryotic promoter structure. Current Biol 12, 1240–
1244.
Mencia M, Moqtaderi Z, Geisberg JV, Kuras L &
Struhl K (2002) Activator-specific recruitment of TFIID
and regulation of ribosomal protein genes in yeast. Mol
Cell 9, 823–833.
Hilton TL & Wang EH (2003) Transcription factor IID
recruitment and Sp1 activation. J Biol Chem 278,
12992–13002.

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS


F. Granero et al.

20 Kawai Y, Asai K, Miura Y, Inoue Y, Yamamoto M,
Moriyama A, Yamamoto N & Kato T (2003) Structure
and promoter activity of the human glia maturation
factor-gamma gene: a TATA-less, GC-rich and bidirectional promoter. Biochem Biophys Acta 1625, 246–252.
21 Adachi N & Lieber MR (2002) Bidirectional gene organization: a common architectural feature of the human
genome. Cell 109, 807–809.
22 Quinones S, Bernal D, Garcı´ a-Sogo M, Elena SF &

Saus J (1992) Exon ⁄ Intron structure of the human a3
(IV) gene encompassing the Goodpasture antigen (a3
(IV): NC1): Identification of a potentially antigenic
region at the triple helix ⁄ NC1 domain junction. J Biol
Chem 267, 19780–19784.
´
23 Revert F, Penades JP, Plana M, Bernal D, Johansson
C, Itarte E et al. (1995) Phosphorylation of the Goodpasture antigen by type A protein kinases. J Biol Chem
270, 13254–13261.
24 Polakowska RR, Piacentini M, Bartlett R, Goldsmith
LA & Haake AR (1994) Apoptosis in human skin
development: morphogenesis, periderm, and stem cells.
Dev Dyn 199, 176–188.
25 Maruoka Y, Harada H, Mitsuyasu T, Seta Y,
Kurokawa H, Kajiyama M & Toyoshima K (1997) Keratinocytes become terminally differentiated in a process
involving programmed cell death. Biochem Biophys Res
Commun 238, 886–890.
26 Ren-Long W, Tung-Ti C & Tung-Tien S (1994) Functional importance of an Sp1- and NFjB-related nuclear
protein in a keratinocyte-specific promoter rabbit K3
keratin gene. J Biol Chem 269, 28450–28459.
27 Casciola-Rosen LA, Anhalt G & Rosen A (1994) Autoantigens targeted in systemic lupus erythematosus are

FEBS Journal 272 (2005) 5291–5305 ª 2005 FEBS

TNF transcriptional induction of GPBP

28

29

30

31

32

33

34

35

clustered in two populations of surface structures on
apoptotic keratinocytes. J Exp Med 179, 1317–1330.
Casciola-Rosen L & Rosen A (1997) Ultraviolet lightinduced keratinocyte apoptosis: a potential mechanism
for the induction of skin lesions and autoantibody production in LE. Lupus 6, 175–180.
Pablos JL, Santiago B, Galindo M, Carreira PE, Ballestin C & Gomez-Reino JJ (1999) Keratinocyte apoptosis
and p53 expression in cutaneous lupus and dermatomyositis. J Pathol 188, 63–68.
Utz PJ & Anderson P (1998) Posttranslational protein
modifications, apoptosis, and the bypass of tolerance to
autoantigens. Arthritis Rheum 41, 1152–1160.
Ausubel FM, Brent R, Kingston RE, Moore DD, Deidman JG, Smith JA & Struhl K (1998) Current Protocols
in Molecular Biology. John Wiley & Sons, Inc, New
York.
Abdelrahim M, Samudio I, Smith R III, Burghardt R &
Safe S (2002) Small inhibitory RNA duplexes for Sp1
mRNA block basal and estrogen-induced gene expression and cell cycle progression in MCF-7 breast cancer
cells. J Biol Chem 277, 28815–28822.
Surabhi RM & Gaynor RB (2002) RNA interference
directed against viral and cellular targets inhibits human

immunodeficiency virus type 1 replication. J Virol 76,
12963–12973.
Boyd KE & Farnham PJ (1997) Myc versus USF: discrimination at the cad gene is determined by core promoter
elements. Mol Cell Biol 17, 2529–2537.
Boyd KE, Wells J, Gutman J, Bartley SM & Farnham
PJ (1998) c-Myc target gene specificity is determined by
a post-DNA binding mechanism. Proc Natl Acad Sci
USA 95, 13887–13892.

5305