We used a DNA microarray to survey the expression of genes regulated by IFN-inducible, STAT2-depend-ent DNA binding, and compared the cDNAs of IFN-treated cells over-expressing intact ST
Trang 1genes whose transcription is STAT2-dependent but
ISGF3-independent
Melissa M Brierley1, Katie L Marchington1, Igor Jurisica2 and Eleanor N Fish1
1 Department of Cell and Molecular Biology, Toronto General Research Institute, University Health Network, and Department of
Immunology, University of Toronto, ON, Canada
2 Division of Signaling Biology, Ontario Cancer Institute, University Health Network, and Department of Medical Biophysics,
University of Toronto, ON, Canada
The type I interferons (IFN)-a⁄ b are multifunctional
cytokines that mediate host defense against microbial
challenges, influence both normal and neoplastic
pro-liferation, and modulate innate and adaptive immune
responses [1,2] The binding of type I IFNs to their
shared cognate receptor, type I interferon receptor
(IFNAR), activates multiple intracellular signaling
cascades that coordinate to trigger both the
tran-scriptional activation and translational modifications
necessary to invoke various biological responses [3,4] Arguably the most notable of these cascades is the Janus kinase (Jak)-signal transducer and activator of transcription (STAT) pathway that regulates the transcription of numerous IFN-sensitive genes (ISGs) Upon IFN binding to IFNAR, the receptor-associated kinases tyrosine kinase 2 (Tyk2) and Jak1 phos-phorylate key tyrosine residues within the intra-cellular domains of the receptor subunits [5] These
Keywords
gene regulation; interferon; signal
transduction; transcription factors
Correspondence
E N Fish, Toronto General Research
Institute, 67 College Street, Rm 424,
Toronto, ON M5G 2M1, Canada
Fax: +1 416 340 3453
Tel: +1 416 340 5380
E-mail: en.fish@utoronto.ca
(Received 19 January 2006, revised 6
Febru-ary 2006, accepted 13 FebruFebru-ary 2006)
doi:10.1111/j.1742-4658.2006.05176.x
Signal transducer and activator of transcription 2 (STAT2) is best known
as a critical transactivator component of the interferon-stimulated gene factor 3 (ISGF3) complex that drives the expression of many interferon (IFN)-inducible genes However, STAT2 is also involved in DNA binding
in non-ISGF3 transcriptional complexes We used a DNA microarray to survey the expression of genes regulated by IFN-inducible, STAT2-depend-ent DNA binding, and compared the cDNAs of IFN-treated cells over-expressing intact STAT2 to those of IFN-treated cells overover-expressing mutated STAT2 lacking the DNA binding domain The IFN-inducible expression of genes known to be regulated by ISGF3 was similar in both cases However, a subset of IFN-inducible genes was identified whose expression was decreased in cells expressing the mutated STAT2 Impor-tantly, these genes all contained gamma-activated sequence (GAS)-like ele-ments in their 5¢ flanking sequences Our data reveal the existence of a collection of GAS-regulated target genes whose expression is IFN-inducible and independent of ISGF3 but highly dependent on the STAT2 DNA binding domain This report is the first analysis of the contribution of the STAT2 DNA binding domain to IFN responses on a global basis, and shows that STAT2 is required for the IFN-inducible activation of the full spectrum of GAS target genes
Abbreviations
IFN, interferon; ISG, IFN-sensitive gene; ISGF3, IFN-stimulated gene factor 3; ISRE, IFN-stimulated response element; IFNAR, type I interferon receptor; GAS, gamma-activated sequence; IRF, IFN regulatory factor; BSTVQ, binary tree-structured vector quantization; OPHID, online predicted human protein interaction database; SOM, self-organizing map; STAT2, signal transducer and activator of transcription 2; TSS, transcriptional start site.
Trang 2phosphorylated residues act as recruitment sites for
STAT proteins, whereupon activated Jaks
phosphory-late a single tyrosine residue within the carboxy
termi-nus of the STATs [6,7] The phosphorylated and
activated STATs form both homodimeric and
hetero-dimeric complexes that translocate to the nucleus and
bind specific DNA sequences within the promoter
regions of ISGs to initiate transcription [8]
An important IFN-inducible complex is
IFN-stimu-lated gene factor 3 (ISGF3), comprised of STAT1,
STAT2 and IRF-9 (a member of the IFN regulatory
factor [IRF] family) [9] Upon nuclear import, ISGF3
binds to the IFN-stimulated response element (ISRE)
present in the promoter regionsof a subset of
IFN-inducible genes and triggers transcription As well as
ISGF3, type I IFNs induce the formation of additional
STAT-containing complexes, including STAT1–1,
STAT3–3 and STAT5–5 homodimers as well as
STAT3–1 and STAT2–1 heterodimers [10–12] Rather
than to the ISRE, these homodimers and
heterodi-mers bind to palindromic gamma-activated sequences
(GAS) located in the promoters of a different subset
of ISGs
Studies of human U6A fibroblasts lacking functional
STAT2 have shown that this transcription factor is
necessary for IFN-inducible antiviral and growth
inhibitory responses [13,14] Similarly, while STAT2
knockout mice are viable and show no developmental
defects, they have a compromised IFN response and
are highly susceptible to viral infections [15] These
defects are due, at least in part, to the loss of function
of STAT2-containing ISGF3 complexes that would
normally induce expression of ISRE-containing genes
such as ISG15, 9-27, 6-16, PKR, OAS and MxA
[16,17] However, some of the defects in
STAT2-defici-ent systems appear to be due to the loss of function of
ill-defined STAT2-containing complexes that are
dis-tinct from ISGF3 While it is known that STAT2–1
heterodimers can regulate IFN responses by binding to
specific like elements [14,18], only a few
GAS-containing ISGs, including IRF1 and FccRI, have
been identified to date [10]
Microarray gene expression analyses have led to the
identification of numerous ISGs and have implicated
IFNs in activities as diverse as cell adhesion,
transcrip-tional regulation, apoptosis and lipid metabolism
[19,20] In previous work, we constructed a panel of
fibroblast cell lines (based on the STAT2-deficient cell
line U6A) that overexpress various types of mutated
STAT2 molecules In that study, cells bearing the V453I,
V454I (VV-II) mutation (U6A-2VV-II cells) that
com-promises the STAT2 DNA binding domain, exhibited
intact ISRE-mediated transcriptional activation but
impaired GAS-mediated transcription [14] To precisely determine the transcriptional target genes of ISGF3-independent STAT2-containing complexes, cDNAs from IFN-treated cells overexpressing either intact STAT2 (U6A-2 cells) or the VV-II mutant form of STAT2 (U6A-2VV-II) were hybridized to an Affymetrix DNA microarray containing over 22 000 unique tran-scripts By comparing the IFN-inducible gene expres-sion profiles of these cells, we identified a subset of GAS-dependent ISGs whose activation is exclusively regulated by ISGF3-independent STAT2-containing complexes
Results
ISG expression in the absence of the STAT2 DNA binding domain as revealed by DNA microarray
We used DNA microarray analysis to compare the gene expression profiles of U6A (STAT2-deficient), U6A-2 (intact STAT2), and U6A-2VV-II (mutant STAT2 lacking the DNA binding domain) cells treated with 5 ngÆmL)1 IFN-alfacon-1 for 6 h Differences in mRNA expression among these groups (normalized to untreated controls) were evaluated using the Affyme-trix U-133A GeneChip microarray and genespring software As expected, IFN treatment of U6A cells expressing either intact or mutated STAT2 induced the expression (to varying degrees) of many ISGs (Fig 1) Indeed, IFN-alfacon-1 treatment up-regulated the expression of 232 and 286 genes in U6A-2 and U6A-2VV-II cells, respectively, by greater than two-fold In control U6A cells, only eight genes showed a greater than two-fold increase in expression in response to IFN, confirming the importance of STAT2 function to ISG expression Furthermore, several genes known to be important for mediating the biolo-gical effects of IFN, most notably 2¢5¢OAS1 ⁄ 2, Mx and viperin, were not expressed in U6A cells following IFN stimulation (Table 1) In contrast, in both U6A-2 and U6A-2VV-II cells, IFN treatment induced com-parable levels of expression of several known ISRE-mediated ISGs, including 2¢5¢OAS, Mx, ISG15, 9-27 and MHC class I These results confirm that the activ-ity of ISGF3 complexes is intact in the absence of the STAT2 DNA binding domain Expression levels of several known GAS-driven genes, including GBP1, were also up-regulated to the same degree in both U6A-2 and U6A-2VV-II cells (Table 1) However, the expression levels of several other ISGs including IFIT1, IFIT2, 2¢5¢OAS2 and GIP3, differed markedly between IFN-treated U6A-2 and U6A-2VV-II cells (Table 1)
Trang 3ISG expression in the absence of the STAT2 DNA
binding domain as revealed by real-time PCR
To more quantitatively examine the expression of
IFN-regulated genes in the absence of the STAT2 DNA
binding domain, we treated U6A, 2 and
U6A-2VV-II cells with IFN-alfacon-1 for 6 h and analyzed
gene expression using relative quantitative real-time
PCR We also carried out MatInspector analyses (see
below) of the upstream promoter regions of
IFN-regu-lated genes to determine the presence of ISRE, GAS
and additional regulatory sequences Among the genes
selected for examination were PKR, 2¢5¢OAS and Mx1;
genes whose products are known mediators of the
IFN-inducible antiviral response [16] Comparable
transcriptional activation of the PKR, 2¢5¢OAS and Mx
genes was observed in IFN-stimulated U6A-2 and
U6A-2VV-II cells, and the promoters of all three genes
contained the expected ISRE elements (Fig S1A–C)
These data support our previous findings that ISGF3
activation is intact in U6A-2VV-II cells (above and
[14]) Moreover, while the promoter regions of PKR
and 2¢5¢OAS also contain potential GAS-like elements,
ISGF3-independent STAT2-containing complexes
dependent on a functional STAT2 DNA binding
domain do not appear to play an important role in
mediating the transcription of these genes
The c-fos gene was examined in this system as an example of a GAS-driven gene whose IFN-inducibility
is independent of both ISRE and STAT2 Our analysis confirmed previous findings [21] that the c-fos promoter contains a single GAS-like element but not
an ISRE As well, we found that STAT2 expression was not required to mediate c-fos expression, because comparable (albeit weak) IFN-inducible transcriptional activation of c-fos occurred in U6A, 2 and U6A-2VV-II cells (Fig S1D) This result is consistent with previous studies demonstrating that IFN-indu-cible c-fos expression is mediated by the binding of STAT1–1, STAT1–3 or STAT3–3 complexes to the GAS-like element [22]
Several genes listed in Table 1 were characterized by absent or weak expression in IFN-treated U6A cells but high levels of inducible expression in both U6A-2 and U6A-2-VV-II cells This profile implies that STAT2, but not necessarily its DNA binding domain,
is required for the expression of these genes We more closely examined the expression of the GBP1 gene as
an example of this class of ISG GBP1 expression was only weakly up-regulated in U6A cells but induced to high levels in both U6A-2 and U6A-2-VV-II cells (Fig S1E) Our promoter analysis confirmed the exist-ence of a single ISRE and two GAS-like elements in the GBP1 upstream promoter region (Fig S1E) These
127 8
0 0 0 0
576 172
51 35 20 8
351 120
51 35 18 8
Number of Genes Activated upon IFN Stimulation
U6A
VV-II
U6A-2
> 20.0 fold
> 10.0 fold
> 6.0 fold
> 4.0 fold
> 2.0 fold
> 1.5 fold
Fig 1 IFN-inducible transcriptional activation in the absence of the STAT2 DNA binding domain as determined by Affymetrix DNA microarray analysis Total mRNA samples from U6A-2, U6A-2VV-II and U6A cells either left untreated or treated with IFNa for 6 h were applied to Affymetrix U-133A microarray gene chips Hybridization data from the IFN-treated samples were normalized to the data from the correspond-ing untreated samples The normalized gene expression profiles for each cell category were analyzed as described in Experimental proce-dures to determine the number and expression level of genes up-regulated following stimulation with IFN A total of 286 genes were induced by IFN to a greater than two-fold increase over untreated controls in the absence of the STAT2 DNA binding domain (VV-II).
Trang 4results indicate that, although STAT2 is required for
IFN-inducible GBP1 expression, ISGF3-independent
STAT2-containing complexes do not appear to play a
significant role in the transcription of this gene These
results are in agreement with earlier studies which
implicated IFN-a and IFN-c induced STAT1–1
com-plex interactions with the GAS-like elements [23,24]
Two genes listed in Table 1 were characterized by
IFN-inducibility in U6A-2 cells but absent or weak
expression in both U6A and U6A-2-VV-II cells,
imply-ing that the STAT2 DNA bindimply-ing domain is essential
for the IFN-triggered expression of these genes: TLR3, RBMS3 Closer examination of the IFN-inducible expression of the TLR3 gene using quantitative PCR confirmed its diminished expression in U6A and U6A-2-VV-II cells (Fig S1F) Furthermore, promoter analy-sis confirmed a previous report [25] identifying both ISRE and GAS-like elements in the TLR3 promoter (Fig S1F) Notably, in this study mutational analysis determined that the ISRE is important for mediating TLR3 expression, and competition assays and gene expression studies suggested that STAT1 can bind to
Table 1 IFN-inducible gene expression in U6A, U6A-2 and U6A-2VV-II cells Fold induction values represent the change in mRNA levels in IFN-treated cells compared to untreated cells and were obtained using GENESPRING software.
Affymetrix
accession
Fold induction upon IFN
214453_s_at IFN-induced, hepatitis C-associated microtubular aggregate
protein (44 kDa) (MTAP44)
Trang 5the GAS-like elements and is required for
IFN-indu-cible TLR3 expression Viewed together, these findings
strongly suggest that the binding of
ISGF3-independ-ent STAT2-containing heterodimers to GAS-like
ele-ments within the TLR3 promoter region may
contribute significantly to TLR3 expression
ISG expression in the absence of the STAT2
DNA binding domain as revealed by binary
tree-structured vector quantization analysis
The analysis of the microarray results presented in
Table 1 required the use of an arbitrary ‘cut-off’ value
for level of gene expression, an approach that introduces
an element of bias into the analysis The binary
tree-structured vector quantization (BTSVQ) algorithm (see
below) can be used to analyze microarray data in more
depth and in the absence of such bias We applied the BTSVQ algorithm to our microarray data to identify additional target genes that are transcriptionally regulated by ISGF3-independent STAT2-containing complexes The BTSVQ algorithm sorts data into binary trees based on equality of expression of each mRNA target [26] Samples having progressively dissimilar levels of target gene expression are placed further down the tree The data are then visualized by the means of self-organizing maps (SOMs; see below) to cluster genes into distinct units having similar expression levels When the total gene expression profiles of untreated and IFN-treated U6A, U6A-2 and U6A-2VV-II cells were analyzed using BTSVQ, the resulting binary tree showed that cells expressing intact STAT2 segregated from the U6A and U6A-2VV-II cells at the first level (Fig 2) Interestingly, the data suggested that the gene
Level 1
Child 3 Child 4 Child 1
Child 2
Child 4 Child 3
Child 6 Child 5
Samples
Areas representing genes not expressed Areas representing expressed genes
Index
U6A-2 T
Fig 2 BSTVQ analysis of IFN-inducible gene expression in U6A, U6A-2 and U6A-2VV-II cells The raw data from the Affymetrix U-133A microarray analysis in Fig 1 were analyzed using BSTVQ (see Experimental procedures) to generate a binary tree indicating the progressive degree of dissimilarity of the six cell categories The SOMs (colored regions) visually represent the differences in gene expression profiles amongst the six cell categories Areas identified by visual exploratory analysis are circled and represent genes with the indicated expression pattern.
Trang 6expression profile of cells expressing the VV-II mutant
form of STAT2 was more similar to that of U6A cells
than that of U6A-2 cells (Level 1) IFN treatment led
to the further segregation of the sample types as
evi-denced by the altered gene expression profiles in these
cells Surprisingly, untreated U6A-2VV-II cells were
first to segregate away from the U6A⁄ U6A2VV-II
cluster, suggesting that IFN stimulation of cells
expres-sing the VV-II mutant form of STAT2 inhibited the
expression of certain genes (Level 2)
Examination of the SOMs confirmed that each cell
category had a unique gene expression profile and that
IFN treatment altered the profile in each case These
changes to the profile were visualized as blue zones of
the SOMs of the untreated samples becoming red
in the corresponding treated samples, as
IFN-inducible genes were up-regulated (Fig 2, colored
regions) Significantly, while IFN treatment
down-regulated a relatively small number of genes in U6A
and U6A-2 cells, IFN-treated U6A-VV-II cells showed
the down-regulation of a substantially larger subset of
genes
Identification and characterization of a subset of
ISGF3-independent STAT2-dependent target
genes
To identify those genes whose expression was
exclusively regulated by ISGF3-independent
STAT2-containing complexes, we directly compared the gene
expression profiles of the IFN-treated U6A-2 and
U6A-2VV-II cells shown in Fig 3 Using SOM
exam-ination and the BTSVQ program, we were able to
extract a list of 19 differentially expressed transcripts
that were highly expressed in the IFN-treated U6A-2
sample but absent from the IFN-treated U6A-2VV-II
sample (Table 2) Nine of these transcripts represented
genes encoding well-characterized proteins with defined
functions The remaining 10 transcripts represented
either hypothetical proteins or proteins with unknown
functions When genespring was employed to
calcu-late the fold-increase in expression of these genes upon
IFN treatment, we found that each of these mRNAs
was up-regulated about 20–60-fold in IFN-treated
U6A-2 cells compared to IFN-treated U6A-2VV-II
cells (Table 2)
To investigate the promoters of the nine known
dif-ferentially expressed ISGs, we sequenced a region 1000
bases upstream from the transcriptional start site
(TSS) of each gene and searched for the presence
of various STAT-binding GAS elements and
ISGF3-binding ISREs While three of the genes under study
contained potential ISREs, all exhibited potential
STAT-binding elements with the GAS-like palindromic core motif TTNNNNNAA (Fig S2) It should be noted that, although no GAS-like elements were evi-dent in the 1000 bp immediately upstream of the JUND TSS, the matinspector program was able to identify one GAS-like element between )3809 to )3791 and a second one between )3813 to )3837 (rel-ative to the JUND TSS) Importantly, for each ISG, the region containing the GAS-like elements also con-tained binding sites for known transcription factors, including Sp1, Oct1, CREB and NF-jB This juxta-position strongly suggests that the GAS-like elements probably function as promoter regulators modulating the expression of ISGs
To verify that the genes detected by BTSVQ analysis were indeed highly expressed in IFN-treated U6A-2 cells but not in IFN-treated U6A-2VV-II cells, real-time PCR validation was performed on four of the above nine genes: CLDN4, BF, DGKE and MSR1 The analysis confirmed that these genes were all expressed at substantially higher levels in IFN-treated U6A-2 cells than in U6A-2VV-II cells (Fig 3) Thus, IFN-inducible expression of these genes is impaired in the absence of the STAT2 DNA binding domain, sug-gesting that their IFN-inducible transcription requires ISGF3-independent STAT2-containing complexes Notably, MSR1 exhibited the least difference in IFN-inducible gene expression between U6A-2 and U6A2-VV-II cells Whereas the 1000 bp upstream regions of
0 2 4 6 8 10 12 14 16
Fig 3 Characterization of the induction levels of a subset of ISGF3-independent STAT2-dependent ISGs identified by BSTVQ The differential expression of four of the ISGs examined in Fig 4 was assessed in IFN-stimulated U6A, U6A-2 and VV-II cells using relative quantitative real-time PCR as for Fig 2 For each sample, b-actin was evaluated as a reference gene and used for normaliza-tion For each gene, data are presented as the fold-increase in expression in IFN-treated U6A-2 cells compared to IFN-treated U6A-2VV-II cells Values ± SE were calculated using RELATIVE QUANTI-FICATION software (Roche) and are the mean of three separate react-ions, each performed in triplicate.
Trang 7CLDN4, BF and DGKE contain GAS elements and no
ISREs (Fig S2), both elements are present in the
upstream region of MSR1 This result suggests that
IFN-inducible, dependent (as well as
ISGF3-independent) STAT2-containing complexes make a
con-tribution to the regulation of MSR1 gene expression
To explore the physiological relevance of our
microarray findings, we attempted to link our gene
expression data to potential ISGF3-independent
STAT2-regulated signaling pathways that might
influ-ence IFN-inducible outcomes, by generating pathway
networks downstream of IFNAR that highlighted
genes cited in this study (Fig 4) In addition, we
map-ped our ISGF3-independent STAT2-regulated ISGs
to the OPHID protein–protein interaction network
[27] (http://ophid.utoronto.ca) to examine the
inter-relationship of these targets within multiple pathways This exercise generated a network of 1400 proteins linked by 2261 interactions, with all but 16 interac-tions being from human curated sources (Fig S3) Notably, all of the IFN-inducible, ISGF3-independent STAT2-dependent targets identified in this study were interconnected via signaling networks known to be activated by IFNs-a⁄ b In particular, many of the identified ISGF3-independent STAT2-mediated events were associated with cell growth regulation
Discussion
STAT2 is known to be critical for type I IFN signaling and to play a crucial role in ISGF3-mediated tran-scription of IFN-inducible genes However, STAT2¢s
Table 2 mRNAs identified by BSTVQ analysis of microarray data as highly expressed in IFN-treated 2 cells but not in IFN-treated U6A-2VV-II cells Fold induction values represent the change in mRNA levels in IFN-treated U6A-2 cells compared to IFN-treated U6A-U6A-2VV-II cells and were obtained using GENESPRING software.
Characterized (9)
Macrophage scavenger receptor 1 (MSR1) Macrophage-specific trimeric integral membrane glycoprotein 42.7 Jun D proto-oncogene (JUND) a Component of the AP1 transcription factor complex, role in regulation
of transcription from Pol II promoter
35.0 Desmin (DES) Muscle-specific class II intermediate filament, implicated in
cytoskeleton organization and biogenesis
24.8 Interleukin 20 receptor, alpha (IL-20RA) Receptor for interleukin 20 (IL-20), a cytokine that may be involved
in epidermal function.
24.6 Peptidyl-prolyl cistrans isomerase
NIMA-interacting 1-like (PIN1L)
May be involved in organization of the synaptic cell–cell junction through interaction with the delta-catenin ⁄ NPRAP-N-cadherin complex
24.3 Neuromedin B receptor (NMBR) Binds neuromedin B, a potent mitogen and growth factor for normal
and neoplastic lung and for gastrointestinal epithelial tissue Involved
in G-protein signalling
24.3
Diacylglycerol kinase, epsilon (64 kDa) May be involved mainly in the regeneration of phosphatidylinositol (PI)
from diacylglycerol in the PI-cycle during cell signal transduction.
Role in ATP binding and diacylglycerol kinase activity (DGKE)
24.0
B-factor, properdin (BF) Complement factor B, a component of the alternative pathway of
complement activation
22.5 Hypothetical ⁄ unknown (10)
ESTs, Moderately similar to G02654
ribosomal protein L39
mRNA for KIAA0550 protein Similar to brain-specific angiogenesis inhibitor 3, a seven-span
transmembrane protein
35.0
function unknown
30.5 Unknown clone 12262, mRNA Similar to translocase of inner mitochondrial membrane 8
homolog A (yeast)
27.5 Hypothetical protein PRO2822 Weak similarity to cytokine receptor-like factor 2 precusor 26.9
Clone 24775 mRNA sequence Hypothetical protein BC013764, inferred role in potassium ion transport 25.0
a Identified by two separate probe sets.
Trang 8function with respect to ISGF3-independent
transcrip-tion of ISGs has been unclear Previous work showed
that, in cells with a loss of function mutation in the
DNA binding domain of STAT2, ISRE-mediated
tran-scriptional activation and gene expression were intact
but GAS-driven transcriptional activation was
compromised [14] By comparing the IFN-regulated
gene expression profile of cells expressing intact
STAT2 with that of cells expressing the mutated
STAT2, we have identified a subset of GAS-driven
target genes that are selectively regulated by
ISGF3-independent STAT2-containing complexes We are
confident that our stepwise approach to analyzing our
microarray data has produced results in which bias
has been minimized, ensuring that our results are
bio-logically relevant We first established full gene
expres-sion profiles of main subgroups of individual cells
responding to IFN treatment This unsupervised
clus-tering step was followed by identification of the most
differentially regulated genes Finally, these genes were
validated by real-time PCR and placed into the context
of IFN-related biological pathways
Target gene binding by STAT complexes is determined by DNA motif sequence specificity STAT homo- or heterodimeric complexes recognize and bind promoter sequences containing the GAS-like palin-dromic core motif, TTNNNNN(N)AA Although the preferential GAS element for the STAT2–1 heterodi-mer is ATTTCCCGGAAA [18], the STAT2–1 complex can also bind to the GAS elements within the promo-ters of both IRF1 (ATTTCCCCGAAA) and FccRI (ATTTCCCAGAAA) [28] While the close conserva-tion of these three elements suggests a highly con-served binding motif, other binding site studies have suggested that STAT2-containing complexes can bind
to sequences that are distinct from canonical ISRE and GAS elements [18,29] Thus, there is a degree of promiscuity in binding to various DNA motifs that may facilitate STAT-mediated transcriptional regula-tion across a broader range of genes [30,31] In our
Jak1
STAT1 STAT2
IFN-α
Tyk2
NMBR DGKE MSR1 CLDN4 BF DES PIN1L TLR3 JUND IL-20Rα
CLDN4
membrane tight junctions
DGKE
PKC isoforms
regulates signaling
JUND
IFI-202 / p202
regulation of proliferation
BF
complement cascade
PROLIFERATION
MSR1
lipoprotein uptake
MAPKs
IL-20Rα
DES mitochondrial structural integrity, intracellular signaling
DES nuclear shape
DES
gene expression regulation
TLR3
PI3K
NMBR
PIN1L
proliferation
regulator of mitosis
extracellular
cytoplasm
nucleus
Fig 4 ISGF3-independent STAT2-dependent ISGs in a signaling context Schematic representation of potential pathway interactions between known IFN signaling effectors and factors whose expression was found to be regulated by ISGF3-independent STAT2-containing complexes.
Trang 9study, the 5¢ flanking regions of the identified
ISGF3-independent STAT2-dependent ISGs contained
GAS-like sequences in their promoters in which the core
GAS consensus sequence was uniformly conserved but
the spacer nucleotides varied (Figs 2 and 4)
Two of the genes identified in Table 2 as regulated
by ISGF3-independent STAT2-containing complexes,
namely BF and JUND, have been previously
character-ized as ISGs [19,20] BF is an early component of the
alternative complement activation pathway important
for the cellular antiviral response [32,33] Interestingly,
C1s, an early component of the classical complement
cascade, was up-regulated upon IFN stimulation of
both U6A-2 and U6A-2VV-II cells (Table 1) This
lat-ter observation suggests that ISGF3 complexes mediate
IFN-inducible classical complement activation, while
ISGF3-independent STAT2-containing complexes may
regulate activation of the alternative complement
cascade CLDN4, a component of intracellular
junc-tions that regulate paracellular ion flux, may be a
potential mediator of IFN-induced antitumor
res-ponses Increased levels of this protein have been
detected in various tumor cell lines [34–36] CLDN4 is
negatively regulated by TGF-b, positively regulated
by Ras signaling, and restricts the invasiveness and
metastatic potential of pancreatic cancer cells [34]
The JUND proto-oncogene may also influence the
antiproliferative activity of IFNs JUND has been
implicated in the activation of the IFN-inducible
protein, p202 [37] In association with E2F, p202
inhibits cell growth by abrogating E2F1-mediated
tran-scriptional activation of S-phase genes driving cellular
proliferation [38]
It is less obvious how other genes identified in Table 2
are related to IFN-mediated activities, as none has been
previously characterized as an ISG Nevertheless, the
case can be made for several of these genes to be linked
to different aspects of IFN biology For example,
although its precise function remains unknown, the
alpha chain of the IL-20 receptor, IL-20RA, mediates
the signaling of cytokines that are involved in immune
regulation and inflammatory responses, namely IL-19,
IL-20, IL-24 and IL-26 [39–43] Therefore, IFN
regula-tion of IL-20RA will affect various aspects of the innate
and adaptive immune response DGKE encodes a
diacyl-glycerol kinase that influences the diacydiacyl-glycerol-protein
kinase C pathway [44], associated with CLDN4
assem-bly and membrane integrity [45] As suggested above,
IFN regulation of CLDN4 may be associated with
antiproiferative activity DES encodes a filamentous
protein involved in cytoskeletal organization and the
control of nuclear shape and has also been implicated
in intracellular signaling and the regulation of gene
expression [46,47] How other genes, such as PIN1L, MSR1and NMBR, might function as ISGs is currently
a matter of speculation
As well as the nine known genes cited above, our BSTVQ comparison of the gene expression profiles of IFN-treated U6A-2 and U6A-2VV-II cells revealed an additional 10 differentially expressed transcripts that encode proteins with unknown functions It remains to
be determined how these transcripts influence the bio-activity of IFNs The ongoing challenge is to define the sequence of events occurring postreceptor engagement
by IFNs-a⁄ b that distinguish specific signaling cascades leading to specific biological outcomes Future investi-gations of the nature of the newly identified ISFG3-independent STAT2-dependent ISGs cited in this study may shed light on these issues
Experimental procedures
Cells and reagents
Human fibroblast U6A (null for STAT2) cells were obtained from G Stark (Cleveland Clinic Foundation, Cleveland, OH) U6A-2 (overexpresses wild-type STAT2) cells and U6A-2VV-II (overexpresses STAT2 lacking DNA binding domain function) cells have been described previously [14] Cells were cultured in Dulbecco’s modified Eagle’s medium (Invitrogen, Carlsbad, CA, USA), supplemented with 10% (v⁄ v) fetal bovine serum (HyClone, South Logan, UT, USA),
100 UÆmL)1 penicillin, 100 mgÆmL)1 streptomycin (Invitro-gen) and 250 lgÆmL)1 Hygromycin B (Calbiochem, Missis-sauga, ON, Canada) Human recombinant IFN-alfacon-1 (specific activity 3.0· 109UÆmg)1) was provided by L Blatt (Intermune, Brisbane, CA)
RNA preparation for Affymetrix microarray analysis
To prepare total cellular RNA, cells were either left untreated
or treated with 5 ngÆmL)1 IFN-alfacon-1 for 6 h at 37C Cell pellets were lysed and homogenized using Qiagen (Mis-sissauga, ON, Canada) QIA-shredder columns and RNA isolation was performed using the Qiagen RNeasy mini-kit according to the manufacturer’s protocol The preparation
of cDNAs, sample hybridization and scanning of HG-U-133
A GeneChipArrays (Affymetrix, Santa Clara, CA, USA) was performed at the Centre for Applied Genomics Micro-array Facility (Hospital for Sick Children, Toronto, ON) in accordance with procedures established by Affymetrix
Microarray data analysis
Raw microarray data were normalized and analyzed using both the genespring version 6.1 (Silicon Genetics, Santa
Trang 10Clara, CA, USA) and binary tree-structured vector
quanti-zation (BTSVQ) software programs [26,48] Analysis using
the genespring program was performed as follows: (a)
raw microarray data were first normalized according to
default settings to ensure per chip normalization; (b) data
were filtered to exclude raw data readings lower than 80;
(c) to obtain the IFN-inducible gene profiles of each cell
category, sample–sample normalization was performed
using the untreated sample as the control; (d) these
normal-ized data were then filtered to include only those with
pre-sent or marginal flags
To analyze the complete set of raw microarray data
with-out exclusions, the btsvq method was employed btsvq is
an unbiased computational system that combines partitive
k-means clustering and SOMs to analyze and visualize
microarray gene expression data [26] This tool enables the
analysis and clustering of gene expression data without
pre-conceived bias Partitive k-means clustering is a statistical
method of dividing data into a predefined number of
clus-ters The btsvq program uses k¼ 2 such that, at each level,
the data are partitioned into two groups based on the
degree of similarity of their gene expression profiles This
hierarchical clustering method generates a binary tree that
can be used to determine which sample types have the most
similar gene expression profiles The averaged gene
expres-sions of individual gene clusters are then projected into a
color space to visualize the multidimensional data (SOM
mapping) The SOM algorithm clusters genes with similar
levels of expression and assigns the average level of gene
expression a color value Regions in red represent highly
expressed or present genes and those in blue represent
unexpressed or absent genes The intensity of the color is
also informative as a darker shade indicates a greater
degree of expression of genes represented by that region
than does a paler shade
Complementary DNA synthesis and real-time
PCR
Cells were either left untreated or treated with 5 ngÆmL)1
IFN-alfacon-1 for 6 h at 37C Cells were lysed and
homo-genized using Qiagen QIA-shredder columns and RNA
isolation was performed as described above cDNAs were
synthesized using 1 lg RNA in the presence of random
primers and AMV Reverse Transcriptase (Promega,
Madi-son, WI, USA) for 1 h at 42C
Components for real-time PCR were obtained from the
LightCycler FastStart Plus DNA Master SYBR Green I
kit (Roche) The LightCycler instrument (Roche,
Missis-sauga, ON, Canada) and relative quantification
soft-ware were used for all reactions PCR reactions were
performed in a final volume of 20 lL containing 0.5 lm of
each primer and 5 lL template cDNA (concentration
100 ngÆlL)1) The primer sets used are listed in Table S1
Standard curves were established for each primer set and
reference (b-actin) and target reactions were performed in triplicate for each sample
Promoter analysis
The 5¢ flanking sequences were obtained from the NCBI Entrez Gene database (http://www.ncbi.nlm.nih.gov/entrez/ query.fcgi?db¼ gene) Promoters were assessed for poten-tial STAT-binding sites using the gene2promoter and matinspectorprograms (Genomatix; http://www.genomatix de) [49] The NCBI Gene ID numbers were as follows: c-fos (2353), GBP1 (2633), PKR (5610), Mx1 (4599), 2¢-5¢OAS (4939), TLR3 (7038), CLDN4 (1364), BF (629), NMBR (4829), IL20RA (53832), DES (1674), DGKE (8526), PIN1L (5301), MSR1 (4481) and JUND (3727)
Pathway analysis
Pathway analysis was conducted using pathwayassist soft-ware (Iobion Informatics LLC, Stratagene, La Jolla, CA, USA) and the Online Predicted Human Interaction Data-base (OPHID; http://ophid.utoronto.ca) OPHID is a web-based database of about 40 000 predicted and known human protein–protein interactions [27]
Acknowledgements
This study was supported by Canadian Institutes of Health Research Grant MOP 15094 (to E.N.F.); National Science and Engineering Research Council of Canada (NSERC) Grant 203833-02, the Institute for Robotics and Intelligent Systems, and IBM (to I.J.)
References
1 Stark GR, Kerr IM, Williams BR, Silverman RH & Schreiber RD (1998) How cells respond to interferons Annu Rev Biochem 67, 227–264
2 Pestka S, Krause CD & Walter MR (2004) Interferons, interferon-like cytokines and their receptors Immunol Rev 202, 8–32
3 Brierley MM & Fish EN (2002) Review: IFN-alpha⁄ beta receptor interactions to biologic outcomes: under-standing the circuitry J Interferon Cytokine Res 22, 835–845
4 Uddin S & Platanias LC (2004) Mechanisms of Type-I interferon signal transduction J Biochem Mol Biol 37, 635–641
5 Platanias LC, Uddin S & Colamonici OR (1994) Tyr-osine phosphorylation of the alpha and beta subunits
of the type I interferon receptor Interferon-beta selec-tively induces tyrosine phosphorylation of an alpha subunit-associated protein J Biol Chem 269, 17761– 17764