Download Protein interaction for an interferon

Rheumatology 2003;42:1155–1163
doi:10.1093/rheumatology/keg315, available online at www.rheumatology.oupjournals.org
Advance Access publication 30 May 2003
Protein interaction for an interferon-inducible
systemic lupus associated gene, IFIT1
S. Ye1,4, H. Pang2, Y.-Y. Gu1, J. Hua1, X.-G. Chen1, C.-D. Bao1,
Y. Wang1, W. Zhang1, J. Qian1, B. P. Tsao3, B. H. Hahn3,
S.-L. Chen1,4, Z.-H. Rao2 and N. Shen1,4
Objective. To identify disease-related genes and immune-regulatory pathways in
the pathogenesis of systemic lupus erythematosus (SLE) by using gene expression
profiling and protein–protein interaction analysis.
Methods. Peripheral white blood cell gene expression profiles of 10 SLE patients
were determined by oligonucleotide microarray analysis. Clustering of the gene
expression profile was compared with the clinical immune phenotype. SLE-induced
genes that were over- or under-expressed were determined and independently
validated using a real-time polymerase chain reaction (PCR) method. To study
their potential function and the possible pathways involved, a candidate gene was
cloned and a GST (glutathione S-transferase) fusion protein was expressed in
Escherichia coli. The fusion protein was further purified using the glutathione
Sepharose 4B system, and was treated as bait to capture prey from SLE peripheral white blood cell lysate. MALDI-TOF (matrix-assisted laser desorption/
ionization–time-of-flight) mass spectrometry was then performed to determine the
prey protein.
Results. Similarity was found between the gene expression profile and the immune
phenotype clusters of the SLE patients. More than 20 disease-associated genes
were identified, some of which have not been related to SLE previously. Of these
genes, a cluster of interferon-induced genes were highly correlated. IFIT1
(interferon-induced with tetratricopeptide repeats 1) was one of these genes, and
overexpression of its mRNA was confirmed independently by real-time PCR in a
larger population (40 SLE patients and 29 normal controls). An IFIT1 protein–
protein interaction study showed that IFIT1 may interact with Rho/Rac guanine
nucleotide exchange factor.
Conclusion. The gene expression profile seems to be the molecular basis of the
diverse immune phenotype of SLE. On the basis of the SLE-related genes found in
this study, we suggest that the interferon-related immune pathway is important in
the pathogenesis of SLE. IFIT1 is the first gene described as a candidate gene for
SLE, and may function by activating Rho proteins through interaction with Rho/
Rac guanine nucleotide exchange factor. IFIT1 and the interferon-related pathway may provide potential targets for novel interventions in the treatment of SLE.
KEY WORDS: Systemic lupus erythematosus, Gene expression profile, Interferon,
Interferon-induced protein with tetratricopeptide repeats 1, Protein–protein interaction.
1
Shanghai Clinical Centre of Rheumatic Diseases and Institute of Rheumatology, Department of Rheumatology, Renji Hospital, Shanghai Second
Medical University, Shanghai, 2Structural Biology Laboratory, Division of Biology, Tsinghua University, Beijing, China, 3Division of
Rheumatology, Department of Medicine, University of California-Los Angeles, Los Angeles, CA 90095, USA and 4Health Science Centre, Institute
for Biological Sciences, Chinese Academy of Sciences and Shanghai Second Medical University, Shanghai, China.
Submitted 4 November 2002; revised version accepted 12 February 2003.
Correspondence to: N. Shen, Department of Rheumatology, Renji Hospital, 145, Shandong Mid. Road, 200001, Shanghai, China. E-mail:
[email protected]
1155
Rheumatology Vol. 42 No. 10 ß British Society for Rheumatology 2003; all rights reserved
S. Ye et al.
1156
Systemic lupus erythematosus (SLE) is a prototypical
autoimmune disease and its aetiology remains undetermined. In this era of functional genomics, the
microarray provides an appropriate high-throughput
method of investigating gene regulation in SLE from
an overall and relatively non-biased perspective [1].
Proteomics has also provided powerful tools to
delineate disease pathways in terms of the network of
protein–protein interactions [2]. In this study, we used
these strategies to identify disease-related genes and
the possible pathways by which they function in the
pathogenesis of SLE.
Materials and methods
Patients
Ten SLE patients (one male and nine female, age
28.8 7.9 yr) at the Clinical Centre of Rheumatology at
Shanghai Renji Hospital were enrolled in our study. The 10
patients were identified as A/A0 , B, C, D, E, F, G, H, S1 and
S2; A0 denotes a duplicated microarray test of patient A after
treatment. Patients S1 and S2 were an SLE sib-pair.
All patients except S1 were newly diagnosed and had
received no therapeutic intervention (corticosteroids and
cytotoxic agents). All clinical data were recorded according to
the Systemic Lupus Activity Measure (SLAM) [3] (the scores
of the 10 SLE patients ranged from 3 to 12 and averaged
6.8). Eleven items were added, viz. IFANA, double-stranded
DNA (dsDNA) antibody, Sm, SSA, SSB, u1RNP, Rib P,
anticardiolipin antibodies (ACL), 2-glycoprotein I (2GP1),
histone autoantibodies and hypocomplementaemia. This
made a total of 43 items, which were together designated
‘SLAM plus’. In SLAM plus, dsDNA antibody was recorded
as 0, 1, 2 according to the titre [<20, 20–50 and >50 U/ml
(Farr assay) respectively]. Other additional items were
recorded as 0 or 1, which represented negative and positive
respectively. Hep2 cells were used as substrates in the
immunofluorescence antinuclear antibody (IFANA) test;
1:80 were considered positive (for clinical data see Table
1). Ten chips from age- and sex-matched healthy individuals
served as normal controls. Another 40 SLE patients and 29
healthy individuals (matched for age and sex) were included
for analysis by real-time polymerase chain reaction (PCR).
Pooled white blood cell lysate was prepared from the blood
of an additional 10 SLE patients (sonicated) in 1 mM PMSF/
PBS (phenylmethylsulphonyl fluoride/phosphate-buffered
saline) buffer for use in the protein–protein interaction
study. All SLE patients fulfilled at least four criteria for SLE
(1997 revised criteria of the American College of
Rheumatology). All patients and controls were from the
Chinese Han population.
Microarray gene expression profiling [4]
A sample of 20 ml peripheral blood (anticoagulated with acidcitrate dextrose) was drawn from each patient and control
subject, and was treated with red blood cell lysis buffer.
Total RNA was extracted using TRIzol Reagent (Life
Technologies, Carlsbad, CA, USA).
For the microarray analysis, we used the ExpressChipTM
DNA Oligonucleotide Microarray System HO4 (Mergen, San
Leandro, CA, USA). The chips contain 3360 spots, including
96 blank spots, eight spots for negative controls and 88 spots
for housekeeping genes as positive controls. Genelist
TABLE 1. Clinical features of 10 SLE patients
Patient
Clinical features
Systemic symptoms (fatigue/fever/weight loss)
Malar rash/oral ulcers/photosensitivity
Cutaneous vasculitis
Arthritis/arthralgia
Diffuse lymphoadenopathy/hepatosplenomegaly
Pleuritis/pericarditis
Pulmonary involvement
Gastrointestinal vasculitis
Glomerulonephritis
Anaemia
Leucopenia
Thrombocytopenia
ESR elevation
Hypocomplementaemia
ANA
Anti-dsDNA (U/ml)
Sm
RNP
SSA
SSB
Rib P
ACL/B2GP1
Histone
A
D
G
B
E
H
F
S1
S2
C
þ
þ*
–
–
–
þ*
–
–
þ*
–
–
–
þ*
þ
þ
>50*
þ
þ
–
–
–
–
þ
þ
þ
þ
þ
–
þ
–
–
–
–
–
–
þ
þ
þ
þ
–
þ
–
–
–
–
–
–
þ
þ
–
–
–
–
–
–
–
–
þ
þ
–
þ
þ
þ
–
–
–
þ
þ
–
–
–
–
þ
–
–
–
–
þ
–
þ
þ
–
þ
þ
>50
–
þ
–
–
–
–
–
–
–
–
þ
–
þ
–
–
–
–
–
þ
–
–
þ
>50
–
þ
þ
–
–
–
–
þ
þ
–
þ
–
–
–
–
þ
þ
–
þ
þ
þ
þ
þ
–
–
–
–
þ
–
–
–
þ
–
þ
–
–
–
–
þ
þ
–
þ
þ
–
þ
–
–
–
þ
þ
–
–
–
–
þ
–
þ
–
þ
–
–
–
þ
þ
þ
–
–
þ
>50
–
–
–
–
–
–
–
–
þ
–
þ
–
–
–
–
–
–
–
–
–
–
þ
þ
–
–
–
–
–
–
–
–
þ
–
þ
–
–
–
þ
þ
–
þ
–
–
–
–
þ
–
–
–
–
–
þ
–
Data based on a modification of SLAM (see text). Includes part of the autoantibody profile. The sequence of SLE patients is the same as that
used for the clustering data obtained by the use of SLAM plus (Fig. 1).
*Improvement after treatment.
ESR, erythrocyte sedimentation rate; ANA, antinuclear antibodies; RNP, ribonucleoprotein.
Protein interaction for an interferon-inducible SLE associated gene
1157
FIG. 1. (A) Clustering based on the gene expression profile in SLE patients. (B) Clustering based on the clinical immune
phenotype (the SLAM plus autoantibody spectrum) in the same SLE patients as those represented in (A). The 10 patients are
identified as A/A0 , B, C, D, E, F, G, H, S1 and S2, where A0 represents patient A after therapy, when the microarray analysis
and clinical evaluation were repeated. Patients S1 and S2 are a SLE sib-pair. N represents normal controls. There is a
correlation between the gene expression profile and immune phenotype.
(http://www.mergen.com/HO4/HO4finder.asp) lists most of
the immune-associated genes included. The experiments were
performed according to the manufacturer’s protocol.
Total RNAs were retrotranscribed into cDNA using a
cDNA synthesis kit (Roche, Basel, Switzerland). The primer
was oligo[(dT)24T7 promoter]65, and ds-cDNAs were
synthesized. After in vitro transcription from ds-cDNA
(MEGAscriptTM transcription kit; Ambion, Austen, TX,
USA), biotin-labelled cRNA probes (Biotin-CTP; Gibco
BRL, Carlsbad, CA, USA) were obtained. The microarray
was hybridized with the cRNA probe overnight at 30 C. After
washing, blocking, applying streptavidin and first and second
antibodies (labelled with Cy3), detection was performed with a
ScanArray 5000 laser confocal microarray scanner (GSI
Lumonics, Billerica, MA, USA). The data were exported by
QuantArray microarray analysis software (GSI Lumonics).
GeneSpringTM microarray analysis software version 4.2.1
(Silicon Genetics, Redwood City, CA, USA) was used to
mine the gene expression data. After median normalization for
each chip, the Welch t-test and Welch ANOVA (analysis of
variance) were used to compare SLE patients with the control
group. Differences were considered significant if P < 0.05. Sub-
sequently, conventional two-fold change ratio analysis was
performed separately, and the global error model was used (for
the algorithm, see Fig. 2). The Spearman correlation was used
in gene expression profiling and SLAM plus-based immune
phenotype clustering.
IFIT1 mRNA expression in SLE using real-time
PCR [5]
To re-evaluate the expression pattern of the interferoninduced with tetratricopeptide repeats 1 (IFIT1) gene
independently, total RNA from another 40 SLE patients
was reverse-transcribed into cDNA using the Superscript II
RT kit (Invitrogen Life Technologies, Carlsbad, CA, USA).
The control group comprised 29 healthy individuals. Realtime PCR (ABI 7900, Applied Biosystems, Foster City, CA,
USA) was performed according to the manufacturer’s
protocol. The primers and Taqman probe for IFIT1 were
as follows: IFIT1-F: 50 -GCCTCCTTGGGTTCGTCTATAA30 , IFIT1-R: 50 -TCAAAGTCAGCAGCCAGTCTCA-30 .
IFIT1-TaqMan probe: FAM-AGCCCTGGAGTACTATG
AGCGGGCC-TAMRA. GAPDH (glyceraldehyde 3-phos-
1158
S. Ye et al.
protein was eluted with 5 mmol/l reduced glutathione/PBS.
Then 12% SDS–PAGE (sodium dodecyl sulphate–polyacrylamide gel electrophoresis) was performed.
GST–IFIT1 as bait to capture prey protein
[8–9] and MALDI-TOF (matrix-assisted
laser desorption/ionization–time-of-flight)
mass spectrometry analysis
FIG. 2. Algorithm to detect significant differences in the gene
expression profile between SLE patients (n ¼ 10) and normal
controls (n ¼ 10). *The two-fold change cut-off and global
error model methods show similar results, in that 28 gene
spots obtained with the former method are included in the 29
gene spots obtained with the latter method. Three of these 28
gene spots on the HO4 chip are duplicated. Thus, we
obtained 25 non-redundant genes.
phate dehydrogenase) is treated as internal control. GAPDHF: 50 -GAAGGTGAAGGTCGGAGTC-30 , GAPDH-R: 50 GAPDH-TaqMan
GAAGATGGTGATGGGATTTC-30 .
Probe: FAM-CAAGCTTCCCGTTCTCAGCC-TAMRA. In
each cycle, fluorescent signals of both IFIT1 and GAPDH
were collected during the PCR in order to determine their Ct
values. The value of Ct (IFIT1 Ct minus GAPDH Ct), which
was inversely correlated with the number of copies of the
IFIT1 mRNA, was then calculated and compared between
the two groups. SSPS 9.0 statistics software (SPSS, Chicago,
IL, USA) was used to perform the t-test, and P < 0.05 was
considered significant.
Molecular cloning of IFIT1 and expression of
GST–IFIT1 fusion protein [6–7]
The full-length coding sequence of IFIT1 (1437bp) was
amplified with the primer (forward) 50 -CGCGGATCCA
TGAGTACAAATGGTGAT-30 , (reverse) 50 -TCCGCTCGA
GCTAAGGACCTTGTCTCAC-30 . After purification of the
PCR product and BamHI/XhoI digestion, IFIT1 was inserted
into plasmid pGEX-6P-1. The vector was transfected into the
cloning host Escherichia coli XL-1-Blue. The recombinant
plasmid was screened by PCR and 1% agarose gel
electrophoresis, and was confirmed by automated
DNA sequencing (service provided by Genecore, Shanghai,
China).
pGEX-IFIT1 was transfected into the expression clone
host, E. coli BL21(DE3) cells. Isopropylthiogalactoside
(IPTG) (0.1 mmol/l) was added to induce protein expression.
pGEX-6P-1 vector-transfected BL21 cells [which express
glutathione S-transferase (GST) only] and non-transfected
BL21 cells were treated as controls. Escherichia coli
organisms were then harvested and lysed in PBS pH 7.3/
1 mM PMSF by sonication. The supernatant was added to a
prepacked glutathione Sepharose 4B column (Amersham
Pharmacia Biosciences, Piscataway, NJ, USA). After thorough rinsing with PBS (pH 7.3) to baseline, the fusion
GST–IFIT1 and GST were bound to two glutathione
Sepharose 4B columns separately. Pooled peripheral white
blood cell lysate from SLE patients was made to flow
through the GST column in order to remove non-specific
proteins which bind to GST or matrix. The flow-through was
then collected and added to the GST–IFIT1 column to
interact with the GST–IFIT1. After thorough rinsing with
PBS to baseline, PBS–saline gradient washing was performed,
using 5 volumes of column-bed PBS–NaCl with saline
concentrations of 150, 200, 400, 600, 800, 1000 and 2000 mM;
these were added sequentially to the column to elute the
protein(s) captured by IFIT1. A concentrator tube with 5000
Da molecular weight cut-off (Millipore, Billerica, MA, USA)
was used to reduce the collection volume of each gradient to
50 l. SDS–PAGE (12%) was then performed. After
Coomassie Blue staining and depigmentation, the most
significant new protein band was cut out from the gel and
subjected to digestion and MALDI-TOF mass spectrometry
(MS). The definition of the most significant new protein band
(presumably with a higher specificity) was as follows: (i) it
was a new band other than the fusion protein or GST; (ii) it
was obtained at saline concentration of 400 mM [8]; and (iii)
there was enough protein present for MS identification to be
possible. MALDI-TOF MS was performed by National
Centre of Biomedical Analysis, Beijing.
Results
SLE microarray gene expression profiling
Results of gene expression profiling and the clinical
immune phenotype data showed that when clustering
was found before treatment it also tended to be found
after treatment. In addition, sib-pairs were also
clustered together, i.e. patients A/A0 , S1 and S2 were
close neighbours in the cluster tree (Fig. 1). Moreover,
there seemed to be an association between the gene
expression profile and the clinical immune phenotype.
Patients A and D, who had only cutaneous–mucosal
lesions and serositis, were clustered together at one end
of the gene expression or clinical manifestation
spectrum. Patient C, who had predominantly gastrointestinal vasculitis, was located at the other end of
spectrum. The remaining patients had mainly haematological involvement and/or glomerulonephritis.
Identification of SLE-related genes
Testing the reproducibility of our oligonucleotide
microarray system showed that when probes from the
same RNA sample were hybridized to two chips
separately, the correlation coefficient (r2) for all signals
was 0.93; however, 4.8% of the genes showed a
difference of more than 2-fold. To increase the
reliability of our microarray data and to optimize the
Protein interaction for an interferon-inducible SLE associated gene
1159
TABLE 2. The 25 SLE-related genes included in the study
Fold change (S.D.)
GeneBank ID
Gene symbol
Up-regulated
M87434
M24594*
M83202
AJ225089
M33326
U56145
M31165*
M83667
X04371
J05070
OAS2
IFIT1
LTF
OASL
CEACAM8
LY6E
TNFAIP6
CEBPD
OAS1
MMP9
5.038
4.63
4.606
4.582
3.285
3.157
2.947
2.783
2.736
2.545
(4.300)
(3.283)
(4.489)
(3.871)
(3.031)
(2.041)
(1.881)
(1.162)
(1.930)
(1.973)
U73191
U19251
M62762
AF032387
U27467
AF083470
KCNJ15
NAIP
ATP6C
SNAPC4
BCL2A1
IFIT4
2.486
2.441
2.347
2.29
2.183
2.155
(1.244)
(1.419)
(1.433)
(1.259)
(0.869)
(1.322)
X52015*
U32849
L24122
U26173
U13698
Down-regulated
L08177
IL1RN
NMI
NFE2
NFIL3
CASP1
2.153
2.142
2.071
2.043
2.0
(1.641)
(1.198)
(0.972)
(1.179)
(1.025)
EBI2
0.477 (0.163)
X06948
X15260
AF057557
FCER1A
TCRD
TOSO
0.464 (0.105)
0.391 (0.209)
0.38 (0.124)
Remarks
0
0
2 ,5 -Oligoadenylate synthetase 2
Interferon-induced protein 56
Lactotransferrin
0
0
2 ,5 -Oligoadenylate synthetase-like
Carcinoembryonic antigen-related cell adhesion molecule 8
Lymphocyte antigen 6 complex, locus E
Tumour necrosis factor -induced protein 6
CCAAT/enhancer
binding protein (C/EBP) 0
0
2 ,5 -Oligoadenylate synthetase 1
Matrix metalloproteinase 9 (gelatinase B, 92 kDa
gelatinase, 92 kDa type IV collagenase)
Potassium inwardly rectifying channel, subfamily J, member 15
Neuronal apoptosis inhibitory protein
ATPase, Hþ transporting, lysosomal (vacuolar proton pump) 16 kDa
Small nuclear RNA activating complex, polypeptide 4, 190 kDa
BCL2-related protein A1
Interferon-induced protein with tetratricopeptide repeats 4, between
WI-10247 and WI-6075
Interleukin 1 receptor antagonist
N-myc (and STAT) interactor
Nuclear factor (erythroid-derived 2), 45 kDa
Nuclear factor, interleukin 3-regulated
Caspase 1, apoptosis-related cysteine protease (interleukin 1 convertase)
Epstein–Barr virus-induced gene 2 (lymphocyte-specific
G protein-coupled receptor)
Fc fragment of IgE, high-affinity I, receptor for polypeptide
T-cell receptor (V, D, J, C)
Regulator of Fas-induced apoptosis
Genes were considered significant when P < 0.05 (Welch t-test) in the group comparison of SLE (n ¼ 10) with normal controls (n ¼ 10) and
the relative expression ratio for SLE patients exceeded the 2-fold cut-off.
*Duplicated genes. Ratios are means. Further independently validated genes are indicated by GeneBank numbers in bold type (data not
provided except for IFIT1).
experimental procedure and validation method, we
performed a combination of the Welch t-test with
conventional two-fold change filtering analysis. The
results were similar to those obtained with a method
using the global error model. To some extent, statistical
significance is more important than an artificial twofold cut-off [10]. Ultimately, we obtained 25 nonredundant, over- or under-expressed, statistically significant SLE genes (Table 2). The clustering of these
SLE-induced genes shown in the array results indicated
that there was a strongly correlated gene cluster that
contained genes that were all interferon-inducible.
These genes were IFIT1, IFIT4, OAS2, OAS1, OASL
and Ly6E (Fig. 3). By using real-time PCR in a larger
independent population, we validated the microarray
expression data for IFIT1, IFIT4, OAS2, Ly6E and C/
EBPD (data not provided except for IFIT1). Real-time
PCR confirmed that the level of expression of IFIT1
mRNA was significantly up-regulated in SLE (n ¼ 40)
compared with normal controls (n ¼ 29; P < 0.001)
(Fig. 4). This reinforces the reliability of our microarray
data and algorithm.
Next we focused on the candidate gene IFIT1, whose
function is not clear and has not been related to SLE
in the literature.
Molecular cloning of IFIT1 and expression and
purification of the GST–IFIT1 fusion protein
The recombinant pGEX-IFIT1 plasmid was constructed, and automated DNA sequencing of the
positive clone showed that the inserted DNA was
identical to the human IFIT1 gene sequence, with no
shift in the reading frame.
After IPTG induction, SDS–PAGE showed that
vector pGEX-6P-1 transfection led to expression of a
26 kDa GST fragment. Recombinant pGEX-IFIT1
plasmid transfection led to expression of a 82 kDa
protein that matched the predicted size of the GST–
IFIT1 fusion protein (the molecular weight of IFIT1 is
56 kDa, and adding 26 kDa for GST gives 82 kDa).
The GST–IFIT1 fusion protein was purified with a
glutathione Sepharose 4B column (Fig. 5).
Identifying IFIT1-interacting proteins by
MALDI-TOF MS peptide mass fingerprinting
GST–IFIT1 was treated as a bait to capture prey
protein from a pooled lysate of peripheral white blood
cells from SLE patients. Pretreatment removed proteins
that may have interacted with GST or the matrix.
Because protein interaction occurs mainly through
hydrophobic bonds, the specificity of the IFIT1 protein
1160
S. Ye et al.
FIG. 3. Clustering of 25 SLE-associated genes from the microarray analysis. A cluster of IFN-induced genes, including the
candidate gene IFIT1, is highlighted.
FIG. 4. Level of expression of IFIT1 mRNA in SLE patients
(n ¼ 40) and normal controls (NC; n ¼ 29), determined by real
time-PCR. Ct is inversely correlated with the number of gene
copies. IFIT1 is up-regulated in SLE (P < 0.001). The lines
are the standard deviation.
FIG. 5. GST-IFIT1 fusion protein purification. Lane 1,
protein marker; lane 2, pre-IPTG induction; lane 3, postIPTG induction; lane 4, supernatant; lane 5, precipitants;
lane 6, flow-through; lane 7, purified GST–IFIT1 fusion
protein.
partner would be expected to increase as the saline
concentration of elution was increased. SDS–PAGE
showed that there were no visible protein bands when
the saline concentration reached 1000 mM. We obtained
a significant new band in the 40–50 kDa range at the
concentration of 800 mM. (white arrow in Fig. 6).
In the identification of protein(s) interacting with
IFIT1 by MALDXI-TOF MS, candidate proteins had
to fulfil the following conditions: (i) it must not be a
fragment of the GST–IFIT1 fusion protein; (ii) it must
not be a protein from E. coli; and (iii) common
contaminants in MS analysis, such as keratoprotein,
immunoglobulin G, heat-shock proteins and ribosomal
proteins, had to be excluded [11]. MALDI-TOF MS
peptide mass fingerprinting and database searching
(Mascot, http://www.matrixscience.com) identified the
IFIT1 partner as a set of highly similar Rho/Rac
guanine nucleotide exchange factors (GEFs) (Fig. 7).
The quality of MALDI-TOF data is quite good,
but the database search score was not very high,
ranging from 42 to 43 (this score is a measure of the
statistical significance of a match). There were two
Protein interaction for an interferon-inducible SLE associated gene
1161
possibilities: (i) that our captured protein band was a
mixture of several GEF homology proteins; and (ii)
that the protein was a novel protein of the GEF family.
Discussion
FIG. 6. Using a GST tag to capture proteins that interact
with IFIT1. Lane 1, protein marker; lane 2, rinsing with PBS
plus NaCl 400 mM; lane 3, 600 mM NaCl; lane 4, 800 mM
NaCl; lane 5, 1000 mM NaCl; lane 6, 2000 mM NaCl. In order
to undergo further MS identification, a new protein band
(other than fusion protein or GST) had to have high
specificity, i.e. it had to appear at an NaCl concentration of
400 mM [8], and it had to be sufficiently intense to indicate
that there was enough protein for analysis. The white arrow
indicates a protein band (about 45 kDa) captured at 800 mM
NaCl. Black arrows indicate proteins present in insufficient
quantity (band at about 35 kDa) or GST contamination
(band at about 26 kDa).
Attempts have been made to classify disease subtypes
by microarray gene expression profiling in leukaemia
and lymphoma [12, 13]. In the present study, we found
a correlation between the gene expression profile and
the clinical immune phenotype in SLE patients. As our
sample of patients was small, the data we obtained do
not cover the variation of lupus adequately. An
extensive discussion of disease subtype classification at
the molecular level is not possible here. This can be
considered a pilot study, and our results merely
suggest one possible direction for future research in
rheumatology.
Our results indicate that interferon and its associated
immune regulatory pathway may play an important
role in SLE. Evidence has already been obtained that
FIG. 7. MALDI-TOF MS peptide mass fingerprinting of proteins interacting with IFIT1. A group of Rho/Rac GEF proteins
were identified.
Molecular
weight (Da)
Database
search score
gi|5011974
108 175
43
gi|8089027
108 246
43
gi|9744759
111 402
43
gi|6635129
112 016
42
Protein
Remarks
(NM_004723) Rho/Rac guanine nucleotide exchange
factor 2; proliferating cell nucleolar antigen p40
(BC020567) Rho/Rac guanine nucleotide exchange
factor (GEF) 2
(AF486838) guanine nucleotide exchange factor
GEF-H1
(AB014551) KIAA0651 protein
1162
S. Ye et al.
the interferon family, in particular interferons of type I,
is crucial in the pathogenesis in SLE. Blanco et al. [14],
obtained promising data indicating serum from SLE
patients can induce normal monocytes to be transformed into a dendritic cell phenotype. The effector in
the serum was interferon (IFN-). By adding antiIFN- antibody, this effect will be diminished. As a
strong form of antigen-presenting cell, the dendriticlike cell can capture apoptotic cells and nucleosomes
effectively, and then present self-nuclear antigen to
lymphocytes. This mechanism may result in the
breakdown of self-tolerance and autoimmunity in
SLE. On the other hand, nuclear antigen and the
anti-DNA/nucleosome antibody complex are strong
inducers of interferon [15], and this forms a feedback
circle including the antigen-presenting cell, interferon
and nuclear antigen. Our microarray expression
profiling revealed a related cluster of genes that are
inducible by interferon. These SLE-related genes
include IFIT1, IFIT4, OAS1, OAS2 and OASL. The
20 -50 -oligoadenylate synthetases (OASs) are induced
mainly by interferon type I. The function of OASs is
considered to be antiviral, and these enzymes have
been found to be up-regulated in the serum of SLE
patients [16]. We are particularly interested in IFIT1,
the function of which is not clear and has not been
related to SLE previously. We confirmed independently
that the expression of IFIT1 mRNA is significantly upregulated in a larger SLE population using real-time
PCR. Rozzo et al. [17], using microarrays to study
the congenic lupus murine model (NZB B6 F1),
demonstrated that the interferon-induced gene ifi202
may be a susceptibility gene in lupus, and argued that
ifi202 could be a target for intervention in SLE. Basic
local alignment search tool (BLAST) between human
IFIT1 and mouse ifi202 shows no homology between
the two genes.
How IFIT1 protein functions in SLE has been
studied using a proteomics strategy. Protein function
and disease pathways are all ultimately founded on
protein–protein interaction. Extensive study of protein–
protein interaction networks is the focus of the latest
developments in proteomics. The new strategy is to
clone the target protein in order to capture protein
complexes by using affinity-purification tags. Gavin et
al. [11] used a tandem affinity-purification (TAP) tag to
purify and identify yeast protein complexes. Marcotte
et al. [2] published a similar study, using the FLAGTag approach [FLAG is a hydrophilic octapeptide
(DYKDDDDL)]. Sophisticated two-dimensional gel
techniques and yeast two-hybrid systems were not
used, and MS has been used to identify protein
interactions, and to help find novel proteins. MALDITOF MS has high resolution and mass accuracy that
can meet the technical demands of protein interaction
research, and it has therefore become one of the most
important tools in proteomics. The GST pull-down
assay has been used to analyse interactions between
known candidate proteins for a number of years [8].
We have modified the GST pull-down assay to find
FIG. 8. Structure basis of IFIT1, which contains six TPR
domains (determined using www.smart.embl-heidelberg.de/
smart/show_motifs.pl).
unknown partner(s) of our candidate molecules.
Structure prediction shows that IFIT1 has six
tetratricopeptide repeats (TPR domains) (www.smart.embl-heidelberg.de/smart/show_motifs.pl) (Fig. 8).
The tetratricopeptide repeat is known to be the basis
of protein–protein interaction, and this could be the
structural basis of our findings.
MALDI-TOF MS peptide mass fingerprinting has
shown that IFIT1 may interact with Rho/Rac GEF.
GEFs play an important role in the regulation of Rho
protein activation and the downstream pathway. The
Rho proteins are small G proteins, or GTPases, and
there are several members of the family, including
Rho A, Rho C, Rac 1 and Cdc42. By interacting with
GEFs, Rho proteins switch from an inactivated GDPbinding form to an activated GTP-binding state. Rho
proteins are important components of intracellular
signal transduction, and are involved in cytoskeletal
rearrangement, cell cycle regulation, migration, phagocytosis and stress responses [18]. After activation by
GEFs, Rho proteins can activate the JNK (Jun-NH2terminal kinase) and p38 MAPK (mitogen-activated
protein kinase) pathways (Rho proteins can also
activate NFB (induce apoptosis) and precipitate Fcg
receptor-mediated phagocytosis [19–20]. IFIT1 may
cross-react with Rho/Rac GEF, thereby becoming
involved in SLE immune responses. Recently, on the
basis of the common regulatory pathway shared by Tand B-cell receptors and other immune receptors, some
researchers have proposed the concept of multi-subunit
immune-recognition receptors (MIRRs) [21], and have
claimed that a thoroughly studied member of the GEF
family, vav, may play a critical role in the integration
of immune signalling. However, the detailed characteristics of GEFs are still undetermined.
Our study has highlighted the interferon-related
pathway and an interferon-induced gene, IFIT1, and
the results we obtained may be relevant to the
pathogenesis of SLE. Our data show only that IFIT1
may interact with Rho/Rac GEFs. When this IFIT1
partner protein has been characterized by protein
sequencing, it will be possible to use the standard GST
pull-down assay to validate our results. Also, further in
vitro and in vivo studies are needed to determine the
exact role of IFIT1 in the interferon-related pathway of
SLE. Our approach is not suitable for the study of
protein interactions that depend on post-translational
modification, or for the detection of weak, transient—
but perhaps important—protein interactions. However,
our study does provide some interesting clues about
the immune regulatory pathway in SLE, and may
Protein interaction for an interferon-inducible SLE associated gene
help in the identification of targets for therapeutic
intervention.
Acknowledgements
This work was supported by the Chinese National
Natural Science Foundation (grant numbers 30000154,
30271224), Shanghai Science and Technology
Department Fund (01JC14029, 02QMB1404) and Fok
Ying Tung Education Foundation Young Teacher
Award (81030). We also thank the Chinese National
Centre of Biomedical Analysis (NCBA), Beijing, for
MS analysis. Our colleagues at UCLA are funded by
the USA Public Health Service grants and by
donations from the Dorough Foundation, the Paxson
Family and the Arthritis Foundation.
References
1. Molitor JA, Buckner JH, Nepom GT. Transcript array
analysis in rheumatology. Rheum Dis Clinics North Am
2002;28:155–63.
2. Marcotte EM, Pellegrini M, Thompson MJ et al. A
combined algorithm for genome-wide prediction of
protein function. Nature 2002;402:83–6.
3. Bae SC, Koh HK, Chang DK et al. Reliability
and validity of Systemic Lupus Activity Measure-Revised
(SLAM-R) for measuring clinical disease activity in
systemic lupus erythematosus. Lupus 2001;10: 405–9.
4. Relógio A, Schwager1 C, Richter A et al. Optimization of
oligonucleotide-based DNA microarrays. Nucleic Acids
Res 2002;30:e51.
5. Livak KJ, Schmittgen TD. Analysis of relative gene
expression data using real-time quantitative PCR and the
2 (-delta deltaC(T)) method. Methods 2001;25:402–8.
6. Sambrool J, Russell DW. Molecular cloning: A laboratory
manual, 3rd edn. New York: Cold Spring Harbor
Laboratory Press, 2001.
7. Smith DB, Johnson KS. Single-step purification of
polypeptides expressed in Escherichia coli as fusions with
glutathione S-transferase. Gene 1988;67:31–40.
8. Kaetlin JWG, Pallas DC, DeCaprio JZ et al. Identification
of cellular proteins that can interact specifically with the
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
1163
T/E1A-binding region of the retinoblastoma gene products.
Cell 1991;64:521–32.
Lachner M, O’Carroll D, Rea S et al. Methylation of
histone H3 lysine 9 creates a binding site for HP1 proteins.
Nature 2001;410:116–20.
Jin W, Riley RM, Wolfinger RD et al. The contributions of
sex, genotype and age to transcriptional variance in
Drosophila melanogaster. Nature Genetics 2001;29:389–95.
Gavin AC, Bosche M, Krause R et al. Functional
organization of the yeast proteome by systematic analysis
of protein complexes. Nature 2002;415:141–7.
Golub TR, Slonim DK, Tamayo P et al. Molecular
classification of cancer: Class discovery and class prediction
by gene expression monitoring. Science 1999;286:531–7.
Alizadeh AA, Eisen MB, Dacis RE et al. Distinct types
of diffuse large B-cell lymphoma identified by gene
expression profiling. Nature 2000;403:503–11.
Blanco P, Palucka AK, Gill M et al. Induction of
dendritic cell differentiation by IFN-a in systemic lupus
erythematosus. Science 2001;294:1540–3.
Bave U, Vallin H, Alm GV et al. Activation of natural
interferon-alpha producing cells by apoptotic U937 cells
combined with lupus IgG and its regulation by cytokines.
J Autoimmun 2001;17:71–80.
Preble OT, Rothko K, Klippel JH et al. Interferoninduced 20 -50 adenylate synthetase in vivo and interferon
production in vitro by lymphocytes from systemic lupus
erythematosus patients with and without circulating
interferon. J Exp Med 1983;157:2140–6.
Rozzo SJ, Allard JD, Choubey D et al. Evidence for an
interferon-inducible gene, Ifi202, in the susceptibility to
systemic lupus. Immunity 2001;15:435–43.
Matozaki T, Nakanishi H, Takai Y. Small G-protein
networks: their crosstalk and signal cascades. Cell Signal
2000;12:515–24.
Esteve P, Embade N, Perona R et al. Rho-regulated
signals induce apoptosis in vitro and in vivo by a p53independent, but Bcl2 dependent pathway. Oncogene
1998;17:1855–69.
Patel JC, Hall A, Caron E. Vav regulates activation of
Rac but not Cdc42 during FcgammaR-mediated phagocytosis. Mol Biol Cell 2002;13:1215–26.
Turner M, Billadeau DD. Vav proteins as signal
integrators for multi-subunit immune-recognition receptors. Nature Rev Immunol 2002;2:476–86.