(Received for publication, October 23, 1996, and in revised form, March 25, 1997)
From the Department of Microbiology and Immunology
and Walther Oncology Center, Indiana University School of Medicine,
Indianapolis, Indiana 46202, the
Department of Molecular
Immunology, King of Prussia, Pennsylvania 19406, and ** Human Genome
Sciences, Inc., Rockville, Maryland 20850
The tumor necrosis factor receptor (TNFR) superfamily consists of approximately 10 characterized members of human proteins. We have identified a new member of the TNFR superfamily, TR2, from a search of an expressed sequence tag data base. cDNA cloning and Northern blot hybridization demonstrated multiple mRNA species, of which a 1.7-kilobase form was most abundant. However, TR2 is encoded by a single gene which, maps to chromosome 1p36.22-36.3, in the same region as several other members of the TNFR superfamily. The most abundant TR2 open reading frame encodes a 283-amino acid single transmembrane protein with a 36-residue signal sequence, two perfect and two imperfect TNFR-like cysteine-rich domains, and a short cytoplasmic tail with some similarity to 4-1BB and CD40. TR2 mRNA is expressed in multiple human tissues and cell lines and shows a constitutive and relatively high expression in peripheral blood T cells, B cells, and monocytes. A TR2-Fc fusion protein inhibited a mixed lymphocyte reaction-mediated proliferation suggesting that the receptor and/or its ligand play a role in T cell stimulation.
The members of the tumor necrosis factor receptor (TNFR)1/nerve growth factor receptor (NGFR) superfamily are characterized by the presence of three to six repeats of a cysteine-rich motif that consists of approximately 30-40 amino acids in the extracellular part of the molecule (1). The crystal structure of TNFR-I complexed with its ligand showed that a cysteine-rich motif (TNFR domain) was composed of three elongated strands of residues held together by a twisted ladder of disulfide bonds (2). These receptors contain a hinge-like region immediately adjacent to the transmembrane domain, characterized by a lack of cysteine residues and a high proportion of serine, threonine, and proline, which are likely to be glycosylated with O-linked sugars. A cytoplasmic part of these molecules shows limited sequence similarities, a finding that may be the basis for diverse cellular signaling. At present, the members identified from human cells include CD40 (3, 4), 4-1BB (5), OX-40 (6), TNFR-I (7, 8), TNFR-II (9), CD27 (10), Fas (11), NGFR (12), CD30 (13), and LTBR (14). Viral open reading frames encoding soluble TNFRs have also been identified, such as SFV-T2 (9), Va53 (15), G4RG (16), and crmB (17).
Recent studies have shown that these molecules are involved in diverse biological activities such as immunoregulation (18, 19), by regulating cell proliferation (20-22), cell survival (23-25), and cell death (26-28).
Because of their biological significance and the diverse membership of this superfamily, we predicted that there would be further members of the superfamily. By searching an EST data base, we identified a new member of the TNFR superfamily. We report here the initial characterization of the molecule called TR2.
An EST cDNA data base, obtained from over 500 different cDNA libraries (29, 30), was screened for sequence
homology with cysteine-rich motif of the TNFR superfamily, using the
blastn and tblastn algorithms (31). One EST was identified in a human T
cell line library, which showed significant sequence identity to
TNFR-II at the amino acid level. This sequence was used to clone the
missing 5 end by RACE (rapid amplification of cDNA ends) using a
5
-RACE ends-ready cDNA from human leukocytes
(Clontech, Palo Alto, CA). This sequence matched
three further ESTs (HTOBH42, HTOAU65, and HLHA49). Complete sequencing
of these and other cDNAs indicated that they contained an identical
open reading frame homologous to the TNFR superfamily, and it was named
TR2. Analysis of several other ESTs and cDNAs indicated that some
cDNAs had additional sequences inserted into the open reading frame
identified above and might represent various partially spliced
mRNAs.
The myeloid and B cell lines studied represent cell types at different stages of the differentiation pathway. KG1a and PLB 985 (32, 33) were obtained from Phillip Koeffler (UCLA School of Medicine), BJA-B was from Z Jonak (SmithKline Beecham), and TF 274, a stromal cell line exhibiting osteoblastic features, was generated from the bone marrow of a healthy male donor.2 All of the other cell lines were obtained from the American Type Culture Collection (Rockville, MD). Monocytes were prepared by differential centrifugation of peripheral blood mononuclear cells (PBMC) and adhesion to tissue culture dish. CD19+, CD4+, and CD8+ were isolated from PBMC by immunomagnetic beads (Dynal, Lake Success, NY). Endothelial cells from human coronary artery were purchased from Clonetics (San Diego, CA).
RNA and DNA Blot HybridizationTotal RNA of adult tissues was purchased from CLONTECH or extracted from primary cells and cell lines with TriReagent (Molecular Research Center, Inc., Cincinnati, OH). 5-7.5 µg of total RNA was fractionated in a 1% agarose gel containing formaldehyde, as described (34), and transferred quantitatively to Zeta-Probe nylon membrane (Bio-Rad) by vacuum blotting. The blots were prehybridized, hybridized with 32P-labeled Xhol/EcoRI fragment of TR2 or OX-40 probe, washed under high stringency conditions, and exposed to x-ray films.
High molecular weight human DNA was digested with various restriction enzymes and fractionated in 0.8% agarose gel. The DNA was denatured, neutralized, and transferred to nylon membrane and hybridized to 32P-labeled TR2 or its variant cDNA.
In Situ Hybridization and FISH DetectionThe in situ hybridization and FISH detection of TR2 location in human chromosomes were performed as described previously (35, 36). FISH signals and the DAPI banding pattern were recorded separately by taking photographs, and the assignment of the FISH mapping data with chromosomal bands was achieved by superimposing FISH signals with a DAPI-banded chromosome (37).
Production of Recombinant TR2-Fc Fusion ProteinsThe 5
portion of the TR2 containing the entire putative open reading frame of
extracellular domain was amplified by polymerase chain reaction (38).
For correctly oriented cloning, a HindIII site on the 5
end
of the forward primer and a BglII site on the 5
end of the
reverse primer were created. The Fc portion of human IgG1
was PCR-amplified from ARH-77 (ATCC) cell RNA and cloned in the
SmaI site of the pGem7 vector (Promega, Madison, WI). The Fc
fragment, including hinge, CH2, and CH3 domain
sequences, contained a BglII site at its 5
end and an
XhoI site at its 3
end. The HindIII-BglII fragment of TR2 cDNA was
inserted upstream of human IgG1Fc and an in-frame fusion
was confirmed by sequencing. The TR2-Fc fragment was released by
digesting the plasmid with HindIII-XhoI and
cloned into pcDNA3 expression plasmid.
The TR2-Fc plasmid, linearized with PvuI, was transfected into NIH 3T3 by the calcium phosphate co-precipitation method. After selection in 400 µg/ml G418, neomycin-resistant colonies were picked and expanded. Enzyme-linked immunosorbent assay with anti-human IgG1 and Northern analysis with 32P-labeled TR2 probe were used to select for clones that produce high levels of TR2-Fc in the supernatant. In some experiments, a slightly differently engineered TR2-Fc produced in Chinese hamster ovary (CHO) cells was used. The TR2-Fc was purified by protein G chromatography, and the amino acid sequence of the N terminus was determined by automatic peptide sequencer (ABI).
In Vitro Transcription and TranslationThe full-length TR2 cDNA was inserted into HindIII-XhoI sites of pcDNA 3 vector (Invitrogen, San Diego, CA). TNT-coupled reticulocyte lysate system (Promega) was used to in vitro transcribe and translate the TR2 cDNA in pcDNA 3. The 35S-labeled reaction product was fractionated on a 5-15% gradient SDS-polyacrylamide gel, transferred onto an Immobilon membrane (Millipore, Bedford, MA), and exposed to x-ray film.
Blocking MLR-mediated PBMC ProliferationPBMC were isolated from three healthy adult volunteers by Ficoll gradient centrifugation at 400 × g for 30 min. PBMCs were recovered, washed in RPMI 1640 (Life Technologies, Inc.) supplemented with 10% fetal bovine serum, 300 µg/ml L-glutamine, and 50 µg/ml gentamycin, and adjusted to 1 × 106 cells/ml for two donors and to 2 × 105 cells/ml for the third donor.
Fifty µl of each cell suspension was added to 96-well (round bottom) plates (Falcon, Franklin Lakes, NJ) together with 50 µl of TR2-Fc, IL-5R-Fc, anti-CD4 mAb, or control mAb. Plates were incubated at 37 °C in 5% CO2 for 96 h. One µCi of [3H]methylthymidine (ICN Biomedicals, Costa Mesa, CA) was then added for an additional 16 h. Cells were harvested, and radioactivity was counted.
Fig.
1a shows the amino acid sequence of TR2
deduced from the longest open reading frame of one of the isolated
cDNAs (HLHA49). Comparison with other sequenced cDNAs and with
ESTs in the data base indicated potential allelic variants that
resulted in amino acid changes at positions 17 (either Arg or Lys) and
41 (either Ser or Phe) of the protein sequence.
The open reading frame encodes 283 amino acids with a calculated molecular weight of 30,417. The TR2 protein was expected to be a receptor. Therefore, the potential signal sequence and transmembrane domain were sought. A hydrophobic stretch of 23 amino acids toward the C terminus (amino acids 203-225) was assigned as a transmembrane domain, because it made a potentially single helical span (Fig. 1a), but the signal sequence was less obvious. The potential ectodomain of TR2 was expressed in NIH 3T3 and CHO cells as a Fc-fusion protein, and the N-terminal amino acid sequence of the recombinant TR2-Fc protein was determined in both cases. The N-terminal sequence of the processed mature TR2 started from amino acid 37, indicating that the first 36 amino acids constituted the signal sequence (Fig. 1a).
As shown in Fig. 1b, the in vitro translation product of TR2 cDNA was 32 kDa in molecular size. Since the first 36 amino acids constituted signal sequence, and its calculated molecular size was ~4 kDa, the molecular size of the protein backbone of processed TR2 would be approximately 28 kDa. Recently, Montgomery et al. (39) published a herpesvirus entry mediator (HVEM) whose cDNA sequence was identical to TR2. They found that the transfected HVEM cDNA produced a 32-36-kDa protein. Since it is larger than the in vitro product, this suggests that the protein is modified posttranslationally. Two potential asparagine-linked glycosylation sites are located at amino acid positions 110 and 173, as indicated in Fig. 1a.
Along with the other members of the TNFR family, TR2 contains the characteristic cysteine-rich motifs that have been shown by x-ray crystallography (2) to represent a repetitive structural unit. Fig. 1c shows the potential TNFR domain aligned among TR2, TNFR-I, TNFR-II, CD40, and 4-1BB. TR2 contained two perfect TNFR motifs and two imperfect ones.
The TR2 cytoplasmic tail (TR-2 cy) does not contain the death domain seen in the Fas and TNFR-I intracellular domains, and appears to be more related to those of CD40cy and 4-1bbcy. Signals through 4-1BB and CD40 have been shown to be co-stimulatory to T cells and B cells, respectively (40, 41).
TR2 RNA ExpressionA human tissue RNA blot was used to
determine tissue distribution of TR2 mRNA expression. TR2 mRNA
was detected in several tissues with a relatively high level in the
lung, spleen, and thymus, but was not found in the brain, liver, or
skeletal muscle (Fig. 2a). TR2 was also
expressed in monocytes, CD19+ B cells, and resting or PMA
plus PHA-treated CD4+ or CD8+ T cells. It was
only weakly expressed in bone marrow and endothelial cells (Fig.
2b), although expression was observed in the hematopoietic cell line KG1a. For comparison, the tissue distribution of OX-40, another member of the TNFR superfamily, was examined. Unlike TR2, OX-40
was not detected in the tissues examined and was detected only in
activated T cells and KG1a. Several cell lines were negative for TR2
expression, including TF274 (bone marrow stromal), MG63, TE85
(osteosarcomas), RL 95-2 (endometrial sarcoma), MCF-7, T-47D (breast
cancer cells), BE, HT 29 (colon cancer cells), HTB-11, and IMR-32
(neuroblastomas), although TR2 was found in the rhabdosarcoma HTB-82
(data not shown).
Several cell lines were examined for inducible TR2 expression. HL60,
U937, and THP1, which belong to the myelomonocytic lineage, all
increased TR2 expression in response to the differentiating agents PMA
or Me2SO (Fig. 2c). Increases in expression in
response to these agents were also observed in KG1a and Jurkat cells.
In contrast, PMA did not induce TR2 expression in MG63, but
unexpectedly TNF- did.
In almost all cases, the predominant mRNA was approximately 1.7 kilobases in size, although several higher molecular weight species could be detected in some tissues (Fig. 2a), and many cDNAs and ESTs that were sequenced contained insertions in the coding region indicative of partial splicing. The abundance of higher molecular weight mRNAs raises the possibility that TR2 may in part be regulated at the level of mRNA maturation.
TR2 Maps at 1P36.2-P36.3The FISH mapping procedure was
applied to localize the TR2 gene to a specific human chromosomal
region. The assignment of a hybridization signal to the short arm of
chromosome 1 was obtained with the aid of DAP I banding. A total of 10 mitotic figures were photographed, one of which is shown in Fig.
3a. The double fluorescent signals are
indicated on the schematic diagram of chromosome 1 as shown in Fig.
3b. This result indicated that the TR2 gene is located on
the chromosome 1 region p36.2-p36.3. The TR2 position is in close
proximity with CD30 (42), 4-1BB (43, 44), OX-40 (45), and TNFR-II
(46), suggesting that it evolved through a localized gene duplication
event. Interestingly, all of these receptors have stimulatory
phenotypes in T cells in response to cognate ligand binding, in
contrast to Fas and TNFR-I, which stimulate apoptosis. This prompted us
to test if TR2 might be involved in lymphocyte stimulation.
TR2-Fc Interferes with MLR-mediated Proliferation of PBMC
To
determine the potential involvement of cell surface TR2 with its ligand
in lymphocyte proliferation, we examined allogeneic MLR proliferative
responses. As shown in Fig. 4, a and
b, when TR2-Fc was added to the culture, a significant
reduction of maximal responses was observed (p < 0.05). The addition of TR2-Fc at 100 µg/ml inhibited the
proliferation up to 53%. No significant inhibition of proliferation
was observed with the control IL-5R-Fc. Surprisingly, at high
concentrations (10-100 µg/ml) IL-5R-Fc was shown to enhance proliferation. The concentrations of TR2-Fc required to inhibit MLR
proliferation (1-100 µg/ml) are comparable with those of CD40-Fc required for inhibition in other lymphocyte assays (47-50). An anti-CD4 mAb assayed simultaneously inhibited MLR-mediated
proliferation up to 60%, whereas a control anti-IL-5 mAb failed to
inhibit the proliferation. It is well known that a major component of
the MLR proliferative response is T cell-dependent; hence,
it would appear that inhibiting the interaction of TR2 with its ligand prevents optimal T lymphocyte activation and proliferation.
Hence, we have identified an additional member of the TNFR superfamily that either plays a direct role in T cell stimulation or binds to a ligand which can stimulate T cell proliferation through one or more receptors, which may include TR2. We are currently trying to identify this ligand to which TR2 binds to clarify its role.