and François Mallet
Bouton, Frédéric Bedin, Hervé Perron, Bernard Mandrand
Jean-Luc Blond, Frédéric Besème, Laurent Duret, Olivier
Endogenous Retrovirus Family
Expression of HERV-W, a New Human
Molecular Characterization and Placental
Updated information and services can be found at:
This article cites 49 articles, 24 of which can be accessed free
more»articles cite this article),
Receive: RSS Feeds, eTOCs, free email alerts (when new
Information about commercial reprint orders:
To subscribe to to another ASM Journal go to:
on May 30, 2013 by guest
JOURNAL OF VIROLOGY,
Copyright © 1999, American Society for Microbiology. All Rights Reserved.
Feb. 1999, p. 1175–1185Vol. 73, No. 2
Molecular Characterization and Placental Expression of HERV-W,
a New Human Endogenous Retrovirus Family
JEAN-LUC BLOND,1FRE´DE´RIC BESE`ME,1LAURENT DURET,2OLIVIER BOUTON,1
FRE´DE´RIC BEDIN,1HERVE´PERRON,1BERNARD MANDRAND,1
AND FRANC ¸OIS MALLET1*
Unite ´ Mixte 103 CNRS-bioMe ´rieux, Ecole Normale Supe ´rieure de Lyon, 69364 Lyon, Ce ´dex 07,1and
Laboratoire de Biome ´trie Ge ´ne ´tique et Biologie des Populations, UMR CNRS 5558,
Universite ´ Claude Bernard-Lyon 1, 69622 Villeurbanne Cedex,2France
Received 15 July 1998/Accepted 10 November 1998
The multiple sclerosis-associated retrovirus (MSRV) isolated from plasma of MS patients was found to be
phylogenetically and experimentally related to human endogenous retroviruses (HERVs). To characterize the
MSRV-related HERV family and to test the hypothesis of a replication-competent HERV, we have investigated
the expression of MSRV-related sequences in healthy tissues. The expression of MSRV-related transcripts
restricted to the placenta led to the isolation of overlapping cDNA clones from a cDNA library. These cDNAs
spanned a 7.6-kb region containing gag, pol, and env genes; RU5 and U3R flanking sequences; a polypurine
tract; and a primer binding site (PBS). As this PBS showed similarity to avian retrovirus PBSs used by
tRNATrp, this new HERV family was named HERV-W. Several genomic elements were identified, one of them
containing a complete HERV-W unit, spanning all cDNA clones. Elements of this multicopy family were not
replication competent, as gag and pol open reading frames (ORFs) were interrupted by frameshifts and stop
codons. A complete ORF putatively coding for an envelope protein was found both on the HERV-W DNA
prototype and within an RU5-env-U3R polyadenylated cDNA clone. Placental expression of 8-, 3.1-, and 1.3-kb
transcripts was observed, and a putative splicing strategy was described. The apparently tissue-restricted
HERV-W long terminal repeat expression is discussed with respect to physiological and pathological contexts.
A few years ago, the presence of extracellular particles as-
sociated with reverse transcriptase (RT) activity was observed
in leptomeningeal cells (LM7) (42) and monocyte cultures (44)
from patients with multiple sclerosis (MS). Viral RT-associ-
ated activity plus electron microscopy examination and the
presence in MS patients of antibodies cross-reacting with hu-
man immunodeficiency virus type 1 and type 2 RT and human
T-cell leukemia virus type 1 CA as well (41) led to the hypoth-
esis of the retroviral nature of these viral particles. By mainly
a PCR-based approach, a partial molecular identification of a
novel retrovirus isolated from cerebrospinal fluid, plasma, or
cell culture supernatant of MS patients was recently obtained
(40). Retroviral RNAs with strong similarities to MS-associ-
ated retrovirus (MSRV) were also found in plasma of rheu-
matoid arthritis (RA) patients (17). Furthermore, expression
of related mRNA sequences (8) was found in both MS and
control brain tissues (29).
All together, these observations might result from three
different situations. The extracellular particles may be pro-
duced by a replication-competent endogenous provirus. Alter-
natively, MSRV could represent a virion-producing exogenous
member of an endogenous family, as described for the mouse
mammary tumor virus and murine leukemia virus retroviral
families of mice (23) and in the Jaagsiekte retroviral family of
sheep (37). Lastly, a more complicated process would involve
separated defective retroviral entities cooperating via trans-
MSRV was found to be related to the endogenous retroviral
sequence ERV-9 (26), and Southern blot analysis with an
MSRV pol probe showed hybridization with a multicopy en-
dogenous family (40). Therefore, we have studied the MSRV-
related human endogenous retroviruses (HERVs), which could
represent either distant members of the ERV-9 family or an as
yet undescribed HERV family. As (i) the HERV family was
shown to be a multicopy family (40) and (ii) we focused on the
hypothesis that MSRV could be a replication-competent HERV,
we followed a strategy based on the characterization of mRNA
isolated from a healthy human tissue, by using MSRV- and
ERV-9-derived probes. We describe a new multicopy endog-
enous retroviral family tentatively named HERV-W. All mem-
bers of the family are apparently not competent for replica-
tion. However, a functional U3 promoter and an open reading
frame (ORF) putatively coding for an envelope protein were
identified. HERV-W mRNA expression is restricted to placen-
tal tissue in Northern blot analysis. The strategy of transcrip-
tion is addressed, and the promoter tissue specificity is dis-
MATERIALS AND METHODS
Probe definition and labelling. The 678-bp Ppol-MSRV probe was RT-PCR
amplified from RNA extracted from particles collected from the supernatant of
synoviocyte cultures of an RA patient. These particles were gradient concen-
trated as previously described (43), and RNA was extracted by the method of
Chomczynski and Sacchi (10) and DNase treated as previously described (54). A
first round of RT-PCR amplification was performed with primers AP2942 (5?-
AGGAGTAAGGAAACCCAACGGAC-3?) (forward) and AP2152 (5?-TAA
GAGTTGCACAAGTGCG-3?) (reverse). A nested PCR was then performed
with primers AP2522 (5?-TCAGGGATAGCCCCCATCTAT-3?) (forward) and
AP2510 (5?-AACCCTTTGCCACTACATCAATTTC-3?) (reverse). The 649-bp
Ppol-ERV-9 probe was PCR amplified from a B-lymphocyte cDNA library of an
MS patient with primers AP2522 and AP2510 mentioned above. Discrimination
between the Ppol-MSRV- and Ppol-ERV-9-cloned PCR fragments was per-
formed with an enzyme-linked oligosorbent assay (ELOSA) (35). Capture and
detection oligonucleotide probes used in this nonisotopic sandwich technique are
depicted in Fig. 1A.
The Pgag-LB19 (536 bp), Ppro-E (364 bp), and Penv-C15 (591 bp) probes
* Corresponding author. Mailing address: Unite ´ Mixte 103 CNRS-
bioMe ´rieux, Ecole Normale Supe ´rieure de Lyon, 46 alle ´e d’Italie,
69364 Lyon, Ce ´dex 07, France. Phone: 33 472 728 358. Fax: 33 472 728
533. E-mail: email@example.com.
on May 30, 2013 by guest
corresponded to PCR fragments of the longest ORFs of LB19, E, and C15
clones, respectively. LB19 and E clones (37a) were obtained from RNA extracted
from gradient-concentrated particles of B-lymphocyte cultures of an MS patient
or from the plasma of an MS patient, respectively. The C15 clone (22a) was
obtained from RNA extracted from the pellet of centrifuged supernatants of
synoviocyte cultures of an RA patient.
Fifty nanograms of each probe was radioactively labelled by random priming
with the Ready-to-Go DNA labelling kit from Pharmacia Biotech, Inc., accord-
ing to the manufacturer’s protocol with [?-32P]dCTP (3,000 Ci/mmol). The
specific activity of probes was ?5.0 ? 108cpm/?g.
Placental cDNA library screening, 5? rapid amplification of cDNA ends (RACE),
and sequencing. A placental cDNA library constructed in the bacteriophage
lambda vector was used (human placenta 5?-Stretch Plus cDNA library from
Clontech Laboratories, Inc., Palo Alto, Calif.). Screening of the library was
performed according to the manufacturer’s instructions with the Ppol-MSRV
probe for initial screening, and probes were derived from the resulting clones for
additional screening. Briefly, for each screening the plating density was about 5 ?
105PFU for two 220- by 220-mm plates. After growth, plaques were transferred
onto a nylon membrane (Hybond N?; Amersham) and hybridized with probes
radioactively labelled as described above (106cpm/ml of hybridization solution).
Prehybridization was performed overnight at 42°C in a freshly prepared solution:
50% formamide, 5? SSPE (20? SSPE is 3 M NaCl, 0.2 M NaH2PO4, and 20 mM
EDTA, pH 7.4), 5? Denhardt’s solution (100? Denhardt’s solution is 10 g of
Ficoll 400, 10 g of polyvinylpyrrolidone, and 10 g of bovine serum albumin for
500 ml), and 0.1% sodium dodecyl sulfate (SDS) with 100 ?g of heat-denatured
herring sperm DNA per ml. Hybridization was performed overnight at the same
temperature, in a renewed solution freshly prepared in the same way. After
hybridization, filters were washed once for 20 min at room temperature with 2?
SSC (20? SSC is 3 M NaCl and 0.3 M sodium citrate)–0.5% SDS followed by
one wash for 25 min at 60°C with 1? SSC–0.1% SDS. Washed filters were
exposed to X-ray films (Kodak) for 72 h at ?80°C. Isolated positive plaques were
picked and eluted in the appropriate buffer.
A rapid method was used for subcloning of the cDNA clones. A PCR ampli-
fication using a primer generated according to the bacteriophage lambda vector
sequence and a specific primer designated according to the probe sequence was
performed in a 9600 Perkin-Elmer machine. PCR was carried out with initial
denaturation at 94°C for 5 min followed by 30 cycles of 94°C for 1 min, 54°C for
1 min, and 72°C for 4 min; terminal extension was performed at 72°C for 7 min.
After 1% agarose gel analysis, the longest fragments were subcloned into pCR
2.1 vector (TA cloning; Invitrogen, San Diego, Calif.) and sequenced.
A 5? RACE technique was used to obtain the 5? end of the retrovirus genome;
the kit, used strictly according to the manufacturer’s recommendations, was
provided by Life Technologies, Inc. (Bethesda, Md.). Reverse transcription,
dC-tailing, and amplifications were performed on a commercially available pla-
cental mRNA [human placental poly(A)?RNA; Clontech], and fragments were
subcloned into pCR 2.1 vector (TA cloning).
Sequencing reactions were performed, in both directions for each clone, with
the Prism Ready Reaction kit and Dye Deoxyterminator cycle sequencing kit
(Applied Biosystems). Automatic sequence analysis was performed on an auto-
matic sequencer (Applied Biosystems).
Hybridization of Northern blots and RNA dot blots. Several Northern blots of
different registration lot numbers were used (multiple-tissue Northern blot, cat-
alog no. 7760-1 from Clontech Laboratories, Inc.). Prehybridization was per-
formed overnight at 42°C in the following freshly prepared solution: 50% form-
amide, 5? SSPE, 10? Denhardt’s solution, 2% SDS, and heat-denatured herring
sperm DNA (100 ?g/ml). Hybridization was performed overnight at 42°C in a
renewed solution with a probe labelled as described above (106cpm/ml of
hybridization solution). Filters were washed twice with 1? SSC–0.1% SDS at
room temperature for 5 min and once with 0.1? SSC–0.1% SDS at 50°C for 10
min. Washed filters were exposed to X-ray films (Kodak) for 5 days at ?80°C.
RNA dot blot (human RNA Master Blot from Clontech Laboratories, Inc.)
was used according to the manufacturer’s instructions. Briefly, prehybridization
was performed at 65°C in the ExpressHyb hybridization solution from Clontech
for 30 min. Hybridization of a probe labelled as described above was performed
overnight at 65°C in a renewed solution. The membrane was washed twice at
65°C with 2? SSC–1% SDS for 20 min and once at 55°C with 0.1? SSC–0.5%
SDS and exposed to X-ray film (Kodak) for about 12 h.
Hybridization of Southern blots and DNA dot blots. Human genomic DNA
was digested with EcoRI, HindIII, and PstI. Ten micrograms of each reaction
mixture was electrophoresed on a 0.8% agarose gel and transferred onto a
charged nylon membrane (Hybond N?[Amersham]). Prehybridization was per-
formed overnight at 42°C in a freshly prepared solution: 50% formamide, 5?
SSPE, 1% (wt/vol) nonfat dried milk, 1% SDS, and 50 ?g of heat-denatured
herring sperm DNA per ml. Hybridization was performed overnight at 42°C in a
renewed solution, with a probe radioactively labelled as mentioned above (106
cpm/ml of hybridization solution). Filters were washed once with 1? SSC–0.1%
SDS at room temperature for 5 min and once with 0.1? SSC–0.1% SDS at 50°C
for 10 min. Washed filters were exposed to X-ray films (Kodak) for 5 days at
For the DNA dot blot, successive dilutions of each probe (2.5, 5, 10, 25, 50, and
100 pg) and 0.5 ?g of genomic DNA were blotted on a charged nylon membrane
after denaturation. For each probe, the dot blot and the Southern blot were
handled in the same vessel.
CAT assay. The putative promoter region was cloned into pCAT3 Enhancer
reporter vector purchased from Promega Biotec (Madison, Wis.). HeLa 60%
confluent cells cultured in Dulbecco modified Eagle medium–10% fetal calf
serum (Life Technologies) were transfected with the Superfect transfection kit
(Qiagen GmbH) with 2 ?g of purified recombinant plasmid. After a 48-h incu-
bation, the cells were harvested in order to evaluate chloramphenicol acetyl-
transferase (CAT) activity by the use of the CAT enzyme assay system (Promega
Biotec). For this purpose, the liquid scintillation counting protocol was followed
as recommended by the manufacturer. Briefly, cell extracts were obtained by
several freezes-thaws and a subsequent heating at 60°C for 10 min. Then, 10 ?l
of this extract was added to 0.25 M Tris-HCl, pH 8.0 (125 ?l, final volume),
containing 0.15 ?Ci of [14C]chloramphenicol (NEN, Boston, Mass.) and 5 ?g of
n-butyryl coenzyme A, and incubated for 4 h at 37°C. Five hundred microliters
of a mix of xylene isomers was added, and the butyrylated chloramphenicol
present in this organic phase was extracted twice with 100 ?l of 0.25 M Tris-HCl.
This 200 ?l was mixed with 5 ml of scintillation fluid and counted in a liquid
In vitro transcription-translation. In vitro transcription-translation was per-
formed with a kit from Promega Biotec (TNT coupled reticulocyte lysate sys-
tems). It was used strictly according to the manufacturer’s instructions. Briefly,
DNA templates were obtained by PCR with a 5? primer designed with a T7
TTATCAT-3?). The radiolabelled amino acid was [35S]methionine, provided by
Amersham. Glycosylation was studied by the translation in vitro in the presence
of canine pancreatic microsomal membranes (also provided by Promega). Post-
translational analysis was performed by SDS-polyacrylamide gel electrophoresis,
and the gel was exposed to X-ray films (Kodak), usually for 12 h.
Sequence analysis. Individual cDNA clones were used for phylogenetic anal-
ysis. Sequences were filtered for low-complexity regions and repeat sequences
(Alu-like and microsatellites) with XBLAST (12). A BLASTN (or BLASTX) (2)
query for GenBank (release 101, June 1997) was performed, and sequence
matches with scores greater than 200 were retained. Homologous fragments were
extracted from GenBank and aligned with our cDNA clones. Alignments were
performed semimanually, with the SEAVIEW multiple alignment editor (16)
and the Clustal W (52), Dialign (36), or MABIOS (1) multiple alignment pro-
gram. Regions corresponding to putative coding regions were determined with
Blixem (51), a program allowing the visualization of protein similarities on a
nucleic acid sequence from the alignment. We selected in the alignment three
regions from long terminal repeat (LTR), Pol, and Env that are conserved in all
members of the family of human endogenous viruses that we have identified.
Phylogenetic trees were computed from these three regions with Phylo_win (16)
with the neighbor-joining method (49). Five hundred bootstrap replicates were
performed to evaluate the robustness of the trees.
Promoter prediction was performed with PROSCAN v1.7 (47) and SIGSCAN
v4.05 (46). A searching for protein sequence similarities was performed with
LFASTA (39). The prediction of leader peptide was made with GeneWorks 2.5.1
software. Prediction of the fusion domain and the transmembrane region was
performed with the PHD package (48).
Nucleotide sequence accession numbers. The sequences described in this
paper have been submitted to GenBank under the following accession num-
bers. The probe numbers are as follows: Ppol-MSRV (AF072494), Ppol-ERV-
9 (AF072495), Pgag-LB19 (AF072496), Ppro-E (AF072497), and Penv-C15
(AF072498). The placental cDNA clones numbers are as follows: cl.6A1
(AF072499), cl.6A2 (AF072500), cl.7A16 (AF072501), cl.Pi5T (AF072502),
cl.Pi22 (AF072503), cl.44.4 (AF072504), cl24.4 (AF072505), cl.PH74 (AF072506),
cl.PH7 (AF072507), and cl.C4C5 (AF072508).
Design of probes. Pol probes were designed to evaluate the
expression of MSRV-related sequences in healthy human tis-
sues. The choice of Pol probes relied on the observations that
(i) the previously described Pol region of MSRV was related to
the ERV-9 HERV and (ii) ERV-9 sequences were occasion-
ally detected with MSRV in plasma or cell culture superna-
tants of MS patients (40). Because (i) the 2,304-bp MSRV Pol
region (40) was a consensus sequence resulting from overlap-
ping clones obtained from different sources and (ii) the MSRV
and ERV-9 codetected sequences were only 120 bp long, we
chose to amplify two continuous 650-bp MSRV- and ERV-9-
related fragments, defined by the same borders and including
the 120-bp codetected region. MSRV- and ERV-9-related Pol
regions were RT-PCR amplified as described in Materials
and Methods, and mixed resulting fragments were cloned and
checked (data not shown) with the ELOSA procedure (Fig.
1176BLOND ET AL.J. VIROL.
on May 30, 2013 by guest
1A), which was previously shown to discriminate between the
MSRV and ERV-9 Pol 120-bp coamplified fragments (40).
The ELOSA MSRV-related probe, which was called Ppol-
MSRV, has 87 and 69% similarity with MSRV and ERV-9
reference sequences, respectively. The ELOSA ERV-9-related
probe, which was called Ppol-ERV-9, has 70 and 87% similar-
ity with MSRV and ERV-9, respectively.
In order to improve the likelihood of finding sequences
containing ORFs, consistent with the hypothesis that MSRV
could be a replication-competent HERV, three other probes
were derived from putatively packaged extracellular mRNAs
detected in pathological contexts. These mRNAs were RT-
PCR amplified with primers designed according to the ob-
tained placental cDNA clones (see below) and/or to the
conserved region of retroviruses and/or to 5? and 3? RACE
protocols. Although no amplified fragments supported coding
capacities compatible with a replication-competent retrovirus,
Pgag-LB19, Ppro-E, and Penv-C15 probes were designed, cor-
responding to the longest ORFs found in PCR products de-
rived from gag, pro, and env, respectively.
Expression of a novel HERV in placenta. Ppol-MSRV, Ppol-
ERV-9, Pgag-LB19, and Penv-C15 probes were then used for
hybridization of a multiple-tissue Northern blot under strin-
gent conditions. The Ppol-MSRV probe revealed an 8-kb tran-
script that was expressed in placenta but not in heart, brain,
lung, liver, skeletal muscle, kidney, or pancreas (Fig. 1B). No
signal was detected with the Ppol-ERV-9 probe (Fig. 1B). The
placenta-restricted expression was confirmed for Pgag-LB19
and Penv-C15 probes (data not shown). Furthermore, 48 tissue
samples including cerebral, muscular, endocrine, exocrine,
lymphoid, visceral, and fetal tissues were not revealed by Pgag-
LB19 and Penv-C15 probes by RNA dot blot, but both placen-
tal and kidney mRNAs were found to be positive (data not
shown). As the kidney mRNA expression was not revealed by
Northern blotting, with four different commercial lots, we
interpreted the kidney mRNA positive signal as an artifact
resulting from the method per se, including tissue-specific
mRNA preparation or DNA contamination.
A BLASTN query on the EST (expressed sequence tag)
database, with placental cDNA clones (see below) derived
from the above probes, showed hundreds of related transcripts
in the human tissues, most in the opposite direction. With a
search criterion of 90% identity over 100 nucleotides (nt) or
more, these sequences were found predominantly expressed in
the placenta (53%), but also in fetal liver-spleen (28%). The
relative abundance of HERV-W sequences among placenta
ESTs was found to be three times higher than that among fetal
Isolation and sequencing of overlapping cDNA fragments. A
placental cDNA library was screened with Ppol-MSRV, Pgag-
LB19, and Penv-C15 probes. Probes derived from the resulting
clones were then used for additional screening. The clones
obtained with all three screenings exhibited the same charac-
teristics. All together, the overlapping cDNA clones covered
7.6 kb. Nine overlapping clones obtained with the Ppol-MSRV
screening are shown in Fig. 2A. The percentages of similarity
among the overlapping parts of these Ppol-MSRV-derived
cDNA clones, a polyadenylated clone (cl.C4C5) probed with
Penv-C15, and the probes used for the screening as well are
presented in Fig. 2B. The similarity between cDNA clones
ranged from 97 to 100% with the exception of the cl.7A16
clone, which was about 90%. The similarity between the cDNA
clones and probes derived from MSRV-related extracellular
RNA sequences ranged from 85 to 94%. Similarly, this per-
centage was near 90% with the MSRV Pol region previously
described (40) but was lower than 70% with the ERV-9 Pol
With BLASTX, sequences of the different fragments showed
extensive homology with Gag, Pol, and Env retroviral proteins.
However gag and pol genes did not support ORFs correspond-
ing to functional proteins. An ORF putatively coding for a
retroviral envelope protein was observed on the 3? cDNA
clones (cl.24.4, cl.C4C5, and cl.PH74). This Env ORF was
totally contained in the cl.PH74 cDNA clone. Flanking un-
translated regions (UTRs) were observed upstream and down-
stream from gag and env genes, respectively, containing re-
peated sequences (R) as described for LTRs. The length of R
was estimated as about 120 bp by comparison of the 5? end of
the 5? most cDNA clone (cl.6A2) and the 3? end of a polyad-
enylated clone (cl.PH74), both clones showing 98% similarity
in their 5? UTR overlap. Three clones contained 3? UTRs
followed by a poly(A) tail (cl.Pi5T, cl.PH74, and cl.C4C5).
Phylogenetic analysis. A phylogenetic analysis was per-
formed at the nucleic acid level on 11 different subregions of
the region spanned by the cDNAs and at the protein level on
two different Env subregions, as described in Materials and
Methods. All the trees had the same topology regardless of the
region addressed. Notably, the human BAC clone RG083M05
was represented in all trees. These results were illustrated at
the nucleic acid level within the most conserved regions of the
LTR (R-U5) and the pol gene (RT catalytic domain) between
the obtained sequences and ERV-9 and RTVL-H (Fig. 3). The
trees clearly showed that those sequences described a new
family related to, but distinct from, ERV-9 and even more
distant from RTVL-H as highlighted by the bootstrap analysis.
FIG. 1. (A) Capture and detection probes used in ELOSA to detect PCR-
amplified ERV-9 and MSRV pol-related sequences. Discrimination between
ERV-9 and MSRV cloned PCR fragments was performed at the capture step
with CpV1A and CpV1B-D, respectively. Colorimetric detection was achieved
with the DpV1A-D peroxidase-labelled probe located within the most conserved
region of ERV-9 and MSRV pol genes. The selected PCR fragments were called
Ppol-MSRV and Ppol-ERV-9 according to ELOSA characterization. The refer-
ence (ref) ERV-9 (gb X57147, gb M85205, and gb M37638) and MSRV (gb
AF009668) sequences are indicated as well as the consensus (cons) sequences.
These consensus sequences resulted from RT-PCR amplification of the con-
served region of the pol gene present in all retroviruses (40, 50); the RNA
material was obtained from MS patients. (B) Northern blot analysis of poly(A)?
cellular RNAs from eight human tissues with the Ppol-MSRV and Ppol-ERV-9
probes characterized by ELOSA.
VOL. 73, 1999CHARACTERIZATION AND EXPRESSION OF HERV-W 1177
on May 30, 2013 by guest
Those sequences were found on several chromosomes, includ-
ing 5, 7, 14, 16, 21, 22, and X, with an apparent high concen-
tration of LTR in chromosome X.
Comparison at the protein level between the most conserved
regions of Env retroviral proteins (immunosuppressive to
transmembrane [TM] domain) resulted in trees showing a se-
quence distribution similar to the one observed at the nucleic
acid level (Fig. 3). The two proximal coding sequences corre-
sponded to the translated env ORF of RG083M05 and a trun-
cated ORF located downstream from the GTP-binding protein
RAB7 on the same mRNA. This group of sequences was more
closely related to simian type D and reticuloendotheliosis avian
retroviruses than to type C mammalian retroviruses.
Genomic complexity of the family. In order to characterize
the complexity of this new endogenous retroviral family, we
used several probes spanning the entire genome. Southern blot
analysis of human DNA digested with three different restric-
tion enzymes showed complex patterns with an apparent in-
crease of complexity from env to gag and protease regions (Fig.
4A). A more precise determination of the copy number of each
region was done by a dot blot analysis of serial dilutions of the
plasmids that contained the corresponding probes (data not
shown). The calculated number of copies per haploid genome
(Fig. 4A) emphasized the heterogeneous distribution of the
retroviral subregions. A 6- and 20-fold increase of Gag and
protease signals, respectively, versus the Env signal confirmed
the gradation suspected in Southern blot analysis.
To characterize some genomic sequences at the molecular
level, a BLASTN query on several databases was performed,
with the placental cDNA clones. The four most significant
sequence hits are depicted in Fig. 4B. They consisted of the
human BAC clone RG083M05 from 7q21-7q22, the human
BAC378 corresponding to Homo sapiens T-cell receptor alpha
delta locus whose chromosomal location is 14q11-12, the H. sa-
piens chromosome 21q22.3 cosmid Q11M15, and the human
DNA sequence from cosmid U134E6 on chromosome Xq22
containing NIK-like and thyroxin-binding globulin precursor.
Repeated sequences were found located at both ends of three
of these clones. All the cDNA sequences fell entirely within the
clone RG083M05 (10.2 kb) with a similarity of about 98 to
99%, except the clone cl.7A16 (90%). However, in addition to
gag, pol, and env genes, RG083M05 exhibited a 2-kb insert
located just downstream from the 5? UTR, this insert being
found also in clones BAC378 and Q11M15. These two clones
and a third one (U134E6) presented a strictly conserved 2.3-kb
deletion just upstream from the 3? UTR. No clone contained
all three gag, pol, and env ORFs. The clone RG083M05 exhib-
ited a 538-amino-acid ORF corresponding to a full-length en-
velope. The cosmid Q11M15 contained two large contiguous
ORFs of 413 (frame 0) and 305 (frame ?1) aa corresponding
to a truncated Pol polyprotein. All together, these data sug-
gested a quite complex family from which the RG083M05
clone could represent a genomic prototype.
RNA elements: LTR, polypurine tract, and primer binding
site (PBS) characterization. In order to identify a potential
promoter downstream from the cl.PH74 putative env ORF, a
predictive analysis was performed. Proscan 1.7 did not predict
any obvious promoter, but Signalscan 4.05 indicated the pres-
ence of putative transcription factor binding sites, including a
CCAAT box and a TATA box. The 2,364- to 2,720-nt cor-
responding region of the cl.PH74 clone (2,764 bp) was PCR
amplified and subcloned into pCAT3 reporter vector. The
CAT assay showed a significant expression level (Fig. 5A). A
promoter activity was also found in BeWo human choriocar-
cinoma cells and Jurkat human T cells (data not shown).
In order to characterize the retroviral LTR, the 5? and 3?
UTRs of cDNA experimental clones were aligned with the 5?
and 3? repeated sequences of the most proximal human DNA
sequence, RG083M05. Nonretroviral flanking cellular se-
quences allowed us to delineate a 783-bp putative LTR (Fig.
5B). A duplicated 4-bp sequence (CAAC) was found flanking
5? and 3? LTRs. The characteristic TG-CA base pairs were
found juxtaposed with nonretroviral sequences at each end of
the integrated provirus. The 5? end of U3 was confirmed by the
presence, on the cl.PH74 and cl.C4C5 clones, of a polypurine
tract located immediately upstream from U3 and 46 bp down-
stream from the env stop codon. Determination of the U3-R
junction was quite complex. Several possible cap sites were
found by 5? RACE experiments on placental mRNA (data not
shown), but surprisingly, all these sites were found included
within the cl.6A2 clone (Fig. 5B). More surprisingly, the pre-
dicted TATA box fell within the R sequence defined according
FIG. 2. (A) Clones isolated from a placental cDNA library first with the
Ppol-MSRV probe and probes derived from the resulting clones for additional
screening. The length of the clone is indicated in parentheses. Black, white, and
grey boxes represent regions showing extensive homology with gag, pol, and env
genes, respectively; AAAAAAAA indicates poly(A) tail; flanking striped boxes
represent UTRs; and repeats are indicated by black arrowheads. (B) Percentages
of similarity between the overlapping regions of placental cDNA clones and
between the cDNA clones and the probes used in this study. For the cDNA clone
analysis, percentages are indicated in boldface and italics for overlapping frag-
ments larger than 1 kb and 490 bp, respectively. The smallest overlapping frag-
ment is 205 bp long. For the comparison between cDNA clones and probes,
percentages labelled with asterisks indicate a partial overlap. Probes used for the
placental cDNA library screening are underlined.
1178BLOND ET AL. J. VIROL.
on May 30, 2013 by guest
to the cl.6A2 clone. To simplify the analysis, a similar proce-
dure was applied to a single U3-R region, with total RNA
extracted from pCAT3-3?LTR-transfected HeLa cells (data
not shown). It indicated a potential cap site downstream from
the TATA box (Fig. 5B). The 3? end of R was precisely defined
according to the presence of a poly(A) tail on cl.PH74 and
cl.C4C5 clones. The polyadenylation signal was found within
the R region, 13 bp upstream of U5. The U5 LTR remaining
sequence was found to be 455 bp long. The 3? end of U5 was
confirmed by the presence, on the cl.6A2 (gag), cl.PH74, and
cl.24.4 (env) clones, of a putative PBS located 4 bp down-
stream. This PBS showed extensive homology with the avian
retrovirus PBS (53) used by tRNATrpfor minus-strand DNA
synthesis (34). As the tRNATrpwas not described for humans
and this PBS sequence did not correspond to any other tRNA,
we tentatively named this new family HERV-W.
Transcription and putative splicing strategy. The expression
of the HERV-W family in placenta was analyzed by Northern
blotting (Fig. 6A) with a set of probes spread over the 7.6-kb
region containing the cDNA clones (Fig. 6B) and consisting of
two U5 probes, U5(g) and U5(e), derived from an RU5-gag
cDNA clone (cl.6A2) and a U5-env cDNA clone (cl.24.4),
respectively, as well as gag, pro, pol, and env probes as de-
scribed in Materials and Methods, and a U3-R probe, U3(e),
derived from an env–U3-R–poly(A)?clone (cl.C4C5). A sim-
ple pattern consisting of three bands at 8, 3.1, and 1.3 kb was
observed. Probes from the gag, pro, and pol genes hybridized
only to the 8-kb transcript which was also revealed by env, U5,
and U3 probes and thus may correspond to a putative full-
length transcript. Probe from the env gene detected both the
full-length transcript and an abundant 3.1-kb transcript which
also hybridized to the U5 and U3 probes. Thus, this 3.1-kb
transcript exhibited the characteristic features of a singly
spliced subgenomic retroviral env mRNA. The U5 and U3-R
probes revealed an additional transcript of 1.3 kb, lacking de-
tectable gag, pro, pol, and env sequences, which may result from
alternative splicing or transcription from defective HERV-W
No trivial correspondence between the observed transcripts
and the isolated cDNA clones could be found, as the 5? end of
R was not unambiguously defined. However, the cl.PH74 clone
containing an env ORF and repeats at both ends was a good
candidate for the 3.1-kb subgenomic mRNA. Likewise, the
cl.PH7 clone which contained U5 and U3 sequences at the 5?
and 3? ends, respectively, but no structural gene, might repre-
sent a large spliced subgenomic mRNA. In addition, if we
assumed (i) that the cap site location was as defined by the 3?
LTR transfection experiment and (ii) the region spanned by
FIG. 3. Phylogenetic analyses were performed at the nucleic acid level for LTR and pol regions and at the protein level for Env. The included sequences were
selected as described in Materials and Methods. Experimental clones are indicated in shaded boxes, and MSRV sequences are underlined. LTR, POL, and ENV
neighbor-joining trees based upon 123-nt LTR (nt 6 to 143 of clone cl.6A2), 575-nt pol (nt 2309 to 2933 of clone cl.6A1), and 113-aa Env (translation of nt 1852 to
2211 of clone cl.PH74) fragments are presented from left to right. Pairwise gap removal was used for the LTR tree, and all-gap removal was used for POL and ENV
trees. Bootstrap values for 500 replicated trees are indicated. “c-” means that the sequence is presented in reverse orientation. Chromosomal assignments are indicated
in brackets. Upstream (5) and downstream (3) LTRs are indicated when required. Sequences used in the LTR tree are as follows: clone RG083M05 (AC000064);
IMMDL, DNA sequence of the human immunoglobulin D segment locus (X97051 S64822); cosmid U221F2 (Z75893); cosmid cU96H11 (Z70042); PAC 107N3
(Z75741); cosmid E110C7 (Z68223); P135H5C8, H. sapiens DNA sequence on chromosome 21 (L35660); cosmid cN74G7 (Z69715); cosmid U101D3 (Z85998); BAC
clone CIT987SK-29B12 (U95738); PAC clone H74 (AC000352); cosmid 315B17 (Z73967); cosmid U61F10 (Z75895); cosmid U85B5 (Z69724); cosmid U116E9
(Z95333); human DNA sequence from clone J428A131 (Z82209); PAC 49C23 (Z93019); PAC 162C6 containing endogenous retrovirus pHE.1 (ERV-9) (Z84475);
ERV-9LTR3 (M92648); ERV-9LTR2 (M92647); and ERV-9LTR-ZNF80, H. sapiens DNA for ZNF80-linked ERV-9 LTR (X83497). Additional sequences included
in the POL tree are as follows: MSRV-pol reconstructed consensus (AF009668); Ppol-MSRV probe; ERV-9, HERV pHE.1 mRNA sequence (X57147, M85205, and
M37638); Ppol-ERV-9 probe; PAC pDJ239b22 (AC003969 and U90583); and RTVL-H, HERV type H (D11078). Sequences appearing in the ENV tree are as follows:
SNV, spleen necrosis virus (EMBL M87666); REV-A, reticuloendotheliosis virus strain A (X01455 K02537); SMRV-H, simian sarcoma virus (M23385); SRV-1 (L47.1),
simian SRV-1 type D retrovirus (M11841); MPMV, simian Mason-Pfizer type D retrovirus (M12349); SRV-2, simian SRV-2 type D retrovirus (M16605); SRV-1, simian
type D virus 1 (U85505); RD114, cat endogenous retrovirus RD114 env gene (X87829); BaEV, baboon endogenous virus virion (M16550); GALV, gibbon ape leukemia
virus (M26927); MERV-Fv4, mouse endogenous retrovirus in Fv4 locus (M33884); Cas Br E MuLV, murine leukemia virus (Cas-Br-E MuLV) (M14702); MoMLV,
Moloney murine leukemia virus (J02255, J02256, J02257, and M76668); MuLV, murine leukemia virus (M93052); and human sequences RTVL-H (NBRF B44282),
ERV-9 (M85205 and M37638), and RAB7 (H. sapiens mRNA) (X93499).
VOL. 73, 1999CHARACTERIZATION AND EXPRESSION OF HERV-W 1179
on May 30, 2013 by guest
the cDNAs reflects a genomic RNA organization as suggested
by the Northern blot analysis, the lengths of genomic, sub-
genomic, and small mRNAs deduced from the cDNA clones
would be 7,541, 2,812, and 1,148 nt, consistent with the 8-, 3.1-,
and 1.3-kb observed transcripts, respectively.
Although no band larger than 8 kb was observed in the
Northern blot, the alignment of the experimental clones with
the 10.2-kb DNA sequence of the RG083M05 clone was per-
formed in order to understand whether these RNA transcripts
result from splicing events or indicate expression of defective
HERV-W proviruses. This strategy permitted us to identify
several apparently well-conserved splice donor and acceptor
sites (Fig. 6B). The comparison of the 5? cDNA clones cl.44.4
and cl.6A2 with the RG083M05 sequence identified a splice
donor (DS1) and a splice acceptor (AS1) site, strictly flanking
the 2,075-bp insertion previously observed (Fig. 4). They were
found on RG083M05, just 564 and 2,640 nt, respectively, down-
stream from R, as defined above. The absence of a major splice
donor site in the 5? cDNA clones cl.44.4 and cl.6A2 suggested
that the 8-kb apparent genomic RNA could result from a
splicing strategy. The comparison of cl.PH74 and cl.PH7 clones
with the RG083M05 sequence identified two putative splice
acceptor sites, AS2 and AS3, located 7,338 and 9,001 bp down-
stream from R, respectively (Fig. 6B). AS3 was also found in
the cl.Pi5T clone, which contained, in addition, a second pu-
tative donor site (DS2), 7,545 bp downstream from R and 207
bp downstream from the env acceptor site but upstream from
the env ATG codon. The occurrence of DS1/AS1, DS1/AS2,
and DS1/AS3 splicing events on a putative RG083M05 precur-
sor RNA would lead to 7,491-, 2,793-, and 1,130-nt transcripts
whose sizes are remarkably close to those of genomic, sub-
genomic, and small mRNAs deduced from the cDNA clones.
Coding capacities. On the cl.PH74 clone, the initiation co-
don of the long ORF putatively encoding the envelope protein
was found 227 bp downstream from the putative DS1/AS2
splice junction. It was the first ATG downstream from this
junction, although not the first from the 5? end of the sub-
genomic RNA as five ATGs preceding five small ORFs (less
than 27 aa) were situated ahead. Nevertheless, this ATG was in
a relatively favorable context (CCCATGG) although slightly
different from the known (A/G)CCATGG favorable context
for translation (24, 25). Figure 7A presents the amino acid
sequence of a putatively functional envelope protein. This se-
quence was derived from three clones: cl.PH74 presented a 5?
LTR, a splice junction, a coding sequence of 538 aa, and a 3?
LTR poly(A)?; cl.24.4 presented a 5? LTR, a splice junction,
and a coding sequence of 410 aa; and cl.C4C5 presented a
coding sequence of 242 aa, a 3? LTR, and poly(A)?. This ORF
exhibited the characteristic features of the precursor polypep-
tide of retroviral envelope proteins. It presented a leader pep-
tide at the amino terminus and a carboxy-terminal hydropho-
bic segment which could anchor the protein in the membrane.
A furin cleavage site (RNKR) separated the two characteristic
subdomains consisting of the surface protein (SU) and the
transmembrane protein (TM). The TM contained, in addition
to the membrane-spanning segment, a hydrophobic fusion do-
main and a putative immunosuppressive region homologous to
the immunosuppressive p15E retroviral peptide conserved
among murine, feline, and human retroviruses (11). The SU
and TM regions contained seven and one potential glycosyla-
tion site, respectively. Furthermore, these potential glycolysa-
tion sites are conserved among the three considered clones and
in the RG083M05 clone, which might code for the same pro-
tein too. An in vitro transcription-translation assay has been
performed on this env gene. The results are shown in Fig. 7B.
The measured molecular mass of the precursor is 60 kDa, in
agreement with the calculated molecular mass of 59,565 Da of
the Env ORF based on its amino acid composition. Glycosyl-
ation of the Env precursor was observed, consistent with the
prediction of carbohydrate addition sites. The measured mo-
lecular mass of the glycosylated protein is 80 kDa.
Two initiation codons potentially directing the synthesis of
52-aa (ORF1) and 48-aa (ORF2) peptides were found 22 and
95 bp downstream from the putative AS3 splice acceptor
site, respectively. Both ATGs were not in a favorable context.
ORF1 consisted of the carboxy-terminal part of Env, and
ORF2 was translated in a different but overlapping frame. No
obvious homology was found by using BLAST query. However,
with an LFASTA query on the restricted Retroviridae subda-
tabase, ORF1 and ORF2 showed about 35% identity with Rex
from primate and human T-lymphotropic virus and Tat from
simian immunodeficiency virus, respectively (data not shown).
Although unusual, the presence of such regulatory proteins in
HERVs has been described elsewhere for HERV-K10 (33).
FIG. 4. (A) Southern blot analysis of human genomic DNA digested sepa-
rately with EcoRI (E), HindIII (H), and PstI (P) and probed with Pgag-LB19,
Ppro-E, and Penv-C15. The number of bands detected on the blot is indicated at
the bottom for each restriction enzyme. The absolute copy numbers deduced
from dot blot analysis, as well as confidence intervals, are also indicated at the
bottom. Un., undigested. (B) Alignment of the placental cDNA clones (cDNAs;
schematic representation of the 7.6-kb region defined with the overlapping
cDNA clones) with the four sequences showing the higher scores with BLASTN
query. Shown are human BAC clone RG083M05 (gb AC000064), human
BAC378 (gb U85196 and gb AE000660), H. sapiens cosmid Q11M15 (gb
AF045450), and human DNA sequence from cosmid U134E6 (EMBL Z83850).
Locations of the aligned region of each clone are indicated. Chromosomal
assignments are indicated in brackets. The percentages of similarity (without
gaps) between the four sequences and the consensus sequence (cDNAs) are
given (values for individual clones are indicated in the text). The presence of
repeats at both ends of the genome is stated.
1180 BLOND ET AL.J. VIROL.
on May 30, 2013 by guest
FIG. 5. Definition of a putative LTR. (A) The 2,364- to 2,720-nt region of the cl.PH74 clone, identified as a putative promoter region with Signalscan 4.05, was PCR
amplified and subcloned into pCAT3 reporter vector. The resulting pCAT3-3?LTR plasmid and pCAT3 control plasmid were used to transfect HeLa cells prior to CAT
activity determination. Cells, HeLa cell extract; pCAT3 and pCAT3-3?LTR, extract of HeLa cells transfected with control plasmid and promoter plasmid, respectively.
(B) Alignment of 5? and 3? UTRs of placental experimental clones with the 5? (5-RG-28000-28872) and 3? (3-RG-37500-38314) repeated sequences of the 28,000- to
38,314-nt fragment of human DNA sequence RG083M05. The end of the 3?-most ORF is indicated (env orf). The tandemly repeated CAAC flanking sequences are
doubly underlined on DNA sequences. The 783-bp LTR consensus sequence is positioned at the bottom. The polypurine tract (PPT) upstream from the 5? end of the
LTR and the tRNATrpPBS downstream from the 3? end of LTR are indicated. U3, R, and U5 subparts are indicated, except for the U3-R junction. Transcription factor
sites determined with Signalscan 4.05 are underlined and labelled from I to VI. The sites V and VI may correspond to CCAAT box and TATA box, respectively. The
3? region of clone cl.PH74 used in the CAT assay is underlined on the cl.PH74 sequence. ? and Œ, potential cap sites obtained in 5? RACE experiments with human
placental poly(A)?RNA and total RNA extracted from pCAT-3?LTR-transfected HeLa cells, respectively. [polyA], polyadenylation signal.
on May 30, 2013 by guest
Here we report the molecular characterization of HERV-W,
a new multicopy family of HERVs whose expression in healthy
tissues seemed restricted to the placenta. The phylogenetic
trees within the Pol region showed that the HERV-W family is
related to ERV-9 and RTVL-H families and thus belongs to
the class I endogenous retroviruses (5). Phylogenetic analysis
of the env ORF showed that it was closer to simian type D and
avian reticuloendotheliosis retroviruses than to murine type C
retroviruses. The homologies within the pol and env genes with
the murine type C and simian type D retroviruses, respectively,
suggest a chimeric genome structure as described for baboon
endogenous virus (22). Based on the size criteria, such a chi-
merism seemed to exist within the LTR: the 247-nt U3 and the
79- to 81-nt R elements were comparable to avian or type D
retrovirus U3 and mammalian type C R elements, respectively,
although the 410- to 455-nt U5 element remained unclassified
as unusually long (57). The bush-like topology that is observed
in the phylogenetic trees for the HERV-W and ERV-9 se-
quences suggests that in both families most elements were
fixed in the germ line during a relatively short period of evo-
lution and probably derive from one or a few active elements.
The phylogenetic tree, supported by high bootstrap values,
shows that ERV-9 and HERV-W families derive from two
independent bursts of insertions. Thus, the active element(s) at
the origin of the HERV-W family is distinct from the one(s)
from which the ERV-9 family is derived. Moreover, HERV-W
PBS is predicted to use tRNATrp, whereas ERV-9 probably
uses tRNAArg. Finally, members of the HERV-W family are
expressed in the placenta, whereas we have not detected ERV-
9 RNAs in this tissue. However, by RNase protection assay, the
detection of protected fragments smaller than the expected
one suggested that ERV-9-related sequences could be ex-
pressed at a low level in placenta (26).
The persistence of flanking duplicated sequences generated
during the integration process suggests either that a selection
process persists within the LTR or that the integration oc-
curred recently. The 5? and 3? LTRs of RG083M05 HERV-W
DNA prototype showed 4% divergence and 6% gaps. This is
comparable to the 0.2 to 4.3% divergence described for
HERV-K, but smaller than the 5 to 12% divergence described
for a number of cloned proviruses (5). Given an evolution rate
of 3.5 ? 10?9per site per year (30), the 4% divergence cor-
responds to a relatively recent integration event 6 million years
ago. The HERV-W distribution in different primates will be
studied to determine the first introduction into the germ line.
Like other HERVs, most (if not all) current members of the
HERV-W family are defective for replication, due to muta-
tions that disrupt one or more of the gag, pol, and env ORFs
(38, 58). The observation of more gag-pro than env sequences
in the human genome may reflect the loss of the env gene.
Several genomic clones seemed to reflect such a loss, which
appears to be a common feature among endogenous elements
(5). However, a complete env ORF was observed on the
cl.PH74 placental cDNA clone and the genomic RG083M05
clone which was nevertheless defective for replication. These
results are not compatible with the hypothesis that particles in
cell culture supernatants of MS patients (42, 44) or the pres-
ence of specifically packaged mRNA in plasma of MS and RA
patients (17, 40) may result from a replication-competent
HERV. However, although no replication-competent genome
has been observed in the databases, it cannot be excluded that
a nonidentified one exists. Furthermore, no HERV single ge-
netic unit has yet been shown experimentally to be a source of
retroviral particles (58), even in the HERV-K family, which
corresponds to the HTDV particles (6, 32). On the other hand,
a trans-complementation process between distinct but individ-
ually defective loci cannot be excluded. Indeed, queries on
nucleic acid databases showed that some large regions were
conserved and confirmed a widespread distribution on numer-
ous chromosomes. To address these two hypotheses, a new
approach will be developed to detect potential coding se-
quences, irrespective of the existence of a putative single ge-
netic element. A procedure based on probe hybridization cou-
FIG. 6. (A) Northern blot analysis of poly(A)?RNAs from placental tissues with the U5(g) (nt 115 to 717 of cl.6A2) and U5(e) (nt 1 to 491 of cl.24.4) probe, gag
(Pgag-LB19), pro (Ppro-E), pol (Ppol-MSRV), and env (Penv-C15) probes as described in Materials and Methods and the U3(e) (nt 732 to 1116 of cl.C4C5) probe.
(B) Hypothetical splicing strategy. The RG083M05 prototype DNA clone as well as six experimental clones (cl.6A2 and cl.44.4, cl.24.4 and cl.PH74, cl.PH7, and cl.Pi5T)
isolated from the cDNA library is depicted. The nucleotide sequences of regions overlapping putative splice sites are boxed. Nucleic acid sequences of RG083M05 and
all placental cDNA clones are indicated with lowercase and capital letters, respectively. AG and GT bordering splice sites are in boldface. Splice donor (DS) and
acceptor (AS) sites are indicated by rightward and leftward arrows, respectively. U3 (hatched from lower left to upper right), R (light grey), U5 (hatched from upper
left to lower right), 2-kb insert (vertically striped), Gag (black), Pol (white), and Env (dark grey) are shown as boxes. The positioning of the seven probes used in the
Northern blot analysis and the placental region spanned by the cDNAs (cDNAs) are illustrated at the bottom of the figure.
1182 BLOND ET AL.J. VIROL.
on May 30, 2013 by guest
pled with in vitro transcription-translation will be applied to
isolated human chromosomes in heterokaryon somatic hybrid
Although HERV-W is a quite complex family at the geno-
mic level, a relatively simple pattern of mRNA expression was
observed by Northern blot analysis, consisting of three major
bands of 8, 3.1, and 1.3 kb. This pattern resembles the one of
lentiviruses or oncoviruses such as mouse mammary tumor
virus and human T-cell leukemia virus (45). It is also similar to
the one described for ERV-9 in undifferentiated embryonal
carcinoma cells (28) and the HTDV/HERV-K family in ter-
atocarcinoma cell lines (32), in which three to four transcripts
were derived from a single locus by alternative usage of splic-
ing signals. Whether the observed transcripts resulted from a
splicing strategy of one or several genetic entities or originated
from different defective entities remains unresolved. Due to (i)
the overall behavior of U5, gag, pro, pol, env, and U3 probes in
Northern blot analysis and (ii) the identification of conserved
splice sites on the RG083M05 DNA prototype and the cDNA
clones, it is probable that the three observed bands represent
genomic, subgenomic, and large spliced mRNAs. The situation
resulting from the proposed splicing strategy was quite unusual
as the genomic mRNA would result from a splicing event.
Furthermore, no 9.6-kb precursor mRNA was detectable by
Northern blotting, which may reflect a very short half-life of
such a transcript. A query of the EST database with the 2-kb
insert of the RG083M05 clone showed only smaller internal
fragments, suggesting either a complex splicing strategy or the
presence of shorter unidentified genomes. The existence of an
alternative splicing strategy was confirmed by the isolation of
clone cl.Pi5T, for which no trivial corresponding transcript was
revealed on the Northern blot; thus, this clone may represent
a poorly expressed mRNA. Nevertheless, we cannot exclude
that some if not most of the placental cDNA clones, although
highly homologous, were obtained from different genetic enti-
ties. The most divergent clone, Pol cl.7A16, may reflect such a
situation. The contradiction between the 5? end of the cl.6A2
clone and the positioning of the predicted TATA box may also
result from such a context. This clone may be derived from
transcription driven by a nonretroviral promoter upstream
from a complete or altered LTR and an extension through the
R polyadenylation site as described for HERV-R (ERV-3)
(20). Taken as a whole, this suggests that the complex HERV-
W family observed at the DNA level could include well-con-
served genomic elements (subset) coexpressed in the placenta.
According to the results of the Northern blot and the
RNA dot blot analyses, a significant LTR functionality among
healthy tissues seemed restricted to the placenta. Thus, the ex-
pression of HERV-W could be regulated in a hormone-depen-
dent manner as suggested for ERV-9 with the ZNF80 zinc
finger gene (15) or for HERV-R with the H-plk gene (21).
However, by more sensitive techniques, it may be possible to
find much lower expression of HERV-W in nonplacental tis-
sue. Such related mRNA sequences were found in brains of
healthy and MS individuals (8, 29), by an RT-PCR-based
method, and a complex mRNA expression was observed in
various tissues by Northern blotting with the amplified product
as a probe (29). This discrepancy with our experimental data,
obtained with six probes spanning all the regions of a retroviral
genome, may result from different experimental conditions,
including the length of the probe and the hybridization strin-
gency. Furthermore, the EST query showed a significant con-
centration of HERV-W expression in fetal liver-spleen, in ad-
dition to placenta. As the RNA dot blots were negative for
both fetal liver and fetal spleen, one may suspect the influence
of the differentiation stage or a low level of expression. Partial
LTR tags were found in other fetal, adult, and tumoral tissues,
most in the opposite direction. This may reflect expression
driven by solitary LTRs or nonretroviral promoters. Interest-
ingly, the 85 to 94% similarities observed between the placen-
tal cDNA clones and the probes derived from extracellular
RNA isolated in a pathological context compared with the 90
to 100% similarities obtained between overlapping placental
cDNAs indicate that at least some of the probes were derived
from transcriptional units distinct from those expressed in the
placenta. This interpretation was supported by the reproduc-
ible isolation of env interrupted ORFs from extracellular
mRNA in MS and RA patients (22b) versus an env-containing
ORF from the placental cDNA library screening. This suggests
a placenta-restricted expression of an HERV-W subset versus
a different or probably broader HERV-W expression in a
pathological context (or in other tissues). In agreement with
this, the expression in different tissues at different times of
different subsets of one given family of elements was described,
e.g., the expression of HTDV/HERV-K (27), ERV-9 (31), and
HERV-R (13, 19) mRNAs differs quantitatively and qualita-
tively with respect to the tissue considered. In pathological
situations, silent LTRs could be reactivated by several factors
such as virus, e.g., members of the herpesvirus family (3, 9, 18),
or by local immune activation.
FIG. 7. (A) Sequence of the putative HERV-W envelope polypeptide de-
rived from three different placental cDNA clones. The leader peptide (L), the
surface protein (SU), and the transmembrane protein (TM) are indicated be-
tween arrows. The hydrophobic fusion peptide and the carboxy-transmembrane
region are singly and doubly underlined, respectively. The immunosuppressive
region is indicated by italics. Potential glycosylation sites are indicated by dots.
Divergent amino acids are indicated on the lower line. (B) In vitro transcription-
translation assay for HERV-W envelope gene product, without (RR?) or with
(RR?) canine microsomes, and control (Cont) without template.
VOL. 73, 1999CHARACTERIZATION AND EXPRESSION OF HERV-W1183
on May 30, 2013 by guest
The finding in placental cDNA clones of a large ORF coding
for a putatively functional envelope protein suggests some
physiological functions related to pregnancy, such as a role in
the creation of the syncytiotrophoblast layer of the placenta (7,
55) and a role in suppressing the maternal immune response
against the fetal allograft (7, 55, 56) as suggested for the en-
velope protein of HERV-R. In pathological situations, an
HERV Env might protect against exogenous retroviral in-
fections by a receptor interference mechanism (4) or alter
the local cellular immunity via, for example, a superantigen-
encoded region, as was recently proposed for type I diabe-
tes (insulin-dependent diabetes mellitus) (14). Conversely,
HERV-W expression may represent only a consequence of
physiological or pathological events. Whatever the situation is,
the finding of HERV-W sequences both in a physiological
tolerogenic immune context such as pregnancy and in autoim-
mune diseases such as MS and RA deserves further investiga-
We thank F. L. Cosset (UCB-Lyon I, Lyon, France), C. Guillon
(Erasmus University, Rotterdam, The Netherlands), and Doran Spen-
cer for critical reading of the manuscript. We also thank C. Gautier
and all the staff of the Biometry Laboratory (UCB-Lyon I, Lyon,
France) for helpful discussions and assistance with nucleic acid analysis
and F. Penin and C. Gourgeon (IBCP, Lyon, France) and F. Horn
(EMBL, Heidelberg, Germany) for helpful assistance in protein anal-
ysis. Most of the sequence analyses were performed in the PBIL (Po ˆle
Bio-Informatique Lyonnais) facilities (http://pbil.univ-lyon1.fr). We
are grateful to Florence Komurian-Pradel and Glaucia Parhanos-Bac-
cala for providing C15 and LB19 and E clones, respectively.
J.-L.B. was supported by a doctoral fellowship from the Ministe `re de
l’Enseignement Supe ´rieur et de la Recherche and bioMe ´rieux.
1. Abdeddaim, S. 1996. Fast and sound two-step algorithms for multiple align-
ment of nucleic sequences, p. 4–11. In Proceedings of the IEEE Interna-
tional Joint Symposia on Intelligence and Systems, 4 to 5 November 1996,
Los Alamos, N.Mex.
2. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990.
Basic local alignment search tool. J. Mol. Biol. 215:403–410.
3. Bergstro ¨m, T., O. Andersen, and A. Vahlne. 1989. Isolation of herpes simplex
virus type 1 during first attack of multiple sclerosis. Ann. Neurol. 26:283–285.
4. Best, S., P. R. Le Tissier, and J. P. Stoye. 1997. Endogenous retroviruses and
the evolution of resistance to retroviral infection. Trends Microbiol. 5:313–
5. Boeke, J. D., and J. P. Stoye. 1997. Retrotransposons, endogenous retrovi-
ruses, and the evolution of retroelements, p. 343–435. In J. M. Coffin,
S. H. Hughes, and H. E. Varmus (ed.), Retroviruses. Cold Spring Harbor
Laboratory Press, Plainview, N.Y.
6. Boller, K., H. Ko ¨nig, M. Sauter, N. Mueller-Lantzsch, R. Lo ¨wer, J. Lo ¨wer,
and R. Kurth. 1993. Evidence that HERV-K is the endogenous retrovirus
sequence that codes for the human teratocarcinoma-derived retrovirus
HTDV. Virology 196:349–353.
7. Boyd, M. T., C. M. R. Bax, B. E. Bax, D. L. Bloxam, and R. A. Weiss. 1993.
The human endogenous retrovirus ERV-3 is upregulated in differentiating
placental trophoblast cells. Virology 196:905–909.
8. Brahic, M., and J. F. Bureau. 1997. Multiple sclerosis and retroviruses. Ann.
9. Challoner, P. B., K. T. Smith, J. D. Parker, D. L. MacLeod, S. N. Coulter,
T. M. Rose, E. R. Schultz, J. L. Bennett, R. L. Garber, M. Chang, P. A.
Schad, P. M. Stewart, R. C. Nowinski, J. P. Brown, and J. P. Burmer. 1995.
Plaque-associated expression of human herpesvirus 6 in multiple sclerosis.
Proc. Natl. Acad. Sci. USA 92:7440–7444.
10. Chomczynski, P., and N. Sacchi. 1987. Single step method of RNA isolation
by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Bio-
11. Cianciolo, G. J., T. D. Copeland, S. Oroszlan, and R. Snyderman. 1985.
Inhibition of lymphocyte proliferation by a synthetic peptide homologous to
retroviral envelope protein. Science 230:453–455.
12. Claverie, J. M., and D. J. States. 1993. Information enhancement methods
for large scale sequence analysis. Comput. Chem. 17:191–201.
13. Cohen, M., N. Kato, and E. Larsson. 1988. ERV3 human endogenous pro-
virus mRNAs are expressed in normal and malignant tissues and cells, but
not in choriocarcinoma tumor cells. J. Cell. Biochem. 36:121–128.
14. Conrad, B., R. N. Weissmahr, J. Bo ¨ni, R. Arcari, J. Schu ¨pbach, and B. Mach.
1997. A human endogenous retroviral superantigen as candidate autoim-
mune gene in type I diabetes. Cell 90:303–313.
15. Di Cristofano, A., M. Strazzullo, L. Longo, and G. La Mantia. 1995. Char-
acterization and genomic mapping of the ZNF80 locus: expression of this
zinc-finger gene is driven by a solitary LTR of ERV9 endogenous retroviral
family. Nucleic Acids Res. 23:2823–2830.
16. Galtier, N., M. Gouy, and C. Gautier. 1996. SEA VIEW and PHYLO_WIN:
two graphic tools for sequence alignment and molecular phylogeny. Comput.
Appl. Biosci. 12:543–548.
17. Gaudin, P., H. Perron, G. Favre, B. Mandrand, R. Juvin, F. Marcel, F.
Beseme, F. Bedin, F. Mallet, B. Mougin, J. Fauconnier, J. M. Seigneurin,
and X. Phelip. 1997. Detection of retrovirus RNA in plasma from rheuma-
toid arthritis. Arthritis Rheum. 40:S245.
18. Haahr, S., M. Sommerlund, T. Christensen, A. W. Jensen, H. J. Hansen, and
A. Moller-Larsen. 1994. A putative new retrovirus associated with multiple
sclerosis and the possible involvement of Epstein-Barr virus in this disease.
Ann. N. Y. Acad. Sci. 724:148–156.
19. Kato, N., E. Larsson, and M. Cohen. 1988. Absence of expression of a
human endogenous retrovirus is correlated with choriocarcinoma. Int. J.
20. Kato, N., S. Pfeifer-Ohlsson, M. Kato, E. Larsson, J. Rydnert, R. Ohlsson,
and M. Cohen. 1987. Tissue-specific expression of human provirus ERV3
mRNA in human placenta: two of the three ERV3 mRNAs contain human
cellular sequences. J. Virol. 61:2182–2191.
21. Kato, N., K. Shimotohno, D. VanLeeuwen, and M. Cohen. 1990. Human
proviral mRNAs down regulated in choriocarcinoma encode a zinc finger
protein related to Kru ¨ppel. Mol. Cell. Biol. 10:4401–4405.
22. Kato, S., K. Matsuo, N. Nishimura, N. Takahashi, and T. Takano. 1987. The
entire nucleotide sequence of baboon endogenous virus DNA: a chimeric
genome structure of murine type C and simian type D retroviruses. J. Genet.
22a.Komurian-Pradel, F. Unpublished data.
22b.Komurian-Pradel, F. Personal communication.
23. Kozak, C. A., and S. Ruscetti. 1992. Retroviruses in rodents, p. 405–481. In
J. A. Levy (ed.), The Retroviridae, vol. 1. Plenum Press, New York, N.Y.
24. Kozak, M. 1989. The scanning model for translation: an update. J. Cell Biol.
25. Kozak, M. 1992. A consideration of alternative models for the initiation of
translation in eukaryotes. Crit. Rev. Biochem. Mol. Biol. 27:385–402.
26. La Mantia, G., D. Maglione, G. Pengue, A. Di Cristofano, A. Simeone, L.
Lanfrancone, and L. Lania. 1991. Identification and characterization of
novel human endogenous retroviral sequences preferentially expressed in
undifferentiated embryonal carcinoma cells. Nucleic Acids Res. 19:1513–
27. La Mantia, G., G. Pengue, D. Maglione, A. Pannuti, A. Pascucci, and L.
Lania. 1989. Identification of new human repetitive sequences: character-
ization of the corresponding cDNAs and their expression in embryonal
carcinoma cells. Nucleic Acids Res. 17:5913–5922.
28. Lania, L., A. Di Cristofano, M. Strazzullo, G. Pengue, B. Majello, and G. La
Mantia. 1992. Structural and functional organization of the human endog-
enous retroviral ERV9 sequences. Virology 191:464–468.
29. Lefebvre, S., B. Hubert, F. Tekaia, M. Brahic, and J. F. Bureau. 1995.
Isolation from human brain of six previously unreported cDNAs related to
the reverse transcriptase of human endogenous retroviruses. AIDS Res.
Hum. Retroviruses 11:231–237.
30. Li, W. H., and D. Graur. 1991. Fundamentals of molecular evolution, p. 67–
73. Sinauer Associates, Inc., Sunderland, Mass.
31. Lindeskog, M., P. Medstrand, and J. Blomberg. 1993. Sequence variation of
human endogenous retrovirus ERV9-related elements in an env region cor-
responding to an immunosuppressive peptide: transcription in normal and
neoplastic cells. J. Virol. 67:1122–1126.
32. Lo ¨wer, R., K. Boller, B. Hasenmaier, C. Korbmacher, N. Mueller-Lantzsch,
J. Lo ¨wer, and R. Kurth. 1993. Identification of human endogenous retrovi-
ruses with complex mRNA expression and particle formation. Proc. Natl.
Acad. Sci. USA 90:4480–4484.
33. Lo ¨wer, R., R. R. To ¨njes, C. Korbmacher, R. Kurth, and J. Lo ¨wer. 1995.
Identification of a Rev-related protein by analysis of spliced transcripts of the
human endogenous retrovirus HTDV/HERV-K. J. Virol. 69:141–149.
34. Mak, J., and L. Kleiman. 1997. Primer tRNAs for reverse transcription.
J. Virol. 71:8087–8095.
35. Mallet, F., P. Cros, and B. Mandrand. 1995. Enzyme-linked oligosorbent
assay for detection of PCR-amplified HIV-1, p. 19–28. In Y. Becker and G.
Darai (ed.), PCR: protocols for diagnosis of human and animal virus dis-
eases. Springer-Verlag, Berlin, Germany.
36. Morgenstern, B., A. Dress, and T. Werner. 1996. Multiple DNA and protein
sequence alignment based on segment-to-segment comparison. Proc. Natl.
Acad. Sci. USA 93:12098–12103.
37. Palmarini, M., C. Cousens, R. G. Dalziel, J. Bai, K. Stedman, J. C. DeM-
artini, and J. M. Sharp. 1996. The exogenous form of Jaagsiekte retrovirus
1184BLOND ET AL.J. VIROL.
on May 30, 2013 by guest
is specifically associated with a contagious lung cancer of sheep. J. Virol. 70:
37a.Parhanos-Baccala, G. Unpublished data.
38. Patience, C., D. A. Wilkinson, and R. A. Weiss. 1997. Our retroviral heritage.
Trends Genet. 13:116–120.
39. Pearson, W. R. 1988. Searching protein sequence libraries: comparison of
the sensitivity and selectivity of the Smith-Waterman and FASTA algo-
rithms. Proc. Natl. Acad. Sci. USA 85:2444–2448.
40. Perron, H., J. A. Garson, F. Bedin, F. Beseme, G. Paranhos-Baccala, F.
Komurian-Pradel, F. Mallet, P. W. Tuke, C. Voisset, J. L. Blond, B. Lalande,
J. M. Seigneurin, B. Mandrand, and The Collaborative Research Group on
Multiple Sclerosis. 1997. Molecular identification of a novel retrovirus re-
peatedly isolated from patients with multiple sclerosis. Proc. Natl. Acad. Sci.
41. Perron, H., C. Geny, O. Genoulaz, J. Pellat, J. Perret, and J. M. Seigneurin.
1991. Antibody to reverse transcriptase of human retroviruses in multiple
sclerosis. Acta Neurol. Scand. 84:507–513.
42. Perron, H., C. Geny, A. Laurent, C. Mouriquand, J. Pellat, J. Perret, and
J. M. Seigneurin. 1989. Leptomeningeal cell line from multiple sclerosis with
reverse transcriptase activity and viral particules. Res. Virol. 140:551–561.
43. Perron, H., B. Gratacap, B. Lalande, O. Genoulaz, A. Laurent, C. Geny, M.
Mallaret, P. Innocenti, E. Schuller, P. Stoebner, and J. M. Seigneurin. 1992.
In vitro transmission and antigenicity of a retrovirus isolated from a multiple
sclerosis patient. Res. Virol. 143:337–350.
44. Perron, H., B. Lalande, B. Gratacap, A. Laurent, O. Genoulaz, C. Geny, M.
Mallaret, E. Schuller, P. Stoebner, and J. M. Seigneurin. 1991. Isolation of
a retrovirus from patients with multiple sclerosis. Lancet 337:862–863.
45. Petropoulos, C. 1997. Retroviral taxonomy, protein structures, sequences,
and genetic maps, p. 757–805. In J. M. Coffin, S. H. Hughes, and H. E.
Varmus (ed.), Retroviruses. Cold Spring Harbor Laboratory Press, Plain-
46. Prestridge, D. S. 1991. SIGNALSCAN: a computer program that scans DNA
sequences for eukaryotic transcriptional elements. CABIOS 7:203–206.
47. Prestridge, D. S. 1995. Predicting polII promoter sequences using transcrip-
tion factor binding sites. J. Mol. Biol. 249:923–932.
48. Rost, B., R. Casadio, P. Fariselli, and C. Sander. 1995. Prediction of helical
transmembrane segments at 95% accuracy. Protein Sci. 4:521–533.
49. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method
for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425.
50. Shih, A., R. Misra, and M. G. Rush. 1989. Detection of multiple, novel
reverse transcriptase coding sequences in human nucleic acids: relation to
primate retroviruses. J. Virol. 63:64–75.
51. Sonnhammer, E. L. L., and R. Durbin. 1994. A workbench for large-scale
sequence homology analysis. CABIOS 10:301–307.
52. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W:
improving the sensitivity of progressive multiple sequence alignment through
sequence weighting, position specific gap penalties and weight matrix choice.
Nucleic Acids Res. 22:4673–4680.
53. Thoraval, P., P. Savatier, J. H. Xiao, F. Mallet, J. Samarut, G. Verdier, and
V. Nigon. 1987. Partial nucleotide sequence of the avian erythroblastosis
virus (AEV ES4). Nucleic Acids Res. 15:9612.
54. Tuke, P. W., H. Perron, F. Bedin, F. Beseme, and J. A. Garson. 1997.
Development of a pan-retrovirus detection system for multiple sclerosis
studies. Acta Neurol. Scand. 95:16–21.
55. Venables, P. J. W., S. M. Brookes, D. Griffiths, R. A. Weiss, and M. T. Boyd.
1995. Abundance of an endogenous retroviral envelope protein in placental
trophoblasts suggests a biological function. Virology 211:589–592.
56. Villareal, L. P. 1997. On viruses, sex, and motherhood. J. Virol. 71:859–865.
57. Vogt, V. M. 1997. Retroviral virions and genomes, p. 27–69. In J. M. Coffin,
S. H. Hughes, and H. E. Varmus (ed.), Retroviruses. Cold Spring Harbor
Laboratory Press, Plainview, N.Y.
58. Wilkinson, D. A., D. L. Mager, and J. A. C. Leong. 1994. Endogenous human
retroviruses, p. 465–535. In J. A. Levy (ed.), The Retroviridae, vol. 3. Plenum
Press, New York, N.Y.
VOL. 73, 1999CHARACTERIZATION AND EXPRESSION OF HERV-W1185
on May 30, 2013 by guest