INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Endogenous retroviruses are present in multiple copies dispersed throughout the genomes of host species (9, 47). These retroelements have been vertically transmitted through the germ line as Mendelian genes for millions of years and may represent the endogenous counterparts of exogenous retroviruses that infected ancestral species long ago. It is estimated that up to 10% of the human genome is composed of sequences that were reverse transcribed and that up to 1% of the human genome is made up of endogenous retroviruses that are very similar to exogenous retroviruses both in sequence and in structure (3, 5, 47). Both the prevalence and maintenance of these elements suggest that they may play a role in the biology of the host species.

Endogenous retroelements may affect their hosts via multiple mechanisms (46, 47): (i) they can act in trans by expressing viral gene products that could interfere with or modulate cellular activities (20); (ii) they can act in cis by activating neighboring cellular genes with their regulatory sequences (11) or by retrotransposing and leading to insertional mutagenesis (31, 32); (iii) they can promote genomic plasticity as well as contribute to allelic variations (42, 48); (iv) and they have been shown, in some cases, to potentially play a role in autoimmune diseases such as glomerulonephritis (18, 37, 46).

One human endogenous retrovirus of interest is HERV-K (human endogenous retrovirus K), where the K denotes the lysine tRNA used by this element as a primer for reverse transcriptase (35). An HERV-K family member, HERV-K10, was first isolated by using a Syrian hamster intracisternal A particle (IAP) pol probe and was shown to have some homology to the mouse mammary tumor virus (MMTV) env gene (34). HERV-K elements entered the primate lineage after the divergence of New World monkeys and Old World monkeys more than 30 million years ago and underwent amplifications as well genomic rearrangements (40). Human-specific integration events have been reported and indicate relatively recent HERV-K activity (30).

HERV-K is present in about 30 to 50 copies in the human genome (34), along with an estimated 10,000 solitary long terminal repeats (LTRs) (21). Unlike many other HERV family members, some of the HERV-K proviruses have retained open reading frames for their viral genes (13, 17, 25, 27, 28, 33, 44). Although no one provirus has been demonstrated to produce the HERV-K viral particles, an almost intact provirus with all open reading frames has been found on chromosome 7 (29). Noninfectious particles derived from as of yet unidentified HERV-K proviruses have been observed in testicular teratocarcinoma cell lines and tumors, as well as in the placenta, but not in the vast majority of other tumor types or cell lines (4, 8, 24, 39).

Testicular teratocarcinomas are germ cell tumors (GCTs) that are composed of an undifferentiated embryonal carcinoma (EC) stem cell population and a variety of differentiated cell populations (1, 2). The human teratocarcinoma-derived viruses (HTDVs), which are expressed in these tumors, are arrested at late budding stages and lack the electron-lucent space between the envelope and the core shell indicative of mature particles (4). These particles seem to no longer be observed in the differentiated derivatives of GCTs such as differentiated embryonal carcinoma cells, cultures of yolk sac carcinomas, and teratomas (8), although it has also been suggested that HTDVs are actually budding from the trophoblastic component of differentiating teratocarcinoma cell cultures (22, 24).

High levels of expression of HERV-K members appear to be restricted to GCTs (including testicular teratocarcinoma cell lines) and their testicular precursor lesions (14, 15, 23). Mature and immature teratomas and spermatocytic seminomas with no embryonal carcinoma cell component show no HERV-K expression (15). Furthermore, many other tissues, tumor types, and cell lines do not demonstrate detectable levels of HERV-K expression (15, 23). Occasionally, a low level of expression has been found in chronic myeloid leukemias (6), leukocytes (7), placenta (38), peripheral blood mononuclear cells, and brain tissues of healthy persons and multiple sclerosis patients (36). The significance of these sporadic and peculiar expression patterns remains elusive.

In this study, an active HERV-K LTR has been isolated and used to study the expression patterns of the viral promoter-enhancer(s) in order to determine when the transcriptional regulatory elements of a typical HERV-K virus direct gene expression. The isolated HERV-K LTR was demonstrated to drive high-level expression of a reporter gene in human and murine teratocarcinoma cell lines but not in the differentiated counterparts or in various other tumor cell lines tested. To study the expression pattern of this viral LTR in vivo, transgenic mice that harbor the HERV-K LTR upstream of the -galactosidase reporter gene were generated. Both Northern and reverse transcription-PCR (RT-PCR) analyses of mouse tissues indicate that expression is limited primarily to the more undifferentiated spermatocytes of the adult testes and, to a lesser extent, the brain. These observations suggest that some HERV-K members harboring similarly active LTRs might be developmentally regulated, being active primarily in adult gonocytes and in undifferentiated GCTs, but not in their differentiated counterparts nor in a variety of other tissue and tumor types.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Sequences. The accession number for the 5' LTR of HERV-K10 is M12851. The accession number for HERV-K LTRE3 is AF148679. The BLAST program from NCBI was used to compare the two LTRs.

Cell lines. The cell lines used were Tera 1 (human teratocarcinoma ATCC HTB 105), Tera 2 (human teratocarcinoma ATCC HTB 106), 1218E (human teratocarcinoma cells obtained from P. A. Andrews), F9 (murine embryonal carcinoma ATCC CRL 1720), P19 (murine embryonal carcinoma ATCC CRL 1825), SaOs2 (human osteogenic sarcoma ATCC HTB 85), H1299 (human lung adenocarcinoma cells obtained from M. Murphy), MDA231 (human breast adenocarcinoma ATCC HTB 26), MDA435 (human breast carcinoma ATCC HTB 129), T-47D (human breast carcinoma ATCC HTB 133), 293 (transformed human embryonic kidney cells ATCC CRL 1573), and SW480 (human colon adenocarcinoma ATCC CCL 228).

Isolation of HERV-K LTRs. Primers specific for HERV-K LTRs were designed according to the published HERV-K10 LTR sequence (32): N5LTR1Kpn (5'-CGGGGTACCTGTTACTGTGTCTGTGTAG-3'; nucleotides [nt] 28 to 46) and N5LTR2Xho (5'-CCGCTCGAGTACACACCTGTGGGTGTT-3'; nt 949 to 932). The human teratocarcinoma cell line Tera 1 was used for PCR amplification of LTRs, most of which would presumably come from the more abundant solitary LTRs. The resulting PCR products were cloned upstream (Kpn-Xho sites) of a luciferase vector that includes a simian virus 40 (SV40) early promoter, PGL2-Promoter (Promega). To isolate LTRs from full-length proviral elements, a WI38 human fibroblast genomic library in the lambda FIXII vector (Stratagene) was screened with using an HERV-K gag probe (nt 1559 to 2037 from the HERV-K10 sequence). Lambda DNA from 6 of 10 isolated clones generated a PCR product with LTR-specific primers. Some of the LTRs were also cloned upstream of a LacZ vector devoid of any promoter (Promega parent pGEMZ vector with deleted promoter region).

Luciferase and lacZ reporter gene assays. The appropriate reporter plasmids (1 to 2 µg) were transfected into various cell lines by either electroporation (Bio-Rad gene pulser with 0.4-cm cuvettes and 0.24 V) or LipofectAMINE (Gibco BRL) as specified by the supplier. For transient transfections, cells lysates were isolated 24 h after transfection and assayed for luciferase and lacZ light units by using the Dual Light system (Tropix) on a Perkin-Elmer microplate luminometer as specified by the manufacturer. Relative light unit levels were normalized for transfection efficiency with the converse reporter gene (Promega Luciferase PGL2-Control for lacZ; and Clontech LacZ pUC19-SV40 for luciferase). Pooled stable transfectants were established by cotransfecting a neo plasmid with the reporter gene of interest at a 1:10 molar ratio (pCI-neo mammalian expression vector; Promega) and selecting with G418 (Geneticin; Gibco BRL) for 1 week or until control cells with no neo plasmids died.

X-Gal staining. Tissue culture cells were rinsed with phosphate-buffered saline (PBS) and fixed (2% formaldehyde, 0.2% gluteraldehyde, 1× PBS) for 5 min at 4°C. The cells were washed again with PBS and overlaid with the staining-reaction mixture (1 mg of 5-bromo-4-chloro-3-indolyl--D-galactopyranoside [X-Gal] per ml, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, and 2 mM MgCl₂ in PBS). The cells were then incubated in a bacterial incubator with the reaction mixture for 12 to 18 h (the color can be seen in a few hours).

Transgenic mice. The most active isolated LTR, LTRE3, was cloned upstream of the lacZ gene in a promoterless pGEM vector, and the LTRE3-lacZ sequences (devoid of vector sequences) were microinjected into fertilized eggs resulting from matings between B6D2F₁ mice (C57BL/6J and DBA/2J hybrids) (16). Transgenic mice were identified by Southern blotting with 10 µg of restriction enzyme-digested tail DNA. The blots were hybridized with HERV-K LTR and -galactosidase (-gal) probes. The tail DNA was also used for PCR analysis with primers for the HERV-K LTR and -gal sequences (data not shown). Four separate transgenic lines were identified (Tg2, Tg4, Tg7, and Tg9), and two of those lines (Tg7 and Tg9) were subsequently found to express the -gal gene by Northern or RT-PCR analysis or both. The nontransgenic line, 129Sv (Jackson Laboratory), was used as a negative control, while a transgenic line expressing lacZ in a ubiquitous fashion, ROSA-geo26 (12), was used as a positive control.

Northern blots. Total RNA was extracted from frozen mouse testis, brain, lung, kidney, thymus, heart, ovary, uterus, skeletal muscle, spleen, intestine, and liver by using TRIzol (Gibco BRL) or RNEasy (Qiagen) as specified by the supplier. Poly(A)⁺ mRNA was isolated with the oligo(dT) cellulose MessageMaker (Gibco BRL) as recommended by the manufacturer. Northern blot analyses were performed by standard methods with 10 to 20 µg of total RNA or 10 µg of poly(A)⁺ mRNA. Formaldehyde-agarose gels (1.2% formaldehyde) were electrophoresed overnight, stained with ethidium bromide to examine 18S/28S RNA integrity, and transferred with Turboblotters onto Nytran nylon membranes as specified by the manufacturer (Schleicher & Schuell). The UV-crossed-linked membranes were then hybridized overnight in Church buffer (0.5 M Na₂ HPO₄/NaH₂ PO₄, 7% sodium dodecyl sulfate [SDS], 1 mM EDTA, 1% bovine serum albumin) with labeled HERV-K LTR probe. Probes were labeled with Stratagene Prime-it kits and added to the hybridization buffer at 10⁶ cpm/ml at 65°C. The Northern blots were rinsed with 1× SSC (0.15 M NaCl, 0.015 M sodium citrate)-0.1% SDS, washed with the same solution once at room temperature for 30 min, and washed with 0.1× SSC-0.1% SDS twice at 65°C for 30 min.

RT-PCR. One microgram of total RNA was treated with DNase as specified by the manufacturer (Gibco BRL) and reverse transcribed with the GeneAmp RNA PCR kit with oligo(dT) primers as specified by the vendor (Perkin Elmer). RT cycles consisted of 10 min at room temperature, 15 min at 42°C, 5 min at 99°C, and 5 min at 4°C with dT primers. The cDNAs were then subjected to 35 rounds of PCR amplification with primers (5'-GGGATGGCGTGGGACGCGGC-3' and 5'-GGGACTGGGTGGATCAGTCGC-3') specific for -gal (the initial denaturation cycle was 94°C for 4 min; the amplification cycles were 45 s at 94°C, 45 s at 57°C, and 1 min at 72°C; and the extension cycle was 10 min at 72°C). RT-PCR controls included no reverse transcriptase added to the RT reaction mixture for one set of samples, to ensure that no contaminating DNA was present, and a different set of primers for the PCR (a gene expressed in most tissues, tpi), to assess the integrity of the starting RNA (5'-CACAAACACAATACAGGGGCTTTGGCACCGTG-3' and 5'-GAACTGCTACAAAGTGACCAATGGGCCTTTC-3').

Southern blots. Some of the RT-PCR products as well as tail DNA were run on agarose gels by Tris-acetate-EDTA gel electrophoresis. The DNA was denatured or neutralized in the gels, which were then transferred onto nylon membranes overnight by standard methods (Turboblotter and Nytran membranes). The labeled LTR probe (the same as in the Northern blot analyses) was added to the hybridization solution (5% Denhardt's solution, 0.5% SDS, 6× SSC) with the UV-cross-linked membrane for overnight hybridizations at 65°C. The Southern blots were rinsed and washed three times with 1× SSC-0.1% SDS for 30 min each at 65°C. Additional washes with 0.1× SSC-0.1% SDS were done when necessary.

Immunofluorescence. Testis tissues were fixed in 4% paraformaldehyde (in PBS) overnight and then passed through a sucrose gradient (12, 16, and 18% sucrose solutions for 3 h each at 4°C). The tissues were embedded in tissue freezing medium (Triangle Biomedical Sciences) and flash frozen in liquid nitrogen. They were then sectioned in a cryostat at a thickness of 10 µm and used for immunofluorescence with a polyclonal -gal antibody (Cortex Biochem). The sections were blocked with 10% goat serum in PBSBT (PBS, 0.5% bovine serum albumin, 0.05% Tween 20) for 30 min at room temperature. The polyclonal -gal antibody was added (1:250), and the slides were incubated for 45 min at room temperature. The slides were washed three times for 5 min in PBSBT. A secondary antibody directly conjugated to Texas Red (1:500) was then added (goat anti-rabbit immunoglobulin G [Jackson ImmunoResearch]), and the slides were incubated for 30 min at room temperature. The sections were again washed three times for 5 min with PBSBT. Mounting solution (30% glycerol, 70% PBS) was used to mount the slides with coverslips.

Nucleotide sequence accession numbers. The sequence determined in this study have been deposited in GenBank as follows: HERV-K LTRE3, AF148679.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of active HERV-K LTRs. To isolate HERV-K LTRs from human genomic DNA, primers were designed from the published HERV-K10 LTR sequences (34). LTR PCR products from a teratocarcinoma cell line (Tera 1) were cloned upstream of a luciferase reporter gene vector (PGL2-Promoter), which includes its own SV40 early promoter, to identify LTRs harboring functional enhancers (data not shown). It had been previously shown (by investigating endogenous viral transcripts by Northern and RT-PCR analyses) that HERV-K family members can transcribe their viral genes as well as some adjacent cellular genes to high levels in testicular teratocarcinoma cell lines (GCT cell lines) but only to lower or nondetectable levels in most other tumors and cell lines tested to date (23, 25, 26, 45). However, no detailed studies investigating the expression patterns with isolated active LTRs have been done.

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Nucleotide sequence homology between HERV-K LTRE3 (KE3: nt 1 to 921) and HERV-K10 5' LTR (K10: nt 78 to 999). Identities are depicted as dots. The TATA box and polyadenylation signal are capitalized. The potential glucocorticoid/enhancer site is underlined.

HERV-K LTR expression is restricted to teratocarcinoma cell lines. It has been previously observed that HERV-K expression (as determined by mRNA and protein expression as well as particle formation) is highest in testicular teratocarcinomas as well as in the cell lines derived from these tumors and is either undetectable or present at lower levels in a number of normal tissues and tumor types. Results with the isolated LTRE3 corroborate this conclusion. The enhancer(s) in LTRE3 increased luciferase expression relative to promoter alone from 10- to 70-fold in three human GCT cell lines (Tera 1, Tera 2, and 1218E) and in two murine GCT cell lines (F9 and P19) (Fig. 2A), but it did not enhance luciferase expression by more than 2-fold in several non-GCT human cell lines including a transformed embryonic kidney cell line, an osteogenic sarcoma, a lung adenocarcinoma, three breast carcinomas, and a colon adenocarcinoma (293, SaOs2, H1299, MDA231, MDA435, T47D, and SW480, respectively) (Fig. 2B).

View larger version (28K): [in this window] [in a new window]   FIG. 2.   LTRE3 expression of a reporter gene is high in GCT cell lines (A) and low in non-GCT cell lines (B). The HERV-K LTRE3 was cloned upstream of the luciferase reporter gene and was transiently transfected into various cell lines along with a -gal control vector to normalize for transfection efficiency. The relative luciferase level value is the light unit ratio between the luciferase vector (Pro-Luc) with promoter alone (SV40 early promoter) and the luciferase vector with promoter and LTRE3 (LTR-Pro-Luc).

RA differentiation of a teratocarcinoma cell line leads to a decrease in LTR expression. Previous observations (8) have suggested the possibility that HERV-K LTR expression is dependent on the state of differentiation of GCTs. EC cells are the undifferentiated, rapidly dividing stem cell population of teratocarcinomas. The murine EC cell line F9 has been successfully used as an in vitro model to study the regulation of differentiation. These EC cells do not differentiate spontaneously but can be induced to do so upon treatment with the acidic form of vitamin A, retinoic acid (RA) (41). These EC cells can differentiate into primitive endoderm cells (when treated with RA), into visceral endoderm (upon aggregation and treatment with RA), and into parietal endoderm (with RA and dibutyryl cyclic AMP) (41).

6

View larger version (40K): [in this window] [in a new window]   FIG. 3.   LTRE3 expression is downregulated upon differentiation of an EC cell line. (A) The HERV-K LTRE3 was cloned upstream of a lacZ reporter gene in a vector devoid of any promoter sequences and transiently transfected into the murine EC cell line, F9, along with a luciferase control vector to normalize for transfection efficiency. The relative lacZ level value is the ratio of the promoterless lacZ vector with that of the other vectors: no lacZ (no DNA as a negative control), LTR-lacZ (LTRE3-lacZ), and SV40 lacZ (lacZ vector with the SV40 promoter as a positive control). (B) F9 cells were also stably transfected with the LTRE3-lacZ vector and a neo vector to select for G418-resistant colonies. Colonies were pooled and stained with X-Gal to look for lacZ staining. Panel 1 shows the F9 cells stained with X-Gal. Panel 3 shows the same field by phase-contrast microscopy to show cell morphology. This pooled stable F9 cell line was also differentiated for 1 week with 106M RA. Panel 2 shows the differentiated F9 cells stained with X-Gal. Panel 4 shows the same field by phase-contrast microscopy to show the differentiated cell morphologies.

Generation of transgenic mice harboring the LTRE3-lacZ sequence. Since LTRE3 was shown to be active in murine cells (Fig. 3), it was postulated that this human LTR could be tested for tissue specificity in its ability to drive transcription in a transgenic mouse system. The same construct used in the RA differentiation experiment in Fig. 3 was used to create transgenic mice harboring the human HERV-K LTRE3 driving the expression of the bacterial lacZ gene. Four mice were shown to be transgenic for the human endogenous viral LTR and the bacterial lacZ gene by both Southern blot and PCR analyses of tail DNA (data not shown). The four independent transgenic lines (Tg2, Tg4, Tg7, and Tg9) were then crossed to 129Sv mice to generate mouse lines and homozygotes of all four independent transgenic lines.

Transgene expression is restricted mainly to the more undifferentiated spermatocytes of adult testes. Total RNA was extracted from mouse tissues (brain, liver, lung, thymus, heart, intestines, testis, ovaries, uterus, skeletal muscle, kidney, and spleen) and used for RT-PCR as well as Northern blot analysis. Of the four transgenic lines obtained, two (Tg7 and Tg9) were shown to express the lacZ gene in a few tissues and to various degrees. Tg7 and Tg9 were both found to express lacZ in adult testes when analyzed by RT-PCR (Fig. 4A). The control experiment with no reverse transcriptase enzyme (RT) showed no bands even when the RT-PCR gel was assayed by Southern blotting and long exposures were taken. This demonstrates that there is no DNA contamination in the RT-PCR reactions. The ROSA-geo26 strain (R+) was used as a positive control since it constitutively expresses lacZ in most of its organs and tissues. The 129Sv strain (S) was used as a negative control since it does not contain any lacZ sequences. Northern blot analysis of total RNA from the two expressing transgenic lines revealed that Tg7 expresses discrete sized transcripts in adult testes while Tg9 does not (expression was detectable only by RT-PCR analysis), reflecting differences in the levels of expression between the two independently derived transgenic lines. The lacZ transgenic line ROSA-geo26 and the nontransgenic line 129Sv were used to control for the specificity of the lacZ probe (Fig. 4A and C). mRNA was also isolated from both transgenic lines but, as with Northern blot analysis with total RNA, no detectable expression was seen for Tg9 (Fig. 4D).

View larger version (36K): [in this window] [in a new window]   FIG. 4.   Two independent transgenic lines express the lacZ gene in adult testes, but at different levels. (A) RT-PCR analysis followed by Southern blotting reveals that only lines Tg7 and Tg9 express the reporter gene. R+ refers to the ROSA-geo26 lacZ line, S refers to the nontransgenic control line, while Tg2, Tg4, Tg7, and Tg9 refer to the transgenic lines. Testis RNA was DNase treated and reverse transcribed in the +RT gel (but not in the RT gel) and PCR amplified with lacZ primers. (B) RNA integrity was checked by using the RT reactions to PCR amplify a ubiquitous gene, tpi. (C) Total testis RNA was used for Northern blot analysis and hybridized with a lacZ probe. (D) mRNA was isolated for Tg7 and Tg9 and hybridized to a lacZ probe.

View larger version (88K): [in this window] [in a new window]   FIG. 5.   RT-PCR analysis of tissue RNA from transgenic mice reveals the strongest expression in adult testes and, to a lesser extent, in adult brain. Tissues selected from the two expressing transgenic lines Tg7 and Tg9 were Br (brain), Ht (heart), Ts (testis), It (intestines), Kd (kidney), Lg (lung), Lv (liver), Sk (skeletal muscle), Sp (spleen), Th (thymus), Ut (uterus), and Ov (ovary). Controls included negative control tissues from 129Sv nontransgenic mice (S) and positive control tissues from a constitutive lacZ-expressing ROSA-geo26 strain (R+). All isolated RNAs were DNase treated for 30 min, and 1 µg was used for each RT-PCR. The first row (+RT -gal) is RNA that was reverse transcribed with dT primers and PCR amplified with -gal primers. The second row (RT -gal) is RNA that was not reverse transcribed but was PCR amplified with -gal primers. The third row (+RT Tpi) is the same reverse-transcribed RNA as in the first row but PCR amplified with tpi primers (a ubiquitously expressed gene).

View larger version (115K): [in this window] [in a new window]   FIG. 6.   Immunofluorescence studies reveal staining in the more undifferentiated spermatocytes of the testes of transgenic line Tg7. Frozen sections of adult testes were incubated with a polyclonal antibody against -gal to detect lacZ staining. (A) ROSA+ control with primary -gal antibody and secondary Texas Red antibody. (B) ROSA+ with only primary -gal antibody. (C) ROSA+ with only secondary Texas Red antibody. (D to F) Three sections of nontransgenic testes stained with -gal and Texas Red antibodies. (G to I) Three sections of transgenic line Tg7 stained with -gal and Texas Red antibodies. ROSA+ shows strongest staining in the more differentiated cell types of the seminiferous tubules (found near the lumen of the seminiferous tubules), while the negative control 129Sv has a faint amount of background. Sections of testes from Tg7 exhibit staining which is restricted primarily to the more undifferentiated spermatocytes found on the periphery of the seminiferous tubules.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This family of viruses is thought to be the most active HERV group in the human genome due to its high levels of expression in some tumor cell lines and its ability to code for functional proteins that can give rise to particles, albeit immature noninfectious ones. These proviruses have been maintained in the human genome for more than 30 million years. Although these proviruses may be regarded as vestiges of the past, they remain active, with the potential to impact differentiated or pathological states. It remains possible that the expression of at least one of the HERV-K viral genes or some cellular genes driven by the retroviral LTRs has been selected for and that this is the reason why some have not been mutated out of existence.

It was previously observed that HERV-K proviruses preferentially express their viral genes in GCTs such as testicular teratocarcinomas. HERV-K expression has also been found in some other tissue and tumor types, but it has been sporadic and has been present at much lower levels. No thorough studies had been undertaken to determine the expression patterns dictated by isolated active HERV-K LTRs. To address how this family of virus regulates the expression of its viral genes or neighboring cellular genes, it was necessary to isolate and characterize an active LTR.

The results in this paper demonstrate that an isolated active HERV-K LTR confers the same types of expression patterns that had been observed for the viral genes. An active HERV-K LTRE3 was able to drive the expression of reporter genes in testicular teratocarcinoma cell lines but not in a variety of other cell lines including a transformed kidney cell line, breast carcinomas, an osteosarcoma, and lung and colon adenocarcinomas. Furthermore, this LTR was downregulated upon differentiation of a teratocarcinoma cell line, indicating that it might be developmentally regulated. A transgenic mouse model was used to determine that expression in the adult tissues was limited to the testes and, to a lesser extent, the brain. Although the expression of some HERV-K elements was previously noted in the placenta (39), no expression was found in the placenta of near-term LTRE3 transgenic pups by RT-PCR (data not shown). The strong possibility remains that there are some HERV-K LTRs which generate other expression patterns (such as placenta expression or differentiated trophectoderm expression) due to sequence variations that are still uncharacterized.

All expression patterns were noted in two independent transgenic lines and were stable over all generations tested. A few other tissues did, in some RT-PCR experiments, generate faint signals, but the irregularity of these signals and their low intensities suggest that they might be considered low basal levels of expression. In contrast, RT-PCR-detected expression in testis was very reproducible and always generated strong signals in Tg7, and the expression was also seen by Northern blot analysis. Tg9 expressed the reporter gene to lower levels in testes and to slightly higher ones in brain than did Tg7. The significance of the expression in the brain is unclear, but it is worthwhile to note that a transgenic mouse harboring a murine endogenous retrotransposon LTR (IAP) directing lacZ expression was also found to only express the reporter gene in testes and brain (10).

IAPs are murine proviral elements that produce particles which assemble on the membranes of the endoplasmic reticulum and subsequently bud into the cisternae (19). These particles, like the HERV-K HTDVs, are noninfectious and cannot seemingly be transmitted horizontally. As with HERV-K particles, IAPs seem to be present in many testicular teratocarcinomas with an EC component but absent or present in fewer numbers in more highly differentiated derivatives of these GCTs (19, 43). HERV-K and IAP elements are not evolutionarily related and yet are expressed in very similar patterns. Furthermore, many other HERVs, such as HERV-H and ERV-9 family members, are likewise expressed primarily in GCTs (25). The significance of these independently derived similar expression patterns is unclear, although germ line expression can ensure their survival.

Since endogenous retroviruses such as HERV-K are transmitted vertically and do not seemingly have an exogenous phase where they would be able to infect other cells and other hosts, it is critical for their amplification or movement in the genome that they be expressed in germ cells, where any reintegration events can be passed on to the next generation. It is therefore not surprising that HERV-K does indeed seem to be preferentially expressed in germ cells, as noted by its high levels of expression in undifferentiated spermatocytes and testicular teratocarcinomas. Since LTRs drive gene expression by interacting with a variety of ubiquitous and cell-type-specific transcription factors, additional HERV-K LTR mutational and deletion studies to identify proteins that are involved in the regulation of proliferation and differentiation of germ cells and other stem cells are under way.

Testicular teratocarcinomas are the primary malignancy that have long been associated with high levels of HERV-K expression as well as particle formation. However, the significance of these findings with respect to HERV-K gene expression (i.e., whether it is involved in or merely a consequence of the development of these tumors) remains unclear. The data presented in this paper support the previous observations that HERV-K proviruses are expressed primarily in GCTs containing EC components. We contributes the novel finding that the differentiation of a GCT leads to a dramatic decrease in HERV-K LTR expression. Furthermore, the LTR is also found to be preferentially expressed in the more undifferentiated spermatocytes of adult testes. Taken together, it can now be concluded that active HERV-K members can be expressed not only in GCTs but also in healthy adult germ cells. Experiments to cross the LTRE3 transgenic mice with strains of mice that have a high predisposition to testicular cancer are under way. This should be useful in determining the extent to which this virus is expressed as the cancer develops.