Mouse mammary tumor virus (MMTV) is a betaretrovirus that encodes three accessory and regulatory proteins, a superantigen (Sag) 123, a dUTPase 4 and an RNA export protein, Rem 5. Rem is a 33 kDa protein that is encoded by a doubly spliced mRNA 56. The N-terminal portion of Rem contains nuclear and nucleolar localization signals as well as an arginine-rich motif similar to the RNA export proteins, Rev, Rex, and Rec, produced by the complex retroviruses, human immunodeficiency virus (HIV), human T-cell leukemia virus (HTLV), and human endogenous retrovirus type-K (HERV-K), respectively 56. Our previous data have shown that Rem is larger than other retroviral export proteins due to a unique C-terminus, which negatively regulates Rem-mediated RNA export activity 5. Negative regulation of MMTV transcription also occurs during viral replication in several cell types 78910. MMTV has a complex life cycle that allows transmission through maternal milk to susceptible offspring using dendritic cells as well as B and T cells 11. Amplification of MMTV in various lymphoid cell types requires virally encoded Sag to effectively transfer virus from lymphocytes to mammary epithelial cells during puberty 1213. Both the sophisticated mode of transmission and production of multiple accessory and regulatory proteins imply that MMTV is a complex retrovirus 5.

MMTV may interact with human complex retroviruses. Multiple laboratories previously have reported that MMTV sequences are detectable in human breast cancer or lymphomas, but not most normal tissues, using PCR to amplify one or more regions of the viral genome 1415161718. However, not all studies agree 1920. Recent data indicate that MMTV can infect and integrate into chromosomal DNA of cultured human cells 2122, suggesting that zoonotic infections can occur. Furthermore, MMTV is highly related to HERV-Ks [also known as human MMTV-like proviruses (HMLs)] 23. Some intact HERV-K/HML-2 proviruses have been described, consistent with their relatively recent acquisition in the human genome, yet none of these proviruses are known to be infectious 242526. A number of HERV-Ks are highly expressed in specific tissues 2327. In addition, a recent report indicates that antibodies to HERV-K/HML-2 are detectable in the plasma of breast cancer and lymphoma patients, and these titers dropped when the cancers were treated. HERV-K reverse transcriptase activity, viral RNA, processed viral proteins, and virus-like particles also could be detected in patient plasma 28. Together, these experiments suggest that sporadic MMTV infections of human cells may result in interactions with HERV-Ks or the generation of recombinant infectious viruses.

Prior experiments indicate that HIV Rev and HTLV Rex can activate expression from reporter plasmids containing the HERV-K Rec-responsive element (RcRE) 29. Because of sequence and organizational similarities between MMTV and HERV-K and the potential for MMTV infection of human cells, we have tested for interactions between heterologous retroviral export proteins and the Rem-responsive element (RmRE) using our previously described reporter vector, pHMRluc 5. Surprisingly, Rev and Rex, but not Rec, could activate MMTV-based reporter gene expression in human T cells and was dependent on the presence of the RmRE. Cell fractionation experiments followed by RNA quantitation suggested that each of the regulatory proteins, including Rem, did not affect RNA export or stability using an MMTV-based reporter vector. Rem activity was undetectable using heterologous response elements. These results suggest that retroviral export elements recognize similar features of RNA structure and support the idea that MMTV is a complex murine retrovirus that may interact with other retroviruses in human cells.

To determine if the RNA export proteins from known complex retroviruses function on the MMTV RmRE, we used the reporter vector, pHMRluc (Figure 1) 5. This vector contains the cytomegalovirus (CMV) promoter upstream of the 3' end of the MMTV genome as well as the Renilla luciferase gene between splice donor and acceptor sites 5. The RmRE appears to span the envelope-3' LTR junction 530. Detection of luciferase activity in transfected cells indicates cytoplasmic export of unspliced transcripts since splicing would remove the reporter gene. Rem expression gave a 25 to 30-fold increase in luciferase expression in Jurkat human T cells relative to cells co-transfected with the reporter plasmid and empty vector (pEGFP), whereas no increase was observed using the reporter lacking a functional RmRE (Figure 2A). Interestingly, Rev expression in Jurkat cells also increased luciferase activity approximately 3-fold compared to control cells expressing only the reporter plasmid (Figure 2A). This result is statistically significant and has been repeated in multiple experiments. Further, Rev-mediated enhancement of reporter activity required the RmRE since deletion of this element eliminated the effect (compare results using pHMRluc or pHMΔeLTRluc reporter plasmids) (Figure 2A). The Rev effect on reporter gene expression in Jurkat cells, which is abolished by the Δ3 mutation in the leucine-rich nuclear export sequence (data not shown) 31, is believed to interact with the cellular export protein Crm1 3233. In addition, Rev and Rem-mediated induction of pHMRluc activity was tested by transient transfections of 293T human kidney cells (Figure 2B). Although Rem gave a small, but statistically significant, increase in reporter activity, Rev did not. Both Rem and Rev were expressed as GFP-fusions to allow comparison of the relative expression of these regulatory proteins, and similar amounts were detected using a GFP-specific antibody and Western blotting after transfection of both cell types compared to the actin loading control (Figure 2C and data not shown).

The effect of Rev on the MMTV-based luciferase vector also was determined in HC11 mouse mammary cells since breast epithelial cells are permissive for MMTV replication 34. Rem gave a 4 to 5-fold increase in HC11 cells and was dependent on the presence of the RmRE (Figure 3A), whereas Rev gave no detectable effect in the presence or absence of the response element. Western blots verified protein expression (Figure 3B). Therefore, the heterologous export protein, Rev, appears to function in a cell-type and/or species-specific manner on the MMTV RmRE.

Transfection experiments also were performed after expression of HTLV Rex in the presence of pHMRluc (Figure 4). Rex1 and Rex2 from HTLV-1 and -2, respectively, stimulated luciferase activity in Jurkat cells in a RmRE-dependent manner, although the magnitude of the effect (ca. 5 to 7-fold) was greater than that observed for Rev (Figure 4A). Rex also was tested for stimulation of reporter gene expression in human kidney epithelial cells (293T) 35. Rem and Rex stimulated reporter expression 2-fold and 4-fold, respectively, and was dependent on the RmRE (Figure 4B). Western blotting showed similar expression of these proteins (Figure 4C and data not shown). Thus, the stimulation was dependent on the presence of the RmRE in human cells.

No RmRE-dependent effect of Rex1 or 2 was observed in mouse HC11 epithelial cells (Figure 5A), although Rex has been shown to function in mouse fibroblast cells 36. However, 2- to 4-fold increases were observed using both the pHMRluc and pHMΔeLTRluc vectors, which we attribute to cell-type-specific effects of Rex since greatly diminished activity was observed in human cells after RmRE deletion (Figure 4A). Furthermore, vectors that substituted the RmRE with the RxRE gave a 7- to 9-fold increase in luciferase expression in HC11 cells, which was dependent on Rex, but not Rem (see Figure 8). Expression of GFP-fusion proteins was equivalent in mouse and human cells as determined by Western blotting (Figures 4C and 5B). Therefore, mouse epithelial cells may lack cell-type specific proteins that allow Rex function on the MMTV RmRE. Like Rev, these results suggest that the enhancement of reporter activity by Rex is species or cell-type specific, whereas the effect of Rem is not.

Although HIV and HTLV are only distantly related to MMTV, the human endogenous retrovirus type K (HERV-K) has sequence similarity to MMTV and is a betaretrovirus that encodes the RNA export protein, Rec 37. Both HERV-K Rec and MMTV Rem are translated from the same open reading frame as their respective envelope signal peptides, but Rec lacks the extensive autoregulatory region found in the Rem C-terminus 529. Rec expression in either Jurkat (Figure 6A) or HC11 (Figure 6C) cells failed to enhance the basal luciferase activity of pHMRluc despite demonstrable expression of the GFP-fusion proteins (Figure 6B and data not shown). Together, these results indicated that Rev and Rex, but not Rec, could stimulate expression of unspliced RNA containing the RmRE.

To determine whether the HIV Rev and HTLV Rex proteins enhanced luciferase expression from the MMTV-based reporter vector through increases in RNA export or another mechanism, transfected Jurkat cells were subjected to cellular fractionation. After different detergent concentrations were tested to optimize the integrity of the fractions, nuclear and cytoplasmic RNAs were obtained, and samples were subjected to Northern blotting and staining to confirm the isolation of intact RNA and absence of rRNA precursors in the cytoplasmic fractions (Figure 7A). Subsequently, Jurkat cells were subjected to electroporation with the HIV-based reporter vector, pDM128, in the presence and absence of a RevGFP expression plasmid. Transfected cells were used to obtain nuclear and cytoplasmic fractions. RNA samples from these fractions then were subjected to semi-quantitative reverse transcription (RT)-PCRs using primers specific for the cat reporter gene (Figure 7B). As expected 383940, increased cytoplasmic levels of unspliced RNA containing the chloramphenicol acetyl transferase (cat) gene in pDM128 are observed in the presence of Rev (compare lanes 6 and 8 as well as 10 and 12 with two different amounts of cDNA). Although controls indicated that the nuclear fractions in this experiment were contaminated with DNA (Figure 7B, lanes 1 and 3), the absence of contaminating DNA in the cytoplasmic fractions further substantiated the integrity of the cellular fractionations. RT-PCRs using gapdh-specific primers confirmed similar levels of intact RNA (lanes 13–20). PCR conditions did not appear to be saturated since higher product levels could be observed for the more abundant gapdh mRNA than for reporter transcripts.

Additional experiments then were performed in Jurkat cells using the MMTV-based reporter vector (pHMRluc) and co-transfected expression vectors for GFP-tagged Rem, Rev, or Rex. RNAs from cytoplasmic and nuclear fractions were treated with DNase I and subjected to semi-quantitative RT-PCRs using primers specific for the Renilla luciferase gene or gapdh 5 (Figure 7C). PCRs without added reverse transcriptase showed that DNA contamination was absent (data not shown). Cytoplasmic RNA levels then were quantitated using ImageJ software after normalization for gapdh expression. Unexpectedly, these experiments indicated that Rem, Rev, or Rex had little effect on the levels of RNA in the nucleus or cytoplasm in Jurkat cells, suggesting that these proteins do not affect intron-containing transcript stability or export (Figure 7D).

We also tested the ability of Rem to affect expression using heterologous response elements in both Jurkat and HC11 cells (Figure 8). The response elements from HERV-K, HTLV-1, HTLV-2, and HIV were cloned into pHMΔeLTRluc, which lacks a functional RmRE, resulting in the reporter plasmids pRcRERluc, pRxRE1Rluc, pRxRE2Rluc, and pRRERluc, respectively (Figure 1). As expected, each of these plasmids gave significantly increased (100 to 1300-fold) luciferase activity in human Jurkat T cells in the presence of their homologous export protein compared to the activity of the reporter plasmid in the presence of unfused GFP (Figure 8A). Similar transfections also were performed in HC11 mouse mammary cells (Figure 8B). Since Rev had no demonstrable activity on the RmRE in HC11 cells, Rev and the pRRERluc construct were not tested in these cells. Rec, Rex1, and Rex2 showed increased reporter activity with their homologous response elements in HC11 mammary cells, but function was substantially reduced compared to that observed in Jurkat cells (Figure 8A). Further, Rem failed to enhance expression from any of the tested response elements in either cell line. Western blotting with GFP-specific antibody confirmed similar levels of each export protein (data not shown). These results suggest that the RmRE secondary and/or tertiary structure does not duplicate other retroviral export elements. However, the ability of Rem, Rev and Rex to function on the RmRE in human cells suggests common features of RNA recognition.

We have previously shown that Crm1 is required for Rem function in HC11 mouse mammary cells 5. Rev and Rex also use Crm1 for the export of intron-containing homologous RNAs 3241. To test the involvement of Crm1 in human cells, we tested whether a dominant-negative nucleoporin involved in the Crm1 export pathway (ΔCAN) 42 would affect the increased luciferase activity mediated by Rem, Rev, or Rex in Jurkat T cells. The dominant-negative protein gave a statistically significant suppression of Rem, Rev, Rex1 and Rex2 activation of the HMRluc vector, although the greatest effect was observed with Rem (Figure 9). Suppression by the ΔCAN mutant did not appear to be toxic for the cells under these conditions (data not shown). Furthermore, our previous data indicated that overexpression of a dominant-negative Tap/NXF1 mutant (TapA17) had no effect on Rem-induced reporter activity in HC11 mouse cells 5. Similarly, TapA17 overexpression in human Jurkat cells did not affect Rev, Rex, or Rem-mediated stimulation of HMRluc activity (not shown). Neither ΔCAN nor TapA17 had a dramatic effect on reporter activity in the absence of a regulatory protein (not shown). Together with previous results, these data suggest that enhancement of reporter activity by Rem requires Crm1 and that the regulatory proteins facilitate a post-export function that depends on the RmRE.

Our previous experiments have shown that MMTV Rem functions in nuclear export of unspliced viral RNA in rodent cells 5. In this manuscript, we have shown that Rem functions in human cell lines. Our results also indicate that Rev and Rex can increase reporter gene expression by interaction with the MMTV RmRE in human Jurkat T cells (Figures 2 and 4). Rex also could function on the RmRE in 293T HEK cells. Prior data indicate that some retroviral export proteins function on heterologous retroviral RNAs. For example, Rex can bind and function on both the RRE and the RxRE 4344. However, the interaction is not reciprocal since Rev cannot act on the RxRE 43. In this respect, the RmRE is quite permissive since it is required for enhancement of luciferase activity by Rem, Rev and Rex in human T cells. No effect of Rem was observed on the HIV and HTLV response elements (Figure 8). Surprisingly, the Rec export protein from the human retrovirus most closely related to MMTV, HERV-K/HML, had no effect on expression from the MMTV RmRE (Figure 6). Further, no effect of Rem was observed on the RcRE, although both HIV Rev and HTLV Rex have been reported to increase expression from the HERV-K response element 29. However, polymorphisms have been observed in different HERV-K proviruses 29, and it is possible that other RcRE variants might function with MMTV Rem or that other Rec variants may be functional on the RmRE. Given that the regulatory proteins require formation of specific secondary structures rather than a simple primary sequence 2945464748, Rec also may need secondary or tertiary structures not found in the RmRE.

The effect of Rev and Rex on the MMTV RmRE appears to be specific in human cells by several criteria. First, increases in reporter gene activity that were dependent on the RmRE were only observed in human cells with Rex or Rev. Although different results have been reported 3649, Rev appears to function in both mouse and human cell lines using vectors with a design similar to that of pHMRluc, but based on the 3' end of the HIV genome 50. Rex also has been reported to function in both human and mouse cells 36, although Rev and Rex have primarily been tested in fibroblasts 3650, which are not natural target cells for HIV, HTLV or MMTV. Second, a Rev mutant defective in the nuclear export sequence gave no specific effect in the pHMRluc assay, similar to the effect observed with RRE-containing vectors 31. Third, a dominant-negative mutant nucleoporin in the Crm1 pathway inhibited Rem, Rev, and Rex activation of reporter expression through the RmRE in human cells. Rem previously has been shown to require Crm1 in rodent cells 5, whereas Rev and Rex use Crm1 in human cells 3341. Fourth, no effect of Rec was observed on the RmRE in either mouse or human cells. Fifth, insertion of different response elements in the pHMRluc vector yielded the expected increases in luciferase activity after expression of the homologous export protein. These results indicate that the MMTV-based vector allows the activity of other response elements and that each of the GFP-fusion proteins is functional (Figure 8). Although we may have lowered the sensitivity for detection of regulatory protein function in mouse cells by testing fusion proteins, Western blotting using an antibody that recognizes all of the fusion proteins allowed us to verify that similar amounts of each protein were expressed in transfection assays. Prior experiments by Dangerfield et al. suggest that Rev can bind to the MMTV LTR and stimulate luciferase expression from constructs containing the MMTV LTR in monkey cells 51. Our studies differ significantly since their data were obtained by insertion of MMTV sequences into an HIV-based vector, and the ability of Rem to function on heterologous response elements was not determined. Furthermore, only the MMTV LTR, which lacks a portion of the RmRE 30 (our unpublished data), was present in the HIV vector 51. Thus, our data argue for a specific effect of HIV and HTLV regulatory proteins on the MMTV RmRE in human cells.

Previous experiments from our laboratory have shown that human Jurkat T cells can produce mature MMTV particles after transfection of a cloned provirus, and these particles are infectious for mice 5253. Consistent with this observation, our current data indicate that Rem can function in human cells. The reports of MMTV infection of human cells and detection of MMTV sequences in breast cancers and lymphomas 1415161718 appear to be feasible since most steps of viral replication occur in human cells. Cell entry would provide the primary barrier to infection 54. Although human cell infections appear to be inefficient and infrequent, certain cell types may have an additional entry receptor, which is dependent on cellular activation or differentiation state. The ability of Rev and Rex to function on the MMTV RmRE in human T cells suggests that rare interactions of these viruses could occur.

Rev is known to have multiple functions, including enhancement of RNA encapsidation of HIV and SIV-based vectors 55. Our previous results indicated that export of unspliced MMTV RNA and Gag expression from a transfected MMTV provirus requires Rem in rat fibroblast cells 5; encapsidation was not measured. The reporter vector pHMRluc is based on the 3' end of the MMTV provirus and has been shown to be responsive to Rem only in the presence of the RmRE in rat, mouse, and human cells 5 (this study). Further, the use of the Renilla luciferase gene in the vector provides both a sensitive and highly quantitative assay, which is difficult to achieve using RNA fractionation experiments and Northern blotting. Rev/RRE interactions also have been shown to affect Gag trafficking and HIV assembly, and it has been suggested that export elements facilitate "marking" of RNAs in the nucleus for particular events in the cytosol 56. Our experiments show that Rev and Rex function through Crm1 on pHMRluc (Figure 9). Nevertheless, cell fractionation experiments with the pHMRluc vector indicate that the regulatory proteins primarily lacked effects on nuclear RNA export and RNA stability (Figure 7). Since effects on cytosolic RNA levels and Gag production were clearly demonstrable using an MMTV proviral clone with a transposon insertion into the rem coding sequence 5, our results with the pHMRluc vector suggest that different sequence elements at the 5' end of the full-length MMTV RNA allow additional effects of Rem on RNA stability and/or export.

Published experiments indicate a wide variability (0 to 10-fold) in Rev function on RNA export 5557585960. Suboptimal splicing appears necessary to allow the accumulation of genomic HIV RNA and the export effects of Rev 61. Efficiency of splicing of full-length MMTV RNA versus pHMRluc vector RNA appears to be an unlikely explanation for differences in observed nuclear export. The splice donor and acceptor sites found in pHMRluc are those normally used to generate either the rem or sag fully spliced mRNAs, and the low abundance of these RNAs in MMTV-infected cells 662 suggests that splicing at these sites is suboptimal compared to those used to produce MMTV env RNAs from genomic RNAs. Rev also appears to overcome effects of several cis-acting repressive sequences, including sequences within gag-pol 6364 as well as env sequences that overlap with the RRE 6566. The repressive sequences in HIV gag-pol appear to be AU-rich, and mutation led to increased steady state RNA levels 64. The pHMRluc vector lacks gag-pol sequences (Figure 1), but our previous work with Rem-deficient MMTV genomic clones was consistent with defective RNA export, rather than a stabilization effect.

The cell fractionation data with pHMRluc (Figure 7) and MMTV genomic length RNA 5 argue that Rem has multiple functions, including both export and post-export activities. Rev and Rex have been reported to function at the level of translation 596768. Specific cis-acting elements found in the RU5 and gag regions of several retroviruses appear to affect translation 6970717273, but such sequences are absent in the pHMRluc vector. Since the post-export function of Rem with pHMRluc is sensitive to competition with a Crm1-binding site on Nup214 (ΔCAN) (Figure 9), it is possible Crm1 dictates Rem protein export independent of the vector RNA. Nevertheless, Rem binding to the 3' RmRE, perhaps in the cytoplasm or after binding of a cellular protein in the cytoplasm, may promote a post-export step, such as translation. Rem binding through sequence elements at the 5' end of the MMTV RNA may increase Crm1-dependent export, but such 5' elements may not be necessary for detection of the post-export activity of the pHMRluc vector. Our current data indicate that the RmRE maps to the junction of the envelope gene and the 3' LTR using deletion analysis with the pHMRluc vector and co-transfection of a Rem expression vector (see below). Interestingly, these results suggest that all MMTV mRNAs contain the 3' RmRE, unlike the RRE, which would be removed from completely spliced HIV mRNAs, such as those encoding Tat, Nef, and Rev 74. Previously published data indicate that export of unspliced genomic MMTV RNA, but not partially spliced envelope RNA, is leptomycin B and, by implication, Crm1-dependent 6. Such experiments suggest that only unspliced MMTV RNA is selectively exported. Therefore, it is possible that the MMTV genome contains two RmREs, one at the 5' end of viral RNA present only in unspliced RNA and a second element at the 3' end present in all MMTV RNAs. The 3' element may facilitate translation of all mRNAs, whereas the 5' element would specifically facilitate nuclear export of genomic RNA. Cell-type specific effects also may occur. Characterization of the molecular mechanisms of Rem function will require further investigation.

Both the pHMRluc vector and MMTV genomic RNAs contain a RmRE that spans the envelope-LTR junction 30 (Mertz et al., in preparation). Published data indicate that retroviral export/regulatory proteins bind to complex RNA structures that have multiple stems and loops 29487576. Rev and Rec appear to bind to RNA stems with a bulge, and recognition of heterologous elements may not occur through the same primary sequence as the homologous protein 29. Our current data using RmRE susceptibility to several RNases is consistent with a complex structure containing multiple stems and bulges, which encompasses a region of ca. 500 bases (Mertz et al., in preparation) rather than the single stem with multiple bulges previously proposed 30. The export of unspliced retroviral RNA is known to require specific cellular proteins, such as hnRNPs and Sam68 7778, and binding of these cellular proteins may determine the cell-type specificity observed in our experiments. Since retroviral export/regulatory proteins recognize certain RNA secondary structures 4879, one or more of these proteins may bind to and function on specific cellular RNAs as reported for Rex 80.

RNA-binding proteins appear to regulate several steps following transcription, leading to coordinated regulation of cellular RNAs with related functions called RNA regulons 81. MMTV replication in the mouse requires several different cell types, including lymphocytes and mammary epithelial cells 34. We previously have shown that MMTV replication is controlled at the transcriptional level during mammary gland development coordinately with several milk-specific genes 8283. Therefore, post-transcriptional control of MMTV expression also may be modulated by Rem depending on the cell type and state of differentiation. Our results provide additional evidence that MMTV is a murine complex retrovirus with the potential to interact with human retroviruses 5.

The authors declare that they have no competing interests.

JAM, MML, and JPD performed the experiments. JAM and JPD conceived the experiments, and all authors participated in writing the manuscript. JAM, MML and JPD have approved the manuscript.