Primer Binding Site-Dependent Restriction of Murine Leukemia Virus Requires HP1 Binding by TRIM28

Abstract

TRIM28 is a transcriptional corepressor which is required for primer binding site (PBS)-dependent restriction of murine leukemia virus (MLV) replication in embryonic stem and embryonic carcinoma (EC) cells. PBS-dependent restriction of MLV leads to transcriptional silencing of the integrated provirus and has been shown to correlate with TRIM28-mediated recruitment of HP1 to the silenced loci. Here we show, using a cell line with a point mutation in the HP1 binding domain of TRIM28, that interaction with HP1 is absolutely required for the PBS-dependent restriction of MLV in the F9 EC cell line.

Murine leukemia viruses (MLVs) are unable to replicate in mouse embryonic carcinoma (EC) and embryonic stem (ES) cells (2, 6, 7, 18). MLVs are able to establish integrated proviral DNA normally in these cell types, but the DNA is subsequently transcriptionally silenced. This transcriptional silencing is in part due to reduced transcription factor binding to the viral enhancers (9, 12) and in larger part due to repressive trans-acting factors in these cell types (1, 4, 19, 20). The primer binding site (PBS) of MLV is a major target of such repression. The PBS of the MLV genome is complementary to 18 nucleotides at the 3′ end of the host proline tRNA and is a critical sequence for virus replication. The proline tRNA is annealed to the PBS in the RNA genome at the time of virus assembly and, upon infection, is used as the primer for minus-strand DNA synthesis during reverse transcription (8). The restriction exerted by the PBS is at the transcriptional level, and we have recently shown that it is dependent on the recruitment of the transcriptional corepressor TRIM28 (Kap-1, Tif1-beta) to the integrated MLV provirus (20). TRIM28 functions as the universal corepressor of Krüppel-associated box (KRAB) zinc finger DNA-binding proteins (5) and acts by bridging the KRAB domain of the zinc finger proteins to several known transcriptional repressors, including the NuRD histone deacetylase complex, the histone H3 K9 methyltransferase ESET, and HP1 (11, 15, 16). We have previously shown that TRIM28 recruitment to the MLV PBS during restriction is correlated with the recruitment of HP1γ (20). It has also been shown that the interaction between HP1 and TRIM28 is required for TRIM28 transcriptional repressor function in other settings and for its role in orchestrating differentiation (3, 10, 17). We sought to determine whether HP1 recruitment by TRIM28 is also required for the PBS-directed restriction of MLV.

The PBS-directed restriction of MLV replication in embryonic cells correlates with the presence of a high-molecular-weight complex which binds to the DNA sequence corresponding to the MLV PBS, as visualized by electrophoretic mobility shift assay (EMSA) using a 28-bp ³³P-labeled probe containing the MLV PBS sequence (14). The introduction of a single point mutation (know as the B2 mutation) into this DNA probe abrogates both the mobility shift and also the restriction of MLV in embryonic cells (2, 14, 20). Figure 1A shows EMSA reactions with nuclear extracts from a panel of cell lines incubated with either the wild-type (WT) PBS sequence (PRO) or the mutated B2 probe (B2). Extracts from the EC cell lines F9 and PCC4, as well as the ES cell line JM1, cause a robust shift of the PRO probe but not the mutated B2 probe. The depletion of TRIM28 from PCC4 cells with a small interfering RNA (RNAi) expression construct targeted to TRIM28 [PCC4 RNAi TRIM28 (111)] caused a dramatic reduction in the level of the shift, whereas a control small RNAi (PCC4 RNAi Scrambled) did not (Fig. 1A and reference 20). Differentiated cell lines which do not restrict MLV, such as RAT2, 293A, and HeLa cells, showed no shift of the PRO probe. NIH 3T3 cells, however, which do not show restriction, did show low levels of shift activity. This result suggests that the levels of the repressor complex in this cell line are too low to induce restriction or, alternatively, that the complex in these cells for some reason is not active (Fig. 1A). These same nuclear extracts were probed with an anti-TRIM28 antibody, and all cell lines with the exception of PCC4 RNAi TRIM28 (111) express TRIM28 at high levels, showing that TRIM28 is not limiting for the PBS-mediated restriction of MLV (Fig. 1B and reference 20). Anti-β-actin Western blotting performed on the same samples confirmed equal loadings (Fig. 1B).

FIG. 1.

(A) Nuclear extracts from F9, PCC4, PCC4 SCRAM, PCC4 RNAi TRIM28 (111) (20), HeLa, NIH 3T3, RAT2, and 293A cells were prepared and used in EMSA reaction mixtures with a ³³P-labeled 28-bp probe corresponding to either the WT (PRO) or the mutated B2 MLV PBS sequence, as indicated. RBS, repressor binding site. (B) Western blots of the same nuclear extracts probed with anti-TRIM28 ab22553 (ABCAM) and anti-β-actin antibodies.

We wished to determine whether TRIM28 that is no longer able to bind to HP1 is still able to bind to the PBS sequence and subsequently restrict MLV. To address this question, we made use of two engineered cell lines, the TRIM28^HP1box/− and TRIM28^+/− F9 cell lines. The TRIM28^HP1box/− cell line is an F9 EC line which has one copy of the TRIM28 gene deleted and has two point mutations in the HP1 binding motif (V488L490/AA) of the second copy of TRIM28 which render it unable to bind HP1 (3). This line is viable but is not able to undergo normal differentiation into parietal endoderm-like cells upon the addition of retinoic acid and dibutyryl cyclic AMP (3). The TRIM28^+/− cell line is the control progenitor of the TRIM28^HP1box/− cell line and has one copy of TRIM28 deleted and one WT copy of the gene. To test whether these cells still contain the PBS binding complex, nuclear extracts were prepared from F9, TRIM28^HP1box/−, and TRIM28^+/− cell lines and used in EMSA reactions with the PRO probe (Fig. 2A). Similar levels of the PBS binding complex and similar mobilities were observed in all of these cell lines. If anti-TRIM28 antiserum was added to these reaction mixtures, it was possible to supershift each of these complexes (Fig. 2A). These results suggest that even without the TRIM28-HP1 interaction, the PBS silencing complex is able to form on DNA and still contains TRIM28. The fact that the EMSA shifts are all of the same mobility suggests that HP1 is not an integral component of this complex (Fig. 2A). This is perhaps surprising as the HP1-TRIM28 interaction is known to occur at very high affinity, with a dissociation constant of approximately 60 nM (10). The absence of HP1 from the complex was also suggested by the observation that HP1 antibodies were unable to supershift the TRIM28 complex in WT F9 cells (data not shown). The levels of TRIM28 in each nuclear extract were assessed by Western blot analysis to confirm that each cell line still expressed similar levels of TRIM28. The blots were also probed with an anti-β-actin antibody to ensure equal loadings (Fig. 2B).

FIG. 2.

(A) Nuclear extracts from F9, TRIM28^+/−, and TRIM28^HP1box/− cell lines were prepared and used in EMSAs with a ³³P-labeled 28-bp probe corresponding to the WT MLV PBS. In the last three lanes, EMSAs were performed with the addition of 3 μg of anti-TRIM28 antibody ab22553 (ABCAM). RBS, repressor binding site. (B) Western blots of the same nuclear extracts probed with anti-TRIM28 and anti-β-actin antibodies.

These results indicated that mutated TRIM28 could form the minimal PBS silencing complex in the TRIM28^HP1box/− cells without interacting with HP1. To determine whether the TRIM28^HP1box/− cells were still able to restrict MLV in a PBS-dependent fashion, RAT2, F9 TRIM28^+/−, and TRIM28^HP1box/− cells were infected with two MLV-based retroviral vectors, LJ-PAdMLPEnh− (WT) and LJB2-ADMLPEnh− (B2). These vectors express the neomycin resistance marker from the adenoviral major late promoter and are identical except that the former has a WT proline PBS and the latter a mutated B2 PBS (13). MLV particles pseudotyped by the vesicular stomatitis virus G protein (VSV-G) and containing either the LJ-PAdMLPEnh− or the LJB2-ADMLPEnh− vector were generated, and the infectivity of each of these virus preparations was determined by colony formation assays in medium containing 0.5 mg/ml G418 (20). The WT vector is repressed by PBS binding activity, and the B2 vector is not, and therefore the ratio of the titers of these viruses in a particular cell line is a measure of the PBS-mediated restriction activity in that cell line. Rat2 cells lack PBS-mediated repression activity, and therefore the WT and B2 viruses infect these cells with very similar efficiencies. To correct for variations in titers between virus preparations, we chose to use this cell line as the standard, and the ratios of the titers of the B2 and WT viruses were all normalized to the ratio of titers in this cell line. Both the F9 parental cell line and the TRIM28^+/− cell line showed a high level of PBS-mediated restriction, manifesting as high B2/WT ratios of transduction efficiency of 38- and 53-fold, respectively. In contrast, the TRIM28^HP1box/− cells showed no PBS-mediated restriction, with a B2/WT ratio of 1.1 (Fig. 3A). The individual results for each experiment are shown in Table S1 in the supplemental material.

FIG. 3.

(A) Repression of B2 versus that of PRO MLV. RAT2, F9, TRIM28^+/−, and TRIM28^HP1box/− cell lines were infected with VSV-G-pseudotyped MLV containing either LJ-PAdMLPEnh− (WT) or LJB2-ADMLPEnh− (B2) constructs. Infection efficiency in each cell line was monitored by colony counting after 2 weeks of G418 selection. The graph shows ratios of B2 to WT infection efficiency in each cell line, normalized with RAT2, defined as equal to 1. Error bars show ± standard errors (n = 5). (B) F9, TRIM28^+/−, and TRIM28^HP1box/− cell lines were infected with either WT or B2 VSV-G-pseudotyped MLV as described above. Two weeks later, genomic DNA was extracted from these cells and PCR was performed using primers specific for either the MLV PBS or GAPDH (glyceraldehyde-3-phosphate dehydrogenase), as described previously (20).

As the TRIM28^HP1box/− cell line had a lower colony formation efficiency than the F9 and TRIM28^+/− cell lines, we wished to ensure that the observed relief of restriction in the TRIM28^HP1box/− cells was not due to a difference in DNA integration efficiency between the B2 and WT viruses. Therefore, each cell line was transduced with equal numbers of CFU (as measured in RAT2 cells) of viruses containing LJ-PAdMLPEnh− or LJB2-ADMLPEnh−, and 2 weeks postinfection, genomic DNA was extracted from these cells and the level of newly integrated proviruses was detected by PCR (Fig. 3B). In each cell line, the ratios of proviral DNA between the PRO and B2 viruses were comparable, suggesting that there was no difference in efficiency of infection in theses cells. These results as a whole suggest that TRIM28-HP1 binding is required for PBS-mediated repression of MLV in F9 cells.

The results presented here strengthen the argument that TRIM28 activity is required for PBS-mediated silencing in ES and EC cells, demonstrating that two point mutations in TRIM28 are sufficient to cause a complete loss of silencing in these cells. The results also suggest that HP1 binding by TRIM28 is required for this repression, though HP1 does not appear to be required for the formation of the core PBS binding complex. This is implied by the fact that TRIM28^HP1box/− cells still show DNA binding activity for PRO PBS DNA as measured by EMSA (Fig. 2A). These observations are consistent with the observation that HP1γ is specifically recruited to integrated proviruses that are being silenced by this complex and suggest that HP1 is responsible for orchestrating the transcriptional silencing of this locus (20). It remains possible, however, that the two amino acid changes in the TRIM28-HP1 interaction domain may also have other unknown functions which could also lead to a relief of PBS-mediated repression that is independent of HP1 binding.

These results also strongly suggest that TRIM28 acts in the PBS-mediated silencing complex as the key component recruiting transcriptional silencing machinery. Thus, attenuating the ability of TRIM28 to silence transcription could lead to the relief of PBS-mediated restriction. Cells containing a PBS binding complex but without restriction activity, such as NIH 3T3 cells, may lack some downstream mediator of TRIM28 function.