Depletion of hnRNP A2/B1 overrides the nuclear retention of the HIV-1 genomic RNA

Abstract

hnRNP A2 is a cellular protein that is important for nucleocytoplasmic and cytosolic trafficking of the HIV-1 genomic RNA. Both hnRNP A2’s interaction with HIV-1 RNA and its expression levels influence the activities of Rev in mediating nucleocytoplasmic export of the HIV-1 genomic RNA. While the lack of Rev expression during HIV-1 gene expression results in nuclear retention of HIV-1 genomic RNA, we show here by fluorescence in situ hybridization and fractionation studies that the genomic RNA translocates to the cytoplasm when hnRNP A2/B1 are depleted from cells. Polyribosome analyses revealed that the genomic RNA was shunted into a cytoplasmic, dense polyribosomal fraction. This fraction contained several RNA-binding proteins involved in viral gene expression and RNA trafficking but did not contain the translation initiation factor, eIF4G1. Amino acid incorporation into nascent polypeptides in this fraction was also greatly reduced, demonstrating that this fraction contains mRNAs that are poorly translated. These results demonstrate that hnRNP A2/B1 expression plays roles in the nuclear retention of the HIV-1 genomic RNA in the absence of Rev and in the release of the genomic RNA from translationally inactive, cytoplasmic RNP complexes.

Keywords: :

• HIV-1 RNA
• RNA trafficking
• nuclear retention
• hnRNP A2/B1

Introduction

The trafficking of HIV-1 RNAs is characterized by the overlapping activities of several host and viral proteins during their transit from the nucleus into the cytoplasm, in the cytoplasm, and eventually, in the case of the 9 kilobasepair (kb) genomic, unspliced genomic RNA (vRNA), from the cytoplasm into progeny virions.¹ Previous work has demonstrated a critical role for the HIV-1 protein, Rev in the regulated export of the vRNA from the nucleus. This RNA is unspliced, harboring introns, and might be as a consequence retained in the nucleus or degraded by the cellular RNA surveillance machinery. Nuclear retention may also rely on the presence of cis-acting instability (INS) and nuclear retention sequences in the HIV-1 RNA that are bound by cellular proteins to promote nuclear retention.^2-4 However, the vRNA (as well as the singly-spliced HIV-1 RNA species) harbors a Rev-responsive element (RRE) to overcome nuclear retention of the RRE-containing RNAs by its association with Rev.⁵ This event is critical for efficient expression of HIV-1 structural and auxiliary proteins encoded by these mRNAs.^6-8

Several host proteins have been shown to influence Rev’s activity by direct interaction or by influencing the ribonucleoprotein (RNP) context of the RNA.^9-12 Rev also has effects on splicing regulation^13,14 as well as on downstream gene expression events at the translation level.^15-17 Efforts to understand Rev’s functions during viral replication cycle have provided several key leads in the identification of potential therapeutic strategies and targets^18,19 and have provided critical information on the processes governing nuclear RNA export in general.^20-22

Proteins of the heterogeneous nuclear ribonucleoprotein (hnRNP) family have been shown to influence Rev activity. hnRNP A1 was shown to act synergistically with Rev to contribute to nucleocytoplasmic transport of RRE-containing HIV-1 RNAs.²³ The capacity of hnRNP A1 relied on the presence of an hnRNP A1-binding site within the INS in the N-terminal Matrix domain of HIV-1 Gag.²³ Additional roles for hnRNP A1 and other hnRNPs in retroviral splicing regulation rely largely on their ability to bind sequences adjacent to and within splicing enhancer and negative splicing regulatory sequences.^18,24,25 These roles in retroviral RNA splicing and nucleocytoplasmic trafficking might also be functionally linked²⁶ a notion that is supported by newer work that shows that several functionally distinct hnRNPs interact with Rev.²⁷ Although principally nuclear, several hnRNPs indeed have cytoplasmic functions that contribute to the localization, the utilization, and fate of RNA in the cell.^28-31 Recent work for example emphasizes the importance for hnRNPs that translocate to the cytoplasm in an HIV-1-dependent manner to contribute to vRNA stability and translation in the cytoplasm.^32-34

Transcription of the HNRPA2B1 gene generates a primary transcript that is alternatively spliced to generate two mRNAs. One of these mRNAs encodes the 36 kDa, hnRNP A2. hnRNP A2 expression levels have also been shown to influence HIV-1 Rev activity and function to export vRNA into the cytoplasm. For example, the disruption of the interaction between hnRNP A2 with its cognate binding sequence in the HIV-1 RNA, the hnRNP A2 response element (A2RE) leads to nuclear retention of the vRNA despite the expression of Rev.³⁵ This result suggested that hnRNP A2’s association to the vRNA through the A2RE plays a permissive role in enabling Rev’s overriding function in mediating nuclear RNA export of unspliced or incompletely spliced viral mRNAs. The activity of hnRNP A2/B1 was confirmed by demonstrating that efficient depletion of hnRNP A2 in these conditions (i.e., Rev^-) enabled the A2RE proviral mutant vRNA to be released into the cytoplasm.³¹ These studies revealed two possible roles for hnRNP A2, one in nuclear retention that relies on its association to the vRNA and another in nucleocytoplasmic trafficking that depends on its expression levels in cells. As such, these data underscore a potentially important association between hnRNP A2 expression levels and Rev function during the expression phases of HIV-1.

Our later work revealed that the depletion of hnRNP A2 by small interfering (si)RNA (and the splice variant hnRNP B1 due to the identical targeted sequence in the mRNAs) from cells resulted in a juxtanuclear accumulation of the vRNA at the microtubule organizing center (MTOC) but had no detectable effects on other aspects of HIV-1 gene expression such as splicing or translation^31,35 although this was met by a modest increase in vRNA encapsidation and lowered virus production.^31,36 Expression levels of the major structural protein, 55 kDa precursor group-specific antigen or pr55^Gag that is encoded by the vRNA were not detectably modified when hnRNP A2/B1 were depleted.³¹

In this work, we characterize additional roles for hnRNP A2/B1 in HIV-1 vRNA biogenesis (hnRNP A2/B1 is used herein since we target both isoforms with the siRNA). Expression of a Rev-negative provirus (Rev^-) resulted in predominant nuclear localization of the vRNA. However, when hnRNP A2/B1 were depleted from cells in these conditions, the vRNA translocated to the cytoplasm. The vRNA was found almost exclusively in the cytoplasm, yet pr55^Gag expression remained undetectable. Thus, in the absence of Rev, hnRNP A2/B1 expression is necessary to maintain the nuclear localization of vRNA. The cytoplasmic accumulation of the vRNA was reflected by its presence in a dense polyribosome fraction. This fraction contained resident RNA trafficking granule proteins such as Staufen1, PABP1, eEF1α, ribosomal protein L7,^33,37-39 was virtually devoid of the translation initiation factor eIF4G1 and was severely compromised in translational activity. Our data support the notion that hnRNP A2/B1 expression has an important accessory role to play with Rev in promoting nuclear RNA retention and nucleocytoplasmic RNA transport. Moreover, we demonstrate that hnRNP A2/B1 expression promotes the recruitment of HIV-1 RNAs from a translationally deficient polyribosomal fraction that could impact on viral assembly and vRNA encapsidation.

Results

hnRNP A2/B1 depletion promotes cytoplasmic localization of vRNA in Rev^--expressing cells

In cases when the A2RE is mutated, vRNA is retained in the nucleus despite the expression of Rev, while in cells in which hnRNP A2/B1 is depleted in cells, the vRNA is localized to the cytosol.³¹ To explore this interaction further, we examined the localization of the HIV-1 vRNA in cells depleted of hnRNP A2/B1 during the expression a Rev- provirus that results in the retention of vRNA in the nucleus. Cells were mock transfected or transfected with provirus, HxBRU (Rev+ provirus) or a Rev- provirus [pcMRev(-)]. Cells were treated with either a non-silencing control siRNA (siNS) or one that results in the depletion of hnRNP A2 (and hnRNP B1) from cells (siA2/B1). Gels were run long enough to distinctly separate hnRNPs and their isoforms on western blots. In mock- and siNS-treated cells, the expression of hnRNP A1 (34 kDa, and its isoform hnRNP A1B, 39 kDa) and hnRNP A2 (36 kDa, and its isoform hnRNP B1, 38 kDa) is readily detectable by western blotting using a pan-specific antibody that recognizes a common epitope in all of these hnRNPs (Fig. 1A). The depletion of hnRNP A2, as well as hnRNP B1 by siA2/B1, was consistently as high as 90–97% as shown in this blot related to the loading control protein, γ-tubulin. When we examined the localization of vRNA by fluorescence in situ hybridization (FISH) as described in Materials and Methods, siNS-treated cells showed a punctate distribution of vRNA and siA2/B1 treatment resulted in a phenotype identical to that observed previously with the vRNA accumulating at a juxtanuclear position, identified as the MTOC in our earlier work with Gag distributed throughout the cell and at the periphery in wild-type, Rev+-expressing cells (Fig. S1A; ref. 31). In Rev^--expressing cells, Gag expression was undetectable by both immunofluorescence and western blotting analyses but Gag expression was readily rescued by supplying Rev in trans using the expression vector pCMV-Rev (ref. 40 [not shown; Fig. S1B]). In the context of pcMRev(-) expression, cells were co-stained for hnRNP A2/B1 using a specific rabbit anti-hnRNP A2/B1 antiserum (Act-2) by immunofluorescence and vRNA by FISH. In Rev^--expressing and siNS-treated cells (Rev^- + siNS), all cells abundantly expressed hnRNP A2/B1 and the vRNA was predominantly found in the nucleus as expected in HIV-1-expressing cells (white arrowheads in Fig. 1B). In siA2/B1-treated and Rev^--expressing cells (Rev^- + siA2/B1), hnRNP A2/B1-depleted cells were identified by those in which red fluorescence was faint or undetectable (yellow arrowheads in Fig. 1C). Surprisingly, in hnRNP A2/B1-depleted cells, the localization of vRNA was principally cytoplasmic, whereas cells in which a detectable signal for hnRNP A2/B1 was found exhibited the characteristic nuclear localization of vRNA (white arrowhead in upper portion of panels of Fig. 1C). There was a tight correlation between cells that were depleted of hnRNP A2/B1 and the cytoplasmic localization of vRNA in four independently performed experiments (n = 110 cells, 86 ± 5%, S.D.). A minor proportion of cells exhibited intermediate phenotypes indicated by low but detectable hnRNP A2/B1 expression levels and nuclear and cytoplasmic localization of vRNA. We then performed cell fractionation studies to confirm the changes in the localization of vRNA in siA2/B1 conditions. Cells were transfected with pcMRev(-) in siNS or siA2/B1 followed by nuclear and cytoplasmic fractionation. We used hnRNP A2/B1 and GAPDH as nuclear (principally) and cytoplasmic marker proteins in this analysis. hnRNPs were found predominantly in the nuclear fraction and GAPDH was found in the cytoplasmic fraction demonstrating efficient biochemical separation of these compartments (Fig. 1D). In siA2/B1 conditions, hnRNP A2/B1 were noticeably reduced by 80% in both compartments due to siRNA treatment (Fig. 1D). When we assessed the distribution of HIV-1 vRNA in these fractions, this revealed that whereas it was predominantly nuclear in Rev- conditions as expected (90%, average from two experiments), there was a 5–6-fold enhancement in vRNA abundance in the cytoplasmic fraction (Fig. 1D). This result correlated well with our in situ data shown above (Fig. 1C) and demonstrates that hnRNP A2/B1 depletion promotes nuclear export of retained vRNA. gapdh mRNA fractionated almost exclusively to the cytoplasmic fraction in both conditions (Fig. 1D).

Figure 1. Depletion of hnRNP A2/B1 leads to cytoplasmic localization of HIV-1 vRNA in Rev- conditions. HeLa cells were mock transfected or transfected with nonsilencing siRNAs (siNS) or siRNA to deplete hnRNP A2/B1 (siA2/B1) in cells. HxBru (Rev+) or pcMRev(-) (Rev-) proviral DNAs were then transfected. (A) The expression levels of hnRNP A1, A2, B1, and A1B and γ-tubulin (as loading control) were determined by western blot analysis. The distribution of hnRNP A2 (and hnRNP B1 due to the common epitope recognized in these protein isoforms; red fluorescence) and HIV-1 vRNA (green fluorescence) in Rev^--expressing cells (siNS or siA2/B1 conditions) is shown and was determined by immunofluorescence using an hnRNP A2/B1-specific antiserum (Act-2) and FISH using a pol-specific digoxigenin-labeled RNA probe. All siNS-treated cells in (B) express hnRNP A2/B1 abundantly. White arrowheads in (B and C) identify cells in the vRNA localizes in the nucleus due to a lack of Rev expression. (C) shows cells treated with siA2/B1. Yellow arrowheads identify hnRNP A2/B1-depleted cells with noticeable reduction in hnRNP A2/B1 staining (red). White bar = 10 µm, Cells were fractionated into nuclear (NUC) and cytoplasmic (CYT) fractions following transfection with pcMRev(-) and siNS or siA2/B1 (D). Extracts from each of these fractions were analyzed for GAPDH and hnRNP A2/B1 content as well as for gapdh and genomic mRNAs. The average determination for the abundance of vRNA (presented as % RNA in each fraction) in each subcellular fraction is shown from two experiments. This figure is supplemented by Figure S1 in supplemental material.

hnRNP A2/B1 depletion promotes the accumulation of HIV-1 RNA in a dense polyribosome population

In earlier work, Rev was shown to enhance polyribosomal loading of HIV-1 RNAs to enhance Gag expression.¹⁷ Consistently, Gag expression was not detectable in either immunofluorescence or western blotting experiments when Rev was not expressed (ref. 35 and Fig. S1B). Nevertheless, we were interested in determining the state of translational readiness of the vRNA since it was now found in the cytoplasm in cells depleted of hnRNP A2/B1 in Rev^- conditions. Importantly, because Gag was not expressed despite cytoplasmic localization of its cognate mRNA, we suspected that HIV-1 vRNA must also be found as a cytoplasmic RNP complex that is translationally incompetent/silent in Rev^- conditions. Polyribosome profile analysis has been used to identify this type of translationally inactive, cytoplasmic RNP in cells that were shown to sediment as dense RNP complexes.⁴¹ Therefore, cells were treated with siNS or siA2/B1 and transfected with pcMRev(-) as described in Materials and Methods. At 30–36 h, cells were harvested for polyribosome profile analyses exactly as described.⁴² Ten fractions from the bottom of each gradient were collected, fraction 1 being the most dense and fraction 10 the least. Fractions 7–8 represent the monosome sedimentation peak as determined by spectrophotometry (M, in Figs. 2 and 3) or by the sedimentation of ribosomal protein L7 in small to larger polyribosomes (see Fig. 4). RNA was isolated from each fraction and separated on denaturing agarose gels and transferred to nylon membranes for northern blotting. HIV-1 RNAs were identified using a radiolabelled TAR cDNA probe to identify all three RNA species (2, 4, and 9 kb; Fig. 2B) and gapdh mRNA was identified as described before (Fig. 2C; ref. 43). In Rev^- + siNS cells, the vRNA was predominantly found in the nucleus and little is detectable in the cytoplasm by our in situ methods (Fig. 1B and C) and only about 10% is found to be cytoplasm using biochemical fractionation techniques (Fig. 1D). Importantly, RRE-containing HIV-1 RNAs have not been shown to be strictly dependent on Rev for nuclear export and are indeed found in the cytoplasm in undefined complexes.¹⁶ They are nevertheless dependent on Rev for efficient expression (Fig. S1). This early observation is also corroborated by the weak signal obtained for the 9 kb vRNA that was found in the least dense fraction of the polyribosome gradient (Fig. 2, left panels). This likely represents a soluble, non-polyribosome bound cytoplasmic mRNP fraction consistent with the undetectable Gag expression levels in these cells. In all experiments (n = 6), the RRE-containing 4 kb as well as the fully spliced 2 kb RNAs were predominantly found in fractions 1–6, within polyribosomal populations (Fig. 2B). When we depleted hnRNP A2/B1 from these cells (Fig. 2, right panels), the cytoplasmic, resident vRNA shifted from fraction 10 to the most dense fraction of the gradient (fraction 1) as shown in a long exposure of the northern blot, although this did not result in any change in Gag expression levels (Fig. S1B). These results suggest that hnRNP A2/B1 expression levels in cells dictate the compartmentalization of the vRNA between a non-polyribosomal compartment (fraction 10) and a distinctly dense, polyribosomal fraction (fraction 1).

Figure 2. siA2/B1 treatment results in the redistribution of vRNA to a dense polyribosomal fraction. Cells were treated with siNS or siA2/B1 and transfected with pcMRev(-) and harvested for polyribosome analyses as described in Materials and Methods. RNA was purified from each gradient fraction and separated on agarose gels followed by northern blotting. (A) Continuous OD₂₅₄ polyribosome profiles are shown. The monosome peak is indicated (M). (B) Northern blotting was performed to identify the three HIV-1 RNA size species (9, 4, and 2 kb). (C) The gapdh mRNA polyribosome profile is shown. Fraction 1 represents the most dense fraction (bottom of gradient) and fraction 10, the least dense, likely containing soluble RNA and proteins and material that has not entered sucrose gradients.

Figure 3. siA2/B1 treatment results in the redistribution of vRNA to a dense polyribosomal fraction in Rev+ conditions. Cells were transfected as described in the legend of Figure 2 except that the proviral DNA, HxBRU (Rev+) replaced pcMRev(-) in transfections. (A) Continuous OD₂₅₄ polyribosome profiles are shown. The monosome peak is indicated (M). (B) Short exposure of northern blots to identify 9, 4, and 2 kb HIV-1 RNA species. A longer exposure of the autoradiogram (OE) is shown below to visualize more clearly the 9 kb vRNA. (C) gapdh mRNA was identified as a loading control as well as a polyribosomal distribution control in these analyses.

Figure 4. siA2/B1 treatment results in the redistribution of several proteins to a dense polyribosomal fraction (DF). Cells were treated as described in the legend to Figure 3. Following sucrose polyribosome gradient fractionation, proteins in all fractions were precipitated with TCA and precipitated proteins were resuspended in SDS-containing buffer. Extracts from each fraction were resolved on SDS-PAGE gels and probed for PABP, Gag (pr55^Gag, pr41^Gag, p25, and p24), γ-tubulin, ribosomal protein L7, eIF1α, and CRM1. The location of the monosome peak (M) is indicated on top.

Our data demonstrate that the vRNA is translated when cells are depleted of hnRNP A2/B1 with no apparent changes in Gag synthesis levels during the expression of a Rev+ proviral DNA. In siA2/B1-knockdown conditions, the vRNA also localizes to the MTOC.³¹ We therefore performed polyribosome profile analysis in these conditions. The OD₂₅₄ profiles were identical in Rev+ and Rev^--expressing cells (Fig. 3A). The distributions of the 4 and 2 kb HIV-1 RNAs were similar between Rev+ and Rev- condition with the bulk of these RNA species sedimenting in fractions 1–6 (Fig. 3B). The distribution of the vRNA, while not clearly visible in the exposure shown in the top panel in Figure 3B, was found throughout the gradient in siNS-treated cells (see overexposed [OE] image in bottom inset). gapdh mRNA was found primarily in the polyribosomal fractions 1–6 and did not markedly move with siRNA treatment. A population of vRNA was indeed found in the dense fraction in this gradient, and a smaller proportion was found in the least dense fraction 10. In siA2/B1-treated cells, the distribution of 4 and 2 kb HIV-1 RNAs and gapdh mRNA were nearly identical. However, again there was a striking accumulation of vRNA in the densest polyribosome fraction 1 in hnRNP A2/B1-depleted cells, similar to what we observed in the Rev^- conditions (Fig. 2B). In this case, there was a 2.8 (n = 4, ± 0.4, S.D.)-fold increase in the proportion of vRNA found in fraction 1 when related to the total RNA signal from all fractions in siA2/B1 conditions (Fig. 3B). Thus, a population of vRNA is shunted into this dense polyribosomal population when hnRNP A2/B1 is depleted from cells revealing that at steady-state, hnRNP A2/B1 impacts on the polyribosomal distribution of vRNA. In Rev- conditions, this appears to occur via the recruitment of vRNA from a soluble compartment (noticeable at the top of the sucrose gradients, fraction 10, Fig. 2B). Treatment of cells with puromycin to release ribosomes from polyribosomes or of cell extracts with EDTA prior to loading led to the expected accumulated material in monosomes and RNPs, respectively as determined by spectrophotometry (Fig. S2). However, whereas puromycin led to the sequestration of the HIV-1 vRNA into translationally silent stress granules,⁴⁴ the vRNA was found to be refractory to EDTA treatment (ref. 45; and Valiente-Echeverria et al., submitted). The vRNA found in the DF is therefore polyribosomal and can be dissociated with puromycin but not entirely with EDTA. For experiments performed in the presence of Rev, there was also a loss of vRNA from fraction 10 and a concomitant increase in the relative level of vRNA into fraction 1 (Fig. 3B, OE). Because the distribution of RNA is widespread in the gradient, it is not possible to determine from where RNA is being recruited into fraction 1, although there is quantitatively less vRNA in fraction 10 (Fig. 3B, OE).

siA2/B1 depletion causes a redistribution of RNA-binding proteins

In our attempts to define the dense fraction 1 in which vRNA accumulates under siA2/B1 conditions, we performed polyribosome profile analysis again but examined protein content in each fraction. Following sedimentation in sucrose gradients, each polyribosome fraction was precipitated using TCA as described in Materials and Methods. Following resuspension, proteins were loaded on SDS-PAGE gels for western blotting. Gag proteins, except p25 or p24, were found in the top 5–6 fractions whereas p24 was found primarily in the soluble fractions 9 and 10. We posited that the dense fraction 1 was a translationally incompetent/silent complex whose composition would be reflected by the presence of multiple RNA-binding proteins contained in this type of complex.^1,37,46 We therefore probed for PABP, Tubulin, L7, and eEF1α, most of which are components of typical RNA trafficking granules that are considered to be translationally silent until RNA is trafficked to the right destination in cells.⁴⁷ When we assessed their sedimentation pattern in polyribosome gradients (Fig. 4A and B), we noted that most of these proteins had wide sedimentation patterns, indicating that they are either ribosome/polyribosome-bound proteins, sediment in complexes of similar density, and/or are involved in various aspects of ribosome function and translation. Ribosomal protein L7 and this protein was found within monosomes and adjacent to the monosome peak likely in small polyribosomes of the gradient (peak at fraction 6), as shown earlier.⁴⁸ However, while many of these proteins sedimented throughout the gradients, hnRNP A2/B1 depletion also led to the appearance of several of these proteins in the dense fraction 1 (Fig. 4B). In fact, pr55^Gag (but not any of the smaller processed or mature forms of Gag), PABP, Tubulin, L7, eIF1α, as well as CRM1 accumulated in dense fraction 1. This recruitment corresponded to the marked shift of vRNA to this fraction in Rev^- and Rev⁺ conditions. Thus, hnRNP A2/B1 depletion favored the accumulation of several proteins in the dense fraction 1. CRM1 was probed because it has been found associated with Rev-mediated nucleocytoplasmic transport as well, it has been localized to the MTOC region,⁴⁹ a region at which RNAs accumulate in siA2/B1 conditions, with and without Rev (Fig. 1 and ref. 31). PABP has also been found in translationally silent RNA trafficking granules and intracellular stress granules.^38,44,46

The dense polyribosomal fraction 1 contains poorly translated mRNAs

In order to investigate if the fraction 1 represented a translationally silent, dense polyribosomal fraction/RNP complex, we blotted for eIF4G1, a critical translation initiation factor that bridges 5′ and 3′ regions of the RNA to favor translational initiation.⁵⁰ This protein was found in the polyribosome gradients in all conditions tested in this manuscript and was distributed between fraction 4–10 (Fig. 5A). We never observed eIF4G1 in the dense fraction 1 in any condition tested indicating that translation initiation is limited or absent on RNAs found in this fraction. We confirmed this result by performing acute metabolic labeling of cells immediately prior to harvesting cells for polyribosome analysis. Cells were pulsed with radiolabeled amino acids for 7 min, harvested, and polyribosome profile analysis was performed. The distribution of TCA-precipitable radiolabeled polypeptides were counted in an aliquot from each fraction of the gradient. A peak of amino acid incorporation was found in fractions 4–6 in each gradient and there was little difference found in Rev^-/Rev⁺, siNS, or siA2/B1 conditions (Fig. 5B). Most importantly is the low levels of TCA-precipitable counts in the dense polyribosome fractions 1 and 2, representing 7–12% TCA precipitable counts, demonstrating that these dense fractions are translationally deficient, even if they contain PABP, translation factors, ribosomal proteins, and mRNAs (Fig. 4). The lack of eIF4G1 and amino acid incorporation in this dense polyribosomal fraction 1 both demonstrate that HIV-1 vRNA and other RNAs found in this dense fraction are not actively translated, consistent with earlier observations.⁴¹

Figure 5. The dense fraction 1 (DF) contains little if any eIF4G and contains translationally silent RNAs. Cells were transfected with siNS or siA2/B1 and HxBRU (Rev+) and polyribosome analyses were performed. Proteins were precipitated from gradient fractions as described above and blotted for eIF4G. The dense fraction 10 is indicated by DF; the monosome peak is identified with M on top. Lane C represents 5–10% input control before sucrose gradient centrifugation. (B) Seven minutes before cell harvesting, cells were pulsed with radiolabelled amino acids. Polyribosome profile analysis was performed as described above. An aliquot from each gradient fraction was TCA precipitated and radiolabelled polypeptides were collected on glass fiber filters and counted. Radioactivity was plotted as the percentage of radioactivity in each fraction related to the total. The background was corrected as described in Materials and Methods. The inset on right shows expression levels for hnRNP A2/B1 (identified using a mouse monoclonal anti-hnRNP A2/B1), pr55^Gag and GAPDH (as loading control) in siNS- and siA2/B1-treated cells.

Staufen1—but not hnRNP—is found in dense polyribosome gradient fraction

The polyribosome fraction to which vRNA is recruited when hnRNP A2/B1 is depleted represents not only a dense RNP as determined by sucrose gradient analyses, but is also deficient in translational activity. Because these are both characteristics of RNA trafficking granules, we attempted to identify if any of the classical RNA trafficking granule proteins, including the hnRNPs and Staufen1, were present. Thus, cells were transfected with siNS or siA2/B1 in Rev+ or Rev- conditions and cells were harvested for polyribosome profile analyses. Protein extracts were prepared from each gradient fraction as described earlier and loaded onto SDS-PAGE gels for Western analysis. We blotted for hnRNPs and Staufen1 using antisera that recognize several isoforms of each (A1, A2, B1, A1B, Staufen1^{55 kDa}, and Staufen1^{63 kDa}). In Rev⁺ and Rev^- conditions (Fig. 6A and B), hnRNPs A1, A2, B1, and AIB were found to be distributed mostly in fractions 5–10, hnRNP A1 and hnRNP A2 were the most readily detectable proteins using our pan-specific antibody. In siA2/B1 conditions in both Rev- and Rev+ conditions, a noticeable gap between hnRNP A1 and A1B protein signals was observed (compare with Fig. 1A), again demonstrating effective depletions of hnRNP A2 and hnRNP B1 proteins in both conditions. siA2/B1 treatment did not appear to significantly modify the sedimentation of the remaining hnRNP proteins in these gradients. However, we did detect the appearance of hnRNP A1 in more dense fractions of the gradient, likely due to the reciprocal increase in the abundance of hnRNP A1 in these siA2/B1 knockdown conditions as noted in earlier work.^31,51

Figure 6. The distribution of hnRNPs and Staufen1 in polyribosome gradients. Cells were transfected with siNS or siA2/B1 and HxBRU (REV+) or pcMRev(-) (REV-) and polyribosome profile analyses were performed. Proteins were precipitated from gradient fractions as described above and hnRNPs (A2, B1, A1, and AIB) and Staufen1 (55 kDa and 63 kDa isoforms) were identified in western blots using a rabbit pan-specific and mono-specific antisera, respectively (identified on right of each autoradiogram). The dense fraction 1 is indicated by DF; the location of the monosome peak is identified with M on top. Lane C represents 5–10% input control before sucrose gradient centrifugation.

We then assessed the abundance of Staufen1 in polyribosomal gradient fractions because Staufen was found to be a major component of translationally silent complexes in neuronal cells.⁴¹ The smaller, more abundant isoform of Staufen1 (Staufen1^{55 kDa}) was found in virtually all gradient fractions, including dense fraction 1, peaking at fractions 8/9, coincident with ribosomes, consistent with earlier reports^52,53 (Fig. 6A and B). The larger, less abundant Staufen1 isoform, Staufen1^{63 kDa}, appeared only in the fractions 4/5–10, and tapered off toward the denser gradient fractions, in all conditions. The distribution of Staufen1 was unaffected by siA2/B1 treatment, however. While we now understand that there are differential roles for Staufen1 isoforms during HIV-1 replication,^44,54 as well as for other Staufen proteins,^55-57 only the Staufen1^{55 kDa} isoform sedimented to the dense fraction 1 in the polyribosome gradients. The presence of Staufen1^{55 kDa} in the dense fractions, in contrast to what we observed for the Staufen1^{63 kDa} isoform, implicates the lower molecular isoform in HIV-1 RNP biogenesis. Indeed, this would be consistent with the roles for this isoform in HIV-1 assembly as well as in vRNA packaging that we have defined earlier.^33,44 It is possible that Staufen1^{55 kDa} governs these processes in the context of a Staufen1^{55 kDa} containing and dense RNP that only functions during limited translational activity, as proposed in earlier work.^41,44,64

Discussion

The results presented in this manuscript contribute to a more complete understanding on how HIV-1 uses hnRNP A2/B1 during viral replication. Even though vRNA retention is believed to depend on threshold levels of Rev at sites of proviral DNA transcription during the late expression phase of HIV-1,¹⁴ hnRNP A2/B1 appears to play a role in this event in two experimental situations. In the first, if the association between hnRNP A2/B1 is blocked by mutating a cognate hnRNP A2 (/B1)-response element (A2RE) found at the N terminus of the vpr coding region, vRNA is partially retained in the nucleus, similar to a Rev- phenotype³⁵ suggesting a functional importance for the hnRNP A2/B1-vRNA association. The second scenario is presented in this manuscript in which we show that when hnRNP A2/B1 are depleted, nuclear retained vRNA in Rev- conditions is released to the cytoplasm (Fig. 1). This could be achieved by a loss of a direct vRNA interaction and/or by the deficiency of a Rev-interacting partner.²⁷ While we have proposed that the hnRNP A2/B1-A2RE association is important for both nuclear and cytoplasmic vRNA trafficking,³⁵ hnRNP A2/B1 depletion could also indirectly affect the expression or localization of critical nuclear vRNA retention factors, a scenario that will deserve further study (Zolotukhin et al. 2003). In light of the well-characterized roles in post-transcriptional regulation, these results promote the notion that hnRNP A2/B1’s several functions in the control of vRNA fate are coupled.^1,58 Furthermore, this work demonstrates that hnRNP A2/B1’s functional role extends into the cytoplasm in which hnRNP A2/B1 are important for the recruitment of the HIV-1 vRNA from cytoplasmic RNPs such that when hnRNP A2/B1 are depleted from cells, there is an accumulation of vRNA in dense, cytoplasmic polyribosomal fractions in both Rev^- and Rev⁺ conditions that correspond to translation-deficient RNP complexes or RNA trafficking granules.^37,41,59 These complexes bear many of the hallmarks of RNA trafficking granules because they sediment to dense fractions in sucrose gradients,⁴¹ they are translationally silent and they contain classical RNA trafficking granule components.^37,41,59,60 It is noteworthy that many of the components of these granules have functions during HIV-1 gene expression stages or are found associated in purified virus particles,^1,33 suggesting that these types of granules contribute to viral assembly.

The recruitment of the vRNA from translationally deficient RNPs might imply that hnRNP A2/B1 have effects on gag mRNA (vRNA) translation and Gag synthesis consistent with the reported activities of hnRNP A2 on translation enhancement.^61,62 However, hnRNP A2/B1 depletion does not lead to detectable changes to steady-state Gag levels as shown earlier^31,35 and in this manuscript (see inset in Fig. 5B; Fig. S1). The shift of a proportion of vRNA in polyribosomes may have a more subtle function than we appreciated in our earlier work, however. Because only a fraction of the total cellular pool of vRNA has been shown to be encapsidated into progeny virions⁶³ and translational silencing of the vRNA was suggested to be a signal immediately prior to the encapsidation of vRNA into new virus particles,⁶⁴ a polyribosomal shift of vRNA to a translationally silent pool may be critical in viral assembly in the selection of vRNA for encapsidation. Consistently, depleting hnRNP A2/B1 not only leads to MTOC localization of the vRNA but could explain why a modest increase in vRNA encapsidation is also observed.³¹ Our results support such a model in the control of vRNA encapsidation that involves subtle changes to the composition and trafficking of the HIV-1 RNP. hnRNP A2/B1 is a common component of several relevant HIV-1 complexes that have been recently characterized at the proteomics level^9,33,65 and these hnRNPs likely exert functions during the transit of HIV-1 complexes from the nucleus to and within the cytoplasm, perhaps acting in concert with Rev at multiple levels.

Polyribosomal dynamics were explored for HIV-1 RNAs in earlier work in which Rev was shown to be a principal player in promoting the loading of most, but not all HIV-1 transcripts on polyribosomes.¹⁶ The results shown here highlight the fact that the cytoplasmic location of HIV-1 RNAs (at least that of gag mRNA) is insufficient to promote translation and expression of Gag and support important earlier findings that Rev expression is essential for viral structural protein synthesis in the cytoplasm. The nature of the requirement for Rev expression likely stems in part from a highly structured 5′UTR, which is a strong impediment to translation initiation factors and ribosomes.⁶⁶ Rev could interact with the RNA packaging signal⁶⁷ to modify the conformation of this region and the 5′-end of HIV-1 RNA so that it becomes more readily amenable to the cellular translation machinery leading to ribosome loading on HIV-1 RNAs.^15,68,69 These types of conformational switches are important for HIV-1 and MLV, for example, in the encapsidation process.^70-72 Furthermore, one major finding suggested that a fraction of the gag mRNA (vRNA) is retained in distinct cytoplasmic compartments in both the absence and presence of Rev.¹⁷ Our results confirm this finding and also reveal how cellular proteins, like hnRNP A2/B1, influence the dynamic formation of these HIV-1 RNA pools by influencing tightly regulated nuclear and cytoplasmic trafficking events. hnRNP A2/B1’s influence on the polyribosomal distribution of HIV-1 RNA also denotes an important cytoplasmic role for hnRNP A2/B1 in RNA trafficking, viral gene expression, assembly, and encapsidation.

Materials and Methods

Cell culture

HeLa cells were cultured at 37 °C in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and containing 1% penicillin/streptomycin. Sixteen hours prior to transfection, cells were trypsinized, counted, and replated. For sucrose gradient experiments, cells were seeded in 150 cm² flasks or 6-well plates at a density of 3.5 × 10⁴ cells per cm². The quantity of cells was doubled for some experiments in order to enhance the detection of endogenous proteins in gradient fractions.

For FISH/IF co-analyses experiments, cells were seeded in 6-well plates (NUNC) containing autoclaved glass coverslips at a density of 1.5 × 10⁴ cells per cm². Cells collected for western blotting in FISH/IF experiments were plated at a density of 3.5 × 10⁴ cell per cm². These cells were collected for western blotting in order to determine the efficiency of siRNA-mediated gene silencing.

Transfections and siRNAs

Transfections were performed using Lipofectamine 2000 (Invitrogen) in OptiMEM serum-reduced medium (Invitrogen), according to the manufacturer’s protocol. Knockdown of hnRNP A2/B1 was achieved using a double transfection method. On day 1, cells were transfected with either non-silencing (siNS) 5′ AATTCTCCGA ACGTGTCACG A or hnRNP A2/B1-specific (siA2/B1) 5′ AAGCTTTGAA ACACAGAAGA siRNA duplexes at 25 nM (Qiagen-Xeragon). Twenty-four hours later, cells were transfected a second time with both siRNA and proviral DNA. The proviral DNA construct used in these experiments was HxBru.^31,35 The Rev-minus proviral construct, pcMRev(-) was used as described previously (kindly provided via the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, from Dr Maria Sadaie^35,73,74). In Rev rescue experiments, pCMV-Rev was used at 0.5 µg per transfection⁴⁰ (kindly provided via the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, [NIH] from Dr Marie-Louise Hammarskjöld and Dr David Rekosh⁴⁰). Mock-transfected cells were transfected with the empty vector pcDNA3 and siNS (25 nM). Four hours after each transfection, the culture medium, containing Lipofectamine and OptiMEM, was removed and replaced with fresh DMEM/FBS without antibiotics. hnRNP A2/B1 knockdowns were effective as of 8 h post-second transfection as described,³¹ and cells were harvested or fixed at time points as indicated in the text. Knockdown efficiencies were calculated by quantitation of the autoradiographic signals obtained in western blotting with the Molecular Analyst software (Bio-Rad) or using ImageJ software from the National Institutes of Health as described.³⁵

Fluorescence in situ hybridization/immunofluorescence (FISH/IF) co-analyses

Procedural details on FISH/IF co-analyses are completely described elsewhere.⁷⁵ Briefly, at 30 h post-transfection, the medium was carefully removed and wells containing coverslips were washed twice with PBS. They were then incubated in 4% paraformaldehyde for 20 min at room temperature. Fixed cells were then incubated for 10 min at room temperature with gentle agitation in a solution of 0.1 M glycine in PBS. This step ensures that any remaining aldehyde groups are quenched. Lastly, cells were permeabilized in a 0.2% solution of Triton-X-100 in PBS for 10 min at room temperature, also with gentle agitation. Between fixation, quenching, and permeabilization steps, coverslips were gently washed with PBS. Coverslips were stored in 70% ethanol in sealed plates at 4 °C until examined. FISH/IF and microscopy were performed exactly as described using an Olympus BX-51 fluorescence microscope equipped with an UPlanFI 100X oil objective. Images were captured in black and white with the Spot camera (Diagnostics Instruments) using Spot Advanced Software and Image-Pro-Plus version 4.0.1. Images were pseudo-colored using Adobe Photoshop CS5 (Adobe Systems) in RGB mode and then merged as described.^31,35

Polyribosome profile analyses

Polyribosome profile analysis experiments were performed with HeLa cells and were performed exactly as described.⁴² Thirty to 36 h post-transfection, cells were washed twice and collected in PBS, then spun for 10 min at 1 000 × g at 4 °C to pellet the cells. Supernatant was removed and the cells were resuspended in Polysome Buffer (250 mM KCl, 10 mM MgCl₂, 20 mM HEPES [pH 7.5], 0.25 M sucrose, 0.1 mM DTT, 150 μg/mL cycloheximide, 1μL/mL RNase Out (Invitrogen), and 0.5% NP-40). Cells were homogenized mechanically using a sterile Eppendorf pestle and spun at 500 × g for 5 min at 4 °C, in order to pellet the nuclei. The supernatant was then removed and added to a 20% solution of sodium deoxycholic acid in order to liberate membrane-associated proteins in the cell homogenate. Following this, post-mitochondrial supernatants were prepared by centrifugation for 25 min at 16 000 × g (4 °C). Post-mitochondrial supernatants were then transferred to clean Eppendorf tubes, snap frozen in liquid nitrogen and stored at -80 °C. Prior to freezing, an aliquot was removed from each sample for quantitation by spectrophotometry and for western blot analysis to ascertain the efficiency of the knockdown. Continuous sucrose density gradients (10% to 50% wt/vol) were prepared in buffer containing 500 mM KCl, 10 mM MgCl₂, 20 mM HEPES (pH 7.5), 2 mM dithiothreitol, 150 μg/mL cycloheximide, and 10 U/mL sodium heparin. Gradients were prepared in 5 mL polyallomer tubes by gently layering 2.2 mL of 10% sucrose in buffer over 2.2 mL of 50% sucrose in buffer. Tubes were then sealed, turned on their sides and left to equilibrate overnight at 4 °C. The next day, 600 μL of sample lysate (containing equal quantities of material as normalized by spectrophotometry, λ = 260 nm) was layered gently on to the gradients and ultracentrifuged in a Beckman Ti55 swing rotor at 83 000 × g for 6 h (4 °C). Continuous OD₂₅₄ readings for gradients were read from the bottom by piercing and fractions were collected using an ISCO fractionator (Teledyne, ISCO). Ten fractions of 0.5 mL were collected from each gradient. In some experiments, a 250 μL aliquot was taken for RNA analysis while the remaining 250 μL was used for protein analysis. The ribosomal protein L7 was used in western blotting analyses in some experiments and this protein sediments in monosomes and in small polyribosomes, as shown.⁴⁸

RNA precipitation and northern blotting

RNA was purified from sucrose density gradient fractions exactly as described before using proteinase K digestion, phenol-chloroform extraction and ethanol precipitation.⁴² Alternatively, TRIzol® LS (Invitrogen) was employed to isolate RNA from the collected fractions according to the manufacturer’s protocol. In both procedures, 5 μg of glycogen (Roche Diagnostics, ON) was used as a carrier in each sample during ethanol precipitation to enhance the yield of RNA. RNA pellets were resuspended in 10 μL of RNase-free water and denatured for 30 min at 65 °C in a solution of 15% sample, 50% formamide, 7.4% formaldehyde, and 5% ethidium bromide in MOPS buffer (20 mM 3-[N-morpholino] propanesulfonic acid, 2 mM sodium acetate, 2 mM EDTA). Samples were then resolved by electrophoresis in a denaturing 0.8% agarose gel containing 1.85% formaldehyde in 1X MOPS buffer. RNA was transferred overnight by capillary action to a Biodyne® B membrane (Pall Corporation) in 20X SSC (3 M NaCl, 300 mM sodium citrate, pH 7.0), then covalently linked using UV radiation. Membranes were hybridized overnight at 65 °C in Church’s Buffer (250 mM sodium phosphate [pH 7.0], 7% SDS, 1 mM EDTA [pH 8.0], 1% bovine serum albumin), and developed by autoradiography. A ³²P-labeled cDNA probe specific to the TAR region (this probe recognizes all HIV-1 transcripts) and gapdh mRNA were used as described previously.^43,76

For protein content analysis in each fraction, following sedimentation in sucrose gradients, each fraction was mixed with 0.1 volumes of 70% trichloroacetic acid (TCA), and stored on ice for 2 h in order to precipitate the protein components from the sucrose solution in each fraction. After this, samples were spun at 16 000 × g for 15 min (4 °C) and protein pellets were washed twice in ice cold acetone, then dried at room temperature. Dried pellets were resuspended in a solution of 5% SDS in PBS 1X, and stored at -20 °C. Resolution of the protein components in each fraction was accomplished using SDS PAGE (SDS-PAGE). Stacking gels were 4% polyacrylamide while the resolving gel was 12% acrylamide. Proteins were transferred to a nitrocellulose membrane (Pall Corporation) and blocked with 10% fat-free milk for 2 h.

Immunoblotting was performed with the following antibodies: rabbit anti-p24 (1:5000, Trinity Biotech); pan-specific rabbit polyclonal anti-hnRNP, recognizing hnRNP A1, A1b, A2, and B2 (1:5000) generously provided by Benoit Chabôt (Université de Sherbrooke); rabbit anti-eIF4G1 (1:1000), and rabbit anti-PABP1 (1:1000), both generously provided by Dr Nahum Sonenberg (McGill University); rabbit anti-γ-tubulin (1:5000, Sigma-Aldrich), mouse anti-L7 (1:5000, Novus Biologicals), mouse anti-eEF1α (1:1000), rabbit anti-CRM1 (1:5000, Santa Cruz Biotechnologies), and anti-GAPDH (1:5000, Sigma-Aldrich). Horseradish peroxidase-conjugated goat anti-rabbit (1:5000) and sheep anti-mouse (1:5000) were used as secondary antibodies. Finally, signals were exposed by autoradiography following a 1 min incubation with Western Lightning® Chemiluminescence Reagents, as described by the manufacturer (Perkin-Elmer). Antisera to hnRNP A2 (rabbit anti-hnRNP A2 [Act-2, used for immunofluorescence]), EF67 (mouse anti-hnRNP A2; see below) recognize both hnRNP A2 and hnRNP B1 due to common C-terminal epitopes in these isoforms.³⁵

Cell fractionation analyses

HeLa cells transfected with pcMRev(-) and siNS or siA2/B1 and cells were harvested at 30–36 h post-transfection. Cells were fractionated into nuclear and cytoplasmic fractions. Cells were allowed to swell for 5 min in 400 µL hypotonic buffer (20 mM Tris, 10 mM KCl, 1 mM MgCl2, 1 mM dithiothreitol, 0.5 mM PMSF, 0.2% NP-40) and were then dounce homogenized using an Eppendorf pestle. Nuclei were collected at 500 × g for 5 min at 4 °C and the supernatant was recentrifuged and put on ice (cytoplasmic extract). Nuclei were washed once with NP-40-containing buffer (+5 mM EDTA) and once without NP-40 (+5 mM EDTA) and the wash supernatants were discarded. Nuclei were then lysed in 10 mM Tris-100 mM NaCl-1mM EDTA containing 0.5% NP-40, centrifuged at 14 000 × g to remove chromosomal DNA. Aliquots from nuclear and cytoplasmic extracts were taken for western blot analysis. For RNA extraction, extracts were immediately transferred to a tube containing 400 µL phenol-chloroform-isoamyl alcohol (50:50:1) and 200 µL urea-SDS extraction buffer as described previously.^43,77 RNA was collected by ethanol precipitation using glycogen as a carrier. RNA was slot-blotted onto nylon membranes and then probed using a HIV-1 vRNA-specific probe as described³¹ or one that recognizes gapdh mRNA.⁴³ RNA gel blotting was performed as described above.⁴³

Amino acid incorporation into de novo synthesized polypeptides

Cells were transfected with HxBRU or pcMRev(-) proviral DNA, and siNS or siA2/B1. Cells were washed with pre-warmed PBS and before harvesting cells for polyribosome analyses, cells were pulsed with 1150 µCi Express (S³⁵) Protein labeling mix (Perkin Elmer; 8 mCi/mL) for 7 min in methionine- and cysteine-free DMEM, washed three times with ice-cold PBS, and processed for polyribosome analyses as described above. Following collection of gradient fractions, a 200 µL aliquot was precipitated with 10% TCA, filtered through GF/C Whatman glass fiber filters. Filters were washed extensively with 5% TCA, dried, and radioactivity was counted using a liquid scintillation counter. Background counts in each fraction were derived from mock transfected cells incubated without radiolabelled amino acids. For this set of experiments, a mouse anti-hnRNP A2/B1 monoclonal antibody (EF67) was used to detect hnRNP A2/B1 proteins by western blotting analysis.

Abbreviations:
HIV-1	=	human immunodeficiency virus type 1
hnRNP	=	heterogeneous ribonucleoprotein
RNP	=	ribonucleoprotein
kb	=	kilobasepair
vRNA	=	HIV-1 genomic RNA
RRE	=	Rev-responsive element
eIF	=	eukaryotic initiation factor
siNS	=	non-silencing siRNA
Gag	=	group specific antigen
MTOC	=	microtubule organising center
S.D.	=	standard deviation
PABP	=	polyA-binding protein
CRM1	=	chromosome region maintenance 1

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We especially thank Dr Anne Monette, Melanie Halvorsen, and Martin Lehmann for contributions to experiments reported in this manuscript, Drs Benoit Chabôt, Nahum Sonenberg, Graciella Boccaccio for antibodies, Drs Sadaie, Hammarskjöld, Rekosh, and the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH for reagents and Alan Cochrane for advice on the fractionation assay. AJM was supported by a Canadian Institutes of Health Research (CIHR) New Investigator Award and this work is supported by grants from the Canadian Foundation for AIDS Research and the CIHR (MOP-56974 & MOP-38111).

10.4161/rna.26542

Supplemental Materials

Supplemental materials may be found here: www.landesbioscience.com/journals/rnabiology/article/26542