Results
Evolution of Pre-pVII in siadenoviruses and identification of NLSs within Frog siadenovirus 1 (FrAdV1). Nuclear transport of AdV pVII has been partially characterized for HAdV. Several studies reported that HAdV5 pVII utilizes several different NLSs, enabling transportation to the host cell nucleus via various import pathways, whereas very little is known for other AdVs. While HAdV2 Pre-pVII localized to the nucleolus, we have recently shown that the psittacine siadenovirus F (PsSiAdV) Pre-pVII mainly localized in the nucleoplasm, when expressed in the absence of any other viral proteins, implying important differences in nuclear localization among diverse AdVs47. To gain insights into the evolutionary aspects of pVII among siadenovirus members, we performed a phylogenetic analysis using their primary amino acids sequences and identified FrAdV1 pVII as the most ancient form within the genus Siadenovirus (Fig. 1A). In order to better understand the nuclear transport of FrAdV1 Pre-pVII, we analyzed its primary sequence bioinformatically with cNLSmapper, and by visual inspection for putative nuclear localization signals (NLSs). Such analysis revealed four putative NLSs (NLSa:43GYWRRKR SKK A53, NLSb:74RKK IPK TPVGVVGPLWQGTRKR 95; NLSc:111GRK IKK AR AP120, NLSd: 141PPRKRRR VA149), of which NLSb was identified a putative bipartite NLS (Fig. 1B). FrAdV1 and HAdV5 Pre-pVII share very little sequence identity (26.1%). Accordingly, the identified NLSs have no clear counterparts in HAdV5 pVII (Fig. 1C).
Since the structure of Pre-pVII proteins is yet to be determined, we employed AlphaFold2 to generate models that may provide preliminary insights into their structure (Supplementary Fig. S1). Predictions suggest HAdV5 Pre-pVII to be disordered, with an unstructured N-terminus (comprised of a loose string of five alpha helices, and two beta strands separated by flexible loops) and a disordered C-terminus (Supplementary Fig. S1A-C). AlphaFold2 also predicts FrAdV1 Pre-pVII to be an intrinsically disordered protein with no structured regions (Supplementary Fig. S1D-E). Such disorder would potentially enable exposure of the putative NLS regions, making them accessible to the nuclear import receptors. Such a model would likely be flexible and dynamic and could adopt multiple conformations and be accessible to bind nuclear import receptors.
FrAdV1 Pre-pVII protein strongly accumulates in the nucleolus in the absence of other viral proteins . To investigate the subcellular localization of FrAdV1 Pre-pVII, we transiently expressed a Pre-pVII GFP-fusion protein, and compared its subcellular localization to similar fusions with HAdV2 Pre-pVII and PsSiAdV Pre-pVII, which are known to accumulate into the nucleoli and nucleoplasm, respectively47. As further controls, GFP alone and GFP fused to human cytomegalovirus (HCMV) DNA polymerase processivity factor UL44 were also expressed 48, while DsRed-fibrillarin was used as a marker for nucleolar localization (Fig. 2A). As expected, GFP alone localised with a diffuse pattern distributed between the nuclei and cytoplasm, whereas UL44 strongly accumulated in the cell nucleus, and both proteins were largely excluded from the nucleoli (Fig. 2B). As recently shown, Pre-pVII from PsSiAdV exhibited a localization reminiscent of UL44, while HAdV2 accumulated in the nucleoli and colocalised with fibrillarin, although a considerable fraction of the protein was detectable in the nucleoplasm (Fig. 2B). Intriguingly, FrAdV1 Pre-pVII strongly accumulated in the nucleoli, with limited detection in the nucleoplasm, displaying a different localization pattern to its HAdV2 and PsSiAdV orthologues (Fig. 2B).
FrAdV1 Pre-pVII NLSs can bind to different cellular transporters. Given their ability to interact with multiple IMPs, HAdVs pVIIs are believed to be transported into the nucleus by multiple pathways 10. We therefore assessed the binding of each FrAdV-1 putative NLS to IMPα2 and IMPβ1 using a quantitative fluorescence polarization (FP) assay, commonly used in the field23,25,34,35,47 (Figure 3 and Supplementary Table S3). Interestingly, NLSa interacted with IMPβ1 with high affinity, NLSd interacted with high affinity with IMPα2, while NLSb and NLSc bound with low affinity to both IMPα2 and IMPβ1 (Figure 3 and Supplementary Table S3). These results suggest that FrAdV1 Pre-pVII can interact with different IMPs via distinct NLSs, consistent with its ability to be transported into the nucleus via several nuclear import pathways.
FrAdV1 Pre-pVII NLSs can confer IMPα/β dependent and independent nuclear import to heterologous proteins. Since the FrAdV1 Pre-pVII strongly accumulated in the nucleoli of transfected cells and contains four putative NLSs (Fig. 1BC) which can mediate binding to several IMPs (Fig. 3), we analysed their ability to alter the subcellular localization of a reporter protein. To this end, we analysed the subcellular localization of fusion proteins between GFP and FrAdV1 Pre-VII putative NLSs when transiently expressed in mammalian cells (Fig. 4A), followed by quantitative analysis on nuclear and nucleolar accumulation levels (Fig. 4BC). We also expressed GFP alone and GFP fused to Simian Virus 40 (SV40) large tumour antigen (LTA) NLS, as negative and positive controls for nuclear accumulation respectively. As expected, GFP alone equally distributed between nucleus and cytoplasm (Fig. 4B) with a Fn/c of 1.2 (Fig. 4C); GFP-SV40-NLS strongly accumulated in the cell nucleus (Fig. 4B), with a Fn/c of 6.9 (Fig. 4C); and both proteins were partially excluded from the nucleoli with a Fno/Fn < 1 (Fig. 4D). Importantly, fusion to each FrAdV1 Pre-pVII-NLS affected GFP subcellular localization, although to different extents. The highest levels of nuclear accumulation were observed for GFP-FrAdV1 pVII-NLSd, with a Fn/c of 7.2 (Fig. 4C), and similarly to GFP alone, it was mainly excluded from nucleoli (Fno/n < 1, Fig. 4D). On the other hand, while nuclear accumulation mediated by either NLSa or NLSc was considerably lower as compared to NLSd, both sequences could significantly increase GFP targeting to the nucleolus (Fig. 4D). Finally, NLSb slightly increased GFP nuclear targeting (Fn/c of. c. 2, see Fig. 4C), but not its nucleolar accumulation (Fno/n < 1, see Fig. 4D). Our results suggest that FrAdV1 Pre-pVII contains four sequences which could contribute to its nuclear localization (NLSa-d), and two (NLSa and NLSc) which could contribute to its nucleolar targeting. Since we and others have extensively shown that the peptide inhibitor Bimax2 can abolish IMPα/β dependent nuclear import 21,25,49, we decided to test its effect on the activity of FrAdV1 Pre-pVII NLSs. We therefore tested the effect of co-expression with mCherry-Bimax2 on the nuclear accumulation of the FrAdV1 Pre-pVII NLSs GFP fusion proteins (Supplementary Fig. S2). As expected, expression with Bimax2 completely abolished nuclear targeting of GFP-SV40 NLS (Supplementary Fig. S2), consistent with its ability to bind exclusively to IMPα/β via IMPα. Importantly, Bimax2 significantly reduced nuclear accumulation mediated by NLSd but not by NLSa, NLSb and NLSc (Supplementary Fig. S2B). These results are consistent with FrAdV1 Pre-pVII NLSd being a cNLS functionally interacting with IMPα.
High-resolution crystal structure uncovers the classical binding interface of IMPα FrAdV1 Pre-pVII NLSd. Based on our observation that the FrAdV1 Pre-pVII NLSd bound to with high affinity to IMPα2, we undertook X-ray crystallography of this complex to gain a deeper understanding of the interactions between FrAdV1 Pre-pVII and host cellular transporters. The protein crystallization process was carried out using the hanging-drop vapor diffusion method, resulting in the formation of large rod-shaped crystals within a three-day incubation period (Fig. 5). These crystals diffracted to 2.2 Å at the Australian Synchrotron on the MX2 beamline, and the data were indexed in P212121, with unit cell parameters of a = 77.29, b = 89.29, and c = 95.80. The structure was solved using Phaser 41 through molecular replacement with the model derived from PDB entry 3UKX. The resolved structure identified one chain of FrAdV1 Pre-pVII NLSd (PPRKRRR VA-149) bound to IMPα2ΔIBB (residues 72-498) with a well-characterized lysine at the P2 site. Lys144 is the predominant binding determinant of the major binding site forming hydrogen bonds with Gly150, Thr155, and Asp192 at the P2 binding pocket of IMPα2, with a salt bridge formed between Asp192 of IMPα2 and Lys144 of the NLSd peptide. Arg145 of NLSd interact with the P3 binding site of IMPα2 at Asn188 and Asn288, via hydrogen bonding whereas Arg146 and Arg147 bound at P4 and P5 positions (Fig. 5) 39,40. The full data collection and refinement statistics is given in Supplementary Table S4. Furthermore, when an excessive amount of peptide is employed, NLSd was found at IMP⍺ minor binding site (Fig. 5B). Our results are therefore consistent with NLSd being a cNLS functionally interacting with IMPα major binding site.
FrAdV1 Pre-pVII can accumulate in the nucleolus in a Ran independent process. Since nuclear import of HAdV2 Pre-pVII and mature pVII has been shown to rely on IMPα/β, and IMPβ2, respectively10, we analysed the effect of inhibitors of the IMPα/β and IMPβ2 nuclear import pathways on the subcellular localization of FrAdV1 Pre-pVII. Inhibitors tested include mcherry-Bimax230, which impairs IMPα/β dependent nuclear import, M9M-RFP 29, which inhibits IMPβ2 dependent nuclear import, and DsRed-RanQ69L 31, affecting Ran dependent nuclear import (Fig. 6A). GFP-UL44, which is imported into the nucleus by IMPα/β, was also expressed as a control 48. As previously, in the absence of nuclear import inhibitors, both GFP-UL44 and FrAdV1 Pre-pVII strongly localised to the cell nucleus (Fig. 6B), with a Fn/c > 20 (Fig. 6C). Co-expression with either mCherry-Bimax2 (Fn/c 0.3) or DsRed-RanQ69L (Fn/c 1.1) strongly impaired nuclear accumulation of GFP-UL44, while no significant reduction was observed in the presence of RFP-M9M (Fig. 6C). Surprisingly, GFP-FrAdV1 Pre-pVII nuclear import was not significantly impaired by any inhibitor tested. The lack of response to either Bimax2 or M9M can be interpreted as the consequence of the ability of Pre-pVII to interact with multiple IMPs, while nuclear localization in the presence of RanQ69L could be the consequence of its ability to passively enter the nucleus due to its low molecular weight, followed by interaction with nuclear components.
Molecular dissection of the role of FrAdV1 individual in nuclear and nucleolar targeting. To verify these hypotheses, we dissected the contribution of each NLS identified here to FrAdV1 Pre-VII nuclear and nucleolar targeting. Therefore, we quantitatively analysed the subcellular localization of transiently expressed GFP fusion proteins with FrAdV1 Pre-pVII and several substitution derivatives thereof, whereby NLS basic residues were replaced either by A or T (Fig. 7A), using DsRed-Fibrillarin as a nucleolar marker. Importantly, individual inactivation of single NLSs did not significantly reduce nuclear accumulation (Fig. 7B and C), with barely no signal being detectable in the cytosol of cells expressing GFP-FrAdV1 Pre-pVII wt (Fn/c of 17.7), mNLSa (Fn/c of 14.7), mNLSb (Fn/c of 14.8), mNLSc (Fn/c of 17.9) and mNLSd (Fn/c of 15.69). Simultaneous inactivation of either NLSa and NLSc (mNLAac, Fn/c of 11), NLSa and NLSd, (mNLSad, Fn/c of 12) or NLSc and NLSd (mNLScd, Fn/c of 15) only mildly reduced nuclear localization (Fig. 7C). These results confirm that, when any of FrAdV1 Pre-VII NLS is individually inactivated, the others can functionally compensate for its absence. However, simultaneous inactivation of NLSa, NLSc and NLSd was sufficient to significantly reduce nuclear accumulation (mNLSacd, Fn/c of 7.7). This data suggests that NLSb is functional in the context of full-length protein, but not sufficient to mediate optimal nuclear targeting. Accordingly, further inactivation of NLSb completely abolished nuclear accumulation (mNLSabcd, Fn/c 1.5). Therefore, FrAdV1 contains multiple functional NLSs responsible of active nuclear import, which are apparently functionally redundant.
On the other hand, almost any substitution introduced in FrAdV1 Pre-pVII NLSs significantly affected nucleolar accumulation, although to a different extent (Fig. 7B). Quantitative analysis revealed that GFP-FrAdV1 Pre-pVII nucleolar accumulation (Fno/n of c. 20) was minimally affected by substitutions within NLSa (mNLSa, Fno/n of 13.5) or NLSd (mNLSd, Fno/n of 10.8), while it was clearly affected by substitution of either NLSb (mNLSb, Fno/n of 3.7) and especially NLSc (mNLSc, Fno/n of 1.2). Overall, all derivatives containing substitutions within NLSc failed to accumulate in the nucleoli, confirming that NLSc (111GRK IKK AR AP120) has a major role in nucleolar accumulation, while NLSb (RKK IPK TPVGVVGPLWQGTRKR -95) also contributes to nucleolar targeting.
FAdV1 Pre-pVII is transported into the nucleus by IMPα/β and IMPβ, specifically recognising individual NLSs. To more precisely characterize the contribution of each NLS identified in this study to the functional interaction with specific IMPs, we analyzed the effect of overexpressing inhibitors of specific nuclear import pathways on the subcellular localization of the FrAdV Pre-pVII substitution derivatives. To this end, mammalian cells were transfected to express GFP-fusion proteins (Fig. 8A) in the presence and absence of mcherry-Bimax2, M9M-RFP or DsRed-RanQ69L, and microscopic images were captured by CLSM (Fig. 8B), followed by quantitative analysis of the levels of nuclear (Fn/c; Fig. 8C) and nucleolar (Fno/n; Fig 8D) accumulation. Despite the nuclear accumulation of FrAdV1 Pre-pVII not being affected by any inhibitor tested, nuclear import of all substitution derivatives lacking NLSc (mNLSc, mNLSac, mNLScd and mNLSabc) was significantly inhibited by RanQ69L (Fig. 8B and C, see Supplementary Figs S3-S5). This suggests that Pre-pVII is capable of passively diffusing into the nucleus, likely due to its small size, and accumulating in the nucleolus in the absence of active transport by binding to nucleolar components dependent on NLSc, which functions as a NoLS. In the absence of a NoLS, nuclear accumulation is dependent on Ran-dependent active nuclear import, and therefore inhibited by RanQ69L. Intriguingly, M9M did not inhibit nuclear import of any FrAdV1 derivative (Fig. 8BC, Supplementary Figs S6 and S7), despite the reduced accumulation of a GFP-GST-FUS fusion protein, which has been shown to be functionally depend on transportin (also called IMPβ2; Supplementary Fig. S8). These findings suggest that transportin does not play a major role in FrAdV1 Pre-pVII nuclear import. Finally, nuclear accumulation of most FrAdV1 Pre-pVII derivatives was not affected by Bimax2 co-expression, unless both NLSa and NLSc were simultaneously inactivated, such as in the case of FrAdV1 Pre-pVII mNLSac, and Pre-pVII mNLSacd (Fig. 8B and C, Supplementary Figs S9 and S10). Taken together, our data suggest that each signal identified in this study plays a specific and independent role in mediating FrAdV1 Pre-pVII subcellular trafficking. NLSa confers IMPβ1 dependent nuclear import, NLSb is an atypical sequence which contributes both to IMPα/β nuclear import and nucleolar targeting, and NLSc is the main NoLS, mediating nucleolar localization by interacting with intranuclear components, while NLSd is a cNLS conferring IMPα/β nuclear import (Fig. 9).