An interchangeable prion-like domain is required for Ty1 retrotransposition

Abstract

Retrotransposons and retroviruses shape genome evolution and can negatively impact genome function. Saccharomyces cerevisiae and its close relatives harbor several families of LTR-retrotransposons, the most abundant being Ty1 in several laboratory strains. The cytosolic foci that nucleate Ty1 virus-like particle (VLP) assembly are not well-understood. These foci, termed retrosomes or T-bodies, contain Ty1 Gag and likely Gag-Pol and the Ty1 mRNA destined for reverse transcription. Here, we report a novel intrinsically disordered N-terminal prion-like domain (PrLD) within Gag that is required for transposition. This domain contains amino-acid composition similar to known yeast prions and is sufficient to nucleate prionogenesis in an established cell-based prion reporter system. Deleting the Ty1 PrLD results in dramatic VLP assembly and retrotransposition defects but does not affect Gag protein level. Ty1 Gag chimeras in which the PrLD is replaced with other sequences, including yeast and mammalian prionogenic domains, display a range of retrotransposition phenotypes from wildtype to null. We examine these chimeras throughout the Ty1 replication cycle and find that some support retrosome formation, VLP assembly, and retrotransposition, including the yeast Sup35 prion and the mouse PrP prion. Our interchangeable Ty1 system provides a useful, genetically tractable in vivo platform for studying PrLDs, complete with a suite of robust and sensitive assays, and host modulators developed to study Ty1 retromobility. Our work invites study into the prevalence of PrLDs in additional mobile elements.

Introduction

Retrotransposons are pervasive across diverse eukaryotes and influence genome evolution and affect host fitness. The budding yeast Saccharomyces cerevisiae contains Ty1–5 long-terminal repeat (LTR)-retrotransposons, with Ty1 as the most abundant element in many laboratory strains (1, 2). LTR-retrotransposons are the evolutionary progenitors of retroviruses; Ty1 elements share many structural hallmarks with retroviral genomic RNA and undergo an analogous replication cycle but lack an extracellular phase. Ty1 is transcribed from LTR-to-LTR and contains two partially overlapping open reading frames: GAG and POL. Ty1 RNA serves as a template for protein synthesis and reverse transcription. Translation of Ty1 POL requires a programmed +1 frameshift near the C-terminus of GAG, resulting in a large Gag-Pol precursor (p199) (3). Like retroviral RNA, Ty1 RNA is specifically packaged into virus-like particles (VLPs) where RNA is present in a dimeric form (4–7). Proteolytic protein maturation occurs within VLPs by a protease (PR) encoded within GAG and POL. Ty1 PR cleaves the Gag-p49 precursor near the C-terminus to generate p45, the capsid protein, and Gag-Pol-p199 to form mature PR, integrase (IN), and reverse transcriptase (RT) (8, 9). Reverse transcription occurs within mature VLPs and, like HIV-1, requires a complex formed between RT and IN (10, 11). Ty1 preferentially integrates upstream of genes actively transcribed by RNA Polymerase III (Pol III) due to interactions between IN and Pol III subunits (12–14).

Ty1 Gag performs the same functions as retroviral capsid and nucleocapsid. Amino acids 159–355 encode NTD and CTD capsid folds, assembling VLPs (15), and a C-terminal domain of Gag displays nucleic acid chaperone (NAC) activity (16, 17). Sequences in the Ty1 RNA encoding the Gag protein are required for packing, dimerization, and reverse transcription (3). The N-terminal region of Gag has unknown function, and it is not known whether it is required for transposition.

While several steps of retrotransposon life cycles have been investigated, it is not wellunderstood how their RNA genomes and protein machinery associate within the cellular milieu to facilitate VLP assembly and replication. Retroviral particle assembly often occurs in subcellular domains, referred to as “viral factories” or “viral inclusions” (18, 19). The sites of Ty1 VLP assembly are cytoplasmic foci termed retrosomes, or T-bodies, which contain Ty1 RNA, Gag, Gag-Pol, and perhaps additional cellular proteins (20–22). What drives the biogenesis of retrosomes is not understood. Mounting evidence suggests liquid-liquid phase separation (LLPS) underlies many examples of membraneless compartments (23, 24). Aggregation-prone proteins that drive LLPS have overlapping properties with prions, and both are implicated in age-related disease (25–34). Spontaneous demixing in these systems is often facilitated by intrinsically disordered domains, multivalent proteins, and scaffolding around nucleic acids. Indeed, prion-like and LLPS mechanisms provide intriguing models for retroelement assembly steps. Ty1 retrosomes contain Ty1 RNA and Gag oligomers associated with the RNA. Several viruses utilize LLPS in replication and assembly, including rabies virus (35), influenza A (36), herpes simplex virus 1 (37), measles virus (38), HIV-1 (39), and SARS-CoV-2 (40). Also, the human retrotransposon LINE-1 has been reported to phase separate in vitro (41). Here, we present evidence that the Ty1 Gag protein contains a prion-like domain required for VLP assembly and transposition, raising the possibility that Ty1 Gag facilitates prion-like or phase separating behaviors within retrosomes.

Results

Bioinformatic analyses reveal a prion-like domain in Ty1 Gag.

Ty1 Gag contains several protein features, including capsid and nucleic acid chaperone domains (15, 17). The N-terminal region of the protein, meanwhile, is predicted to be unstructured and does not have previously reported function. We analyzed Ty1 Gag (Fig. 1A) and Gag-Pol (Fig. S1) using several bioinformatic tools designed to predict protein disorder, amyloidogenic secondary structures, and amino acid composition similarity to known yeast prions (42–45). For comparison, we ran the well-studied yeast prions Sup35 and Ure2, the mouse prion protein PrP, and Alzheimer’s disease-associated human Aβ_1–42 through the same bioinformatic analyses (Fig. 1B–E). Ty1 Gag contains a 71-amino acid domain with strikingly similar amino acid composition to yeast prions in its disordered N-terminus, comparable to Sup35 and Ure2. This Gag prion-like domain (PrLD) is predicted to be unstructured by AlphaFold (46) and no published structures of the region are available, similar to canonical prions (15, 47–50) (Fig. S2). Given the computational predictions and the requirement for Gag in forming Ty1 retrosomes, we further investigated prionogenic properties of the Gag PrLD, which we define as amino acid residues 66–136.

Fig 1.

The Ty1 retrotransposon Gag contains a prion-like domain. Schematic of the Ty1 retrotransposon gene organization, with a detailed view of domains of the Gag protein (A), yeast prion Sup35 (B), yeast prion Ure2 (C), mouse prion protein (PrP) (D), and human amyloid beta (Aβ) (E); PrLD = prion-like domain, capsid domain (CA) and nucleic acid chaperone domain (NAC) are defined in (15). Below are bioinformatic analyses of each protein aligned with the schematic above: yeast prion-like amino acid composition (PLAAC), predicted amyloidogenic regions (ArchCandy), predicted protein disorder (PrDOS), predicted disordered (green) and globular (grey) regions (GlobPlot2.3).

Prionogenic properties of the Gag_PrLD.

We used a well-characterized Sup35-based in vivo reporter system to assess the ability of the Gag PrLD to promote prionogenesis in a yeast strain harboring a mutant allele of the adenine biosynthesis gene, ade1–14, which contains a premature stop codon (Fig. 2A) (51–53). Soluble Sup35 functions as a translation termination factor, resulting in a truncated non-functional Ade1 protein. Yeast fails to grow on media lacking adenine and appears red due to the buildup of a metabolic intermediate. However, formation of a prion state (termed [PSI⁺]) aggregates Sup35 away from the ribosome, allowing for translational readthrough. This can be detected by adenine prototrophy and yeast colonies appearing white. Fusion of the PrLD of interest to the Sup35 N or NM domains promotes prion nucleation and has previously been used to study mammalian PrP and Aβ (53). Expression of Sup35NM-Gag_PrLD fusion under the CUP1 copper-inducible promoter stimulates prionogenesis, as detected by increased papillation on SC-Ade when compared to the reporter alone (Fig. 2B). Growth is copper responsive, however, we found that the Sup35N reporter construct displays a high background growth independent of induction (Fig. S3A–B). We next biochemically monitored prion aggregation using semi-denaturing detergent-agarose gel electrophoresis (SDD-AGE) (54). Gag_PrLD fusions formed large, slow-migrating, copper-inducible aggregates with both Sup35N (Fig. 2C) and Sup35NM (Fig. S2C) above reporter alone. Finally, we verified prion nucleation specifically, as opposed to colony growth due to accumulating suppressor mutations, by curing colonies of the prion after passaging cells on guanidine hydrochloride (GdHCl) (55, 56). Representative cells are shown for the naïve [psi⁻], induced [PSI⁺], and cured states for Sup35NM fusions to Gag_PrLD or positive control Aβ, both HA-tagged (Fig. 2D) and untagged (Fig. S2D). A large fraction of Sup35NM-Gag_PrLD Ade⁺ colonies were curable by GdHCl.

Fig 2.

Gag_PrLD nucleates a Sup35-based prion reporter. (A) Schematic of the prionogenesis assay using the ade1–14 allele containing a premature stop codon. Soluble Sup35 terminates translation at the premature stop codon, yielding a non-functional, truncated Ade1 (N-succinyl-5-aminoimidazole-4-carboxamide ribotide synthetase); yeast cannot grow on media lacking adenine (SC-Ade) and a red pigment develops. Sup35 aggregated into the prion state allows for translational readthrough and production of functional Ade1; yeast grow on SC-Ade and appear white. (B) Qualitative prionogenesis of Sup35NM fusions; growth on SC-Ade indicates either a suppressor mutation or [PSI⁺] prionogenesis. Expression of Sup35 fusions were induced with 150 μM CuSO₄. A representative image of at least 3 experiments is shown. (C) SDD-AGE analysis of Sup35N-HA with and without Gag_PrLD fusion. Expression of Sup35 fusions were induced with 100 μM CuSO₄. Monomers and high-molecular weight aggregates of chimeric proteins were detected with anti-HA antibody. A representative image of at least 3 experiments is shown. (D) Curing of Ade⁺ colonies by guanidine hydrochloride (GdHCl) of Sup35NM-HA chimeras. One [psi⁻] Sup35NM-Aβ fusion control strain is shown induced to [PSI⁺] and cured. Three independent inductions of a [psi⁻] Sup35NM-Gag_PrLD fusion are shown induced to [PSI⁺] and cured. [PSI⁺] yeast grow on SC-Ade while [psi⁻] and cured yeast do not. The table below shows the guanidine curability of Ade⁺ colonies induced by chimeric constructs.

The Gag_PrLD is required for Ty1 transposition.

Given that the Gag_PrLD promotes prionogenesis of a Sup35-based reporter, we investigated its functional role in Ty1 transposition. In a Saccharomyces paradoxus strain lacking genomic Ty1 elements (57, 58), we first deleted the PrLD from Gag in a Ty1 element provided on a plasmid and tagged with the robust and sensitive his3-AI retrotranscript indicator gene (59) (Fig. 3A). This marker contains a mutant his3 gene split by an antisense artificial intron (AI) that is inserted at the 3’ untranslated region of Ty1 in the opposite transcriptional orientation. The AI is in the correct orientation to be spliced only in Ty1his3-AI RNA; cDNA reverse transcribed from this product results in a functional HIS3 allele. Insertion into the genome, either by integration or recombination, allows cells to grow on media lacking histidine. The frequency of His⁺ prototrophy is a direct measure of Ty1his3-AI retrotransposition or cDNA recombination, collectively known as retromobility. Deletion of the Gag_PrLD in a complete Ty1his3-AI element overexpressed under the GAL1 promoter completely abolished retromobility (Fig. S4A), despite retaining normal Gag protein levels (Fig. S4B). However, the PrLD region of Gag contains cis-acting RNA signals required for efficient reverse transcription (60, 61). To distinguish between a functional role in retrotransposition of the PrLD in the Gag protein versus the role of the RNA sequences that encode for the PrLD, we used a two-plasmid system to separate Ty1 RNA and protein functions (Fig. 3B). A helper-Ty1 encodes a functional mRNA, providing protein products, but lacks a 3’ LTR thus disrupting cis-acting signals required for reverse transcription. Mini-Ty1his3-AI lacks complete open-reading frames (ORFs) but contains cis-acting signals for dimerization, packaging, and reverse transcription of mini-Ty1his3-AI RNA (61, 62). Retromobility is monitored through the his3-AI reporter. In the two-plasmid assay, deletion of the Gag_PrLD also inhibits retromobility (Fig. 3C–D), despite producing normal levels of Gag protein (Fig. 3E), confirming a critical contribution from the PrLD in the Gag protein to retromobility.

Fig 3.

Ty1 Gag chimeras containing known PrDs produce stable Gag but have a range of transposition and proteolytic maturation phenotypes. (A) Schematic of Ty1 Gag constructs. The Ty1 Gag PrLD is intact in wildtype (WT), deleted in PrLDΔ, and replaced with known PrDs in the chimeras. (B) Schematic illustrating the two-plasmid system separating Ty1 RNA and protein functions. Helper-Ty1 encodes a functional mRNA, providing protein products, but lacks a 3’ LTR thus disrupting cis-acting signals required for reverse transcription. Mini-Ty1his3-AI lacks complete ORFs but contains cis-acting signals for dimerization, packaging, and reverse transcription of mini-Ty1his3-AI RNA. The his3-AI indicator gene detects retromobility of mini-Ty1HIS3 cDNA. (C) Qualitative retromobility of chimeric Gag constructs in the two-plasmid system. Colony growth on a medium lacking histidine indicates a retromobility event. A representative image of at least 3 replicates is shown. (D) Quantitative mobility assay of galactose-induced cells. Each bar represents the mean of at least eight independent measurements, displayed as points, and the error bar ± the standard deviation. Error bars are omitted for PrLDΔ, Ure2, and Aβ chimeras that did not transpose; one retromobility event was observed in one replicate of PrLDΔ. Adjusted retromobility frequency is indicated above the bars. For Ure2 and Aβ, frequencies are indicated as less than the calculated frequency if one retromobility event had been observed. Significance is calculated from a two-sided Student’s t-test compared with WT (n.s. not significant, ***p < 0.001. Exact p-values are provided in Supplementary Table 1). (E) Protein extracts prepared from galactose-induced cells expressing the indicated Gag constructs in the two-plasmid system were immunoblotted for the protein indicated on left. Polypeptide precursors are bracketed and mature RT and IN sizes are noted on right. Pgk1 serves as a loading control. Migration of molecular weight standards is shown alongside the immunoblots. A representative image of at least 3 replicates is shown.

Ty1 mobility of Gag chimeras containing foreign PrLDs.

To better understand the nature of the PrLD’s contribution to retromobility, we asked whether the Gag_PrLD sequence is uniquely capable of facilitating retromobility. Since the Gag_PrLD has prionogenic properties and sequence similarity to prions, we created chimeric Ty1 Gags in which the PrLD is replaced with prion domains from well-studied prions and aggregating proteins (Fig. 3A). We chose the yeast prions Sup35 and Ure2, the mouse prion protein PrP, and the Alzheimer’s disease-associated human Aβ_1–42 using domains predicted computationally (Fig. 1) (52, 53, 63). Chimeric Ty1 elements on the helper-Ty1 plasmid were co-expressed with mini-Ty1his3-AI, and the level of Ty1 mobility was determined. Remarkably, substitution of the Gag_PrLD with the prion domain from yeast Sup35 or mouse PrP supported Ty1 retromobility in qualitative (Fig. 3C) and quantitative retromobility assays (Fig. 3D). Gag_Sup35N retromobility is not significantly different from wildtype, whereas Gag_PrP is an order of magnitude lower, although still readily detectable on a qualitative plate assay. Replacing the PrLD sequence disrupts RNA signals which is reflected in the single plasmid assay, in which Gag_Sup35N and Gag_PrP chimeras have dramatically reduced retromobility (Fig. S4A), despite producing similar Gag protein levels (Fig. S4B), highlighting the importance of separating protein and RNA function with the two-plasmid assay.

Retromobility measured as the frequency of His⁺ prototrophs formed from his3-AI tagged elements includes both new chromosomal integrations likely created via retrotransposition, and recombination of the spliced cDNA with homologous sequences present on the mini-Ty1his3-AI plasmid. To assess whether the chimeras support retrotransposition or merely recombination, we distinguished the two by monitoring histidine prototrophy after segregating the helper and mini-Ty1his3-AI plasmids (Fig. S4C). In our strain background with the wildtype two-plasmid system, 4% of retromobility events were due to recombination with either of the plasmids. The Gag_Sup35N and Gag_PrP chimeras had modestly increased recombination events, although only Gag_Sup35N reached statistical significance (p=0.024) (Fig. S4D). We conclude that the Gag chimeras support de novo retrotransposition and cDNA recombination remains a minor pathway (64, 65).

Effect of Gag_PrLD chimeras on Ty1 protein level and maturation.

The result that the Gag_PrLD can be replaced by foreign prion sequences indicates its function is not unique to the PrLD sequence and may be the same as provided in aggregation-prone proteins. However, not all the disordered domains tested in Gag chimeras supported transposition. Ty1 chimeras containing the domains from yeast Ure2 or human Aβ did not transpose (Fig. 3C–D). All the chimeric Gags were expressed at similar levels (Fig. 3E), arguing against different transposition phenotypes due to effects on protein stability from the foreign prion domains. The substituted prion domains are of various sizes, and Gag chimeras had predicted electrophoretic mobilities. Gag proteolytically matures from p49 to p45 and is subject to post-translational modifications, often resulting in multiple bands observed by western blot (3). To determine whether the Gag chimeras affected protein maturation, we assessed the relative levels of mature RT and IN by western blotting with antibodies specific to each protein. Deletion of the PrLD results in dramatically reduced mature RT and IN levels (Fig. 3E). The Gag_Sup35N chimera transposed as well as wildtype and produced equivalent levels of mature RT and IN. The transposition-deficient chimeras, Gag_Ure2 and Gag_Aβ, have very reduced levels, comparable to Gag_PrLDΔ. Interestingly, Gag_PrP supports transposition, although reduced from wildtype, and has low levels of mature RT and IN. These results raise the possibility that Gag chimeras can block PR function and production of mature RT and IN that are essential for Ty1 mobility.

Ty1 Gag_PrLDΔ and Gag chimeras fused to GFP affect aggregation and localization.

Proteolytic maturation of RT and IN via PR occurs within VLPs (66), which are believed to be assembled in retrosomes (20–22). Gag fused to green fluorescent protein (GFP) has been used as a reporter for retrosome assembly and location (67), therefore we examined formation of cytoplasmic foci of wildtype, mutant, and chimeric Gag-GFP in the Ty-less background. Wildtype Gag-GFP fusions formed discrete cytoplasmic foci, as previously reported using this construct, but deleting the PrLD resulted in diffuse localization throughout the cytoplasm (Fig. 4). We found that a 24 hr galactose induction, shorter than 48 hr-induction used above, was ideal for live-cell microscopy and GFP-detection as yeast cultures are in log-phase growth (Fig. S5). 24 hr induced Gag_Sup35N formed similarly discrete foci patterns as wildtype Gag, whereas Gag_Ure2 had diffuse localization similar to Gag_PrLDΔ. Gag_PrP supports transposition and predominately formed foci similar to wildtype, but also had a modest fraction of cells containing a visually distinct fluorescent morphology that appears as a single, large, very bright focus. Ty1 Gag_Aβ does not transpose, yet formed foci and an even larger fraction of cells contained these single, large foci. Forming Gag-GFP foci correlates with a requirement for transposition but, as Gag_Aβ shows, is not sufficient.

Fig 4.

Foci detected in cells expressing wildtype Gag, the GagPrLDΔ mutant, and Gag-PrLD chimeras fused to GFP. (A) Live-cell yeast fluorescence microscopy of strains expressing chimeric Gag-GFP after 24 hr galactose induction. Normaski (DIC) and GFP channels are shown with cell outlines added to GFP channels based on DIC images. The strain labels are colored to match the most common foci observed. White arrows indicate cells with a single large focus. Scale bars represent 5 μm. (B) Quantitation of categories of foci observed as a percentage in at least 300 cells. The multiple foci category includes cells with multiple large foci, one or more small foci, or a combination of both sizes. Cell counts are provided in Supplementary Table 2.

In addition, we investigated the structures formed by Gag-GFP chimeras in fixed yeast cells by thin section transmission electron microscopy (TEM) (Fig. S6) using methods similar to those used for detecting Ty1 VLPs (22). Wildtype Gag-GFP produced electron-dense structures that appear similar to VLPs but look incomplete or incorrectly assembled, lacking a circular shell with a hollow interior. Gag_PrLDΔ did not form any VLP-like structures detectable in micrographs. The Gag_Sup35N strain produced tubular or filamentous structures, also not resembling proper VLPs. And strikingly, the Gag_Aβ strain formed large densities in defined regions of the cell, instead of clusters of particles or filaments across the cytoplasm, perhaps corresponding to the single large foci seen by fluorescent microscopy. These results suggest that Gag-GFP can reveal severe assembly defects as evidenced by Gag_PrLDΔ but GFP may confer aberrant VLP assembly properties when wildtype or chimeric Gag-GFP fusions are produced in cells.

The Ty1 Gag_PrLDΔ and Gag chimeras affect VLP assembly.

To evaluate VLP assembly in the chimeras using the two-plasmid system, we examined Gag sedimentation profiles of yeast lysate run through a 7–47% continuous sucrose gradient, as previously reported (15, 58, 68). Wildtype VLPs accumulated in more dense sucrose fractions near the bottom half of the gradient, with peak fractions indicated by a bar (Fig. 5). Gag_PrLDΔ appears unable to assemble complete VLPs, as Gag in these mutants accumulated in less dense sucrose fractions near the top of the gradient. The transposition-competent chimeras Gag_Sup35N and Gag_PrP had similar sedimentation profiles as wildtype, whereas transposition-deficient Gag_Ure2 accumulated near the top of the gradient like Gag_PrLDΔ. Gag_Aβ does not support retrotransposition, but peaked in similar fractions as wildtype, although somewhat more broadly distributed across the gradient.

Fig 5.

Transposition-incompetent Gag chimeras disrupt VLP assembly. Protein extracts from galactose-induced yeast cells (Input) were fractionated over a 7–47% continuous sucrose gradient and immunoblotted for Gag. Expression plasmids and molecular weight standards are noted alongside the blots. The bars at the bottom of blots denote peak Gag fractions containing more than 1/9 of the Gag signal across the gradient, as determined by densitometric analysis. A representative image of at least 3 replicates is shown.

To further examine the VLPs assembled by each chimera, we visualized thin sections of fixed yeast cells by TEM. Cells overexpressing the wildtype two-plasmid Ty1 system produced large clusters of VLPs (Fig. 6). VLPs are characteristically round with an electron dense shell and their interior appears hollow in micrographs. Importantly, these particles were not observed in the parental yeast strain expressing empty vectors. Ty1 VLPs are heterogeneously sized and are approximately 30–80 nm in diameter, based on previous measurements of purified particles (69, 70). In thin section TEM, particles may be in different Z-planes when sectioned, therefore masking the diameter of a roughly spherical particle, and preventing quantitative particle size data collection from thin section TEM. With this limitation in mind, we measured particle diameters from several cells in multiple micrographs to estimate an approximate size, and found wildtype particles ranging from 40–80 nm, with a median diameter of 59 nm (Fig. S7), largely agreeing with previous reports of purified particles.

Fig 6.

Transposition competent Gag chimeras support VLP production. Thin-section TEM of galactose-induced cells expressing Gag chimeras. Representative cells are shown, those containing VLP clusters include zoomed in cutouts to highlight VLPs. The black bars represent 500 nm, the white bars represent 100 nm.

We did not observe any cells producing VLPs in the Gag_PrLDΔ mutant, in agreement with Gag_PrLDΔ-GFP imaging and sucrose sedimentation profiles. Taken together, these data lead us to conclude that the PrLD is required for Ty1 VLP assembly. The transposition-deficient Gag_Ure2 chimera also did not assemble VLPs as monitored by thin section TEM, again agreeing with sucrose sedimentation results. The two transposition-competent chimeras, Gag_Sup35N and Gag_PrP, assembled VLPs similar in size and appearance to wildtype. These chimeras also produced large numbers of particles in each cell, although consistently appearing somewhat more dispersed throughout the cell than wildtype particle clusters. Interestingly, Gag_Aβ does not support retrotransposition, but has a similar sucrose sedimentation profile as wildtype, suggesting it may assemble particles that are defective for transposition. In thin section TEM, we observed particles in cells expressing Gag_Aβ that are visually distinct from wildtype. The most striking difference is that these particles do not have the characteristic hollow center and instead appear electron-dense throughout. They are smaller than wildtype with a median diameter of 42 nm (Fig. S7), and, like the Gag_Sup35N and Gag_PrP chimeras, are produced in large numbers of particles but are dispersed throughout the cell. Together, these results illustrate the robust and flexible nature of VLP assembly. However, our data also underscore the requirement for PrLD functionality as yeast and mammalian Gag-prionogenic chimeras form VLPs in vivo whereas the Gag_PrLDΔ mutant does not.

Discussion

The data presented here permit several conclusions about prionogenic domains, the functional organization of Ty1 Gag, and VLP assembly. Our results demonstrate that the Ty1 Gag protein contains a novel prion-like domain that is required for VLP assembly and retrotransposition. The Gag_PrLD has intrinsic prionogenic properties demonstrated by a cell-based Sup35 reporter assay, and its function in Ty1 transposition can be replaced by certain yeast and mammalian bona fide prion domains. Our findings also raise interesting unresolved questions about sequence constraints of PrLDs and how widespread PrLD functions are across retroelements. Finally, this work suggests using Ty1 as an in vivo screening platform to study intrinsically disordered domains.

Prion properties of the Ty1 Gag_PrLD.

We have examined prionogenic properties of the Ty1 Gag_PrLD using an established cell-based assay in which the newly discovered Gag_PrLD is fused to the N- and NM- domains of Sup35. Nonsense readthrough is measured by auxotrophic growth and colony color, aggregate formation is monitored biochemically with SDD-AGE, and curability is assessed after GdHCl treatment. It will be informative to further characterize prionogenic properties of the Ty1 Gag_PrLD using additional assays on various Gag_PrLD fusion constructs, including fused to Sup35C, and measuring binding of the amyloid-sensitive dye thioflavin-T, non-Mendelian inheritance, aggregation in SDD-AGE, and GFP localization patterns (71, 72).

RNA-contributions to Gag_PrLD function.

To disentangle protein-level effects of mutations in the Ty1 Gag PrLD from mutations of cis-acting RNA sequences, we separated RNA and protein function in a two-plasmid system. Retromobility is considerably lower in the two-plasmid system (Fig. 3C) than a single-plasmid expressing the intact transposon (Fig. S4A). Nonetheless, the two-plasmid system provides a wide dynamic range allowing for sensitive measurement of the impact of Gag_PrLD chimeras on retromobility. The Gag_Sup35N chimera restored retromobility in the two-plasmid system but had a severe retromobility defect in the single-plasmid assay. Based on our current knowledge of functional contributions from Ty1 cis-acting RNA sequences to retrotransposition, this is likely due to disruption of the pseudoknot sequences in the RNA region that encodes for the PrLD (60, 61). It may be possible to engineer an equivalent pseudoknot sequence in the Sup35N-encoding RNA and restore retromobility in a single-plasmid Gag_Sup35N chimera. Our chimeric proteins may provide a useful platform to interrogate RNA requirements.

Sequence requirements of the Ty1 Gag_PrLD.

We replaced the Gag_PrLD with exogenous prion domains selected based on computational predictions and the literature. We chose the entirety of Aβ_1–42 and the complete N-terminal domain of Sup35_2–123. We introduced the highest scoring 60 amino acid stretch predicted by PLAAC, Ure2_17–76, which is within the established prion domain reported as the first 89 amino acids (52, 63). The infectious PrP 27–30 isoform initially isolated is roughly 142 amino acids long and spans from approximately residues 90 to 230 (73), but shorter truncations still display prion phenotypes (74–76) and PrP_90–159 is sufficient to induce prionogenesis in a yeast-based assay (53). The PrP_121–231 fragment is soluble, and its structure has been determined using solution NMR (49, 77). We introduced PrP_90–159 as a Ty1 Gag chimera based on prior success in yeast. It will be interesting to examine other regions of PrP for function when present in Gag.

The sequence features constraining Ty1 PrLD function are not yet well-defined. Intriguingly, both the transposition-competent Gag chimeras (Sup35 and PrP) are from proteins with oligopeptide repeats associated with prionogenesis (78, 79). However, the PrP sequence introduced as a Ty1 Gag chimera in this study does not contain these repeats. Moreover, the Ty1 Gag_PrLD does not have equivalent repeats of 8–10 amino acids. Instead, like other reported prion domains, the Gag_PrLD is Q/N-rich and is depleted of charged residues. Additionally, a large number of prolines in the Gag_PrLD likely prevents secondary structure formation and starkly contrasts with the highly alpha-helical folding of the Gag capsid domain (15, 68). Further investigation will be required to understand the sequence parameters, such as length, amino acid composition, charge, or oligorepeats, that govern function of the Gag_PrLD. Transposition-deficient Gag_PrLD chimeras may be analyzed by reversion analysis to select mutations that restore transposition. Characterizing the revertants could reveal incompatibility with PR or other Pol proteins, rather than early VLP assembly steps.

Interaction of the Gag_PrLD with the host and environment.

Many host genes have been identified that activate or restrict Ty1 transposition (80–83), and examining genetic interactions with the Gag_PrLD may provide regulatory insights. Prion domains have been proposed to be protein-specific stress sensors that allow cells to respond to environmental conditions (84–86). Substituting the Gag PrLD may therefore change Ty1 regulation and create genetic interaction partners selective for specific Gag chimeras. For example, modulators of Sup35 prionogenesis could specifically modulate Gag_Sup35N but not wildtype Ty1. Conversely, Gag chimeras may no longer be subject to regulation by Ty1 modulators.

Gag chimeras reveal varied deficiencies across the Ty1 life cycle.

Different Gag_PrLD chimeras had different phenotypes across the Ty1 life cycle. Gag_PrP supported retromobility, although less well than wildtype or Gag_Sup35N. Whereas Gag_PrP produces VLPs that appear to have wildtype morphology by TEM (Fig. 6) and have similar sedimentation profiles to wildtype (Fig. 5), Gag_PrP accumulates low levels of mature RT and IN (Fig. 3E). This could indicate an incompatibility of Gag_PrP as a substrate for PR. Another possibility is that Gag_PrP VLPs inefficiently incorporate Gag-Pol or are partially defective in ways not detectable by TEM or sedimentation. The reduced retromobility of Gag_PrP may be explained by impaired RT and IN protein maturation. Meanwhile, Gag_Aβ produces particles that do not support retrotransposition or Pol maturation. These particles lack the characteristic hollow center observed in TEM of wildtype VLPs (Fig. 6) and are noticeably smaller in diameter (Fig. S7). These observations highlight that VLP assembly is robust but underscores the point that simply assembling particles is not sufficient for transposition and that assembling correct VLPs is required for proteolytic maturation. It will be informative to measure packaging of the mini-Ty1 RNA into chimeric VLPs. Ultimately, cDNA synthesis requires both the mature enzymes and the RNA substrate to be present in VLPs. Our sedimentation and TEM results presented here build upon previously published sedimentation experiments (15, 58, 68), and strengthen the value of sedimentation as a proxy for VLP assembly. Nonetheless, the value of TEM is exemplified by the Gag_Aβ chimera, which sediments similarly to wildtype but TEM reveals aberrant particle morphology.

Gag-GFP fusions have been used as a proxy for Ty1 retrosomes (67), although we have not formally tested for Ty1 RNA co-localization in our specific system. We used a previously published wildtype GFP-fusion construct that contains the mature Gag (p45) and not a full-length element. The utility of Gag-GFP is shown by the cellular mislocalization observed in Gag chimeras. However, GFP is a 26 kD protein and fusion impaired proper VLP formation (Fig. S6), perhaps interfering with Gag-Gag contacts that must be made to assemble the complete particle structure. Examining the PrLD fused to GFP alone, without the full Gag protein, or testing a Gag truncation that lacks the NAC domain, will indicate the minimal region that promotes foci formation and if RNA recruitment is required. Ty1 Gag contains a NAC and binds Ty1 RNA, but also binds diverse RNAs in vitro and cellular mRNAs associate with Ty1 VLPs (16, 17, 62, 87–89). Whether Ty1 RNA, specifically, is required to form foci or to nucleate VLP assembly, or if there is an RNA requirement at all, will require further study. It remains to be determined whether the GFP foci and VLP nucleation site is associated with any subcellular locales, as has been previously proposed at the endoplasmic reticulum (67).

The Gag-GFP foci may mature over time, as an increased percentage of Gag_PrLDΔ cells had foci after 48 hr of induction compared to 24 hr. These foci are proxies for retrosomes and therefore represent an early step of the Ty1 life cycle preceding VLP assembly, protein maturation, and transposition. After 48 hr of galactose induction, cultures enter stationary-phase growth and have higher levels of GFP-negative cells which often appear to have a wrinkled, potentially senescent morphology (Fig. S5). We, therefore, chose to examine Gag chimeras by fluorescent microscopy after 24 hr, but our pilot experiments at 48 hr anecdotally suggested more bright, single large foci. This observation would be consistent with a kinetic component to retrosome formation and may represent a progression from LLPS towards hydrogel formation. Whether such a gel would be an irreversible phase that is unable to dissolve and proceed with VLP formation remains to be determined.

Does the Ty1 retrosome constitute a phase-separated compartment?

Wildtype cells assemble discrete VLPs that can be found throughout the cell but are often observed in a particular region of the cytoplasm, and even the wildtype Gag-GFP assembled discrete structures, observed by TEM. However, the Gag_Aβ-GFP strain produced large densities that may correspond to large foci observed by fluorescence microscopy. These assemblies would be consistent with LLPS compartments containing high concentrations of Gag-GFP that stall and cannot complete VLP assembly; however, we have not examined LLPS properties such as concentration-dependence, droplet merging, or internal mixing (23). Prion-like domains can drive formation of a gradient of assemblies, from LLPS to hydrogels and amyloid-like fibers. The Ty1 Gag chimeras may exhibit a spectrum of these morphologies. The filamentous assemblies formed by Gag_Sup35N-GFP are potentially similar to Sup35 amyloid fibers observed in vitro, and Gag_Aβ-GFP may form liquid droplets. Sup35, while canonically known for its ability to form amyloid fibers as a prion, has more recently been appreciated to undergo LLPS upon a decrease in cytosolic pH and can mature over time into a gel-like condensate (85, 90). Whereas wildtype Gag allows for VLP assembly to proceed and supports transposition, perhaps transiently existing in an LLPS state, chimeras may become blocked along the retrosome and VLP assembly pathway, resulting in the striking structures observed by fluorescence microscopy and TEM. Further work will be required for the rigorous characterization necessary to declare the Ty1 retrosome or other assemblies formed by Gag chimeras an LLPS compartment. Ty1 provides a promising system to unite studies of prion and LLPS pathways.

An interchangeable platform to study PrLD and LLPS domains in living cells.

The condensate-forming property, but not the prion-forming property, of Sup35 is conserved across 400 million years from S. cerevisiae to Schizosaccharomyces pombe, emphasizing the evolutionary importance of this ancient phenotype (85). Our discovery of the Ty1 PrLD raises the possibility that LLPS may, too, be widespread among retroelements. Our preliminary computational analyses of Pseudoviridae (Ty1/copia) retroelement family members reveal predicted PrLDs in not only the closely-related yeast Ty2, but also in distantly-related plants in the Oryza element Retrofit and the Arabidopsis elements Evelknievel and AtRE1. The human retrotransposon LINE-1 phase separates and retrotransposition is associated with cancer (91) and age-associated inflammation (92, 93). A condensate-hardening drug was found to block human respiratory syncytial virus replication which occurs in virus-induced inclusion bodies (94), highlighting the potential of the Ty1 platform to contribute to new anti-viral and other humanhealth therapeutics. The Ty1 Gag chimera strategy developed here may prove to be a useful platform to study prion-like and LLPS-forming domains due to the genetic tractability of yeast and the suite of robust and sensitive in vivo assays developed for Ty1.

Materials and Methods

Bioinformatic analyses.

PLAAC (http://plaac.wi.mit.edu/) (42) was used with default settings: core length of 60 and 100% S. cerevisiae background probabilities. ArchCandy (https://bioinfo.crbm.cnrs.fr/index.php?route=tools&tool=7) (43) was used with a score threshold of 0.500 and the transmembrane regions filter off; the sum of scores data is presented. PrDOS (https://prdos.hgc.jp/) (44) was used with the default 5% FDR and the disordered probability threshold set to 0.5. GlobPlot2.3 (http://globplot.embl.de/) (45) was used with default settings with Russell/Linding propensities. Bioinformatic outputs were uniformly plotted using a custom script using the base plot() and rect() functions in R version 3.5.2. Structure analysis was performed using PyMOL v1.5.0.5 with the “align” command.

Yeast strains and media.

Yeast strains with full genotypes are listed in Supplementary Table 3. Standard yeast genetic and microbiological techniques were used in this work (95). Prion nucleation experiments were performed in GT409, an S. cerevisiae strain that is [psi⁻ pin⁻] and harbors the ade1–14 allele which contains a premature stop codon (kindly provided by Y. Chernoff) (53). Ty1 assays were performed in the DG3582 background, a Ty-less S. paradoxus derivative of DG1768 (57, 58). For galactose induction in liquid media, starter cultures were grown overnight at 30 °C in synthetic complete (SC) dropout media containing 2% raffinose, diluted 1:20 into media containing 2% galactose, and grown at 22 °C for 48 hours.

Plasmids and cloning.

Plasmids, primers, and gene fragments are listed in Supplementary Tables 4–6. Detailed descriptions of plasmids and cloning are provided in SI Materials and Methods.

Prion nucleation and curing.

[PSI+] induction was assayed in a [psi−] strain for chimeric plasmids under a P_CUP1 promoter; yeast cells were grown at 30 °C. Yeast were grown on SC-Ura for 2 days, replica plated to SC-Ura ± 150 μM CuSO₄ and grown for 2 days, then replica plated to SC-Ade and grown for approximately 10 days until imaged. Following prion nucleation, Ade⁺ colonies were cured of [PSI⁺] by guanidine hydrochloride (GdHCl). First, the induction plasmid was counter selected on FOA and single colonies were isolated. Then, Ade⁺/Ura⁻ colonies were passaged as single colonies on YPD spotted with 10 or 25 μL of 5 M GdHCl until red-pigmented colonies developed.

SDD-AGE.

Semi-denaturing detergent-agarose gel electrophoresis (SDD-AGE) was adapted from published methods (53, 54). Detailed protocols are described in SI Materials and Methods.

Ty1his3-AI mobility assays.

Ty1 retromobility events were detected using the his3-AI retromobility indicator gene (59) by qualitative and quantitative assays as previously described (15, 58). Detailed protocols are described in SI Materials and Methods.

Immunoblotting.

Total yeast protein was prepared by trichloroacetic acid precipitation and immunoblotted using standard techniques (58, 96). Detailed protocols are described in SI Materials and Methods.

Yeast microscopy.

Detailed protocols for live-cell fluorescence microscopy and transmission electron microscopy preparation and imaging of yeast cells are described in SI Materials and Methods.

Sucrose gradient sedimentation.

Sucrose gradient sedimentation was performed as previously described (15). Detailed protocols are described in SI Materials and Methods.

Significance.

Retrovirus-like retrotransposons help shape the genome evolution of their hosts and replicate within cytoplasmic particles. How their building blocks associate and assemble within the cell is poorly understood. Here, we report a novel prion-like domain (PrLD) in the budding yeast retrotransposon Ty1 Gag protein that builds virus-like particles. The PrLD has similar sequence properties to prions and disordered protein domains that can drive the formation of assemblies that range from liquid to solid. We demonstrate that the Ty1 PrLD can function as a prion and that certain prion sequences can replace the PrLD and support Ty1 transposition. This interchangeable system is an effective platform to study additional disordered sequences in living cells.

Data Availability

All data is presented within this article and supplementary information.