Phase separation of the LINE-1 ORF1 protein is mediated by the N-terminus and coiled-coil domain

Abstract

Long interspersed nuclear element-1 (L1) is a retrotransposable element that autonomously replicates in the human genome, resulting in DNA damage and genomic instability. Activation of L1 in senescent cells triggers a type I interferon response and age-associated inflammation. Two open reading frames encode an ORF1 protein functioning as messenger RNA chaperone and an ORF2 protein providing catalytic activities necessary for retrotransposition. No function has been identified for the conserved, disordered N-terminal region of ORF1. Using microscopy and NMR spectroscopy, we demonstrate that ORF1 forms liquid droplets in vitro in a salt-dependent manner and that interactions between its N-terminal region and coiled-coil domain are necessary for phase separation. Mutations disrupting blocks of charged residues within the N-terminus impair phase separation, whereas some mutations within the coiled-coil domain enhance phase separation. Demixing of the L1 particle from the cytosol may provide a mechanism to protect the L1 transcript from degradation.

Figure 1

ORF1 phase separation is driven by electrostatics and is reversible. (A) Domain organization of L1 and ORF1 functional units. ORF1 consists of three folded domains: a coiled-coil region responsible for trimerization, an RRM, and a CTD, which act in concert to bind to L1 RNA. The intrinsically disordered N-terminal region is denoted as IDR. The sequence of the N-terminal 53 residues is shown color coded by charge. (B) Composite model of the ORF1 trimer. The extended coiled-coil domain (Protein Data Bank entry PDB: 6FIA) is superposed on the ORF1 core structure (PDB: 2YKO) in the boxed region. Domains are colored as shown in (A). The stammer region is highlighted in red. An enlarged view of the coiled-coil domain shows the charge distribution on the solvent-exposed surface, with basic residues shown in blue and acidic residues in red. (C) The extent and morphology of the ORF1 phase-separated state is dependent on the NaCl concentration in the buffer solution. In low salt concentrations, ORF1 forms amorphous aggregates while forming spherical droplets in ranges of 200–400 mM NaCl. In 500 mM NaCl or higher, ORF1 remains dispersed. All samples were imaged at room temperature in 20 mM Tris (pH 8.0) and 1 mM DTT with the noted NaCl concentration and 50 μM ORF1_1–338. (D) ORF1 phase separation is reversible. Increasing the NaCl concentration from 300 to 500 mM while maintaining protein concentration (300 μM) solubilizes phase-separated ORF1. Phase-separated droplets reappear when the NaCl concentration is reduced to 300 mM. Data were collected at room temperature in 20 mM Tris (pH 8.0) and 1 mM DTT in either 300 or 500 mM NaCl as noted on the micrograph. To see this figure in color, go online.

Figure 2

ORF1_1–152 readily phase separates, and the N-terminal domain and coiled-coil domain are both required for phase separation. (A) Constructs of the N-terminal region (1–53), the coiled-coil domain (53–152), and truncation of the coiled-coil domain (1–131) in a tandem construct remain soluble at a lower salt concentration (300 mM NaCl). In contrast, the full-length tandem construct lacking the RRM and CTD (1–152) phase separates to the same extent as the full-length protein, indicating that both the N-terminus and coiled-coil domains are necessary for phase separation. All data were collected with 300 μM ORF1 at room temperature in 20 mM Tris (pH 8.0), 300 mM NaCl, and 1 mM DTT. (B) The narrow chemical shifts of ORF1_1–53 in the ¹⁵N-¹H HSQC experiment demonstrate that this region is intrinsically disordered. Resonances beyond 53 correspond to the histidine tag. (C) The secondary structure propensity (SSP (39)) score and δ2D-helical propensity scores (40) show that residues 41–49 have slight α-helical character. Open circles and open bars correspond to residues from the histidine tag.

Figure 3

The ORF1 N-terminal domain interacts with the C-terminal 22 residues of the coiled-coil domain. (A) ORF1_1–53 can disrupt phase separation of the full-length ORF1_1–338 protein in 300 mM NaCl, suggesting it competes for interaction sites found in ORF1_1–338 but cannot promote phase separation of the coiled-coil domain in trans, as shown in (B). Arrows in the 3:1 sample in (A) highlight the position of smaller droplets in the sample. This suggests that multivalency and interdomain interactions provided by the fusion of these domains is important for LLPS. All data were collected with 300 μM ORF1_1–338 or ORF1_53–152 at room temperature in 20 mM Tris (pH 8.0), 300 mM NaCl, and 1 mM DTT with increasing concentrations of ORF1_1–53. (C) Comparison of ¹⁵N–¹H HSQC correlation spectra of 700 μM ORF1_1–53 alone (black) and 700 μM ORF1_1–53 mixed at an equimolar ratio with 700 μM ORF1_53–152 (red) at pH 6.0 (see detail below) shows line broadening of ORF1_1–53 in the presence of the coiled-coil domain with resonance attenuation consistent with binding of ORF1_53–152 around residues 47–53 of ORF1_1–53 and attenuations distributed across the entire sequence, shown in (D), consistent with binding at multiple sites. Open bars correspond to residues from the histidine tag. Elevated values for the ¹⁵N R₂ relaxation rate constants (E) are observed in the presence of ORF1_53–152 at pH 7.25 (black) and pH 6.0 (red). Open circles correspond to residues from the histidine tag. (F) The difference in R₂ in the presence and absence of ORF1_53–152, ΔR₂, provides further evidence that residues 47–53 are perturbed by the addition of ORF1_53–152 because of slowed or conformational exchange because of binding to the structured, slow-moving coiled coil. (G) Quantification of the ORF1 protein concentration remaining in the dispersed phase after centrifugation demonstrates that ORF1_1–131 phase separates much less readily than full-length ORF1_1–338, whereas ORF1_1–141 LLPS is intermediate. N-terminal domain deletion ORF1_65–338 also significantly disrupts LLPS. NMR experiments were collected in 20 mM HEPES (pH 7.25), 200 mM NaCl, 1 mM DTT, 5% D₂O or in 20 mM MES (pH 6.0), 200 mM NaCl, 1 mM DTT, and 5% D₂O to optimize the signal/noise ratio. Error bars denote the standard deviation of three technical replicates. To see this figure in color, go online.

Figure 4

Mutations in both the N-terminus and coiled-coil domain modulate phase separation of the full-length protein. (A) Altering the charge at the extreme N-terminus (K3E/K4E and K3E/K4E/R7E/K8E) impairs the ability of ORF1 to phase separate in 300 mM NaCl. Some mutations in the N-terminus and coiled-coil domain also appear to enhance the quantity and size of phase-separated droplets, as demonstrated by mutations S27D and L93P. All data were collected with 300 μM ORF1 at room temperature in 20 mM Tris (pH 8.0), 300 mM NaCl, and 1 mM DTT. (B) Quantification of ORF1 protein concentration in the dilute phase shows that mutations that introduce a positive charge at the extreme N-terminus significantly impair phase separation in comparison to the wild-type protein, whereas the S27D and L93P mutations enhance phase separation. Error bars denote the standard deviation of three technical replicates. To see this figure in color, go online.