Cytoplasmic protein aggregates interfere with nucleocytoplasmic transport of protein and RNA

See allHide authors and affiliations

Science  08 Jan 2016:
Vol. 351, Issue 6269, pp. 173-176
DOI: 10.1126/science.aad2033

Location, location, location

Aggregates of certain disease-associated proteins are involved in neurodegeneration. Woerner et al. now show that the exact location of these aggregates in the cell may be the key to their pathology (see the Perspective by Da Cruz and Cleveland). An artificial aggregate-prone protein caused problems when expressed in the cytoplasm but not when expressed in the nucleus. Cytoplasmic aggregates interfered with nucleocytoplasmic import and export. Perhaps if we can shunt pathological aggregates to the nucleus in the future, we will be able to ameliorate some forms of degenerative disease.

Science, this issue p. 173; see also p. 125


Amyloid-like protein aggregation is associated with neurodegeneration and other pathologies. The nature of the toxic aggregate species and their mechanism of action remain elusive. Here, we analyzed the compartment specificity of aggregate toxicity using artificial β-sheet proteins, as well as fragments of mutant huntingtin and TAR DNA binding protein–43 (TDP-43). Aggregation in the cytoplasm interfered with nucleocytoplasmic protein and RNA transport. In contrast, the same proteins did not inhibit transport when forming inclusions in the nucleus at or around the nucleolus. Protein aggregation in the cytoplasm, but not the nucleus, caused the sequestration and mislocalization of proteins containing disordered and low-complexity sequences, including multiple factors of the nuclear import and export machinery. Thus, impairment of nucleocytoplasmic transport may contribute to the cellular pathology of various aggregate deposition diseases.

Cellular protein homeostasis (proteostasis) is controlled by a complex network of factors, including molecular chaperones, proteases, and their regulators (1, 2). Misfolded proteins are recognized and either refolded, degraded, or sequestered to distinct cellular sites. However, when these proteostasis machineries become compromised, as is increasingly the case during aging (1, 3), aberrant proteins tend to accumulate as toxic aggregate species. This process is associated with numerous neurodegenerative diseases and other disorders (4).

Intracellular protein aggregation in disease occurs predominantly in the cytoplasm and nucleus, with toxic effects possibly arising in both locations (57). Here, we investigated the basic mechanisms by which aggregates exert cytotoxicity in a compartment-specific manner. We used authentic disease proteins and artificial β-sheet proteins known to form prefibrillar and fibrillar aggregates (8, 9). The artificial proteins are members of a combinatorial library designed to form β strands. They have no evolved biological function, and their mRNA does not contain tri- or hexanucleotide repeat regions (fig. S1, A to C), which may contribute to neurodegenerative pathology (10, 11). We analyzed two of these proteins, β17 and β23, which vary in aggregation efficiency and toxicity. An artificial protein forming a soluble α-helical bundle (αS824) served as a nontoxic control (8, 9).

β17 and β23 form aggregates in the cytoplasm and nucleus of human embryonic kidney 293T (HEK293T) cells (9). To identify the cellular compartment in which toxicity arises, we added a nuclear export sequence (NES) or nuclear localization sequence (NLS) to the proteins (fig. S1A). NES-β17 and NES-β23 formed large cytoplasmic inclusions that stained with the amyloid-specific dye NIAD-4 (Fig. 1A and fig. S2A), suggesting the presence of cross-β structure (12, 13). NLS-β17 and NLS-β23 formed multiple nuclear inclusions in close proximity to the nucleolus (Fig. 1, A and B). These inclusions stained only weakly with NIAD-4 and displayed reduced detergent solubility compared to the cytoplasmic aggregates (Fig. 1C and fig. S2A). The β proteins in both cytoplasmic and nuclear inclusions were highly immobile, as measured with β17–green fluorescent protein (GFP) fusion proteins (fig. S2B).

Fig. 1 Targeting aggregation-prone proteins to cytoplasm or nucleus modulates their properties.

HEK293T cells 24 hours after transfection of the indicated NES- and NLS-proteins. (A) Anti-Myc (red), anti-NPC (nuclear pore complex) proteins (green), and DAPI (4′,6-diamidino-2-phenylindole) (blue). Scale bar, 10 μm. (B) Anti-Myc (red), nucleoli labeled with anti-NPM1 (green) and DAPI (blue). Scale bar, 10 μm. (C) Solubility of β proteins directed to cytoplasm and nucleus. S, soluble fraction; P, pellet fraction; H3, histone H3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (D) Viability of HEK293T cells expressing αS824 and β proteins, as measured by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay. All values relative to untransfected control cells. Data are mean + SD, n = 4 independent experiments. *P ≤ 0.05, ***P ≤ 0.001 by unpaired Student’s t test. (E) HEK293T cells were transfected with the indicated constructs; after cell lysis, anti-Myc was used for immunoprecipitation (IP). T, total.

The toxicity of the nuclear β proteins was significantly reduced compared to their cytoplasmic counterparts and the proteins lacking a localization signal (Fig. 1D and fig. S3, A and B). This reduced toxicity was not due to lower expression levels (Fig. 1C and fig. S3C) or the presence of an NLS. Point mutations in the NLS prevented nuclear targeting and increased toxicity (fig. S3, A and B). Thus, otherwise virtually identical proteins form biochemically distinct aggregate species in the cytoplasm and nucleus that differ markedly in toxicity.

The nuclear aggregates coimmunoprecipitated with the abundant, negatively charged nucleolar protein nucleophosmin-1 (NPM1) (Fig. 1E) (14). During mitosis, when the nuclear envelope breaks down and nucleoli are disassembled, NPM1 distributed diffusely throughout the cell, but maintained its apparent association with the aggregates, as shown for NLS-β17 (fig. S4). Thus, NPM1 may have a chaperone-like function in shielding the surfaces of potentially toxic aggregates (15), and this may have prevented access of the NIAD-4 dye. Notably, cells with cytoplasmic aggregates underwent mitosis only very rarely.

The expression of β proteins in the cytoplasm altered the distribution of nuclear pore complex (NPC) proteins and partially dislocated NPC proteins to the cytoplasmic inclusions of NES-β17 and NES-β23 (Fig. 1A). In contrast, expression of NLS-β17 and NLS-β23 had no visible effect on NPC integrity (Fig. 1A and fig. S3A). To test whether the cytoplasmic aggregates interfered with nuclear transport processes, we used a GFP reporter carrying NES and NLS signals (Shuttle-GFP or S-GFP). In control cells, S-GFP accumulated predominantly in the cytoplasm (Fig. 2, A and B). Expression of NLS-β17 did not alter the distribution of S-GFP. However, NES-β17 caused a significant retention of S-GFP in the nucleus (Fig. 2, A and B). This inhibition of nuclear export was even more pronounced in cells containing cytoplasmic inclusions of polyQ-expanded Htt exon 1 (Htt96Q) (Fig. 2, A and B). Upon addition of the exportin inhibitor leptomycin B (LMB) (16) to control cells, S-GFP accumulated in the nucleus within 15 min (Fig. 2, A and B), reflecting rapid nuclear import. Again, this import was significantly inhibited by cytoplasmic aggregates of NES-β17 and Htt96Q, but not by nuclear NLS-β17. Thus, the cytoplasmic aggregates analyzed interfere with both import and export of proteins through the nuclear pore.

Fig. 2 Cytoplasmic aggregates interfere with nuclear protein transport.

(A) HEK293T cells cotransfected with S-GFP (green) and either empty vector (Control), NES-β17, NLS-β17, or Htt96Q (red); DAPI (blue). Leptomycin B (LMB; 10 ng/ml) was added for 15 min where indicated. Scale bar, 10 μm. (B) Quantification of S-GFP distribution from data in (A). The x axis shows enrichment of S-GFP in the nucleus relative to the cytoplasm. Data are mean + SD, n = 3 independent experiments. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 from unpaired Student’s t test. (C) HEK293T cells transfected with empty vector (Control), NES-β17, or NLS-β17 (red) were analyzed for NF-κB p65 localization (green) with and without TNFα treatment (30 min). Nuclear DNA (blue). Scale bar, 10 μm.

We also analyzed the nuclear translocation of the endogenous protein nuclear factor κB (NF-κB) subunit p65 upon activation with the cytokine tumor necrosis factor–α (TNFα) (17). p65 readily translocated in control cells and in cells expressing nuclear β-sheet proteins. In contrast, cells containing cytoplasmic aggregates failed to support p65 transport (Fig. 2C), even though phosphorylation of p65 and degradation of inhibitor of nuclear factor κB (IκB) were not altered (fig. S5).

Next we tested whether cytoplasmic aggregates also affected the transport of RNA. In control cells, polyadenylate [poly(A)] RNA (mRNA) was distributed throughout the cytoplasm and was present in small nuclear ribonucleic particles (Fig. 3A) (18). Expression of the nuclear β proteins did not influence this pattern. However, we observed a substantial nuclear accumulation of mRNA in cells expressing either the cytoplasmic β proteins, Htt96Q, or a GFP-tagged C-terminal fragment of TAR DNA binding protein–43 (TDP-F4) (Fig. 3, A and B, and fig. S6A). Similar fragments of TDP-43 aggregate in the cytoplasm of neuronal cells in amyotrophic lateral sclerosis and frontotemporal dementia (19).

Fig. 3 Protein aggregates in the cytoplasm inhibit mRNA export.

(A) HEK293T cells were analyzed for polyA RNA (green), NES-β17, NLS-β17, Htt96Q, or TDP-F4 (red); DAPI (blue). Scale bar, 10 μm. (B) Quantification of (A). Data are mean + SD, n = 3 independent experiments. (C) Quantification of protein biosynthesis from data in fig. S6B. 35S-Met incorporation in control cells was set to 100%. Data are mean + SD, n = 3 independent experiments. *P ≤ 0.05, **P ≤ 0.01 from unpaired Student’s t test. (D) Distribution of polyA RNA (green) in brain slices of 9-week-old wild-type (Wt) and R6/2 mice, transgenic for human polyQ-expanded Htt exon 1. PolyQ expanded Htt (red); nuclear DNA (blue). The middle panel shows an example of a cell with nuclear accumulation of mRNA and mRNA remaining in the cytoplasm. The right panel shows a cell with strongly reduced overall mRNA levels. (E) Quantification of fraction of cells from Wt and R6/2 mice with abnormal polyA RNA distribution from data in (D). At least 1000 cells were analyzed per animal. Data are mean + SD, n = 3 independent experiments. **P ≤ 0.01 from unpaired Student’s t test.

Impairment of mRNA export from the nucleus would explain the reported reduction in protein synthesis capacity of β-protein–expressing cells (9). Indeed, expression of the cytoplasmic β proteins, but not the nuclear β proteins, resulted in a significant reduction in protein biosynthesis (Fig. 3C and fig. S6B). Quantitative analysis of mRNA levels confirmed that the strong reduction in cytoplasmic mRNA content was due to inhibition of export from the nucleus and not to reduced RNA synthesis or increased turnover (fig. S6, C and D).

Inhibition of mRNA export by cytoplasmic aggregates was reproduced in neuroblastoma cells and primary neurons (figs. S7 and S8). Mutant Htt preferentially forms nuclear inclusions in primary neurons (7), and these did not interfere with mRNA export (fig. S8). However, in the striatum and cortex of the R6/2 mouse, which is transgenic for human polyQ-expanded Htt exon 1 (20), mutant Htt forms cytoplasmic and nuclear aggregates (21). We observed significant alterations in mRNA distribution in brain slices of 9-week-old R6/2 mice. More than 10% of the neurons analyzed showed either an accumulation of mRNA in the nucleus or an overall reduction in mRNA levels (Fig. 3, D and E, and fig. S9A).

Protein aggregates may exert toxicity through aberrant interactions with other proteins, resulting in their functional impairment and sequestration (9, 22). Quantitative interactome analysis of the β proteins in HEK cells previously identified several proteins involved in nuclear transport, including importin subunit alpha-1/KPNA2 and THOC2 (9), a subunit of the THO (suppressor of the transcriptional defects of hpr1Δ by overexpression) complex involved in mRNA export (23, 24). A similar interactome analysis in primary neurons expressing either β17-GFP or GFP alone identified the seven-subunit THO complex as a highly enriched β-protein interactor (Fig. 4A and tables S1 to 3). Indeed, expression of NES-β17 caused mislocalization and, in some cases, aggregation of THOC proteins in the cytoplasm, as shown with antibodies against THOC2 (Fig. 4B). Notably, the nuclear inclusions of NLS-β17 did not coaggregate with THOC2, although most of the THOC2 was present in the nucleus. Mislocalization of THOC2 to the cytoplasm also occurred in cells containing cytoplasmic Htt96Q or TDP-F4 aggregates (Fig. 4B), and in the brains of R6/2 mice (fig. S9B). THOC2 often formed separate inclusions that did not colocalize with these aggregates (Fig. 4B), consistent with the prion-like behavior described for certain RNA binding proteins when dislocated to the cytoplasm (25). Importin α-1 and importin α-3 were also mislocalized to cytoplasmic β-protein aggregates (fig. S10, A and B).

Fig. 4 Cytoplasmic aggregates cause mislocalization of nuclear transport factors.

(A) The β17-GFP interactome was determined by anti-GFP pull-down and quantitative mass spectrometry, with cells expressing GFP alone as a control. Significant interactors were analyzed for highly enriched annotations. The y axis indicates the fold enrichment of annotations; the x axis depicts the significance of the respective enrichment. The THO complex was the most strongly enriched interactor of β17-GFP in primary neurons. (B) Immunofluorescence analysis of HEK293T cells transfected with empty vector (Control), NES-β17, NLS-β17, Htt96Q, or TDP-F4. THOC2 (green), protein aggregates (red), DAPI (blue). Scale bar, 10 μm.

Besides THOC, the β-protein interactome in primary neurons contained splicing factors and several other RNA binding proteins (Fig. 4A and table S1), suggesting that the cytoplasmic aggregates affect not only mRNA export, but also nuclear mRNA processing. Indeed, cells containing cytoplasmic aggregates exhibited a more pronounced accumulation of mRNA in the nucleus than cells in which THOC2 was down-regulated (fig. S10C) (26).

Our findings provide insight into common mechanisms that are likely to contribute to aggregate toxicity in neurodegenerative diseases and other disorders. Cytoplasmic aggregates of artificial β-sheet proteins and authentic disease proteins caused a pronounced impairment of nucleocytoplasmic transport and a redistribution of nuclear shuttle factors to the cytosol. Several of these transport proteins, including THOC2 (fig. S11), contain disordered and low-complexity sequences that may render them vulnerable to interactions with the interactive surfaces of cytoplasmic aggregates (table S2). Thus, the inhibition of nuclear transport observed in this system can be assigned to proteotoxicity, rather than to aberrant interactions between RNA repeat sequences and RNA binding proteins (10). Such repeat RNAs occur in the coding region or in untranslated regions of disease genes and are associated with amyotrophic lateral sclerosis, frontotemporal dementia, and CAG repeat disorders (11). Their coexistence with protein aggregates has confounded the analysis of toxicity mechanisms (10, 27, 28).

Surprisingly, otherwise identical aggregation-prone proteins did not interfere with nucleocytoplasmic transport when directed to the nucleus. How the nuclear environment alters the interaction properties of the β proteins remains to be investigated in detail, but our findings suggest that the negatively charged, nucleolar protein NPM1 is involved in shielding interactive aggregate surfaces. Indeed, recent reports that misfolded cytoplasmic proteins are actively imported into the nucleus for degradation (29) support a protective role for the intranuclear sequestration of aberrant proteins (7). However, specific aggregation-prone proteins may escape recognition by the nuclear quality-control machinery. For example, nuclear aggregates of polyQ expansion proteins have been shown to interfere with transcriptional regulation by engaging transcription factors containing glutamine repeats (30). A better understanding of nuclear proteostasis may help in developing new strategies for the treatment of proteinopathies.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

Tables S1 to S3

References (3147)

References and Notes

  1. Acknowledgments: We thank R. Klein, K. Schulz-Trieglaff, and I. Dudanova from the Max Planck Institute of Neurobiology for help with the preparation of primary neurons and brain slices, and J. Cox, F. Hosp, and M. Duerrbaum for support with proteomic data analysis. We thank C. Klaips and D. Balchin for critically reading the manuscript. The research leading to these results has received funding from the European Commission under grant FP7 GA ERC-2012-SyG_318987ToPAG (to A.C.W., D.H., M.M., F.U.H., and M.S.H.), the Munich Cluster for Systems Neurology (K.F.W, M.M, F.U.H. and M.S.H.), the German Research Foundation (J.T., K.F.W.), and the Hans and Ilse Breuer Foundation (M.P.). F.U.H has an advisory position with Proteostasis Therapeutics Inc.. Data from the mass spectrometry analysis described in this manuscript can be found in the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article