RBM45 associates with nuclear stress bodies and forms nuclear inclusions during chronic cellular stress and in neurodegenerative diseases

Abstract

The RNA binding protein (RBP) RBM45 forms nuclear and cytoplasmic inclusions in neurons and glia in amyotrophic lateral sclerosis (ALS), frontotemporal lobar degeneration with TDP-43 proteinopathy (FTLD-TDP), and Alzheimer's disease (AD). The normal functions of RBM45 are poorly understood, as are the mechanisms by which it forms inclusions in disease. To better understand the normal and pathological functions of RBM45, we evaluated whether the protein functions via association with several membraneless organelles and whether such an association could promote the formation of nuclear RBM45 inclusions. Under basal conditions, RBM45 is diffusely distributed throughout the nucleus and does not localize to membraneless organelles, including nuclear speckles, Cajal bodies, or nuclear gems. During cellular stress, however, nuclear RBM45 undergoes a reversible, RNA-binding dependent incorporation into nuclear stress bodies (NSBs). Chronic stress leads to the persistent association of RBM45 with NSBs and the irreversible accumulation of nuclear RBM45 inclusions. We also quantified the cell type- and disease-specific patterns of RBM45 pathology in ALS, FTLD-TDP, and AD. RBM45 nuclear and cytoplasmic inclusions are found in both neurons and glia in ALS, FTLD-TDP, and AD but are absent in non-neurologic disease controls. Across neurodegenerative diseases, RBM45 nuclear inclusion pathology occurs more frequently than cytoplasmic RBM45 inclusion pathology and exhibits cell type-specific variation. Collectively, our results define new stress-associated functions of RBM45, a mechanism for nuclear RBM45 inclusion formation, a role for NSBs in the pathogenesis of ALS, FTLD-TDP, and AD, and further underscore the importance of protein self-association to both the normal and pathological functions of RBPs in these diseases.

Keywords: Amyotrophic lateral sclerosis; Frontotemporal lobar degeneration; Nuclear stress bodies; Protein inclusions; RBM45; RNA binding proteins.

Fig. 1

RBM45 association with membraneless nuclear organelles. a-c HEK293 cells were immunostained for endogenous RBM45 and marker proteins of the indicated organelles. Under basal conditions, RBM45 is diffusely localized throughout the nucleus and does not associate with nuclear speckles, Cajal bodies, or nuclear gems (a-c, respectively). d Sodium arsenite (0.5 mM for 1 h) was used to induce cytoplasmic stress granule formation. RBM45 does not associate with G3BP1-positive cytoplasmic stress granules during cellular stress, but coalesces into nuclear puncta. e Heat shock (42 °C for 1 h) led to a redistribution of SAFB (top) and HSF1 (bottom) into nuclear stress bodies (NSBs). Stress-induced RBM45 nuclear foci correspond to NSBs, indicated by co-localization of RBM45 with the NSBs. For all images, scale bar = 5 μm

Fig. 2

RBM45 association with nuclear stress bodies is a general response to cellular stress. HEK293 cells were treated as indicated and the distribution of RBM45 and SAFB-positive nuclear stress bodies (NSBs) was evaluated by immunocytochemistry. a In untreated cells, the distribution of RBM45 and SAFB is diffuse and nuclear. b-e Treatment with the indicated stressors results in the formation of RBM45- and SAFB-positive NSBs. For all images, scale bar = 5 μm

Fig. 3

Domains required for RBM45 incorporation into nuclear stress bodies (NSBs). a Schematic showing functional domains and their position in the full-length RBM45 protein. RRM = RNA recognition motif, HOA = homo-oligomer assembly domain, NLS = nuclear localization sequence. HEK293 cells were transfected with constructs encoding N-terminally HA-tagged wild-type (WT) or domain-modified forms of RBM45 as indicated to determine which domains of the protein are necessary for incorporation into NSBs. b In unstressed cells, the distribution of WT RBM45 is diffuse and nuclear and no NSBs are observed. c Heat shock (42 °C for 2 h) leads to the robust formation of NSBs that are positive for wild-type (WT) RBM45 and the NSB marker HSF1. d Removal of the RBM45 NLS (∆NLS) leads to its sequestration in the cytoplasm and prevents its association with NSBs during heat shock. e Removal of RRM1 (∆RRM1) does not alter the association of RBM45 with NSBs. f Removal of RRM2 (∆RRM2) prevents the association of RBM45 with NSBs. g Removal of RRM3 (∆RRM3) prevents the association of RBM45 with NSBs. h Removal of the RBM45 HOA domain (∆HOA) does not alter the association of RBM45 with NSBs. For all images, scale bar = 5 μm

Fig. 4

Chronic stress promotes RBM45 nuclear inclusion formation. HEK293 cells were treated with three separate cellular stressors for varying durations to examine the effects of chronic and acute cellular stress on the subcellular distribution of RBM45 and nuclear stress body (NSB) formation. a In untreated cells, the distribution of RBM45 is diffuse and nuclear and no SAFB-positive NSBs are visible. b Acute treatment (6 h) with the genotoxic stressor mitoxantrone (MTX; 5 μM) induces the formation of RBM45- and SAFB-positive NSBs. c Chronic treatment with MTX (24 h; 1 μM) leads to the formation of RBM45 inclusions via NSB formation. At 24 h, RBM45 nuclear inclusions are no longer positive for the NSB marker SAFB. (d and e) Similar results were obtained with acute (d) and chronic (e) treatment of cells with the oxidative stressor cadmium sulfate (CdSO₄; 30 μM [acute] or 5 μM [chronic]) and the NSB marker SAFB. f and g Similar results were also obtained for acute and chronic treatment of HEK293 cells with sodium arsenite (Ars; 1 mM [acute; g] or 0.1 mM [chronic; h]). For all images, scale bar = 5 μm

Fig. 5

RBM45 solubility during conditions of cellular stress. a Workflow to obtain soluble and insoluble total protein extracts from untreated and stressed HEK293 cells expressing endogenous, N-terminally FLAG-tagged RBM45. “SN” = supernatant. b Western blotting for FLAG-tagged RBM45 (α-FLAG), GAPDH (negative insoluble fraction control), and TDP-43 (positive insoluble fraction control) following sodium arsenite treatment (1 mM for 1 or 2 h). c FLAG-RBM45, GAPDH and TDP-43 solubility following heat shock (1 h at 42 °C)

Fig. 6

Live-cell imaging of RBM45 nuclear stress body (NSB) and nuclear inclusion formation during conditions of cellular stress. HEK293 cells induced to express N-terminal GFP-tagged RBM45 were used to visualize the subcellular distribution of RBM45 during normal conditions and during acute and chronic cellular stress. a In untreated cells, RBM45 is predominantly nuclear and diffusely localized throughout the nucleus. b Treatment with 0.5 mM sodium arsenite for 1 h causes the formation of numerous RBM45-positive nuclear stress bodies (NSBs; arrowheads). Cells were treated with 0.5 mM sodium arsenite at t = 0 h. Under conditions of acute cellular stress, NSB formation is reversible, most RBM45-positive NSBs have dissipated by t = 6 h, and RBM45 remains diffusely localized throughout the nucleus for the remainder of the experiment. c Chronic cellular stress leads to the entrapment of RBM45 in nuclear inclusions. Cells were treated with 0.1 mM sodium arsenite for 12 h to assess the effect of chronic cellular stress on RBM45 subcellular distribution. Because of the lower concentration of sodium arsenite, the time course of NSB formation is altered. Following removal of sodium arsenite (panel “Washout”), RBM45 remains in nuclear inclusions for the remaining 12 h duration of the experiment (arrowheads). For all panels, scale bar = 5 μm

Fig. 7

RBM45 subcellular distribution and inclusion pathology in FTLD-TDP, ALS, and AD. Immunohistochemistry for RBM45, SAFB, and TDP-43 was performed in hippocampal dentate gyrus from non-neurologic disease control, FTLD-TDP, ALS, and AD subjects. a Control subjects showed strong immunoreactivity for SAFB, TDP-43, and RBM45, with RBM45 immunoreactivity concentrated in the perinucleolar region (arrowheads) and SAFB and TDP-43 diffusely localized throughout the nucleus. b RBM45 immunoreactivity was predominantly in nuclear inclusions in FTLD dentate gryus cells, with attendant loss of cytoplasmic and perinucleolar immunoreactivity. Nuclear RBM45 inclusions lack SAFB and TDP-43. Dentate gyrus cells containing RBM45 nuclear inclusions also exhibit reduced SAFB immunoreactivity. Several TDP-43 and RBM45-positive cytoplasmic inclusions were found in FTLD subjects (arrow). c As in (b), but for ALS subjects. Nuclear RBM45 inclusions lack SAFB or TDP-43, and cytoplasmic inclusions contain RBM45 and TDP-43 (arrow). d As in (b), but for AD subjects. Nuclear RBM45 inclusions lack SAFB and TDP-43, and cytoplasmic inclusions contain RBM45 and TDP-43 (arrow). e Boxplot showing the number of RBM45 nuclear inclusions by group (# = p < 0.01, * = p < 1 × 10^− 6). f Stacked bar plot showing the number of RBM45 nuclear inclusions per cell by subject group. Most dentate gyrus cells from control subjects have no inclusions, while FTLD-TDP, ALS, and AD subjects have significantly more granules per cell across the range of observed values. g The relationship between RBM45 nuclear inclusions and SAFB immunoreactivity is shown, along with the regression line. Regardless of disease state, with increasing numbers of RBM45 nuclear inclusions, SAFB immunoreactivity decreases, particularly at ≥3 granules per cell. The relationship between these two variables is statistically significant (p < 1 × 10^− 16). In the scatterplot shown, jitter is added to the points to more clearly visualize the data, but the separation between inclusion integer numbers is still seen. The inset image shows an example of adjacent cells in an FTLD-TDP subject where one cell with no RBM45 granules has strong SAFB nuclear immunoreactivity, while the neighboring cell has several nuclear RBM45 inclusions and low SAFB nuclear immunoreactivity. In (a-d), the scale bar = 20 μm, while in the inset in (g) the scale bar = 5. For all panels, FTLD = FTLD-TDP

Fig. 8

RBM45 subcellular distribution and inclusion pathology in lumbar spinal cord. Immunohistochemistry for RBM45 with either SAFB or TDP-43 was performed in lumbar spinal cord sections from ALS and non-neurologic disease control subjects. a RBM45, SAFB, and TDP-43 immunoreactivity were measured in motor neurons and glial cells of non-neurologic disease control subjects. RBM45 exhibited strong nucleolar immunoreactivity in motor neurons (arrowheads), while SAFB and TDP-43 immunoreactivity were diffusely nuclear. b In ALS patients, motor neuron nucleolar RBM45 immunoreactivity was reduced (arrowheads) and large, skein-like cytoplasmic inclusions were observed. These inclusions were positive for TDP-43 but negative for SAFB. ALS patient glial cells frequently contained multiple RBM45 nuclear inclusions (arrows). c Glial cells from non-neurologic disease control (top) and ALS (bottom) subjects. Glial cells from ALS subjects frequently contained 1 or more nuclear RBM45 inclusions. The presence of glial RBM45 nuclear inclusions was associated with reduced SAFB immunoreactivity. d Boxplot showing the number of glial nuclear RBM45 inclusions in control and ALS subjects. For the comparison shown, * = p < 1 × 10^− 8. e Stacked bar plot showing the proportion of cells (Y axis) with the given number glial nuclear RBM45 inclusions (X axis) across groups. ALS subjects have significantly more inclusions per glial cell than control subjects, who rarely exhibit RBM45 glial cell pathology. f The relationship between glial RBM45 nuclear inclusions and SAFB immunoreactivity is shown, along with the regression line. Increasing numbers of RBM45 nuclear inclusions are associated with reduced SAFB immunoreactivity, particularly at ≥3 granules per cell. The relationship between these two variables is statistically significant (p < 1 × 10⁻⁶). In the scatterplot shown, jitter is added to the points to more clearly visualize the data, but the separation between inclusion integer numbers can still be seen. The inset image shows an example of adjacent glial cells in an ALS patient where one cell without RBM45 nuclear inclusions has strong SAFB nuclear immunoreactivity, while the neighboring cell with RBM45 nuclear inclusions shows reduced SAFB immunoreactivity. In (a and b) the scale bar = 20 μm, in (c) the scale bar = 5 μm, and for the inset in (f) the scale bar = 2.5 μm