Fragile X related protein 1 clusters with ribosomes and messenger RNAs at a subset of dendritic spines in the mouse hippocampus

Abstract

The formation and storage of memories in neuronal networks relies on new protein synthesis, which can occur locally at synapses using translational machinery present in dendrites and at spines. These new proteins support long-lasting changes in synapse strength and size in response to high levels of synaptic activity. To ensure that proteins are made at the appropriate time and location to enable these synaptic changes, messenger RNA (mRNA) translation is tightly controlled by dendritic RNA-binding proteins. Fragile X Related Protein 1 (FXR1P) is an RNA-binding protein with high homology to Fragile X Mental Retardation Protein (FMRP) and is known to repress and activate mRNA translation in non-neuronal cells. However, unlike FMRP, very little is known about the role of FXR1P in the central nervous system. To understand if FXR1P is positioned to regulate local mRNA translation in dendrites and at synapses, we investigated the expression and targeting of FXR1P in developing hippocampal neurons in vivo and in vitro. We found that FXR1P was highly expressed during hippocampal development and co-localized with ribosomes and mRNAs in the dendrite and at a subset of spines in mouse hippocampal neurons. Our data indicate that FXR1P is properly positioned to control local protein synthesis in the dendrite and at synapses in the central nervous system.

Figure 1. FXR1P is expressed in neurons of the developing hippocampus.

A. Hippocampal lysates were prepared from mice at different developmental stages (P2 = postnatal day 2) and analyzed for FXR1P, FXR2P, FMRP, L7 ribosomal protein and GAPDH. Isoforms of FXR1P (a.b,c,d), FXR2P, FMRP and ribosomal protein L7 were all highly expressed during early postnatal development in the hippocampus. B. FXR1P (isoforms c, d) expression across postnatal development was quantified and normalized against GAPDH expression. FXR1P levels decreased relative to GAPDH. C. We immunostained cryostat sections prepared from a P14 mouse with #ML13 and imaged the hippocampus at 10X (left panel). We found that FXR1P was highly expressed in neurons at P14. A 60X image of pyramidal neurons in area CA1 of the hippocampus (box in 10X image) showing FXR1P expression in the cell body and proximal dendrites of CA1 neurons (right panel). Arrow points to a proximal dendrite found in the plane of the image. Scale bar = 60 µm (low magnification) and 10 µm (high magnification)

Figure 2. FXR1P is associated with polyribosomes in mouse brain extracts.

Aliquots of native polyribosomes and EDTA treated polyribosomes were loaded onto linear 15–45% (w/w) sucrose gradients and centrifuged for 2 hr at 34 000 rpm at 4°C in a Beckman SW40 rotor. Each collected fraction was assayed for the presence of FXR1P and L7 ribosomal protein. Fractions from the top to the bottom of the gradient are shown from left to right and the position of the 80 S ribosome monomer is indicated. SS, LS: ribosomal small and large subunits, respectively.

Figure 3. FXR1P forms clusters along the dendrite and at a subset of spine-like protrusions.

A. We fixed dissociated hippocampal neurons at different developmental time-points and immunostained them with antibodies against FXR1P (#ML13; green) and MAP2 (dendritic marker; magenta). FXR1P formed clusters along dendrites at all developmental time-points. High magnification views of the segments outlined in white are shown below each image. We noted an increase in cluster size and intensity over time. Scale bar = 20 µm (low magnification) and 5 µm (high magnification). B. We transfected hippocampal neurons at 14 days in vitro with a plasmid encoding membrane targeted red fluorescent protein (RFPf) and immunostained for FXR1P. The single plane FXR1P image was thresholded to highlight the brightest clusters. FXR1P was found in clusters along the dendrite and at a subset of spine-like protrusions. Scale bar = 10 µm. C. (I) High magnification view of the segment of dendrite boxed in white in B. FXR1P clusters were found in the base, neck or head of a subset of dendritic spine-like protrusions. Arrows denote spine-like protrusions with an FXR1P cluster; arrowheads denote spine-like protrusions without an FXR1P cluster. DIV = days in vitro. Scale bar = 5 µm C. (II) High magnification view of the FXR1P positive spine-like protrusions labeled in C (I). Scale bar = 2.5 µm (high magnification).

Figure 4. FXR1P colocalizes with ribosomes in clusters along the dendrite.

A. Immunostaining of dissociated hippocampal neurons at 14 days in vitro with anti-FXR1P (#ML13) and anti-P0 shows a high degree of colocalization between FXR1P and P0 (white signal). Scale bar = 10 µm. B. High magnification view of the dendritic segment outlined in A showing colocalization between FXR1P and P0 in clusters along the dendrite. Scale bar = 2.5 µm. C. Graph demonstrating the covariance in the fluorescence intensities of FXR1P and P0 along the dendritic segment traced in B. D, E. Example of the results obtained from the Intensity Correlation Analysis (ICA). Images showing FXR1P, P0 and merged staining (D). Arrows point to colocalized clusters of FXR1P/P0, whereas arrowheads point to bright FXR1P clusters lacking P0. Scale bar = 5 µm. In E, the fluorescence intensity of FXR1P and P0 was plotted against the Products of the Differences from the Mean (PDM) of that pixel. Pixels where fluorescence intensities are correlated are shown to the right of the red line; uncorrelated pixels are shown on the left. These graphs show that a large number of high intensity P0 and FXR1P pixels are correlated. However, a fraction of high intensity FXR1P pixels are not correlated with P0 intensity, whereas a fraction of low intensity P0 pixels are not correlated with FXR1P intensity. (Inset) Image showing the positive PDM produced using the ICA plugin in ImageJ. For clarity, only the PDMs for pixels with intensities above the mean are shown. An intensity lookup table has been applied to the image and is shown on the right. Scale bar = 5 µm.

Figure 5. FXR1P colocalizes with mRNAs in clusters along the dendrite.

A. We performed fluorescence in situ hybridization on dissociated hippocampal neurons at 14 days in vitro using a digoxigenin-labeled poly(dT) probe to detect polyadenylated mRNAs. In situ hybridization was followed by immunostaining using anti-FXR1P (#ML13) and anti-P0 antibodies (data not shown). This merged image shows a high degree of colocalization between FXR1P and poly(dT) (white signal). Scale bar = 10 µm. B. High magnification view of the dendritic segment outlined in A showing colocalization between FXR1P and poly(dT) in clusters along the dendrite. Scale bar = 2.5 µm. C. Graph showing covariance in the fluorescence intensities of FXR1P and poly(dT). D, E. Example of results obtained from the Intensity Correlation Analysis (ICA). D. Images showing FXR1P, poly(dT) and merged staining. Scale bar = 5 µm E. The fluorescence intensity of poly(dT) and FXR1P was plotted against the Product of the Differences from the Mean (PDM) of that pixel. Pixels where fluorescence intensities are correlated are shown to the right of the red line; uncorrelated pixels are shown on the left. These graphs show that the majority of FXR1P and poly(dT) pixels are correlated. Inset. Image showing the positive PDMs produced using the ICA plugin in ImageJ. For clarity, only the PDMs for pixels with intensities above the mean are shown. An intensity lookup table has been applied to the image and is shown to the right. Scale bar = 5 µm.

Figure 6. FXR1P does not colocalize with PSD95.

A. We immunostained dissociated hippocampal neurons at 14 days in vitro using anti-FXR1P (#ML13) and anti-PSD95 antibodies. Single channel and merged images show the lack of colocalization between FXR1P and PSD95. Scale bar = 10 µm. High magnification view of the dendritic segment outlined above are shown below. B. Graph showing the lack of covariance in the fluorescence intensities of FXR1P and PSD95 along the drawn line shown in A. C. Intensity correlation analysis of the segment shown in A. The fluorescence intensity of each PSD95 and FXR1P pixel was plotted against the Product Difference of the Mean (PDM) of that pixel. Pixels where fluorescent intensities are correlated are plotted to the right of the red line; uncorrelated pixels are plotted on the left. These graphs show that most of the pixels lie to the left of the red line, demonstrating a lack of colocalization between FXR1P and PSD95.

Figure 7. eGFP-FXR1P forms clusters along the dendrite and at spine-like protrusions in cultured neurons.

We co-transfected hippocampal neurons grown for either 7 or 14 days in vitro with plasmids encoding membrane-targeted red fluorescent protein (RFPf) and eGFP-FXR1P. RFPf was used to visualize filopodia and spine-like protrusions. Here we show both low magnification and high magnification images of RFPf and eGFP-FXR1P at 7 and 14 days in vitro. We find that similar to endogenous FXR1P, overexpressed eGFP-FXR1P forms clusters of different sizes all along the dendritic shaft, with some of these clusters found close to filopodia and spine-like protrusions. Arrowheads point to filopodia and spines that are closely apposed by a bright eGFP-FXR1P cluster. Scale bar = 20 µm (low magnification) and 5 µm (high magnification). D.I.V = days in vitro.

Figure 8. eGFP-FXR1P colocalizes with ribosomes.

Dissociated hippocampal neurons were transfected with eGFP-FXR1P at 7 days in vitro. Cells were fixed after 24 hours and immunostained using an antibody against P0, a marker of the large ribosomal subunit (A), S6, a marker of the small ribosomal subunit (B), and TIA-1, an RNA-binding protein and marker of stress granules (C). In all cases, neurons were also immunostained with an antibody against MAP2 to delineate the proximal dendrites. We find that the majority of eGFP-FXR1P clusters contain strong signals for P0 and S6, but not TIA-1. The same results are seen at 14 days in vitro (data not shown). Results of the colocalization analyses are shown in Table 2. Scale bar = 10 µm.

Figure 9. eGFP-FXR1P colocalizes with mRNAs.

Dissociated hippocampal neurons were transfected with eGFP-FXR1P at 14 days in vitro. Cells were fixed after 24 hours and hybridized with a digoxigenin-labeled poly(dT) probe to detect polyadenylated mRNAs. In situ hybridization was followed by immunostaining for GFP and P0 (data not shown). We found that the majority of eGFP-FXR1P clusters contain mRNAs. Results of the colocalization analyses are shown in Table 2. Scale bar = 20 µm (low magnification) and 10 µm (high magnification).

Figure 10. eGFP-FXR1P clusters are found at the base of a subset of dendritic spines.

We transfected organotypic hippocampal slices at 7 days in vitro with plasmids encoding eGFP-FXR1P and membrane targeted red fluorescent protein (RFPf). The slices were fixed after 48 hours and CA1 apical dendrites were imaged using confocal microscopy. We quantified the subcellular localization of eGFP-FXR1P with respect to the dendrite and dendritic spines. A. A representative image of an apical dendrite of a CA1 cell. eGFP-FXR1P clusters are found along the dendrite and at a subset of spines. Arrows point to spines with a closely apposed eGFP-FXR1P cluster. B. We found that the density of eGFP-FXR1P clusters was variable and averaged 0.67±0.25 clusters/µm (mean±standard deviation(SD)). C. eGFP-FXR1P clusters were found at a subset of dendritic spines. On average, eGFP-FXR1P clusters were found at 23.6±19.34% of spines (mean ± SD). D. The majority of clusters were found in the dendritic shaft ( = 66.3%, spine = 33.7%). E. eGFP-FXR1P spine clusters are more likely found at the base and neck of the dendritic spine versus the spine head (base/neck = 63.5%, head = 14.4%). Each dendritic segment is color coded to allow comparison between the different measurements. The black dot and vertical bar represent mean ± standard deviation (SD). Data represent 17 dendrites imaged from 4 independent slice cultures.