Subcellular Fractions

The eFP Browser is implemented in Python and makes use of the Python Imaging Library (PIL) Build 1.1.5 (www.python.org), which we modified to provide an optimized flood pixel replacement function called replaceFill, and other Python modules, as described on the eFP Browser development homepage. The inputs for the eFP Browser are illustrated in Figure 1. A pictographic representation of the sample collection as a Targa-based image is required, as is an XML control file, shown in detail in Figure 1B. Two other inputs are a database of gene identifiers and their appropriate microarray element lookups and annotations, and a database of gene expression values for the given samples. In the case of the Arabidopsis, Cell and Mouse eFP Browsers, we have mirrored publicly-available microarray data from several sources – described in the Data Sources and subsequent two sections – in our Bio-Array Resource [10] (link). These inputs are used by the eFP Browser algorithm to generate an output image for a user's gene identifier.
The eFP Browser algorithm itself is programmed in an object-oriented manner. The main program, efpWeb.cgi, is responsible for the creation of the HTML code for the user interface and presentation of the output image. It calls on four modules to complete the task. These modules are 1) efp.py, which performs most of the functions for the generation of the output image, including the parsing of the XML control file, average and standard deviation calculations, fold-change relative to control value calculations, and image map HTML code; 2) efpDb.py, which connects to the gene expression, microarray element and annotation databases, and returns the appropriate values upon being called; 3) efpImg.py, which formulates the actual colour replace calls on the Targa input image; and 4) efpXML.py, which identifies the XML control files that are present in the eFP Browser's data directory. These are displayed to the user in the Data Source drop-down, thus obviating the need to have them hard-coded in the main efpWeb.cgi program.
In the case of the Cell eFP Browser, data in the SUBA database indicate the presence of a given protein in a particular subcellular location, either based on computational methods or as molecularly documented by mass spectrometric analysis of subcellular fractions, GFP fusions etc. [11] (link). We have used a simple heuristic to turn these data into a confidence score for a given gene product's presence in a given subcellular compartment:
where
m = molecular method index of 5 possible methods
p = prediction algorithm index of 10 possible algorithms
s = weighting for molecular method = 1
s′ = weighting for prediction algorithm = 0.2
D = presence in the subcellular compartment for a given method or algorithm (1 or 0).
The maximum value the confidence score can be for a given compartment is 7 if all methods call a given gene product present in that compartment. While we have arbitrarily given a weighting to prediction algorithm calls for a particular subcellular compartment one fifth that for a molecular method, it would also be possible to incorporate the quality scores for each prediction algorithm instead.

Differences between CHT ubiquitination levels of STs and GTs were analyzed using two-sided t tests. For comparisons across subcellular fractions, obtained from the same total synaptosomal preparation, α was set at 0.05/2. Basal cytokine levels in STs and GTs were compared using nonparametric Mann–Whitney tests because for several analytes, all, or nearly all, measures from GTs were at the assay’s detection threshold and thus data were not normally distributed. The effects of LPS (vehicle and two doses) on CHT ubiquitination levels were analyzed using two-way ANOVAs on the effects of treatment and phenotype, followed by one-way ANOVAs (where applicable) and pairwise comparisons (uncorrected Fisher’s LSD test) as permitted by ANOVA results. For parametrically analyzed data, graphs depict individual values, means and 95% confidence intervals (CI). Statistical analyses were performed using SPSS for Windows (version 17.0; SPSS) and GraphPad Prism (version 9.4.1). Exact P values were reported (Greenwald et al., 1996 (link); Sarter and Fritschy, 2008 (link); Michel et al., 2020 (link)). For major results derived from parametric tests, effect sizes (Cohen’s d) were indicated (Cohen, 1988 ).

According to the manufacturer’s instructions, subcellular fractions were prepared using an endoplasmic reticulum isolation kit (Sigma-Aldrich). All procedures were performed at 4 °C. Mouse midbrains were isolated, cut into small pieces, and homogenized in four volumes of ice-cold Isotonic Extraction Buffer (10 mM HEPES, pH 7.8, 250 mM sucrose, 25 mM KCl, 1 mM EGTA, and 1× Protease and Phosphatase Inhibitor Cocktail) with Dounce homogenizer (12 strokes). Cultured cells were harvested, washed with ten volumes of PBS, and spun down at 600 × g for 5 min. The cell pellet was suspended and incubated in three volumes of ice-cold Hypotonic Extraction Buffer (10 mM HEPES, pH 7.8, 25 mM KCl, 1 mM EGTA, and 1× Protease and Phosphatase Inhibitor Cocktails) for 20 min to allow the cells to swell. Swollen cells were centrifuged at 600 × g for 5 min. The new cell pellet was homogenized in two volumes of ice-cold Isotonic Extraction Buffer with Dounce homogenizer (10 strokes). The homogenate (referred to as total lysate) from brain tissues or cultured cells was spun at 1000 × g for 10 min. The supernatant (S1) was collected, and the pellet (P1) was saved as crude nuclei fraction. S1 was centrifuged at 12,000 × g for 15 min. The supernatant (S2) was collected, and the pellet (P2) was saved as crude mitochondria fraction. S2 was further centrifuged at 100,000 × g for 60 min. The supernatant (S3) was collected as cytosol fraction, and the pellet (P3) was saved as ER microsomes fraction. Crude nuclei fraction, crude mitochondria fraction, and ER microsomes fraction were lysed in 1% SDS buffer. SDS was added to the total lysate and cytosol fraction to 1% final concentration. Equal amounts of protein from total lysate and each fraction were resolved in SDS-PAGE and applied to western blot analysis.

Protein extracts from either total homogenates or subcellular fractions were used for Western blot assays with standard procedures. Nuclear and cytoplasmic fractions were separated using a Nuclear Extract Kit (Active Motif, Carlsbad, CA, USA) according to the manufacturer’s instructions. Fifteen micrograms of protein concentrate were separated on 12% SDS–PAGE gels at 120 V and transferred onto PVDF membranes (Pall Life Sciences) at 250 mA for 90 min. After transfer, the PVDF membranes were blocked with 3% BSA/0.1% Tween-TBS buffer for 1.5 h and incubated overnight at 4°C with the following primary antibodies: mouse ELAVL1/HuR (dilution 1:1,000; Santa Cruz) and mouse HSP70 (dilution 1:2,000; Santa Cruz). As a secondary antibody, we used an HRP-conjugated goat anti-mouse antibody (dilution 1:10,000, Abcam). An HRP-conjugated alpha tubulin antibody (dilution 1:2,000, Rockland) was used as a loading control. The membrane signals were detected using chemiluminescence (ChemiDoc MP, Bio-Rad). Protein bands were quantified using ImageJ software with the Band/Peak Quantification Tool (see text footnote 1).