Extended Data Fig. 4: Endogenous TREX–mRNP complex purification strategies, biochemical characterization, and negative stain EM.

a. Endogenous TREX–mRNP complexes were obtained via affinity purification of ectopically overexpressed THOC1-GFP in K562 cell nuclear extract (NE), which underwent a mild nuclease treatment. Purified TREX–mRNPs sediment at ~90–100 S in a sucrose density gradient. Individual fractions were analyzed by SDS-PAGE and S-values were estimated using CowSuite. This experiment was done more than ten times. b. Mass spectrometry analysis of endogenous TREX–mRNP complexes shows the 11 members of TREX and EJC within the top 12 hits. The relative abundance of each protein was estimated by summing up the peak areas of the top three peptides. Asterisks indicate tubulin proteins, which are abundant cellular proteins that are common purification contaminants. See Supplementary Table 1 for a complete list of identified proteins. c. TREX–mRNP purification yields the same protein composition using different strategies: (i) two different cell lines, ectopic THOC1-GFP overexpression (Lenti O/E) versus endogenous GFP-THOC5 CRISPR/Cas9-tagging (Endo), (ii) nuclear extract preparation methods, rapid cell fractionation (RCF) versus the standard nuclear extract preparation protocol (see Methods for details) or (iii) without and with mild nuclease digestion with benzonase. SDS-PAGE gels after affinity purification using GFP-trap resin and elution with 3C protease are shown. The experiment comparing RCF versus standard nuclear extract preparation protocols was done once. The comparison between THOC1-3C-GFP Lenti O/E and GFP-3C-THOC5 Endo nuclear extracts was done twice. The comparison between benzonase and non-benzonase treatments was done eight times. For gel source data, see Supplementary Fig. 5. d. SRSF1 is phosphorylated in endogenous TREX–mRNP complexes. Western blot analysis of SRSF1 in purified TREX–mRNPs before (lane 1) and after (lane 2) treatment with lambda phosphatase. Phosphorylated SRSF1 migrates slower during SDS-PAGE-PAGE and is less efficiently recognized by the anti-SRSF1 antibody. This experiment was done four times. For gel source data, see Supplementary Fig. 6. e. NXF1 is absent from purified TREX-mRNPs. Western blot showing protein levels of THOC1, NXF1, EIF4A3 and the proteasome subunit PSMA7 control in input (standard nuclear extract) and affinity purified TREX–mRNPs. While THOC1 and EIF4A3 are enriched in TREX–mRNPs, NXF1 and the proteasome are not. NXF1–NXT1 may be absent from TREX–mRNPs either due to a low affinity interaction with TREX–mRNPs or because it associates after an additional mRNP remodelling step. The experiment was done twice. For gel source data, see Supplementary Fig. 7. f. Mild nuclease treatment is required to obtain well-separated TREX–mRNP particles for electron microscopy. The nuclease activity of benzonase was reduced by omitting Magnesium from the buffer. Negative stain EM micrographs of TREX–mRNPs purified from nuclear extract either without (left) or with (right) mild nuclease treatment show that non-treated TREX–mRNP particles more frequently clump together. This experiment was done once. Scale bar, 200 Å. g. Purified TREX–mRNPs without (top) or with (bottom) mild nuclease treatment show identical negative stain EM 2D class averages. TREX complexes are indicated on the 2D classes using green arrow heads, showing that in both conditions single and multiple TREX complexes bound to a globular mRNP density. Scale bar, 200 Å. h. Purified TREX–mRNPs without (left) or with (right) mild nuclease treatment show identical negative stain EM 3D reconstructions. Scale bar, 200 Å. i. Nuclease treatment does not affect TREX–mRNP particle diameter or shape when visualized with negative stain EM. Left: Violin plot of TREX–mRNP particle diameters measured on negative stain electron micrographs. Horizontal bars indicate 25^th (grey), 50^th (black) and 75^th (grey) percentiles. Nuclease-treated (n = 259) or untreated (n = 245) particles are not significantly different (Welch’s t-test, p = 0.91). Right: Particle roundness, calculated by dividing the length of the shortest axis of each particle by the length of the longest axis, is also not significantly different (Welch’s t-test, p = 0.82).