Assembly and regulation of the mammalian mRNA processing body

Abstract

Messenger RNA processing bodies (P-bodies) are cytoplasmic membrane-free organelles that contain proteins involved in mRNA silencing, storage and decay. The mechanism by which P-body components interact and the factors that regulate the stability of these structures are incompletely understood. In this study, we used a fluorescence-based, two-hybrid assay to investigate interactions between P-body components that occur inside the cell. LSm14a, PATL1, XRN1, and NBDY were found to interact with the N-terminal, WD40-domain-containing portion of EDC4. The N-terminus of full-length PATL1 was required to mediate the interaction between EDC4 and DDX6. The C-terminal, alpha helix-domain- containing portion of EDC4 was sufficient to mediate interaction with DCP1a and CCHCR1. In the absence of endogenous P-bodies, caused by depletion of LSm14a or DDX6, expression of the portion of EDC4 that lacked the N-terminus retained the ability to form cytoplasmic dots that were indistinguishable from P-bodies at the level of UV light microscopy. Despite the absence of endogenous P-bodies, this portion of EDC4 was able to recruit DCP1a, CCHCR1 and EDC3 to cytoplasmic dots. The results of this study permit the development of a new model of P-body formation and suggest that the N-terminus of EDC4 regulates the stability of these structures.

Fig 1. NLS-EDC4 localizes to Cajal bodies, and adjacent to PML-nuclear bodies.

A plasmid encoding green fluorescent protein (GFP) fused to the nuclear localization sequence (NLS) of SV40 T antigen and full length EDC4 was transfected into HEp-2 cells. GFP-NLS-EDC4 (panel a) localized adjacent to PML-nuclear bodies (panels b, c). The GFP-NLS-EDC4 fusion protein was detected using mouse monoclonal anti-GFP antibodies and PML nuclear bodies were identified using rabbit anti-Sp100 antiserum. Transfection of GFP-NLS-EDC4 and subsequent staining with mouse anti-GFP antibodies and rabbit anti-p80 coilin antibodies revealed that NLS-EDC4 (panel d) localized to Cajal bodies (panels e, f). Transfection of HEp-2 cells with a plasmid encoding GFP-NLS-EDC4(630–1437) (panel g) and subsequent staining with mouse anti-GFP antibodies and human serum containing anti-p80 coilin antibodies (panels h, i), revealed that in the absence of the N-terminus, EDC4 was no longer able to localize to Cajal bodies. GFP-NLS-EDC4(630–1437) (panel j) retained the ability to localize adjacent to PML nuclear bodies (panels k, l). Merge of panels a and b, d and e, g and h, and j and k is shown in panels c, f, i, and l respectively. DAPI staining (blue) indicates the location of nuclei in c, f, i and l.

Fig 2. Interaction between PATL1 and EDC4 requires the presence of the C-terminus of PATL1 and the N-terminus of EDC4.

After transfection of a plasmid encoding GFP-PATL1 into HEp-2 cells, endogenous EDC4 (panel a) and GFP-PATL1 (panels b, c) localized to cytoplasmic dots. Co-expression of NLS-EDC4 and GFP-PATL1 in HEp-2 cells resulted in localization of NLS-EDC4 (panel d) in nuclear dots, while GFP-PATL1 remained in cytoplasmic dots (panels e, f). Expression of a plasmid encoding NLS-EDC4 (panel g) together with a plasmid encoding Myc fused to PATL1 amino acids 398–770 (panels h, i) resulted in the two proteins co-localizing in nuclear dots in 97 +/-3% of cells that expressed both proteins. When cells were transfected with GFP-NLS-EDC4(630–1437) (panel j) and Myc-PATL1(398–770) (panels k, l), GFP-NLS-EDC4(630–1437) localized to nuclear dots while Myc-PATL1(398–770) remained in cytoplasmic dots in all cells that expressed both proteins. Note that the anti-Myc antiserum produces faint diffuse nuclear staining. EDC4 in panels a, d, and g was detected using human serum containing anti-EDC4 antibodies. GFP-PATL1 in panels b and e, and GFP-NLS-EDC4(630–1437) in panel j were detected using mouse monoclonal anti-GFP antibodies. Myc-PATL1 in panels h and k was detected using rabbit anti-Myc antiserum. Merge of panels a and b, d and e, g and h, and j and k is shown in c, f, i and l respectively. DAPI staining (blue) indicates location of nuclei in c, f, i and I. White arrows in j-l indicate a representative cell that contains NLS-EDC4(630–1437) in nuclear dots, while Myc-PATL1(398–770) localized to cytoplasmic dots.

Fig 3. Full-length PATL1 interacts with EDC4.

Transfection of HEp-2 cells with GFP-PATL1 followed by treatment with leptomycin B (LMB) resulted in localization of PATL1 in one of three nuclear patterns: homogeneous (not shown), finely speckled and nuclear dot (panel a). Endogenous EDC4 remained in cytoplasmic dots, despite treatment with LMB (panels b, c). White arrows in a-c indicate a representative cell with GFP-PATL1 in nuclear dots, while endogenous EDC4 remained in P-bodies. Co-expression of GFP-PATL1 (panel d) and NLS-EDC4 (panels e, f) and subsequent treatment with LMB resulted in the two proteins localizing to nuclear dots in all cells that expressed both proteins. NLS-EDC4 recruits NLS-LSm14a to nuclear dots and the N-terminus of EDC4 is required for the observed interaction. Co-expression of NLS-EDC4 (panel g) and NLS-LSm14a (panel h) resulted in localization of the two proteins to nuclear dots (panel i) in 86 +/-4% of cells with both proteins in the nucleus. In cells transfected with GFP-NLS-EDC4(630–1437) and NLS-LSm14a, GFP-NLS-EDC4(630–1437) was detected in nuclear dots (panel j), while NLS-LSm14a (panel k, l) was distributed diffusely throughout the nucleus. NLS-LSm14a did not co-localize with GFP-NLS-EDC4(630–1437) in nuclear dots in any of the cells that contained both proteins in the nucleus. Mouse monoclonal anti-GFP antibody was used to detect GFP-PATL1 in a and d, and GFP-EDC4(630–1437) in panel j. Human serum was used to detect EDC4 in b, e, and g. Rabbit anti-LSm14a antiserum was used to detect NLS-LSm14a in h and k. Note that the rabbit antiserum detected endogenous LSm14a in cytoplasmic dots (panels h, k). Merge of panels a and b, d and e, g and h, and j and k is shown in c, f, i and l respectively. DAPI staining (blue) indicates location of nuclei in c, f, i, and l.

Fig 4. EDC4 interacts with XRN1.

Transfection of HEp-2 cells with GFP-NLS-EDC4 (panel a) and monomeric cherry (mCh) fused to SV40 T antigen NLS and XRN1 amino acids 1232–1706 (mCh-NLS-XRN1(1232–1706) (panel b) resulted in localization of both proteins in nuclear dots (panel c) in 94+/-2% of cells expressing both proteins. Co-expression of GFP-NLS-EDC4(630–1437) and mCh-NLS-XRN1(1232–1706) resulted in localization of GFP-NLS-EDC4(630–1437) to nuclear dots (panel d) while mCh-NLS-XRN1(1232–1706) remained in cytoplasmic dots (panels e, f) in all cells that expressed both proteins. EDC4 interacts with NBDY. Co-expression of NLS-EDC4 (panel g) and mCh-NBDY (panels h, i) resulted in both proteins localizing diffusely throughout the nucleus in 59+/-2% of cells expressing both proteins. In the remaining cells that expressed both proteins, mCh-NBDY co-localized with NLS-EDC4 in nuclear dots. A single white arrow in g-i indicates a cell expressing both NLS-EDC4 and mCh-NBDY with both proteins in a homogeneous nuclear staining pattern. Double white arrows in g-i indicate a cell with both proteins in nuclear dots. Localization of GFP-NLS-EDC4(630–1437) to nuclear dots (panel j) was not affected by expression of mCh-NBDY (panels k, l). Human serum was used to detect NLS-EDC4 in panels a and g, and mouse monoclonal anti-GFP detected GFP-EDC4(630–1437) in d and j. Rabbit anti-mCh antiserum was used to detect mCh-NLS-XRN1(1232–1706) in b and e and mCh-NBDY in h and k. Merge of panels a and b, d and e, g and h, and j and k is shown in c, f, I and l respectively. DAPI staining (blue) indicates the location of nuclei in c, f, i and l.

Fig 5. The N-terminus of EDC4 is not required for interaction with DCP1a.

Co-expression of NLS-EDC4 (panel a) and NLS-DCP1a (panel b) resulted in co-localization of the proteins to nuclear dots (panel c) in all cells that had both proteins in the nucleus. White arrows in a-c indicate a representative cell containing both NLS-EDC4 and NLS-DCP1a in nuclear dots. GFP-NLS-EDC4(630–1437) (panel d) was also able to recruit NLS-DCP1a (panels e, f) to nuclear dots in all cells that expressed both proteins. The C-terminus of EDC4 is sufficient to mediate interaction with CCHCR1. Co-expression of NLS-EDC4 (panel g) and CCHCR1-Myc (panel h) resulted in co-localization of the proteins in nuclear dots (panel i) in all cells that expressed both proteins. The C-terminus of EDC4 (GFP-NLS-EDC4(935–1437)) (panel j) was sufficient to recruit CCHCR1-Myc (panels k, l) to nuclear dots. CCHCR1-Myc co-localized with GFP-NLS-EDC4(935–1437) in 96+/-4% of cells expressing both proteins. White arrows indicate the location of representative cells that expressed NLS-EDC4 and CCHCR1-Myc (g-i) and NLS-EDC4(935–1437) and CCHCR1-Myc (j-l). Human serum was used to detect NLS-EDC4 in a and g. Rabbit antiserum was used to detect NLS-DCP1a in b and e. Rabbit anti-Myc antiserum was used to detect CCHCR1-Myc in h and k. Mouse monoclonal anti-GFP antibody detected GFP-NLS-EDC4(630–1437) and GFP-NLS-EDC4(935–1437) in d and j respectively. Merge of panels a and b, d and e, g and h, and j and k is shown in c, f, i and l respectively. DAPI staining (blue) indicates the location of nuclei in c, f, i and l.

Fig 6. DDX6 interacts with PATL1.

Co-expression of NLS-EDC4 (panel a), GFP-PATL1 (panel b) and monomeric cherry (mCh)-NLS-DDX6(289–483) (panel c) followed by treatment with leptomycin B (LMB) resulted in co-localization of the proteins in nuclear dots in 81+/-5% of cells expressing all three proteins. Faint, diffuse nuclear staining of mCh-NLS-DDX6(289–483) was also detected. Co-expression of NLS-EDC4, GFP-PATL1(398–770) and mCh-NLS-DDX6(289–483 followed by treatment with LMB resulted in localization of NLS-EDC4 (panel d) and GFP-PATL1(398–770) (panel e) in nuclear dots. However, in cells expressing all three proteins, mCh-NLS-DDX6(289–483) was not detected in nuclear dots (panel f). Human serum containing anti-EDC4 antibodies and coumarin-conjugated donkey anti-human IgG antiserum were used to detect NLS-EDC4 in a and d. Mouse monoclonal anti-GFP antibody detected GFP-PATL1 (panel b) and GFP-PATL1(398–770) (panel e). Rabbit anti-mCh antiserum detected mCh-NLS-DDX6(289–483) in c and f.

Fig 7. After depletion of LSm14a and loss of endogenous P-bodies, GFP-EDC4(630–1437), but not GFP-EDC4, localizes to cytoplasmic dots.

siRNA directed against LSm14a was transfected into HEp-2 cells, followed 24 hours later by transfection of GFP-EDC4 or GFP-EDC4(630–1437). LSm14a was not detected in siLSm14a-treated cells (panel a). GFP-EDC4 (panels b, c) was unable to localize to P-bodies and was distributed throughout the cytoplasm of transfected cells. In cells depleted of LSm14a (panel d) and expressing GFP-EDC4(630–1437) (panels e, f), GFP-EDC4(630–1437) was detected in cytoplasmic dots that were indistinguishable from P-bodies at the level of UV light microscopy. After disruption of endogenous P-bodies, neither full-length EDC4 nor EDC4(630–1437) was able to recruit DDX6 to cytoplasmic dots. After depletion of LSm14a using siRNA, GFP-EDC4 (panel g) did not form cytoplasmic dots and was unable to recruit endogenous DDX6 (panel h) to cytoplasmic dots. After depletion of LSm14a, GFP-ED4(630–1437) (panel j) was able to form cytoplasmic dots, but was unable to recruit endogenous DDX6 (panels k, l) to these structures. Rabbit antiserum was used to confirm the absence of LSm14a in a and d. Mouse anti-GFP antibodies detected GFP-EDC4 (b, g) and GFP-EDC4(630–1437) (e and j). Rabbit anti-DDX6 antiserum was used to detect endogenous DDX6 in h and k. Merge of panels a and b, d and e, g and h, and j and k is shown in c, f, i, and l, respectively. DAPI staining (blue) in c, f, i and l indicate the location of nuclei.

Fig 8. EDC4(630–1437) can recruit DCP1a, EDC3 or CCHCR1 to cytoplasmic dots despite depletion of LSm14a and loss of endogenous P-bodes.

After treatment with scrambled siRNA, LSm14a (panel a), EDC4(630–1437) (panel b) and GFP-DCP1a (panel c) all localized in cytoplasmic dots. siRNA directed against LSm14a resulting in depletion of LSM14a from HEp-2 cells (panel d), but EDC4(630–1437) (panel e) and GFP-DCP1a (panel f) localized to cytoplasmic dots. Depletion of LSm14a (panels g, j) did not affect localization of EDC4(630–1437) (panels h, k) and EDC3 (panel i) or CCHCR1 (panel l) to cytoplasmic dots. Rabbit antiserum was used to detect LSm14a in panel a and to confirm the absence of LSm14a and endogenous P-bodies in d, g, and j. Human serum reacted with EDC4(630–1437) in b, e, h and k. Mouse anti-GFP antibody detected GFP-DCP1a in c and f, GFP-EDC3 in i, and CCHCR1-GFP in l.

Fig 9. Model of the mammalian mRNA P-body.

Interactions between EDC4 and LSm14a, PATL1, XRN1, MARF1, NBDY and DDX6 (via PATL1) require the N-terminal, WD40-containing domain in EDC4. The C-terminus of EDC4 is sufficient to mediate interaction between EDC4 and DCP2, DCP1a, CCHCR1 and EDC3 (via DCP1a). RNA binding proteins DDX6, LSm14a, PATL1, and MARF1 may be present on the “outer surface” of the P-body, while the decapping protein (DCP2) and enhancers of decapping may be “buried”, and perhaps inactive, inside P-bodies. Phosphorylation of NBDY or depletion of LSm14a or DDX6 results in disruption of P-bodies.