Werner A et al. (2015), Cell-fate determination by ubiquitin-dependent ...

XB-ART-51292

Nature September 24, 2015; 525 (7570): 523-7.

Cell-fate determination by ubiquitin-dependent regulation of translation.

Werner A , Iwasaki S , McGourty CA , Medina-Ruiz S , Teerikorpi N , Fedrigo I , Ingolia NT , Rape M .

Abstract
Metazoan development depends on the accurate execution of differentiation programs that allow pluripotent stem cells to adopt specific fates. Differentiation requires changes to chromatin architecture and transcriptional networks, yet whether other regulatory events support cell-fate determination is less well understood. Here we identify the ubiquitin ligase CUL3 in complex with its vertebrate-specific substrate adaptor KBTBD8 (CUL3(KBTBD8)) as an essential regulator of human and Xenopus tropicalis neural crest specification. CUL3(KBTBD8) monoubiquitylates NOLC1 and its paralogue TCOF1, the mutation of which underlies the neurocristopathy Treacher Collins syndrome. Ubiquitylation drives formation of a TCOF1-NOLC1 platform that connects RNA polymerase I with ribosome modification enzymes and remodels the translational program of differentiating cells in favour of neural crest specification. We conclude that ubiquitin-dependent regulation of translation is an important feature of cell-fate determination.

PubMed ID: 26399832
PMC ID: PMC4602398
Article link: Nature
Grant support: [+]

Species referenced: Xenopus
Genes referenced: acta2 acta4 actc1 arrb1 arrb2 brd2 casp3.2 cul3 kbtbd8 nolc1 rpl28 smo tcof1 tp53
Morpholinos: cul3 MO1 kbtbd8 MO1 kbtbd8 MO2

Article Images: [+] show captions

	Extended Data Figure 2. KBTBD8 controls neural crest formationa. Stable depletion of KBTBD8 from H1 hESCs, as determined by Western. b. KBTBD8 depletion does not significantly change the cell cycle profile of hESCs, as determined by propidium iodide staining and FACS. c. Control or KBTBD8-depleted hESCs were counted at indicated times after seeding. (mean of 3 biological replicates, +/− s.d.). d. KBTBD8 depletion does not induce apoptosis in hESCs, as shown by immunostaining against cleaved caspase 3 (red) or DNA (blue). (200 cells/condition; scale bar 10μm) e. KBTBD8 depletion does not affect the gene expression profile of hESCs, as determined by microarray analysis (genes > 2.5-fold change, n>30,000; mean of 3 biological replicates, Anova p –value <0.05). f. Loss of KBTBD8 causes a decrease in the expression of neural crest cell markers during EB formation, as shown by comparative microarray analysis (genes > 2.5-fold change, n>30,000; mean of 3 biological replicates, Anova p –value <0.05). g. mRNA levels of pluripotency and differentiation markers in EBs stably expressing control- or KBTBD8-shRNAs were measured by qRT-PCR (3 technical replicates +/− s.e.m.).
	Extended Data Figure 3. KBTBD8 controls neural crest specificationa. Depletion of KBTBD8 from hESCs subjected to neural conversion results in loss of neural crest cells, as determined by immunofluorescence against HNK1, TFAP2, and p75 (n>200 cells, mean of 3 biological replicates +/− s.d.). b. H1 hESCs transduced with control- (green) or KBTBD8-shRNAs (red) were subjected to neural conversion, and expression of neural crest markers SOX10 (circles) and SNAIL2 (boxes) was monitored by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). c. H1 hESCs described above were subjected to neural conversion and abundance of CNS precursor markers SOX2 (circles) and PAX6 (boxes) was measured by qRT-PCR. d. H1 hESCs described above were subjected to neural conversion and abundance of telencephalon markers SIX3 (circles) and FOXG1 (boxes) was measured by qRT-PCR. e. Expression of OCT4 was monitored by qRT-PCR during neural conversion in the presence or absence of KBTBD8. f. hESCs stably expressing control- or KBTBD8-shRNAs were subjected to neural conversion and analyzed for expression of pluripotency (OCT4, CDH1), neural crest (SOX10, SNAIL2, AP2), or CNS precursor markers (PAX6) by Western blotting. To provide consistency, samples were taken from the same experiment as shown in Figure 5d (* marks blots also shown in Figure 5d). g. Loss of neural crest occurs in response to KBTBD8 depletion by two independent shRNAs, as shown by Western. h. hESCs were subjected to neural conversion and analyzed by immunofluorescence microscopy against SOX10- (neural crest), PAX6- (CNS precursor), and OCT4 (pluripotency).
	Extended Data Figure 4. KBTBD8 is required for differentiation into functional neural crest cellsa. H1 hESCs stably expressing control- or KBTBD8-shRNAs were subjected to neural conversion for 43 days and analyzed by immunofluorescence microscopy against GFAP (glia), smooth muscle actin (SMA; mesenchymal cells), and neurofilament L (neurons). b. Control H1 hESCs or hESCs depleted of KBTBD8 were subjected to neural conversion for 43 days and expression of markers for glia (GFAP), mesenchyme (smooth muscle actin, SMA), melanocytes (TYRP1, DCT), chondrocytes (COL2A1), or CNS derivatives (PAX6, NESTIN, neurofilament L) was analyzed by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). c. X. tropicalis embryos were injected at the two-cell stage with splice-blocking morpholinos (sMO) against CUL3 or KBTBD8, or with a dominant-negative construct of CUL3 that allows KBTBD8 to bind, but not ubiquitylate substrates. Neural crest formation was monitored by SOX10 in situ hybridization. Quantification included experiment shown in Fig. 1d (mean of 3 biological replicates +/− s.d; ~20 embryos per condition and replicate).
	Extended Figure 5. Biochemical characterization of the substrate adaptor role of KBTBD8a. Domain structure of KBTBD8, including the residues mutated to generate ubiquitylation- (Y74A) and substrate-binding-deficient KBTBD8 (F550A; W579A). b. Effects of point mutations in predicted KELCH domain loops on binding of KBTBD8 to candidate substrates were determined by affinity-purification and Western. c. Effects of point mutations in BTB domain on binding of KBTBD8 to CUL3 were determined by affinity-purification and Western. Dimerization of FLAGKBTBD8 with KBTBD8HA was analyzed in the same experiment to provide a folding control. d. Binding of recombinant CUL3 to immobilized recombinant MBPKBTBD8 variants was analyzed by Coomassie. e. Binding of in vitro-transcribed/translated 36S-NOLC1 to immobilized recombinant KBTBD8 variants was analyzed by autoradiography. f. Binding of in vitro-transcribed/translated 36S-TCOF1 to immobilized recombinant KBTBD8 variants was analyzed by autoradiography. g. Endogenous β-arrestin proteins in reticulocyte lysates binds immobilized, recombinant KBTBD8, as detected by Western. h. 293T cells were transfected with control- or β-arrestin 1/2-siRNAs and reconstituted with FLAGKBTBD8. Binding of KBTBD8 to endogenous TCOF1 and NOLC1 was analyzed by αFLAG-affinity purification and Western. i. Ubiquitylation of HATCOF1 in 293T cells depleted of β-arrestin 1/2 and reconstituted with KBTBD8 was determined after denaturing NiNTA-purification by Western blotting as described above. i. Ubiquitylation of HANOLC1 was detected in 293T cells depleted of β-arrestins and reconstituted with KBTBD8, as described above.
	Extended Data Figure 6. KBTBD8 specifies neural crest fate through TCOF1 and NOLC1a. mRNA levels of KBTBD8, NOLC1, and TCOF1 were determined in hESCs or differentiating cells transduced with lentiviruses expressing the indicated shRNAs by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). b. hESCs stably depleted of KBTBD8 and reconstituted with either wt-KBTBD8, KBTBD8W579A, or KBTBD8Y74A were subjected to neural conversion (9d) and analyzed for the expression of marker proteins by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). c. hESCs stably depleted of KBTBD8, TCOF1, or NOLC1 were subjected to neural conversion (9d) and analyzed for marker expression by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). d. Depletion of TCOF1 or NOCL1 from hESCs results in loss of neural crest cells, as determined by triple staining immunofluorescence against the neural crest markers HNK1, TFAP2, and p75 (n>200 cells, mean of 3 biological replicates +/− s.d.). e. hESCs were transduced with lentiviruses expressing control- or BRD2-hRNAs, subjected to puromycin selection for 7d, and analyzed by Western. f. Depletion efficiency for shRNAs against various KBTBD8 binding partners, as determined by qRT-PCR (mean of 3 technical replicates +/− s.e.m.).
	Extended Data Figure 7. Characterization of TCOF1-regulation by ubiquitylationa. hESCs depleted of either KBTBD8 or TCOF1 were subjected to neural conversion and analyzed for expression of indicated proteins by Western. b. Control or KBTBD8-depleted hESCs were fixed and subjected to indirect immunofluorescence analysis against endogenous TCOF1 or NOLC1. Scale bar = 10 μm. c. Total spectral counts of proteins associated with TCOF1/NOLC1-complexes purified by sequential immunoprecipitation in the presence of KBTBD8 compared to single TCOF1 affinity-purification in the absence of KBTBD8, as determined by mass spectrometry (sum of 3 biological replicates) d. 293T cells were reconstituted with CUL3KBTBD8 and depleted of TCOF1 and NOLC1 by siRNAs. Endogenous RNA polymerase I was immunoprecipitated and binding to the SSU processome (NOP58, CSK2A) was analyzed by Western.
	Extended Data Figure S8. KBTBD8 is not required for general ribosome biogenesisa. hESCs stably depleted of KBTBD8 were subjected to neural conversion and levels of 5S rRNA, 18S rRNA, and mRNAs encoding RPS6, RPS28, RPL10A, and RPL28 were measured by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). b. hESCs stably depleted of KBTBD8 were subjected to neural conversion, and total RNA was subjected to a bioanalyzer assay to monitor processing of ribosomal RNAs. c. hESCs stably depleted of KBTBD8 were subjected to neural conversion (3d), and nucleoli were analyzed by αfibrillarin immunofluorescence microscopy. d. Quantification of nucleolar analysis described above (mean of 3 technical replicates +/− s.e.m.). e. hESCs stably depleted of KBTBD8 were analyzed for localization of 5.8S rRNA by α5.8S rRNA immunofluorescence microscopy. f. hESCs depleted of KBTBD8 were subjected to neural conversion and analyzed by α5.8S rRNA immunofluorescence microscopy. g. Polysomes were purified from control or KBTBD8-depleted hESCs and differentiated counterparts subjected to neural conversion via sucrose gradient centrifugation followed by fractionation and UV detection. h. KBTBD8-depleted hESCs were subjected to neural conversion for 9 days and analyzed for apoptosis by immunofluorescence analysis against cleaved caspase 3 (red) and DNA (Hoechst, blue). Cells with active caspase 3 staining were quantified. (~200 cells/condition; scale bar 10μm)
	Extended Data Figure 9. Characterization of KBTBD8- and TCOF1-depleted hESCs during neural conversiona. hESCs were treated with increasing concentrations of rapamycin, subjected to neural conversion for 9 days, and analyzed for expression of neural crest or CNS precursor markers by qRT-PCR. For comparison, effects of KBTBD8, TCOF1, or NOLC1 depletion (extracted from Fig. 3a) are shown. b. hESCs were treated with rapamycin, subjected to neural conversion, and analyzed for marker expression by Western blotting. c. hESCs were depleted of KBTBD8 or TCOF1, subjected to neural conversion for 3 days, and analyzed for expression of 5S and 18S rRNA by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). d. hESCs depleted of KBTBD8 or TCOF1 were subjected to neural conversion for 3 days and analyzed for p53 activation by RNA seq against p53 targets. e. hESCs were depleted of KBTBD8 or TCOF1, subjected to neural conversion for 3 days, and analyzed for apoptosis by immunofluorescence microscopy against cleaved caspase 3. Quantification shown below (~200 cells/condition). f. hESCs depleted of KBTBD8 were subjected to neural conversion for 9 days and analyzed for expression levels of 5S and 18S rRNA by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). g. hESCs stably depleted of NOLC1 or TCOF1 were subjected to neural conversion for 9 days and analyzed by immunofluorescence microscopy against cleaved caspase 3 (red) or DNA (Hoechst, blue). Quantification is shown below (~200 cells/condition; scale bar 10μm).
	Extended Data Figure 10. KBTBD8 controls translationa. hESCs stably depleted of KBTBD8 were subjected to neural conversion for 3 days, and hESCs and differentiating cells were analyzed by RNA deep sequencing and ribosomal profiling to determine translation efficiency. Distribution of translation efficiency changes for 7725 mRNAs brought about by KBTBD8 depletion is shown. b. hESCs stably depleted of either TCOF1 or KBTBD8 were subjected to neural conversion for 3 days, and translation efficiency was determined by RNAseq and ribosome profiling. c. Translation efficiency blot of differentiating hESCs transduced with control- or KBTBD8-shRNAs were labeled for significantly affected transcripts in general (blue), with links to CNS precursor formation (gold), or with links to neural crest formation (green). d. hESCs stably depleted of KBTBD8 or TCOF1 were subjected to neural conversion for 3 days, and expression levels of indicated proteins were analyzed by Western. e. hESCs stably depleted of KBTBD8 were subjected to neural conversion for 3 days, and levels of ATRX1 and PCM1 mRNA were determined by qRT-PCR (mean of 3 technical replicates +/− s.e.m.). f. hESCs stably depleted of KBTBD8 were subjected to neural conversion for 3 days, and protein stability of ATRX1 and PCM1 was determined by cycloheximide chase and Western (mean of 3 biological replicates +/− s.d., ATRX1 and PCM1 levels were normalized relative to actin levels and 0h time point set to 100%).
	Figure 1. CUL3KBTBD8 drives neural crest specificationa. hESCs stably depleted of KBTBD8 were subjected to neural conversion and analyzed by qRT-PCR. (mean of 3 technical replicates, +/− s.e.m) b. Depletion of KBTBD8 results in loss of neural crest cells, as determined by Western analysis (full scans in Supplementary Fig. 1). c. KBTBD8-depleted hESCs were subjected to neural conversion and analyzed by immunofluorescence microscopy (mean of 3 biological replicates, +/− s.e.m; ~1500 cells/condition). d. X. tropicalis embryos injected with translation-blocking morpholinos against KBTBD8 were analyzed by in situ-hybridization. e. Model of the CUL3KBTBD8-controlled developmental switch.
	Figure 2. CUL3KBTBD8 monoubiquitylates TCOF1 and NOLC1a. High-confidence interactors of wt- or mutant KBTBD8. Left: normalized TSCs per interactor of wt-KBTBD8 (sum of 3 biological replicates/condition). Right: heatmap depicting binding relative to wt-KBTBD8. b. Verification of KBTBD8 interactions in 293T cells by αFLAG-immunoprecipitation and Western. c. Immunoprecipitation of KBTBD8 from hESCs (full scans in Supplementary Fig. 1). d. Ubiquitylated HATCOF1 detected after denaturing Ni-NTA purification in 293T cells reconstituted with KBTBD8 variants e. Monoubiquitylation of HANOLC1 by CUL3KBTBD8 in 293T cells. f. Monoubiquitylation of endogenous TCOF1 and NOLC1 in 293T cells reconstituted with KBTBD8 variants and HISubiquitinL73P.
	Figure 3. CUL3KBTBD8 controls neural crest specification through TCOF1- and NOLC1a. hESCs were reconstituted with shRNA-resistant KBTBD8 variants or depleted of KBTBD8-binding partners; subjected to neural conversion (9d); and analyzed by qRT-PCR and unsupervised clustering. b. Protein expression during neural conversion of hESCs reconstituted with shRNA-resistant KBTBD8 variants (full scans in Supplementary Fig. 1). c. Protein expression in hESCs stably depleted of KBTBD8, TCOF1, or NOLC1 and subjected to neural conversion. d. hESCs were stably depleted of the indicated combinations of KBTBD8, TCOF1, or NOLC1, subjected to neural conversion (9d), and analyzed by qRT-PCR.
	Figure 4. Ubiquitylation-dependent TCOF1/NOLC1-complexes couple RNA polymerase I to ribosome modification enzymesa. Interactors of TCOF1 in 293T cells reconstituted with KBTBD8 or KBTBD8Y74A (sum of 3 biological replicates/condition). b. Validation of CUL3KBTBD8-dependent formation of TCOF1/NOLC1 complexes. c. CompPASS mass spectrometry analysis of sequential immunoprecipitation of FLAGTCOF1/HANOLC1-complexes. d. Validation of sequential affinity-purification of KBTBD8-dependent TCOF1/NOLC1-complexes (full scans in Supplementary Figure 1). e. Immunoprecipitation of RNA polymerase I from 293T cells reconstituted with KBTBD8 variants. f. Immunoprecipitation of RNA polymerase I from hESCs depleted of KBTBD8. g. Model of ubiquitin-dependent formation of a TCOF1/NOLC1-platform.
	Figure 5. CUL3KBTBD8 remodels translational programs during differentiationa. 36S-methionine incorporation into newly synthesized proteins in hESCs lacking KBTBD8. (3 biological replicates +/− s.e.m.) b. RNA seq and ribosomal profiling of hESCs lacking KBTBD8 (blue: transcripts significantly changed in ribosome footprints but not mRNA; n>7725, q<0.1). c. Translation efficiency of ATRX1 and PCM1 in hESCs lacking KBTBD8. Orange: translation efficiency change; grey inserts: mRNA abundance change (mean of two biological replicates). d. Protein expression in hESCs lacking KBTBD8 during neural conversion (full scans in Supplementary Figure 1). e. Model of CUL3KBTBD8-dependent neural crest specification.

References [+] :

, Positional cloning of a gene involved in the pathogenesis of Treacher Collins syndrome. The Treacher Collins Syndrome Collaborative Group. 1996, Pubmed