The super elongation complex (SEC) mediates phase transition of SPT5 during transcriptional pause release

Abstract

Release of promoter-proximally paused RNA Pol II into elongation is a tightly regulated and rate-limiting step in metazoan gene transcription. However, the biophysical mechanism underlying pause release remains unclear. Here, we demonstrate that the pausing and elongation regulator SPT5 undergoes phase transition during transcriptional pause release. SPT5 per se is prone to form clusters. The disordered domain in SPT5 is required for pause release and gene activation. During early elongation, the super elongation complex (SEC) induces SPT5 transition into elongation droplets. Depletion of SEC increases SPT5 pausing clusters. Furthermore, disease-associated SEC mutations impair phase properties of elongation droplets and transcription. Our study suggests that SEC-mediated SPT5 phase transition might be essential for pause release and early elongation and that aberrant phase properties could contribute to transcription abnormality in diseases.

Keywords: SPT5; paused Pol II; phase separation; super elongation complex (SEC); transcription elongation.

Figure EV1. Pausing factors DSIF and NELF form clusters

1. Confocal images showing co‐localization of Pol II‐S5P with NELFA and SPT5 in nuclear puncta in HCT 116 cells in serum starvation condition. Results shown are representative images from three independent experiments.

2. Line intensity profiling of representative sections of cells in Fig EV1A. is indicated by yellow dashed lines.

3. Confocal images showing localization of MED1 with NELFA and SPT5 in nuclear puncta in HCT 116 cells in serum starvation condition. Representative images from three independent experiments.

4. Line intensity profiling of representative sections of cells in Fig EV1C. is indicated by yellow dashed lines.

5. Boxplots showing the mean values of the Pearson correlation coefficient of co‐localization ratios. Central band is median; boxes represent 1st and 3rd quartile (25th and 75th percentile, respectively) and whiskers 1.5× interquartile range. Results are representative of three biological replicates, each n > 20.

6. Confocal images showing SPT5, NELFE, NELFA, Pol II‐S5P, and Pol II‐S2P in nuclear puncta in HCT 116 cells in serum starvation and serum treatment conditions.

Figure 1. NELF and DSIF form stable clusters at the paused Pol II genes

1. A, B

   (A) Confocal imaging of FOS DNA FISH with concurrent SPT5 and Pol II‐S5P IF showing that SPT5 and Pol II‐S5P occupied the FOS loci after serum starvation. Zoomed‐in views of the merged regions are indicated by the white arrow. (B) Same as Fig 1A. but for FOS DNA FISH with concurrent SPT5 and NELFE IF.

2. C

   Percentage of FISH foci overlap with IF puncta in Fig 1A and B. Results are plotted with four technical repeats from two biological replicates. Error bars represent standard deviations.

3. D

   Violin plot showing the number of cells containing nuclear puncta in Fig EV1F. Results are representative of three biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. For Pol II‐S5P, **P = 0.0034. For Pol II‐S2P, ***P = 0.0004. For SPT5, P = 0.8631. For NELFE, **P = 0.0015. For NELFA, ***P < 0.0001.

4. E

   Confocal images showing that SPT5, NELFA, Pol II‐S5P, and AFF4 formed nuclear puncta in HCT 116 cells before and after treated with 1% 1,6‐hexanediol (abbreviation as 1,6‐hex) for 30 min.

5. F

   Violin plot showing the number of cells containing nuclear puncta in Fig 1E. Results are representative of three biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. For SPT5, P = 0.1301. For NELFA, P = 0.2756. For Pol II‐S5P, P = 0.8179. For AFF4, ***P < 0.0001.

6. G–I

   ChIP‐qPCR showing that the occupancies of Pol II (G), SPT5 (H), and NELFE (I) at the promoters and gene bodies of HSP70, GAPDH, and MYC in HeLa cells. The HEMO gene serves as a negative control for ChIP‐qPCR. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. For Pol II ChIP, HSP70_promoter, **P = 0.0014; GAPDH_promoter, P = 0.8385; GAPDH_gene body, ***P < 0.0001; MYC_promoter, P = 0.1765; MYC_gene body, ***P = 0.0002. For SPT5 ChIP, HSP70_promoter, **P = 0.0033; GAPDH_promoter, P = 0.9603; GAPDH_gene body, **P = 0.0012; MYC_promoter, P = 0.9898; MYC_gene body, ***P < 0.0001. For NELFE ChIP, HSP70_promoter, P = 0.3913; GAPDH_promoter, P = 0.9347; MYC_promoter, P = 0.8558.

Figure 2. SPT5 carries an intrinsic property to form clusters

1. An NCPR (net charge per residue) analyses (top) of SPT5. The scores are assigned between −1 and 1. The score above 0 indicates positive charge (blue), while the score below 0 indicates negative charge (red). The intrinsic disorder calculated by PONDR (Predictor of Natural Disordered Regions) VSL2 algorithm (bottom). The intrinsic scores are shown on the y‐axis, and amino acid positions are shown on the x‐axis. The scores are assigned between 0 and 1, and the score above 0.5 indicates disorder. The gray regions represent the indicated IDRs under investigation in this study.

2. Fluorescence microscopy images showing that FLAG affinity purified NELFE and SPT5 formed core‐shell‐like clusters in 100 mM NaCl‐containing buffer.

3. Normalized FRAP curves for eGFP‐SPT5. The bleaching events occurred at 0 s. Confocal images were embedded within the plot showing FRAP of the in vitro eGFP‐SPT5 clusters. Results shown are from three biological replicates. The arrowhead represents the bleached region.

4. Domain organization of SPT5. NTR, N‐terminal IDR; NGN, NusG N‐terminal domain; KOW, Kyprides, Ouzounis, Woese domains; CTR, C‐terminal IDR. NLS, nuclear localization signal. The residuals phosphorylated by P‐TEFb and the Pol II interacting regions in the SPT5 sequence are indicated.

5. Live cell imaging of HEK‐293 T cells expressing eGFP‐SPT5 together with mCherry‐SPT5, mCherry‐SPT5_ΔNTR, or mCherry‐SPT5_ΔCTR close to endogenous level.

6. Violin plot showing the clusters' area (top) and circularity (bottom) of SPT5 with SPT5, SPT5_ΔNTR, or SPT5_ΔCTR in Fig 2E. Each n > 20. Results are representative of three biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. ***P < 0.001.

7. Normalized FRAP curves for eGFP‐SPT5 (green) and mCherry‐SPT5 (purple) in the SPT5 homotypic clusters, while eGFP‐SPT5 (yellow) and mCherry‐SPT5_ΔNTR (blue) in the SPT5/SPT5_ΔNTR heterotypic aggregates. The bleaching events occurred at 0 s. Results shown are from three biological replicates.

8. Confocal images showing FRAP analysis of the NELFA‐IDR‐eGFP and SPT5‐CTR‐mCherry clusters in vitro. The arrowhead represents the bleached and recovered cluster.

9. Normalized FRAP curves for NELFA‐IDR‐eGFP and SPT5‐CTR‐mCherry clusters. The bleaching events occurred at 0 s. Results shown are from two biological replicates.

Figure EV2. SPT5 carries an intrinsic property to form clusters

1. A

   Dot plot showing the percentage of intrinsic disorder regions of the DSIF subunits, SPT5 and SPT4, calculated by different algorithms including IUPred, VLXT, XL1_XT, VL3, and VSL2. The gray regions represent the percentage of residuals greater than 0.4.

2. B, C

   HEK‐293 T cell expressing FLAG‐tagged eGFP, eGFP‐SPT5, and eGFP‐NELFE and co‐expressing eGFP, mCherry or eGFP‐NELFE, and mCherry‐SPT5, the recombinant proteins were purified using FLAG affinity purification and analyzed by silver staining (B) and SDS–PAGE (C). Arrows indicate FLAG‐tagged proteins.

3. D

   Fluorescence microscopy images showing the pausing factors NELFE and SPT5 clusters in 100 mM NaCl‐containing buffer.

4. E

   Time‐lapse fluorescence images showing that the homotypic eGFP‐SPT5 clusters slowly fused upon contact. The image was subjected to illuminate every 30 s for the times indicated. The two SPT5 clusters that underwent spontaneous contact are indicated by arrows.

5. F

   Western analyses showing expression of mCherry‐SPT5, mCherry‐SPT5_ΔNTR, or mCherry‐SPT5_ΔCTR. Results are representative of two biological replicates.

6. G

   Live cell imaging of HEK‐293 T cells expressing mCherry‐SPT5, mCherry‐SPT5_ΔNTR, or mCherry‐SPT5_ΔCTR.

7. H

   Western analyses showing co‐expression of eGFP‐SPT5 with mCherry‐SPT5, mCherry‐SPT5_ΔNTR, or mCherry‐SPT5_ΔCTR in Fig 2E. Results are representative of two biological replicates.

8. I

   I HEK‐293 T cells expressing SPT5, SPT5_ΔNTR, or SPT5_ΔCTR after SPT5 depletion; and RPL13, RPS8 mRNA levels were assessed by RT–qPCR and normalized to U6. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. *P < 0.05, **P < 0.01, ***P < 0.01, n.s., not significant. For RPL13, P‐value (from left to right, 0.0144, 0.0165, 0.0040, 0.0346, 0.5948, 0.8402, 0.4175, 0.6308); For RPS8, P‐value (from left to right, 0.0167, 0.0099, 0.0222, 0.0152, 0.3870, 0.3798, 0.7867, 0.5143).

9. J

   ChIP‐qPCR showing that the occupancies of Pol II at the promoters and gene bodies of HSP70 in HEK‐293 T cells expressing SPT5, SPT5_ΔNTR, or SPT5_ΔCTR after SPT5 depletion. The HEMO gene serves as a negative control for ChIP‐qPCR. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. *P < 0.05, **P < 0.01, n.s., not significant. For HSP70_promoter, P‐value (from left to right, 0.0170, 0.0138, 0.0019, 0.0057, 0.6895, 0.7905, 0.5126, 0.9187).

Figure 3. Intrinsic disordered regions of SPT5 regulate Pol II pause release

1. A

   Western blot analysis of re‐expression of mCherry‐SPT5, mCherry‐SPT5_ΔNTR, or mCherry‐SPT5_ΔCTR after SPT5 depletion. α‐Tubulin was used as a loading control. Results are representative of two biological replicates.

2. B

   Confocal imaging of nascent RNA labeled by EU in HEK‐293 T cells expressing SPT5, SPT5_ΔNTR, or SPT5_ΔCTR after SPT5 depletion.

3. C

   Violin plot showing the intensity of cells in Fig 3B. Results are representative of two biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. ***P < 0.001.

4. D, E

   GAPDH/MYC (D) RPL10/RPL29/RPS19 (E) mRNA levels were assessed by RT–qPCR and normalized to U6 snRNA in HEK‐293 T cells expressing SPT5, SPT5_ΔNTR, or SPT5_ΔCTR after SPT5 depletion. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant. For GAPDH, P‐value (from left to right, <0.0001, 0.0002, <0.0001, 0.0025, 0.3390, 0.8664, 0.4291, 0.7180); For MYC, P‐value (from left to right, 0.0001, <0.0001, 0.0002, <0.0001, 0.9656, 0.3931, 0.7182, 0.4345); For RPL10, P‐value (from left to right, 0.0028, 0.0039, 0.0024, 0.0098, 0.6381, 0.9173, 0.4390, 0.8740); For RPL29, P‐value (from left to right, 0.0002, 0.0002, 0.0046, 0.0021, 0.6530, 0.8222, 0.6608, 0.6877); For RPS19, P‐value (from left to right, 0.0038, 0.0056, 0.0073, 0.0111, 0.4555, 0.5773, 0.8044, 0.8576).

5. F

   HEK‐293 T cells expressing SPT5, SPT5_ΔNTR, or SPT5_ΔCTR after SPT5 depletion were heat shocked for 2 h or not; and HSP70 mRNA levels were assessed by RT–qPCR and normalized to U6. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. **P < 0.01, ***P < 0.001, n.s., not significant. For nonheat shock, P‐value (from left to right, 0.0043, 0.0031, 0.0009, 0.0030, 0.3069, 0.4297, 0.7652, 0.5788); for heat shock, P‐value (from left to right, 0.0016, 0.0011, 0.0004, <0.0001, 0.0600, 0.3142, 0.1750, 0.1610).

6. G, H

   ChIP‐qPCR showing the occupancies of Pol II at the promoters and gene bodies of GAPDH (G), and MYC (H) in HEK‐293 T cells expressing SPT5, SPT5_ΔNTR, or SPT5_ΔCTR after SPT5 depletion. The HEMO gene serves as a negative control for ChIP‐qPCR. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. *P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant. For GAPDH_promoter, P‐value (from left to right, <0.0001, <0.0001, <0.0001, 0.0044, 0.8873, 0.9310, 0.6855, 0.9615); GAPDH_gene body, P‐value (from left to right, 0.0370, 0.0214, 0.0574, 0.0256, 0.9536, 0.8038, 0.7623, 0.8773); For MYC_promoter, P‐value (from left to right, 0.0002, 0.0014, 0.0004, 0.0048, 0.0670, 0.7501, 0.3482, 0.3284); MYC_gene body, P‐value (from left to right, 0.0004, 0.0013, 0.0081, 0.0299, 0.1332, 0.5839, 0.4404, 0.6170).

Figure EV3. Phase transition from the pausing clusters to the liquid‐like elongation droplets

1. A

   Confocal imaging (left) of HeLa cells expressing eGFP‐AFF4, eGFP‐ENL, and eGFP‐SPT5. Quantitative analyses (right) showing the percentage of the cells containing the phase‐separated droplets that indicated in the left. Central band is median; boxes represent 1st and 3rd quartile (25th and 75th percentile, respectively) and whiskers 1.5× interquartile range.

2. B

   Western blot analysis showing the expression of eGFP‐SPT5 after transfection with different amounts of plasmids (top). Live cell imaging for indicated concentration of proteins was shown (bottom).

3. C

   Western blot analysis showing the expression of related proteins after transfection with different amounts of plasmids.

4. D–F

   Live cell imaging of HeLa cells expressing AFF4/SPT5 (D), AFF4/NELFA (E, top), AFF4/TBP (E, bottom), ENL/SPT5 (F, top), and CCNT1/SPT5 (F, bottom). The arrowhead represents the shell‐like clusters.

5. G

   Live cell imaging of HeLa cells co‐transfected with eCFP‐SPT5, mCherry‐NELFE and increasing amount of eGFP‐AFF4. The concentration ratios are indicated as C_AFF4/C_{(SPT5 + NELFE)}.

6. H

   Graph plots (left) showing the intrinsic disorder regions across the entire length of AFF4 calculated by PONDR VSL2 algorithm. The intrinsic scores are shown on the y‐axis, and amino acid positions are shown on the x‐axis. The scores are assigned between 0 and 1, and the score above 0.5 indicates disorder regions. The gray rectangle represents previously studied IDR (Guo et al, 2020a). Live cell imaging (right) of HeLa cells expressing eGFP‐AFF4_ΔIDR. The deletion region is indicated as gray rectangle.

7. I

   Live cell imaging of HeLa cells co‐transfected with eCFP‐SPT5, mCherry‐NELFE, and increasing amounts of eGFP‐AFF4_ΔIDR. The concentration ratio is indicated as C_AFF4ΔIDR/C_{(SPT5 + NELFE)}.

8. J

   Curve showing the percentage of cell with droplet (top), average droplet number per cell (mid), and average droplet circularity (bottom) per nucleus in Fig EV3G and I. Error bars represent standard deviations. Results are representative of three biological replicates.

9. K

   The recombinant SPT5‐CTR‐mCherry and AFF4‐IDR‐eGFP proteins were purified and analyzed by SDS–PAGE followed by Coomassie blue staining. The arrowhead represents the purified protein.

Figure 4. SPT5 relocates from the pausing clusters to the liquid‐like elongation droplets

1. Fluorescence microscopy images showing the phase‐separated droplets formed in 50 mM NaCl‐containing buffer with 2 μM NELFA‐IDR‐eGFP and 10 μM SPT5‐CTR‐mCherry and 2 μM BFP or AFF4‐IDR‐BFP in the presence of 10% PEG‐8000.

2. Fluorescence microscopy images showing the phase‐separated AFF4‐IDR‐eGFP and SPT5‐CTR‐mCherry droplets formed in 50 mM NaCl‐containing buffer in the presence of 10% PEG‐8000. The concentration ratios are indicated as C_AFF4‐IDR/C_SPT5‐CTR.

3. Dot plot showing the droplet area in Fig 4B. Fields per condition n = 5.

4. Live cell confocal images showing FRAP of the heterotypic eGFP‐AFF4 + ENL and mCherry‐SPT5 droplets. The arrowhead represents the bleached and recovered droplet.

5. Normalized FRAP curves for eGFP‐AFF4 + ENL (green) and mCherry‐SPT5 (purple) in the AFF4/ENL/SPT5 heterotypic droplets. The bleaching events occurred at 0 s. Results shown are from three biological replicates.

6. Time‐lapse fluorescence images showing that the heterotypic AFF4‐IDR‐eGFP and SPT5‐CTR‐mCherry droplets rapidly fused into one spherical droplet upon contact. The images were subjected to illuminate every 2 s. The droplet formation buffer contained 10% PEG‐8000 and 50 mM NaCl.

7. Time‐lapse fluorescence images of the nucleus of the HeLa cell expressing eGFP‐AFF4, eGFP‐ENL, and mCherry‐SPT5 subjected to illuminate every 1 min for the times indicated. The SEC and SPT5 heterotypic droplets underwent spontaneous fusion as indicated by arrows.

Source data are available online for this figure.

Figure 5. Disruption of the SEC clusters increases the SPT5‐containing pausing clusters

1. A

   Confocal imaging of FOS RNA FISH with concurrent SPT5 and AFF4 IF showing that SPT5 and AFF4 co‐occupied the FOS loci after serum treatment. Zoomed‐in views of the merged regions are indicated by the white arrow.

2. B

   The three columns showing average FOS FISH signal and average SPT5 and AFF4 immunofluorescence signal in Fig 5A. centered on the FISH foci (see Methods). Data shown were analyzed from at least four biological replicates.

3. C

   Confocal imaging of FOS RNA FISH with concurrent SPT5 IF in wild‐type (WT) and ENL KO HCT116 cells after serum treatment. Zoomed‐in views of the merged regions are indicated by the white arrow.

4. D

   The two columns showing average FOS FISH signal and average SPT5 immunofluorescence signal in Fig 5C. centered on the FISH foci (see Methods). Data shown were analyzed from at least four biological replicates.

5. E

   Confocal imaging of nascent RNA with EU staining in WT and ENL KO HCT116 cells.

6. F

   Violin plot showing the intensity of cells in Fig 5E. Results are representative of two biological replicates, each n > 70. Two‐tailed, unpaired Student's t‐test was performed. ***P < 0.001.

7. G

   Western blot analysis of AFF4, SPT5, and ENL in HCT 116 cells after KL‐1 treatment. α‐Tubulin was used as a loading control.

8. H

   Confocal images showing SPT5 in nuclear puncta in WT or ENL KO cells treatment with KL‐1.

9. I

   Violin plot showing the number of cells containing nuclear puncta in Fig 5H. Results are representative of three biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. ***P < 0.001.

10. J–M

    (J) Confocal images showing co‐localization of Pol II‐S5P with SPT5 in nuclear puncta in KL‐1 treated HCT 116 cells. Results shown are representative images from three independent experiments. (K) Boxplots showing the mean values of the Pearson correlation coefficient of co‐localization ratios of SPT5 and Pol II‐S5P in Fig 5J. Central band is median; boxes represent 1st and 3rd quartile (25th and 75th percentile, respectively) and whiskers 1.5× interquartile range. Results are representative of three biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. KL‐1 vs. DMSO, **P = 0.0050. (L) Same as Fig 5J. but for SPT5 and NELFA. (M) Same as Fig 5K. but for SPT5 and NELFA. Two‐tailed, unpaired Student's t‐test was performed. KL‐1 vs. DMSO, *P = 0.0100.

Figure EV4

SEC promotes SPT5 relocation into elongation droplets.

1. Confocal imaging (top) of FOS RNA FISH with concurrent SPT5 and AFF4 IF after serum starvation. Confocal imaging (mid) or (bottom) was same as (top), but for AFF4 or SPT5 and Pol II‐S2P.

2. Confocal imaging (top) of FOS RNA FISH with concurrent AFF4 and Pol II‐S2P IF showing that AFF4 and Pol II‐S2P co‐occupied the FOS loci after serum treatment. Zoomed‐in views of the merged regions are indicated by the white arrow. Confocal imaging (bottom) was same as (top), but for SPT5 and Pol II‐S2P.

3. The three columns showing average FOS FISH signal and average AFF4, SPT5, and Pol II‐S2P immunofluorescence signal in Fig EV4B. centered on the FISH foci (see Methods). Data shown were analyzed from at least four biological replicates.

4. Average FISH signals and immunofluorescence signals centered on the random are shown as a control (Figs 5A and EV4B; see Methods). Data shown were analyzed from at least four biological replicates.

5. RT–qPCR analysis showing RNA levels of FOS, JUNB, EGR2 in HCT 116, or ENL KO cells before or after serum stimulation. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. For normal state, ENL KO vs. WT, FOS, P = 0.3638; JUNB, P = 0.1549; EGR2, P = 0.4051. For serum stimulation, ENL KO vs. WT, FOS, **P = 0.0020; JUNB, *P = 0.0109; EGR2, *P = 0.0258.

6. Western blot analysis of SPT5 in HCT 116 or ENL KO cells. α‐Tubulin was used as a loading control.

7. Average FISH signals and immunofluorescence signals centered on the random are shown as a control (Fig 5C) (see Methods). Data shown were analyzed from at least four biological replicates.

8. Confocal images showing AFF4 (top) or ENL (bottom) in nuclear puncta in KL‐1 treated HCT 116 or ENL KO cells.

9. Violin plot showing nuclear puncta of cells in Fig EV4H. Results are representative of three biological replicates, each n > 20. Two‐tailed, unpaired Student's t‐test was performed. ***P < 0.001, n.s., not significant.

10. Confocal imaging of FOS RNA FISH with concurrent SPT5 and PAF1 (left) or SF3B1 (right) IF showing that SPT5 and PAF1 or SF3B1 co‐occupied the FOS loci after serum treatment. Zoomed‐in views of the merged regions are indicated by the white arrow.

Figure 6. SPT5 phosphorylation promotes the incorporation of SPT5 into the elongation droplets

1. The recombinant SPT5‐CTR‐eGFP proteins were purified and analyzed by SDS–PAGE followed by Coomassie blue staining. The arrowhead represents the purified protein.

2. Fluorescence microscopy images showing that 2 μM SPT5‐CTR‐eGFP formed phase‐separated droplets in the NaCl‐containing buffer.

3. Droplet number (top) and droplet area (bottom) analyses are shown. Central band is median; boxes represent 1^st and 3^rd quartile (25^th and 75^th percentile, respectively) and whiskers 1.5× interquartile range. Results are representative of two biological replicates, each n > 6. Two‐tailed, unpaired Student's t‐test was performed. For droplet number, SPT5‐CTR + P‐TEFb (− ATP) vs. SPT5‐CTR, **P = 0.0015. SPT5‐CTR + P‐TEFb (+ vs. − ATP), **P = 0.0032. For droplet area, SPT5‐CTR + P‐TEFb (+ vs. − ATP), ***P < 0.001.

4. The recombinant SPT5‐CTR‐mCherry and SPT5‐CTR‐TE‐mCherry proteins were purified and analyzed by SDS–PAGE followed by Coomassie blue staining. The arrowhead represents the purified protein.

5. Fluorescence microscopy images showing the phase‐separated droplets formed in 50 mM NaCl‐containing buffer with 10 μM AFF4‐IDR‐eGFP and 10 μM SPT5‐CTR‐mCherry or SPT5‐CTR‐TE‐mCherry in the presence of 10% PEG‐8000.

6. Droplet number (top) and droplet area (bottom) analyses are shown. Central band is median; boxes represent 1^st and 3^rd quartile (25^th and 75^th percentile, respectively) and whiskers 1.5× interquartile range. Results are representative of two biological replicates, each n > 6. Two‐tailed, unpaired Student's t‐test was performed. AFF4‐IDR + SPT5‐CTR‐TE vs. AFF4‐IDR + SPT5‐CTR, **P = 0.0029.

7. Fluorescence microscopy images showing the phase‐separated droplets formed in 50 mM NaCl‐containing buffer with 2 μM NELFA‐IDR‐eGFP and 10 μM SPT5‐CTR‐mCherry or SPT5‐CTR‐TE‐mCherry in the presence of 10% PEG‐8000.

Source data are available online for this figure.

Figure EV5. Disease‐causing AFF4 mutants promote phase separation of elongation factors

1. Sanger sequencing results of the AFF4 point mutation cell line.

2. RT–qPCR analysis showing RNA levels of AFF4 in wild‐type (WT) and R258W cells. Error bars represent standard deviations. Results are representative of two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. n.s., not significant. P = 0.8322.

3. Western blot analysis of AFF4, CDK9, and SPT5 in WT and R258W cells. α‐Tubulin was used as a loading control. Results are representative of three biological replicates.

4. The AFF4‐ALF‐mCherry, AFF4‐T254A‐mCherry, AFF4‐T254S‐mCherry, and AFF4‐R258W‐mCherry proteins were purified from recombinant BL21, and analyzed by SDS–PAGE followed by Coomassie blue staining. The arrowhead represents the purified protein.

5. Live cell imaging for indicated concentration of eGFP‐AFF4_mut proteins was shown.

6. Normalized FRAP curves for eGFP‐AFF4 + ENL (green/yellow) and mCherry‐SPT5 (purple/blue) in the heterotypic droplets in the presence or absence of FLAG‐MLL‐ENL. The bleaching events occurred at 0 s. Results shown are from three biological replicates.

7. Fluorescence microscopy images showing that the AFF4‐ALF‐mCherry, AFF4‐T254A‐mCherry, AFF4‐T254S‐mCherry, and AFF4‐R258W‐mCherry droplets in buffer containing 50 mM NaCl were sensitive to 3% 1,6‐hexanediol.

8. Fluorescence microscopy images showing that the AFF4 droplets in WT and R258W cells were sensitive to 1% 1,6‐hexanediol.

9. Confocal imaging (left) of nascent RNA with EU staining in WT and R258W cells. Violin plot (right) showing the intensity of cells. Results are representative of two biological replicates, each n > 50. Two‐tailed, unpaired Student's t‐test was performed. **P < 0.01.

10. ChIP‐qPCR showing the occupancies of Pol II (left), AFF4 (mid), and SPT5 (right) at the promoter of MYC in U937 (non‐MLL fusion) and MOLM‐13 (MLL‐AF9 fusion) cell lines. The HEMO gene serves as a negative control for ChIP‐qPCR. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐tests were performed. Pol II, ***P = 0.0002; AFF4, ***P = 0.0004; SPT5, **P = 0.0080.

Figure 7. Disease‐causing AFF4 mutations reduce droplet fluidity, but promote the accumulation of elongation factors at the target genes

1. A

   Confocal images showing co‐localization of AFF4 with SPT5 in nuclear puncta in AFF4 wild‐type (WT) or R258W cells.

2. B

   Dot plot (left) showing the AFF4 or SPT5 puncta area in AFF4 WT or R258W cells (Fig 7A). Two‐tailed, unpaired Student's t‐test was performed. AFF4 R258W vs. WT, *P = 0.0254. Violin plot (right) showing the Pearson correlation coefficient of co‐localization ratio (Fig 7A). Each n > 20. Results are representative of three biological replicates. Two‐tailed, unpaired Student's t‐test was performed. AFF4 R258W vs. WT, *P = 0.0166.

3. C

   Droplet experiments showing that CHOPS syndrome patient missense mutations reinforced AFF4‐ALF domain droplet formation.

4. D

   Dot plot showing the droplet areas in Fig 7C. Fields per condition n = 5. Two‐tailed, unpaired Student's t‐test was performed. T254S, T254A, or R258W vs. WT, *P < 0.05, **P < 0.01, ***P < 0.001.

5. E

   AFF4‐ALF‐mCherry condensed fraction as a function of AFF4‐ALF‐mCherry concentration for experiments in Fig 7C. Error bars represent standard deviations. Fields per condition n = 5. Two‐tailed, unpaired Student's t‐test was performed. T254S, T254A, or R258W vs. WT, *P < 0.05, **P < 0.01.

6. F

   Western blot analysis for the eGFP‐AFF4_mut was expressed in a concentration‐dependent manner. α‐Tubulin was used as a loading control.

7. G

   Normalized FRAP curves for eGFP‐AFF4 (green) and mCherry‐ENL (purple/blue) and eGFP‐AFF4_mut (yellow) in the AFF4/ENL and AFF4_mut/ENL heterotypic droplets. The bleaching events occurred at 0 s. Results shown are from three biological replicates.

8. H

   Normalized FRAP curves for eGFP‐AFF4 + ENL (green) and mCherry‐SPT5 (purple/blue) and eGFP‐AFF4_mut + ENL (yellow) in the AFF4/ENL/SPT5 and AFF4_mut/ENL/SPT5 heterotypic droplets. The bleaching events occurred at 0 s. Results shown are from three biological replicates.

9. I

   AFF4 WT and R258W HEK‐293 T cells were heat shocked by incubation at 42°C for indicated times (minutes), and HSP70 mRNA levels were assessed by RT–qPCR and normalized to GAPDH. Error bars represent standard deviations. Results are from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. For 90 min, AFF4 R258W vs. WT, **P = 0.0033. For 120 min, AFF4 R258W vs. WT, ***P = 0.0005.

10. J

    AFF4 WT and R258W HEK‐293 T cells were heat shocked by incubation at 42°C for indicated times (minutes), and R258W HEK‐293 T cells were treated with indicated concentrations of 1,6‐hexanediol (1,6‐hex) for 30 min, and HSP70 mRNA levels were assessed by RT–qPCR and normalized to GAPDH. Error bars represent standard deviations. Results are from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. For 90 min, AFF4 R258W + 1% 1,6‐hex vs. AFF4 R258W, **P = 0.0016; AFF4 R258W + 3% 1,6‐hex vs. AFF4 R258W, **P = 0.0011. For 120 min, AFF4 R258W + 1% 1,6‐hex vs. AFF4 R258W, ***P < 0.0001; AFF4 R258W + 3% 1,6‐hex vs. AFF4 R258W, ***P < 0.0001.

11. K–N

    ChIP‐qPCR showing the occupancies of Pol II (K), AFF4 (L), SPT5 (M), and PAF1 (N) at the promoter and gene body of HSP70 in AFF4 WT and R258W HEK‐293 T cells after heat shock for 2 h. The HEMO gene serves as a negative control for ChIP‐qPCR. Error bars represent standard error of means. Results are plotted with four technical repeats from two biological replicates. Two‐tailed, unpaired Student's t‐test was performed. For Pol II ChIP, HSP70_promoter, ***P < 0.0001; HSP70_gene body, ***P = 0.0002. For AFF4 ChIP, HSP70_promoter, ***P < 0.0001; HSP70_gene body, ***P < 0.0001. For SPT5 ChIP, HSP70_promoter, ***P < 0.0001; HSP70_gene body, **P = 0.0086. For PAF1 ChIP, HSP70_promoter, ***P = 0.0007; HSP70_gene body, **P = 0.0021.

Source data are available online for this figure.