Feeders facilitate telomere maintenance and chromosomal stability of embryonic stem cells

Feeder cells like mouse embryonic fibroblasts (MEFs) have been widely applied for culture of pluripotent stem cells, but their roles remain elusive. Noticeably, ESCs cultured on the feeders display transcriptional heterogeneity. We investigated roles of feeder cells by examining the telomere maintenance. Here we show that telomere is longer in mESCs cultured with than without the feeders. mESC cultures without MEF feeders exhibit telomere loss, chromosomal fusion, and aneuploidy with increasing passages. Notably, feeders facilitate heterogeneous transcription of 2-cell genes including Zscan4 and telomere elongation. Moreover, feeders produce Fstl1 that together with BMP4 periodically activate Zscan4. Interestingly, Zscan4 is repressed in mESCs cultured in 2i (inhibitors of Mek and Gsk3β signaling) media, associated with shorter telomeres and increased chromosome instability. These data suggest the important role of feeders in maintaining telomeres for long-term stable self-renewal and developmental pluripotency of mESCs.


Supplementary Figure 2 Elevated expression of Zscan4 and Tbx3 in ESCs
cultured with feeders. a, Histogram shows distribution of relative telomere length displayed as TFU by Q-FISH analysis of J1 and F1 ESCs. Green line indicates medium telomere length. Mean ± s.d. of telomere length is shown above each panel. Telomere length was measured in ESCs at passage 5 cultured under feeder or feeder-free conditions. Wilcoxon-Mann-Whitney rank sum test. b, Relative expression levels of Tert, Terc and c-Myc in -F and +F J1 ESCs by qPCR. c, Telomerase activity measured by TRAP assay. Lysis buffer served as negative control. d, Relative gene expression levels in -F and +F J1 ESCs. e, Flow cytometry quadrantal diagram indicates percentage and mean fluorescence intensity of Zscan4 + cells. f, Western blot analysis of Zscan4 protein levels. b-actin served as loading control. g, Expression of naive genes in -F and +F N33 ESCs. h, Western blot analysis of Tbx3. i, Immunofluorescence staining of Tbx3 in -F and +F ESCs. Scale bar, 10 µm. Mean ± s.e.m. from three independent experiments. *p<0.05, **p<0.01, ns, not significant (p>0.05). Student's t-test. -F, without feeders; +F, with feeders.

Supplementary Figure 3 Expression of pluripotency marker genes in
ESCs cultured with or without feeders. a, Immunofluorescence staining of pluripotency-associated markers Nanog and Oct4. N33 ESCs were cultured on feeders or feeder-free gelatin-coated, or E-cad-fc-coated plates for 8 passages. Scale bar, 100 µm. b, Expression of Oct4, Sox2, Nanog and Utf1 of N33 ESCs by qPCR. *p<0.05; ns, not significant (p>0.05), compared with ESCs cultured on feeders. ANOVA with Fisher's protected least-significant difference (PLSD) analysis. c, Immunofluorescence of Oct4, Nanog and SSEA1 under higher magnification. Scale bar, 20 µm. Figure 4 Signal pathways by RNA-seq data analysis. a, Cytokines (e.g. ILs and IFNs) and receptors (e.g. ILRs) involved in LIF pathway up-regulated in feeder+ESCs compared with feeder-free ESCs. Most of LIF targeting genes are not changed in their expression levels except for CycD. b, Cytokines, ECMs, some growth factors and their corresponding receptors upstream PI3K-Akt are activated in ESCs cultured with feeders. However, serine/threonine-protein phosphatase PP2A is also up-regulated, which may partly explain the decreased phosphorylation levels of Akt at Thr308 and Ser473 in feeder+ESCs. c, Both activating factors (e.g. Wnts) and inactivating ones (e.g. PEDF and Dkk) are up-regulated in feeder+ESCs compared with feeder-free ESCs, and expression of β-catenin not altered. d, Notch signal pathways. Notch targeting genes, Hes1 and Hes5 are up-regulated in feeder+ESCs. e, In Mek-ERK pathway, growth factors (e.g. EGFs and FGFs) and Ras are activated, however, MKPs (including DUSP1, DUSP9, DUSP10) also up-regulated in ESCs cultured with feeders. f, Levels of p-Erk1/2 and Erk1/2 protein in -F and +F ESCs. Two repeats. b-actin served as loading control. g, BMP-Smad pathway. BMP-Smad targeting Id genes mostly are activated in ESCs cultured with feeders. Figure 5 Signal transduction in ESCs cultured without (-F) or with feeders (+F). a, Schematic diagram of trans-well assay. b, Expression of Zscan4 by qPCR. c, Western blot analysis of Zscan4 protein levels. d, Expression level of Zscan4 is increased in feeder-conditioned medium (CM). +/-, CM was diluted with same volume of ESC medium. e, mRNA and protein levels (bottom panel) of Zscan4 in ESCs treated with LIF from 1000 U/ml (1×), 2000 U/ml (2×) to 3000 U/ml (3×) for 24 h and 48 h. f, LY294002 (10 µM) inhibits expression of Nanog and Zscan4, but not Oct4. g,h, Decreased phosphorylation levels of Akt at Thr308 (g) and Ser473 (h) in +F ESCs. i, Expression of Wnt targeting genes, Cdx1 and Axin2. j,k, Protein levels of total (j) and nuclear (k) β-catenin in -F and +F ESCs by Western blot analysis. Bottom panel, Quantity using ImageJ software of β-catenin protein levels relative to b-actin or H3 served as internal control. l, Relative expression level of Zscan4 by qPCR analysis following treatment of N33 ESCs with CHIR99021 (CHIR, 3 µM) for three passages. Data represent mean ± s.e.m. from three independent experiments. *p<0.05, **p<0.01, ***p < 0.001, ns, not significant (p>0.05). ANOVA with PLSD for b,d,e,l, and Student's t-test for i-k. Figure 6 Notch signaling and up-regulation of Zscan4 following knockdown of Notch4 by shRNAs. a-b, Relative expression level by qPCR analysis of Notch receptors (a) and ligands (b) in MEFs, and ESCs cultured with or without feeders. c, Expression of Zscan4 in ESCs after treatment with Jagged1 peptides. d, Expression by qPCR analysis of Zscan4 and Notch4 after Notch4 knockdown. e, Representative images of ESCs after transfection with a control shRNA construct (shGFP) or shRNAs targeting Notch4. Scale bar, 100 µm. f, Protein levels of Zscan4 and Histone H3 acetylation by Western blot. Data represent Mean ± s.e.m. from three independent experiments. *p<0.05, **p<0.01, ***p < 0.001, ns, not significant (p>0.05), compared between two selected groups for a and b, or compared with control groups for c and d. ANOVA with PLSD was used for the statistical analysis.

Supplementary Figure 7 Feeders decrease Dnmt3b expression in ESCs. a,
Expression of Dnmt3a and Dnmt3b by qPCR. b, Western blot analysis of Dnmt3a and Dnmt3b protein levels in N33 and J1 ESCs. c, Relative protein levels of Dnmt3a and Dnmt3b normalized to β-actin by ImageJ software. d, ChIP-qPCR analysis of Dnmt3b occupancy at three proximal regions of Zscan4c promoter, subtelomere of chr7 and chr13. The β-actin locus served as control. e, Bisulfite sequencing analysis of DNA methylation at subtelomeric regions of chr7 and chr13. White circles indicate unmethylated cytosines; Black circles methylated cytosine and crosses mutation of cytosine. f, MeDIP-qPCR analysis of 5mC levels at subtelomeres of chr7, chr13 and Zscan4c promoter in -F and +F ESCs. The β-actin locus served as control. g, Expression of Dnmt3b and Zscan4 after stable knockdown (KD) of Dnmt3b for 9 passages. Two different shRNAs sequences against Dnmt3b were used. #, clones picked. h, Western blot analysis of Dnmt3b and Zscan4 protein levels. i, Relative protein levels of Dnmt3b and Zscan4 in mock KD and Dnmt3b KD ESCs normalized to β-actin by ImageJ software. j,k, Expression levels of Zscan4 and Dnmt3b in Zscan4 + and Zscan4 -ESCs sorted by flow cytometry followed by qPCR (j) and by Western blot analysis (k). J1 ESCs and MEF served as positive and negative controls for Zscan4 expression, respectively. l, Representative IF-FISH images of Zscan4 (green) and telomere (red). Scale bar, 10 µm. m, Relative telomere lengths (fluorescence intensity) of Zscan4+ and Zscan4-ESCs estimated by ImageJ. 50 cells were counted for each group. n, Relative telomere length shown as T/S ratio by qPCR. o, Changes of Dnmt3b protein levels following overexpression of Zscan4. Overexpression of Zscan4 using two different vectors (Pbase or Pcmv) for 48 h immediately reduces Dnmt3b protein levels and stable overexpression of Zscan4 also leads to reduced levels of Dnmt3b after several passages. Mean ± s.e.m. from three independent experiments. *p < 0.05, **p<0.01, ***p < 0.001, ns, not significant (p>0.05). Student's t-test. Figure 8 2i repress Dnmt3a/3b, SSEA1 and Zscan4. 2i, PD0325901 (Mek inhibitor) and CHIR99021 (Gsk3β inhibitor). a-h, ESCs cultured under serum+feeder conditions added with 2i (+2i) or DMSO control (-2i) for 5 passages. a, Expression by qPCR analysis. b, Western blot analysis. c, Relative protein levels of Dnmt3a/b normalized to β-actin by ImageJ software. d, Immunofluorescence of Dnmt3a and Dnmt3b. Scale bar, 10 µm. e, Relative fluorescence intensity of Dnmt3a and Dnmt3b by ImageJ. f, Immunofluorescence of DNA methylation by 5mC and DNA hydroxymethylation by 5hmC. Scale bar, 5 µm. g, Gene expression by qPCR of three ESC lines. P values, compared with ESCs cultured without 2i (-2i). h, Western blot of Dnmt3a, Dnmt3b and Zscan4. i, N33 and F1 ESCs cultured for 5 and 3 passages, respectively, in serum+feeder+LIF or N2B27+2i+LIF conditions and protein levels of Dnmt3a/3b by Western blot. j, Immunofluorescence of Oct4, Nanog and SSEA1 in F1 ESCs (3 passages). Scale bar, 10 µm. k, Immunofluorescence of Nanog by Confocal Laser Scanning Microscope indicating more homogenous expression of Nanog in 2i culture. Scale bar, 10 µm. l, Immunofluorescence of co-staining of Oct4 and Zscan4 in F1 ESCs. Scale bar, 10 µm. Right panel, Percentage of Zscan4 + cells in approximately 1,500 ESCs counted. Mean ± s.e.m. from three independent experiments. *p < 0.05, **p<0.01, ***p < 0.001, ns, not significant (p>0.05). χ2 test for l, and Student's t-test for a, c, e, g. Figure 9 Zscan4 is repressed in ESCs cultured in N2B27+2i/L medium. a, Expression by qPCR of indicated genes of N33 ESCs cultured for 5 passages in three conditions. b, Morphology of N33 ESC colonies under bright field with phase-contrast optics. Scale bar, 100 µm. c, Relative expression of Zscan4 as measured by qPCR analysis of ESCs cultured for four passages. d, Zscan4 protein levels by Western blot. e, Flow cytometry quadrantal diagram indicating proportion and mean fluorescence intensity of Zscan4 + cells in a ESC population. f, qPCR analysis of Zscan4 mRNA levels of J1 ESCs cultured in serum+feeder and 2i/L conditions for four passages. g, Western blot analysis of Dnmt3a, Dnmt3b and Zscan4 protein levels. b-actin served as loading control. Mean ± s.e.m. from three independent experiments. **p <0.01, ***p < 0.001, ns, not significant (p>0.05). ANOVA with PLSD for a,c, and Student's t-test for f.

Supplementary Figure 10 Shorter telomeres in N33 ESCs cultured in
N2B27+2i/L media compared with cultures with feeders. a, Expression levels of pluripotency-associated genes by qPCR of ESCs cultured for five passages. Mean ± s.e.m. from three independent experiments. *p < 0.05, **p < 0.01. ANOVA with PLSD. b, Representative telomere Q-FISH images. Blue, chromosomes stained by DAPI; Green dots, telomeres. Red arrowhead indicates chromosome fusion. c, Histogram displaying distribution of relative telomere length shown as TFU. Green line indicates medium telomere length. Mean ± s.d. of telomere length is shown at right corner of each panel. Wilcoxon-Mann-Whitney rank sum test. d, Frequency of chromosomal fusion per metaphase. ns, not significant (p>0.05). ANOVA with PLSD.

Supplementary Figure 11 Zscan4 expression and telomere elongation in 129xC57/B6 ESCs cultured with feeders compared to those without feeders.
ESCs were initially maintained in serum+feeder+LIF condition for 4 passages, and then were split onto plates with or without feeders for 5 passages (P9) or 11 passages (P15). Three independent ESC lines (1#, 3#, 5#) were used for analysis of gene expression and telomere lengths. 129xC57/B6 ESCs exhibited high developmental pluripotency as determined by TEC assay. At passage 5, 23 embryos were transferred following injection of the ESCs into tetraploid albino embryos, and of 8 live pups born, four ESC-pups live to adults (over 5 weeks). At passage 14, 7 live pups were born from 40 embryos and 6 of them were still alive. a, Immunofluorescence co-staining of Oct4 and Zscan4. Scale bar, 10 µm. b, Percentage of Zscan4 + cells for each cell line (shown are number of ESCs counted). c, Percentage of Zscan4 + cells for all cell lines. Mean ± s.e.m, ***p<0.001. Student's t-test. d, Relative expression levels by qPCR of Zscan4, Tcstv1 and Tcstv3. Mean ± s.d, *p<0.05, **p<0.01. Student's t-test. e, Western blot of Zscan4 in ESCs cultures with or without feeders. f, Representative telomere Q-FISH images of ESCs cultured on gelatin or feeders at P9. Blue, chromosomes stained by DAPI; Green dots, telomeres; Red arrowhead indicate telomere signal-free ends. g, Histogram shows distribution of relative telomere length displayed as TFU by Q-FISH analysis. Green line indicates medium telomere length. Mean ± s.d. of telomere length is shown above each panel. Scale bar, 10 µm. h, Frequency of telomere signal-free ends indicative of shortest telomere per metaphase at P9 and P15. Mean ± s.e.m. i, Telomere length distribution shown as TRF by Southern blot analysis at P9.