Graphical Abstract

   graphic file with name fx1.jpg
   [41]Open in a new tab

Highlights

     * •
       Conditional DICER1 knockouts enable the study of its essential
       requirement in hESCs
     * •
       DICER1 deletion leads to increased extrinsic (receptor-mediated)
       apoptosis
     * •
       miRNAs rescue the DICER1 knockout apoptotic phenotype
     * •
       This work provides a platform for interrogating miRNA function in
       hESCs
     __________________________________________________________________

   Huangfu and colleagues report an unexpected requirement for DICER1 in
   preventing death-receptor-mediated apoptosis in hESCs. This essential
   pro-survival role is largely mediated by mir-302-367 and mir-371-373
   cluster members that bear the seed sequence “AAGUGC” shared by the
   ESC-specific cell-cycle-regulating family of miRNAs. This work provides
   a robust platform for interrogating miRNA function in hESC pluripotency
   and differentiation.

Introduction

   MicroRNAs (miRNAs) are ∼20- to 24-nucleotide long noncoding RNAs that
   bind target mRNAs through partial complementarity with their 3′ UTRs to
   promote target mRNA decay and/or inhibit translation ([42]Ebert and
   Sharp, 2012). The RNase III enzyme Dicer1 is required for the
   processing of precursor miRNA hairpins into their mature functional
   products ([43]Yang and Lai, 2011). Dicer1 knockout mouse embryos
   exhibit post-implantation morphological abnormalities at around
   embryonic day 7.5 (E7.5) ([44]Bernstein et al., 2003, [45]Spruce
   et al., 2010), emphasizing the importance of miRNAs for early mammalian
   development. Due to overlapping miRNA requirements, no individual miRNA
   mouse knockouts have recapitulated the Dicer1 knockout phenotype
   ([46]Kuhnert et al., 2008, [47]Liu et al., 2008, [48]Park et al., 2010,
   [49]Ventura et al., 2008, [50]Zhao et al., 2007). Mouse embryonic stem
   cells (mESCs) have been used to gain further understanding of miRNA
   functions in the pre-implantation period. mESCs lacking either Dicer1
   or Dgcr8, factors essential for miRNA biogenesis, can self-renew but
   show reduced proliferation and a relative accumulation of cells in the
   G[1] phase of the cell cycle ([51]Kanellopoulou et al., 2005,
   [52]Murchison et al., 2005, [53]Wang et al., 2007, [54]Wang et al.,
   2008). miRNA rescue experiments in Dgcr8 knockout mESCs have shown that
   the rapid cell cycle of mESCs is regulated by members of the
   mir-290-295 (homologous to human mir-371-373), mir-302-367, and
   mir-17-92b clusters, known collectively as the ESC-specific
   cell-cycle-regulating (ESCC) family of miRNAs ([55]Wang et al., 2008).

   Thus far the exact requirements for DICER1 in human embryonic stem
   cells (hESCs) have not been established. One study used a knockdown
   approach in hESCs but DICER1 mRNA was only reduced about 5-fold ([56]Qi
   et al., 2009). Compared with mESCs, hESCs reflect a later developmental
   stage ([57]Brons et al., 2007, [58]Tesar et al., 2007), and there could
   be human-specific DICER1 and miRNA requirements. Here we report the
   generation and characterization of DICER1-deficient hESCs. We show that
   in hESCs, unlike in mESCs, DICER1 is essential for self-renewal and
   inhibition of death receptor-mediated apoptosis. Furthermore, ESCC seed
   sequence-containing members of the mir-302-367 and mir-371-373 clusters
   could reverse the apoptotic phenotype. Our study provides a valuable
   platform for the discovery of individual miRNA functions during early
   development of human embryos and hESC differentiation.

Results

Generation of DICER1 Conditional Knockout hESCs

   To investigate the role of DICER1 in human pluripotency, we attempted
   to generate DICER1 knockout hESC lines. Despite using the highly
   efficient iCRISPR platform ([59]Gonzalez et al., 2014), we were unable
   to generate homozygous null mutant hESC lines after screening 464 lines
   generated with nine guide RNAs (gRNAs) ([60]Figures S1A–S1C). We
   retargeted a line already carrying a heterozygous DICER1 frameshift
   mutation, but still failed to generate a frameshift mutation in the
   second allele ([61]Figure S1D). These results suggest that
   DICER1-deficient hESCs have self-renewing defects. To overcome the
   apparent cell lethality issues, we designed a conditional knockout
   strategy. We first selected two CRISPRs (cr4 and cr6) based on their
   ability to efficiently generate indel (insertion/deletion) mutations
   ([62]Figure S1C). We then generated two Tet-On transgenic lines
   expressing DICER1 transgenes that have a silent mutation in the PAM
   sequence and thus are immune to cr4 or cr6 gRNA targeting (referred to
   as TRE-DICER1^∗ lines) ([63]Figures S1E and S1F). We next used
   CRISPR/Cas9 to knock out endogenous DICER1 while maintaining expression
   of exogenous DICER1^∗ by doxycycline treatment ([64]Figure 1A). This
   approach enabled us to create 32 conditional biallelic frameshift
   knockout lines out of 187 clonal lines analyzed ([65]Figure 1C). We
   focused further analysis on three conditional knockout lines: B2
   (generated using cr4), and D11 and F2 (generated using cr6)
   ([66]Figures 1B and 1D). When maintained with doxycycline treatment,
   these conditional DICER1 knockout hESC lines were indistinguishable
   from wild-type hESCs and exhibited normal karyotypes ([67]Figure S1G).

Figure 1.

   [68]Figure 1
   [69]Open in a new tab

   Generation of DICER1 Conditional Knockout hESC Lines

   (A) Schematic of the procedure to generate conditional DICER1 knockout
   hESCs. CRISPR/Cas9 was used to generate DICER1 knockout mutations in
   cells infected with a gRNA-immune doxycycline (DOX)-inducible
   TRE-DICER1^∗ transgene. Cas9 and DICER1^∗ protein expression was
   induced by doxycycline treatment. DICER1 knockout mutants were
   maintained in doxycycline until the start of experiments. An asterisk
   is used to distinguish the exogenous modified DICER1^∗ from endogenous
   DICER1.

   (B) Schematic representation of DICER1 protein with domains. Orange
   arrows represent gRNA-targeting loci upstream of the RNase IIIa
   functional domain. cr, CRISPR; DUF, domain of unknown function; dsRBD,
   double-stranded RNA-binding domain.

   (C) CRISPR/Cas9 targeting results of conditional DICER1 hESC lines.
   White, wild-type lines; yellow, monoallelic mutant lines; red,
   biallelic mutant lines. Stripes indicate DICER1 knockout lines with
   frameshift mutations in both alleles. fs, frameshift allele.

   (D) Representative endogenous DICER1 sequences of three selected
   conditional DICER1 knockout clones with PAM sequences labeled in red.
   WT, wild-type.

   Unless otherwise indicated, scale bars represent 100 μm in all figures.
   See also [70]Figure S1.

DICER1 Is Essential for hESC Self-Renewal

   We removed doxycycline to investigate the effects of DICER1 deletion in
   conditional DICER1 knockout hESCs ([71]Figure 2A). Six days after
   doxycycline withdrawal, no DICER1^∗ mRNA or protein expression was
   detected, nor did we observe significant expression of the endogenous
   DICER1 protein ([72]Figures 2B and 2C). The DICER1-depleted hESCs
   survived and maintained comparable protein levels of pluripotency
   markers OCT4 (POU5F1), NANOG, and SOX2 based on immunofluorescence
   staining, although some reduction of mRNA levels was observed by
   qRT-PCR ([73]Figures S2A and S2B). We also tested the levels of several
   mature miRNAs that are abundant in wild-type hESCs including miR-373
   and miR-302a. All miRNAs examined were significantly reduced in
   DICER1-depleted hESCs by qRT-PCR and northern blot, and there was a
   concomitant increase in precursor miRNAs ([74]Figures 2D and [75]S2C).
   These results confirmed that miRNA processing was compromised in DICER1
   null hESCs. However, miRNAs were not fully depleted after 6 days, which
   is consistent with the notion that miRNAs are highly stable molecules
   ([76]Guo et al., 2015).

Figure 2.

   [77]Figure 2
   [78]Open in a new tab

   DICER1 Is Essential for hESC Self-Renewal

   (A) Schematic representation of experimental procedure in feeder-based
   condition. Cells were split at 28,000 cells/cm^2. DOX, doxycycline; WB,
   western blot; AP, alkaline phosphatase.

   (B) Western blot of FLAG-DICER1^∗ and total DICER1 levels of three
   conditional DICER1 knockout lines (B2, D11, F2) in feeder-based
   condition with and without DOX treatment for 6 days. WT, wild-type.

   (C) qRT-PCR analysis of DICER1^∗ transgene expression in three
   conditional DICER1 knockout lines in feeder-based condition with and
   without doxycycline treatment for 6 days. Levels relative to GAPDH.
   Statistical analysis was performed in comparison with +DOX samples. n =
   3 independent experiments. n.d., not detected.

   (D) Mature miRNA qRT-PCR analysis of eight hESC-expressed miRNAs in
   DICER1 conditional knockout line B2 after 6 days of DOX withdrawal in
   feeder-based condition. Levels are normalized to +DOX. Statistical
   analysis was performed in comparison with +DOX samples. n = 3
   independent experiments.

   (E) qRT-PCR analysis of four hESC-abundant miRNAs (miR-373, miR-302a,
   miR-92a, and miR-200c) at days 6 through 9 of DOX withdrawal. miRNA
   levels in the +DOX samples are used for comparison. Analysis was done
   on the B2 DICER1 conditional knockout line in feeder-based condition.
   For each time point, statistical analysis was performed in comparison
   with +DOX samples. Day-6 data are reused from (D) for comparison
   purposes. n = 3 independent experiments.

   (F) AP staining of three conditional DICER1 knockout lines in
   feeder-based conditions with and without DOX treatment for 12 days. n =
   6 independent experiments. Scale bar, 5 mm.

   (G) AP staining of three conditional DICER1 knockout lines in
   feeder-free condition with and without doxycycline treatment for
   8 days. n = 3 independent experiments. Scale bar, 5 mm.

   Error bars indicate SD, and significance is indicated as ^∗p < 0.05,
   ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001; ns, not significant
   (p ≥ 0.05). See also [79]Figure S2.

   After passaging these DICER1-depleted cells and maintaining them in the
   absence of doxycycline for another 6 days ([80]Figure 2A), we observed
   further reduction in the accumulation of mature miRNAs ([81]Figure 2E).
   Coinciding with this reduction in miRNA levels, although the
   DICER1-depleted cells were properly attached to the dish after
   replating, they began to show compromised growth 10 days after
   doxycycline withdrawal as determined by staining for alkaline
   phosphatase activity and live cell counting ([82]Figures S2D and S2E).
   By day 12 of doxycycline withdrawal there were barely any cells
   detected in all three conditional DICER1 knockout lines ([83]Figures 2F
   and [84]S2F). The few remaining DICER1-depleted cells could not be
   propagated without doxycycline treatment.

   In addition to the original feeder-based condition, we also tested the
   requirements for DICER1 in the Essential 8 (E8) chemically defined
   feeder-free culture condition ([85]Chen et al., 2011). Upon doxycycline
   withdrawal, DICER1-depleted cells could not be propagated in the E8
   condition either, and almost no cells were detected by alkaline
   phosphatase staining after 8 days of doxycycline withdrawal
   ([86]Figure 2G). Therefore, DICER1 is required for self-renewal of
   hESCs cultured in feeder-based and feeder-free conditions.
   DICER1-depleted cells survived longer in the feeder-based condition
   presumably due to pro-survival effects of the feeder cells. In
   comparison, Dicer1-depleted mESCs are capable of self-renewal although
   they proliferate at a slower rate than wild-type mESCs ([87]Bernstein
   et al., 2003, [88]Kanellopoulou et al., 2005, [89]Murchison et al.,
   2005, [90]Wang et al., 2007, [91]Wang et al., 2008).

Loss of DICER1 in hESCs Causes Apoptosis

   Unlike Dicer1-deficient mESCs, we did not observe an apparent mitotic
   defect based on flow-cytometry analysis of the mitotic marker
   phosphohistone H3 ([92]Figures S3A and S3B), nor defects in cell-cycle
   progression based on cell-cycle analysis by propidium iodide staining
   ([93]Figure S3C). Instead, we detected significantly increased numbers
   of cleaved caspase-3 (C-CSP3)-positive cells by flow cytometry starting
   from day 8 after doxycycline removal and peaking at day 10 in both
   DICER1 conditional knockout lines (B2 and D11) cultured in the
   feeder-based condition ([94]Figures 3A, 3B, [95]S3D, and S3E). The
   increased apoptotic rate coincided with the compromised cell growth
   that first became apparent around days 9–10 ([96]Figures S2D and S2E).
   C-CSP3-positive cells lost OCT4 expression and showed nuclear
   condensation and fragmentation typical of apoptotic cells when examined
   by immunofluorescence staining at days 9 and 10 ([97]Figure 3C).
   Similarly, DICER1 conditional knockout cells in the E8 condition also
   showed significantly increased numbers of apoptotic cells by
   flow cytometry for C-CSP3 expression and annexin-V staining (another
   apoptosis assay) after doxycycline removal ([98]Figures 3E–3G and
   [99]S3F). Therefore, unlike its pro-proliferation role in mESCs,
   DICER1's main function in hESCs is pro-survival.

Figure 3.

   [100]Figure 3
   [101]Open in a new tab

   DICER1 Depletion in hESCs Causes Caspase-3-Mediated Apoptosis

   (A) Schematic representation of the experimental design in feeder-based
   condition. Cells were split at 28,000 cells/cm^2. DOX, doxycycline.

   (B) Flow cytometry of cleaved caspase-3 (C-CSP3) from days 6 through 11
   in the D11 conditional DICER1 knockout line with and without DOX in
   feeder-based condition. n = 3 independent experiments.

   (C) Representative immunofluorescence pictures of C-CSP3 and OCT4 in
   the D11 conditional DICER1 knockout line maintained with and without
   DOX at days 9 and 10 in feeder-based condition. Inset represents a
   close-up view showing fragmentation (arrowheads) of nuclei and loss of
   OCT4 in a C-CSP3-positive cell. Scale bars, 30 μm (main images) and
   7 μm (close-up). n = 3 independent experiments.

   (D) Schematic representation of the experimental design in feeder-free
   condition. Cells were split at a density of 14,000 cells/cm^2.

   (E) Flow-cytometry detection of C-CSP3 expression in B2 and D11
   conditional DICER1 knockout lines in feeder-free condition maintained
   with and without DOX for 7 days. n = 4 independent experiments.

   (F) Flow cytometry of annexin-V in B2 and D11 conditional DICER1
   knockouts cultured in feeder-free condition maintained with or without
   DOX for 7 days. n = 4 independent experiments for B2, and n = 5
   independent experiments for D11.

   (G) Representative histograms of annexin-V flow cytometry on D11
   conditional DICER1 knockout line maintained with or without DOX in
   feeder-free condition. n = 5 independent experiments.

   (H) KEGG pathway enrichment analysis was performed in R (v3.4.2) with
   clusterProfiler (v3.6.0) on RNA-seq data of B2 conditional DICER1
   knockout line cultured in feeder-based condition with or without DOX
   for 9 days. n = 3 independent experiments.

   (I) Heatmap of apoptosis-related genes. Data are derived from RNA-seq
   of B2 conditional DICER1 knockout hESCs with and without DOX for 9 days
   in feeder-based condition. The heatmap was generated by cluster
   analysis using the online analysis tool CARMAweb
   ([102]https://carmaweb.genome.tugraz.at/genesis) with the methods
   normalized gene. The variation between the biological repeats could be
   due to the rapid nature of apoptosis. n = 3 independent experiments.

   Error bars indicate SD, and significance is indicated as ^∗p < 0.05,
   ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, not significant (p ≥ 0.05). See also
   [103]Figure S3.

   RNA sequencing (RNA-seq) analyses identified 692 significantly
   upregulated and 277 significantly downregulated genes in
   DICER1-depleted hESCs on day 9 of doxycycline withdrawal in the
   feeder-based condition compared with the doxycycline-treated control
   ([104]Figure S3G). As expected, DICER1 was the most significantly
   downregulated gene ([105]Figure S3G). KEGG pathway enrichment analysis
   of these differentially expressed genes identified pathways involved in
   apoptosis, pluripotency, and cell adhesion ([106]Figure 3H), while gene
   ontology analysis revealed mostly development-related terms
   ([107]Figure S3I). In particular, DICER1 deletion led to increased
   expression of pro-apoptotic genes of the extrinsic (receptor-mediated)
   pathway, notably FAS, NIK (MAP3K14), and TRAILR4 (TNFRSF10D) as
   verified by qRT-PCR analyses ([108]Figures 3I and [109]S3H). Key genes
   involved in the intrinsic (mitochondria-mediated) apoptotic pathway
   were generally not upregulated in our RNA-seq analysis or qRT-PCR
   validation, except for BIM ([110]Figures 3I and [111]S3H).
   DICER1-depleted hESCs did not show increased expression of BNIP3L, an
   intrinsic pathway pro-apoptotic factor upregulated upon miR-302/367
   knockdown in hESCs ([112]Figures 3I and [113]S3H) ([114]Zhang et al.,
   2015). This difference could be due to perturbing global versus
   specific miRNAs, or the knockdown versus the knockout approach.
   Together, these results demonstrate a previously unappreciated role for
   DICER1 in preventing receptor-mediated apoptosis in hESCs.

miRNA Rescue of DICER1 Deficiency

   To further understand the DICER1-dependent hESC survival, we attempted
   to rescue the DICER1 knockout phenotype by transfecting DICER1-depleted
   hESCs with selected miRNA families ([115]Figure 4A). We tested the
   ESCC-family of miRNAs (mir-302-367, mir-371-373, and mir-17-92) since
   they have been previously implicated in mESC proliferation and they are
   among the most abundant miRNAs in both mouse and human ESCs ([116]Greve
   et al., 2013, [117]Wang et al., 2008). The most striking rescues were
   observed for members of the mir-302-367 and mir-371-373 clusters based
   on alkaline phosphatase staining and cell number counts ([118]Figures
   4B and [119]S4B). The transfected cells expressed pluripotency markers
   OCT4 and NANOG ([120]Figure S4A). Although members of the mir-17-92
   cluster were sufficient to rescue the proliferation defect in
   miRNA-depleted mESCs ([121]Wang et al., 2008), their effects were
   relatively mild in hESCs ([122]Figures 4B and [123]S4B), suggesting
   differences between mouse and human ESCs in miRNA requirements.
   Interestingly, not all members of the mir-302-367 and mir-371-373
   clusters rescued the DICER1 knockout hESC phenotype. In the mir-302-367
   cluster, miR-302a, miR-302b, miR-302c, and miR-302d showed significant
   rescue but miR-367 did not ([124]Figures 4B and [125]S4B). Likewise, in
   the mir-371-373 cluster, miR-372 and miR-373 but not miR-371 showed
   significant rescue ([126]Figures 4B and [127]S4B). Notably, the miRNAs
   with significant rescue effects all shared the ESCC seed sequence
   “AAGUGC” ([128]Figure 4C).

Figure 4.

   [129]Figure 4
   [130]Open in a new tab

   ESCC Seed Sequence-Containing miRNAs Rescue DICER1 Knockout Defects

   (A) Schematic representation of miRNA mimic transfection procedure in
   feeder-based condition. DOX, doxycycline.

   (B) Representative AP-stained D11 conditional DICER1 knockout hESCs
   with or without transfection of miRNAs in feeder-based condition. In
   green are the six miRNAs that partially rescue the DICER1 knockout
   phenotype. n = 5 independent experiments. Scale bar, 5 mm.

   (C) List of mature miRNAs (3p or 5p arms) transfected in D11
   conditional DICER1 knockout line. In green are the mature miRNAs that
   significantly rescued the DICER1 knockout phenotype. The box encloses
   the seed sequence shared among those mature miRNAs that rescue the
   DICER1 knockout phenotype.

   (D) Schematic representation of miRNA mimic transfection procedure in
   feeder-free condition.

   (E) Flow cytometry of annexin-V in D11 conditional DICER1 knockout line
   cultured in feeder-free conditions for 8 days and transfected with
   mature miRNAs or control miRNA as shown in (D). For comparison between
   multiple experimental groups in (E) and (F), one-way ANOVA was used
   followed by multiple comparisons with Tukey correction. n = 3
   independent experiments.

   (F) qRT-PCR analysis of representative pro-apoptotic genes belonging to
   the extrinsic apoptotic pathway (FAS, NIK, TRAILR4) as well as
   intrinsic marker BIM in feeder-free condition with and without
   doxycycline treatment for 8 days and transfected with miR-302b,
   miR-372, or miR control as shown in (D). n = 3 independent experiments.

   Error bars indicate SD, and significance is indicated as ^∗p < 0.05,
   ^∗∗p < 0.01, and ^∗∗∗p < 0.001; ns, not significant (p ≥ 0.05). See
   also [131]Figure S4.

   We also observed similar rescue of cell survival in the E8 feeder-free
   condition, and annexin-V staining showed a significant reduction of
   apoptotic rates in DICER1-deficient hESCs transfected with miR-372,
   miR-373, miR-302b, and miR-302d ([132]Figures 4D, 4E, and [133]S4C). In
   silico analysis using TargetScan and MirTarget ([134]Agarwal et al.,
   2015, [135]Wong and Wang, 2015) showed that FAS, NIK, TRAILR4, and BIM
   genes, significantly upregulated in DICER1-depleted hESCs, were
   predicted targets of the ESCC seed sequence-bearing miRNAs ([136]Table
   S1). Therefore, we examined whether their upregulation could be
   reversed in hESCs transfected with miR-372 or miR-302b. By qRT-PCR
   analysis, the expression levels of FAS, NIK, and TRAILR4, but not BIM,
   were significantly rescued to levels close to those of
   doxycycline-treated control hESCs expressing DICER1 ([137]Figure 4F).
   These results suggest that the ESCC seed sequence-bearing members of
   mir-302-367 and mir-371-373 clusters are responsible for preventing
   apoptosis in DICER1 knockout hESCs primarily through the regulation of
   death receptor-mediated apoptosis pathway genes such as FAS.

Discussion

   Overlapping functions of miRNAs pose a technical challenge to fully
   understanding their exact roles during early embryonic development
   ([138]Greve et al., 2013). Our conditional DICER1 knockout hESCs
   provide an excellent platform to investigate the roles of individual
   miRNAs ([139]Park et al., 2010). For instance, we find that
   reintroduction of ESCC seed sequence-containing miRNAs of the
   mir-302-367 and mir-371-373 clusters is sufficient to promote cell
   survival in DICER1 knockout hESCs, whereas mir-17-92 does not play a
   significant role in hESCs in this context despite its prominent role in
   mESC cell-cycle progression ([140]Wang et al., 2008). Our platform is a
   valuable tool for the discovery of novel miRNA requirements for both
   hESC pluripotency maintenance and stem cell differentiation.

   In contrast to the essential requirement for DICER in hESCs, Dicer1 is
   not required for mESC survival. This could be due to the different
   developmental stages thought to be represented by mouse and human ESCs.
   Supporting this idea, recent studies have uncovered essential
   requirements for EZH2 and DNMT1 in hESCs, which are not observed in
   mESCs ([141]Collinson et al., 2016, [142]Liao et al., 2015, [143]Shen
   et al., 2008, [144]Tsumura et al., 2006). In this regard, it may be
   more appropriate to compare DICER1 knockout phenotypes in hESCs with
   those observed in mouse epiblast stem cells (mEpiSCs). In mEpiSCs,
   Dicer1 deletion causes increased Bim expression, leading to activation
   of the intrinsic apoptotic pathway (the mitochondrial pathway); these
   effects are mitigated by reintroduction of miR-20, miR-92, and miR-302
   family members into Dicer1-depleted cells ([145]Pernaute et al., 2014).
   Although we also observed some upregulation of BIM expression in
   DICER1-depleted hESCs, there was no clear rescue of BIM expression upon
   reintroduction of miR-302b or miR-372. Instead, DICER1-depleted hESCs
   upregulate a cohort of extrinsic apoptotic pathway genes.
   Reintroduction of miR-302b or miR-372 in DICER1 knockout cells rescued
   the cell-survival defect and reversed the aberrant upregulation of
   extrinsic pathway genes FAS, NIK, and TRAILR4. It is of interest to
   note that the human, but not the mouse, FAS and TRAILR4 3′ UTR
   sequences were predicted by in silico analyses to contain the ESCC
   targets ([146]Table S1). These differences may explain the differing
   roles for DICER1 in hESCs compared with mESCs or mEpiSCs.

   In addition to the human-mouse species difference, the pro-survival
   requirement for DICER1 also depends on the cellular context. Mouse
   studies have shown that Dicer1 is essential for the survival of a
   number of developmental cell types such as limb mesoderm, αβ T cells,
   and melanocyte stem cells and differentiated melanocytes ([147]Cobb
   et al., 2005, [148]Harfe et al., 2005, [149]Levy et al., 2010). On the
   other hand, Dicer1 is not required for survival of the γδ T cells
   ([150]Cobb et al., 2005), and it is dispensable in the specification
   and survival of the fetal adrenal cortex ([151]Krill et al., 2013). In
   HCT116 human colorectal cancer cells, DICER1 is also not essential for
   survival ([152]Kim et al., 2016). Combined with directed
   differentiation, our DICER1 conditional knockout hESC lines can be used
   in the future to determine cell-context-dependent DICER1 and miRNA
   functions, and inform us about human-specific aspects of
   post-implantation development.

Experimental Procedures

hESC Culture

   All experiments were performed on HUES8 hESCs (NIHhESC-09-0021). HUES8
   iCas9 cells were generated as previously described ([153]Gonzalez
   et al., 2014). For experiments conducted in the feeder-based condition,
   irradiated mouse embryonic fibroblasts were used as feeder cells, and
   hESCs were cultured in DMEM/F12 (without HEPES) supplemented with 20%
   KnockOut Serum Replacement, 1× nonessential amino acids, 1× GlutaMAX,
   100 U/mL penicillin/100 μg/mL streptomycin (Gemini), 0.055 mM
   2-mercaptoethanol, and 10 ng/mL recombinant human basic fibroblast
   growth factor. Cultures were passaged at a 1:12 split ratio every
   4–6 days using TrypLE to dissociate the cells. Rho-associated protein
   kinase inhibitor Y-27632 (Selleck Chemicals, S1049) (5 μM) was added to
   the culture medium when passaging cells.

   For experiments conducted in the feeder-free condition, hESCs were
   cultured in the complete E8 medium (Thermo Fisher Scientific, A1517001)
   on vitronectin (VTN-N; Thermo Fisher Scientific, A14700) pre-coated
   plates. Cells were replated at 125,000 cells/mL in 6-well plates and
   passaged every 4 days after cell dissociation using 0.5 mM EDTA (KD
   Medical, RGE-3130). The hESC culture work was conducted as per NIH
   guidelines and approved by the Tri-SCI Embryonic Stem Cell Research
   Oversight Committee.

miRNA Mimics Rescue Assay

   For the miRNA mimics rescue experiments, DICER1 knockout hESCs (B2 and
   D11 lines) were cultured with or without doxycycline for 6 days
   (feeder-based condition) or 4 days (feeder-free condition). They were
   replated at 125,000 cells/mL in 24-well plates in feeder-based
   conditions and at 125,000 cells/mL in 12-well plates in feeder-free
   conditions, and transfected while in suspension with miRNA mimics
   (mirVana miRNA mimics; Thermo Fisher, 4464066) using Lipofectamine
   RNAiMAX (Thermo Fisher Scientific, 13778150) as per manufacturer's
   instructions. For the positive control, doxycycline was added back. For
   the negative control, a control miRNA mimic that contains a nonspecific
   sequence that does not bind to any mRNAs was transfected. The final
   concentration of each miRNA mimic was 10 nM, as previously described
   ([154]Wang et al., 2008). The transfected cells were allowed to grow
   without doxycycline for 6 days (feeder-based condition) or 4 days
   (feeder-free condition), at which point the cells were harvested for
   analyses.

Statistical Analysis

   All values are shown as mean ± SD. Data were analyzed using unpaired
   two-tailed Student's t test in GraphPad Prism unless otherwise
   indicated. p values of ≥0.05 were considered not significant (ns) and p
   values of <0.05 (^∗), <0.01 (^∗∗), <0.001 (^∗∗∗), and <0.0001 (^∗∗∗∗)
   were considered significant.

Author Contributions

   V.T. led, designed, and performed most experiments; D.Y. performed most
   of the revision experiments and assisted with flow cytometry; S.M.
   performed northern blots; F.G. cloned the DICER1^∗ vectors; R.K. and
   C.X. analyzed the RNA-seq data; R.W.R. and N.V. assisted with
   additional experiments; D.H. and E.C.L. supervised the study; V.T. and
   D.H. wrote the manuscript; all other authors provided editorial advice.

Acknowledgments