As the aim of this study was to elucidate alternative mechanisms of translation initiation that are independent of the cap, we were looking for a model system which allows us to reliably create a condition that favors non-canonical translation initiation. Further, we were interested in a system that also naturally switches to physiological conditions in which repression of cap-dependent
148
translation is thought to be of importance. Hence, we generated clonal mESC lines with stable integration of a known suppressor of cap-dependent translation, dominant negative 4E-BP1, and stable integration of a proneuronal factor, ASCL1, strongly promoting neuronal differentiation.
3.1.1 Working principle of dominant negative 4E-BP1 in cap-dependent translation inhibition To investigate cap-independent translation mechanisms, a model system with inducible expression of dominant negative 4E-BP1 was established. 4E-BP1 is a translational repressor whose activity is regulated via phosphorylation by the mTOR signaling pathway. Human 4E-BP1 contains in total seven phosphorylation sites: Thr37, Thr46, Ser65, Thr70, Ser83, Ser101 and Ser112 [249]. The first five of these are evolutionary conserved and phosphorylation of the two most N-terminal threonine residues is required for subsequent phosphorylation of the residual C-terminal sites [238], [239]. The phosphorylation state of 4E-BP1 determines whether 4E-BP1 is bound to eIF4E. In the hypophosphorylated form 4E-BP1 interacts with eIF4E through a similar site as eIF4G, thereby inhibiting eIF4F complex assembly and preventing translation initiation.
Hyperphosphorylation, especially phosphorylation of Thr70, Ser65 and Ser112, causes a release of 4E-BP1 from eIF4E [240]. This in turn permits the association of eIF4E with eIF4G which is needed for eIF4F assembly at the mRNA 5’end to recruit ribosomes.
The 4E-BP1(4Ala) mutant constitutively binds eIF4E, as four of its phosphorylation sites (Thr37, Thr46, Ser65, Thr70) were replaced with alanine [849]. Like this, phosphorylation of 4E-BP1(4Ala) is abrogated and cap-dependent translation is constantly repressed by 4E-4E-BP1(4Ala) binding eIF4E. To generate a model system with inducible inhibition of cap-dependent translation, 4E-BP1(4Ala) was stably integrated into the genome of mESCs.
3.1.2 ASCL1 induces differentiation of mouse embryonic stem cells into neurons
To study cap-independent translation mechanisms in steady state mESCs and during the process of differentiation into neurons, the transcription factor ASCL1 was, in addition to 4E-BP1(4Ala), stably integrated into the mESC genome. ASCL1 is a proneural transcription factor which functions in proliferation and differentiation of neural stem and progenitor cells. It further regulates cell migration, axon guidance and synapse formation during neurogenesis [850]. ASCL1 is exceptional among proneural factors, because it acts as a pioneer factor that alone is sufficient to reprogram mouse and human fibroblasts or ESCs into induced neuronal cells [851], [852]. The induction of ASCL1 expression in our model system leads to differentiation of mESCs into induced neurons (iNs) within only one week. After 24 h of ASCL1 induction cells are phenotypically still close to mESCs (Fig. S1) but it was shown that already after 12 h differentiating cells can express neuronal Tubb3 and that about 780 genes are upregulated on
149
transcriptional level with nervous system development being the most significantly enriched GO biological process term [853]. At 48 h, cells already express genes that are linked to specific neuronal cell types, especially noradrenergic and cortical interneuron markers, but these are heterogeneously expressed across the cell population [853]. Hence, ASCL1 induction leads to a rapid transition from an embryonic towards a neuronal cell identity already at early time points of the differentiation process.
3.1.3 Recombinase mediated cassette exchange to generate stable cell lines expressing dominant negative 4E-BP1
Both, dominant negative 4E-BP1(4Ala) and ASCL1 were integrated into the genome of mESCs via inducible cassette exchange (ICE). In detail, the male A17 mESC line which carries the ICE locus upstream of the HPRT gene on the X chromosome was used for transgenesis [854]. The ICE locus contains the tetracycline response element upstream of the Cre transgene which is floxed by heterologous loxP sites and which is followed by a Δneo gene that misses a promoter and the start codon (Fig. 4). Upon doxycycline induction, Cre is expressed and after p2Lox plasmid transfection, ASCL1 and 4E-BP1(4Ala) are inserted by Cre-Lox recombination in between the two loxP sites, exchanging Cre for the two new transgenes (Fig. 4). Further, the recombination introduces a promoter and a start codon upstream of the Δneo gene so that successfully modified recombinants are G418 resistant.
As a control, another mESC line was generated with stable integration of ASCL1 only. In the following, this mESC cell line is referred to as ‘wt’, although it carries the inducible ASCL1 transgene in the ICE locus.
Fig. 4: Inducible cassette exchange stably integrates dominant negative 4E-BP1 and ASCL1 into mESCs. Doxycycline induces tetracycline response element (TRE) driven expression of Cre recombinase, which after transfection of p2Lox plasmid, mediates recombination at heterologous LoxP sites (green & purple triangles) such that PGK promoter and ATG start codon are integrated upstream of the neomycin resistance gene that lacks a promoter and a start codon (Δneo) and that ASCL1, F2A cleavage peptide and 4E-BP1(4Ala) are integrated downstream of TRE at the ICE locus on the X chromosome. Adopted from Iacovino et al, 2014 [840].
150
3.1.4 Expression of dominant negative 4E-BP1 reduces cap-dependent translation and promotes cap-independent translation
After antibiotic selection, successful integration of 4E-BP1(4Ala) and ASCL1 into the genome of mESCs was validated by western blot and qPCR for the V5 peptide tag, which is C-terminally attached to 4E-BP1(4Ala). Three different monoclonal cell lines were selected and 4E-BP1(4Ala) protein expression could be confirmed in all clones tested after 24 h of doxycycline induction (Fig. 5A). 4E-BP1(4Ala) expression was also readily detectable by qPCR after 7 d of doxycycline induction and stem cell differentiation into iNs. Like wt mESCs, monoclonal 4E-BP1(4Ala)-expressing cell lines exhibited a severe decrease in expression of stem cell markers Oct4 and Sox1, while expression of the neuronal marker Tubb3 was increased (Fig. 5B). Hence, stable integration and inducible expression of 4E-BP1(4Ala) could be demonstrated on RNA and protein level. Further, the overexpression of 4E-BP1(4Ala) in synergy with ASCL1 drove neuronal differentiation and similar marker gene expression as in wt cells expressing ASCL1 alone, suggesting that monoclonal cell lines can differentiate into iNs in the presence of dominant negative 4E-BP1.
To determine the effects on cap-dependent and cap-independent translation by 4E-BP1(4Ala) overexpression, we performed dual luciferase reporter assays using bicistronic reporter constructs. These bicistronic reporters contained an EMCV IRES in between the 5’-terminally located FLuc and 3’-terminally located RLuc ORFs. That way, cap-dependent FLuc expression and IRES-mediated RLuc expression are derived from the same transcript (Fig. 5C). By taking the ratio of RLuc/FLuc expression, one can quantify the ability of a certain sequence element to mediate IRES driven translation. To do so, reporter plasmids were transfected into wt and monoclonal mESCs that were subsequently induced with doxycycline for 24 h before luciferase values were analyzed. All selected transgenic clones displayed an increase in RLuc/FLuc expression ratios upon 4E-BP1(4Ala) overexpression (Fig. 5D). Likewise, wt mESC showed a slight increase in RLuc/FLuc expression after doxycycline induction; however, the changes of RLuc/FLuc ratios after overexpression of ASCL1 alone were small and not significant.
A mutated non-functional EMCV* IRES reporter was used as negative control and displayed only very low levels of RLuc/FLuc expression (~3 % of EMCV). Doxycycline induction did not significantly alter RLuc/FLuc ratio of EMCV* reporters in wt and clonal cell lines (Fig. 5D).
In sum, results of the bicistronic reporter assays indicate that overexpression of 4E-BP1(4Ala) promotes EMCV IRES-mediated translation over cap-dependent translation.
Overexpression of ASCL1 also slightly stimulates cap-independent translation initiation, although too much lesser and not significant extend. Hence, the generated model system can be used to
151
investigate cap-independent translation mechanisms after 4E-BP1(4Ala) induced inhibition of cap-dependent translation.
3.2 Identification of upregulated genes upon dominant negative 4E-BP1 expression in mouse