GAM modulates E2F1 and Ras levels, increases cell apoptosis and reduces cell proliferation. (A) HEK-293 cells were transfected with either pCMV-HA or HA-GAM-2. GAM effects on E2F1 and Ras were analyzed by western blotting. (B) HEK-293 cells were transfected with either pCMV-HA or HA-GAM-4 along with a control RNA or pre-miR-92a-1. GAM effects on Ras were analyzed by western blotting. (C) Ras levels in HEK-293 cells transfected with either a pre-miR-control, pre-miR-92a-1 or a miR-92a antisense inhibitory RNA (92-I) were determined by western blotting. (D) HEK-293 cells transfected as indicated were assayed 24 h later for apoptosis or proliferation. Values represent the mean ± SD (n = 3). # and ##, Significantly different from apoptosis control, ^#P = 0.063, ^##P < 0.005. * and **, Significantly different from proliferation control, *P < 0.02, **P < 0.005.

GAM impairs the activation of the miR-17-92 promoter. (A) Schematic representation of the clones in pGL3-Basic containing different parts of the genomic region upstream of the miR-17-92 cluster. The first nucleotide of pre-miR-17 was arbitrarily used as nucleotide +1 for convenience. The ends of the DNA fragments used to generate the Clu-1 to Clu-5 constructs are indicated. The positions of one c-Myc-binding site (10) and of E2F-binding sites 1, 2 and 3, 4 (12) are indicated by an arrow. The approximate locations of TSS1 (transcription start site 1) (13) and TSS2 (21) are indicated by arrowheads. (B and C) Effects of GAM on the miR-17-92 promoter either alone (B) or in conjunction with c-Myc (C). MCF7 cells were transfected with the empty pGL3-Basic vector or the Clu-1 to Clu-5 constructs as indicated along with the empty pCMV-HA vector or a construct expressing GAM and/or c-Myc. The Firefly luciferase activity was measured 48 h after transfection and then normalized to the Renilla luciferase activity. Values represent the mean ± SD (n = 4). (D) Effects of siRNAs directed against GAM (siGAM) or against c-Myc transcripts (siMyc) on the miR-17-92 promoter expression either alone or in conjunction. MCF7 cells were transfected with the empty pGL3-Basic vector or the Clu-1 to Clu-4 constructs as indicated along with a control RNA, siGAM and/or siMyc. The Firefly luciferase activity was measured 48 h after transfection and then normalized to the Renilla luciferase activity. Values represent the mean ± SD (n = 4).

GAM opposes the upregulation of miR-17-92 miRNAs by TGFβ effectors. (A–C) HEK-293 (A and C) or HepG2 cells (B) were transfected with either a control RNA or siGAM (A and B), or with the empty pCMV-HA vector or pCMV-HA-GAM (C). After 34 h, they were either mock-treated (A and B) or treated with TGFβ (A–C) as indicated. The relative levels of the indicated miRNAs were determined by qRT–PCR. Values represent the mean ± SD (n = 3). The results obtained with mock-treated samples (M) in (A) are the same as those in Figure 3B. They are presented again along with the results obtained with TGFβ-treated samples (Tβ) for easier comparison, as they were drawn from the same experiment. Panel A: * and **, siGAM significantly different from Control RNA for each treatment, *, P < 0.005, **, P < 0.0005; #, TGFβ-treated Control RNA significantly different from mock-treated Control RNA, P < 0.0002; o and oo, TGFβ-treated siGAM significantly different from mock-treated siGAM, o, P < 0.003, oo, P < 0.0004. Panel B: * and **, siGAM significantly different from Control RNA for each treatment, *, P < 0.05, **, P < 0.002; #, TGFβ-treated Control RNA significantly different from mock-treated Control RNA, P < 0.02; o and oo, TGFβ-treated siGAM significantly different from mock-treated siGAM, o, P < 0.02, oo, P < 0.01. Panel C: *, Significantly different from Empty control vector, P < 0.05. (D) The relative levels of miR-17, miR-18a and miR-20a were determined by RNase-protection assays. Total RNAs extracted from HepG2 cells (10 µg per assay) treated and transfected as indicated (four left lanes) were hybridized with a radiolabelled RNA antisense probe complementary to the indicated miRNAs. The lengths of the antisense miR-17, miR-18a and miR-20a probes were made equal to that of the U6snRNA antisense probe by addition of a 3′-poly(A)-tail. Namely, the RNases used in the assay (RNase 1 and RNase T1) are not able to cut phosphodiester bonds following a non-paired A nt. The relative intensities of the protected fragments, given under each panel in percent of the control sample (lane 1), were calculated using the U6snRNA as an internal control. The efficiency of RNase digestion was assessed in parallel following the incubation of each probe with 10 µg yeast tRNAs (two rights lanes). ‘-RNase’ controls correspond to a 15-fold dilution of the undigested samples. Exposure time at −80°C was 36 h for miR-17, miR-18a and miR-20a probes, and 2 h for U6snRNA probe. Nd, non-digested, Ns, non-specific. (E,F) HepG2 cells were transfected with either a pCMV-HA/Control RNA mix (Control), a pCMV-HA-GAM/Control RNA mix (GAM) or a pCMV-HA/siGAM mix (siGAM). 34 h later, they were either mock-treated or treated with TGFβ. The relative levels of let-7d and let-7f (E) and of let-7e (F) were determined by qRT–PCR. Values represent the mean ± SD (n = 3). Panel E: #, mock-treated siGAM significantly different from mock-treated Control, P = 0.01; *, TGFβ-treated siGAM significantly different from mock-treated siGAM, P = 0.03. Panel F: *, Significantly different from the corresponding mock-treated sample, *, P < 0.02.

Working model depicting the intricate interaction loops between GAM, TGFβ effectors, Drosha, miR-17-92 and let-7 miRNAs. For clarity, only the main interactions have been represented. Arrows indicate activating effects, and T-barred arrows inhibitory effects. The effects of c-Myc on the miR-17-92 cluster were previously described (10). See text for more details.

GAM transcripts are direct targets of miR-17-92 and let-7 miRNAs. (A) Schematic representation of GAM 3′-UTR constructs in pGL3-Control (Luc-1, -2, -3 and -4) and pCMV-HA (HA-GAM-1, -2, -3 and -4). Conserved putative target sites for miR-17/20a (17,20), miR-92a-1 (92) and let-7 (L7) miRNAs are indicated. The positions of the 5′- and 3′-limits of the inserts with respect to GAM 3′-UTR are indicated over the lines, and their respective lengths are given in bp below the lines. (B–D) Luciferase assays were done with Luc-1 to Luc-4 as indicated, both in sense (S) and antisense (AS) orientation. Mutations in constructs used in D are given in Supplementary Tables S1 (miR-17-92 miRNA target sites) and S2 (let-7 miRNA target sites). Bars show the ratios of the Firefly luciferase (normalized to Renilla luciferase) activity measured following transfection with the indicated pre-miRNAs to that obtained following transfection with the pre-miR Control for the same construct. Values represent the mean ± SD (n = 4). (B and C) *P < 0.001, significantly different from pre-miR Control. (D) *P < 0.005, significantly different from wild type construct. (E and F) Endogenous GAM levels in HEK-293 cells transfected with either a pre-miR-Control, the indicated pre-miRNAs or antisense inhibitory RNAs to miR-17, miR-20a or miR-92a (17-I, 20-I and 92-I, respectively) were determined by western blotting.

GAM downregulates miR-17-92 miRNAs. (A and B). The relative levels of miR-17-92 miRNAs in HEK-293 cells transfected with either the empty pCMV-HA vector or pCMV-HA-GAM (A), or with either a control RNA or siGAM (B), were determined by qRT–PCR. Values represent the mean ± SD (n = 3). (A) * and **, Significantly different from empty control vector, *P < 0.004, **P < 0.0004. (B) *P < 0.005, significantly different from control RNA. (C) HEK-293 cells were transfected with either a control pCMV-HA/Control RNA mix or a construct in the pCMV vector expressing the whole miR-17-92 cluster (Clu-81) in combination with either a pCMV-HA/Control RNA mix, a pCMV-HA-GAM/Control RNA mix, or a pCMV-HA/siDrosha mix. Clu-81 contains a 3837-bp insert bearing the whole miR-17-92 cluster sequence (787 bp) plus 216-bp upstream of the first nucleotide of pre-miR-17 and 2834-bp downstream of the last nucleotide of pre-miR-92a-1. The relative levels of the indicated miRNAs were determined by qRT–PCR. Values represent the mean ± SD (n = 3). *P < 0.01, significantly different from control mix. (D) HEK-293 cells were transfected with Clu-81 along with either the empty pCMV-HA vector or pCMV-HA-GAM. The relative levels of RNAs containing unprocessed miR-92a-1 (42-nt long protected band) were determined by RNase-protection assays using 2 µg total RNA per assay. The efficiency of RNase digestion was assessed in parallel following the incubation of each probe with the same amount of yeast tRNAs (two left lanes). ‘-RNase’ controls correspond to a 15-fold dilution of the undigested sample. The relative intensities of the protected fragments, calculated using the U6snRNA as an internal control, are given in percent of the pCMV-HA control. Nd, non-digested, Ns, non-specific.

GAM downregulates Drosha, a critical effector of miRNA maturation, and GAM-Drosha interaction is RNA-dependent. (A and C)The relative levels of Drosha transcripts in HEK-293 transfected with either the empty pCMV-HA vector or HA-GAM (A), or with either a Control RNA or siGAM (C), were determined by qRT–PCR. Values represent the mean ± SD (n = 3). *P < 0.05, significantly different from control RNA. (B and D) Drosha levels in extracts prepared from HEK-293 cells transfected with either the empty pCMV-HA vector or HA-GAM-2 (B), or with either a Control RNA or siGAM (D), were analyzed by western blotting. The blots in B were from the same experiment as the blots of Figure 2A. (E) The relative levels of GAM transcripts in HEK-293 transfected with either the Control RNA or siRNAs directed against Drosha transcripts (siDrosha) were determined by qRT–PCR. Values represent the mean ± SD (n = 3). *P < 0.001, significantly different from Control RNA. (F–H) HEK-293 cells were transfected with the empty pCMV-HA vector or pCMV-HA-GAM. In (F), cells were cotransfected with a Flag-Drosha construct. In (H), extracts were submitted to RNase digestion before immunoprecipitation as indicated. Immunoprecipitations were conducted using anti-Flag (F) or anti-HA antibody (G and H). Immunocomplexes were analyzed by western blotting with the indicated antibodies. In (G), the left blot (WCL) shows the levels of endogenous Drosha in whole cell lysates of HEK-293 cells transfected as indicated, before immunoprecipitation (these blots were prepared using 1/25th of the whole lysate subsequently used for immunoprecipitation).

TGFβ and miR-17-92 miRNAs downregulate GAM transcripts, while GAM impairs the canonical TGFβ signaling pathway. (A) The levels of GAM transcripts in HepG2 cells transfected with a pre-miR Control (Control) or the indicated pre-miRNAs, and either mock-treated or treated with TGFβ, were determined by qRT–PCR. Values represent the mean ± SD calculated from triplicate qRT–PCR reactions of the same cDNAs. #, TGFβ-treated sample significantly different from the corresponding mock-treated sample, P < 0.003. * and **, Pre-miR mock-treated significantly different from mock-treated pre-miR Control, *, P < 0.002, **, P < 0.001. (B) GAM levels in Hep-G2 cells mock-treated or treated with TGFβ were analyzed by western blotting. (C,D) The relative levels of GAM transcripts in HEK-293 cells transfected with either the empty pCMV vector (Empty) or Clu-81 (C), or either mock-treated or treated with TGFβ (D) were determined by qRT–PCR. Values represent the mean ± SD calculated from three independent reverse transcription reactions, each analyzed by qRT–PCR in triplicates. Panel C: *, Significantly different from Empty control vector, P < 0.001. Panel D: *, Significantly different from mock-treatment, P < 0.001. (E) After cotransfection with the SMAD2/SMAD3/SMAD4 luciferase reporter plasmid along with either a pCMV-HA/Control RNA mix, a pCMV-HA-GAM/Control RNA mix or a pCMV-HA/siGAM mix, HepG2 cells were mock-treated or treated with TGFβ. The Firefly luciferase activity was measured 48 h after transfection and then normalized to the Renilla luciferase activity. Values represent the mean ± SD (n = 6). *, Significantly different from Control, P < 0.005. (F) The effects of GAM or siGAM on p21/CDKN1A expression in HepG2 cells mock-treated or treated with TGFβ were analyzed by western blotting.