Noncanonical role of ALAS1 as a heme-independent inhibitor of small RNA-mediated silencing

Abstract

microRNAs (miRNAs) and small interfering RNAs (siRNAs) are 21- to 22-nucleotide RNAs that guide Argonaute-class effectors to targets for repression. In this work, we uncover 5-aminolevulinic acid synthase 1 (ALAS1), the initiating enzyme for heme biosynthesis, as a general repressor of miRNA accumulation. Although heme is known to be a positive cofactor for the nuclear miRNA processing machinery, ALAS1-but not other heme biosynthesis enzymes-limits the assembly and activity of Argonaute complexes under heme-replete conditions. This involves a cytoplasmic role for ALAS1, previously considered inactive outside of mitochondria. Moreover, conditional depletion of ALAS activity from mouse hepatocytes increases miRNAs and enhances siRNA-mediated knockdown. Notably, because ALAS1 is the target of a Food and Drug Administration-approved siRNA drug, agents that suppress ALAS may serve as adjuvants for siRNA therapies.

Fig. 1.. Human ALAS1 represses miRNA biogenesis independently of Microprocessor.

(A) Western blot validation of single and double knockouts of ALAS1 and CPOX in HEK293T cells. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; WT, wild type. (B) Primary miRNA transcripts were not affected in ALAS1-KO cells. (C) A miRNA processing sensor, consisting of a luciferase transcript with a pri-miRNA in its 3′UTR, was unaffected in ALAS1-KO and CPOX-KO cells grown in conventional media. However, the sensor increased upon the depletion of heme from FBS. ns, not significant. (D) Validation that ALAS1-KO and CPOX-KO cells are intracellularly compromised for heme biogenesis. (E) In conventional heme-replete media, ALAS1-KO cells increase multiple mature miRNAs. (F) ALAS1-KO increases miRNAs isolated from Ago2 complexes. The loading volume of each eluate was adjusted to ensure similar Ago2 levels across samples. (G) Small RNA-seq normalized to spike-in controls shows increases in the majority of miRNAs in ALAS1-KO cells. (H) Effects on mature miRNAs are selective to ALAS1-KO and not CPOX-KO. Analysis of variance (ANOVA) with Dunnett’s test [(B), (C), and (D)] was applied for statistical analysis. P values for each comparison are indicated.

Fig. 2.. ALAS1 inhibits RISC assembly independently of other heme biogenesis factors.

(A) Validation of Drosha CRISPR knockout in wild-type and independent ALAS1-KO HEK293T cell lines cultured in serum. (B) In vitro RISC assembly using cell lysate from (A). Limiting cellular miRNAs by Drosha-KO slightly increased both pre- and holo-RISC in wild-type cells, with a more robust enhancement observed in ALAS1-KO cells. (C) Quantification of holo-RISC assembly in reactions from (B). The latest time point (60 min) was assessed by ANOVA with Dunnett’s test. The error bars represent the standard deviation; **P < 0.01 and ***P < 0.001 compared with the wild-type control. (D) Design of a tRNA-mir-451 vector that generates miRNAs independently of Drosha and Dicer RNase III enzymes. In some experiments, this miRNA construct is coupled to an RFP-miR-451 sensor. (E) ALAS1 is epistatic to CPOX with respect to the inhibition of miRNA function, downstream of RNase III enzymes. ANOVA with Dunnett’s test was applied for statistical analysis. P values for each comparison are indicated. (F) Transient CRISPR-Cas9–mediated mutagenesis of heme biosynthesis factors in the tRNA-mir-451/RFP-4xmiR-451 sensor clonal cell line. Only knockout of ALAS1 generates a population with reduced miRNA sensor expression, reflecting enhanced silencing. The incomplete ALAS1 depletion in these transient experiments (fig. S6E) explains why only a subpopulation of cells exhibits enhanced silencing. (G) Heme (hemin) administration assay. Wild-type and ALAS1-KO cells were cultured with the indicated concentrations of additional hemin for 16 hours, and ALAS1 expression and miRNAs were analyzed by either Western blotting or Northern blotting, respectively. In wild-type cells, culturing with high concentrations of hemin (5 and 10 μm) markedly reduced ALAS1 protein expression and increased miRNA levels, whereas hemin administration in ALAS1-KO cells had no further effect.

Fig. 3.. Ago2 protein levels are regulated by cytoplasmic ALAS1.

(A) siRNA-mediated transient knockdown of ALAS1 in HEK293T cells increases Ago2 protein levels. (B) Quantification of Ago2 protein levels from three independent ALAS1 knockdown experiments. (C) Schematic of the ALAS1 protein structure. Three heme regulatory motifs (CP motifs) are located in the N-terminal region of cytoplasmic precursor ALAS1. Two of these motifs (CP1 and CP2) mediate mitochondrial translocation, whereas the remaining motif (CP3) is retained in mature mitochondrial ALAS1. The catalytically active lysine residue (K445) binds to substrate glycine to initiate 5-aminolevulinic acid synthesis for heme production. aa, amino acid. (D) The increased Ago2 protein levels in ALAS1-KO cells are rescued by both wild-type full-length ALAS1 and truncation mutants lacking the N-terminal mitochondrial localization signals (CP1 and CP2 motifs). However, mutation of the catalytically active lysine to alanine fails to rescue Ago2 protein levels in either construct. (E) Quantification of Ago2 protein levels from three independent experiments, as shown in (D). (F) miRNA analysis by Northern blotting shows that the increase in miRNA levels in ALAS1-KO cells is rescued by the same manipulations described in (D). ANOVA with Dunnett’s test [(B) and (E)] was applied for statistical analysis. P values for each comparison are indicated.

Fig. 4.. ALAS1 knockout enhances mRNA targeting by Ago2-miRNA complexes.

(A) Meta-profiles of Ago2 CLEAR-CLIP signals at miRNA target sites show enhanced occupancy in ALAS1-KO HEK293T cells in heme-replete conditions. (B) Directional increase in occupancy of miRNAs and target-engaged miRNAs in Ago2 complexes in ALAS1-KO cells. CPM, counts per million. (C) Example of locus (SERTAD3) with enhanced Ago2 occupancy at a conserved miRNA site for a seed family; chimeric reads identify the guide miRNA specifically as miR-92a-3p. bp, base pair. (D) Validation of enhanced association of Ago2 with SERTAD3 in ALAS1-KO cells. RT-qPCR, reverse transcription quantitative polymerase chain reaction. (E) Repression of a luciferase-SERTAD3 3′UTR sensor is enhanced in ALAS1-KO cells and lost in DICER-KO cells. ANOVA with Dunnett’s test [(D) and (E)] was applied for statistical analysis. P values for each comparison are indicated.

Fig. 5.. Endogenous ALAS represses miRNA accumulation in hepatocytes.

(A) Design of a conditional mouse allele of Alas1. nt, nucleotide; fs, frame shift. (B) Scheme to deplete endogenous Alas1 and Alas2 from mouse hepatocytes through SA-Cre-ERT2–mediated conditional inactivation of Alas1 combined with siRNA-mediated suppression of Alas2 (i.e., Alas1/2-mut). i.p., intraperitoneal; s.c., subcutaneous. (C) Validation that the Alas1/2-mut liver lacks ALAS1 and ALAS2 proteins. The asterisk indicates a nonspecific band. (D) Northern blotting of RNAs shows substantial increase in mature miRNAs in hepatocytes purified from Alas1/2-mutants; each lane represents hepatocytes purified from an independent control or mutant liver. (E) Small RNA-seq normalized to spike-in controls shows general increase in most mature miRNAs in ALAS-depleted hepatocytes. (F) Enhanced silencing by ectopic miR-144 expressed in ALAS1-KO HEK293T cells grown in serum. (G) Enhanced silencing by Hcp1 siRNA in Alas1/2-depleted hepatocytes [mouse protocol as in (B), except that 0.0001 mg/kg Hcp1 siRNA was introduced subcutaneously at day 5, before harvest at day 6]. (H) Enhanced silencing by Hcp1 siRNA in hepatocytes from Alas1/2-siRNA injected animals. ANOVA with Dunnett’s test [(F), (G), and (H)] was applied for statistical analysis. P values for each comparison are indicated.