Auto-regulation of miRNA biogenesis by let-7 and Argonaute

Recent studies from our lab provided a global map of interactions between Argonaute and endogenous mRNAs in C. elegans at the fourth larval (L4) stage of development⁶. Using cross-linking immunopurification with high-throughput sequencing (CLIP-seq), over 3,000 mRNA transcripts were found to have sequences bound by the C. elegans Argonaute Like Gene 1 (ALG-1). Surprisingly, the non-coding let-7 primary transcripts (pri-let-7), which are processed into the mature let-7 miRNA, were among the RNAs targeted by Argonaute, as indicated by the presence of an ALG-1 binding site towards the 3′ end of the transcripts (Fig. 1a and Supplementary Fig. 1). To verify that Argonaute associates with pri-let-7, we performed RNA Immunopurification (RIP) assays, using an anti-ALG-1 antibody, in L4 stage lysates from wild-type (WT) worms and alg-1(gk214) mutants, which lack the anti-ALG-1 epitope and exhibit slightly delayed development (Supplementary Fig. 2). A robust signal for pri-let-7 sequences was observed in WT but not alg-1(gk214) extracts (Fig. 1b and Supplementary Fig. 3a). Two unspliced (A and B) and one SL1 trans-spliced primary let-7 isoforms are produced by the let-7 gene in C. elegans⁷, and they all appear to associate with Argonaute (Supplementary Fig. 3a). Control actin mRNA or pri-mir-2 transcripts were not detected in ALG-1 immunopurifications, demonstrating the specificity of Argonaute association with primary let-7 transcripts (Fig. 1b and Supplementary Fig. 3a). Human Argonaute (Ago) proteins also associate with a subset of let-7 primary transcripts, suggesting that this interaction is conserved in higher organisms (Fig. 1c and Supplementary Fig. 3b).

Argonaute-mediated post-transcriptional regulation can result in mRNA degradation^4,5, and many established miRNA targets in C. elegans are up-regulated in alg-1(gk214) mutant worms⁶. To test if alg-1 also regulates the levels of primary let-7 transcripts, we performed Northern blotting on RNA extracted from WT and alg-1(gk214) worms at every two hours of development from mid L3, when mature let-7 starts to accumulate, until early adulthood. The three primary let-7 transcripts were detected at similar levels in WT and alg-1(gk214) worms at 24 hours of development (Fig. 1d). However, by four hours later the total let-7 primary transcript levels in WT had decreased to almost undetectable levels whereas the levels in alg-1(gk214) remained high (Fig. 1d–e). While the expression of let-7 primary transcripts in both strains still followed the previously described oscillation behaviour⁸, the cycling was muted in the alg-1(gk214) worms. Overall the pattern is consistent in independent biological replicates though the rapid cycling of let-7 can shift the expression peaks of individual time courses (Supplementary Fig. 3d–e). In contrast to let-7, the levels of other primary miRNA transcripts exhibited modest, if any, changes in alg-1(gk214) versus WT at the L4 stage, and none of these primary miRNAs were detected in ALG-1 RIP assays (Supplementary Fig. 3c and data not shown). At all time points, precursor let-7 (pre-let-7) accumulated to higher levels in alg-1(gk214) compared to WT, whereas mature let-7 was more abundant in WT (Fig. 1d,f–g and Supplementary Fig. 3d,f–g). An accumulation of precursor and diminished level of mature was observed for all miRNAs tested (mir-58, mir-90, lin-4), regardless of changes in primary transcript levels, in alg-1(gk241) compared to WT worms (Supplementary Fig. 3d and data not shown). These observations are consistent with previous reports of a general role for Argonaute in precursor processing and mature miRNA stabilization^9–12.

The Argonaute-bound region in let-7 primary transcripts is about 500 nucleotides downstream of the mature let-7 sequence and covers about 100 nucleotides, as mapped by ALG-1 CLIP-seq⁶ (Fig. 1a). To test if this site is required for interaction of ALG-1 with let-7 primary transcripts, transgenic strains with a single copy insertion of the let-7 locus lacking the ALG-1 binding site were created. These animals express the endogenous primary let-7 transcripts from the X Chromosome (WT pri-let-7) and primary let-7 transcripts with the ALG-1 binding site deleted (Δalg-1 pri-let-7) from Chromosome II. Both the WT and Δalg-1 pri-let-7 transcripts were expressed in the transgenic animals, but only the WT pri-let-7 with the ALG-1 binding site intact immunopurified with Argonaute (Fig. 2a and Supplementary Fig. 4a).

Compared to the control strain, worms harbouring the Δalg-1 pri-let-7 transgene expressed overall higher levels of primary let-7 (Fig. 2b and Supplementary Fig. 4b). In these assays, primary transcripts expressed from the endogenous let-7 locus are detected in addition to those of transgenic origin. To specifically analyse transgenic let-7 expression, we crossed the transgenic strains into the let-7(mn112) background. The let-7(mn112) mutation removes 190 base pairs just upstream and including the first nucleotide of the let-7 precursor, resulting in undetectable precursor or mature miRNA production¹³. Northern blotting showed higher levels of primary let-7 transcripts lacking the ALG-1 binding site compared to WT for most time points in the let-7(mn112) background (Fig. 2c and Supplementary Fig. 4c). Using primers that detect the unspliced versus SL1-spliced primary transcripts derived from the transgenes in qRT-PCR assays, we found that the SL1-spliced isoform was responsible for the increase detected in the Δalg-1 pri-let-7 transgenic worms (Fig. 2d and Supplementary Fig. 4d). Surprisingly, although the Δalg-1 pri-let-7 levels were elevated, the precursor and mature let-7 levels in those animals were 2- to 5-fold lower than the levels in their WT counterparts (Fig. 2c,e–f and Supplementary Fig. 4c,e–f). This deficiency reduced the rescue activity of the Δalg-1 pri-let-7 transgene in the let-7(mn112) background (Supplementary Fig. 4g).

We found a potential let-7 complementary site (LCS) within the ALG-1 binding region of pri-let-7 that is conserved in other Caenorhabditis species (Fig. 3a and Supplementary Fig. 5a,b). To test if let-7 miRNA mediates association of ALG-1 with primary let-7 transcripts, we performed RIP assays, using late L3 stage WT and let-7(n2853) mutant animals. The let-7(n2853) mutant animals harbour a point mutation (G→A) at the fifth nucleotide of the mature miRNA (Fig. 3a), which disrupts base-pairing to target mRNAs^13–17. Primary let-7 immunopurified with ALG-1 in WT, but not in let-7(n2853) extracts (Fig. 3b and Supplementary Fig. 6a), indicating that the association of Argonaute with primary let-7 transcripts requires the mature let-7 miRNA.

Northern blot analysis of primary let-7 in staged let-7(n2853) and WT animals during development from mid L3 until the mid-L4 stages showed similar or higher levels in let-7(n2853) compared to WT animals (Fig. 3c–d and Supplementary Fig. 6b–d). Despite the generally higher levels of primary let-7, precursor and mature let-7 were substantially reduced in let-7(n2853) compared to WT worms (Fig. 3c–f and Supplementary Fig. 6b–f).

Our cumulative data indicate that binding of ALG-1 via mature let-7 miRNA to pri-let-7 results in reduced transcript levels and increased mature levels, consistent with a role for ALG-1 in promoting the processing of primary let-7 transcripts. To further test this idea, we asked if introduction of WT let-7 would boost the levels of mature let-7 (n2853). Since the let-7(n2853) primary transcript has an intact LCS, the presence of WT mature let-7 miRNA is predicted to recruit ALG-1 to the transcript to promote processing and increase mature let-7(n2853) levels. Strains containing the WT let-7 gene integrated on chromosome II and the let-7(n2853) allele on the X chromosome [let-7(n2853); let-7(+)] were tested for effects on the levels of mature let-7(n2853) RNAs. Compared to the let-7(n2853) strain, the levels of mature let-7(n2853) were increased in the let-7(n2853); let-7(+) strain (Fig. 3g). This up-regulation is likely muted by destabilization of the mature miRNA attributed to the n2853 mutation, which impairs the target-mediated protection conferred to mature miRNAs by their target mRNAs¹⁵.

Since processing of miRNA primary transcripts typically occurs in the nucleus, we investigated the association of ALG-1 with pri-let-7 in this compartment. About 20% of ALG-1 protein localized to the nuclear fraction (Fig. 4a). Notably, almost half of the mature let-7 miRNA was associated with the nuclear fraction, which was double the amount of three other tested miRNAs (Fig. 4a). Using ALG-1 RIP assays of whole cell and nuclear fractions from WT animals, we observed strong and specific association of pri-let-7 with ALG-1 in both extracts (Fig. 4b).

Recently, the nuclear transport receptor Exportin-1 (XPO-1) was implicated in regulating primary miRNA processing through an unknown mechanism¹⁸. As previously reported, depletion of xpo-1 by RNAi resulted in a two-fold decrease in mature let-7 levels (Fig. 4c)¹⁸. Moreover, xpo-1(RNAi) reduced the nuclear localization of ALG-1 by a factor of 3.5 ± 0.1 (mean ± s.e.m., n = 5, P < 0.001), but did not affect total ALG-1 levels, compared to WT controls (Fig. 4d). Additionally, RIP assays revealed reduced association of ALG-1 with pri-let-7 in xpo-1 depleted animals (Fig. 4e). Although Exportin-1 is best characterized as a nuclear export factor, it has been shown to interact with Argonaute proteins in mammalian cells¹⁹ and, thus, it remains to be determined how direct the role is for XPO-1 in mediating nuclear interactions between ALG-1 and pri-let-7.

Our studies reveal a novel paradigm of ncRNAs as targets of miRNA induced silencing complexes. We find that Argonaute associates with let-7 miRNA primary transcripts in C. elegans and with a subset of pri-let-7 RNAs in human cells, indicating that transcripts other than mRNAs can also be miRNA targets and should be considered in target prediction endeavours. Argonaute is best characterized for its role in directing deadenylation and translational repression of bound mRNAs through incompletely resolved mechanisms^4,5. Although the mechanism by which ALG-1 regulates let-7 primary transcripts is yet to be fully elucidated, it appears to depend on the subcellular localization of Argonaute, as regulated by XPO-1, and to be independent of LIN-28 (Supplementary Fig. 7). Recently, miR-709 was reported to inhibit the processing of pri-miR-15a/16-1 in mouse cells²⁰. The discovery that the miRNA complex can regulate the processing of primary transcripts reveals a new role for Argonaute in the miRNA pathway. Moreover, this is the first example of a direct miRNA auto-regulatory loop, whereby mature let-7 binds and promotes processing of its own primary transcript. This amplification mechanism may be important for efficient production of mature let-7 from the oscillating levels of primary transcript substrates during C. elegans development. Our demonstration of direct miRNA regulation of let-7 biogenesis adds to the growing list of factors that impose post-transcriptional control on production of this important miRNA^8,21–27. Since misregulation of let-7 leads to disease²⁸, the let-7 positive feedback loop presents a new target for therapeutic interventions designed to restore appropriate miRNA levels.