lin-4 and the NRDE pathway are required to activate a transgenic lin-4 reporter but not the endogenous lin-4 locus in C. elegans

As the founding member of the microRNA (miRNA) gene family, insights into lin-4 regulation and function have laid a conceptual foundation for countless miRNA-related studies that followed. We previously showed that a transcriptional lin-4 reporter in C. elegans was positively regulated by a lin-4-complementary element (LCE), and by lin-4 itself. In this study, we sought to (1) identify additional factors required for lin-4 reporter expression, and (2) validate the endogenous relevance of a potential positive autoregulatory mechanism of lin-4 expression. We report that all four core nuclear RNAi factors (nrde-1, nrde-2, nrde-3 and nrde-4), positively regulate lin-4 reporter expression. In contrast, endogenous lin-4 levels were largely unaffected in nrde-2;nrde-3 mutants. Further, an endogenous LCE deletion generated by CRISPR-Cas9 revealed that the LCE was also not necessary for the activity of the endogenous lin-4 promoter. Finally, mutations in mature lin-4 did not reduce primary lin-4 transcript levels. Taken together, these data indicate that under growth conditions that reveal effects at the transgenic locus, a direct, positive autoregulatory mechanism of lin-4 expression does not occur in the context of the endogenous lin-4 locus.


Introduction
The lin-4 miRNA is the founding member of the miRNA gene family, and a critical regulator of developmental timing in C. elegans [1]. lin-4 loss-of-function mutants display severe developmental phenotypes, including abnormal seam cell division and differentiation patterns, and a complete failure in vulval morphogenesis [2]. lin-4 is strongly upregulated toward the end of the first larval (L1) stage, resulting in the suppression of its key target, lin-14, to promote L2-specific developmental events [3]. Previous reports have identified the FLYWCH (FLH) family of transcription factors [4], as well as the Period homolog LIN-42 [5][6][7], as repressors of lin-4 expression during C. elegans development. However, FLH transcription factors were found to primarily affect embryonic lin-4 expression, whereas lin-4 was only mildly derepressed in lin-42 mutant larvae. Thus, during larval development, the molecular mechanisms that regulate the timing and activation of lin-4 expression remain unclear. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 miRNA promoter::GFP gene fusions have been instrumental in uncovering the transcriptional regulation and expression patterns of numerous miRNAs. A~500bp promoter region is sufficient to drive lin-4 expression and rescue the lin-4(e912) null phenotype [1]. We and others have shown that animals carrying a GFP reporter driven by this~500bp lin-4 promoter (Plin-4::GFP) begin expressing GFP in the seam cells in late L1, consistent with the reported timing of lin-4 upregulation as measured by Northern blotting and quantitative PCR (qPCR) [8,9]. Moreover, we discovered that a lin-4 complementary element (LCE), as well as lin-4 itself, was necessary for Plin-4::GFP expression [10]. This suggested that the lin-4 miRNA may function in a highly non-canonical manner to transcriptionally activate its own expression. miRNAs typically function in the RNA interference (RNAi) pathway, acting as specificity factors to recruit Argonaute proteins to silence targeted transcripts [11]. In contrast, the term RNA activation, or "RNAa", has been used to describe a phenomenon by which small RNAs complementary to promoter regions induce the transcriptional activation of the downstream gene [12]. While the mechanisms of RNAa remain poorly understood, one mechanistic similarity between RNAa and RNAi appears to be the requirement of an Argonaute protein as the effector of gene regulation [13]. In this study, we report that the four major nuclear RNAi (nrde) factors in C. elegans, including the nuclear Argonaute NRDE-3 (nuclear RNAi defective 3) [14,15], are necessary for Plin-4::GFP expression. Taken together with our previous work [10], these findings strongly supported the model of a direct, positive feedback loop in the regulation of lin-4 expression.
However, we further show that this potential autoregulatory mechanism is not active at the endogenous lin-4 locus under the conditions examined here. nrde mutants did not display significantly altered lin-4 expression, nor any detectable lin-4 phenotypes. Similarly, CRISPR--Cas9-mediated mutations of the LCE and of the mature lin-4 sequence did not result in any measurable effects on endogenous lin-4 promoter activity. Thus, our work describes a gene regulatory mechanism active in a transgenic context, but not at the endogenous gene locus. These results emphasize the importance of validating the endogenous relevance of cis-regulatory elements identified through reporter-based experiments.

lin-4 reporter expression is activated by nuclear RNAi factors
The lin-4 miRNA and an LCE in the lin-4 promoter are required for the expression of a Plin-4:: GFP reporter [10]. Given that miRNAs function within Argonaute complexes, we hypothesized that lin-4 may bind to the LCE and activate Plin-4::GFP expression through the action of the nrde genes. To test this, we depleted nrde-2 and nrde-3 by RNAi in the zaIs1(Plin-4::GFP) line, and measured the effect on seam cell GFP expression. Both RNAi treatments significantly reduced seam cell GFP in L2 and L3 animals, suggesting that nrde-2 and nrde-3 are positive regulators of Plin-4::GFP activity (Fig 1A). To validate these results, and to investigate additional nrde genes, we crossed the zaIs1(Plin-4::GFP) line into nrde-1, nrde-2, nrde-3 and nrde-4 mutant animals. We found that seam cell GFP expression was completely abolished in each of the four nrde mutant backgrounds (Fig 1B and 1C). We conclude that the nuclear RNAi pathway is required for the expression of a lin-4 reporter in C. elegans seam cells.
lin-4 overexpression is sufficient to upregulate Plin-4::GFP activity [10]. To test if this was dependent on NRDE-3, we overexpressed lin-4 in the zaIs1;nrde-3 mutant line. Similar to what we previously observed in a wild-type background, we found that lin-4 overexpression resulted in a~four-fold increase in seam cell GFP expression ( Fig 1D). These data suggest that the nuclear Argonaute NRDE-3 is not required for the lin-4-mediated activation of Plin-4::GFP expression. Thus, the precise mechanisms through which both lin-4 and the nrde pathway  activate lin-4 reporter expression remain to be determined. While the nuclear RNAi pathway is well known to silence transgenes [16], our data represents, to our knowledge, the first example of nrde factors in positively affecting the expression of a transgene. How this occurs in the context of the lin-4 reporter may provide new insights into small RNA pathways and the balance between silencing and allowing/inducing the expression of different foreign sequences.

Nuclear RNAi factors are not required for mature lin-4 expression
Do the nrde genes also positively regulate endogenous lin-4 expression? To test this, we generated a nrde-2;nrde-3 double mutant, and performed qPCR to measure primary and mature lin-4 levels in synchronized (12h) L1's-a time at which lin-4 expression begins to increase during wild-type development. We observed only a mild decrease in primary lin-4 levels in nrde-2; nrde-3 mutants compared to wild-type, without a significant decrease in mature lin-4 ( Fig 1E  and 1F). To determine whether the NRDE pathway may regulate endogenous lin-4 expression specifically in seam cells, we examined two seam cell phenotypes which manifest with 100% penetrance in lin-4(e912) null mutants. First, we examined the L2-specific division in V-lineage seam cells, which fail to occur in the absence of lin-4. Second, we examined the formation of adult alae, a cuticular structure that does not form in lin-4(e912) mutants due to a failure of seam cell differentiation. We found that the timing of L2 seam cell divisions was unaffected in nrde-3 mutants, while adult alae formation was also unaffected in nrde-2;nrde-3 mutants (Table 1). Taken together, these results suggest that while nrde-2 and nrde-3 are required for seam cell Plin-4::GFP activity, they do not regulate endogenous lin-4 expression.

Deletion of the endogenous LCE does not affect mature lin-4 expression
The LCE in the lin-4 promoter is essential for the expression of a Plin-4::GFP reporter [10]. To test the regulatory importance of the LCE in the endogenous lin-4 locus, we targeted the LCE using CRISPR-Cas9 [17] and generated two LCE mutant C. elegans lines. lin-4-LCE(za25) harbors a 25nt deletion that removed the entire 17nt LCE and 4nt on either side; lin-4-LCE(za26) harbors a 2 nucleotide (TT) deletion (Fig 2A). These mutants were backcrossed into a wildtype (N2) background three times before further analysis. The wIs79(ajm-1::GFP; scm-1::GFP) seam cell marker strain was used to visualize seam cell divisions. Synchronized L1's were plated on op50 and were grown at 25˚C for 15h prior to scoring. Animals were scored positive as long as at least one of the six V-lineage seam cell completed the L2-specific symmetrical division prior to a second, asymmetrical division, producing a maximum of six additional seam cells. To examine the presence of alae in young adults, synchronized L1's were plated on op50 and scored 54h post-feeding. Both LCE mutants appeared morphologically wild-type, and displayed no obvious lin-4 phenotypes, including no significant changes in brood size (Fig 2B). In wild-type animals, we found lin-4 to be upregulated~1500-fold between 10h and 16h of post-embryonic development ( Fig 2C); however, the timing and magnitude of this upregulation was unaffected in lin-4-LCE(za25) mutants. We conclude that deletion of the endogenous LCE does not affect mature lin-4 miRNA expression.

The LCE is not required for endogenous lin-4 promoter activity
The steady-state level of mature miRNAs can be influenced by numerous regulatory mechanisms beyond transcriptional control, including RNA degradation pathways and the processing of the primary and precursor miRNAs [18]. Thus, it was possible that the endogenous LCE did indeed function as a transcriptional regulatory element, but that its effect on lin-4 expression was masked by post-transcriptional regulatory mechanisms. The lin-4 gene resides in an intronic region of the F59G1.4 host gene, a poorly characterized gene expressed at relatively low and invariant levels throughout C. elegans development. lin-4 is located in the 9 th intron of the F59G1.4a isoform,~300bp downstream of an antisense transcript (F59G1.12), and~200bp upstream of an exon and alternative transcriptional start site of its host gene (F59G1.4b) ( Fig  3A). We profiled the temporal expression profile of F59G1.4a, F59G1.4b, F59G1.12, host intron 9, and pri-lin-4, in wild-type and lin-4-LCE(za25) mutants. This analysis provided a unique overview of the relationship between host gene, antisense RNA, and pri-miRNA expression, in a defined time frame in which the developmentally programmed upregulation of lin-4 occurs.
First, we found that deletion of the LCE does not affect the steady-state levels of any of these transcripts, at any of the time points examined (Fig 3B, i-v). Second, the trend in expression changes over time was very similar between all transcripts: RNA levels begin to increase around 10h post-embryonic development, peak at 16-18h, and then decrease to approximately their original starting levels by 24h. However, the magnitude of these changes are different. Pri-lin-4 levels show the highest fluctuation, increasing over 10-fold at peak expression ( Fig  3B, iv). Together with the expression profile of mature lin-4, these data support a previously suggested, host gene-independent mechanism of lin-4 expression.
However, our investigation of the host intron and flanking exons revealed a moderate, but previously unrecognized, contribution of host gene transcription to lin-4 upregulation. We found the intron, and both isoforms of the F59G1.4 host gene mRNA, to be upregulated between~1.5 and 3-fold between 10h-16h of post-embryonic development, suggesting that the activation of lin-4 expression in late L1's is not solely due to independent transcription, but partly involves host gene activation as well (Fig 3B, i, ii, iv).
A study by Bracht et al. [9] reported a~4-fold increase at peak expression in pri-lin-4 levels relative to the host intron, by semi-quantitative RT-PCR. Our act-1-normalized qPCR data reveals a~2.5-fold increase in host gene intron levels and a~10-fold increase in pri-lin-4 levels ( Fig 3B, ii, iv). Thus, relative to the host intron, we also find pri-lin-4 expression to be increased 4-fold. The remarkable consistency between our results not only adds confidence to the accuracy of our findings but is likely a reflection of the highly regulated nature of lin-4 miRNA biogenesis.
As a final test of LCE function, we performed chromatin immunoprecipitation (ChIP)-qPCR to assay RNA polymerase II (RNAPII) recruitment to the lin-4 locus in synchronized L1 animals. While our results in wild-type larvae confirm ModEncode RNAPII ChIP-Seq data, we found no significant changes in RNAPII occupancy between wild-type and lin-4-LCE (za25) mutants at any of the intronic DNA regions examined (Fig 3C). We conclude that the LCE not required for the transcriptional activity of the endogenous lin-4 promoter.

The temporal expression profile of F59G1.12, a lin-4 promoter-associated antisense RNA, is not affected by the LCE
We also examined the temporal expression profile of F59G1.12, a lin-4 promoter-associated antisense RNA, for three reasons. First, it contains the LCE sequence. Second, antisense transcripts have been implicated in both activating and repressing roles in regulating proximal gene expression [19]. Third, we wished to determine whether its expression was separately regulated from that of lin-4, which would suggest a potential regulatory function. However, the nearly identical temporal expression profile of the F59G1.12 antisense RNA with that of prilin-4 indicates that it is likely a product of bidirectional transcription, where high levels of RNA polymerase activity in the sense direction often produces a measurable amount of antisense transcripts. The fact that F59G1.12 expression is unaffected in the lin-4-LCE(za25) mutant further confirms that the LCE does not influence the transcriptional activity of the endogenous lin-4 promoter.

lin-4 autoactivation is not an endogenous regulatory mechanism
We previously showed that Plin-4::GFP expression is reduced in a lin-4(e912) mutant background [10]. The lin-4(e912) mutant carries a large deletion that removes lin-4 as well as~5kb of upstream sequence [1], making it impossible to measure its effects on the endogenous lin-4 promoter activity. To disrupt lin-4 miRNA function while preserving its promoter, we used CRISPR-Cas9 to generate small indel mutations within the mature lin-4 sequence. We generated three mutant lines, each displaying a fully penetrant lin-4(e912) phenotype (Fig 4A and  4B). We suspected that these mutations may also impair the processing of the pre-and pri-miRNA due to abnormal hairpin structures [20], potentially confounding our interpretation of pri-miRNA levels as a readout of lin-4 promoter activity. Indeed, all three mutants displayed elevated primary lin-4 expression compared to wild-type ( Fig 4C). However, since the lin-4 (za24) mutant displayed the smallest relative accumulation of primary lin-4 transcripts compared to wild-type (a~two-fold increase at 12h post-embryonic development), we compared the temporal expression profile of pri-lin-4 in wild-type and lin-4(za24) mutants. Despite a possible mild defect in pri-lin-4 processing, the similarity in the magnitude and timing of the peak in pri-lin-4 levels in wild-type and lin-4(za24) mutants suggests lin-4 is not required for the activity of its own endogenous promoter (Fig 4D).

Conclusions
In summary, together with our previous work, we have identified the three core components of a miRNA regulatory module (namely lin-4, the LCE, and the nuclear Argonaute NRDE-3 and its co-factors), as positive regulators of Plin-4::GFP expression. However, here we show that none of these three components function as activators of the endogenous lin-4 promoter. How might we explain this discrepancy? It is possible that lin-4 may somehow positively regulate the translation of Plin-4::GFP-derived transcripts; indeed, examples of miRNA-mediated translational activation have been reported [21,22]. It is also possible that the Plin-4::GFP reporter construct, which is integrated in multiple copies in an unknown genomic location in the zaIs1 line [8], is subject to locus-specific and/or multicopy gene-specific regulatory mechanisms that are not active at the endogenous lin-4 promoter. We do note that we have generated multiple independent strains carrying non-integrated, extragenic copies of the Plin-4::GFP reporter, which behave similarly to the zaIs1 line, suggesting that the integration site itself is not likely the culprit. Finally, specifically regarding the importance of the LCE, it is possible Differential regulation of endogenous and transgene-derived lin-4 miRNA expression previously described. Presumably, this is partly due to the difficulty of testing candidate regulatory sequences in endogenous contexts, at least prior to the advent of CRISPR-Cas9-mediated gene editing tools. Our work provides a cautionary tale for the interpretation of promoter elements within transgenic reporter constructs, and underscores the importance of validating the endogenous relevance of regulatory sequences identified in reporter-based systems.

Brood size assay
Individual N2 and LCE mutant~L3 animals were picked onto 6cm NGM plates with op50. After 7 days, the number of adult F1s were counted. In this time none of the F2 progeny became adults, ensuring an accurate count of the F1 brood size.

Microscopy
For seam cell GFP expression, V-lineage divisions, and alae production, animals were immobilized in 1mM levamisole and examined using an upright Zeiss Axioplan microscope under 40x and 63x magnification.
For qPCR analyses of transcripts around the endogenous lin-4 locus, animals were bleached and embryos were allowed to hatch overnig ht in M9. The following day, the time at which synchronized L1's were plated onto op50 plates was considered 0hr and the start of the time course. Total RNA was collected, DNase-treated, and reverse transcribed as described above, with two exceptions: (1) The F59G1.12 antisense transcript was reverse transcribed using a single, gene-specific primer F59G1.12 RT1: 5'-CGTCTCTGTGGCACCTAACA-3'; (2) U18 and mature lin-4 RNA were reverse transcribed with Taqman probes RT00176 and RT00258, respectively, according to the manufacturer's protocol. All qPCR primer efficiencies were tested and only those with an efficiency between 85-115% were used. All gene expression was normalized to act-1 mRNA, with the exception of mature lin-4 miRNA, which was normalized to U18. Fold changes were calculated using the ΔΔCt method. Primers used in this study are shown in Table 2.