Inducible and reversible inhibition of miRNA-mediated gene repression in vivo

Although virtually all gene networks are predicted to be controlled by miRNAs, the contribution of this important layer of gene regulation to tissue homeostasis in adult animals remains unclear. Gain and loss-of-function experiments have provided key insights into the specific function of individual miRNAs, but effective genetic tools to study the functional consequences of global inhibition of miRNA activity in vivo are lacking. Here we report the generation and characterization of a genetically engineered mouse strain in which miRNA-mediated gene repression can be reversibly inhibited without affecting miRNA biogenesis or abundance. We demonstrate the usefulness of this strategy by investigating the consequences of acute inhibition of miRNA function in adult animals. We find that different tissues and organs respond differently to global loss of miRNA function. While miRNA-mediated gene repression is essential for the homeostasis of the heart and the skeletal muscle, it is largely dispensable in the majority of other organs. Even in tissues where it is not required for homeostasis, such as the intestine and hematopoietic system, miRNA activity can become essential during regeneration following acute injury. These data support a model where many metazoan tissues primarily rely on miRNA function to respond to potentially pathogenic events.

MicroRNAs (miRNAs) are short non-coding RNAs that in Metazoa repress gene expression at the post-transcriptional level by binding to partially complementary sequences on target mRNAs [1][2][3][4] . MiRNAs act as part of a large ribonucleoprotein complex known as the miRNA-Induced Silencing Complex (miRISC). In mammals, the Argonaute protein family (AGO1-4) and the Trinucleotide Repeat-Containing gene 6 protein family (TNRC6A/GW182, TNRC6B and TNRC6C) are the core components of the miRISC. AGO binds to the miRNA, and facilitates its interaction with target mRNAs 5 . In turn, TNRC6 binds to AGO and recruits the decapping and deadenlyation complexes, leading to degradation of target mRNAs [6][7][8][9][10][11][12][13][14][15][16] . Although miRNAs are abundantly expressed in embryonic and adult mouse tissues, and computational and experimental analyses indicate that they target components of virtually every cellular process 17 , animals harboring targeted deletion of single miRNA genes are often indistinguishable from their wild type counterparts [18][19][20][21][22][23][24][25] . One explanation for these observations is that the redundant functions of related miRNAs may buffer the emergence of obvious phenotypes in mutant animals 1,3 . Interestingly, however, clear phenotypes often emerge in mutant adult animals when exposed to external or internal perturbations 19,23,26 . These observations suggest that, at least in some contexts, miRNA function is conditionally, rather than constitutively, required to carry on cellular processes. Previous efforts to investigate the consequences of global inhibition of miRNA function have relied upon the targeted deletion of the core miRNA biogenesis factors DICER, DROSHA, and DGCR8 (reviewed in ref. 27 ). Several animal models harboring conditional or constitutive knockout alleles of these genes have been generated [28][29][30][31][32][33][34][35][36] . Although these strategies have provided important insights into miRNA biology, they suffer from several limitations. First, inactivation of these gene products is known to have other consequences in addition to impairing miRNA biogenesis. For instance, DICER is involved in epigenetic regulation in the nucleus in a miR-NA-independent manner [37][38][39][40][41][42] , and is essential to metabolize transcripts from short interspersed nuclear elements, predominantly Alu RNAs in humans and B1 and B2 RNAs in rodents 43 . DROSHA, on the other hand, regulates the expression of several coding and non-coding RNAs by directly cleaving stem-loop structures embedded within the transcripts 44 . Furthermore, DICER and DROSHA are also involved in ribosomal RNA biogenesis 45 and in the DNA-damage response 46 47 , and DGCR8 regulates the maturation of small nucleolar RNAs and of some long non-coding RNAs 48,49 . Consequently, the phenotypes observed in these models cannot be solely attributed to inhibition of miRNA activity. Another limitation of conditional ablation of miRNA-biogenesis genes in vivo is that due to their high stability mature miRNAs can persist for several days after their biogenesis is inhibited. For example, four weeks after near complete conditional ablation of Dicer1 in the muscle, the levels of the highest expressed miRNAs were found to be only reduced by 30-40% and their expression remained substantial even 18 months later 24 . This complicates the interpretation of experiments based on temporally controlled conditional ablation of these biogenesis factors, especially in non-proliferating tissues. Third, a subset of mammalian miRNAs does not rely on the canonical biosynthesis pathway, and therefore their expression and activity are not affected by inactivation of the core miRNA biogenesis factors 44,[50][51][52][53][54][55] . Finally, these genetic approaches are not reversible and therefore these animal models cannot be used to study the effects of transient inhibition of miRNA function. To circumvent these limitations, we have generated a novel genetically engineered mouse strain that allows inducible and reversible disassembly of the miRISC, thereby achieving controllable inhibition of miRNA-mediated gene repression in vivo without affecting small RNA biogenesis. To address the reliance of adult tissues on miRNA-mediated gene repression, we have used this novel strain to investigate the consequences of acute inhibition of the miRISC under homeostatic conditions, and during tissue regeneration.

Results
Inhibition of the miRNA pathway through peptide-mediated disruption of the miRISC. Multiple motifs within the N-terminal domain of 1 TNRC6 proteins contain regularly spaced tryptophan residues which mediate the interaction between AGO and TNRC6 by inserting into conserved hydrophobic pockets located on AGO's Piwi domain 56,57 . A peptide encompassing one of the AGO-interacting motifs of human TNRC6B has been previously employed as an alternative to antibody-based approaches to efficiently pull down all AGO family members from cell and tissue extracts 58,59 . This peptide, named T6B, competes with endogenous TNRC6 proteins for binding to AGOs. However, as it lacks the domains necessary for the recruitment of de-capping and de-adenylation factors, it prevents the assembly of the full miRISC, thus resulting in effective inhibition of miRISC-mediated gene repression in cells 58,60 . Based on these results, we reasoned that temporally and spatially controlled expression of a T6B transgene in animals would offer the unprecedented opportunity to study the consequences of acute and reversible inhibition of miRNA function in vivo without interfering with miRNA biogenesis or abundance (Fig. 1a).
To test the suitability of this approach, we first investigated the dynamics of interaction between T6B and the miRISC in mouse and human cell lines. We employed a previously reported size exclusion chromatography (SEC)-based assay 61,62 to analyze the molecular weight of AGO-containing complexes in lysates from cells expressing either a doxycycline-inducible FLAG-HA-T6B-YFP fusion protein (hereafter referred to as T6B), or a mutant version (hereafter referred to as T6B Mut ) incapable of binding to AGO (Extended Data Fig. 1). We reasoned that if T6B expression prevents AGO from stably binding to TNRC6 and its targets, AGO proteins should be detected in fractions corresponding to approximately 120-130 kDa, the sum of the molecular weights of AGO (approximately 95 kDa) and the T6B fusion protein (approximately 30 kDa). In contrast, unperturbed AGO complexes that are part of the fully assembled miRISC bound to mRNAs should elute in the void of the column, which contains complexes larger than 2 MDa (Fig. 1b). As expected, in lysates from cells expressing no T6B or T6B Mut AGO2 and TNRC6A were mostly detected in the high molecular weight frac- Change (T6B or T6B Mut vs control) of predicted targets for each conserved miRNA family was calculated, converted to a z-score and is plotted on the x-axis against the miRNA family abundance (log of the sum of read counts for each member of the family). The size of each circle is proportional to the number of predicted targets. A positive z score indicates that the targets for that family are preferentially upregulated upon T6B expression, while a negative score would indicate preferential downregulation. Expression of T6B, but not of T6B Mut , causes preferential upregulation of miRNA targets of the most miRNA families and the effect is roughly proportional to each miRNA family abundance. vs control) of predicted targets for each conserved miRNA family was calculated, converted to a z-score and is plotted on the x-axis against the miRNA family abundance (log of the sum of read counts for each member of the family). The size of each circle is proportional to the number of predicted targets. A positive z score indicates that the targets for that family are preferentially upregulated upon T6B expression, while a negative score would indicate preferential downregulation. Expression of T6B, but not of T6B Mut , causes preferential upregulation of miRNA targets of the most miRNA families and the effect is roughly proportional to each miRNA family abundance. 2 tions, indicating the presence of target-bound miRISC (Fig. 1c). In contrast, AGO2 and TNRC6A were nearly completely depleted from the high molecular weight fractions in lysates from cells expressing T6B (Fig. 1c). Moreover, while AGO2, TNRC6A and the polyA-binding protein 1 (PABP1) cofractionated in lysates from control cells, they eluted in different fractions in lysates from T6B-expressing cells (Fig. 1c), indicating that T6B leads to loss of interactions between the miRISC components and mRNAs. As expected based on the strong evolutionary conservation of human and mouse AGO and TNRC6 proteins 59, 63, 64 , we obtained identical results when human T6B was expressed in mouse embryo fibroblasts (MEFs) (Extended Data Fig. 2).
To test whether the redistribution of AGO-containing complexes induced by T6B expression was mirrored by a loss of miRNA-mediated gene repression, we performed RNAseq analysis on MEFs expressing T6B or T6B Mut . Cells expressing T6B displayed marked and selective de-repression of predicted mRNA targets for expressed miRNAs (Fig.  1d). The extent of de-repression was roughly proportional to the abundance of individual miRNA families, with predicted targets of poorly expressed miRNAs collectively showing modest de-repression compared to targets of more abundantly expressed miRNA families (Fig.  1d). Importantly, de-repression of miRNA targets was not accompanied by a global change in mature miRNAs levels ( Fig. 1e), consistent with the role of T6B in perturbing the effector step of the miRNA pathway, without affecting miRNA processing. Of the four mammalian AGO proteins, AGO2 is the only one that has endo-ribonucleolytic activity, which does not require TNRC6 65 and is triggered when the AGO2-loaded small RNA and the target are perfectly complementary [66][67][68] . AGO2's catalytic activity is essential for gene regulation in the germline. For example, in mouse oocytes, AGO2 loaded with endogenous small-interfering RNAs (endo-siRNAs) mediates the cleavage of coding and non-coding transcripts bearing perfectly complementary sequences 42,69 . In metazoan somatic tissues, in contrast, AGO2 catalytic activity is mainly involved in the biogenesis of miR-486 and miR-451 in the hematopoietic system 50,70 , and in occasional instances of miRNA-directed cleavage of mRNAs 71 . Importantly, T6B expression does not interfere with the ability of synthetic siRNAs to cleave perfectly complementary endogenous targets (Fig. 1f), indicating that AGO2's catalytic function is not affected by the binding of T6B, and implying that the loading of small RNAs onto AGOs is also not perturbed by T6B.
Collectively these results demonstrate that ectopic T6B expression in mammalian cells causes global inhibition of miRISC function with minimal perturbation of the expression of mature miRNAs, and with preservation of AGO2's endo-nucleolytic activity.
Generation of a mouse strain with inducible expression of a T6B transgene. To apply this general strategy to an in vivo setting, we next generated mouse embryonic stem cells (mESCs) expressing a doxycycline-inducible T6B transgene. We used a knock-in approach in which the doxycycline-inducible transgene is inserted into the Col1A locus of mESC expressing the reverse tetracycline-controlled transactivator (rtTA) under the control of the endogenous Rosa26 (R26) promoter 72 (Fig. 2a). Targeted mESCs were tested for the ability to express the T6B transgene in response to doxycycline (Extended Data Fig. 3) and then used to generate mice with genotype R26 rtTA/rtTA ; ColA1 T6B/T6B (hereafter R26 T6B ). R26 rtTA/rtTA ; ColA1 +/+ mice, with untargeted ColA1 loci but expressing rtTA served as negative controls (hereafter R26 CTL ). Upon doxycycline administration we observed strong expression of T6B in R26 T6B mice and across most adult tissues (Fig. 2b). Notable exceptions were the central nervous system (Fig. 2b and Extended Data. Fig.  4), probably due to low blood-brain barrier penetration of doxycycline, and the skeletal muscle and the heart, most likely due to low expression of the rtTA transgene in these tissues 73 .
When doxycycline was administered in the diet, T6B became detectable after 24h, reached a plateau after three days, and completely disappeared four days after doxycycline removal from the diet (Fig. 2c).
Because colon and liver expressed uniformly high levels of T6B in response to doxycycline, we used these tissues to test the effects of T6B expression on miRISC activity in vivo. Co-IP experiments using antibodies directed to T6B confirmed the interaction between AGO and T6B in these tissues ( Fig. 2d and Extended Data Fig. 5). Expression of T6B resulted in nearly complete disassembly of the miRISC, as indicated by the elution shift of AGO from the high molecular weight to low molecular weight fractions in both tissues ( Fig. 2e and Extended Data Fig.  6). Importantly, doxycycline removal from the diet led to a complete reconstitution of the miRISC, as indicated by the reappearance of AGO2 in the high molecular weight fractions (Fig. 2e).
To test whether T6B expression also resulted in inhibition of miR-NA-mediated gene repression in vivo, we performed RNAseq on total RNAs extracted from the liver and colon of R26 T6B and R26 CTL mice kept on doxycycline-containing diet for one week. As shown in Fig. 2f, T6B expression resulted in marked de-repression of miRNA targets in both tissues.
Based on these results we conclude that T6B expression allows acute and reversible disruption of the miRISC, and concomitant inhibition of miRNA function in vivo.
Consequences of miRISC disruption in adult tissues under homeostatic conditions. Given the central role of miRNAs in gene regulatory networks, one might expect widespread phenotypes emerging when miRISC function is systemically inhibited. Consistent with this hypothesis continuous doxycycline administration starting at conception caused embryonic lethality (Fig. 3a), while inhibition of miRISC starting at mid-gestation caused developmental defects and perinatal lethality in R26 T6B mice ( Fig. 3b and Extended Data Fig. 7). Surprisingly, however, adult R26 T6B mice kept on doxycycline diet for up to two months remained healthy and appeared normal upon macroscopic and histopathologic examination. Detailed examination of the intestine confirmed extensive T6B expression in the epithelium and in the mesenchymal compartment (Extended Data Fig. 8) but no architectural abnormalities were observed (Fig. 3c). Cells in the crypts showed no significant changes in expression pattern of Ki67 protein (Extended Data Fig. 9), suggesting that the proliferation and turnover of the epithelium is maintained even in absence of a functional miRISC. No significant change in the number of goblet cells was detected throughout the intestine (Extended Data Fig. 10), and mice maintained normal body mass throughout the period of doxycycline treatment (Extended Data Fig. 11), suggesting that general intestinal functions were not affected. Although no obvious macroscopic, functional, or architectural abnormalities were caused by T6B expression in the intestine, we observed a reduction in lysozyme expression in Paneth cells in the crypts (Fig. 3d, upper row). However, this phenotype was reversible, as lysozyme signal in the crypts returned to normal levels when doxycycline was removed from the diet (Fig. 3d, lower row), suggesting that T6B expression did not affect neither the viability of intestinal stem cells, nor their self-renewal ability.
Complete blood counts showed a modest, but significant, decrease in erythrocytes volume (MCV) and hemoglobin content (MCH) in R26 T6B RBCs ( Fig. 3e and table 1), analogously to what reported in mice harboring targeted deletion of miR-451 74 . Flow cytometric analysis of bone marrow showed a 3-fold depletion in Pre-B cells as well as a significant decrease in immature and mature circulating B cells in R26 T6B mice. We also observed a reciprocal increase in the frequency of Pro-B cells in the bone marrow of these animals ( Fig. 3f and Extended Data Fig. 12).
These results are reminiscent of the partial block in B cell differentiation observed upon deletion of the miR-17~92 cluster 75 . Further characterization of hematopoietic stem cells showed that the number of long-term repopulating hematopoietic stem cells (LT-HSC) was unaffected after 3 weeks of doxycycline exposure. However, we observed a modest decrease in short-term repopulating HSCs (ST-HSCs) and a concomitant increase in multipotent progenitors (MPPs) relative to controls ( Fig. 3g and Extended Data Fig. 13).
Collectively, these data suggest that in a subset of adult tissues miRISC function can be suppressed with minimal or no consequences on the ability of these tissues to maintain homeostasis.

miRISC disruption impairs the regeneration of injured colon epithelium.
Several studies have shown that the phenotype caused by targeted deletion of individual miRNAs often manifests only after the mutant animals are subjected to "stress" 19,23,26,76 . For example, ablation of miR-143/145 causes no apparent phenotype under homeostasis but severely impairs the ability of the mutant animals to respond to acute damage to the intestinal epithelium 19 . Prompted by these reports, and by our initial observation that prolonged T6B expression does not substantially affect intestinal homeostasis, we tested the consequences of miRISC disruption on the regenerating intestine. A cohort of R26 T6B and R26 CTL mice were kept on doxycycline-containing diet for ten days, after which they were treated with dextran sulfate sodium (DSS), which induces severe colitis in mice 19,77 . A significant and progressive loss of body mass was observed in both groups during DSS treatment and two days following DSS removal (Fig.  4a). However, R26 T6B mice lost body mass more rapidly than controls and reached critical health conditions seven days after DSS removal. Three days after DSS removal, control animals started to regain weight, reaching the initial body mass within five days after DSS removal (Fig. 4a). In contrast, R26 T6B mice failed to fully recover (Fig. 4a), and all reached a humane endpoint within five days after DSS removal from the diet (Fig. 4b).
Histological analysis confirmed that DSS treatment induced the disruption of the architecture of the epithelium, and the appearance of ulcerative areas to a similar extent in both R26 T6B and R26 CTL control mice, (Fig. 4c and Extended Data Fig. 14). In contrast, although five days after DSS removal the integrity of the colonic epithelium of control mice was largely reestablished with the exception of isolated dysplastic areas (Extended Data Fig. 15), extensive ulcerated regions persisted in the colon of R26 T6B mice (Fig. 4c). Importantly, we observed the presence of dysplastic epithelium in R26 T6B mice during and after DSS treatment, indicating that miRISC disruption does not completely abolish the potential of cells to proliferate, as also confirmed by Ki67 staining (Fig. 4d). Therefore, we speculate that other factors, such as impaired stem cell maintenance or differentiation, may be responsible for the increased susceptibility of T6B-expressing colon to DSS treatment. Chivukula and colleagues have shown that defective intestinal regeneration in the colon of miR-143/145-deficient mice is associated with upregulation of the miRNA-143 target IGFBP5 in the mesenchymal compartment. The increased levels of IGFBP5 protein cause the inhibition of IGF1R signaling in the epithelium through a non-cell autonomous mechanism, which ultimately prevented epithelial regeneration 19 . Consistent with their findings, in situ hybridization analyses in the colon of DSS-treated R26 T6B mice showed a significant upregulation of IGFBP5 mRNA in the mesenchymal compartment compared to controls (Fig. 4e). The extent of de-repression of IGFBP5 was comparable to that previously observed in miRNA-143/145 knockout mice 19 , providing further evidence that T6B-mediated miRISC disassembly is an effective strategy to globally inhibit miRNA function in vivo. Collectively, these results support a model whereby miRNA-mediated gene regulation, while dispensable to maintain normal colon homeostasis, becomes critical for its regeneration following acute damage. miRISC disruption impairs regeneration of the hematopoietic system. To further characterize the consequences of miRISC inhibition during tissue regeneration, we explored the possibility that other tissues may adopt a similar dynamic reliance on miRNA function. Along with the intestinal epithelium, blood is one of the most rapidly turned over tissues in mice. Hematopoietic stem cells (HSCs) reside as a    predominantly quiescent population in the bone marrow and are rapidly induced to re-enter the cell cycle in response to external cues, such as infection or injury 78 . Furthermore, HSCs can be readily isolated by flow cytometry and transplanted, allowing the study of mechanisms underlying regeneration at the single cell level.
To test the consequences of miRISC disruption in the regenerating hematopoietic system, we treated R26 T6B and R26 CTL mice on doxycycline-containing diet with a single dose of the cytotoxic drug 5-fluorouracil (5FU). 5-FU selectively depletes rapidly proliferating hematopoietic progenitors and leads to a compensatory increase in LT-HSC proliferation. Flow cytometry analysis of the bone marrow seven days after 5FU-injection showed that T6B expression prevented this compensatory increase in LT-HSC. We observed an identical phenotype when R26 T6B and R26 CTL mice that were bled repeatedly over a 3-week period to induce LT-HSC to re-enter the cell cycle (Fig. 5a). The decreased number of HSCs in the bone marrow of R26 T6B mice after a single 5-FU challenge compared to controls, suggested that miRISC disruption impaired HSCs' ability to re-enter the cell cycle and regenerate the hematopoietic compartment. Consistent with this hypothesis, when injected with repetitive 5-FU doses, R26 T6B mice showed significantly shorter survival compared to controls (Fig. 5b).
To more directly measure the regenerative capacity of HSCs in a context where T6B would only be expressed in hematopoietic cells, we performed competitive transplantation of T6B-expressing (CD45.2 + ) and wild-type (CD45.1 + ) bone marrows (1:1 ratio) into lethally irradiated hosts. The recipient animals were divided into four groups as shown in Fig. 5c: (i) a control group that was never administered doxycycline; (ii) a group maintained on a doxycycline-containing diet throughout the duration of the experiment (8 weeks); (iii) a group treated with doxycycline starting 4 weeks after transplant; and (iv) a group that was on doxycycline for only the first 4 weeks after transplant. Blood samples were taken at 4 and 8 weeks following the start of the experiment for analysis (Fig. 5c). This experiment was designed to test the prediction that expression of T6B during the first 4 weeks following transplant, when the regenerative demand is highest and when we hypothesize miRNA-mediated gene repression is required, would more severely affect the ability of donor cells to contribute to the recipient hematopoietic reconstitution compared to T6B expression after homeostasis is reestablished. Consistent with this prediction, mice that were administered doxycycline in the first 4 weeks post-transplant had significantly fewer CD45.2 + peripheral blood mononuclear cells (PBMCs) (Fig. 5d). Contribution to the B cell population was particularly impaired by T6B expression but this was reversed once the recipients were taken off of doxycycline, consistent with the developmental block described earlier (Fig. 3d and Extended Data Fig. 13). Interestingly, the decrease in total CD45.2 + PBMCs and CD45.2 + myeloid cells was not reversed by doxycycline withdrawal, which suggested the T6B-expressing CD45.2 + HSCs might have been outcompeted by wild-type CD45.1 + HSCs in these recipients (Fig. 5d). Consistent with this hypothesis we observed a significant reduction in CD45.2 + HSCs only in the bone marrow of recipient animals that were fed a doxycycline-containing diet in the first 4 weeks post-transplant (Fig. 5e).
Taken together, these results support a model where the miRNA-mediated gene regulation is conditionally essential for the maintenance of hematopoietic stem cells during acute regeneration but is largely dispensable under homeostasis.
An essential role for miRNA-mediated gene repression in the skeletal muscle and in the heart. As previously discussed, we observed low or no expression of T6B in the heart and skeletal muscle of R26 T6B mice treated with doxycycline (Extended Data Fig. 4), consistent with previous reports indicating that rtTA expression from the endogenous R26 promoter is tissue restricted 73 . To extend the analysis of the phenotype caused by the loss of miRISC activity to these tissues, we crossed T6B transgenic mice with the Rosa26-CAGs-rtTA3 strain 79 in which the modified chicken beta-ac- tin with CMV-IE enhancer (CAG) promoter 80 drives a more ubiquitous expression of the rtTA variant rtTA3 (hereafter CAG T6B ). As expected, the pattern and intensity of T6B expression upon dox administration in CAG T6B mice and R26 T6B mice were largely overlapping, except for the heart and the skeletal muscle, for which significant T6B expression was only observed in CAG T6B mice ( Fig. 6a and Extended Data Fig. 4). RNAseq analyses confirmed inhibition of miRNA function in both heart and skeletal muscle of CAG T6B mice upon dox administration (Fig. 6b).
In contrast to R26 T6B mice, CAG T6B mice fed a doxycycline-containing diet showed a progressive decline in body mass (Extended Data Fig. 16), and died or reached a humane endpoint within 4-6 weeks (Fig. 6c). The decrease in body mass was not caused by intestinal malabsorption as, similarly to what observed in R26 T6B mice, we found no evidence of architectural defects throughout the intestine. In contrast, histopathologic examination of heart and skeletal muscle showed severe alterations in both organs, including dilated cardiomyopathy and diffuse muscular degeneration (Fig. 6d). All mice also showed necro-inflammatory changes in the liver, variable alterations in the pancreas, and increased urea nitrogen and alanine aminotransferase levels in the serum (data not shown). Such alterations are likely secondary to congestive heart failure, and/or to severe muscle catabolism as they were not observed in R26 T6B mice. The emergence of severe cardiac and skeletal muscle phenotypes, as opposed to the lack of obvious structural and functional abnormalities in the majority of T6B-expressing tissues, points toward the existence of sig-nificant differences among adult tissues in their reliance on the miRNA pathway during homeostasis.

Discussion
We report the generation of a novel genetically engineered mouse strain in which miRISC assembly and function can be temporally and spatially controlled in a reversible manner by a doxycycline-inducible transgene encoding a T6B-YFP fusion protein to address the role(s) miRNA-mediated gene regulation plays in vivo in adult tissues. Surprisingly, in most adult tissues, we do not find an essential role for miR-NA-mediated gene repression in organ homeostasis. A notable exception are the heart and the skeletal muscle, where miRISC inactivation in adult mice results in acute tissue degeneration and death even in the absence of tissue damage or exogenous stress. Even though miRISC function is not overtly required for the homeostasis of other tissues, we have investigated the consequences of miRNA inhibition in the intestine and in the hematopoietic system of adult mice under homeostatic conditions and during tissue regeneration. These are tissues that periodically respond to external/internal stresses. In both tissues we have found that miRISC activity is dispensable for homeostasis. However, miRNA function becomes essential during tissue regeneration following acute injury. These results lend experimental support to the hypothesis that a major role for miRNA-mediated gene repression is to support tissue adaptation to stress. four cardiac chambers in hearts of CAG T6B mice compared to controls (n = 9 for each genotype). Despite having thinner walls, the histomorphology of ventricular cardiomyofibers were within normal limits. Bottom row: representative H&E staining showing degenerative and regenerative changes in the skeletal muscle of the hind limbs of CAG T6B mice compared to controls (n = 9 for each genotype).

Skeletal muscle
In previous studies where Dicer1 was conditionally ablated in the skeletal muscle of adult mice, muscle regeneration was impaired after acute injury, but no effect on muscle morphology or function was observed during homeostasis 24,81,82 . An explanation for this difference is that in the Dicer1 conditional knockout experiments miRNA levels were only partially reduced even weeks after Dicer1 ablation, likely reflecting the high stability of these short non-coding RNAs. The T6B mouse strain we describe here overcomes this major limitation and allows the rapid and effective inhibition of miRNA-activity independently from the half-life of these molecules.
In this manuscript we have focused on the role of miRNA-mediated gene repression in adult mice. The same strategy for the acute inhibition of miRISC-activity can in principle be applied to other organisms. We have found that expression of T6B in embryos of both sea urchin (Paracentrotus lividus) and zebrafish (Danio rerio), induces developmental defects and gene expression changes consistent with the essential role of the miR-NA pathway during development [83][84][85][86][87][88][89] (Extended Data Fig. 17). Considering that in vitro T6B efficiently binds to AGO proteins from different non-mammalian organisms 58 , these findings are not unexpected, yet they highlight the usefulness of the T6B system for dissecting the miRNA pathway in a variety of animal models. Despite its many advantages, the T6B mouse strain has also some unique limitations that need to be considered when designing and interpreting experiments. First, although our biochemical and computational analysis of cells and tissues expressing T6B indicate that the peptide can effectively impair miRISC function, we cannot exclude some residual miRISC activity even in cells expressing high levels of the T6B transgene. The observation that we can recapitulate phenotypes observed in mice harboring complete targeted deletion of miR-143/145 miRNAs in the intestine 19 and of miR-17~92 and miR-451 in the hematopoietic system 74,75,90 is reassuring in this respect. For example, consistent with observations made in the regenerating intestine of miRNA-143/145 knockout mice 19 , we did not record any abnormalities or toxicity during the normal intestinal homeostasis of R26 T6B mice, whereas T6B expression became lethal during intestinal regeneration. Moreover, in the hematopoietic system, abnormalities were mostly restricted to B cell maturation, which are consistent with a developmental block at the Pro-B to Pre-B transition found in mir17~92 knockout mice 75 . Finally, we also observed a statistically significant decrease in hematocrit, erythrocyte volume and hemoglobin content in adult T6B-expressing mice, analogous to what reported in mice harboring targeted deletion of miR-451 74 .
In contrast, some of our results markedly differ from results obtained by conditional ablation of Dicer1 in mice. For example, conditional knockout of Dicer1 in the hematopoietic system has been reported to result in the rapid depletion of HSCs 91 . Furthermore, the lack of an overt phenotype in the intestine contrasts with previous reports showing that post-natal, conditional deletion of Dicer1 results in depletion of Goblet cells 92,93 , in addition to abnormal vacuolation and villous distortion in the small intestine 31,93 . We cannot exclude that these differences are due to an incomplete inactivation of the miRISC pathway in T6B mice, but an alternative explanation is that they reflect the well-characterized miRNA-independent functions of DICER. Another limitation to be considered is the possibility that T6B expression impairs the activity of other complexes in addition to the miRISC. Although RNAseq analysis of cells expressing T6B has not revealed changes that are not explained by loss of miRNA-mediated gene repression and the phenotypes observed are consistent with loss of miRNA activity, this possibility cannot be formally excluded at this time. Further studies to experimentally identify T6B interactors in cells and tissues will be important to formally address this possibility.
In conclusion, we have developed a novel mouse strain that enables investigating the role of miRNA-mediated gene repression in adult organisms. The body of data presented here suggest that in adult animals miRNAs primarily provide for the ability to adaptively change gene expression in response to the physiologic and pathologic stresses that accompany metazoans' life. It is likely that the specific miRNAs and stresses differ based on the adult organ or tissue being studied and the model we have generated will be useful address these important aspects of miRNA biology.

Methods
Animal models. The Rosa26 rtTA/rtTA ; ColA1 T6B/T6B (R26 T6B ) mice were generated by site-specific integration of the transgene coding for the FLAG-HA-T6B-YFP fusion protein within the Col1a locus of KH2 embryonic stem cells (Col1A-frt/ Rosa26 rtTA) 72 . Briefly, the FLAG-HA-T6B-YFP (FH-T6B-YFP) DNA fragment was subcloned into the targeting vector, as described in "Vectors and molecular cloning". A mixture of 5mg of the targeting vector and 2.5mg of the pCAGGS-flpE-puro (Addgene #20733), Flippase recombinase-expressing vector were electroporated into KH2 cells, using 4D-Nucleofector core unit (Lonza), following manufacturer's "Primary cells P3" protocol. Selection of targeted clones was initiated 48h after electroporation, using 150mg hygromycin per mL of culture medium. 10 days later, individual hygromycin-resistant ES cell clones were analyzed by PCR to confirm correct integration of the knock-in allele. Clones carrying the correctly integrated knock-in allele were genotyped using a three-primer PCR, with the following primers: 1) 5'-AATCATCCCAGGTGCACAGCATTGCGG-3'; 2) 5'-CTTTGAGGGCTCATGAACCTCCCAGG-3'; 3) 5'-ATCAAGGAAAC-CCTGGACTACTGCG-3' . A 287pb-long PCR product indicates successful integration of the transgene into the Cola1, while a 238bp-long PCR product indicates a wild type, untargeted locus. Two independent ES clones were injected into C57BL/6J albino blastocysts and backcrossed the resulting chimeras to C57BL/6J mice to achieve germline transmission of the recombinant allele. F1 animals were then intercrossed to generate animals expressing rtTA from the R26 locus under control of the R26 endogenous promoter, while expressing the T6B fusion protein from the Col1a locus under control of the tetracycline-responsive element (TRE) and the minimal CMV promoter. Animals were genotyped as follows: to assess the presence of the transgene in the ColA1 locus, PCR was carried out as for the genotyping of KH2 cells. To assess the presence of the rtTA transgene in the Rosa26 locus, a three-primer PCR was performed, using the following primers: 1) 5'-AAAGTCGCTCTGAGTTGTTAT-3'; 2) 5'-GCGAA-GAGTTTGTCCTCAACC-3'; 3) 5'-CCTCCAATTTTACACCTGTTC-3' . A 350pb-long PCR product indicates the presence of the rtTA transgene into the Rosa26 locus, while a 297bp-long PCR product indicates the presence of a wild type locus. CAG rtTA/rtTA ; Col1A T6B/T6B (CAG T6B ) mice were generated by backcrossing R26 T6B with Rosa26-CAGs-rtTA3 mice (a gift from Scott Lowe, MSKCC). In the Rosa26-CAGs-rtTA3 mice, the knock-in allele has the CAG promoter driving the expression of the third-generation reverse tetracycline-regulated transactivator gene (rtTA3), all inserted into the Gt(ROSA)26Sor locus. In vivo doxycycline-dependent expression of the FLAG-HA-T6B-YFP transgene was achieved by feeding mice chow that contained doxycycline at the concentration of 625mg/Kg (Envigo #TD01306). Mice were maintained and euthanized in accordance with a protocol approved by the Memorial Sloan-Kettering Cancer Center Institutional Animal Care and Use Committee.
Immunohistochemistry. For IHC, deparaffinized sections were subjected to antigen retrieval and processed with the EnVision+ HRP kit (K401111-2, DAKO, Glostrup, Denmark) according to the manufacturer's instructions. A primary polyclonal antibody against Ki67 (Cell Signaling #12202) at 1:400 dilution was diluted in Antibody Diluent (DAKO #S0809) and incubated overnight at 4°C. Next, sections were incubated in the provided anti-rabbit HRP-labeled polymer reagent and detection was performed according to the manufacturer's protocol. Images were acquired using an Olympus BX-UCB slide scanner.
In situ hybridization. 5μm sections were obtained from formalin-fixed, paraffin-embedded (FFPE) colons from age/sex-matched mice. Before staining, tissue slides were deparaffinized, rehydrated and permeabilized according to standard procedures. Detection was carried out using RNAscope 2.5 HD Detection Reagent, BROWN (ACD # 320771), with a specific RNAScope Igfbp5 Probe (ACD #425738, according to the manufacturer's instructions.
Serum chemistry and hematology. For serum chemistry, blood was collected into tubes containing a serum separator, the tubes were centrifuged, and the serum was obtained for analysis. Serum chemistry was performed on a Beckman Coulter AU680 analyzer and the concentration of the following analytes was determined: alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, creatine kinase, gamma-glutamyl transpeptidase, albumin, total protein, globulin, total bilirubin, blood urea nitrogen, creatinine, cholesterol, triglycerides, glucose, calcium, phosphorus, chloride, potassium, and sodium. Na/K ratio, albumin/globulin ratio were calculated. For hematology, blood was collected retro-orbitally into EDTA microtainers. Automated analysis was performed on an IDEXX Procyte DX hematology analyzer.

Dextran sulfate sodium (DSS) treatment and post DSS treatment quantitative analyses.
Mice kept in doxycycline-containing chow were treated for 5 days with 4% w/v DSS (FW 40.000) (Cayman Chemical #23250) dissolved in drinking water. Body mass was monitored daily. Measurements of colon length, aggregated length of ulcers, percentage of colon with ulcers, area of ulcers, the number of immune nodules and the area of immune nodules were obtained using OMERO (https:// www.openmicroscopy.org/omero/). Measurements of these parameters were used to estimate the extent of damage and colitis induced by DSS treatment. All measurements were acquired from H&E-stained colon sections. Ulcer was defined as regions of colon with complete/partial loss of formal epithelial structural, accompanied by massive immune infiltrates. Colon length was measured by tracing the length of muscular layer of each colon. Length of ulcer was measured as the added length of each ulcerated region along the colon. Ulcer percentage was calculated as the length of ulcer/length of colon. The area of each individual ulcer was also measured and summed for each animal. Clear immune nodules are visible, showing aggregates of immune cells with high nucleus/cytoplasm ratio. Number and area of the immune nodules were summarized for each animal.
Tissue isolation and total lysates preparation. Organs extracted from 8-to 12-week-old mice, perfused with PBS, were snap-frozen in liquid nitrogen and stored at −80 °C until further processing. To prepare total extract from solid tissues, tissues were pulverized using a mortar, resuspended in 1mL of lysis buffer per cm3 of tissue, and dounce-homogenized with a tight pestle until completely homogenized. Next, extracts were cleared by centrifugation at 20,000 × g for 5 min followed by a second step of centrifugation at 20,000 × g for 5 min. To prepare total extracts from cultured cells, pelleted cells were snap frozen in liquid nitrogen and stored at −80 °C until further processing. Pellets were then resuspended in lysis buffer, incubated for 10 minutes on ice, and cleared by centrifugation at 20,000 × g. Two different lysis buffers were used, depending on the specific downstream application. For IP and size exclusion chromatography, lysates were prepared in SEC buffer (150 mM NaCl, 10 mM Tris-HCl pH 7.5, 2.5 mM MgCl2, 0.01% Triton X-100). For Western blotting applications, lysates were prepared in RIPA buffer (Sigma-Aldrich # R0278). Upon usage, both buffers were supplemented with the addition of EDTA-free complete protease inhibitors (Sigma-Aldrich #11836170001), phosphate inhibitors (Roche #04906837001), and 1mM DTT.
Flow cytometry. Analysis of bone marrow populations was performed by harvesting femurs and tibiae from euthanized mice. Bone marrow was isolated by centrifugation (REF), resuspended in FACS buffer (PBS with 2% fetal calf serum) and passed through a 40µm cell strainer to make a single cell suspension. Nonspecific antibody binding was blocked by incubation with 10µg/ml Rat IgG (Sigma #I-8015) for 15 min on ice. Antibodies used to identify HSCs included a cocktail of biotinylated lineage antibodies (Gr1, CD11b, TER119, B220, CD3, CD4, CD8), CD117 (c-kit) APC (2B8), Sca-1 (D7) PE-cy7, CD150 PE, and CD48 Pacific Blue. B cell progenitors were identified with the following antibodies: B220, CD19, CD25, CD43, IgM, IgD and c-kit. For analysis of peripheral blood mononuclear cells, blood was collected retro-orbitally from live mice into EDTA microtainers. Whole blood was lysed in ACK buffer for 5 min at room temperature, washed with FACS buffer and pelleted prior to antibody staining. Mature blood populations were identified with the following antibodies: CD45.1, CD45.2, Gr1, CD11b, B220, CD3. Cells were incubated with primary antibodies for 45 min, washed once with FACS buffer and incubated with BV711 streptavidin conjugate for 15 min. All incubations were carried out on ice and protected from light. Antibodies were purchased from Biolegend or eBioscience.
Size exclusion chromatography (SEC). SEC was performed using a Superose 6 10/300 GL prepacked column (GE Healthcare) equilibrated with SEC buffer. A total of 400μL (1.5-2 mg) of precleaned total extracts either from cultured cells or tissues were run on the SEC column at a flow rate of 0.3 mL/min. 1mL fractions were collected. Proteins were extracted from each fraction by TCA precipitation following standard procedures, and run on SDS-PAGE gels for Western blotting analysis.
Western blotting and antibodies. Western blotting was performed using the Novex NuPAGE SDS/PAGE gel system (Invitrogen). Total cell lysates were run either on 3-8% Tris-acetate or 4-12% Bis-Tris precast gels, transferred to nitrocellulose membranes, and probed with antibodies specific to proteins of interest. Detection and quantification of blots were performed on Amersham hyperfilm ECL (Cytiva #28906839) and developed on film processor SRX-101A (Konica). Antibodies used for Western blots were obtained from commercial sources as follows: anti-GW182 (Bethyl #A302-

Immunoprecipitation (IP).
For IP of AGO-T6B complexes from human HCT116 cells, 500μg of lysates in 500 μL of SEC buffer were incubated for 3 hours with primary antibodies directed to either AGO proteins (WAKO anti-AGO2 #011-22033, EMD Millipore anti-panAGO #MABE56) or directed to T6B-fusion protein (Cell Signaling anti-FLAG #8146S, Cell Signaling anti-HA #2367S) or mouse IgG1 isotype control (Cell Signaling #5415). Next, lysates were incubated with 20µl of protein A/G PLUS-Agarose beads (Santa Cruz #2003) for 1 hour. For IP of AGO-T6B complexes from mouse tissues, 500μg of lysates in 500 μL of SEC buffer were incubated for 2 hours with GFP-trap magnetic agarose beads (Chromotek #gtma-10) or binding control beads (Chromotek #bmab-20). The immune complexes were run on SDS-PAGE and analyzed by Western blotting.
Vectors and molecular cloning. The targeting vector expressing the FH-T6B-YFP under control of TRE and CMV minimal promoter, was generated from a modified version of the pgk-ATG-frt plasmid (Addgene plasmid #20734), in which the region of pgk-ATG-frt comprised between the EcoRI site and the PciI site was substituted with the rabbit β-globin polyadenylation signal (RBG pA). The FH-T6B-YFP DNA insert was generated by PCR using the plasmid pIRES-Neo-FH-T6B-YFP 58 as a template. PCR was carried out using the following primers: Forward: 5'-GACTACAAGGACGACGATGACAAG-3' , Reverse: GTTACTTG-TACAGCTCGTCCATG. Next, the modified pgk-ATG-frt, was cut with NcoI, filled-in to produce blunt ends, dephosphorylated and ligated to the PCR-generated FH-T6B-YFP DNA fragment, according to standard subcloning procedures. Below, a scheme of the cloning strategy: To generated cell lines expressing either FH-T6B-YFP or FH-T6B Mut -YFP fusion proteins in a doxycycline-inducible manner, a modified version of the retroviral vector pSIN-TREtight-HA-UbiC-rtTA3-IRES-Hygro (hereafter TURN vector, a gift from Scott Lowe) was used to transduce commercially available HCT116 and MEFs cell lines. TURN is an all-in-one Tet-on vector that includes: 1) The rtTA3 gene under the human ubiquitin C promoter; 2) The transgene of interest driven by a tetracycline-responsive element (TRE)/CMV promoter. We used the pIRES-Neo-FH-T6B-YFP described in Hauptmann ae al. 58  Small RNA sequencing. Total RNA was extracted from T6B and induced as well uninduced T6B mut MEFs and 1 µg was used as input for sRNA-seq library preparation as described in ref. 94 . Briefly, 1 µg total RNA was ligated to nine distinct pre-adenylated 26-nt 3'-adapters with a 5-nt barcode using a mutated and truncated Rnl2 followed by urea gel purification and size selection and 5'-adapter ligation with Rnl1. This ligation reaction was again gel purified and size-selected for fully ligated product and reverse transcribed using SuperScript III RT followed by PCR amplification using Taq polymerase for 25 cycles. The final PCR product was separated on a 2% agarose gel in TBE buffer and extracted using the QIAgen gel extraction kit according to the manufacturer's instructions including all optional steps. After high-throughput sequencing, small RNA reads were aligned to a miRNA genome index built from 1,915 murine pre-miRNA sequences from miRbase version 21 95  Z-score calculation. For conserved miRNA families, the mean log2-fold change of predicted targets compared to the rest of the transcriptome (back-ground) was calculated. The means were converted to z-scores using an approach developed by Kim and Volsky 99 . Z-score = (Sm -m)3m1/2/ SD, where Sm is the mean of log2-fold changes of genes for a given gene set, m is the size of the gene set, and mand SD are the mean and the standard deviation of background log2-fold change values.  Measurements of these parameters were obtained using OMERO (https://www.openmicroscopy.org/omero/) and used to estimate the extent of damage and colitis induced by DSS treatment. Plots show that no significant differences between R26 CTL and R26 T6B mice were observed, suggesting that both groups experienced similar level of DSS-induced colitis.