MYC is a transcription factor that directly regulates 20%–30% of the genome, and indirectly influences many metabolic processes (Fernandez et al., 2003; Li et al., 2003; Patel et al., 2004; Dang et al., 2006). Deregulation of MYC is implicated in 60%–70% of all human cancers, including Burkitt’s lymphoma, breast cancer, osteosarcoma, and hepatocellular carcinoma (Greer et al., 2013). MYC has been proposed to act as a master regulator of metabolism, cell growth, and cell division (Miller et al., 2012). MYC thus appears to be a central point of metabolic regulation, integrating intrinsic growth factor signals with nutrient signals from the environment in order to determine whether the cell should grow, divide, differentiate, or undergo apoptosis (Grandori et al., 2000).

While upregulation of MYC has been extensively implicated in the context of cancer, downregulation of MYC is associated with increased health span and lifespan (Greer et al., 2013; Hofmann et al., 2015). Homozygous deletion of MYC is embryonic lethal; however, Myc heterozygous (Myc ^+/− ) mice show a 10% lifespan extension in males, and 20% in females (Hofmann et al., 2015). These findings are in agreement with other lifespan extending interventions which have shown that reduction of translation, energy production, oxidative phosphorylation, and ribosome biogenesis, all of which are under positive regulation by MYC, extend lifespan (Brown et al., 2008; Gems and Patridge, 2013; Johnson et al., 2013). Furthermore, upregulation of MYC results in increased generation of reactive oxygen species (ROS) and DNA damage, which are both associated with aging (Vafa et al., 2002; Hoeijmakers, 2009). Together, this large body of evidence indicates that MYC upregulation promotes cancer and aging, while downregulation promotes healthy aging and increased lifespan.

Myc ^+/− mice are 10%–20% smaller than their wild-type siblings, but show no changes in developmental timing or reproductive ability despite an approximately 50% reduction in MYC levels across all analyzed tissues (Hofmann et al., 2015). Myc ^+/− mice have increased health span, evidenced by significant amelioration of age-related phenotypes such as cardiac fibrosis, bone density loss, dysregulation of lipid metabolism and immunosenescence, and increased rotarod performance. They also display significantly higher metabolic rates and activity at both young and old ages (Hofmann et al., 2015). These observations indicate a strong impact of decreased MYC activity on age-regulated pathways.

Interestingly, liver gene expression patterns in Myc ^+/− mice do not overlap strongly with other life-extending interventions such as caloric restriction, resveratrol, and metformin (Hofmann et al., 2015; Ma and Gladyshev, 2017). However, several age-associated pathways are downregulated in Myc ^+/− mice, including insulin-like growth factor 1 (IGF1), protein kinase B (AKT), and mechanistic target of rapamycin (mTOR) signaling pathways. While these pathways are canonically upstream of MYC activity, we recently showed that MYC regulates mTOR activity by modulating glutamine uptake through direct transcriptional regulation of the amino acid transporters Slc1a5 and Slc7a5, suggesting that negative feedback loops are present within these systems (Zhao et al., 2019).

IGF1 is a downstream effector of the somatotropic axis which regulates organismal growth and development in response to environmental clues such as nutrient availability, sleep, daylight, and exercise through modulation of growth hormone levels (Kato et al., 2002). IGF1 modulates somatic growth and cellular proliferation through both endocrine and autocrine/paracrine effects, with most of the endocrine-functioning hormone produced in the liver in response to growth hormone stimulation (Sullivan et al., 2002). IGF1 produced in the liver is secreted into the serum where it is found in circulation in a complex with one of seven IGF binding proteins (IGFBP1-7) and the acid-label subunit (ALS) (Rosenfeld et al., 2000). IGF1 is also produced by other tissues in both a growth hormone dependent and independent manner, but this tissue-specific production does not contribute significantly to overall IGF1 serum levels, suggesting an alternative purpose for extrahepatic IGF1 production (Le Roith et al., 2001).

Reduced IGF1 signaling is associated with increased longevity in many animal models, including nematodes, Drosophila, and mice, and has been correlated with longer lifespan in humans (Kenyon et al., 1993; Clancy et al., 2001; Tatar et al., 2001; Junnila et al., 2013). Decreased IGF1 signaling is however also associated with several age-related diseases such as osteoporosis, cardiovascular disease, skeletal muscle wasting and atrophy, as well as neurological ailments such as dementia (Liu et al., 2008; Elis et al., 2011). Many of these aging-related diseases can be alleviated through administration of either growth hormone or IGF1, suggesting a causal link between IGF1 decrease and development of these diseases (Yakar and Isaksson, 2015). This seeming contradiction between decreased IGF1 being simultaneously associated with increased lifespan and increased risk of age-related diseases has not been resolved but suggests that optimal health and lifespan rely on tight regulation of IGF1.

Myc heterozygous mice have decreased serum IGF1 levels, are long-lived, and are resistant to the development of osteoporosis, thus presenting a unique model system to address whether reducing IGF1 signaling can increase lifespan without deleterious effects on health span. We provide evidence that this effect is caused by the upregulation of specific miRNAs, which are normally repressed by Myc, and that increased levels of these miRNAs reduce the translation of the IGF-1 mRNA.

In circulation, ∼95% of IGF1 is bound by one of seven IGF binding proteins (IGFBPs) and the acid-labile subunit (ALS) (Rosenfeld et al., 2000). This tight binding of IGF1 to IGFBPs impedes detection by antibodies used in common ELISA assays, thus additional steps are required to assess total IGF1 levels. In order to assess whether total IGF1 levels were decreased in Myc ^+/− mice, we prepared sample dilutions in a buffer containing an excess of IGF2, which is not expressed at significant levels in adulthood, but has equal affinity for IGFBPs. This allows for IGF2 to outcompete IGF1 in binding to the present IGFBPs, thus freeing IGF1 to allow detection by anti-IGF1 antibodies. Using this approach, we determined that consistent with previously published results (Hofmann et al., 2015), total plasma IGF1 levels were decreased in young and old female Myc ^+/− mice by ∼20%–30% (Figures 1A, B).

Interestingly, we observed no difference in IGF1 plasma levels in male Myc ^+/− mice, suggesting a sexual dimorphism in the effect of MYC on IGF1. We next analyzed Igf1 transcript and protein levels in liver, which is the main site of synthesis of circulating IGF1 (Sjogren et al., 1999). Consistent with the plasma data, we saw an ∼20% decrease in IGF1 protein levels in young and old female Myc ^+/− mouse liver tissue compared to Myc ^+/+ mice, and no significant change in male Myc ^+/− mouse liver (Figures 1C, D). However, Igf1 transcript levels were unchanged in either sex in both young and old mice (Figures 1E, F). In line with this, IGF1 protein corresponded to Myc transcript abundance in female, but not male, mice in both Myc ^+/− and Myc ^+/+ genotypes (Figures 1G, H). While we find that Myc transcript levels exhibit significant overlap between the two genotypes, particularly in males, our previously published data shows that Myc ^+/− mice exhibit an approximately 50% decrease in MYC protein levels in liver of male and female mice (Hofmann et al., 2015). Taken together, these results suggest that MYC positively regulates IGF1 protein levels post-transcriptionally in a sex-specific manner, and that decreased MYC expression in female Myc ^+/− mice results in lower IGF1 protein levels in the liver, which results in decreased levels of circulating IGF1.

To assess whether IGF1 translation is regulated in female Myc ^+/− mice, we isolated polysome-bound mRNA from liver extracts using sucrose density centrifugation. We found that while related transcripts such as Igfbp3 and Igf1r showed no changes in polysome association, the Igf1 transcript was reduced by ∼20% in the polysome-associated mRNA fraction in the Myc ^+/− mice (Figure 1I). This result suggests that MYC regulates IGF1 by inhibiting its translation. MYC is a known regulator of genes involved in translation, and Myc ^+/− mice do show a slight decrease in overall rates of translation (Hofmann et al., 2015). However, the lack of significant translational repression on transcripts related to and regulated by the same pathways as Igf1, such as its binding proteins and receptor, suggests that the regulation of Igf1 translation by MYC is a targeted effect, rather than a global one.

To determine whether hepatocytes from female Myc ^+/− mice respond to growth hormone stimulation as efficiently as Myc ^+/+ hepatocytes, mouse primary hepatocytes (MHC) were isolated using a two-step perfusion protocol and allowed to adhere for 24 h in cell culture. Hepatocytes from both genotypes were then treated with 100 nM of mouse recombinant growth hormone for 2 h prior to harvest and extraction of mRNA and protein. We found that Igf1 transcript levels were upregulated to a similar extent in Myc ^+/− and Myc ^+/+ hepatocytes (Figure 1J). However, while Myc ^+/+ hepatocytes show a 30% increase in IGF1 protein in response to growth hormone stimulation, treatment of Myc ^+/− hepatocytes with growth hormone did not result in significant IGF1 upregulation (Figure 1K). Together, these data suggest that low MYC levels impede the translation of the Igf1 transcript.

Given the known role of MYC in regulating the expression of miRNA genes, as well as the ∼6 kb length of the Igf1 3′UTR, we next assessed whether MYC-regulated miRNAs could target the Igf1 transcript. TargetScan prediction identified 117 conserved miRNA binding sites in the Igf1 3′UTR, including those of 11 of the most expressed miRNAs (Lewis et al., 2005). 56 of these miRNAs have been previously shown to be repressed by MYC and are thus predicted to be downregulated in Myc ^+/− mice (Chang et al., 2008). To assess whether miRNAs known to be repressed by increased MYC levels and predicted to target the Igf1 transcript are indeed upregulated in Myc ^+/− mice, we analyzed their expression in total RNA extracts from livers of 3-month-old Myc ^+/− and Myc ^+/+ mice using the Nanostring platform.

Of the 56 candidate miRNAs identified above, 45 were found to have detectable expression in liver tissue of 3-month-old mice (greater than five counts in at least one sample). We found 10 candidate miRNAs to be significantly (FDR-adjusted p-value <0.05) upregulated in female, but not male Myc ^+/− mice (Figures 2A–C). Interestingly, no downregulated miRNAs were found for either sex. These results are consistent with the literature on miRNA regulation by MYC, which documents that the majority of miRNA genes are downregulated in the context of MYC overexpression (Chang et al., 2008).

In order to validate that upregulated candidate miRNAs in Myc ^+/− female mice lead to increased targeting of the Igf1 transcript in vivo, we performed AGO CLIP-Seq on liver tissue of both genotypes and sexes (Moore et al., 2014). Briefly, flash frozen liver was pulverized and cross-linked using UV-light prior to protein extraction in the presence of RNase inhibitors. Argonaute immunoprecipitation was then carried out, and Argonaute:miRNA:mRNA complexes were isolated by gel electrophoresis using a radiolabeled 3′ linker for visualization. A 5′ linker containing a degenerate sequence was ligated to the isolated RNA tags to allow for PCR amplification and subsequent sequencing and filtering of duplicate reads. After filtering for duplicate and low-quality reads, sequences were aligned to the mm10 genome, and MACS was used to call peaks (Zhang et al., 2008). miRNA seed sequences were identified within 50 bp of peak centers using Targetscan (Lewis et al., 2005).

We found that in female Myc ^+/− mice, AGO binding of the Igf1 transcript was significantly enriched at the target sites of miR-122, let-7, and to a more modest degree miR-29 (Figures 2D, E). These three miRNA families are among the most highly expressed miRNAs in the liver, and all three were found to be significantly upregulated in female Myc ^+/− mice in our Nanostring analysis. Given these results, we validated our Nanostring results of let-7i and miR-122 expression by RT-qPCR modified for miRNA detection (Balcells et al., 2011; Busk, 2014). We found that both let-7i and miR-122 are upregulated by ∼30% in liver of female, but not male Myc ^+/− mice (Figures 2F, G), consistent with our previous analysis.

miR-122 is a liver-specific miRNA that compromises ∼70% of the total miRNA species in mouse liver (Jopling, 2012). Furthermore, as decreased miR-122 expression has been linked to hepatocellular carcinoma, its effects have been investigated in the liver, and an AGO CLIP-Seq dataset is available from a miR-122 liver-specific knockout mouse model (Luna et al., 2017). We analyzed this available dataset and found that miR-122 knockout significantly reduces AGO binding at the predicted miR-122 target site in the Igf1 transcript, thus further validating this site as a bona fide miR-122 target (Figure 2H). Given that our AGO-CLIP-Seq data identified the target sites for let-7 and miR-122 as the most enriched for AGO binding in female Myc ^+/− mouse liver, we chose these two miRNAs for further analysis.

MYC has previously been implicated as a positive regulator of the enhancer of zeste homologue 2 (EZH2) (Ito et al., 2018). EZH2 is a methyltransferase that, as part of the Polycomb Repressive Complex 2 (PRC2), di-/tri-methylates histone 3 lysine 27 (H3K27) to promote the heterochromatization of target regions (Bracken and Heln, 2009). We thus analyzed available Encode ChIP-Seq datasets from the human HepG2 hepatocyte cell line and found that the promoter regions of Igf1-targeting miRNAs (including members of the let-7 family and miR-122) are enriched for EZH2 and H3K27me3, and are frequently found in CpG-rich chromatin regions (Figure 2I). These results suggest that many of the miRNAs that target Igf1 might be regulated by polycomb group repression. Quantification of EZH2 protein levels in aged females showed decreased EZH2 expression in liver, but not muscle in Myc ^+/− versus Myc ^+/+ mice (Figure 2J). Although these results point to a possible involvement of EZH2 in the regulation of miRNA genes by MYC, more work remains to be done to confirm this hypothesis.

To further elucidate the effects of upregulating miR-122 and let-7 in the liver, we transfected LNA-modified miRNA mimics of these miRNAs, both alone and in combination, into primary hepatocytes isolated from wild type C57Bl/6 mice. Transfection resulted in a significant intracellular increase in both let-7i and miR-122 as measured by RT-qPCR. Specifically, transfection with 20 nM miRNA mimics resulted in a 30 to 40-fold increase in let-7i (Figure 3A). Transfection with miR-122 increased its expression to 1.7-fold over scrambled control, though due to the very high expression of miR-122 in hepatocytes, this increase translates to a significant upregulation of the miR-122 miRNA (Figure 3B). Transfection with let-7i or miR-122 decreased IGF1 protein levels by ∼20% but did not achieve significance, while transfection with both let-7i and miR-122 in combination significantly decreased IGF1 protein levels by ∼40% relative to scrambled control (Figure 3C). Transfection with let-7i, miR-122, or let-7i/miR-122 combined did not affect Igf1 transcript levels as assessed by RT-qPCR in primary hepatocytes (Figure 3D). These results are consistent with our in vivo data in Myc ^+/− vs. Myc ^+/+ female mice which showed a decrease in IGF1 protein, but not transcript, levels (Figures 1A–C). Together, these results show that upregulation of let-7i and miR-122 can decrease IGF1 protein levels while not significantly affecting Igf1 mRNA levels.

To assess whether upregulation of let-7 or miR-122 in vivo can mediate translational repression of Igf1 we injected wild-type C57Bl/6 mice with 1 nmol of each miRNA either alone or in combination by tail-vein injection using invivofectamine as a carrier. Tissues were harvested at 4 days post-injection, with retrograde perfusion of the liver prior to harvest. Injection of 1 nmol of let-7i increased its expression 12-fold relative to scrambled control, while injection of the same amount of miR-122 increased its expression 1.2-fold (Figures 3E, F). In line with our data in primary hepatocytes, upregulation of let-7i alone resulted in a slight decrease in IGF1 protein levels which did not achieve statistical significance. Compared to scrambled control, injection of miR-122 either alone or in conjunction with let-7i significantly reduced IGF1 protein levels by almost 50% in the liver, without significantly affecting Igf1 transcript levels (Figures 3G, H). Furthermore, treatment of mice with miR-122 or combined let-7i/miR-122 significantly decreased plasma levels of IGF1 (Figure 3I). In turn, growth hormone levels were increased in the pituitary, in line with the known negative-feedback loop between these two hormones (Figure 3J).

While decreased IGF1 levels have been associated with longevity in numerous model organisms, as well as in humans, decreased IGF1 expression with age has also been correlated with increased risk of osteoporosis, muscle-wasting, and dementia (Obermayr et al., 2005; Perrini et al., 2010; Westwood et al., 2014). Several mouse models with decreased growth hormone (and consequently IGF1) levels have been shown to have decreased bone mineral density, as well as increased trabecular spacing in old age (Palmer et al., 2009; Yakar Isaksson, 2015). Interestingly, mice with a liver-specific Igf1 deletion that was essentially complete by 10 days of age did not show decreased femoral or body length (Sjogren et al., 1999; Yakar et al., 1999), suggesting that local, rather than endocrine, levels of IGF1 may be more important for the preservation of bone health into old age. In contrast to this data, Igf1 liver-specific deletion at 1 year of age reduced circulating IGF1 levels by 70% and significantly reduced trabecular number, but not bone density (Gong et al., 2014). Together, these results suggest that the decrease of circulating IGF1 levels with age, rather than decreased lifetime levels of IGF1, may be responsible for age-related bone loss while low levels of circulating IGF1 from birth do not significantly impair skeletal health.

Female Myc ^+/− mice, with decreased IGF1 at all ages, show a remarkable resistance to age-related bone loss, and in fact show no decrease in bone density, trabecular number, or trabecular spacing at 22 months of age compared to Myc ^+/+ mice (Hofmann et al., 2015). Though osteoporosis is a disease which typically affects females, we nevertheless extended our analysis to include a measurement of these parameters in male Myc ^+/− mice. While micro-CT analysis of the L4 vertebrae in wild-type male mice showed small trends for a decrease in bone density, increase in trabecular spacing, and decrease in trabecular number, none reached significance (Figures 4A–D). In contrast, these changes were exacerbated male Myc ^+/− mice and reached significance in two out of three parameters. Thus, while reduced levels of MYC positively impact skeletal health with age in females, the opposite was true for males.

To assess whether local levels of IGF1 in bone tissue were affected by Myc heterozygosity, we extracted protein from femurs of male and female Myc ^+/+ and Myc ^+/− mice at 3 and 24 months of age. Interestingly, in both sexes, Myc ^+/− mice IGF1 levels in the bone tended to be higher, as opposed to the decreases we found in the liver (above), although these trends were not significant (Figures 4E, F). The lack of decreased IGF1 expression in Myc ^+/− mouse bone tissue suggests that MYC regulation of IGF1 is liver-specific. However, we found a significant decrease of IGF1 protein levels in males (Figure 4F), which might explain the decrease in bone health.

Mice with liver-specific ablation of Igf1, resulting in greater than 70% reduction in circulating IGF1, show significantly increased levels of circulating growth hormone (Gong et al., 2014). Although the details of this negative feedback mechanism remain unclear, it has been reported that both growth hormone releasing hormone (GHRH) and growth hormone inhibiting hormone (GHIH) are regulated by IGF1 levels (Obal et al., 1991; Romero et al., 2012). We thus assessed the expression of these two factors in the hypothalamus at the mRNA level, and found a modest but significant increase in Ghrh in the hypothalamus of female, but not male, Myc ^+/− mice (Figure 4G), whereas Ghih levels were unaffected (Fig. 7H). Consistent with this, transcript and as well as protein levels of growth hormone in the pituitary were significantly increased in female but not male Myc ^+/− mice (Figures 4I, J).

Circulating growth hormone levels are difficult to assess in vivo due to their circadian as well as feeding-dependent fluctuations, the pulsatile, time of day-dependent fluctuations, which can range by as much as 100-fold in a matter of hours. However, through repeated measurement at consistent timepoints, it is possible to determine an average baseline level for each animal (Steyn et al., 2011). We thus collected blood via the saphenous vein at 10 a.m. every other day for a total of 3 time points from each animal and GH concentrations in each sample were measured by ELISA. We found that average growth hormone levels were significantly increased in female, but not male, Myc ^+/− mice (Figure 4K). These results indicate that decreased liver production of IGF1 in female Myc ^+/− mice, and thus reduced circulating IGF1, through a negative feedback loop increase GHRH levels in the hypothalamus, and consequently increase growth hormone production and secretion by the pituitary.

Myc ^+/− mice show global upregulation of miRNA expression, in line with other reports showing that MYC represses most of its target miRNA genes. Many of these miRNAs have been implicated in various human diseases, including Alzheimer’s, osteoporosis, muscle wasting, and hepatocellular carcinoma. As upregulation of MYC has been associated with more than 50% of all tumors, as well as other aging-related diseases, the modulation of miRNA expression by MYC is likely a significant mechanism by which MYC dysregulation leads to impaired human health (Levens, 2010). Interestingly, caloric restriction has also been shown to globally increase miRNA expression, suggesting that miRNAs may play a significant role in the regulation of lifespan (Zhang et al., 2019). Here, we have shown that downregulation of MYC in female mice induces the expression of multiple miRNAs that target the Igf1 transcript, and that at least one of these, miR-122, significantly reduces IGF1 translation upon ectopic upregulation in vivo.

While the mechanisms by which MYC affects transcriptional activation are well characterized, those mediating its role in transcriptional repression are more poorly understood. Several hypotheses have been proposed and supported by experimental results, including that MYC can upregulate the expression of some transcriptional repressors (Philipp et al., 1994; Lee et al., 1996), that MYC is recruited to the promoters of its repression targets through protein-protein interactions with transcriptional regulators such as TFII-I, YY-I, Sp-I, and MIZ-1 (Schneider et al., 1997; Seoane et al., 2001), and that MYC can regulate the expression of chromatin silencing factors such as the Polycomb Repressive Complex (PRC) (Knoepfler et al., 2006).

MYC has been found to regulate both the transcription of Ezh2 and EZH2 protein activity through phosphorylation (Bhandari et al., 2011; Neri et al., 2012), but the genes that are regulated through this PcG mechanism are not fully characterized. In line with MYC’s role as a chromatin regulator, genes repressed by MYC show considerable overlap with PcG and HDAC repressed genes (Kaur and Cole, 2013; Bhadury et al., 2014). Furthermore, knockdown of Ezh2 in glioma cells was found to upregulate 85 miRNA genes, many of which are also known to be repressed by MYC (Wang et al., 2013). In fact, many of the miRNAs upregulated in Myc ^+/− mice have already been documented to be regulated by EZH2 (So et al., 2011; Liu et al., 2013; Vella et al., 2015).

We found that, consistent with literature showing that MYC overexpression upregulates EZH2, female Myc ^+/− mice showed reduced liver EZH2 protein levels compared to Myc ^+/+ mice. Analysis of available Encode ChIP-Seq datasets from the human HepG2 hepatocyte cell line showed that the promoter regions of Igf1-targeting miRNAs, including most members of the let-7 family and miR-122, are enriched for EZH2 and H3K27me3. These results suggest that many of the miRNAs that target Igf1 may be regulated by PcG repression, which is regulated by MYC.

While EZH2-mediated repression is a plausible mechanism for MYC-mediated repression of miRNA genes, it was unclear why these miRNA genes were upregulated specifically in female Myc ^+/− mice. Many miRNA genes have been shown to be regulated in a sexually dimorphic manner. For example, in rat liver the IGF1-targeting miR-193a, miR-29b, and miR-122 miRNAs were found to be expressed at higher levels in females relative to males (Cheung et al., 2009). Another group found that 37% of miRNA genes were differentially expressed when MCF-7 human breast cancer cells were exposed to estrogen (Hah et al., 2011). These results were corroborated by another group which showed that estrogen induces the expression of 21 miRNAs while down-regulating 7 in MCF-7 cells (Bhat-Nakshatri et al., 2009). The miRNAs upregulated by estrogen in these cell types (the let-7, miR-30, miR-23 families), also overlap extensively with the miRNAs that target the Igf1 transcript, as well as those upregulated in Myc ^+/− females. Interestingly, it has been shown that EZH2 can be recruited to target gene promoters through estrogen receptor alpha, and that testosterone administration to female mice, which downregulates ER-alpha expression, results in the upregulation of several miRNAs, including members of the let-7 family, miR-122, and miR-30d (Delic et al., 2010; Ariazi et al., 2017).

Estrogen also plays a significant role in the regulation of MYC targets, and miR-122 expression has been previously shown to be upregulated in females compared to males (Cheung et al., 2009). miR-122 is the predominant miRNA expressed in liver tissue, compromising more than half of the total miRNA pool in the liver, according to our analyses and those of others (Jopling, 2012). miR-122 is regulated by multiple liver-enriched transcription factors, including HNF3b (FOXA2), HNF4a, and HNF6 (Xu et al., 2010). The miR-122 regulating liver-enriched transcription factor FOXA1/2 alternatively regulates MYC depending on the presence of estrogen or androgen (Li et al., 2012). This sex-hormone specific regulation is believed to be an important contributor to the sexual dimorphism of hepatocellular carcinoma, which has a prevalence in males 2–4 times higher than in females (Hsu et al., 2012; Li et al., 2012). In fact, downregulation of several oncogenic genes by FOXA1/2 is dependent on the presence of estrogen (Li et al., 2012). While estrogen-dependent regulation of miR-122 by FOXA1/2 has not been explicitly described, it is a plausible explanation for the sex-specific upregulation of miR-122 in Myc ^+/− female mice.

Of the many miRNAs that are predicted to target the Igf1 transcript, our results showed that miR-122 had the most profound effects on IGF1 translation in Myc ^+/− female mice. Given that miR-122 is a liver-specific miRNA, this suggests that MYC regulates IGF1 in a liver-specific manner. In line with this, we found that female Myc ^+/− mice did not have decreased IGF1 levels in bone tissue, in contrast to the significant reduction seen in the liver. In fact, IGF1 levels in femur tissue of Myc ^+/− mice trended towards increased expression both at young and old age.

IGF1 and its main regulator, growth hormone, function in a negative feedback loop such that reduced IGF1 levels in the circulation trigger an increase in growth hormone levels in the pituitary, and its secretion into circulation (Romero et al., 2012). We found increased GH expression and circulating levels in Myc ^+/− female mice, which is likely explained by the known negative-feedback loop between circulating IGF1 and GH. Increased growth hormone in circulation of Myc ^+/− female mice may then signal to peripheral tissues such as the bone to stimulate local production of IGF1, thereby mitigating bone loss with age, and possibly explaining the remarkable resistance of Myc ^+/− female mice to age-related bone loss.

Together, these results suggest that liver-specific reduction in IGF1 does not affect local production of IGF1 in peripheral tissues, and that in fact it may enhance it. Several mouse models with liver-specific disruption of IGF1 have been generated. Constitutive liver-specific Igf1 knockout resulted in a >60% reduction in circulating IGF1 levels and had minimal effects on body weight, organ weight, and femoral length (Yakar et al., 1999). Liver-specific ablation of IGF1 at 3 weeks of age resulted in a ∼75% decrease in circulating IGF1 levels, extended lifespan by 16% in females, but resulted in only minor reduction in femoral length (Sjogren et al., 1999; Svensson et al., 2011). Unfortunately, local IGF1 levels were not measured in bone tissue of either mouse model, nor was trabecular morphology assessed with age. However, the results of these studies are consistent with our data, and suggest that liver-specific disruption of IGF1 at young age does not significantly affect bone health.

In aggregate, we propose a model in which MYC-mediated activation of EZH2 causes the upregulation of multiple miRNAs (namely, miR-29, let-7, and miR-122), in an estrogen-dependent manner, which then target the Igf1 transcript and reduce its translation. The regulation of Igf1 translation by miR-122, which is a liver-specific miRNA that showed the most pronounced effect on Igf1 translation, results in the reduction of IGF1 protein levels in the liver of female Myc ^+/− mice. Since circulating IGF1 is produced predominantly in the liver, the systemic effect of decreased MYC expression is reduced plasma levels of IGF1. Given that IGF1 and GH function in a negative feedback loop, such that reduced circulating IGF1 triggers an increased production and secretion of GH by the pituitary, decreased MYC expression concomitantly increases plasma levels of GH. Increased GH expression, in turn, increases the local production of IGF1 in tissues such as the bone, thereby evading some of the negative consequences of global IGF1 reduction.