Male and female mice of strain C57BL/6N were obtained from the National Institute on Aging (NIA) Aged Rodent Colonies (www.nia.nih.gov/research/dab/aged-rodent-colonies-handbook). Liver tissue from male Ames dwarf, a long-lived strain containing a mutation in the Prop1 gene resulting in growth hormone deficiency, and wild type control mice; and male calorie restricted and ad lib fed control C57BL/6 mice were obtained from the NIA Aged Rodent tissue bank (https://www.nia.nih.gov/research/dab/aged-rodent-tissue-bank-handbook). All procedures were approved by the Brown University IACUC and carried out in compliance with IACUC guidelines. This study was carried out in compliance with the ARRIVE guidelines.

MicroRNA mimics for mmu-mir-465b-5p were purchased from Exiqon (Woburn, MA, United States). Negative Control 5 miRCURY LNA miRNA Mimic (GAU​GCU​ACG​GUC​AAU​GUC​UAA​G) was used as a control (Exiqon, Woburn, MA, United States).

Primary hepatocytes were transfected with miRNA mimics using Fugene HD transfection reagent (Promega, Madison, WI, United States) at a 4:1 ratio of the transfection reagent to µg of nucleic acid following the manufacturer’s instructions. miRNA mimics were prepared for injection into mice by complexing 2 pmoles of mimic with Invivofectamine 3.0 (ThermoFisher, United States) following the manufacturer’s recommendations.

Mice were gently warmed under a heat lamp, placed in a restraining device, and the tail warmed by wrapping it in a paper towel wetted with warm tap water. The tail was dried and wiped with 70% ethanol. The tail vein was visualized by illuminating the underside of the tail using Veinlite R (Veinlite). Each mouse received a 200 µL injection of containing 2 pmoles of the miRNA mimic complexed with Invivofectamine 3.0.

RNA was isolated using a combination of Trizol reagent and the RNeasy kit (Qiagen, Valencia, CA, United States). 30–50 mg of tissue was homogenized in 1 mL Trizol and extracted with 200 µL of chloroform. Following centrifugation at 12,000 x g for 15 min the RNA in the aqueous phase was precipitated by adding 500 µL isopropanol. The precipitated RNA in solution was loaded onto the RNeasy column and purified according to the manufacturer’s instructions. The RNA was quantified using a NanoDrop spectrophotometer.

Total RNA was isolated as above and diluted to a concentration of 33.3 ng/μL. 3 μL of each sample was run on the Nanostring platform using the nCounter Mouse v1.5 miRNA panel according to manufacturer’s instructions. Data was normalized to the top 100 expressed miRNAs on the panel using the nCounter software (Nanostring).

The full-length transcripts encoding the miRNA cluster on the X chromosome were obtained using the SMARTer RACE 5’/3’ Kit following the instructions provided by the kit (Takara Bio United States Inc., San Jose, CA, United States). Briefly, total RNA was converted to cDNA using the kit reagents and instructions. The cDNA was used as a template for PCR reactions using gene specific primers (Table 1) and primers to the adaptors provided in the kit and SeqAmp polymerase (Takara Bio United States, Inc.). The PCR products were gel purified and either cloned into pRACE provided with the kit and sent for Sanger sequencing, or the purified PCR product was sent to the MGH CCIB DNA Core for complete amplicon sequencing. For the latter, reads were aligned to the mm10 genome using RNA star and assembled using Cufflinks.

For quantification of mRNA expression 1 μg of total RNA was transcribed into cDNA in 50 μL reactions using random hexamers and the TaqMan kit (Applied Biosystems, Foster City, CA, United States), according to the manufacturer’s protocol. A volume of 0.1–1.5 μL of this reaction was used in subsequent qPCR reactions. For quantification of miRNA expression cDNA was made following the procedure outlined in Balcells et al. (2011). A quantity of 1 µg of total RNA was polyadenylated at the 3′ end by incubating for 30 min at 37°C with 0.25 U/µL polyA polymerase (New England Biolabs, Ipswich, MA, United States), 1 mM ATP (New England Biolabs, Ipswich, MA, United States), 5.5 mM MgCl₂ in 1x reverse transcription (RT) buffer from the TaqMan kit (Applied Biosystems, Foster City, CA, United States) in a 25 µL volume. A degenerate RT-primer was added to a final concentration of 200 nM and allowed to anneal at 60°C for 5 min. Finally, the RNA was reverse transcribed to cDNA by incubating for 1 h at 42°C with 500 µM (each) dNTP mix (Applied Biosystems, Foster City, CA, United States), 1.25 U/µL MultiScribe reverse transcriptase (Applied Biosystems, Foster City, CA, United States), 5.5 mM MgCl₂ in 1x RT buffer in a final volume of 50 µL.

qPCR was performed using the SYBR Green system (Applied Biosystems, Foster City, CA, United States) on an ABI ViiA 7 Real Time System instrument (Applied Biosystems, Foster City, CA, United States), according to the manufacturer’s specifications. Primers for mRNAs were designed using the Primer-BLAST software. Primers for miRNA quantification were designed using miRprimer (Busk, 2014). Primer sequences can be found in Table 2 . All primers were tested on serial dilutions of target cDNA to ensure they amplified quantitatively, i.e., had a slope between −3.0 and −3.6 and an R-squared value above 0.95 (Dorak, 2006). Only primers meeting these criteria were used in this study. Due to the short length of the miRNAs, we were not able to identify suitable primers for all of the miRNAs present in the X-chromosomal cluster. The forward and reverse primers were used at a concentration of 300 nM each. GAPDH and HPRT were used as normalization controls for mRNA expression. The small nucleolar RNAs SNORD66 and SNORD70 were used as normalization controls for miRNA expression. Significance was determined using 2-tailed Student’s t-test.

Four-month-old male and female mice were injected with 2 nmoles of biotinylated mir-465b via the tail vein as described above. After 4 days the mice were sacrificed and the liver and blood were collected. 100 mg of frozen liver was homogenized on dry ice, thawed in PBS + 1.5% paraformaldehyde, and incubated for 20 min. The fixation was quenched with glycine for 5 min and the tissue washed with PBS + 60 U/mL of the RNase inhibitor Superase (ThermoFisher) + cOmplete protease inhibitor cocktail (Roche) added at the manufacturer’s recommended concentration and centrifuged for 5 min at 500 x g. The pellet was resuspended in lysis buffer (20 mM Tris + 100 mM KCl + 5 mM MgCl₂ + 0.3% v/v NP40) + 60 U/mL Superase + 1x protease inhibitor cocktail. The suspension was incubated on ice for 20 min, while pipetting up and down 20x every 5 min with a p1000 pipet. Debris was removed by centrifugation at 10,000 x g for 30 min. 35 µl of the supernatant was removed as input. The remaining supernatant was incubated with streptavidin coated magnetic beads at 4°C for 4 h with rotation. Beads were subsequently washed 5x with 1 mL lysis buffer then incubated for 20 min at 37°C with 100 µL DNase (Qiagen). Beads were washed once with 500 µL of lysis buffer, then incubated in 100 µL lysis buffer + proteinase K (New England Biolabs) at 55°C for 20 min. One hundred µl of lysis buffer + proteinase K was added to the input and incubated at 55°C for 20 min. 750 µl of Trizol LS was added to the beads and input, and RNA was isolated following the manufacturer’s recommendations, except the samples were incubated at −20°C overnight after isopropanol addition, and 2 µL of GlycoBlue (ThermoFisher) was added to each sample to visualize the RNA pellet. mir-465b enrichment was determined by qPCR as described above using primers specific for mir-465b.

Hepatocytes were isolated following the protocol outlined by Zhang (http://www.mouselivercells.com/Documents/Hepatocyte%20Isolation%20Protocol.pdf). Briefly, the liver was perfused with ∼70 mL HBSS + 0.5 mM EGTA (without magnesium and calcium) through the portal vein to remove the blood, followed by ∼70 mL digestion medium (DMEM-low glucose w/L-Gln and Na-pyruvate + 1x Penn-Strep + 15 mM HEPES containing 100 CDU/mL collagenase). Once digestion was complete, the liver was excised and placed in a 10 cm dish containing ∼ 20 mL of the digestion medium. The liver was torn apart to release the hepatocytes and cell clumps were disrupted by trituration with a 25 mL pipet. The cell suspension was filtered with a 70 µm filter (Fisher Scientific). Cells were collected by centrifugation at 50 × g for 2 min at 4°C and washed 3x in 25 mL isolation medium (DMEM-high glucose w/L-Gln and Na-pyruvate + 1 mM Na-lactate + 2 mM L-glutamine + 15 mM HEPES + 1x Penn/Strep + 0.1 µM Dexamethasone +10% FBS), collecting the cells by centrifugation at 50 × g for 2 min at 4°C between washes. The final pellet was resuspended in 25 mL cold isolation medium. Viable cells were counted on a hemocytometer and the percent viability was determined by staining 10 µL of the cell suspension with 10 µL 0.4% trypan blue. Cells were diluted to 3 × 10⁵ viable cells/mL and cells were distributed by plating 10 mL of cells in 10 cm collagen-coated dishes. Once the cells attached to the plates, the medium was changed to culture medium (DMEM-low glucose w/L-Gln and Na-pyruvate + 10 mM Na-lactate + 2 mM L-glutamine + 5 mM HEPES + 1x Penn/Strep + 0.01 Dexamethasone).

Heparin stabilized plasma was collected from the mice at the time of sacrifice. Liver protein extracts were prepared by homogenizing 5–10 mg of tissue in 60 μL/mg tissue of tissue extraction buffer containing 100 mM Tris (pH 7.4) + 150 mM NaCl + 1 mM EGTA + 1 mM EDTA + 1% v/v Triton X-100 + 0.5% sodium deoxycholate + 1 mM phenylmethylsulfonyl fluoride + 1X cOmplete mini protease inhibitor cocktail. Samples were homogenized using either the Fisher PowerGen 125 motorized homogenizer, or by passage through a 26 G needle. Homogenized extracts were incubated on ice for 20 min, then centrifuged at maximum speed for 10 min at 4°C. The resulting supernatant was diluted 1:20 with MilliQ water prior to assessment of concentration using the Qubit Protein Assay kit (Qiagen, Q33211).

Liver and plasma total IGF-1 protein abundance was quantified using the IGF-1 (Mouse/Rat) ELISA kit (ALPCO) as per manufacturer’s protocol. Liver extracts and plasma samples were diluted 1:1 and 1:100, respectively, in assay buffer. All samples were run in duplicate. Absorbance readings were normalized to standard curve generated from readings of standard solutions of known IGF-1 concentration. IGF-1 measurements for liver extracts were normalized to protein concentration as determined by the Qubit Protein Assay kit.

Statistical significance of the miRNA expression data was determined using the nSolver 4.0 software (Nanostring). Statistical significance of qPCR data was determined using a Student’s t-test for data with a normal distribution and a Mann-Whitney test for data without a normal distribution using either GraphPad Prism version 9.4.1 or Microsoft Excel 2019. Error bars in Figures 1A, B are standard deviation. The error bars on all remaining graphs are standard error of the mean (S.E.M.).

Our previous studies showed an age-associated increase in expression of a cluster of 18 miRNAs located in a ∼65 kb region between the Fmr1 and Slitrk2 genes on the X-chromosome in male mouse liver (Elias et al., 2019). The updated mm10 genome annotation reveals there are 19 miRNAs in this cluster. To determine if this upregulation occurs in both sexes, or is restricted to males, we performed Nanostring analysis on young (3 months) and aged (24 months) male and female mice. Analysis of the expression of approximately 600 miRNAs confirmed that the X-chromosomal cluster of miRNAs that are present on the Nanostring card were upregulated at 24 months of age in both male and female liver (Figures 1A, B). Eighteen of the 19 miRNAs are represented on the Nanostring card, with the newly annotated 19th miRNA being absent. Since mir-465b and mir-465c are duplicated in the cluster, and the duplicates contain the identical sequence, it is impossible to distinguish mir-465b-1 from mir-465b-2 and mir-465c-1 from mir-465c-2. These miRNAs are represented as mir-465b and mir-465c on the card, respectively. These results indicate that despite being expressed in a testes-specific manner in males during development (Zhang et al., 2019a), the age-associated upregulation of the miRNAs in this cluster was a generalized phenomenon that occurs in both males and females. The level of increase was similar in each sex, indicating that the inactivated X chromosome was likely not involved in the upregulation seen in females and that similar age-associated mechanisms occurred in both sexes. In addition, we did not see the same increase in mir-122, a liver-specific miRNA (Figures 1A, B), showing that the increase in expression seen for the miRNAs in the X-chromosomal cluster were specific for that cluster and not a generalized phenomenon.

To confirm that the upregulation of this cluster of miRNAs at 24 months resulted from physiological changes that occurred with age, we examined an intervention known to delay the development of aging phenotypes to determine if it attenuated or prevented the age-associated increase in expression of the X-chromosomal miRNA cluster. We examined expression of these miRNAs in liver from 24-month-old mice subjected to calorie restriction. We found reduced expression in liver from calorie restricted mice for the miRNAs located at the 5′ end of the cluster (based on the direction of transcription, as the miRNAs in this cluster are transcribed from the reverse strand) (Figure 1C). There was significantly lower expression of mir-465b and mir-465c in the calorie restricted animals. There was also reduced expression of mir-871 and mir-471, both located in the center of the cluster. Interestingly, calorie restriction led to an increase in mir-883a, suggesting that expression of this miRNA was induced by calorie restriction. There was no attenuation of expression levels of mir-470, mir-878, or mir-743a, the latter located at the 3′ end of the cluster. These results indicated that calorie restriction attenuated the rise in expression of a subset of miRNAs located towards the 5′ end of the X-chromosomal miRNA cluster, and that there may be differential regulation of expression of the miRNAs along the cluster, especially towards the 3’ end.

We investigated whether genetic backgrounds that are associated with an extended lifespan also show attenuation of the de-repression of the miRNA cluster with age. The Ames dwarf mouse is homozygous for a single mutation in the Prop-1 locus resulting in disruption of the somatotropic axis leading to an 50%–64% increase in lifespan (Sornson et al., 1996). When compared to 4-month-old animals on the same genetic background, wild type littermates had a robust increase in expression of the miRNA cluster at 24 months of age, whereas the Ames dwarf littermates showed attenuation of this increase (Figure 1D). These results indicated that lifespan extending genetic alterations may attenuate or delay the onset of upregulation of this cluster of miRNAs.

Many miRNA clusters are transcribed as a polycistronic transcript. To investigate whether this cluster of miRNAs is expressed as a single transcript, or if the miRNAs are expressed on multiple transcripts and therefore may be subject to different regulatory processes, we investigated transcripts originating from this region. We identified a long, non-coding RNA (lncRNA) that was transcribed from the negative strand that could function as a primary miRNA. This lncRNA encoded 12 miRNAs found in the X-chromosomal cluster, including the mir-465 family. This primary miRNA overlapped with a previously published primary miRNA sequence found in the Neuro2A cell line (Chang et al., 2015), however the transcript that was expressed in liver was longer. Using the Neuro2A primary miRNA sequence as a starting point, we performed 5′ and 3′ RACE to identify the 5′ and 3′ ends of the transcript in mouse liver. We found that the transcription start site of the primary miRNA is approximately 62 kb upstream of the transcription start site of the first miRNA in the cluster, mir-465a (Figure 2A). We used whole amplicon sequencing to determine the 3’ end of the transcript. We found 2 transcripts that are expressed from this region (Figure 2A, GenBank accession numbers OQ974188 and OQ974189). There was a shorter transcript that stopped upstream of the miRNA cluster, and a longer transcript that terminated just downstream of the mir-463 gene. The longer transcript encoded 12 of the 19 miRNAs in the cluster, all of which were located in introns. In total, the full length lncRNA sequence encoding the mir-465 family in mouse liver was approximately 124 kb in length containing 14 exons, and was spliced into a 3746 bp non-coding transcript. The shorter lncRNA sequence was around 44 kb in length containing 12 exons, including a larger alternate exon 12, that stopped upstream of the miRNAs. The shorter transcript was spliced into a 3673 bp non-coding transcript. In order to determine if one, both, or none of these transcripts were transcribed and spliced in mouse liver, we performed PCR to identify the spliced products of each transcript using primers designed to the unique regions of each spliced transcript (Table 1; Figure 2B). We verified that both the long primary miRNA transcript encoding the miRNAs and the second, smaller transcript were produced and spliced in aged mouse liver and not in young mouse liver (Figure 2C), consistent with the expression pattern of the miRNAs.

To determine if the reduction in GH signaling with age could be related to an increase in expression of the miRNA cluster, we validated that the mir-465 family targets the GHR in mouse liver hepatocytes. Previously, we established that all 3 members of the mir-465 family (mir-465a, mir-465b, and mir-465c) target the GHR 3′UTR equally in the immortalized AML12 mouse liver cell line, leading to a reduction in GHR at the mRNA and protein level (Elias et al., 2019). This resulted in an attenuation of JAK2/STAT5 signaling, and IGF-1 and IGFBP3 expression following GH stimulation (Elias et al., 2019). To verify that the mir-465 family targets the GHR leading to an attenuation of GH signaling in the liver we isolated primary hepatocytes from young mouse liver, since these cells are more physiologically relevant compared to the AML12 cell line used in our previous experiments, and transfected them with either a mir-465b mimic or a scrambled, control miRNA mimic. We investigated only 1 of the mir-465 family members since all 3 members act equally on the GHR mRNA (Elias et al., 2019). We found a significant 30% reduction in GHR mRNA expression in hepatocytes transfected with the mir-465 mimic when compared with non-transfected controls, whereas hepatocytes transfected with the control miRNA mimic did not show a significant reduction in GHR mRNA expression (Figure 3A). Forty-8 hours post-transfection the hepatocytes were exposed to 500 ng/mL of mouse growth hormone and followed over a 2-h period. Non-transfected hepatocytes and hepatocytes transfected with the control miRNA mimic had a significant increase in IGF-1 expression 1 h after GH stimulation. However, there was an attenuation of the increase in IGF-1 expression in response to GH stimulation in hepatocytes transfected with mir-465 (Figure 3B). These results showed that mir-465 family expression in primary hepatocytes led to a reduction in GHR expression and an attenuation of GH signaling.

Our in vitro data showed that the mir-465 family targets the GHR 3′UTR resulting in reduced GHR levels and an attenuation of GH signaling. To determine if the mir-465 family targets the GHR mRNA and leads to an attenuation of GH signaling in vivo, young male and female mice were injected with 2 nMole of a biotinylated mir-465b mimic, or a biotinylated scramble miRNA as a control, via the tail vein. Since we previously showed all 3 mir-465 family members target the GHR equally (Elias et al.), we only investigated one of the family members, mir-465b. At day 4 post injection, the liver was perfused through the portal vein with PBS to remove the blood and liver samples were taken for analysis. We found a significant increase in mir-465b in the liver in both male and female animals injected with the mir-465b mimic, indicating the miRNA was targeted to the liver (Figures 4A, B). There was no increase in mir-465b in animals injected with the scrambled miRNA control. A streptavidin pulldown of biotinylated miRNA mimics revealed an increase in GHR mRNA in the mir-465b mimic samples (Figure 4C), when compared to the scramble miRNA mimic control. These results suggested that mir-465b bound directly to the GHR mRNA in the mouse liver.

To verify that an increase in mir-465b levels results in decreased GH signaling, we examined several components of the GH signaling pathway. We found a significant reduction in GHR mRNA in both males and females (Figure 4D), indicating that mir-465b had similar activity in both sexes. GHR activation results in an increase in IGF-1 expression and in the associated proteins IGFBP3 and IGFALS. Therefore, a reduction in GHR expression should result in a reduction in expression of these factors as well. IGF-1 showed a 49% and 24% reduction in mRNA expression in males and females, respectively (Figure 4E). IGFBP3 showed a 27% and 34% reduction in mRNA levels in males and females, respectively (Figure 4F). We found that the IGFALS was also decreased in mRNA expression, by 17% and 26% in males and females, respectively (Figure 4G). These data indicated that an increase in levels of the mir-465 family in the liver in vivo led to a reduction in mRNA expression of the GHR, IGF-1, IGFBP3, and IGFALS, analogous to what is seen in vitro, and this pattern was similar in both males and females. The reduction in GHR and IGF-1 expression suggested a reduction in GH signaling, which has been shown to affect healthspan and lifespan.

A reduction in mRNA levels of components of the GHR signaling pathway should translate to a reduction in the protein levels. To confirm this, we investigated total liver and plasma IGF-1 protein levels in the same mice. The majority of IGF-1 is normally bound by one of the IGF binding proteins (IGFBP) in a complex with ALS, preventing its detection by antibodies found in most ELISA kits. To circumvent this problem, the kit used here used low pH to dissociate IGF-1 from the IGFBP/ALS complex, and buffers with excess IGF-2 to out compete IGF-1 from re-binding to the IGFBPs. This allowed us to assay total IGF-1 levels, not just free IGF-1 levels. We found a 45% reduction in IGF-1 protein levels in the liver and a 15% reduction in the plasma of male mice injected with the mir-465b mimic, compared to mice injected with the scrambled sequence control miRNA mimic (Figures 5A, B). Female mice showed a 42% reduction in IGF-1 protein levels in the liver following mir-465b injection, however while there was a trend towards reduced plasma levels in females, the reduction was not significant (Figures 5A, B). These results indicated that there may be differential regulation of IGF-1 protein levels in the blood between males and females. We did note that overall, females had relatively higher IGF-1 levels in both the liver and the blood, and levels in females showed greater variation between individuals than in males.