Prmt5 is a regulator of muscle stem cell expansion in adult mice

Skeletal muscle stem cells (MuSC), also called satellite cells, are indispensable for maintenance and regeneration of adult skeletal muscles. Yet, a comprehensive picture of the regulatory events controlling the fate of MuSC is missing. Here, we determine the proteome of MuSC to design a loss-of-function screen, and identify 120 genes important for MuSC function including the arginine methyltransferase Prmt5. MuSC-specific inactivation of Prmt5 in adult mice prevents expansion of MuSC, abolishes long-term MuSC maintenance and abrogates skeletal muscle regeneration. Interestingly, Prmt5 is dispensable for proliferation and differentiation of Pax7+ myogenic progenitor cells during mouse embryonic development, indicating significant differences between embryonic and adult myogenesis. Mechanistic studies reveal that Prmt5 controls proliferation of adult MuSC by direct epigenetic silencing of the cell cycle inhibitor p21. We reason that Prmt5 generates a poised state that keeps MuSC in a standby mode, thus allowing rapid MuSC amplification under disease conditions.

O rgan-specific adult stem cells enable continuous regeneration of various tissues throughout adult life. Most adult stem cells are assumed to undergo constant turnover to ascertain self-renewal and tissue homeostasis, although dormant adult stem cells have been described in the haematopoietic system, which remain in a quiescent state for most of their lifetime and divide only rarely under severe stress conditions 1,2 . Adult skeletal muscle stem cells (MuSC) are represented by a specialized subset of myofibre-associated cells called satellite cells and own a remarkable regenerative potential, which enables them to continuously replace myofibres by undergoing repeated rounds of activation and expansion under persisting disease conditions. Satellite cells originate during early embryonic development from a population of proliferating cells of the paraxial mesoderm. While most cells activate myogenic genes and form skeletal muscle fibres, some remain undifferentiated, adopt a satellite cell position in postnatal muscle and acquire a quiescent state [3][4][5] . After injury or excessive exercise, Pax7 þ cells exit quiescence, proliferate and differentiate to generate new myofibres or fuse with existing myofibres, thereby fully restoring damaged skeletal muscle tissues [6][7][8][9] . Several recent studies demonstrated that ablation of Pax7 þ MuSCs prevents muscle repair under pathological conditions and during ageing (review in ref. 10). However, the molecular mechanisms that control satellite cell functions during skeletal muscle regeneration are only partially understood, although several factors directing the fate of MuSC have been identified (reviewed in refs 2,11).
To identify new regulators of MuSC activation, self-renewal, expansion and differentiation, we establish a systematic screening approach taking advantage of mass spectrometry-based protein profiling of FACS (fluorescence-activated cell sorting)-sorted MuSC 6,12,13 combined with short hairpin RNA (shRNA)mediated knockdown of MuSC-specific genes. Several genes are identified that strongly affect activities of MuSC in vitro, of which the histone arginine methyltransferase Prmt5 is chosen for indepth functional analysis. We find that Prmt5 is essential for adult MuSC proliferation and muscle regeneration by restricting p21 expression via direct epigenetic silencing, thereby allowing rapid expansion of MuSC. Since the lack of Prmt5 does not affect embryonic myogenesis, we postulate that prenatal muscle development and adult muscle regeneration use distinct genetic and epigenetic mechanisms for the control of muscle progenitor cell expansion.

Identification of novel regulators controlling MuSC homeostasis.
To determine the proteome of MuSCs, we isolated GFP-labelled stem cells (SC GFP ) from skeletal muscles of Pax7 ICN /ZEG mice 6,14 via FACS ( Supplementary Fig. 1a), which all expressed Pax7 protein and readily differentiated into myocytes ( Supplementary  Fig. 1b,c). Protein extracts of freshly isolated MuSCs were subjected to mass spectrometry analysis (n ¼ 3) resulting in the identification of 135,341 peptides in all samples combined corresponding to 5,031 proteins in MuSC with at least one unique peptide (Supplementary Data 1). Notably, we detected numerous proteins known to be highly expressed in MuSCs including CD34, integrin a7, caveolin-1, Numb and b1-integrin, but not haematopoietic or endothelial cell markers such as CD45 or CD31 (refs 12,13,15). Comparison with proteome data sets obtained from myofibres, Pax7mononuclear cells and the MuSC PG fraction (satellite cells after percoll gradient and before FACS sorting) allowed us to identify 441 proteins that are exclusively present in MuSC but not in differentiated myofibres ( Fig. 1a and Supplementary Data 2). Mass spectrometry data of randomly selected proteins (Wdr61, ABCC4, Lxn, Mustn1, P2RX4 and Prmt5) were validated by immunofluorescence staining of freshly isolated myofibres (Fig. 1b).
To analyse the function of MuSC-specific proteins, we generated a custom-arrayed lentiviral shRNA library (400 genes, one shRNA per well, on average five different shRNAs per gene; Fig. 1c and Supplementary Data 2). FACS-purified Pax7 ZsGreen MuSC 12 were transduced with shRNA expressing lentiviruses and analysed by high-throughput fluorescent microscopy 96 h post transduction for the ratio of Pax7 þ versus total 4,6-diamidino-2-phenylindole (DAPI) þ cells (Fig. 1d), providing a read-out for genes affecting self-renewal, proliferation and differentiation of MuSC. shRNAs targeting Pax7 and Nf1 were included as quality controls (n ¼ 4 wells for each plate). After selecting for genes, which yielded a strong shift of the ratio of Pax7 þ versus total DAPI þ cells after knockdown, we ended up with 30 genes inducing and 90 genes decreasing Pax7/DAPI þ cell ratios after knockdown (Fig. 1e, Supplementary Fig. 1d and Supplementary Data 3).
Prmt5 prevents depletion of the MuSC pool in mdx mice. To further explore the role of Prmt5 in replenishing the MuSC pool, we utilized mdx mice, which lack functional dystrophin resulting in continuous degeneration/regeneration of myofibres, accompanied by repeated activation and enhanced turnover of satellite cells 20 . Treatment of 8-week-old Prmt5 sKO /mdx compound mutant mice for 3 weeks with TAM ( Fig. 3a) resulted in progressive loss of body weight, whereas Prmt5-deficient and mdx mice gained weight similar to wild-type littermates (Fig. 3b,c). Prmt5 sKO /mdx mice had a markedly lower body weight 4 months after initiation of the TAM treatment (Fig. 3b,c), and the diaphragm was markedly thinner compared with controls ( Fig. 3d). Magnetic resonance imaging (MRI) measurements revealed a massive decrease of the total muscle mass normalized to tibia length of Prmt5 sKO /mdx (87.5 ± 27.2 mm 3 per mm, n ¼ 4) but not of Prmt5 mutants (211.5±10.0 mm 3 per mm, n ¼ 3), wild-type mice (224.9 ± 32.3 mm 3 per mm, n ¼ 5) and mdx mice (332.0±30.8 mm 3 per mm, n ¼ 5), which gained muscle mass due to myofibre hypertrophy (Fig. 3e). Furthermore, Prmt5 sKO (4.67±2.52, n ¼ 3) and Prmt5 sKO /mdx mice (2.80±2.49, n ¼ 5) displayed a significant reduction of Pax7 þ MuSC at 6 months of age compared with controls (Ctrl 15.20±2.39, n ¼ 5; mdx 32.67 ± 18.72, n ¼ 3; Fig. 3f). We concluded that Prmt5 plays an essential role to maintain the MuSC pool and enables continuous muscle regeneration under chronic disease conditions. Prmt5 controls proliferation and differentiation of MuSC. To investigate the cellular mechanisms responsible for the loss of the satellite cell pool and impaired muscle regeneration in Prmt5 sKO mice, we first analysed FACS-purified MuSCs from Prmt5 sKO and control mice in vitro. Prmt5-deficient MuSCs showed a virtually complete arrest of cell proliferation as reflected by a marked reduction of the number of Pax7 þ , 5-ethynyl-2 0deoxyuridine (EdU)-incorporating cells (Ctrl 32.55 ± 5.94%; Prmt5 sKO 8.29±4.24%, each n ¼ 6), which is in line with the results from Prmt5 knockdown experiments (Fig. 4a). Conversely, lentiviral overexpression of human Prmt5 stimulated proliferation of MuSC indicated by increased EdU incorporation (Ctrl GFP 36.90 ± 2.52%, n ¼ 8; Prmt5 OE 43.43 ± 4.13%, n ¼ 6; Fig. 4b). Moreover, we found a virtually complete absence of the formation of myogenic colonies on single myofibres from FDB muscles of TAM-treated Prmt5 sKO mice (1.00±1.00, n ¼ 3) despite the presence of Pax7 þ satellite cells (Fig. 4c). Genetic labelling of MuSC using a Rosa26 nlacZ reporter in which removal of a stoplox cassette by Pax7 CreERT2 resulted in activation of nlacZ expression uncovered a marked reduction of lacZ-positive MuSC in regenerating muscle of Prmt5 sKO mice 3 days after CTX injection (Ctrl 925±104; Prmt5 sKO 212±6, each n ¼ 3; Fig. 4d), indicating that Prmt5 is required for MuSC proliferation during the early phase of injury-induced muscle regeneration. Additional lineage tracing of MuSC using a Rosa26 YFP reporter revealed that Prmt5-deficient MuSC cells activated MyoD expression both on isolated myofibres and in single-cell cultures despite the failure to proliferate, suggesting that activation and proliferation of MuSC are not necessarily linked (Fig. 5a,b). However, activated, nonproliferative Prmt5-mutant MuSC failed to differentiate properly.  Fig. 5c). In addition, isolated MuSC from Prmt5 sKO mice expressed lower levels of MyoG (Ctrl 72.58±5.30%; Prmt5 sKO 11.13±1.50%, each n ¼ 4) and did not form MF20 þ myotubes efficiently (Ctrl 257.50±33.61 mm 2 ; Prmt5 sKO 56.53±8.27 mm 2 , each n ¼ 3; Fig. 5d,e). To further investigate the role of Prmt5 for myogenic differentiation, we inactivated the Prmt5 gene in vitro by treatment with 4-hydroxy-TAM (4-OH) (Fig. 6a) after amplification of isolated Prmt5 sKO MuSC and induction of differentiation. Although we observed expression of the early differentiation marker MyoG in this experimental setting (Fig. 6b), differentiation of MuSC into MF20 þ myotubes was essentially abrogated (Fig. 6c). We concluded that Prmt5 plays an additional role at a late stage of myogenic differentiation, independent of its function in proliferation and regulation of MyoG expression. Intriguingly, we also detected an increase of apoptosis in isolated Prmt5 sKO MuSCs after induction of differentiation (Ctrl 0.27 ± 0.12%; Prmt5 sKO 9.00±1.75%, each n ¼ 3), but not under conditions stimulating proliferation (Ctrl 0.08 ± 0.07%; Prmt5 sKO 0.11±0.11%, each n ¼ 3; Fig. 6d), suggesting that either Prmt5 promotes cell survival during differentiation of MuSC or that lack of proliferation before differentiation favours apoptosis.
Prmt5 represses the cell cycle inhibitor p21 in MuSC. To gain a better mechanistic understanding of the action of Prmt5 in MuSC and to identify genes that might be directly regulated by Prmt5, we performed transcriptome analysis in 4-OH-treated MuSC from control and Prmt5 sKO mice by RNA-sequencing (RNA-seq; Supplementary Data 4). Gene ontology (GO)-term analysis revealed an up-or downregulation of B500 genes (false discovery rate o0.05) involved in cell cycle control, DNA metabolism and replication after inactivation of Prmt5 ( Supplementary Fig. 3a,b), which is consistent with the proliferation defects observed in Prmt5-deficient MuSC 17,21 . The upregulation of the cell cycle inhibitor p21 attracted our particular attention [22][23][24] . qRT-PCR analysis of freshly isolated FACS-sorted MuSC confirmed a strong upregulation of p21 expression in Prmt5-deficient MuSCs indicating a transcriptional inhibition of p21 by Prmt5 (Fig. 7a).
In addition, we detected a clear upregulation of p21 in 4-OHtreated MuSC from Prmt5 sKO mice ( Fig. 7b) together with downregulation of CyclinB1, a p21 target gene, while transcription of myogenic factors including Pax7, MyoD and Myf5 was not altered (Fig. 7b). To investigate a potential direct repression of the p21 gene by Prmt5, we performed chromatin immunoprecipitation (ChIP) assays concentrating on four wellcharacterized regulatory regions of the murine p21 gene: upstream enhancer like region (En), p53 binding site (p53BS), transcriptional start site (TSS) and downstream intronic CpG island (CpG; Fig. 7c; refs 25,26). In control MuSC, Prmt5 was highly enriched at the En and p53BS but not at the TSS and CpG sites, which was lost after treatment of Prmt5 sKO MuSC with 4-OH ( Fig. 7d). Loss of Prmt5 binding caused a significant reduction of H3R8me2s at the p53BS site in Prmt5-deficient MuSCs (Fig. 7e), a significant loss of nucleosome occupancy at the TSS site ( Fig. 7f) and increased H3K4 trimethylation at the CpG site (Fig. 7g), which is all consistent with suppression of p21 by Prmt5. Binding of Prmt5 to the p53BS prompted us to ask whether Prmt5 suppresses p21 expression by preventing recruitment of p53. Surprisingly, inactivation of Prmt5 prevented binding of p53 to the p53BS in the p21 locus, thereby suggesting p53-independent upregulation of p21 in Prmt5-deficient MuSCs (Fig. 7h). This conclusion was also supported by normal p53 mRNA and protein levels in Prmt5 mutant compared with control MuSCs, although we detected accumulation of Mdm4 splicing variants that were shown to stabilize p53 in Prmt5-deficient NPCs (Fig. 7i,j; ref. 19).
Prmt5 is dispensable for embryonic muscle development. It is widely assumed that embryonic and adult myogenesis are regulated by similar molecular cues, although a number of important differences are apparent 10 . Hence, we wanted to know whether Prmt5 does not only control MuSC and muscle regeneration but is also involved in the formation of skeletal muscles during development, in particular, since Prmt5 has been claimed to regulate expression of Myf5, MyoD, Myogenin and Mef2c during zebrafish myogenesis 27 . Therefore, we deleted the Prmt5 gene in the myogenic lineage using the constitutively active Pax7Cre knock-in mouse strain (Pax7 Cre ) 14 . qRT-PCR analysis of FACSsorted Pax7 ZsGreen myogenic cells from Pax7 Cre /Prmt5 loxP/loxP mutant embryos (hereafter referred to as Prmt5 mKO ) verified efficient inactivation of Prmt5 expression in embryonic muscle progenitor cells (data not shown). Analysis of control and Prmt5 mKO mutant embryos at E9.5, E12.5 and E16.5 revealed no obvious defects in skeletal muscle formation (Fig. 9a). The normal presence of Pax7 þ , MyoG þ and MF20 þ cells in embryonic forelimbs muscle of E12.5 and E14.5 Prmt5 mKO embryos (Fig. 9b,c) suggested that loss of Prmt5 in Pax7 þ myogenic progenitor cells neither affects their expansion nor differentiation during embryonic muscle development. Similarly, lack of Prmt5 had no effects on Pax7, MyoD and MyoG expression in forelimb and hindlimb muscles at E16.5 when Pax7 þ muscle progenitor cells play an essential role for fetal muscle growth 28 or on prenatal muscle growth until birth (Fig. 9d,e). Despite the absence of an apparent skeletal muscle phenotype, most Prmt5 mKO mutants died around birth, which we attributed to the activity of Pax7 Cre and consecutive loss of Prmt5 in the central nervous system (CNS) 19 .

Discussion
Our screen identified several novel potential regulators together with molecules that have already been documented to control the fate of MuSC. Prominent examples include Smad3 (ref. 29) and syndecan-4 (ref. 30). We also identified several epigenetic modifiers including Wdr91, the poly (ADP-ribose) polymerase Parp12 and Ash2l, a component of the Mll2 complex that mediates H3K4 methylation, which has been shown to form a complex with the transcription factor Pax7 to regulate Myf5 expression and satellite cell proliferation 31,32 . The histone arginine methyltransferase Prmt5 attracted our particular attention also because we identified several known interaction partners of Prmt5 in the screen including Myd88 (ref. 33) and Mapk13 (also known as p38delta), a component of the mitogenactivated protein (MAP) kinase pathway 34 , suggesting that Prmt5-dependent mechanisms play a preeminent role in the regulation of MuSC proliferation and differentiation. During adulthood MuSC mostly exist in a resting, quiescent state but must be able to expand rapidly in order to regenerate damaged muscle tissue. Relaxed control of quiescence might lead ARTICLE to over-proliferation, depletion of the stem cell pool and might favour tumour formation. Failure to respond appropriately to proliferative cues will impair self-renewal of MuSC and compromise regeneration. Prmt5 seems to be a decisive component of the regulatory network that maintains this intricate balance and keeps MuSC in a poised standby mode (Fig. 10). In contrast, embryonic myogenesis is characterized by the rapid expansion of myogenic progenitor cells, which need to form skeletal muscles in a relatively short time period alleviating the need to enter a quiescent, non-proliferative state. Hence, it makes sense that the role of Prmt5 in the regulation of cell proliferation differs significantly between embryonic and adult myogenesis, whereas the control of muscle lineage determination and differentiation seems to follow a similar pattern 10,35-37 . A major function of Prmt5 for conferring a reversible resting state to MuSC is apparently the restriction of p21 expression. Reduced expression of Prmt5 in MuSC will result in upregulation of p21, which increases the threshold for cell cycle re-entry (Fig. 10). Although we detected Prmt5 by immunofluorescence in virtually all MuSC, its level of activity and hence regulation of p21 might vary, thereby contributing to the heterogeneity of MuSC. MuSC with lower Prmt5 activity might constitute a reserve population that is only activated under severe stress conditions. Alternatively, differential regulation of Prmt5 activity in asymmetrically dividing MuSC might distinguish cells returning to quiescence from those that undergo rapid expansion. Careful quantitative evaluation of Prmt5 activity in single MuSC will solve these questions in the future.
We do not claim that the epigenetic repression of p21 is the only mechanism by which Prmt5 arrests MuSC proliferation, in particular, since RNA-seq analysis identified several additional cell cycle regulators that might also be regulated by Prmt5 either directly or indirectly. Inactivation of p21 in Prmt5-deficient MuSC failed to rescue muscle regeneration fully, although proliferation of MuSC could be partially restored, indicating different modes of action of Prmt5 independent of p21. In fact, additional functions of Prmt5 in the regulation of progenitor cell behaviour have been reported previously. In embryonic stem (ES) cells, Prmt5 promotes pluripotency by modulating the cytoplasmic LIF/Stat3 signalling pathway, indirectly suppressing genes ARTICLE that are associated with ES cell differentiation 17 . In NPCs, Prmt5 regulates alternative splicing of Mdm4 that in turn stabilizes p53, which causes upregulation of p21 and inhibition of cell cycle progression 19 . Superficially, the findings in NPC appear to partially recapitulate the situation in MuSC, but a more careful analysis reveals fundamental differences in the mode of action. Inactivation of Prmt5 in MuSC does not change the mRNA and protein level of p53, although alternative splicing of Mdm4 was altered. Furthermore, we found that binding of p53 to the p21 locus was lost after inactivation of Prmt5, indicating that activation of p21 in Prmt5-deficient MuSC does not depend on p53.
Although suppression of MuSC expansion by upregulation of p21 dominated the phenotype of Prmt5 sKO mice, lineage-tracing experiments revealed that Prmt5-deficient MuSC failed to express myogenin, indicating that Prmt5 mutant MuSCs are unable to differentiate and form myofibres in vivo. This conclusion was also supported by the failure of MuSC to form myotubes even when Prmt5 was deleted after initiation of MyoD expression, a phenomenon that was also observed in C2C12 myoblasts 38 . In addition, the timed inactivation of Prmt5 in differentiating MuSC suggests an additional role for terminal myogenic differentiation after expression of MyoG has commenced. Interestingly, induction of differentiation of Prmt5 mutant MuSC triggered apoptosis, which might be related to the differentiation block and contribute to the loss of MuSC during regeneration and ageing.
Our study revealed that inactivation of Prmt5 in MuSC of mdx mice resulted in a severe loss of muscle volume and recapitulated several symptoms of human Duchenne muscular dystrophy within 90 days. The findings emphasize the pivotal role of MuSC in maintaining muscle mass under disease conditions. A similar phenotype was described recently using mdx mice completely lacking telomerase activity (mdx/mTR 2G mice) 20 . However, in mdx/mTR 2G mice, a massive atrophy of the diaphragm was only visible after 60 weeks, indicating that the lack of Prmt5 had more severe consequences in MuSC function than loss of telomerase activity. We believe that Prmt5 sKO /mdx mice might serve as a valuable model to study effects of therapeutic interventions on dystrophin-deficient myofibres without the interference of MuSC constantly replenishing lost or damaged myofibres.
Remarkably, muscle mass remained rather stable in Prmt5 sKO mice under physiological conditions for at least 3 months despite a significant decline of the number of MuSC and the failure of MuSC to expand. This finding allows two conclusions: (i) under physiological conditions, MuSC contribute only to a minor degree to the maintenance of muscle mass; and (ii) a significant proportion of MuSC undergoes self-renewal during a 3-month period. However, the second conclusion has to be viewed with caution, since it is possible that the lack of Prmt5 induces cell death of MuSC without prior activation and induction of proliferation, although we did not find evidence for such a scenario in our experiments. In the future, it will be interesting to further exploit the Prmt5 sKO model (Fig. 10) to study the role of

Pax7
Myogenin    Table 3. For RT-qPCR assays, total RNA from satellite cells and muscles was isolated using Trizol reagent (Invitrogen) according to the manufacturer's protocol. An amount of 1 mg of purified RNA was subjected to reverse transcriptase reaction in the presence of 25 ng ml À 1 random primers and 2.5 mM dA/C/G/TTP with 10 U ml À 1 SuperScript II Reverse Transcriptase (Invitrogen). Primers used for RT-qPCR are shown in Supplementary Table 4.
Western blot analysis. For western blot assays in vitro cultured satellite cells were harvested, washed with ice-cold PBS and lysed in cell lysis buffer (20 mM Tris (pH 7.5), 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM b-glycerophosphate, 1 mM Na 3 VO 4 and 1 mg ml À 1 leupeptin). Whole-cell lysates (10 mg) were subjected to SDS-PAGE and western blotting using antibodies are shown in Supplementary Table 2. Protein expression was visualized using an enhanced chemiluminescence detection system (GE Healthcare, Little Chalfont, UK) and quantified using a ChemiDoc gel documentation system (Bio-Rad). Uncropped images of original western blots are presented in (Supplementary Fig. 4).  46 . Peptides were purified by stop and go extraction tips 47 , and were run on a LC system coupled to a LTQ-Orbitrap XL or a LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) equipped with a nanoelectrospray source (Proxeon). Full-survey scan spectra (m/z ¼ 300-1,650) were acquired in the Orbitrap with a resolution of R ¼ 60,000 after accumulation of 1,000,000 ions. Raw data were analysed using the MaxQuant software package 48 . Database searches were performed with the Mascot search engine against a murine FASTA database (IPI 3.54). A false discovery rate of 1% was used, and only peptides with a minimum of six amino-acid length with at least one unique peptide were included for data analysis.

RNA
High-throughput screen and hit validation. A customized shRNA library against selected genes was robotically re-arrayed from a murine genome-wide The RNAi Consortium (TRC) shRNA library 49   GenElute HP 96-Well Miniprep Kits (Sigma-Aldrich) and tested for integrity with a PvuII digest. High-throughput production of lentivirus was performed by Ca 3 (PO 4 ) 2 transfection of HEK293T cells with helper plasmids pMD2.G and psPAX2. Four-hundred fifty freshly isolated satellite cells per well were seeded into 384-well tissue culture plates freshly coated with Matrigel (Greiner). Twenty-four hours later, cells were transduced with lentiviral supernatants supplemented with 8 mg ml À 1 polybrene for 6 h. After media exchange, cells were incubated for 72 h. Each 384-well plate contained 56 controls including 12 individual GFP-producing lentiviruses to monitor transduction efficiency, four positive controls for Pax7 knockdown (shRNA Pax7) and four positive controls for Nf1 knockdown (shRNA Nf1). Cells were fixed in 4% PFA and whole-well images were acquired and analysed using an ImageXpress Micro automated high-throughput fluorescence microscope and MetaXpress software (Molecular Devices). Pax7/DAPI ratios were determined for each individual well. Only plates with Z 0 values 40.5 according to 1 À (2s pos þ 2s neg /m neg À m pos ) were used for further processing. Values as reference for Pax7 expression in percentage were then calculated according to z ¼ (x/m) ( À 1) (x, value of particular sample; m, mean of plko1 empty vector; n ¼ 4). Target genes qualified as hits if percentiles were higher or lower than 25% compared with control.
Magnetic resonance imaging. All MRI experiments were performed on a 7.0-T superconducting magnet (Bruker Biospin, Pharmascan, 70/16, 16 cm; Ettlingen, Germany) equipped with an actively shielded imaging gradient field of 300 mT m À 1 (ref. 50). The frequency for the 1 H isotope is 300.33 MHz. A 60-mm inner diameter linear-polarized 1 H volume resonator was used for RF pulse transmission and signal reception (Bruker Biospin). Localized images were acquired using a spin-echo sequence and corrections of slice angulation were performed, if necessary. RARE (Rapid Acquisition with Relaxation Enhancement) sequences (repetition time (TR) ¼ 2,500 ms, echo time (TE) ¼ 36.7 ms, slice thickness ¼ 1 mm) in axial and coronal orientation were used to determine exact positioning of the lower part of the mouse body. A coronal MSME (Multi-Slice-Multi-Echo)-spin-echo-sequence with an echo time TE ¼ 8.6 ms, repetition time TR ¼ 453 ms, a field of view FOV ¼ 7 Â 7 cm 2 , matrix size MTX ¼ 512 Â 256 and a slice thickness of 1 mm was recorded. Volumetric quantification of fat and muscle tissue from images was processed by software ImageJ. A list of anatomically defined landmarks was used to derive tissue-specific signal intensity thresholds and to define the region of interest for intensity sensitive region growing segmentation. The resulting tissue voxel volumes inside the region of interest were determined as cubic millimetres for each tissue class. Mice were measured under volatile isoflurane (1.5-2.0% in oxygen and air with a flow rate of 1.0 l min À 1 ) anaesthesia; the body temperature was maintained at 37°C by a thermostatically regulated water flow system during the entire imaging protocol.