3647RESEARCH ARTICLE

INTRODUCTIONHeritable epigenetic changes in gene expression have now beenobserved in a recurrent manner in the mouse. Formally comparable tothe plant paramutations (Brink, 1956), they were efficiently inducedin early mouse embryos by RNAs with sequence homology to thetarget locus, either transcript fragments or microRNAs. Carried by thesperm and ovocytes of modified parents, these RNAs act as signals ofthe transgenerational transfer of modified phenotypes. Paramutationof the Kit locus, first observed in the progeny of heterozygotescarrying a disrupted allele, was induced by microinjection in one-cellembryos of the cognate microRNAs miR-221 and -222(Rassoulzadegan et al., 2006). The notion of RNA-mediatedepigenetic heredity was then extended to a pathological hypertrophyof the heart owing to an elevated expression of the Cdk9 transcriptionfactor and was supported by microinjection of transcript fragmentsand of the cognate microRNA miR-1 into fertilized eggs (Wagner etal., 2008). In both cases, the long-term effect initiated by either themicroRNA or the transcript sequences was identified as an increasedtranscriptional activity of the target gene, distinct from the knownpost-transcriptional regulations exerted by microRNAs. Takentogether, these observations led to the concept of a surveillancemechanism, which, in the early embryo, detects abnormal profiles oftranscripts and initiates a hereditary program enhancing expression ofthe normal allele at the epigenetic level.

We extended this approach to a microRNA with a distinct organspecificity, miR-124, expressed in the brain and important in thedevelopment of the central nervous system (Cao et al., 2007; Lagos-Quintana et al., 2002; Makeyev et al., 2007; Visvanathan et al.,2007). Unexpectedly, every pup born after microinjection (referredto herein as miR-124*) showed an unusually large body size, a‘giant’ phenotype maintained to adulthood and subsequentlyinherited over several generations. The accelerated growth rate wasin fact established at the most early developmental stages (morulato blastocyst). Among several transcripts upregulated in the variantembryos, our attention was drawn to Sox9 as a possible target of theparamutation. The high mobility group (HMG)-box transcriptionfactor Sox9 is a pleiotropic actor in a number of terminaldifferentiation processes, including heart development, sexdetermination, chondrogenesis, neural crest differentiation,gliogenesis, hair follicle function, pancreas development, prostatedevelopment and retina development (Lefebvre et al., 2007; Pochéet al., 2008; Thomsen et al., 2008). A crucial function of Sox9 inproliferation control in the first embryonic stem cells would beconsistent with its known function in various postnatal and adultstem cells and progenitors.

MATERIALS AND METHODSRNA microinjection in fertilized eggsRNA microinjection into the male pronucleus of fertilized B6D2 mouse eggsafter spontaneous ovulation was performed by the established methods ofDNA transgenesis (Hogan et al., 1994). Oligoribonucleotides and theirfluorescein (FITC)-labelled derivatives were obtained from Sigma-Prolabo.The sequences used are presented in Table 1. As controls, fertilized eggswere microinjected with a 20-nt RNA oligonucleotide with an irrelevantsequence, either a fragment of the mouse Sycp1 coding sequence or othermicroRNAs. For studies of embryos during gestation, the control and miR-124-microinjected embryos were separately re-implanted in the left and rightuterine horns of the same foster mother. Investigations were conducted inaccordance with French and European rules for the care and use oflaboratory animals.

The miR-124–Sox9 paramutation: RNA-mediated epigeneticcontrol of embryonic and adult growthValérie Grandjean1,2, Pierre Gounon3,*, Nicole Wagner4,5,*, Luc Martin1,2, Kay D. Wagner4,5, Florence Bernex6,François Cuzin1,2 and Minoo Rassoulzadegan1,2,†

The size of the mammalian body is determined by genetic and environmental factors differentially modulating pre- and postnatalgrowth. We now report a control of growth acting in the mouse from the first cleavages to the postnatal stages. It was evidencedby a hereditary epigenetic modification (paramutation) created by injection of a miR-124 microRNA into fertilized eggs. From theblastocyst to the adult, mouse pups born after microinjection of this miRNA showed a 30% increase in size. At the blastocyst stage,frequent duplication of the inner cell mass resulted in twin pregnancies. A role of sperm RNA as a transgenerational signal wasconfirmed by the giant phenotype of the progeny of transgenic males expressing miR-124 during spermiogenesis. In E2.5 to E8.5embryos, increased levels of several transcripts with sequence homology to the microRNA were noted, including those of Sox9, agene known for its crucial role in the progenitors of several adult tissues. A role in embryonic growth was confirmed by the largesize of embryos expressing a Sox9 DNA transgene. Increased expression in the paramutants was not related to a change in miR-124expression, but to the establishment of a distinct, heritable chromatin structure in the promoter region of Sox9. While theheritability of body size is not readily accounted for by Mendelian genetics, our results suggest the alternate model of RNA-mediated heritable epigenetic modifications.

KEY WORDS: One-cell embryo, Blastocyst, Sox genes, Twins, Non-Mendelian inheritance, Mouse

Development 136, 3647-3655 (2009) doi:10.1242/dev.041061

1Inserm U636, F-06108 Nice, France. 2Université de Nice-Sophia Antipolis,Laboratoire de Génétique du Développement Normal et Pathologique, F-06108Nice, France. 3Centre Commun de Microscopie Appliquée, Université de Nice-SophiaAntipolis, F-06108 Nice, France. 4Inserm-Avenir U907, F-06107 Nice, France.5Faculté de Médecine, Université de Nice-Sophia Antipolis, F-06107 Nice, France.6Institut National de la Recherche Agronomique, Ecole Nationale Vétérinaire d’Alfort,F-94704 Maisons-Alfort, France.

*These authors contributed equally to this work†Author for correspondence (minoo@unice.fr)

Accepted 2 September 2009 DEVELO

PMENT

3648

Determination of the male sexual maturityStarting at day 30 postpartum, individual males were isolated with twoyoung females in each cage. The dates of the first parturitions and, bycounting backward, of the first mating were recorded. Values are the averageof twelve experimental animals and twelve controls.

RT-PCR analysisAnalysis of early embryos was performed with the Gene Expression Cells-to-CT kit (Ambion). For late embryos and adult tissues, RNA was extracted withthe Trizol Reagent (Invitrogen). 0.5 g RNA samples were reverse-transcribedto cDNA by using random primers, hexamers and MMLV reversetranscriptase (Invitrogen). qPCR was performed using the Platinum SYBRGreen qPCR SuperMix-UDG kit (Invitrogen). Sequences of oligonucleotideprimers, obtained from Sigma-Prolabo, are presented Table 1.

Northern blot hybridization10 g of total RNA was loaded onto a 15% denaturing polyacrylamide geland electrophoresed until the bromophenol blue marker reached the bottomof the gel. The separated RNA was electrotransferred to a Hybond N+membrane (Amersham). Hybridization was carried out in the presence of a32P-end-labeled DNA oligonucleotide probe complementary to the maturemiR-124 sequence.

Chromatin immunoprecipitation (ChIP) analysisChIP was carried out using the LowCell ChIP kit (Diagenode, pAb-056-050).ChIP assays were performed with at least five embryos at day 7.5 postcoitus.After dissection, the embryos were rinsed twice in PBS. Proteins and DNAwere then crosslinked with 1% formaldehyde for 8 minutes at roomtemperature and crosslinking was stopped with 125 mM glycine for 5 minutes.Embryos were centrifuged at 470 g in a swing-out rotor with soft decelerationsettings for 10 minutes at 4°C and washed twice in 0.5 ml ice-cold PBS-butyrate solution by gentle vortexing and centrifugation as described above.Preparation of chromatin fragments (~500 bp), immunoprecipitation and DNArecovery were performed as described in the manufacturer’s procedures.

ImmunohistochemistryParaffin sections (5 m) were stained for BrdU with a monoclonal mouseantibody (Roche Molecular Biochemicals, 11170376001) diluted 1:25 inPBS, 0.1% Triton X-100 and 3% BSA; subsequent antibody detection was

performed with the M.O.M. Kit (Vector Laboratories, PK-2200) and DAB(Vector Laboratories, SK-4100) as a substrate. Staining of Sox9 (Millipore,1:100 dilution, AB5535) was performed with a biotinylated anti-rabbitantibody (Vector Laboratories), followed by incubation with peroxidase-coupled Streptavidin (Sigma) and DAB as a substrate. Nuclei werecounterstained with Hematoxylin.

Scanning electron microscopyEmbryos were fixed with 1.6% glutaraldehyde in 0.1 M cacodylate bufferthen rinsed in same buffer, carefully sectioned with a razor blade, dehydratedin ethanol and treated with HMDS (hexamethyldisilazane) before air drying.Samples were coated with gold palladium and observed with a Jeol 6700FSEM.

BrdU labelling of actively cycling cellsPregnant female mice received 50 g/g of 5-bromo-2�-deoxyuridine(BrdU, Roche Molecular Biochemicals) by intraperitoneal injection andwere sacrificed 6 hours later. Embryos were stained by immunolabellingwith an anti-BrdU antibody (Roche Molecular Biochemicals) asdescribed above.

Transgene constructionTo construct the Prm1-miR-124 transgene, pre-miR-124 was amplified byPCR using genomic DNA as a template (oligonucleotide primers revmir124and fwdmir124). DNA oligonucleotides used to construct the Prm1-miR-124 were as follows: fwdmir124, GGACTAGTAGGCCT CTCTCT -CTCCGTGTTCAC; revmir124, ATAAGAATGCGGCCGCCAGCCCCA -TTCTTGGCATTCA. The resulting 101-bp fragment was cloned into thepGEM-T Easy vector (Promega) and sequenced. The construct was digestedwith NotI and ligated into NotI digested pNASSb vector (Clonetech). Theprotamine-1 promoter, a gift of M. Jasin, was then cloned in front of the pre-miR-124 sequence. The construct was digested with EcoRI and injected intothe male pronucleus of fertilized mouse eggs (Hogan et al., 1994).

Statistical methodsData are expressed as mean±s.e.m. ANOVA with the Bonferroni test as apost hoc test was used versus the control. Differences between two groupswere tested using the Mann-Whitney test for non-parametric samples. A P-value less than 0.05 was considered statistically significant.

RESEARCH ARTICLE Development 136 (21)

Table 1. Oligonucleotides used in this studyGenes Accession number Sequence Strand Localization (nucleotide)

DNA oligonucleotides used for quantitative RT-PCR

Acaa2 NM_177470 AAATGTGCGCTTCGGAAC FCGTTAATCCTGCCCACAAAG R

Cd164 NP_058594 CAGCTAAGCCCACAACTCG FGATGTGACAACTGAGGGAGTAGG R

Igf1 NM_010512 CAAAAGCAGCCCGCTCTA FTCGATAGGGACGGGGACT R

Lamc1 NM_010683.2 GGCCGAGTGCCTACAACTT FCAGTGGCAGTTACCCATTCC R

Sox8 NM_011447 CAAGACCCTAGGCAAGCTGT FCTGGGTGGTCTTTCTTGTGC R

Sox9 Q04887 GAAGCTGGCAGACCAGTACC FGGTCTCTTCTCGCTCTCGTTC R

Beta-actin X03765 CTAAGGCCAACCGTGAAAAG FACCAGAGGCATACAGGGACA R

Gapdh BC083149 TGTCCGTCGTGGATCTGAC FCCTGCTTCACCACCTTCTTG R

RNA oligonucleotides used in microinjection experiments

Acaa2 NM_177470 GCGGAAUAGCUGAGCUUCGC 18-35LamC1 BC032194 GAGAUCGCCUCCAGGGAGCTC 51-70Sox9 (5� region) NM_011448 AGUUUCAGUCCAGGAACUUUUC 72-96Sox9 NM_011448 GUUCCUAGAACAUUCACUGUGC 2896-2916Sox8 NM_011447 GGCAACCUUGGAUUCUAGAGUG 72-91Igf1 BC012409 UGCUUGCUCACCUUCACCAGCU 61-80miR-124 MMU459733 UAAGGCACGCGGUGAAUGCC 62-83

F, forward; R, reverse. DEVELO

PMENT

RESULTSThe ‘giant’ phenotype induced by microinjectionof miR-124 RNA in the one-cell embryoIn the first series of experiments, a total of 78 mice were born fromfertilized eggs after microinjection in the male pronucleus of asynthetic single-stranded 22-nt oligoribonucleotide with thesequence of the mature product, identical for the three loci encodingthe microRNA in the mouse genome (Mir124a-1, a-2 and a-3,Mouse Genome Informatics). Quite remarkably, all these miR-124*mice showed a large size at birth, maintained to adulthood with bodyweights ~30% greater than that of the controls (Fig. 1A,B). Such aneffect had not been observed after microinjection of othermicroRNAs (miR-221, -222, -1, -16, -92) (Rassoulzadegan et al.,2006; Wagner et al., 2008) (our unpublished results). As shown inFig. 1C,D, increased growth rates were steadily maintained both bymales and by females during the successive phases of postnatalgrowth (Eisen, 1976). Most organs showed a size proportional tothat of the body, with the exception of an additional increment in sizeof the kidney, pancreas and vertebral axis of 10-20% relative to bodysize (data not shown). As indicated in Fig. 1D, the males reachedsexual maturity an average of 10 days before the controls.

Increased embryonic growth ratesAccelerated growth was not limited to postnatal development butwas already noticeable during embryonic life. As shown in Fig. 2A-C, embryonic day 7.5 (E7.5) embryos exhibited the sameproportional increase in size as the newborns. In order to minimizevariations in developmental timing between foster mothers, allcomparative studies were performed between miR-124- and mock-injected embryos, separately transferred into the two uterine hornsof the same female. Intraperitoneal injection of BrdU andimmunochemical staining performed 6 hours later in miR-124* E7.5embryos (Fig. 2D) evidenced a larger number of labelled cells andthe darker stain indicative of two successive S phases in theserapidly dividing cells (Hogan et al., 1994). Despite their larger sizes,scanning electron microscopy did not evidence a more advanceddevelopmental stage of the E7.5 paramutant embryos (Fig. 2E).

The phenotype was already evident at the very beginning ofdevelopment by the possession of a greater number of cells perblastocyst (Fig. 3A), with a frequent abnormal dispersion of theinner cell fraction, and, in some instances (~5% of the embryos), acomplete duplication of the inner cell mass (Fig. 3B). Consistentwith this observation was the occurrence of a similar proportion oftwin embryos linked to a common placental structure (Fig. 3C-F).

InheritanceCrosses of either male or female miR-124* with wild-type partnersgenerated progenies with body weights significantly greater thanthose of the controls (Fig. 4A). In crosses with wild-type partners,transmission by both genders was efficient until the secondgeneration, with the average weight returning to normal in the F3progeny. The same pattern of heredity was observed in intercrossesbetween paramutant animals of the F0 to F2 generations (data notshown). Based on our previous observations, we considered apossible role of the gametes in the hereditary transmission of RNAmolecules. Electron microscopy examination after EDTA reversestaining (Biggiogera and Fakan, 1998) evidenced increased RNAcontents in miR-124* sperm nuclei (see Fig. S1 in the supplementarymaterial). miR-124 sequences were detected in the testis, butquantitative limitations in the recovery of sperm RNA, together withminute amounts of microRNA, made a search in spermatozoonRNA less reliable and provided variable results.

In order to evaluate the potential role of the transfer of microRNAby sperm, we developed a different strategy by first generating twotransgenic families in which miR-124, expressed at the latepostmeiotic stages under the control of the Prm1 (Protamine 1)promoter, accumulates in spermatozoa (Fig. 4B). All progenies ofthe transgenic males exhibited increased postnatal growth rates andearly sexual maturity similar to those of the miR-124* paramutants,independently of the transmission of the hemizygous transgene (Fig.4C). In this case, however, the ‘giant’ phenotype of the offspring wasnot further transmitted to the progeny by the male or female ‘giants’.Lack of transmission was correlated with the absence of detectablemicroRNA in the sperm of the non-transgenic male offspring (datanot shown).

Effector(s) of the paramutationIn order to identify gene(s) whose expression would be modulatedin the paramutated state, we first considered the known actors ofembryonic and postnatal growth, namely Gh (growth hormone),Igf1 (insulin-like growth factor), Igf2 and their receptors(Fowden, 2003), but none of them was overexpressed at any stagein the paramutant embryos. We then considered the loci withsequence similarities to miR-124 previously identified as targetsof the post-transcriptional regulation by the microRNA using the

3649RESEARCH ARTICLEParamutation of Sox9 in embryonic progenitors

Fig. 1. The ‘giant’ phenotype of the miR-124* paramutant.(A)Adult (2-month-old) mice born from miR-124-injected embryos andmock-injected embryos. Note that the size of the body is increasedboth in length (left) and width (right). (B)2-day-old miR-124* (right)and control (left) pups. (C)Postnatal growth: body weights of the male(n8) and female (n12) paramutants compared with controls (blacksymbols: miR-124*, white symbols: controls, squares: males, circles:females). (D)Logarithmic representation (shown only for males forclarity) illustrates a constant ratio between the growth rates ofparamutants and controls throughout their successive periods ofgrowth (Koops and Grossman, 1991). Arrows indicate the time of thefirst reproductive activity, measured as described in the Materials andmethods. Values (days post-partum) were 34±0.4 for miR-124*founders and 51±1 for controls (n12 for each series).

DEVELO

PMENT

3650

TargetScanHuman and miRBase databases (Karginov et al., 2007;Cheng et al., 2009). Among them, PCR assays for expression inE4.5 to E7.5 embryos led us to retain Sox9 (high-mobility-groupbox transcription factor), LamC1 (laminin g1 subunit), Acaa2(acetyl-Coenzyme A acyltransferase 2) (see Fig. S2 in thesupplementary material), Vamp3 (Vesicle-associated membraneprotein 3) and Cd164 (Cd164 antigen; data not shown). Of specialinterest were LamC1, required for embryonic development(Smyth et al., 1999), and Sox9, extensively studied for its crucialrole in proliferation and differentiation controls in the progenitorsof various organs, including those differentially affected in theparamutant pancreas, cartilage-derived skeletal structures andkidney (Lefebvre et al., 2007; Seymour et al., 2007) (A.Reginensi, M. C. Chaboissier and A. Schedl, personalcommunication).

Transcripts of Sox9 (Fig. 5A,B), LamC1 and Acaa2 (Fig. 5B)were expressed at a low, but significant, level in the wild-typecontrols from as early as E2.5 up to E6.5 and showed a markedincrease in miR-124* embryos. The transcript levels were at least5-fold higher during the pre-implantation and immediate post-implantation stages (E6.5). By contrast, accumulation of Cd164and Vamp3 transcripts, also direct targets of the microRNA, wasidentical to the controls at all stages (Fig. 5B). At later stages(E15.5 to postnatal period), Sox9 RNA levels returned to levelsidentical in paramutants and controls. However, localized regions

of increased expression of the protein were still evident in thepancreas and kidneys (Fig. 5C), the organs in which growth wasmore markedly increased.

Confirmatory evidence of Sox9 as a target of the paramutation wasprovided by microinjection assays performed with twooligoribonucleotides with sequences randomly chosen from thetranscript (Table 2), as previously done in the first two instances ofparamutation analyzed (Rassoulzadegan et al., 2006; Wagner et al.,2008). Both Sox9 sequences induced the oversized phenotype. NeitherLamC1 nor Acaa2 transcript sequences (one randomly chosensequence each) generated the same effect, a negative result difficult tointerpret at the present stage. Further studies were then conducted onthe function of Sox9 in the normal and the modified embryo.

Unchanged expression of the microRNATo investigate further the mechanism of the permanent modificationof Sox9 expression, we first checked whether maintenance and/oroverexpression of miR-124 was involved. Not unexpectedly, givenour previous observations (Rassoulzadegan et al., 2006; Wagner etal., 2008), the altered phenotype was not associated with a modifiedlevel of the microRNA. The expression of miR-124 remainedidentical to that of the controls, both in the embryo and in the adult(Fig. 6A,B). In the adult, expression was essentially only detected inthe brain, with no significant values registered in other organs,including the kidney, in which growth was markedly affected.

RESEARCH ARTICLE Development 136 (21)

Fig. 3. Increased cell number in miR-124* early embryo and twinpregnancies. (A)Increased cell number in miR-124* blastocysts. After4 days in culture, embryos were fixed (3% paraformaldehyde, 1 hour,then methanol:acetic acid 70:30) and stained with DAPI. Cells werecounted under the microscope (n7 for miR-124* and n6 for controls;*, P<0.0001). (B)Although the structure in ‘a’ is that of a normalblastocyst with a unique inner cell mass, in a fraction of embryos (5 to10%), the cells form multiple aggregates, as shown in b-d (arrows).(C-F)Twin miR-124* embryos at successive developmental stages.(C)E7.5, with arrows pointing to duplicated egg-cylinder-stageembryos. (D)E9.5 (same as in C). (E)E12.5, embryos connected to aunique placental structure (*). (F)E17.5.

Fig. 2. Increased growth rate during embryonic development.(A,B)E7.5 miR-124* (top) and control (bottom) embryos taken fromthe two uterine horns of the same foster mother, still enclosed in thedeciduum (A) and dissected (B). (C)Compared embryo lengthsmeasured as indicated in B (n5 for controls and n15 for miR-124*;*, P<0.01). Arrows in B indicate how measurements were taken.(D)BrdU labelling by intraperitoneal injection during pregnancy (E7.5).(E)In spite of their larger sizes, scanning electron microscopy of E7.5miR-124* embryos (left) shows the same developmental stage as thecontrols (right) (Theiler, 1989): al: allantois; am: amniotic cavity;e: embryonic ectoderm; ee: extra-embryonic ectoderm; n: notochord.Note the different sizes indicated by the scaling bars. Scale bars:0.1 mm in A,B,E; 50m in D.

DEVELO

PMENT

Quantitative PCR determination at the first stages followingmicroinjection showed the expected high values in the zygote,quickly followed by a return to the basal level of the control (Fig. 6C).After microinjection of fluorescent FITC-tagged miR-124oligonucleotides into the male pronucleus, the bulk of the microRNAwas excluded from the nucleus within the first hour (Fig. 6D).

A heritable modification of the chromatinstructure of the Sox9 locusA heritable modification of the structure of the chromatin wasevidenced at the Sox9 locus at day E6.5 after miR-124 injection byan increase in the methylated forms of histone H3 (H3K9me2 andme3), a post-translational modification previously associated with avariety of changes in genome expression, most often with silenced

regions (Lachner and Jenuwein, 2002). The modification (Fig. 7)affects a putative upstream regulatory region [region ‘–3K’ of Panet al. (Pan et al., 2009)] in a manner characteristic of the modifiedembryos. The proximal promoter (+1) is loaded with the modifiedhistones characteristic of active promoters H3K4me3 and H3Ac, butto the same extent as in the controls and modified embryos.Interestingly, the modification of the –3K region was transmittedfrom the founders to their progeny.

Sox9 as a regulator of embryonic growthEmbryo overgrowth associated with the locus-specific modulationof Sox9 expression suggests that the gene plays a central role in thecontrol of proliferation of the embryonic stem cells in the firstdevelopmental period. Although not reported so far, such a function

3651RESEARCH ARTICLEParamutation of Sox9 in embryonic progenitors

Fig. 4. RNA-mediated inheritance of the miR-124* giantphenotype. (A)Body weights of the progenies (4-week-old) in serialcrosses with wild-type partners. Squares: male transmission; circles:female transmission. F0 refers to three male founders and three femalefounders born after miR-124 microinjection; F1 to their offspring incrosses with wild-type partners; F2 to F4 to the pooled offspring, eithermale or female, of two crosses with wild-type partners of two to threemiR-124* mice, male or female, randomly chosen in the previousgeneration. Ordinate: ratio of body weight to control wild type, namely19.9±1.5 g (n13) for the males and 16.8±1.0 g (n11) for thefemales. Values of body weights at each generation are the average of10 to 14 individual animals, with standard errors of the mean (s.e.m.)ranging between 5 and 10% (F0 to F2 values significant at P<0.001, F3at P<0.01 vs controls). (B)Expression of miR-124 measured by RT-qPCRnormalized to Gapdh in Prm1miR-124 transgenic and controlspermatozoa (**, P<0.001). (C)Transmission of a giant phenotype byPrm1miR-124 transgenic males. The flow chart summarizes the crossesbetween transgenic males and their progeny with wild-type partners.Symbols indicate the progeny of each cross; the total number of micein each class (n) and the average body weights are indicated in theinsert. Crossed symbols indicate the carriers of the transgene. Theexperiment was repeated with identical results by two independentlyestablished founders. Complete values are shown in Table S1 in thesupplementary material.

Fig. 5. Early expression of Sox9, LamC1 and Acaa2 is increased inmiR-124* embryos. (A)Sox9 RNA in the E2.5, E6.5 and E15.5 embryo.RT-qPCR values are normalized to -actin (means±s.e.m.) in six or moretotal RNA preparations. **, P=0.001. (B)Transcript levels measured forSox9, Lamc1, Acaa2, Cd164 and Igf1 in miR-1* embryos at successivedevelopmental stages, in comparison with the levels in control embryosand mice of the same age. (C)In the E15.5 embryo (b,d), levels of theSox9 protein are still elevated in comparison with the controls (a,c) in thekidney (K) and pancreas (P) but not in other organs such as the lung (L).

DEVELO

PMENT

3652

is not completely unexpected, as a series of terminal differentiationpathways in the adult are thought to be dependent on Sox9 activityin early progenitors.

Independent evidence of a role of Sox9 in early embryonic growthwas acquired by modulating its expression at the first embryonicstages. Because of the lethality at birth of heterozygotes carryingnull mutations (Bi et al., 2001), we first attempted to downregulateexpression by microinjecting si-RNA molecules directed againstSox9 mRNA into zygotes. Embryos were of a small size and theirdevelopment was largely abnormal, leading to extensivemalformations at E10.5 and the early death expected from the E11.5lethality of inactivating mutations. While it was clear that Sox9 isnecessary during early development, a specific effect on cellproliferation and growth control could not be ascertained in this way(data not shown).

Conversely, to achieve increased levels of expression, we resortedto microinjection of a construct in which the complete Sox9 cDNAsequence is inserted downstream of the early CMV1 promoter,which drives high levels of expression in the one-cell embryo (ourunpublished results). Embryos collected at E7.5 possessed, with anormal morphology, a size larger than that of the controls (Fig. 8A).Contrary to the paramutant situation, development was againeventually arrested at a later stage (E11.5), likely owing to Sox9expression not being correctly regulated past the egg cylinder stage.

Growth control in the late embryonic period andpostnatal growthAs embryogenesis proceeded, the levels of Sox9 transcripts in totalRNA declined to the levels of the controls (Fig. 5B). Local sites ofhigh expression were still noted in the organs that had a differentiallyhigher growth rate, such as the kidney and pancreas (Fig. 5C), butthis was not the case for other parts of the body, as illustrated for thelung in Fig. 5C. Increased growth rates and cell proliferation werenevertheless maintained in a coordinated manner for the whole body,raising the question of the genetic determination of the giantphenotype in late embryogenesis and postnatal growth.

One could argue that large pools of stem cells generated byaccelerated growth during the first part of the pregnancy aresufficient to produce the adult ‘giant’ phenotype. An alternativehypothesis, however, is that, as in other developmental processes

(Lefebvre et al., 2007), Sox9 expression initiates a cascade of geneexpression and regulatory mechanisms. Interestingly, it does notinvolve the genes known for their function in growth control, Gh andIgf1, whose expression was neither modified in the young nor in theadult. Two lines of evidence point to Sox8 as a possible candidate.Studies on testicular differentiation had suggested that Sox8 takesover some of the functions of its close homologue Sox9 (Chaboissieret al., 2004), as Sox8-negative mutants were characterized by theirsmall size (Sock et al., 2001). RT-qPCR assays performed on RNAfrom the testis, kidney and brain RNA detected a significant increaseof Sox8 transcripts in the young (16.5 days post-partum ) and in theadult (Fig. 8B). Although further studies are clearly required, andother genes could be involved, we hypothesize that secondaryinduction of Sox8 may play a role in the maintenance of acceleratedgrowth rates up to adulthood.

DISCUSSIONThe epigenetic change and the striking increase in body sizeestablished upon microinjection in the one-cell embryo of miR-124RNA are mitotically stable during development and transmittedthrough meiosis and fertilization in a non-Mendelian manner. Wetherefore used the term ‘paramutation’, with its initial meaning of

RESEARCH ARTICLE Development 136 (21)

Fig. 6. Levels of the microRNA are not quantitatively altered inthe miR-124* paramutants. (A)miR-124 expression in the adult brainand kidney of miR-124* and control animals was analyzed by northernblotting and hybridizing with the miR-124 oligonucleotide probecorresponding to the complement of the mature miR-124 sequence.Purified RNA was analyzed on a 15% denaturing polyacrylamide gelstained with Ethidium Bromide. miR-221 microRNAs were loaded onthe left part of the gel as a size marker. tRNA was used as a loadingcontrol (left panel). The top band corresponds to the pri-miR124 andthe bottom one to the microRNA (right panel). (B)Quantification ofmiR-124 normalized to Gapdh RNA in the brain and kidney of E6.5embryos and adult miR-124* and control mice. Results are themean±s.e.m. for at least six samples. (C)After microinjection, copynumbers of miR-124 return quickly to the value of the control. miR-124-injected and control embryos were cultured in vitro in M16 culturemedium (Sigma) and harvested every 24 hours. miR-124 copy numberwas estimated by RT-qPCR. Results are the means of three or moreexperiments performed with a minimum of seven embryos.(D)Visualization of FITC-labeled-miR-124 oligoribonucleotide injectedinto fertilized eggs.

Table 2. Oligonucleotide microinjection assaysWeight

Microinjected oligonucleotide† n E7.5‡ (mg±s.e.m.) P18§ (g±s.e.m.)

Experiment 1

Control 8 13.7±0.01Igf1 nt 61-80 7 13.8±0.02 nsmiR-124 7 19.3±0.04*Sox9 nt 2881-2900 8 16.3±0.02*

Experiment 2

Control 16 7.8±0.20miR-124 7 11.6±0.30*Sox9 nt 2881-2900 12 10.8±0.33*Sox9 nt 61-80 12 9.8±0.27*LamC1 nt 50-68 11 7.2±0.18 ns†Controls microinjected with buffer; oligonucleotides reproduce the indicatedregions of the mRNAs; sequences are provided in Table 1. ‡Embryos collected at E7.5 from foster mothers carrying both mock and RNAmicroinjected embryos; ns, not significant. §Measurements under the same experimental conditions during postnatal growth(18 dpp). *, P<0.001.

DEVELO

PMENT

hereditary epigenetic change, as initially observed in plants (Brink,1956). This does not, however, imply that the plant and animalphenomena are alike in every respect. Indeed, there are significantdifferences. The plant paramutation is essentially a gene-silencingeffect (Chandler, 2007), whereas, in the mouse, it refers totranscriptional activation at the chromatin level. While non-codingRNAs were reported to control epigenetic states in plants, as in otherorganisms, including in the Drosophila germ line (Brennecke et al.,2008; Chambeyron et al., 2008), transgenerational determination ofan epigenetic state by gametic RNA is, so far, unique to the mouseparamutation.

As in previously analyzed instances of mouse paramutation,homologous RNAs appear as the initial signal, in this case,microRNA miR-124. The subsequent maintenance of the phenotypefrom the blastocyst to the adult does not appear to result from apermanent activity of the microRNA. Our results, rather, indicatethat initial exposure to the microRNA, and more generally to RNAwith sequence homology to the transcript, results in a change in thechromatin structure of the promoter. Although the known markersof transcriptional activity, acetylation of histone H3 andtrimethylation of H4 (H3K4me3), were indeed detected in thepromoter region (in the same amounts in both the paramutants andthe controls), a discriminating and hereditary modification observed

was the increase in H3K9me2 and H3K9me3 at a potentialregulatory element in the upstream region of the promoter. Thisresult was not that expected, as these variations had been associatedwith cases of silencing, most notably X chromosome inactivation(Lachner and Jenuwein, 2002). Several hypotheses might beconsidered, such as that of a repressor-RNA encoded in the upstreamregion. A peculiar histone modification such as H3K9 methylationmight also serve different purposes in different biological contexts.A crucial piece of information that only the genetic analysis couldprovide is whether the upstream region is necessary for theestablishment of the paramutated state.

These observations leave open the important question of themechanism by which RNA directs the establishment of the locus ofthe modified chromatin structure. As recently described for X-chromosome inactivation (Zhao et al., 2008), one likely model isthat the inducing RNAs target a chromatin remodelling system tothe affected locus, possibly including one of the trithorax groupproteins associated with transcriptional activation (Schuettengruberet al., 2007). The fact that this first encounter occurs in the secrecyof the one-cell embryo makes a search for molecular aspectstechnically demanding. To face this problem, we are currentlyworking on the development of cell culture systems more amenableto molecular analysis (H. Ghanbarian, V.G. and M.R., unpublished).

A role of RNA as a transgenerational determinant was confirmedby the induction of the epigenetic state in eggs fertilized by the miR-124-loaded sperm of Prm1-transgenic males. In this case, however,the ‘giant’ phenotype was not transmitted further by the ‘giant’ non-transgenic progeny, whose sperm did not maintain copies of the

3653RESEARCH ARTICLEParamutation of Sox9 in embryonic progenitors

Fig. 7. Covalent histone modification in the Sox9 promoter ofthe paramutants and their progeny. Chromatin from at least fiveembryos at E7.5 was precipitated with antibodies directed againstdimethyl H3-K9 (Millipore, 17-648), trimethyl H3-K9 (Millipore, 17-625)and acetylated H3 (Millipore, 06-590). After DNA recovery, theprecipitates were evaluated by real-time PCR for the level ofenrichment, over the negative control antibody, at the 5� upstream(–3K) and promoter (+1) region of Sox9, which have previously beenshown to be enriched in H3K9me2 and me3 and in H3K4me3 andH3Ac, respectively (Pan et al., 2009). To validate the ChIP assays, wemonitored the association of each antibody with the Gapdh promoter(data not shown) (O’Neill et al., 2006). F0 refers to E7.5 miR-124* andcontrol embryos taken from the two uterine horns of the same fostermother, and F1 to their offspring in crosses with wild-type femalepartners.

Fig. 8. (A)Forced expression of Sox9 in the E7.5 embryo. Microinjectionof a CMV-Sox9 construct results in a larger size of the E7.5 embryos.Note the normal morphology of the transgenic embryos. (B)Length(mm) of transgenic and control embryos (n7, **, P<0.001). (C)Sox9upregulation in embryo and Sox8 upregulation during postnatalgrowth. Ratio of miR-124* to control values of Sox9 and Sox8 RNA (RT-qPCR normalized to -actin, mean±s.e.m. of six or more samples).Sox8, not detected in the E2.5 embryo, is expressed at the control levelat later stages, but is increased during postnatal growth and in theadult. Results shown for the testis are representative of the otherorgans analyzed (brain, kidney).

DEVELO

PMENT

3654

microRNA. Not excluding other possibilities, such as differentialmarking of the RNA molecules or different nucleoproteincomplexes, the most likely difference between the transgenic systemand the paramutant mice is a smaller number of copies transferredto the oocyte via the transgenic sperm as compared withmicroinjection. This observation might provide a clue regarding theconditions necessary for hereditary transmission by distinguishingthe requirements for epigenetic determination in the embryo and thesubsequent transfer to the germ line.

Epigenetic modulation of Kit and Cdk9 expression was inducedby oligoribonucleotides with sequences of their respectivetranscripts as efficiently as by the cognate microRNAs(Rassoulzadegan et al., 2006; Wagner et al., 2008). A searchperformed on this basis for genes with partial similarities to the miR-124 sequence provided a shortlist of candidates. These candidatesnecessarily included several of the known targets of the microRNApost-transcriptional regulation. On the basis of their early expressionin development, five of them were retained for further studies,among which, three showed an elevated level of expression in theparamutant embryos. Interestingly, this was not the case of everytarget of the microRNA, as neither Vamp3 nor CD164 was affected.None of the three candidates at this stage, Acaa2, LamC1 and Sox9,had been reported to act in the control of embryonic growth. Themitochondrial acetyl-Coenzyme A acyltransferase 2 encoded byAcaa2 is not expected to act as a central regulator of growth,although the development of a large body implies increasedmetabolic activity and a corresponding contribution ofmitochondria. LamC1, encoding the gamma 1 laminin subunit, anda physiological target of miR-124 during neural development (Caoet al., 2007), is necessary for basement membrane formation anddevelopment past the blastocyst stage (Smyth et al., 1999). Thedevelopment of a giant embryo certainly requires basal membraneformation, but again, a central regulatory role did not appear to belikely. In fact, neither LamC1 (Table 2) nor Acaa2oligoribonucleotides (data not shown) induced the ‘giant’ phenotypewhen injected into the embryo.

The situation was different for Sox9, although its expression hasnot so far been recorded at the early developmental stages byanalysis of expressed sequence tags (EST). Its level of expressionmeasured by quantitative PCR in the early embryo was indeed low,but clearly significant, and even more so in the paramutated form.Epigenetic upregulation of Sox9 as a primary modification in themiR-124* embryos would be compatible with a correspondingincrease of other transcripts in a complex network of regulations, notunexpected given the multiple roles of the gene in development. Acommon theme of Sox9 analysis in various differentiation pathways(Lefebvre et al., 2007) is the ability to function at an early stage,possibly in the control of the proliferation and differentiation ofprogenitors. This is precisely the role suggested for themultipotential embryonic stem cells by the present results. It is ofinterest to consider that the heritability of body size in human is notfully accounted for by Mendelian determinants (Maher, 2008).Inherited epigenetic determinations might offer an alternativemodel, especially considering the significant concentration of RNAof human sperm (Miller et al., 2005).

As it is, and as it was for mutations in classical genetics,paramutation could lead to the identification of genes involved incomplex developmental phenotypes. Microinjection assays performedwith other RNAs, microRNAs and their combinations might allow thediscovery of new regulatory pathways. Finally, note that previouslywe considered (Wagner et al., 2008) that paramutation can be viewedbasically as a surveillance program, recognizing an abnormal RNA

profile that indicates alteration of the locus, by either illegitimaterecombination or transposon insertion, and maintaining a crucialgene function by an increased expression of the correct allele.

AcknowledgementsWe thank M. C. Chaboissier and A. Schedl for the communication ofunpublished results and for critical reading of the manuscript. The experttechnical assistance of F. Paput, J. Paput, M. Radjkhumar, C. Vannetti, T. Amineand M. Bossert is gratefully acknowledged. This work was funded by grantsfrom Ligue Contre le Cancer as ‘Equipe Labellisée’ and from Agence Nationalede la Recherche (ANR-06-BLAN-0226 PARAMIR), France. N.W. was therecipient of a fellowship from the Fondation de France.

Supplementary materialSupplementary material for this article is available athttp://dev.biologists.org/cgi/content/full/136/21/3647/DC1

ReferencesBi, W., Huang, W., Whitworth, D. J., Deng, J. M., Zhang, Z., Behringer, R. R.

and de Crombrugghe, B. (2001). Haploinsufficiency of Sox9 results in defectivecartilage primordia and premature skeletal mineralization. Proc. Natl. Acad. Sci.USA 98, 6698-6703.

Biggiogera, M. and Fakan, S. (1998). Fine structural specific visualization of RNAon ultrathin sections. J. Histochem. Cytochem. 46, 389-395.

Brennecke, J., Malone, C. D., Aravin, A. A., Sachidanandam, R., Stark, A.and Hannon, G. J. (2008). An epigenetic role for maternally inherited piRNAs intransposon silencing. Science 322, 1387-1392.

Brink, R. A. (1956). A genetic change associated with the R locus in maize whichis directed and potentially reversible. Genetics 41, 872-879.

Cao, X., Pfaff, S. L. and Gage, F. H. (2007). A functional study of miR-124 in thedeveloping neural tube. Genes. Dev. 21, 531-536.

Chaboissier, M. C., Kobayashi, A., Vidal, V. I., Lutzkendorf, S., van de Kant,H. J., Wegner, M., de Rooij, D. G., Behringer, R. R. and Schedl, A. (2004).Functional analysis of Sox8 and Sox9 during sex determination in the mouse.Development 131, 1891-1901.

Chambeyron, S., Popkova, A., Payen-Groschene, G., Brun, C., Laouini, D.,Pelisson, A. and Bucheton, A. (2008). piRNA-mediated nuclear accumulationof retrotransposon transcripts in the Drosophila female germline. Proc. Natl.Acad. Sci. USA 105, 14964-14969.

Chandler, V. L. (2007). Paramutation: RNA-mediated instructions passed acrossgenerations. Cell 23, 641-645.

Cheng, L. C., Pastrana, E., Tavazoie, M. and Doetsch, F. (2009). miR-124regulates adult neurogenesis in the subventricular zone stem cell niche. Nat.Neurosci. 12, 399-408.

Eisen, E. J. (1976). Results of growth curve analyses in mice and rats. J. Anim. Sci.42, 1008-1023.

Fowden, A. L. (2003). The insulin-like growth factors and feto-placental growth.Placenta 24, 803-812.

Hogan, B., Costantini, F. and Lacy, L. (1994). Manipulating the Mouse Embryo –a Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor LaboratoryPress.

Karginov, F. V., Conaco, C., Xuan, Z., Schmidt, B. H., Parker, J. S., Mandel, G.and Hannon, G. J. (2007). A biochemical approach to identifying microRNAtargets. Proc. Natl. Acad. Sci. USA 104, 19291-19296.

Koops, W. J. and Grossman, M. (1991). Multiphasic growth and allometry.Growth Dev. Aging 55, 203-212.

Lachner, M. and Jenuwein, T. (2002). The many faces of histone lysinemethylation. Curr. Opin. Cell Biol. 14, 286-298.

Lagos-Quintana, M., Rauhut, R., Yalcin, A., Meyer, J., Lendeckel, W. andTuschl, T. (2002). Identification of tissue-specific microRNAs from mouse. Curr.Biol. 12, 735-739.

Lefebvre, V., Dumitriu, B., Penzo-Mendez, A., Han, Y. and Pallavi, B. (2007).Control of cell fate and differentiation by Sry-related high-mobility-group box(Sox) transcription factors. Int. J. Biochem. Cell Biol. 39, 2195-2214.

Maher, B. (2008). Personal genomes: The case of the missing heritability. Nature456, 18-21.

Makeyev, E. V., Zhang, J., Carrasco, M. A. and Maniatis, T. (2007). TheMicroRNA miR-124 promotes neuronal differentiation by triggering brain-specific alternative pre-mRNA splicing. Mol. Cell 27, 435-448.

Miller, D., Ostermeier, G. C. and Krawetz, S. A. (2005). The controversy,potential and roles of spermatozoal RNA. Trends Mol. Med.11, 156-163.

O’Neill, L. P., VerMilyea, M. D. and Turner, B. M. (2006). Epigeneticcharacterization of the early embryo with a chromatin immunoprecipitationprotocol applicable to small cell populations. Nat. Genet. 38, 835-841.

Pan, Q., Wu, Y., Lin, T., Yao, H., Yang, Z., Gao, G., Song, E. and Shen, H.(2009). Bone Morphogenetic Protein-2 induces chromatin remodeling andmodification at the proximal promoter of Sox9 gene. Biochem. Biophys. Res.Commun. 379, 356-361.

RESEARCH ARTICLE Development 136 (21)

DEVELO

PMENT

Poché, R. A., Furuta, Y., Chaboissier, M.-C., Schedl, A. and Behringer, R. R.(2008). Sox9 is expressed in mouse multipotent retinal progenitor cells andfunctions in Müller Glial cell development. J. Comp. Neurol. 510, 237-250.

Rassoulzadegan, M., Grandjean, V., Gounon, P., Vincent, S., Gillot, I. andCuzin, F. (2006). RNA-mediated non-mendelian inheritance of an epigeneticchange in the mouse. Nature 441, 469-474.

Schuettengruber, B., Chourrout, D., Vervoort, M., Leblanc, B. and Cavalli, G.(2007). Genome regulation by polycomb and trithorax proteins. Cell 128, 735-745.

Seymour, P. A., Freude, K. K., Tran, M. N., Mayes, E. E., Jensen, J., Kist, R.,Scherer, G. and Sander, M. (2007). SOX9 is required for maintenance of thepancreatic progenitor cell pool. Proc. Natl. Acad. Sci. USA 104, 1865-1870.

Smyth, N., Vatansever, H. S., Murray, P., Meyer, M., Frie, C., Paulsson, M.and Edgar, D. (1999). Absence of basement membranes after targeting theLAMC1 gene results in embryonic lethality due to failure of endodermdifferentiation. J. Cell Biol.144, 151-160.

Sock, E., Schmidt, K., Hermanns-Borgmeyer, I., Bosl, M. R. and Wegner, M.(2001). Idiopathic weight reduction in mice deficient in the high-mobility-grouptranscription factor Sox8. Mol. Cell. Biol. 21, 6951-6959.

Theiler, K. (1989). The House Mouse Atlas of Embryonic Development. Berlin:Springer-Verlag.

Thomsen, M. K., Butler, C. M., Shen, M. M. and Swain, A. (2008). Sox9 isrequired for prostate development. Dev. Biol. 316, 302-311.

Visvanathan, J., Lee, S., Lee, B., Lee, J. W. and Lee, S. K. (2007). The microRNAmiR-124 antagonizes the anti-neural REST/SCP1 pathway during embryonic CNSdevelopment. Genes Dev. 21, 744-749.

Wagner, K. D., Wagner, N., Ghanbarian, H., Grandjean, V., Gounon, P.,Cuzin, F. and Rassoulzadegan, M. (2008). RNA induction and inheritance ofepigenetic cardiac hypertrophy in the mouse. Dev. Cell 14, 962-969.

Zhao, J., Sun, B. K., Erwin, J. A., Song, J. J. and Lee, J. T. (2008). Polycombproteins targeted by a short repeat RNA to the mouse X chromosome. Science322, 750-756.

3655RESEARCH ARTICLEParamutation of Sox9 in embryonic progenitors

DEVELO

PMENT