Deletion of the MBII-85 snoRNA Gene Cluster in Mice Results in Postnatal Growth Retardation

1The human genetic locus 15q11-q13 is subject to genomic imprinting that is controlled by a bipartite imprinting center (IC). Imprinting defects or chromosomal rearrangements/deletions within this locus (Figure 1) are responsible for the development of two clinical disorders—Angelman (AS) and Prader-Willi (PWS) syndromes. AS (MIM 105830) is a complex neurogenetic disorder characterized by mental retardation, severe limitations in speech and language and abnormal behavior, and results from loss of maternal expression of the Ube3A gene. PWS (MIM 176270) is a complex neurogenetic disorder with a population prevalence of 1 in 10 000 to 50 000 that is characterized by decreased fetal activity, muscular hypotonia, failure to thrive, short stature, obesity, mental retardation, and hypogonadotropic hypogonadism, for review, see. PWS results from lack of paternal expression of one or several imprinted genes within the PWS/AS locus. Several paternally expressed protein-coding genes map to this locus, including NECDIN (NDN), MAGEL2, MKRN3, and the bi-cistronic SNURF-SNRPN (Figure 1B). There are also numerous paternally expressed C/D box snoRNA genes located downstream from the SNURF-SNRPN gene. Most of them are organized into two main clusters of HBII-85 and HBII-52 snoRNAs, containing 29 and 47 copies, respectively. Other snoRNAs are present as single (HBII-436, HBII-13 and HBII-437) or double copy (HBII-438a/438b) genes (Figure 1B). Most, if not all, snoRNAs are processed from a long, non-protein-coding RNA (npcRNA) transcript designated U-UBE3A-ATS in human and Lncat (large paternal non-protein-coding RNA, encompassing Snurf-Snrpn and Ipw exons together with the Ube3a antisense transcript) in mouse. U-UBE3A-ATS extends ∼450 kb from the untranslated U exons upstream of the small nuclear ribonucleoprotein N (SNURF/SNRPN) gene to the UBE3A gene (Figure 1B).

2In mouse the syntenic PWS/AS locus is located on chromosome 7C and contains all the aforementioned protein coding and non-protein-coding gene orthologs, except for the presence of protein-coding gene Frat 3 and absence of HBII-437 and HBII-438a/438b snoRNA genes (Figure 1C). Several mouse models for PWS have already been generated (Figure 1D). They can be divided into 3 groups: 1) transgenic mouse lines with disruptions of the PWS/AS locus; 2) mice with targeted elimination of the imprinting center (IC) controlling transcription of PWS genes, or targeted elimination of individual, single genes from the PWS locus; and 3) mice with uniparental paternal disomy (UPD).

3In all existing PWS mouse models (Figure 1D) that involve large deletions comprising several paternally expressed, imprinted genes, severe phenotypes (failure to thrive and early postnatal lethality) were observed. Targeting of the single Mkrn3 and Snrpn genes, or some of the Ipw exons together with the MBII-52 snoRNA genes cluster or deletion of an analogous region in human all produced no PWS-like phenotypes. Elimination of the Magel 2 gene caused altered behavioral rhythmicity (Figure 1D8). However this gene is unlikely to be a main “PWS player”. Currently, only elimination of the single Necdin gene leads to the development of early postnatal lethality and neurological abnormalities resembling the PWS, although the phenotypic effect depends on the targeted region and genetic background of the mice, some of which had no apparent phenotype. The PWS critical region (PWScr) was narrowed to the locus containing the MBII-85 small nucleolar RNA (snoRNA) gene cluster based on existing mouse models and gene expression analysis in PWS patients with chromosomal translocations. These studies indicate that SNURF/SNRPN, MKRN3, NECDIN and MAGEL2 genes are unlikely to play a primary role in the pathogenesis of PWS. However, the question of their possible functional contribution to more severe phenotypic expression seen in typical PWS patients remains open. We have applied the “chromosome engineering” technique to delete the PWScr in mice. When the deleted allele is inherited maternally (PWScrm−/p+), no phenotypic abnormalities are visible. When it is inherited paternally (PWScrm+/p−), we consistently observe postnatal growth retardation in mice and less than 15 percent postnatal lethality in 129SV x C57BL/6 genetic crosses.

4To test the hypothesis that the PWScr is the most probable candidate region for neonatal lethality, failure to thrive and postnatal growth retardation, we devised the following strategy for producing PWScrm+/p− mice (Figure 1E and 1F). Hypoxanthine-guanine phosphoribosyltransferase (HPRT)-deficient ES cells, AB2.2, were modified through homologous recombination (HR) using targeting constructs 5′HPRT/PWScr_targ and 3′HPRT/PWScr_targ to place loxP sites proximal to the 5′ flanking region of the MBII-85 snoRNA gene cluster and distal to the Ipw exon C, respectively (Figure 1E). Deletion of the PWScr harboring the entire cluster of MBII-85 snoRNA genes together with Ipw exons A-C was accomplished by expressing CRE recombinase in ES-targeted cells (Figures 1F and 2A) and injecting these into blastocysts. We identified two chimaeras derived from one of the PWScr-deleted ES clones with successful germ-line transmission. Deletion of the PWScr allele was confirmed several ways: 1) Resistance of ES cells to HAT media requires a functional HPRT gene. Restoring the intact HPRT gene through deletion of the PWScr leads to resistance of ES cells to HAT media. 2) Southern blot analysis of all HAT-resistant ES colonies (data not shown), as well as all PWScr-deleted mice, using 5′HR and 3′HR probes (Figure 2A and 2B), revealed identical bands corresponding to the correctly deleted PWScr allele (19998 bp). 3) We PCR amplified and sequenced flanking regions of the PWScr locus together with the inserted HPRT gene (Figures 1G, 2C, and S1) using PCR primers MB85seqD1 and MB85seqR1 (Table S1), and confirmed the deleted PWScr allele as well.

5PWScrm+/p− pups born from chimaeras were significantly smaller than their wild-type siblings, such that on postnatal day 10, the PWScrm+/p− individuals (Figure 3) could be reliably predicted prior to DNA analysis. We monitored the weights of all mice over several weeks after continuous breeding and found that only the PWScrm+/p− and PWScrm−/p− mice displayed postnatal growth retardation compared to the PWScrm+/p+ siblings. Statistical analyses were performed on mice separated by genotype, genetic background and gender (Figures 4A–4D and S2), as well as separately for litters (data provided upon request). Postnatal growth retardation in PWScrm+/p− mice was observed, thus far, over six generations, independent of genetic background [e.g., 129SV x C57BL/6 (>85% C57BL/6 genetic background contribution, Figure 4A and 4B), 129SV x C57BL/6 x FVB/N (∼50% FVB/N contribution, Figure S2A and S2B), and 129SV x C57BL/6 x BALB/c (∼50% BALB/c contribution, Figure S2C and S2D)]. Differences in growth dynamics between PWScrm+/p+ and PWScrm−/p− or PWScrm+/p− mice continued to be statistically significant into adulthood (up to 1 year in BALB/c crosses) (Figure 4A, 4B, and S2). Moreover, not a single case of obesity was detected. Interestingly, when growth dynamics for mice were analyzed by gender, deficiencies in the PWScrm+/p− or PWScrm−/p− female mice tended to be less pronounced than those in the male deletion mice (Figure 4A, 4B, S2C, and S2D). This observation is well correlated with a statistical analysis of PWS patients indicating that the degree of short stature is more prominent in males than in females, irrespective of ethnic groups (genetic background). Furthermore, mice with the maternally transmitted PWScr-deleted allele were indistinguishable from their wild type littermates (Figure 4C). In addition, males (5′HPRT) derived from 5′HPRT/PWScr_targ-targeted ES cells were crossed with C57BL/6 females. As expected, we did not observe growth retardation in the 5′HPRTm−/p+ mice (p=0.275) (Figure 4D).

6Notably, the observed postnatal growth retardation phenotype became apparent during the first week of life starting from postnatal day 5 in males and 6 in females (Figure 4A and 4B; Table S2), while there were no growth differences at early postnatal ages P1-P4 (Figure 4A and 4B; Table S2). In agreement with the absence of early postnatal growth retardation, no weight differences were observed during embryonal development at E12.5, E15.5 and E18.5, or in late gestation (E15.5 and E18.5) placentas (Figure 4E, and data not shown). Hence, one possible explanation for our observations might be poor breast-feeding in PWScrm+/p− pups as is the case with postnatal PWS patients. Insufficient milk intake resulting in growth retardation is consistent with the Holland hypothesis and observations in TgPWS mice (transgenic deletion PWS mouse model; Figure 1D1) suggesting, that the main basis of the PWS syndrome is not obesity and uncontrollable craving but early postnatal starvation.

7The PWScrm+/p− mice exhibit postnatal growth retardation, but contrary to early predictions postnatal lethality was observed in only 15 percent of the cases in 129SV x C57BL/6 genetic crosses (Figure 4F). On the other hand, we did not observe postnatal lethality in the FVB/N crosses. PWScrm+/p− mice were dying from postnatal days 1 to 22. The surviving PWScrm+/p− and PWScrm−/p− mice are alive and apparently well for at least 1 year. In future experiments, it would be interesting to test the possible role of MBII-13 and MBII-436 snoRNAs and/or the genomic region between the Snrpn and MBII-85 genes for their contributions to more severe phenotypes with higher rates of postnatal lethality. In addition, our PWScrm+/p− mouse model is in good agreement with two recently identified patients with a balanced chromosomal translocation involving SNRPN. Both patients lack HBII-85 expression and exhibit a mild PWS phenotype, (e.g., they did not require gavage feeding, but were growth retarded.)

8PWS is also characterized by hypogonadotropic hypogonadism and infertility in patients. We studied the fertility of the PWScrm+/p− and PWScrm−/p+ mice and observed that PWScrm+/p− males and females transmitted the PWScr deleted allele to offspring. To further extend our observations, we established 10 breeding pairs from each of the PWScrm+/p− and PWScrm−/p+ males with 80 wild-type females. All matings resulted in pregnancies leading to successful live births with litter sizes around eight (7.93 ± 0.46 for PWScrm+/p− and 8.33 ± 0.48 for PWScrm−/p+ males), and always included wild-type mice and mice carrying the PWScr-deleted allele in a ratio of 1:1 (4.17 ± 0.34: 3.66 ± 0.38 for PWScrm+/p− and 4.00 ± 0.54: 4.17 ± 0.53 for PWScrm−/p+; Table S3). Thus, male mice containing the PWScr-deleted allele inherited maternally or paternally, are transmitting this allele in a Mendelian fashion.

9The PWS locus includes several paternally expressed, protein coding genes, including Necdin, Magel2, Mkrn3, Frat3, and the bi-cistronic Snurf-Snrpn. To examine whether the deletion of the paternal PWScr from mouse chromosome 7C perturbed the expression of the aforementioned imprinted genes, we analyzed their expression levels in our mice by RT-PCR and real-time PCR. We failed to detect any significant differences in the expression levels of the investigated genes in PWScrm+/p− mice compared to control littermates PWScrm+/p+ (Figure 5A; Table 1). In addition, the controversial data concerning the involvement of Necdin in the PWS phenotype prompted us to also examine its expression in more detail by Northern blot hybridization (Figure 5B). Consistent with the RT-PCR and real-time PCR results, there were no differences in the levels of Necdin mRNA in brains of the PWScrm+/p−, PWScrm−/p+ or control PWScrm+/p+, 5′HPRTm+/p+ and 3′HPRTm+/p+ mice (Figure 5B; Table 1 and data not shown).

10We also examined the levels of the maternally expressed, imprinted Ube3a and Atp10a protein coding genes, located at the end of the PWS locus and involved in the development of AS (Figure 1B and 1C). The RT-PCR (Figure 5A), real-time PCR (Table 1), and Northern blot hybridization (Figure 5C and 5D) analyses revealed similar expression levels of both genes in brains of PWScrm+/p− and PWScrm−/p+, as well as in control mice.

11The mouse PWS locus on chromosome 7C also contains numerous neuronal, paternally expressed C/D box snoRNA genes, including two single copies of MBII-436 and MBII-13, and two multiple copy clusters, MBII-85 and MBII-52. Vertebrate snoRNAs are embedded in introns of protein coding genes, which are posttranscriptionally processed to yield mature mRNA and snoRNA. Occasionally the spliced exons are devoid of open reading frames, as if the sole function of the transcript is the expression of the snoRNA. This is apparently the case for most, if not all, snoRNAs from the PWS locus that are co-transcribed with the large paternally expressed polycistronic Lncat npcRNA. The Lncat transcript is complex and generates, by alternative splicing and other processing events, numerous RNA products (e.g., those represented by expressed sequenced tags (ESTs) and mature snoRNAs). The corresponding Ipw exons are a subset of Lncat-derived ESTs and map to the MBII-85 and MBII-52 snoRNA clusters that are interspersed with repeated exons A1/A2 and G1/G2, respectively. The Ipw exons B, C, H, E, F map between both clusters. Northern blot analyses of RNA samples extracted from brains of PWScrm+/p− mice revealed the complete absence of MBII-85 snoRNA while expression of all other snoRNA genes in the PWS locus were unaffected (Figure 6A).

12We then analyzed the expression of the Ipw exons using Northern blot hybridization with a cDNA probe containing the F and G exons, and found it to be slightly decreased in PWScrm+/p− mice (Figure 6B). However, expression of MBII-52 snoRNA was not altered, presumably because the primary Lncat transcript can undergo different processing pathways to yield mature MBII-52 snoRNA (Figure 7). Recently it was reported that lack of HBII-52 snoRNA genes together with most of the IPW exons did not result in a PWS phenotype in either human individuals or in a mouse model. Therefore, it is less likely that deletion of alternatively spliced Ipw exons A1/A2, B and C are responsible for the phenotype we obtained here, although we cannot completely exclude the possibility that lack of those exons in a long npcRNA can contribute to it. In future experiments, we will address the question, whether expression of MBII-85 snoRNA in a different host gene is sufficient to compensate the observed phenotype.

13Lack of MBII-85 snoRNA expression in PWScrm+/p− compared to PWScrm−/p+ mice, along with the unaltered expression of paternally and maternally expressed genes from the PWS/AS locus in both heterozygous mice, indicates that deletion of PWScr does not affect imprinting of neighboring genes. Moreover, patients with translocations leading to a similar deletion encompassing the genes encoding HBII-85 and HBII-438A snoRNA exhibit normal biparental methylation patterns. Thus, absence of MBII-85 snoRNA is the most likely cause for the phenotype observed in PWScrm+/p− and PWScrm−/p− mice (Figure 6A).

14In Eukarya, most C/D box npcRNAs guide site-specific 2′-O-methylation in rRNAs and small nuclear RNAs (snRNAs) by complementarity to defined sites within these RNA targets. However, most if not all C/D box small npcRNAs that map to the PWS locus, including MBII-85 and MBII-52 snoRNAs, lack significant complementarities to any rRNA or snRNA targets. Although a major role for MBII-52 in the etiology of the PWS can be excluded, cell culture experiments have suggested that MBII-52 snoRNA might play a role in A to I editing and/or alternative splicing of the 5HT-2c serotonin receptor pre-mRNA. In spite of these discrepancies, one might still be tempted to propose that MBII-85 snoRNA interacts (by complementarity) with a yet unidentified RNA target. However, other less orthodox functions for MBII-85 snoRNA and its role in postnatal growth retardation must also be seriously entertained. In any event, MBII-85 is the first example of a C/D box snoRNA gene, whose deletion results in an obvious phenotypic change in a multicellular organism. Future experiments might reveal additional, less obvious defects and deficiencies in PWScrm+/p− mice. Our mouse model will serve as an important tool for further investigations of the molecular pathogenesis of PWS in man.

15For construction of the 5′HPRT/PWScr_targ and 3′HPRT/PWScr_targ targeting vectors, we isolated clones RPCIP711K19517Q6 and RPCIP711J18414Q6, respectively, from the RPCI21 mouse PAC library (RZPD German Resource Center for Genome Investigation) using the MBII-85 oligonucleotide (Table S1). The 5′HPRT/PWScr_targ construct was generated using the 5′ flanking region of the MBII-85 gene cluster as a template for PCR-amplification of 1075 bp and 7387 bp DNA fragments with primer pairs 5′FLAdir/5′FLArev and 5′FLBdir/5′FLBrev, respectively. The PCR fragments were used as homologous arms in the targeting vector. The cassette containing the 5′ portion of the HPRT gene, the loxP site, and the neomycin resistance gene was subcloned from a λ phage vector kindly provided by A. Bradley (Baylor College of Medicine, Houston, USA). The thymidine kinase (TK) gene was placed outside of the homologous arm (Figure 1E).

16The 3′HPRT/PWScr_targ construct was cloned in a similar way. The 3′ flanking region of the MBII-85 gene cluster was used as a template to PCR-amplify 1098 bp and 6720 bp DNA fragments using primer pairs 3′FLBdirN/3′FLBrev and 3′FLAdir/3′FLArev, respectively. The insertion cassette, containing the 3′ portion of the HPRT gene, the loxP site, and the puromycin resistance gene was subcloned from the λ phage vector kindly provided by A. Bradley. The TK gene was placed outside of the homologous arm.

17The 5′ HR probe for Southern blot hybridization was generated by using the PAC DNA containing the 5′ region of the MBII-85 gene cluster in two PCR reactions with oligonucleotide pairs 5′PRAdir/5′PRArev and 5′PRBdir/5′PRBrev, and ligating and cloning the resulting PCR products into the pSL300 vector. The 3′ HR probe was cloned in a similar fashion. PCR products were obtained with oligonucleotide pairs 3′PRAdir/3′PRArev and 3′PRBdir/3′PRBrev (Table S1).

18HPRT-deficient embryonic stem cells AB2.2 (from A. Bradley), passage 17, were expanded in HEPES-buffered, Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum (HyClone), nonessential amino acids, L-glutamine, β-mercaptoethanol, 1000 U/ml recombinant LIF (Chemicon) and antibiotics (penicillin 100 U/ml and streptomycin 100 μg/ml) on a γ-irradiated monolayer of SNL6.7 cells (from A. Bradley) or mouse primary fibroblasts. For electroporation, 2 × 107 ES cells were resuspended in 20 mM HEPES pH 7.4, 173 mM NaCl, 5 mM KCl, 0.7 mM Na2HPO4, 6 mM dextrose, and 0.1 mM β-mercaptoethanol. The NotI linearized replacement targeting vectors 5′HPRT/PWScr_targ, 3′HPRT/PWScr_targ, and intact CRE expressing cassette pOG231 (55 μg DNA of each) were electroporated at 25 μF and 400V (Gene Pulser; Bio-Rad). After electroporation, cells were plated onto 100 mm culture dishes containing a γ-irradiated monolayer of primary, G-418-resistant, or SNL6.7 (G-418, puromycin- and HAT-resistant) fibroblast feeder cells. Thirty-eight hours later, 350 μg/ml G418 (Invitrogen) and 0.2 μM 2′-deoxy-2′-fluoro-β-D-arabinofuranosyl-5-iodouracil (FIAU) (Moravek Biochemicals and Radiochemicals, USA) or 1.0 - 0.5 μg/ml puromycin (Sigma) and 0.2 μM FIAU, or HAT (Sigma) were added to the culture medium. The medium was replaced every day and colonies were picked and analyzed eight days after plating. 5′HPRT loxP-targeted ES cells were analyzed using oligonucleotide pairs 85–5′screen2d/85–5′screen2r and 85–5′screen3d/85–5′screen3r for a nested PCR approach. For analysis of 3′HPRT loxP-targeted cells we employed the 85–3′screen1d/85–3′screen1r and 85–3′screen2d/85–3′screen2r nested PCR primer pair combinations. DNA blot analysis was performed as described. Membranes were hybridized with 32p−labeled 5′HR and 3′HR probes (Figure 2A). Several independent ES clones containing the CRE-mediated PWScr-deletion were injected into 3.5-day-old B6D2F1 (C57BL/6 x DBA) blastocysts, and the resulting embryos were transferred to CD-1 foster mice. Chimeras were identified by their agouti coat color.

19The 5.7 kb DNA fragment containing flanking regions of PWScr together with the inserted HPRT gene was amplified and sequenced using PCR primers MB85seqD1 and MB85seqR1. Sequencing reactions were performed using the BigDye terminator cycle sequencing reaction kit (PE Applied Biosystems) and resolved on an ABI Prism 3100 (Perkin Elmer) capillary sequencing machine.

20To genotype mice we performed PCR analysis of genomic DNA from tail biopsies using the primer pair MB85deld1/MB85delr1 (Figure 2A and 2C; Table S1). PCR cycling was done at 93 °C – 2 min; 7 cycles (93 °C – 40 sec, 70 oC – 20 sec, touch down −1 °C, extension 67 oC – 1 min 40 sec); 35 cycles (93 °C – 40 sec, 55 oC – 20 sec, 67 oC – 1 min 40 sec). The final extension was performed at 67 °C for 5 min.

21Total RNA was isolated from mouse brains using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. RNA samples, 20 μg each, were treated with RNase-free DNase I (Roche). First strand cDNA synthesis was performed using Transcriptor reverse transcriptase (Roche) and hexamer oligonucleotides, followed by PCR amplification with gene specific oligonucleotides (Table S1). The cDNA probes for necdin, Ube3A, Atp10a mRNAs, and Ipw exons were PCR amplified, cloned in the pCRII vector (Invitrogen), and subsequently sequenced using gene specific oligonucleotides (Table S1). Approximately 20 μg of total RNA was denatured, fractionated on 1.2% agarose formaldehyde gels, and transferred to GeneScreen nylon membranes (NEN DuPont). Hybridization was performed with 32P-labeled cDNA probes. Northern blot analysis of snoRNAs was performed with specific oligonucleotides (Table S1) as described.

22A PWScr-deficient mouse line was established by breeding male chimeras nos. 2 and 5 from one mutant ES cell line with female C57BL/6 mice to produce heterozygous mice. Subsequently, heterozygous mice were interbred or bred to C57BL/6 mice. All breeding occurred at the ZMBE animal facility of the University Clinics, Münster in a controlled (21 °C, 30–50% humidity) room with a 12:12 hour light-dark cycle, and mice were housed under non-enriched, standard conditions in individually ventilated (36 (l) x 20 (w) x 20 (h) cm) cages for up to five littermates. Pups were weaned 19 – 23 days after birth and females were kept separately from males.

23Statistical analysis was performed using the StatView software package. Body weight was analyzed using Student's t-test, or ANOVA for each day of postnatal age. Weights of placentas and embryos were analyzed using Mann-Whitney nonparametric statistics. Postnatal lethality was analyzed with the chi-square test.