www.elsevier.com/locate/ygeno

Genomics 85 (20

A nonimprinted Prader–Willi Syndrome (PWS)-region gene regulates a

different chromosomal domain in trans but the imprinted PWS

loci do not alter genome-wide mRNA levels

Mihaela Stefana, Toni Portisb, Richard Longneckerb, Robert D. Nichollsa,*

aDepartment of Psychiatry, and Department of Genetics, Center for Neurobiology and Behavior, University of Pennsylvania, 415 Curie Boulevard,

Philadelphia, PA 19104-6140, USAbDepartment of Microbiology and Immunology, Feinberg School of Medicine, Northwestern University, Ward 6-231, 303 East Chicago Avenue,

Chicago, IL 60611, USA

Received 22 November 2004; accepted 5 February 2005

Abstract

Prader–Willi syndrome (PWS) is a complex neurobehavioral disorder that results from loss of function of 10 clustered, paternally

expressed genes in a 1.5-Mb region of chromosome 15q11–q13. Many of the primary PWS region genes appear to have nuclear RNA

regulatory functions, suggesting that multiple genetic pathways could be secondarily affected in PWS. Using a transgenic mouse model of

PWS (TgPWS) with an ¨4-Mb chromosome 7C deletion of paternal origin that models the neonatal phenotype of the human syndrome we

compared by oligonucleotide microarrays expression levels of ¨12,000 genes and ESTs in TgPWS and wild-type brain. Hybridization data

were processed with two distinct statistical algorithms and revealed a dramatically reduced expression of 4 imprinted genes within the

deletion region in TgPWS mice, with 2 nonimprinted, codeleted genes reduced twofold. However, only 3 genes outside the deletion were

significantly altered in TgPWS mouse brain, with ¨1.5-fold up-regulation of mRNA levels. Remarkably, these genes map to a single

chromosome domain (18B3), and by quantitative RT-PCR we show that 8 genes in this domain are up-regulated in TgPWS brain. These

18B3 genes were up-regulated in an equivalent manner in Angelman syndrome mouse (TgAS) brain, which has the same deletion but of

maternal origin. Therefore, the trans-regulation of the chromosome 18B3 domain is due to decreased expression of a nonimprinted gene

within the TgPWS/AS mouse deletion in mouse chromosome 7C. Most surprisingly, since 48–60% of the genome was screened, it appears

that the imprinted mouse PWS loci do not widely regulate mRNA levels of other genes and may regulate RNA structure.

D 2005 Elsevier Inc. All rights reserved.

Keywords: Disease mechanisms; Gene expression; Microarray; Quantitative mRNA transcription; Transregulation

Prader–Willi syndrome (PWS) is a complex devel-

opmental and multisystem disorder affecting ¨1/15,000

live births [1,2]. Clinically, PWS is characterized by a

two-stage natural history. PWS infants have failure to

thrive with muscle hypotonia, genital hypoplasia, respira-

tory problems, and feeding difficulties due to poor suck

and swallowing reflexes [1,3], but between 2 and 4 years

of age obesity sets in as a consequence of an uncon-

trolled eating behavior [4]. In addition, PWS subjects

0888-7543/$ - see front matter D 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.ygeno.2005.02.004

* Corresponding author. Fax: +1 215 898 0273.

E-mail address: robertn@mail.med.upenn.edu (R.D. Nicholls).

have an abnormal body composition with decreased

muscle mass and an increase in adiposity even prior to

onset of obesity, a lowered metabolic rate, short stature,

and mild to moderate mental retardation with behavioral

disorders [1–5]. To date, most of the hormonal and

metabolic abnormalities reported in PWS patients, inclu-

ding growth hormone (GH) and insulin-like growth factor

(IGF-1) deficiencies, have been assumed to be associated

with hypothalamic dysfunction but this is as yet

unproven.

PWS is caused by absence of expression of a cluster

of paternally expressed genes in chromosome 15q11–

q13, by paternal chromosomal deletion, maternal unipar-

05) 630 – 640

YGENO-07463; No. of pages: 11; 4C:

M. Stefan et al. / Genomics 85 (2005) 630–640 631

ental disomy (UPD), or imprinting defect mechanisms

[6]. De novo 4-Mb deletions account for ¨70% of PWS

cases [6–8], and span the 2-Mb imprinted domain and

several nonimprinted genes. There are 10 known pater-

nally expressed loci potentially involved in PWS etiology

(Fig. 1A). This includes NDN, encoding NECDIN, a

MAGE protein family member proposed to act as a

growth suppressor and antiapoptotic protein in postmitotic

neurons [11–13]. While NECDIN was initially described

as a nuclear protein [14,15] able to bind to specific DNA

sequences [16], recent data suggest a primarily cytoplas-

mic localization [13]. MAGEL2 is a related MAGE

family member of unknown function but with several

potential nuclear localization signals and is highly

expressed in the hypothalamus and other brain regions

[17,18]. A third locus, MKRN3, encodes a putative

ribonucleoprotein [19]. In addition, the complex IC–

SNURF –SNRPN –snoRNA polycistronic locus codes for

several functions including a cis-acting regulatory region

(imprinting center or IC), two nuclear proteins (SNURF

and SmN) [20,21], five classes of box C/D small

nucleolar RNA (snoRNA) [9,22,23], and UBE3A-AS

antisense transcript [23,24]. SmN is a spliceosomal

protein expressed primarily in brain, where it replaces

the otherwise constitutively expressed SmBV/B proteins.

In the absence of SmN in PWS patients and mice, SmBV/B levels increase significantly, which may largely

compensate for splicing functions [25]. SNURF is rich

in arginine residues and hence might bind RNA or have

an unknown function [21]. Each brain-specific snoRNA

in the PWS region lacks the usual rRNA complementar-

ity and may target cellular mRNAs for methylation or

alternative splicing [9,22,23]. Combined, these data

suggest that many of the ‘‘PWS primary genes’’ have

transcriptional or RNA regulatory functions and hence

Fig. 1. Genetic structure of (A) human chromosome 15q11–q13 and (B) the ho

flanked by PWS/AS common deletion breakpoints (BP, zigzag lines) is shown. The

[9]. The TgPWS/AS deletion (del) in mouse 7C [10,26] is equivalent to class I del

are circles, genes; black, paternal; gray, maternal; open, biparental; IC, imprinting c

of genes; *, mouse PWS/AS-region genes present on the MG-U74Av2 array; bar

that whole pathways of secondary and perhaps farther

downstream genes could be defective in PWS [5].

Human chromosome 15q11–q13 is homologous to the

central part of mouse chromosome 7, with conservation of

most gene content and imprinted features [6] (Fig. 1B).

Since mouse 7C imprinted loci lack a homolog to the

human HBII-438 snoRNA but have gained the retrotrans-

posed Frat3 gene [26], these are unlikely to play a

significant role in PWS. To date, targeted deletion studies

have not linked any individual mouse PWS-region gene to

any major clinical feature of PWS [27]. For example, Ndn-

deficient mice have a respiratory deficit with partial

neonatal lethality [28–30], mice harboring intragenic

deletions in Snurf –Snrpn were phenotypically normal

[31,32] although behavioral studies have not been reported,

while paternal deletion from Snurf to Ube3a results in

growth retardation and ¨80% lethality before weaning

[32]. In contrast, PWS mouse models with UPD, an

imprinting defect, or a paternally derived chromosomal

deletion share a very similar phenotype including hypo-

tonia, respiration difficulties, and failure to thrive, with

early postnatal lethality, modeling the first stage of the

human syndrome [6,10,31].

Given the neurobehavioral phenotype of PWS and the

putative regulatory features of many of the PWS-region

genes that are predominantly expressed in brain [33,34],

we have examined brain gene expression profiles in

transgenic deletion PWS (TgPWS) vs wild-type (WT)

mice at postnatal day 1 (P1) to identify candidates for

downstream gene pathways. Imprinted and nonimprinted

genes within the TgPWS mouse deletion were reduced in

TgPWS mouse brain, as expected, whereas there was

minimal change in global gene expression. Intriguingly,

expression of a cluster of eight genes outside the deletion

from chromosome 18B3 was significantly up-regulated in

mologous mouse chromosome 7C. The human 4-Mb 15q11–q13 region

HBII-85 and HBII-52 snoRNAs are repeated 24 and 47 times, respectively

etions in PWS and AS that occur in ¨40% of deletion patients [8]. Symbols

enter; cen, centromere; tel, telomere; line arrows, transcriptional orientation

at bottom, LMP2A transgene (Tg)-induced PWS/AS del mouse model [10].

M. Stefan et al. / Genomics 85 (2005) 630–640632

TgPWS brain as well as in mice with maternal inheritance

of the same deletion (an Angelman syndrome, or AS,

mouse model). These results suggest coordinate regulation

of expression of the 18B3 domain in trans by a non-

imprinted PWS/AS-region gene product.

Results

Gene expression in TgPWS vs WT brain at P1

We examined total RNA expression levels in the TgPWS

deletion vs WT mouse brain at P1 in two groups consisting

of five animals each, using high-density oligonucleotide

arrays with probes for ¨12,000 genes and ESTs. P1 was

chosen because at this age TgPWS mice do not display the

physiological changes of severe failure to thrive that

Fig. 2. Microarray analysis of brain gene expression in TgPWS vs WT mice. Gen

the MG-U74Av2 array with ¨12,000 genes/ESTs. (A) To evaluate expression prof

was performed in MAS 5.0 and generated a total of 16 transcripts as potentially der

log ratio (TSD). (B) A two-class unpaired analysis (SAM) yielded 20 transcripts

Both MAS 5.0 and SAM identified as the most significant (*) decreases in TgPWS

appears twice on the array, see Table 1). Other significant decreases are nonimprinte

were identified by both statistical methods as significantly changed in TgPWS, with

(Tcerg1), leucyl-tRNA synthetase (Lars), and protein phosphatase 2A, regulatory

commence at or after P2 (M.S., H. Ji, R.A. Simmons, D.E.

Cummings, K.H. Kaestner, R.S. Ahima, M.I. Friedman,

R.D.N., manuscript in preparation), thereby avoiding detec-

tion of changes secondary to failure to thrive. To detect

significant changes in gene expression between the two

groups, hybridization data were analyzed using two software

packages. The first comparison analysis was carried out

using Affymetrix MicroArray Suite 5.0 (MAS 5.0). After

applying the defining filtering criteria and establishing a

consistency of change in the same direction for at least three

of five replicate pair experiments, we identified 16 tran-

scripts as potentially deregulated in TgPWS brain with a fold

�1.4 (Fig. 2A). The second comparison was performed

using Significance Analysis of Microarrays (SAM) and

generated 20 potentially abnormally expressed genes in

TgPWS vs WT mice with a median false discovery rate

(FDR) of �45% (D = 0.3) (Fig. 2B). As expected, combined

e expression was compared for five TgPWS and five WT males at P1 using

iles from Gene Chip arrays for TgPWS and WT mice, a comparison analysis

egulated in TgPWS vs WT mice. Values are presented as the means of signal

with potentially abnormal expression in TgPWS vs WT brain (FDR � 45).

brain the PWS-region imprinted genes within the deletion (black bars; Ndn

d genes within the TgPWS deletion (open bars). Only three other transcripts

¨1.5-fold increases (striped light gray bars): transcription elongation factor

subunit B, h isoform (PP2ABb).

Fig. 3. Representation of mRNA expression levels of Cyfip1 in TgPWS and

TgAS vs WT mice at P1 and P4. Relative quantitation by QRT-PCR of

Cyfip1 mRNA showed a twofold decreased expression in TgPWS and

TgAS vs WT mice at all tested postnatal days. All values are the means

from five (P1) or four animals (P4) T SD. *p � 0.05; **p � 0.001; ***p �0.0001, significant differences between WT and TgPWS or TgAS at the

indicated time points (independent samples t test).

M. Stefan et al. / Genomics 85 (2005) 630–640 633

analysis of the two datasets identified the PWS-region

imprinted genes (Mkrn3, Magel2, Ndn, Snurf–Snrpn) as

the most significant decreases in gene expression in TgPWS

mice (Figs. 2A and 2B; Table 1). In addition, two non-

imprinted genes (Herc2, Cyfip1) within the deletion dem-

onstrate an ¨2-fold decrease in expression in TgPWS pups

(Figs. 2A and 2B; Table 1), which is in agreement with a

prediction of 50% reduction for these genes in TgPWS brain.

These six genes constitute built-in ‘‘controls’’ that give great

confidence in the microarray data, including the reproduci-

bility (all five TgPWS vs all five WT mice) and sensitivity

both for genes that have a low level of expression (e.g.,

Mkrn3) and for those with a small change in expression level

between WT and TgPWS mice (e.g., Herc2).

To confirm the reliability and sensitivity of detecting

subtle changes (less than or equal to twofold) in gene

expression in TgPWS mice, mRNA levels of Cyfip1, a

nonimprinted gene within the TgPWS/AS deletion, were

quantified by quantitative real-time PCR (QRT-PCR) in

brain tissue at P1 and P4. Consistent with the microarray

and the expectation for a deleted gene, the average mRNA

expression was decreased by twofold for Cyfip1 in TgPWS

vs WT mice and equivalent results were found for TgAS

vs WT mice (Fig. 3). Furthermore, since Cyfip1 transcript

levels were normalized to Gapdh levels, these data

demonstrate that the levels of Gapdh are not changed by

parameters of phenotype or age in this analysis, justifying

the use of this control gene in the following studies.

Despite the sensitivity of the microarray experiments,

only 3 genes not in the TgPWS deletion were identified

by both statistical methods with a consistent difference in

expression between the TgPWS and the WT groups:

Tcerg1, PP2ABb, and Lars (Figs. 2A and 2B; Table 1).

All 3 genes exhibit up-regulation in TgPWS vs WT brain

Table 1

Changes in brain gene expression in TgPWS vs WT mice at P1 by

microarray identified by comparison analysis in both MAS 5.0 and SAM

GenBank Gene MAS (5.0) SAM

D76440 Ndn –274.3 p < 0.0001 0.008 q = 20%

AJ00236a Ndn –137.1 p < 0.0001 0.005 q = 20%

X60388 Snurf–Snrpn –26.9 p < 0.0001 0.02 q = 20%

AJ243608 Magel2 –26 p < 0.0001 0.03 q = 20%

U19106 Mkrn3 (Zfp127) –10.4 p < 0.0001 0.06 q = 20%

AF072697 Cyfip1 (Shyc) –1.8 p < 0.0001 0.5 q = 36.3%

AF061529 Herc2 –1.7 p < 0.0001 0.6 q = 36.3%

B023485 Tcerg1 (CA150b) 1.5 p = 0.0002 1.5 q = 36.3%

AI844089 Lars (LeuRS) 1.4 p < 0.0001 1.5 q = 36.3%

AW048155 PP2ABb (PR55B/b) 1.4 p = 0.001 1.5 q = 36.3%

For pair-wise comparison analysis (MAS 5.0), all microarray values are

presented as the ratio of mean expression fold changes for TgPWS vs WT.

The p values were computed for expression levels (Signal) of individual

genes in TgPWS vs WT using a t test for independent samples. For two-

class unpaired analysis (SAM), all microarray results are presented as fold

change for TgPWS vs WT. The q values (%) represent the lowest FDR at

which the gene is called significant.a TFIIH EST but coligated with Ndn.

with the same fold change, ¨1.5 (Table 1). Moreover,

these genes show the maximum consistency of change (5/

5 experiments for PP2ABb and Lars and 4/5 experiments

for Tcerg1) by MAS 5.0. Microarray results were

confirmed by QRT-PCR at P1 for all 3 genes (Table 2).

Intriguingly, Tcerg1, Lars, and PP2ABb are clustered

within a 700-kb domain in mouse chromosome 18B3 and,

based on Fisher_s exact test, the probability of colocaliza-

tion of 3 of 12,000 genes spotted on MG-U74Av2 (with

23 genes within chromosome 18B) is 1.79 � 10�7. This

observation suggests that these 3 genes are regulated as a

domain.

Identification of eight genes in chromosome 18B3 that are

deregulated in TgPWS mice

Protein phosphatase 2A (PP2A) comprises a family of

serine/threonine phosphatases composed of catalytic (C),

structural (A), and regulatory subunits (B) [35]. There are

a large number of regulatory B subunits that independ-

ently associate with the constitutive core of the enzyme

consisting of subunits A and C. The B family of subunits

(PR55) is encoded by four genes: Ba, Bh, Bg, and By,with two major splice variants for PP2ABb generated by

use of alternate promoters, Bb.1 and Bb.2 [36]. Since our

studies above examined the 3V end of PP2ABb, we

performed QRT-PCR to distinguish the Bb.1 and Bb.2mRNA isoforms. This demonstrated that the Bb.1 isoform

was up-regulated 1.5-fold in PWS brain at P1 (Fig. 4A;

Table 2) and 1.5-fold at P4 (Table 2). In contrast, Bb.2expression remains unchanged between TgPWS and WT

mice (Fig. 4A; Table 2).

Five other genes were identified in the chromosome

18B3 cluster based on analysis of genomic sequence (Fig.

4B) and were not present on the MG-U74Av2 array:

Pou4f3, encoding a member of the Brn-3 family of POU

transcription factors [37]; Rbm27, encoding an RNA

binding polypeptide; Gpr151, encoding a member of the

Table 2

Chromosome 18B3 gene expression in TgPWS vs WT mouse brain at E18, P1, and P4

Genea QRT-PCR (E18)b QRT-PCR (P1)c QRT-PCR (P4)b

WT TgPWS p TgPWS/

WT ratio

WT TgPWS p TgPWS/

WT ratio

WT TgPWS p TgPWS/

WT ratio

PP2ABbd 1.08 (T0.09) 1.48 (T0.28) 0.03 1.4 1.08 (T0.09) 1.48 (T0.29) 0.04 1.4 1.17 (T0.15) 1.64 (T0.12) 0.003 1.4

PP2ABb.1 0.93 (T0.13) 1.14 (T0.10) 0.04 1.2 1.15 (T0.16) 1.69 (T0.18) 0.001 1.5 1.06 (T0.05) 1.55 (T0.09) <0.0001 1.5

PP2ABb.2 0.94 (T0.07) 1.06 (T0.13) 0.1 1.1 1.28 (T0.33) 1.03 (T0.17) 0.2 0.8 0.94 (T0.12) 0.82 (T0.11) 0.2 1

Tcerg1 1.23 (T0.20) 1.65 (T0.31) 0.07 1.3 1.11 (T0.07) 1.56 (T0.29) 0.01 1.4 0.90 (T0.08) 1.24 (T0.10) 0.003 1.4

Lars 1.16 (T0.15) 1.44 (T0.32) 0.2 1.2 1.27 (T0.46) 1.97 (T0.28) 0.02 1.6 0.93 (T0.14) 1.23 (T0.16) 0.03 1.3

Pou4f3 1.62 (T0.56) 2.90 (T1.03) 0.008 1.8e 1.73 (T0.95) 1.85 (T0.91) 0.8 1e 1.00 (T0.16) 1.42 (T0.30) 0.003 1.4e

Rbm27 1.17 (T0.13) 1.76 (T0.16) 0.001 1.5 1.04 (T0.11) 1.55 (T0.15) 0.0004 1.5 0.99 (T0.14) 2.17 (T0.55) 0.006 2.1

Gpr151 1.06 (T0.31) 1.08 (T0.32) 0.9 1 1.02 (T0.21) 1.11 (T0.14) 0.3 1e 0.93 (T0.13) 1.53 (T0.23) 0.004 1.6

C330008R14Rik 1.00 (T0.22) 1.50 (T0.43) 0.01 1.5e 1.87 (T0.82) 2.21 (T0.94) 0.4 1.2e 1 (T0.24) 1.7 (T0.23) 0.006 1.7

Sh3rf2 0.82 (T0.21) 1.16 (T0.08) 0.02 1.4 1.00 (T0.40) 1.41 (T0.24) 0.01 1.4e 1.09 (T0.20) 1.25 (T0.18) 0.06 1.1e

For each developmental stage (E18, embryonic day 18; P1, P4, postnatal days 1 and 4), QRT-PCR data are presented as the means of expression fold changes

(TSD) (first two columns), with two-tailed p values (third column), and the ratio of mean fold changes for TgPWS vs WT (fourth column).a GenBank accession numbers are in Table 1, also AK07905 (PP2ABb.1), AK013600 (PP2ABb.2), S69352 (Pou4f3), XM-128924 (Rbm27), NM_181543

(Gpr151), AK021193 (C330008R14Rik), and NM_172966 (Sh3rf2).b TgPWS, n = 4; WT, n = 4.c TgPWS, n = 5; WT, n = 5.d Probe and primers are in the 3Vend of the gene and do not differentiate between the b.1 and the b.2 isoform.e QRT-PCR results are pooled from two repeated reactions.

M. Stefan et al. / Genomics 85 (2005) 630–640634

G-protein-coupled receptor family expressed in the central

nervous system, predominantly in the trabecular complex

[38,39]; C330008R14Rik, encoding a putative protein

with a PRELI/MSF1 domain; and Sh3rf2, encoding a

polypeptide with SH3 domains and a RING Zn-finger

domain. QRT-PCR analysis revealed no change in

expression levels of Pou4f3, Gpr151 , and C33-

0008R14Rik in TgPWS mouse brain at P1, but an

¨1.5-fold significantly increased expression of Rbm27

and Sh3rf2 (Table 2). To investigate further the expres-

sion changes for chromosome 18B3 candidate genes

during the failure-to-thrive period and in embryonic life

of the TgPWS mouse, mRNA expression of these genes

was determined by QRT-PCR at P4 and E18, respectively.

For the most part, the few genes showing discordant data

at any one age (e.g., Pou4f3 at P1, Sh3rf2 at P4) show

wide standard deviations in expression levels within

groups, possibly due to low levels of expression

(Supplemental Data Fig. S1). At E18, five genes exhibit

significant expression variance between TgPWS and WT

mice (PP2ABb.1, Pou4f3, Rbm27, Sh3rf2, C330008R14-

Rik), at P1 the number of deregulated genes is also 5

(PP2ABb.1, Tcerg1, Rbm27, Lars, Sh3rf2), and by P4 all

genes within this domain except for Sh3rf2 (and the

PP2ABb.2 isoform) were up-regulated in TgPWS vs WT

mice (Table 2; Fig. 4B).

A nonimprinted PWS/AS region gene is responsible for

deregulation of the 18B3 domain

The data to date suggested that a gene associated with

the LMP2A transgene insertion and deletion in TgPWS

mice [10] regulates in trans a chromosome 18B3 domain

of eight genes, but does not distinguish between an

imprinted (paternally expressed) and a nonimprinted

PWS-region gene as being causative nor whether this

could be an effect of the LMP2A transgene insertion. To

distinguish between the first two possibilities, mRNA

levels for genes within the 18B3 region were assessed by

QRT-PCR in brain tissue of TgAS vs WT mice at P1 and

P4 (Table 3). At P1, with the exception of Lars, the same

genes displaying up-regulation in TgPWS brain show

between 1.4- and 1.8-fold significantly increased mRNA

levels in TgAS vs WT brain. At P4, the number of

deregulated genes within the 18B3 region increases to 5

following a similar pattern as in TgPWS brain (Table 3,

Fig. 4B). These data rule out an imprinted gene in the

chromosome 18B3 effect in these mice.

To distinguish a role for a nonimprinted, deleted gene or

the LMP2A transgene, expression of three genes within the

18B3 domain that showed increased mRNA levels in

TgPWS and TgAS brain was assessed by QRT-PCR in

liver tissue from EALMP2A vs WT mice at P1. Liver was

chosen as control since EALMP2A mice do not have

detectable expression of the transgene in brain, but do have

significant LMP2A expression in B cells of neonatal liver.

All three genes (Tcerg1, Rbm27, Sh3rf2; Supplemental

Data Table S1) are significantly up-regulated in TgPWS

liver and one (Rbm27) in TgAS (the other two genes have a

large SD in expression levels, possibly due to low level of

expression; Supplemental Data Fig. S1). In contrast, no

significant expression variance for Tcerg1, Rbm27, and

Sh3rf2 was observed for EALMP2A and WT liver (Supple-

mental Data Table S1). Similarly, expression of PP2ABb,Tcerg1, Lars, Rbm27, and Sh3rf2 in P1 brain does not differ

between EALMP2A and WT mice (data not shown).

Therefore, the transgene does not affect chromosome

18B3 expression.

Fig. 4. Abnormal expression of chromosome 18 domain in TgPWS and TgAS mice. (A) Representation of mRNA expression levels of PP2ABb.1 and

PP2ABb.2 in TgPWS vs WT mice. Relative quantitation of expression of Bb.1 and Bb.2 isoforms was performed in two groups of five TgPWS males (open

bars) and five WT males (black bars) at P1 and in four TgPWS and four WT mice at E18 and P4 by QRT-PCR. The average mRNA expression for Bb.1 was

increased by 1.2-fold at E18 (*p � 0.001), 1.5-fold at P1 (**p � 0.001), and 1.5-fold at P4 (***p � 0.0001) in TgPWS vs WT, whereas no change in the

average expression of Bb.2 was detected. (B) Model for trans-regulation of a chromosome 18B3 domain by a nonimprinted mouse TgPWS/AS locus. A

nonimprinted gene present in the TgPWS/AS deletion region is hypothesized to encode a factor that activates specific regulatory DNA elements that inhibit

transcription at various developmental time points for C330008R14Rik, Sh3rf2, Lars, Rbm27, Pou4f3, Tcerg1, Gpr151, and PP2ABb.1, but not PP2ABb.2.Alternatively, the PWS/AS nonimprinted gene may control a gene pathway leading to a factor that regulates the 18B3 domain. In TgPWS and TgAS mice,

haploinsufficiency of the nonimprinted gene results in increased expression of the chromosome 18B3 target genes. For the eight genes in this domain, the first

and last exons (black boxes) and transcriptional orientation (horizontal arrows) are shown. Symbols used: +, activation; j or – , increase or no change in

expression level, respectively (black j or – , TgPWS; gray j or – , TgAS).

M. Stefan et al. / Genomics 85 (2005) 630–640 635

Candidate gene assessment in TgPWS brain

Genes that could be considered a priori as candidates

for expression changes in TgPWS mice were individually

inspected in the microarray data set, if present on the chip,

and either exhibited no change in TgPWS vs WT or were

not expressed at detectable levels (Supplemental Data

Table S2). This analysis identified a few notable examples,

such as Ube3a, which maps within the TgPWS/AS deletion

but with maternal-only expression in brain and, as expected,

does not change in TgPWS vs WT (mRNA signal levels:

WT 96, PWS 84). Another candidate, Snrpb, was identified

as unchanged by microarray (WT 1718, PWS 1780),

suggesting that the up-regulation in protein levels observed

in PWS patient and TgPWS mouse model brain [25] occurs

posttranscriptionally. Other candidates, including Gh and

Igf-1, had unchanged mRNA levels in TgPWS vs WT brain,

as did genes implicated in food intake and energy balance

regulation, such as agouti-related protein (Agrp) and

neuropeptide Y (Npy). Secretogranin II (SCG2), a gene

that exhibits strong down-regulation in clonal fibroblast cell

lines from a PWS patient who is mosaic for normal cells and

cells with maternal uniparental disomy 15 [40], showed no

difference in brain mRNA levels in the TgPWS mouse

model. Therefore, SCG2 is unlikely to be relevant to the

PWS phenotype.

Discussion

Minimal gene expression changes in brain of the TgPWS

mouse model

In this study we used oligonucleotide microarrays and

QRT-PCR to ascertain the primary transcriptional changes

in the brain of a PWS mouse model, prior to onset of the

severe neonatal failure-to-thrive phenotype. Our microarray

screen reliably identified 4 imprinted and 2 nonimprinted

Table 3

Chromosome 18B3 gene expression in TgAS vs WT mouse brain at P1 and P4

Genea QRT-PCR (P1)b QRT-PCR (P4)c

WT TgAS p TgAS/WT ratio WT TgAS p TgAS/WT ratio

PP2ABbd 1.04 (T0.03) 1.84 (T0.28) 0.001 1.8 1.13 (T0.12) 1.85 (T0.17) <0.0001 1.6

PP2ABb.1 0.95 (T0.05) 1.39 (T0.17) 0.003 1.4 1 (T0.17) 1.5 (T0.16) 0.001 1.5

PP2ABb.2 1.08 (T0.10) 1 (T0.12) 0.3 1 0.91 (T0.10) 0.88 (T0.11) 0.6 0.9

Tcerg1 1.12 (T0.19) 1.70 (T0.22) <0.0001 1.5e 1.02 (T0.14) 1.24 (T0.17) 0.008 1.2e

Lars 1.13 (T0.34) 1.39 (T0.46) 0.2 1.2e 1.04 (T0.57) 1.36 (T0.51) 0.4 1.3

Pou4f3 0.79 (T0.21) 1.03 (T0.27) 0.2 1.3 1.11 (T0.32) 1.50 (T0.46) 0.1 1.4

Rbm27 1.03 (T0.12) 1.51 (T0.10) 0.001 1.5 1.06 (T0.13) 1.41 (T0.19) 0.01 1.4

Gpr151 1.10 (T0.15) 1.14 (T0.22) 0.8 1 1.28 (T0.32) 2.16 (T0.25) 0.001 1.6

C330008R14Rik 1 (T0.2) 1.3 (T0.18) 0.1 1.3 0.86 (T0.15) 1.09 (T0.21) 0.09 1.2

Sh3rf2 1.3 (T0.27) 2.1 (T0.57) 0.03 1.6 0.8 (T0.14) 1.1 (T0.23) 0.03 1.3

For each developmental stage (P1, P4, postnatal day 1 and 4), QRT-PCR data are presented as the means of expression fold changes (TSD) (first two columns),

with two-tailed p values (third column), and the ratio of mean fold changes for TgAS vs WT (fourth column).a GenBank accession numbers are in Table 2.b TgAS, n = 4; WT, n = 4.c TgAS, n = 5; WT, n = 5.d Probe and primers are in the 3Vend of the gene and do not differentiate between the b.1 and the b.2 isoform.e QRT-PCR results are pooled from two or three repeated reactions.

M. Stefan et al. / Genomics 85 (2005) 630–640636

genes from within the ¨4-Mb deletion as reduced in

TgPWS mouse brain, illustrating the sensitivity and

reproducibility of this approach. Combined with other work

defining the deletion endpoints ([10,26]; M.S., K. Claiborn,

J.M. Greally, R.D.N, unpublished data), the TgPWS/AS

mouse model has a deletion of 13 known imprinted and 10

nonimprinted genes (Fig. 1B) that is equivalent to class I

deletions in PWS and AS [6–8]. An additional set of

clustered genes from a different chromosome was identified

by microarray analysis as deregulated in TgPWS mouse

brain with ¨1.5-fold increased expression and is discussed

below. Although additional candidate genes may be

identified when whole genome mRNA arrays become

available, we screened between 48 and 60% of the

expressed mouse genome [41]. Therefore, despite the

neonatal lethality of the TgPWS mouse model, we conclude

that very few genes have significantly altered brain

expression levels as a primary genetic change in TgPWS

mice.

That few genes have significant changes in expression

levels in the newborn TgPWS mouse brain was unexpected

and may have several alternative explanations. Due to the

small size of the hypothalamus, key genes might not be

detectable by microarray analysis of whole brain, even if

expression is altered. Some but not all hypothalamic-

specific genes were detectably expressed (Supplemental

Data Table S2). Nevertheless, the role of hypothalamus in

PWS is unknown and remains hypothetical, particularly as

the mouse PWS-region genes are expressed more widely in

the brain [6,33,34]. In addition, developmentally determined

gene expression changes may be more readily detected in

older mice. However, since physiological and metabolic

changes associated with failure to thrive become evident in

the TgPWS mouse beginning at P2, many mRNA changes

in older TgPWS pups will be secondary to glucose

homeostasis and/or hepatic energy metabolism impairments.

We previously observed alterations in the orexigenic Agrp

and anorexigenic Pomc mRNA levels in the hypothalamic

arcuate nucleus in P3 TgPWS mice using in situ hybrid-

ization [42], but in the current study Pomc levels were

below detection while Agrp had no change in P1 mice,

suggesting the change at P3 of the latter may be secondary

to failure to thrive in PWS. Another possible basis for

finding few genes with altered expression in TgPWS mice is

that the brain may be a secondary mediator of primary

changes in peripheral organs, despite the predominant

neuronal expression of PWS-region imprinted genes

[6,33,34].

Alternatively, the PWS-region genes may encode factors

that regulate mRNA modification or structure rather than

transcriptional levels. Indeed, SmN (encoded by Snrpn)

replaces SmBV/B in the core spliceosome in neurons and

may play a role in neuronal-specific mRNA alternative

splicing patterns [6,20,21]. Moreover, the C/D box snoR-

NAs in the PWS region may have roles in cellular RNA

modification (e.g., site-specific 2V-O-ribose methylation) or

alternative splicing [9,22,23]. Changes in modification and/

or splicing of target mRNAs may change transcript levels

only minimally or not at all. Nevertheless, alterations in

mRNA modification or splicing could subsequently lead to

significant changes in protein structure and/or expression.

Therefore, proteomics approaches in TgPWS mice will

complement future studies of splicing patterns both globally

and for candidate genes to provide a more comprehensive

picture of the molecular changes underlying the TgPWS

mouse neonatal phenotype.

Minimal gene expression changes in congenital and

acquired diseases

Similar to our study on a PWS mouse model,

surprisingly subtle transcriptional changes have also been

M. Stefan et al. / Genomics 85 (2005) 630–640 637

identified by microarray in mouse models for other

developmental and neurological syndromes. For example,

Mecp2 mutant mice, which exhibit phenotypic similarities

to Rett syndrome, display very few genes with signifi-

cantly changed brain expression despite the expected role

of MECP2 in modulating methylcytosine silencing of gene

expression [43]. Although genes with 1.5-fold increased

expression have been reliably identified from the trisomic

region of a Down syndrome (DS) mouse model, support-

ing a role for gene dosage in DS, no changes in gene

expression levels from other regions of this or other

chromosomes were identified [44]. Similarly, significant

expression changes for oxidative phosphorylation genes in

diabetic and prediabetic muscle are altered by only 0.2- to

0.5-fold [45–47]. Along with our data on TgPWS mice,

these findings imply significant relevance of minimal gene

expression changes for the molecular pathogenesis of

several congenital syndromes and acquired diseases.

Therefore, we suggest that major changes in expression

of genetic pathways in vivo may be limited to severe

pathological states such as cancer or embryonic lethal

conditions.

A PWS/AS-region nonimprinted gene trans-regulates a

cluster of chromosome 18B3 genes

Another intriguing finding from our experiments was

identification of a cluster of genes from chromosome

18B3 with coordinate abnormal expression in the TgPWS/

AS mouse model. The next genes centromeric (Sb140, 2

Mb) and telomeric (DpysI3, Eif3alpha, ¨0.8 Mb) to this

domain are unchanged in TgPWS by microarray (data not

shown). Indeed, a boundary to the coregulated 18B3 gene

cluster likely exists within the PP2ABb locus, since there

is up-regulation of the promoter (Bb.1) closest to the

cluster of genes that are also up-regulated, whereas the

considerably more distal promoter (Bb.2) is not affected in

either TgPWS or TgAS mice (see Fig. 4). Our data also

suggest a temporal regulation for genes in the 18B3

domain, since fewer genes have abnormal regulation in

TgPWS brain at E18 and at P1 than at P4 (Fig. 4B). A

largely similar expression pattern of the chromosome

18B3 genes was identified in the TgAS mouse model at

P1 and P4, excluding a role of a PWS imprinted locus in

the regulation of this domain. We also ruled out an effect

of the LMP2A transgene with an estimated ¨80 copies in

TgPWS [10] on chromosome 18B3 gene expression by

analysis of control EALMP2A transgenic mice that have

B-cell-lineage expression of LMP2A. Therefore, we

conclude that reduced expression of a PWS/AS-region

nonimprinted gene is responsible for the coordinated

deregulation of an eight-gene expression domain in

chromosome 18B3.

The finding of up-regulation of a cluster of genes from

chromosome 18B3 in both TgPWS and TgAS mouse brain

raises the question of whether deregulation of the

homologous genes from human chromosome 5q31 may

play any role in PWS and AS in the human. Identifying the

specific nonimprinted PWS/AS-region gene involved in the

trans-effect (see below) will determine whether this is

likely to be a conserved mechanism, and future studies of

gene expression on autopsy material from PWS and AS

patients can address whether equivalent molecular changes

do occur in human. Since PWS and AS deletion patients

share few overlapping phenotypic characteristics, other

than hypopigmentation associated with hemizygous dele-

tion of at least the OCA2 gene [6], it appears likely that

up-regulation of the 5q31 genes does not play a major

phenotypic role in PWS and AS. Nevertheless, the greater

neurobehavioral severity of AS compared to PWS may

mask otherwise shared clinical characteristics. Therefore,

clinical and molecular studies of individuals with 5q31

duplications and increased dosage of the same genes as

potentially up-regulated in PWS and AS may prove

revealing.

Based on the observed gene expression patterns, we

suggest a model by which coordinate regulation of the 18B3

domain involves a chromatin protein or RNA encoded by a

PWS/AS-region nonimprinted gene, which binds directly a

cis-silencing element in 18B3 and has domain-wide silenc-

ing effects (Fig. 4B). This genetic control mechanism would

be disrupted in TgPWS and TgAS deletion mice, with up-

regulation of the target genes in chromosome 18B3.

Alternatively, the PWS/AS-region nonimprinted gene may

encode a molecule that impacts a silencing mechanism

indirectly. As noted above, there are currently 10 known

nonimprinted genes within the TgPWS/AS deletion that are

candidates for the putative silencing factor. An additional

possibility is reduced level of nonimprinted Ube3a or

Atp10c expression in TgPWS and TgAS mouse brain,

although this was not detected by microarray for Ube3a.

Several of the nonimprinted loci encode receptors, trans-

porters, or channels (Gabrb3, Gabra5, Gabrg3, p/Oca2,

Nipa1, Nipa2, Atp10c, Chrna7) ([6,48,49]; M.S., K. Clai-

born, J.M. Greally, R.D.N, unpublished data) and appear

unlikely to be involved. Gcp5 encodes a component of the

g-tubulin complex required for microtubule nucleation at

the centrosome [50], while Herc2 putatively is involved in

protein trafficking and degradation [51,52], roles that can

likely be ruled out as direct, but indirect roles remain

possible. Cyfip1 encodes a protein that interacts with the

small GTPase Rac1, implicated in neuronal development

and maintenance [53], and also interacts with the fragile X

mental retardation 1 (FMR1) protein [54]. FMR1 is

implicated in negative regulation of translation, possibly

by functioning in RNAi- or miRNA-related pathways [55].

Ube3a encodes an E3 ubiquitin ligase involved in protein

degradation [6,56] and which can impact gene expression

through p53-, steroid hormone receptor-, and NFX-related

mechanisms [6,56]. It is also possible that additional

nonimprinted candidate genes in the PWS/AS domain await

discovery. Further studies making use of mouse models with

M. Stefan et al. / Genomics 85 (2005) 630–640638

specific gene mutations for candidate genes ([51,52]; L. Li,

S. Mertzer, R.D.N, D. Carpenter, E.M. Rinchik, D.K.

Johnson, Y. You, unpublished data) and/or mice that carry

radiation-induced chromosomal deletions spanning the p

locus [57,58] will likely identify the nonimprinted gene

responsible for the chromosome 18B3 trans effect. Addi-

tional studies of temporal changes in chromatin structure in

the 18B3 domain in TgPWS or TgAS mice will shed light

on the molecular mechanisms of coordinate trans-regulation

by a gene in the PWS/AS domain.

Materials and methods

Animals

The PWS/AS deletion mouse model has an LMP2A-

transgene-induced ¨4-Mb deletion (Fig. 1B) [10]. Maternal

transmission of the transgene leads to the TgAS mouse

model, with a viable, fertile phenotype and presumably with

similar mild neurobehavioral features compared to AS

knockout mouse models [6], as well as a late-onset obesity

(unpublished data). In contrast, transgene transmission

through males results in the TgPWS mouse phenotype.

Construction and characterization of additional LMP2A

transgenic mice with B-cell-lineage expression of LMP2A

(EALMP2A) have been described previously [59]. Brain

tissues from five TgPWS and five WT male CD-1 sibs at P1

were used for microarray studies. Brain and liver tissues

from at least four TgPWS, TgAS, and EALMP2A and

littermate WT mice were used for QRT-PCR experiments.

Brain gene expression was determined by QRT-PCR in

TgPWS vs WT mice at E18, P1, and P4, whereas for TgAS

vs WT and EALMP2A vs WT mice QRT-PCR was

performed at P1 and P4. Liver mRNA levels were evaluated

by QRT-PCR at P1 and P4 in all three mouse models. All

animals were bred and genotyped as described [10,59]. The

University of Pennsylvania and Northwestern Institutional

Animal Care and Use Committees approved all animal

experiments.

Microarray analysis

Total RNA was extracted using TRIzol (Invitrogen,

Carlsbad, CA, USA), treated with RQ1 RNase-free DNase

(Promega, Madison, WI, USA), and checked for integrity

by gel electrophoresis. Microarray hybridization and

staining were performed at the Penn Microarray Facility

using the Affymetrix Murine Genome U74Av2 array. This

array represents all sequences (¨6000) in the Mouse

UniGene database (build 74) that were functionally

characterized, with an additional ¨6000 EST clusters

also represented. Target preparation, hybridization, and

initial data analysis were carried out as described (http://

www.affymetrix.com/index.affx: GeneChip Analysis Tech-

nical Manual). To evaluate expression profiles from Gene

Chip arrays for TgPWS and WT mice, a comparison

analysis was performed in MAS 5.0 (Affymetrix, Santa

Clara, CA, USA) and Excel (Microsoft, Redmond, WA,

USA). Five biological replicates of the pair-wise compar-

ison were done using 10 microarrays, 1 for each of the P1

animals described above. The results were organized by

filtering out genes called No Change or Absent on both

Gene Chips, sorted by Change p value and Signal Log

Ratio, and then converted to Fold Change. An average fold

change derived from all pair-wise comparisons of �1.4 was

used as the cut-off for potential significant differences in

gene expression and only genes that showed a consistent

change in the same direction for at least three of five

replicate pair experiments were further considered.

Normalized microarray hybridization data were further

imported into and analyzed by SAM (http://www-stat-class.

stanford.edu/SAM/servlet/SAMServlet) to identify genes

with highly reproducible expression changes. SAM assigns

a score to each gene in a gene expression profile based on

its change in expression relative to the standard deviation

of repeated measurements for that gene [60]. A gene is

considered significantly changed if it exceeds an arbitrary

adjusted threshold (D). SAM calculates a FDR, which

represents the median percentage of genes that are

incorrectly identified as significant [60]. Two-class

unpaired analysis was applied to compare gene expression

profiles of five TgPWS and five WT samples. For D = 0.3,

the permutated dataset generated 20 significant genes with

FDR � 45%.

QRT-PCR

Total RNA was isolated as above and cDNA was

synthesized using random hexamers with the Superscript

First-Strand Synthesis System (Invitrogen). For TgPWS vs

WT (P1), RT-PCR mRNA quantification was carried out

on the same samples as for microarray. QRT-PCR was

performed on an Applied Biosystems, Inc. (ABI) (Foster

City, CA, USA) Prism 7000 sequence detection system

(version 1.6) using SYBR Green dye. Matching primers

for each gene (Supplemental Data Table S3) were

designed using the Primer Express program (ABI). Relative

quantification of gene expression data was performed using

a comparative CT method (http://home.appliedbiosystems.

com/: User Bulletin 777802-002). Gapdh was used as an

endogenous control for comparison of test gene expression,

with DCT values calculated for each sample. One WT DCT

value for each developmental time point (E18, P1, P4) was

set as a calibrator value of 1 and all WT and TgPWS or

TgAS values from the same time point were normalized to

this. The DDCT values were then calculated by subtracting

the calibrator value from each DCT value. These were then

converted to expression fold changes (TSD) as described

(ABI). Mean expression fold changes (TSD) for WT and

TgPWS or TgAS samples and two-tailed p statistics were

calculated using a t test for independent samples (Analyze-

M. Stefan et al. / Genomics 85 (2005) 630–640 639

It, Microsoft Excel). A ratio of mean fold changes was then

determined for the two groups (TgPWS or TgAS vs WT).

BAC contig assembly

The chromosome 18 contig data were generated by stand-

ard Ensembl (http://www.ensembl.org/Mus_musculus/) and

BLAST (http://www.ncbi.nlm.nih.gov/BLAST/) analyses.

Acknowledgments

We thank Brande Latney for technical assistance, Dr.

Don Baldwin for assistance and training with the microarray

experiments, Dr. Sue Keilbaugh for training in QRT-PCR

methodology, Melissa Fazzari, Drs. Hong Ji and John

Tobias for helpful discussions, and Drs. John M. Greally

and Don Baldwin for comments on the manuscript. This

work was supported by the National Institutes of Health

(HD31491 and HD36079 to R.D.N.). R.L. is supported by

Public Health Service Grants CA62234, CA73507, and

CA93444 from the National Cancer Institute and DE13127

from the National Institute of Dental and Craniofacial

Research. M.S. is supported by an award from the American

Heart Association. T.P. is a Special Fellow of the Leukemia

and Lymphoma Society of America.

Appendix A. Supplementary data

Supplementary data for this article may be found on

ScienceDirect.

References

[1] S.B. Cassidy, Prader–Willi syndrome, J. Med. Genet. 34 (1997)

917–923.

[2] P. Burman, E.M. Ritzen, A.C. Lindgren, Endocrine dysfunction in

Prader–Willi syndrome: a review with special reference to GH,

Endocr. Rev. 22 (2001) , 787–799.

[3] D.J Aughton, S.B. Cassidy, Physical features of Prader –Willi

syndrome in neonates, Am. J. Dis. Child. 144 (1990) 1251–1254.

[4] C.J. Partsch, C. Lammer, G. Gillessen-Kaesbach, R. Pankau, Adult

patients with Prader–Willi syndrome: clinical characteristics, life

circumstances and growth hormone secretion, Growth Horm. IGF

Res. (2000) S81–S85 (Suppl. B).

[5] R.D. Nicholls, et al., Mouse models for Prader–Willi and Angelman

syndromes offer insights into novel obesity mechanisms, Prog. Obes.

Res. 9 (2003) 313–319.

[6] R.D. Nicholls, J.L. Knepper, Genome organization, function, and

imprinting in Prader–Willi and Angelman syndromes, Annu. Rev.

Genom. Hum. Genet. 2 (2001) 153–175.

[7] S.L. Christian, et al., Molecular characterization of two proximal

deletion breakpoint regions in both Prader–Willi and Angelman

syndrome patients, Am. J. Hum. Genet. 57 (1995) 40–48.

[8] J.M. Amos-Landgraf, et al., Chromosome breakage in the Prader–

Willi and Angelman syndromes involves recombination between

large, transcribed repeats at proximal and distal breakpoints, Am. J.

Hum. Genet. 65 (1999) 370–386.

[9] J. Cavaille, et al., Identification of brain-specific and imprinted small

nucleolar RNA genes exhibiting an unusual genomic organization,

Proc. Natl. Acad. Sci. USA 97 (2000) 14311–14316.

[10] J.M. Gabriel, et al., A transgene insertion creating a heritable

chromosome deletion mouse model of Prader–Willi and Angelman

syndromes, Proc. Natl. Acad. Sci. USA 96 (1999) 9258–9263.

[11] H. MacDonald, R. Wevrick, The necdin gene is deleted in Prader–

Willi syndrome and is imprinted in human and mouse, Hum. Mol.

Genet. 6 (1997) 1873–1878.

[12] P. Jay, et al., The human necdin gene, NDN, is maternally imprinted

and located in the Prader–Willi syndrome chromosomal region, Nat.

Genet. 17 (1997) 357–361.

[13] M. Niinobe, K. Koyama, K. Yoshikawa, Cellular and subcellular

localization of necdin in fetal and adult mouse brain, Dev. Neurosci.

22 (2000) 310–319.

[14] T. Aizawa, K. Maruyama, H. Kondo, K. Yoshikawa, Expression of

necdin, an embryonal carcinoma-derived nuclear protein, in develop-

ing mouse brain, Brain. Res. Dev. Brain. Res. 68 (1992) 265–274.

[15] K. Maruyama, M. Usami, T. Aizawa, K. Yoshikawa, A novel brain-

specific mRNA encoding nuclear protein (necdin) expressed in

neurally differentiated embryonal carcinoma cells, Biochem. Biophys.

Res. Commun. 178 (1991) 291–296.

[16] K. Matsumoto, H. Taniura, T. Uetsuki, K. Yoshikawa, Necdin acts as a

transcriptional repressor that interacts with multiple guanosine

clusters, Gene 272 (2001) 173–179.

[17] I. Boccaccio, et al., The human MAGEL2 gene and its mouse

homologue are paternally expressed and mapped to the Prader–Willi

region, Hum. Mol. Genet. 8 (1999) 2497–2505.

[18] S. Lee, et al., Expression and imprinting of MAGEL2 suggest a role in

Prader –Willi syndrome and the homologous murine imprinting

phenotype, Hum. Mol. Genet. 9 (2000) 1813–1819.

[19] M.T. Jong, et al., A novel imprinted gene, encoding a RING zinc-

finger protein, and overlapping antisense transcript in the Prader–

Willi syndrome critical region, Hum. Mol. Genet. 8 (1999) 783–793.

[20] C.C. Glenn, et al., Gene structure, DNA methylation, and imprinted

expression of the human SNRPN gene, Am. J. Hum. Genet. 58 (1996)

397–413.

[21] T.A. Gray, S. Saitoh, R.D. Nicholls, An imprinted, mammalian

bicistronic transcript encodes two independent proteins, Proc. Natl.

Acad. Sci. USA 96 (1999) 5616–5621.

[22] T. de los Santos, J. Schweizer, C.A. Rees, U. Francke, Small

evolutionarily conserved RNA, resembling C/D box small nucleolar

RNA, is transcribed from PWCR1, a novel imprinted gene in the

Prader–Willi deletion region, which is highly expressed in brain, Am.

J. Hum. Genet. 67 (2000) 1067–1082.

[23] M. Runte, et al., The IC-SNURF-SNRPN transcript serves as a host

for multiple small nucleolar RNA species and as an antisense RNA for

UBE3A, Hum. Mol. Genet. 10 (2001) 2687–2700.

[24] M. Landers, et al., Regulation of the large (approximately 1000 kb)

imprinted murine Ube3a antisense transcript by alternative exons

upstream of Snurf/Snrpn, Nucleic Acids Res. 32 (2004) 3480–3492.

[25] T.A. Gray, et al., Concerted regulation and molecular evolution of the

duplicated SNRPBV/B and SNRPN loci, Nucleic Acids Res. 27 (1999)

4577–4584.

[26] J.-H. Chai, D.P. Locke, T. Ohta, J.M. Greally, R.D. Nicholls,

Retrotransposed genes such as Frat3 in the mouse chromosome 7C

Prader–Willi syndrome region acquire the imprinted status of their

insertion site, Mamm. Genome 12 (2001) 813–821.

[27] R.D. Nicholls, Incriminating gene suspects, Prader–Willi style, Nat.

Genet. 23 (1999) 132–134.

[28] M. Gerard, L. Hermandez, R. Wevrick, C.L. Stewart, Disruption of the

mouse necdin gene results in early post-natal lethality, Nat. Genet. 23

(1999) 199–202.

[29] F. Muscatelli, et al., Disruption of the mouse Necdin gene results in

hypothalamic and behavioral alterations reminiscent of the human

Prader–Willi syndrome, Hum. Mol. Genet. 9 (2000) 3101–3110.

[30] J. Ren, et al., Absence of Ndn, encoding the Prader–Willi syndrome-

M. Stefan et al. / Genomics 85 (2005) 630–640640

deleted gene necdin, results in congenital deficiency of central

respiratory drive in neonatal mice, J. Neurosci. 23 (2003) 1569–1573.

[31] T. Yang, et al., A mouse model for Prader –Willi syndrome

imprinting-centre mutations, Nat. Genet. 19 (1998) 25–31.

[32] T.-F. Tsai, Y. Jiang, J. Bressler, D. Armstrong, A.L. Beaudet, Paternal

deletion from Snrpn to Ube3a in the mouse causes hypotonia, growth

retardation and partial lethality and provides evidence for a gene

contributing to Prader–Willi syndrome, Hum. Mol. Genet. 8 (1999)

1357–1364.

[33] S. Lee, C.L. Walker, R. Wevrick, Prader–Willi syndrome transcripts

are expressed in phenotypically significant regions of the developing

mouse brain, Gene Expression Patterns. 3 (2003) 599–609.

[34] A.I. Su, et al., A gene atlas of the mouse and human protein-encoding

transcriptomes, Proc. Natl. Acad. Sci. USA 101 (2004) 6062–6067.

[35] V. Janssens, J. Goris, Protein phosphatase 2A: a highly regulated

family of serine/threonine phosphatases implicated in cell growth and

signalling, Biochem. J. 353 (2001) 417–469.

[36] R.K. Dagda, J.A. Zaucha, B.E. Wadzinski, S. Strack, A developmen-

tally regulated, neuron-specific splice variant of the variable subunit B

beta targets protein phosphatase 2A to mitochondria and modulates

apoptosis, J. Biol. Chem. 278 (2003) 24976–24985.

[37] L. Erkman, et al., Role of transcription factors Brn-3.1 and Brn-3.2 in

auditory and visual system development, Nature 381 (1996) 603–606.

[38] D.K. Vassilatis, et al., The G protein-coupled receptor repertoires

of human and mouse, Proc. Natl. Acad. Sci. USA 100 (2003)

4903–4908.

[39] A. Ignatov, I. Hermans-Borgmeyer, H.C. Schaller, Cloning and

characterization of a novel G-protein-coupled receptor with homology

to galanin receptors, Neuropharmacology 46 (2004) 1114–1120.

[40] B. Horsthemke, et al., Somatic mosaicism for maternal uniparental

disomy 15 in a girl with Prader–Willi syndrome: confirmation by cell

cloning and identification of candidate downstream genes, Hum. Mol.

Genet. 12 (2003) 2723–2732.

[41] International Human Genome Consortium, Finishing the euchromatic

sequence of the human genome, Nature 431 (2004) 931–945.

[42] Y. Ge, T. Ohta, D.J. Driscoll, R.D. Nicholls, S.P. Kalra, Anorexigenic

elanocortin signaling in the hypothalamus is augmented in association

with failure-to-thrive in a transgenic mouse model for Prader–Willi

syndrome, Brain Res. 957 (2002) 42–45.

[43] M. Tudor, S. Akbarian, R.Z. Chen, R. Jaenisch, Transcriptional

profiling of a mouse model for Rett syndrome reveals subtle

transcriptional changes in the brain, Proc. Natl. Acad. Sci. USA 99

(2002) 15536–15541.

[44] K. Amano, et al., Dosage-dependent over-expression of genes in the

trisomic region of Ts1Cje mouse model for Down syndrome, Hum.

Mol. Genet. 13 (2004) 1333–1340.

[45] V.K. Mootha, et al., PGC-1alpha-responsive genes involved in

oxidative phosphorylation are coordinately downregulated in human

diabetes, Nat. Genet. 34 (2003) 267–273.

[46] V.K. Mootha, et al., Erralpha and Gabpa/b specify PGC-1alpha-

dependent oxidative phosphorylation gene expression that is altered in

diabetic muscle, Proc. Natl. Acad. Sci. USA 101 (2004) 6570–6575.

[47] M.E. Patti, et al., Coordinated reduction of genes of oxidative

metabolism in humans with insulin resistance and diabetes: potential

role of PGC1 and NRF1, Proc. Natl. Acad. Sci. USA 100 (2003)

8466–8471.

[48] C.T. Culiat, et al., Deficiency of the beta 3 subunit of the type A

gamma-aminobutyric acid receptor causes cleft palate in mice, Nat.

Genet. 11 (1995) 344–346.

[49] J.-H. Chai, et al., Identification of four highly conserved genes

between breakpoint hotspots BP1 and BP2 for the Prader–Willi/

Angelman syndromes deletion region that have undergone evolu-

tionary transposition mediated by flanking duplicons, Am. J. Hum.

Genet. 73 (2003) 898–925.

[50] S.M. Murphy, et al., GCP5 and GCP6: two new members of the human

gamma-tubulin complex, Mol. Biol. Cell. 12 (2001) 3340–3352.

[51] A.L. Lehman, et al., A very large protein with diverse functional

motifs is deficient in rjs (runty, jerky, sterile) mice, Proc. Natl. Acad.

Sci. USA 95 (1998) 9436–9441.

[52] Y. Ji, et al., The ancestral gene for transcribed, low-copy repeats in the

Prader–Willi/Angelman region encodes a large protein implicated in

protein trafficking, which is deficient in mice with neuromuscular and

spermiogenic abnormalities, Hum. Mol. Genet. 8 (1999) 533–542.

[53] K. Kobayashi, et al., p140Sra-1 (specifically Rac1-associated protein)

is a novel specific target for Rac1 small GTPase, J. Biol. Chem. 273

(1998) 291–295.

[54] A. Schenck, et al., A highly conserved protein family interacting with

the fragile X mental retardation protein (FMRP) and displaying

selective interactions with FMRP-related proteins FXR1P and FXR2P,

Proc. Natl. Acad. Sci. USA 98 (2001) 8844–8849.

[55] H. Siomi, A. Ishizuka, M.C. Siomi, RNA interference: a new

mechanism by which FMRP acts in the normal brain? What can

Drosophila teach us?, Ment. Retard. Dev. Disabil. Res. 10 (2004)

68–74.

[56] L. Gewin, H. Myers, T. Kiyono, D.A. Galloway, Identification of a

novel telomerase repressor that interacts with the human papilloma-

virus type-16 E6/E6–AP complex, Genes Dev. 18 (2004) 2269–2282.

[57] L.B. Russell, C.S. Montgomery, N.L. Cacheiro, D.K. Johnson,

Complementation analyses for 45 mutations encompassing the

pink-eyed dilution (p) locus of the mouse, Genetics 141 (1995)

1547–1562.

[58] D.K Johnson, et al., Molecular analysis of 36 mutations at the mouse

pink-eyed dilution (p) locus, Genetics 141 (1995) 1563–1571.

[59] R.G. Caldwell, J.B. Wilson, S.J. Anderson, R. Longnecker,

Epstein–Barr virus LMP2A drives B cell development and survival

in the absence of normal B cell receptor signals, Immunity 9

(1998) 405–411.

[60] V.G. Tusher, R. Tibshirani, G. Chu, Significance analysis of micro-

arrays applied to the ionizing radiation response, Proc. Natl. Acad.

Sci. USA 98 (2001) 5116–5121.