GH1 Splicing Is Regulated by Multiple Enhancers Whose ... - GenScript

0013-7227/04/$15.00/0 Printed in U.S.A.

Endocrinology 145(6):2988 –2996 Copyright © 2004 by The Endocrine Society doi: 10.1210/en.2003-1724

GH1 Splicing Is Regulated by Multiple Enhancers Whose Mutation Produces a Dominant-Negative GH Isoform That Can Be Degraded by Allele-Specific Small Interfering RNA (siRNA) ROBIN C. C. RYTHER, ALEX S. FLYNT, BRYAN D. HARRIS, JOHN A. PHILLIPS III, JAMES G. PATTON

AND

Department of Biological Sciences (R.C.C.R., A.S.F., B.D.H., J.G.P.), Vanderbilt University; and Department of Pediatrics (J.A.P.), Vanderbilt University School of Medicine, Nashville, Tennessee 37235 The majority of mutations that cause isolated GH deficiency type II affect splicing of GH1 transcripts, leading to the production of a dominant-negative GH isoform. Because numerous mutations and polymorphisms throughout the GH1 gene have not yet been tested for aberrant splicing, we used a deletion mutagenesis screen across intron 2-exon 3-intron 3 to identify splicing regulatory sequences. These analyses identified a new enhancer element, ESE2, upstream of the cryptic splice site in exon 3 and further defined a previously described enhancer (ESE1) to include the first seven nucleotides

S

PLICING IS THE process by which introns are removed and exons are joined to generate mature mRNA. It is mediated by the complex interplay of several cis- and transacting factors. In constitutive splicing, the cis-acting elements include the 5⬘ splice site, branch point, polypyrimidine tract, and the 3⬘ splice site. These are identified and acted on by the RNA and protein components of the macromolecular complex termed the spliceosome (reviewed in Ref. 1). Additional sequences are often needed to help recognize weak splice sites that only loosely conform to consensus sequences. These elements are often composed of purine rich or A/C rich sequences called splicing enhancers (2, 3). They can occur in exons, termed exonic splicing enhancers (ESEs), or in introns, intronic splicing enhancers (ISEs), and are typically recognized by the serine-arginine rich (SR) family of splicing regulators (1, 4, 5). The human GH gene, GH1, is composed of five exons and four introns with fairly weak splice sites across introns 2 and 3 (6). Removal of all four introns generates the full-length mature transcript, which is translated to produce the 22-kDa GH isoform. However, aberrant splicing of wild-type transcripts can produce minor amounts of smaller GH isoforms (reviewed in Ref. 7). The two most common aberrantly Abbreviations: ASLV, Avian sarcoma-leukosis virus; dsRNA, double-stranded RNA; ESE, exonic splicing enhancer; FL, full-length; GHD, GH deficiency; IGHD, isolated GH deficiency; ISE, intronic splicing enhancer; siRNA, small interfering RNA; SR, serine-arginine rich; wt, wild-type. Endocrinology is published monthly by The Endocrine Society (http:// www.endo-society.org), the foremost professional society serving the endocrine community.

of exon 3. Besides enhancers, the overall size of intron 3 is also crucial for exon inclusion. Given the deleterious effects of the dominant-negative 17.5-kDa isoform, these and previous studies underscore the extent to which splicing regulatory elements serve to prevent exon skipping. Importantly, we show here that small interfering RNAs can be used to specifically degrade exon 3-skipped transcripts, potentially a new avenue of therapeutic intervention in isolated GH deficiency II and other dominant disorders. (Endocrinology 145: 2988 –2996, 2004)

spliced products involve exon 3. Activation of an in-frame cryptic splice site generates a transcript identical to that of the full-length transcript but missing the first 45 nt of exon 3, producing a 20-kDa isoform. Complete skipping of exon 3 generates transcripts encoding a 17.5-kDa isoform that lacks the protein linker domain between the first two helices of GH and a cysteine residue involved in an internal disulfide bridge (8). The 17.5-kDa isoform exhibits a dominant-negative effect on secretion of the 22-kDa isoform in both tissue culture cells (9, 10) and transgenic mice (11). In addition, the 17.5-kDa isoform is retained in the endoplasmic reticulum, disrupts the Golgi apparatus, impairs both GH-releasing hormone and TRH protein trafficking (12), and partially reduces the stability of the 22-kDa isoform (10). This likely explains why transgenic mice overexpressing the 17.5-kDa isoform exhibit a defect in the maturation of GH secretory vesicles and anterior pituitary hypoplasia due to loss of the majority of somatotrophs (6, 11). Mutations in patients with isolated GH deficiency type (IGHD) II, an autosomal dominant form of GH deficiency (GHD), cause increased levels of exon 3 skipped transcripts encoding the 17.5-kDa isoform. The majority of the known mutations that cause IGHD II are located at the 3⬘ and 5⬘ splice sites bordering exon 3 (13–20). In contrast, a second group of IGHD II mutations that affect splicing of GH1 are located at positions distinct from the canonical splice sites. These include two single base pair substitutions (exon 3 ⫹ 5 A3 G, ESEm and IVS3 ⫹ 28 G3 A, ISEm1) and a 17-nt deletion in intron 3 (IVS3⌬28 – 45, ISEm2). These mutations were found to lie within purine-rich sequences and cause increased levels of exon 3 skipped transcripts (14, 21, 22),

2988

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

most consistent with the notion that they disrupt splicing enhancer elements (ESE1 at the 5⬘ end of exon 3 and an ISE in intron 3) (6, 21, 23). That splicing of GH1 might require enhancers to aid appropriate recognition of exon 3 is consistent with the finding that splicing of wild-type GH1 normally produces minor amounts of non-full-length transcripts and because the intron 2 and intron 3 splice sites are weak (6, 7). Previous work has demonstrated the critical importance of avoiding exon skipping, illustrating that regulation of GH1 splicing is essential to GH secretion and function (6, 9 –12). In this study, we sought to identify additional cis-acting regulatory elements that control GH1 splicing to facilitate mutation interpretation, generate a clearer model of GH1 splicing regulation, and better understand the pathogenesis of IGHD II. Because numerous variations throughout the GH1 gene have not been tested for their effects on splicing, we screened intron 2-exon 3-intron 3 by generating deletion mutants across these regions to identify splicing regulatory sequences. This analysis led to three discoveries. First, a new splicing enhancer was identified adjacent to the cryptic splice site in exon 3. Second, the overall size of intron 3 is crucial to exon 3 inclusion. Third, we refined and expanded the ESE1 region to include the first seven nucleotides of exon 3. Importantly, we also demonstrate that small interfering RNAs (siRNAs) can be used to specifically degrade exon 3 skipped transcripts and thus provide a potential new avenue of therapeutic intervention in IGHD II and other dominant disorders. Materials and Methods Plasmid constructs Wild-type GH1, ESEm, ISEm1, and ISEm2 were cloned into pXGH5 as previously described (21). Deletion and point mutants were generated from wild-type GH1 by reverse PCR (24) using mismatch primers followed by sequence verification of all constructs. DSX⌬E and DSX-avian sarcoma-leukosis virus (ASLV) were generous gifts from Dr. Brenton R. Graveley. Annealed and phosphorylated DNA primers corresponding to each GH enhancer sequence (see Fig. 4A) were inserted into DSX⌬E at the BstBI in exon 2 for the doublesex GH constructs. The 17.5-si construct was generated by inserting the annealed and phosphorylated DNA primers (5⬘-GATCCCCCCAGGAGTTTAACCTAGAGTTCAAGAG-ACTCTAGGTTAAACTCCTGGTTTTTGGAAA-3⬘, 5⬘-AGCTTTTCCAAAAACCAGGAGTTTAACCTAGAG-TCTCTTGAACTCTAGGTTAAACTCCTGGGGG-3⬘) into the BglII and HindIII sites in pSuper and pSuper.retro (Oligoengine, Seattle, WA) (25). Transcripts were then transcribed from a histone H1 promoter-generating RNA that folds back upon itself to form a hairpin loop structure that is processed in vivo to double-stranded RNAs (dsRNAs) corresponding to the exon 2-exon 4 junction. The control-si construct was generated using the DNA primers (5⬘-GATCCCCCAGTCGCGTTTGCGACTGGTTCAAGAGACCAGTCGCAAACGCGACTGTTTTTGGAAA-3⬘ and 5⬘-AGCTTTTCCAAAAACAGTCGCGTTTGCGACTGGTCTC-TTGAACCAGTCGCAAACGCGACTGGGG-3⬘) as above. The GH1 nucleotides are numbered according to the transcription initiation site as ⫹1 for GenBank accession no. J03071.

In vivo GH splicing analysis (GH3 cells) One thousand nanograms of wild-type or mutant GH1 constructs were transfected into GH3 cells using the Mirus TransIT system (Panvera, Invitrogen Life Technologies, Carlsbad, CA). RNA isolation was performed 48 h post transfection using TRI-Reagent (Molecular Research Center, Cincinnati, OH), treated with DNase I, washed with 75% ethanol, and precipitated. cDNA was synthesized with avian myelo-

Endocrinology, June 2004, 145(6):2988 –2996 2989

blastosis virus-reverse transcriptase (Promega, Madison, WI) and a human GH-RT2 primer located in the 3⬘-untranslated region (5⬘-GGACAAGGCTGGTGGGCACTGG-3⬘). GH1 exons 2–5 were PCR amplified with 32P-labeled primers (exon 2: 5⬘-CCATCGTCTGCACCAGCTGGC-3⬘ and exon 5: 5⬘-CACAGCTGCCCTCCACAGA-GC-3⬘). PCR products were separated with 6% denaturing PAGE and exposed to phosphor imager analysis.

ESEFinder analysis The entire GH1 sequence was analyzed using the ESEFinder (release 2.0, available at http://exon.cshl.org/ESE) (26), and the motif score was plotted against GH1 sequence surrounding E3⌬5.

In vitro splicing analysis (cell free) All doublesex constructs were linearized with Mlu I, capped, and in vitro transcribed with T7 RNA polymerase. Transcripts were incubated for 2 h under splicing conditions in 60% HeLa cell nuclear extract at 30 C. The resulting products were separated on 6% denaturing polyacrylamide gels and exposed to phosphor imager analysis.

In vivo siRNA analysis (GH3 cells) One thousand nanograms of plasmids expressing either wild-type (wt) GH1 or GH1 enhancer mutants (ESEm and ISEm1) were cotransfected with either 2000 ng of a plasmid expressing the 17.5-si, a random sequence (control-si), or a combination of the two using the Mirus TransIT system (Panvera). RNA was isolated 48 h post transfection using TRI-Reagent (Molecular Research Center), treated with DNase I, washed with 75% ethanol, precipitated, and divided into two pools. RT-PCR was performed on the first pool to analyzed GH1 splicing patterns as described above. For the second pool, first-strand cDNA was generated with an oligo-dT primer and, as a control, actin-specific primers 5⬘-CTTAATGTCACGCACGATTTCC-3⬘ and 5⬘-CGTGATGGTGGGCATGGGTCAG-3⬘. The resulting products were analyzed on 6% denaturing polyacrylamide gels and exposed to phosphor imager analysis.

Results Deletion analysis of GH1 intron 2-exon 3-intron 3

To identify splicing regulatory elements in GH1, mammalian expression constructs were generated containing either the entire wt GH1 sequence or constructs in which small regions of intron 2, exon 3, or intron 3 were deleted. Deletion constructs were designed to avoid regions of known splicing regulatory elements (5⬘ splice sites, 3⬘ splice sites, putative branch points, and polypyrimidine tracts) as well as the cryptic 3⬘ splice site within exon 3. We also maintained the reading frame in all constructs to avoid any effects of nonsense-mediated decay or nonsense-mediated altered splicing (reviewed in Ref. 27). As shown in Fig. 1, each deletion mutant was named according to the intron/exon name and the deletion number (i.e. the second deletion in intron 3 will be referred to as I3⌬2). Constructs were transiently transfected into rat somatotroph cells (GH3 cells), and total RNA was isolated 48 h post transfection. GH1 splicing patterns were then analyzed for the full-length (FL), cryptic, and skipped isoforms by RT-PCR. All deletion mutants that increased the percent of either the cryptic or skipped isoforms were then analyzed further for possible splicing enhancer activity. For intron 2, none of the deletion mutants displayed any increase in the percent of the cryptic or skipped isoforms (Fig. 1B). Within exon 3, two of the deletions (E3⌬1 and E3⌬5) displayed increased amounts of aberrantly spliced isoforms (Fig. 1C). E3⌬1, a 3-nt deletion immediately

2990

Endocrinology, June 2004, 145(6):2988 –2996

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

FIG. 1. GH1 deletion analysis of intron 2-exon 3-intron 3. A, The position and relative lengths of each deletion is shown by the numbered lines beneath each region. Exons are in boxes and introns are single lines. The location of the cryptic splice site in exon 3 is indicated by a dashed line and the locations of ESE1 and the ISE are indicated with boxes shaded with a forward slash and a checkerboard, respectively. Exon 3 has been enlarged to show the positions and relative lengths of the indicated deletions. GH3 cells were transiently transfected with vectors containing either wt GH1 or GH1 with ESEm, ISEm1, or the indicated deletion mutants from mutations in intron 2 (B), exon 3 (C), or intron 3 (D). Total RNA was isolated followed by RT-PCR analysis of splicing. Products representing the FL isoform, activation of the cryptic splice site (cryptic), or the exon 3 skipped isoform (skipped) are indicated to the left.

downstream of ESE1, exhibited increased levels of both the cryptic and skipped isoforms. In contrast, E3⌬5, a 15-nt deletion upstream of the cryptic splice site in exon 3, increased only the percent of skipped transcripts and showed no increase in cryptic splicing. E3⌬10 occasionally showed marginal increases in skipped transcripts that were inconsistent between experiments, and we have therefore chosen not to explore the E3⌬10 splicing pattern further. The larger deletions in intron 3 (I3⌬1, I3⌬3, and I3⌬4; Fig. 1D) also displayed increased levels of skipped isoforms, greatest in I3⌬1 and I3⌬4. In sum, deletion analysis of intron 2, exon 3, and intron 3 defined four regions for further analysis: E3⌬1, E3⌬5, I3⌬1, and I3⌬4. ESE1 consists of the first seven nucleotides in exon 3

ESE1 was originally identified in a family with IGHD II that had an A3 G transition (ESEm) at the fifth nucleotide

of exon 3 (21). This mutation is within a (GAA)2 repeat that resembles the (GAR)N splicing enhancer motif (1, 28). When ESEm was analyzed in vivo for aberrant splicing, increased levels of the cryptic and skipped isoforms were observed (Fig. 2B and Ref. 21). Subsequent analysis determined that point mutations across the region also led to increased percentages of the cryptic and skipped transcripts (21). Given that E3⌬1 lies immediately 3⬘ of what was thought to constitute ESE1 and because the splicing patterns are nearly identical for both mutants, it seems likely that one or more of the nucleotides deleted in E3⌬1 (Fig. 2A) contributes to the overall splicing activity of ESE1. As such, we mutated each nucleotide within E3⌬1 and analyzed the resulting splicing patterns (Fig. 2B). The G709T mutation led to increased percentages of both the cryptic and skipped isoforms, similar to E3⌬1 and ESEm. In contrast, C710T and C711T did not increase either the cryptic or skipped isoforms, displaying wt

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

splicing. These results suggest that the sequence of ESE1 consists of the first seven nucleotides in exon 3, 5⬘-GAAGAAG-3⬘ (Fig. 2A, diagonal box).

FIG. 2. ESE1 consists of the first seven bases in exon 3. A, Diagram of the 5⬘ end of exon 3. Intron 2 sequences are lowercase, and exon 3 sequences are in capital letters and boxed. The sequence representing the three nucleotide deletion E3⌬1 is shaded in gray. The newly defined position of ESE1 is indicated by the box with diagonal hatches. As indicated, ESEm is an A3 G transition just upstream of position 709. B, GH3 cells were transiently transfected with vectors containing either wt GH1 or various GH1 mutations including an in-frame threenucleotide deletion (E3⌬1) and four point mutations (ESEm, G709T, C710T, and C711T). Total RNA was isolated followed by RT-PCR analysis of splicing. Products representing the FL isoform, activation of the cryptic splice site in exon 3 (cryptic), or the exon 3 skipped isoform (skipped) are indicated to the left.

FIG. 3. Enhancer activity in exon 3. A, The entire GH1 sequence was analyzed using ESEFinder (26). The sequence (xaxis) immediately surrounding E3⌬5 (underlined) and the motif scores for each sequence beginning at the letter on the x-axis is represented on the y-axis of the bar graph. Position 721 is indicated by a carrot and the K41R mutation (A731G) is indicated by an arrow. B, GH3 cells were transiently transfected with vectors containing wt GH1, a 15nucleotide deletion in exon 3 (E3⌬5), or the A731G GHD patient mutation. Total RNA was isolated followed by RTPCR analysis of splicing. PCR products representing the FL isoform, activation of the cryptic splice site (cryptic), or the exon 3 skipped isoform (skipped) are indicated to the left. C, Point mutations generated across E3⌬5 and transiently transfected into GH3 cells, and the resulting splicing products were analyzed by RT-PCR.

Endocrinology, June 2004, 145(6):2988 –2996 2991

Sequences within E3⌬5 strongly resemble an ESE and contain GHD patient mutations

E3⌬5, a 15-nt deletion, is located 12 nt upstream of the cryptic splice site in exon 3 (Fig. 1A). Because E3⌬5 increased the percent of skipped transcripts and is purine rich, we hypothesized that it might contain an ESE. To initially address this, we analyzed the sequences within E3⌬5 using ESEFinder (Ref. 26; http://exon.cshl.org/ESE/). ESEFinder is designed to identify SR protein binding sites, generating a motif score that reflects binding site strength. E3⌬5 contains several overlapping SR protein binding sites including sites for SRp40 and SF2/ASF (Fig. 3A). To determine whether mutations within this region would lead to aberrant splicing, we generated point mutations across E3⌬5 and analyzed their splicing patterns (Fig. 3C). Each nucleotide was mutated to a pyrimidine whenever possible to increase the likelihood of disrupting purine-rich enhancers. In addition, none of the constructs contained any nonsense codons to avoid any effects of nonsense-mediated decay or nonsense-mediated altered splicing (reviewed in Ref. 27). As shown in Fig. 3C, many of the point mutations from nucleotides 721 to 732 caused aberrant splicing, increasing either the cryptic or skipped isoforms, or both, suggesting that the region deleted in E3⌬5 contains a previously undiscovered ESE. One of the potential benefits of studying splice site selection in the GH1 gene is the presence of a number of patient mutations that cause GHD. We therefore sought to determine whether any patients with GHD had mutations within E3⌬5. Recently Millar et al. (20) found additional GH1 gene lesions by modulating the clinical selection criteria to include all patients with physical findings suggestive of possible GHD. They identified an individual at the low end of the normal range of stature (⫺1.8 sd), reduced height velocity (⫺0.9), and delayed bone age who was heterozygous for an A3 G transition at nucleotide 731 (A731G) that results in a

2992

Endocrinology, June 2004, 145(6):2988 –2996

lysine to arginine substitution (K41R) (20). Because this mutation lies within E3⌬5 (Fig. 3A), we determined the effect of this mutation on GH1 splicing. The A731G point mutation increased the amount of skipped transcripts without any corresponding increase in cryptic splicing (Fig. 3B). This suggests that aberrant splicing could play a role in this form of GHD and that this region contains a novel ESE in exon 3 that we will refer to as ESE2. ESE2 activates splicing in a heterologous setting

One hallmark for defining splicing enhancers is the ability of these sequences to activate splicing in heterologous settings. To determine whether ESE2 could activate splicing, we used a construct derived from the D. melanogaster doublesex (dsx) gene whose splicing is enhancer dependent (DSX⌬E, Fig. 4A) (29 –32). As a positive control, an enhancer from the ASLV was inserted into the downstream exon. To test enhancer activity of GH sequences, a series of constructs were created. These included constructs in which each GH enhancer was inserted in the forward orientation in the down-

FIG. 4. ESE2 activates splicing in a heterologous setting. A, Diagram of the doublesex (DSX) minigene construct (29) containing two exons (boxes) and an intron (line). Various sequences inserted at the BstBI site in exon 2 (dotted box) are listed below. These include an ASLV enhancer (29), a sequence containing back-to-back copies of ESE1, ISE, and ESE2 called multi-GH, two copies of this same insert (multi-x2), and multi-GH in the reverse orientation (multirev). Transcripts were subjected to in vitro splicing in nuclear extracts for either 0 or 2 h. Spliced products were separated on 6% denaturing polyacrylamide gels (B and D) and the percent of mRNA (spliced products/premRNA*100) was quantified using phosphor imager analysis software (C and E). B, Two copies of each individual component of the multi-GH enhancer were inserted into DSX⌬E and spliced as above (D, E, and data not shown).

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

stream exon (DSX-ESE1, DSX-ESE2, and DSX-ISE) or, as a control, in the reverse orientation (DSX-ESE1-rev, DSX-ESE2rev, and DSX-ISE-rev). In addition, because splicing enhancers often require multiple sequence elements to activate splicing, a multi-GH enhancer was generated containing the ESE1, ESE2, and the ISE elements in the forward orientation (multi-GH), reverse orientation (multi-rev), or a tandem repeat of these elements in the forward orientation (multi-x2). Transcripts from all five constructs were subjected to splicing nuclear extracts (in vitro splicing) and analyzed on 6% denaturing polyacrylamide gels. DSX⌬E exhibited very little activation of splicing, whereas insertion of the ASLV enhancer strongly increased mRNA levels (Fig. 4, B and C). The GH multi enhancers (multi and multi-x2) activated splicing to a similar degree as the ASLV enhancer, whereas the multi-rev construct was mostly inactive (Fig. 4C). We conclude that the increase in splicing is attributable to splicing enhancer activity from the GH1 sequences. To determine which of the three GH enhancers contributed to the splicing activation observed with the DSX-multi construct, we ana-

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

Endocrinology, June 2004, 145(6):2988 –2996 2993

lyzed the activation ability of each enhancer sequence by itself. The DSX-ESE1 and DSX-ISE constructs only weakly activated splicing (⬃1.5- to 2-fold), compared with DSX⌬E and the appropriate reverse construct (data not shown). In contrast, DSX-ESE2 strongly activated splicing (Fig. 4D), an activation that was not observed with the DSX-ESE2-rev construct. This demonstrates that ESE2 activates splicing in a heterologous setting. It is interesting to note that ESE2 appears to contribute the greatest splicing activation of all three GH enhancers in the DSX-multi construct, suggesting that ESE2 might also regulate the majority of GH1 splicing in cells (in vivo splicing). This is supported by observation that point mutations within ESE2 (Fig. 3C) cause more aberrant splicing than point mutations in either ESE1 (21) or the ISE (23). Decreasing the size of intron 3 in GH1 leads to aberrant splicing

Intron 3 of GH1 is 92 nt in length, well above the minimum intron length required for spliceosome assembly and efficient splicing (33–38). We found it intriguing that all of the large deletions in intron 3, ranging in size from 12- to 14-nt, (I3⌬1, I3⌬3, and I3⌬4; Fig. 1D) showed an increase in the skipped isoform, whereas the smaller 4-nt deletion (I3⌬2; Fig. 1D) did not, suggesting that decreasing the size of intron 3 might prevent its accurate identification by the spliceosome. We chose I3⌬1 and I3⌬4 for further analysis because they exhibited the greatest amount of exon skipping. To examine the apparent size effect observed in these mutants, we generated nonoverlapping subdeletions within I3⌬1 and I3⌬4. These deletions were 3, 4, or 6 nt in length, a size analogous to the smaller I3⌬2 deletion that did not show any increased exon skipping (Fig. 1D). The subdeletions are named according to the original deletion and position (i.e. I3⌬1–1 for the first subdeletion within I3⌬1). All of the subdeletions for both I3⌬1 (I3⌬1–1, I3⌬1–2, and I3⌬1–3) and I3⌬4 (I3⌬4 –1, I3⌬4 –2, and I3⌬4 –3) displayed wt splicing patterns (Fig. 5B), suggesting that decreasing the size of intron 3 by as little as 12 nt caused increased exon 3 skipping. To verify this, constructs were generated in which the same number of nucleotides deleted in I3⌬1 and I3⌬4 were replaced with sequences from the 5⬘ end of I2⌬5 to generate I3⌬1-ins and I3⌬4-ins, respectively. We chose the I2⌬5 sequence because deletion of this sequence showed no increase in any of the aberrantly spliced GH isoforms, suggesting that it does not contain any enhancer elements. Inserting this sequence into I3⌬4 completely restored the wt splicing pattern (Fig. 5C), demonstrating that the aberrant splicing pattern observed with the I3⌬4 mutant was the result of a decrease in overall intron size. Inserting this sequence into I3⌬1 only partially restored splicing (approximately 20% skipped transcripts remain in I3⌬1-ins; Fig. 5C) showing that the decrease in intron size plays an important role in the aberrant splicing observed in I3⌬1. However, because I3⌬1-ins still demonstrates a small amount of exon skipping, the sequences within I3⌬1 may also play a role in splicing.

FIG. 5. Intron 3: reduced size increases exon skipping. A, The position of the known ISE (hatched box) and a series of deletions and subdeletions in intron 3 (lines) are shown. B and C, GH3 cells were transiently transfected with vectors containing either wt GH1 or various GH1 deletions. Total RNA was isolated followed by RT-PCR analysis of splicing. Products representing the FL isoform, activation of the cryptic splice site (cryptic), or the exon 3-skipped isoform (skipped) are indicated to the left. B, Subdeletions in I3⌬1 and I3⌬4 all demonstrated a GH splicing pattern comparable with wt splicing. C, The regions deleted in I3⌬1 and I3⌬4 were replaced with sequences from the 5⬘ end of I2⌬5 identical in length to the sequences originally deleted (Fig. 1B).

Allele-specific siRNA mediated degradation

The presence of multiple enhancers contributing to exon 3 definition and the enhancer-dependent activation of the intron 2 3⬘ splice site at the expense of the stronger, nearby cryptic splice site (6) demonstrates that multiple mechanisms have evolved to avoid production of skipped GH1 transcripts, especially those encoding the dominant-negative 17.5-kDa isoform. Because numerous in vivo (6, 11) and in vitro experiments (9, 10, 12) suggest that normal GH secretion depends, in part, on avoiding increased levels of the 17.5-kDa isoform, therapies that specifically target the 17.5-kDa isoform would be useful in patients with IGHD II. To exclusively target skipped transcripts for degradation, we employed a siRNA strategy (Fig. 6A; Refs. 39 – 42). We selected the exon 2-exon 4 junction as the siRNA target because it represents the only sequence within GH1 that is unique to transcripts encoding the 17.5-kDa isoform. Plasmids expressing this siRNA (17.5-si) or a random dsRNA sequence (control-si) were cotransfected in conjunction with either wt GH1 or GH1 enhancer mutants (ESEm and ISEm1) that display increased percentages of skipped transcripts. Splicing patterns and actin expression levels were then analyzed by RT-PCR. The 17.5-siRNA construct reduced the percent of skipped transcripts by greater than 95%, whereas having no effect on either the FL or cryptic isoforms (Fig. 6B). Similar analyses with dsRNA identical to the 17.5-si sequence re-

2994

Endocrinology, June 2004, 145(6):2988 –2996

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

Intron 3: size and sequence requirements

Intron 3 of GH1 is 92 nt in length, well above the approximately 50 nucleotide minimum intron length required for spliceosome assembly (33–38). Nevertheless, deletion analysis within intron 3 suggests that the exon skipping observed with deletions at least 12 nt long is due to decreasing the size of intron 3, rather than the actual sequences deleted. Intron 3 also contains a previously characterized ISE (6, 23). Two families with IGHD II are heterozygous for mutations in this ISE: ISEm1 (IVS3 ⫹ 28 G3 A) and ISEm2 (IVS3⌬28 – 45) (14, 22). In light of our findings that decreasing the size of intron 3 results in aberrant splicing, it is interesting to note that ISEm2, a 17-bp deletion, exhibits a greater percent of exon skipping (6) than ISEm1, a point mutation in the same region (Figs. 1D and 6). This suggests that the deletion mutant (a deletion larger than any of the mutations made in this study) causes aberrant splicing through two distinct mechanisms, both decreased intron size and disruption of an enhancer. ESE2 regulates exon 3 inclusion

FIG. 6. Allele-specific siRNA. A, siRNAs were designed to target the exon 2-exon 4 junction. B, GH constructs (wt, ESEm, and ISEm1) were transiently cotransfected with vectors expressing either siRNAs targeting the exon-skipped isoform (17.5-si) or control siRNAs containing a scrambled sequence (control-si). Total RNA was isolated followed by RT-PCR analysis of GH splicing products or actin (shown below). Products representing the FL isoform, activation of the cryptic splice site (cryptic), the exon 3 skipped isoform (skipped), or actin are indicated to the left.

sulted in a greater than 90% decrease in the percent of the skipped transcripts without affecting the full-length or cryptic transcripts (data not shown). Thus, siRNAs can specifically degrade transcripts encoding the 17.5-kDa isoform, suggesting that this strategy could be effective in preventing production of the dominant-negative 17.5-kDa GH isoform. Discussion

Multiple experiments in cultured cell lines and transgenic mice, as well as numerous IGHD II patient mutations, have demonstrated the importance of avoiding increased levels of the dominant-negative 17.5-kDa GH isoform. Because GH1 has several weak splice sites and exhibits small amounts of aberrant splicing even in normal patients, cis-acting splicing regulatory elements are essential to maintain full-length splicing of GH1. In this study, we localized and defined the cis-acting elements that regulate exon 3 inclusion. To this end, intron 2, exon 3, and intron 3 were screened for splicing enhancer activity by generating deletion mutants across these regions. This analysis resulted in three new discoveries regarding GH1 splicing regulation. First, a new splicing enhancer, ESE2, was identified, exhibiting the strongest splicing activity of all three enhancers. Second, the overall size of intron 3 is crucial to exon 3 inclusion. Third, ESE1 includes the first seven nucleotides of exon 3.

E3⌬5 is an in-frame, 15-nt deletion in exon 3 that showed increased exon skipping. Sequences in E3⌬5 are very purine rich and contain several overlapping SR protein binding sites, suggesting that they may contain a previously undiscovered ESE. Point mutations within this region resulted in aberrant splicing, and this sequence could also activate splicing in a heterologous setting, demonstrating that the sequences within E3⌬5 contain an ESE. As described above, an individual has been identified who is heterozygous for an A3 G transition within E3⌬5 and has normal GH secretion, but reduced height (K41R) (20). The A731G construct led to approximately 20% exon skipping, the weakest aberrant splicing pattern observed in our analyses. In addition, Millar et al. (20) also found that in transfected cells, the A731G mutation exhibited reduced GH secretion and receptor activation and was associated with a low-expressing promotor haplotype. Furthermore, although this mutation was not found in a group of normal controls, strict dominant inheritance can be eliminated in this family because the A731G mutation was maternally inherited from a woman of normal stature (0.9 sd) (20). Thus, this mutation seems to exert pleiotropic effects on height, a mutational mechanism that may be the norm rather than the exception in the genetics of stature. This is consistent with the finding that small amounts of exon skipping can apparently be tolerated because low levels of exon skipping are often observed in individuals of normal height (Fig. 1 and Refs. 6 and 7). The lowest percent of exon 3 skipping that has been observed to date in a family with dominant GHD and reduced GH secretion is 45% from that allele (ESEm) (6, 21). This suggests that increases in exon 3 skipping and the amount of the 17.5 kDa might have to reach a certain threshold to exhibit dominant effects on GH secretion. In addition, this agrees with the dose dependence of dominant-negative effects on GH secretion seen in rat somatotroph cells (MtT/S) (9), suggesting that it is the ratio of the 17.5-kDa isoform to the 22-kDa GH isoform that is important in determining GH secretion. Observations in transgenic mice also support this hypothesis because mouse lines with high transgene numbers expressing the 17.5-kDa iso-

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

form (⌬3 allele) exhibit almost undetectable amounts of GH in comparison with only reduced GH levels in the mice with low copy numbers (11). The observable differences in GH levels are also reflected phenotypically because high copy number mice display a more severe phenotype and earlier age of onset. In light of these findings, it is interesting to note that patients with splice site mutations and greater amounts of the 17.5-kDa isoform appear to have more severe GHD and an earlier age of onset than do patients with the enhancer mutations and less 17.5-kDa isoform (21, 43). Allele-specific RNAi

Current GH replacement therapy involves sc injections of recombinant GH through puberty (reviewed in Ref. 44). As successful as this therapy has been in helping GHD patients attain more normal stature, such injections cannot replicate the normal, pulsatile pattern of GH secretion (45). In addition, although GH replacement therapy is generally well tolerated, side effects still remain including insulin resistance, benign intracranial hypertension, slipped capital femoral epiphysis, prepubertal gynecomastia, arthralgia, edema, myalgia, headache, and hypertension (44, 46 –50). Although these side effects can often be mitigated with individualized dose adjustment, novel methods to restore endogenous GH secretion could be more efficacious and safe. As such, therapies that specifically target the 17.5-kDa isoform could be useful in patients with IGHD II and GH1 splicing defects. To this end, we targeted the exon 3 skipped transcripts for degradation by designing siRNAs for the exon 2-exon 4 junction. In cultured GH3 cells, siRNAs delivered either as dsRNAs or using a plasmid expression system efficiently and specifically degraded the exon 3 skipped transcripts. Our results suggest that siRNAs could be designed to target dominant, aberrantly spliced isoforms or other dominant alleles for degradation, a new field of therapeutics (51). Clinical use of siRNAs will require an effective and safe delivery system, one of many obstacles to their potential therapeutic use. IGHD II may be an ideal system in which to study the use of therapeutic siRNA because the pituitary is apparently not protected by the blood-brain barrier and may be reasonably accessible through its blood supply. In addition, delivery of even small amounts of the 17.5-si may prove efficacious because the 17.5-si is very effective at low concentrations (data not shown). Thus, delivery of even a small amount of siRNAs to the pituitary might possibly restore more normal GH secretion in IGHD II patients. In conclusion, splicing of GH1 exon 3 is controlled by the complex interplay of at least three splicing enhancers in addition to normal splicing regulatory elements. Additional regulatory elements likely remain undefined in other regions of GH1 that could act combinatorially to ensure accurate splicing. Avoidance of aberrantly spliced exon-skipped transcripts encoding the 17.5-kDa isoform is critical to normal GH function and secretion. In patients with GHD due to splicing mutations, allele-specific siRNA could specifically and efficiently degrade exon 3 skipped transcripts, possibly providing a new therapeutic approach that, if successful, could potentially have broad therapeutic implications in other dominant disorders.

Endocrinology, June 2004, 145(6):2988 –2996 2995

Acknowledgments The authors thank Drs. Brenton R. Graveley, Reuven Agami, David S. Millar, David N. Cooper, and Annie M. Procter for the kind donation of various reagents. We also thank Drs. Brenton R. Gravely, Massimo Caputi, Adrian R. Krainer, Michael Q. Zhang, and Melissa Prince for helpful discussions. Thanks also to Drs. S. K. Dey and Joy Cogan for critical reading of the manuscript. Received December 19, 2003. Accepted February 17, 2004. Address all correspondence and requests for reprints to: James G. Patton, Box 1820 Station B, Department of Biological Sciences, Vanderbilt University, Nashville, Tennessee 37235. E-mail: james.g.patton@ vanderbilt.edu or [email protected]. This work was supported by National Institutes of Health (NIH) Grant DK035592. R.C.C.R. was supported by NIH Grant 5T32 GM0347.

References 1. Black DL 2003 Mechanisms of alterative pre-messenger RNA splicing. Annu Rev Biochem 72:291–336 2. Coulter LR, Landree MA, Cooper TA 1997 Identification of a new class of exonic splicing enhancers by in vivo selection. Mol Cell Biol 17:2143–2150 3. Smith CWJ, Valcarcel J 2000 Alternative pre-mRNA splicing: the logic of combinatorial control. Trends Biochem Sci 25:381–388 4. Fu X-D 1995 The superfamily of arginine/serine-rich splicing factors. RNA 1:663– 680 5. Hastings ML, Krainer AR 2001 Pre-mRNA splicing in the new millennium. Curr Opin Cell Biol 13:302–309 6. Ryther RCC, McGuinness L, Phillips III JA, Moseley CT, Magoulas CB, Robinson ICAF, Patton JG 2003 Disruption of exon definition produces a dominant-negative growth hormone isoform that causes somatotroph death and IGHD II. Hum Genet 113:140 –148 7. Procter A, Phillips III JA, Cooper D 1998 The molecular genetics of growth hormone deficiency. Hum Genet 103:255–272 8. De Vos A, Ultsch M, Kossiakoff A 1992 Human growth hormone and extracellular domain of its receptor: crystal structure of the complex. Science 255:306 –312 9. Hayashi Y, Yamamoto M, Ohmori S, Kamijo T, Ogawa M, Seo H 1999 Inhibition of growth hormone (GH) secretion by a mutant GH-I gene product in neuroendocrine cells containing secretory granules: an implication for isolated GH deficiency inherited in an autosomal dominant manner. J Clin Endocrinol Metab 84:2134 –2139 10. Lee M, Wajnrajch M, Kim S, Plotnick L, Wang J, Gertner J, Leibel R, Dannies P 2000 Autosomal dominant growth hormone (GH) deficiency type II: the Del32–71-GH deletion mutant suppresses secretion of wild-type GH*. Endocrinology 141:883– 890 11. McGuinness L, Magoulas C, Mathers K, Carmignac D, Manneville J-B, Christian H, Phillips III JA, Robinson ICAF 2003 Autosomal dominant growth hormone deficiency disrupts secretory vesicles: in vitro and in vivo studies in transgenic mice. Endocrinology 144:720 –731 12. Graves TK, Patel S, Dannies PS, Hinkle PM 2001 Misfolded growth hormone causes fragmentation of the Golgi apparatus and disrupts endoplasmic reticulum-to-Golgi traffic. J Cell Sci 114:3685–3694 13. Binder G, Ranke M 1995 Screening for growth hormone (GH) gene splice-site mutations in sporadic cases with severe isolated GH deficiency using ectopic transcript analysis. J Clin Endocrinol Metab 80:1247–1252 14. Cogan J, Ramel B, Lehto M, Phillips III J, Prince M, Blizzard R, de Ravel T, Brammert M, Groop L 1995 A recurring dominant negative mutation causes autosomal dominant growth hormone deficiency—a clinical research center study. J Clin Endocrinol Metab 80:3591–3595 15. Fofanova O, Evgrafov O, Polyakov A, Poltouraus A, Peterkova V, Dedov I 2000 Molecular heterogeneity of familial isolated growth hormone deficiency, type II: a novel IVS3⫹2 T-C splicing mutation in the GH-1 gene. Horm Res 53(Suppl 2):P1-187 16. Hayashi Y, Kamijo T, Yamamoto M, Ohmori S, Phillips III J, Ogawa M, Igarashi Y, Seo H 1999 A novel mutation at the donor splice site of intron 3 of the GH-I gene in a patient with isolated growth hormone deficiency. Growth Horm IGF Res 9:434 – 437 17. Missarelli C, Herrera L, Merica V, Carvallo P 1997 Two different 5⬘ splice site mutations in the growth hormone gene causing autosomal dominant growth hormone deficiency. Hum Genet 101:113–117 18. Phillips III J, Cogan J 1994 Genetic basis of endocrine disease. 6. Molecular basis of familial human growth hormone deficiency. J Clin Endocrinol Metab 78:11–16 19. Fofanova O, Evgrafov O, Polyakov A, Poltaraus A, Peterkova V, Dedov I 2003 A novel IVS2–2A⬎T splicing mutation in the GH-1 gene in familial isolated growth hormone deficiency type II in the spectrum of other splicing mutations in the Russian population. J Clin Endocrinol Metab 88:820 – 826 20. Millar D, Lewis M, Horan M, Newsway V, Easter T, Gregory J, Fryklund L,

2996

21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.

Endocrinology, June 2004, 145(6):2988 –2996

Norin M, Crowne E, Davies S, Edwards P, Kirk J, Waldron K, Smith P, Phillips III J, Scanlon M, Krawczak M, Cooper D, Procter A 2003 Novel mutations of the growth hormone 1 (GH1) gene disclosed by modulation of the clinical selection criteria for individuals with short stature. Hum Mut 21:424 – 440 Moseley C, Mullis P, Prince M, Phillips III J 2002 An exon splice enhancer mutation causes autosomal dominant GH deficiency. J Clin Endocrinol Metab 87:847– 852 Cogan J, Prince M, Lekhakula S, Bundey S, Futrakul A, McCarthy E, Phillips III JA 1997 A novel mechanism of aberrant pre-mRNA splicing in humans. Hum Mol Genet 6:909 –912 McCarthy EMS, Phillips III JA 1998 Characterization of an intron splice enhancer that regulates alternative splicing of human GH pre-mRNA. Hum Mol Genet 7:1491–1496 Coolidge CJ, Patton JG 1995 Run-around PCR: a novel way to create duplications using polymerase chain reaction. Biotechniques 18:763–764 Brummelkamp T, Bernards R, Agami R 2002 A system for stable expression of short interfering RNAs in mammalian cells. Science 296:550 –553 Cartegni L, Wang J, Zhu Z, Zhang M, Krainer A 2003 ESEFinder: a Web resource to identify exonic splicing enhancers. Nucleic Acids Res 31:3568 –3571 Maquat LE 2002 NASty effects on fibrillin pre-mRNA splicing: another case of ESE does it, but proposals for translation-dependent splice site choice live on. Genes Dev 16:1743–1753 Cartegni L, Chew S, Krainer A 2002 Listening to silence and understanding nonsense: exonic mutations that affect splicing. Nat Rev Genet 3:285–298 Caputi M, Kendzior Jr R, Beemon K 2002 A nonsense mutation in the fibrillin-1 gene of a Marfan syndrome patient induces NMD and disrupts an exonic splicing enhancer. Genes Dev 16:1754 –1759 Graveley BR, Hertel KJ, Maniatis T 1998 A systematic analysis of the factors that determine the strength of pre-mRNA splicing enhancers. EMBO J 17: 6747– 6756 Tian M, Maniatis T 1992 Positive control of pre-mRNA splicing in vitro. Science 256:237–240 Tian M, Maniatis T 1994 A splicing enhancer exhibits both constitutive and regulated activities. Genes Dev 8:1703–1712 Zhang MQ 1998 Statistical features of human exons and their flanking regions. Hum Mol Genet 7:919 –932 Wang LL, Worley K, Gannavarapu A, Chintagumpala MM, Levy ML, Plon SE 2002 Intron-size constraint as a mutational mechanism in RothmundThomson syndrome. Am J Hum Genet 71:165–167 Berget SM 1995 Exon recognition in vertebrate splicing. J Biol Chem 270: 2411–2414

Ryther et al. • Disruption of GH1 Enhancers in GHD Patients

36. Sterner DA, Carlo T, Berget SM 1996 Architectural limits on split genes. Proc Natl Acad Sci USA 93:15081–15085 37. Smith CWJ, Nadal-Ginard B 1989 Mutually exclusive splicing of a-tropomyosin exons enforced by an unusual lariat branch point location; implications for constitutive splicing. Cell 56:749 –758 38. Smith CWJ, Porro EB, Patton JG, Nadal-Ginard B 1989 Scanning from an independently specified branch point defines the 3⬘ splice site of mammalian introns. Nature 342:243–247 39. Elbashir S, Harborth J, Lendeckel W, Yalcin A, Weber K, Tuschl T 2001 Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411:494 – 498 40. Elbashir S, Harborth J, Weber K, Tuschl T 2002 Analysis of gene function in somatic mammalian cells using small interfering RNAs. Methods 26:199 –213 41. Zamore PD 2001 RNA interference: listening to the sound of silence. Nat Struct Biol 8:746 –750 42. Hammond S, Caudy A, Hannon GJ 2001 Post-transcriptional gene silencing by double-stranded RNA. Nat Rev Genet 2:110 –119 43. Binder G, Keller E, Mix M, Massa G, Stokvis-Brantsma W, Wit J, Ranke M 2001 Isolated GH deficiency with dominant inheritance: new mutations, new insights. J Clin Endocrinol Metab 86:3877–3881 44. Drake W, Howell S, Monson J, Shalet S 2001 Optimizing GH therapy in adults and children. Endocr Rev 22:425– 450 45. Romijn J, Smit J, Lamberts S 2003 Intrinsic imperfections of endocrine replacement therapy. Eur J Endocrinol 149:91–97 46. Monson J 2003 Long-term experience with GH replacement therapy: efficacy and safety. Eur J Endocrinol 148:S9 –S14 47. Tanaka T, Cohen P, Clayton P, Laron Z, Hintz R, Sizonenko P 2002 Diagnosis and management of growth hormone deficiency in childhood and adolescence. Part 2: growth hormone treatment in growth hormone deficient children. Growth Horm IGF Res 12:323–341 48. Wit J 2002 Growth hormone therapy. Best Pract Res Clin Endocrinol Metab 16:483–503 49. Radetti G, Buzi F, Paganini C, Pilotta A, Felappi B 2003 Treatment of GHdeficient children with two different GH doses: effect on final height and cost-benefit implications. Eur J Endocrinol 148:515–518 50. Bramnert M, Segerlantz M, Laurila E, Daugaard J, Manhem P, Groop L 2003 Growth hormone replacement therapy induces insulin resistance by activating the glucose-fatty acid cycle. J Clin Endocrinol Metab 88:1455–1463 51. Miller V, Xia H, Marrs G, Gouvion C, Lee G, Davidson B, Paulson H 2003 Allele-specific silencing of dominant disease genes. Proc Natl Acad Sci USA 100:7195–7200

Endocrinology is published monthly by The Endocrine Society (http://www.endo-society.org), the foremost professional society serving the endocrine community.