Phenotypic Mutation 'triaka' (pdf version)
Alleletriaka
Mutation Type missense
Chromosome4
Coordinate11,379,300 bp (GRCm38)
Base Change T ⇒ C (forward strand)
Gene Esrp1
Gene Name epithelial splicing regulatory protein 1
Synonym(s) 2210008M09Rik, Rbm35a
Chromosomal Location 11,331,933-11,386,783 bp (-)
MGI Phenotype Mice homozygous for an ENU-induced allele exhibit hyperactivity and circling with no detectable hearing deficits.
Accession Number

NCBI RefSeq: NM_194055; MGI: 1917326

Mapped Yes 
Amino Acid Change Methionine changed to Valine
Institutional SourceBeutler Lab
Ref Sequences
M161V in Ensembl: ENSMUSP00000103947 (fasta)
Gene Model not available
SMART Domains

DomainStartEndE-ValueType
signal peptide 1 20 N/A INTRINSIC
Blast:EXOIII 29 192 N/A BLAST
RRM 226 298 2.6e-2 SMART
RRM 327 402 1.75e-5 SMART
low complexity region 420 434 N/A INTRINSIC
RRM 446 521 1.03e-2 SMART
low complexity region 542 556 N/A INTRINSIC
Predicted Effect probably benign

PolyPhen 2 Score 0.047 (Sensitivity: 0.95; Specificity: 0.82)
(Using Ensembl: ENSMUSP00000103947)
Phenotypic Category behavior/neurological, hearing/vestibular/ear, nervous system
Penetrance unknown 
Alleles Listed at MGI

All alleles(1) : Gene trapped(1)

Lab Alleles
AlleleSourceChrCoordTypePredicted EffectPPH Score
IGL01320:Esrp1 APN 4 11384374 missense possibly damaging 0.47
IGL02251:Esrp1 APN 4 11361202 missense probably damaging 1.00
IGL02669:Esrp1 APN 4 11386324 missense possibly damaging 0.61
Barley UTSW 4 11365205 missense probably benign 0.13
R1109:Esrp1 UTSW 4 11365205 missense probably benign 0.13
R1531:Esrp1 UTSW 4 11379375 missense probably damaging 0.98
R2189:Esrp1 UTSW 4 11357603 missense probably benign 0.05
R2255:Esrp1 UTSW 4 11365211 missense probably benign 0.21
R5919:Esrp1 UTSW 4 11344146 missense probably damaging 0.96
R5924:Esrp1 UTSW 4 11361174 missense probably damaging 1.00
R6042:Esrp1 UTSW 4 11357580 missense possibly damaging 0.93
Mode of Inheritance Autosomal Recessive
Local Stock Sperm, gDNA
MMRRC Submission 032823-MU
Last Updated 03/29/2017 2:07 PM by Katherine Timer
Record Created unknown
Record Posted 10/28/2013
Other Mutations in This Stock Stock #: 2107 Run Code:
Validation Efficiency: 79/99

GeneSubstitutionChr/LocMutationPredicted EffectZygosity
Afmid T to A 11: 117,835,561 N198K probably damaging Homo
Fam184a T to C 10: 53,641,057 E977G probably damaging Homo
Mypn C to T 10: 63,203,751 possibly damaging Homo
Nme7 T to A 1: 164,345,353 I211N possibly damaging Het
Tmod4 G to A 3: 95,130,168 probably null Homo
Wfs1 C to T 5: 36,967,273 R758H probably damaging Het
Zfp112 T to C 7: 24,126,841 C749R probably damaging Het
Phenotypic Description
The triaka phenotype was initially identified among G3 mice in an in vivo screen for mutants with defects in natural killer (NK) cell and CD8+ T cell cytotoxicity (In Vivo NK Cell and CD8+ T Cell Cytotoxicity Screen).  The original triaka stock displayed a defect in NK cell cytotoxicity as well as hyperactivity and circling behavior. The NK cell and neurobehavioral phenotypes segregated independently in subsequent breeding; the designation triaka was kept for the neurobehavioral phenotype.
 
Triaka mice displayed hyperactivity and circling behavior of variable intensity. The circling phenotype was incompletely penetrant among homozygous mutants.  The phenotype may be more penetrant on a mixed C3H x C57BL/6J background than on a pure C57BL/6J background.  Triaka mice displayed normal hearing in tests for auditory brain stem response (ABR) and distortion product otoacoustic emissions (DPOAE).
Nature of Mutation
The triaka mutation was mapped based on the hyperactivity and circling phenotype to proximal Chr. 4 with a peak LOD of 2.71 at D4mit235.  Fine mapping defined a critical region from the centromere to 13.7 Mbp.  Chd7 was identified as a candidate in the critical region because mice with Chd7 mutations exhibit hyperactivity, circling behavior, ear defects, and tail-kinks.  However, no mutation was found after 100% coverage of the gene (coding region and 5’ UTR) by DNA sequencing.  Whole genome SOLiD sequencing of a homozygous triaka mouse revealed an A to G transition at position 828 of the Esrp1 mRNA (NM_194055, GRCm38), located within the critical region, for which 86.3%, 74.9%, and 62.3% of coding/splicing sequence was covered at least 1X, 2X, or 3X, respectively. Validation sequencing of the critical region covered all nucleotides for which discrepancies were seen at 1x or greater coverage, with 31 of 41 discrepancies successfully processed. The Esrp1 mutation is located in exon 4 of 16 total exons. Multiple Esrp1 transcripts generated by alternative splicing are annotated in the Ensembl (7) and Vega (6) databases.
813 CTGGATGTCGCAGCGATGACAGAAAGCCTGAAC
156 -L--D--V--A--A--M--T--E--S--L--N-

The mutated nucleotide is shown in red within the Esrp1 mRNA sequence, and results in a methionine to valine substitution at amino acid 161 of the ESRP1 protein.

Protein Prediction
ESRP1 and ESRP2 are RNA binding proteins with homology to heterogeneous nuclear ribonucleoprotein (hnRNP) F and hnRNP H1 (1).  The hnRNP family consists of 24 proteins that stably associate with high molecular weight nuclear RNA (heterogeneous nuclear RNA or pre-mRNA) and contain one or more RNA binding domains (2;3).  Several types of RNA binding domains are known, including the RNA recognition motif (RRM; also known as consensus RNA-binding domain, CS-RBD; and ribonucleoprotein motif, RNP), quasi-RRM (qRRM), and K homology (KH) motif. The hnRNP H subfamily consists of the four proteins hnRNP F, hnRNP H1, hnRNP H2, and hnRNP 2H9, which contain two or three qRRMs. ESRP1 and ESRP2 each contain three domains with similarity to the qRRMs of hnRNP F and hnRNP H1 (Figure 1). However, ESRP1 and ESRP2 are not classical hnRNPs because they do not copurify with hnRNA. ESRP1 and ESRP2 are 51.6% identical and 63.7% similar overall. Both proteins are conserved in vertebrates and invertebrates, with RRM1 displaying the highest level of conservation between ESRP1 and ESRP2 in examined species (mouse, human, chicken, flies, worms).
 

Figure 1. Domain structures of ESRP1 and ESRP2 (mouse).  The amino acid sequences of motifs RNP1 and RNP2 are shown for each of the RRMs of ESRP1, with the residues matching the consensus sequence indicated in a bold font.  None of the three key RNA-binding aromatic residues within RNP1 ([F/Y]3, [F/Y]5) and RNP2 ([F/Y]2) (highlighted in blue) are conserved in the RRMs of ESRP1. The position of the triaka mutation is shown with a red asterisk.

 

 

 

Figure 2. NMR structure of hnRNP F qRRM1.  Helix α3 and a β-hairpin consisting of strand β3’ and β3” are additional secondary structural elements not found in classical RRMs.  α3 interacts with the surface of the β-sheet, preventing RNA recognition in the prototypical manner.  Instead, residues in the β-hairpin, helix α3, and in the β1/α1 and β2/β3 loops mediate RNA binding.  Alanine mutations of Y82 and W20 completely abolish RNA binding.  Model produced using UCSF Chimera based on 2HGL, Dominguez and Allain, Nucleic Acids Res. 34, 3634-45 (2006). Click on the 3D structure to  view it rotate.

 

The RRM, the most common RNA binding domain, is an approximately 90 amino acid domain containing two short, conserved RNA-binding sequences designated RNP1 and RNP2 (4;5).  The RNP consensus sequence or RNP1 is located near the center of the RRM and consists of eight residues that are mainly aromatic and positively charged, and follows the pattern [K/R]-G-[F/Y]-[G/A]-[F/Y]-[V/I/L]-X-[F/Y].  RNP2 is a sequence of six amino acids near the N-terminus of the domain, [I/V/L]-[F/Y]-[I/V/L]-X-N-L.  Determination of the three dimensional structures of more than 40 isolated RRMs revealed that the RRM folds into an αβ sandwich structure with β1-α1-β2-β3-α2-β4 topology [reviewed in (4;5)].  The β strands form an antiparallel β-sheet that makes the primary contacts with RNA, and two α-helices pack against one surface of the β-sheet.  RNP1 and RNP2 motifs are located in the central strands (β3 and β1, respectively) of the β-sheet.  The classic RRM-nucleic acid binding mode involves three key aromatic residues within RNP1 ([F/Y]3, [F/Y]5) and RNP2 ([F/Y]2), in addition to other positively charged residues on the surface of the β-sheet (4;5).  These positively charged and aromatic residues mediate RNA binding through hydrogen bonding and base stacking interactions.  Other less conserved residues apart from the amino acids at the surface of the β-sheet can also contribute or completely mediate RNA binding.  In particular, loops connecting β-strands and α-helices can be crucial for nucleic acid recognition. Loops β1/α1, β2/β3, and α2/β4 have been shown to participate in RNA binding in several RRM- or qRRM-containing proteins, including RBMY (6), Fox-1 (7), and hnRNP F (8).  Some RRMs have been shown to mediate interactions with proteins, such as those of PTB and p14, a component of the human U2 snRNP (9;10).
 
The qRRMs of hnRNP H family members specifically recognize poly(G) sequences abundant in both DNA and RNA.  In DNA, these poly(G) sequences are mainly located in telomeres where they often form G-quadruplex structures (11).  In RNA, poly(G) sequences are important for 5’ splice site recognition.  They also frequently function as splicing regulatory elements in introns or exons (12-14), and are abundant downstream from mammalian polyadenylation signals (15).  qRRMs of hnRNP H family proteins were distinguished from classical RRMs because although they bind to RNA, their RNP1 and RNP2 motif sequences deviate from the consensus (16).  In particular, the aromatic Phe/Tyr residues involved in RNA binding are poorly conserved in hnRNP H family members.  Analysis of the NMR structure of the three qRRMs of hnRNP F demonstrated that they adopt a classical RRM fold (8).  However, additional secondary structural elements in the form of a small β-hairpin between helix α2 and strand β4, and an α-helix C-terminal to strand β4, are also present in the two N-terminal qRRMs (qRRMs 1 and 2) of hnRNP F (Figure 2).  qRRMs 1 and 2, but not qRRM3, mediate RNA binding by hnRNP F, but interactions between the extra C-terminal helix and amino acids on the surface of the β-sheet mask the RNA binding site used by classical RRMs.  Instead, qRRM1 and qRRM2 use positively charged or aromatic residues in the extra β-hairpin and C-terminal helix, and in the β1/α1 and β2/β3 loops, to interact with RNA.  These include the critical residues W20 and F120 in the β1/α1 loop, and Y82 and Y180 in the β-hairpin of qRRM1 and qRRM2, respectively (8), which are conserved in qRRM1 and qRRM2 of ESRP1 (1).  The RRMs of ESRP1 are 41%, 50%, and 37% identical to those of hnRNP F, and 41%, 49%, and 37% identical to those of hnRNP H1 (mouse sequences).  No 3D structures of ESRP1 or its RRMs have been reported.
 
In general, RRMs are found in multiple copies within a protein, as is the case for ESRP1, which may allow continuous recognition of a long nucleotide sequence and increase binding affinity (4).  hnRNPs often contain one or more protein-protein interaction domains, such as proline-rich, glycine-rich, ‘serine, arginine, glycine, tyrosine-rich’ (SRGY), and acidic domains.  Members of the hnRNP H family each have a region rich in glycine, tyrosine, and arginine between the C-terminal qRRMs, as well as a C-terminal glycine- and tyrosine-rich domain (16).  In addition to three RRMs, ESRP1 is predicted by the SMART program to possess an N-terminal signal peptide; the significance of this feature is unknown.  No signal peptide is predicted for ESRP2.
 
Multiple isoforms of ESRP1 are generated by alternative splicing, with some reportedly displaying greater activity in assays for splicing of a reporter minigene dependent on FGFR2 intron 8 sequences (1) (see Background).  A 606 amino acid isoform encoded by a transcript lacking exons 14 and 15 displayed reduced splicing activity compared to the full length protein containing 681 amino acids. 
 
The triaka mutation is an M to V substitution at amino acid 161, which occurs in a region of ESRP1 for which no functional domains have been identified.  The mutated methionine is conserved in ESRP1 sequences of mouse, rat, pig, chimpanzee, human, and zebrafish.
Expression/Localization
Epithelial cell-specific expression of Esrp1 mRNA in diverse tissues and organs was reported for postnatal day 1 (P1) and adult mouse tissue sections as detected by in situ hybridization (1).  Skin, gastrointestinal, and olfactory epithelia had strong expression.  Esrp1 mRNA was also detected in the ear (including in the Organ of Corti) at P1, and in epithelia of the developing brain at embryonic day 13.5 (E13.5) and P0 (1;17)
 
Esrp1 and Esrp2 expression is developmentally regulated.  Both Esrp1 and Esrp2 mRNA were expressed more abundantly in mouse placenta than embryo tissue at midgestation (E9.5 and E11.5) (18).  In addition, Esrp1 was ubiquitously expressed in the embryo at E5.5 and E6.5, but became restricted to the chorion and definitive endoderm starting at E7.5 (18;19).  This endodermal expression was subsequently maintained in most derivatives of the endoderm, including the salivary gland, lung epithelium, pancreas, and intestines (18). Esrp1 was also found in the the nephrogenic cord (precursor to metanephric kidney and urogenital sinus), and developing tooth buds of E8.5-E10.5 mouse embryos (18;20)
 
The subcellular localization of ESRP1 has not been reported.  However, two rabbit polyclonal ESRP1 antibodies displayed nuclear but not nucleolar staining in two human cell lines, U-2OS and A-431 (Novus Biologicals, NBP1-82201 and NBP1-82202).  This staining pattern is typical of hnRNPs, which usually show general nucleoplasmic localization with little or no staining in nucleoli or in the cytoplasm when visualized with immunofluorescent microscopy (3).  Most of the nuclear signal is thought to represent hnRNPs bound to nascent RNA polymerase II transcripts, with the remaining signal from hnRNPs bound to transcripts that are being processed or mature transcripts awaiting transport to the cytoplasm.  hnRNPs are not usually found in nuclear “speckles,” interchromatin granules where small nuclear ribonucleoprotein particles (snRNPs), key components of the spliceosome, are concentrated (21).
Background
Alternative splicing is the process by which exons are differentially incorporated (and introns are removed) to produce two or more distinct mature mRNAs from a single pre-mRNA, and represents one mechanism for generating protein diversity in complex organisms (22).  In a genome-wide microarray study monitoring splicing at every exon-exon junction, more than 74% of human multi-exon genes were alternatively spliced (23).
 
Cells control constitutive and alternative splicing by regulating the assembly of the pre-mRNA splicing machinery, a giant macromolecular complex (containing 100-200 proteins and approximately 3 megadaltons in size) collectively known as the spliceosome (24).  The U1, U2, U4/U6, and U5 small nuclear ribonucleoprotein complexes (snRNPs), named for the uridine-rich small nuclear RNAs (U snRNAs) they contain, are the key subunits of the spliceosome, associating/dissociating with the pre-mRNA substrate at discrete steps of the spliceosome cycle to mediate intron removal and exon joining.  In addition to the U2-dependent major spliceosome, metazoan cells contain a U12-dependent minor spliceosome responsible for removal of approximately 0.25% of all introns (25).  Intron excision and joining of flanking exons is directed by four required cis-acting RNA elements within the intron that are recognized by base pairing with the U snRNAs (26).  The 5’ splice site (donor site) marks the exon/intron junction at the 5’ end of the intron, and consists of nine bases including an invariant GU dinucleotide.  Near the 3’ end of the intron are the three other elements: a branch point adenosine (A) residue, polypyrimidine tract, and the 3’ splice site (acceptor site) consisting of about 15 bases including an invariant AG dinucleotide at the intron/exon junction.  The spliceosome catalyzes the two transesterification steps of the splicing reaction (Figure 3) (22).  In the first step, the 2’-OH group of the branch point A attacks the phosphate at the 5’ splice site, ligating the first nucleotide of the intron and the branch point A to form a lariat structure, and cleaving the 5’ exon from the intron.  In the second step, the 3’-OH of the detached exon attacks the phosphate at the 3’ end of the intron, ligating the two exons and releasing the intron, still in the form of a lariat.
 

Figure 3. The two transesterification steps of the splicing reaction. The first step yields the free 5’ exon and the intron/3’ exon in a lariat structure.  The products of the second step are the ligated exons and the intron lariat, which is subsequently degraded.

 

In mammalian splice sites, the sequences adjacent to the invariant dinucleotides are not strictly conserved, with humans having a median splice site consensus value of 82 and 80 for 5’ and 3’ splice sites, respectively (100 being perfect consensus and 0 being worst consensus based on the relative contributions of each position compiled for thousands of introns) (27).  Splice site choice is controlled by assembly of the spliceosome.  The degree to which splice sites conform to the best consensus sequence determines their efficiency in recruitment and assembly of the spliceosome, and thus their intrinsic capacity to direct splicing.  However, additional elements modulate splice site selection, activating or repressing certain splice sites.  ESEs and ISEs (exonic and intronic splicing enhancers) and ESSs and ISSs (exonic and intronic splicing silencers) are short sequences within the pre-mRNA that recruit RNA binding proteins to promote or repress spliceosome assembly at overlapping or adjacent sites.  Splicing silencers and enhancers are important in the regulation of both constitutive and signal-induced splicing.  Whereas the SR proteins typically bind to splicing enhancers and stimulate the binding of U2AF, U2 snRNP, or U1 snRNP (28), the family of hnRNP proteins bind to splicing silencers and repress specific splice sites (2).
 
More than half of the major hnRNPs have been demonstrated to play a role in splicing, one which is typically negative.  Study of these proteins has revealed several strategies utilized by hnRNPs to inhibit the splicing reaction, either constitutively or at specific splice sites (i.e. splice site selection).  hnRNP binding can directly block the binding of splicing factors, spliceosome components, or positive splicing regulators to the pre-mRNA.  For example, the binding of hnRNPA1 to an ESE in exon 2 of the tat transcript of HIV-1 interferes with recruitment of the U2 snRNP (29-32) and the SR protein SC35 (33).  hnRNP I (also known as polypyrimidine tract binding protein, PTB) can also interfere with the binding of U2AF at the polypyrimidine tract of α-tropomyosin pre-mRNA (34;35).  hnRNPs can cooperatively bind to adjacent sites on the pre-mRNA, creating a zone of local repression.  This may occur when one high affinity site is flanked by several lower affinity sites, allowing propagative binding of the hnRNP over a distance.  Spliceosome assembly at splice sites distant from the high affinity site can then be blocked (36).  Rather than physically blocking their binding, another means by which hnRNPs may repress splicing is by interfering with interactions between spliceosome components across introns or exons.  PTB binding to ISEs in the c-src pre-mRNA does not prevent binding of either the U1 snRNP or U2AF (a key factor in spliceosome assembly; recognizes the 3’ splice site and polypyrimidine tract), but instead prevents the U1 snRNP from making productive cross-intron interactions with U2AF (37).  Finally, hnRNPs bound to different introns have been proposed to interact with each other, thereby looping out the intervening RNA sequence and promoting exon skipping (38;39).  Whether hnRNPs function to promote or inhibit particular splicing events appears to be context-dependent, as has been demonstrated for hnRNP H [(40) and references therein].
 
ESRP1 and ESRP2 were identified in a genome-wide high-throughput cDNA expression screen for factors that promote the epithelial pattern of FGFR2 splicing (1).  FGFR2 transcripts undergo one of two mutually exclusive alternative splicing events that result in inclusion in the mature mRNA of either exon 8, also known as exon IIIb, or exon 9, also known as exon IIIc.  Exons IIIb and IIIc encode the C-terminal half of the membrane-proximal immunoglobulin-like domain of FGFR2, and their differential usage results in proteins that exhibit distinct ligand binding properties (41;42).  The FGFR2-IIIb splice variant is expressed exclusively in epithelial cells, whereas the FGFR2-IIIc variant is mesenchymal (43); this compartmentalization is essential for proper regulation of cell proliferation and differentiation during development (44-50).
 
Multiple cis elements within or flanking exons IIIb or IIIc have been demonstrated to control expression of the epithelial-specific FGFR-IIIb by recruiting several splicing regulatory factors (Figure 4).  These include the upstream intronic splicing silencer (UISS), downstream intronic splicing silencer (DISS), and exon IIIc ESSs that bind to PTB (51-54); ISE-1 that binds to TIA-1 (55); and the exon IIIb ESS bound by hnRNP A1 (56).  ISE-2 and ISAR (intronic splicing activator and repressor) have been shown to base pair with each other to form a stem structure that contributes to inclusion of exon IIIb and exclusion of exon IIIc (57-59).  The Fox family of splicing regulators silences exon IIIc by binding to UGCAUG elements upstream and within exon IIIc (60).  hnRNP F and hnRNP H also silence exon IIIc through interactions with GGG motifs in exon IIIc that function as ESSs (40).  An element designated ISE/ISS-3 activates the epithelial splicing pattern of FGFR2 in epithelial cell types by both promoting splicing to exon IIIb and repressing splicing to exon IIIc, but has no effect on splicing when expressed in mesenchymal cell types, and appears to function independently of the other regulatory cis elements (61).  In order to repress exon IIIc, ISE/ISS-3 requires the use of G as the branch point nucleotide during the first step of splicing.

Figure 4. Alternatively spliced variants of FGFR2 and RNA cis elements shown to regulate splicing.  The portion of the protein encoded by exons IIIb or IIIc is shown at the top.  The pre-mRNA with regulatory elements and splicing events that lead to production of FGFR2-IIIb in epithelial cells or FGFR2-IIIc in mesenchymal cells is below.  TM, transmembrane domain; TK, tyrosine kinase domains.  Hatched boxes represent intronic and exonic splicing enhancers and silencers.  Figure adapted from reference (61) Figure 1.

 

 In 293T cells that normally express the mesenchymal FGFR2-IIIc isoform, ectopic expression of ESRP1 or ESRP2 induced a substantial switch to expression of the exon IIIb-containing isoform (1).  Conversely, RNAi-mediated knockdown of ESRP1, but not ESRP2, in a prostate epithelial cell line that expresses FGFR2-IIIb caused a switch from the epithelial to mesenchymal isoform, demonstrating that FGFR2-IIIb expression requires ESRP1.  When overexpressed in 293T cells, ESRP1 and ESRP2 are found bound to ISE/ISS-3 by UV crosslinking, suggesting that they mediate their effects through this cis element. 
 
ESRP1 and ESRP2 appear to have broad splicing regulatory function that controls the epithelial-to-mesenchymal transition (EMT).  Induction of an EMT in a mammary epithelial cell line resulted in downregulation of ESRP1 and ESRP2 (62).  Microarray analysis of exon junctions of alternatively spliced human genes in cells overexpressing ESRP1 or in which ESRP1 and ESRP2 were knocked down revealed an extensive ESRP-regulated splicing network; loss of this network induced cellular phenotypic changes characteristic of the EMT (62).  The epithelial-specific targets of ESRP1 and ESRP2 include CD44 (63), p120-Catenin (CTNND1) (64), and hMena (ENAH), as well as others that function in cell-cell adhesion, cell motility, and GTPase signaling (62;65;66).  ESRP1 and ESRP2 bound to hexamers containing repeats of UGG or GUU, promoting exon inclusion when UGG-rich motifs were present downstream of the enhanced exon, and exon exclusion when UGG-rich motifs were present within silenced exons (66).  Several transcriptional repressors reportedly downregulate Esrp1 expression during EMT, including δEF1, SIP (67), and Snail (68), whereas the transcription factors OVOL1 and OVOL2 promote Esrp1 expression during the mesenchymal to epithelial transition (MET) (69)
 
During EMT, epithelial cells undergo changes in protein expression that promote migratory and invasive behavior while inhibiting cell polarization and adhesion.  Thus, EMT is thought to play an important role in tumor invasion and metastasis (70).  The downregulation of ESRP1 leading to expression of the mesenchymal pattern of alternative splicing events is essential for EMT and cancer progression (63;67).  In particular, expression of the standard isoform of CD44 (CD44s) was shown to occur upon downregulation of ESRP1 and was required for cells to undergo EMT and for the formation of breast tumors with EMT characteristics in mice (63).  CD44s expression appears to be correlated with cancer progression in human breast tumor samples (63), although others have reported conflicting findings (71).  ESRP1 has also been proposed to function as a tumor suppressor in colon cancer cells by controlling translation via the 5’ UTR of mRNAs (72).  Notably, ESRP1 has recently been implicated in pluripotency regulation of stem cells through inhibition of Oct4, Nanog, and Sox2 expression, in part by decreasing their polysome loading (73).
 
Mutations in the cis- and trans-acting factors that control splicing are a primary cause of many diseases (74).  The effects of cis- versus trans-acting mutations can be quite different: in the case of cis-acting mutations, only the mutated allele is affected, but when a trans-acting factor is mutated, the expression of many genes may be affected.  Recent estimates are that 50-60% of disease-causing mutations disrupt cis-acting splicing regulatory elements (75;76).  Many of these elements are exonic, mutations of which are typically assumed to directly affect protein function rather than splicing and subsequent expression and isoform expression.  For example, one fourth of synonymous mutations in exon 12 of the cystic fibrosis transmembrane conductance regulator (CFTR) gene result in exon skipping and an inactive CFTR protein, causing cystic fibrosis (OMIM #219700) (77;78).  Trans-acting mutations affect the splicing machinery or its regulatory proteins and cause disease.  An example is spinal muscular atrophy (OMIM #253300), caused by mutations in the survivor of motor neuron-1 (SMN1) gene, which is required for assembly of snRNPs (79).  Retinitis pigmentosa (OMIM #600138, #600059, #601414) is caused by mutations in pre-mRNA processing factor gene homologues PRPF31, PRPF8, and PRPF3, encoding proteins required for assembly and function of the U4/U5/U6 snRNP (80).  Finally, both cis- and trans-acting splicing mutations can cause cancer (81-83).
Putative Mechanism
FGFR2 signaling is critical for normal mouse development, with null mutations causing embryonic lethality.  Isoform IIIb-deficient mutants die at birth with defects in multiple organs and tissues, whereas isoform IIIc-deficient mutants have defects in osteoblast and chondrocyte lineages resulting in dwarfism.  The triaka mice show none of these phenotypes, suggesting that other splicing factors can compensate for the loss of ESRP1 function in vivo.  FGFR2-IIIb signaling is required for inner ear morphogenesis (48), such that a mouse deficient in ear-specific FGFR2-IIIb would be expected to have profound deafness.  However, triaka mice display normal hearing in ABR and DPOAE tests.  Thus, it appears unlikely that defective FGFR2-IIIb signaling is responsible for the triaka phenotype.
 
It remains possible that the circling phenotype of triaka mice is caused by a central nervous system defect.  Indeed, impaired alternative splicing is responsible for many neurological diseases (84), but defects in known neuronal alternative splicing events have not been shown to cause circling in mice.  Furthermore, no gross morphological abnormalities were detected in the brains of triaka mice, based on neuron staining in sagittal sections.
Primers Primers cannot be located by automatic search.
Genotyping
Triaka genotyping is performed by amplifying the region containing the mutation using PCR, followed by sequencing of the amplified region to detect the single nucleotide insertion.
 
Primers
triaka (F): 5’- GCTTAATCGGAAGGCACAGTGACC -3’
triaka (R): 5’-CACACTTGGCAGGAACACTCTGAC -3’
 
PCR program
1) 95°C             2:00
2) 95°C             0:30
3) 56°C             0:30
4) 72°C             1:00
5) repeat steps (2-4) 29X
6) 72°C             7:00
7) 4°C              ∞
 
Primers for sequencing
triaka_seq(F): 5’- ACCCTAAGGTTCCTGGTTACTAGAG -3’
triaka_seq(R): 5’- CAAGGCATTTTTCTTGAGCCAG -3’
 
The following sequence of 625 nucleotides (from Genbank genomic region NC_­­­­­000070 for linear genomic sequence of Esrp1, sense strand) is amplified:
 
4722                                              gcttaatcg gaaggcacag
4741 tgaccatgtt actctttctg ttaaaatgag atggcttttg taaaactgct ctaggattta
4801 atgcaatatg gtggccaagg attaaaaata aagccaggcc acattgagtt taatatttct
4861 aacaactgtg gtaacatgtt aactatattc taccctaagg ttcctggtta ctagagaaat
4921 acggaccaga ctcgcctagt taaccacctc taccttgtcc tccctcagaa tgtgttactg
4981 cctgagtgct tctattcctt tttcgatctt cggaaagagt tcaagaagtg ctgcccgggc
5041 tctcccgata tcgacaagct ggatgtcgca gcgatgacag aaagtatcct ttcaggtcac
5101 ctaagagtgc ttcctgtccc agccttgcac tttgctttcc ttcctgtttg ttttttaggt
5161 tttggagaca gaacttggtc tggcctggaa ctcaggtagt ggaaggaggc taccttcaga
5221 cttacaggta tccccctgcc tcaggcctac acagtgctag aattaaatgt gttagccaca
5281 ggacctggct caagaaaaat gccttgaagt gtcagtgtgt gggtcagagt gttcctgcca
5341 agtgtg
 
Primer binding sites are underlined; sequencing primer binding sites are highlighted in gray; the mutated A is indicated in red.
References
Science Writers Eva Marie Y. Moresco
Illustrators Eva Marie Y. Moresco, Diantha La Vine
AuthorsPhilippe Krebs, Bruce Beutler