Original Article

Exon definitive regions for MPC1 microexon

splicing and its usage for splicing modulation
Eunjin Koh,1 Daye Shin,1 and Kyung-Sup Kim1
1Department of Biochemistry and Molecular Biology, Institute of Genetic Science, Yonsei University College of Medicine, Seoul 03722, Korea

Alternative splicing of microexons (3–30 base pairs [bp]) is cancers.9–15 Thus far, the regulatory mechanisms on microexon
involved in important biological processes in brain develop- splicing have been studied with multiple of 3 bp (3x bp) microexons.
ment and human cancers. However, understanding a splicing RBPs, which regulate alternative splicing, are involved in microexon
process of non-3x bp microexons is scarce. We showed that splicing in tissue and disease-specific manners. RBFOX and PTBP1
4 bp microexon of mitochondrial pyruvate carrier1 (MPC1) is proteins act as enhancers and suppressors of microexon splicing,
constitutively included in mRNA. Based on our studies with respectively.9 SRRM4, a brain-specific RBP, stimulates the inclusion
minigene and exon island constructs, we found the strong of microexon in the brain and various cancers.10,12,15,16 QUAKING
exon definition region in the proximal introns bordering (QKI) plays important roles in microglia homeostasis by regulating
MPC1 microexon. Ultimately, we defined a nucleotide frag- alternative splicing of Rho GTPase pathway-related microexons.14
ment from the 30 ss 67 bp of MPC1 microexon to the 50 ss Ubiquitously present RBPs, Srsf11, and Rnps1 identified by
consensus sequence, as a core exon island, which can concate- genome-wide CRISPR-Cas9, preferentially regulate neuronal micro-
nate its microexon and neighboring exons by splicing. Further- exons.13 In addition, cis elements that reside in flanking introns
more, we showed that insertion of the core exon island into a are important regulators of microexon splicing. Replacement of
target exon or intron induced skip the target exon or enhance purines in the upstream polypyrimidine tract with pyrimidines can
the splicing of an adjacent exon, respectively. Collectively, we recover the skipped microexons,5 and intronic splicing enhancer,
suggest that the exon island derived from MPC1 microexon “GGGGCUG,” located downstream of 50 ss activates microexon inclu-
modifies genuine splicing patterns depending on its position, sion through SF1 binding.17 The sequence at the branchpoint is also
thereby providing insights on strategies for splicing-mediated involved in microexon partial inclusion.18 Despite the importance of
gene correction. microexons in abnormal brain development and cancers, the mecha-
nism by which microexons are spliced and the functions of amino
acid sequences encoded by these microexons are largely unknown.
INTRODUCTION
Splicing of precursor mRNA (pre-mRNA) into mRNA is essential In contrast to 3x bp exons, aberrant splicing of non-3x exons can lead
for gene expression in eukaryotic cells. Splicing is governed by to the formation of premature termination codon (PTC) resulting in
many cis- and trans-elements, such as RNA-binding proteins degradation of transcripts through nonsense-mediated decay (NMD)
(RBPs) and heterogeneous nuclear ribonuclear proteins (hnRNPs). followed by compromised protein levels in the cell.19 In the same
Thus, its complexity is sufficient to explain that the misregulation light, a recent study demonstrated that the inclusion of a 20 bp micro-
of splicing leads to many abnormal cellular functions.1 In fact, aber- exon of Bak1 is a target of NMD that contributes to the reduction of
rant splicing has been implicated in many human diseases.2 Bak1 during brain development.11 However, studies on the splicing
mechanism and functions of non-3x microexons are still greatly
In contrast to common exons with sizes of approximately 130 base in need.
pairs (bp),3 microexons under 50 bp are easily skipped due to their
short length and lack of exonic splicing enhancers (ESEs), as well as MPC1 is a component of the mitochondrial pyruvate carrier (MPC)
spatial limitations for spliceosome assembly at 30 and 50 splice sites complex, which mediates important metabolic processes by trans-
(ss) to fulfill exon definition.4–6 Furthermore, genomic mapping porting cytosolic pyruvate into the mitochondria.20,21 Since the
and annotation analysis showed that the number of constitutively expression of MPC1 is essential for stabilizing the MPC complex,
spliced microexons markedly drops with exon size.7–9 Moreover, the regulation of MPC1 mRNA levels has been investigated, including
there are alternatively spliced (AS) microexons that are used in partic-
ular tissues such as the brain, heart, muscle, and pituitary gland,
although their relevance needs further investigation.7,9 Recently, the Received 17 October 2022; accepted 20 January 2023;
https://doi.org/10.1016/j.omtn.2023.01.010.
function of microexons and their splicing processes have gained sub-
Correspondence: Kyung-Sup Kim, Department of Biochemistry and Molecular
stantial attention because microexons exhibit important biological
Biology, Institute of Genetic Science, College of Medicine, Yonsei University, Seoul
functions despite their small size. The misregulation of microexons 03722, Korea.
is involved in abnormal brain development, autism, and various E-mail: kyungsup59@yuhs.ac

398 Molecular Therapy: Nucleic Acids Vol. 31 March 2023 ª 2023 The Author(s).
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
www.moleculartherapy.org

in our recent study.22 In particular, exon 2 of MPC1 is a very short Next, we reconfirmed whether the 4 bp MPC1 microexon is constitu-
non-3x bp microexon (4 bp) flanked with long introns on both sides. tively spliced. To exclude the possibility that microexon-skipped tran-
However, no study on MPC1 microexon has been conducted to our scripts may be targets of NMD, 786-O cells were pre-treated with
knowledge. NMD inhibitor, cycloheximide (CHX), and transcripts were exam-
ined by RT-PCR (Figure 1C). Noticeably, there is no detectable
Here, we demonstrated that a 4 bp microexon of MPC1 is constitu- microexon-skipped transcript even in the presence of CHX. These re-
tively included during splicing through exon definition at its proximal sults suggest that the reason why microexon-skipped transcripts are
flanking introns. We have defined a nucleotide fragment from the not detected is due to the absolute splice-in of the microexon, not
upstream 67 bp of MPC1 microexon to the downstream 6 bp, as a because of its clearance by NMD. In addition, the same results were
core exon island with a strong exon definition enough to splice observed when MPC1 was transcriptionally enhanced by the overex-
various microexon sequences into the final mRNA. We also presented pression of PGC-1a.22 The same experiment was done in several
promising gene-editing actions of this defined exon island based on other cell lines derived from mice and humans, and it was confirmed
various genetic models. that the microexon was spliced almost completely (Figures S2C and
S2D). Taken together, we could conclude that the microexon of
RESULTS MPC1 is highly conserved in mRNA under normal and transcription-
Microexon of MPC1 is constitutively included during the splicing ally active states.
process
We investigated the human genome and counted microexons under 30 ss of intron 1 of MPC1 is critical for the inclusion of its
27 bp in length (Figure 1A). The number of microexons of each length microexon
was counted based on the values of Chr_Accession, Exon_Start, To assess the potential role of flanking introns as cis elements for
Exon_End, and Exon_length_bp from the Gene_Table.xlsx spread- MPC1 microexon splicing, we generated a variety of MPC1 minigene
sheet file provided by Piovesan et al.23 Most microexons under constructs (Figure 2A). All the constructs were transfected into HeLa
12 bp have 3x bp microexons (77%). Noticeably, there are very limited cells, and total RNA prepared to perform quantitative real-time PCR.
numbers of non-3x bp microexons under 12 bp. Two genes (GRK6 MPC1 microexon minigene construct harboring the MPC1 genome
and SEPT7) have the shortest microexons with 2 bp in the human spanning exon 1 to exon 3, shortened by removing the middle part
genome. Based on the information of transcript variants in the of introns (MPC1-wt). The upstream or downstream intron of
NCBI gene web site (https://www.ncbi.nlm.nih.gov/gene), microexon exon 2 microexon was swapped with a foreign intron (MPC1 intron
in GRK6 is found in minor transcript at the second to last exon, which 3), which has well-conserved splice sites, to make Swap-1/3 or Swap-
produces the C-terminal variant. SEPT7 has AS microexon located in 2/3, respectively. As shown in Figure 2B, the Swap-1/3 completely lost
50 UTR, which does not affect amino acid sequence. Four and 5 bp microexons during splicing, while Swap-2/3 produced a considerable
microexons are found in eight genes (TNNI1, MPC1, SPINK2, amount of mRNA harboring microexons, indicating that upstream
DCTD, KSR1, TMEM237, SPARCL1, and SEPT7). Microexons of introns are critical for the splicing-in of the MPC1 microexon. There-
DCTD and SPARCL1 were AS in the 50 UTR in minor transcripts, fore, we next constructed intronic hybrid mutants, which the 50 or 30
not affecting the amino acid sequence. SEPIN7 has AS microexons region of intron 1 is replaced with the 50 and 30 region of intron 3,
at the second exon generating the alternative start codon resulting respectively, named Hybrid-3/1 and Hybrid-1/3. Notably, Hybrid-
in several N-terminal variants. Microexons of the remaining 3/1 stimulated microexon inclusion, whereas Hybrid-1/3 failed
genes (TNNI1, MPC1, KSR1, and TMEM237) and 7 bp microexon microexon splicing, indicating that the 30 ss region of the intron 1 is
of MUSK are in the coding region and did not show skipped critical for microexon splicing.
variants in the NCBI gene bank, so they were tested in this experi-
ment. In addition, it is observed that the microexon and its proximal Next, to determine the minimal 30 ss region of intron 1 required for
intron sequences of MPC1 are highly conserved for cross-species MPC1 microexon splicing, 329 bp of intron 1 in Hybrid-3/1 were
(Figure S1), and those of the four genes we have tested (TNNI1, serially deleted, as depicted in Figure 2D (top). The inclusion of mi-
KSR1, TMEM237, and MUSK) also showed high homology across croexon was maintained until the 30 ss region of the upstream intron
species. was removed up to 67 bp. However, the microexon was markedly lost
when the 30 region was reduced to 46 bp. These results indicate that
Alternative splicing of microexons contributes to isoform production; 67 bp of the 30 region of the upstream intron is a minimal region
however, aberrant splicing of non-3x microexons can cause a frame- required for the splicing of MPC1 microexon. As denoted in Figure 2D
shift, resulting in PTC formation followed by defects in gene expres- (bottom), 67 bp of the 30 region of the upstream intron has a pyrim-
sion through NMD.19 We amplified the region where microexons are idine-rich region between 67 and 43 (22 pyrimidines out of 25).
in the transcripts of the five genes and directly sequenced RT-PCR The further deletion to 23 bp completely abolished the splicing,
products. As shown in Figure 1B the sequences of each microexon where direct repeat (“TGTCTG”) is located. It would be further study
from five genes exhibited no overlapped peak by alternative spliced to investigate about RBP binding or RNA secondary structure at the
transcripts, indicating that no microexons of these genes were skipped 67 bp minimal region that shows very similar sequences between spe-
in the cell lines we tested. cies (Figure S1).

Molecular Therapy: Nucleic Acids Vol. 31 March 2023 399

 Molecular Therapy: Nucleic Acids

Figure 1. Microexon of MPC1 is constitutively included in mRNA during the splicing process
(A) Analysis of microexons less than 27 bp in length. The numbers of microexons were counted based on the Gene_table.xlsx provided by Piovesan et al.23 (B) Comparison of
mRNA reference sequences from NCBI vs. those of RT-PCR amplified products. The regions of microexons from MPC1 (4 bp), TNNI1 (4 bp), TMEM237 (5 bp), KSR1 (5 bp),
and MUSK (7 bp) are shaded (light blue). (C) RT-PCR analysis of MPC1 mRNA. “I” and “S” indicate microexon-included and skipped mRNA, respectively. Arrows indicate the
locations of primers used for the amplification of “I” and “S” (see Table S2). The specificity of the primers was validated (Figures S1A and S1B); 50 mM of cycloheximide (CHX)
was treated for 6 h before total RNA extraction. Parental 786-O cells and 786-O cells with lentiviral stable expressed PGC-1a were used.

If the exon island is defined as a DNA fragment containing the exon could play a role as exon island. To evaluate the exon island activity,
and its surrounding intron sequences, which has a strong exon defi- the reporter vector was constructed, in which an upstream exon con-
nition enough to concatenate neighboring exons by splicing, we taining ATG start codon, intron 3 of MPC1, and a downstream exon
investigated that MPC1 microexon with its flanking intron sequences were sequentially placed between the SV40 promoter and the EGFP

400 Molecular Therapy: Nucleic Acids Vol. 31 March 2023

www.moleculartherapy.org

F G

(legend on next page)

Molecular Therapy: Nucleic Acids Vol. 31 March 2023 401

 Molecular Therapy: Nucleic Acids

gene. MPC1 microexon with its flanking intron sequences (light precisely sequenced without overlapping with skipped transcripts.
green) were placed in the middle of intron of the reporter vector as Thus, we could conclude that the core exon island from 67 bp 30 ss
shown in Figure 2E. The 96 bp and 67 bp upstream introns along to 6 bp 50 ss of MPC1 microexon is sufficient for the exon definition
with 139 bp downstream intron showed full splicing activity as of MPC1 microexon in our experimental settings.
Hybrid-3/1 (67 bp) construct. Noticeably, no loss of splicing activity
was observed until the downstream intron was reduced to 6 bp, the Splicing process is maintained until MPC1 microexon length is
consensus 50 ss sequence in the intron. Based on these results, we reduced to zero
could conclude that the 67 bp 30 ss of the upstream intron and 6 bp In the human genome, we could not find the microexon with a length
50 ss of the downstream intron were sufficient for exon definition of 1 bp, although this might come from the recent technical limita-
recognizing 4 bp (“TACG”) as an exon. tion. These facts raised the question of whether the splicing efficiency
of the MPC1 microexon could be preserved when the length of the
The fact that non-3x microexons of four genes (TNNI1, KSR1, microexon is shortened to less than 4 bp. Therefore, we generated
TMEM237, and MUSK) were constitutively conserved in endogenous MPC1 microexon minigene constructs as illustrated in Figure 3A.
mRNA intrigued us to explore whether their proximal flanking in- Since the constructs with 4-, 2-, and 1-bp microexons were designed
trons also accurately splice respective microexons. Reporter con- to express EGFP in frame when its corresponding microexon is
structs of each microexon were designed in two forms, one expresses correctly spliced into the final mRNA their splicing activities were
EGFP when the microexon is spliced in, and the other expresses evaluated by EGFP fluorescence intensity, as described in materials
EGFP when the microexon is skipped (Figure 2F). Therefore, and methods. Compared with 4 bp (wt) MPC1 microexon, shortened
the EGFP intensity of MPC1 “splicing-in” reporter was the same as 2 bp or 1 bp microexon showed no marked difference in their splicing
that of “blank,” whereas its skipping reporter did not show any activ- activity, which indicates that these microexons were successfully
ity, indicating that the 30 ss and 50 ss of MPC1 microexon (239 bp) led spliced in mRNA (Figure 3B). RT-PCR and sequencing analysis
to the complete inclusion of microexons. On the contrary, nucleotide confirmed that 4-, 2-, and 1-bp microexon sequences were included
fragments of TNNI1 (468 bp), KSR1 (409 bp), TMEM237 (477 bp), in the mRNA (data not shown).
and MUSK (418 bp) are noticeably ineffective in microexon
splicing-in and showed significant levels of microexon skipping, In addition, it was intriguing to examine whether the splicing occurs
meaning that they might require other elements in addition to the when all the exon sequence is removed (0 bp) or when the consensus
proximal introns around the microexon for maximal inclusion of of 30 ss and 50 ss were compromised (2 bp) (Figure 3A). As shown in
their microexons during splicing. Figure 3C, there was no noticeable difference of the spliced transcripts
level in three minigene constructs, indicating that their splicing suc-
Next, we tried to ascertain whether the 30 ss (67 bp) and 50 ss (6 bp) of cessfully proceeded in all constructs. To further investigate whether
MPC1 microexon could act as the effective exon island in different the splicing of the RNA from these constructs is processed with
genomic contexts. Therefore, the 77 bp nucleotide fragment of two steps at the junction of the upstream and downstream intron
MPC1 core exon island was inserted into the middle of the first intron or both introns are removed at once, the lariats generated during
of EF1a and the second intron of rabbit HHB2; both are commonly the splicing process were examined with total RNA prepared from
used in conventional expression vectors, such as pEF1a-EGFP and HeLa cells transfected with minigene constructs (Figure 3D). We first
pSG5-EGFP, respectively (Figure 2G). RT-PCR products were analyzed the level of lariat “b,” which originated from the downstream
sequenced to confirm whether the 4 bp of microexon was spliced intron (Figure 3A). Four bp minigene construct produced lariat “b” as
into the final mRNA. Remarkably, 4 bp microexon sequences were expected, but 2 bp minigene construct in which 50 ss and 30 ss is

Figure 2. Inclusion of the microexon of MPC1 is facilitated by the 30 splice site of intron 1
(A) Schematic diagram explains the structure of MPC1 minigene constructs used in experiments. MPC1-wt minigene contains MPC1 microexon and its adjacent exon and
shortened flanking introns. Swap 1/3 and 2/3 constructs have intron 3 (688 bp) instead of intron 1 or intron 2, respectively. Hybrid-3/1 and 1/3 constructs are generated by
replacing the 50 or 30 regions of intron1 with those of intron 3 (purple or orange, respectively). All constructs have the following downstream sequences of EGFP and poly-A
after exon 3. (B) and (C) MPC1 microexon splicing activities. The level of the microexon-included mRNA was evaluated by quantitative real-time RT-PCR using E1E2-s and
E3-as. Real-time RT-PCR values of the EGFP region by EGFP1-s and EGFP1-as were used as the total transcript level for normalization of the spliced-in of microexon (see
Table S2 for primer sequences). (D) Intron 1 30 end was serially deleted from Hybrid-3/1 construct (top). The level of microexon inclusion was measured by quantitative real-
time PCR. The values were normalized in the same way as in (B) and (C). Sequence from 67 bp to 28 bp upstream of microexon was depicted (bottom). The pyrimidine
tract and the direct repeat of “TGTCTG” are denoted as black and red lines. (E) and (F) The level of microexon inclusion was measured by EGFP fluorescence with whole-cell
extracts of HeLa cells. For internal control of transfection efficiency, mCherry expression vector was co-transfected. EGFP intensities were divided by mCherry expression
level. (E) The constructs were generated by inserting the fragments harboring 4 bp MPC1 microexon and its adjacent 30 ss and 50 ss in different lengths into intron 3 (purple) of
the reporter vector (top). (F) Exon island activities of various microexons. The microexon islands of MPC1 (239), TNNI1 (468 bp), KSR1 (409 bp), TMEM237 (477 bp), and
MUSK (418 bp) were amplified by PCR and inserted into the middle of intron in the reporter construct. (G) The 30 ss (67 bp) and 50 ss (6 bp) directed 4 bp MPC1 microexon into
mRNA. The 77 bp-length nucleotide fragments harboring a 4 bp MPC1 microexon was inserted in the intron of either pEF1 a-EGFP or pSG5-EGFP plasmids. Constructs
were transfected into the HeLa cells, followed by RT-PCR and sequence analysis. The inserted sequences of 4 bp, TACG, are shaded in red. Data are representative of three
experiments. *p < 0.05 and **p < 0.01 by Student’s two-tailed t test. All bar graphs are plotted as mean ± SE.

402 Molecular Therapy: Nucleic Acids Vol. 31 March 2023

www.moleculartherapy.org

A B

D E F

G H I

Figure 3. MPC1 microexon splicing efficiently occurs when its length is shortened to 2, 1, and 0 bp
(A) MPC1 minigene constructs containing various lengths of microexons. The structure of MPC1 minigene is same as MPC1-wt in Figure 2A except the microexon sequences
(capital letters). The dotted line within intron 1 indicates a high GC region. The absence of exon and disruption of splicing sites are depicted as 0 and 2 bp. All constructs
were designed to express EGFP when the microexon is correctly spliced into the final mRNA. (B) Splicing activities of MPC1 minigene constructs containing microexons of
various lengths. EGFP intensities were measured with whole-cell extracts of HeLa cells transiently transfected with each minigene construct. Values are normalized by the
level of mCherry fluorescence expression considering transfection efficiency. (C) Comparison of the quantified spliced products among 4, 0, 2 bp minigene constructs by
real-time RT-PCR using primers UE-s and EGFP-as (Table S2). The real-time RT-PCR values of the EGFP region were used for normalization of the spliced-in of microexon.
(D) Agarose gel analysis of lariats by RT-PCR. Lariats a and c were amplified from Hybrid-3/1 minigene-transfected cells and lariat b from MPC1-wt. The lariats derived from
MPC1 minigene construct are depicted as “a,” “b,” and “c” in (A). Each lariat derived from 4 bp, 0 bp, 2 bp minigene construct was amplified by PCR and analyzed by
agarose gel as described in the materials and methods. (E) The sequence around branch points in 30 ss of intron 1 and 2. Branchpoints (blue capital letters) are determined by
sequencing 20 plasmids cloned from amplified lariat. The major branch sites of intron 1 were A at 68 and 64 and minor site was G at 57. The major branchpoints of intron
2 were A at 41 and 25. (F) MPC1 minigene constructs of splice sites of 30 ss and 50 ss mutated to AG>AA and GU>AU, respectively. (G) to (I) Real-time RT-PCR analysis of
transcripts synthesized with total RNA of HeLa cells transiently transfected with reporter constructs. Primers shown as arrows are E1E2-s and E3 primers for (G), E1E3-s1 and
E3-as primers for (H), and I3-s and E3-as are used in (I). Primer sequences are listed in Table S2. Data are representative of three experiments. All bar graphs are plotted as
mean ± SE. *p < 0.05 by Student’s two-tailed t test.

Molecular Therapy: Nucleic Acids Vol. 31 March 2023 403

 Molecular Therapy: Nucleic Acids

abrogated did not. Next, we analyzed lariat “a” and lariat “c” by We further recapitulated these data with glycerol-3-phosphate dehydro-
using the Hybrid3/1 minigene constructs (Figure 2A) because we genase 1 (GPD1) (Figure 4D). As shown in Figure 4E, all reconstituted
had difficulties in PCR to amplify these two lariats where the high microexons (20, 8, and 5 bp) from normal exon 5 (113 bp) were skipped
GC contents (70%–80%) at the upstream 50 region (Figure 3A, dotted during splicing. The agarose gel electrophoresis of RT-PCR revealed the
line). As expected, lariat “a” was produced from the 4 bp minigene expected sizes of exon-included or -skipped RT-PCR products, which
construct but not from the 2 bp minigene construct. Only lariat were also confirmed by sequencing analysis (Figure 4F).
“c” was detected in 2 bp minigene construct transfected cells indi-
cating that its splicing occurred at once between 50 ss of upstream As in the case of ACACA, it was expected that switching of the 30 ss re-
and 30 ss of the downstream intron. It is also noteworthy to observe gion of an upstream intron of all the GPD1 minigene constructs with
that 0 bp minigene construct produced lariat “a” as well as lariat that of MPC1 microexon would completely rescue their splicing activity
“b,” indicating that even in the absence of exon, each downstream (Figure 4G). As shown in Figure 4H, all the GDP1-microexon-30 ss con-
and downstream intron were spliced through transesterification reac- structs showed a marked increase in EGFP expression, indicating that
tions mediated by their 50 ss and 30 ss. Collectively, these results suggest 30 ss of MPC1 greatly recovered the inclusion of reconstituted microex-
that splicing occurs regardless of the length of microexons, even in the ons. However, the agarose gel electrophoresis of RT-PCR products
absence of an exon, if 30 ss and 50 ss around MPC1 microexon are exhibited unexpected upper bands (marked as *) produced from
intact. each GDP1-microexon-30 ss, while GPD1-113-30 ss exhibited only one
single band (Figure 4I). Interestingly, sequencing data revealed that
Next, we explored the effect of the disruption of 30 ss or 50 ss on lower bands contained each microexon spliced-in, whereas upper
exon splicing. We prepared MPC1 microexon minigene constructs bands were aberrant transcripts spliced at downstream cryptic 50 ss
with 30 ss and 50 ss disrupting mutations as 30 ss (AG>AA) or 50 ss depicted as “GT” in Figure 4I, indicating that cryptic 50 ss rather than
(GU>AU) (Figure 3F). Both mutations completely inhibited the genuine 50 ss was preferred when reconstituted microexons were
splicing of their microexons (Figure 3G). However, there was an spliced. These data led us to compare 50 ss regions (26 bp) of MPC1,
apparent difference in their splicing patterns. In the case of 50 ss mu- ACACA, and GPD1 microexons, and it turned out that the GPD1
tation, the skipping of microexons was markedly induced, like that 50 ss region shows higher GC, lower AT, and lower pyrimidines than
observed when both 30 ss and 50 ss were compromised (Figure 3H). those of ACACA and MPC1 (Figure S2A). Based on these observations,
In contrast, 30 ss mutation led to the splicing of downstream introns it is conceivable that downstream intron 50 ss of MPC1 microexon may
resulting in intron 1 retention (Figure 3I). These results indicate play a role in microexon splicing. As shown in Figure 4J, we generated
that the cis elements of proximal upstream introns may act as splicing various constructs with each exon flanked with 30 ss (96 bp) and 50 ss
enhancers for downstream intron splicing, even when 30 ss is (27 bp) from MPC1 microexon flanking introns (3’50 ss), followed by
disrupted. the measurement of splicing activities. Surprisingly, the splicing activity
of microexons was greatly enhanced in 30 50 ss constructs (Figure 4K).
Splicing of reconstituted microexon is facilitated by 30 ss and 50 ss Moreover, RT-PCR showed a single major product without cryptic
of MPC1 microexon splicing in all minigene constructs (Figure 4L). The sequencing analysis
The exon definitive action of 67 bp 30 ss and 6 bp 50 ss of MPC1 revealed that 113 bp and all microexons were included in the final
microexon was further supported by the experiments with reconsti- mRNA. Collectively, these data indicate that both 30 ss and 50 ss of
tuted microexon minigene constructs. Previous papers, shortening MPC1 microexon exert stimulatory effects on GPD1-reconstituted mi-
of constitutively included normal exon to less than 50 bp led to their croexon inclusion.
complete exclusion during splicing.5 Therefore, we prepared the
minigene constructs containing normal size exon 25 (125 bp) Exon island of MPC1 microexon regulates the splicing process
of acetyl-CoA carboxylase (ACACA), or its reduced 8-bp-length through its exon definitive action
microexon, and their splicing activities were evaluated (Figures 4A The finding that the minimal region of MPC1 microexon flanking
and 4B). As expected, the inclusion of reconstituted microexon introns acts as an exon definition prompted us to test whether it could
was almost completely prevented in contrast with 125-bp-length modulate canonical splicing processes. As shown in Figure 5A, the
normal exon. Given that the 30 ss region in the upstream intron of length of microexon within the exon island was adjusted to 5 bp so
MPC1 microexon is critical for 4 bp microexon splicing, it is that the reading frame could be maintained when the microexon
speculated that this region stimulates the inclusion of reconstituted replaces the target exon. Then, we observed the effect of the exon
microexon. To test this prediction, the 30 ss region of an upstream island (5 bp) on the splicing process when it is positioned in the
intron of ACACA minigene construct was switched to 30 ss middle of exon 5 (113 bp) of GPD1- and in the 50 ss intron 25 of
(96 bp) of MPC1 microexon, and its splicing activity was investi- ACACA minigene construct (Figures 5B and 5C, left). These con-
gated. Remarkably, the inclusion of 8 bp microexon was completely structs were introduced into HeLa cells, and total RNA was extracted
recovered by replacing its 30 ss with that of the MPC1 microexon to perform RT-PCR. Noticeably, a reduced size of the PCR band was
(Figure 4B). The electrophoresis and sequence analysis of the observed when exon island (5 bp) was introduced at each target site
RT-PCR product confirmed the inclusion of ACACA reconstituted (Figures 5B and 5C, right). Sequencing analysis revealed that the
microexon in final mRNA (Figure 4C). lower band of each group contained the 5 bp (“TAACG”), which

404 Molecular Therapy: Nucleic Acids Vol. 31 March 2023

www.moleculartherapy.org

A B C

D E F

G H I

J K L

Figure 4. Splicing of reconstituted microexons is recovered by 30 ss and 50 ss of MPC1 microexon

(A) Wild-type (wt) and 30 ss ACACA minigene constructs. The 30 ss of wt ACACA constructs was switched with the 30 ss 96 bp of MPC1 intron 1. (B), (H), and (K) Splicing
activities by measurement of GFP fluorescence. GFP intensities are normalized by mCherry expression level. (C) Acrylamide gel (8%) analysis of RT-PCR products to check
whether the microexon is included or not. (D) GPD1 minigene construct was generated by insertion of DNA fragment from the exon 4 to exon 6 including exon 5 (113 bp) of
GPD1 gene between SV40 promoter and EFGP gene. Then, the length of exon 5 is shortened as indicated in gray boxes. (E) Quantification of exon inclusion and skipping of
minigene constructs by real-time RT-PCR. (F), (I), and (L) Agarose gel analysis of RT-PCR products from mRNA transcribed from each minigene construct. HeLa cells 48 h
after transfection. (G) The 30 ss GPD1 minigene constructs. The 30 ss of GPD1 intron 4 was switched with that of 30 ss 96 bp of MPC1 intron 1. (J) The 30 50 ss GPD1 minigene
constructs. Both the 30 ss and 50 ss of GPD1 intron 4 and intron 5 were replaced by those of MPC1 intron 1 (30 ss 96 bp) and intron 2 (50 ss 27 bp), respectively. Data are
representative of three experiments. *p < 0.05 and **p < 0.01 by Student’s two-tailed t test.

derived from the exon island, indicating splicing occurred through netic model with SMN protein deficiency due to SMN1 mutation.
the 30 ss and 50 ss within the exon island resulting in skipping of its Since the SMN2 is almost identical to SMN1, it would be expected
target exon. These data demonstrate that the exon island of MPC1 mi- to compensate for the SMN protein production. However, the nucle-
croexon could be used to force exon skipping while maintaining the otide difference at exon 7 of SMN2 skips exon 7 during splicing, which
reading frame. makes SMN2 unable to produce the proper amount of SMN protein.24

Next, we tested the possibility that the exon island might stimulate Therefore, it was tempting to examine if the exon island of MPC1 mi-
neighboring exon splicing. We used the spinal muscular atrophy ge- croexon could induce exon 7 inclusion in mRNA of SMN2. We

Molecular Therapy: Nucleic Acids Vol. 31 March 2023 405

 Molecular Therapy: Nucleic Acids

B C

D E

Figure 5. Exon island of MPC1 acts as a regulator of splicing processes

(A) Structure of the exon island (5 bp) which has “TAACG” in-between 67 bp of 30 ss of MPC1 microexon flanking upstream intron and the 6 bp 50 ss consensus. (B) Exon island
(5 bp) was inserted in the middle of exon 5 (113 bp) of GPD1 minigene construct. Primers used in RT-PCR are marked as arrows (left). Agarose gel analysis of RT-PCR
products (right). (C) Exon island (5 bp) was inserted in the 50 ss of intron 25 of ACACA minigene construct (left). Agarose gel analysis of RT-PCR products amplified with
primers marked as arrows (right). (D) and (E) Splicing modulation of SMN2 exon7 by exon island. SMN1 and SMN2 minigene were placed between the SV40 promoter and
EGFP gene in the pSV40-EGFP backbone. Asterisk (*) indicates C>T mutation in exon 7 of SMN2. SMN2 minigene vector was modified by the insertion of exon island (0 bp)
into either intron 6 or intron 7. The exon island (0 bp) has no exon sequence. Constructs were transfected into HeLa cells followed by total RNA extraction and RT-PCR. The
RT-PCR products amplified with the primers shown in (E) were analyzed in 6% acrylamide gel electrophoresis. For evaluating the efficiency of exon 7 inclusion in final mRNA,
quantitative real-time RT-PCR was performed. Primers are listed in Table S2. All bar graphs are plotted as mean ± SE. **p < 0.01 by Student’s two-tailed t test.

designed wt SMN1 and SMN2 minigene constructs encompassing inal sequences. Collectively, these data suggest that the exon island of
exons 6 to 8 of respective genes (Figure 5D). Furthermore, it was veri- MPC1 has strong splicing-inducing strength, which can modify ca-
fied that C to T mutation at exon 7 of SMN2 led to exon 7 skipping nonical splicing processes depending on where it is inserted.
while exon 7 was included in SMN1 mRNA (Figure 5E lanes 1 and
2). To test the ability of the exon island to facilitate the inclusion of DISCUSSION
exon 7 of SMN2, two modified SMN2 minigene constructs were pre- There is growing evidence on the biological importance of microex-
pared by inserting the exon island (0 bp) into either upstream (intron ons alongside their splicing mechanisms. However, few studies have
6) or downstream intron (intron 7), respectively (Figure 5D). These elucidated non-3x microexon splicing. Here, we showed how non-
constructs were transfected into HeLa cells, and the splicing variants 3x microexon of MPC1 (4 bp) was spliced through the experiments
from each SMN minigene were visualized by electrophoresis and using various minigene and exon island reporters. We also found
quantified by real-time RT-PCR. Remarkably, the insertion of the that the proximal intron region of 67 bp 30 ss and 6 bp 50 ss of the
exon island (0 bp) at intron 6 and intron 7 strongly stimulated the MPC1 microexon has strong exon definitive activity and raised the
substantial inclusion of exon 7 in the final mRNA (Figure 5E). Further possibility of its usage as a modifier for splicing processes in target
sequencing analysis revealed that no additional sequence was detected genes.
at exon junctions since the exon island (0 bp) used in this experiment
did not have an exon sequence between its 30 ss and 50 ss, indicating Considering the reading frame, alternative splicing of non-3x bp mi-
that the exon definition of the exon island (0 bp) acts as an enhancer croexons should be precisely regulated. Our observation in Figure 1B
for potentiating neighboring exon inclusion without perturbing orig- shows that non-3x bp microexons under 10 bp from five genes

406 Molecular Therapy: Nucleic Acids Vol. 31 March 2023

www.moleculartherapy.org

exhibited constitutively spliced in the final mRNA. We obtained GPD1 microexon was achieved by the replacement of the 50 ss region
consistent data with differentiated and undifferentiated 3T3L1 of the GPD1 microexon with that of the MPC1 (27 bp), which possibly
(mouse preadipocyte) and C2C12 (mouse skeletal muscle) (Figure S1). indicates that the subtle difference in the strength of 50 ss could deter-
However, the transient expression in HeLa cells with artificial re- mine the complete inclusion of microexon although exact underlying
porter studies did not implement constitutive splicing of microexons mechanisms are currently unclear.
of TNNI1, TMEM237, KSR1, and MUSK except for the MPC1 micro-
exon (Figure 2F). These discrepancies in the microexon splicing Previous studies demonstrated that flanking introns have information
pattern observed in two different experimental settings may be that codes for the correct splicing of microexons in a variety of con-
because the artificially shortened intron does not have sufficient in- texts within the cell.5,9,13,17,18,32–34 In addition, the functional associ-
tronic elements for complete microexon splicing except for the ation between specific histone modifications and microexon splicing
MPC1 microexon. In addition, the lack of tissue-specific RBPs in events has been elucidated.35,36 Moreover, there are complexities in
HeLa cells might lead to incomplete inclusion of microexons of combined actions among the location and numbers of intronic cis el-
four genes except for MPC1, which could be supported by the fact ements as well as temporal- and tissue-specific availability of RBPs for
that, unlike MPC1, the four genes are tissue-specifically expressed. the accurate splicing of microexons. There is a high sequence similar-
ity at the upstream proximal intron between species as shown in Fig-
Despite experimental limitations with reporter constructs, the inclu- ure S1. The pyrimidine-rich region of MPC1 microexons is located
sions of MPC1 microexon in mRNA seems quite consistent in various farther upstream than its common exons, probably to overcome
experimental settings. Usually, constitutively spliced microexons are spatial limitations in microexon splicing through exon definition.
predicted to have stronger splice-site tendencies, shorter flanking in- Direct repeat of “TGTCTG” and highly conserved sequences
trons, and a higher ESE density than those of longer exons.9 However, downstream of pyrimidine tract needs further study to elucidate the
MPC1 microexon, 4 bp, is extremely short to accommodate sufficient interaction of the RBPs or other regulators for splicing.
ESE and has long upstream and downstream flanking introns, over
12 and 3 kbp, respectively. Moreover, MPC1 microexon is highly Since we verified a strong exon definition of the minimal region of
expected to skip because it has a high skipping score (S = 4.22) MPC1 microexon flanking introns, we presented the possibilities of
when assessed using computational methods using MaxEntScan.25,26 using it as a modifier of genuine splicing processes (Figure 5). The
target exon can be skipped during splicing by the insertion of the
In this study, we first demonstrated that both splicing sites bordering exon island into the target exon or splice site, maintaining the reading
the MPC1 microexon have strong splicing activity sufficient to splice 4, frame. Strikingly, we also showed that the exon island (0 bp), which
2, and 1 bp, and even 0 bp (Figure 3). Particularly, our studies of has no exon sequence, positioned in intron 6 or intron 7 of SMN2
splicing intermediates and mutagenesis of splice sites strongly minigene markedly enhanced exon 7 inclusion without any sequence
indicate that the MPC1 microexon is spliced by its exon definition addition or reading frameshift. The observation that the 50 ss destruc-
in a manner different from common exons where the exon junction tion of the exon island did not enhance exon 7 inclusion strongly sug-
complex deposited after the splicing reaction strongly inhibits recur- gests that complete exon definition of the exon island is essential for
sive splicing in its vicinity,27–30 thereby preventing cryptic splicing its splicing enhancing action (Figure S4). Future investigations in vivo
from producing microexons. Interestingly, our lariat studies identified models are necessary to clarify the efficacy of the exon island (0 bp)
several branchpoints (Figure 3E). It turned out that two “A’s” at 68 usage regarding SMN protein expression and to compare its potency
and 63 nucleotides from microexon are the major branching with that of other splicing-editing tools such as antisense oligonucle-
positions near the pyrimidine-rich region, which are unusually distant otide, small molecules, and targeting splicing regulating elements.37
locations compared with regular branchpoints positioned at a median
of 28 bp upstream of the 30 ss.31 This distal pyrimidine-rich region Overall, we identify the possible usages of the exon island of MPC1 in
might help to overcome the proximity between 30 ss and 50 ss bordering various genetic contexts. This study may provide insights for effective
the 4 bp microexon, allowing the accurate splicing of microexons. and precise splicing-mediated gene correction.

We also provide evidence that the 67 bp 30 ss of MPC1 microexon MATERIALS AND METHODS
flanking intron is critical for its exon definition through various recon- Cell culture and transfection
stituted microexon and mutagenesis studies. This minimal region HeLa cells were grown in Dulbecco’s modified Eagle medium con-
(67 bp 30 ss) is sufficient not only for the MPC1 microexon but also taining 10% heat-inactivated fetal bovine serum, 100 unit/mL peni-
for the correct inclusion of reconstituted microexons of ACACA cillin, and 100 g/mL streptomycin. For ViaFect (Promega, Madison,
with the exceptions of those of GPD1 that were partially recovered WI) was used in transient expression according to the manufacturer’s
(Figure 4I). Assuming that the role of 50 ss 6 bp of MPC1 microexon introductions.
on splicing is sufficient, only one nucleotide in 50 ss of GPD1 was
substituted with that of MPC1 (guGaga > guAaga). However, the Generation of minigene constructs
splicing-in of reconstituted GPD1 microexons was incomplete MPC1 minigene construct containing its microexon (exon 2) is con-
(Figures S2B and S2C). The complete inclusion of the reconstituted structed by the insertion of the 30 region of exon 1 to 50 region of

Molecular Therapy: Nucleic Acids Vol. 31 March 2023 407

 Molecular Therapy: Nucleic Acids

intron 1 (901 bp) (chr6: 166,381,901-166,382,878), 30 region of intron 25 to 50 region of exon 26 (Chr17: 37,226,641-37,226,421) in order
1 (329 bp) to 50 region of intron 2 (541bp) (chr6: 166,369,677- between the SV40 promoter and the EGFP gene of pSV40-EGFP
166,370,544), and 30 region of intron 2 (259 bp) to 50 region of backbone.
exon 3 (496b0) (chr6: 166,366,843-166,366,794) in order between
the SV40 promoter and the EGFP gene of pSV40-EGFP backbone. Generation of SMN1 and SMN2 minigene constructs
MPC1 Swap-1/3 and 2/3 constructs are prepared by replacing the SMN1 and SMN2 minigene construct containing exon 7 is produced by
intron 1 and intron 2 with intron 3 (688 bp) (chr6: 166,366,107- the insertion of the 30 region of exon 6 to the 50 region of intron 6 (hg38;
166,370,544), respectively. QuickChange reaction was performed us- chr5: 70,946,149-70,946,593 and hg38; chr5: 70,070,724-70,071,168,
ing the amplicon of intron 3 of which both ends contain the sequences respectively), and 30 region of intron 6 to 50 region of exon 8 (chr5:
of either side of the site to be inserted. MPC1 hybrid-3/1 and 1/3 70,951,501-70,952,510 and chr5: 70,076,081-70,077,090, respectively)
constructs are prepared by replacing the 50 region of intron 1 in order between the SV40 promoter and the EGFP gene in pSV40-
(901 bp) with the 50 region of intron 3 (349 bp) (chr6: 166,366,446- EGFP backbone.
166,366,794) and the 30 region of intron 1 (329 bp) with the 30 region
of intron 3 (339 bp) (chr6: 166,366,107-166,366,445) in MPC1 mini-
Measurement of splicing activity
gene construct, respectively. Hybrid-3/1 serial deletion constructs and
We transiently transfected the minigene construct into HeLa cells. Af-
MPC1 minigene constructs containing different lengths (0, 1, 2, and
ter 48 h of transfection, whole-cell lysates were prepared with 1x Pas-
3 bp) of microexon were generated by QuickChange reaction using
sive lysis buffer (Promega, Madison, WI, USA). EGFP and mCherry
the respective pair of primers (Table S1).
intensities were measured with supernatant after centrifugation at
8,000 g (TECAN, infinite F200 pro, Männedorf, Switzerland). The
Generation of “exon island” reporter constructs
value of GFP intensity was divided by mCherry intensity for normal-
Plasmid SV40-EGFP backbone is constructed by the ligation of
ization according to transfection efficiency.
SV40 promoter, EGFP gene without ATG start codon, followed by
SV40 poly(A) signal sequence. For microexon island construction,
the amplicon of MPC1 gene from the 30 region of exon 3 to the 50 region RNA extraction and quantitative real-time RT-PCR
of exon 4 (hg38:chr6: 166,366,102-166,366,809) was inserted between For quantitative real-time RT-PCR (qPCR), cDNAs were synthesized
the SV40 promoter and the EGFP gene. Then, ATG start codon from 4 mg of total RNA using random hexamer primers and
having Kozak sequence and three kinds of the reading frame was SuperScript reverse transcriptase III (Themo Fisher Scientific, Wal-
made in MPC1 exon 3 region and EcoRV restriction site was introduced tham, MA) following the manufacturer’s instructions. Diluted cDNAs
at the middle of intron 3 (chr6: 166366432) (pSV40-EGFPa, b, c). were used as a template for qPCR using the primers (Table S2) with
Microexon island constructs were prepared by insertion of the the SYBR Green Master Mix (Applied Biosystems; Foster City, CA). Re-
microexon and both flanking regions of TNNI1 (chr1: 201,416,899- actions are performed using the ABI PRISM 7300 RT-PCR System
201,417,360), KSR1 (chr17: 27,585,451-853), TMEM237 (chr2: 201, (Applied Biosystems).
639,991-201,640,463), and MUSK (chr9: 110,762,013-110,762,442) at
EcoRV site. Analysis of lariats
Ten micrograms of total RNA was treated with ribonuclease R (Bio-
Generation of reconstituted microexon minigene constructs search Technologies, Hoddesdon, UK) for 1 h at 37 C and purified
GPD1 minigene construct containing exon 6 is constructed by with Monarch RNA extraction kit (New England Biolabs, Ipswich,
the insertion of the 30 region of exon 5 to the 50 region of exon 7 MA). First PCR was done with Super-Script IV One-step RT-PCR sys-
(hg38; chr17: 50,106,394-50,107,573) between the SV40 promoter tem (Themo Fisher Scientific, Waltham, MA). The primers were as fol-
and the EGFP gene of pSV40-EGFP backbone. GPD1 microexon lows: lariat “a” forward (f) 50 -CAGGTGTGAGGTTG-CAGTAACCT-
vectors are generated by QuickChange Kit using the forward and 30 and reverse (r) 50 -GACACAGTC-TTCAGTTCAGTGTGC-3’; lariat
reverse primers of GPD-20, GPD8, and GPD-5 (Table S1). To “b” (f) 50 - GACTGCTGAATATCTGACAGCG-30 and (r) 50 -AGGAG
replace intron 5 30 ss in GPD1 minigene constructs with 96 bp of CATGGCTGTCACAGA-30 ; lariat “c” (f) 50 -GACTGCTGAATATCT
MPC1 intron 1 30 ss, MPC1 intron 1 30 ss region was amplified GACAG-CG-30 , (r) 50 -CACTCTTCCATGCCATCACATTCTGTAT
using GPD-3ss-fwd and respective reverse primers (GPD-113-3ss- GT-30 . Second PCR was done using LaboPass IP-Taq DNA Polymerase
rev, GPD-20-3ss-rev, GPD-8-3ss-rev, and GPD-5-3ss-rev), and (Cosmogenetech, Seoul, South Korea). One-10th diluted first PCR
PCR products were used in QuickChange reactions. Modification products were used as PCR templates. The primers were as follows:
of 30 ss and 50 ss in GPD1 minigene constructs was done by lariat “a” primer (f) 50 -GTTGCAGTAACCTAATAAGACCAA-30
QuickChange with the primers listed in Table S1 and respective and (r) 50 -CACTCTTCCATG-CCATCACATTCTGTATGT-3’; lariat
GPD 30 ss constructs as used as a template for PCR. ACACA mini- “b” (f) 50 - CCAAACCCTTTCTAGCTCTG-30 , (r) 50 - TGCAGTAGGC
gene construct containing exon 25 is generated by the insertion of ACCCTTCACA-30 ; lariat “c” (f) 50 -AGGAGCATGGCTGTCACAGA-
the 30 region of exon 24 to the 50 region of intron 24 (hg38; 30 and (r) 50 -TCTTGAAATGCAAGCAGGAGC-30 . PCR products are
chr17: 37,240,524-37,240,230), 30 region of intron 24 to 50 region cloned into pBluescript vector followed by sequencing to analyze
of intron 25 (chr17: 37,235,326-37,234,785), and 30 region of intron branchpoints.

408 Molecular Therapy: Nucleic Acids Vol. 31 March 2023

www.moleculartherapy.org

Statistical analysis 11. Lin, L., Zhang, M., Stoilov, P., Chen, L., and Zheng, S. (2020). Developmental atten-
All results are expressed as the mean ± standard error (SE). The Stu- uation of neuronal apoptosis by neural-specific splicing of Bak1 microexon. Neuron
107, 1180–1196.e8.
dent’s t test was used for comparison between two groups. A p
12. Head, S.A., Hernandez-Alias, X., Yang, J.S., Ciampi, L., Beltran-Sastre, V., Torres-
value < 0.05 was considered statistically significant. Méndez, A., Irimia, M., Schaefer, M.H., and Serrano, L. (2021). Silencing of
SRRM4 suppresses microexon inclusion and promotes tumor growth across cancers.
DATA AVAILABILITY PLoS Biol. 19, e3001138.
The data that support the findings of this study are available in the 13. Gonatopoulos-Pournatzis, T., Wu, M., Braunschweig, U., Roth, J., Han, H., Best, A.J.,
supplemental information of this article. Raj, B., Aregger, M., O’Hanlon, D., Ellis, J.D., et al. (2018). Genome-wide CRISPR-
cas9 interrogation of splicing networks reveals a mechanism for recognition of
autism-misregulated neuronal microexons. Mol. Cell 72, 510–524.e12.
SUPPLEMENTAL INFORMATION 14. Lee, J., Villarreal, O.D., Chen, X., Zandee, S., Young, Y.K., Torok, C., Lamarche-Vane,
Supplemental information can be found online at https://doi.org/10. N., Prat, A., Rivest, S., Gosselin, D., et al. (2020). QUAKING regulates microexon
1016/j.omtn.2023.01.010. alternative splicing of the Rho GTPase pathway and controls microglia homeostasis.
Cell Rep. 33, 108560.
15. Irimia, M., Weatheritt, R.J., Ellis, J.D., Parikshak, N.N., Gonatopoulos-Pournatzis, T.,
ACKNOWLEDGMENTS
Babor, M., Quesnel-Vallières, M., Tapial, J., Raj, B., O’Hanlon, D., et al. (2014). A
This work was supported by the National Research Foundation of Korea highly conserved program of neuronal microexons is misregulated in autistic brains.
(NRF) grant funded by the Korean Government (MSIT) [NRF- Cell 159, 1511–1523.
2020R1A2C1004833, NRF-2017R1A2B4007462, NRF-2017R1A2B400 16. Li, Y., Zhang, Q., Lovnicki, J., Chen, R., Fazli, L., Wang, Y., Gleave, M., Huang, J., and
9674]. This work was supported by a faculty research grant of Yonsei Dong, X. (2019). SRRM4 gene expression correlates with neuroendocrine prostate
cancer. Prostate 79, 96–104.
University College of Medicine (6-2022-0063). Funding for open access
charge: National Research Foundation of Korea. 17. Carlo, T., Sierra, R., and Berget, S.M. (2000). A 5’ splice site-proximal enhancer binds
SF1 and activates exon bridging of a microexon. Mol. Cell Biol. 20, 3988–3995.
18. Dominski, Z., and Kole, R. (1992). Cooperation of pre-mRNA sequence elements in
AUTHOR CONTRIBUTIONS splice site selection. Mol. Cell Biol. 12, 2108–2114.
E.K. and K.K. conceived and designed the experiments. E.K. and D.S.
19. Lykke-Andersen, S., and Jensen, T.H. (2015). Nonsense-mediated mRNA decay: an
conducted most of the cell and biochemical experiments. K.K. per- intricate machinery that shapes transcriptomes. Nat. Rev. Mol. Cell Biol. 16, 665–677.
formed bioinformatics analyses. E.K. and K.K. analyzed and inter- 20. Bricker, D.K., Taylor, E.B., Schell, J.C., Orsak, T., Boutron, A., Chen, Y.C., Cox, J.E.,
preted results and wrote the manuscript with input from all authors. Cardon, C.M., Van Vranken, J.G., Dephoure, N., et al. (2012). A mitochondrial py-
All authors contributed to editing the manuscript. ruvate carrier required for pyruvate uptake in yeast, Drosophila, and humans.
Science 337, 96–100.

DECLARATION OF INTERESTS 21. Herzig, S., Raemy, E., Montessuit, S., Veuthey, J.L., Zamboni, N., Westermann, B.,
Kunji, E.R.S., and Martinou, J.C. (2012). Identification and functional expression of
The authors declare no competing interests. the mitochondrial pyruvate carrier. Science 337, 93–96.
22. Koh, E., Kim, Y.K., Shin, D., and Kim, K.S. (2018). MPC1 is essential for PGC-
REFERENCES 1a-induced mitochondrial respiration and biogenesis. Biochem. J. 475, 1687–1699.
1. Ule, J., and Blencowe, B.J. (2019). Alternative splicing regulatory networks: functions, 23. Piovesan, A., Antonaros, F., Vitale, L., Strippoli, P., Pelleri, M.C., and Caracausi, M.
mechanisms, and evolution. Mol. Cell 76, 329–345. (2019). Human protein-coding genes and gene feature statistics in 2019. BMC Res.
2. Scotti, M.M., and Swanson, M.S. (2016). RNA mis-splicing in disease. Nat. Rev. Notes 12, 315.
Genet. 17, 19–32. 24. Lefebvre, S., Bürglen, L., Reboullet, S., Clermont, O., Burlet, P., Viollet, L., Benichou,
3. Hawkins, J.D. (1988). A survey on intron and exon lengths. Nucleic Acids Res. 16, B., Cruaud, C., Millasseau, P., Zeviani, M., et al. (1995). Identification and character-
9893–9908. ization of a spinal muscular atrophy-determining gene. Cell 80, 155–165.

4. Berget, S.M. (1995). Exon recognition in vertebrate splicing. J. Biol. Chem. 270, 25. Ye, Z., Chen, Z., Lan, X., Hara, S., Sunkel, B., Huang, T.H.M., Elnitski, L., Wang, Q.,
2411–2414. and Jin, V.X. (2014). Computational analysis reveals a correlation of exon-skipping
events with splicing, transcription and epigenetic factors. Nucleic Acids Res. 42,
5. Dominski, Z., and Kole, R. (1991). Selection of splice sites in pre-mRNAs with short 2856–2869.
internal exons. Mol. Cell Biol. 11, 6075–6083.
26. Yeo, G., and Burge, C.B. (2004). Maximum entropy modeling of short sequence mo-
6. Black, D.L. (1991). Does steric interference between splice sites block the splicing of a tifs with applications to RNA splicing signals. J. Comput. Biol. 11, 377–394.
short c-src neuron-specific exon in non-neuronal cells? Genes Dev. 5, 389–402.
27. Pai, A.A., Paggi, J.M., Yan, P., Adelman, K., and Burge, C.B. (2018). Numerous recur-
7. Volfovsky, N., Haas, B.J., and Salzberg, S.L. (2003). Computational discovery of inter- sive sites contribute to accuracy of splicing in long introns in flies. PLoS Genet. 14,
nal micro-exons. Genome Res. 13, 1216–1221. e1007588.
8. Wu, T.D., and Watanabe, C.K. (2005). GMAP: a genomic mapping and alignment 28. Joseph, B., Kondo, S., and Lai, E.C. (2018). Short cryptic exons mediate recursive
program for mRNA and EST sequences. Bioinformatics 21, 1859–1875. splicing in Drosophila. Nat. Struct. Mol. Biol. 25, 365–371.
9. Li, Y.I., Sanchez-Pulido, L., Haerty, W., and Ponting, C.P. (2015). RBFOX and PTBP1 29. Blazquez, L., Emmett, W., Faraway, R., Pineda, J.M.B., Bajew, S., Gohr, A., Haberman,
proteins regulate the alternative splicing of micro-exons in human brain transcripts. N., Sibley, C.R., Bradley, R.K., Irimia, M., et al. (2018). Exon junction complex shapes
Genome Res. 25, 1–13. the transcriptome by repressing recursive splicing. Mol. Cell 72, 496–509.e9.
10. Capponi, S., Stöffler, N., Irimia, M., Van Schaik, F.M.A., Ondik, M.M., Biniossek, 30. Boehm, V., Britto-Borges, T., Steckelberg, A.L., Singh, K.K., Gerbracht, J.V., Gueney,
M.L., Lehmann, L., Mitschke, J., Vermunt, M.W., Creyghton, M.P., et al. (2020). E., Blazquez, L., Altmüller, J., Dieterich, C., and Gehring, N.H. (2018). Exon junction
Neuronal-specific microexon splicing of TAF1 mRNA is directly regulated by complexes suppress spurious splice sites to safeguard transcriptome integrity. Mol.
SRRM4/nSR100. RNA Biol. 17, 62–74. Cell 72, 482–495.e7.

Molecular Therapy: Nucleic Acids Vol. 31 March 2023 409

 Molecular Therapy: Nucleic Acids

31. Pineda, J.M.B., and Bradley, R.K. (2018). Most human introns are recognized via mul- 35. Luco, R.F., Pan, Q., Tominaga, K., Blencowe, B.J., Pereira-Smith, O.M., and Misteli, T.
tiple and tissue-specific branchpoints. Genes Dev. 32, 577–591. (2010). Regulation of alternative splicing by histone modifications. Science 327, 996–1000.
32. Black, D.L. (1992). Activation of c-src neuron-specific splicing by an unusual RNA 36. Hu, Q., Kim, E.J., Feng, J., Grant, G.R., and Heller, E.A. (2017). Histone posttransla-
element in vivo and in vitro. Cell 69, 795–807. tional modifications predict specific alternative exon subtypes in mammalian brain.
33. Carlo, T., Sterner, D.A., and Berget, S.M. (1996). An intron splicing enhancer contain- PLoS Comput. Biol. 13, e1005602.
ing a G-rich repeat facilitates inclusion of a vertebrate micro-exon. RNA 2, 342–353. 37. Nurputra, D.K., Lai, P.S., Harahap, N.I.F., Morikawa, S., Yamamoto, T.,
34. Modafferi, E.F., and Black, D.L. (1997). A complex intronic splicing enhancer from Nishimura, N., Kubo, Y., Takeuchi, A., Saito, T., Takeshima, Y., et al. (2013).
the c-src pre-mRNA activates inclusion of a heterologous exon. Mol. Cell Biol. 17, Spinal muscular atrophy: from gene discovery to clinical trials. Ann. Hum.
6537–6545. Genet. 77, 435–463.

410 Molecular Therapy: Nucleic Acids Vol. 31 March 2023