Investigating mRNA Sequence Features That Regulate Nuclear ... · chromosome architecture and/or 3) RNA regulatory sequences that impact alternative splicing and other RNA processing
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Investigating mRNA Sequence Features That Regulate Nuclear Export
by
Abdalla Mahmoud Akef
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Investigating mRNA Sequence Features That Regulate Nuclear
Export
Abdalla Mahmoud Akef
Doctor of Philosophy
Department of Biochemistry University of Toronto
2016
Abstract In eukaryotes, the nucleus divides the cell into two compartments: the nucleoplasm, where RNA
is synthesized, processed and packaged, and the cytoplasm, where mature mRNA is translated
into proteins. The mechanisms that determine whether an mRNA should be exported from the
nucleus to the cytoplasm are poorly understood. The best established model postulated that
mRNAs need to be spliced or to contain nuclear export-promoting sequences so as to be
efficiently transported to the cytoplasm. However, this model suffered several problems since it
was known that many intronless mRNAs are efficiently exported to the cytoplasm independently
of splicing and they do not contain any known nuclear export-promoting sequences. In this
thesis, I sought to dissect the sequence features within a transcript that determine its cellular
localization. I discovered that intronless β-Globin (βG) mRNA contains a nuclear retention
element that inhibits its transport to the cytoplasm. This nuclear retention element can be
overcome when βG mRNA is either spliced or its length extended. The mRNA nuclear export
factor UAP56 binds to several intronless mRNAs including βG mRNA independently of
splicing. I also discovered that UAP56 is required for the egress of intronless mRNAs from
nuclear speckles. My results suggest that most mRNAs are exported to the cytoplasm
independently of splicing unless they contain a nuclear retention element.
iii
Acknowledgments I acknowledge Dr. Alex Palazzo. I would really like to thank my thesis committee
members, Dr. Craig Smibert and Dr. Fritz Roth for their time and feedback. Finally, I would like
to thank my mom.
iv
Table of Contents Abstract ............................................................................................................................... ii
Acknowledgments .............................................................................................................. iii
List of Tables ................................................................................................................... viii
List of Figures .................................................................................................................... ix
List of Abbreviations ......................................................................................................... xi
Palazzo, A.F., and Akef, A. (2012). Nuclear export as a key arbiter of “mRNA identity” in
eukaryotes. Biochim. Biophys. Acta 1819, 566–577.
2
1.1 Introduction
1.1.1 The nucleus and the expansion of non-coding sequences: two features that distinguish eukaryotes from prokaryotes The nucleus is the defining feature of the eukaryotic cell. It compartmentalizes the
cellular space into two distinct regions: the nucleoplasm, where RNA is synthesized, processed
and packaged, and the cytoplasm, where mature messenger RNA (mRNA) is translated into
proteins. This is in striking contrast to prokaryotes, where transcription and translation occur
concurrently in the same compartment. Another important difference between these two groups
is the percentage of their genomes that encode protein. In multicellular eukaryotes, protein-
coding sequences account for a fraction of the genome, varying from 1.5 to 36 percent (Gregory,
2005; Lynch, 2007; Lynch et al., 2011), while in prokaryotes, the majority of the genome is
protein-coding (Lynch, 2006). Eukaryotes have experienced a vast expansion in genomic
sequences that does not code for proteins (Gregory, 2005; Palazzo and Gregory, 2014). It has
been assumed by many that this increase was a consequence of natural selection acting to expand
the amount of functional information and organismal complexity (ENCODE Project Consortium,
2012; Mattick et al., 2010; Taft et al., 2007; Ecker et al., 2012; Pennisi, 2012; Kapusta and
Feschotte, 2014), which could have taken the form of an amplification in 1) functional non-
coding transcriptional products, 2) DNA regulatory sequences that direct RNA transcription and
chromosome architecture and/or 3) RNA regulatory sequences that impact alternative splicing
and other RNA processing events.
In support of the expansion of functional non-coding transcripts, several large-scale
analyses have indicated that most of the eukaryotic genome is transcribed, albeit at a low level
(ENCODE Project Consortium et al., 2007; Johnson et al., 2005; Djebali et al., 2012). However,
it has been suggested that these non-coding transcripts do not have a specific function since most
of the non-protein coding transcribed regions are poorly conserved (ENCODE Project
Consortium et al., 2007; Marques and Ponting, 2009; Wang et al., 2004; Doolittle et al., 2014)
and are rapidly degraded (Chekanova et al., 2007; Davis and Ares, 2006; Neil et al., 2009; Preker
et al., 2008; Thiebaut et al., 2006; Vasiljeva et al., 2008; Wyers et al., 2005; Xu et al., 2009;
Tisseur et al., 2011). An analysis of mouse nascent transcripts by RNA-Sequencing (RNA-Seq)
has revealed that the number of reads that map to intergenic regions of the genome is almost
3
equivalent to the number of reads that align to exonic regions (Menet et al., 2012). On the other
hand, most of the steady state polyadenylated RNA population mapped to exonic regions with
only a small percentage aligning to intergenic regions (Menet et al., 2012). However, it should be
pointed out that a short-lived transcript is not necessarily non-functional. Moreover, many
examples have been found that contradict these general features. Non-coding RNAs have been
found to regulate transcription (Martens et al., 2004), mRNA translation and stability (Fire et al.,
1998), histone modification (Tsai et al., 2010; Rinn et al., 2007), DNA methylation (Bartolomei
et al., 1991), DNA recombination (Kobayashi and Ganley, 2005), and even cross-regulate other
non-coding RNAs (Salmena et al., 2011).
In parallel to these studies, advances in population genetics have uncovered many of the
evolutionary forces that shape genomic content. One important principle derived from these
analyses is that the ability of natural selection to weed out mildly deleterious mutations, such as
the insertion of non-functional DNA sequences (i.e., introns and intergeneic sequences),
increases with the number of breeding individuals (Lynch, 2006; Lynch et al., 2011). As a
consequence, species that have a low population size, as is the case with most eukaryotes, cannot
effectively select out these genetic alterations. It appears that these entities are being eliminated
in certain eukaryotes whose effective population size must have recently increased (such as
Saccharomyces cerevisiae (S. cerevisiae), which has experienced a recent loss of introns (Csuros
et al., 2011; Irimia and Roy, 2008; Irimia et al., 2007; Rogozin et al., 2003) and intergenic
sequences (Chen et al., 2011; Kuo and Ochman, 2009). However, it is likely that these lineages
are the exception rather than the rule, as the level of intergenic sequence in most unicellular
eukaryotic genomes is about 50% (Lynch, 2007).
It has been proposed that intergenic regions are likely to contain cryptic transcriptional
start sites, whose sequences tend to be highly degenerate (Carninci et al., 2006; Froula and
Francino, 2007; Hahn et al., 2003; Lynch et al., 2005). This would explain why both the mouse
and human genomes contain about an order of magnitude more promoter regions as compared to
protein-coding genes (Carninci et al., 2006; ENCODE Project Consortium et al., 2007). It is also
worth noting that although RNA polymerase II (Pol II) does not efficiently initiate transcription
at non-promoter sites, the proliferation of non-functional DNA may also increase the frequency
of spurious transcription initiation by increasing the amount of non-specific substrate. Indeed,
spurious initiation of Pol II-driven transcription has been observed at nucleosome-free sites in
vivo (Carninci et al., 2005; Cheung et al., 2008) and likely accounts for a substantial fraction of
4
all nascent RNAs (Menet et al., 2012; Palazzo and Gregory, 2014). As a result of all these forces,
we can begin to understand why a large fraction of active Pol II is associated with intergenic
regions throughout the yeast and human genomes (Buratowski, 2008; Cheung et al., 2008).
1.1.2 A key role of the nucleus is to sort transcripts that contain “mRNA identity” features from spurious transcripts From this vantage point the real question that we should be asking is how functional
transcripts are extracted from all the transcriptional noise resulting from the eukaryotic genome.
The answer appears to be that transcripts bearing hallmarks of protein-coding genes are
identified by an extensive network of feedback and feed-forward loops between various
machineries present at different steps of the gene expression pathway (Fig. 1.1). Thus at each
step, features associated with mRNA identity are acted on by one set of factors, and these
directly promote the activity of other machines responsible for subsequent and previous steps.
This phenomenon, generally known as “coupling” (Maniatis and Reed, 2002; Moore and
Proudfoot, 2009; Perales and Bentley, 2009), was previously viewed as a method for either
enhancing the efficiency of gene expression or distinguishing properly processed from
unprocessed mRNAs. However, through this extensive coupling network, a system for
identifying mRNAs from spurious transcription also emerges. It should be noted that the concept
of “mRNA identity” was originally used to describe how protein-coding transcripts are
differentiated from snRNA, tRNAs and other nuclear exported transcripts (Masuyama et al.,
2004; Ohno et al., 2002; Ullman, 2002). I would like to expand this concept to explain how cells
differentiate protein-coding mRNAs from transcriptional noise through coupling. The role of the
nuclear envelope within this context becomes apparent: by segregating RNA production and
processing from the translation machinery, it ensures that this coupling network completes its
task of separating mRNAs from transcriptional noise before allowing these products to be
translated into proteins (Martin and Koonin, 2006). As a consequence, the nuclear mRNA export
machinery is one of the central nodes of this coupling network (Fig. 1.1) and acts as a key arbiter
of mRNA identity. The coupling between the mRNA nuclear export machinery and the various
gene expression complexes will be discussed in the next section.
5
Figure 1.1. Coupling between mRNA nuclear export and various steps of gene expression.
Green arrows represent a positive feed-forward or feedback regulation while red lines represent a
negative feedback relationship.
6
1.2 The coupling of mRNA processing with mRNA nuclear export 1.2.1 The mRNA nuclear export machinery
Since the Transcription Export (TREX) sits at the center of many of the coupling
reactions outlined below, I will begin by reviewing its constituents and how it regulates mRNA
nuclear export (Table 1.1). In S. cerevisiae, the TREX complex is composed of the THO
complex (Hpr1p, Tho2p, Mft1p and Thp2p), the DEAD box RNA helicase Sub2p and the RNA
recognition motif containing protein, Yra1p (Chávez et al., 2000; Jimeno et al., 2002; Strässer
and Hurt, 2001; Strässer et al., 2002; Zenklusen et al., 2002). In metazoans, the THO complex
contains homologs for Hpr1p (THOC1), Tho2p (THOC2) and at least three proteins not found in
budding yeast (THOC5, THOC6 and THOC7) (Rehwinkel et al., 2004; Zenklusen et al., 2002).
Mammals have two homologs for Sub2p (UAP56 and URH49) (Masuda et al., 2005) and one
predominant Yra1p-like protein (Aly) (Luo et al., 2001).
The mammalian THO complex and UAP56 are thought to be assembled onto the mRNA
in nuclear speckles, foci where transcription, splicing, polyadenylation and general messenger
ribonucleoparticle (mRNP) assembly are thought to take place (Dias et al., 2010; Kota et al.,
2008). The mRNA nuclear export adaptor, Aly is then recruited to TREX by physically
interacting with UAP56 in its ATP-bound form (Dufu et al., 2010; Taniguchi and Ohno, 2008).
Then in response to an unknown stimulus, UAP56 hydrolyzes ATP, and the ADP-bound form is
then released from TREX and the mRNA (Taniguchi and Ohno, 2008). This exposes a domain
on Aly that can recruit the TAP/p15 heterodimer (Mex67p/Mtr2p in S. cerevisiae) to the
transcript (Dufu et al., 2010; Hautbergue et al., 2008; Strässer and Hurt, 2000; Strässer et al.,
2002; Stutz et al., 2000), and in mammalian cells this likely occurs within nuclear speckles
(Schmidt et al., 2006). After disrupting the association of Aly and the transcript, TAP/p15 ferries
the transcript across the nuclear pore complex (NPC) by interacting with nucleoporins
(Hautbergue et al., 2008; Katahira et al., 1999; Santos-Rosa et al., 1998; Segref et al., 1997).
Although this picture paints an extremely defined sequential series of events, there are
likely multiple mechanisms by which TREX mediates nuclear export of mRNA. For example, in
budding yeast Hpr1 can directly recruit TAP/p15 (Hobeika et al., 2007), while in
Schizosaccharomyces pombe (S. pombe), UAP56 recruits the conserved Rae1p to the transcript,
which then promotes a TAP/p15-independent mRNA export (Thakurta et al., 2007; Yoon et al.,
2000). In other systems Rae1p has been shown to bind to both TAP/p15 and nucleoporins, and is
7
Table 1.1. Major role of different mRNA nuclear export factors.
Export Factor Complex Major Proposed Role
Yeast Metazoan
Hpr1p THOC1 THO-TREX Recruits UAP56 to mRNA transcripts.
Tho2p THOC2 THO-TREX
Mft1p - THO-TREX
Thp2p - THO-TREX
- THOC5 THO-TREX Binds TAP.
- THOC6 THO-TREX
- THOC7 THO-TREX
Sub2p UAP56 TREX Recruits Aly to mRNA transcripts.
URH49
Yra1p Aly TREX Recruits TAP/p15 to transcripts.
Mex67p TAP Mex67/Mtr2 (TAP/p15)
Ferries transcripts across the nuclear pore.
Mtr2pa p15 Mex67/Mtr2 (TAP/p15)
Ferries transcripts across the nuclear pore.
Nab2 - Assists release of Aly from the mRNP.
Gle2p Rae1 Associates with TAP at the nuclear pore.
Sac3p GANP TREX-2/THSC Delivers TAP containing mRNPs to the nuclear pore.
Dbp5p Dbp5 Remodeling of the mRNP after crossing the pore.
Gle1p Gle1 Stimulates ATPase activty of Dbp5.
Npl3 SR proteins (ASF/SF2,
9G8, SRp20)
Recruits TAP/p15 to transcripts.
- UIF Recruits TAP/p15 to transcripts.
- Iws1 Recruits ALY to mRNA transcripts. a Mtr2p and p15 share structural but no sequence similarity.
8
essential for mRNA export (Bailer et al., 1998; Blevins et al., 2003; Pritchard et al., 1999; Yoon
et al., 2000). It should be also noted that additional mRNA export adaptors including UAP56
interacting factor (UIF) and Chtop have been shown to be redundant with Aly (Chang et al.,
2013; Hautbergue et al., 2009) likely providing an explanation for the fact that the depletion of
Aly by RNA interference (RNAi) neither impairs the nuclear export of bulk poly(A) mRNA nor
cell viability (Akef et al., 2013; Hautbergue et al., 2009). In addition, there are other NPC-
associated proteins whose roles are not fully understood. In S. cerevisiae, Gle1p stimulates the
helicase activity of Dbp5p at the cytoplasmic face of the pore (Alcázar-Román et al., 2006;
Weirich et al., 2006) where Dbp5p regulates the remodeling of the mRNP complex by removing
TAP/p15 from the mRNA (Lund and Guthrie, 2005). Dbp5p also interacts with Aly (Schmitt et
al., 1999), although the role of this interaction is not understood. Interestingly, while Dbp5p was
initially characterized as a cytoplasmic protein, Dbp5p appears to shuttle between the nucleus
and cytoplasm (Cáceres et al., 1998). Additionally, other “nuclear” proteins involved in mRNA
export, such as Hpr1, THOC5, UAP56, URH49 and Aly, also shuttle to the cytoplasm (Katahira
et al., 2009; Meignin and Davis, 2010; Thomas et al., 2011; Zhou et al., 2000). Finally, a group
of proteins that has been linked to mRNA nuclear export is the serine/arginine-rich (SR)
proteins. These recognize distinct RNA motifs and play various roles in RNA metabolism
including splicing, 3’end processing and export (Long and Caceres, 2009). In S. cerevisiae,
TREX components can recruit SR proteins to the transcript (Hurt et al., 2004), and in
mammalian cells, certain SR proteins promote TAP/p15-dependent export (Hautbergue et al.,
2008; Huang and Steitz, 2001; Huang et al., 2003) and likely accompany the mRNA into the
cytoplasm (Cáceres et al., 1998).
Thus in summary, it appears that there are several ways that components of the TREX
complex can mediate export. These alternatives to the canonical TREX pathway probably reflect
the plasticity of the entire mRNA export system. Importantly, the TREX activity has been shown
to be coupled to several different mRNA processing machineries, each of which are associated
with distinct identity elements.
1.2.1.1 CRM1 regulates the nuclear export of snRNAs and a subset of mRNAs
The Chromosomal Region Maintenance 1 (CRM1) is a member of the family of nuclear
transport receptors that interact with nucleoporins and transport cargoes across the NPC (Köhler
9
and Hurt, 2007). In the nucleus, CRM1 binds its cargo and the small GTPase Ran. Upon
reaching the cytoplasm, the concerted action of the cytoplasmically localized proteins Ran
GTPase Activating Protein 1 (RanGAP1) and Ran Binding Protein 1 (RanBP1) stimulate Ran to
hydrolyze its bound GTP to GDP (Köhler and Hurt, 2007). The conformational change of Ran
promotes the dissociation of the cargo from CRM1. Subsequently, CRM1 and Ran shuttle back
to the nucleus where the nuclear localized Ran Guanine nucleotide Exchange Factor (RanGEF)
protein catalyzes the exchange of GDP for GTP on Ran (Köhler and Hurt, 2007). As the complex
assembles in the nucleus and disassembles in the cytoplasm, the latter compartment acts as a sink
for nuclear export substrates.
CRM1 was initially shown to promote the nuclear export of small nuclear RNA (snRNA)
and nuclear export signal (NES)-containing proteins (Fornerod et al., 1997). While both mRNAs
and snRNAs share many similarities such as being Pol II-mediated transcription products and
having a 5’ cap, snRNAs and mRNAs are funneled through distinct nuclear export pathways.
The two RNA species are differentiated by the ability of the snRNA export adaptor PHAX to
selectively bind RNA species that are typically shorter than mRNAs and range from 200 to 300
nucleotides (Masuyama et al., 2004). Subsequently, PHAX recruits CRM1 which in turn ferries
the snRNA across the NPC to the cytoplasm. A recent study has shown that the heterogenous
nuclear ribonucleoproteins C1/C2 (hnRNPC1/C2) compete with PHAX for binding to mRNAs
(McCloskey et al., 2012). Moreover, depletion of hnRNPC1/C2 by RNAi enhances the
recruitment of PHAX to mRNA transcripts leading to their nuclear retention through an
unknown mechanism (McCloskey et al., 2012). These authors proposed that the hnRNPC1/C2
complex acts as a measuring stick in order to identify long RNAs (McCloskey et al., 2012).
These studies lent credence to the notion that the length of a transcript can serve as one potential
avenue for eukaryotic cells to distinguish various RNA species in the nucleus.
CRM1 and RanGTP were thought to be dispensible for mRNA nuclear export (Clouse et
al., 2001; Neville and Rosbash, 1999) However, several studies have suggested a role for CRM1
in mediating the nuclear export of several mRNAs in mammalian cells (Kimura et al., 2004;
Culjkovic et al., 2006). In one example, it was reported that the nuclear export of human
interferon-α mRNA required CRM1 (Kimura et al., 2004) while the TREX complex was
required in another study (Lei et al., 2011). This raises the possibility that CRM1 may in some
cases interact with canonical mRNA nuclear export factors such as the TREX complex and the
heterodimer TAP/p15 to regulate mRNA export to the cytoplasm.
10
1.2.2 Sorting promoter-driven transcripts from spurious transcription Cells need to identify transcripts emanating from Pol II rather than other polymerases.
This is accomplished through a particular feature of the largest subunit of Pol II, its carboxy-
terminal domain (CTD). This flexible extension is composed of a heptad amino acid sequence
(YSPTSPS) that is repeated in tandem (there are 52 repeats in the human Pol II) and acts as a
modifiable scaffold which recruits various complexes to Pol II, depending on the enzyme’s
history (for reviews see (Buratowski, 2009; Moore and Proudfoot, 2009)). For example, the
promoter-associated TFIIH complex promotes the phosphorylation of the fifth serine (Ser5) of
the CTD heptad repeats (YSPTSPS) (Feaver et al., 1994). This modified CTD recruits the
enzymes responsible for adding a 7-methylguanine cap to the nascent mRNA transcript (Cho et
al., 1997; McCracken et al., 1997; Yue et al., 1997). The 5’cap allows these transcripts to be
distinguished from those originating from Pol II initiation events at non promoter sequences. The
5’cap then recruits the nuclear cap binding complex (CBC), which in turn physically interacts
with Aly, thereby loading the TREX complex on the 5’ end of the transcript (Cheng et al., 2006).
Dbp5p has also been shown to associate with the 5’ end of newly transcribed Balbiani ring
transcripts in Chironomus tetans (Zhao et al., 2002) and may be recruited there through
interactions with Aly (Schmitt et al., 1999) or transcription factors such as TFIIH (Estruch and
Cole, 2003).
Additional promoter-proximal DNA elements may act to further identify protein-coding
transcripts. For example, upon gene activation in S. cerevisiae and Drosophila melanogaster (D.
melanogaster), a physical bridge is formed between upstream activation sequences (UAS) and
the NPC in a process referred to as gene gating (Blobel, 1985; Casolari et al., 2004; Kurshakova
et al., 2007). This phenomenon is thought not only to enhance the rate of gene expression, but
also to help reactivate genes at a later time point in a process known as transcriptional memory
(Brickner, 2009). In certain cases, gene gating in both S. cerevisiae and D. melanogaster is
mediated by the TREX-2/THSC complex, which forms a bridge between the UAS-associated
transcription co-activator SAGA and the NPC (Cabal et al., 2006; Fischer et al., 2002;
Kurshakova et al., 2007; Luthra et al., 2007). Importantly, the S. cerevisiae TREX-2/THSC
complex has been shown to be required for nuclear mRNA export (Fischer et al., 2002, 2004; Lei
et al., 2003; Rodríguez-Navarro et al., 2004). A key component of the S. cerevisiae TREX-
2/THSC complex is Sac3p, which can associate with both TAP/p15 and nucleoporins through
two distinct domains (Fischer et al., 2002). Despite the presence of at least one Sac3p homolog
11
in humans (GANP) (Jani et al., 2009; Wickramasinghe et al., 2010, 2014), there is no evidence
for gene gating in vertebrates (Perales and Bentley, 2009). GANP, however, is required for
mRNA export (Wickramasinghe et al., 2010, 2014). It is thus likely that the TREX-2/THSC has
other functions beyond gene gating, but these have yet to be elucidated.
It is thus clear that the majority of promoter-driven transcription events result in an RNA
molecule with associated factors that are fundamentally different from the products of spurious
transcription. In this context, conserved promoters with the right proximal-promoter elements
recruit machinery that then marks the nascent transcripts. These identity marks include the 5’cap
and the loading of proteins (such as Aly, Dbp5 and perhaps Sac3p/GANP) onto the transcript.
1.2.3 Coupling transcription elongation to mRNA nuclear export One important event that occurs during Pol II transcription is promoter clearance.
Promoter clearance allows initiated Pol II to progress to the elongation phase of transcription and
is likely a step where the gene expression machinery evaluates mRNA identity (Svejstrup, 2004).
This process requires phosphorylation of Ser2 in the Pol II CTD heptad repeats (YSPTSPS) by
the Ctk1p kinase in budding yeast and P-TEFb (positive transcription elongation factor b) in
mammalian cells (Bartkowiak and Greenleaf, 2011; Lenasi and Barboric, 2010). The successful
progression to this phase of transcription is coupled to the budding yeast homolog of Aly, which
interacts with the Pol II CTD that is phosphorylated on both Ser2 and Ser5 (MacKellar and
Greenleaf, 2011). A second important coupling event is the recruitment of the THO complex to
the transcript during elongation (Chávez et al., 2000; Strässer et al., 2002), although the exact
molecular details remain elusive. In S. cerevisiae, the THO complex may form a direct physical
interaction with Ctk1p kinase, and these two factors in turn recruit SR proteins to the CTD (Hurt
et al., 2004). PAF, another complex that is involved in Pol II elongation, also directly associates
with the THO complex (Chang et al., 1999). In D. melanogaster, the THO complex is recruited
to elongating Pol II by ENY2, a component of both TREX-2/THSC and SAGA complexes
(Kopytova et al., 2010). Interestingly, the THO complex is also required to promote transcription
elongation in S. cerevisiae, especially through regions containing high GC-content (Chávez et
al., 2001), and this function in transcription appears to be conserved in mammals (Wang et al.,
2006). Thus, once the nascent transcript has acquired sufficient identity marks and has recruited
the THO complex, not only is its elongation promoted, but the now localized THO complex may
feedback to promote the generation of subsequent transcripts from that same genetic locus.
12
Another set of factors that associate with transcribed regions and can thus serve to impart
identity marks onto the RNA are nucleosome remodeling complexes. These include the
facilitates chromatin transcription (FACT) complex, which disassembles nucleosomes and thus
enhances Pol II transcription elongation, and Spt6, which associates with Phospho-Ser2 CTD and
helps to reassemble these dissociated nucleosomes (Belotserkovskaya and Reinberg, 2004; Yoh
et al., 2007). Interestingly, disruption of either factor promotes spurious transcription, indicating
that they help to distinguish protein-coding regions from the rest of the genome (Kaplan et al.,
2003; Mason and Struhl, 2003). This may have to do with the maintenance of histone
modifications at protein-coding loci, which can serve as additional identity marks for protein-
coding genes. Importantly, these nucleosome remodeling factors also interact with TREX in
mammalian systems. Thus Spt6 binds to Aly through its interacting partner Iws1 (interacts with
Spt6) (Yoh et al., 2007, 2008), while FACT recruits UIF to the mRNA (Hautbergue et al., 2009).
UIF and Aly may work in redundant pathways to recruit TAP/p15 to the transcript (Hautbergue
et al., 2009).
Thus it appears that histone modifications, which provide an identity mark for protein-
coding genes, are coupled to TREX-dependent export through these remodeling factors. This
coupling could also work in the reverse direction: once a transcript has satisfied sufficient criteria
of mRNA identity and has thus acquired TREX, this may help Pol II acquire nucleosome
remodeling complexes that further promotes transcription.
1.2.4 Coupling splicing to mRNA nuclear export Splicing is a highly effective mRNA identity mark. Although the vast majority of human
protein-coding transcripts are spliced, most non-coding intergenic transcripts are not (van Bakel
et al., 2010). This mark may not be so critical in organisms that have few introns and small
intergenic regions, such as S. cerevisiae. Indeed, in most cases the elimination of introns from a
budding yeast gene has only a small impact on its expression and on the fitness of the mutant
strain (Parenteau et al., 2008), although exceptions do exist (Juneau et al., 2006; Parenteau et al.,
2008, 2011). However, the situation in organisms with low population numbers is quite different.
These organisms tend to have larger intergenic regions and more introns. Indeed in vertebrates
where many promoters are bidirectional, it was shown that recognition sites of the U1 snRNP
component of the splicing machinery are enriched in the correct forward transcript and depleted
from the reverse transcript (Almada et al., 2013). This pattern contrasted that seen with
polyadenylation sites (PAS) which are more enriched in the reverse transcripts. Interestingly,
13
disrupting the function of U1 snRNP caused a spike in promoter proximal cleavage and
polyadenylation of forward transcripts suggesting that U1 snRNP recruitment to nascent
transcripts inhibits premature 3’ cleavage (Almada et al., 2013). Moreover, splicing has a major
impact on several other RNA processes, such as 3’ cleavage (McCracken et al., 2002),
polyadenylation (Lutz et al., 1996), and nuclear export (Luo and Reed, 1999). Splicing was also
proposed to be required for promoting mRNA nuclear export in metazoans (Cheng et al., 2006;
Masuda et al., 2005). This last activity is mediated by the interaction of UAP56 with components
of the spliceosome (Fleckner et al., 1997; Luo et al., 2001; Masuda et al., 2005). It should also be
noted that UAP56 was originally identified as being required for proper splicing (Fleckner et al.,
1997) and spliceosome maturation (Shen, 2009). Thus, the recruitment of the TREX complex to
a transcript by other identity elements may in turn enhance the processing of proper mRNAs
over spurious transcripts. It is also likely that the mRNA nuclear export machinery evaluates
splicing post-transcriptionally. For example, the TREX complex is required for the splicing-
dependent export of microinjected intron-containing pre-mRNAs in both Xenopus lavis (X. lavis)
oocytes (Luo and Reed, 1999) and human cells (Palazzo et al., 2007). It should be noted that
many intronless mRNAs are efficiently exported in mammalian cells (Nott et al., 2003; Lu and
Cullen, 2003) and this is dependent on TREX activity (Hautbergue et al., 2009; Taniguchi and
Ohno, 2008). These results suggest that even in organisms that have large amounts of intergenic
DNA, splicing is not absolutely required for the export of an mRNA from the nucleus to the
cytoplasm.
Splicing also leads to the deposition of the exon junction complex (EJC), which is
involved in nonsense-mediated decay (Le Hir et al., 2001a), temporal regulation of mRNA
stability (Giorgi et al., 2007), the enhancement of translation (Ma et al., 2008; Nott et al., 2004)
and the proper cytoplasmic localization of mRNA (Hachet and Ephrussi, 2001, 2004; Le Hir et
al., 2001b). Although it has been reported that the EJC also interacts with TREX components (Le
Hir et al., 2001b; Schmidt et al., 2006) and SR proteins (Singh et al., 2012), the EJC is not
required for mRNA nuclear export (Nott et al., 2003; Palazzo et al., 2007). However, it is
possible that the physical interaction of these two complexes may again give an advantage to the
processing, export and translation of protein-coding mRNAs over spurious transcripts.
1.2.5 Coupling of 3’ processing to mRNA nuclear export The final identity mark that is associated with a protein-coding mRNA is the 3’ poly(A)
tail. This hallmark of protein-coding transcripts is added to the 3’ end due to the action of various
14
other mRNA identity cues. In S. cerevisiae, the remaining phospho-Ser2 CTD recruits the
cleavage factor (CF1A) and cleavage polyadenylation factor (CPF), which are required for
proper 3’processing. A second cue is the polyadenylation signaling elements, found near the end
of the transcript. These two identity elements are not only used to mark the transcript with a
poly(A) tail, but are directly coupled to TREX activity. In early experiments it was noted that as
polyadenylation factors were recruited to a transcript, the THO complex fell off, indicating that
the former displaced the latter from the mRNA (Kim et al., 2004). In addition, direct links have
been made between TREX and 3’processing machines. For example, in S. cerevisiae, 3’ end
processing signals are required to recruit yeast Aly to certain transcripts (Lei and Silver, 2002),
and this is likely mediated by the CF1A component Pcf11p (Johnson et al., 2009). Mutations of
TREX components in the budding yeast cause the accumulation of CF1A components and
TAP/p15 in mRNA-chromatin-protein complexes that are trapped at the NPC (Rougemaille et
al., 2008). Thus it is likely that TREX communicates with the 3’ end processing machinery and
if this fails, due to insufficient TREX-recruitment or weak 3’processing signals, nuclear export is
blocked.
TREX-recruitment also affects the degree of 3’ end processing. S. cerevisiae cells
harboring deletions in the TREX components sometimes show defects in the 3’ end cleavage of
certain transcripts (Libri et al., 2002) and defects in polyadenylation in others (Saguez et al.,
2008). Thus, other identity cues may feed forward to enhance polyadenylation through TREX. In
this light, the poly(A) tail is an mRNA identity mark that is not only responsive to 3’processing
cues and Pol II status, but also to other identity elements. One would thus expect that like
splicing, long poly(A) tails should strongly promote export, and this is exactly what was
observed in X. lavis oocyte mRNA microinjection experiments (Fuke and Ohno, 2008).
Although the mechanism in vertebrates remains unclear, the nuclear poly(A) binding protein
(Pabp2) is required for mRNA export in D. melanogaster (Farny et al., 2008). In S. cerevisiae,
the yeast-specific poly(A) binding protein Nab2p has been implicated in export (Green et al.,
2002; Hector et al., 2002) and forms a complex with yeast Aly and TAP/p15 on the
nucleoplasmic face of the NPC (Fasken et al., 2008; Iglesias et al., 2010).
Again, these various coupling events allow TREX to assess 3’end identity marks.
Moreover, it is possible that gene looping, or the proximity of the 5’ and 3’ ends of a highly
transcribed gene (Lainé et al., 2009), may allow the TREX complex to monitor identity marks
simultaneously on the 5’ and 3’ ends of a given transcript. However, it is apparent that there is a
15
complicated relationship between the TREX complex and the 3’ end. It appears that at this stage
TREX may play a role in mRNP quality control both positively and negatively. This topic will
be explored in the next section.
1.2.6 Nuclear export as an arbiter of mRNA quality control Most spurious transcripts are quickly degraded by the TRAMP poly(A) polymerase
complex and the nuclear exosome (Chekanova et al., 2007; Davis and Ares, 2006; Neil et al.,
2009; Preker et al., 2008; Thiebaut et al., 2006; Vasiljeva et al., 2008; Wyers et al., 2005; Xu et
al., 2009). It is thus likely that the failure to recruit the TREX complex redirects the mRNA to
this degradation machinery. This is supported by the observation that TREX inhibition in S.
cerevisiae causes the degradation of nuclear localized mRNAs by several complexes, which
include not only TRAMP, and the nuclear exosome, but also the CCR4-Not deadenylase
complex (Assenholt et al., 2011; Rougemaille et al., 2007). Indeed, several large-scale genetic
experiments have identified strong genetic interactions between TREX and components of the
TRAMP and exosome complexes (Milligan et al., 2008; Wilmes et al., 2008; Zenklusen et al.,
2002). However it appears that the TREX complex also plays an active role in this rerouting. For
example Yra1p, Nab2p and Dbp5p physically interact with the Ccr4-Not complex (Kerr et al.,
2011).
Another critical quality control step is the evaluation of mRNP composition and RNA
processing by Mlp1p/2p, two large proteins present at the nucleoplasmic face of the NPC (Galy
et al., 2004; Iglesias et al., 2010). If the mRNP passes this filter, Aly is ubiquitinated and
released from the Nab2p/TAP/p15 complex (Iglesias et al., 2010). It appears that this remodeling
occurs in response to quality control assessment and is required for export. In the event that
export fails it is likely that these transcripts are degraded by a specialized endonuclease Swt1p
(synthetic with TREX 1) (Skruzný et al., 2009). Since Mlp1 and 2 also interact with the Ccr4-
Not (Kerr et al., 2011), it is likely that this complex is also involved in the degradation of
aberrant transcripts. It is thus clear that mRNA nuclear export factors cross talk to mRNA
degradation complexes in order to partition transcripts that fail to meet a certain threshold of
mRNA identity.
Moreover, TREX activity has been linked to small interfering RNA (siRNA) mediated
heterochromatin formation and gene silencing. In S. pombe, siRNAs are generated from highly
repetitive heterochronic regions of the genome and are required for their transcriptional silencing
and chromatin remodeling (Moazed, 2009). This process is likely conserved in metazoans and
16
used to prevent the proliferation of transposons. Interestingly, S. pombe cells harboring
mutations in the Aly homolog Mlo3 show defects in siRNA production and no longer silence
heterochronic regions (Zhang et al., 2011). In plants, the THO complex is also required for the
production of siRNAs that are involved in post-transcriptional silencing (Jauvion et al., 2010;
Yelina et al., 2010). This putative role of TREX in heterochromatin silencing is likely conserved
and extends beyond siRNA generation, as the mutation, deletion or overexpression of the S.
cerevisiae homolog of UAP56 also affects siRNA-independent gene silencing at heterochronic
loci (Fan et al., 2001; Lahue et al., 2005; West and Milgrom, 2002), and the deletion of UAP56
in D. melanogaster suppresses the spreading of heterochromatin to subtelomeric regions (Eberl
et al., 1997). These studies lead to the provocative model that TREX components act to evaluate
transcripts and determine whether they contain sufficient mRNA identity elements, or whether
they are generated from heterochromatic loci or transposons and should be shunted towards a
silencing pathway.
1.2.7 Coupling mRNA nuclear localization to mRNA export The gene gating hypothesis postulated that genes are positioned in a non-random pattern
in proximity to nuclear pores (Blobel, 1985). As a result, mRNAs transcribed from a given gene
are transported to the cytoplasm through the nearest NPC to that gene (Blobel, 1985). Indeed, it
has been shown that transcriptional activation targets certain genes to the NPC in S. cerevisiae
(Casolari et al., 2004). Moreover, previous studies in S. cerevisiae indicate that certain mRNPs
are rendered export-competent in the vicinity of the nuclear pore (Cabal et al., 2006; Fasken et
al., 2008; Pascual-García et al., 2008). Intriguingly, gene gating has not been observed in mammalian cells (Hocine et al., 2010),
suggesting that the majority of mRNP formation occurs elsewhere in the nucleoplasm. Several
nonmembranous nuclear compartments have been discovered in mammalian cell nuclei
including nucleoli, nuclear speckles (also known as interchromatin granule clusters (IGC)),
paraspeckles and cajal bodies (Spector and Lamond, 2011). Among the various nuclear
compartments, the role of nuclear speckles as a potential site for transcription and mRNP
processing is particularly interesting. Indeed, it was shown that the chromosomal gene-poor
regions (G-bands) are less likely to be present along the edges of nuclear speckles than the gene-
rich regions (R-bands) (Shopland et al., 2003). Furthermore, components of the transcription
eukaryotic cells distinguish mRNAs derived from protein-coding loci. In addition, sequence
features within a transcript confer information as to whether this transcript should be exported to
the cytoplasm or not. While none of these mRNA identity marks in isolation can be used to
separate protein-coding RNA from spurious transcripts; I propose that if a particular transcript
meets a critical threshold of these identity marks, then it will be efficiently transported to the
cytoplasm and translated to protein. Based on this model, it is likely that each transcript is
independently analyzed, and depending on its particular features, a subset of mRNA export
factors is engaged to export that transcript to the cytoplasm. This might explain why each
budding yeast and fly mRNA requires only a subset of RNA export factors for their efficient
export (Hieronymus and Silver, 2003; Kim Guisbert et al., 2005; Rehwinkel et al., 2004). It
should also be noted that there are other important quality control mechanisms that take
advantage of the coupling between gene expression steps to eliminate mis-processed mRNAs
such as the nonsense-mediated decay (NMD) pathway (Le Hir et al., 2001a; Lejeune and
Maquat, 2005). Indeed, it is likely that many biological processes operate as networks of
feedback and feed-forward loops rather than strictly linear pathways, and it is through these
interconnections that true biological function can emerge.
21
1.5 Rationale of thesis research My thesis research investigated the molecular mechanisms that regulate mRNA nuclear
export in mammalian cells. I sought to examine the sequence features that control the transport
of an mRNA to the cytoplasm. I also sought to unravel how these sequence features define the
nuclear localization where the early steps of mRNP assembly occur.
1.5.1 The TREX complex is required for the egress of export-competent ftz mRNPs from nuclear speckles Rationale: Previous studies have demonstrated that several pre-mRNA splicing factors and
nuclear export factors localize to nuclear speckles (Dias et al., 2010; Kota et al., 2008; Masuda et
al., 2005). These results supported that pre-mRNA splicing and the recruitment of mRNA
nuclear export factors are spatially coupled in nuclear speckles. The crucial role of nuclear
speckles in proper mRNP assembly was further supported by the fact that depleting several
components of the TREX complex not only inhibited mRNA transport to the cytoplasm but also
resulted in trapping the mRNA in speckles (Dias et al., 2010). In contrary to the splicing-
dependent mRNA nuclear export model, several groups have reported that splicing is not a strict
requirement for mRNA nuclear export (Lu and Cullen, 2003; Nott et al., 2003) and that
components of the TREX complex are recruited to intronless mRNAs (Taniguchi and Ohno,
2008). However, it was unknown whether the assembly of intronless mRNAs into nuclear
export-competent mRNPs also occurs in nuclear speckles. I sought to investigate some of the
early steps of intronless mRNP assembly in the nuclei of mammalian cells.
Hypothesis: I hypothesized that nuclear export-competent intronless mRNAs also traffic through
nuclear speckles and that speckle association is dependent on sequence features present within
the mRNA.
Results: I discovered that the intronless fushi tarazu (ftz) mRNA traffics through the nuclear
speckle compartment on its route to being exported to the cytoplasm. I also discovered that
depleting the expression of the TREX components UAP56 and URH49 led to ftz mRNA
entrapment in speckles. In contrast, intronless human β-Globin (βG) mRNA was not efficiently
exported to the cytoplasm as previously reported (Valencia et al., 2008), and did not traffic
through speckles. The splicing of βG mRNA was sufficient to both promote the localization of
βG mRNA to nuclear speckles and promote its transport to the cytoplasm. Moreover, I also
investigated the role of CRM1 in mediating the nuclear export of ftz mRNA. CRM1 inhibition
22
led to the retention of ftz mRNAs in nuclear speckles, although it remains unclear if this is a
direct or indirect effect. These results suggest i) that a subset of intronless mRNAs traffic
through nuclear speckles, and ii) that components of the TREX complex and CRM1 are likely
required for mRNP maturation and release from these intra-nuclear domains.
1.5.2 Splicing promotes the nuclear export of β-Globin mRNA by overcoming nuclear retention elements Rationale: There has been an ongoing debate as to whether splicing is required for efficient
mRNA nuclear export in metazoans (Luo and Reed, 1999; Nott et al., 2003; Valencia et al.,
2008). The evidence supportive of splicing being required for an mRNA to be efficiently
exported to the cytoplasm was partially based on studies which used human βG mRNA as a
model reporter (Valencia et al., 2008). While spliced βG mRNA (βG-i) is efficiently exported to
the cytoplasm, intronless βG mRNA (βG-∆i) is retained in the nucleus (Valencia et al., 2008). In
contrast to βG mRNA, several intronless transcripts were shown to be efficiently exported to the
cytoplasm independently of splicing (Nott et al., 2003; Lu and Cullen, 2003).
Hypothesis: I hypothesized that βG-∆i mRNA contained an RNA element that impeded its
transport to the cytoplasm and that this element can be overcome when βG mRNA is spliced.
Results: I discovered that βG mRNA contained nuclear retention elements in the 3’ end of the
mRNA that inhibited its transport to the cytoplasm. These nuclear retention elements could be
overcome when βG mRNA was either spliced or when the length of the mRNA was extended.
These results provided an illustration of how mRNA identity, and thus its “exportability”, is
defined by the aggregate sum of the various nuclear export-promoting and export-inhibitory
features present within an mRNA.
23
Chapter 2
The TREX complex is required for the egress of export-competent
ftz mRNPs from nuclear speckles
Parts of this chapter were published in:
Akef, A., Zhang, H., Masuda, S., and Palazzo, A.F. (2013). Trafficking of mRNAs containing
ALREX-promoting elements through nuclear speckles. Nucleus 4, 326–340.
I generated all the data presented in this chapter.
24
2.1 Summary
The nuclear speckle is a compartment within the nucleus where pre-mRNA splicing is
thought to occur. In addition, the TREX complex localizes to speckles where it is thought to be
recruited to the mRNA in a splicing dependent manner. Here, I demonstrate that the efficiently
exported intronless ftz mRNP is trafficked through nuclear speckles. In contrast, intronless βG
mRNA, which is not efficiently exported to the cytoplasm, did not co-localize with speckles.
Depletion of two TREX-associated RNA helicases, UAP56 and its paralog URH49, not only
inhibits ftz transport to the cytoplasm but also appears to trap these mRNAs in nuclear speckles.
Moreover, I show that intronless ftz mRNAs associate with UAP56 in vivo. Inhibiting the activity
of nuclear export receptor CRM1 also impairs the transport of ftz mRNAs to the cytoplasm and
causes their entrapment in nuclear speckles. In contrast, CRM1 inhibition did not affect the
export of spliced βG mRNA. By analyzing ftz-βG fusion constructs, I determined that the
splicing of the βG transcript renders the mRNA insensitive to CRM1. These results suggest that
RNA sequences within a transcript determine nuclear speckle-association where mRNA nuclear
export factors are recruited to the mRNA and thus promote its transport to the cytoplasm.
25
2.2 Introduction
In eukaryotes, mRNAs are synthesized in the nucleus where they are capped, spliced, and
polyadenylated. During these maturation steps, the mRNAs are loaded with several proteins to
form mRNP particles that are capable of crossing the nuclear pore to reach the cytoplasm where
the mRNAs are translated into proteins (Palazzo and Akef, 2012). It is believed that the majority
of mRNAs in vertebrate cells are exported from the nucleus in a splicing dependent export
pathway (Luo and Reed, 1999). This pathway is initiated during splicing of the first intron where
the spliceosome and the nuclear cap binding complex collaborate to deposit the TREX complex
at the 5’end of the mRNA (Cheng et al., 2006; Masuda et al., 2005). TREX is a multiprotein
complex that is composed of the THO subcomplex, the RNA helicase UAP56, the adaptor
molecules Aly and Chtop (Chang et al., 2013; Luo et al., 2001; Strässer et al., 2002; Zhou et al.,
2000). This complex acts to recruit the heterodimeric mRNA export receptor TAP/p15, which
directly binds to the mRNA (Viphakone et al., 2012) and ferries it across the nuclear pore to the
cytoplasm (Katahira et al., 1999; Segref et al., 1997; Strässer and Hurt, 2000).
Although it is generally believed that the nuclear export receptor CRM1 is not required
for the transport of mRNAs to the cytoplasm, many lines of evidence suggest that this might be
an oversimplification. The export of certain mRNAs, such as interferon-α mRNA, is sensitive to
both CRM1 (Kimura et al., 2004) and TAP (Lei et al., 2011) inhibition. This raises the
possibility that these two export pathways collaborate to promote nuclear export. Interestingly,
members of the Palazzo lab have observed that the nuclear export of ftz mRNA is sensitive to the
inhibition of RanGTP (H. Zhang and A.F. Palazzo, unpublished data). Since TAP/p15-dependent
export does not require the Ran gradient (Clouse et al., 2001), but CRM1-dependent export does,
these observations raise the interesting possibility that the nuclear export of ftz mRNA is partially
dependent on CRM1.
It is unclear where in the nucleoplasm the assembly of these mRNPs takes place. One
potential subnuclear compartment where mRNP assembly may occur is nuclear speckles. These
are nuclear aggregates that contain active Pol II, spliceosomal components, and splicing
cofactors such as SR proteins (Spector and Lamond, 2011). In addition, these structures contain
TREX complex components (Dufu et al., 2010; Kota et al., 2008; Masuda et al., 2005; Zhou et
al., 2000). Many mRNAs appear to be recruited to nuclear speckles by the act of splicing (Dias et
al., 2010; Tokunaga et al., 2006; Wang et al., 1991). In the vicinity of speckles, splicing is
26
completed and the mRNA acquires components of the exon junction and likely TREX
complexes (Daguenet et al., 2012; Schmidt et al., 2006; Vargas et al., 2011). Depletion of
UAP56, enhances the association of these spliced mRNAs with nuclear speckles (Dias et al.,
2010), suggesting that the TREX complex is required for mRNAs to exit these structures.
Finally, it is likely that TAP/p15 is itself loaded onto spliced transcripts within or in the vicinity
of nuclear speckles (Schmidt et al., 2006; Teng and Wilson, 2013). In contrast to spliced
mRNAs, it is less clear whether intronless mRNPs also associate with nuclear speckles prior to
their export to the cytoplasm.
In this current work, I characterized some of the earliest events in the assembly of ftz
mRNP. My data show that ftz mRNP traffics through nuclear speckles prior to being exported to
the cytoplasm. My data suggest that within speckles, mRNAs undergo a series of mRNP
maturation steps and that the TREX-associated RNA helicases UAP56 and URH49, and CRM1,
are required for the eventual release of ftz mRNAs from these structures.
27
2.3 Materials and Methods
Plasmid constructs. The MHC-ftz-Δi, c-ftz-Δi, c-ftz-i and INS-Δi constructs in pCDNA3 were
described previously (Palazzo et al., 2007). Human βG introns were amplified from U2OS
genomic DNA and inserted into pcDNA3 mammalian expression vector containing βG cDNA
(Valencia et al., 2008) by restriction-free cloning (van den Ent and Löwe, 2006) using the
following primer sequences, forward primer:
GTGGTGAGGCCCTGGGCAGGTTGGTATCAAGGTTACAAG and the reverse primer:
GACCAGCACGTTGCCCAGGAGCTGTGGGAGGAAGATAAG. The mouse MHC SSCR,
which comprises the first 66 nucleotides from the mouse H2kb gene, was first constructed using
the following primers—forward primer:
CAAAAACTCATCTCAGAAGAGGATATGGTACCGTGCACGCTGCTCCTGCTGT and the
reverse primer 3: CTCCTCAGGAGTCAGATGCACCGCGCGGGTCTGAGTCGGAGC. This
product was then used in a subsequent PCR reaction to insert the MHC SSCR into the βG-
containing vector by restriction-free cloning. Subsequent to PCR amplification, products were
treated with DpnI (New England Biolabs) and incubated at 37 °C for 3–12 h. Products were then
purified using PCR purification kits (Qiagen). DH5α E. coli cells were transformed with the
cloned plasmids. MHC-ftz-βG-Δi and MHC-ftz-βG-i were constructed by amplifying MHC-ftz-Δi
using a reverse primer that contained a HindIII site just upstream of the stop codon. This PCR
product was digested with HindIII and ligated into either βG-Δi or βG-i pCDNA3 respectively
that was cut with the same enzyme. c-ftz-i-βG-i constructed by amplifying c-ftz-i using a reverse
primer that contained a HindIII site just upstream of the stop codon. This PCR product was
digested with HindIII and ligated into βG-i pCDNA3 that was cut with the same enzyme.
Cell lines and antibodies. Both human osteosarcoma (U2OS) and embryonic kidney 293T
(HEK293T) were maintained in high glucose DMEM (Wisent) containing 10% FBS (Wisent)
and antibiotics (Sigma). For CRM1 inhibition experiments, U2OS cells were treated with 20 nM
Leptomycin B (LMB).
The following antibodies were used: rat polyclonal anti-UAP56 (Yamazaki et al., 2010),
rabbit polyclonal anti-UAP56 (Sigma), rat polyclonal anti-URH49 (Yamazaki et al., 2010),
rabbit polyclonal anti-Aly (Zhou et al., 2000), rabbit polyclonal anti-PHAX (Masuyama et al.,
Each column represents the average of three experiments, each consisting of 100 SC35-positive
speckles (see methods section for more details). Error bars represent standard error of the mean.
(E) For each mRNA, FISH images from one set of nuclei (1 hour post-microinjection), were
superimposed over SC35 images from a separate set of un-injected nuclei to determine the rate
of random colocalization. The data was analyzed and plotted as in (D). (F) The percentage of
speckles that colocalized with different transcripts was plotted for cells, 1 hour after
microinjection of plasmids. Again, the data was analyzed and plotted as in (D). (G-H) Plasmids
containing the MHC-ftz-Δi gene were microinjected into the nuclei of U2OS cells and mRNA
was allowed to express for 20 minutes. Cells were then treated with α-amanitin and the
colocalization of MHC-ftz-Δi mRNA with SC35 over time (post-drug treatment, indicated on the
x-axes) was monitored. (G) The percentage of nuclear speckles that demonstrate different levels
of colocalization with MHC-ftz-Δi. Again, the data was analyzed and plotted as in (D). (H) The
amount of MHC-ftz-Δi mRNA that is present in nuclear speckles (as defined by the brightest
20% pixels in the nucleus, using SC35 immunofluorescence) as a percentage of either the
cellular or nuclear level. Each data point represents the average and standard error of the mean of
37
10 cells. (I) In vitro transcribed MHC-ftz-Δi mRNA was microinjected into the nuclei of U2OS
cells. Cells were left for various amounts of time to allow mRNA export. The amount of MHC-
ftz-Δi mRNA present in nuclear speckles was monitored as described in Figure 2.2H.
38
Figure 2.3. MHC-ftz-Δi mRNA, but not dextran, colocalizes with SC35-containing nuclear
speckles. (A) U2OS were microinjected with DNA plasmid that contains the MHC-ftz-Δi
construct along with the microinjection marker 70kDa Dextran. After 1 hour, cells were probed
for ftz mRNA, immunostained for the speckle marker SC35 and stained for DNA using DAPI.
All panels are from a single field of view. Overlays of either MHC-ftz-Δi or Dextran (red) and
SC35 (green) are shown. Scale bar = 5 µm. (B) The fluorescence intensities (y-axis) of either
MHC-ftz-Δi or Dextran (red) and SC35 (green) were plotted along the length of the arrow (x-
axis) as seen in the overlay images in (A).
39
between dextran and SC35 (Fig. 2.2D), and colocalization between βG FISH and SC35 from
different nuclei (Fig. 2.2E)].
While both intronless ftz mRNAs (MHC-ftz-Δi and c-ftz-Δi) associated with speckles,
neither intronless βG transcripts (MHC- βG-Δi and βG-Δi) localized to speckles 1 hour after the
plasmids were microinjected (Fig. 2.2F). Intronless human insulin mRNA (INS-Δi), an intronless
mRNA which is efficiently exported (Palazzo et al., 2007) also localized to speckles (Fig. 2.2F)
suggesting that this distribution was not unique to ftz. From these observations, I concluded that
various reporter mRNAs appear to have different abilities to localize to nuclear speckles. These
data suggest that mRNAs that associate with nuclear speckles were more efficiently exported to
the cytoplasm. Furthermore, robust speckle targeting was dependent on particular features within
the transcript.
2.4.3 MHC-ftz-Δi traffics through nuclear speckles I next investigated whether these mRNAs could target to the speckles post-
transcriptionally. To test this idea, plasmids containing MHC-ftz-Δi were microinjected and after
20 minutes the transcriptional inhibitor α-amanitin was added to the cells. This treatment
completely inhibits transcription of microinjected plasmids within 5 minutes (Gueroussov et al.,
2010). The distribution of MHC-ftz-Δi at various time points after transcriptional shut-down was
monitored. Our analysis indicated that over time, newly synthesized MHC-ftz-Δi transcripts
increased their degree of colocalization with SC35 (Fig. 2.2G), suggesting that they can target to
these structures post-transcriptionally.
Although colocalization studies can indicate whether a particular mRNA is enriched in
speckles, they do not indicate how much of that particular transcript partitions into these
structures. Moreover, monitoring the total level of speckle-associated mRNA over time may
provide insights into the kinetics of this process. To examine whether the localization to speckles
represented a transient event, we used the SC35 immunofluorescence signal to subdivide nuclei
into nuclear speckle regions and non-speckle regions (for details see the materials and methods
section) and monitored what fraction of MHC-ftz-Δi was present in these zones after α-amanitin
treatment. To limit the amount of variation between measurements, nuclear speckles were
defined by thresholding the brightest 20% (+/- 0.5%) of pixels in each nucleus using SC35
immunofluorescence. Generally, the amount of speckle-associated MHC-ftz-Δi mRNA decreased
over time (Fig. 2.2H, see Speckle/ Total mRNA). This result implied that this mRNA was
trafficking out of the nuclear speckles over the time course, although the possibility of enhanced
40
degradation of speckle-associated mRNAs can not be definitively ruled out. Interestingly, when
only the nuclear MHC-ftz-Δi mRNA levels were assessed, the amount associated with speckles
only slightly decreased over the same period (Fig. 2.2H, Speckle/Nuclear mRNA). From these
measurements, we could not determine whether any of the mRNA in the speckle was targeted
post-transcriptionally. However this data suggested that the partitioning of mRNA between the
non-speckle and speckle regions was close to equilibrium.
To obtain a clearer picture of post-transcriptional mRNA trafficking through nuclear
speckles, we microinjected in vitro synthesized, capped, and polyadenylated mRNA into nuclei
and measured the partitioning of this into speckles over time. Note that microinjected mRNA is
exported at a higher rate than endogenously transcribed transcripts. For example, the half-time of
export for microinjected MHC-ftz-Δi mRNA is about 15 minutes, while the figure for the same
mRNA that is transcribed endogenously from plasmids is 40–50 minutes (Palazzo et al., 2007). I
found that microinjected MHC-ftz-Δi mRNA very rapidly accumulated into nuclear speckles, and
this peaked at about 10 minutes post injection (Fig. 2.2I, Speckle/Nuclear mRNA). After this
point the amount of mRNA in nuclear speckles decreased. This result confirmed that mRNA was
likely trafficking through nuclear speckles and that this localization could occur post-
transcriptionally.
In summary these results suggest that intronless mRNAs traffic through nuclear speckles.
In light of the role of nuclear speckles in mRNA metabolism (Spector and Lamond, 2011), it is
likely that this speckle trafficking is linked to mRNP assembly. It should be noted that our data
could not exclude the possibility that only a fraction of the MHC-ftz-Δi mRNA transits through
nuclear speckles.
2.4.4 The nuclear export of MHC-ftz-Δi requires UAP56 and URH49 A previous study suggested that TREX components are required for spliced mRNAs to
exit nuclear speckles (Dias et al., 2010). In light of my localization findings, I investigated the
effect of depleting the expression of several components of the TREX complex on intronless and
spliced ftz mRNAs. I took advantage of a lentiviral delivery system to transduce U2OS with
plasmids that contain small hairpin RNA (shRNA) constructs that are complementary in
sequence to both UAP56 and URH49 mRNAs. As shown in Fig. 2.4A, treating cells with these
two viruses for three days caused a decrease in the level of UAP56 and URH49 protein levels in
comparison to cells which were treated with control viruses (UAP56 levels decreased by 73 +/-
41
Figure 2.4. UAP56 and URH49 are required for the export of intronless ftz mRNA.
(A) U2OS cells were treated with lentiviruses that mediate the delivery of either plasmids that
contain shRNAs directed against UAP56 and URH49 or an empty control plasmid. Cell lysates
were collected after 72 hours, separated by SDS-PAGE and analyzed by immunoblot using
antibodies against UAP56, URH49 and α-tubulin. (B-D) U2OS cells depleted of various proteins
72 hours post-infection were microinjected with plasmids containing MHC-ftz-Δi or c-ftz-i. After
42
allowing expression for 20 minutes, cells were treated with α-amanitin and allowed to export the
mRNA for an additional 2 hours. Cells were then fixed, and the mRNA was stained by FISH (B).
Scale bar = 20 µm. (C) Quantification of the fraction of MHC-ftz-Δi and c-ftz-i mRNA in the
cytoplasm. Each bar represents the average and standard error of at least three independent
experiments, each consisting of 15-30 cells. Note that for each experiment, depleted and control
cells were assayed in parallel to control for day-to-day variations in nuclear export levels. (D-E)
U2OS cells were treated with lentiviruses that mediate the delivery of shRNAs directed against
THOC1 (D), ALY (E) or an empty plasmid (D-E). Cell lysates were collected after 96 hours,
separated by SDS-PAGE and analyzed by immunoblot with antibodies against THOC1, ALY,
GAPDH and tubulin. Asterisk represents a nonspecific band that is absent from HeLa nuclear
extract. To account for unequal cell confluency several volumes of the knock-down lysate were
loaded (0.75x, 1x or 1.25x of the control cell lysate). (F) Quantification of the fraction of poly(A)
mRNA in the cytoplasm in cells co-depleted of UAP56 and URH49, or depleted of UAP56,
URH49, THOC1 or ALY, or treated with control lentiviruses. Each bar represents the average
and standard error of at least three independent experiments, each consisting of 15-30 cells. (G)
U2OS cells co-depleted of UAP56 and URH49, or treated with control lentiviruses, were
microinjected with DNA plasmids containing MHC-ftz-Δi. Cells were fixed two hours after
injection without the prior addition of the transcription inhibitor α-amanitin and the mRNA was
stained by FISH. Quantification of the fraction of cytoplasmic mRNA is shown. Each bar
represents the average of at least three independent experiments, each consisting of 11-30 cells.
Error bars represent standard error of the mean. (H) U2OS cells co-depleted of UAP56 and
URH49, or treated with control lentiviruses, were microinjected with either in vitro transcribed
and polyadenylated MHC-ftz-Δi or c-ftz-i mRNA. Cells were fixed 1 hour after microinjection
and the mRNA was stained by FISH. Quantification of the cytoplasmic mRNA distribution is
shown. Each bar represents the average and standard deviation of two independent experiments,
each consisting of 9-25 cells.
43
7%, while URH49 decreased 87 +/- 3%, n = 3). These cells were microinjected with plasmids
containing MHC-ftz-Δi or c-ftz-i. After allowing transcription to proceed for 20 minutes, α-
amanitin was added to shut off transcription, and the newly synthesized mRNA was allowed to
export for an additional 2 h. I observed that when UAP56 and URH49 were co-depleted, both
MHC-ftz-Δi and c-ftz-i mRNAs were fully retained in the nucleus (Fig. 2.4B, quantitation
Fig. 2.4C). In contrast, depletion of either helicase alone had only slight effects on export
(Fig. 2.4C). As previously published (Kapadia et al., 2006), co-depletion of UAP56 and URH49
also caused a drastic accumulation of poly(A) mRNA in the nucleus (Fig. 2.4F). In contrast,
depletion of either UAP56 or URH49 alone did not cause a change in poly(A) mRNA
distribution (Fig. 2.4F). Depletion of the THO complex member THOC1 (also known as hHpr1
and p84) or the adaptor Aly had little to no effect on the export of either MHC-ftz-Δi or c-ftz-i
mRNA (Fig. 2.4C-E). Furthermore, depletion of either THOC1 or Aly also had no effect on bulk
mRNA export (Fig. 2.4F).
Previous experimental results demonstrated that the nuclear export of the SSCR-
dependent in vitro synthesized MHC-ftz-Δi mRNA was independent of UAP56 and URH49 in
HeLa cells (Palazzo et al., 2007). One difference between the results reported in Palazzo et al.
(2007) and my current results is the use of α-amanitin in my experiments. Nevertheless, when α-
amanitin treatment was omitted from DNA injection experiments, MHC-ftz-Δi mRNA was still
retained in the nucleus in U2OS cells depleted of UAP56 and URH49 (Fig. 2.4G). It is possible
that microinjected mRNA, which is exported more rapidly than its in vivo transcribed
counterpart, is more efficient at utilizing the low levels of UAP56/URH49 remaining in cells,
regardless of the cell line. Indeed, the export of microinjected MHC-ftz-Δi mRNA was only
partially inhibited in U2OS cells that were depleted of UAP56 and URH49 (Fig. 2.4H). In
contrast, the export of microinjected c-ftz-i mRNA was more sensitive to the depletion of these
two factors. These differences could be due to the effect of the SSCR itself, which seems to
greatly affect the export of microinjected transcripts but not of endogenously produced mRNA.
These results suggest that endogenously transcribed intronless MHC-ftz-Δi mRNA is
dependent on the TREX complex components UAP56 and URH49.
2.4.5 UAP56 and URH49 are required for MHC-ftz-Δi to exit out of nuclear speckles
44
In the course of my experiments I observed that the depletion of UAP56 and URH49
caused MHC-ftz-Δi mRNAs to accumulate into large nuclear foci and these colocalized with
several nuclear speckle markers such as SC35 and Aly (Fig. 2.5A-B). In agreement with
previous findings (Dias et al., 2010), c-ftz-i also accumulated in nuclear speckles in
UAP56/URH49-depleted cells (Fig. 2.5C). The speckle localization of MHC-ftz-Δi was much
more pronounced in the UAP56/URH49 knockdown cells than in control cells, whether this was
calculated by Pearson correlation or by computing the total amount of mRNA associated with
these structures (Fig. 2.5D-E). Indeed, in the knockdown cells practically every SC35-positive
speckle had an enrichment of MHC-ftz-Δi mRNA (Fig. 2.5D). In contrast, the depletion of these
two helicases had minor effects on the speckle-association of βG-Δi mRNA (Fig. 2.5D and F).
These experiments indicate that intronless MHC-ftz-Δi mRNA requires UAP56/URH49
for its exit from speckles. Although it is possible that inhibition of components of the TREX
complex directly promote the targeting of mRNA to speckles, we favor the model that under
normal conditions these proteins enhance the rate of egress from nuclear speckles.
2.4.6 UAP56 associates with MHC-ftz mRNA In U2OS cells, UAP56 is distributed throughout the nucleoplasm with a slight enrichment in
nuclear speckles (Fig. 2.6A). However, upon UAP56/URH49 co-depletion, the remaining
UAP56 was predominantly associated with nuclear speckles (Fig. 2.6A). Interestingly this
change in distribution was seen across almost the entire cell population (Fig. 2.6B). Since this
shRNA treatment also promoted the enrichment of MHC-ftz-Δi in speckles, I determined whether
UAP56 associates with this mRNA in vivo. Previously, it had been demonstrated that UAP56 is
recruited to mRNAs in a splicing dependent manner using an in vitro splicing reaction (Masuda
et al., 2005). I thus immunoprecipitated UAP56 from cells expressing either intronless or spliced
ftz constructs and analyzed the interacting RNAs by reverse transcription-quantitative PCR (RT-
qPCR). To control for nonspecific binding, I repeated these experiments with rat non-immune
serum. UAP56 immunoprecipitates contained the UAP56-interacting protein Aly (Fig. 2.7A),
suggesting that the isolated complexes are stable throughout the purification procedure. Indeed, I
observed a higher level of MHC-ftz-Δi mRNA enrichment in UAP56 immunoprecipitates in
comparison to c-ftz-i (Fig. 2.7B). To further confirm my results, I compared the enrichment of
MHC-ftz-Δi mRNA and the 7SL RNA in UAP56 immunoprecipitates. 7SL is a very abundant
non-coding RNA that is part of the Signal Recognition Particle (SRP) and is exported to the
cytoplasm independently of the TREX-TAP mRNA export pathway (Takeiwa et al., 2015).
45
Figure 2.5. Depletion of UAP56 and URH49 causes an enrichment of MHC-ftz-Δi mRNA
but not βG-Δi mRNA in nuclear speckles.
(A-C, F) U2OS cells were treated with lentiviruses that either mediate the delivery of shRNAs
against UAP56 and URH49 or control plasmids. Three days post-infection, cells were
microinjected with plasmid containing MHC-ftz-Δi (A-B), c-ftz-i (C) or βG-Δi (F). After allowing
the plasmid to be transcribed for 20 minutes, cells were treated with α-amanitin and incubated
for an additional 2 hours. Cells were then fixed, probed for either ftz (A-C) or βG (F) mRNA and
immunostained for the nuclear speckle markers SC35 (A, C, F) or ALY (B). Each row represents
a single field of view. Overlays of mRNA (red) and SC35 (A, C, F) or ALY (B) (green) are
shown in the right panels. Scale bar = 5 µm. (D) The percentage of SC35-positive speckles that
colocalize with MHC-ftz-Δi, βG-Δi mRNA or dextran in control cells or cells depleted of UAP56
46
and URH49. The data was analyzed and plotted as in Figure 2.2D. (E) The percentage of total
cellular and nuclear MHC-ftz-Δi mRNA that is present in nuclear speckles in control cells or cells
depleted of UAP56 and URH49 as described in Figure 2.2H. Each bar represents the average and
standard error of the mean of 10 cells.
47
Figure 2.6. Co-depletion of UAP56/URH49 causes the remaining levels of UAP56 to
associate with speckles. (A) U2OS cells were treated with lentiviruses to deliver either shRNA
directed against UAP56 and URH49 or an empty control plasmid. Cells were then microinjected
with plasmids containing MHC-ftz-Δi. After 20 minutes, cells were treated with α-amanitin and
mRNA export was allowed to proceed for 2 hours. Cells were then fixed, probed for ftz mRNA
by FISH, and UAP56 by immunofluorescence (A). Each Column is a single field of view. Scale
bar = 20 µm. (B) Quantification of the percentage of cells showing either nucleoplasmic or
speckle distribution for UAP56. Each bar represents the average and standard error of three
experiments, each consisting of at least 80 cells.
48
Figure 2.7. MHC-ftz-Δi associates with UAP56 in vivo.
(A) UAP56 was immunoprecipitated from U2OS lysates using rat anti-UAP56 antibodies pre-
bound to protein G sepharose. The immunoprecipitates were analyzed by immunoblot using
rabbit polyclonals against UAP56 and ALY. Non-immune rat serum was used in the mock
immunoprecipitation reaction. (B-C) U2OS cells were transfected with plasmids containing
MHC-ftz-Δi (B-C) or c-ftz-i (B). One day after transfection cell lysates were collected and
immunoprecipitated with rat anti-UAP56 antibodies or rat non-immune serum. RNA was
collected from fractions and converted to cDNA using ftz specific primers (B) or random
hexamers (C). The fold enrichment of mRNAs in anti-UAP56 over non-immune serum
precipitates was quantified by RT-qPCR. Each bar represents the average of five (B) and three
(C) independent experiments. Error bars represent standard error of the mean.
49
Indeed, while MHC-ftz-Δi was enriched in the UAP56 immunoprecipitates, 7SL was not
(Fig. 2.7C). From these experiments, I conclude that MHC-ftz-Δi associates with UAP56 in vivo.
2.4.7 CRM1 is required for the speckle egress and nuclear export of ftz mRNAs In addition to the canonical TREX-TAP/p15 mRNA nuclear export pathway, several
mRNAs are exported by a distinct pathway that is mediated by the karyopherin CRM1
(Culjkovic et al., 2006). Preliminary observations from the Palazzo lab indicated that the nuclear
export of ftz mRNA was sensitive to inhibitors of the Ran gradient (H. Zhang and A.F. Palazzo,
unpublished data), upon which CRM1 dependent export relies. In order to test whether CRM1
plays a role in the nuclear export of ftz and βG mRNAs, I inhibited the activity of CRM1 by
treating the cells with the CRM1 inhibitor Leptomycin B (LMB) as previously described (Nishi
et al., 1994). DNA plasmids that contain MHC-ftz-∆i, c-ftz-i and βG-i genes were microinjected
in LMB treated cells. As expected, LMB treatment inhibited the nuclear export of microinjected
NES-GFP fusion protein, which contains the NES of HIV REV protein and is known to be
dependent on CRM1 for mediating its transport to the cytoplasm (Fornerod et al., 1997).
Inhibiting the function of CRM1 also caused a block in the nuclear export of MHC-ftz-∆i and c-
ftz-i (Fig. 2.8A-B). Moreover, MHC-ftz-∆i mRNA accumulated in nuclear speckles in LMB
treated cells (Fig. 2.8C). In contrast, inhibiting the activity of CRM1 had a negligible effect on
the nuclear export of βG-i mRNA (Fig. 2.8A-B).
These results suggest that additional factors such as CRM1 cooperate with UAP56 and
URH49 towards mediating the egress of ftz mRNAs from nuclear speckles.
2.4.8 Examining the RNA sequence features that define the requirement for CRM1 in mRNA nuclear export
Next, I investigated the RNA elements within a transcript that determine whether CRM1
activity is required for a transcript to be efficiently exported to the cytoplasm. Since the nuclear
export of MHC-ftz-∆i and c-ftz-i was dependent on CRM1 but βG-i was independent of CRM1, I
compared the dependence of the fusion mRNAs: MHC-ftz-∆i-βG-∆i, MHC-ftz-∆i-βG-i and c-ftz-
i-βG-i on CRM1 (Fig. 2.9A). I reasoned that if ftz contained a specialized RNA element that
required CRM1 activity, then the fusion constructs should also be dependent on CRM1. DNA
50
Figure 2.8. CRM1 inhibition causes the accumulation of ftz mRNAs in nuclear speckles.
Plasmids containing the indicated constructs were microinjected into the nuclei of human U2OS
cells treated with 20 nM LMB. After 20 minutes, cells were treated with α-amanitin and mRNA
export was allowed to proceed for 2 hours. Cells were then fixed, probed for ftz or βG mRNA by
FISH, imaged (A) and nuclear export was quantified (B). Each bar represents the average and
standard error of three independent experiments, each consisting of 15-60 cells. (C) Plasmids
51
containing MHC-ftz-∆i were microinjected into the nuclei of human U2OS cells treated with 20
nM LMB. Cells were then fixed, probed for ftz mRNA by FISH, and SC35 by
immunofluorescence. Each row represents a single field of view. Overlays of mRNA (red) and
SC35 (green) are shown in the right panels.
52
Figure 2.9. CRM1 activity is required for the nuclear export of MHC-ftz-∆i-βG-∆i.
(A) Schematic representation of the different constructs used in this figure. (B) U2OS cells were
treated with 20 nM LMB or Ethanol either immediately, “cotreatment”, or 1 hour prior,
“pretreatment”, to microinjecting DNA plasmids containing MHC-ftz-∆i-βG-∆i, MHC-ftz-∆i-βG-
i or c-ftz-i-βG-i. Cells were fixed two hours after injection without the prior addition of the
transcription inhibitor α-amanitin and the mRNA was stained by FISH. Quantification of the
fraction of cytoplasmic mRNA is shown. Each bar represents the average of at least three
independent experiments, each consisting of 15-25 cells. Error bars represent standard error of
the mean.
53
plasmids that contain the indicated constructs were microinjected in U2OS cells either
immediately after the cells were treated with 20 nM LMB, “cotreatment” or 1 hour after LMB
treatment, “pretreatment”. I found that that inhibiting the activity of CRM1 caused the nuclear
retention of MHC-ftz-∆i-βG-∆i. Moreover, the level of nuclear retained MHC-ftz-∆i-βG-∆i
mRNA increased when CRM1 was inhibited for 1 hour prior to microinjecting the DNA
plasmid. In contrast to MHC-ftz-∆i-βG-∆i, inhibiting the activity of CRM1 had a minor effect on
the nuclear export of either MHC-ftz-∆i-βG-i or c-ftz-i-βG-i mRNAs (Fig. 2.9B).
These results suggested that the splicing of βG-i mRNA rendered its nuclear export to be
less dependent on CRM1. Moreover, the longer the period of time that CRM1 was inhibited
before the plasmids were microinjected, the more MHC-ftz-∆i-βG-∆i mRNAs were retained in
the nucleus.
2.4.9 PHAX depletion is not sufficient to cause a block in the nuclear export of ftz mRNPs
A previous study has shown that the binding of the snRNA nuclear export adaptor PHAX
to mRNAs can inhibit their nuclear export (McCloskey et al., 2012). I therefore tested if PHAX
was responsible for the CRM1 mediated nuclear retention of ftz mRNAs. I depleted the levels of
PHAX using RNAi (Fig. 2.10A). Subsequently, the cells were treated with LMB and DNA
plasmids that contain either MHC-ftz-∆i or c-ftz-i were microinjected (Fig. 2.10B). In cells with
functional CRM1, depleting the levels of PHAX was insufficient to impair the nuclear export of
either MHC-ftz-∆i or c-ftz-i mRNAs (Fig. 2.10B). Since CRM1 activity is required for the
nuclear export of PHAX, I hypothesized that inhibiting the activity of CRM1 would cause the
nuclear retention of PHAX which would subsequently bind a subset of nuclear mRNAs and
inhibit their transport to the cytoplasm. To test this hypothesis, I inhibited the activity of CRM1
by treating cells with LMB and compared the nuclear export of MHC-ftz-∆i and c-ftz-i in cells
where PHAX was either expressed or depleted. However, the CRM1-mediated nuclear retention
of MHC-ftz-∆i and c-ftz-i mRNAs was not rescued when cells were depleted of PHAX (Fig.
2.10B).
These results suggest that CRM1 mediates the nuclear retention of ftz mRNAs through a
PHAX-independent pathway.
54
Figure 2.10. Depleting the expression of PHAX is not sufficient to rescue the CRM1-
mediated nuclear retention of ftz mRNAs.
(A) U2OS cells were treated with lentiviruses that mediate the delivery of either plasmids that
contain shRNAs directed against PHAX or an empty control plasmid. Cell lysates were collected
after 72 hours, separated by SDS-PAGE and analyzed by immunoblot using antibodies against
PHAX and α-tubulin. (B) U2OS cells depleted of PHAX 72 hours post-infection were treated
with 20 nM LMB and microinjected with plasmids containing MHC-ftz-Δi or c-ftz-i. After
allowing expression for 20 minutes, cells were treated with α-amanitin and allowed to export the
mRNA for an additional 2 hours. Cells were then fixed, and the mRNA was stained by FISH.
Quantification of the fraction of cytoplasmic mRNA is shown. Each bar represents the average
and standard error of three independent experiments, each consisting of 15-30 cells.
55
2.5 Discussion
Here I investigated some of the early steps of intronless ftz mRNP assembly. My results
suggest that the trafficking of mRNA through nuclear speckles is sensitive to certain features of
the reporter transcript. Two mRNAs, ftz, and insulin, traffic through nuclear speckles, while a
third, βG, shows weak association with these compartments. My results are consistent with the
idea that both egress from speckles and mRNA nuclear export require the RNA helicases UAP56
and URH49 and the nuclear transport receptor CRM1. From this data I propose that mRNAs are
targeted to speckles by several possible routes such as splicing or by virtue of the presence of
certain sequence features in the mRNA (Fig. 2.11). Within speckles, it is likely that components
of the TREX complex are recruited to the mRNA to help assemble the mRNA into an export-
competent mRNP. My data supports the model that egress of export-competent ftz mRNPs from
speckles requires UAP56 and URH49 paralogs and CRM1 activity. Egress from speckles may be
coupled with the release of UAP56/URH49 from the mRNP (Hautbergue et al., 2008; Taniguchi
and Ohno, 2008), although it is possible that UAP56/URH49 may stay on the mRNA and
accompany it to the cytoplasm (Thomas et al., 2011). Work from others in the Palazzo lab
suggests that the egress from speckles also requires a poly(A) tail and TAP/p15 activity (Akef et
al., 2013).
The sequence features within a transcript that determine speckle association are yet to be
defined. It has also been reported that a handful of naturally intronless mRNA do not associate
with speckles (Lei et al., 2011, 2012), and this may be due to the requirements of different
mRNA export pathways, although it could also be attributable to the fact that these transcripts
have a low level of speckle association, but that it is so transient that it is not detectable under
normal circumstances. It has also been documented that certain spliced mRNAs do not traffic
through speckles yet are still exported to the cytoplasm (Smith et al., 1999), although again it is
hard to determine whether these mRNAs have a low level of speckle association that is not
normally detectable.
My results also demonstrate that while ftz requires CRM1 activity, the splicing of βG-i
renders the mRNA less dependent on CRM1 for their export to the cytoplasm. Moreover,
inhibiting the activity of CRM1 for 1 hour prior to expressing MHC-ftz-∆i-βG-∆i led to higher
levels of MHC-ftz-∆i-βG-∆i mRNA retention in the nucleus than if CRM1 was inhibited
immediately prior to plasmid microinjection. In contrast, the NES-GFP protein is completely
56
retained in the nucleus even when CRM1 was inhibited immediately prior to microinjecting the
protein. Since CRM1 has a well-established role in mediating protein nuclear export, I propose
that inhibiting the activity of CRM1 might lead to the nuclear accumulation of one or more NES-
containing proteins that inhibit the export of certain mRNAs. Subsequently, the accumulation of
these proteins in the nucleus leads to ftz mRNA nuclear retention through a mechanism that is yet
to be identified.
Previous studies in S. cerevisiae indicate that certain mRNPs are rendered export-
competent in the vicinity of the nuclear pore (Cabal et al., 2006; Fasken et al., 2008; Pascual-
García et al., 2008). This process is initiated by the anchoring of the transcribed gene to the pore
in a process known as gene-gating (Brickner, 2009; Casolari et al., 2004; Hocine et al., 2010).
Interestingly, this phenomenon has never been observed in mammalian cells (Hocine et al.,
2010), suggesting that the majority of mRNP formation occurs elsewhere in the nucleoplasm in
these organisms. My data lends support to the notion that in mammalian cells, the formation of
certain mRNPs occurs within speckles. Importantly, this may not be universally true for all
mRNPs. Of course, to understand this process would require further work to identify all the
critical steps in mRNP assembly and the RNA elements that modulate this process.
57
Figure 2.11. Model linking mRNP formation with the trafficking of mRNAs through
nuclear speckles.
The association of mRNAs with nuclear speckles is promoted by both splicing, and sequence
features within the mRNA. Within the speckle, mRNP maturation occurs through the activity of
the RNA helicases UAP56/URH49 and additional factors such as CRM1. This process primes
the mRNP for export to the cytoplasm.
58
Chapter 3
Splicing promotes the nuclear export of β-Globin mRNA by
overcoming nuclear retention elements
Parts of this chapter were published in:
Akef, A., Lee, E.S., and Palazzo, A.F. (2015). Splicing promotes the nuclear export of β-globin
mRNA by overcoming nuclear retention elements. RNA 21, 1908–1920.
I generated all the results presented in this chapter.
59
3.1 Summary
Most current models of mRNA nuclear export in vertebrate cells assume that an mRNA
must have specialized signals in order to be exported from the nucleus to the cytoplasm. Under
such a scenario, mRNAs that lack these specialized signals would be shunted into a default
pathway where they are retained in the nucleus and eventually degraded. These ideas are based
on the use of selective model mRNA reporters. For example, it has been shown that splicing
promotes the nuclear export of certain model mRNAs, such as human βG, and that in the
absence of splicing, the cDNA-derived mRNA is retained in the nucleus and degraded. Here I
provide evidence that βG mRNA contains an element that actively retains it in the nucleus where
it is degraded. Interestingly, this nuclear retention activity can be overcome by increasing the
length of the mRNA or by including an intron. These results suggest that, contrary to many
current models, the default pathway for most intronless RNAs is to be exported from the nucleus,
unless the RNA contains elements that actively promote its nuclear retention.
60
3.2 Introduction
Eukaryotic cells contain two major compartments, the nucleoplasm where mRNA is
synthesized and processed, and the cytoplasm where this mRNA is translated into proteins. It is
currently believed that in vertebrate cells, mRNAs contain specialized cis-acting elements that
recruit nuclear export factors and permit their efficient export to the cytoplasm (Palazzo and
Akef, 2012). This contrasts to the situation in S. cerevisiae where nuclear export factors are
recruited to the transcript during transcription, regardless of their sequence or the presence of any
specialized cis-elements (Palazzo and Akef, 2012).
Much of the current thinking on this subject has been derived from studies of the main
export complex in eukaryotic cells, the TREX complex. In vertebrate cells, it is believed that the
TREX complex is primarily loaded onto the RNA in both a splicing and cap dependent manner
(Masuda et al., 2005; Strässer and Hurt, 2001; Zhou et al., 2000). TREX then recruits the
heterodimeric nuclear transport receptor TAP/p15 (also known as NXF1/NXT1), which ferries
the mRNA across the nuclear pore (Katahira et al., 1999; Stutz et al., 2000). Thus introns act as
de facto export-promoting cis-elements and this has been validated by the observation that
certain model mRNAs such as ftz and βG are only exported when they contain introns and thus
spliced (Luo and Reed, 1999; Valencia et al., 2008). In contrast, yeast TREX components are
loaded onto mRNAs co-transcriptionally by the action of RNA Pol II (Chávez et al., 2000;
Strässer et al., 2002; Jimeno et al., 2002; Zenklusen et al., 2002). The lack of a splicing
requirement in yeast is not surprising since the vast majority of their protein-coding genes are
intronless.
Within this context, it was assumed that human protein-coding genes that are naturally
intronless, would need some substitute cis-element to recruit the TREX complex in the absence
of splicing. This led to the identification of putative CAR-Es by the analysis of a handful of
human intronless genes (Lei et al., 2011, 2012). This idea was validated by fusing 16 copies of
the putative CAR-E to the 5’end of the intronless βG mRNA (βG-∆i) (Lei et al., 2012).
Inexplicably, the mutation of putative CAR-Es from intronless mRNAs did not affect their
nuclear export (Lei et al., 2012), suggesting that the export of naturally intronless mRNAs may
not be dependent on these elements.
61
In parallel with these studies, several groups have mapped and identified cis-elements
within lncRNAs that promote their nuclear retention. Interestingly, when these elements are
eliminated from these lncRNAs, even the intronless lncRNAs are efficiently exported to the
cytoplasm (Miyagawa et al., 2012; Zhang et al., 2014). This was true of many different small
fragments of lncRNAs (Miyagawa et al., 2012; Zhang et al., 2014). These observations would
indicate that either export-promoting elements are quite plentiful, or that in the absence of any
cis-elements, the default pathway for any given RNA is to be exported to the cytoplasm.
Recently, an RNA element present in certain ftz reporter plasmids has been found to
promote nuclear mRNA retention (Lee et al., 2015). This element, which is identical to the
consensus 5’ splice site motif, was present in the multi-cloning region of the plasmid,
downstream of the ftz gene and upstream of the 3’ cleavage site and was thus incorporated into
the 3’ UTR of the mature ftz mRNA. When this motif was eliminated, the resulting intronless ftz
mRNA was efficiently exported. These observations indicated that either the ftz mRNA has an
additional nuclear export- promoting element, or that when nuclear retention elements are
eliminated, all mRNAs become substrates for nuclear export. These two possibilities were
supported by the finding that UAP56, a central component of the TREX complex, is efficiently
loaded onto ftz without the requirement for splicing (Taniguchi and Ohno, 2008; Akef et al.,
2013).
Here I analyzed the export requirements of ftz and βG mRNAs. My results are consistent
with the idea that the 3’ end of the βG gene inhibits nuclear export. The data suggest that the
activity of this nuclear retention element can be over-ridden by extending the RNA at the 5’end
or by the inclusion of introns in the transcript. Importantly, my findings suggest that, in the
absence of any cis-element, an mRNA is exported to the cytoplasm. This is contrary to most
fibroblasts (NIH 3T3) were maintained in high glucose DMEM containing 10% calf serum
(Wisent) and antibiotics (Sigma). The following drugs were used: puromuycin (Sigma) was used
at 200 µM, homoharringtonine (HHT) (Tocris Bioscience) was used at 5µM and α-amanitin
(Sigma) was used at 1µg/mL.
63
Table 3.1. A description of the constructs used in Chapter 3.
Construct Description MHC-ftz-∆i See Palazzo et al. (2007) c-ftz-∆i See Palazzo et al. (2007) MHC-βG-Δi See Akef et al. (2013) βG-∆i See Valencia et al. (2008) βG-i See Akef et al. (2013) c-ftz-i See Palazzo et al. (2007) F1-βG-∆i Sequence -20 to 99 of c-ftz-∆i inserted downstream of nucleotide 3 of βG-∆i (A
of ATG as position 1) F2-βG-∆i Sequence 80 to 199 of c-ftz-∆i inserted downstream of nucleotide 3 of βG-∆i F3-βG-∆i Sequence 180 to 299 of c-ftz-∆i inserted downstream of nucleotide 3 of βG-∆i F4-βG-∆i Sequence 280 to 399 of c-ftz-∆i inserted downstream of nucleotide 3 of βG-∆i MHC-ftz-∆i Del 1 Sequence 67 to 116 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 2 Sequence 117 to 166 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 3 Sequence 167 to 216 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 4 Sequence 217 to 266 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 5 Sequence 267 to 316 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 6 Sequence 317 to 366 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 7 Sequence 367 to 416 deleted from MHC-ftz-∆i MHC-ftz-∆i Del 8 Sequence 417 to 468 deleted from MHC-ftz-∆i MHC-ftz-∆i-βG-∆i Sequence -20 to 465 of MHC-ftz-∆i inserted at the HindIII site upstream of βG-∆i RC-MHC-ftz-∆i-βG-∆i
The reverse complement of sequence -20 to 465 of MHC-ftz-∆i inserted at the HindIII site upstream of βG-∆i
βG-∆i-βG-∆i Sequence 1 to 441 of βG-∆i inserted at the HindIII site upstream of βG-∆i B1-βG-∆i Sequence -44 to 285 of βG-∆i inserted at the KpnI site upstream of βG-∆i B2-βG-∆i Sequence 61 to 390 of βG-∆i inserted at the KpnI site upstream of βG-∆i B3-βG-∆i Sequence 166 to 495 of βG-∆i inserted at the KpnI site upstream of βG-∆i MHC-βG-Δi Del 1 Sequence -44 to 2 deleted from MHC-βG-Δi MHC-βG-Δi Del 2 Sequence 64 to 120 deleted from MHC-βG-Δi MHC-βG-Δi Del 3 Sequence 121 to 230 deleted from MHC-βG-Δi MHC-βG-Δi Del 4 Sequence 231 to 340 deleted from MHC-βG-Δi MHC-βG-Δi Del 5 Sequence 341 to 450 deleted from MHC-βG-Δi MHC-βG-Δi Del 6 Sequence 429 to 555 deleted from MHC-βG-Δi βG-∆i-ftz intron EJ 1 Sequence 302 to 448 of c-ftz-i inserted downstream of nucleotide 92 of βG-∆i βG-∆i-ftz intron EJ 2 Sequence 302 to 448 of c-ftz-i inserted downstream of nucleotide 315 of βG-∆i B1-∆i Sequence 286 to 500 deleted from βG-Δi
64
Transfection, Microinjection, FISH, and Immunostaining. Cells were plated on 22x22 mm
acid-washed cover slips (VWR) in 35 mm mammalian tissue culture dishes (Thermo Scientific)
for 24 hours prior to injection or transfection. Cells were transfected using GenJet in vitro DNA
Transfection Reagent for U2OS (SignaGen Laboratories) according to the manufacturer’s
protocol. DNA microinjections were performed as previously described (Akef et al., 2013).
Fluorescence in situ hybridization (FISH) mRNA staining was performed as previously
described (Akef et al., 2013). The probe oligonucleotide sequences used included anti-ftz