Functions of hUpf3a and hUpf3b in nonsense-mediated
mRNA decay and translation
JOACHIM B. KUNZ,1,2GABRIELE NEU-YILIK,1,2MATTHIAS W. HENTZE,2,3ANDREAS E. KULOZIK,1,2
and NIELS H. GEHRING1,2
1Department for Pediatric Oncology, Hematology and Immunology, University of Heidelberg, Heidelberg, Germany
2Molecular Medicine Partnership Unit, University of Heidelberg and European Molecular Biology Laboratory, Heidelberg, Germany
3European Molecular Biology Laboratory, Gene Expression Unit, Heidelberg, Germany
The exon–junction complex (EJC) components hUpf3a and hUpf3b serve a dual function: They promote nonsense-mediated
mRNA decay (NMD), and they also regulate translation efficiency. Whether these two functions are interdependent or
independent of each other is unknown. We characterized the function of the hUpf3 proteins in a lN/boxB-based tethering
system. Despite the high degree of sequence similarity between hUpf3b and hUpf3a, hUpf3a is much less active than hUpf3b to
induce NMD and to stimulate translation. We show that induction of NMD by hUpf3 proteins requires interaction with Y14,
Magoh, BTZ, and eIF4AIII. The protein region that mediates this interaction and discriminates between hUpf3a and hUpf3b in
NMD function is located in the C-terminal domain and fully contained within a small sequence that is highly conserved in Upf3b
but not Upf3a proteins. Stimulation of translation is independent of this interaction and is determined by other regions of the
hUpf3 protein, indicating the presence of different downstream pathways of hUpf3 proteins either in NMD or in translation.
Keywords: exon–junction complex; nonsense-mediated mRNA decay; Upf3; tethering
Nonsense-mediated mRNA decay (NMD) is a gene expression
surveillance mechanism reported in all eukaryotic organ-
isms analyzed. In humans, the activity of NMD modulates
the phenotype of acquired and hereditary disorders (Hall
and Thein 1994; Wilschanski et al. 2003; Holbrook et al.
2004) by degrading transcripts with premature termination
codons arising from errors in transcription or splicing,
from mutated genes, or from genes that are physiologically
regulated by NMD (He et al. 2003; Mendell et al. 2004).
Thus, NMD is of general biological and medical impor-
tance. Mechanistically, NMD depends on splicing, trans-
lation, and a number of trans-acting protein factors, such as
the Upf and Smg proteins (Gonzalez et al. 2001; Schell et al.
2002; Maquat 2004).
According to current models of mammalian NMD, the
exon–junction complex (EJC) is deposited on the mRNA
during splicing at a position 20–24 nt upstream of exon–
exon boundaries (Le Hir et al. 2001). Components of the
EJC, namely Y14 (RBM8A), Magoh, RNPS1, eIF4AIII
(DDX48, Nuk34), and Barentsz (BTZ, MLN51, CASC3),
are involved in the definition of the exon–intron junctions
during NMD (Lykke-Andersen et al. 2001; Fribourg et al.
2003; Gehring et al. 2003; Chan et al. 2004; Degot et al.
2004; Ferraiuolo et al. 2004; Palacios et al. 2004; Shibuya
et al. 2004). Recently, we have shown that these EJC
components define two functionally distinguishable EJC
subgroups: RNPS1-type EJCs require normal levels of
UPF2 to trigger efficient NMD, whereas Y14-Magoh-
eIF4AIII-BTZ-type EJCs even activate NMD in UPF2-
depleted cells and thus tolerate very low UPF2 levels (Gehring
et al. 2005). In the cytoplasm, the translating ribosome
removes EJCs and associated NMD factors within the open
reading frame, which validates the mRNA for the pool of
stable translated mRNAs. If, however, the ribosome termi-
nates at a stop codon upstream of an EJC, it is thought to
recruit hUpf1, possibly via an interaction of hUpf1 with the
peptide release factors. hUpf1 is required for the degrada-
tion of all known NMD substrates, which is thought to
involve both decapping and deadenylation (Lejeune et al.
EJCs play a role not only in NMD, but in several other post-
transcriptional steps of gene expression (Tange et al. 2004).
Reprint requests to: Niels H. Gehring, Children’s Hospital, Im
Neuenheimer Feld 156, 69120 Heidelberg, Germany; e-mail: niels.gehring@
med.uni-heidelberg.de; fax: +49-6221-56-4580.
Article published online ahead of print. Article and publication date are
RNA (2006), 12:1015–1022. Published by Cold Spring Harbor Laboratory Press. Copyright ? 2006 RNA Society.
The EJC’s role in nucleocytoplasmic transport is indicated by
the interaction of REF/Aly with export factors such as TAP/
p15 (Stutz et al. 2000; Zhou et al. 2000). The EJC has also been
shown to enhance translation when deposited within an ORF
by increasing the proportion of mRNAs associated with
polysomes (Lu and Cullen 2003; Nott et al. 2003, 2004).
In vertebrates, two homologs of the yeast Upf3 protein
exist, termed hUpf3a and hUpf3b (Lykke-Andersen et al.
2000) or hUpf3 and hUpf3X, respectively (Serin et al. 2001).
Two splice variants of hUpf3a, hUpf3aL and hUpf3aS, have
been described that either retain or skip exon 4. Each hUpf3
mRNA isoform and splice variant is expressed in all adult
tissue types examined (Serin et al. 2001). hUpf3a and hUpf3b
are predominantly nuclear and shuttle between nucleus and
cytoplasm (Lykke-Andersen et al. 2000; Serin et al. 2001).
Both hUpf3 proteins preferentially associate with spliced
rather than unspliced mRNA, indicating that the process of
splicing and EJC recruitment facilitates their interaction with
mRNAs (Lykke-Andersen et al. 2000). hUpf3aL and hUpf3b
can bind hUpf2 (Lykke-Andersen et al. 2000; Serin et al.
2001), an interaction that is conserved from yeast to humans.
The cocrystal structure of the interacting domains of hUpf2
and hUpf3b revealed that the N-terminal RNP domain
(ribonucleoprotein-type RNA-binding domain, also known
as RNA recognition motif, RRM) of hUpf3b interacts
hydrophilically with the C-terminal MIF4G (middle portion
of eIF4G) domain of hUpf2 (Kadlec et al. 2004). Although
the interaction between Upf3 and Upf2 is evolutionarily
widely conserved, it is not required for the NMD activity of
tethered hUpf3b, suggesting either that tethering of hUpf3b
to the RNA bypasses the hUpf2-requiring step of NMD or
that the interaction with hUpf2 is a redundant step within
the NMD process (Gehring et al. 2003, 2005). By contrast,
the C-terminal domain of hUpf3b is essential for NMD and
is needed to assemble a complex containing the Y14/Magoh
heterodimer. This domain is also present in hUpf3a, explain-
ing its coprecipitation with Y14 (Kim et al. 2001; Lykke-
Andersen et al. 2001). Functionally, both hUpf3a and
hUpf3b destabilize a reporter mRNA if tethered downstream
of a physiological termination codon (Lykke-Andersen et al.
2000). This mRNA degradation reflects characteristic fea-
tures of NMD such as dependency on translation and hUpf1
(Lykke-Andersen et al. 2000; Gehring et al. 2003). In
contrast to hUpf3aL, hUpf3aS lacks a part of the N-
terminal RNP fold and does not coprecipitate hUpf2
(Ohnishi et al. 2003). In gel filtration experiments, hUpf3aL
cofractionates with hUpf2 and elutes at a higher molecular
weight than hUpf3aS, which associates with Smg7 but not
hUpf2 (Ohnishi et al. 2003; Schell et al. 2003). These
differences in the protein–protein interactions suggest
different functions of the various hUpf3 isoforms. There-
fore, we have analyzed the different activities of hUpf3a and
hUpf3b in NMD and in translation stimulation. Immuno-
precipitated complexes of hUpf3a and hUpf3b contain
a similar set of proteins, including Y14, Magoh, eIF4AIII,
and BTZ. However, hUpf3a is much less efficient than
hUpf3b in triggering NMD and in stimulating translation.
NMD and translation stimulation by hUpf3 proteins re-
quire different protein regions and interacting factors,
representing separable and likely independent functions
of hUpf3a and hUpf3b.
RESULTS AND DISCUSSION
The NMD activities of hUpf3a and hUpf3b
are markedly different
hUpf3aL has previously been shown to destabilize a reporter
mRNA if tethered downstream of the physiological termi-
nation codon employing a MS2-based tethering system
(Lykke-Andersen et al. 2000). Similarly, tethering of
hUpf3a with the lN/boxB system decreased the abundance
of the globin 5boxB reporter mRNA (Fig. 1B, lanes 3,4).
We compared the effects of tethering hUpf3b with hUpf3aL
and hUpf3aS. Other hUpf3a variants, which we identified
from HeLa cDNA (data not shown) but cannot safely be
assigned to a protein product by Western blotting, were
not analyzed. In keeping with previous studies (Lykke-
Andersen et al. 2000; Gehring et al. 2003), tethered hUpf3b
decreased the amount of the reporter mRNA to less than
one-third compared to the unfused lN-peptide (Fig. 1B,
lanes 7,8). In contrast, tethered hUpf3aL displayed weak
NMD activity and the abundance of reporter mRNA was
reduced to only 71% (Fig. 1B, lane 3). Neither deletion of
the hUpf2-interacting domain between amino acids 66
and 140 (Fig. 1B, lane 5) nor of exon 4 (hUpf3aS; Fig. 1B,
lane 4) changed the weak activity of hUpf3a. If, however,
a C-terminal stretch of 14 amino acids (residues 434–447,
construct hUpf3aDY14) was deleted, the abundance of
reporter RNA remaining with tethered hUpf3aDY14 was
clearly increased compared to tethered full-length hUpf3aL
(Fig. 1B, lanes 3,6): In each of nine independent experi-
ments the ratio of reporter RNA was at least 1.45 times
higher with hUpf3aDY14 than with hUpf3aL (mean, 1.70-
fold; SD, 0.38), indicating that the small reduction in
mRNA abundance with hUpf3aL is a specific albeit weak
effect of hUpf3aL. This stretch of amino acids (434–447) is
highly homologous to residues 421–434 of hUpf3b, which
are required to form a complex containing Y14 (Gehring
et al. 2003). We conclude that hUpf3a is only marginally
NMD active in tethered function analysis, despite its high
degree of homology with hUpf3b. Nonetheless, the same
protein module is required for the partial NMD activity of
hUpf3a and the strong NMD activity of hUpf3b.
The different NMD activities of hUpf3a and hUpf3b
are determined by their C termini
To find the basis for the observed differences in NMD
activity between hUpf3a and hUpf3b, we compared
Kunz et al.
RNA, Vol. 12, No. 6
C-terminal sequences of Upf3 proteins from several verte-
brates (Fig. 2). In all Upf3b proteins deposited in databases,
a C-terminal sequence required for complex formation
with Y14 is highly conserved (Gehring et al. 2003). The
corresponding sequence is more variable and in some
species, such as rat and mouse, even absent in Upf3a
proteins. Single amino acid changes within this region of
hUpf3b impair its NMD activity (Gehring et al. 2003).
Notably, the homologous position of arginine R419, which
has been shown to be important for hUpf3b activity, is
replaced by an alanine residue in hUpf3a (A432). We
therefore reasoned that substituting A432 in hUpf3a for
an arginine, i.e., changing hUpf3a to the Upf3b consensus,
might improve the NMD activity of hUpf3a. Indeed, the
hUpf3aL 432R point mutant is more active than the
hUpf3a wild type and displays a NMD activity similar to
that of tethered hUpf3b (Fig. 3B, lanes 2,3,5). When we
exchanged the C-terminal 150 amino acids of hUpf3aL
for the corresponding sequence of hUpf3b, we obtained
a hybrid protein that was as active as hUpf3b (Fig. 3B, lanes
4,5). These data demonstrate that the NMD activity of
hUpf3 proteins is determined by the C-terminal sequence
in general and is largely affected by the single amino acid
position R419 in hUpf3b and A432 in hUpf3a, respectively.
hUpf3a-dependent NMD requires interaction with
Y14/Magoh but not with hUpf2
In order to test the functional relevance of known protein–
protein interactions of hUpf3 proteins, we analyzed the
ability of mutant Upf3a proteins to coprecipitate hUpf2
and Y14/Magoh (Fig. 4A), proteins that have previously
been shown to interact with hUpf3 proteins (Lykke-
Andersen et al. 2000, 2001; Kim et al. 2001; Serin et al.
2001; Gehring et al. 2003). Full-length hUpf3aL and
hUpf3b precipitated both hUpf2 and Y14/Magoh. By
contrast, hUpf3aS and the mutant hUpf3aDhUpf2 (amino
acids 66–140 deleted) did not precipitate hUpf2, probably
because they lack the N-terminal or the C-terminal part of
the Upf2-binding RNP-fold, respectively (Kadlec et al.
2004). hUpf3aL, hUpf3aS, and the hUpf3aDhUpf2 mutant
all display a similar partial NMD activity when compared
to hUpf3aDY14 (Fig. 1, lanes 3–5,6), which demonstrates
that hUpf3a’s ability to bind hUpf2 does not significantly
modulate its NMD function. In contrast, the ability to form
a complex with Y14/Magoh correlates with NMD activity:
All hUpf3 proteins containing the C-terminal Y14/Magoh
interaction sequence were at least partially active in
triggering NMD. When Y14/Magoh binding is abrogated,
NMD activity is completely lost. However, this does not
explain the difference in activity between hUpf3aL, hUp-
f3aL 432R, hUpf3a+3bCterm, and hUpf3b. Although in our
immunoprecipitation experiments hUpf3a proteins had
a tendency to coprecipitate less Y14, Magoh, and eIF4AIII
compared to hUpf3b (Fig. 4A, lanes 2,3,7; Fig. 5, lanes 2–4),
this effect was only gradual and could not reflect the clear
differences in NMD activity. We conclude that the in-
teraction of Upf3 proteins with Y14/Magoh is necessary but
not sufficient for NMD function. This has previously been
shown for tethered hUpf3b using siRNA-mediated de-
pletion of Y14 (Gehring et al. 2003). However, the marginal
reduction of reporter mRNA did not permit us to apply the
same strategy for tethered hUpf3a.
We next tested several hUpf3 mutants for their ability
to coprecipitate hUpf1, hSmg5, and RNPS1 (Fig. 4B).
Two of the proteins used are NMD-active (hUpf3b,
hUpf3bDhUpf2 with amino acids 49–143 deleted), two
FIGURE 1. hUpf3a elicits NMD less efficiently than hUpf3b. (A)
Schematic representation of hUpf3a mutants used in tethered func-
tion analysis: The Y14 interacting site homologous to hUpf3b is
shown in black, and the other conserved region containing the hUpf2-
interacting site and the RRM-like domain is depicted in gray. The
alternatively spliced exon 4 is in light gray (e4). The D1–24 construct
corresponds to an N-terminally truncated variant of hUpf3a reported
by Serin at al. (2001). Full-length hUpf3a and the splice variant
lacking exon 4 are designated hUpf3aL and hUpf3aS, respectively.
hUpf3aDhUpf2 lacks part of the proposed hUpf2 interaction site
(amino acids 66–140); hUpf3aDY14 lacks the proposed Y14-interac-
tion site (amino acids 434–447). (B) HeLa cells were transfected with
the globin 5boxB reporter, a control for transfection efficiency and the
indicated lN-tagged hUpf3-construct. As a positive control, hUpf3b
constructs were used (right panel). The amount of reporter mRNA
was quantified relative to the unfused lN-peptide. Data are means
from five independent experiments. To control for protein expression
levels, lysates were analyzed by immunoblotting with an anti-lN-
Different functions of hUpf3a and hUpf3b
are partially active (hUpf3aL and -S), and one is in-
active (hUpf3bDY14 with amino acids 421–414 deleted);
some mutants cannot interact with hUpf2 (hUpf3aS,
hUpf3bDhUpf2) or with Y14/Magoh (hUpf3bDY14). All
of the proteins tested coprecipitate hUpf1, hSmg5, and
RNPS1, indicating that the interactions of hUpf3 with any
of these proteins does not suffice to elicit NMD. Notably,
we observed that hUpf3aS precipitated the most hUpf1
when compared to the other hUpf3 protein variants. These
data demonstrate that efficient immunoprecipitation of
hUpf1 by hUpf3 does not depend on hUpf2 and that
binding of the NMD factors hUpf1, hSmg5, and RNPS1
does not discriminate between NMD-active and NMD-
inactive hUpf3 variants.
NMD-active hUpf3 proteins form a complex
containing Y14/Magoh, eIF4AIII, and BTZ
PYM/p29, eIF4AIII, and BTZ are proteins that have been
reported to interact with Y14/Magoh (Bono et al. 2004;
Ferraiuolo et al. 2004; Palacios et al. 2004; Shibuya et al.
2004). We analyzed whether they are part of the NMD
activating complex containing hUpf3, Y14, and Magoh. All
hUpf3 proteins that show at least partial NMD activity and
precipitate Y14/Magoh also precipitate eIF4AIII and BTZ
(Fig. 5, lanes 2–4,6,7). In contrast, the NMD-inactive
C-terminal deletion mutant hUpf3bDY14 fails to precipitate
Y14/Magoh, eIF4AIII, or BTZ (Fig. 5, lane 5). Thus, only
NMD-active hUpf3 proteins engage in a large complex
containing Y14/Magoh, eIF4AIII, and BTZ. All deletions of
hUpf3 that disrupt complex formation with Y14/Magoh/
eIF4AIII/BTZ also cause a loss of NMD function of hUpf3a
and hUpf3b (Gehring et al. 2003; Fig. 1, lanes 6,9). The
difference in NMD activity between hUpf3a and hUpf3b
might be due to a gradual difference in affinity to the
Y14/Magoh/eIF4AIII/BTZ complex that cannot be detected
in immunoprecipitation experiments. Alternatively, hUpf3a
might fulfill functions unrelated to NMD or represent a
regulatory component of the hUpf3/Magoh/eIF4AIII/BTZ
Interestingly, PYM/p29, which has previously been
shown to interact with the Y14/Magoh heterodimer, is
not precipitated by any hUpf3 protein or by eIF4AIII, but
coprecipitates with Y14 (Fig. 5, cf. lanes 2–7,9 and lane 8)
FIGURE 2. The C terminus of Upf3b but not Upf3a is conserved. C-terminal sequences for Upf3 proteins from rat (GenBank accession nos.
Upf3a NM_001012159, Upf3b XP_233312), mouse (Upf3a XP_356061, Upf3b XM_110787), man (Upf3a NM_080687, Upf3b AY013251), and
zebrafish (Upf3a XP_694916, Upf3b NP_957248) were aligned (ClustalW [http://www.ebi.ac.uk], Boxshade [http://www.ch.embnet.org]).
Sequences were chosen to contain the Y14 interaction site (boxed) or its corresponding sequences in the other species. The arrow indicates
position 432 in hUpf3a and position 419 in hUpf3b, respectively. The zebrafish Upf3a sequence is part of an unusually short Upf3 protein of
unknown biological function.
FIGURE 3. The differences in NMD activity of hUpf3a and hUpf3b
reside in the C terminus. (A) Schematic representation of hUpf3a
constructs used for tethered function analysis. Wild-type hUpf3aL
(432A) was mutated to hUpf3aL 432R to match the Upf3b consensus.
In construct hUpf3a+3b Cterm the entire C terminus (from position
332 in hUpf3a) was replaced by that of hUpf3b (333–483). (B) HeLa
cells were transfected with the b-globin 5boxB reporter, a control for
transfection efficiency, and the indicated lN-tagged hUpf3-construct.
As a positive control, hUpf3b was transfected (lane 5). The amount of
reporter mRNA was quantified relative to the unfused lN. Data are
means from five independent experiments. To control for expression
levels, protein lysates were analyzed by immunoblotting with an anti-
lN-antibody for transfected proteins.
Kunz et al.
RNA, Vol. 12, No. 6
as previously demonstrated (Bono et al. 2004). Hence, it is
unlikely that PYM/p29 plays a role in NMD activated by
proteins of the Upf3 family.
Different activities of hUpf3a and hUpf3b in
The EJC is a key regulator of post-transcriptional mRNA
metabolism (Tange et al. 2004). As such, it plays a central
function in NMD, but also affects the translational effi-
ciency of spliced transcripts. The EJC, together with the
Upf proteins, stimulates translation and enhances the
polysome association of the mRNA (Nott et al. 2003,
2004; Wiegand et al. 2003). Since hUpf3 proteins need to
interact with Y14/Magoh/eIF4AIII/BTZ, but not with
hUpf2, in order to trigger NMD, we tested whether the
translation stimulation shows equivalent requirements. We
coexpressed lN-hUpf3 proteins with an intronless reporter
mRNA, coding for Renilla luciferase containing 4 boxB
sites within the open reading frame (Fig. 6A). As a negative
control for possible nonspecific effects of coexpressed
NMD factors, we used a Renilla luciferase construct with-
out boxB sites; transfection efficiency was controlled by
cotransfecting a firefly luciferase reporter without boxB
sequences. Firefly and Renilla luciferase activities were
normalized against their respective mRNA levels. This
analysis showed that both hUpf3aL and hUpf3aS enhance
translation of the Renilla reporter mRNA by a factor of
z1.5-fold, whereas hUpf3b stimulates translation approx-
imately threefold without significantly affecting mRNA
levels (Fig. 6B). Surprisingly, hUpf3bDY14, which does
not recruit the Y14/Magoh/eIF4AIII/BTZ complex, and
even hUpf3bD371–470, which lacks all of the C terminus,
also stimulate translation with an efficiency similar to that
of full-length hUpf3b. Thus, in contrast to its NMD
activity, the translation stimulation activity of hUpf3b
is independent of complex formation with Y14/Magoh/
We tested a set of hUpf3b mutants (Fig. 5B) for their
activity in translation stimulation, in order to define the
region of the hUpf3b protein that mediates translation
FIGURE 5. The NMD-active hUpf3 complex contains Y14, Magoh,
eIF4AIII, and BTZ, but not p29/PYM. (A) FLAG fusions were coex-
pressed with V5-BTZ (top part) or V5-eIF4AIII (bottom part) and
precipitated using anti-FLAG (M2) agarose beads. Unfused FLAG
served as a negative control (lane 1). Coprecipitated proteins were
detected by Western blotting using antibodies directed against V5,
PYM/p29, Y14, and Magoh. To control for equal loading, the
membranes were reprobed with a FLAG-specific antibody (M2). (B)
Schematic representation of hUpf3b constructs used in Figures 5 and 6.
FIGURE 4. The NMD activity of hUpf3 proteins is determined by
their interaction with Y14 and Magoh, but not with hUpf2, hUpf1,
hSmg5, or RNPS1. (A) FLAG-hUpf3 proteins were expressed in HeLa
cells and precipitated using anti-FLAG (M2) agarose beads; copreci-
pitated endogenous proteins were detected by Western blotting with
antibodies directed against Y14, Magoh, and hUpf2. Unfused FLAG
served as negative control (lane 1). To control for equal loading, the
membrane was reprobed with a FLAG-specific antibody (M2). (B)
Ten micrograms (10 mg) of FLAG-hUpf3 fusion constructs were
cotransfected with V5-hUpf1 (2 mg), V5-Smg5 (10 mg), or V5-RNPS1
(10 mg) and FLAG-proteins precipitated using anti-FLAG (M2)
agarose beads. Coprecipitated V5 fusions were detected by Western
blotting using an anti-V5 antibody. To control for equal loading, the
membranes were reprobed with a FLAG-specific antibody (M2).
Different functions of hUpf3a and hUpf3b
(Fig. 6B). Significantly decreased translation activity was
only observed for hUpf3bD49–279, which lacks the RNP
domain including the hUpf2 interaction site. This mutant
displayed the same weak activity as hUpf3a. Since the
deleted region of hUpf3bD49–279 spans more than 200
amino acid residues, this manipulation could cause the
mutant protein to adopt a misfolded, nonfunctional struc-
ture. However, hUpf3bD49–279 coprecipitates Y14 from
HeLa extracts (data not shown), indi-
cating that it is still able to interact with
the EJC. hUpf2 binding does not seem
to be required for hUpf3b function in
translation because hUpf3bD154–232,
which displays very weak hUpf2 bind-
ing (data not shown), does not signifi-
cantly differ from hUpf3b wild type in
hUpf3bD117–279, and hUpf3aL432R
have an intermediate activity between
hUpf3a and hUpf3b, indicating that
changes in these protein regions have
a minor effect on the translational
activity of the resulting mutant protein.
Another function of EJCs within the
ORF, especially if located close to the 59
end, is enhancing mRNA synthesis by
increasing 39 end processing efficiency
(Nott et al. 2003). We designed a re-
porter construct with the tethering sites
close to the 39 end of the ORF to be able
to examine translational effects without
interfering with RNA synthesis effects.
Consequently, the effect of tethered
hUpf3 proteins on the abundance of
the reporter RNA is small compared to
the effect on reporter protein levels.
The data presented here document
that hUpf3a in its splice variants is not
a major NMD activating component of
theEJC. The differences
hUpf3a and hUpf3b in NMD function
are determined by their respective
C-terminal sequences that induce as-
sembly of a protein complex containing
necessary but not sufficient for NMD.
Interestingly, the role of Upf3 proteins
in stimulating translation does require
neither the assembly of a complex con-
taining Y14/Magoh/eIF4AIII/BTZ nor
binding of hUpf2. However, hUpf3a
and hUpf3b differ in their translation
stimulation activity as they do in NMD
activity, indicating that the function of
hUpf3a is fundamentally different from
the activity of hUpf3b and not redundant as was previously
We conclude that enhancing translation and activating
NMD by Upf3 proteins do not involve identical protein
complexes but require the recruitment of different down-
stream factors. Thus, different aspects of post-transcrip-
tional mRNA metabolism represent separable functions of
the EJC in general and of Upf3 proteins in particular.
FIGURE 6. Activation of translation by hUpf3 proteins does not depend on Y14/Magoh
interaction. (A) Schematic representation of reporter and control constructs (for details, see
Materials and Methods). (B) Relative translational yields of Renilla luciferase. HeLa cells were
transfected with the Renilla 4boxB reporter or the Renilla reporter without boxB sites, the
firefly luciferase control, and the indicated lN-tagged hUpf3 construct (1 mg). Seventy-two
hours after transfection, luciferase activities were measured, and RNA levels of reporter and
control were determined by real-time PCR. Renilla mRNA levels and Renilla luciferase
activities were normalized to firefly mRNA levels and firefly luciferase activity, respectively, and
are represented relative to lN alone. Corresponding Renilla reporter mRNA levels are given
below each bar. Data are means from at least four independent experiments; error bars
represent standard errors.
Kunz et al.
RNA, Vol. 12, No. 6
MATERIALS AND METHODS
Beta-globin 5boxB, beta-globin WT+300+e3 (control), pCIlN,
pCIFLAG, and hUpf3b constructs were described previously in
Gehring et al. (2003). pCIV5 was created by inserting the V5
sequence (GKPIPNPLLGLDST) into the NheI/XhoI sites of
pCIneo. hUpf3a, Magoh, eIF4AIII, BTZ, RNPS1, hSmg5, hUpf1,
and hUpf2 were amplified from HeLa cell cDNA by PCR (primer
sequences are available upon request). The PCR products were
inserted between the XhoI/NotI sites of pCIlN, pCIFLAG, or
pCIV5. All deletion and point mutants were introduced by site-
directed mutagenesis (sequences are available upon request).
Renilla and firefly luciferase coding sequences were amplified by
PCR from pCREL (Creancier et al. 2000) and inserted between the
NheI/XhoI sites of pCIneo. The 4boxB sequence was amplified by
PCR with XhoI/SalI sites on the primers and inserted into the SalI
site of pCIneorenilla. Finally, firefly, Renilla, and renilla4boxB
were subcloned from pCIneo into pcDNA3.1(?) (Invitrogen)
using NheI and NotI, resulting in pcDNArenilla, pcDNArenilla4-
boxB, and pcDNAfirefly.
Cell culture and transfection
HeLa cells were grown in Dulbecco’s modified eagle medium
under standard conditions. Calciumphosphate transfection of
HeLa cells was performed as previously described in Thermann
et al. (1998).
For experiments shown in Figures 1 and 3, HeLa cells were
cotransfected in 6-well-plates (2 mL medium) with 2 mg of beta-
globin 5boxB, 0.6 mg of the control plasmid (WT+300+e3), 0.4 mg
of a GFP-expression vector, and 1 mg of a lN-expression plasmid
(e.g., pCI lN hUpf3a). Forty hours after transfection, cells were
harvested for analysis of RNA and protein expression. Northern
blotting was performed as described previously in Gehring et al.
For experiments shown in Figure 6, HeLa cells were cotrans-
fected in 6-well-plates with 1 mg of pcDNArenilla4boxB or
pcDNArenilla, 0.1 mg of pcDNAfirefly, 0.2 mg of a GFP-expression
vector, and 1 mg of a lN-expression plasmid.
Seventy-two hours after transfection, cells were harvested for
analysis of luciferase activity and RNA abundance: Cells were
washed two times in PBS and suspended in 350 mL of cold Passive
Lysis Buffer (Promega, Dual-Luciferase Reporter Assay System).
Five microliters of the lysate were used for measuring luciferase
activities according to the manufacturer’s recommendations in
a Centro LB 960 luminometer (Berthold) with the Dual-Luciferase
Reporter Assay System (Promega). The remainder of the lysate
was used for RNA extraction.
cDNA synthesis and real-time PCR
cDNA was synthesized from 2 mg of cytoplasmic RNA using a mix
of primers (20 mM each) (Danckwardt et al. 2004):
In order to avoid amplification of plasmid DNA contaminating
the cDNA preparations, an anchor-specific antisense primer
(59-TGCTGGTCGATGCACGTGACGC-39) and a specific sense
primer were used for real time PCR:
for Renilla luciferase 4boxB: 59-TAGAGTCGACGCTCGCTTC-39,
for renilla luciferase: 59-TCGGACCCAGGATTCTTTTC-39,
for firefly luciferase: 59-GTGGATTACGTCGCCAGTCA-39.
For quantitative PCR, the SYBR Green I Kit and the Light-
cycler 2000 (Roche) were used according to the manufacturer’s
Immunoprecipitation and Western blot
For immunoprecipitation experiments, HeLa cells were trans-
fected in 10-cm dishes with 10 mg of a pCIFLAG plasmid, 2 mg
of a GFP expression vector, and 10 mg of a pCIV5 plasmid, if
required. Cells were harvested 40 h after transfection for immu-
noprecipitation: Cells were washed once with PBS, lysed in 1 mL
lysis buffer (50 mM Tris/HCl, 150 mM NaCl, 1 mM EDTA, 0.5%
Triton-X 100, 0.25% deoxycholate, 13 complete protease inhib-
itors, 1 mM PMSF, 20 mg/mL RNase A at pH 7.2) by passing
several times through a syringe, and centrifuged (10 min, 4°C,
16,000g). The supernatant was assayed for protein concentration
by the Bradford method. Equal protein amounts were incubated
for 1 h on a rotator with 15–20 mL of anti-FLAG-agarose (Sigma).
The beads were washed with 4 3 1 mL lysis buffer without
protease inhibitors. Bound proteins were eluted with 40 mL
200 mM glycin (pH 2.5) and analyzed by SDS-PAGE and Western
blotting. Antibodies against FLAG and V5 were from Sigma.
Secondary antibodies (HRP-conjugated anti-mouse and anti-
rabbit) were from Sigma and used at 1:10,000 dilution for
We thank Nicole Echner for technical assistance and the members
of the Molecular Medicine Partnership Unit for helpful discus-
sions. We also acknowledge the following colleagues for kindly
providing reagents: Ennio DeGregorio and Julie Baron (anti-lN
antibodies), Jens Lykke-Andersen (anti-UPF2 antibodies), and
David Gatfield and Elisa Izaurralde (anti-Y14, Magoh, and p29/
PYM antibodies). J.B.K was supported by the Young Investigator
Award of The Medical Faculty of Heidelberg. This study was
funded by grants KU563/7-1 and KU563/8-1 from the Deutsche
Forschungsgemeinschaft and grant 1999-1076 from the Fritz Thyssen
Received January 10, 2006; accepted February 7, 2006.
Bono, F., Ebert, J., Unterholzner, L., Guttler, T., Izaurralde, E., and
Conti, E. 2004. Molecular insights into the interaction of PYM
with the Mago-Y14 core of the exon junction complex. EMBO
Rep. 5: 304–310.
Chan, C.C., Dostie, J., Diem, M.D., Feng, W., Mann, M.,
Rappsilber, J., and Dreyfuss, G. 2004. eIF4A3 is a novel component
of the exon junction complex. RNA 10: 200–209.
Different functions of hUpf3a and hUpf3b
Creancier, L., Morello, D., Mercier, P., and Prats, C. 2000. Fibroblast
growth factor 2 internal ribosome entry site (IRES) activity ex vivo
and in transgenic mice reveals a stringent tissue-specific regula-
tion. J. Cell Biol. 150: 275–281.
Danckwardt, S., Gehring, N.H., Neu-Yilik, G., Hundsdoerfer, P.,
Pforsich, M., Frede, U., Hentze, M.W., and Kulozik, A.E. 2004.
The prothrombin 39end formation signal reveals a unique archi-
tecture that is sensitive to thrombophilic gain-of-function muta-
tions. Blood 104: 428–435.
Degot, S., Le Hir, H., Alpy, F., Kedinger, V., Stoll, I., Wendling, C.,
Seraphin, B., Rio, M.C., and Tomasetto, C. 2004. Association of
the breast cancer protein MLN51 with the exon junction complex
via its speckle localizer and RNA binding module. J. Biol. Chem.
Ferraiuolo, M.A., Lee, C.S., Ler, L.W., Hsu, J.L., Costa-Mattioli, M.,
Luo, M.J., Reed, R., and Sonnenberg, N. 2004. A nuclear trans-
lation-like factor eIF4AIII is recruited to the mRNA during
splicing and functions in nonsense-mediated decay. Proc. Natl.
Acad. Sci. 101: 4118–4123.
Fribourg, S., Gatfield, D., Izaurralde, E., and Conti, E. 2003. A novel
mode of RBD-protein recognition in the Y14-Mago complex. Nat.
Struct. Biol. 10: 433–439.
Gehring, N.H., Neu-Yilik, G., Schell, T., Hentze, M.W., and
Kulozik, A.E. 2003. Y14 and hUpf3b form an NMD-activating
complex. Mol. Cell 11: 939–949.
Gehring, N.H., Kunz, J.B., Neu-Yilik, G., Breit, S., Viegas, M.H.,
Hentze, M.W., and Kulozik, A.E. 2005. Exon-junction complex
components specify distinct routes of nonsense-mediated mRNA
decay with differential cofactor requirements. Mol. Cell 20: 65–75.
Gonzalez, C.I., Bhattacharya, A., Wang, W., and Peltz, S.W. 2001.
Nonsense-mediated mRNA decay in Saccharomyces cerevisiae.
Gene 274: 15–25.
Hall, G.W. and Thein, S. 1994. Nonsense codon mutations in the
terminal exon of the b-globin gene are not associated with
a reduction in b-globin-mRNA accumulation: A mechanism for
the phenotype of dominant b-thalassemia. Blood 83: 2013–2017.
He, F., Li, X., Spatrick, P., Casillo, R., Dong, S., and Jacobson, A. 2003.
Genome-wide analysis of mRNAs regulated by the nonsense-
mediated and 59 to 39 mRNA decay pathways in yeast. Mol. Cell
Holbrook, J.A., Neu-Yilik, G., Hentze, M.W., and Kulozik, A.E. 2004.
Nonsense-mediated decay approaches the clinic. Nat. Genet. 36:
Kadlec, J., Izaurralde, E., and Cusack, S. 2004. The structural basis for
the interaction between nonsense-mediated mRNA decay factors
UPF2 and UPF3. Nat. Struct. Mol. Biol. 11: 330–337.
Kim, V.N., Kataoka, N., and Dreyfuss, G. 2001. Role of the nonsense-
mediated decay factor hUpf3 in the splicing-dependent exon–exon
junction complex. Science 293: 1832–1836.
Le Hir, H., Gatfield, D., Izaurralde, E., and Moore, M.J. 2001. The
exon–exon junction complex provides a binding platform for
factors involved in mRNA export and nonsense-mediated mRNA
decay. EMBO J. 20: 4987–4997.
Lejeune, F., Li, X., and Maquat, L.E. 2003. Nonsense-mediated mRNA
decay in mammalian cells involves decapping, deadenylating, and
exonucleolytic activities. Mol. Cell 12: 675–687.
Lu, S. and Cullen, B.R. 2003. Analysis of the stimulatory effect of
splicing on mRNA production and utilization in mammalian cells.
RNA 9: 618–630.
Lykke-Andersen, J., Shu, M.D., and Steitz, J.A. 2000. Human Upf
proteins target an mRNA for nonsense-mediated decay when
bound downstream of a termination codon. Cell 103: 1121–1131.
———. 2001. Communication of the position of exon–exon junc-
tions to the mRNA surveillance machinery by the protein RNPS1.
Science 293: 1836–1839.
Maquat, L.E. 2004. Nonsense-mediated mRNA decay: Splicing, trans-
lation and mRNP dynamics. Nat. Rev. Mol. Cell Biol. 5: 89–99.
Mendell, J.T., Sharifi, N.A., Meyers, J.L., Martinez-Murillo, F., and
Dietz, H.C. 2004. Nonsense surveillance regulates expression of
diverse classes of mammalian transcripts and mutes genomic
noise. Nat. Genet. 36: 1073–1078.
Nott, A., Meislin, S.H., and Moore, M.J. 2003. A quantitative anal-
ysis of intron effects on mammalian gene expression. RNA 9:
Nott, A., Le Hir, H., and Moore, M.J. 2004. Splicing enhances
translation in mammalian cells: An additional function of the
exon junction complex. Genes & Dev. 18: 210–222.
Ohnishi, T., Yamashita, A., Kashima, L., Schell, T., Anders, K.R.,
Grimson, A., Hachiya, T., Hentze, M.W., Anderson, P., and
Ohno, S. 2003. Phosphorylation of hUPF1 induces formation of
mRNA surveillance complexes containing hSMG-5 and hSMG-7.
Mol. Cell 12: 1187–1200.
Palacios, I.M., Gatfield, D., St Johnston, D., and Izaurralde, E. 2004.
An eIF4AIII-containing complex required for mRNA localization
and nonsense-mediated mRNA decay. Nature 427: 753–757.
Schell, T., Kulozik, A.E., and Hentze, M.W. 2002. Integration of
splicing, transport and translation to achieve mRNA quality
control by the nonsense-mediated decay pathway. Genome Biol.
Schell, T., Kocher, T., Wilm, M., Seraphin, B., Kulozik, A.E., and
Hentze, M.W. 2003. Complexes between the nonsense-mediated
mRNA decay pathway factor human upf1 (up-frameshift protein
1) and essential nonsense-mediated mRNA decay factors in HeLa
cells. Biochem. J. 373: 775–783.
Serin, G., Gersappe, A., Black, J.D., Aronoff, R., and Maquat, L.E.
2001. Identification and characterization of human orthologues to
Saccharomyces cerevisiae Upf2 protein and Upf3 protein (Caeno-
rhabditis elegans SMG-4). Mol. Cell. Biol. 21: 209–223.
Shibuya, T., Tange, T.O., Sonenberg, N., and Moore, M.J. 2004.
eIF4AIII binds spliced mRNA in the exon junction complex and
is essential for nonsense-mediated decay. Nat. Struct. Mol. Biol. 11:
Stutz, F., Bachi, A., Doerks, T., Braun, I.C., Seraphin, B., Wilm, M.,
Bork, P., and Izaurralde, E. 2000. REF, an evolutionary conserved
family of hnRNP-like proteins, interacts with TAP/Mex67p and
participates in mRNA export. RNA 6: 638–650.
Tange, T.O., Nott, A., and Moore, M.J. 2004. The ever-increasing
complexities of the exon junction complex. Curr. Opin. Cell Biol.
Thermann, R., Neu-Yilik, G., Deters, A., Frede, U., Wehr, K.,
Hagemaier, C., Hentze, M.W., and Kulozik, A.E. 1998. Binary
specification of nonsense codons by splicing and cytoplasmic
translation. EMBO J. 17: 3484–3494.
Wiegand, H.L., Lu, S., and Cullen, B.R. 2003. Exon junction com-
plexes mediate the enhancing effect of splicing on mRNA
expression. Proc. Natl. Acad. Sci. 100: 11327–11332.
Wilschanski, M., Yahav, Y., Yaacov, Y., Blau, H., Bentur, L., Rivlin, J.,
Aviram, M., Bdolah-Abram, T., Bebok, Z., Shushi, L., et al. 2003.
Gentamicin-induced correction of CFTR function in patients with
cystic fibrosis and CFTR stop mutations. N. Engl. J. Med. 349:
Zhou, Z., Luo, M.J., Straesser, K., Katahira, J., Hurt, E., and Reed, R.
2000. The protein Aly links pre-messenger-RNA splicing to
nuclear export in metazoans. Nature 407: 401–405.
Kunz et al.
RNA, Vol. 12, No. 6