Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Archives of Biochemistry and Biophysics 523 (2012) 151–160

Contents lists available at SciVerse ScienceDirect

Archives of Biochemistry and Biophysics


journal homepage: www.elsevier.com/locate/yabbi

Expression profile and interactions of hnRNP A3 within hnRNP/mRNP complexes


in mammals
Christina Papadopoulou, Georgios Boukakis, Vassiliki Ganou, Meropi Patrinou-Georgoula,
Apostolia Guialis ⇑
RNA Processing Program, Institute of Biological Research and Biotechnology, National Hellenic Research Foundation, 48 Vas. Constantinou Avenue, 11635 Athens, Greece

a r t i c l e i n f o a b s t r a c t

Article history: The hnRNP A/B family contains abundant nuclear proteins with major roles in alternative splicing and the
Received 28 November 2011 ability for nucleo-cytoplasmic shuttling. Compared to the best known members of this family (hnRNP A1,
and in revised form 27 March 2012 A2/B1), hnRNP A3 is a relatively less known protein. We report herein immunochemical studies with the
Available online 22 April 2012
hnRNP A3 isoforms (A3a and A3b) that provided evidence for species-specific expression. The unspliced
A3a was found in human and murine cells, while the spliced A3b was a unique and abundant isoform in
Keywords: mouse/rat. In addition, a tissue-specific variation was observed in mice, as the brain was the only tissue
hnRNP A3
found to overexpress hnRNP A3a. Both hnRNP A3a and A3b were able to stably associate with immunos-
hnRNP A/B proteins
hnRNP/mRNP complexes
elected hnRNP and mRNP complexes. Use of the auxiliary domain of hnRNP A3 in pull-down assays on
Protein–protein interactions human cell extracts revealed its unique ability to form a network of interactions not only with other
Post-transcriptional control A/B proteins but also with additional hnRNPs. All interactions, except those of hnRNP A1, were highly
enhanced by previous RNase A digestion of the extracts. Our findings revealed novel characteristics of
hnRNP A3 and supported its extensive involvement in the many aspects of mRNA maturation processes
along with the other hnRNP A/B proteins.
Ó 2012 Elsevier Inc. All rights reserved.

Introduction protein–protein interactions. These proteins all contain a combina-


tion of known RNA-binding domains referred to as the RRM/RBD
Proteins with the ability to bind hnRNA1 (pre-mRNA) and mRNA (RNA recognition motif or RNA-binding domain), the KH (K homol-
constitute a large group of RNA-binding proteins (RBPs) with a major ogy) motif and the RGG (Arg-Gly-Gly) box, with the latter also func-
role in the post-transcriptional regulation of RNA pol II genes. In par- tioning in protein–protein interactions [2]. The RRM/RBD is the most
ticular, hnRNP proteins represent a special group of RBPs that are common and best analyzed motif both structurally and functionally
stable components of hnRNP complexes, the sites of mRNA matura- [3]. The hnRNP proteins have a major nucleoplasmic localization
tion. More than 20 distinct hnRNP proteins have been identified, with several of them (A, D, E, I, K, L) capable of nucleo-cytoplasmic
ranging from 32 to 110 kDa and designated (in order of increasing shuttling, while others (C and U) do not exit the nucleus except un-
molecular weight) as hnRNP A–U [1]. The majority of hnRNPs con- der certain cellular conditions [4,5]. In addition to the major nuclear
tain alternatively spliced variant species as well as a number of role of hnRNPs in mRNA processing (splicing, polyadenylation) and
post-translational modifications (mainly phosphorylation and argi- transport, they are known to participate in several other events,
nine methylation). In addition to their generalized role as RNA pack- including transcription, DNA repair and telomere DNA formation.
aging proteins, hnRNPs have specialized functions that are The limited cytoplasmic localization of some hnRNPs is also of func-
dependent on their ability to form specific RNA-protein and tional importance and affects the stability, translation and localiza-
tion of their associated mRNAs. Thus, hnRNPs are considered
important gene regulators, and any alteration in their expression
⇑ Corresponding author. Fax: +30 210 7273755. pattern is expected to affect many cellular functions (for recent re-
E-mail address: aguial@eie.gr (A. Guialis).
1
views, see [6,7]).
Abbreviations used: hnRNP, heterogeneous nuclear ribonucleoprotein; RBP, RNA-
binding protein; RRM, RNA recognition motif; RBD, RNA binding domain; KH, K
In particular, the members of the hnRNP A/B-type family (A1,
homology; RGG, Arg-Gly-Gly box; CBF-A, CArG box-binding factor-A; A2RE, A2 A2/B1, A3, and A0) constitute a special family of proteins that share
response element; 30 UTR, 30 untranslated region; 2-D, two-dimensional; NEPHGE, sequence and structural similarity. These proteins have two N-ter-
non-equilibrium pH gradient electrophoresis; SDS–PAGE, SDS–polyacrylamide gel minal RRM/RBD domains in tandem followed by a C-terminal aux-
electrophoresis; GST, glutathione S-transferase; IPTG, isopropyl b-D-1-thiogalactopy-
iliary domain that is rich in glycine (2xRBD-Gly). Their high
ranoside; ECL, enhanced chemiluminescence; PAS, protein A-sepharose; ssDNA,
single-stranded DNA; IP, immunoprecipitation. sequence homology is mainly due to their shared N-terminal

0003-9861/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.abb.2012.04.012
152 C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160

RNA-binding domain, while their functional diversity originates A3 that support its extensive participation in a network of interac-
from the less conserved auxiliary domain that is mainly responsi- tions for the assembly of hnRNP complexes.
ble for protein–protein interactions [8,9]. Additional members of
the hnRNP A/B family of murine and human origin, referred to as
Materials and methods
the D-subgroup (hnRNP AB, D, DL), are also known [10]. A tran-
scriptional regulator belonging to this subgroup, named CBF-A
Cell culture and extract preparation
(CArG box-binding factor-A), has been identified as the mouse
orthologue of the human hnRNP AB [11] and has since attracted
Two human cell lines, HeLa (cervix carcinoma) and A549 (lung
a great deal of interest [12].
adenocarcinoma), were used interchangeably to prepare cell ex-
The hnRNPs A/B and C1/C2 are highly expressed hnRNPs that,
tracts. Monolayer cultures of exponentially growing cells were
together with their associated RNAs, compose stable components
maintained in D-MEM medium with 10% FCS. The rat Novikoff hep-
of the 40S core particles as isolated from partially degraded nuclear
atoma cell line was similarly grown in D-MEM. Additionally, three
extracts [13]. The most related A/B proteins are hnRNP A1, A2/B1
mouse cell lines (LA-4, CMT64/61 and PVDC57) were employed.
and A3, the latter being a relatively new and less studied member.
The lung cell lines LA-4 (adenoma) and CMT64/61 (carcinoma)
The participation of hnRNP A1 (the prototype A/B-type protein)
were purchased from the European collection of cell cultures
and A2/B1 in many cellular functions has been documented by a
(ECACC). The LA-4 cells were grown in Ham’s F12 medium supple-
series of studies. These proteins primarily act as splicing regulators
mented with 2 mM glutamine, 1% non-essential amino acids, 15%
(usually splicing repressors), thus affecting the expression of their
FCS, and CMT64/61 in Waymouth’s MB 752/1 medium with
associated mRNA targets [9,14 for Refs. therein]. Downstream
2 mM glutamine and 10% FCS. The third mouse cell line of skin ori-
mRNA targets involved in a variety of cellular functions have been
gin, PDVC57 (squamous cell carcinoma), was derived as described
determined for hnRNP A1, A2 and A3 in human cells, and a signif-
in [29] and provided by VK Zoumpourlis (IBRB-NHRF). These cells
icant number of which are shared among these hnRNPs [9]. All
were grown in D-MEM with 10% FCS. Penicillin and streptomycin
three proteins are, together with additional hnRNPs and other
(1% v/v) were added to all cultures. All culture media and FCS were
RBPs, members of a group of proteins able to associate with the
obtained from Gibco Co.
AU-rich cis-elements in the 30 UTR regions of many mRNAs and
The preparation of the nuclear and cytoplasmic extracts from
contribute to their stability and translation [15,16]. Other func-
cultured cells was performed as outlined previously [30]. Affinity
tions shared by hnRNP A2/B1 (but not A1) [17], A3 [18] and CBF-
chromatography on ssDNA-cellulose beads was employed to purify
A [19] include their ability to act as mRNA trafficking trans-factors
hnRNP proteins from whole cell extracts that had been previously
and to directly bind a 21-nucleotide-long element, called the
treated with micrococcal nuclease to remove nucleic acids [31].
A2-response element (A2RE), that is present in several mRNAs of
The proteins that eluted from the beads were resolved on the
neural cells. Moreover, hnRNP A1, A2/B1 and A3 bind to both sin-
two-dimensional (2-D) non-equilibrium pH-gradient electrophore-
gle-stranded telomere DNA and telomerase RNA, regulating their
sis (NEPHGE/SDS–PAGE) gel system as described previously [26].
interactions [20–22].
The whole cell extracts were obtained by lysing a cell pellet
Our interest in hnRNP A3 arose from the detection of autoanti-
containing approximately 1–2  107 cells/ml in NET-2 buffer
bodies that react against discrete hnRNP A/B-type proteins in the
(10 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.05% Nonidet P-40), in
sera of patients with systemic autoimmune diseases. A specific
the presence of 2 lg/ml protease inhibitor mix (Sigma Co.) on
class of these autoantibodies targeted a major protein component
ice. Following sonication (6, 20 s) and centrifugation at
(named mBx) of the rat liver 40S hnRNP complexes, which was
10,000 rpm for 10 min at 4 °C, the resulting supernatant was col-
nearly undetected in human (HeLa) cells [23]. The subsequent iso-
lected and stored at 80 °C.
lation and characterization of a partial cDNA clone coding the en-
tire auxiliary domain of the rat mBx identified it for the first
time as the murine-specific hnRNP A3 protein [24]. We have since Antibodies
referred to this hnRNP A/B protein that is abundant in murine cells
as hnRNP A3/mBx [25,26]. Thereafter, Ma et al. [18] reported the The monoclonal antibodies targeting hnRNP A1 (4B10), hnRNP
isolation from rat and human brain of a full-length cDNA encoding A2/B1 (DP3B3) hnRNP C1/C2 (4F4), hnRNP L (4D11), hnRNP M
hnRNP A3 that, with the exception of one amino acid, had an iden- (1D8) and a-tubulin (TU-02) were purchased from Santa Cruz Bio-
tical sequence between human and rat species. This isolation led to tech. The monoclonal antibody against b-actin (1501) was pur-
the characterization of the two main variants: a 41 kDa unspliced chased from Chemicon. The antibody against CBF-A, which was
form that incorporates the N-terminal exon consisting of 22 amino raised in rabbits as described in [19], was the kind gift of P. Percip-
acids (hnRNP A3a) and its 38 kDa spliced variant (hnRNP A3b) alle (Sweden). The rabbit polyclonal antibody against hnRNP A3
[18,27]. An apparent functional redundancy between hnRNP A3 [anti-A3(N)] that was raised against the N-terminal exon unique
and A1 has been suggested based on their high sequence similarity to the unspliced hnRNP A3a variant was obtained from Abcam Lim-
[18] as well as on RNAi suppression studies [28]. ited, UK (code No. 50949).
The present study focused on the immunochemical character- The rabbit anti-A3aux antibody was produced in our lab by using
ization of hnRNP A3 in mammals with the aim to describe novel the E. coli-expressed auxiliary domain of hnRNP A3 fused to GST as
biochemical properties that might relate to its biological function. an immunogen. The recombinant GST-A3aux protein was first re-
Aided by the production of a homemade antibody able to react solved by SDS–PAGE and localized on the gel by Coomassie staining.
with both the unspliced and spliced forms of hnRNP A3 (A3a and The corresponding gel area was excised, pulverized in liquid nitro-
A3b, respectively), we have verified their differential expression gen and homogenized in phosphate-buffered saline (PBS). The
patterns in human and murine cells. We also investigated the tis- immunization protocol was performed as reported in [32]. The ob-
sue-specific expression of the hnRNP A3 isoforms in mice. The ma- tained rabbit serum was clarified by centrifugation at 6500g for
jor isoform in all examined mouse tissues was hnRNP A3b, with the 10 min at 4 °C, mixed with an equal volume of PBS and brought to
notable exception of the brain, in which the unspliced A3a isoform a final concentration of 50% in (NH4)2SO4. After stirring on ice for
was overexpressed. In addition, the combination of immunopre- 6 h, the solution was centrifuged as described above, and the col-
cipitation studies and pull-down assays using the GST-fused auxil- lected pellet was resuspended in PBS and dialyzed extensively
iary domain of hnRNP A3 revealed novel characteristics of hnRNP against the same buffer. The affinity purification of the antibodies
C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160 153

contained in the dialyzed material was performed against the re- The expression of the full-length hnRNP M as a GST-fused pro-
combinant GST-A3aux protein that was first resolved by SDS–PAGE tein and the subsequent cleavage of the GST tag were performed as
and then immobilized on a nitrocellulose membrane. The elution of described in [34].
the membrane-bound antibodies was performed as in [26].
Results
Protein resolution and western blot analysis
Differential expression of hnRNP A3 isoforms in cells of human and
The protein samples were resolved by SDS–PAGE (10% acrylam- murine origin
ide gel) and then transferred to nitrocellulose membranes. Wes-
tern blotting with the application of the appropriate antibody To establish the previously described presence of discrete
dilution and the visualization of the antigen–antibody reaction hnRNP A3 isoforms in human and rat/mouse cells [23], we per-
with ECL was performed as in [30]. formed a direct comparison of all hnRNP A/B protein species from
a human (HeLa) and a rat (Novikoff) cell line. Whole cell extracts
Mouse tissue lysate preparation were subjected to affinity chromatography on ssDNA-cellulose col-
umns to greatly enrich for the hnRNP proteins that were then re-
Animal tissues were obtained from a male young adult Balb/c solved by 2-D (NEPHGE/SDS–PAGE) gel electrophoresis and
mouse. The dissected tissues were immediately snap-frozen in li- transferred to a nitrocellulose membrane. The Ponceau S staining
quid nitrogen and kept frozen until use. of the resolved protein species is presented in Fig. 1A (upper part),
For a total tissue lysate, a small piece (in the range of 50–60 mg) and the position of hnRNP A/B is marked with a square box. The
was cut from the frozen tissue and homogenized in 5 vol. of immunochemical identification of these proteins was performed
homogenization buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, after the western blotting with monoclonal antibodies against
1 mM EDTA, 1% Nonidet P-40 and 2 lg/ml protease inhibitor hnRNP A2/B1 or A1 as well as with the commercial rabbit antibody
mix; Sigma Co.) as described in [33]. The NE-PER nuclear and raised against the N-terminal exon of the unspliced hnRNP A3a iso-
cytoplasmic extraction reagents kit (Pierce Biotechnology) was form [anti-A3(N)] (Fig. 1A). As observed, similar protein spots for
employed with weighted portions of liver (Li) and brain (Br) tissue hnRNP A1, A2/B1 and A3 were detected in both cell lines. In addi-
(50 mg each), and the corresponding nuclear and cytoplasmic ex- tion, the parallel application of one of our previously characterized
tracts were isolated following the exact instructions provided by human autoimmune sera with a mixed specificity against hnRNP
the manufacturer in the presence of the protease inhibitor mix. A2/B1 and the murine-specific hnRNP A3/mBx (denoted a-B2/Bx,
The concentration of the soluble protein corresponding to the a-A2) [26] verified the presence within the Novikoff hnRNP A/B
whole tissue lysates as well as of the brain and liver nuclear and proteins of at least three major spots corresponding to hnRNP
cytoplasmic extracts was determined by the Bradford assay. A3/mBx that were undetected in HeLa cells, in clear opposition
to the detection of hnRNP A2 in both cell lines (Fig. 1A).
Immunoprecipitation and pull-down assays The immunochemical identification of the hnRNP A/B proteins
presented in Fig. 1A permitted the labeling of almost every spot
The application of the immunoprecipitation (IP) reactions was in the HeLa and Novikoff hnRNP A/B blots, as shown in Fig. 1B. This
performed as described previously [30]. Protein A-Sepharose assignment in the rat (Novikoff) cell line was in perfect agreement
(PAS) beads were incubated with the appropriate antibody, and with our previous report on the rat liver hnRNP A/B proteins [26].
the resulting PAS-IgG fraction was mixed with the cellular extract. The application of two discrete anti-hnRNP A3 antibody specifici-
As a negative control, IP reactions were performed either with ties (rabbit polyclonal and human autoantibodies) provided the
beads alone (mock IP) or with mouse IgG of the same isotype (in first evidence for the differential expression of two hnRNP A3 iso-
equivalent amounts to the antibody applied). The bound proteins forms: the hnRNP A3 isoform, which is common to both human
were released from the beads directly in SDS-sample buffer and re- and mouse, and hnRNP A3/mBx, which is the major isoform unique
solved by SDS–PAGE followed by western blotting with appropri- to murine cells. To simplify the hnRNP A3 protein labeling and to
ate antibodies. conform with the recently reported major hnRNP A3 isoforms of
In vitro pull-down assays were performed as in [30] on nuclear 41 and 38 kDa [18], we shall hereafter refer to the isoform common
and cytoplasmic extracts that had been pre-cleared by incubation to human and mouse as the unspliced hnRNP A3a and to the major
with glutathione beads. The subsequent incubation of the cellular murine isoform, hnRNP A3/mBx, as the spliced hnRNP A3b isoform.
extracts was performed with either GST or GST-fused A3aux re- With the exception of the commercially available rabbit anti-
combinant proteins bound to glutathione beads. The resolution body against hnRNP A3a, we had no access to a comparable anti-
of all of the proteins bound to the beads was performed as for body for hnRNP A3b, which would be a desirable tool for further
the IP reactions. studies. As the use of the autoantibodies in human sera that reacted
with hnRNP A3 had limited application due to their polyspecific
Construction and expression of recombinant proteins nature, we produced a new antibody. For this, we utilized our pre-
viously reported partial cDNA clone of rat origin that codes for the
The isolation of the partial cDNA clone encoding the auxiliary entire C-terminal auxiliary domain of hnRNP A3 that is common to
domain of the rat hnRNP A3/mBx was previously described by both isoforms [24] and expressed it as a GST-fused product (GST-
our lab [24]. To generate constructs for the expression of the cod- A3aux). The specificity of this new, affinity-purified rabbit antibody
ing auxiliary domain of hnRNP A3 (A3aux) fused to glutathione (anti-A3aux) and its direct comparison with the commercial anti-
S-transferase (GST), the cDNA was excised from a pBluescript plas- A3(N) (applied above) is shown in Fig. 2A in parallel with the immu-
mid and re-inserted into the expression vector pGEX4T3 (Amer- nodetection of hnRNP A1, which served as a reference hnRNP A/B
sham). The fused GST-A3aux protein and GST alone were used to protein, and b-actin. This comparison involved the western blotting
transform the bacterial strain BL21(DE3), and their expression of whole cell extracts that were prepared from two human (HeLa
was induced with isopropyl b-D-1-thiogalactopyranoside (IPTG). and A549) and three mouse (CMT, LA-4 and PVDC) cell lines. As ex-
Glutathione–agarose beads (Sigma) were then used to purify the pected (see also Fig. 1A), the application of the anti-A3(N) antibody
recombinant proteins from the bacterial lysates as described in led to reaction with a single protein species common to both human
[34]. and mouse cell lines, which corresponded to the hnRNP A3a
154 C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160

A
HeLa Novikoff
NEPHGE
kDa H+ OH- H+ OH-
B
116
SDS-PAGE

97 HeLa
80
66 B2
A1B
55 A3 (A3a)
B1
45

A2
A1
30

hnRNP A/B
WB: Novikoff
hnRNP B1
α-A2/B1
hnRNP A2 B2
A3/mBx (A3b)
A1B
A3 (A3a)
B1

α-A1 hnRNP A1B A2


hnRNP A1 A1

α-A3(N) hnRNP A3

AutoAbs:
hnRNP A3/mBx
α-B2/Bx, α-A2 hnRNP A2

Fig. 1. Direct comparison of the hnRNP A/B proteins between human (HeLa) and murine (rat Novikoff) cells. The hnRNP proteins were isolated from whole cell extracts by
affinity chromatography on ssDNA-cellulose columns, and equal protein amounts (100 lg) were resolved on 2-D NEPHGE/SDS–PAGE gels. (A) Following transfer to
nitrocellulose membranes, the resolved proteins were visualized by reversible staining with Ponceau S. The position of hnRNP A/B is indicated by a box. The membranes were
then western blotted using monoclonal antibodies against hnRNP A2/B1 and A1, the a-A3(N) rabbit antibody targeting hnRNP A3 and a human autoimmune serum with
mixed specificity for hnRNP A2 and the A3/mBx variant form exclusive to Novikoff. (B) The assignment of the immunochemically identified hnRNP A/B protein species,
including the unspliced hnRNP A3(A3a) and the spliced A3/mBx(A3b) isoforms, is indicated in the enlarged picture of the proteins boxed in A.

isoform. This reaction was clearly distinct from the specificity of the tissues, including the kidney (Ki), spleen (Sp), lung (Lu), heart
anti-A3aux antibody, which recognized the hnRNP A3b isoform as (He), liver (Li) and brain, with the latter corresponding to the bulk
the major protein species almost exclusive to the mouse cell lines of brain tissue (cerebrum; Br) and to the cerebellum alone (Br/ce).
and also reacted with the hnRNP A3a isoform common to both hu- A similar amount of total protein extracted from the different
man and mouse. An unexpected finding with the anti-A3aux anti- mouse tissues was resolved by SDS–PAGE followed by the immu-
body was the reaction with an additional protein of nodetection of the antigenic proteins, as shown in Fig. 2C. With re-
approximately 75 kDa (and a probable degraded fraction of it) that spect to the hnRNP A3 species, the first observation was the major
was unique to the mouse cell extracts (Fig. 2A). As this antibody difference in their relative enrichment in the extracted tissues. The
population was affinity purified against the same recombinant pro- highest levels were observed in the spleen and lung and the lowest
tein used as an immunogen, this presently unidentified cross-reac- (almost background levels) in the kidney and heart tissues; this
tive murine-specific protein is likely to share some degree of pattern is very similar to that of hnRNP A1. Moreover, the spleen,
homology with the auxiliary domain of hnRNP A3. lung, liver and kidney (the latter referring to a higher concentration
Because the combined use of the anti-A3(N) and anti-A3aux of a tissue homogenate; data not shown) as well as the brain cer-
polyclonal antibodies could clearly distinguish between the unsp- ebellum, although not the brain (cerebrum) tissue, exhibited a
liced and spliced hnRNP A3 isoforms (A3a and A3b, respectively), higher A3b–A3a ratio, in line with the mouse/rat cell lines shown
these antibodies were applied to 2-D gels of the same type pre- in panels A and B of Fig. 2. The A3b–A3a ratio was reversed only
sented above in Fig. 1A. This final analysis, shown in Fig. 2B, veri- in the case of the brain tissue (excluding cerebellum), which is
fied the presence of the common hnRNP A3a isoform in human an interesting observation that agreed with the recently reported
and murine cells, in contrast to hnRNP A3b, which appeared almost findings upon the application of whole brain extracts [27].
exclusively in the murine cells. The 2-D gel resolution of hnRNP To investigate the sub-cellular distribution of the hnRNP A3a
A3b revealed (as in Fig. 1) three major and at least two minor spots, and A3b isoforms, we selected the liver and brain tissues to isolate
which were most likely the products of extensive post-transla- their corresponding nuclear and cytoplasmic extracts (as indicated
tional modification (e.g., phosphorylation, methylation). in Materials and methods). The liver represented all of the tissues
We then performed the immunodetection of hnRNP A3 with a higher A3b–A3a ratio, while the brain tissue was employed
isoforms (A3a and A3b) in extracts prepared from several mouse because it was the sole tissue that exhibited a reversed ratio of the
C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160 155

A mouse human mouse human B


A549

PDVC
Novikoff

PDVC
HeLa

A549

HeLa

A549
LA-4

LA-4
CMT
CMT
NEPHGE
OH-

SDS-PAGE
kDa kDa H+ H+ OH-

97 80
66
80
55
66 * 45
hnRNP A3a
55
hnRNP A3a 30
45
hnRNP A3b
WB: anti-A3(N)
30

WB: anti-A3(N) WB: anti-A3aux


*
hnRNP A3a
hnRNP A1
hnRNP A3b

β-actin
WB: anti-A3aux

C D Liver (Li) Brain (Br)


Ki Sp Lu He Li Br Br/ce C N C N

hnRNP A3a hnRNP A3a

short exp long exp


hnRNP A3a hnRNP A3a

hnRNP A3b hnRNP A3b

hnRNP A3a hnRNP A3a

hnRNP A3b hnRNP A3b


short exp long exp

hnRNP A1 hnRNP A1

β-actin α-tubulin

Fig. 2. Immunochemical characterization of hnRNP A3 isoforms in mouse and human cells. (A) Equal amounts (100 lg) of the proteins extracted from the three mouse
(CMT64/61, LA-4 and PVDC57) and the two human (HeLa and A549) cell lines were resolved by SDS–PAGE and transferred to nitrocellulose. Duplicate membranes were
incubated with either the commercial antibody targeting the unspliced hnRNP A3a variant alone [anti-A3(N)] or the new homemade affinity-purified anti-A3aux antibody
able to react with both A3a and A3b isoforms. The detection of a cross-reactive protein of roughly 75 kDa in the mouse extracts by the anti-A3aux antibody is marked (⁄). The
parallel immunodetection of hnRNP A1 and b-actin is also shown. (B) The 2-D resolution (NEPHGE/SDS–PAGE) of the hnRNP proteins isolated from the human (A549) and the
rat (Novikoff) cell extracts was performed as in Fig. 1A, followed by western blotting with either the anti-A3(N) or the anti-A3aux rabbit antibodies. The immunodetection of
the spots corresponding to the hnRNP A3a and A3b isoforms is shown. (C). Equal amounts (100 lg) of the soluble proteins obtained from mouse tissue homogenates
corresponding to the kidney (Ki), spleen (Sp), lung (Lu), heart (He), liver (Li) and two brain sub-areas, the cerebrum (Br) and cerebellum (Br/ce), were resolved by SDS–PAGE
and transferred to nitrocellulose. The immunodetection of the hnRNP A3a and A3b isoforms, hnRNP A1 and b-actin was performed by blotting with the appropriate
antibodies. (D). The nuclear (N) and cytoplasmic (C) extracts were obtained from portions of the liver (Li) and brain (Br) tissue (50 mg each), and an amount of the total
nuclear and cytoplasmic protein that corresponded to the same percentage (1/10th) of mouse tissue homogenate was resolved by SDS–PAGE. For this experiment, a much
higher concentration of cytoplasmic protein was loaded onto the gel than the nuclear protein (roughly 10-fold for the brain and 30-fold for the liver). The immunodetection of
the hnRNP A3a and A3b isoforms was performed on duplicate gels, and the immunodetection of hnRNP A1 and a-tubulin was performed by the sequential blotting of the
same gels. The 4-pointed star symbol is used to indicate the presence of a reactive protein impurity (of approximately 50 kDa) in the liver cytoplasmic extract. Either a short
or long ECL exposure of part of the gel (as indicated in panels C and D) that refers to the application of the anti-A3aux antibody is also included to enable a better resolution
between the hnRNP A3a and A3b isoforms within the mouse tissues.

isoform types, as observed in Fig. 2C. To estimate the relative and brain (also in both cellular compartments), there were no
enrichment of the hnRNP A3 isoforms in the nuclear and cytoplas- other substantial differences observed with respect to their sub-
mic compartments, we based our comparison on tissue equiva- cellular distributions.
lents. For this experiment, the same percentages of the extracted The present study provides the first comparison of the hnRNP
nuclear (N) and cytoplasmic (C) total proteins were resolved on A3 variants within different mouse tissues, which led to the iden-
duplicate gels. Western blotting was then performed to identify tification of the brain as the sole tissue thus far to contain a larger
hnRNP A3a and A3b in parallel with hnRNP A1 and a-tubulin amount of A3a than A3b.
(the latter served as a cytoplasmic index protein). The results of
this analysis are presented in Fig. 2D. As clearly observed, hnRNP Presence of hnRNP A3 within immunoselected hnRNP/mRNP
A3a and A3b were highly enriched in the nuclear compartment complexes from mammalian cell extracts
in both the liver and brain tissues, with a minor fraction of both
isoforms also detected in the cytoplasmic fraction, similar to the In comparison to hnRNP A1 and A2/B1, hnRNP A3 remains the
cellular distribution of hnRNP A1. Aside from the clear reproduc- least studied member of the hnRNP A/B family in human cells. In
tion of the differential ratio of A3a and A3b between the liver Fig. 3A, the immunodetection of the human hnRNP A/B is observed
156 C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160

in the nuclear (N) and cytoplasmic (C) fractions from the HeLa and monoclonal antibody that had previously enabled the detection
A549 cell lines in parallel with the exclusively nuclear hnRNP C1/ of hnRNP A3 among the hnRNP A1-associated protein species
C2 [35], shown for comparison, and with b-actin. The hnRNP A3a [30]. Fig. 3B(a) shows the anti-A1 IP assays of the A549 nuclear
isoform unique to human cells was mainly concentrated in the nu- and cytoplasmic fractions with or without their prior RNase A
clear compartment, as observed for hnRNP A1 and A2/B1. However, digestion (±RNase) in parallel with mock reactions. As observed
similar to hnRNP A1 (but not A2/B1), a minor fraction of hnRNP by the protein resolution of the immunoprecipitated protein and
A3a was clearly detected in the cytoplasmic fractions. These results the subsequent western blotting with antibodies for hnRNP A3a,
were in good agreement with the nuclear and cytoplasmic distri- A1 and C1/C2 (the latter also served as a nuclear protein index),
bution of hnRNP A3a in the mouse tissues as presented in Fig. 2D. hnRNP A3a was clearly associated with the hnRNP A1-selected nu-
We then examined the ability of the anti-hnRNP A3 antibodies clear and cytoplasmic RNP complexes. In both cellular compart-
to immunoselect hnRNP/mRNP complexes from fractionated ments, this association was sensitive to the prior RNase A
human (A549) cell extracts. However, both the anti-A3(N) and digestion of the extracts, indicative of an RNA-dependent interac-
anti-A3aux polyclonal antibodies performed very poorly in immu- tion, which was similar to the nuclear hnRNP C1/C2-A1 association.
noprecipitation (IP) studies (data not shown). Due to this limita- The above IP studies were expanded to include antibodies against
tion, we relied solely on the reciprocal use of the anti-hnRNP A1 CBF-A, which is another hnRNP A/B type protein. This choice was

PDVC LA-4
C

IgG-co

IgG-co
α- Α1

α- Α1
A PDVC LA-4
C N C N
HeLa A549 kDa
80
WB: anti-A3(N)

C N C N 66 IgGH
55
hnRNP A1
45 hnRNP A1
hnRNP A3a
30
hnRNP A3a
WB: anti-A3aux

80
hnRNP A2/B1
66 * IgGH
55 hnRNP A3a
45 hnRNP A3a
hnRNP C1/C2
30 hnRNP A3b

β-actin 80
IgGH
66 hnRNP A3a
55 α- tubulin
hnRNP A3b
a. b.
B
Α549: Nuclear Ex. Cytoplasmic Ex.
Input

Input

mock IP α-Α1 mock IP α-Α1 HeLa Nuclear Ex.


RNase:
α-CBF-A

α-CBF-A

- + - + - - + - + -
mock IP

hnRNP A1
A1 fr.
RNase: - - +
hnRNP A3a CBF-A

hnRNP A3a
hnRNP C1/C2

a. b.
Fig. 3. Cellular distribution and detection of hnRNP A3 isoform species within hnRNP complexes immunoselected from human and murine cell extracts. (A) The nuclear (N)
and cytoplasmic (C) extracts were prepared from two human cell lines (HeLa and A549), and equal protein amounts (50 lg) were resolved by SDS–PAGE and transferred to
nitrocellulose. The immunodetection of the hnRNP A1, A3a, A2/B1 and C1/C2 proteins is shown in parallel with b-actin following blotting with the appropriate antibodies.
(Ba) The monoclonal antibody against the hnRNP A1 protein (a-A1) was employed in immunoprecipitation (IP) assays using nuclear and cytoplasmic extracts from human
(A549) cells. For the extracts used, the protein amount corresponded to the same cell equivalent (1.7  107 cells/IP) with (+) or without () prior treatment with RNase A.
Parallel mock IP reactions in the absence of the antibody were performed as negative controls. The proteins present in the immune pellets were resolved by SDS–PAGE,
transferred to nitrocellulose and blotted with antibodies against hnRNP A1, A3a and C1/C2. The input corresponds to an aliquot of the extract applied in IP (A1fr. indicates the
degradation of hnRNP A1 that occurred in this experiment). (Bb) Similar IP assays as those in Ba were performed on a HeLa nuclear preparation using an antibody against CBF-
A. The immunodetection of CBF-A and hnRNP A3a in the immune pellets is shown. (Ca) The nuclear (N) and cytoplasmic (C) extracts were prepared as in panel A using two
mouse cell lines (PDVC and LA-4), and the immunodetection of the hnRNP A3a and A3b variants followed the application of the anti-A3(N) or anti-A3aux antibodies. The
parallel immunodetection of a-tubulin, a known cytoplasmic index protein, is also shown, as is the presence of the as yet uncharacterized cross-reactive protein (⁄) initially
presented in Fig. 2A. (Cb) The nuclear extracts corresponding to the PVDC and LA-4 mouse cells were subjected to IP reactions with the application of the a-A1 antibody (as in
Ba) in parallel with a control IP (IgG-co). Following the protein resolution of the immune pellets and transfer to nitrocellulose, the presence of the immunodetected hnRNP A1
and A3a and A3b variants was indicated. IgGH refers to the heavy IgG chain.
C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160 157

based on our previous findings that showed the ability of nuclear and A2 [36], recombinant A3aux was able to make homomeric
actin to associate with a subset of the rat liver hnRNP A/B proteins interactions with the endogenous hnRNP A3a and to form het-
[25]. This subset includes CBF-A as well as the minor forms of ero-complexes with the other hnRNP A/B, A1 and A2. However,
hnRNP A2 and A3 (excluding their major variant forms). In in clear opposition to hnRNP A1 [37], GST-A3aux could also associ-
Fig. 3B, panel b, the outcome of the application of the anti-CBF-A ate stably with at least two more hnRNPs, the hnRNP M and L pro-
antibody to an RNase A-treated or -untreated human (A549) nucle- teins, in addition to the A/B proteins.
ar extract is presented. The western blotting identification of CBF-A The availability of a full-length recombinant hnRNP M [34] al-
and hnRNP A3a in the immune pellets, in direct comparison to the lowed us to use a pull-down assay to test whether the interaction
mock IP, revealed the co-precipitation of hnRNP A3a with CBF-A. between GST-A3aux and hnRNP M was direct and occurred in the
Interestingly, this association was unaffected by RNase A digestion, absence of any endogenous proteins. The outcome of this ap-
in clear opposition to the RNase-sensitive hnRNP A1-A3a interac- proach, shown in Fig. 4C, provided the novel identification of the
tion shown in panel 3B (a). ability of these two recombinant species (corresponding to hnRNP
A similar procedure to that applied to the human (A549) cell ex- A3 and M) to form a direct protein–protein interaction in vitro that
tracts was also performed with mouse cell extracts, and both was stable under the high-salt washing of the beads.
hnRNP A3a and A3b isoforms could be identified. The nuclear Additional pull-down assays were performed to examine the
and cytoplasmic extracts obtained from the two mouse cell lines properties of the observed association of hnRNP A3 with the
(PDVC and LA-4) were blotted with either anti-A3(N) or anti- endogenous hnRNPs. In this experiment (see Fig. 4D), GST-A3aux
A3aux antibodies to detect the hnRNP A3a or A3b isoforms, respec- or GST alone was incubated with the fractionated nuclear and cyto-
tively (Fig. 3C(a)). Both isoforms exhibited a major nuclear plasmic extracts (from A549 cells) with or without their prior
localization with a minor fraction of hnRNP A3b detected in the RNase A-digestion. After washing the beads in the low-salt buffer,
cytoplasmic compartment (see detection of a-tubulin as a cyto- the immunodetection of the bound hnRNP protein species was per-
plasmic index). We note in this aspect the similar distribution of formed. A major finding of this approach was that, with the excep-
the as yet unidentified cross-reactive 75 kDa protein. Thereafter, tion of hnRNP A1, the interactions between GST-A3aux and the
the application of the anti-hnRNP A1 antibody in IP assays endogenous hnRNPs A3a, A2/B1, L and M in the nuclear extract ap-
performed on the mouse nuclear extracts provided additional evi- peared to be insensitive to prior RNase digestion, with a notable
dence for the ability of both hnRNP A3 isoforms (A3a and A3b) to RNase-induced enhancement. This result was most likely due to
associate specifically with the hnRNP A1-selected RNP complexes the release of hnRNP proteins from the RNase-treated hnRNP com-
in mammals (Fig. 3C(b)). plexes and, thus, their increased accessibility for interaction with
GST-A3aux. An overall similar pattern of interactions as that ob-
Interaction of hnRNP A3 with other protein members of hnRNP served with the nuclear extract was observed in the cytoplasmic
complexes in vitro extract, albeit with much lower intensity and without the detec-
tion of hnRNP A2/B1. The RNA-sensitive association of hnRNP A1
With the notable exception of hnRNP A3, the pattern of interac- with GST-A3aux, in contrast to that observed with hnRNP L, was
tions operating amongst most human hnRNP proteins has been also observed in the HeLa nuclear extract, as presented in Fig. 4E,
extensively analyzed both in vivo and in vitro [36,37]. To investi- in agreement with the IP experiment (Fig. 3B) that showed an
gate the type and nature of the interactions hnRNP A3 could have RNA-dependent association of hnRNP A3a with A1.
with the other hnRNP proteins in vitro, the recombinant GST- Based on the above findings, the ability of the auxiliary domain
A3aux protein was bound to agarose–glutathione beads and em- of hnRNP A3 to participate in stable protein–protein interactions
ployed in pull-down assays using fractionated human (A549) cell with a number of hnRNP proteins in both the nuclear and cytoplas-
extracts. Following the incubation of the beads with either the nu- mic compartments was clearly demonstrated in vitro.
clear or cytoplasmic extracts, the association of the endogenous
hnRNP proteins with the recombinant GST-A3aux was investi-
gated, using recombinant GST alone as a negative control. Because Discussion
hnRNPs are mainly nuclear proteins, a higher cytoplasmic (com-
pared to nuclear) protein concentration was used in the pull-down One major outcome of the present study was the establishment
assays, which also improved the detection of any interactions be- of the differential protein expression of hnRNP A3 between human
tween the endogenous hnRNPs and the recombinant proteins. and murine (rat/mouse) cells. This conclusion was based on our
The incubated beads were then extensively washed either with investigation of hnRNP A3 in a number of human and murine cell
the standard low buffer (150 mM NaCl) or a higher stringency lines (including mouse tissue lysates) that extended and verified
wash buffer (300 mM NaCl). The proteins that bound to the beads our initial work on this protein [23]. In the present study, the
were resolved by SDS–PAGE and transferred to a membrane. In spliced A3b isoform was verified as the major A3 form in murine
Fig. 4B, the presence of the recombinant GST-A3aux (the approxi- (mouse/rat) cells but was practically undetected in human cells.
mately 42 kDa intact form as well as its fragmented forms) and GST In contrast, the unspliced A3a, which was the only isoform de-
proteins that were added in the pull-down assays can be observed tected in humans, was found to be a relatively minor hnRNP A3
by an initial Ponceau S staining of the membrane. Subsequent wes- isoform in mouse/rat. In addition, the examination of a number
tern blotting using antibodies that targeted selected hnRNPs was of mouse tissues (spleen, lung, liver, kidney, and brain cerebellum)
used to test their ability to co-precipitate with the recombinant revealed the expected enrichment of A3b relative to A3a. However,
proteins. In the low stringency wash (W/150), the western blot re- an important finding was that mouse whole brain lysate [the main
vealed the specific association of GST-A3aux with the endogenous brain tissue (cerebrum)] exhibited higher protein levels of A3a
nuclear hnRNP A/B (A3a, A1, A2) as well as with the L and M pro- than A3b, indicative of a tissue-specific variation of hnRNP A3
teins. The same hnRNP proteins were also identified in the cyto- expression. Our findings support and explain the published results
plasmic extract with the exception of A2, which was below on the identification of hnRNP A3 isoforms based on the use of
detection levels (see Fig. 3A). In both cellular compartments, the neuronal cells of human and murine origin [18]. Furthermore, in
association of the hnRNP A/B proteins was sensitive to the higher a recent study [27], a detailed examination of the species-specific
stringency wash (W/300), in clear opposition to hnRNP L and M. variation in the splicing patterns and the protein expression of
We note here that, similar to the auxiliary domain of hnRNP A1 hnRNP A/B (A1, A2/B1 and A3) in brain extracts of human and
158 C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160

D Nuclear Ex. Cytoplasmic Ex.


A

GST-A3aux

GST-A3aux
RNA binding domain Aux. domain

Ex. input
Ex. input
RRM RRM

GST

GST
N- -C
RNase: - + - + - + - +
A3aux
hnRNP A3a

B hnRNP A1
Nuclear Ex. Cytoplasmic Ex.
GST-A3aux
GST-A3aux

hnRNP A2/B1
GST
GST

hnRNP L
Rec. input
Rec. input

Ex. input

Ex. input
W/150
W/300

W/150
W/300
W/300
W/150

W/150
W/300

kDa hnRNP M
64
Ponseau S

49
GST-A3aux
C E

GST + Rec. M
37 HeLa nuclear ex.

GST-A3aux
GST Rec. M
GST-A3aux + Rec. M - RNase + RNase

W/150

W/300
26

GST-A3aux

GST-A3aux
GST

Ex. input
hnRNP A3a kDa

GST

GST
115
hnRNP A1 82 hnRNP M

β-actin 64 M fr.
WB

hnRNP A3a
49
hnRNP A2/B1
37
hnRNP L
hnRNP A1
hnRNP M 26
WB: anti-hnRNP M
hnRNP L

Fig. 4. Interaction of the auxiliary domain of hnRNP A3 fused to GST (GST-A3aux) with endogenous hnRNP proteins in vitro. (A) Schematic diagram of the RNA binding and
auxiliary (aux) domains common to the hnRNP A/B-type proteins. (B) The nuclear and cytoplasmic extracts (corresponding to 1 and 2 mg of total protein, respectively) were
isolated from human (A549) cells and allowed to incubate with glutathione beads that had been previously bound to recombinant GST-A3aux or GST alone in an in vitro pull-
down assay. The beads were washed in a buffer of 150 or 300 mM NaCl (W/150 or W/300). The bound proteins were resolved by SDS–PAGE and transferred to nitrocellulose.
The Ponceau S staining of the entire blot is shown. The lanes labeled as Rec. input represent an aliquot of the recombinant GST-A3aux or GST proteins, and the Ex. input lanes
represent an aliquot of the extract as applied in the pull-down assays. The membrane was then western blotted to detect the presence of hnRNP A3a, A1, A2/B1, L and M as
well as b-actin using the corresponding antibodies. (C) Pull-down assays were performed by incubating the GST-untagged full-length hnRNP M with glutathione beads that
had bound to GST-A3aux or GST alone in the absence of any extract. The beads were then washed under low (W/150) or high (W/300) salt conditions. The lane labeled as Rec.
M represents an aliquot of the recombinant hnRNP M. Western blotting was performed with a monoclonal antibody specific to hnRNP M. The presence of some fragmented
recombinant M is indicated (M fr.). (D) Pull-down assays similar to those presented in panel B were performed on A549 nuclear and cytoplasmic extracts with (+) or without
() prior incubation with RNase A. The proteins eluted from the beads (washed in low salt buffer) and an aliquot of the extract (Ex. Input) were immunoblotted with
antibodies against the hnRNP proteins A3a, A1, A2/B1, L and M. (E) Presentation of a pull-down assay (as in D) performed on a HeLa nuclear extract after western blotting with
antibodies against hnRNP A3a, A1 and L.

murine origin has been reported. An interesting observation made present work]. At this stage, only some suggestions can be made
in this report is that, in rat whole brain lysate, A3a is the predom- about a presumed requirement for increased hnRNP A3a levels in
inantly expressed isoform, despite the fact that the A3a transcript rat/mouse brain function as well as in humans. hnRNP A/B pro-
is the minor alternatively spliced form [27]. In this context, our teins, with their modular 2xRBD-Gly structure, participate in a
present analysis employing a number of mouse tissues revealed range of nucleic acid (RNA and ssDNA) and protein–protein inter-
the tissue-specific protein expression of the hnRNP A3a and A3b actions that provide for their broad range of nuclear and cytoplas-
isoforms (shown in Fig. 2C), which deserves special attention. Thus, mic functions in mRNA maturation [8,9]. The presence or absence
in addition to species-specific variations, we also uncovered the of an alternate exon in an A/B isoform is expected to modify its par-
novel tissue-specific expression of hnRNP A3 isoforms in murine ticipation in the above interactions. In the Introduction, we noted
cells. the ability of hnRNP A2/B1 (in particular, its minor A2b isoform)
The fact that brain (cerebrum) is, thus far, the only murine and A3, but not A1, to act as mRNA trafficking trans-acting factors
tissue found to overexpress the unspliced hnRNP A3a variant is ex- in both oligodendrocytes and neurons [18,19]. In addition, it has
pected to be of biological significance and requires further investi- been proposed that the inclusion of the 22-nucleotide-long N-ter-
gation. This finding should be considered along with the observed minal exon 1b that is next to the start codon in hnRNP A3a is ex-
lack of hnRNP A3b in human cells as well as the fact that such ma- pected to affect translational efficiency [27]. In the case of the rat
jor species-specific and tissue-specific variations have not been re- brain, the application of affinity pull-down assays using A2RE
ported for the other hnRNP A/B (A1 and A2/B1) members [27], RNA and telomere-specific DNA sequences resulted in the
C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160 159

over-representation of A3b as the selected variant, which was in hnRNP A3 to the assembly and stability of hnRNP complexes in
contrast to the overall enrichment of A3a over A3b in the brain ex- mammals. We also wish to refer in this context to our recent study
tract [27]. It may be inferred from the above findings that the unsp- employing (similar to human) pull-down assays on the extracts of
liced hnRNP A3a isoform is not likely to play a major role in the mouse embryonic fibroblast cells that revealed the ability of GST-
trafficking and localization of specific brain mRNAs or in telomere A3aux to make an RNA-independent interaction with HuR [30].
formation. This isoform might, rather, have functions related to the This finding extended the repertoire of feasible hnRNP A3 interac-
species- and tissue-specific requirements for nuclear events tions and is of relevance to the current work because HuR is an-
(mainly splicing) as well as in the stability and translation of other RNA-binding protein known to function in the stability/
mRNAs in the cytoplasm. translation of mRNAs containing 30 -UTR ARE elements [33 for Refs.
The difference in the overall amounts of hnRNP A3 between hu- therein]. HuR has also been reported to extensively associate with
man and murine cells should also be of functional significance. As the hnRNP proteins in immunoselected hnRNP and mRNP com-
clearly presented in our studies [23,26 and present work], A3b was plexes [30].
a major A/B type protein in murine cells, almost as abundant as The broad participation of hnRNP A3 in many cellular functions
hnRNP A2/B1, but it was undetected in human cells. We believe either by itself or in conjunction with the other A/B type proteins is
that the apparent functional redundancy of the hnRNP A/B proteins currently supported by a number of recent studies. Of special inter-
A1, A2 and A3 [18,28] should compensate for this great imbalance est is its contribution to age-related gene expression based on the
in the cellular concentration of hnRNP A3, an issue worth consider- identification of hnRNP A3 as the major mouse liver protein that
ing in future studies. binds to age-related elements in the 30 UTR of both the human
In our work, we have not examined the anticipated involvement and mouse factor IX (FIX) mRNA and regulates their expression
of hnRNP A3 in protein–RNA interactions that should be mainly [38]. The increase of the nuclear hnRNP A3 protein level (but not
based on its N-terminal RNA-binding domain. We have only indi- its mRNA) with age and the dependence on its phosphorylation
rectly addressed this issue in our IP studies by the detection of status has also been reported [38]. hnRNP A3 (together with the
the RNA-labile association of hnRNP A3 with the immunoselected D-like and K) is among the small group of growth hormone-regu-
hnRNP A1-RNP and mRNP complexes from both human and mouse lated proteins in rat liver, thus affecting sexual dimorphism [39].
cell extracts (Fig. 3), which indicated its direct or indirect binding Additionally, the anti-proliferative action of the glucocorticoid hor-
to precursor and mature mRNA. The bulk of hnRNP A3 was found mones in rat glioma cells has been shown to be related to the
in the nucleus with a minor fraction detected in the cytoplasmic abundance of a few proteins, including hnRNP A3 [40]. The latter
compartment, suggesting participation in the cytoplasmic as well is an apparent contradiction to the increase of hnRNP A3 protein
as in the nuclear events of mRNA metabolism. levels, in concert with those of A1 and A2/B1, in human lung cancer
A limited number of studies is available concerning the protein– biopsies that we reported [41], indicative of the operation of com-
protein interactions of hnRNP A3, in contrast to the other major plex regulatory mechanisms involving hnRNP A3.
hnRNP A/B proteins, A1 and A2, which have been studied both In summary, our findings on the novel characteristics of hnRNP
in vivo and in vitro [36,37]. The existence of RNA-independent A3, along with its recently reported broad participation in mRNA
homomeric and heteromeric interactions between hnRNP A1 and maturation and many other cellular processes, underscore the role
A2 that are dependent on the presence of the Gly-rich auxiliary do- of this protein as an important constituent of hnRNP/mRNP com-
main of the protein has been clearly established. We presented plexes in mammals.
here for the first time in vitro studies that investigate the contact
of the Gly-rich auxiliary domain of hnRNP A3 with the other com- Acknowledgments
ponents of the hnRNP and mRNP particles within the nuclear and
cytoplasmic extracts of human (A549, HeLa) cells. Based on the This work was supported by a research Grant (PENED) co-fi-
findings shown in Fig. 4, the auxiliary domain of hnRNP A3, as in nanced by E.U.-European Social Fund (75%) and the Greek Ministry
the case of hnRNP A1 and A2, could also form homo-complexes of Development-GSRT (25%), to A.G.
as well as hetero-complexes with hnRNP A1 and A2. However, in We thank Dr. Piergiorgio Percipalle (Dept. of Cell and Molecular
clear opposition to the findings with hnRNP A1 and C [37], GST- Biology, Karolinska Institute, Stockholm) for providing the anti-
A3aux could also associate with other than A/B-type proteins, as CBF-A rabbit antibody.
observed by its stable and RNA-independent association with
hnRNP M and L. It is interesting to note the persistence of this nu-
References
clear association in the cytoplasmic fraction (Fig. 4). As the forma-
tion of homo- and hetero-complexes (with the exception of A1) [1] G. Dreyfuss, J.M. Matunis, S. Pinol-Roma, G.C. Burd, Annu. Rev. Biochem. 62
was greatly enhanced in the RNase-treated extracts, compared to (1993) 289–321.
the untreated extracts and did not require the presence of the [2] A.M. Krecic, M.S. Swanson, Curr. Opin. Cell Biol. 11 (1999) 363–371.
[3] C. Maris, C. Dominguez, F.H.T. Allain, FEBS J. 272 (2005) 2118–2131.
RNA-binding domain of the recombinant A3 protein, these com- [4] S. Pinol-Roma, G. Dreyfuss, Nature 355 (1992) 730–732.
plexes are considered to require mainly protein–protein interac- [5] E.L.P. Kneller, J.H. Connor, D.S. Lyles, J. Virol. 83 (2009) 770–780.
tions. Although RNase digestion largely removed all detectable [6] T. Glisovic, J.I. Backorik, J. Yong, G. Dreyfuss, FEBS Lett. 582 (2008) 1977–1986.
[7] S.P. Han, Y.H. Tang, R. Smith, Biochem. J. 430 (2010) 379–392.
RNA from the extract (data not shown), we cannot exclude the pos- [8] G. Biamonti, S. Riva, FEBS Lett. 340 (1994) 1–8.
sibility that a minor fraction of protected oligoribonucleotides re- [9] Y. He, R. Smith, Cell. Mol. Life Sci. 66 (2009) 1239–1256.
mained bound to the associated proteins and contributed to this [10] A.A. Akindahunsi, A. Bandiera, G. Manzini, Comput. Biol. Chem. 29 (2005) 13–
23.
interaction. [11] J.L.E. Dean, G. Sully, R. Wait, L. Rawlinson, A.R. Clark, J. Saklatvala, Biochem. J.
The ability of the auxiliary domain of hnRNP A3 to make stable 366 (2002) 709–719.
associations with hnRNP M and L represents a feature that is thus [12] P. Percipalle, C.S. Raju, N. Fukuda, RNA Biol. 6 (2009) 171–174.
[13] L. Lothstein, H.P. Arenstorf, S.Y. Chung, B.M. Walker, J.C. Wooley, W.M.
far unique to an A/B type protein and suggests the involvement of
LeStourgeon, J. Cell Biol. 100 (1985) 1570–1581.
hnRNP A3 in an extensive network of protein–protein interactions [14] L.R. Fried, S.P. Han, J.A. Rothnagel, R. Smith, Biochim. Biophys. Acta 1783
within the hnRNP and, likely, mRNP complexes. As this network (2008) 1972–1980.
could facilitate the bridging within the hnRNP particles of the core [15] S. Rousseau, N. Morrice, M. Peggie, D.G. Campbell, M. Gaestel, P. Cohen, EMBO
J. 21 (2002) 6505–6514.
proteins (A/B and C) with the less stably associated hnRNP proteins [16] S.J. Cok, S.J. Acton, A.E. Sexton, A.R. Morrison, J. Biol. Chem. 279 (2004) 8196–
(hnRNP D–U), a possibility worth considering is the contribution of 8205.
160 C. Papadopoulou et al. / Archives of Biochemistry and Biophysics 523 (2012) 151–160

[17] T.P. Munro, R.J. Magee, G.J. Kidd, J.H. Carson, L. Barbarese, L.M. Smith, R. Smith, [29] K.D. Katsanakis, V. Gorgoulis, A.G. Papavassiliou, V.K. Zoumpourlis, Mol. Med. 8
J. Biol. Chem. 274 (1999) 34389–34395. (2002) 624–637.
[18] A.S.W. Ma, K. Moran-Jones, J. Shan, T.P. Munro, M.J. Snee, K.S. Hoek, R. Smith, J. [30] C. Papadopoulou, M. Patrinou-Georgoula, A. Guialis, Biochim. Biophys. Acta
Biol. Chem. 277 (2002) 18010–18020. 1804 (2010) 692–703.
[19] C.S. Raju, C. Goritz, Y. Nord, O. Hermanson, C. Lopez-Iglesias, N. Visa, G. [31] S. Pinol-Roma, M.S. Swanson, M.J. Matunis, G. Dreyfuss, in: J.E. Dahlberg, J.N.
Castelo-Branco, P. Percipalle, Mol. Biol. Cell 19 (2008) 3008–3019. Abelson (Eds.), Methods in Enzymology (RNA Processing, Part B, Specific
[20] L.P. Ford, W.E. Wright, J.W. Shay, Oncogene 21 (2002) 580–583. Methods), vol. 181, Academic Press, 1990, pp. 326–331.
[21] E. Tanaka, H. Fukuda, K. Nakashima, N. Tsuchiya, H. Seimiya, H. Nakagama, [32] V. Aidinis, C.E. Sekeris, A. Guialis, Nucleic Acids Res. 23 (1995) 2742–2753.
Biochem. Biophys. Res. Commun. 358 (2007) 608–614. [33] J.-Y. Lu, R.J. Schneider, J. Biol. Chem. 279 (2004) 12974–12979.
[22] P.R. Huang, S.C. Hung, T.C. Wang, Biochim. Biophys. Acta 1803 (2010) 1164– [34] M. Marko, M. Leichter, M. Patrinou-Georgoula, A. Guialis, Exp. Cell. Res. 316
1174. (2010) 390–400.
[23] A. Dangli, A. Plomaritoglou, E. Boutou, N. Vassiliadou, H.M. Moutsopoulos, A. [35] S. Nakielny, G. Dreyduss, J. Cell Biol. 134 (1996) 1365–1373.
Guialis, Biochem. J. 320 (1996) 761–767. [36] L. Cartegni, M. Maconi, E. Morandi, F. Cobianchi, S. Riva, G. Biamondi, J. Mol.
[24] A. Plomaritoglou, T. Choli-Papadopoulou, A. Guialis, Biochem. Biophys. Acta Biol. 259 (1966) 337–348.
1490 (2000) 54–62. [37] J.H. Kim, B. Hahm, Y.K. Kim, M. Choi, S.K. Jang, J. Mol. Biol. 298 (2000) 395–405.
[25] P. Percipalle, A. Jonsson, D. Nashchekin, C. Karlsson, T. Bergman, A. Guialis, B. [38] T. Hamada, S. Kurachi, K. Kurachi, PLoS One 5 (2010) 12971–12982.
Daneholt, Nucleic Acids Res. 30 (2002) 1725–1734. [39] E.V. Laz, C.A. Wiwi, D.J. Waxman, Mol. Cell. Proteomics 3 (2004) 1170–1180.
[26] S. Siapka, M. Patrinou-Georgoula, P.G. Vlachoyiannopoulos, A. Guialis, [40] M.A. Demasi, W.R. Montor, G.B. Ferreira, D.C. Pimenta, L. Labriola, M.C.
Autoimmunity 40 (2007) 223–233. Sogayar, J. Steroid Biochem, Mol. Biol. 103 (2007) 137–148.
[27] S.P. Han, K.S. Kassahn, A. Skarshewski, M.A. Ragan, J.A. Rothnagel, R. Smith, [41] G. Boukakis, M. Patrinou-Georgoula, M. Lekarakou, C. Valavanis, A. Guialis,
RNA 16 (2010) 1760–1768. BMC Cancer 10 (2010) 434–446.
[28] Y. He, M.A. Brown, J.A. Rothnagel, N.A. Saunders, R. Smith, J. Cell Sci. 118 (2005)
3173–3183.

You might also like