Relative amounts of antagonistic splicing factors, hnRNP A1 and ASF/SF2, change during neoplastic lung growth: implications for pre-mRNA processing.

Abstract

Pre-mRNA processing is an important mechanism for globally modifying cellular protein composition during tumorigenesis. To understand this process during lung cancer, expression of two key pre-mRNA alternative splicing factors was compared in a mouse model of early lung carcinogenesis and during regenerative growth following reversible lung injury. Heterogeneous nuclear ribonucleoprotein (hnRNP) A1 and alternative splicing factor/splicing factor 2 (ASF/SF2) act antagonistically to modulate splice site selection. Both hnRNP A1 and ASF/SF2 contents rose in adenomas and during injury-induced hyperplasia compared to control lungs, as measured by immunoblotting. While both proteins increased similarly during compensatory hyperplasia, hnRNP A1 increased to a much greater extent than ASF/SF2 in tumors, resulting in a 6-fold increase of the hnRNP A1 to ASF/SF2 ratio. Immunohistochemical analysis showed that hnRNP A1 localized exclusively within tumor nuclei, while ASF/SF2 appeared in cytoplasm and/or nuclei, depending on the growth pattern of the tumor cells. We also demonstrated cancer-associated changes in the pre-mRNA alternative splicing of CD44, a membrane glycoprotein involved in cell-cell and cell-extracellular matrix interactions. hnRNP A1 and ASF/SF2 expression is thus differentially altered in neoplastic lung cells by mechanisms that do not strictly arise from increased cell division. These changes are influenced by tumor histology and may be associated with production of variant CD44 mRNA isoforms.

Full text

Page 1

Relative Amounts of Antagonistic Splicing
Factors, hnRNP A1 and ASF/SF2, Change
During Neoplastic Lung Growth:
Implications for Pre-mRNA Processing
Laura K. Zerbe,1{Irene Pino,2{Ruben Pio,2Pippa F. Cosper,1Lori D. Dwyer-Nield,1Amy M. Meyer,3
J. David Port,3Luis M. Montuenga,2and Alvin M. Malkinson1*
1Department of Pharmaceutical Sciences, University of Colorado Health Sciences Center, Denver, Colorado
2Departments of Histology, Pathology, Biochemistry and Center of Applied Medical Research,
University of Navarra, Pamplona, Spain
3Department of Pharmacology, University of Colorado Health Sciences Center, Denver, Colorado
Pre-mRNA processing is an important mechanism for globally modifying cellular protein composition during
tumorigenesis. To understand this process during lung cancer, expression of two key pre-mRNA alternative
splicing factors was compared in a mouse model of early lung carcinogenesis and during regenerative growth
following reversible lung injury. Heterogeneous nuclear ribonucleoprotein (hnRNP) A1 and alternative splicing factor/
splicing factor 2 (ASF/SF2) act antagonistically to modulate splice site selection. Both hnRNP A1 and ASF/SF2 contents
rose in adenomas and during injury-induced hyperplasia compared to control lungs, as measured by immunoblotting.
While both proteins increased similarly during compensatory hyperplasia, hnRNP A1 increased to a much greater
extent than ASF/SF2 in tumors, resulting in a 6-fold increase of the hnRNP A1 to ASF/SF2 ratio. Immunohistochemical
analysis showed that hnRNP A1 localized exclusively within tumor nuclei, while ASF/SF2 appeared in cytoplasm and/or
nuclei, depending on the growth pattern of the tumor cells. We also demonstrated cancer-associated changes in the
pre-mRNA alternative splicing of CD44, a membrane glycoprotein involved in cell-cell and cell-extracellular matrix
interactions. hnRNP A1 and ASF/SF2 expression is thus differentially altered in neoplastic lung cells by mechanisms that
do not strictly arise from increased cell division. These changes are influenced by tumor histology and may be
associated with production of variant CD44 mRNA isoforms.associated with production of variant CD44 mRNA isoforms. ? ? 2004 Wiley-Liss, Inc.
Pre-mRNA processing is an important mechanism for globally modifying cellular protein composition during
tumorigenesis. To understand this process during lung cancer, expression of two key pre-mRNA alternative
splicing factors was compared in a mouse model of early lung carcinogenesis and during regenerative growth
following reversible lung injury. Heterogeneous nuclear ribonucleoprotein (hnRNP) A1 and alternative splicing factor/
splicing factor 2 (ASF/SF2) act antagonistically to modulate splice site selection. Both hnRNP A1 and ASF/SF2 contents
rose in adenomas and during injury-induced hyperplasia compared to control lungs, as measured by immunoblotting.
While both proteins increased similarly during compensatory hyperplasia, hnRNP A1 increased to a much greater
extent than ASF/SF2 in tumors, resulting in a 6-fold increase of the hnRNP A1 to ASF/SF2 ratio. Immunohistochemical
analysis showed that hnRNP A1 localized exclusively within tumor nuclei, while ASF/SF2 appeared in cytoplasm and/or
nuclei, depending on the growth pattern of the tumor cells. We also demonstrated cancer-associated changes in the
pre-mRNA alternative splicing of CD44, a membrane glycoprotein involved in cell-cell and cell-extracellular matrix
interactions. hnRNP A1 and ASF/SF2 expression is thus differentially altered in neoplastic lung cells by mechanisms that
do not strictly arise from increased cell division. These changes are influenced by tumor histology and may be
2004 Wiley-Liss, Inc.
Key words: 3-methylcholanthrene; alternative splicing; butylated hydroxytoluene; CD44; lung cancerKey words: 3-methylcholanthrene; alternative splicing; butylated hydroxytoluene; CD44; lung cancer
INTRODUCTION INTRODUCTION
Lungcancer,theprimarycauseofcancerdeathsin
the US and Europe [1], is seldom diagnosed prior to
metastatic spread. It is, therefore, critical to under-
stand the mechanisms that guide early stages of
tumor development in order to identify biomarkers
and develop improved molecular-targeted chemo-
preventiveandchemotherapeuticstrategies.Thereis
growing evidence that modulation of gene expres-
sion at the pre-mRNA level is an important regula-
tory mechanism in tumorigenesis [2]. Pre-mRNA
processing proteins mediate many aspects of prepar-
ing a final mRNA transcript for polysomal transla-
tion, including alternative splicing, stability, and
transport of pre-mRNA [3,4], and thus can influence
the translation into proteins of numerous mRNAs,
including those that facilitate carcinogenesis. Se-
quencing of the human genome has emphasized the
importance of pre-mRNA processing, particularly
alternativesplicing,innormalaswellaspathological
tissues. The discovery of far fewer genes than anti-
cipated based on mammalian proteome complexity
indicates that alternative splicing and posttransla-
tional modifications are critical for protein diversity
[5]. In fact, it is estimated that 40–60% of genes are
alternativelysplicedinto twoormoretranscripts[6].
Computational analysis of mRNAs derived from
tumorornormalsamplesidentified845alternatively
splicedisoformssignificantlyassociatedwithhuman
cancer [7].
At least 20 hnRNP proteins in human cells form a
diverse set of large, complexes with heterogeneous
MOLECULAR CARCINOGENESIS 41:187–196 (2004)
? 2004 WILEY-LISS, INC.
Abbreviations: AC, adenocarcinoma; ASF/SF2, alternative splicing
factor/splicing factor 2; BHT, butylated hydroxytoluene; hnRNP,
heterogeneous nuclear ribonucleoprotein; IHC, immunohistochem-
istry; MCA, 3-methylcholanthrene; SR, serine/arginine.
{Laura K. Zerbe and Irene Pino contributed equally to this study.
*Correspondence to: Department of Pharmaceutical Sciences,
University of Colorado Health Sciences Center, Box C238, 4200 East
Ninth Ave., Denver, CO 80262.
Received 8 April 2004; Revised 1 July 2004; Accepted 20 July
2004
DOI 10.1002/mc.20053

Page 2

nuclear RNA during pre-mRNA processing [3].
The pre-mRNA processing protein, heterogeneous
nuclear ribonucleoprotein (hnRNP) A2/B1, has been
proposed as a promising biomarker for lung cancer.
In prospective studies, hnRNP A2/B1 expression in
epithelial cells exfoliated into human sputum pre-
dicted lung cancer in approximately 70% of the
subjects [8]. We demonstrated previously by immu-
nohistochemistry (IHC) that several members of
this family (hnRNP A1, A2, B1, C1, C2, and K) are
increased in tumors compared to normal tissue from
lungcancer patientbiopsies[9].Wehavealsoshown
that several pre-mRNA processing proteins, hnRNP
A2/B1 [10], hnRNP D, and Hu antigen R (HuR) [11],
increase in mouse lung tumors and related neo-
plastic cell lines, which are preclinical models of
pulmonary adenocarcinoma [12].
Serine/arginine (SR)-rich proteins, another impor-
tant class of pre-mRNA processing proteins involved
inconstitutiveandalternativesplicing,interactwith
the hnRNP proteins. One member of this family,
alternative splicing factor/splicing factor 2 (ASF/
SF2), acts antagonistically to hnRNP A1 to regulate
alternative splicing; increased concentrations of
ASF/SF2 select proximal 50sites, while high hnRNP
A1 levels promote distal 50site selection [13,14].
Several SR proteins, including ASF/SF2, increase
during mouse mammary gland tumorigenesis [15].
Because of the increased expression patterns of
hnRNP proteins in human lung cancer and the
known importance of interactions between hnRNP
A1andASF/SF2inregulatingalternativesplicing,we
evaluated expression of these two proteins in mouse
models of lung AC and lung regenerative repair. Our
results indicate that both hnRNP A1 and ASF/SF2
increase in benign tumors and in nonneoplastic
proliferating lung compared to normal lung tissue.
In tumors, however, hnRNP A1 increased substan-
tially more than ASF/SF2, while both proteins
increased similarly during regenerative hyperplasia.
To demonstrate that alternative splicing indeed
occurs in this lung cancer model, we evaluated the
mRNA isoform pattern of CD44, a gene whose
alternative splicing is associated with many cancers
[16–18]. Interestingly, the increased ratio of hnRNP
A1relativetoASF/SF2intumorscorrespondedtothe
presence of metastasis-associated CD44 splice var-
iants relative to control lung, while no changes
in CD44 splicing were observed in nonneoplastic
proliferating cells.
MATERIALS AND METHODSMATERIALS AND METHODS
Mice and Treatments Mice and Treatments
Male BALB/cByJ mice were obtained from Jackson
Laboratories (Bar Harbor, ME) at 5–7 weeks of age,
andallowedtoacclimatefor10–14dayspriortotheir
use in experiments. The mice were given access
to Teklad-8640 standard laboratory chow (Harlan
Teklad, Madison, WI) and water ad libitum, and
maintained on a hardwood bedding under a 12 h
light/12 h dark cycle. All injections were given
prior to 10 AM because of circadian effects on BHT
pneumotoxicity (Malkinson, A.M., unpublished
results). Mice were sacrificed by lethal injection of
100 ml 90 mg/ml sodium pentobarbital containing
1000 U/ml heparin (Sigma, St. Louis, MO).
MCA/BHT 2-Stage Lung Carcinogenesis in BALB MiceMCA/BHT 2-Stage Lung Carcinogenesis in BALB Mice
Mice were injected IP once with 15 mg/kg body wt
MCA (Sigma) dissolved in MazolaTMcorn oil (ACH
Food Companies, Memphis, TN), followed by 6
weekly injections of BHT (Sigma; first injection,
150 mg/kg body wt; the second through sixth
injections, 200 mg/kg body wt) dissolved in corn
oil, as described [19]. Mice were sacrificed 26 wks
after MCA injection.
BHT-Induced Acute Lung Injury in BALB MiceBHT-Induced Acute Lung Injury in BALB Mice
Mice were injected IP once with 200 mg/kg body
wt BHT as described [20], and sacrificed 1 h to 6 days
after BHT administration.
Preparation of Extracts for ImmunoblotttingPreparation of Extracts for Immunoblottting
Dissected lungs from untreated or BHT-treated
mice,normal-appearinguninvolvedtissueintumor-
bearing lungs, and tumors were Dounce-homoge-
nized in 100–500 ml solubilization buffer (20 mM
HEPES, pH¼7.5, 2 mM EDTA, 2 mM EGTA, 1 mM
DTT, 10% glycerol, 5 mg/ml aprotinin, and 10 mM
leupeptin), and centrifuged at 16000?g for 25 min
at 48C. After the supernatant (soluble fraction) was
removed, the pellet (particulate fraction) was resus-
pended in solubilization buffer containing 0.1%
NP40, sonicated and an aliquot (10–20 ml) was
removed to determine protein concentration (based
on the method of Lowry et al. [21]). FourX gel-
loading buffer was added to the remaining sample at
a4-fold dilution(finalconcentrations: 10%glycerol,
1% SDS, 50 mM Tris, pH¼6.8, 4% b-mercaptoetha-
nol, trace amount of bromophenol blue) and
incubated at 958C for 5 min. All samples were stored
at ?708C until immunoblot analysis.
ImmunoblottingImmunoblotting
Proteins (25–100 mg) were resolved on a 15% SDS-
polyacrylamide gel followed by transfer to an
Immobilon-P membrane (Millipore Corp., Billerica,
MA) and blocked in 3% nonfat dried milk dissolved
in TNS (15 mM Tris, pH¼7.4, 150 mM NaCl) for
30 min. The membranes was then incubated in
primary anti-hnRNP A1 monoclonal antibodies
(4B10 or 9H10, kindly supplied by Dr. G. Dreyfuss,
HHMI,UniversityofPennsylvania)diluted1:1000in
TNS or anti-ASF/SF2 goat polyclonal antibody P-15,
sc-10254 (Santa Cruz Biotechnology, Santa Cruz,
CA) diluted 1:100 in TNS for 1 h and washed in TNS
188188
ZERBE ET AL.ZERBE ET AL.

Page 3

with 3% milk and 0.1% Tween 20. For hnRNP A1,
a secondary goat, antimouse antibody (Bio-Rad,
Hercules, CA) was diluted 1:5000 in TNS, while for
ASF/SF2, a rabbit antigoat antibody (A-5420, Sigma)
was diluted 1:25000 in TNS and incubated for 1 h
followed by washing in TNS with 3% milk and 0.1%
Tween 20. Protein bands were visualized by chemi-
luminiscence after exposure to CL-Xposure film
(Pierce, Rockford, IL) with Immun-StarTMAP sub-
strate (Bio-Rad) for hnRNP A1 and Western Lighting
(PE Applied Biosystems, Foster, CA) for ASF/SF2.
Immunoreactive proteins on Western blots were
quantified by UN-SCAN-ITTM(Silk Scientific, Orem,
UT). Ponceau-staining (0.1% ponceau in 5% acetic
acid) of proteins transferred to the membrane
confirmed equal protein loading.
Lung Fixation Lung Fixation
Immediately after the mice were sacrificed, the
lungs were gently inflated through a canulated
trachea with formalin for 1 h. Lungs were then
removed from the chest cavity and individual lobes
dissected and submerged in formalin overnight.
Fixed lungs were transferred to 95% ethanol and
embedded in paraffin.
ImmunohistochemistryImmunohistochemistry
Tissue sections (3 mm) were deparaffinized and
endogenous peroxidase activity blocked by treat-
ment with 3% hydrogen peroxide. For hnRNP A1,
after pretreatmentwith
Blocker Kit (Zymed Laboratories, Inc., SanFrancisco,
CA), 4B10 primary antibody was used at a 1:1000
dilution overnight at 48C. Slides were rinsed and
incubated with the ENVISIONTMreagent for mouse
monoclonal antibodies (Dakocytomation, Glostrup,
Denmark) for 30 min at room temperature. For ASF/
SF2immunostaining,slideswereimmersedin10mM
citrate buffer, pH 6, and microwaved at two 15-min
cyclesto retrieveantigenbefore blocking withrabbit
normal serum (1:20; Dakocytomation). The sections
were incubated with a primary goat polyclonal
antibodyP-10254 (Santa
diluted 1:100 at 48C overnight. Addition of an
antigoat rabbit secondary biotinylated antibody
(1:300; Dakocytomation) was followed by ABC
complexincubation(1:100;
Immunostaining was developed with diaminoben-
zidine (Dakocytomation) and counter-stained with
hematoxylin.
the Histomouse-SPTM
CruzBiotechnology)
Dakocytomation).
RT-PCRRT-PCR
Freshly prepared dissected tumor, uninvolved
lung from tumor-bearing mice or control lung tissue
was submerged in 0.5–1 mL ice-cold RNAlater
(Ambion,Austin,TX) andstoredat?208until future
analysis. RNA was subsequently isolated with either
TRIzol reagent (Invitrogen, Carlsbad, CA) or RNeasy
mini kit (Qiagen, Valencia, CA). RNA was reverse-
transcribed to DNA with the SuperScriptTMfirst-
strand synthesis system for RT-PCR (Invitrogen).
CD44 isoforms were amplified with Platinum1Taq
DNA polymerase (Invitrogen) and primers 50-
ACCCCAGAAGGCTACATTTTGC-30(forward)
50-CTCATAGGACCAGAAGTTGTGG-30(reverse) as
previously described [15] at 35 rounds of PCR
consisting of 45 s 948C, 45 s 608C, and 2 min 728C.
PCR products were separated on a 2% agarose gel,
visualized with UV light and quantified by UN-
SCAN-ITTM(Silk Scientific) software.
and
Statistical Analysis Statistical Analysis
Data were represented in graphs as mean?SEM
and analyzed byStudent’s unpaired t-test withPrism
3.0 software (GraphPad, San Diego, CA) In all
analyses, significance was accepted at P<0.05.
RESULTSRESULTS
hnRNP A1 and ASF/SF2 Protein Expression
During 2-Stage CarcinogenesisDuring 2-Stage Carcinogenesis
hnRNP A1 and ASF/SF2 Protein Expression
2-stage carcinogenesis consists of initiation by the
tobaccocarcinogen,MCA,followedbypromotionof
the growth of initiated cells by BHT. BHT undergoes
extensive pulmonary metabolism in mice to gener-
ate highly reactive oxidative species that cause a
reversible pneumotoxicity consisting of peripheral
epithelial cell damage, regenerative growth, and
inflammation; these properties promote tumor
growth in some mouse strains including BALB when
BHT is administered after a carcinogen [19,20,22–
25]. This experimental model affords the opportu-
nity to compare hnRNP A1 and ASF/SF2 expression
during early neoplasia after 2-stage carcinogenesis
(MCAþBHT) with the nonneoplastic proliferation
that follows BHT treatment alone.
hnRNP A1 and ASF/SF2 protein contents were
measured by immunoblot analysis of dissected
tumors and uninvolved lungs (i.e., normal-appear-
ing tissue adjacent to tumors), isolated from mice
treated with MCAþBHT or whole lungs from age-
matched control mice (Fig. 1). hnRNP A1 levels were
elevated substantially in tumor versus control parti-
culate samples (20300?2260 and 700?230 densi-
tometric values, respectively) resulting in a 30-fold
increase. ASF/SF2 levels also increased in tumor
versus control samples (2100?146 and 400?49
densitometric values, respectively), but to a lesser
extent than hnRNP A1. This differential increase in
hnRNP A1 compared to ASF/SF2 resulted in nearly a
6-fold elevation of the hnRNP A1 to ASF/SF2 ratio.
Soluble fractions were also analyzed, but negligible
alterations in hnRNP A1 or ASF/SF2 contents were
observedintumorrelativetocontrollungtissue(0.9-
fold to 1.4-fold increases, data not shown). Both
hnRNP A1 and ASF/SF2 also rose in uninvolved
compared to control tissue (3-fold and 2-fold,
respectively, Fig. 1).
hnRNP A1 AND ASF/SF2 IN MOUSE LUNG TUMORShnRNP A1 AND ASF/SF2 IN MOUSE LUNG TUMORS
189189

Page 4

hnRNP A1 and ASF/SF2 Cellular and Intracellular
Localization During 2-Stage CarcinogenesisLocalization During 2-Stage Carcinogenesis
hnRNP A1 and ASF/SF2 Cellular and Intracellular
TodeterminethecellularlocalizationofhnRNPA1
and ASF/SF2, we performed IHC on lung tissue
sections obtained from BALB mice treated with
MCAþBHT (Fig. 2). hnRNP A1 was more highly
expressedintumorcellsthaninadjacentuninvolved
alveolar and bronchiolar cells (Fig. 2A), while ASF/
SF2 levels were moderately higher in tumors com-
pared to surrounding pulmonary tissue (Fig. 2B),
similar to immunoblotting experiments (Fig. 1).
Mouse lung adenomas grow in either solid or
papillary patterns; papillary tumors are more aggres-
sive with a greater tendency to progress to malig-
nancy [26]. In both solid and papillary tumors,
hnRNP A1 localized to tumor nuclei (Fig. 2C and
E). In contrast, ASF/SF2 localization varied between
the two pathologies; the solid phenotype contained
both nuclear and cytoplasmic ASF/SF2 (Fig. 2D),
whereas papillary tumors exhibited only cytoplas-
mic localization (Fig. 2F).
hnRNP A1 and ASF/SF2 Expression During
Nonneoplastic Pulmonary Growth Induced by BHT Nonneoplastic Pulmonary Growth Induced by BHT
hnRNP A1 and ASF/SF2 Expression During
The contents of hnRNP A1 and ASF/SF2 in tumor
cells may be upregulated consonant with the
increasedRNAmetabolismfoundinrapidlydividing
cells. To examine this, we measured hnRNP A1 and
ASF/SF2 levels during nonneoplastic proliferation.
BHT damage to alveolar type I cells is repaired by a
regenerative growth response of Clara cells and type
2 cells; the latter cell type then differentiates into
type 1 cells to repair the alveolar epithelium [25].
BALB mice were given a single dose of BHT in the
absence of MCA, and lung extracts assessed by
immunoblotting 6 days later in particulate fractions
(Fig.3).ThecontentsofhnRNPA1andASF/SF2both
Figure 1. Expression of hnRNP A1 and ASF/SF2 protein in mouse
lung tumorigenesis. Immunoblotting of hnRNP A1 (A) and ASF/SF2
(C) was conducted on particulate fractions of age-matched control
lungs (Con), uninvolved lung (UI; normal-appearing tissue in tumor-
bearinglungs) and tumors(Tum) isolatedfromBALB mice. Micewere
treated with 15 mg/kg body wt MCA, followed by 6 weekly
injections of BHT (first injection, 150 mg/kg body wt; injections 2–
66¼ ¼200 mg/kg body wt) or 7 corn oil injections and sacrificed after
Figure 1. Expression of hnRNP A1 and ASF/SF2 protein in mouse
lung tumorigenesis. Immunoblotting of hnRNP A1 (A) and ASF/SF2
(C) was conducted on particulate fractions of age-matched control
lungs (Con), uninvolved lung (UI; normal-appearing tissue in tumor-
bearinglungs) and tumors(Tum) isolatedfromBALB mice. Micewere
treated with 15 mg/kg body wt MCA, followed by 6 weekly
injections of BHT (first injection, 150 mg/kg body wt; injections 2–
200 mg/kg body wt) or 7 corn oil injections and sacrificed after
26 wk. The bands correspond to the expected 34 and 30 kDa
molecular weights of hnRNP A1 and ASF/SF2, respectively. hnRNP A1
(B) and ASF/SF2 (D) immunoreactive protein levels were quantified by
UN-SCAN-IT software. Data represent mean UN-SCAN-IT software. Data represent mean? ?SEM from four mice
per group. For both hnRNP A1 and ASF/SF2, ***per group. For both hnRNP A1 and ASF/SF2, ***P P? ?0.0001 tumor
vs. control samples. Ponceau-staining of proteins transferred to the
membrane confirmed equal protein loading.membrane confirmed equal protein loading.
26 wk. The bands correspond to the expected 34 and 30 kDa
molecular weights of hnRNP A1 and ASF/SF2, respectively. hnRNP A1
(B) and ASF/SF2 (D) immunoreactive protein levels were quantified by
SEM from four mice
0.0001 tumor
vs. control samples. Ponceau-staining of proteins transferred to the
Figure 2. Expression and localization of hnRNP A1 and ASF/SF2 in
mouse lung tumorigenesis and in nonneoplastic proliferating lung
cells.Immunohistochemistry(IHC)ofhnRNPA1insolidtumors(A,C),
papillary tumor (E), control lung (G), and BHT-treated lung (I). IHC of
ASF/SF2 in solid tumors (B, D) and papillary tumors (F), control lung
(H), and BHT-treated lung (J). Solid and papillary tumors were derived
from BALB mice treated with 15 mg/kg body wt MCA, followed by 6
weekly injections of BHT (first injection, 150 mg/kg body wt; weekly injections of BHT (first injection, 150 mg/kg body wt;
injections 2–6injections 2–6¼ ¼200 mg/kg body wt) and sacrificed 26 wk after
MCA administration. BHT-treated and control lung samples were
obtained from BALB mice injected with 200 mg/kg body wt BHT or
corn oil, respectively, and sacrificed after 6 days. hnRNP A1 or ASF/
SF2 positive cells are brown. Hematoxylin counterstain. Original
magnification for A, B, C, D, E, and F, 1250 magnification for A, B, C, D, E, and F, 1250? ? and for G, H, I, and J,
250250? ?. .
Figure 2. Expression and localization of hnRNP A1 and ASF/SF2 in
mouse lung tumorigenesis and in nonneoplastic proliferating lung
cells.Immunohistochemistry(IHC)ofhnRNPA1insolidtumors(A,C),
papillary tumor (E), control lung (G), and BHT-treated lung (I). IHC of
ASF/SF2 in solid tumors (B, D) and papillary tumors (F), control lung
(H), and BHT-treated lung (J). Solid and papillary tumors were derived
from BALB mice treated with 15 mg/kg body wt MCA, followed by 6
200 mg/kg body wt) and sacrificed 26 wk after
MCA administration. BHT-treated and control lung samples were
obtained from BALB mice injected with 200 mg/kg body wt BHT or
corn oil, respectively, and sacrificed after 6 days. hnRNP A1 or ASF/
SF2 positive cells are brown. Hematoxylin counterstain. Original
and for G, H, I, and J,
190 190
ZERBE ET AL.ZERBE ET AL.

Page 5

Figure 2.Figure 2.

Page 6

rose in BHT-treated lungs relative to control lungs to
a similar extent (5.7-fold and 3.2-fold, respectively).
While significant, these increases were smaller than
that for hnRNP A1 in tumors relative to control lung
(Fig. 1). Similar to neoplastic tissue, hnRNP A1 or
ASF/SF2 expression in soluble fractions of BHT-
treated versus control lung did not change (data
not shown). Because hnRNP A1 and ASF/SF2 each
increased to similar extents during repair of BHT-
induced lung injury, the differential increases of
hnRNP A1 and ASF/SF2 levels in neoplastic lung
tissueprobablydoesnotsimplyresultfromincreased
cell division.
hnRNP A1 and ASF/SF2 Localization During
Hyperplasia Induced by BHT Hyperplasia Induced by BHT
hnRNP A1 and ASF/SF2 Localization During
To determine how the cellular localization of
hnRNP A1 and ASF/SF2 in nonneoplastic proliferat-
ing cells compared with their expression in tumors,
IHC was performed on lungs isolated from BHT-
treated BALB mice. We assayed hnRNP A1 and ASF/
SF2 expression between 1 h and 6 days after BHT
exposure. In control mice, hnRNP A1 was not
detectable by IHC in alveolar type 2 or bronchiolar
Clara cells; ASF/SF2 was not detected in type 2 cells
but was expressed in Clara cells (Fig. 2G and H).
Moderate increases in alveolar hnRNP A1 and ASF/SF2
levelsoccurredearlyafterBHT-treatment;cytoplasmic
ASF/SF2wasdetectedinalveoliwithin1handnuclear
hnRNP A1 in alveoli by 24 h (data not shown). A few
daysafterBHT-treatment,correspondingtothetimeof
maximal cellular proliferation and repair [20,23,25],
highnuclearandcytosolichnRNPA1weredetectedin
alveolarandbronchiolarcells(Fig.2I).ASF/SF2content
also reached peak cytosolic and nuclear levels in
alveolar and bronchiolar cells by 6 days, but bronch-
iolarstainingwasmoreintense(Fig.2J).Ingeneral,the
intercellular localization of ASF/SF2 was similar in
nonneoplastic proliferating lung tissue and tumors,
displaying both nuclear and cytosolic expression. In
contrast, hnRNP A1 localization differed between the
two pathologies; expression in repairing lungs was
both cytoplasmic and nuclear while tumor expression
was primarily nuclear.
Alternative Splicing of CD44 Pre-mRNA in Lung
Tumors and BHT-Treated Lung—A Candidate
Substrate for hnRNP A1 and ASF/SF2Substrate for hnRNP A1 and ASF/SF2
Alternative Splicing of CD44 Pre-mRNA in Lung
Tumors and BHT-Treated Lung—A Candidate
The relative ratio of hnRNP A1 relative to ASF/SF2
alters the alternative splicing of several pre-mRNAs
Figure 3. Expression of hnRNP A1 and ASF/SF2 protein in
nonneoplastic proliferating lung cells. Immunoblotting of hnRNP
A1 (A) and ASF/SF2 (C) was conducted on particulate fractions of
lung tissue isolated from BALB mice treated with corn oil control
(Con) or 200 mg/kg body wt BHT (BHT). Mice were sacrificed 6 days
after BHT administration. The bands correspond to the expected
34 and 30 kDa molecular weights of hnRNP A1 and ASF/SF2,34 and 30 kDa molecular weights of hnRNP A1 and ASF/SF2,
respectively. hnRNP A1 (B) and ASF/SF2 (D) immunoreactive protein
levels were quantified by UN-SCAN-IT software. Data represent
meanmean? ?SEM from five mice per group. For hnRNP A1 **
and ASF/SF2 *** and ASF/SF2 ***P P< <0.0001 BHT-treated vs. control samples.
Ponceau staining of proteins transferred to the membrane confirmed
equal protein loading. equal protein loading.
Figure 3. Expression of hnRNP A1 and ASF/SF2 protein in
nonneoplastic proliferating lung cells. Immunoblotting of hnRNP
A1 (A) and ASF/SF2 (C) was conducted on particulate fractions of
lung tissue isolated from BALB mice treated with corn oil control
(Con) or 200 mg/kg body wt BHT (BHT). Mice were sacrificed 6 days
after BHT administration. The bands correspond to the expected
respectively. hnRNP A1 (B) and ASF/SF2 (D) immunoreactive protein
levels were quantified by UN-SCAN-IT software. Data represent
SEM from five mice per group. For hnRNP A1 **P P¼ ¼0.0027
0.0001 BHT-treated vs. control samples.
Ponceau staining of proteins transferred to the membrane confirmed
0.0027
192 192
ZERBE ET AL. ZERBE ET AL.

Page 7

in vitro [13,14]. Given the differential increases of
hnRNP A1 and ASF/SF2 in tumors relative to control
tissuereportedhere,itseemedprobablethatsplicing
of many pre-mRNAs would be affected, including
thoseinvolvedintumorigenesis.Tobegintotestthis
hypothesis, the mRNA expression pattern of CD44,
a gene for which alternative splicing has been im-
plicated in many types of cancer including lung
cancer [16,17], was compared in MCA-BHT lung
carcinogenesis, BHT-induced nonneoplastic prolif-
erating lung tissue, and control lung. CD44 is a cell
surface molecule that regulates cell-cell and cell-
extracellular matrix interactions during inflamma-
tion and cancer, and is composed of 20 exons [18].
Particularvariantsareassociatedwithmetastasis;the
CD44standard (CD44s) isoform results from splicing
S1-S5 and S6-S10 exons, while metastasis-associated
CD44variant (CD44v) isoforms include 10 variant
exons in various combinations (Fig. 4A). CD44
isoforms were measured by RT-PCR with primers
that bound S5 and S6 CD44s exons and flanked the
10 CD44 variant exons V1–V10 (Fig. 4A).
In MCAþBHT studies, the mRNA expression
pattern of CD44 switched from predominantly the
single 221 nucleotide long CD44s exon in control
and uninvolved tissues to a series of larger CD44
splice variants in tumors (Fig. 4B), similar to that
observed in other tumor types [15,16,18,27]. In fact,
58% of the total CD44 isoforms were variant in the
lungtumorextractscomparedtoonly14%incontrol
tissue(Fig.4D).ThefivemajorCD44variantisoforms
in tumors displayed a ladder pattern indicating
sizes ranging from approximately 492 to 1230 nuc-
leotides long (based on DNA ladder standards) that
Figure 4. Expression pattern of CD44 mRNA isoforms in mouse
lung pathology. The lung pathology. The CD44 CD44 gene structure showing the 10 variant
exons (V1–V10) flanked by standard exons S5 and S6 (A). RT-PCR
analysis of CD44 mRNA isoforms during 2-stage carcinogenesis (B)
and reparative hyperplasia (C) in BALB mice with primers specific for
standard exons S5 and S6. RNA was isolated from dissected tumors
(Tum), uninvolved lung (UI; normal-appearing tissue in tumor-
bearing lungs), and age-matched control lungs (Con) in tumorigen-
esis studies (B) and from BHT-treated mice (BHT) and age-matched
control lungs (Con) in reparative hyperplasia studies (C). Incontrol lungs (Con) in reparative hyperplasia studies (C). In
Figure 4. Expression pattern of CD44 mRNA isoforms in mouse
gene structure showing the 10 variant
exons (V1–V10) flanked by standard exons S5 and S6 (A). RT-PCR
analysis of CD44 mRNA isoforms during 2-stage carcinogenesis (B)
and reparative hyperplasia (C) in BALB mice with primers specific for
standard exons S5 and S6. RNA was isolated from dissected tumors
(Tum), uninvolved lung (UI; normal-appearing tissue in tumor-
bearing lungs), and age-matched control lungs (Con) in tumorigen-
esis studies (B) and from BHT-treated mice (BHT) and age-matched
tumorigenesis studies mice were treated with 15 mg/kg body wt
MCA, followed by 6 weekly injections of BHT (first injection, 150 mg/
kg body wt; injections 2–6 kg body wt; injections 2–6¼ ¼200 mg/kg body wt) or 7 corn oil
injections and sacrificed after 26 wk and in reparative hyperplasia
studies mice were treated with 200 mg/kg BHT or corn oil and
sacrificed after 6 days. CD44 mRNA isoforms were quantified by UN-
SCAN-IT software. Data represent CD44variant isoforms/CD44total
isoformsisoforms? ?100% and are the mean
group (D, E). For tumor versus control ***group (D, E). For tumor versus control ***P P? ?0.0001 (D).
tumorigenesis studies mice were treated with 15 mg/kg body wt
MCA, followed by 6 weekly injections of BHT (first injection, 150 mg/
200 mg/kg body wt) or 7 corn oil
injections and sacrificed after 26 wk and in reparative hyperplasia
studies mice were treated with 200 mg/kg BHT or corn oil and
sacrificed after 6 days. CD44 mRNA isoforms were quantified by UN-
SCAN-IT software. Data represent CD44variant isoforms/CD44total
100% and are the mean? ?SEM from 3 to 5 mice perSEM from 3 to 5 mice per
0.0001 (D).
hnRNP A1 AND ASF/SF2 IN MOUSE LUNG TUMORShnRNP A1 AND ASF/SF2 IN MOUSE LUNG TUMORS
193193

Page 8

differed in size by approximately 115 nucleotides,
which is comparable to previous reports [15,27]. In
contrast, nonneoplastic proliferating lung in BHT-
treated mice did not exhibit a pattern of CD44
variant isoforms; rather, the CD44s isoform predo-
minated, similar to control lung tissue (Fig. 4C and
E). These data showed that differential increases of
hnRNP A1 and ASF/SF2 in mouse lung tumors
correspond to an altered CD44 mRNA splice pattern,
although a causal relationship is not proved.
DISCUSSION DISCUSSION
Expressionofantagonistic splicingfactors,hnRNP
A1 and ASF/SF2, is substantially altered in early
mouse lung neoplasia. The contents of both factors
roseintumorscomparedtocontrollung,buthnRNP
A1increasedmorethanASF/SF2,resultingina6-fold
increase of the hnRNP A1 to ASF/SF2 ratio. Other
groups have shown in vitro that the relative levels of
hnRNP A1 compared to ASF/SF2 affects 50splice site
selection [13,14], these differential changes in vivo
probably modulate alternative splicing of many
genes, including those involved in carcinogenesis.
The increase in hnRNP A1 and ASF/SF2 is not
exclusively due to the high proliferative activity of
tumor tissues. Both factors rose in regenerating
mouse lung tissue, but did not exhibit an altered
hnRNPA1toASF/SF2ratio.Thisisinagreementwith
our previous results showing that mRNA expression
of several hnRNPs including hnRNP A1 is not
correlated strictly with proliferative index in a panel
of cultured human lung cell lines [9].
The aberrantexpression of hnRNPA1 and ASF/SF2
occurred in early-stage tumors of mouse lung
tumorigenesis (26 weeks after MCA injection) that
are typically small with uniform-looking cells.
Adenocarcinomas with increased disorganization
of tumor cells and invasion into the stroma are not
usually observed until 35 weeks after MCA treat-
ment. This early appearance of splice factors
provides evidence that they may regulate initial
stages of neoplastic transformation. In fact, both
ASF/SF2andhnRNPA1arealsosignificantlyelevated
in uninvolved tissue surrounding tumors relative to
control lung, suggesting that they play a role
in the transition of normal cells to neoplastic. If
this is the case, both factors may prove to be
promising biomarkers for lung cancer, similar to
hnRNP A2/B1 [8].
hnRNP A1 localized to the nucleus of tumor cells
while it exhibited both nuclear and cytoplasmic
localization in nonneoplastic proliferating lung
(Fig. 2). ASF/SF2 was observed in both nuclear and
cytoplasmic compartments in tumors as well as
regenerative lung. There may be trafficking mechan-
isms that localize hnRNP A1 to the nucleus during
tumorigenesis. Indeed, other investigators have
shown that the phosphorylation status of ASF/SF2
affects its intracellular trafficking between the
nucleus and cytoplasm and determines in part its
nuclear content relative to hnRNP A1 [28,29].
Because pre-mRNA processing proteins elicit glo-
bal changes in mRNA expression, upregulation of
this class of proteins in precancerous lesions can
dramatically affect the neoplastic phenotype. Diver-
sity in the proteome resulting from changes in
alternative splicing would enhance the possibilities
for selecting variant cells capable of overcoming
normal homeostatic controls, and encourage neo-
plastic growth. Protein isoforms spliced from the
same pre-mRNA display unique biological activities
and are subject to distinct regulatory controls. For
example, the short and long isoforms of Bcl-x have
opposing functions in apoptosis [30]. Genes whose
protein isoforms change during tumorigenesis reg-
ulate major hallmarks of neoplasia, including:
oncogenes such as Kras [31] and Wnt [32]; tumor
suppressor genes such as Brca1 [33] and Mdm2 [34];
genes regulating apoptosis such as Bcl2, Fas, and
caspase 2 [35]; genes regulating angiogenesis such as
VEGF[36];genesinvolvedinDNArepairsuchasDNA
polymerase b [37]; genes regulating cell adhesion and
intercellular communication such as CD44 [16]
and cadherin-11 [38]; and genes involved in cellular
senescence such as the INK4a locus (p14ARFand
p16INK4a) [39].
The best-characterized example of how hnRNP A1
and ASF/SF2 may modulate mRNA alternative spli-
cing is CD44, a membrane glycoprotein that binds
hyaluronic acid and osteopontin and is involved in
cell-cell and cell-extracellular matrix interactions
[16,18]. CD44 has multiple alternative splice forms,
with particular variants associated with metastasis.
These variant species influence primary tumor
growth and enable tumor cells circulating in the
vasculature to adhere to the vascular endothelium,
thereby facilitating tumor seeding at distant sites. In
vitro studies show that hnRNP A1 binds regulatory
splice elements in CD44 variant exon v5 and modu-
lates its cancer-associated alternative splicing [40].
Changes in the concentrations of ASF/SF2 corre-
spond to altered CD44 splicing in mouse mammary
gland tumorigenesis [15] and human colon adeno-
carcinoma [27]. We describe herein a substantial
increase in the metastatic-associated CD44 mRNA
isoforms in lung tumors compared to regenerative
proliferating lung, uninvolved lung, and control
tissue. The parallel time course of increased hnRNP
A1 and ASF/SF2 expression with appearance of the
variant CD44 isoforms suggests that hnRNP A1 and
ASF/SF2 may be involved in these splicing changes.
There are numerous additional splicing factors,
however, and changes in one or several of these
may also impact expression of variant CD44 iso-
forms. We are currently conducting in vitro studies
todelineateanycausalrolebetweenaberranthnRNP
A1 and ASF/SF2 expression and alternative splicing
of CD44 as well as other cancer-associated genes.
194194
ZERBE ET AL. ZERBE ET AL.

Page 9

In summary, numerous enhancing and repressing
proteinsinteractinacombinatorialfashionwithpre-
mRNA and the spliceosome to direct splice-site
selection [3,4]. Because of thiscomplexity, very little
isknownaboutthisprocessinawholeanimalmodel
or how it becomes dysregulated during disease states
including cancer. The intriguing changes in hnRNP
A1 and ASF/SF2 in mouse lung tumors described
herein are the beginnings of an effort to characterize
the role of pre-mRNA alternative splicing in lung
neoplasia. This unique neoplastic expression profile
of splicing factors could serve as a diagnostic tool for
identifying early lung cancer as well as distinguish-
ing lung cancer subtypes.
ACKNOWLEDGMENTS ACKNOWLEDGMENTS
We thank Ms. Katherine A. Peebles for her helpful
comments on this manuscript. This study was
supported by USPHS CA33497 and CA 96133 (to
A.M.M).; FIS/00/0835, PI021116, and RTIC C03/10
(to L.M.M. and R.P.), and University of Colorado
Cancer Center Seed Grant (to L.K.Z).
REFERENCESREFERENCES
1. Stewart BW, Kleihues P. World Cancer Report. Lyon:
IARCPress; 2003.
2. Mercatante D, Kole R. Modification of alternative splicing
pathways as a potential approach to chemotherapy.
Pharmacol Ther 2000;85:237–243.
3. Dreyfuss G, Kim VN, Kataoka N. Messenger-RNA-binding
proteins and the messages they carry. Nat Rev Mol Cell Biol
2002;3:195–205.
4. Black DL. Mechanisms of alternative pre-messenger RNA
splicing. Annu Rev Biochem 2003;72:291–336.
5. Graveley BR. Alternative splicing: Increasing diversity in the
proteomic world. Trends Genet 2001;17:100–107.
6. Modrek B, Resch A, Grasso C, Lee C. Genome-wide detec-
tion of alternative splicing in expressed sequences of human
genes. Nucleic Acids Res 2001;29:2850–2859.
7. Wang ZN, Lo HS, Yang H, et al. Computational analysis
and experimental validation of tumor-associated alternative
RNA splicing in human cancer. Cancer Res 2003;63:655–
657.
8. Tockman MS, Mulshine JL, Piantadosi S, et al. Prospective
detectionofpreclinicallungcancer:Resultsfromtwostudies
of heterogeneous nuclear ribonucleoprotein A2/B1 over-
expression. Clin Cancer Res 1997;3:2237–2246.
9. PinoI,PioR,ToledoG,etal.Alteredpatternsofexpressionof
members of the heterogeneous nuclear ribonucleoprotein
(hnRNP) family in lung cancer. Lung Cancer 2003;41:131–
143.
10. PeeblesKA,DuncanMW,RuchRJ,MalkinsonAM.Proteomic
analysis of a neoplastic mouse lung epithelial cell line whose
tumorigenicity has been abrogated by transfection with the
gap junction structural gene for connexin 43, Gjal. Carcino-
genesis 2003;24:651–657.
11. Blaxall BC, Dwyer-Nield L, Bauer AK, Bohlmeyer TJ,
Malkinson AM, Port JD. Differential expression and localiza-
tion of the mRNA binding proteins, AU-rich element mRNA
bindingprotein(AUF-1)andHuantigenR(HuR),inneoplastic
lung tissue. Mol Carcinog 2000;28:76–83.
12. Malkinson AM. Primary lung tumors in mice as an aid for
understanding, preventing, and treating human adenocar-
cinoma of the lung. Lung Cancer 2001;32:265–279.
13. Mayeda A, Krainer AR. Regulation of alternative pre-mRNA
splicing by hnRNP A1 and splicing factor SF2. Cell 1992;68:
365–375.
14. CaceresJF,StammS,HelfmanDM,KrainerAR.Regulationof
alternative splicing in vivo by overexpression of antagonistic
splicing factors. Science 1994;265:1706–1709.
15. Stickeler E, Kittrell F, Medina D, Berget SM. Stage-
specificchangesinSRsplicingfactorsandalternativesplicing
in mammary tumorigenesis. Oncogene 1999;18:3574–
3582.
16. Cooper DL, Dougherty GJ. To metastasize or not? Selection
of CD44 splice sites. Nat Med 1995;1:635–637.
17. Hirata T, Fukuse T, Naiki H, Hitomi S, Wada H. Expression of
CD44 variantexon 6 in stage I non-small cell lung carcinoma
as a prognostic factor. Cancer Res 1998;58:1108–1110.
18. Ponta H, Sherman L, Herrlich PA. CD44: From adhesion
molecules to signalling regulators. Nat Rev Mol Cell Biol
2003;4:33–45.
19. Malkinson AM, Koski KM, Evans WA, Festing MF. Butylated
hydroxytoluene exposure is necessary to induce lung tumors
inBALBmicetreatedwith3-methylcholanthrene.CancerRes
1997;57:2832–2834.
20. Malkinson AM. Prevention of butylated hydroxytoluene-
induced lung damage in mice by cedar terpene administra-
tion. Toxicol Appl Pharmacol 1979;49:551–560.
21. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein
measurement with the folin phenol reagent. J Biol Chem
1951;193:265–275.
22. Malkinson AM, Beer DS. Major effect on susceptibility to
urethan-induced pulmonary adenoma by a single gene in
BALB/cBy mice. J Natl Cancer Inst 1983;70:931–936.
23. Witschi H, Malkinson AM, Thompson JA. Metabolism and
pulmonary toxicity of butylated hydroxytoluene (BHT).
Pharmacol Ther 1989;42:89–113.
24. Malkinson AM, Thaete LG, Blumenthal EJ, Thompson JA.
Evidenceforaroleoftert-butylhydroxylationintheinduction
of pneumotoxicity in mice by butylated Hydroxytoluene.
Toxicol Appl Pharmacol 1989;101:196–204.
25. Adamson IY, Bowden DH, Cote MG, Witschi H. Lung injury
induced by butylated hydroxytoluene: Cytodynamic and
biochemical studies in mice. Lab Invest 1977;36:26–32.
26. ThaeteLG,GunningWT,StonerGD,MalkinsonAM.Cellular
derivation of lung-tumors in sensitive and resistant strains of
mice—Resultsat28and56weeksafterurethanetreatment.
J Nat Cancer Inst 1987;78:743–749.
27. Ghigna C, Moroni M, Porta C, Riva S, Biamonti G. Altered
expression of heterogenous nuclear ribonucleoproteins and
SR factors in human colon adenocarcinomas. Cancer Res
1998;58:5818–5824.
28. Xiao SH, Manley JL. Phosphorylation of the ASF/SF2 RS
domain affects both protein-protein and protein-RNA inter-
actions and is necessary for splicing. Genes Dev 1997;11:
334–344.
29. Caceres JF, Screaton GR, Krainer AR. A specific subset of SR
proteins shuttles continuously between the nucleus and the
cytoplasm. Genes Dev 1998;12:55–66.
30. Boise LH, Gonzalezgarcia M, Postema CE, et al. Bcl-X, A Bcl-
2-related gene that functions as a dominant regulator of
apoptotic cell-death. Cell 1993;74:597–608.
31. Wang Y, You M, Wang Y. Alternative splicing of the K-ras
gene in mouse tissues and cell lines. Exp Lung Res 2001;27:
255–267.
32. KatohM,KirikoshiH,SaitohT,SagaraN,KoikeJ.Alternative
splicing of the WNT-2B/WNT-13 gene. Biochem Biophys Res
Commun 2000;275:209–216.
33. OrbanTI,OlahE.ExpressionprofilesofBRCA1splicevariants
in asynchronous and in G1/S synchronized tumor cell lines.
Biochem Biophys Res Commun 2001;280:32–38.
34. Lukas J, Gao DQ, Keshmeshian M, et al. Alternative and
aberrant messenger RNA splicing of the mdm2 oncogene in
invasive breast cancer. Cancer Res 2001;61:3212–3219.
hnRNP A1 AND ASF/SF2 IN MOUSE LUNG TUMORS hnRNP A1 AND ASF/SF2 IN MOUSE LUNG TUMORS
195 195

Page 10

35. Jiang ZH, Wu JY. Alternative splicing and programmed cell
death. Proc Soc Exp Biol Med 1999;220:64–72.
36. Yuan A, Yu CJ, Kuo SH, et al. Vascular endothelial growth
factor 189 mRNA isoform expression specifically correlates
withtumorangiogenesis,patientsurvival,andpostoperative
relapse in non-small-cell lung cancer. J Clin Oncol 2001;19:
432–441.
37. Thompson TE, Rogan PK, Risinger JI, Taylor JA. Splice
variants but not mutations of DNA polymerase beta are
common in bladder cancer. Cancer Res 2002;62:3251–
3256.
38. Feltes CM, Kudo A, Blaschuk O, Byers SW. An alternatively
spliced cadherin-11 enhances human breast cancer cell
invasion. Cancer Res 2002;62:6688–6697.
39. Zhu DG, Xu G, Ghandhi S, Hubbard K. Modulation of the
expressionofp16(INK4a)andp14(ARF)byhnRNPA1andA2
RNA binding proteins: Implications for cellular senescence.
J Cell Physiol 2002;193:19–25.
40. MatterN,MarxM,Weg-RemersS,PontaH,HerrlichP,Konig
H. Heterogeneous ribonucleoprotein A1 is part of an exon-
specific splice-silencing complex controlled by oncogenic
signaling pathways. J Biol Chem 2000;275:35353–35360.
196196
ZERBE ET AL.ZERBE ET AL.