Mice Deficient for the Ets Transcription Factor Elk-1 Show Normal Immune Responses and Mildly Impaired Neuronal Gene Activation

Abstract

The transcription factor Elk-1 belongs to the ternary complex factor (TCF) subfamily of Ets proteins. TCFs interact with serum response factor to bind jointly to serum response elements in the promoters of immediate-early genes (IEGs). TCFs mediate the rapid transcriptional response of IEGs to various extracellular stimuli which activate mitogen-activated protein kinase signaling. To investigate physiological functions of Elk-1 in vivo, we generated Elk-1-deficient mice by homologous recombination in embryonic stem cells. These animals were found to be phenotypically indistinguishable from their wild-type littermates. Histological analysis of various tissues failed to reveal any differences between Elk-1 mutant and wild-type mice. Elk-1 deficiency caused no changes in the proteomic displays of brain or spleen extracts. Also, no immunological defects could be detected in mice lacking Elk-1, even upon infection with coxsackievirus B3. In mouse embryonic fibroblasts, Elk-1 was dispensable for c-fos and Egr-1 transcriptional activation upon stimulation with serum, lysophosphatidic acid, or tetradecanoyl phorbol acetate. However, in brains of Elk-1-deficient mice, cortical and hippocampal CA1 expression of c-fos, but not Egr-1 or c-Jun, was markedly reduced 4 h following kainate-induced seizures. This was not accompanied by altered patterns of neuronal apoptosis. Collectively, our data indicate that Elk-1 is essential neither for mouse development nor for adult life, suggesting compensatory activities by other TCFs.

Ternary complex factors (TCFs) represent a subclass of the Ets family of transcription factors (reviewed in references 38, 44, and 50). TCF was discovered in HeLa cell nuclear extracts that formed ternary complexes with the serum response factor (SRF) on the c-fos serum response element (SRE) (15, 39). To date three TCFs have been identified: Elk-1 (Ets-like protein 1) (6, 17), Sap-1 (SRF accessory protein 1) (also called Elk-4) (7), and Net (also called Sap-2, Erp, or Elk-3) (10, 26). The TCFs contain four conserved domains. The N-terminal ETS domain mediates DNA binding. The B box enables the physical interaction with SRF (12, 22, 40). The D domain acts as a docking site for mitogen-activated protein kinases (MAPKs) (8, 19, 55, 56), and the C domain functions as a MAPK-inducible transcription activation domain (5, 16, 23, 30). The transcriptional activation potential of the TCFs is activated by C-domain phosphorylation at multiple Ser and Thr residues. The MAPKs Erk1/2, JNK, and p38 can catalyze these modifications as a function of extracellular stimulation. TCFs have been implicated in transducing mitogen and stress signals to immediate-early genes (IEGs), such as c-fos and Egr-1, to elicit a transient and rapid response (reviewed in reference 20). Additionally, Elk-1 phosphorylation by the Erk MAPK pathway can lead to the recruitment of the mSin3A-histone deacetylase I corepressor complex, and this event correlates kinetically with the postinductional shutoff of the c-fos promoter (54). The TCFs appear to respond selectively to different subsets of MAPK cascades. Elk-1 activity has been shown to be regulated by Erk1/2, JNK, and p38 MAPK pathways (42, 55). Sap-1, instead, is efficiently targeted by Erk and p38 pathways but appears to be a poor JNK substrate (32, 41, 51, 52) except when JNK is overexpressed (21). Net differs from the other TCFs in its ability to strongly repress transcription, whereby activation of the Erk signaling pathway results in loss of this repressive function (10). Repression is mediated by two domains of Net, the Net inhibitory domain (28) and the CtBP interaction domain (4). A novel evolutionarily conserved repression domain, the R motif, has been identified in Elk-1, which apparently reduces the basal transcriptional activity of Elk-1 and its response to mitogenic signals (53).

The TCF-SRF-SRE ternary complex apparently confers a tightly controlled, temporal, and cell-type-specific regulation upon SRE-containing promoters. At the same time, the three TCFs are coexpressed in many cell types and tissues (2, 10, 26, 27, 31, 37). Therefore, it remains unclear to what extent individual TCFs fulfill specific cellular functions. Sap-1^−/− mice are viable and fertile but display immunological defects, including in T-cell function. They display a phenotype similar to that of Castleman's disease (R. Treisman, personal communication). Net, on the other hand, as determined based on phenotypic consequences of expression of the Netδ allele in Net-deficient mice, has been implicated in angiogenesis and formation of the lymphatic network (3). These animals survive to birth, develop postnatal chylothorax, and die due to respiratory failure. Furthermore, the expression of Egr-1 is elevated in some vascular structures in the thoracic region of the Netδ mice.

To investigate the function of Elk-1 in vivo, we generated mice lacking Elk-1. This was achieved via deletion by homologous recombination of the entire Elk-1 gene in mouse embryonic stem (ES) cells, followed by blastocyst injection and breeding of chimeric mice. Elk-1-deficient mice are viable, appear phenotypically normal, and exhibit normal life spans. For most parameters tested, they are indistinguishable from their wild-type littermates. We could not detect any defect in induction of the IEGs c-fos and Egr-1 in mouse embryonic fibroblasts (MEFs) lacking Elk-1 upon stimulation with serum, tetradecanoyl phorbol acetate (TPA), or lysophosphatidic acid (LPA). Interestingly, however, c-fos expression, but not that of Egr-1 or c-Jun, was reduced in the amygdalae, the hippocampi, and the cortices of Elk-1 null mice compared to wild-type animals at 4 h following kainic acid application. Taken together, these results indicate that Elk-1 is not essential for murine development, growth, and adult life, but they rather suggest a possible role of Elk-1 in selected activities of the brain.

MATERIALS AND METHODS

Elk-1 targeting vector.

The mouse Elk-1 locus was subcloned from a strain 129/SV genomic library (Lambda Fix II; Stratagene). The Elk-1 targeting vector (pSK-delElk1.HygTk.DT) was designed to achieve deletion of the entire Elk-1 coding region on the X chromosome and replace it with an expression cassette bearing the hygromycin B phosphotransferase-thymidine kinase gene (HygTk) positive-negative selection marker flanked by a wild-type (F) and a modified (F3) Flp recognition target site (FRT) (35) (Fig. 1A). The components of the targeting vector were a 5′ arm of homology corresponding to an 5-kb genomic EcoRI-SpeI fragment upstream of the first Elk-1 coding exon, the F3HygTkF selection cassette, a 3′ arm of homology of 1.1 kb which was generated by PCR, and a diphtheria toxin A cassette (DT) for negative selection. The targeting vector was linearized with SalI before transfection of ES cells.

FIG. 1.

Targeted disruption of the mouse Elk-1 gene. (A) Schematic diagram of the murine Elk-1 locus, the targeting vector for homologous recombination, and the structure of the recombined null allele (named Elk1-137). Blue boxes represent the Elk-1 coding exons. In the targeting vector, with its genomic sequences marked as black bars, the complete coding region and intronic sequences contained therein were replaced by the positive-negative selection genes HygTk, flanked by two nonidentical sites of recognition for the Flp recombinase (F/F3) in direct orientation. DT, additional diphtheria toxin cassette. The green bar marks the location of the 3′ external Southern probe used in panel B. Restriction site abbreviations: BHI, BamHI; Bg2, BglII; BsE, BstEII; RI, EcoRI. WT, wild type. (B) Southern blot analysis of wild-type (E14-1) and recombined (Elk1-137) Elk-1 alleles in ES cells and progeny from Elk1-137 mice (gender and Elk-1 genotype as indicated). The lengths of the expected genomic EcoRI restriction fragments, representative of the wild-type (13.5 kb) and targeted (7.9 kb) Elk-1 alleles, are given on the right.

Generation of Elk-1 knockout mice.

Linearized Elk-1 targeting vector DNA was electroporated into E14-1 ES cells cultured on a layer of mitotically inactive feeder cells carrying a genomic hygromycin B resistance gene (45). ES cell colonies resistant to 200 μg of hygromycin B per ml were expanded and screened for homologous recombination by PCR and Southern blot analysis of EcoRI digests by using a 1.1-kb 3′ probe external to the targeting vector (Fig. 1B). Southern data were confirmed by using a 5′ probe external to the targeting vector (not shown). One Elk-1^−/0 ES cell clone (named Elk1-137), exhibiting a normal karyotype, was injected into C57BL/6N (Charles River, Sulzfeld, Germany) blastocysts to obtain chimeric animals. Chimeric male mice with a high contribution of agouti coat color were bred to C57BL/6N females, and germ line transmission of the Elk-1 disrupted allele was obtained. Genotypes of these mice were confirmed by Southern blotting of tail DNA (Fig. 1B). Subsequent genotyping was carried out by PCR with allele specific primers (data not shown; experimental details are available upon request). The primers used were as follows: Elk-1 allele, GGCTATGGGGACTGCGCAAGAACAAGACC (forward) and GCCATAAACTACCGAGGACAGAAAGCACAGG (reverse); Elk1-137 allele, GAGGGCGTGGATATGTCCTGCGGGTAAATAGC (forward) and CGCCTCGCTCCAGTCAATGACCGCTGTTAT (reverse).

Animals were maintained under specific-pathogen-free conditions and were backcrossed into the C57BL/6N genetic background for at least five generations.

Isolation and culture of MEFs.

MEFs were isolated from day 13.5 homozygous wild-type and mutant embryos. After removal of the heads and internal organs, the remaining tissue was dissociated into single cells by trypsinization. Cells were then cultured in Dulbecco modified Eagle medium containing 4.5 g of glucose per liter and 3.7 g of NaHCO₃ per liter and supplemented with 2 mM l-glutamine, 100 U of penicillin per ml, 100 g of streptomycin per ml, and 10% fetal calf serum (FCS). Cultivation was at 37°C in a humidified atmosphere at 5% CO₂, and cells were used at passages 3 to 5.

For IEG induction studies, cells were serum starved in Dulbecco modified Eagle medium with 2 mM l-glutamine, 100 U of penicillin per ml, 100 g of streptomycin per ml, and 0.1% FCS for 24 h. Cells were then stimulated with 15% FCS, 100 ng of the phorbol ester tetradecanoyl phorbol acetate (TPA)/ml, or 20 μM LPA for different times, as indicated in Fig. 3.

FIG. 3.

IEG induction of c-fos and Egr-1 by serum, TPA, or LPA is not impaired in Elk-1-deficient MEFs. Quantitative RT-PCR measuring expression of c-fos (left panels) and Egr-1 (right panels) in MEFs of the indicated Elk-1 genotypes is shown. Cells were either serum starved for 24 h (0 min) or serum starved and subsequently stimulated with 15% serum for 40, 60, or 180 min (upper panels); with 100 ng of TPA per ml for 30, 60, or 180 min (middle panels); or with 20 μmol of LPA per liter for 30, 60, or 180 min (lower panels). mRNA levels were normalized to those of the Hprt gene. Each kinetic measurement was performed in duplicate. One representative data set from three independent experiments is shown.

Expression analysis by RT-PCR.

Preparation of total RNA (RNeasy kit; Qiagen) and first-strand cDNA synthesis (Superscript II; Gibco) were done according to the manufacturers' instructions. Quantitative reverse transcription-PCR (RT-PCR) analysis of IEGs has been described previously (36). The primers used were as follows: c-fos, CCTGCCCCTTCTCAACGAC (forward) and GCTCCACGTTGCTGATGCT (reverse); Egr-1, GCCGAGCGAACAACCCTAT (forward) and TCCACCATCGCCTTCTCATT (reverse); Elk-1, TGCTCCCCACACATACCTTGA (forward) and ACTGGACGGAAACTGGAAGGA (reverse); Hprt, GCCTAAGATGAGCGCAAGTTG (forward) and TACTAGGCAGATGGCCACAGG (reverse); and Gapdh, TGGCATGGCCTTCCGT (forward) and TCTCCAGGCGGCACGT (reverse).

Brain extracts and electrophoretic mobility shift assay (EMSA).

Brain tissue was homogenized in lysis buffer at 4°C by using a Polytron PT2100 homogenizer (Kinematika). Lysates were obtained by centrifugation for 20 min at 4°C at 16,000 × g. Aliquots were frozen in liquid nitrogen and stored at −80°C. Full-length human Elk-1 protein was synthesized with the TNT-coupled reticulocyte lysate system (Promega) with human Elk-1 cDNA as the template (pGEM-hElk1). The control reaction (TNT mock) was made in the absence of template DNA. Purified human SRF protein was previously described (13).

EMSA reaction mixtures (10 μl) contained 10 mM Tris (pH 7.5), 50 mM NaCl, 1 mM EDTA, 5% glycerol, 0.05% low-fat milk, 1 mM dithiothreitol (DTT), 2 μg of salmon sperm DNA, and 30,000 cpm of ³²P-labeled double-stranded oligonucleotides. These mixtures were supplemented with 100 ng of purified SRF protein and/or 5 μl of TNT reaction product (TNT Elk-1 or mock) or 50 μg of brain protein extracts. When used, the Elk-1 (PAb512) (28) and SRF [SRF(G20); Santa Cruz] antisera were added to the reaction mixture before addition of the extracts. After a 30-min incubation period at room temperature, the DNA-protein complexes were separated on a 5% polyacrylamide gel containing 0.5× Tris-borate buffer at 1 mA/cm. Gels were dried and exposed to phosphorimaging screens (Fuji). The DNA probes used were as follows: c-fos SRE sense strand, 5′-AGCTTACACAGGATGTCCATATTAGGACATCTGCGTCAG-3′; c-fos SRE antisense strand. 5′-TCGACTGACGCAGATGTCCTAATATGGACATCCTGTGTA-3′; E74 sense strand, 5′-GATCTCTAGCTGAATAACCGGAAGTAACTCATCCTTG-3′; E74 antisense strand, 5′-CTAGGTTCCTACTCAATGAAGGCCAATAAGTCGATCT-3′.

2D gel electrophoresis.

Whole brains and spleens from two adult Elk-1^+/0 and two Elk-1^−/0 mice were rapidly dissected and homogenized in lysis buffer {9 M urea, 4% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate [CHAPS], 1% Pharmalytes, and 1% DTT} at 4°C with a Polytron PT2100 homogenizer (Kinematika). Insoluble material was removed by centrifugation at 10,000 × g for 5 min. The supernatant was precipitated with acetone for 1 h at −20°C and then resuspended in 2 ml of lysis buffer. After centrifugation for 5 min at 10,000 × g, the supernatant was used for two-dimensional (2D) gel analysis. Isoelectric focusing was performed by using IPG 4-7 strips (18 cm) from Amersham Biosciences. The second-dimension sodium dodecyl sulfate-polyacrylamide gel electrophoresis on a 14% running gel was performed without a stacking gel. The equilibrated IPG was placed on top of the second-dimension gel (20 by 20 by 1.5 cm), and electrophoresis was carried out at a constant 250 mA. Each organ extract was separated repeatedly (three to five times) by 2D gel electrophoresis. After silver staining (L. G. Vuong et al., submitted for publication), the stained gels were scanned and images were analyzed by using Progenesis software (Perkin-Elmer Life Sciences). In this analysis, the Expression Window function was used to highlight matched spots in the gels that showed two- to threefold or higher differences in normalized volume. Difference maps were used to investigate expression changes between gels. No statistically significant differences were detected between gels displaying proteins from Elk-1^+/0 and Elk-1^−/0 organ extracts.

Coxsackievirus infection and in situ hybridization.

The coxsackievirus B3 (CVB3) infection protocol and CVB3 RNA detection by radioactive in situ hybridization were described previously (24). Briefly, dewaxed paraffin-embedded tissue sections (6 μm) from the heart, pancreas, spleen, liver, skeletal muscle, and lung, mounted on 3-aminopropyl-3-ethoxysilane-coated microscopic slides, were incubated for 18 h at 42°C in hybridization buffer containing ³⁵S-labeled CVB3 RNA probe (500 ng/ml) or pSPT18 control probe (500 ng/ml). The probes were dissolved in 10 mM Tris-HCl (pH 7.4)-50% (vol/vol) deionized formamide-600 mM NaCl-1 mM EDTA-0.02% polyvinylpyrrolidone-0.02% Ficoll-0.05% bovine serum albumin-10% dextran sulfate-10 mM DTT-200 mg of denaturated sonicated salmon sperm DNA per ml-100 mg of rabbit liver tRNA per ml. Slides were washed, autoradiographed as described, and stained with hematoxylin-eosin.

Flow cytometry.

For intracellular cytokine staining, freshly isolated splenocytes were cultured overnight at 5 × 10⁶ cells per well in a 24-well-plate in α-MEM containing 10% fetal calf serum, 2 mM glutamine, 50 U of penicillin per ml, and 50 μg of streptomycin per ml in the presence or absence of 2.5 μg of concanavalin A (ConA) (Sigma) per ml. GolgiStop (Becton Dickinson) was added according to the manufacturer's instructions and left for 5 h. The cells were then fixed, permeabilized, and stained at 200,000 cells per well of a 96-well-plate with fluorochrome-labeled antibodies (all from Becton Dickinson) as indicated in the instructions to the intracellular cytokine-staining kit (Becton Dickinson).

Similarly, thymocytes were isolated, counted, and stained with fluorochrome-labeled anti-CD8, anti-CD4, anti-CD25, and anti-CD44 antibodies for 30 min on ice. Flow cytometry was performed on a four-color FACS-Calibur (Becton Dickinson) with CellQuestPro software (Becton Dickinson) for data acquisition and analysis. Triplicates of individual experiments, as well as means of all experiments, were subjected to a standard two-tailed Student t test for assessment of significance.

For cell surface staining of murine splenocytes (10⁶/ml) from CVB3-infected and noninfected mice, the cells were incubated with a mix of fluorochrome-conjugated antibodies (either anti-CD3-allophycocyanin (APC), anti-CD4-peridinine chlorophyll-a protein (PerCP), anti-CD45R [B220]-R-phycoerythrin (R-PE), and anti-CD69-fluorescein isothiocyanate or anti-CD3-APC, anti-CD8-PerCP, anti-CD28-R-PE, and anti-CD25-fluorescein isothiocyanate [all from Becton Dickinson]) for 30 min at 4°C. Flow cytometry was done as described above.

Kainic acid induction of IEGs.

To investigate IEG induction following seizure activity, 10 Elk-1^−/0 mice and 10 Elk-1^+/0 control animals (backcross 10 in C57BL/6N) were injected with kainic acid (Tocris; 30 mg/kg intraperitoneally [i.p.]) and kept for 4 h before being sacrificed. Symptomatic seizure development was analyzed within the next 30 min and in a further follow-up according to the ranking used previously (34): stage 1, immobility (animals do not walk around); stage 2, forelimb and/or tail extension and rigid posture; stage 3, repetitive movements and head bobbing; stage 4, rearing and falling; stage 5, continuous rearing and falling; and stage 6, severe tonic-clonic seizures. If an animal did not surpass stage 2 to 3 after 30 min, an additional 3 mg of kainic acid per kg was injected i.p. every 15 min until the animal reached at least stage 4.

For immunohistochemistry, mice were deeply anesthetized with chloral hydrate (600 mg/kg i.p.) and killed by intracardial perfusion with ice-cold 4% paraformaldehyde. Brains were removed, postfixed, and dehydrated in a 30% sucrose solution. Subsequently, brains were cryoprotected and cut into 40-μm free-floating sections. The following primary antibodies were used: c-Fos (1:300; rabbit polyclonal; Oncogene), c-Jun (1:400; rabbit polyclonal; Calbiochem/Oncogene), Egr-1 (1:6,000; rabbit polyclonal; characterized by R. Bravo) (14, 25), and Iba1 1:2,000 (rabbit polyclonal as characterized previously [18]).

Cell death was determined by cresyl violet staining and by detection of terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL)-positive DNA fragmentation with the Roche kit according to the manufacturer's instructions.

To determine cell death and differences in the expression profiles in the cortex, amygdala, or hippocampus, maximally induced immunoreactivity was recorded by photography for each animal (Leica Qwin software), and nuclear immunoreactivities were quantified in microscopic frames of 500 by 500 μm in three independent brain slices of each animal. Significances (P ≤ 0.05) were calculated with Student's t test.

RESULTS

Generation of Elk-1-deficient mice.

In order to assess the function of Elk-1 in mammals, a null mutation of the Elk-1 gene was generated in the mouse genome by homologous recombination in murine ES cells. The Elk-1 targeting vector was designed to delete the entire Elk-1 coding sequence within the genomic Elk-1 locus and to simultaneously replace this sequence with an expression cassette carrying the HygTk fusion gene (Fig. 1A). This positive-negative selection marker was flanked by two different FRT recognition sequences (F and F3) for the Flp recombinase (35). The two FRT sites were used in a separate study to modify the tagged Elk-1 locus by recombinase-mediated cassette exchange (35) (F. Cesari et al., unpublished data).

Since murine Elk-1 is located on the X chromosome (11), its targeted deletion in the genotypically male E14-1 ES cells led to the generation of an Elk-1^−/0 ES cell clone (named Elk1-137), as confirmed by genomic PCR (data not shown) and Southern blot analysis (Fig. 1B, lane 1). ES cells lacking Elk-1 showed no phenotypic abnormalities regarding cell morphology, proliferative capacities, or differentiation potential (data not shown). Upon blastocyst injection, the Elk1-137 ES cell clone transmitted the Elk-1 mutation into the germ line of male chimeric mice, which were subsequently mated to C57BL/6N wild-type females. Elk-1^+/− females were then crossed with C57BL/6N wild-type males to generate animals with the Elk-1^−/0 genotype, as determined by genomic PCR (data not shown) and Southern blot analysis (Fig. 1B) with tail DNA. The full viability of Elk-1-deficient animals demonstrated that Elk-1 is not an essential gene in the mouse.

The Elk1-137 allele represents a null allele.

To confirm that Elk1-137 represented a null allele, RT-PCR (Fig. 2A to C) and protein assays (Fig. 2D) were performed. No Elk-1-specific RT-PCR product could be obtained by using RNA from either brain tissue of Elk-1^−/0 mice (Fig. 2B, lanes 1 and 2) or from Elk-1^−/0 primary MEFs (Fig. 2C, lane 2), whereas the expected RT-PCR products were generated from RNAs derived from Elk-1^+/0 mice (Fig. 2A, lanes 1 and 2) or the corresponding MEFs (Fig. 2C, lane 1). Control RT-PCR with Gapdh primers resulted in the expected PCR product in Elk-1^+/0 and Elk-1^−/0 mice (Fig. 2A and B, lanes 7 and 8) and MEFs (Fig. 2C, lanes 3 and 4). Control reactions lacking reverse transcriptase verified the absence of genomic DNA in the RNA preparations (Fig. 2A and B, lanes 3 and 4).

FIG. 2.

The mutated Elk-1 allele represents a bona fide null allele. (A and B) RT-PCR analysis of Elk-1 mRNA from brains of Elk-1^+/0 (A) and Elk-1^−/0 (B) mice. Duplicate RT-PCR experiments were performed with primers specific for Elk-1 (lanes 1 and 2) or Gapdh (lanes 7 and 8). Lanes 3 and 4 represent chromosomal DNA contamination controls; lanes 5 and 6 represent water-only (no cDNA) controls. (C) RT-PCR analysis of Elk-1 (lanes 1 and 2) and Gapdh (lanes 3 and 4) mRNAs in MEFs from Elk-1^+/0 (lanes 1 and 3) and Elk-1^−/0 (lanes 2 and 4) mouse embryos. (D) EMSA with SRE (lanes 1 to 11) or E74 (lanes 12 to 17) probes with brain extracts from Elk-1^+/0 and Elk-1^−/0 animals, as indicated. A radiolabeled c-fos SRE DNA probe was incubated with recombinant SRF alone (lane 2) or with additional in vitro-synthesized Elk-1 protein (lanes 3 to 5). The SRE probe was also incubated with brain extracts from wild-type (lanes 6 to 8) or Elk-1^−/0 (lanes 9 to 11) animals. SRF-containing DNA-protein complexes (cI and cII) were supershifted by a polyclonal anti-SRF antiserum (αSRF) (lanes 5, 8, and 11). The polyclonal anti-Elk-1 antiserum (αElk-1) (lanes 4, 7, and 10) recognized Elk-1-containing DNA-protein complexes (cII) (lanes 4 and 7). A radiolabeled E74 oligonucleotide probe, which can be bound directly by Elk-1 in the absence of SRF, was incubated with in vitro-synthesized Elk-1 (lanes 12 and 13). The same probe was incubated with brain extracts from wild-type (lanes 14 and 15) or Elk-1^−/0 (lanes 16 and 17) animals. The polyclonal anti-Elk-1 antiserum added in lanes 13, 15, and 17 identified Elk-1-containing DNA-protein complexes (dc) (lanes 13 and 15). No Elk-1-containing complexes were detected in Elk-1^−/0 brain extracts (lanes 9, 10, 16, and 17). ni, DNA-protein complexes generated by nonidentified E74 binding proteins in brain extracts, presumably non-Elk-1 Ets proteins; uc, nonreproducible, unspecific complex.

Protein extracts from brains of wild-type or Elk-1^−/0 mice were analyzed for Elk-1 DNA binding activity, as measured in EMSA with either SRE or E74 DNA probes (Fig. 2D). The Elk-1 protein cannot bind to the SRE sequence alone (17), whereas the SRF-Elk-1-SRE ternary complex (cII) was formed in the presence of added SRF protein (Fig. 2D, lane 3). The cII complex, which was recognized by antibodies specific for either SRF or Elk-1 (Fig. 2D, lanes 4 and 5), was observed in the Elk-1^+/0 but not in the Elk-1^−/0 brain extracts (Fig. 2D, compare lanes 6 and 9). The absence of Elk-1 DNA binding activity in the Elk-1^−/0 brain extracts was confirmed by using the E74 EMSA probe, which can be bound directly by Elk-1 alone (Fig. 2D, lanes 12 and 14) and can be supershifted by anti-Elk-1 antiserum (Fig. 2D, lanes 13 and 15). The Elk1-E74 direct complex was seen only with the Elk-1^+/0 and not with the Elk-1^−/0 brain extracts (Fig. 2D, lanes 14 and 16). These results confirmed that the Elk1-137 allele is a null allele of the murine Elk-1 gene.

Elk-1 is dispensable for efficient induction of the IEGs c-fos and Egr-1 in MEFs.

Transcription of the IEGs c-fos and Egr-1 is rapidly induced in response to many different extracellular stimuli, including growth factors, cytokines, neurotransmitters, phorbol esters, or radiation. Associated activation of signaling cascades, primarily MAPK signaling, targets the TCF components of TCF-SRF-SRE ternary complexes located in some IEG promoters (43). Therefore, we examined the consequence of Elk-1 deficiency on c-fos and Egr-1 induction in MEFs obtained from wild-type or Elk-1-deficient mouse embryos. Serum-withdrawn MEFs were stimulated for up to 3 h with either serum, TPA, or LPA, and the resulting c-fos and Egr-1 mRNA levels were measured by quantitative RT-PCR analysis (Fig. 3). Both wild-type and Elk-1-deficient MEFs were able to mount a robust IEG response, as judged by c-fos and Egr-1 transcription induction profiles. Fully congruent results were obtained with Elk-1^−/0 ES cells (not shown). These results demonstrate that Elk-1 is not essential for IEG induction in MEFs, probably due to functional compensation by other TCFs (Sap-1 and/or Net) still present in Elk-1-deficient cells. Of note, Elk-1 deficiency did not lead to increased expression levels of Sap1a or Net mRNAs in MEFs or in various organs (brain, heart, lung, liver, and kidney) of Elk-1^−/0 animals (data not shown).

Characterization of Elk-1-deficient mice.

Breeding of Elk-1^+/− females with wild-type males generated equal numbers of Elk-1^−/0 and Elk-1^+/0 male offspring. Thus, the absence of Elk-1 protein does not compromise male embryonic development. Elk-1^−/0 males carrying the HygTk gene of the Elk1-137 allele, however, were sterile, most probably due to aberrant expression of the Tk gene in the testis (1, 9). This hypothesis was supported by the phenotypic correction obtained with a second Elk-1 null allele (named Elk1-RMCE16) that did not contain the Tk gene in this locus. In this case, removal of the HygTk cassette from the Elk1-137 locus by recombinase-mediated cassette exchange generated fertile Elk-1-deficient mice (Cesari et al., unpublished data). In order to analyze the phenotype of the Elk-1-deficient mice carrying the Elk1-137 allele, Elk-1^+/− females were backcrossed with C57BL/6N wild-type males for at least five generations. Elk-1-deficient mice were phenotypically indistinguishable from their wild-type littermates. Growth rates and life expectancies were comparable to those of wild-type animals. In addition, histological analyses of brains, spleens, livers, lungs, hearts, kidneys, stomachs, intestines, pancreas, thymuses, and lymph nodes of Elk-1^−/0 animals did not show any phenotypic abnormalities (data not shown). Furthermore, organs from Elk-1^−/0 and Elk-1^+/0 animals were used to perform proteome analyses. Protein extracts from brains and spleens of two Elk-1^−/0 and two wild-type littermates were prepared, and the proteins were separated by 2D gel electrophoresis. Although about 2,600 protein signals could be detected on a brain 2D image with the aid of the Progenesis software (Perkin-Elmer Life Sciences), no statistically significant qualitative or quantitative differences in protein signals could be identified between the brain extracts of Elk-1^−/0 and Elk-1^+/0 mice (data not shown). A similar result was obtained for the spleen samples (not shown).

T-cell maturation and activation are not impaired in Elk-1-deficient mice.

We investigated the effect of Elk-1 deficiency on the immunological activities of T-cell maturation and activation, using flow cytometry. We first analyzed the overall distribution of thymocytes and found that the respective cell populations of CD4⁻ CD8⁻, CD4⁺ CD8⁺, CD4⁺ CD8⁻, and CD4⁻ CD8⁺ cells showed comparable percentile representations in both Elk-1^+/0 and Elk-1^−/0 animals (Fig. 4A, upper panel). Second, the subpopulations of the CD4⁻ CD8⁻ cell pool were determined with regard to the expression of the CD25 and CD44 markers. From both Elk-1^+/0 and Elk-1^−/0 animals equal numbers (data not shown) and percentages of (i) T-cell precursor cells entering the thymus as CD25⁻ CD44⁺ cells, (ii) intermediate stages of T cells expressing both CD25 and CD44 molecules, and (iii) the subsequently generated CD25⁺ CD44⁻ and CD25⁻ CD44⁻ populations were obtained (Fig. 4A, lower panel).

FIG. 4.

Elk-1 deficiency does not affect activation and development of T-cells. (A) The percentages of cells at different stages of T-cell development, as determined by fluorescence-activated cell sorter analysis, do not differ substantially between Elk-1^+/0 and Elk-1^−/0 mice. (B) After stimulation with the mitogen ConA, splenocytes from Elk-1^+/0 and Elk-1^−/0 mice do not show differences in the induction of gamma interferonproduction by CD8⁺ T cells, IL-2 production by CD4⁺ T cells, or tumor necrosis factor alpha production by CD4⁺ T cells or in upregulation of CD25 by CD4⁺ T cells, CD69 upregulation by CD8⁺ T cells, or CD44 upregulation by CD4⁺ T cells. The CD25, CD69, and CD44 upregulations, although shown for only one T-cell subset (CD4⁺ or CD8⁺), are representative of those for all T cells (data not shown). Representative results obtained from two mice are shown for each genotype. Error bars indicate standard deviations.

Taken together, these data show that in Elk-1^−/0 animals, T-cell maturation is not impaired. Irrespective of the Elk-1 genotype, comparable numbers of T cells downregulated CD25 to develop into CD4⁺ CD8⁺ double-positive thymocytes, of which 95% die by positive and negative selection. Surviving naive T cells either recognize major histocompatibility complex class I and are CD8⁺ or recognize major histocompatibility complex class II and are CD4⁺. In agreement, the CD4⁺/CD8⁺ ratio was not affected in Elk-1-deficient mice (not shown).

We then analyzed the mitogenic and activating effects of the lectin ConA on T cells derived from Elk-1^+/0 and Elk-1^−/0 animals. Activated T cells produce cytokines, which can be visualized by flow cytometry. Gamma interferon is preferentially produced by CD8⁺ T cells, and no difference could be detected between wild-type and Elk-1 knockout cells (Fig. 4B). Similarly, interleukin-2 (IL-2) production by CD4⁺ T cells and tumor necrosis factor alpha production by all T cells were not affected by the absence of Elk-1 (Fig. 4B). The first activation marker, visible on the T-cell surface already at 2 h after activation, is CD69, followed by the IL-2 receptor α-chain (CD25) and later by CD44. All of these activation markers were expressed by similar numbers of Elk-1-deficient and wild-type T cells (Fig. 4B) and at comparable levels (data not shown). The decreasing numbers of cells expressing CD69, CD25, and CD44 reflect the activation kinetics of these surface markers, with the markers being expressed on fewer cells representing later upregulation. The percentages of B cells, natural killer (NK) cells, and γδ T cells did not differ in cells isolated from either the spleens or lymph nodes of Elk-1^−/0 and Elk-1^+/0 mice (data not shown).

Elk-1 knockout mice do not differ from wild-type animals regarding infection with CVB3.

To address a potential function of Elk-1 in fighting viral infection, Elk-1^−/0 and Elk-1^+/0 mice were inoculated intraperitoneally with CVB3. CVB3 induced a mild myocarditis in Elk-1 wild-type and knockout mice, with mononuclear cell infiltrates and cytotoxic effects of infected myocytes. The virus destroyed the major part of the exocrine pancreas in both mouse strains. In the spleens of both Elk-1^−/0 and Elk-1^+/0 mice, primarily cells within lymph follicles contained viral RNA. No infection or inflammation was observed in skeletal muscle, lung, or liver of Elk-1-deficient and wild-type animals (Fig. 5A). In addition, no significant differences in the splenic lymphocyte subpopulations or in the expression of functional markers on T cells could be observed between CVB3-infected wild-type and Elk-1-deficient mice (Fig. 5B).

FIG. 5.

CVB3 infection of Elk-1^−/0 and Elk-1^+/0 animals. (A) Radioactive in situ hybridization for detection of enteroviral RNA in different organs of acutely CVB3-infected Elk-1^+/0 and Elk-1^−/0 mice. Patterns of infection reveal comparable amounts of CVB3 RNA-positive cells in heart, pancreas, and spleen of both mouse strains. Skeletal muscle, liver, and lung were negative for viral RNA in both Elk-1^+/0 and Elk-1^−/0 mice. (Magnification ×50). (B) Quantitation of lymphocyte subpopulation profiles (T cells, TH cells, cytotoxic T lymphocytes [CTLs], and B cells) (upper left panel) in CVB3-infected Elk-1^−/0 and Elk-1^+/0 mice. Subgroups within these T-cell subpopulations are further quantified according to functional marker protein expression (Cd25, CD69, and CD28) (upper right and lower panels). *, defined as CD3⁺ CD8⁻; **, defined as CD3⁺ CD4⁻.

Reduced neuronal expression of c-fos, but not of Egr-1 and c-Jun, in Elk-1 knockout mice following kainic acid injection.

The role of Elk-1 in brain is not yet well understood. However, Elk-1 has been described as being expressed in neuronal cell types of the rat brain and as being phosphorylated in response to glutamate receptor activation (37, 48, 49). In addition, the activation of Elk-1 has been correlated to the activation of IEGs, such as c-fos, in these cells. To address Elk-1 function in the brain, we studied the ability of the glutamate receptor agonist kainic acid to induce IEG expression in Elk-1-deficient and control mice.

We first investigated IEG activation in the hippocampi of Elk-1^+/0 and Elk-1^−/0 animals 4 h after kainate administration. Treatment with kainic acid (30 mg/kg i.p.) led to severe seizures in the majority of mice. According to the suggested seizure intensity scale, 80% of the mice reached stage 4 within 30 min following kainic acid injection. In both the Elk-1^+/0 and Elk-1^−/0 groups, the numbers of (i) nonresponding individuals (below stage 4) and (ii) mice dying from the kainic acid treatment did not differ (two or three animals per group). Additionally, no differences in the onset or duration of seizure activities were observed.

c-fos expression was not detectable in brains of untreated mice (Fig. 6). At 4 h after kainic acid application, c-fos expression was strongly induced in the hippocampi of Elk-1^+/0 mice, specifically in granule cells of the CA1, the CA3, the dentate gyrus, the amygdala, and the cortex (but not in the cerebellum and the hypothalamus or the midbrain regions). This induction of c-fos was significantly impaired in Elk-1^−/0 mice (Fig. 6A, upper panel, and B). A reduced intensity of c-fos was found in the granule cell layer of CA1, CA3, and the dentate gyrus of Elk-1^−/0 mice. Lower numbers of c-fos-expressing neurons were found in Elk-1^−/0 mice than in controls, as evidenced in the amygdala (18 ± 8 versus 32 ± 6 neurons per 2.5 mm²) (data not shown), the hippocampus in CA1 and CA3 (25 ± 5 versus 33 ± 4 neurons per 2.5 mm²), and the cortex (37 ± 15 versus 54 ± 17 neurons per 2.5 mm²) (Fig. 6).

FIG.6.

Expression of neuronal IEGs (c-fos, Egr-1, and c-Jun) following kainic acid challenge of mice. (A) Induction of inducible transcription factors in the rostral hippocampi of untreated mice and at 4 h following kainic acid injection. In untreated Elk-1^−/0 and wild-type Elk-1^+/0 animals (left column), there was no c-fos expression detectable. Basal expression of Egr-1 was moderate in CA1 and weak in both CA3 and the dentate gyrus (DG). Weak basal expression of c-Jun was found in the dentate gyrus prior to treatment. Following kainic acid-induced seizures (middle and right columns), strong c-fos induction was seen in the hippocampi of wild-type mice, whereas in Elk-1^−/0 mice c-fos induction was less efficient with respect to both the number of positive cells and the intensity of c-fos staining. Egr-1 is induced by kainic acid in the dentate gyrus and CA3 of the rostral hippocampi of both Elk-1^+/0 and Elk-1^−/0 mice. c-Jun expression is also induced by kainate in the dentate gyrus, CA1, and CA3 regions, without significant differences between Elk-1^+/0 and Elk-1^−/0 mice. Magnification, ×50. For orientation, the hippocampal regions CA1, CA3, and dentate gyrus are indicated in the upper middle picture. (B) Cortical and hippocampal CA1 expression of c-fos. Whereas in untreated Elk-1^+/0 and Elk-1^−/0 mice, low basal c-fos expression was seen, a strong neuronal induction was observed in CA1 and the cortex of Elk-1^+/0 mice. This induction was markedly less pronounced in Elk-1^−/0 mice. Magnification, ×200. For orientation, regions of cortex and CA1 are indicated. Asterisks delineate the granule cell layer of CA1.

Egr-1 expression, on the other hand, was found in many cortical neurons in untreated mice already (43 ± 17 versus 47 ± 13 neurons per 2.5 mm²). In the hippocampi of untreated mice, Egr-1 was almost absent in the dentate gyrus, except for single labeled neurons. A weak immunoreactivity was found in CA3 neurons, with a wide variation and a strong basal expression of Egr-1 being detected in the granule cell layer of CA1 neurons (Fig. 6A). Differences between Elk-1^+/0 and Elk-1^−/0 mice regarding Egr-1 induction upon kainate treatment were not observed. After 4 h of kainic acid treatment, the cortical and the hippocampal CA1 expression patterns of Egr-1 did not change, whereas Egr-1 induction was observed in the granule cell layer of the hippocampal CA3 and dentate gyrus regions at similar levels in both Elk-1^+/0 and Elk-1^−/0 animals (Fig. 6A). Other brain areas, such as the amygdala, also did not show differences in Egr-1 expression patterns between strains.

c-Jun expression was observed at a low level in granule neurons of the dentate gyrus in all untreated animals. At 4 h after kainic acid treatment, very strong c-Jun expression was found in the amygdala, in the cortex, and in all neurons of the granule cell layer of the hippocampus (Fig. 6A). No differences in this strong c-Jun activation response were observed between Elk-1^−/0 and Elk-1^+/0 mice. Two days after kainic acid treatment, IEG expression of Egr-1, c-fos, and c-Jun in the hippocampus, amygdala, and cortex was markedly reduced compared to that at 4 h after treatment, without any differences between Elk-1^+/0 and Elk-1^−/0 animals (data not shown).

In conclusion, Elk-1 deficiency affected the efficiency of rapid kainate-induced c-fos expression 4 h after treatment, but not that of Egr-1 or c-Jun expression, in the forebrains of mice.

Second, we investigated cell survival in the forebrains of mice at 4 h and 2 days after kainic acid treatment. After 4 h of kainate treatment, no TUNEL-positive DNA fragmentation or cresyl violet staining of pyknotic cells was observed in any of the experimental animals (data not shown). After 2 days after kainate treatment, organized neuronal clusters in the hippocampal CA3 and CA1 regions showed pyknotic cell death and TUNEL-positive DNA fragmentation. These signals displayed a wide range of interindividual variation and did not differ between Elk-1^+/0 and Elk-1^−/0 mice (data not shown). Furthermore, we addressed the impact of Elk-1 deficiency on microglial cells by staining for the microglia-macrophage-specific antibody Iba1. The microglial morphology, indicative of microglial activation, did not differ between Elk-1^−/0 and Elk-1^+/0 mice 2 days after kainate treatment (data not shown).

DISCUSSION

This paper describes the generation of mice, ES cells, and MEFs lacking the transcription factor Elk-1. It is demonstrated that Elk-1 does not fulfill an essential function in mouse development or adult life or in mediating the IEG response in the cultured cells tested. Since there are two close relatives of Elk-1 with significantly overlapping expression patterns, Sap-1 and Sap-2/Erp/Net, we speculate that these three Ets-type TCF proteins can functionally compensate for each other's activities to a significant degree. Indeed, preliminary experiments designed to lower either Sap-1 or Net mRNA expression in Elk-1-deficient MEFs, using small interfering RNA strategies, indicate that reduction of expression levels of two of the three TCFs severely impairs c-fos immediate-early induction (A. Kettenbach et al., unpublished data).

In light of the structural similarities exhibited by the three TCF proteins Elk-1, Sap-1, and Sap-2/Erp/Net, a certain degree of functional compensation between them was to be anticipated. However, given their differences regarding (i) relative expression levels (10, 11, 33), (ii) responsiveness to different types of MAPK (21), and (iii) ability to act as a transcriptional repressor and/or activator (4, 28), the observed weak biological effects of Elk-1 deficiency are surprising. A possible mechanism for functional compensation of Elk-1 deficiency would be compensatory changes in Sap-1/Net expression levels. Our comparison of the RNA expression levels of Sap-1 and Net in wild-type and Elk-1-deficient organs (brain, kidney, liver, heart, and lung) and cells (MEFs and ES cells) failed to detect any difference (not shown). Differences in expression at the protein level cannot be analyzed with certainty, given the lack of appropriate immunoreagents. However, at the level of sensitivity of EMSA activity, we observed a strong reduction in ternary complex-forming activity in brain extracts from Elk-1^−/− animals compared to wild-type mice (Fig. 2, lanes 6 and 9). This argues against compensatory upregulation at the protein level of non-Elk-1 TCF activity in the brains of Elk-1-deficient mice. Should such compensatory adaptation have occurred during development, this might be revealed in future experiments by conditional knockout strategies which permit removal of Elk-1 at adult stages.

In contrast to the normal antiviral defense and T-cell activation profile of Elk-1^−/− mice described here, Sap-1^−/− animals display immunological defects, including impairment in the generation of single-positive T cells and a phenotype similar to that in Castleman's disease (R. Treisman, personal. communication). This may suggest either that Elk-1 does not contribute in a major way to T-cell activation or that Elk-1 functions can be compensated for by other TCFs, whereas Sap-1 functions cannot be. Future generation of double and triple knockout mice that are deficient for more than one TCF will help to solve the issue of functional redundance within the TCF group of Ets family proteins and will clarify the activities of individual TCFs in vivo.

The only difference between Elk-1-expressing and Elk-1-deficient mice observed to date relates to the inefficiency of c-fos induction in the forebrains of kainate-treated Elk-1^−/− mice (Fig. 6). This kainate treatment, however, did not elicit different patterns of neuronal cell death in the two mouse strains after 2 days. c-Fos was found previously to affect cell death upon kainic acid treatment (57). However, no clear correlation between levels of Fos expression in different regions of the brain and sensitivity to kainate-induced cell death could be established (29).

Another SRE-containing IEG, Egr-1, was induced normally in the absence of Elk-1, revealing a significant induction in CA3 and dentate gyrus at 4 h after kainate treatment. Although 4 h is known to be the time point at which maximal Egr-1 induction is displayed in wild-type hippocampi, we presently cannot formally rule out some effects of Elk-1 deficiency on the precise kinetic profile of Egr-1 induction. The differential requirements of c-fos and Egr-1 for Elk-1 in the process of kainate induction may be based on the different promoter structures of these two genes, which are endowed with one and four SREs, respectively. The expression and function of Elk-1 in the brain had previously been studied primarily in rat (33, 37). Elk-1 was found selectively in neurons, where the protein was observed not only in nuclei but also in soma, dendrites, and axon terminals (37). Elk-1 activation, as measured by specific phosphorylation, accompanied stimulatory treatments eliciting long-term potentiation (48) or drug addiction (46, 47). Future behavioral studies will reveal whether Elk-1 deficiency in mice causes any effects on brain functions relating to learning, memory, and drug addiction.