DUSP6 (MKP3) Null Mice Show Enhanced ERK1/2 Phosphorylation at Baseline and Increased Myocyte Proliferation in the Heart Affecting Disease Susceptibility*

The strength and duration of mitogen-activated protein kinase signaling is regulated through phosphorylation and dephosphorylation by dedicated dual-specificity kinases and phosphatases, respectively. Here we investigated the physiological role that extracellular signal-regulated kinases 1/2 (ERK1/2) dephosphorylation plays in vivo through targeted disruption of the gene encoding dual-specificity phosphatase 6 (Dusp6) in the mouse. Dusp6-/- mice, which were viable, fertile, and otherwise overtly normal, showed an increase in basal ERK1/2 phosphorylation in the heart, spleen, kidney, brain, and fibroblasts, but no change in ERK5, p38, or c-Jun N-terminal kinases activation. However, loss of Dusp6 did not increase or prolong ERK1/2 activation after stimulation, suggesting that its function is more dedicated to basal ERK1/2 signaling tone. In-depth analysis of the physiological effect associated with increased baseline ERK1/2 signaling was performed in cultured mouse embryonic fibroblasts (MEFs) and the heart. Interestingly, mice lacking Dusp6 had larger hearts at every age examined, which was associated with greater rates of myocyte proliferation during embryonic development and in the early postnatal period, resulting in cardiac hypercellularity. This increase in myocyte content in the heart was protective against decompensation and hypertrophic cardiomyopathy following long term pressure overload and myocardial infarction injury in adult mice. Dusp6-/- MEFs also showed reduced apoptosis rates compared with wild-type MEFs. These results demonstrate that ERK1/2 signaling is physiologically restrained by DUSP6 in coordinating cellular development and survival characteristics, directly impacting disease-responsiveness in adulthood.

The strength and duration of mitogen-activated protein kinase signaling is regulated through phosphorylation and dephosphorylation by dedicated dual-specificity kinases and phosphatases, respectively. Here we investigated the physiological role that extracellular signal-regulated kinases 1/2 (ERK1/2) dephosphorylation plays in vivo through targeted disruption of the gene encoding dual-specificity phosphatase 6 (Dusp6) in the mouse. Dusp6 ؊/؊ mice, which were viable, fertile, and otherwise overtly normal, showed an increase in basal ERK1/2 phosphorylation in the heart, spleen, kidney, brain, and fibroblasts, but no change in ERK5, p38, or c-Jun N-terminal kinases activation. However, loss of Dusp6 did not increase or prolong ERK1/2 activation after stimulation, suggesting that its function is more dedicated to basal ERK1/2 signaling tone. In-depth analysis of the physiological effect associated with increased baseline ERK1/2 signaling was performed in cultured mouse embryonic fibroblasts (MEFs) and the heart. Interestingly, mice lacking Dusp6 had larger hearts at every age examined, which was associated with greater rates of myocyte proliferation during embryonic development and in the early postnatal period, resulting in cardiac hypercellularity. This increase in myocyte content in the heart was protective against decompensation and hypertrophic cardiomyopathy following long term pressure overload and myocardial infarction injury in adult mice. Dusp6 ؊/؊ MEFs also showed reduced apoptosis rates compared with wild-type MEFs. These results demonstrate that ERK1/2 signaling is physiologically restrained by DUSP6 in coordinating cellular development and survival characteristics, directly impacting disease-responsiveness in adulthood.
Mitogen-activated protein kinase (MAPK) 3 signaling pathways consist of a sequence of successively acting kinases that ultimately result in the dual phosphorylation and activation of terminal kinases p38, c-Jun N-terminal kinases (JNKs), and extracellular signal-regulated kinases (ERKs) (1). The major upstream activators of ERK1/2 are two MAPK kinases (MAPKK), MEK1, and MEK2, which directly phosphorylate a dual site in the activation loop of ERK1/2 kinases (Thr-Glu-Tyr). p38 kinases are directly activated by MKK6 and MKK3, JNKs are directly activated by MKK4 and MKK7, whereas ERK5 is directly activated by MKK5. Upstream of the MAPKKs, multiple MAPKKKs and even MAPKKKKs exist that form a complex network of kinases that either directly sense stress stimulation, or are directly regulated by effectors such as low molecular weight G-proteins (Ras, Rac, Rho, and others). For example, mitogen-induced activation of the insulin-like growth factor receptor 1 results in Ras activation through the action of Grb2 and Sos, which directly leads to Raf (a MAPKKK) activation and then MEK1/2 (MAPKK) activation followed by ERK1/2 (1).
The duration and extent of MAPK phosphorylation are crucial features in determining the physiological response to any given mitogenic or stress stimulation. The terminal MAPKs (ERK1/2, JNK, ERK5, and p38) are each inactivated through dephosphorylation of either the threonine or the tyrosine residue, or both, within the activation loop. Threonine dephosphorylation and partial inactivation of ERK1/2 and p38 can be achieved through protein phosphatase 2A (2)(3)(4)(5)(6), although paradoxically PP2A has also been shown to enhance Ras-Raf1-MEK1/2-ERK1/2 signaling (7,8). Perhaps of even greater physiological importance, a specialized family of phosphatases has evolved that can dephosphorylate both serine/threonine and tyrosine residues, known as dual-specificity protein phosphatases (DUSPs). There are 38 dual-specificity phosphatases (DUSP) genes in the human genome, 11 of which are highly specific for the MAPKs and are referred to as MAPK phosphatases (MKPs) (9 -11). A unique feature of many MKP-DUSPs is their regulation at the level of gene expression following stress or mitogen stimulation, providing a negative feedback loop to dampen the extent and duration of MAPK signaling (9 -11).
Once expressed, this subclass of DUSPs is constitutively active and capable of direct binding to the activation loop in MAPKs, resulting in dephosphorylation. Each of the 11 MKP-DUSPs family members differs with respect to subcellular localization (cytoplasmic versus nuclear), tissue expression pattern, or exact specificity for ERK1/2/5, JNK1/2, or p38 (9 -11). For example, DUSP6 (MKP3) is cytoplasmic and highly specific for ERK1/2 at endogenous levels of expression. Indeed, detailed biochemical analysis shows that ERK1/2 directly bind and induce closure of the catalytic "general acid" loop in DUSP6 (12)(13)(14). Detailed mutagenesis and domain swapping showed that Arg-65 in DUSP6 makes direct contacts with ERK2 and that DUSP6 contains the ERK-specific FXFP motif that is also found in Elk-1 (13,(15)(16)(17). The binding domains between ERK2 and DUSP6 have also been carefully mapped with hydrogen/deuterium exchange mass spectrometry (18).
The Ras-Raf-1-MEK1/2-ERK1/2 signaling pathway responds to a very wide array of mitogenic and stress stimuli, often in a cell type-dependent manner. Not surprisingly, ERK1/2 signaling participates in an extremely diverse range of cellular responses subsequently affecting proliferation, differentiation, survival versus apoptosis, and hypertrophy (19). With respect to regulation of cell proliferation, ERK1/2 can translocate to the nucleus and phosphorylate the transcription factors Elk1, c-Fos, p53, Ets1/2, and c-Jun, leading to enhanced cell cycle activity and/or facilitating tumorigenesis (20,21). Indeed, the transforming capacity of Ras and Raf depend on intact MEK1 and ERK1/2 signaling (20,21). Analysis of Erk1 and/or Erk2 gene-targeted mice revealed a function for these kinases in controlling proliferation of T cells at select stages in vivo (22,23). Moreover, Erk2 Ϫ/Ϫ mice die during early embryonic development, most likely due to defects in proliferation of trophectoderm or cells that eventually contribute to the placenta, likely leading to secondary defects in mesoderm development as well (24 -26). Knockdown of ERK2 by RNA interference in wild-type (Wt) or Erk1 Ϫ/Ϫ hepatocytes also reduced proliferation, further supporting the importance of ERK in regulating cell cycle (27). With respect to dephosphorylation of ERK1/2, Dusp6 Ϫ/Ϫ mice were recently generated by another group and shown to be mostly normal, except for a lowly penetrant phenotype of postnatal lethality, skeletal dwarfism, coronal craniosynostosis, and hearing loss (28).
The mammalian heart initially grows through a combination of myocyte proliferation and hypertrophy, although early in the postnatal developmental period proliferation ceases and all subsequent heart growth is due to hypertrophy. In response to various congenital or acquired stimuli the adult heart can undergo an acute hypertrophic response that is largely devoid of cellular proliferation (29). Induction of cardiac hypertrophy is initially a compensatory response that temporarily augments cardiac output, although prolonged hypertrophy can be deleterious and eventually leads to heart failure and/or sudden death (30,31). The hypertrophic response is initiated by neural, humoral, and intrinsic stimuli resulting in the activation of membrane-bound receptors that in turn activate specific signaling pathways such as MAPK. Specifically, ERK1/2 have been implicated as important regulators of the cardiac growth response downstream of nearly all hypertrophy-inducing stim-uli characterized to date (32). Consistent with this hypothesis, cardiac-specific transgenic mice expressing activated MEK1 showed highly specific activation of only ERK1/2 in the heart and a phenotype of stable concentric hypertrophy (33). However, analysis of Erk1 or Erk2 gene-targeted mice, or transgenic mice overexpressing DUSP6 in the heart, were not defective in their ability to hypertrophy following multiple disease inducing stimuli (34). Thus, MEK1-ERK1/2 signaling appears to be sufficient but not necessary for cardiac growth in vivo. To further examine the function of ERK1/2 signaling in vivo we deleted the Dusp6 gene in mice. We identified a consistent increase in basal ERK1/2 signaling in multiple tissues. The effect associated with enhanced ERK1/2 activity was analyzed in depth in the heart, demonstrating a phenotype of greater cellular proliferation during development that rendered the heart with more cells and resistance to select disease insults as adults, in association with reduced apoptosis.

MATERIALS AND METHODS
Dusp6 Gene Targeting Strategy-A targeting vector that replaces all three exons of the Dusp6 gene with neomycin was generated and electroporated into mouse embryonic stem cells. Arms for the targeting vector were generated by PCR from SV129j genomic DNA to match the SV129j-based embryonic stem cells that were used. Correctly targeted embryonic stem cells were identified by Southern blotting and subsequently injected into C57Bl/6 blastocysts to generate chimeric mice, which were bred with C57Bl/6 mice to obtain germ line and homozygous null Dusp6 mice with a final hybrid background of C57Bl/6-SV129j. Wt and targeted alleles in mouse genomic DNA were detected by PCR using the primers: 5Ј-cagtccatcagcagaagccgatag-3Ј (exon 1), 5Ј-gctctatggcttctgaggcg-3Ј (3Ј neomycin), and 5Ј-cattgactcggagagtgatctggt-3Ј (reverse common primer).
Animal Models-All experimental procedures with animals were approved by the Institutional Animal Care and Use Committee. Transverse aortic constriction (TAC) of the transverse aorta was performed in 8-week-old mice as previously described (35). Transthoracic echocardiography to measure cardiac dimensions and pressure gradients across the aortic constriction was performed as described previously (36). Pressure gradients were calculated as 4 ϫ V max 2 (m/s) where V max is the velocity of the blood flow across the aortic constriction measured by Doppler. Fractional shortening (FS) was calculated using left ventricle dimensions in end of systole and diastole (LVES and LVED, respectively) according to the formula: FS ϭ [(LVED Ϫ LVES)/LVED] ϫ 100 (%). Azlet 1002 osmotic minipumps (Cupertino, CA) either filled with a mixture of angiotensin II/phenylephrine (AngII 432 g/kg/day, PE 100 mg/kg/day in 150 mM NaCl, 0.01 N acetic acid) or isoprenaline (60 mg/kg/day in phosphate-buffered saline) were implanted under the skin for 2 weeks following a routine surgical procedure (37). Organ weights normalized to body weights were measured at the indicated times and expressed in milligrams/g. For quantification of ventricular proliferation, pregnant females were injected intraperitoneally with bromodeoxyuridine (BrdUrd) labeling reagent according to the manufacturer's instructions (Zymed Laboratories Inc., San Francisco, CA).
The surgical procedure for ischemia-reperfusion (I/R) injury and myocardial infarction (MI) injury in the mouse were described previously (38). Briefly, a suture was tied with a slipknot around the left coronary artery, and mice were revived by removal from anesthetic during 60 min of ischemia, after which the knot was released and the heart was reperfused for 24 h (I/R). The MI procedure was identical except that the ligature was permanent with no reperfusion. Mice were sacrificed by CO 2 asphyxiation, and hearts were analyzed as previously described using 2% triphenyltetrazolium chloride in saline and 2% Evan's blue dye infusion to identify area at risk, infarct area, and area of perfusion for the I/R procedure (38).
Cell Culture-Mouse embryonic fibroblasts (MEFs) were isolated from embryos at day 12.5 of gestation. They were immortalized on a 3T3 protocol (9 ϫ 10 5 cells transferred at 3-day intervals) as previously described (39). When indicated, serum-starved confluent cells were stimulated with 1 M PE for the indicated amount of time. Some experiments were also conducted in serum-containing media, such as shown in Fig. 7.
Histological Analysis, Cell Size Measurement, TUNEL, and Immunohistochemistry-For histological analysis, adult hearts were fixed in 10% formalin/phosphate-buffered saline and dehydrated for paraffin embedding. Fibrosis was detected with Masson's Trichrome staining on 5-m paraffin sections. Blue collagen staining was quantified using Metamorph software (40). For cell surface area measurements, membranes were stained with fluorescein isothiocyanate-or TRITC-labeled lectin from Triticum vulgaris (Sigma), and nuclei were labeled with TO-PRO 3 iodine (Molecular Probes, Carlsbad, CA). Cellular areas were measured across cells with centered nuclei and quantified with ImageJ 1.33 software (Scion Corp., Frederick, MD). TUNEL rates were determined using the TMR Red In Situ Death Detection Kit according to the manufacturer's instructions (Roche Diagnostics) (36). For quantification of ventricular proliferation, 13.5-day-old embryos were harvested after 4 h of BrdUrd labeling and fixed in 10% formalin/phosphate-buffered saline. BrdUrd incorporation was identified on 5-m paraffin sections using the Brdu Staining kit (Zymed Laboratories Inc., San Francsico, CA) where peroxidase detection was replaced with an Alexa-fluor 568 mouse anti-BrdUrd antibody (1/50, Molecular Probes). Slides were counterstained with anti-GATA4 antibody (1/50, Santa Cruz Biotechnology, Santa Cruz, CA) to identify myocyte nuclei, coupled to Alexa-Fluor 488 anti-goat antibody (Molecular Probes). BrdUrd-and GATA4-doubled positive cells were quantified against total number of GATA4 cells. Anti-phospho-Ser-10-histone H3 staining (1/50, Upstate Biotechnology) and anti-phospho-ERK1/2 (1/100, Cell Signaling) was performed as previously described (36). Myocyte surface areas were also quantified after disassociation from 8-week-old adult hearts as described previously (41). Isolated adult myocytes were immediately fixed with ethanol, plated on polylysine-coated slides (Sigma), and incubated for 1 h with bisbenzimide (100 g/ml) to show nuclei.
Statistical Analysis-Means Ϯ S.E. are presented for all data analysis. Differences between two independent experimental groups with normal distribution were analyzed by unpaired Student's t test; non-normal samples were tested with the Mann-Whitney U test. For differences between three variables, parametric analysis of variance (coupled to a Tukey's post-test) was applied on normally distributed values. Otherwise, Kruskal-Wallis analysis of variance was applied (Dunn's posttest) (SigmaStat3.5 Software). p Ͻ 0.05 was considered significant.

Generation of Dusp6
Ϫ/Ϫ Mice-To examine the function of DUSP6 as a negative regulator of ERK1/2 signaling we inactivated the Dusp6 locus in embryonic stem cells. The Dusp6 gene contains three exons that were all deleted by homologous recombination using a targeting vector (Fig. 1A). Correctly targeted embryonic stem cells were used to generate chimeric mice, which were then used to generate germ line-containing heterozygous mice for the Dusp6-targeted locus. We re-confirmed that all three exons were absent from the Dusp6 locus in both the ES cells and targeted mice, indicating that a partial protein product for DUSP6 is not possible. Intercrosses of Dusp6 ϩ/Ϫ mice yielded homozygous null mice at weaning at predicted Mendelian frequencies, indicating no developmental or perinatal lethality associated with loss of this gene (data not shown). Western blotting of protein extracts generated from neonatal lung, spleen, thymus, and heart showed a complete absence of DUSP6 protein (Fig. 1, B and C). DUSP6 mRNA was previously shown to be relatively ubiquitous in expression (43).
Loss of Dusp6 Increases Basal ERK1/2 Phosphorylation-Given that DUSP6 is exclusive for ERK1/2 (9 -11), we first investigated the effect that deletion of this gene had on ERK1/2 phosphorylation in MEFs, which were first incubated overnight in serum-free media. Remarkably ERK1/2 phosphorylation levels were consistently elevated in Dusp6 Ϫ/Ϫ MEFs compared with Wt control MEFs generated in parallel ( Fig. 2A). Although Dusp6 Ϫ/Ϫ MEFs appeared to have slightly more total ERK1/2 protein, careful quantitation of phosphorylation levels normalized to total ERK1/2 showed a significant increase (p Ͻ 0.05, data not shown). Phosphorylation levels of JNK1/2 were unaffected by loss of Dusp6 in MEF cultures, although we sometimes observed a subtle decrease in p38 MAPK phosphorylation in the absence of Dusp6 ( Fig. 2A), which was only observed in the MEF culture system and not in primary tissues (see below). In 1-month-old mice, baseline ERK1/2 phosphorylation levels were increased by 2-fold in Dusp6 Ϫ/Ϫ hearts compared with Wt hearts (Fig. 2, B and C). This increase in baseline ERK1/2 phosphorylation in the absence of Dusp6 was also observed in spleen, kidney, and brain at 1 month of age (Fig.   2D). Thus, loss of the Dusp6 gene renders ERK1/2 slightly more active at baseline, suggesting that DUSP6 protein normally functions as a negative regulator of ERK1/2 signaling "tone" in multiple tissues.
Loss of Dusp6 Does Not Affect Stimulus-induced ERK1/2 Phosphorylation-Although deletion of Dusp6 affected basal phosphorylation of ERK1/2, we also hypothesized that ERK1/2 activation would be prolonged following agonist stimulation in the absence of this dual-specificity phosphatase. However, in contrast to our hypothesis, loss of Dusp6 did not prolong or otherwise affect ERK1/2 phosphorylation kinetics after agonist stimulation. For example, stimulation of Wt or Dusp6 Ϫ/Ϫ MEFs with the ␣-adrenergic agonist PE produced a similar profile and magnitude of ERK1/2 activation and inactivation (Fig.  3A). In vivo, systemic injection of PE in mice produced a similar profile of ERK1/2 activation and inactivation between hearts from Wt and Dusp6 Ϫ/Ϫ animals (Fig. 3B). Loss of Dusp6 did not affect p38 or JNK1/2 phosphorylation in the adult heart at base-  line or following PE stimulation (Fig. 3B). Wt and Dusp6 Ϫ/Ϫ mice were also subjected to cardiac pressure overload stimulation by TAC for 2 weeks, which normally induces abundant cardiac ERK1/2 phosphorylation (34). However, here again loss of Dusp6 did not enhance or prolong stimulus-induced ERK1/2 activation in this tissue (Fig. 3C). The only observable difference between Wt and Dusp6 Ϫ/Ϫ mice was an increase in basal ERK1/2 phosphorylation (i.e. compare Wt sham versus Dusp6 Ϫ/Ϫ sham) (Fig. 3, B and C). Immunohistochemistry for phospho-ERK1/2 in Dusp6 Ϫ/Ϫ MEFs consistently showed a change in subcellular localization, such that staining was observed around the perinucleus, with greater total nuclear accumulation compared with Wt MEFs (Fig. 3D, and data not shown). Histological sections from neonatal hearts from Dusp6 Ϫ/Ϫ mice also showed the same relative increase in perinuclear and nuclear phospho-ERK1/2 compared with Wt (data not shown). These results suggest that DUSP6 not only controls basal ERK1/2 phosphorylation, but it may also regulate ERK1/2 activity in a subcellular compartment-specific manner.
Dusp6 Ϫ/Ϫ Mice Show Greater Proliferation and Myocyte Cell Numbers in the Heart-Dusp6 Ϫ/Ϫ mice showed no obvious phenotypic defects, they bred normally, had body weights comparable to strain-matched Wt controls, and they lived well past 1 year of age (data not shown). However, Dusp6 Ϫ/Ϫ mice at 2 and 3 months of age showed significantly larger hearts and spleens normalized to body weight, yet liver, lung, and kidney weights were normal (Table 1). Given the prominent growth effect observed in the heart this tissue was examined in greater detail. Analysis of heart-weight normalized to body-weight (HW/BW) at 1 month and 1 year of age also showed a significantly greater increase (Fig. 4A). This increase in heart weight was not pathological, because cardiac function measured by echocardiography in 1-year-old null mice was comparable to Wt mice (FS ϭ 28.2% in Wt versus 28.6% in Dusp6 Ϫ/Ϫ ). Careful histological analysis of myocyte surface areas showed no difference at 1 month of age between Wt and Dusp6 Ϫ/Ϫ hearts, or from myocytes isolated from 2-month-old hearts, suggesting that the increase in heart weight (size) was due to greater cellularity (Fig. 4B). Consistent with this interpretation, hearts from 3-day-old Dusp6 Ϫ/Ϫ neonates showed a 30% increase in phospho-histone H3 labeling by immunohistochemistry compared with Wt hearts, whereas myocyte surface areas were again invariant (Fig. 4, C and D). We also observed a 30% increase in BrdUrd incorporation into the hearts of 13.5-day-old Dusp6 Ϫ/Ϫ embryos (Fig. 4E). However, no change was observed in the percentage of monoversus bi-nucleated myocytes isolated from 2-month-old hearts, further suggesting that differences in DNA synthesis rates likely produced cell number dif- A, quantitation of heart weight normalized to body weight (HW/BW) at 1, 2 and 12 months of age in Wt and Dusp6 Ϫ/Ϫ mice (n ϭ 4 or more each). *, p Ͻ 0.05 versus Wt control at the same age. B, quantitation of myocyte surface area from histological heart sections at 1 month of age or total surface area of individual myocytes isolated from hearts of 2-month-old Wt or Dusp6 Ϫ/Ϫ mice. Wt values were normalized to 100 arbitrary units (AU). C, immunohistochemical quantitation of phospho-histone H3 levels in myocytes from 3 day-old Wt and Dusp6 Ϫ/Ϫ mice (n ϭ 2000 nuclei counted each); D, with corresponding cell cross-sectional areas. *, p Ͻ 0.05 versus Wt. E, immunohistochemical quantitation of BrdUrd incorporation in the hearts of day 13.5 embryos from Wt and Dusp6 Ϫ/Ϫ mice (n ϭ 4000-6000 nuclei counted from two different embryos). *, p Ͻ 0.05 versus Wt. F, phospho-histone H3 levels measured by immunocytochemistry in cultures of Wt and Dusp6 Ϫ/Ϫ MEFs expressed as a percentage of total nuclei (n ϭ 500 nuclei counted for each). *, p Ͻ 0.05 versus Wt. ferences (data not shown). Thus, loss of Dusp6 and the concomitant increase in basal ERK1/2 signaling had a developmental effect that rendered the heart with greater proliferation rates and more myocytes. Indeed, even MEFs generated from Dusp6 Ϫ/Ϫ embryos had greater proliferation rates in culture (Fig. 4F). Adult Dusp6 Ϫ/Ϫ Mice Are Partially Protected from Stimuli That Induce Heart Disease-The increase in myocyte number associated with Dusp6 deletion also impacted the disease responsiveness of the adult heart. Two-month-old Wt and Dusp6 Ϫ/Ϫ mice were first subjected to agonist stimulation with angiotensin II and PE infusion over 2 weeks with Alzet minipumps, resulting in a similar cardiac hypertrophic response in both groups (Fig. 5A). Similarly, infusion of isoproterenol for 2 weeks generated a significant hypertrophic response and relative percent increase from baseline in both Wt and Dusp6 Ϫ/Ϫ mice, although the Dusp6 Ϫ/Ϫ hearts were significantly larger after isoproterenol than Wt controls (Fig. 5B). However, when an even greater hypertrophic stimulus was used, 8 weeks of pressure overload by TAC, both Wt and Dusp6 Ϫ/Ϫ mice produced a similar robust hypertrophic response (Fig. 5C). However, despite starting out with larger hearts, Dusp6 Ϫ/Ϫ mice subjected to TAC only hypertrophied to the same ultimate levels as Wt mice (not more). Importantly, the transaortic pressure gradients across the constriction were not different between the groups (68.5 mmHg in Wt versus 80 mmHg in Dusp6 Ϫ/Ϫ mice). While the overall growth of the heart was not different after TAC between the groups, analysis of myocyte cross-sectional surface areas from histological sections revealed less cellular hypertrophy in Dusp6 Ϫ/Ϫ mice compared with Wt (Fig. 5D). Thus, increased ERK1/2 basal signaling associated with the Dusp6 gene deletion reduced the growth of individual myocytes during long term pressure overload, an effect that is likely associated with increased cellularity in the heart (see "Discussion").
To more carefully examine the hypothesis that Dusp6 Ϫ/Ϫ mice might be protected from cardiac disease-inducing stimuli we performed long term pressure overload stimulation and measured cardiac function by echocardiography every 2 weeks for up to 8 weeks compared with sham. Wt and Dusp6 Ϫ/Ϫ mice subjected to TAC showed significant reductions in FS at 4, 6, and 8 weeks, although FS was significantly greater in TAC Dusp6 Ϫ/Ϫ mice at 6 and 8 weeks compared with Wt (Fig. 6A). These data were further strengthened by analysis of lung weights normalized to body weight as an indicator of circulatory failure and associated pulmonary edema, which tended to be greater in Wt compared with Dusp6 Ϫ/Ϫ mice (Fig. 6B). Moreover, the degree of cardiac fibrosis and induction of apoptosis measured by TUNEL were significantly increased in Wt FIGURE 5. Examination of the cardiac hypertrophic response in Dusp6 ؊/؊ mice following stimulation. A, quantitation of HW/BW in 2-month-old Wt or Dusp6 Ϫ/Ϫ mice implanted with control (saline) or AngII/PE containing Alzet mini-pumps (432 g/kg/day, 100 mg/kg/day, respectively) for 2 weeks (n ϭ 6 or more mice in each group). *, p Ͻ 0.05 versus Wt saline; NS, not significant. B, quantitation of HW/BW in 2-month-old Wt or Dusp6 Ϫ/Ϫ mice implanted with control (saline) or isoprenaline (Iso) containing Alzet mini-pumps (60 mg/kg/day) for 2 weeks (n ϭ 6 or more mice in each group). *, p Ͻ 0.05 versus Wt saline; # , p Ͻ 0.05 versus Wt isoprenaline; NS, not significant. C, quantitation of HW/BW in 2-month-old Wt or Dusp6 Ϫ/Ϫ mice subjected to TAC for 8 weeks (n ϭ 8 Wt saline, 6 Dusp6 Ϫ/Ϫ saline, 13 Wt TAC, and 11 Dusp6 Ϫ/Ϫ TAC). *, p Ͻ 0.05 versus Wt sham; NS, not significant. D, myocyte surface area from histological sections of the mice shown in C. *, p Ͻ 0.05 versus Wt sham; # , p Ͻ 0.05 versus Wt TAC; NS, not significant. hearts after 8 weeks of TAC, but not in Dusp6 Ϫ/Ϫ hearts (Fig. 6,  C and D). Collectively, these results suggest that deletion of Dusp6 from the heart partially protects from long term pressure overload-induced dysfunction and disease. We hypothesize that this protection was due, in part, to a developmental effect associated with greater number of myocytes (see below).
Loss of Dusp6 Affects Apoptosis Susceptibility-As previously suggested, it is possible that increased basal ERK1/2 signaling, such as in the absence of DUSP6, is inherently anti-apoptotic and protective to the adult heart (32). Alternatively, it is possible that greater ERK1/2 activity during development, which rendered the heart with more cells, was protective simply due to this increase in cellularity. In an attempt to distinguish between these two mechanisms we first analyzed the susceptibility of MEFs to apoptosis following staurosporine treatment. Interestingly, Dusp6 Ϫ/Ϫ MEFs showed significantly lower levels of TUNEL at 100 and 300 nM staurosporine compared with Wt MEFs (Fig. 7A). Similarly, Dusp6 Ϫ/Ϫ MEFs showed less staurosporine-induced cleavage of PARP, suggesting less caspase activation associated with apoptosis (Fig. 7B). These data suggest that enhanced basal ERK1/2 phosphorylation can protect fibroblasts from apoptotic signals in culture.
Consistent with the data obtained in MEFs, Dusp6 Ϫ/Ϫ mice subjected to permanent ligation of the left coronary artery (MI) showed significantly less decompensation 4 weeks after injury compared with Wt mice subjected to MI (Fig. 7C). Indeed, analysis of MI scar size using standard histological methods showed a greater degree of infarct expansion in Wt mice compared with Dusp6 Ϫ/Ϫ mice, suggesting less ongoing apoptosis in the absence of DUSP6 (Fig. 7D). These results suggest that loss of DUSP6 protein might also protect the heart by preventing ongoing apoptosis following injury. However, loss of the Dusp6 gene did not protect the heart from acute injury following I/R compared with Wt mice, suggesting that increased basal ERK1/2 signaling was not inherently protective in this context (Fig. 7, E and F). As a possible explanation for this seemingly incongruent result, I/R injury is an acute response and both Wt and Dusp6 Ϫ/Ϫ mice showed the same maximal activation of ERK1/2, whereas more chronic stimuli, such as TAC and MI, may be impacted by the increase in basal tone ERK1/2 phosphorylation given the protracted nature of this stimulation (see "Discussion").

DISCUSSION
Here we generated mice lacking Dusp6, an ERK1/2-specific inactivating phosphatase that is ubiquitously expressed. Dusp6 Ϫ/Ϫ or Dusp6 ϩ/Ϫ mice did not show embryonic, postnatal, or adult lethality, nor did they fail to thrive at any age. In contrast, another group independently generated and characterized mice targeted for Dusp6, which they reported to have a variable to lowly penetrant phenotype of dominant-postnatal lethality and skeletal dwarfism (28). Of 580 F 1 intercross offspring analyzed, Li et al. reported a Mendelian inheritance ratio of 21.5% (25% is the expected value) in homozygous null mice during the late postnatal period, which achieved a statistical significance of p ϭ 0.04. This minor postnatal lethal phenotype was enhanced in the F 2 and F 3 generations when backcrossed into the pure C57Bl/6 genetic background (28). Our failure to identify this subtle lethal phenotype during the postnatal period may reflect our strategy of backcrossing into the same hybrid genetic background (C57Bl/6Sv129j). Another interesting disparity between the two studies was our inability to identify "small" mice in the postnatal period from either Dusp6 ϩ/Ϫ or Dusp6 Ϫ/Ϫ offspring. In contrast, Li et al. identified 7 and 43% of Dusp6 ϩ/Ϫ and Dusp6 Ϫ/Ϫ offspring, respectively, that were classified as smaller than normal with skeletal dwarfism, coronal craniosynostosis, and hearing loss (28). Our lack of identifying these smaller mice may be due to the manner in which each genetic locus was targeted. We completely deleted the entire coding region of the Dusp6 gene (all three exons), whereas Li et al. generated an insertional mutation with a LacZ cassette that left the three Dusp6 exons mostly intact. Moreover, Li et al. reported the presence of a new higher species mRNA from the mutant Dusp6 locus that could theoretically generate a truncated form of DUSP6 with interfering activity. Despite these differences we both identified an increase in ERK1/2 phosphorylation in the absence of Dusp6, we both failed to observe any embryonic defects, and the majority of the mice in both studies appeared to be normal.
DUSP6 is a member of a large family of MAPK phosphatases that bind with varying specificity to ERK1/2, JNK1/2, p38␣/␤/ ␥/␦, ERK5, and presumably ERK3/4 (9 -11). DUSP6 is perhaps the best characterized of all the MKPs, and its specificity has been examined in detail. DUSP6 is known to form a tight complex with ERK2 through a series of carefully crafted studies that delineated the exact amino acids of interaction within two general binding surfaces between these proteins (12-15, 17, 18). These same amino acids are conserved in ERK1, but not ERK5 or ERK3/4, suggesting that DUSP6 is highly specific for ERK1/2. Indeed, we failed to observe any difference in ERK5 phosphorylation from Dusp6 Ϫ/Ϫ mice (data not shown). More importantly, we recently generated transgenic mice overexpressing DUSP6 protein in the heart (34). These mice showed a complete absence of ERK1/2 phosphorylation at baseline or after any applied stress stimulation, yet ERK5, p38, and JNK phosphorylation were unaffected (34). Thus, we believe that loss of Dusp6 only affects ERK1/2 within the greater MAPK family, although we cannot rule out the possibility that DUSP6 can affect other unknown proteins.
Loss of Dusp6 in our mice significantly enhanced baseline ERK1/2 phosphorylation in multiple tissues and in cultured MEFs, but it did not affect stimulus-induced ERK1/2 activation or its decay. DUSP-MKPs are typically negative feedback regulators of MAPK activity, where they become transcriptionally induced after a stress or mitogenic stimulation, leading to relatively rapid protein production that inactivates MAPKs within 20 -40 min. It was rather unexpected that the strength of stimulus-induced ERK1/2 activation and the timing of inactivation were unaffected in MEFs or hearts lacking Dusp6, yet baseline ERK1/2 phosphorylation was enhanced. Li et al., also identified an increase in baseline ERK1/2 phosphorylation in the developing limbs of four of eight embryos, although they did not employ a stimulation protocol (28). The most likely explanation for the lack of prolongation or enhanced ERK1/2 activity after stimulation in the absence of Dusp6 is that other DUSPs compensate and play a greater role following stimulation. Indeed, DUSP2, -4, -5, -7, and -9 also dephosphorylate ERK1/2, suggesting that one of more of these phosphatases could compensate for the lack of DUSP6 in affecting stimulus-induced ERK1/2 phosphorylation kinetics (10). Alternatively, DUSP6 might be more highly specialized for regulating the baseline tone in ERK1/2 signaling across most tissues, whereas other DUSPs regulate the duration and strength of ERK1/2 signaling after stimulation. In this manner compensation by other DUSP family members in the absence of Dusp6 may depend on the extent of overlapping expression in different tissues and even differences in subcellular localization of one DUSP family member compared with another.
One possible mechanism whereby loss of DUSP6 protein might alter the baseline tone in ERK1/2 signaling is suggested by the observation that DUSP6 and MEK1 each appear to interact with the same domains in ERK1/2 in a mutually exclusive manner (13). Such a mechanism could permit greater net activation of ERK1/2 at baseline through a shift in the equilibrium of MEK1 association. Of note, we did not detect any difference in MEK1 phosphorylation levels in Dusp6 Ϫ/Ϫ hearts, suggesting that the upstream activity through this pathway was not altered in the absence of DUSP6 protein (data not shown).
Loss of Dusp6 produced higher rates of myocyte proliferation in the heart during development, resulting in greater myocyte content in adulthood. Although it is difficult to directly quantify myocyte number in the heart at any single time point, the relative difference is easily ascertained by measurement of myocyte surface areas normalized to heart weight, provided the myocyte volume fraction is unchanged. For example, overexpression of calmodulin in the developing mouse heart by transgenesis promoted a 73% increase in myocyte number by embryonic day 19 and a 40% increase in myocyte number in the adult heart (44). Overexpression of either insulin-like growth factor 1 in the heart or Bcl-2 both increased myocyte number as measured by assessment of myocyte size normalized to heart weight and volume fractions (45,46). Interestingly, both of these mouse models with increased cell numbers were protected from cardiomyopathy, potentially suggesting that enhanced cell number in the heart has a protective influence (47,48). Indeed, overexpression of telomerase in the heart by transgenesis also produced greater myocyte content in the heart with some hypertrophy, which was protective against apoptotic insults (49).
Dusp6 Ϫ/Ϫ mice traversed more slowly into heart failure following long term pressure overload, characterized by less fibrosis, less pulmonary edema, less reductions in FS measured by echocardiography, and less TUNEL. Dusp6 Ϫ/Ϫ mice subjected to MI also maintained cardiac function better at 4 weeks and showed smaller scar sizes, suggesting less ongoing apoptosis. Indeed, Dusp6 Ϫ/Ϫ MEFs were partially protected from apoptotic stimuli in culture. Consistent with these data, we have previously shown that cardiac-specific inhibition of ERK1/2 signaling by DUSP6 overexpression predisposes the myocardium to failure following long term pressure overload, similar to Erk2 ϩ/Ϫ mice (34). We believe that these observations are due to an inherent protective effect associated with ERK1/2 signaling in the heart, such as antagonism of apoptosis during ongoing disease-inducing stimuli. However, following acute injury such as after I/R injury ERK1/2 are maximally activated nullifying any effect associated with increased basal ERK1/2 activity. Thus, loss of Dusp6 and the net increase in basal ERK1/2 phosphorylation are also protective by partially antagonizing apoptosis associated with chronic disease stimuli where ERK1/2 activity is dynamically regulated over time.