Cardiac-Specific Activation of IKK2 Leads to Defects in Heart Development and Embryonic Lethality

The transcription factor NF-κB has been associated with a range of pathological conditions of the heart, mainly based on its function as a master regulator of inflammation and pro-survival factor. Here, we addressed the question what effects activation of NF-κB can have during murine heart development. We expressed a constitutively active (CA) mutant of IKK2, the kinase activating canonical NF-κB signaling, specifically in cardiomyocytes under the control of the α-myosin heavy chain promoter. Expression of IKK2-CA resulted in embryonic lethality around E13. Embryos showed defects in compact zone formation and the contractile apparatus, and overall were characterized by widespread inflammation with infiltration of myeloid cells. Gene expression analysis suggested an interferon type I signature, with increased expression of interferon regulatory factors. While apoptosis of cardiomyocytes was only increased at later stages, their proliferation was decreased early on, providing an explanation for the disturbed compact zone formation. Mechanistically, this could be explained by activation of the JAK/STAT axis and increased expression of the cell cycle inhibitor p21. A rescue experiment with an IκBα superrepressor demonstrated that the phenotype was dependent on NF-κB. We conclude that activation of NF-κB is detrimental during normal heart development due to excessive activation of pro-inflammatory pathways.


Introduction
NF-κB is a pleiotropic transcription factor that has been associated with diverse biological functions such as cell proliferation, cell survival, immunity, and inflammation. Several pathological heart conditions, among them myocarditis, cardiac hypertrophy, adverse cardiac remodeling, and heart failure, have been associated with activation NF-κB signaling as well [1].
NF-κB is a dimeric transcription factor composed of different combinations of the subunits p50, p52, RelA, RelB, and c-Rel. Under baseline conditions, these NF-κB dimers remain inactive and sequestered in the cytoplasm by inhibitory κB (IκB) proteins. Different signals, such as engagement of cytokine or Toll-like receptors, trigger a signaling cascade that eventually converges on and activates the IκB kinase complex (IKK). This complex is composed of the catalytic subunits IKK1 (also known as IKKα) and IKK2 (IKKβ), and the regulatory subunit NEMO (IKKγ). IKK in turn can phosphorylate IκB proteins and thereby mark them for ubiquitin-mediated degradation. This releases NF-κB dimers, which are now free to translocate to the nucleus and activate specific target genes [2].
Murine heart development begins with the specification of cardiac progenitor cells at E6.0 and the formation of the cardiac crescent at E7.5 [3]. Beginning at E8.0, a linear heart tube is formed, sarcomeres are established, and a heart beat can be detected. After cardiac looping with the establishment of left-right asymmetry around E9.0, the heart tube differentiates into the four chambers and undergoes trabeculation and expansion (E10.5-E12.5), followed by further maturation of essential components such as the heart valves and the great vessels (E13.5 onwards).
Genetic and genomic approaches have identified a multitude of genes essential for mammalian heart development. In particular, they have revealed a complex network of cardiogenic transcription factors that orchestrate heart development [3]. Components of the IKK/ NF-κB signaling pathway appear not to be essential for heart development: For example, mice deficient in IKK2 (Ikbkb -/-) undergo normal heart development and only die due to hepatocyte apoptosis between E12.5 and E14.5 [4][5][6]. Similarly, a conditional ablation of IKK2 in the heart-using the myosin light chain 2V promoter which is specifically active in cardiomyocytes-does not affect normal heart development [7], neither does direct interference with the NF-κB pathway by expressing a transgenic IκBα superrepressor [8].
Still, IKK/ NF-κB signaling may be of great importance under pathologic conditions, since IKK/ NF-κB is activated during many adverse events that are common during gestation, such as infection. In this study, we wanted to address the consequences of the activation of IKK/ NF-κB in the developing heart as a model for adverse events during gestation.

Materials and Methods Mice
Mice were kept under specific pathogen-free conditions at the animal facility of the University of Ulm. The tetO.IKK2-CA [9], α-MyHC.tTA [10], and IκBα-3M mice [11] have been described previously. Animals were killed by cervical dislocation under isoflurane anesthesia. All experiments were in accordance with institutional guidelines and German animal protection laws and were approved by the regional government authority (Regierungspräsidium Tübingen).

Murine cardiomyocyte isolation
Excised hearts were digested in a calcium-free buffer supplemented with 6.5 mg/mL Liberase DH (Roche) and 3.5 mg/mL trypsin (Sigma). Digestion was stopped with FCS, and the cell suspension was filtered through a 100 μm mesh and centrifuged.

BrdU labeling and flow cytometry
Pregnant females were given BrdU (100 mg/ kg body weight i.p.) 4 hours before harvesting the embryos. FACS of digested whole hearts was performed on a FACSCanto™ II equipped with FACSDiva 6.2 software (BD).

RNA extraction and quantitative PCR
Heart tissue was snap-frozen in liquid nitrogen, pulverized and processed with the Qiagen RNeasy Fibrous Tissue kit. The Transcriptor High Fidelity cDNA Synthesis Kit (Roche) was used for cDNA synthesis. Quantitative PCR was done on a LightCycler 480 system, using the Universal Probe Library (Roche). Rpl13 was used as reference gene for relative quantification. Primer sequences are available on request.

Gene expression profiling
RNA quality was checked on the Agilent 2100 Bioanalyzer. 200 ng total RNA was amplified and labeled with the Whole Transcript Sense Target Labeling Assay (Affymetrix) using the GeneChip protocol. Labeled samples were hybridized to Affymetrix GeneChip 1 Mouse Gene 1.0 ST Array and further processed. Arrays were scanned with an Affymetrix GeneChip Scanner 3000 7G and data were analyzed by the RMA algorithm using the Affymetrix Expression Console and the GeneSifter 1 microarray data analysis system. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-3971 (http://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-3971).

Statistical analysis
Values are given as arithmetic mean + (or +/-) standard error of the mean (SEM). Means of two groups were compared by the student's t test or, when indicated, by the Mann-Whitney test. Welch correction was applied for unequal variances. Means of multiple groups were compared by ANOVA. All tests were two-tailed. p values <0.05 were deemed significant ( Ã p<0.05, ÃÃ p<0.01, ÃÃÃ p<0.001).

Expression of Constitutively Active IKK2 in the Heart Causes Embryonic Lethality
In order to activate NF-κB in the developing myocardium, constitutively active IKK2 (IKK2-CA) was expressed specifically in cardiomyocytes. To this end, mice expressing the tetracycline transactivator (tTA) under the control of the cardiomyocyte-specific α-myosin heavy chain (α-MyHC) promoter [10] were crossed with mice carrying a constitutively active IKK2 transgene (IKK2-CA) and a luciferase reporter gene regulated by a bidirectional, tTA-responsive promoter ( Fig 1A) [9]. As expected, expression of the IKK2 transgene was only seen in hearts of double transgenic (IKK MyHC ) embryos and could be silenced by the administration of doxycycline to pregnant mothers ( Fig 1B).
When mice were bred in the absence of doxycycline, i.e. under conditions that allow the transgene to be expressed, the number of double transgenic animals was strongly reduced in comparison to single transgenic and non-transgenic animals: Only 3.3% (instead of the expected 25%) of the born animals (4 out of 122) carried both transgenes (Table 1). When the transgene expression was shut off during development by administering doxycycline to the pregnant females, the ratio was normalized to the expected 25% [12]. These observations suggested that the expression of constitutively active IKK2 resulted in embryonic lethality.
To determine the time point of lethality, the viability of embryos was analyzed at different time points of gestation. IKK MyHC embryos started to die as early as E9.5 (Fig 2A). The exact time point of death was somewhat variable between individual embryos, but 50% of the animals were dead between E12.5 and E13.5. Less than 4% of IKK MyHC embryos survived longer than E17.5. Closer examination of the heart showed pericardial edema and hemorrhages at the ventricles by E10.5; in several cases the pericardial sac was filled with blood ( Fig 2B).
Since the lethality was observed as early as day E9.5, we confirmed by immunofluorescence that the transgene IKK2-CA was actually expressed at this early stage ( Fig 2C). Transgene expression was also detectable by Western blot at later time points and was stronger than endogenous IKK2 expression ( Fig 2D). Expression was restricted to cardiomyocytes, as shown by immunofluorescence costaining of sections at E11.5 with an antibody against the human IKK2 transgene and an antibody against the cardiomyocyte-specific protein cardiac troponin T ( Fig 2E). Next, we wanted to analyze whether IKK2-CA expression actually activated NF-κB signaling. Indeed, nuclear translocation of RelA, a marker for classical NF-κB activation, was found in cardiomyocytes of IKK MyHC embryos, whereas this was not the case for controls ( Fig 2F).

IKK MyHC embryonic hearts show defects in compact zone formation
Embryonic hearts are characterized by the presence of an inner trabecular and an outer compact zone. Systematic histological examination of the IKK MyHC embryonic hearts at E12.5 The    showed a significant reduction in the thickness of the compact zone: The compact zones of both ventricles comprised fewer cell layers in comparison to wild type animals (Fig 3A and  3B). Furthermore, measurement of ventricle circumference revealed that the hearts of IKK-MyHC animals were slightly enlarged ( Fig 3C).

IKK MyHC embryonic hearts show defects in the contractile apparatus
The enlargement of embryonic IKK MyHC hearts prompted us to examine the contractile apparatus more closely. As evident from sarcomeric α-actinin staining, hearts of IKK MyHC embryos at E12.5 exhibited a defective organization of the contractile filaments ( Fig 4A). Whereas control hearts showed the characteristic striped pattern of sarcomeric α-actinin, the staining of IKK MyHC hearts was much more diffuse, with only few cardiomyocytes retaining the regular striped pattern. Similarly, the staining for the Z disc protein desmin was more diffuse and scattered in IKK MyHC hearts as compared to control hearts (Fig 4B). Ultrastructural analysis by transmission electron microscopy confirmed the morphological abnormalities of IKK MyHC cardiomyocytes ( Fig 4C). Control embryos showed the typical pattern of myofibers with normally developed sarcomeres. In contrast, IKK MyHC animals exhibited generally thinner and more disorganized myofibers.

Embryonic IKK MyHC hearts exhibit inflammation
Since the IKK/ NF-κB axis is one of the master regulators of inflammation in adults, embryonic IKK MyHC hearts were analyzed for features of inflammation. Indeed, IKK MyHC hearts at E11.5 manifested a strong infiltration of macrophages, as visualized by staining for the macrophage marker F4/80 ( Fig 5A). Also, adhesion molecules like ICAM-1 were strongly expressed in cardiomyocytes ( Fig 5B). This indicated that IKK MyHC hearts were characterized by continuous recruitment of inflammatory cells.

Cardiomyocyte apoptosis is enhanced in embryonic IKK MyHC hearts at later stages
The reduced thickness of the compact zone of embryonic IKK MyHC hearts could be the result of enhanced cell death and/ or decreased proliferation. Embryonic hearts were examined for apoptosis at days E10.5, E12.5 and E14.5 using TUNEL (terminal deoxynucleotidyltransferasemediated UTP end labeling) assays (Fig 6A). Apoptosis was prominent in IKK MyHC embryos only at later time points, thus putting into question its causal role for the reduced thickness of (arrow, lower right); the pericardium is filled with blood (lower right   the compact zone. In contrast, proliferation appeared to be reduced already at day E11.5, as revealed by BrdU labeling of cardiomyocytes (Fig 6B).

Gene Expression Analysis of embryonic IKK MyHC hearts
For a more comprehensive understanding of gene expression changes in embryonic IKK MyHC hearts we performed microarray analysis of embryonic ventricles at day E12.5 (S1 Table). Constitutive activation of IKK/ NF-κB resulted in a interferon type 1 signature, with a strong upregulation of interferon-regulatory factors (in particular Irf7) and the excessive expression of interferon-inducible genes like Sca1 (Ly6a/e) and Bst2. The gene expression signature of embryonic IKK MyHC hearts also suggested a strong activation of Toll-like receptor (e.g. Tlr3) and Jak/ Stat signaling (e.g. Stat1, Stat2). Genes encoding chemokines (e.g. Ccl2), as well as cell adhesion molecules (e.g. Madcam1) were strongly expressed, reflecting the inflammation observed in IKK MyHC hearts. Fig 7 gives an overview of the deregulated cytokines and their validation via quantitative PCR. Several genes related to cardiac muscle architecture and function (e.g. triadin, titin, and dystrophin) were downregulated. Both apoptotic (such as Trail) and anti-apoptotic genes were upregulated in IKK MyHC hearts. Transcripts of several cell cycle regulators showed a stronger

Regulators of proliferation and differentiation are disturbed in embryonic IKK MyHC hearts
The activation of the Jak/ Stat axis suggested by the gene expression analysis was indeed confirmed on the protein level: Stat1 expression and phosphorylation was strongly enhanced in the IKK MyHC embryos at E12.5 (Fig 8A). Also, the cell-cycle inhibitor p21, a target of Stat1, was upregulated at the protein level in embryonic IKK MyHC hearts at E12.5 (Fig 8A).
In order to minimize the effect of infiltrating cells on these results, embryonic cardiomyocytes were isolated at E10.5, purified, and analyzed by quantitative PCR. Again, the transcript encoding p21 was upregulated (Fig 8B). In addition, the cyclin D1 transcript was downregulated.
Furthermore, we tried to elucidate the significance of the strong upregulation of the stem cell marker Sca1 (Ly6a) in the gene expression analysis. Indeed, Sca1-positive cells were a prominent feature of embryonic IKK MyHC , but not control hearts at day E11.5 ( Fig 8C). There was only a very limited overlap between Sca1-positive cells and F4/80-positive cells, suggesting that macrophages were not primarily responsible for the observed Sca1 expression (Fig 8C,  upper panel). In part, the Sca1-positive population also expressed the transgene IKK2-CA, suggesting that they constitute or contain a precursor population of cardiomyocytes with a possible function in regeneration (Fig 8C, lower panel). The Embryonic Lethality of IKK MyHC Embryos Is Dependent on NF-κB Finally, we wanted to address the question whether the embryonic lethality of IKK MyHC embryos was in fact due to NF-κB activation or potential other pathways activated by the IKK complex. For this end, we crossed IKK MyHC animals with animals expressing a mutated form of IκBα -(IκBα (S32A, S36A, Y42F) ), which is impervious to phosphorylation-induced degradation and thus acts as a superrepressor of NF-κB signaling [11]. Indeed, whereas only very few IKK MyHC animals survived embryonic and postnatal development until weaning (constituting 1.3% of the surviving offspring instead of the expected 12.5%), IKK MyHC animals expressing the IκBα -superrepressor did survive at almost normal Mendelian ratios (constituting 9.1% of the surviving offspring instead of the expected 12.5%), thus supporting the notion that IKK2 exerts its lethal effects via NF-κB signaling ( Table 2).

Discussion
Previously, we showed that activation of IKK/ NF-κB signaling in the adult animal led to inflammatory cardiomyopathy and heart failure [12]. Our present study now expands on these results by illuminating the consequences of IKK2 activation under the specific conditions of embryonic heart development.
Cardiac-specific expression of a constitutively active IKK2 mutant during mouse development resulted in embryonic lethality. The rather long time window during which the embryos died, ranging from E9.5 (shortly after the beginning of the activity of the α-myosin heavy chain promoter) through E16.5 with half of the embryos dead by E12.5, suggests that IKK2 activation does not interfere with a specific, chronologically strictly defined event, but rather with the general conditions required for normal heart development during a longer period of time. Alternatively, the long time window might be due to variability in the strength of transgene expression and the number of cells that express it. In any case, the rescue experiments with an IκBα superrepressor suggest that the lethality IKK2 exerts is mediated by NF-κB. The most prominent finding in IKK MyHC hearts was certainly the strong inflammatory response observed upon IKK/ NF-κB activation. The inflammatory infiltrate was characterized by F4/80-positive cells and thus monocytes/ macrophages. Gene expression analysis revealed that the signature observed in IKK MyHC hearts very much resembled an interferon type I response as observed in an anti-viral response. This is consistent with the notion that IKK/ NF-κB signaling is central to the development of myocarditis, both virally induced and autoimmune [14][15][16][17]. The activation of the transcription factor STAT1, which is prominent in our model, is also known to be an early event in viral myocarditis [18].
In addition to the inflammation observed in IKK MyHC hearts, the defect in compact zone formation was a prominent finding. Several biological processes could explain this defect-in particular an increase in cell death or a decrease in proliferation. Whereas cell death was indeed increased in IKK MyHC embryos, this was a rather late phenomenon and therefore most likely not responsible for the disease process. In contrast, proliferation was reduced in IKK MyHC embryonic hearts already at an earlier time point, suggesting that it might be responsible for the thinning of the compact zone. Indeed, proliferation is a crucial event during this time of heart development, and any disturbance might have catastrophic consequences for the developing heart [3]. This is of course much in contrast to the observations in the adult animals, where proliferation is of minor or no relevance due to the largely post-mitotic status of the adult heart [12].
We also identified a potential link between the observed inflammation and the reduction in proliferation: Among the pathways activated in this response is the JAK/ STAT pathway, which via its target gene p21 might directly inhibit proliferation [19]. The cause of the decrease in proliferation might thus indeed lie in the inflammatory environment that is so characteristic for IKK MyHC hearts.
An interesting finding in IKK MyHC embryonic hearts is the emergence of a Sca1-positive cell population. Recently, Sca1-positive cells have been identified as a source for myocardial renewal, albeit in the adult heart [20]. It remains to be seen whether the cells observed in IKK-MyHC embryonic hearts fulfill a similar and thus beneficial role, or whether they in contrast represent a lack of differentiation of cardiomyocyte progenitors, or even a dedifferentiation of cardiomyocytes. Dedifferentiation might indeed be one of the effects of IKK2, since the contractile apparatus, a hallmark of cardiomyocyte differentiation, was severely affected upon IKK/ NF-κB activation as shown by electron microscopy and immunofluorescence analyses of typical markers such as sarcomeric actinin.
Since infections are a major cause for adverse events during gestation in humans, our findings might be of relevance for a better understanding of the molecular processes in affected embryos and fetuses. According to our study, IKK/ NF-κB is a key regulator of the anti-viral response in cardiomyocytes, a finding that is very interesting in the light that viral infections (e.g. by CMV, parvovirus B19, and rubella virus) are a major problem during gestation.
Supporting Information S1 Table. Microarray analysis reveals widespread changes in gene expression in IKK MyHC embryos. The table shows the full name, the gene ID and the fold induction of genes, arranged according to their function. (XLS)