Epinephrine (EPI) and norepinephrine (NE) are the major peripheral catecholamines produced in mammals. They are synthesized by means of the enzymatic conversion of dopamine to NE by the action of dopamine β-hydroxylase (Dbh) and, subsequently, of NE to EPI by the action of phenylethanolamine n-methyltransferase (Pnmt; Cooper et al., 1991). In adult mammals, this synthesis occurs primarily in the adrenal medulla and in the sympathetic nervous system. Mice that lack the ability to produce NE and EPI due to targeted disruption of the Dbh gene die in utero from apparent cardiac failure (Thomas et al., 1995).
The vast majority of Dbh−/− mouse embryos die before the adrenal gland even forms and before maturation of the sympathetic nervous system, suggesting that the primary source(s) of catecholamine biosynthesis in the embryo must be different from that in adult mice (Thomas et al., 1995). One potential source is the heart itself, where we and others have shown that intrinsic cardiac adrenergic (ICA) cells are present beginning at relatively early stages (embryonic day [E] 9.5 in rat; E3 in chick) of cardiac development (Ignarro and Shideman, 1968; Ebert et al., 1996; Huang et al., 1996; Ebert and Thompson, 2001). A specific developmental role for these ICA cells remains elusive, but we have shown recently that they are transiently and progressively associated with regions of the heart that become pacemaking and conduction tissue (Ebert and Thompson, 2001). This association was initially observed in the sinoatrial node (SAN) and atrioventricular canal regions at approximately E11.5 in the rat. The clustering of ICA cells in these regions then declined over the next several days of development but reappeared along the crest of the ventricular septum (bundle of His region) and, somewhat more sporadically, trailing down the ventricular septum toward the apex at approximately E16.5 (Ebert and Thompson, 2001). These patterns of ICA cell distribution were transient, lasting not more than 1–2 days in each of these locations (Ebert and Thompson, 2001). It was not clear if this progressive appearance and disappearance of ICA cells in these various cardiac regions was due to migration, differentiation, and/or death of ICA cells.
To help resolve these possibilities, we have developed a novel mouse genetic model. The Cre-recombinase gene was inserted into the endogenous Pnmt locus (Pnmt-Cre) so that Cre-recombinase expression would be directed by Pnmt regulatory DNA sequences. Upon crossing these Pnmt-Cre mice to the reporter R26R strain of mice, we anticipated activation of LacZ expression exclusively in adrenergic cells by means of Cre-mediated recombination of upstream loxP sites flanking a DNA sequence responsible for blocking transcription of the LacZ gene (Soriano, 1999). In this system, lacZ expression does not require the continuous activation of Pnmt-Cre. Rather, the genetic changes induced at the Rosa26 locus by transient expression of the Cre-recombinase permanently mark the altered cells and their descendants as β-galactosidase positive and, therefore, provide a means to follow the distribution of adrenergic-derived cells throughout development. In addition, because the targeted insertion of Cre-recombinase disrupted the expression of the Pnmt gene, we could assess the potential consequences of selective adrenergic deficiency in these mice.
To determine whether epinephrine plays an important role during development and to identify adrenergic cell descendants in the developing mouse, we inserted the Cre-recombinase gene into the mouse Pnmt locus so the expression of Cre-recombinase would be dependent upon Pnmt regulatory sequences (Fig. 1A). Specifically, we inserted the Cre-recombinase gene into exon 1, creating a gene fusion of the 5′ leader sequences from Pnmt directly to Cre-recombinase open reading frame. This strategy was designed to disrupt functional expression of Pnmt while permitting expression of Cre-recombinase exclusively in adrenergic cells. Verification of the correct targeting of the Cre-recombinase insert was confirmed by Southern blotting using 5′ and 3′ probes each external to the targeting vector (Fig. 1B).
Two of the correctly targeted clones were further propagated and introduced into mouse blastocysts, and independent lines of founder mice were produced. These founder mice were crossed with wild-type C57B/6 females to generate F1 animals used in all our subsequent studies. Germ-line transmission of the Pnmt-Cre allele was verified initially by Southern blotting and then by polymerase chain reaction (PCR; Fig. 1C). When F1 heterozygotes were intercrossed, mice harboring the disrupted Pnmt allele were born in normal Mendelian ratios. 35 +/+: 75 +/−: 46 −/− were the actual numbers of mice born as compared with 39, 78, and 39 mice expected for each genotypic class from 156 total progeny (P = 0.410 using the Chi−squared test). The −/− mice appear overtly normal and are fertile even in −/− × −/− crosses.
The absence of Pnmt expression was predicted to result in mice that cannot produce epinephrine. To test this hypothesis, we measured EPI and NE concentrations in adrenal extracts from Pnmt+/+, Pnmt+/Cre, and PnmtCre/Cre mice by radioimmunoassay (Table 1). EPI was not detectable in extracts from PnmtCre/Cre mice, and the concentration of NE was found to be significantly greater in PnmtCre/Cre relative to Pnmt+/+ and Pnmt+/Cre extracts (P < 0.001). In contrast, NE and EPI concentrations were similar in extracts from Pnmt+/+ and Pnmt+/Cre mice. In addition, Pnmt immunofluorescent histochemical staining was virtually absent in the adrenal medullae of PnmtCre/Cre mice but could be readily detected in Pnmt+/Cre and Pnmt+/+ mice (data not shown).
To map the developmental distribution of Pnmt-Cre–expressing cells and their descendants, mice heterozygous for the insertion (Pnmt+/Cre) were mated with R26R mice, homozyogous at the ROSA26 reporter locus. Cells in R26R mice have the ability to robustly express the bacterial LacZ (β-galactosidase) gene if activated by Cre-recombinase (due to removal of a floxed transcriptional block; Soriano, 1999). To determine whether our system was working, we first examined β-galactosidase staining in adrenal glands where the expression patterns of Pnmt are well-documented and relatively simple. As illustrated in Figure 2A, Pnmt immunofluorescent staining is concentrated in the adrenal medulla. An adjacent section stained with XGAL to detect LacZ expression (blue cells) produced a nearly identical pattern of staining (Fig. 2B). When Pnmt immunofluorescent staining was performed on this section after the XGAL staining, little or no fluorescently stained cells were observed, because the XGAL blocked or quenched the fluorescence in costained cells (Fig. 2C). Thus, the localization of Pnmt and β-galactosidase appear to be completely coincident here. As a control, the same sections were also stained for tyrosine hydroxylase (Th), the first enzyme in the catecholamine biosynthetic pathway, to independently identify chromaffin cells (Fig. 2D–F). As expected, Th staining was similar to Pnmt and XGAL staining in these sections, but there were a few cells that stained positively for Th that were not stained with XGAL (Fig. 2E,F, arrows). These Th+LacZ− cells presumably represent the minority of chromaffin cells that are known to be noradrenergic rather than the adrenergic majority (Coupland and Tomlinson, 1989; Ebert and Thompson, 2001). Together, these results demonstrate that LacZ expression was highly selective in adrenal sections where it was confined to adrenergic (Pnmt+) cells.
To evaluate Pnmt-Cre expression in the developing mouse embryo, we performed whole-mount staining with XGAL. The earliest stage of development when LacZ expression could be detected was E8.5. LacZ expression was observed in the heart (Fig. 3A) and along the dorsal ridges of the neural folds in the region of the midbrain–hindbrain junction (Fig. 3B). Within the heart, there appears to be a small cluster of LacZ+ cells in the sinus venosus region (Fig. 3A, arrow), with a few additional LacZ+ cells apparent in the ascending loop and also near the atrioventricular junction (see arrowheads, Fig. 3A).
Over the next 2 days, LacZ expression became much more widespread in the heart, as shown in Figure 4A. By this time, LacZ expression was also evident in the brainstem region, 2nd branchial arch, and sporadically along the dorsal surface of the embryo. LacZ expression was dependent upon Cre-recombinase expression from the Pnmt locus, because none was observed when wild-type embryos were similarly crossed with R26R mice (Fig. 4B).
Although it appears from the whole-mount staining (Fig. 4A) that the entire heart was expressing LacZ at this stage of development, examination of heart sections from these embryos revealed that the true staining pattern was more restricted (Fig. 5). Sporadic staining of cardiac cells was observed in the atrial and ventricular regions, including cells in the trabecular region of the ventricle. As shown in Figure 5, LacZ expression was observed in the SA junction region (Fig. 5A,B); however, the strongest LacZ staining was found near the antrioventricular and conoventricular junction regions (Fig. 5C,D). These LacZ expression patterns were consistently observed at this stage of development, and they appear similar to the pattern of catecholamine-producing cells previously reported at a comparable developmental stage in the rat (Ebert and Thompson, 2001).
The pattern of LacZ expression in the heart appears to be restricted to myocardial cells. To further evaluate this observation, we performed triple-labeling experiments whereby Pnmt-Cre/R26R heart sections (E9.5) were stained for expression of (1) a muscle-specific marker, sarcomeric α-actinin (red fluorescence, Fig. 6A), (2) LacZ as a marker for cells descended from an adrenergic lineage (blue XGAL staining, Fig. 6B), and (3) the adrenergic cell marker Pnmt (green fluorescence, Fig. 6C). Extensive overlap of these three staining patterns could be observed by merging the individual staining panels into a single “overlay” image, as shown in Figure 6D. The yellow regions represent cells that coexpress α-actinin and Pnmt (arrows point to some examples of overlapping expression patterns). Some cells are clearly red or green, but many yellow cells can also been seen in this image. Notably, most of the exclusive green (Pnmt+/α-actinin−) cells in the overlay image (Fig. 6D) were found just outside the heart, in both the anterior region adjacent to the outflow tract as well as in the posterior region adjacent to the atrial chamber (indicated by the arrowheads, Fig. 6C,D). The LacZ+ cells (blue) were again found concentrated at junction regions, specifically between the atrial, ventricular, and outflow tract cavities.
By E15.5, the heart has matured morphologically to assume its adult-like appearance. To determine whether Pnmt-expressing cells contribute to pacemaker cells in the SAN, we performed a preliminary immunofluorescent histochemical staining analysis of wild-type E15.5 mouse embryo sections using an antibody that recognizes the Hyperpolarization-activated cyclic nucleotide-modulated channel isoform 4 (HCN4), a major pacemaker channel protein expressed in the SAN (Garcia-Frigola et al., 2003; Stieber et al., 2003). As shown in Figure 7A, HCN4 staining is highly restricted to the SAN region of the right atrium (arrow) and to a lesser extent, to the atrioventricular node (AVN) region (arrowhead). In contrast, sarcomeric α-actinin was found to be expressed throughout the myocardium in this section (Fig. 7B). These data confirm that HCN4 was localized to pacemaker cells in the developing mouse heart.
As shown in the series of sections from a Pnmt-Cre/R26R embryo, LacZ expression not only persists in the heart at E15.5 but is highly pervasive throughout the myocardium in all four chambers (Fig. 7C–E). By far, the strongest and most extensive labeling was found in the ventricular septum, with the most prominent staining appearing at the crest of the septum and extending laterally into both ventricular chambers along the base as well as caudally down the septum toward the apex. LacZ expression in the free walls of both chambers was considerably more sparse than that found in the septum, with a tendency to be localized in cells near the endocardial surface, although LacZ+ cells were also clearly seen in the mid- and epicardial regions of the ventricular myocardium. There appeared to be little or no LacZ expression in the aorta, aortic valves, and mitral valves, all of which can be clearly seen as nonstained tissue in Figure 7D; however, the myocardial region immediately adjacent to the intake region of each of these valves was strongly labeled with blue XGAL stain.
Within the atria, LacZ expression was sporadic, with the majority of atrial myocytes appearing as nonstained cells. The exception to this pattern of atrial staining appeared in the SAN region where a strong clustering of blue LacZ+ cells was evident (Fig. 7E,F). To show that these cells were SAN myocytes, we performed immunofluorescent histochemical staining of the sections with the anti-HCN4 antibody. As shown in Figure 7G,H, HCN4 expression is robust in this region and overlaps extensively with the XGAL staining. These results demonstrate that many, but not all, of the presumptive pacemaker cells (HCN4+) in the SAN also expressed LacZ in these Pnmt-Cre/R26R mice.
Although LacZ expression is found in pacemaker cells, as we have just shown, it is clear from the data presented thus far that LacZ expression in the heart extends well beyond the SAN region. By E15.5, most of the LacZ-expressing cells are found throughout the myocardium, as indicated above (see Fig. 7C–E).
This pattern of LacZ expression persisted in the heart through at least the early postnatal period. As shown in Figure 8, extensive XGAL labeling was observed throughout the heart, and was especially dense in the ventricular septum. Many of the pacemaker cells (HCN4+) were also stained blue with XGAL (Fig. 8B–D). Endogenous Pnmt expression in the SAN region was extremely weak at this stage of development (Postnatal day 2, P2), being at best confined to a small cluster of cells (see arrowhead, Fig. 8E).
EPI-Deficient Mice Appear Normal
In this report, we have described the initial characterization of a novel mouse model for studying adrenergic cell function and distribution. Targeted disruption of the Pnmt gene led to the production of mice that were deficient in their ability to produce EPI. These mice were still able to produce NE. Thus, these mice are a model for investigating the specific role of epinephrine in mouse biology.
There was no overt developmental phenotype associated with EPI deficiency in that homozygous mice lacking EPI survived to adulthood, were physically indistinguishable from wild-type and heterozygous littermates, and were able to breed effectively. We have not as yet explored more subtle physiological or behavioral phenotypes in these mice, but one might expect that they may have limitations in their ability to mediate stress responses, because EPI is one of the primary physiological “stress” hormones.
Mapping Adrenergic Cells and Their Descendants
The earliest expression of LacZ in Pnmt-Cre/R26R mice occurred around E8.5 in the region of the neural folds and in the heart. This essentially simultaneous activation of LacZ in these two disparate regions of the embryo suggests that adrenergic cells arise independently from neural and cardiac precursors. A neural origin for adrenergic cells is well-established in the literature, as Pnmt-expressing adrenal chromaffin cells are known to be derived from neural crest (Bohn, 1983; Anderson, 1993). In addition, Pnmt-expressing cells are also known to contribute to specific populations of brainstem and retinal neurons (Kopp et al., 1979; Armstrong et al., 1982; Bohn et al., 1986; Sawchenko and Bohn, 1989; Hammang et al., 1992). We did not examine the retina in this study, but we did show that LacZ+ cells were present in the brainstem region as early as E8.5 (Fig. 3). We have also observed LacZ+ cells in the neonatal brainstem (not shown). Taken together, the LacZ+ staining patterns in the developing embryo and neonate are consistent with a neurogenic origin of these cells in the brain and adrenal gland.
In contrast, Pnmt+ cells in the heart appear to have a cardiogenic rather than a neurogenic origin, because they appear independently in the heart and neural fold regions by E8.5. In support of this hypothesis, we demonstrated that Pnmt-expressing cells were found in the presumptive posterior and anterior heart-forming regions at E9.5 (Fig. 6). Although it is generally well-accepted that the vast majority of myocardial cells arise “from paired heart-forming fields at the venous pole” (Kelly and Buckingham, 2002), several recent studies from independent laboratories have shown that there is also an anterior (arterial) pole from which myocardial cells are derived (Waldo et al., 2001; Mjaatvedt et al., 2001; Kelly et al., 2001). As was suggested from early data published in the 1970s (Viragh and Challice, 1973; de la Cruz et al., 1977), myocardial cells derived from the anterior heart-forming region primarily populate and extend the outflow tract. Kelly et al. (2001) showed that at least part of the anterior heart-forming field emanated from the pharyngeal arch region between E8.25 and E10.5 in the mouse. Our data identified LacZ+ cells in the pharyngeal arch region during that window of development and showed that many adrenergic-derived myocardial cells were present in the developing outflow tract. These data suggest that adrenergic cells contribute to myocardial development from the anterior as well as the posterior heart-forming fields. Pnmt expression in extracardiac (α-actinin–negative) cell populations at both poles of the tubular heart may indicate expression of Pnmt just before commitment to the myocardial phenotype, a potentially important observation in light of earlier studies (Claycomb, 1976, 1977; Hoffman, 2001), or merely differential sensitivity of these two antibodies as applied here. Although there may be some contributing Pnmt+ cells derived from neural crest in the heart after approximately E10.5, the LacZ staining patterns reported here are not consistent with established cardiac neural crest patterns in the developing heart (Waldo et al., 1999; Kirby, 1999; Jiang et al., 2000). Clearly, further work is needed to resolve this issue and to definitively determine whether the adrenergic lineage hypothesis of cardiomyocyte development proposed here is valid.
Pnmt-Expressing Cells Become Pacemaking and Conduction System Myocytes
We previously showed that ICA cells were transiently and progressively associated with pacemaking and conduction system regions (SAN, AVN, bundle of His, and Purkinje fibers) within the embryonic rat heart (Ebert and Thompson, 2001). The results from the present study in the mouse confirm and extend these findings by showing that a subset of pacemaking cells in the SAN and AVN were derived from adrenergic cells because approximately half of the pacemaker myocytes that stained positively for the pacemaker channel protein, HCN4, also stained positively with XGAL. Although we do not have an antibody marker that selectively recognizes His bundle or Purkinje fiber myocytes, the extensive XGAL labeling of the ventricular septum, especially along the crest and the descending lateral branches, is consistent with the expected pattern of ventricular conduction system myocytes (Viragh and Challice, 1977; Moorman et al., 1998; Rentschler et al., 2001).
LacZ expression in Pnmt-Cre/R26R embryonic mouse hearts was definitely not confined to the pacemaking and conduction systems. In fact, the vast majority of cardiac cells derived from Pnmt-expressing precursors appear to be nonpacemaking myocytes. At present, there is no evidence to suggest that Pnmt-expressing cells give rise to nonmyocardial cells within the heart, but further costaining experiments with additional cell-specific markers (e.g., for fibroblasts, smooth muscle, neuronal, endocardial, and epicardial cells) needs to be performed in this mouse model before such a determination can be made. This caveat notwithstanding, our data suggest that Pnmt is a novel marker for a subpopulation of developing myocytes in the heart. Cardiac cells that express Pnmt may ultimately become working myocardial cells in both atria and ventricles as well as in specialized myocytes constituting the cardiac pacemaking and conduction systems.
Creation of Pnmt-Cre Knock-In/Knockout Mice
To construct targeting vector pKP506, we first isolated and subcloned a 7-kb SacI fragment from bacterial artificial chromosome 238M20 (Genome Systems). This fragment carries Pnmt exon 1 and includes 3.3 kb upstream of the ATG initiation codon and also 3.7 kb of sequences 3′ of the translation start site. A 1.4-kb SexAI fragment carrying exon 1 was next subcloned and mutagenized by using the Gene Editor Mutagenesis Kit (Promega) to add a unique EcoRI site in place of the normal Pnmt translation initiation codon. A cassette carrying the Cre-recombinase gene fused to a nuclear localization signal was then inserted at this unique EcoRI site so that the expected mRNA product is a fusion of the Pnmt 5′ untranslated leader to sequences encoding the Cre-recombinase protein. The cassette also included NeoR sequences flanked by FRT recombination sites. This modified Sex AI fragment was then exchanged with its wild-type cognate on the SacI clone. Finally the diphtheria toxin A gene was added adjacent to the 5′ Pnmt flank for negative selection.
Forty-five micrograms of linearized vector DNA were electroporated into RI mouse embryonic stem cells. G418-resistant clones were isolated and their DNAs characterized by Southern blotting using probes external to the targeting vector (Fig. 1). Cells from two positive clones were injected into C57BL/6 blastocysts and chimeric mice from these injections were mated to C57BL/6 females. Genotypes at the Pnmt locus were identified by PCR analysis by using the following primers: primer 21, 5′-CAG GCG CCT CAT CCC TCA GCA GCC-3′; primer 22, 5′-CTG GCC AGC GTC GGA GTC AGG GTC-3′; and primer 23, 5′-GGT GTA CGG TCA GTA AAT TGG ACA CCG TCC TC-3′. These amplification reactions yield products of 200 and 160 bp for the wild-type and the mutant alleles, respectively. PnmtCre/+ mice were then intercrossed or mated to R26R reporter mice as described in the text.
Epinephrine and norepinephrine concentrations were measured by radioimmunoassay as described previously (Ebert and Thompson, 2001). Briefly, adrenal glands were isolated from decapitated neonates and immediately frozen on dry ice. Individual adrenal gland extracts were prepared by sonication for 10 sec in 0.1 M HCl, followed by microcentrifugation (14,000 × g, 10 min) to remove residual debris. The extracts were assayed using a commercially available radioimmunoassay (ALPCO Labs, Inc., Wyndham, NH), and protein concentrations were determined by using the Bio-Rad (Hercules, CA) protein assay. Equivalent amounts of protein (5 ng/sample) were assayed in a volume of 0.25 ml. Results are expressed as mean ± SEM and compared for statistically significant differences by one-way analysis of variance with P < 0.05 required to reject the null hypothesis.
Whole-Mount Staining for β-Galactosidase Activity
Embryos were staged by timed matings. Females carrying a vaginal plug were removed from the mating cage, and noon of that day was considered to be E0.5. Isolated embryos were fixed 50 min at 4°C in PBS containing 2% paraformaldehyde (w/v) and 2% glutaraldehyde (v/v). Embryos were rinsed at 4°C with phosphate buffered saline (PBS) three times for 30 min each and then incubated in PBS containing 1 mg/ml XGAL (5-bromo-4-chloro-3-indoyl-beta-D-pyranoside), 5 mM ferricyanide, 5 mM ferrocyanide, 2 mM MgCl2, and 0.2% Ipegal CA-630 (Sigma). After 15 hr at 30°C, the embryos were rinsed in PBS containing 3% dimethyl sulfoxide and then stored at 4°C in PBS.
Immunofluorescent Histochemical Staining
Coimmunofluorescent histochemical staining was performed essentially as described previously (Ebert and Thompson, 2001). Primary antibodies used in this study include rabbit polyclonal antibodies from Chemicon International (Temecula, CA) that selectively recognize either Pnmt (AB110) or HCN4 (AB5808) and mouse monoclonal antibodies from Sigma-Aldrich (St. Louis, MO) that selectively recognize either Th (T1299) or sarcomeric α-actinin (A7811). Fluorescently tagged secondary antibodies were obtained from Jackson Immunolabs (Denver, PA), and include fluorescein isothiocyanate–conjugated donkey anti-rabbit IgG and Texas Red–conjugated donkey anti-mouse IgG. All antibodies were used at 1:100 dilution, except for the anti-HCN4 antibody, which was used at 1:50 dilution.
Embryos were isolated from timed-pregnant females and fixed with freshly prepared 4% paraformaldehyde in 0.1 M PBS, pH 7.3 for 1–2 hr on ice. The samples were then transferred to 30% sucrose solution in PBS, pH 7.3, and allowed to equilibrate at 4°C for at least 48 hr. The sucrose-saturated samples were then shipped to FD Neurotechnologies, Inc. (Catonsville, MD) for sectioning (20 μm) and mounting onto SuperFrost Plus microscope slides (Fisher Scientific, Inc., Pittsburgh, PA). The sections were then frozen and stored at −80°C for subsequent use.
For experiments where XGAL staining and immunofluorescent labeling were combined, we first performed the XGAL staining on the sections overnight as described above. The XGAL solution was then removed, and the sections were washed with PBS (3 × 10 min each) before initiating the immunofluorescent staining as described in the preceding paragraph. Digital images were collected by using a Nikon Eclipse E1000 fluorescence microscope and processed for display by using Adobe Photoshop 5.5 software.