Inhibition of heart formation by lithium is an indirect result of the disruption of tissue organization within the embryo


  • Lisa K. Martin,

    1. Department of Regenerative Medicine and Cell Biology, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Momka Bratoeva,

    1. Department of Regenerative Medicine and Cell Biology, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Nadejda V. Mezentseva,

    1. Departments of Physiology and Medicine, New York Medical College/Westchester Medical Center Stem Cell Laboratory, New York Medical College, Valhalla, New York, USA
    Search for more papers by this author
  • Jayne M. Bernanke,

    1. Department of Regenerative Medicine and Cell Biology, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Mathieu C. Remond,

    1. Departments of Physiology and Medicine, New York Medical College/Westchester Medical Center Stem Cell Laboratory, New York Medical College, Valhalla, New York, USA
    Search for more papers by this author
  • Ann F. Ramsdell,

    1. Department of Regenerative Medicine and Cell Biology, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Carol A. Eisenberg,

    1. Departments of Physiology and Medicine, New York Medical College/Westchester Medical Center Stem Cell Laboratory, New York Medical College, Valhalla, New York, USA
    Search for more papers by this author
  • Leonard M. Eisenberg

    Corresponding author
    1. Departments of Physiology and Medicine, New York Medical College/Westchester Medical Center Stem Cell Laboratory, New York Medical College, Valhalla, New York, USA
    Search for more papers by this author

Author to whom all correspondence should be addressed.


Lithium is a commonly used drug for the treatment of bipolar disorder. At high doses, lithium becomes teratogenic, which is a property that has allowed this agent to serve as a useful tool for dissecting molecular pathways that regulate embryogenesis. This study was designed to examine the impact of lithium on heart formation in the developing frog for insights into the molecular regulation of cardiac specification. Embryos were exposed to lithium at the beginning of gastrulation, which produced severe malformations of the anterior end of the embryo. Although previous reports characterized this deformity as a posteriorized phenotype, histological analysis revealed that the defects were more comprehensive, with disfigurement and disorganization of all interior tissues along the anterior-posterior axis. Emerging tissues were poorly segregated and cavity formation was decreased within the embryo. Lithium exposure also completely ablated formation of the heart and prevented myocardial cell differentiation. Despite the complete absence of cardiac tissue in lithium treated embryos, exposure to lithium did not prevent myocardial differentiation of precardiac dorsal marginal zone explants. Moreover, precardiac tissue freed from the embryo subsequent to lithium treatment at gastrulation gave rise to cardiac tissue, as demonstrated by upregulation of cardiac gene expression, display of sarcomeric proteins, and formation of a contractile phenotype. Together these data indicate that lithium’s effect on the developing heart was not due to direct regulation of cardiac differentiation, but an indirect consequence of disrupted tissue organization within the embryo.


Despite considerable difference in the anatomy and physiology of the mature mammalian, avian, amphibian, and fish heart, the morphological and molecular mechanisms that initiate cardiac development among vertebrate classes are very similar (Lohr & Yost 2000; Mohun et al. 2000; Moorman et al. 2003; Wessels & Sedmera 2003; Gittenberger-de Groot et al. 2005). Heart formation in the vertebrate embryo begins during the onset of gastrulation. Cells fated to become heart localize to paired mesodermal regions within the anterior half of the embryo. In the frog gastrula, the area containing the heart-forming regions is referred to as the dorsal marginal zone (DMZ) (Afouda & Hoppler 2009). As development proceeds, these bilateral precardiac fields merge at the midline to give rise to the primary heart tube (Eisenberg & Eisenberg 2002; Eisenberg et al. 2004; Abu-Issa & Kirby 2007). Initial steps in the cardiac differentiation of the mesoderm are marked by the expression of the transcription factors Nkx2.5, Tbx5, serum response factor (SRF), MEF2C, and GATA6, which are displayed before the heart forming fields merge and myofibrillar proteins are expressed (Lyons et al. 1995; Grow & Krieg 1998; Horb & Thomsen 1999; Mori & Bruneau 2004; Brown et al. 2005; Burch 2005; Srivastava 2006; Warkman & Krieg 2007). These early events in cardiac development are regulated by multiple secreted signaling molecules, including insulin-like growth factors (IGF), transforming growth factor (TGF)-α, cripto, fibroblast growth factors (FGFs), bone morphogenetic proteins (BMPs), and Wnts (Antin et al. 1996; Eisenberg & Markwald 1997; Xu et al. 1999; Brand 2003; Eisenberg & Eisenberg 2006, 2007; Abu-Issa & Kirby 2007; Wagner & Siddiqui 2007).

The positioning of the heart in the upper chest cavity results from developmental events predating gastrulation that determine the proper orientation of the embryonic axis (Lohr & Yost 2000). The animal model that has provided the greatest insights into the process underlying the formation of the primary axis has been Xenopus laevis (Wakahara 1989; Moon & Kimelman 1998), due to the accessibility of the frog embryo at the earliest stages of development. Prominent among the molecular pathways that regulate axis formation is the canonical Wnt pathway, whose activity instructs the polarity of the primary axis (Larabell et al. 1997; Tao et al. 2005). In the canonical Wnt pathway, transduced signal resulting from Wnt/receptor interaction inactivates glycogen synthase kinase 3β (GSK3β), which leads to the cellular accumulation of β-catenin. In turn, β-catenin forms a transcriptional enhancer complex with LEF/TCF DNA binding proteins, which upregulate expression of selected target genes (Eisenberg & Eisenberg 2006, 2007; Rao & Kuhl 2010). The key occurrence in this pathway is the inhibition of GSK3β, from which is elaborated the downstream signals that mediate Wnt regulation (Wu & Pan 2010). Because of the importance that Wnt signal transduction plays in various developmental processes, cancer, stem cell biology, and neural function (Nusse 2008; Cadigan & Peifer 2009; Wend et al. 2010), several pharmacological reagents that regulate GSK3β have been investigated for their therapeutic potential (Luo 2009; Mishra 2010). Among these reagents is lithium, whose biological impact was first recognized over 100 years ago.

In medicine, lithium is commonly used for the treatment of bipolar disorder (Williams & Harwood 2000; Quiroz et al. 2004). At high doses, lithium becomes teratogenic (Smithberg & Dixit 1982; Ghatpande et al. 1993). This latter property has proven advantageous for the study of embryonic development, as lithium has provided a useful tool for uncovering cellular and molecular mechanisms of early vertebrate pattern formation (Moon & Kimelman 1998; Heasman 2006). Exposure of the early frog blastula to high concentrations of lithium produces hyperdorsalized embryos, which consist solely of a radially symmetrical head. Injection of lithium into the ventral side of the early blastula promotes formation of a secondary axis (Kao et al. 1986; Kao & Elinson 1989). In this regard, the action of lithium resembles the effect of Wnt overexpression, which promotes a full secondary axis when ectopically expressed in ventral blastomeres (McMahon & Moon 1989; Sokol et al. 1991).

Several molecules have been shown to be pharmacological targets of lithium, which specifically blocks the enzymatic activities of inositol monophosphate phosphatase (IMPase), inositol polyphosphate phosphatase (IPPase), bisphosphate nucleotidase, fructose 1,6-bisphosphastase, phosphoglucomutase, GSK3α, and GSK3β (Klein & Melton, 1996, Williams & Harwood 2000; Phiel & Klein 2001; Quiroz et al. 2004). The correspondence between effects of lithium and Wnts on Xenopus development has led to the conclusion that the principal enzymatic target of lithium is GSK3β (Klein & Melton, 1996, Hedgepeth et al. 1997). Moreover, exposing the Xenopus blastula to selective IMPase inhibitors had no deleterious effect on the subsequent development of the embryos (Klein & Melton, 1996). However, there is the intriguing result that administration of myo-inositol, whose accumulation within the cell is dependent on the function of IMPase and IPPase, can override the developmental consequences of ectopically inhibited GSK3β activity in the embryo (Livingston & Wilt 1995; Hedgepeth et al. 1997).

In contrast to its effect when used on the early blastula, where lithium anteriorizes the embryo, lithium exposure at the onset of gastrulation produces posteriorized embryos with underdeveloped anterior structures but having fully extended tails (Yamaguchi & Shinagawa 1989; Fredieu et al. 1997). In our efforts to understand the molecular regulation of cardiac specification, we examined the impact lithium exposure during early gastrulation has on heart formation. Our results indicate that exposure to lithium at the onset of gastrulation can prevent heart formation in the whole embryo, but that this effect is indirect as both lithium-treated explants containing the precardiac mesoderm and precardiac tissue freed from lithium-treated embryos gives rise to differentiated cardiac tissue. In addition, sectioning through lithium-treated embryos indicated that the characterization of their phenotype as being posteriorized might be a misnomer, as it was apparent that all interior structures along the anterior-posterior axis were disfigured. Thus, lithium did not selectively target anterior tissues for changing cell fate, but instead acted globally to disrupt tissue organization within the entire embryo. The implications of these findings in regards to heart formation and the effect of lithium on embryogenesis are discussed.

Material and methods

Embryo culture and lithium treatments

Xenopus laevis embryos were obtained using standard procedures (Sive et al. 2000). Mature eggs were produced from Xenopus laevis females injected with 500 U human gonadotropin (Sigma) to induce ovulation. The eggs were fertilized in vitro in 1% modified Barth’s solution (MBS), dejellied in 2% cysteine, pH 7.8, and reared in 0.1% MBS. Developmental stages of the embryos were based on the classifications of Nieuwkoop and Faber (Nieuwkoop & Faber 1994). Early gastrula stage embryos were obtained by incubation at room temperature for approximately 10 h postfertilization. Embryos that exhibited a dorsal blastoporal groove, but did not yet display involution of cells on the ventral side, were identified as stage 10+ (10.25), as previously designated (Hausen & Riebesell 1991; Heasman 2006). Lithium treatments of 32 cell stage and stage 10+ embryos were according to a standardized dose and exposure time (Kao & Elinson 1989; Klein & Melton 1996; Fredieu et al. 1997). Embryos were immersed in 300 mmol/L lithium chloride (LiCl; Sigma) in 0.5× MBS for 10 min, followed by several rinses with 0.1× MBS, and incubated in 0.1× MBS until desired stages were attained. Control embryos were subjected to the same series of washes and media changes without the addition of lithium. Additionally, some embryos were treated for 10 min with 4 mmol/L of the GSK3β inhibitor SB415286 (Tocris Bioscience). All experimentation received prior approval from the Institutional Animal Care and Use Committee.

Microdissection and explant culture

The DMZ region was isolated from stage 10+ Xenopus laevis by dissection using an eyelash knife. Dissections were performed in 0.5× MBS and allowed to heal in 0.5× MBS. After healing, explants were placed in fresh 0.5× MBS containing 1× penicillin/streptomycin (Sigma) and cultured at room temperature in Nunc four-well dishes pre-coated with 2% sterile agarose. Explants were treated by exposing DMZ tissue to 300 mmol/L LiCl for 10 min immediately after their removal from the embryo. Subsequently, tissue was washed several times in 0.5× MBS and then allowed to heal in fresh media. The heart-forming region (HFR) from stage 18 embryos was identified according to the mapping studies of Sater and Jacobson (Sater & Jacobson 1989, 1990). After using an eyelash knife to harvest the HFR from the embryo, the tissue was cultured at room temperature in 0.5× MBS, 1× penicillin/streptomycin in agarose-coated wells.

Histology and immunofluorescent staining

For histological examination, standard procedures were used (Sive et al. 2000; Ramsdell et al. 2006). Embryos were placed in MEMFA fixative, washed multiple times with phosphate-buffered saline (PBS), dehydrated in a graded ethanol series, cleared with Histoclear (National Diagnostics), and embedded in Paraplast Plus (Fisher). Paraffin blocks were serially sectioned at 10 microns and deparaffinized with Histoclear. Immunohistochemistry was performed using previously described protocols (Eisenberg & Eisenberg 1999; Eisenberg et al. 2006; Ramsdell et al. 2006; Remond et al. 2011). Cultures, sectioned tissue, and whole embryos were methanol and/or paraformaldehyde fixed, and then exposed to mouse monoclonal antibodies specific for fibrillin-2 (JB3), and sarcomeric myosin heavy chain (MF20). These antibodies were obtained from the Developmental Studies Hybridoma Bank at The University of Iowa, Iowa City, IA, USA. For whole embryo immunostaining, the ventral dermis that overlaid the developing heart or corresponding area in treated embryos was carefully removed before adding the primary antibody, which allowed these molecules to more fully penetrate the tissue. For immunodetection of phosphorylated-Histone H3 and cleaved Caspase-3, rabbit polyclonal antibodies specific for the modified proteins (Cell Signaling Technology, Inc.) required the use of frozen sections. These various molecules were visualized following incubation with fluorescein-labeled anti-mouse or rabbit IgG secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Selected sections were co-stained with the fluorescent nuclear marker 4′6′-diamidino-2-phenylindole dihydrochloride (DAPI).

RNA isolation and RT–PCR amplification

Cultures and excised embryonic tissue were placed immediately in RNAlater (Ambion) to prevent degradation of the RNA, which was subsequently isolated using RNeasy kits (Qiagen, Valencia, CA, USA). Gene expression was assayed by both endpoint reverse transcription-PCR and quantitative real-time (q)PCR. For the former assay, amplification was carried out using the OneStep RT–PCR kit (Qiagen). PCR primers for the ODC, siamois, Nkx2.5, and cTnI, as well as the corresponding reaction conditions, were carried out as previously described (Agius et al. 2000; Schneider & Mercola 2001). Comparative qPCR analysis was performed with the StepOne plus qPCR system (Applied BioSystems) using TaqMan qPCR Master Mix (Applied BioSystems), as described (Martin et al. 2011). Primer pairs and probes used for qPCR (Martin et al. 2011) were specific for elongation factor 1α (EF1α), GATA6, Tbx5, Tbx20, Nkx2.5, cardiac α actin (Actc), cardiac myosin heavy chain-α (cMHCα), and cardiac troponin I (cTnI). Relative gene expression levels were estimated by ΔΔCt method using the housekeeping gene EF1α.


Embryos treated with lithium do not form hearts

For all experiments reported in this study, we followed the standardized lithium-treatment protocol established for initiating canonical Wnt signaling in Xenopus embryos (Kao & Elinson 1989; Klein & Melton 1996; Fredieu et al. 1997). Embryos and tissues were exposed to 300 mmol/L LiCl for 10 min and then extensively rinsed in 0.1× MBS. The activity of lithium using this protocol was verified by treatment of 32-cell stage blastulas (Fig. 1A,B), which exhibited highly dorsalized phenotypes in response to this chemical (Kao & Elinson 1989, 1998). Upregulation of the canonical Wnt target gene siamois (Lemaire et al. 1995; Brannon & Kimelman 1996; Fagotto et al. 1997) in response to lithium treatment confirmed that this chemical inhibitor can exhibit signal transduction properties that correspond to canonical Wnt signaling (Fig. 1C).

Figure 1.

 Canonical Wnt signaling mediated by lithium. (A) Lithium exposure of blastulas at the 32 cell stage promotes a dorsalized phenotype, as shown in comparison to parallel cultures of control embryos. A total number of 46 lithium-treated and 122 control embryos were counted and scored according to the dorsoanterior index (DAI), as previously described (Kao & Elinson 1989). (B) Individual example of a lithium-mediated dorsalized embryo. (C) Reverse transcription–polymerase chain reaction (RT–PCR) amplification showing expression of the Wnt target gene siamois in stage 10 ventral tissue harvested from lithium-treated and non-treated embryos. In parallel, harvested RNA was amplified for ornithine decarboxylase (ODC) as a control for template concentration. Upregulation of siamois expression in response to lithium is an indicator of activated canonical Wnt signal transduction (Lemaire et al. 1995; Fagotto et al. 1997).

To examine the effect that lithium has on heart formation, Xenopus embryos were exposed to this chemical at stage 10+, which is the point in development where the mesoderm layer first appears by invagination at the DMZ. Subsequently, embryos were allowed to develop (i) to stage 30 and scored for formation of anterior and posterior structures (Fig. 2); and (ii) further to stage 42 where they were examined for heart development either by immunofluorescent staining of cardiomyocyte proteins (Fig. 3A–D) or RT–PCR amplification of expressed cardiac genes (Fig. 3E,F).

Figure 2.

 Lithium treatment at the onset of gastrulation and its effect on Xenopus development. (A) Summary of the phenotypes displayed by frog embryos exposed to lithium at stage 10+ and parallel cultures of non-treated control embryos. The embryos were scored according to a previously defined scale with the designation of a posteriorized or anteriorized phenotype ascribed to embryos suffering preferentially from reduction in either head or tail structures (Yamaguchi & Shinagawa 1989; Fredieu et al. 1997). According to this defined scale, lithium exposure at stage 10+ promotes a posteriorized phenotype. A total number of 81 lithium-treated and 179 control non-treated embryos were tabulated in this chart. (B, C) Examples of lithium-mediated posteriorized phenotype as shown at stage 27 and 35, respectively. Note the absence of definable head structures such as eyes and cement gland.

Figure 3.

 Lithium treatment at the onset of gastrulation prevents formation of the heart. Stage 42 embryos that were either (A) non-treated or (B–D) treated with lithium at stage 10+. (A) Immunostaining of a non-treated stage 42 control embryo for sarcomeric myosin shows the position of the heart (arrow) (B–D) Stage 42 embryos previously treated with lithium at stage 10+. (B, C) Lithium-treated embryo imaged with visible or fluorescent light after immunostaining with the extracellular matrix protein fibrillin-2, which normally marks the endocardial tube of the developing heart (Kolker et al. 2000). The lack of fibrillin-2 in the area where the heart would usually form (arrow) indicates that a cardiac structure has not formed in the treated embryo. (D) The absence of the heart is further demonstrated by the lack of sarcomeric myosin in the normal heart-forming region of the embryo (arrow). (E) Drawing of stage 42 control and lithium-treated embryos, which illustrate the areas used for RT–PCR gene expression analysis. (F) RT–PCR amplification of the housekeeping gene ODC and the cardiac genes Nkx2.5 and cTnI. Note the decrease in both cardiac genes in response to lithium-treatment. (G) Summary of experiments where Xenopus embryos were exposed to 300 mmol/L lithium for 10 min at either stages 9, 10, 11, or 12 (n = 9, 21, 9, 6, respectively). In parallel dishes, embryos were treated at stage 10 with 4 mmol/L SB415286 for 10 min (n = 19), which is the optimized dose of this selective GSK3 inhibitor (Martin et al. 2011). While lithium treatments at stage 9 and 10 abolished formation of a functional heart, neither lithium exposure at stage 12 or SB415286 at stage 10 prevented development of a contractile heart. (# refers to lack of contractility in the scored population of embryos). (H, I) Representative examples of stage 12 lithium and stage 10 SB415286-treated embryos, respectively, with arrow showing the position of a contractile heart.

Visual inspection of control and treated embryos indicated that exposure to lithium at gastrulation produced severe cranial malformations (Fig. 2A–C), which were scored as a posteriorized phenotype according to a previously established classification scale (Yamaguchi & Shinagawa 1989; Fredieu et al. 1997). While control embryos that were allowed to develop to stage 42 displayed well-formed hearts on the ventral side below the head (Fig. 3A), the heart was not discernible in the corresponding area in parallel-incubated lithium-treated embryos (Fig. 3B–D). This was shown by the absence of a three-dimensional structure in the embryonic region where the heart would normally form (Fig. 3B,C) and the lack of cardiomyocyte protein expression (Fig. 3D). Correspondingly, genes expressed in differentiated cardiomyocytes (cardiac troponin I; cTnI) were absent from stage 42 embryos exposed to lithium at gastrulation (Fig. 3E,F). Surprisingly, the early cardiac marker Nkx2.5 was detectable in these lithium-treated embryos, although at levels lower than observed in control embryos. Cardiac ablation in response to lithium was stage dependent, as delaying exposure to this chemical until stage 12 no longer prevented formation of a beating heart (Fig. 3G,H). Thus, the data demonstrate that-exposure of the whole embryo to lithium at the onset of gastrulation (i.e., stage 10) inhibited formation of the heart and differentiation of precardiac mesoderm to myocytes. In contrast, treatment of embryos at stage 10 with high doses of the GSK3 inhibitor SB415286 (Martin et al. 2011) did not hinder development of contractile hearts (Fig. 3G,I).

Lithium-treatment disrupts organogenesis throughout the embryo

Previous studies had shown that lithium-exposure at gastrulation disrupted formation of anterior structures in the developing frog embryo (Yamaguchi & Shinagawa 1989; Fredieu et al. 1997; Kao & Elinson 1998). Accordingly, these treatments were characterized as having posteriorized the embryos. However, sectioning of embryos at stage 42 indicated that the earlier exposure to lithium had a more comprehensive impact on embryogenesis (Fig. 4A–G). Control and lithium-treated embryos were examined by transverse section (Fig. 4A) and either stained with eosin (Fig. 4B,C,F,G) or immunofluorescently stained for sarcomeric myosin heavy chain (Fig. 4D,E). A comparison to control embryos demonstrated that lithium treatment inflicted a nearly complete disruption in tissue formation throughout the organism. Sectioning up from the pericardial cavity through the base of the head revealed a near absence of organized tissue in the lithium-exposed embryos (Fig. 4C). Along with the complete absence of the heart, the otic capsule did not form, the notochord was reduced to a thin hollow tube, and somitic tissue was disorganized and not bilaterally separated (Fig. 4C). However, somitic mesoderm did express sarcomeric proteins, unlike the myocyte-negative tissue of the ventral side where the heart would normally form (Fig. 4D). Among endodermal tissues, there was a complete lack of tissue organization including the absence of the pharyngeal cavity (Fig. 4C). While the neural tube was readily identifiable in the lithium-treated embryos, it exhibited a highly distended shape (Fig. 4C) and lacked the typical chevron anatomy displayed by non-treated control embryos (Fig. 4B).

Figure 4.

 Histology of the Xenopus embryo. (A) Drawing of stage 42 control and lithium-treated embryos displaying the transverse sectional planes shown in the corresponding panels. (B) Cross sectional slice of non-treated embryo from the pericardial cavity through the head region. (C) Transverse section of a lithium-treated embryo at the plane analogous to that shown for the control embryo in panel B. (D, E) Transverse anterior sections of control and lithium-treated embryos, respectively, immunostained for sarcomeric myosin. (F, G) Cross sectional slice through the tail of control and lithium-treated embryos, respectively. da, dorsal aorta; df, dorsal fin; ht, heart; oc, otic capsule; nc, notochord; nt, neural tube; phx, pharynx, ppc, pericardial cavity; som, somites; spc, spinal cord; vf, ventral fin.

Sectioning through the tail end of the embryos demonstrated that disruption in organogenesis in response to lithium-exposure was not limited to the anterior portion of the organism. While the tail region in control embryos displayed well-defined somites, notochord, spinal chord, and dorsal aorta (Fig. 4F), there is little corresponding organization of these structures within the posterior end of the lithium-treated embryos (Fig. 4G) – even though the latter embryos still developed the capability to wriggle their tails to swim. The internal anatomy of these lithium-treated embryos was not distinguishable except for a flattened spinal chord and a small collapsed notochord. In addition, the dorsal and ventral fins of the lithium-embryos were deformed (Fig. 4G), as they lacked the shapely figure of the wild-type tail (Fig. 4F).

Further analysis of the embryos demonstrated that the standard high-dose lithium protocol used to study embryonic patterning, negatively affected cell proliferation and survival in the developing frog (Fig. 5). While every section from non-treated control embryos displayed numerous cells that stained positive with the mitotic marker phospho-histone H3 (Fig. 5A,B), sectioned lithium-treated embryos were mostly devoid of phospho-histone H3 stained cells (Fig. 5C). In comparison, exposure to high doses of the GSK3 inhibitor SB415286 did not decrease cell proliferation, as indicated by positive phospho-histone H3 staining (Fig. 5D). An additional feature of lithium-treated embryos was the high numbers of cells that exhibited the apoptotic marker caspase-3 (Fig. 5E,F). In contrast, neither non-treated control nor SB415286-treated embryos showed a similar display of apoptotic cells (not shown).

Figure 5.

 Immunofluorescent analysis of cell proliferation and apoptosis. Transverse frozen sections of stage 40 embryos that were (A, B) non-treated, or exposed to (C, E, F) lithium or (D) SB415286 at stage 10. (A) Non-treated sectioned embryo stained with the mitotic marker phospho-histone H3 (phosH3). Note the presence of multiple phosH3-positive cells, which indicate cells undergoing mitosis. (B) High magnification view of sectioned non-treated frog embryo, which was stained for phosH3 and co-stained with the nuclear label 4′6′-diamidino-2-phenylindole dihydrochloride (DAPI), demonstrating the nuclear localization of phosH3 antibody staining. (C) Sectioned tissue from lithium-treated embryo, showed an absence of phosH3-staining. (D) Embryos treated with the selective GSK3 inhibitor SB415286 exhibited positive phosH3 staining. (E, F) Brightfield and immunostained sectioned tissue from a lithium-treated embryo that was immunostained for the apoptotic marker capsase-3 and co-stained with DAPI. Exposure to lithium produced embryos that displayed high numbers of apoptotic cells, as indicated by positive staining for caspase-3. Scale bars = 50 μm.

Lithium-treatment does not directly inhibit the myocardial differentiation of precardiac tissue

To test directly whether lithium inhibits cardiogenesis, the DMZ was dissected from stage 10+ embryos and treated for 10 min in the presence or absence of 300 mmol/L lithium, prior to extensive washing and further incubation (Fig. 6A). Stage 10+ DMZ tissue contains the precardiac fields (Jacobson & Sater 1988; Afouda & Hoppler 2009) and will give rise to differentiated cardiac tissue by 48 h of culture and become contractile by the fifth day of incubation (Martin et al. 2011). Parallel control and lithium-treated day 3 explants exhibited cardiac tissue in 92% and 88% of total explants, respectively (Fig. 6B), as determined by the display of a large aggregate of sarcomeric myosin-positive cells within the explant interior (Fig. 6C–F). RT–PCR amplification of RNA harvested from day 3 DMZ cultures confirmed the cardiac differentiation of the tissue, as the cardiac-associated transcription factor Nkx2.5 was readily detectable in both non-treated and lithium-treated explants (Fig. 6G). However, survival rates of tissue exposed to lithium past the third day of culture were far less than the control explants, which raised the possibility that the lower levels of cTnI expression could be an indirect result of chemotoxic effects.

Figure 6.

 Cardiac tissue differentiation of DMZ explants cultured in the presence of lithium. (A) Outline of experimental protocol involving the dissection of DMZ tissue from stage 10+ embryos, which were exposed to either lithium or media alone, and then cultured for 3 days. Subsequently, explants were either (B–F) fixed and immunostained for sarcomeric myosin, or (G) harvested for RNA for reverse transcription–polymerase chain reaction (RT–PCR) amplification. (B) Summary of the immunohistochemical data, with explants displaying a large aggregate of sarcomeric myosin-positive cells within the interior of the tissue being scored for a cardiac phenotype. Examples of sarcomeric myosin-positive (C, D) non-treated control and (E, F) lithium-treated explants, shown in brightfield and fluorescent images. The asterisks overlaid on the brightfield images show position of the sarcomeric myosin positive aggregates revealed in the adjacent panels displaying immunofluorescent staining. (G) RT–PCR analysis of the cultures examined their expression of the housekeeping gene ODC and cardiac genes Nkx2.5 and cTnI. While lithium treatment decreased expression of the late cardiac gene cTnI, levels of Nx2.5 were only moderately reduced in response to lithium exposure.

In the experiments described above, lithium was added to explants that were healing following their microdissection from the embryo. To lessen the potential toxicity of the treatments, an alternative protocol was devised for examining the direct effect of lithium on DMZ tissue. Here, stage 10+ embryos were either lithium or mock-treated for 10 min, and then extensively washed, prior to the excision and incubation of the DMZ (Fig. 7A). In this case, lithium did not prevent the formation of a contractile phenotype, as >50% of the stage 10+ DMZ explants from lithium-treated embryos generated beating cardiac tissue within 5 days of culture (Fig. 7B–E). Lithium exposure did alter gene expression patterns during the period of incubation, with various cardiac genes showing either decreased or increased expression levels (Fig. 7F). Yet precardiac tissue explants harvested from lithium-treated embryos appeared to develop into normal, well-formed cardiac tubes (Figs 7E, S1), which testifies to the relatively mild effect that lithium exposure had on the ex vivo development of excised DMZ tissue as compared with its influence on the whole embryo. Thus, isolation of the precardiac DMZ subsequent to lithium exposure at stage 10+, rescued heart formation from the inhibition that was observed when the whole embryo was allowed to develop following lithium-treatment.

Figure 7.

 Removal of DMZ from lithium-treated embryos rescued the ability to form contractile cardiac tissue. (A) For these experiments, stage 10+ embryos were either mock or lithium-treated, prior to the removal and culture of DMZ tissue. (B) Following 6 days of culture, explants were scored for contractility, as summarized in this graph. Although explants from lithium-treated embryos showed decreased formation of a contractile phenotype as compared to controls, >50% of the lithium-exposed tissue produced beating tissue. This is in contrast to embryos left intact following lithium exposure, which never developed a contractile heart (Figs 2–4). (C, D) Brightfield and fluorescent view of a representative explant taken from a lithium-treated embryo that was subsequently immunostained for sarcomeric myosin. The arrow in both panels shows position of the sarcomeric myosin-positive beating tissue within the explant. (E) Image of a live explant harvested from a lithium-treated embryo, which displays a well-formed heart tube (arrow) that was contractile (see Fig. S1). (F) Cardiac gene expression in DMZ tissue, as expressed as a ratio of relative mRNA levels displayed in tissue obtained from lithium-treated embryos as compared to non-treated embryos. DMZ tissue was cultured for 2, 3, and 5 days prior to RNA isolation and PCR analysis of cardiac transcription factors Nkx2.5, GATA6, Tbx5, Tbx20 and the muscle genes cTnI, Actc, and cMHCα.

As a follow-up to these findings, we delayed the isolation of the heart-forming regions for 10 h of incubation following the stage 10+ lithium exposure (Fig. 8A–G). The heart-forming regions from the resultant stage 18 mock- and lithium-treated embryos were then cultured for 5 days before scoring for contractility and examining their degree of cardiac differentiation by immunofluorescent staining of myocardial proteins or RT–PCR amplification of expressed myocardial genes. In these experiments, there was an obvious divergence in the incidence of contractility displayed by the different cultures, as beating was displayed by 67.9% and 15.4% of the explants from control and lithium-treated embryos, respectively (Fig. 8A). Moreover, levels of cTnI expression were noticeably decreased (although not absent) in the lithium-exposed tissue in comparison to the control explants (Fig. 8C). However, the expression of Nkx2.5 was not reduced in response to lithium exposure (Fig. 8C). In addition, immunofluorescent staining indicated that explants from both control and treated embryos exhibited cardiac structures (Fig. 8D–G). Thus, the cardiogenic differentiation of the heart forming fields can still be partially rescued from an embryo subjected to lithium at gastrulation, even when harvesting of the tissue is substantially delayed after time of treatment.

Figure 8.

 Cardiac tissue differentiation of precardiac tissue removed from embryos 10 h after lithium exposure. (A) For these experiments, stage 10+ embryos were either mock or lithium-treated and incubated for 10 additional hours to stage 18, prior to the removal and culture of the heart forming region (HFR). Explants were cultured for 5 days and scored for beating. (B) Summary of the results, which showed a significant decrease but not complete abolition of contractile tissue formation in response to lithium treatment. The asterisk in the right hand column denotes the nine explants from each group that were subsequently immunostained for sarcomeric myosin. All these explants were scored positive for cardiac tissue, with both control and lithium-exposed tissue exhibiting three-dimensional sarcomeric myosin-positive structure within the explant interior. (C) Reverse transcription–polymerase chain reaction (RT–PCR) amplification of the housekeeping gene ODC and cardiac genes Nkx2.5 and cTnI. The expression of the two heart genes by explants obtained from lithium embryos confirms the cardiac phenotype of the differentiated tissue. Individual examples of sarcomeric myosin-positive explants from (D, E) non-treated control and (F, G) lithium-treated embryos, as imaged in brightfield and fluorescence. Arrows in panels E and G identify cardiac tissue within the explants that stained positive for sarcomeric myosin.


This study was initiated to examine molecular signals that induce formation of the vertebrate heart. An advantage to using the frog as our animal model is that the embryos have the ability to continue to develop to late stages even when heart formation is completely blocked. Our interest in studying the effect of lithium was due to its apparent effect in inhibiting heart formation when administered at the onset of gastrulation, which is the time when the precardiac mesoderm is specified. Moreover, lithium has been shown to phenocopy canonical Wnt signaling in its effect on the early embryo and ability to stimulate signal transduction events ascribed to canonical Wnts (Kao et al. 1986; Kao & Elinson 1989; Klein & Melton, 1996, Hedgepeth et al. 1997; Larabell et al. 1997; Heasman 2006). In the present study, we demonstrated that exposing the frog embryo to lithium at the beginning of gastrulation prevented formation of the heart. Yet, closer examination revealed that developmental deformities brought on by these treatments were not confined to the heart, but affected the entire embryo. Previous reports investigating the effects of exposure to lithium during the onset of gastrulation have designated the resulting deformity as a posteriorized phenotype (Yamaguchi & Shinagawa 1989; Fredieu et al. 1997). This characterization implies that the anterior end of the embryo primarily suffers from malformation, while posterior tissues develop normally. Yes, the external appearance of the lithium-treated embryos, as well as the ability to wriggle their tails, may elicit an initial diagnosis of a posteriorized phenotype. However, sectioning through the embryos exposed to lithium revealed a different story, as it was apparent that all interior structures along the anterior-posterior axis were disfigured. Internal tissues throughout the treated embryos were distended, misshaped, and exhibited a lack of cellular organization. Emerging tissues were poorly segregated and there was decreased cavity formation within the embryo. Overall, lithium-treated embryos displayed reduced cell proliferation and greatly enhanced apoptotic response. While some differentiation did occur – witness the formation of skeletal myocytes – these cells did not coalesce into fully organized tissue.

Unlike skeletal muscle, there was no evidence of cardiac muscle in animals treated at stage 10+ with lithium. Neither was there a hint of a cardiac tube nor a pericardial cavity. Despite the complete absence of cardiac tissue in lithium treated whole embryos, exposure to lithium did not prevent myocardial differentiation of precardiac DMZ explants. Moreover, precardiac tissue freed from the embryo subsequent to lithium treatment at gastrulation was able to give rise to contractile cardiac tissue. Together these data indicate that the effect of lithium on inhibiting cardiac differentiation was not direct.

Why heart formation was rescued by harvesting the precardiac areas following lithium exposure is not yet clear. The extensive washing of the embryos following treatment and their placement back under basal conditions, would appear to discount developmental artifacts due to long-term lithium retention within the intact embryos. While no formal study has been performed on the retentiveness of lithium in the developing frog embryo, pharmacokinetics studies have been described in adult mammals, with the half-life of lithium in the adult mouse being 3.5 h (Wood et al. 1986). The persistence of lithium in the frog embryo should be much less, as its ability to absorb and exchange molecules in the aquatic environment is what allows the embryos to develop without a functional cardiovascular system (Grow & Krieg 1998; Lohr & Yost 2000). In vivo tracing of signal transduction pathways within Xenopus tadpoles indicates that lithium-mediated signaling can persist up to 3 h after treatment (Denayer et al. 2006). That cardiac differentiation could still be salvaged from precardiac areas removed as late as stage 18 – which was approximately 10 h following the 10 min lithium pulse at stage 10 – suggests that the cardiac rescue was not a spurious outcome of relinquishing lithium that had been trapped in the whole embryo. In addition, the lack of cardiac inhibition when lithium exposure was delayed to stage 12 also provides supportive evidence that the developmental abnormalities in response to lithium were not related to the failure of this molecule to be cleared from the embryo at later stages of development.

An alternative explanation for the rescue of cardiac tissue following lithium exposure is related to the disorganized and dense structure of the treated embryos. Stage 19 corresponds to closure of the neural tube, with following stages marked with a burst of organogenesis throughout the embryo (Hausen & Riebesell 1991; Nieuwkoop & Faber 1994). Among the striking consequences of lithium treatment on development is the lack of cavity formation, as these embryos are characterized by a tightly packed architecture. The lack of fluid movement in the embryos may have deprived developing tissues of needed nutrients, proper concentrations of secreted signaling molecules, and important directional cues for promoting tissue polarity. As a consequence of these deficiencies, a supportive environment did not form to efficiently promote cell proliferation and survival. Accordingly, the explant experimentation that freed precardiac areas from lithium-treated embryos may have simply allowed the tissue to develop because it provided necessary fluid access.

It has been accepted that lithium’s influence on development is due to its mimicry of canonical Wnt signaling. Yet a few caveats could be mentioned, the first being that lithium targets several other enzymes besides GSK3β (Williams & Harwood 2000; Quiroz et al. 2004). Second, the lithium doses required to perturb axis formation and organogenesis in the frog embryo is 50–100 fold in excess of half maximal inhibitory concentration (IC50) for GSK3β (Klein & Melton, 1996), which may indicate that proteins having a relatively low affinity for lithium may be greatly affected by this chemical at these high concentrations. Third, lithium exposure at gastrulation generates different developmental outcomes than does treatments with the more selective GSK3 specific inhibitor SB415286, which produces embryos whose defects are primarily at the posterior end (Martin et al. 2011). Yes, embryos exposed to lithium at gastrulation suffer from malformations that may appear similar to the posteriorized phenotype exhibited in response to ectopic expression of Wnt8 plasmid after the midblastula transition (Sokol et al. 1991; Fredieu et al. 1997). However, unlike the global exposure to lithium, the posteriorized phenotype elicited by Wnt8 arose from ectopic expression on the dorsal side only. As shown by our recent data injecting constitutively-active β-catenin into the ventral blastomeres (Martin et al. 2011), ectopic activation of the canonical Wnt pathway can produce either anteriorized or posteriorized phenotypes depending on the placement of the injection. In contrast, the lithium treatments are not side-restricted, but bathe the entire embryo, which suggest that the Wnt and lithium treatments are not analogous. Moreover, our sectioning of the lithium-exposed embryos revealed a very different histology than exhibited by the Wnt8 induced posteriorized embryos. Unlike the lithium embryos, Wnt8 treated embryos exhibited organized tissue, including a fully formed heart surrounded by the pericardial cavity, albeit some anterior structures demonstrate incomplete development (Christian & Moon 1993).

In conclusion, the experiments reported here indicate that exposure of the frog embryo to lithium at the onset of gastrulation produces embryos that only superficially exhibit a posteriorized phenotype as a byproduct of complete tissue disorganization throughout the organism. The absence of a heart in these embryos is due to an indirect influence of lithium’s global effect on development, as cardiac differentiation can be rescued by separation of the precardiac tissue from the embryo. A comparison with embryonic malformations elicited by ectopically expressed Wnt or treatment with selective GSK3 specific inhibitors may imply that developmental perturbations promoted by lithium are not due solely to its mimicry of canonical Wnt signaling. Moreover, the cardiac rescue experimentation suggests that caution may be needed in prescribing direct regulation of biological events in whole embryo experimentation.


This work was supported by NIH RO1HL073190 (L.M.E.).