Tropomodulin1 is Required in the Heart but not the Yolk Sac for Mouse Embryonic Development

Abstract

Tropomodulin1 (Tmod1) caps the pointed ends of actin filaments in sarcomeres of striated muscle myofibrils and in the erythrocyte membrane skeleton. Targeted deletion of mouse Tmod1 leads to defects in cardiac development, fragility of primitive erythroid cells, and an absence of yolk sac vasculogenesis, followed by embryonic lethality at E9.5. The Tmod1 null embryonic hearts do not undergo looping morphogenesis and the cardiomyocytes fail to assemble striated myofibrils with regulated F-actin lengths. To test whether embryonic lethality of Tmod1 nulls results from defects in cardiac myofibrillogenesis and development, or from erythroid cell fragility and subsequent defects in yolk sac vasculogenesis, we expressed Tmod1 specifically in the myocardium of the Tmod1 null mice under the control of the α-myosin heavy chain promoter, Tg(αMHC-Tmod1). In contrast to Tmod1 null embryos, which fail to undergo cardiac looping and have defective yolk sac vasculogenesis, both cardiac and yolk sac morphology of Tmod1-/-^{Tg(αMHC-Tmod1)} embryos are normal at E9.5. Tmod1-/-^{Tg(αMHC-Tmod1)} embryos develop into viable and fertile mice, indicating that expression of Tmod1 in the heart is sufficient to rescue the Tmod1 null embryonic defects. Thus, while loss of Tmod1 results in myriad defects and embryonic lethality, the Tmod1-/- primary defect is in the myocardium. Moreover, Tmod1 is not required in erythrocytes for viability, nor do the Tmod1-/- fragile primitive erythroid cells affect cardiac development, yolk sac vasculogenesis, or viability in the mouse.

Introduction

Tropomodulins (Tmods) are a conserved family of actin-capping proteins that inhibit actin assembly and disassembly at the pointed ends of actin filaments^1,2. First identified in human erythrocytes, the predominant Tmod isoform in both erythrocytes and striated muscle is Tmod1, which controls thin filament length and stability^{1, 3}. Mice lacking Tmod1 display defects in cardiac myofibrillogenesis coupled with aberrant cardiac looping followed by aborted development and embryonic lethality at E9.5^4,5, yet the exact cause of this lethality is uncertain. In addition, mice lacking Tmod1 have mechanically unstable primitive erythroid cells and fail to remodel their yolk sac vasculature⁴. At this stage of embryogenesis (E7.5-E9.5), Tmod1 is expressed only in the myocytes of the developing heart and in the erythroblasts of the blood islands in the yolk sac^4-6. During mouse embryonic development, a continuous vessel network in the capillary plexus of the yolk sac is established when the heart begins to beat, as early as 3 somites^7-9. Both the viscosity of the flowing blood (from the erythroid cells) and the pumping of the heart are required to remodel the yolk sac capillary plexus into a mature vascular network⁸. Although both the heartbeat and blood viscosity contributed by the circulating blood cells are required for vascular remodeling in the yolk sac, it is not clear if one plays a more significant role than the other. Moreover, whether the lethality of the Tmod1 nulls is due to defects in cardiac function, erythroid cell stability, or yolk sac vasculogenesis remains unknown. The Tmod1 null embryo has both mechanically unstable erythroid cells and a non-functioning heart, and thus provides a good system to investigate whether erythroid cell fragility interferes with vascular remodeling, and if Tmod1 is required in erythroblasts for viability.

To discern whether the lethality of Tmod1 null embryos is due to defects in cardiac development⁵ or erythroblast fragility⁴ and subsequent defects in yolk sac vasculogenesis, we performed tissue-specific rescue by reintroducing a cardiac-specific Tmod1-overexpressing transgene [Tg(αMHC-Tmod1)] into the hearts of the Tmod1 null embryos. Here we demonstrate that cardiac-specific expression of Tmod1 is sufficient to completely rescue the Tmod1 null embryonic phenotypes, including cardiac myofibril assembly and thin filament length regulation, looping morphogenesis, yolk sac vasculogenesis, and lethality. We show that the lethality of the Tmod1 null embryos is due directly to loss of Tmod1 in the cardiomyocytes during embryonic development and that sarcomeric actin filament organization with regulated lengths and H-zones are restored upon reintroduction of Tmod1. Moreover, rescue of the Tmod1 null embryonic defects by introduction of exogenous transgenic Tmod1 demonstrates that defects in looping morphogenesis and in myofibrillogenesis are both specifically due to loss of Tmod1. Furthermore, the rescue of both yolk sac vasculogenesis and cardiac function by the cardiac-specific expression of Tmod1 indicates that mechanical stability of erythroid cells is not critical in remodeling of the capillary plexus or in the morphogenesis of the heart. Lastly, despite the loss of Tmod1 from the erythroid cells, cardiac-specific expression of Tmod1 is sufficient to rescue the embryonic lethality of the Tmod1 nulls, indicating that the Tmod1 null erythroblasts are able to confer viability in the mouse. Thus, during development Tmod1 is required in the cardiomyocytes of the embryonic heart for yolk sac vasculogenesis, cardiac looping, cardiac myofibril assembly, and viability.

Materials and Methods

Mice and Genotyping

Tmod1^lacZ-/- mice were previously generated⁵ and maintained on a mixed background of 129SvEv/C57B6J. Tg(αMHC-Tmod1) transgenic mice over-expressing Tmod1 under the control of the αMHC promoter were previously generated on the FVB/N background and identified as “TOT mice”¹⁰. The Tmod1^lacZ-/- mice are referred to as Tmod1-/- and the TOT transgene is referred to as Tg(αMHC-Tmod1) in this manuscript. Overexpression of Tmod1 in mice due to homozygosity for Tg(αMHC-Tmod1) results in dilated cardiomyopathy^{10, 11}. To prevent the complications from gross overexpression of Tmod1 and to avoid additional recombination events, we maintained the Tg(αMHC-Tmod1) strain as a heterozygote. Crossing the Tg(α; MHC-Tmod1) into our Tmod1+/- mice allows for the generation of litters containing the following possible combinations of offspring: Tmod1+/+ (wild-type), Tmod1+/+^{T(gα; MHC-Tmod1)}, Tmod1+/-, Tmod1+/-^{TgαMHC-Tmod1)}, Tmod1-/-, and Tmod1-/-^{Tg(α; MHC-Tmod1)}. All procedures were performed in accordance with The Scripps Research Institute animal care guidelines.

An expanded Materials and Methods section, including primer sequences, embryonic morphological analysis, microscopy, whole-mount immunofluorescence, and western blotting, can be found in the online data supplement at http://circres.ahajournals.org.

Results

Generation of Tmod1 null mice carrying a cardiac-specific Tmod1 transgene

To introduce Tmod1 specifically in the developing mouse heart, we took advantage of the tropomodulin overexpressing transgenic (TOT) mouse [Tg(αMHC-Tmod1)], which expresses Tmod1 in the myocardium under the control of the α-myosin heavy chain (αMHC) promoter¹⁰. The well-characterized αMHC promoter is specifically and highly expressed in the developing myocardium during early mouse development from embryonic day 7.5 (E7.5) through E10.5 at which time the expression levels decrease in the ventricular myocardium but remain high in the atrial myocardium¹². Advantageously, this promoter has a similar tissue specificity and timing of expression in the myocardium as endogenous Tmod1 during cardiac development^4-6 (Online Figure I). The lethal stage of the Tmod1 null embryos (E9.5) coincides with the expression of the αMHC promoter, thus re-introduction of the transgene should be sufficient to rescue the loss of Tmod1 in cardiac muscle during this stage of development. Moreover, at later stages of embryonic development, the αMHC-Tmod1 transgene expression mirrors endogenous Tmod1 protein with high levels in the atria and low, but clearly detectable levels in the ventricles (Online Figure I). By introducing this transgene back into the Tmod1-/- background, we aimed to generate Tmod1-/-^{Tg(αMHC-Tmod1)} embryos that lack Tmod1 in all tissues except for the myocardium.

To ascertain the levels of Tmod1 protein in the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos, we performed Western blot analysis on E8.5 Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} litters, using actin as a loading control. As reported previously, Tmod1 levels are similar in both the Tmod1+/+ and Tmod1+/- embryos, and there is no Tmod1 protein in the Tmod1-/- embryos⁵ (Figure 1A). The transgene expression increases the levels of Tmod1, but total levels of Tg(αMHC-Tmod1) protein plus endogenous Tmod1 protein are similar regardless of the endogenous Tmod1 copy number, i.e. in embryos carrying the transgene, there is a similar amount of Tmod1 protein in the wild-type, heterozygotes, and nulls (Figure 1A). Thus we can re-introduce consistent levels of Tmod1 protein into the heart via expression from the Tg(αMHC-Tmod1) transgene in the Tmod1-/- mouse.

Figure 1. Tmod1 is present in Tmod1-/-^{Tg(αMHC-Tmod1)}, but not Tmod1-/- embryonic hearts.

(A) Western blot of E8.5 mouse embryos generated from a Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} mating. Each lane contains a single embryo. Blots were probed for Tmod1 (top) and Actin as a loading control (bottom, note actin is partially proteolyzed resulting in a doublet). (B) Whole-mount immunofluorescent confocal stacks of E8.5 wild-type (13 somite pairs; i-iii), Tmod1-/- (11 somite pairs; iv-vi), andTmod1-/-^{Tg(αMHC-Tmod1)} (13 somite pairs; vii-ix) embryos stained for Tmod1 (i, iv, vii), sarcomeric α-actinin (ii, v, viii), and actin (iii, vi, ix). The Tmod1 transgene staining is restricted to the heart (vii). Scale bar is 100 μm.

To confirm that the transgene expression is cardiac specific at this stage, whole-mount immunofluorescence was performed on Tmod1-/-^{Tg(αMHC-Tmod1)} embryos at stage E8.5 of development. At this stage of development in wild-type embryos, staining for both Tmod1 and sarcomeric α-actinin is restricted to the myocardium, and F-actin, as shown by phalloidin staining, is enriched in the myocardium (Figure 1B, i-iii). As expected, Tmod1 null embryos do not stain for Tmod1 (Figure 1B, iv), but do exhibit staining for sarcomeric α-actinin and F-actin in the myocardium (Figure 1B, v-vi, and⁵). In turn, Tmod1-/-^{Tg(αMHC-Tmod1)} embryos exhibit specific staining for Tmod1 only in the developing heart (Figure 1B, vii), as expected for expression of the αMHC promoter¹². These data are consistent with cardiac-specific expression of the transgene. It is of note that even at this level of magnification, it is evident that the size and gross morphology of the Tmod1-/-^{Tg(αMHC-Tmod1)} embryo appears to be more similar to the Tmod1+/+ embryo than to the Tmod1-/- littermate, suggesting rescue of developmental defects.

Expression of Tmod1 in the heart rescues the early cardiac looping defects of Tmod1 null embryos

The earliest developmental phenotype we have observed in the Tmod1 null embryos is a defect in looping of the heart tube at embryonic day 8.5 (E8.5). At this stage (7 somite pairs), in both wild-type and Tmod1+/- embryos (which have equivalent levels of Tmod1 and are indistinguishable in phenotype⁵), the linear heart tube begins the first right inward looping (Figure 2A). This process of cardiac looping continues, eventually leading to initial chamber specification at E9.5 (15-20 somite pairs)¹³ (Figure 2C). In contrast, the Tmod1-/- 7 somite pair embryo fails to initiate cardiac looping (Figure 2B). Instead, the embryos lacking Tmod1 form a single bulging ventricle (Figure 2B, 2E)⁴, and never undergo the process of cardiac looping morphogenesis. The Tmod1 null phenotype persists through E10.5 (as embryos develop between 10 and 19 somite pairs, data not shown) at which time the embryos cease developing and become resorbed (Figure 2E).

Figure 2. Tg(αMHC-Tmod1) rescues the cardiac looping defect of the Tmod1 nulls.

(A, B) Gross morphology of E7.5 mouse embryos, ventral view depicting first right inward looping of the embryonic heart tube at 7 somite pairs in Tmod1+/+ (A), but not in Tmod1-/- (B). (C-F) Gross morphology of E9.5 Tmod1+/+ (C), Tmod1+/+^{Tg(αMHC-Tmod1)} (D), Tmod1-/- (E), and Tmod1-/-^{Tg(αMHC-Tmod1)} (F) mouse embryos, side view. The cardiac morphology of the Tmod1-/-^{Tg(αMHC-Tmod1)} (F) is similar to the Tmod1+/+ (C). Note that overexpression of Tmod1 does not affect cardiac looping in the wild-type embryo (D). Scale bar is 200μm in A,B and 500 μm in C,D,E,F.

Examination of Tmod1-/-^{Tg(αMHC-Tmod1)} E8.5 embryos demonstrates that reintroducing transgenic Tmod1 in the heart completely rescues this Tmod1-/- early embryonic looping defect (Figure 2F). This indicates that the looping morphogenesis defect is a direct result of loss of Tmod1 in the heart since Tg(αMHC-Tmod1) is not expressed in any other tissues at this stage of development. Moreover, introduction of the Tg(αMHC-Tmod1) transgene in a wild-type background (leading to increased levels of Tmod1 protein as compared to non-transgenic wild-types, see Figure 1A) does not affect looping morphogenesis (Figure 2D), indicating that precise levels of Tmod1 are not critical during early cardiac development. Taken together, these data demonstrate that the Tg(αMHC-Tmod1) is expressed early enough in development to overcome loss of Tmod1 during cardiac looping, consistent with the tissue-specificity and developmental timing of expression for both Tmod1 and αMHC in the heart^4-6,12. Thus myocardial-specific expression of Tmod1 in the embryo is sufficient to rescue the early embryonic cardiac looping morphogenesis defects of the Tmod1 null mouse.

Expression of Tmod1 in the heart rescues the myofibril assembly defects of Tmod1 null embryonic cardiomyocytes

To test whether the Tg(αMHC-Tmod1) is sufficient to rescue myofibrillogenesis at the looping stage of cardiac development, whole-mount immunofluorescence was performed on progeny derived from Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} matings. Wild-type hearts display well-organized myofibrils with typical striated staining for F-actin and α-actinin (at the Z-lines) (Figure 3A-C). At the pointed ends, where actin filaments are capped by Tmod1, regular gaps in the F-actin staining are apparent at higher magnification (Figure 4A-D), indicating that the actin filament lengths are regulated. In contrast, Tmod1 null embryos fail to make cardiac myofibrils as shown by disorganized F-actin and α-actinin staining (Figure 3D-F). Thus, we never see a striated pattern of F-actin or α-actinin in embryonic hearts lacking Tmod1. Instead, the cardiomyocytes in Tmod1 null embryonic hearts display large non-striated bundles of F-actin at the cell periphery as well as aberrant rod-like aggregates of F-actin and α-actinin in the cytoplasm (Figure 3D-F, arrows). These rods are more prominent in the cardiomyocytes on the outer surface of the developing heart tube and are excluded from the cell nuclei (Figure 3F, arrows). This aberrant rod phenotype of the Tmod1-/- embryonic hearts is evident as early as E7.5 (5 somite pairs) and persists until the lethality of the Tmod1 null embryos (at E9.5-E10.5, as late as 19 somite pairs, data not shown). On the inner wall of the myocardium, the Tmod1 null cardiomyocytes exhibit large bundles of non-striated F-actin, as described previously⁵ (and Online Figure II). Occasionally, these F-actin bundles are decorated with periodic puncta of α-actinin, but they never form true striated myofibrils and thus the actin filaments remain overlapping with unregulated lengths⁵ (Online Figure II).

Figure 3. Tg(αMHC-Tmod1) rescues the cardiac myofibril assembly defect of the Tmod1 nulls.

Immunofluorescence of the cardiomyocytes in the outer wall of the heart from Tmod1+/+ (A-C), Tmod1-/-(D-F), and Tmod1-/-^{Tg(αMHC-Tmod1)} (G-I) E8.5 mouse embryos. Staining for sarcomeric α-actinin (A,D,G, green in merge) and actin (B,E,H, red in merge) show typical striated myofibrils in wild-type embryos (A-C). Tmod1-/- embryos lack any discernable myofibrils and instead display aberrant aggregates of sarcomeric α-actinin and F-actin (D-F, arrows). The Tg(αMHC-Tmod1) restores myofibrillogenesis in the Tmod1-/- embryonic hearts (G-I). Nuclei are stained with Hoechst dye and appear blue in merged images (C,F,I). Note that the cardiomyocytes in the developing heart are a heterogeneous population, images presented are from comparable regions. Images are digital projections of confocal stacks. Scale bar is 20 μm.

Figure 4. Tmod1 protein expressed from the Tg(αMHC-Tmod1) transgene localizes to pointed ends of thin filaments in cardiac myofibrils.

Immunofluorescence of single cardiac myofibrils from Tmod1+/+ (A-D) and Tmod1-/-^{Tg(αMHC-Tmod1)} (E-H) E9.5 mouse embryonic hearts. Embryos are staged similarly at 12-13 somite pairs. Staining for F-actin (B,F, red in merge) and sarcomeric α-actinin (C,G, green in merge) show typical striated myofibrils in both Tmod1+/+ (A-D) and Tmod1-/-^{Tg(αMHC-Tmod1)} (E-H) embryos. Arrowheads indicate the H-zone where actin pointed ends (B,F) are capped by Tmod1 (A,E, blue in merge). Images are single optical sections (0.6 μm thick). Scale bar is 4 μm.

In contrast, the cardiomyocytes in the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos assemble normal appearing striated myofibrils in both the inner and outer walls of the developing heart tube and do not form aberrant F-actin/α-actinin rods (Figures 3G-I, and S2). The ability of exogenous transgenic Tmod1 to rescue myofibril assembly is evidenced further by normal localization of Tmod1 to the pointed ends of the myofibril (Figure 4E-H) and by the reappearance of the organized F-actin with gaps at the H-zone and striated α-actinin (Figures 3G-I, 4E-H), indicative of thin filament length regulation and normal myofibril assembly. Therefore Tmod1 null embryos fail to make striated cardiac myofibrils and instead make aberrant bundles of non-striated F-actin and F-actin/α-actinin rods. Taken together, these data show that myocardial expression of Tg(αMHC-Tmod1) is sufficient to target Tmod1 properly to thin filament pointed ends and rescue the thin filament length regulation and myofibril assembly defects of the Tmod1 null embryonic hearts.

Yolk sac vascular defects in Tmod1 null embryos are secondary to cardiac defects

During development, endogenous Tmod1 is also expressed in the primitive erythroblasts in the blood islands of the embryonic yolk sac and mice lacking Tmod1 have fragile circulating primitive erythroid cells and display defects in yolk sac vasculogenesis^4-6 (Figure 5B). To address whether lethality of the Tmod1 null embryos is due to primary defects in cardiac development versus circulating blood cell driven yolk sac vasculogenesis, yolk sac morphology was assessed in the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos. Wild-type embryos have a highly vascularized yolk sac (Figure 5A), with large vessels containing blood cells (Figure 5A'), whereas the Tmod1-/- embryos display a pale yolk sac with no vessels and very little blood in the form of primitive blood islands (Figure 5B, B', and data not shown), as shown previously⁴. In contrast, yolk sacs from the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos have prominent blood vessels containing blood cells, similar to their wild-type littermates (Figure 5C,C'). These Tmod1-/-^{Tg(αMHC-Tmod1)} rescued yolk sacs continue developing normally and are morphologically indistinguishable from wild-type yolk sacs (Figures 5A,C and S3). Thus, transgenic expression of Tg(αMHC-Tmod1) in the myocardium is sufficient to rescue the yolk sac vasculature defects seen in the Tmod1 null embryos.

Figure 5. Tg(αMHC-Tmod1) rescues the yolk sac vasculogenesis defect of Tmod1 nulls but is not expressed in the yolk sac or in the circulating erythrocytes.

(A-C) Gross morphology of E8.5 embryonic yolk sacs from Tmod1+/+ (A,A'), Tmod1-/- (B,B'), and Tmod1-/-^{Tg(αMHC-Tmod1)} (C,C'). Wild-type embryos have a highly vascularized yolk sac (A'), while Tmod1-/- embryonic yolk sacs do not undergo vascular remodeling (B'). Vascularization of the Tmod1-/-^{Tg(αMHC-Tmod1)} embryonic yolk sac (C') is similar to Tmod1+/+ (A'). (D) Western blot of embryonic yolk sacs; each lane contains a single E8.5 yolk sac. Blots were probed for Tmod1 (top) and Actin as a loading control (bottom). The Tg(αMHC-Tmod1) does not express Tmod1 in the yolk sac (last 3 lanes). Note that the faint band in last lane has altered mobility relative to Tmod1 and is likely a cross-reacting band. Also note that actin shadow is due to a high signal for actin and shifting of blot during exposure. (E) Western blot of isolated erythrocytes from E15.5 embryos. Blots were probed for Tmod1 (top) and α-adducin (bottom) as a loading control. The Tg(αMHC-Tmod1) does not express Tmod1 in the definitive erythrocytes (last lane).

To verify that the Tg(αMHC-Tmod1) transgene does not express Tmod1 in the yolk sac, western blot analysis was performed on yolk sacs isolated from E9.5 embryos derived from Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} matings (Figure 5D). This experiment demonstrates that the levels of Tmod1 protein are similar in yolk sacs from Tmod1+/+, Tmod1+/-, and Tg(αMHC-Tmod1) embryos, and that no detectable Tmod1 protein is present in yolk sacs from Tmod1-/- or Tmod1-/-^{Tg(αMHC-Tmod1)} embryos (Figure 5D and data not shown). To confirm further that Tmod1 is not present in the circulating erythrocytes, western blot analysis was performed on erythrocytes isolated from E15.5 mouse embryos derived from Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} matings (Figure 5E). This experiment demonstrates that Tmod1 protein is not present in Tmod1-/-^{Tg(αMHC-Tmod1)} circulating erythrocytes (Figure 5E). The absence of transgenic Tmod1 protein in the Tmod1-/-^{Tg(αMHC-Tmod1)} rescued yolk sacs and embryonic blood cells is consistent with cardiac-specific expression of the αMHC promoter driving the transgene¹². Thus, while the Tg(αMHC-Tmod1) does not drive expression of Tmod1 in the yolk sac nor in the circulating erythrocytes, yolk sac defects are completely rescued by myocardial-specific expression of Tmod1. These data clearly indicate that the yolk sac vasculogenesis defects seen in the Tmod1 nulls are secondary to the cardiac defect.

Cardiac expression of Tmod1 completely rescues the lethality of the Tmod1 null animals

Mice lacking Tmod1 cease developing and die around day 9.5 of embryonic development (E9.5) (Table 1, ^4,5). To determine whether Tmod1-/-^{Tg(αMHC-Tmod1)} embryos develop normally, we performed litter analyses of staged embryos derived from Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} matings. While we have never recovered Tmod1-/- embryos after E10.5, when the transgene is present we find Tmod1-/-^Tg(αMHC-^Tmod1) embryos up to and well beyond E15.5 (Table 1, Figure 6A-B). Moreover, the αMHC-Tmod1 transgene is clearly expressed at E15.5 in the heart as shown by both immunofluorescence of Tmod1 protein (Online Figure I A-D) and RT-PCR of transgene mRNA (Online Figure I E). In pooled litters of embryos at stage E12.5 up to full-term pups, there are 44 Tmod1+/+ progeny, 89 Tmod1+/-, and 36 Tmod1-/-^{Tg(αMHC-Tmod1)} progeny, demonstrating that both the Tmod1 and the Tg(αMHC-Tmod1) alleles segregate in an appropriate Mendelian distribution (Table 1) (expected numbers of embryos are 42.25, 84.5, and 21.125 respectively, which is well within the acceptable limits of a Chi-Square test, data not shown).

Table 1.

Litter analyses of Tmod1+/- × Tmod1+/-^{Tg(αMHC-Tmod1)} progeny.

	Tmod1 +/+	Tmod1 +/-	Tmod1 -/-	Tmod1-/-;Tg+	Total
E8.5	5	1	1	3	10
E9.5	15	21	4	6	46
E10.5	5	3	2	0	10
E12.5	3	5	0	2	10
E13.5	0	5	0	2	7
E14.5	5	3	0	3	11
E15.5	8	20	0	7	35
E18.5	2	4	0	3	9
Term	26	52	0	19	97

Figure 6. Tg(αMHC-Tmod1) rescues the embryonic lethality of the Tmod1 nulls.

(A, B) Gross morphology of E15.5 littermates demonstrates that Tmod1-/-^{Tg(αMHC-Tmod1)} embryos (B) are indistinguishable from their wild-type counterparts (A). (C) PCR genotyping of a typical Tmod1+/- x Tmod1+/-^{Tg(αMHC-Tmod1)} litter. Tg(αMHC-Tmod1) band (top) indicates that the transgene is present, Tmod1 band (middle) indicates that the wild-type endogenous Tmod1 genomic locus is intact, and LacZ band (bottom) is a marker for the Tmod1 targeting event. The 9th lane shows that the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos are positive for the transgene and positive for the LacZ targeting event, but null for the Tmod1 genomic locus.

Examination of gross morphology of E15.5 embryos indicates no noticeable abnormalities in development between the Tmod1+/+ and the Tmod1-/-^{Tg(αMHC-Tmod1)} littermates (Figure 6A,B). As shown in figure 6B, the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos display apparently normal development with a functioning circulatory system, facial development, and limb formation. Moreover, these Tmod1-/-^{Tg(αMHC-Tmod1)} animals are born and are viable, fertile, and motile. Figure 6C shows tail DNA PCR genotyping of a typical litter derived from a Tmod1+/- x Tmod1+/-^{Tg(αMHC-Tmod1)} mating. The DNA in lane 9 is from a viable mouse that is both Tmod1-/- and Tg(αMHC-Tmod1) positive (note that there are no Tmod1 null animals born without the transgene because the null embryos do not survive without Tmod1). These data show that, unlike their Tmod1-/- littermates, Tmod1-/-^{Tg(αMHC-Tmod1)} embryos continue developing normally well beyond the Tmod1-/- lethal stage. Thus, the lethality of the Tmod1 null mice at E9.5 is a direct result of defects in cardiac development because the myocardial-expressing Tg(αMHC-Tmod1) is sufficient to rescue the embryonic lethality of the Tmod1 null mice. Moreover, the mechanical instability of the primitive erythroid cells⁴ does not affect development or viability in the Tmod1-/-^{Tg(αMHC-Tmod1)} rescued mice.

Discussion

We demonstrate here that expression of Tmod1 in the heart is sufficient to rescue the embryonic lethality of the Tmod1 null mice. The Tmod1 null defects in yolk sac vasculogenesis are completely rescued by the presence of Tmod1 in the myocardium, indicating that both the primitive erythroid cell fragility and the remodeling of the yolk sac vasculature are secondary to cardiac function, and that the primary defect in the Tmod1 nulls is in the myocardium. Embryonic defects in cardiac development are often coupled with yolk sac vasculogenesis phenotypes. For example, mutants in myocardial-specific genes such as Nkx2.5, Mef2c, Connexin45, MLC2a, and N-cadherin which result in defects in cardiac development, frequently display abnormal yolk sac morphology as well^8,14-18. However, unlike Tmod1, these genes are cardiac-specific and not expressed in the yolk sac erythroblasts. It has been proposed that mechanical instability due to loss of Tmod1 in the circulating primitive erythroid cells is a key contributing defect in the Tmod1 null embryonic phenotype⁴. In this scenario, the loss of Tmod1 results in mechanically unstable erythroid cells that contribute to a failure to remodel the yolk sac vasculature causing a cessation of circulation that results in aborted cardiac development⁴, yet our data argue against this. Nevertheless, hemodynamic forces generated by the viscosity of the circulating primitive blood cells are required for remodeling the primary plexus of the yolk sac into mature vessels⁸. However, if the mechanical fragility of the Tmod1-/- blood cells alters this viscosity and affects some subtle aspect of yolk sac vasculogenesis, the data presented here demonstrate that this effect is insignificant for vascular remodeling of the yolk sac, or for viability of the mouse. Moreover, the absence of Tmod1 from the definitive erythrocytes later in development has no effect on viability. In contrast, mutations in other components of the red cell membrane, such as spectrin and Band 3, can result in neonatal lethality due to severe hemolytic anemia¹⁹. While it is possible that loss of Tmod1 could result in a mild anemia, the Tmod1-/-^{Tg(αMHC-Tmod1)} embryos develop normally and the mice are completely viable and fertile. While other Tmod isoforms may compensate for loss of Tmod1 in later developing tissues, allowing for post-natal viability of the Tmod1-/-^{Tg(αMHC-Tmod1)} mice, our data show that Tmod1 is dispensable in both the primitive and definitive erythrocytes during embryonic development. Thus, the increased fragility of the circulating Tmod1-/- primitive erythroid cells observed by Chu et al⁴ does not significantly affect embryogenesis, yolk sac remodeling, or subsequent development.

Tmod1 null embryos die at E9.5 with cardiac defects in looping, myofibril assembly, and function. In the absence of Tmod1 in the embryonic heart, thin filament lengths are not regulated and instead we see bundles of actin filaments that stain continuously for actin. These F-actin structures are non-striated in that they do not form H-zones nor assemble Z-lines with periodic α-actinin. Thus cardiac myofibril assembly in vivo is entirely absent without Tmod1. We show here that transgenic expression of Tmod1 in the myocardium restores thin filament length regulation and the formation of striated myofibrils, completely rescuing myofibrillogenesis. Curiously, while Tmod1 is required for myofibrillogenesis and cardiac function in vivo, it does not seem to be required in culture. ES cell cardiomyocytes lacking Tmod1 both assemble striated myofibrils with regulated actin filament lengths and display beating in culture²⁰. Although it has been suggested that ES cells differentiating into cardiomyocytes in culture have more time to develop myofibrils as compared to the cardiomyocytes differentiating in vivo, the Tmod1 null embryonic hearts fail to assemble myofibrils at the very early E7.5 stage (Figure 4 and data not shown), two days before the lethality of the Tmod1 null at E9.5, indicating an early requirement for Tmod1 in cardiac development. Thus, why myofibril assembly is possible in the absence of Tmod1 in cultured ES cells, but not in cardiomyocytes in the developing heart in vivo, remains unclear.

In addition to myofibril assembly defects, Tmod1 null embryos have unexpected defects in cardiac looping morphogenesis and chamber specification during embryonic heart development^4,5 (see Figure 2). Although it is conceivable that myofibril assembly and contractile function contribute to looping morphogenesis, mutations in other myofibril components exhibit dissimilar phenotypes from the Tmod1 null and develop well beyond the looping stage^8,18,21-23. In turn, many mutants that have defects in looping morphogenesis similar to the Tmod1 null are mutations in transcription factors, such as dHAND, MEF2C, Nkx2.5, and GATA-1^15,16,24,25, or mutations in junctional components, such as N-cadherin, Connexin45, and the Na⁺/Ca²⁺ exchanger^14,17,26. It is possible that Tmod1 has additional roles in gene regulation in cardiomyocytes; Tmod1 has been shown to localize in the nucleus in cultured cardiomyocytes²⁷. Additionally, Tmod1 associates with the cardiomyocyte membrane skeleton²⁸ and may affect cell-cell interactions that could play a role in looping. However, our data rule out the possibility that the targeting event to generate the Tmod1 null mouse altered the expression of another, distinct gene. By reintroducing a Tmod1 transgene into the myocardium and rescuing the Tmod1 null cardiac defects, we have demonstrated genetic rescue of the embryonic development of our Tmod1 null and shown that Tmod1 is the gene directly responsible for all of the cardiac defects seen in the Tmod1 null embryos.

In conclusion, cardiomyocyte-specific expression of Tmod1 is sufficient to rescue the embryonic lethality of the Tmod1 null mutants. The Tmod1 null defects in cardiac looping, myofibrillogenesis, cardiac function, and yolk sac vasculogenesis are rescued by expressing Tmod1 in the cardiomyocytes. Tmod1 null embryos fail to make striated myofibrils with regulated actin filament lengths, a phenotype completely rescued by expression of Tmod1 in the myocardium. Our studies in the mouse embryonic heart have also revealed an early requirement for Tmod1 in the de novo assembly of striated myofibrils, a phenotype not evident in cultured cardiomyocytes. We propose that Tmod1 may have both early and late roles in myofibril assembly, with an early requirement for Tmod1 at the cardiomyocyte membrane and a later requirement for Tmod1 in thin filament length regulation. Moreover, Tmod1 is not required in the primitive erythroid cells for yolk sac vasculogenesis, embryonic development, or viability. Our approach has also succeeded in generating viable mice that express Tmod1 only in the myocardium, but lack Tmod1 in all other tissues, providing us the opportunity to dissect these two roles of Tmod1 and study the functional consequences of loss of Tmod1 in later developing tissues such as definitive erythroid cells and skeletal muscle.