Tunneling nanotube-like structures regulate distant cellular interactions during heart formation

Abstract

In the developing mammalian heart, the endocardium and the myocardium are separated by so-called cardiac jelly. Communication between the endocardium and the myocardium is essential for cardiac morphogenesis. How membrane-localized receptors and ligands achieve interaction across the cardiac jelly is not understood. Working in developing mouse cardiac morphogenesis models, we used a variety of cellular, imaging, and genetic approaches to elucidate this question. We found that myocardium and endocardium interacted directly through microstructures termed tunneling nanotube-like structures (TNTLs). TNTLs extended from cardiomyocytes (CMs) to contact endocardial cells (ECs) directly. TNTLs transported cytoplasmic proteins, transduced signals between CMs and ECs, and initiated myocardial growth toward the heart lumen to form ventricular trabeculae-like structures. Loss of TNTLs disturbed signaling interactions and, subsequently, ventricular patterning.

Fig. 1.. Direct interaction between endocardium and myocardium.

(A) Scheme shows the design principle of the gLCCC system. (B and C) Whole-mount X-gal staining of the control embryo at E8.5 (Cdh5-aGFP-N-tTA; tetO-Lacz), which has the synthetic receptor only, shows no X-gal staining signaling (B), whereas the gLCCC embryo (Tnnt2-mGFP; Cdh5-aGFP-N-tTA; tetO-Lacz) (C) shows signaling in the heart, suggesting a direct myocardium-endocardium interaction. Subpanels (B1), (C1), (C2), and (C3) show zoomed-in views of boxed areas in their corresponding main panels. (D) The X-gal and GFP staining of one transverse section from the gLCCC embryo in (C), and (D1) shows a zoomed-in view. Arrowheads indicate X-gal+ or synthetic Notch-activated ECs. (E and F) N1ECD IF staining in E8.5 heart transverse sections showed its localization to ECs and CMs recognized by cardiac troponin T (cTnT). In addition to ECs, N1ECD also localized to the luminal domain of CMs in the inner layer, indicated by arrowheads, but staining was absent or only weakly present in CMs in the outer layer of the myocardium, indicated by arrows (E1). N1ECD localizes to the myocardial layer at one cell thick (F). CMs and ECs were recognized by cTnT and endomucin, respectively. Scale bars: 50 μm [(B) and (D)], 20 μm (C), and 10μm [(E) and (F)].

Fig. 2.. TNTL is linking EC and CM in vivo and in vitro.

(A) Transverse sections of heart (Nkx2.5^Cre/+; mTmG) at 16 ss (~E8.75). Two types of nanotubes, NC-NT (A1) and TNTL (A2 and A3), are marked with pink and white arrowheads, respectively. (B to D) 3D reconstruction of transverse SBF-SEM sections show TNTL [(B) and (D)] and NC-NT [(C) and (D)] nanotubes, indicated by black and pink arrowheads, respectively. Asterisks point to electron-dense organelles (series sections are shown in B1 to B6) in the TNTL. (E and F) Length (E) and diameter (F) of TNTLs and NC-NTs derived from CMs or ECs were measured based on the SBF-SEM images. (G to L) Cardiac cells from CBF1:H2b-Venus heart were cocultured; the ECs that were single and not connected to any other cells were Venus⁻ (G), whereas those connected to other cells were Venus⁺ [(H) to (J)]. The TNTL indicated by a white arrowhead (J) that links CM to EC is sufficient to activate Notch signaling. Subpanel (J1) shows the zoomed-in view of the dotted boxed area in (J). The percentages of Venus⁺ ECs linked to CMs or ECs are shown in (K). The Venus intensities in ECs (PECAM) linked to CMs (cTnI) are stronger than those in the ECs linked to ECs (L). (M to P) Cardiac cells described in (G) to (L) were subjected to time-lapse imaging combined with tile scanning. DIC images overlapping with fluorescence images were shown. The regions shown in (M) to (P) were cropped from an image composed of 25 tiles. Insets show magnified regions indicated by the dotted boxes, and the black arrowhead points to the TNTL. CMs are tdTomato⁺. A Venus⁻ cell (M), once contacted with the ECs by TNTL (N), becomes Venus⁺, indicated by green arrowheads [(O) and (P)]. Contrast in (M) to (P) was adjusted to enhance the visual clarity. Data are represented as mean ± SEM and are representative of two (L) or three (F) independent experiments. Statistical significance was assessed by a two-tailed Student’s t test; P < 0.05 is considered statistically significant. Scale bar: 10 μm.

Fig. 3.. TNTLs are composed of actin filaments but not microtubules.

(A) Tubulin Tracker Deep Red labels tubulin-mediated structures in the CMs except for the TNTLs. CMs were recognized by Nkx2.5^Cre/+; mTmG. The white arrowhead points to a TNTL. (B) TNTLs were labeled by LifeAct, which binds to filamentous actin. ECs were recognized by IB4 staining. White arrowhead points to TNTL. Subpanels (A1) and (B1) show zoomed-in views of areas marked by white dotted boxes. (C) A low-magnification cryo-EM micrograph showed a TNTL feature between CM and EC. White arrowheads point to a CM (right) and an EC (left). (D) A central slice of the cryo-tomogram displays filamentous structural features consistent with a bundle of actin filaments (red arrowhead) and vesicles (red arrows). (E) Scheme of generating the LifeAct knock-in mouse line at Rosa26 locus (RosaMyoX-LifeAct). (F and G) The right insertion of RosaMyoX-LifeAct was confirmed by Southern blot with two probes (F) and genomic sequencing (G). (H) The cardiac region of the embryo with the genotype of Nkx2.5^cre/+; RosaLifeAct E8.75 was labeled with LifeAct (H), and CM and EC were linked by a LifeAct-labeled TNTL (H1). The white arrowhead points to a TNTL. Scale bar: 5μm in (A) to (C), 100 nm in (D), and 50 μm in (H).

Fig. 4.. Actin filament polymerization is required for the formation of TNTL.

(A and B) Vehicle-treated cells formed TNTL indicated by white arrowhead (A), whereas CK-666 suppressed TNTL formation but formed bleb in cocultured cardiac cells (B). (C and D) The percentage of CMs that contained TNTL (C) and the number of TNTLs per CM (D) were significantly reduced in CK-666–treated CMs compared with the control. (E and F) CMs isolated from the embryonic hearts (Cdc42^flox/flox) were treated with Ade-Cre (E) or Ade-GFP (F), and the Ade-GFP–but not Ade-Cre–treated CMs formed TNTLs, indicated by green arrow-heads between ECs and CMs. (G and H) The percentage of CMs that contained TNTL (G) and the number of TNTLs per CM (H) were significantly reduced in Ad-Cre–treated (Cdc42 null) CMs, compared with the control. Data are represented as mean ± SEM and are representative of at least three independent experiments [(C), (D), (G), and (H)]. Statistical significance was assessed by the Mann-Whitney U rank sum test, and P < 0.05 is considered statistically significant. (I to M) At 16 ss (~E8.75), control and MCKO transverse heart sections contain similar CM numbers per section (K), but TNTL formation in MCKO hearts is significantly reduced compared with that of controls [(I), (J), (L), and (M)] (seven hearts for control and seven for MCKO, and the number of TNTLs was the average of seven sections per heart). White arrowheads indicate TNTLs in (I) and blebs in (J). (N and O) The length and diameter of TNTLs linking ECs and CMs were measured by using the SBF-SEM stack images, and the TNTLs in MCKO hearts were shorter with a larger diameter (19 sections from 3 controls and 6 sections from 3 MCKO hearts). Data are represented as mean ± SEM and are representative of seven [(K) and (L)] or three hearts [(N) and (O)]. Statistical significance was assessed by two-tailed Student’s t test; P < 0.05 is considered statistically significant. Scale bar: 5 μm in (A) to (F) and 10μm in (I) and (J). Subpanels (A1), (B1), (E1), (F1), (I1), and (J1) show zoomed-in views of areas marked by white dotted boxes in their corresponding main panels.

Fig. 5.. TNTL contributes to Notch1 activation in vivo.

(A and B) RNAScope showed that Notch1 transcription level based on the intensity was not obviously different between control (A) and MCKO (B) hearts (Sagittal sections). (C to F) N1ICD level was significantly reduced in the MCKO heart at ~E9.5 (10 sagittal sections from 3 control or MCKO hearts). White arrowheads point to ECs. CMs and ECs were recognized by MF20 and endomucin, respectively [(C) and (D)]. Protein immunoblot analysis shows that the N1ICD was significantly reduced, but the full-length Notch1 was increased in the MCKO [(E) and (F)]. (G to K) Cardiac cells from MCKO hearts at ~E9.5 were cultured on different Notch ligand precoated plates [(G) to (J)], and the individual ECs on ligand-coated plates displayed a significantly higher percentage of Venus⁺ cells than the control (K). CMs were recognized by RFP (Nkx2.5^Cre/+; TdTomato). (L to P) The pregnant female was administered CK-666 when the embryos were at E8.25. Embryos were harvested at ~E8.75 (L) or E9.25. White arrowheads point to blebs (L). At E8.75, transverse heart planes showed that hearts treated with CK-666 exhibited an absence of TNTL formation compared with vehicle-treated controls. The number of TNTLs was averaged from five sections per heart [(M) and (N)]. By E9.25, CK-666–treated hearts demonstrated reduced N1ICD and trabecular (Trab.) formation [(O) and (P)]. CMs were GFP⁺ (Nkx2.5^Cre/+; mTmG). Data are represented as mean ± SEM and are representative of three hearts (F), three or four experiments (K), six hearts (M), five hearts (N), or six sections from three hearts [(O) and (P)]. Statistical significance was assessed by two-tailed Student’s t test; P < 0.05 is considered statistically significant. Scale bar: 50 μm in (A) to (D), 10 μm in (G) (J), and 10 μm in (L). Subpanels (A1), (B1), (C1), (D1), (G1), (H1), (I1), (J1), and (L1) show zoomed-in views of areas marked by white dotted boxes in their corresponding main panels.

Fig. 6.. TNTLs regulate ventricular patterning in a temporal manner.

(A and B) In the heart at ~6 ss, the SBF-SEM image stack shows the NC-NT (asterisk) and TNTLs (arrowheads). The inset in (A) is sectioned deeper and shown at higher magnification in (B). (C and D) In the heart at ~16 ss, CMs at the top of the trabecular-like structure are linked to ECs by TNTLs in the control heart. White arrowhead points to a TNTL. The inset in (C) is sectioned deeper and shown at higher magnification in (D). (E) Quantification of the number of TNTLs in different regions of the heart. TR, trabecular-like structure region. (F and G) In MCKO hearts, the trabecular-like structures were absent at 18 ss. The SBF-SEM image in (G) shows that the ECs do not necessarily contact CMs directly. (H to J) MCKO hearts at ~E9.5 display trabeculation defects with significantly fewer trabeculae per section (nine sections from three control hearts and eight section from three MCKO hearts). (K and L) Cdc42 deletion mediated by SM22-Cre did not cause defects in trabeculation. CMs and ECs were recognized by MF20 (K) and endomucin staining (L). Data are represented as mean ± SEM and are representative of five (E) and three hearts (J). Statistical significance was assessed by a two-tailed Student’s t test; P < 0.05 is considered statistically significant. A transverse section is shown in (A) to (G), and a sagittal section is shown in (H) to (L). Scale bar: 10 μm in (A) and (G) and 50 μm in (C), (F), and (H) to (L).

Fig. 7.. TNTLs provide a highway for molecules transporting.

(A) Adenovirus-expressing Jag1-GFP driven by the cTnT promoter was used to express Jag1-GFP in CMs, and Jag1-GFP was detected in the TNTL. (B) Cardiac cells from ~E9.5 Jag1-eGFP BAC hearts were cultured for FRAP imaging. Jag1-eGFP fluorescence is recovered mainly from signal-sending CMs, indicating unidirectional trafficking. White arrows indicate the trafficking direction. (C) ErbB2-mCherry is trafficked within the TNTL through Time-lapse TIRF imaging. Subpanel (C1) shows six time points from 0 to 60 min. (D) Transverse heart sections from Nkx2.5^cre/+; RosaMyoX-LifeAct embryos were used to examine whether LifeAct-RFP and MyoX-GFP could be transferred from CMs to ECs. In E8.75 hearts, some ECs were LifeAct-RFP⁺ but MyoX-GFP⁻. The white arrowhead points to a TNTL, and the asterisk indicates a LifeAct-RFP⁻ EC. Subpanels (D1) and (D2) show zoomed-in views of areas marked by white dotted boxes in (D). (E and F) This model demonstrates how the TNTL facilitates interactions between ECs and CMs. At an early stage, TNTL-mediated signaling is crucial for Notch activation and the formation of trabecular structures. Disruption of TNTL formation leads to a failure in trabecular development. Scale bar: 10 μm in (A) and (D) and 2 μm in (B) and (C).