G_q-activated fibroblasts induce cardiomyocyte action potential prolongation and automaticity in a three-dimensional microtissue environment

In a three-dimensional microtissue model, which lowers baseline activation of cardiac fibroblasts but enables cell-cell, paracrine, and cell-extracellular matrix interactions, we demonstrate that selective cardiac fibroblast activation by enhanced G_q signaling, a pathophysiological trigger in the diseased heart, modulates cardiomyocyte electrical activity, leading to proarrhythmogenic automaticity.

Abstract

Cardiac fibroblasts (CFs) are known to regulate cardiomyocyte (CM) function in vivo and in two-dimensional in vitro cultures. This study examined the effect of CF activation on the regulation of CM electrical activity in a three-dimensional (3-D) microtissue environment. Using a scaffold-free 3-D platform with interspersed neonatal rat ventricular CMs and CFs, G_q-mediated signaling was selectively enhanced in CFs by Gα_q adenoviral infection before coseeding with CMs in nonadhesive hydrogels. After 3 days, the microtissues were analyzed by signaling assay, histological staining, quantitative PCR, Western blots, optical mapping with voltage- or Ca²⁺-sensitive dyes, and microelectrode recordings of CF resting membrane potential (RMP_CF). Enhanced G_q signaling in CFs increased microtissue size and profibrotic and prohypertrophic markers. Expression of constitutively active Gα_q in CFs prolonged CM action potential duration (by 33%) and rise time (by 31%), prolonged Ca²⁺ transient duration (by 98%) and rise time (by 65%), and caused abnormal electrical activity based on depolarization-induced automaticity. Constitutive G_q activation in CFs also depolarized RMP_CF from –33 to −20 mV and increased connexin 43 and connexin 45 expression. Computational modeling confers that elevated RMP_CF and increased cell-cell coupling between CMs and CFs in a 3-D environment could lead to automaticity. In conclusion, our data demonstrate that CF activation alone is capable of altering action potential and Ca²⁺ transient characteristics of CMs, leading to proarrhythmic electrical activity. Our results also emphasize the importance of a 3-D environment where cell-cell interactions are prevalent, underscoring that CF activation in 3-D tissue plays a significant role in modulating CM electrophysiology and arrhythmias.

NEW & NOTEWORTHY In a three-dimensional microtissue model, which lowers baseline activation of cardiac fibroblasts but enables cell-cell, paracrine, and cell-extracellular matrix interactions, we demonstrate that selective cardiac fibroblast activation by enhanced G_q signaling, a pathophysiological trigger in the diseased heart, modulates cardiomyocyte electrical activity, leading to proarrhythmogenic automaticity.

cardiac fibroblasts (CFs) play a major role in normal cardiac physiology and in the response of the heart to pathological stress and injury. They form a three-dimensional (3-D) cellular scaffold around the cardiac myocytes (CMs) and maintain and remodel the extracellular matrix (ECM), thereby contributing to the integrity and connectivity of the myocardial architecture (19). In response to hemodynamic stress or injury, CFs become activated and assume a myofibroblast (MFb) phenotype with heightened proliferation, contractile properties, and collagen deposition (for reviews, see Refs. 15 and 30). Although CFs are considered nonexcitable cells, they have been shown to modulate active and passive electrical properties in CMs (for reviews, see Refs. 55 and 64).

Two-dimensional (2-D) cell culture models have long been used to investigate the regulation of CMs by CFs in vitro. Immunohistochemistry of heterotypic cell pairs and cocultured monolayers have shown that CFs can directly couple with CMs via gap junctions and adherens junctions (e.g., Refs. 58, 66, and 84). Strips of intervening CFs were also shown to couple otherwise uncoupled CMs and thereby enable action potential (AP) propagation (23), but the conduction velocity was extremely slow, which can greatly facilitate the occurrence of triggered activity and reentry (70, 72). In standard 2-D cell culture, CFs rapidly transition into MFbs, which has been exploited to study MFb effects on CMs by some groups (5, 60), but concerns exist that CF activation induced by substrate stiffness does not fully replicate the activation process in vivo (for reviews, see Refs. 64 and 85). Another approach to studying MFb effects has used activated CFs isolated from infarcted hearts (87). Together, these studies have shown that MFbs slow conduction, change AP duration (APD), and induce ectopic activity in CMs by partial depolarization of CMs (5, 60, 61, 87). Despite longstanding investigations, establishing an active role for CFs in regulating CM electrical function in vivo has been a formidable challenge and matter of considerable debate (for a review, see Ref. 47). Novel technologies enabled recent demonstrations of nonexcitable cardiac cells that follow CM APs in the intact heart which represent strong evidence in support of functional CM-CF coupling in vivo (54, 71).

In contrast to 2-D culture, in which cell-cell interactions are limited to a modest number of planar interactions, cells can interact with each other in multiple directions in a three-dimensional (3-D) tissue environment. Both 3-D scaffold-supported and self-assembled in vitro cultures are important intermediaries between 2-D cultures and in vivo tissues (65). 3-D platforms heighten cell-cell and cell-ECM contact and establish molecular concentration gradients (6). In scaffold-based systems, interactions between cells and external cues dominate. The advantages of self-assembly are organotypic density, architecture and stiffness, maximal cell-to-cell communication and movement among different cell types, self-produced ECM, and cell-driven biomechanical forces (1). Electrical signal in the heart is propagated through gap junctions; thus, interconnectivity is essential to determine the potential for arrhythmogenesis. To determine the full roll of CFs in CM regulation, CM-CF interactions must be evaluated in these controllable but more biomimetic environments.

In this study, we used a scaffold-free 3-D model we previously developed, in which cell-cell interactions between neonatal rat ventricular CMs and CFs dominate but the cells can also interact via paracrine and ECM molecules. We have previously shown self-assembly into viable microtissues with intermingled CMs and CFs, reminiscent of the myocardium, with characteristic CM resting membrane potential (RMP) and AP shape (17). Our original study pointed to the importance of CF inclusion, as AP was prolonged when CFs were interspersed with CMs and showed feasibility for cell-type specific gene manipulation. In the present study, we selectively enhanced G_q signaling in CFs to examine the effect of activated CFs on the structural microenvironment and CM electrical function. We chose the G_q signaling pathway because it is commonly activated in the diseased heart, when the release of angiotensin II (ANG II), endothelin-1 (ET-1), and other factors in response to hemodynamic stress and injury leads to the activation of G_q protein-coupled receptors (G_qPCRs) on both CMs and CFs (49). To examine the effect of enhanced G_q signaling when it is restricted to CFs on CM function in a 3-D microtissue environment, we overexpressed the G_q protein α-subunit or constitutively active Gα_q (Gα_qQ209L) via adenoviral gene transfer in CFs before engineering of the microtissues. Here, we show that enhanced G_q signaling in CFs increased the size of microtissues also containing CMs as well as early profibrotic and prohypertrophic marker gene expression. Optical mapping showed prolongations in CM AP and intracellular Ca²⁺ transient durations and rise times as well as a markedly increased occurrence of abnormal electrical activity based on depolarization-induced automaticity. Increases in the CF resting membrane potential (RMP_CF) and connexin (Cx) expression in CFs with enhanced G_q signaling were computationally modeled for their functional significance, as was the importance of the 3-D environment on the observed arrhythmogenic phenotype. Our findings suggest that elevated CM RMP via enhanced coupling to more depolarized CF_GαqQL, observed in a microtissue context to our knowledge for the first time in this study, may contribute under pathophysiological conditions to electrophysiological anomalies and arrhythmogenicity in the intact heart together with potential paracrine effects.

MATERIALS AND METHODS

Reagents.

The following reagents were purchased from Sigma-Aldrich (St. Louis, MO): trypsin from the porcine pancreas, FBS, 100 U/ml penicillin and 100 mg/ml streptomycin, bromodeoxyuridine (BrdU), glucose, lithium chloride, sucrose, hematoxylin, eosin, BSA, NaCl, KCl, MgCl₂, CaCl₂, NaOH, borax, ammonium formate, sodium formate, and carbenoxolone. The following reagents were purchased from ThermoFisher Scientific (Waltham, MA): DMEM/F-12, HEPES, trypsin-EDTA, formic acid, human embryonic kidney (HEK)293A cells (RRID:CVCL_6910), Superfrost Plus Microscope Slides, TRIzol, agarose, Live/Dead Viability/Cytotoxicity kit, CellTracker Orange CMRA, ProLong Gold antifade reagent containing 4′,6-diamidino-2-phenylindole, TaqMan Reverse Transcription Reagents, TaqManGene Expression Assays, TaqMan Gene Expression master mix, SuperSignal West Pico or Femto Substrate, di-4-ANEPPS, and Rhod2-AM. The sources for all other reagents are specified below.

Isolation and 2-D culture of neonatal rat ventricular CMs and CFs.

The cardiac tissue harvesting procedures conformed with the Guiding Principles in the Care and Use of Vertebrate Animals in Research and Training, and the protocol was approved by the Institutional Animal Care and Use Committee of Rhode Island Hospital. CMs and CFs were isolated from postnatal day 2 Sprague-Dawley rats as previously described (17). Serial enzymatic digestion of isolated ventricles with trypsin (1 mg/ml) and collagenase type 2 (0.6 mg/ml, Worthington Biochemical, Lakewood, NJ) followed by discontinuous Percoll (GE, Fairfield, CT) gradient centrifugation of the cardiac cell suspension yielded CM- and CF-enriched cell fractions. To allow the cells to recover from isolation, CMs and CFs were cultured in 2-D (3 ×10⁶ cells/100-mm dish) in DMEM/F-12 supplemented with 10% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin (referred to as “medium”). The medium for CMs was supplemented with 50 mM BrdU to inhibit the proliferation of the few CFs present in the CM fraction (see below). After 3 days, cells were detached with trypsin-EDTA (3–4 min at 37°C) and replated into nonadhesive hydrogels to generate 3-D microtissues (see below).

Average cell yields per pup were 2.4 ± 0.8 × 10⁶ CMs/pup and 3.4 ± 1.1 × 10⁶ CFs/pup. The purity of CM and CF fractions was 94.9 ± 2.3% and 97.3 ± 1.6%, respectively, as determined by immunostaining of fixed cells after 3 days in 2-D culture with antibodies recognizing cell type-specific markers [i.e., vimentin for CFs and α-sarcomeric actinin (αSA) for CMs] as previously described (17). After trypsinization, 61.2 ± 17.6% of CMs and 50.0 ± 15.3% of CFs were recovered compared with the original cell numbers seeded.

Adenoviral infection of CFs.

Adenoviruses encoding wild-type or constitutively active Gα_qQ209L (Ad-Gα_q and Ad-Gα_qQL, respectively) were kindly provided by Dr. J. Heller Brown (University of California, San Diego, CA). Empty adenovirus (Ad-Ctr) and uninfected cells served as negative controls, and Ad-GFP was used to visualize the effectiveness and uniformity of multiplicity of infection (MOI)-dependent infection. All adenoviruses were amplified in HEK293A cells and titered using an Adeno-X qPCR Titration Kit (Takara Bio USA, Mountain View, CA).

Ad-Gα_q and Ad-Gα_qQL were originally generated by Adams et al. (2) and used to activate Gα_q signaling in neonatal rat ventricular CMs in that study. In the present study, adenoviral infections of neonatal rat ventricular CFs in 2-D culture were achieved at the intended MOIs by cell counting at the day of infection, as previously described for adult rat ventricular CFs (93). To generate cardiac microtissues with adenovirus-infected CFs, before being cell seeded in 3-D culture, CFs were infected in suspension after trypsinization and cell counting in serum-free DMEM/F-12 for 2 h. Cells were then collected by centrifugation and resuspended in medium, which was repeated twice before seeding in 3-D (or 2-D) culture as described below.

Fabrication of hydrogels and 3-D culture.

Scaffold-free 3-D microtissues were generated using nonadhesive agarose gels with cylindrical microwells with hemispherical bottoms to guide self-assembly. Sterilized 2% (wt/vol) agarose was pipetted into molds designed for 6-, 12-, or 24-well plates with 400- or 800-μm-diameter rounded pegs (Microtissues, Providence, RI). After being cooled to room temperature (<5 min), the agarose gels were separated from the molds and transferred to single wells of 6-, 12-, or 24-well plates. For equilibration, medium was added to each well (3 ml for 6-well plates, 2 ml for 12-well plates, or 1 ml for 24-well plates), and air bubbles were removed with a vacuum chamber (400 μm diameter only, Lindberg/Blue M, ThermoFisher Scientific). Hydrogels were equilibrated overnight at 37°C in a humidified incubator with 5% CO₂.

Depending on the specific applications, CFs in suspension were added alone or in combination with CMs (at 1:1 or 1:2 ratios) to the center of the hydrogel seeding chamber with the indicated number of recesses (100 μl for 24-well format with 35 recesses, 225 μl for 12-well format with 81 recesses, and 500 μl for 6-well format with 822 recesses) and allowed to settle into the recesses for 30 min. Medium was then added to each well (3 ml for 6-well molds, 2 ml for 12-well molds, and 1 ml for 24-well molds), and cells were cultured for 2−4 days as indicated.

Phospholipase C-β activity assay.

Total inositol phosphate (IP) formation, an indirect measure of phospholipase C (PLC)-β activity, was assayed in 2-D culture as previously described for adult CFs (93). To determine total IP formation in 3-D cultures, microtissues were collected from 6 hydrogels in the 12-well format (3 × 10⁵ cells each) per condition 48 h after adenoviral infection (see above). They were harvested by inverting hydrogels in PBS and centrifuging the culture plates at 63 g for 2 min. Once removal from the hydrogel was confirmed by light microscopy, the microtissues in PBS from each hydrogel were collected into a microcentifuge tube and centrifuged (19 g for 1 min at room temperature). The supernatant was gently aspirated and the microtissues were pooled. Microtissues were then labeled with myo-[³H]inositol (4 µCi/ml; Perkin Elmer, Waltham, MA) in inositol-free DMEM (MP Biomedical, Solon, OH) for 4 h at 37°C, with inversion every 15–30 min for mixing. They were then briefly centrifuged, 10 mM LiCl was added to each tube, tubes were inverted to mix, and samples were incubated for 1 h at 37°C. After gentle centrifugation (82 g, 1 min, 4°C), media were removed, and the microtissues were washed with PBS once. IPs were extracted with 1 ml of 20 mM formic acid on a nutator at 4°C for 1 h. Extracts were neutralized to pH 7.5 with a solution containing 150 mM KOH and 7.5 mM HEPES and centrifuged at 4°C for 2 min at 16,000 g. Clear supernatants were loaded onto 0.5-ml Dowex AG1-X8 (Bio-Rad, Hercules, CA) anion exchange columns, equilibrated with 2 ml of 1 M NaOH followed by 2 ml of 1 M formic acid and 5 × 5 ml of H₂O. After sample application, columns were washed with 5 ml of H₂O and 5 ml of 5 mM borax-60 mM sodium formate before total IPs were eluted with 3 ml of 0.9 mM ammonium formate-0.1 M formic acid directly into vials containing 15 ml of Ultima Gold XR scintillation cocktail (Perkin Elmer). Each sample was assayed in triplicate.

Live/Dead and CellTracker staining.

Cell viability of microtissues was examined using the Live/Dead Viability/Cytotoxicity kit. Cells were stained with 2 μM calcein-AM and 4 μM ethidium homodimer-1 (in 1 ml PBS) at 37°C for 30 min followed by image acquisition. Before seeding with Ad-GFP-infected CFs, CMs were labeled with CellTracker Orange CMRA (10 μM) as previously described (17).

Image acquisition and processing.

Phase-contrast and epifluorescent images of cells and microtissues were captured with a Nikon TE2000-U and a black and white/color digital camera (MicroVideo Instruments, Avon, MA) and acquired and analyzed with NIS Elements software. Confocal images were acquired with a Nikon C1si confocal (Nikon) using diode lasers at 488 and 561 nm as previously described (17).

Microtissue size analysis.

Stitched 4× phase-contrast images of whole 24-well microtissue hydrogels (0.6 × 10⁶ cells each) were acquired and analyzed. Image thresholding and particle size analysis was used in NIS Elements to determine the top view cross-sectional area of individual microtissues across each mold.

3-D tissue sections.

Microtissues in 24-well hydrogels (0.6 × 10⁶ cells) were fixed with 4% (vol/vol) paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) and 8% (wt/vol) sucrose in PBS overnight at room temperature. Molds were then rinsed twice with PBS and fully equilibrated (as indicated by their sinking, usually over 12 h) with 15% and then 30% (wt/vol) sucrose in PBS. Whole agarose gels containing microtissues were removed from sucrose, blotted dry, and embedded in Tissue-Tek CRYO-OCT Compound (Ted Pella, Redding, CA). The microtissues are free floating in the cylindrical recesses and can be located at different heights within the mold during the embedding process. Blocks were stored at −80°C, sectioned on a Leica CM3050 cryostat microtome (Leica Biosystems, Buffalo Grove, IL) into 10-μm-thick sections, and placed on Superfrost Plus slides. After being air dried for 15 min, sections were postfixed in 4% paraformaldehyde in PBS.

Histological analysis.

Cryosections were stained with hematoxylin and eosin and Sirius Red, after they were thawed for 10 min at 37°C, placed in 95% ethanol for 1 min, and rinsed with distilled H₂O for 30 s. For hematoxylin and eosin, a 4-min incubation in hematoxylin and a rinse with running tap water for 2 min were followed by five dips into 1% HCl in 70% alcohol and washes in running tap water and 95% ethanol (1 min each). Slides were then stained in eosin for 25 s and rinsed with 95% ethanol for 15–20 s (with agitation). For Sirius Red, slides were placed in Sirius Red stain (Electron Microscopy Sciences) for at least 1 h and washed twice for 30 s in 0.5% acetic acid. All slides were then rinsed three times in 100% ethanol for 30 s, cleared in xylene for 5 min, and mounted with SHUR/Mount (Electron Microscopy Sciences).

Western blot analysis.

Microtissues were harvested from two to three hydrogels in the 6-well format (1 × 10⁶ cells each) after 2−4 days in 3-D culture as described above, lysed in 0.2 ml Cell Lysis Buffer (Cell Signaling Technology, Danvers, MA) containing complete mini protease inhibitor cocktail (Roche, Pleasanton, CA), and sonicated for 3 × 10 s on ice. Protein concentrations were measured with a Bradford microassay (Bio-Rad, Hercules, CA) with BSA as the standard. Equal amounts of protein (10 μg/lane) were separated on 4–20% TGX Gels (Bio-Rad), transferred to nitrocellulose membranes, and probed with the following primary antibodies at indicated dilutions: α-smooth muscle actin (αSMA; 1:400, RRID:AB_262054, Sigma), Cx43 (1:5,000, RRID:AB_2294590, Cell Signaling), Cx45 (1:1,000, RRID:AB_91014, Millipore), fibronectin (FN; 1:1,200, RRID:AB_476976, Sigma), GAPDH (1:1,000, RRID:AB_561053, Cell Signaling), Gα_q/11/14 (1:100, RRID:AB_10842057, Santa Cruz Biotechnology), and vimentin (1:500, RRID:AB_477627, Sigma). After incubation with the appropriate peroxidase-coupled secondary antibodies, proteins of interest were visualized by chemiluminescence (SuperSignal West Pico or Femto substrate).

mRNA analysis.

Microtissues were pooled from two to three hydrogels in the 12-well format (1.4 × 10⁶ cells each) for each condition after 3 days in 3-D culture as described above. They were sonicated for 3 × 10 s on ice in 1.0 ml TRIzol before extraction of total RNA. Total mRNA from CFs in 2-D culture (0.875 × 10⁶ cells/6-well plate) was isolated using the RNeasy Plus Mini Kit (Qiagen, Valencia, CA) and pooled from two to three wells. All RNA samples were reverse transcribed using TaqMan Reverse Transcription Reagents and subjected to quantitative PCR (qPCR) using FAM-labeled TaqMan MGB probes for the indicated genes plus 18S and TaqMan Gene Expression Master Mix according to the manufacturer’s instructions. Absence of template during qPCR was used as negative control. qPCR cycling was performed at 95°C for 10 min and then 95°C for 15 s and 60°C for 1 min for a total of 45 cycles. Each sample was assayed in duplicate, and the threshold cycle (C_T) values corresponding to the qPCR cycle number at which fluorescence emission in real time reaches a threshold above the baseline emission were determined using QuantStudio Real-Time PCR software. Data were analyzed by the comparative C_T method (78) and are presented as fold changes in mRNA expression normalized to 18S and relative to microtissues containing Ad-Ctr-infected CFs (CF_Ctr or CM:CF_Ctr).

Optical mapping.

An Olympus MVX10 microscope was used to image 1.2 × 1.2-mm² regions. Microtissues were loaded with voltage-sensitive di-4-ANEPPS (5 μM for 10 min at 35°C) for measurements of membrane potential (V_m) or Rhod2-AM (4 μM) for cytosolic [Ca²⁺]_i. Fluorescence images were acquired at 979 frames/s using a Photometrics Evolve +128 EMCCD camera (2 × 2 binning to 64 × 64 pixels, 18.7 × 18.7-µm² resolution, 1.2 × 1.2-mm² field of view) and an Olympus MXV10 macroview optical system. Fluorescence images were filtered using nonlinear bilateral filter (spatial filter: 5 × 5 window, temporal filter: 13 point window) to preserve AP upstrokes from blurring. Typically, four microtissues were recorded simultaneously per scan at this magnification. A single spheroid is typically covered by ~60 pixels at this magnification. All 35 microtissues in 24-well hydrogels (0.6 × 10⁶ cells) were recorded. To evaluate automaticity, all tissues were initially recorded for 5 s without stimulation. Although longer time recording (20 s) may detect spontaneous activity from CM:CF_CTR, a recording time of 5 s was chosen to avoid potential phototoxic damages caused by extended excitation light exposure that can promote spontaneous activity. Tissues were then field stimulated at a basic pacing cycle length of 1 Hz, with a biphasic stimulation pulse strength of 10 V/cm and duration of 4 ms. Carbenoxolone (CBX; 100 μM) was then added for 10 min to partially inhibit gap junctions, and the same pacing protocol was repeated. For washout experiments, automaticity was recorded without pacing, after perfusion of CBX for 30 min, and after washout for 30 min.

Resting membrane potential recordings.

Standard microelectrode techniques were used for membrane potential recordings with a 15-MΩ microelectrode with pipette filled with 3 mol/l KCl. Because CFs and CMs cannot be distinguished in 3-D cultures, spherical microtissues containing only infected CFs were formed in hydrogels (24-well format, 0.6 × 10⁶ cells) and transferred to coverslips after 2 days of 3-D culture. Tissues were bathed in Tyrode solution containing (in mmol/l) 140 NaCl, 5.4 KCl, 1 MgCl₂, 10 HEPES, 1.8 CaCl₂, and 5.5 glucose (pH 7.4). Recordings were low-pass filtered at 100 kHz and digitized at 200 kHz. V_m was also recorded in infected CFs in standard 2-D cell cultures for comparison using an Axopatch-200B amplifier (Molecular Devices, Foster City, CA) and the perforated current-clamp technique. For V_m recordings, pipettes were filled with (in mmol/l) 120 potassium gluconate, 20 KCl, 5 NaCl, 5 HEPES, and 5 MgATP (pH 7.2). Cells were bathed in Tyrode solution, and β-escin (50 μM) was added to the pipette solution. Pipette resistances were 5–8 MΩ. Recordings were low-pass filtered at 10 kHz and digitized at 20 kHz. All experiments were carried out at room temperature.

Computational modeling of CM-CF coupling.

Effects of RMP_CF and gap junction coupling on CM APs were studied using a computational CM AP model that is electrotonically connected with several CFs. The V_m of a CM or CF in a coupled CM-CF pair is governed by the following equation:

$$C_{\text{CM}}\frac{\text{d}{V}_{\text{CM}}}{\text{d}t}=-I_{\text{ion}}+{\displaystyle \sum }_{k=1}^n{g}_{\text{gap,CM-CF}}\left(V_{\text{CF,}k}-V_{\text{CM}}\right)$$

where C_CM = 125 pF is the membrane capacitance of a CM, V_CM is the V_m of a CM, t is time, I_ion is the corresponding membrane current, n is the number of coupled CFs to a single CM (CM connected to n CFs, and a CF has only one connection with the CM), and g_gap,CM-CF is the gap junction conductance between the CM and the kth connected CF whose V_m is V_CF,k. We used the Luo and Rudy AP model (52) with a modified maximum conductance of inward rectifier K⁺ current (I_K1; G_K1,max = 0.5) to match with our previous experimental results (17). The passive model CF was modeled as follows (90):

$$C_{\text{CF}}\frac{\text{d}{V}_{\text{CF}}}{\text{d}t}=-I_{\text{CF}}+g_{\text{gap,CM-CF}}\left(V_{\text{CM}}-V_{\text{CF}}\right)$$

where C_CF = 25 pF, CF current (I_CF) = g_m,CF (V_CF − RMP_CF), and g_m,CF is the membrane conductance of a CF. In this study, RMP_CF was varied from −50 to 0 mV, and the coupling conductance between CM-CF (g_gap,CM-CF) from 0 to 10 nS to investigate the effects of RMP_CF and g_gap,CM-CF on automaticity. g_m,CF was set to 4 nS based on previous experimental and computational studies (13, 39, 44, 48, 62, 74, 79, 80, 89, 90). Numerical calculation of the differential equations was performed using the Euler method with a time step of Δt = 0.01 ms.

To investigate the effect of CM-CF interactions in a 3-D environment, a cube of 25 × 25 × 25 cells was modeled, which had a sufficient number of cells to reflect a single microtissue. In the 3-D cube, each cell was connected to six neighboring cells through gap junctions, and the CMs and CFs were randomly distributed in the cube with CM:CF = 1:1 or 1:2 ratio. g_gap,CM-CM was set to 10 nS, and g_gap,CM-CF and g_gap,CF-CF were set to 5 nS. This simulation was carried out on a multicore graphic processing unit (Geforce GTX Titan, NVidia) with double precision.

Statistical analyses.

Technical replicates were averaged to a single value. Values are expressed as means with SD. n is reported as the number of microtissues/cells, number of replicates, and number of repeated experiments where appropriate and as indicated. Statistical analyses were performed using one-way ANOVA with Bonferroni post hoc tests, Student’s two-tailed paired and unpaired t-test, or Fisher’s exact test as appropriate.

RESULTS

Microtissue model development with CMs interspersed with G_q-activated CFs.

To study the modulation of CM electrical activity by Gα_q-activated cardiac CFs, we leveraged a scaffold-free cardiac microtissue model we previously developed in which neonatal rat ventricular CMs and CFs self-assemble in nonadhesive hydrogels to form spheroidal microtissues with high interspersion of the two cell types reminiscent of the myocardium (17). G_q activation was achieved in CFs with adenoviruses encoding the α-subunit of heterotrimeric G_q protein (Gα_q) or a constitutively active mutant (Q209L) of Gα_q [Gα_qQL (16)]. First, we confirmed the effectiveness of adenovirus-mediated gene transfer in neonatal rat CFs. Figure 1A shows the uniformity and efficiency of adenoviral infections in CFs in 2-D cultures: although there was variability in fluorescent intensity between individual cells 3 days after infection with Ad-GFP at three different MOIs, the number of infected CFs increased in a MOI-dependent manner. The number of GFP-positive CFs (as a percentage of total cells identified by nuclear Hoechst staining) was 44.1 ± 2.8% at MOI 2.5 (n = 120), 65.2 ± 8.0 at MOI 5 (n = 136), and 73.7 ± 8.7% at MOI 10 (n = 154). We next assessed the effect of adenovirus-mediated gene transfer of wild-type or constitutively active Gα_q in CFs on both Gα_q protein expression (using an antibody recognizing Gα_q and the closely related Gα₁₁ isoform; Fig. 1B) and G_q signaling (using total IP formation as an indirect measure for the activity of PLC-β, the canonical direct downstream target of Gα_q; Fig. 1C): despite a more modest increase in Gα_q protein expression in CFs infected with Ad-Gα_qQL compared with Ad-Gα_q at MOI 10 (3.3-fold vs. 17-fold of Ad-Ctr), Ad-Gα_qQL induced a more pronounced increase in PLC-β activity than Ad-Gα_q (8.6-fold vs. 6.5-fold of Ad-Ctr). This was expected based on prior studies in other cells (see discussion) and is due to the Q209L point mutation that renders the protein constitutively active.

Fig. 1.

Adenoviral infection of neonatal rat ventricular cardiac fibroblasts (CFs). A: representative merged epifluorescent and phase contrast images of neonatal rat ventricular CFs in two-dimensional (2-D) cell culture, 3 days after infection with Ad-GFP (or Ad-Ctr) at indicated multiplicities of infection (MOIs). Cells were stained with Hoechst for 20 min before mounting. Scale bar = 100 μm. B: representative Western blot of Gα_q/11 expression in cell lysates (10 µg protein/lane) collected from CFs in 2-D cultures, 3 days after infection with indicated adenoviruses at MOI 10 (left). GAPDH was used as the loading control. Group data are means ± SD (right). *P < 0.05 vs. Ad-Ctr; #P < 0.05 vs. Ad-Gα_q (ANOVA). C: basal inositol phosphate (IP) production in CFs 72 h after infection with the indicated adenoviruses at MOI 2.5, 5, and 10. Three wells from a 6-well plate were pooled for each sample (n = 3 for each condition). Similar results were obtained in two other experiments testing a total of two and four batches of Ad-Gα_q and Ad-Gα_qQL, respectively. D: representative confocal images from sections of cardiomyocyte (CM):CF microtissues composed of a 10:1 or 1:1 ratio of CellTracker Orange-labeled CMs (CM_CT-Orange) and Ad-GFP-infected CFs (CF_Ad-GFP) 2 days after coseeding in hydrogels. Scale bar = 50 μm.

To generate cardiac microtissues with CF-restricted activation of Gα_q signaling, adenoviral infections of CFs were performed before coseeding with CMs into the hydrogels and self-assembly. To confirm CF-restricted adenoviral infection, Ad-GFP-infected CFs (CF_Ad-GFP) were coseeded with CellTracker Orange CMRA-labeled CMs (CM_CT-Orange). Minimal fluorescent overlap on confocal images in CM_CT-Orange:CF_Ad-GFP microtissues at 10:1 and 1:1 ratios 2 days after coseeding supported Ad-mediated gene transfer predominantly in CFs (Fig. 1D).

We next assessed the effect of CF-restricted Ad-Gα_q and Ad-Gα_qQL infections on total IP formation in 3-D microtissues composed of either infected CFs alone (CF_Ad) and in combination with equal numbers of uninfected CMs (CM:CF_Ad; Fig. 2A). Since microtissues containing uninfected or Ad-Ctr-infected CFs showed similar results, only Ad-Ctr data are shown. In microtissues composed of CFs only, Ad-Gα_q and Ad-Gα_qQL increased PLC-β activity 6.1-fold and 7.3-fold compared with Ad-Ctr, respectively, 3 days after infection (Fig. 2A). CM:CF_Ad microtissues also showed markedly elevated PLC-β activity, but the effect of Ad-Gα_qQL was less pronounced than Ad-Gα_qWT (4.9 ± 1.0-fold and 8.5 ± 1.2-fold, respectively; Fig. 2A). Similar results were observed in 2-D cocultures of CMs and adenovirus-infected CFs (5.0 ± 0.5-fold for Ad-Gα_qQL and 7.9 ± 0.1-fold for Ad-Gα_q; data not shown).

Fig. 2.

Cardiac fibroblast (CF)-restricted Gα_q activation in spherical cardiac microtissues. A: basal inositol phosphate (IP) production in microtissues after adenoviral infection of CFs with the indicated adenoviruses for 2 h before cell seeding in nonadhesive hydrogels alone (CF_Ad) or with cardiomyocytes (CMs) (CM:CF_Ad). Microtissues from 6 hydrogels in the 12-well format were pooled for each sample (n = 3 for each condition). *P < 0.05 vs. control (Ctr); #P < 0.05 vs. Gα_q (ANOVA). B: fluorescent images of representative Live/Dead-stained whole CM:CF_Ad microtissues containing the indicated infected CFs 3 days after infection and cell seeding. Scale bar = 100 µm. C and D: bright-field images of representative hematoxylin and eosin (C)- and Sirius Red (D)-stained cryosections (10 µm thick) of CM:CF_Ad microtissues fixed after 3 days in three-dimensional (3-D) culture. Scale bars = 100 µm. E: area of CM:CF_Ad microtissues after 3 days in 3-D culture. Values are means ± SD; n = 65–70 microtissues/condition. *P < 0.05 (ANOVA).

Microtissues containing Ad-Gα_q-infected CFs were viable, as illustrated by representative Live/Dead-stained CM:CF microtissues (Fig. 2B). Consistent with our prior study (17), immunostaining with antibodies against αSA and vimentin confirmed that CMs and CFs self-assembled to be highly interspersed (data not shown). Hematoxylin and eosin staining of microtissue cryosections did not reveal any obvious differences among CM:CF_Ctr, CM:CF_Gαq, and CM:CF_GαqQL at various depths into the microtissue (Fig. 2C), and representative images of Sirius Red-stained sections suggested no gross differences between CM:CF microtissues containing Gα_q-activated CFs compared with control CFs (Fig. 2D). Although the images shown are likely from different planes because the location of the microtissues within the recesses cannot be controlled during the embedding process (see materials and methods), they are representative of the morphology seen across different planes and spheroids.

Gα_qQL expression in CFs increased the microtissue area of CM:CF_GαqQL microtissues over CM-CF_Ctr (Fig. 2E). CM:CF_Ctr microtissues were 44,900 ± 16,400 μm² (n = 69 microtissues) in cross-sectional area, which equates to a radius of 120 ± 22 μm and a volume of 7.15 × 10⁶ ± 3.91 × 10⁶ μm³, whereas CM:CF_GαqQL microtissues were 55,800 ± 19,100 μm² (n = 70 microtissues) in area (r = 133 ± 23 μm, volume = 9.85 × 10⁶ ± 5.11 × 10⁶ μm³, P < 0.05). The mean values for CM:CF_Gαq microtissues (n = 65 microtissues) were higher than CM:CF_Ctr and lower than CM:CF_GαqQL (49,900 ± 18,900 μm², r = 126 ± 24 μm, volume = 8.38 × 10⁶ ± 4.79 × 10⁶ μm³) but did not reach statistical significance (Fig. 2E).

We next examined the expression of select marker genes for MFb (αSMA) and fibrogenesis (CTGF and FN1) using qPCR analysis. Vimentin was used for comparison because it is expressed in both unactivated and activated CFs in vitro and in vivo (3). First, we compared their expression in 3-D CF microtissues composed of control CFs to cells cultured as a monolayer in conventional 2-D culture for the same duration (3 days): mRNA levels of αSMA, CTGF, and FN1 were markedly reduced in 3-D culture (by 90 ± 4%, 96 ± 2%, and 59 ± 12%, respectively), whereas vimentin expression was comparable to 2-D culture (Fig. 3A). Next, we assessed the effect of CF activation by Gα_q on the expression of these markers in microtissues containing CFs either alone (Fig. 3B) or together with CMs (Fig. 3C). In contrast to vimentin, αSMA mRNA was increased (3.8 ± 1.1-fold) in microtissues composed of Gα_q-expressing CFs but not those composed of Gα_qQL-expressing CFs, whereas CTGF was increased to a comparable level in both (3.3 ± 1.4-fold in CF_Gαq and 3.7 ± 0.9-fold in CF_GαqQL) and FN1 was increased more in CF_GαqQL (4.6 ± 2.0-fold) than CF_Gαq (2.0 ± 0.4-fold; Fig. 3B). We observed similar trends in CM:CF microtissues containing Gα_q-activated CFs (Fig. 3C). The only major difference was a more pronounced increase for CTGF in CM:CF_GαqQL (6.5 ± 0.1-fold) than CM:CF_Gαq (2.6 ± 0.5-fold), which points to a contribution from the CMs. The representative Western blots shown in Fig. 3D demonstrate that vimentin, αSMA, and FN1 protein expression correlated overall with the mRNA levels in CM:CF microtissues. Furthermore, CM:CF microtissues containing Gα_q-activated CFs had comparable αSA mRNA levels to CM:CF_Ctr but increased atrial natriuretic factor (ANF) mRNA (6.4 ± 1.2-fold in CM:CF_Gαq and 9.2 ± 1.4-fold in CM:CF_GαqQL; Fig. 3E).

Fig. 3.

Cardiac fibroblast (CF)-restricted Gα_q overexpression leads to CF activation and atrial natriuretic factor (ANF) expression. CFs were infected with the indicated adenoviruses for 2 h before being seeded. CFs were seeded alone in standard two-dimensional (2-D) culture (A) or were seeded in hydrogels in three-dimensional (3-D) culture alone (CF_Ad; B) or together with CMs (1:1, CM:CF_Ad) (C–E). After the indicated days, cells or microtissues were harvested for mRNA extraction (pooled from 2−3 wells for each condition, 6-well and 12-well plates for 2-D and 3-D, respectively) or protein extraction (pooled from 3 hydrogels in 6-well format for each condition). A: relative mRNA expression for the indicated proteins assessed by quantitative PCR (qPCR) analysis of cells cultured in 2-D monolayers or in 3-D as described. Values are means ± SD; n = 5–6 per condition. *P < 0.05 vs. 2-D (Student’s t-test). B and C: relative mRNA expression for the indicated proteins in CF_Ad (B) or CM:CF_Ad microtissues (C) assessed by qPCR analysis after 3 days in 3-D culture. Values are means ± SD for n = 6 (B) or n = 3 (C) samples/condition. *P < 0.05 vs. control (Ctr); #P < 0.05 vs. Gα_q (ANOVA). D, top: representative Western blots of microtissue lysates after 2 days in 3-D culture (10 µg protein/lane) probed with the indicated antibodies. GAPDH was used as the loading control. Similar changes were observed after 3 and 4 days in culture (not shown). Bottom: pooled group data. Values are means ± SD; n = 4–7 samples/condition. *P < 0.05 vs. Ctr (ANOVA). E: relative mRNA expression of the indicated proteins in CM:CF_Ad microtissues assessed by qPCR analysis after 3 days in 3-D culture. Values are means ± SD; n = 3 samples/condition. *P < 0.05 vs. Ctr (ANOVA). αSA, α sarcomeric actinin; αSMA, α-smooth muscle actin; CTGF, connective tissue growth factor; FN, fibronectin; Vim, vimentin.

APs and Ca²⁺ transients in CM:CF microtissues containing G_q-activated CFs.

To investigate the modulation of CM electrical activity by CFs, optical mapping was used to record APs and [Ca²⁺]_i from paced (1 Hz) CM:CF microtissues using voltage- and Ca²⁺-sensitive dyes, respectively. Voltage and Ca²⁺ could not be recorded in the same microtissues, but the data presented were all recorded from the same batches of isolated cells. Representative AP traces and APD maps from CM:CF microtissues are shown in Fig. 4A. APD at 75% repolarization was comparable for CM:CF_Ctr and CM:CF_Gαq but prolonged in CM:CF_GαqQL by more than 20 ms (95.7 ± 5.6 vs. 71.8 ± 9.1 ms in CM:CF_Ctr). The number of CM:CF_GαqQL microtissues analyzed was lower compared with the other conditions because of automaticity that persisted despite field stimulation (see below). G_q activation in CM:CF microtissues also altered AP rise time, which was prolonged in both CM:CF_Gαq (13.4 ± 3.2 ms) and CM:CF_GαqQL (13.8 ± 1.4 ms) compared with CM:CF_Ctr (10.5 ± 2.6 ms; Fig. 4B). Representative [Ca²⁺]_i traces and duration maps from optical mapping of Rhod2-AM-loaded CM:CF microtissues are shown in Fig. 4C. The Ca²⁺ transient duration nearly doubled in both CM:CF_Gαq (314.1 ± 90.1 ms) and CM:CF_GαqQL (330.1 ± 83.7 ms) over CM:CF_Ctr (166.8 ± 36.6 ms), and the Ca²⁺ transient rise time increased by more than 6 ms in CM:CF_Gαq (25.8 ± 6.7 ms) and by more than 12 ms in CM:CF_GαqQL (31.4 ± 9.6 ms) compared with CM:CF_Ctr (19.0 ± 3.0 ms; Fig. 4D).

Fig. 4.

Gα_q activation in cardiac fibroblasts (CFs) prolongs action potential (AP) and Ca²⁺ transient duration and rise time in cardiomyocyte (CM):CF microtissues. CMs and CFs infected with the indicated adenoviruses were coseeded at a 1:1 ratio in hydrogels. After 3 days in three-dimensional (3-D) culture, microtissues were loaded with di-4-ANEPPS or Rhod2-AM for optical mapping of membrane potentials (V_m; left) or Ca²⁺ transients ([Ca²⁺]_i; right) under paced conditions (1 Hz). A and C: representative traces and duration maps of paced V_m and [Ca²⁺]_i recordings obtained from the indicated microtissues paced at 1 Hz. B and D: quantification of the duration at 75% return (left) and rise time (right) of APs and [Ca²⁺]_i in CM:CF_Ad (1:1) microtissues. Values were averaged from all the pixels of single spheroids (~60 pixels/spheroid). Values are means ± SD for the indicated numbers. *P < 0.05 vs. control (Ctr); #P < 0.05 vs. Gα_q (ANOVA). Data shown are from a representative experiment in which both V_m and [Ca²⁺]_i were recorded in the same batch of microtissues. Comparable recordings and differences between the experimental groups were obtained in two and four additional independent experiments for V_m and [Ca²⁺]_i, respectively. APD, AP duration.

Increased automaticity in CM:CF microtissues containing G_q-activated CFs.

Although neonatal rat CMs spontaneously contract in culture, depolarization-induced automaticity is not typical. Since we observed persistent automaticity in CM:CF_GαqQL when the microtissues were subjected to electrical field pacing (see above), we next examined the occurrence and frequency of spontaneous membrane potential activity in CM:CF microtissues without pacing. As shown in Fig. 5, A and B, none of the interspersed CM:CF microtissues with control CFs showed any spontaneous V_m activity in the 5-s intervals acquired, regardless of the ratio of CM:CF (1:1 or 1:2). In CM:CF_Gαq with a 1:1 ratio, a few microtissues showed depolarization-induced spontaneous V_m activity (16%), and the occurrence of this automaticity increased to 50% with a 1:2 CM:CF_Gαq ratio. In contrast, automaticity was observed in all CM:CF_GαqQL microtissues, even at a 1:1 ratio.

Fig. 5.

Gα_q activation in cardiac fibroblasts (CFs) causes automaticity in unpaced cardiomyocyte (CM):CF microtissues that is mitigated by partial gap junction inhibition. CMs and CFs infected with the indicated adenovirus were coseeded in 1:1 or 1:2 ratios in hydrogels. After 3 days in three-dimensional (3-D) culture, microtissues were loaded with di-4-ANNEPS and optically mapped in the absence of electrical stimulation. A: representative membrane potential (V_m) traces of 1:1 CM:CF_Ad microtissues illustrating automacity in unpaced microtissues. B: group data for the occurrence of automacity within 5-s recordings, expressed as a percentage of total microtissues recorded for each condition per hydrogel. *P < 0.05 (Fisher’s exact test). C: representative V_m traces of unpaced CM:CF_Gαq and CM:CFGα_qQL microtissues (1:1) from A that were optically mapped in the absence of electrical stimulation both before (black) and after (red) administration of carbenoxolone (CBX; 100 μM, 10 min), illustrating that automacity was mitigated by CBX. The traces recorded before and after CBX are from the same microtissue. D: group data for the occurrence of automacity within 5-s recordings before and after 100 μM CBX administration for 10 min (expressed as percentages of total microtissues recorded for each condition). E: unpaced CM:CFGα_qQL microtissues were perfused with 100 μM CBX for 30 min followed by a washout period of 30 min. Averaged frequency of the automaticity observed before CBX (n = 34), after CBX (n = 31), and after CBX washout (n = 27) is shown. Values are means ± SD. *P < 0.05 (Student’s t-test).

To elucidate the potential role of gap junctions in automaticity in CMs regulated by Gα_q-activated CFs, we treated CM:CF (1:1) microtissues with the gap junction inhibitor CBX at a concentration expected to block half of gap junctions [100 μM (68, 81)] for 10 min. CBX treatment abolished automaticity in CM:CF_Gαq microtissues (Fig. 5, C and D). Although the occurrence was not completely suppressed in CM:CF_GαqQL microtissues, their frequency was lowered significantly (from 1.6 ± 0.7 to 1.0 ± 0.3 Hz, n = 20, P < 0.05 by Student’s t-test). The importance of gap junctions in automaticity was further indicated with CBX perfusion and washout (30 min each; Fig. 5E). The frequency of automaticity was partially recovered after the washout (1.56 ± 0.42, 0.57 ± 0.29, and 0.81 ± 0.31 Hz for before, after CBX, and after washout, respectively, P < 0.01). Complete washout of CBX is implausible with our experimental timeline and 3-D tissue environment (21, 34).

Gap junction protein expression in microtissues containing G_q-activated CFs.

We then examined the expression of gap junction proteins at both mRNA and protein levels. To determine Cx expression changes in CFs on enhanced G_q signaling, we first examined CF microtissues without interspersed CMs. Analysis of Cx expression in CFs cultured in 2-D monolayer versus 3-D culture for 3 days revealed marked isoform-specific differences: Cx43 mRNA expression was 5.4 ± 1.8 times higher in 3-D than 2-D, whereas Cx45 expression was not significantly changed, and Cx40 expression was markedly decreased (by 92%) in 3-D (Fig. 6A). We then compared Cx43 and Cx45 mRNA expression changes in 3-D CFs with enhanced G_q signaling compared with CF_Ctr microtissues and observed upregulation of both Cx43 and Cx45 mRNAs, which was more pronounced in CF microtissues that expressed Gα_qQL (Fig. 6B): Cx43 mRNA was increased 2.9 ± 0.4 times with Gα_q overexpression and 5.1 ± 0.9 times with Gα_qQL expression compared with control. Cx45 mRNA expression in CF_Gαq was only 1.4 ± 0.3 times higher than in CF_Ctr but significantly increased (2.7 ± 0.9 times) in CF_GαqQL. Similar trends in mRNA changes were observed in microtissues composed of interspersed CMs and CFs (Fig. 6C), although it should be noted that baseline Cx expression was much higher in CM:CF microtissues than in CF microtissues (data not shown). Although overall less pronounced than in CF-only microtissues (see Fig. 6B), the relative increase in Cx43 and Cx45 mRNA on Gα_q activation in CFs interspersed with CMs was significant in CM:CF_GαqQL (Fig. 6C): Cx43 and Cx45 mRNA levels were increased 1.9 ± 0.1 and 1.6 ± 0.3 times compared with CM:CF_Ctr. Increased Cx43 expression was also observed at the protein level in CM:CF_Gαq and CM:CF_GαqQL compared with CM:CF_Ctr, whereas upregulation of Cx45 protein reached statistical significance in CM:CF_Gαq only (Fig. 6D). Taken together, these results point toward a potential role of enhanced CM-CF coupling with enhanced G_q signaling in CF.

Fig. 6.

Gα_q activation in cardiac fibroblasts (CFs) increases gap junction protein expression in CF_Ad and CM:CF_Ad microtissues. CFs were infected with indicated adenoviruses for 2 h before being seeded in hydrogels alone (CF_Ad) or together with CMs (CM:CF_Ad, 1:1). After 3 days in three-dimensional (3-D) culture, microtissues were harvested for mRNA extraction (pooled from 3 hydrogels in 12-well format per condition) and protein extraction (pooled from 3 hydrogels in 6-well format per condition). A: relative mRNA expression of indicated connexin (Cx) isoforms assessed by quantitative PCR (qPCR) analysis of CFs cultured in two-dimensional (2-D) monolayers for 3 days (pooled from 2−3 wells from a 6-well plate) or in 3-D as described. Values are means ± SD; n = 5–6 samples/condition). *P < 0.05 vs. 2-D (Student’s t-test). B and C: relative mRNA expression of Cx43 and Cx45 in CF_Ad and CM:CF_Ad assessed qPCR analysis after 3 days in 3-D culture. Values are means ± SD; n = 6 samples/condition. *P < 0.05 vs. control (Ctr); #P < 0.05 vs. Gα_q (ANOVA). D: representative Western blots (left) of microtissue lysates after 3 days in 3-D culture (10 µg protein/lane) probed with the indicated antibodies. Vimentin was used as the loading control. Similar results were obtained after 2 days (not shown). Group data (right) are means ± SD; n = 5 samples/condition. *P < 0.05 vs. Ctr (ANOVA).

Resting membrane potential in G_q-activated CFs.

Since the RMP of CFs is an important factor to determine the magnitude of current flow between CMs and CFs, we next measured RMP in Gα_q-activated CFs. Because CFs and CMs cannot be distinguished in 3-D CM:CF microtissues, RMP_CF recordings were acquired from CF microtissues through insertion of a microelectrode. In CF_Ctr and CF_Gαq microtissues, RMP amounted to −33.0 ± 8.4 and −31.9 ± 13.0 mV, respectively, whereas in CF_GαqQL microtissues it was depolarized by more than 10 mV to −20.0 ± 7.2 mV (Fig. 7, right). Additional perforated patch-clamp recordings were acquired in CFs in traditional 2-D culture to validate these findings, and we observed comparable results (Fig. 7, left): the RMP was overall slightly lower in CFs cultured in 2-D, but as in 3-D CF_GαqQL microtissues the RMP in 2-D CF_GαqQL was significantly elevated (−25.8 ± 9.5 mV) compared with 2-D CF_Ctr (−39.8 ± 6.3 mV). In 2-D CF_Gαq, RMP was slightly elevated (−29.7 ± 10.0 mV), but this did not reach statistical significance compared with 2-D CF_Ctr.

Fig. 7.

Elevated resting membrane potential (RMP) in Gα_qQL-activated cardiac fibroblasts (CFs). CFs were infected with the indicated adenoviruses for 2 h before being seeded alone in two-dimensional (2-D) or three-dimensional (3-D) cultures. For electrophysiology experiments, cells were digested after 3 days in 2-D culture and allowed to reattach for 6–12 h before being patch clamped, and microtissues were harvested after 2 days in 3-D culture and allowed to attach to coverslips for 24 h before being probed with a microelectrode. Depicted are resting membrane potentials from individual recordings and group data (means ± SD; n = 8–14 each). *P < 0.05 vs. control (Ctr; ANOVA).

Modeling the effects of gap junctional coupling and RMP on automaticity.

Our experimental results show that Cx43 and Cx45 expression was higher in CF_GαqQL and RMP_CF of CF_GαqQL was more depolarized. We hypothesized that the automaticity in 3-D CM:CF_GαqQL microtissues was associated with RMP_CF and electrotonic coupling conductance (g_gap,CM-CF) between CM and CF through gap junctions. To investigate the conditions of CM-CF coupling promoting automaticity, we used a computational model of CM AP electrically coupled with multiple passive CFs, as shown in Fig. 8A (see materials and methods for detailed model parameters). This computational model replicated the automaticity observed experimentally (Fig. 5). Figure 8B shows V_m traces of a CM (black) and a CF (red) when the CM and CFs were electrically coupled (g_gap,CM-CF = 3.3 nS). A map of automaticity frequency when two CFs were connected to the CM is shown in Fig. 8C. Representative V_m traces of the CM from five parameter sets from the frequency map are shown on the right. When CM-CF coupling strength and RMP_CF were low, RMP_CM remained at −80 mV, and no spontaneous activity occurred (Fig. 8C,a). When RMP_CF and CM-CF coupling strength were increased, AP oscillations appeared (Fig. 8C,b). The oscillations accelerated as g_gap,CM-CF (Fig. 8C,c) and RMP_CF (Fig. 8C,d) increased. Further increase of g_gap,CM-CF or RMP_CF caused RMP_CM to stabilize at close to RMP_CF after few oscillations (Fig. 8C,e). Figure 8D shows frequency maps of automaticity when the CM was connected to one to four CFs, which shows that increasing the number of coupled CFs causes automaticity to occur at lower RMP_CF and g_gap,CM-CF. This agrees with the observation that more frequent automaticity occurred in CM:CF_Gαq microtissues when the CM:CF ratio increased from 1:1 to 1:2 (Fig. 5B). Thus, our computational results suggest that RMP_CF, g_gap,CM-CF, and the number of electrically connected CFs are important parameters for the genesis of automaticity in the CM:CF experimental model, but RMP_CF is the most sensitive one.

Fig. 8.

Computer modeling study predicts that elevated resting membrane potential of cardiac fibroblasts (RMP_CF) and increased cardiac myocyte (CM)-cardiac fibroblast (CF) coupling facilitate automaticity. A: schematics of CM-CF coupling in the computational modeling study. The number of CFs connected to a single CM was increased from n = 1 to n = 4. B: representative membrane potential (V_m) traces from CM (black), CF (red), and reference trace of unconnected CF RMP (dotted blue). V_m of CF also oscillated in parallel with CM through electrotonic current flow between CM and CF through gap junctions. C: map of automaticity frequencies in the parameter space of gap coupling conductance between CM-CF (g_gap,CM-CF; x-axis) and RMP_CF (y-axis) when n = 2 CFs were connected (left). RMP_CF was varied from −50 to 0 mV, and g_gap,CM-CF was varied from 0 to 10 nS. Right, representative V_m traces of CM from the locations marked on the frequency map. The characteristics of V_m traces can be categorized into five groups: without automaticity at low RMP_CF and/or g_gapCM-CF (a), automaticity associated with a slow rise of V_m during the resting period (b), acceleration of automaticity when g_gap,CM-CF increases (c), acceleration of automaticity caused by elevation of RMP_CF (d), and V_m reaches close to RMP_CF with dampening V_m oscillations (e). D: maps of automaticity frequencies in the same parameter space of g_gap,CM-CF (x-axis) and RMP_CF (y-axis) with one to four CFs coupled to CM. The map for two CFs is the same as in C. Increase of connected CFs promoted automaticity at lower RMP_CF and g_gap,CM-CF.

Since cell-cell contact is increased in a 3-D environment through increase of contact surfaces in 3-D, it is possible that automaticity may occur even at lower RMP_CF and g_gap,CM-CF. We further investigated the 3-D effect on automaticity using a 3-D cube (Fig. 9A). In this configuration, cells were considered as a cube of which each surface connects to a neighboring cell (RMP_CF = −20 mV, g_gap,CM-CF = 5 nS). Figure 9B shows an example of activation map of automaticity (see Supplemental Movie S2A in the Supplemental Material; Supplemental Material for this article is available at the American Journal of Physiology-Heart and Circulatory Physiology website). The 3-D cube model demonstrated automaticity originating from the location where a higher number of CFs were present (Fig. 9C). When RMP_CF was set at −30 mV and the CM:CF ratio increased to 1:2, automaticity increased (Fig. 9D), which replicates the experimental results we obtained for CM:CF_Gαq (Fig. 5B). When RMP_CF was increased to −20 mV based on our recordings from CF_GαqQL microtissues (Fig. 7), automaticity also increased (Fig. 9E, top trace). Reduction of g_gap,CM-CF from 5 to 4 nS (Fig. 9E, bottom trace) resulted in a slower oscillation, replicating the lowered frequency (Hz) of automaticity in our experiments of CM:CF_GαqQL microtissues after partial gap junction inhibition with CBX (Fig. 5, D and E). These simulation results underscore the importance of the 3-D environment on arrhythmogenic effects of activated CFs by increasing CM:CF connections.

Fig. 9.

Computational modeling study of the three-dimensional (3-D) effect of cardiomyocyte (CM)-cardiac fibroblast (CF) interactions on automaticity. A: schematics of 3-D cube modeling. CMs and CFs were randomly distributed to have 1:1 or 1:2 ratios in a cube of 25 × 2 5 × 25 cells. B: representative activation map of automaticity from the 3-D cube. The activation maps of each layer labeled as L1−L25 in A are shown in 5 × 5 grids, starting the top layer in the top left corner. In this set of CM:CF distribution, automaticity was originated from two sites (open star). C: the number of CFs connected to a single CM in 3-D cube. The parameter set was chosen based on the single CM-CF simulation shown in Fig. 8. The resting membrane potential of a CF (RMP_CF) was set to −20 mV and gap junction coupling conductance between CM-CF (g_gap,CM-CF) was set to 5 nS. With these parameter settings, automaticity was not observed when n = 1 CF was connected (Fig. 8D). A total of 12 randomly generated CM:CF = 1:1 distributions were tested. The black histogram bars indicate the number of CFs connected to CM from all the CMs in the 3-D cube, which shows that the majority of CMs have connections with three CFs on average. The red histogram bars represent the number of CFs connected only to CMs that initiate automaticity. The initiation sites of automaticity showed a larger number of CFs connected to CMs. D: effect of increased CFs on automaticity. RMP_CF was set to −30 mV and showed no automaticity. When the CM:CF ratio was increased from 1:1 to 1:2, automaticity appeared, suggesting that a higher density of CF promotes automaticity as seen experimentally for CM:CF_Gαq (see Fig. 5B). E: effect of g_gap,CM-CF on automaticity. A reduction of g_gap,CM-CF from 5 to 4 nS decreased the frequency of automaticity, which replicates the effect of carbenoxolone we observed experimentally in CM:CF_GαqQL (see Fig. 5E). V_m, membrane potential.

DISCUSSION

The goal of this study was to investigate the influence of activated CFs on CM electrophysiology in a 3-D microtissue environment. CF activation via G_qPCR stimulation commonly occurs in the diseased heart, but the increased abundance of G_qPCR agonists (such as ANG II and ET-1) has concomitant effects on CMs. To dissect the effect of enhanced G_q signaling in CFs on CM function in a 3-D microtissue environment, we adenovirally overexpressed the α-subunit of G_q protein (wild-type or a constitutively active Q209L mutant) in CFs before microtissue self-assembly with CMs. We showed that CF-restricted overexpression of Gα_qQL leads to the expected increase in G_q signaling in microtissues that are viable, slightly enlarged, and show increased expression of profibrotic and prohypertrophic marker genes. Optical mapping demonstrated prolongations of both the durations and rise times of APs and Ca²⁺ transients during pacing at 1 Hz and increased automaticity in unstimulated microtissues that was mitigated by partial gap junction inhibition. Here, we report an increase in Cx43 and Cx45 expression in Gα_qQL-activated CFs when seeded alone or together with CMs in 3-D cultures and significant RMP elevation in Gα_qQL-activated CFs. Computer modeling revealed that RMP elevation in CM:CF microtissues containing Gα_qQL-activated CFs is a key factor, promoting depolarization-induced automaticity by increasing the current flow from CFs. Computer modeling also revealed that gap junctional conductance between CMs and CFs and the number of electrically connected CFs are important factors for the genesis of automaticity, and these influences underscore the importance of the 3-D environment on arrhythmogenic effects by increasing the connections between CMs and of CFs.

Scaffold-free 3-D microtissues as an experimental model to investigate CM regulation by G_q-activated CFs.

The 3-D tissue-engineered approach used in this study provides several advantages. First, CMs and CFs are highly interspersed in this scaffold-free cardiac microtissue model akin to the native myocardium, as previously reported (17), and the cells can interact with each other through cell-cell, cell-ECM, and cell-paracrine interactions. We used a self-assembly approach rather than a scaffold-based approach because it facilitates this organotypic interspersion (17), with maximized cell-to-cell communication and ECM self-produced by the cells (for a review, see Ref. 1). Second, the CM:CF ratios can be experimentally controlled to recapitulate native ratios. The relative proportion of CMs and CFs varies considerably, depending on the species, developmental and disease states, and methodologies used to identify and count cell numbers that have evolved over time (for reviews, see Refs. 45 and 95). On the basis of an initial report (7) estimating that CFs constitute ~60% of the total cell number in ventricles from adult healthy rats, we primarily focused on 1:1 ratios of CMs and CFs in this study and also used 2:1 ratios when examining automaticity. Third, a large number of individual microtissues can be generated in versatile formats, and they are suitable for molecular and functional investigations. Fourth, cells can be cultured in the absence of unnaturally stiff or adherent substrate. Substrate mechanics are known to influence the morphology and function of cardiac and other cells (18, 82). This is especially true of CFs, which are known to assume an activated MFb phenotype when plated on unnaturally stiff cell culture substrates (for a review, see Ref. 85). Young’s modulus of fibroblast spheroids was shown to be 1.8–3.3 kPa for NIH3T3 cells (37), which much more closely resembles that of native heart tissue [4–500 kPa (for a review, see Ref. 82)] than that of conventional polystyrene, glass, or polydimethylsiloxane [MPa-GPa (56)]. Mechanics along with cell adhesions and diffusible factors constitute the differences between 2-D and 3-D cell culture and native tissue environments (6). In this study, we show that compared with conventional 2-D culture, CFs cultured in self-assembled 3-D microtissues have markedly reduced mRNA levels of αSMA [a contractile intermediate filament-associated protein commonly used as a marker to identify activated MFbs (30)], CTGF [a matricellular protein and well-characterized mediator of fibrotic response (49)], and FN1 (a glycoprotein of the ECM synthetized by fibroblasts), whereas the expression of vimentin was comparable (Fig. 3A). These data indicate that the CFs present in microtissues have lower baseline activation than those on stiff 2-D substrates, which have been mostly used to investigate the effect of CFs on CM electrical properties to date. Marked differences in gene and protein expression between 2-D monolayer and 3-D spheroid cultures have been reported for tumor and nontumor cells (22, 77), but to our knowledge this has not yet been systemically examined for fibroblasts of cardiac or other origin.

Since CFs express several different G_qPCRs including ANG II type 1 (88) and ET-1 (28, 40) receptors, we overexpressed Gα_q protein as the common mediator of their effects to mimic the resulting increased in G_q-mediated signal transduction. Consistent with previous reports in cell lines [COS-1 cells (69)], neonatal rat CMs (2), and transgenic mice with CM-restricted Gα_q overexpression (14, 59), Gα_qQL overexpression in CFs induced a more pronounced increase in PLC-β activity compared with wild-type Gα_q overexpression (Fig. 1C), despite much lower amounts of protein expression at comparable infection conditions (Fig. 1B), because the Q209L point mutation renders the protein constitutively active. Importantly, we provide evidence for the efficacy of adenovirus-mediated gene transfer in CFs (Fig. 1A) and CF-restricted adenoviral infection in our mixed CM:CF microtissues (Fig. 1D), consistent with previous reports (93, 17).

As expected, overexpression of Gα_q or the more active mutant Gα_qQL increased the activity of the direct canonical downstream target PLC-β in microtissues composed of adenovirus-infected CFs only (CF_Ad; Fig. 2A). In the presence of CMs (CM:CF_Ad), PLC-β was markedly increased compared with controls as well, but the effect was less pronounced when CFs were infected with Ad-Gα_qQL than Ad-Gα_q, reflecting a differential contribution from the CMs. Enhanced G_q signaling did not alter microtissue viability (Fig. 2B), gross morphology (Fig. 2C), or Sirius Red-based collagen staining (Fig. 2D), but we observed a modest increase in CM:CF microtissue area on CF G_q activation that was significant for Gα_qQL (Fig. 2E). This could be due to changes in cell sizes and/or ECM deposition but needs to be interpreted with caution, because both packing density and the size of CFs or CMs are difficult to assess in self-assembled interspersed microtissues.

We substantiated that enhanced G_q signaling in CFs leads to CF activation by examining the expression of well-known MFb and profibrotic marker genes in both CF_Ad (Fig. 3B) or CM:CF_Ad (Fig. 3, C and D) microtissues. In contrast to vimentin, αSMA, CTGF, and FN1 were markedly increased when CFs were infected with Ad-Gα_q or Ad-Gα_qQL. However, the regulation of their expression in cardiac microtissues appears to be complex. First, Gα_q-induced upregulation of CTGF mRNA was more pronounced for Gα_qQL versus Gα_q in the presence of CMs (CM:CF_Ad; Fig. 3C) but comparable in their absence (CF_Ad, Fig. 3B), suggesting CTGF expression in CMs in response to CF-restricted Gα_q activation. CTGF is an early profibrotic marker whose upregulation precedes collagen deposition (75), and its expression in response to TGF-β and other stimuli has been reported not only in CFs but also CMs (11, 43). Second, a comparable increase in FN protein levels in CM:CF_Ad microtissues containing Gα_q- or Gα_qQL-activated CFs (Fig. 3D), despite a graded increase in FN1 mRNA expression (Fig. 3C), points toward posttranscriptional regulation. Finally, similarly, αSMA protein expression was upregulated in CM:CF microtissues containing Gα_q- and Gα_qQL-activated CFs (Fig. 3D), but αSMA mRNA was only increased when the CFs were activated by Gα_q (but not Gα_qQL), in both the presence (CM:CF microtissues, Fig. 3C) or absence of CMs (CF microtissues, Fig. 3B). Although αSMA-positive CFs are generally referred to as MFbs (57), a recent single cell expression study (41) showed a lower correlation of αSMA expression to the activated phenotype than previously assumed.

Taken together, CF-restricted enhancement of G_q signaling before their assembly in microtissues alone or in combination with CMs resulted in viable microtissues with increased early profibrotic marker gene expression after 3 days. The additional increase in ANF mRNA expression (Fig. 3E) is consistent with well-established paracrine effects of activated CFs on CM hypertrophy (e.g., Refs. 27, 51, and 67).

G_q-activated CFs prolong CM APs and Ca²⁺ transients.

We previously showed that the APD in our self-assembled cardiac microtissues was overall comparable to other models and that the inclusion of CFs increases APD compared with microtissues composed of only CMs (17). In this study, we show that Gα_qQL-activated CFs further prolong APD compared with control CFs (Fig. 4B). This finding is consistent with APD prolongation reported in neonatal rat ventricular CM monolayers cocultured in a 1:1 ratio with CFs that had been precultured to assume a MFb phenotype (5). However, a reduction in APD can also occur depending on RMP_CF, CF density, and degree of coupling to CMs. For example, this was shown for neonatal rat CM monolayer cocultures with activated CFs obtained from infarcted compared with healthy hearts (87). The effect of CF coupling on APD can be bimodal, potentially due to the impact on activation of L-type Ca²⁺ current as well as repolarization. Computer modeling (90) suggested that increase of CFs coupled with CMs maintains the plateau V_m negative, which, in turn, increases driving force of L-type Ca²⁺ channels and increases APD slightly. However, increase of coupling with CF can suppress the AP plateau and accelerate AP repolarization, shortening APD (53). Computer modeling also predicts that APD is prolonged when RMP_CF is less negative, provided that a sufficient number of CFs are present and the coupling is strong enough (32). This seems to be the case in CM:CF_GαqQL microtissues, as RMP_CF was indeed less negative (Fig. 7), and increased coupling was suggested by increased Cx43 expression (Fig. 6) and the effectiveness of partial gap junction inhibition (Fig. 5). Paracrine effects may contribute to the APD prolongation observed in CM:CF_GαqQL microtissues, as previously shown for CMs in 2-D culture when they were exposed to conditioned medium from CFs in the absence of direct cell-cell contact (42, 67).

AP rise time also increased in microtissues with Gα_q- or Gα_qQL-activated CFs (Fig. 4B). The slow rise of AP (slow AP upstroke velocity) associated with slow conduction has been observed in 2-D culture (5, 31, 60) as well as in computer modeling studies (8, 32, 33, 76) and was attributed to the dissipation of depolarizing currents to the neighboring CFs via a source-sink relationship. The slow AP upstroke could reduce conduction velocity and promote conduction blocks as seen in 2-D culture. Since the size of CM:CF microtissues are relatively small within 150 μm diameter, it was difficult to measure the impact of CF activation on conduction.

The activation of CFs promoted significant changes in CM Ca²⁺ transients (Fig. 4, C and D), as both rise time and duration were prolonged along with APD prolongation. Although we did not quantify the shape of Ca²⁺ transients, the increase of rise time was associated with an altered peak: CM:CF_GαqQL microtissues tended to show a two-step Ca²⁺ rise with a rapid rise phase during the initial rise of Ca²⁺ transients and a slow secondary rise during the peak (Fig. 4C), which contributes to the increase of Ca²⁺ transient rise time. The mechanisms underlying the changes in [Ca²⁺]_i kinetics are not clear. Although slowing of the Ca²⁺ transient upstroke can be explained by slow AP upstroke through electrotonic CM-CF coupling, Ca²⁺ transient prolongation is difficult to explain solely by CM-CF coupling since [Ca²⁺]_i recovery is much slower than AP repolarization (Fig. 4, B and D). Further studies are needed to examine potential paracrine effects from CFs on Ca²⁺ transients in CMs.

G_q-activated CFs promote automaticity in CMs.

Abnormal electrical activity including spontaneous ectopic activity and automaticity has been frequently observed and associated with induction of atrial and ventricular arrhythmias in various cardiac diseases, such as ischemia-reperfusion, myocardial infarction, and heart failure (4, 20, 35, 63). The cause of ectopic activity has been conventionally linked to the excessive inward current during the diastole, which causes slow depolarization leading to the triggering of APs. These extra APs increase risks for ventricular tachyarrhythmias, if they propagate and form reentry. Increasing evidence supports potential roles of activated CFs in generating abnormal automaticity. For example, using micropatterned strands of neonatal rat ventricular CMs coated with CFs that were activated by preculture on a stiff substrate and referred to as MFbs based on αSMA expression, Miragoli et al. showed MFb-induced and density-dependent synchronized spontaneous activity based on depolarization-induced abnormal automaticity (ectopic activity) (61). Similarly, another group showed a marked increase in ectopic activity in conventional 2-D monolayers of neonatal rat ventricular CMs when they were cocultured with similarly preactivated MFbs (5). Involvement of gap junctional coupling was suggested in this study by a reduction in ectopic activity when Cx43 was downregulated in MFbs via RNA inhibition.

In the present study, we examined the role of activated CFs on automaticity in a 3-D environment, both experimentally and via computer simulation, and used a more physiological mode of CF activation. We showed that CF activation by enhanced Gα_q signaling markedly induces automaticity in CM:CF microtissues and that this effect was more pronounced when the number of CFs was increased and when Gα_q was constitutively activated (Fig. 5, A and B). Partial gap junction blockade had an inhibitory effect on the occurrence and frequency of automaticity (Fig. 5, C–E), which together with the increased Cx expression (Fig. 6; discussed below) points toward the possibility of enhanced gap junctional coupling between Gα_q-activated CFs and CMs. Our computer simulations indeed supported the notion that gap junctional coupling (Figs. 8C and 9E) and the number of electrically connected CFs (Fig. 8D and Fig. 9, C and D) are important factors for the generation of automaticity in the CM:CF microtissue model. Other mechanisms of hetercellular interactions, such as paracrine effects (67) and nanotubes (29, 71), may contribute as well.

G_q-activated CFs and the 3-D environment increase Cx expression.

Cell-cell connectivity can be affected by the presence, expression pattern, and functionality of the Cx family of proteins, the building units of functional gap junctions (36). For CFs that were cultured in microtissue or 2-D culture for the same duration, we showed pronounced and isoform-specific differences in mRNA expression for Cx43 [>5-fold upregulated, consistent with Ref. 9, Cx45 (unchanged), and Cx40 (downregulated by 92%); Fig. 6A]. This observation is another example of the marked expression differences between 2-D monolayer and 3-D microtissue cultures that were discussed above. Further investigation of Cx expression in CF microtissues showed up to 5-fold upregulation of Cx43 mRNA in activated CFs and a 2.5-fold increase in Cx45 mRNA in CFGα_qQL (Fig. 6B). Cx upregulation also occurs in CFs activated in vivo postmyocardial infarct, as shown in situ for both Cx43 and Cx45 in a sheep coronary occlusion model (10) and in CFs isolated from the infarct scar of mice (94). In microtissues containing both CMs and CFs activated by Gα_q and/or Gα_qQL, Cx43 and Cx45 upregulation was also observed, with some differences between mRNA and protein expression levels (Fig. 6, C and D). Importantly, Cx expression per se is not an indication for functional gap junction channels, because the oligomerization of Cxs into hemichannels (connexons), the delivery of connexons to the plasma membrane, and the formation of gap junction and their stability and function are complex processes that are subject to various posttranslational modifications and other regulatory processes (83).

In 2-D culture, it has been well established that Cx proteins exist at the sites of heterocellular contact between CMs and CFs (12, 23, 24, 60, 74), and some studies have shown functional coupling (73, 87). However, whether or not gap junctional coupling occurs in the 3-D tissue context and how this may be altered under pathological stimulation and in disease has long been a subject of investigation and debate (for reviews, see Refs. 47, 55, and 64). Strong evidence in support of functional CM-CF coupling in vivo was only recently reported when it was demonstrated that nonexcitable cells in the heart can closely follow CM APs. To that end, expression of a voltage-sensitive fluorescent protein in non-CMs was used to record AP-like signals in cryoinjured scar tissue (71), and current injected into the healthy myocardium could be recorded in adjacent scar tissue, but no signals were recorded when Cx43 was knocked down in non-CMs (54).

Enhanced coupling between CMs and CFs can influence AP characteristics of CMs in several ways depending on RMP_CF: 1) by providing inward currents during the diastolic period to facilitate depolarization of the transmembrane potential of CMs, 2) by providing outward currents during the upstroke and plateau phases of the AP, and 3) by delaying AP repolarization during the late phase of repolarization (for a review, see Ref. 92). Given the challenges in relating Cx expression to connexon formation and gap junction functionality, we used computer modeling to assess the potential impact of functional gap junctional coupling between CM and G_q-activated CFs in our microtissues and were able to replicate key features of the propensity for automaticity in CM:CF_GαqQL microtissues (see above). Our simulations suggests that with coupling above a threshold, automaticity is mostly dependent on the electrophysiological properties of CFs, such as overall ionic conductance, capacitance, and RMP_CF (see below). They also emphasize the importance of a 3-D environment where cell-cell contact increases in all directions. In 3-D, the probability of a number of CFs connecting to CMs is higher because of potential contact over their entire cell surface areas compared with limited side-to-side contacts in 2-D. Despite similar trends in RMP_CF between 2-D and 3-D cultures, the efficacy of triggering automaticity is increased in 3-D simulation, underscoring the influence of CM-CF coupling in a 3-D environment.

Although computer modeling in this study mainly focused on CM-CF coupling, CM-CM or CF-CF coupling may also modulate automaticity. A previous computer modeling study (91) showed that enhancing CM-CM coupling suppressed triggered activity through the source-sink effect, i.e., an enhanced CM-CM coupling strength potentiates the sink effect, making spontaneous depolarization more difficult. Therefore, the increase of CM-CM coupling in 3-D microtissues should have reduced automaticity, but our experimental results showed increased automaticity in Gα_qQL-activated microtissues. However, it is also possible that CM-CM coupling can increase automaticity via synchronization of automaticity of CMs, as seen in sinoatrial pacemaker. Further studies are needed to delineate homo/heterocellular coupling of CMs and CFs in modulating automaticity.

Elevated RMP in G_q-activated CFs.

RMP_CF is an important factor because the difference between V_m,CM and V_m,CF determines the driving force of electrotonic current flow between CMs and CFs, and the importance of RMP_CF elevation on triggered activity has been also shown in computer simulations of CM-CF interactions (96). Since fluorescence recordings by optical mapping do not provide the absolute V_m, we used high-resistance microelectrodes to determine the RMP_CF in 2-D culture or 3-D microtissues. Consistent with previous reports in neonatal (26, 60, 74) and adult ventricular CFs (13), the RMP of control CFs (Fig. 7) was much less negative compared with CMs. Importantly, RMP in CFs infected with Ad-Gα_qQL was markedly elevated (by 13–14 mV) in both 2-D and 3-D culture conditions (Fig. 7). RMP recorded in CF_GαqQL microtissues (−20.0 ± 7.2 mV) was comparable to that reported for neonatal rat ventricular CFs that were activated by 8- to 9-day culture on rigid 2-D substrates [−20.4 + 0.4 mV (60) and −17 + 6 mV (5)]. Therefore, our present study showed that a biological trigger common to the diseased heart (i.e., enhanced G_q signaling) is sufficient to elevate RMP_CF and can do so in a soft microtissue context.

CFs express multiple ion channels (50), with heterogeneity reported depending on the cardiac region (chamber), activation (disease) state, and recording conditions (acutely isolated or cultured) (for reviews, see Refs. 55 and 86). The ionic mechanisms determining RMP_CF are still poorly understood. Indepth electrophysiological studies will be needed to delineate the determinants of the RMP_CF under baseline conditions and the mechanism underlying RMP depolarization in activated CFs, which may depend on the nature of the stimulus. Mechanical strain also influences RMP_CF (26, 38, 48), which needs to be considered when interpreting RMP_CF values obtained in freshly isolated, cultured CFs and those in microtissue context.

Potential relevance and limitations.

Establishing an active role for CFs in regulating CM electrical function in the intact heart has been a formidable challenge because of the intricate interspersion of the two cell types [virtually every CM borders one or more CFs (46)] and exposure to the same environment, so that the contribution of individual cell types to integrated tissue function cannot be easily determined. Previous investigations of the effect of activated CFs on CM electrical activity in vitro were mainly performed on rigid 2-D substrates that can mechanically activate CFs. The present study extended this question to a 3-D microtissue environment that is associated with a lower baseline activation state of the CFs and enables 3-D cell-cell, paracrine, and cell-ECM interactions as present in the cardiac tissue. Using this model, we present experimental and modeling evidence suggesting that G_q-activated CFs can contribute to arrhythmogenesis by direct electrotonic modulation of CMs and highlighting the importance of the 3-D environment on arrythmogenic effects.

A limitation of our experimental model is that Gα_q activation in CFs is initiated shortly after self-assembly of the bioengineered microtissues and not within already established tissue as is the case in the intact diseased heart. The present study also used a simplified CM AP model and passive CF computational model. Although I_K1 conductance was adjusted to the previous voltage-clamp data in the same 3-D microtissue construct, this model does not incorporate potential electrical remodeling caused by paracrine effects from G_q-activated activated CFs. The CF model in this study is also limited without incorporation of ionic currents such as Clcn-3 and Kir2.1. Further studies are needed to understand how potential CM remodeling and homo-/heterocellular couplings contribute to depolarization-induced abnormal automaticity. Transgenic mice with CF-restricted Gα_q expression will be required to extrapolate our findings from the in vitro microtissue environment to the in vivo setting, along with characterizations of activated CFs in diseased hearts with regard to their size, their relative density and localization, their changes in G_q signaling, their cellular electrophysiology and ability to form homocellular and heterocellular gap junctions, and the resulting effect on the integrated function of the ventricular myocardium. There is growing interest to modulate CF activation to mitigate fibrosis (25). Our results demonstrate that CF activation in response to enhanced G_q signaling, a common pathophysiological trigger in the diseased heart, can greatly influence AP and Ca²⁺ handling of CMs, leading to arrhythmogenic ectopic activity, suggesting potential antiarrhythmic benefits of targeting CFs in cardiac disease.

GRANTS

This work was funded by National Heart, Lung, and Blood Institute (NHLBI) Postdoctoral Fellowships T32-HL-094300 and 1-F32-HL-126311 and American Heart Association Grant 15POST22740017 (to C.M. Kofron) and by NHLBI Grants R21-HL-1139181 and R01-HL-114784 (to U. Mende).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

C.M.K., T.Y.K., B.-R.C., and U.M. conceived and designed research; C.M.K., T.Y.K., M.E.K., A.X., F.F., and E.P. performed experiments; C.M.K., T.Y.K., M.E.K., A.X., B.-R.C., and U.M. analyzed data; C.M.K., T.Y.K., M.E.K., A.X., Z.Q., B.-R.C., and U.M. interpreted results of experiments; C.M.K., T.Y.K., M.E.K., B.-R.C., and U.M. prepared figures; C.M.K., T.Y.K., B.-R.C., and U.M. drafted manuscript; C.M.K., T.Y.K., M.E.K., A.X., Z.Q., B.-R.C., and U.M. edited and revised manuscript; C.M.K., T.Y.K., M.E.K., A.X., Z.Q., B.-R.C., and U.M. approved final version of manuscript.