H9c2 myoblasts (ATCC (H9c2(2-1); ATCC, CRL-1446), derived from embryonic rat cardiac tissue (Brandt et al., 1976; Kimes and Brandt, 1976) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; ATCC, 30-2002) supplemented with 10% fetal bovine serum (FBS; Fisher, FB12999102) and 1% penicillin/streptomycin (P/S; Life Technologies, 15140- 122) in 5% CO2 at 37°C in a humidified incubator. Primary human dermal fibroblasts (ATCC, PCS-201-010) were cultured under the same specifications.

The CTCs utilized in this experiment were fabricated similarly to previous iterations (Budhathoki et al., 2021) using polydimethylsiloxane (PDMS; QSil 216, Quantum Silicones, Richmond, VA, United States) and soft lithography. Square frames (5 cm × 5 cm × 1 cm in height) with a central circular cavity (2.8 cm in diameter) were generated using replica molding and PDMS mixed at a 10:1 base-to-cross-linker ratio followed by heat-curing. Flexible membranes were made by spin-coating PDMS mixed at a 15:1 base-to-crosslinker ratio onto silicon wafers and then heat-curing. Next, six PDMS posts (1 cm in height × 0.1 cm in diameter) were attached to each membrane using 3D-printed stencils for precise positioning. The posts were arranged into three pairs, measuring 8 mm from center-to-center with a 6 mm space between adjacent pairs.

Following cleaning, the frames and post-laden membranes were bonded with oxygen plasma (Harrick Plasma Systems, Ithaca, NY, United States) using 700 mTorr pressure, 30 W, and 30 s exposure time. The freshly bonded devices were heated further to strengthen the bond and eliminate any remaining air pockets between the frame and membrane. Lastly, the devices were carefully peeled from the silicon wafers and excess membranes were trimmed.

3D-printed dumbbell-shaped molds were placed in each device (one per pair of posts) and then the device chambers were filled with 3.5% molten agarose (Figure 1A). After agarose gelling at room temperature, the molds were removed, leaving behind three empty troughs (Figure 1B). Following a PBS wash, the devices were filled with a 1% w/v solution of bovine serum albumin (BSA; Sigma-Aldrich, A7906) in PBS and stored overnight at 4°C to generate a non-adherent coating. Prior to seeding, the devices were warmed to 37°C and the BSA solution was removed.

With the devices now prepared, the H9c2s were dissociated with 0.25% Trypsin/EDTA (Gibco, 25200-056) and resuspended with the human dermal fibroblasts, which were dissociated with 0.05% Trypsin/EDTA (Gibco, 25300-054). Each 200 μL cardiac tissue construct contained 2 × 106 H9c2s and 0.667 × 106 fibroblasts and was generated as follows: first, thrombin (Millipore, 605195) was added to 40 mM CaCl2, and 10 μL of the resulting 0.048 units/μL thrombin solution was pipetted into a microcentrifuge tube (one tube per tissue construct). After the cells had been combined in the appropriate ratio and pelleted, they were resuspended in a 10% v/v solution of growth factor reduced Matrigel (Corning, 354230) in DMEM supplemented with 10% FBS and 1% P/S. 180 μL of this cell/hydrogel suspension was then added to each thrombin-containing tube. Lastly, proceeding one tube at a time to prevent premature gelling, 10 μL of 40 mg/mL fibrinogen (Millipore, 341576) was added to each tube, the suspension was gently pipetted to homogenize the mixture, and the 200 μL suspension was pipetted into one of the empty troughs (Figure 1C). Each tissue construct had a final concentration of 2 mg/mL fibrin, 1.2 units thrombin/mg fibrin, and 2 mM CaCl2. After seeding, the CTCs were placed into the 37°C humidified incubator at 5% CO2 for 60-75 min, allowing the tissue constructs to gel (Figure 1D). Following this, the agarose was removed, the device chamber was washed with PBS to remove any agarose debris, and DMEM containing 10% FBS, 1% P/S, and 2 mg/mL aminocaproic acid (ACA; Acros, 103301000) was added to each (Figure 1E). The tissue constructs were maintained in static culture for 72 h with daily media changes prior to experimentation to allow for compaction and ECM deposition (Figure 1F).

A modified version of the Biomimetic Cardiac Tissue Model (BCTM) designed by our laboratory (Rogers et al., 2019; Budhathoki et al., 2021) was used in these experiments to mimic left intraventricular pressure and volume changes. First, the seeded CTC was sandwiched between an upper polycarbonate piece (5 cm × 5 cm × 1 cm in height) with media inlet and outlet ports, and a bottom polycarbonate piece of equal size that had a 3 cm × 3 cm × 0.5 cm deep centrally-located cavity with inlet and outlet ports for pressurized air. A 2 cm × 0.9 cm × 0.4 cm hemicylindrical PDMS fragment was also placed in the cavity of the bottom piece. Next, flexible tubing with one-way valves to ensure anterograde flow was used to connect the sandwiched CTC to a media reservoir, and a peristaltic pump was employed to prime the device/reservoir loop. Following priming, the CTC loops were moved to a 37°C humidified incubator filled with 5% CO2, and the media reservoirs were clamped to ring stands and positioned at appropriate heights. The end-diastolic pressure in this system (Figure 2), is dependent upon the hydrostatic pressure generated by the height of the media column as well as the compliance of the PDMS membrane. In order to facilitate greater stretch in the volume overload (VO) devices in comparison to the normal load (NL) devices, the VO device membranes were fabricated with 200 μL of PDMS, whereas the NL devices were fabricated with 500 μL. End-diastolic pressures of 10 mmHg were established by raising the media column to a height of 32 cm for the NL devices and 55 cm for the VO devices. Next, the bottom polycarbonate piece of each CTC was connected to a pneumatic pump (LB Engineering GbR, Berlin, Germany) set at 60 bpm and pulse duration of 0.4 s, which generated peak systolic pressures of ∼100 mmHg. This systolic pulse below the membrane returns it to a position closer to the non-deformed baseline, propelling the media forward, and the ensuing 0.6 s filling phase pushes the membrane downward, stretching it and the tissue constructs around the hemi-cylindrical obstacle below. While these LV pressures are consistent with those seen in mice (Joho et al., 2007), a sub-physiological heart rate of 60 bpm was chosen due to technical limitations of the employed pneumatic pumps and to allow for full stretch and relaxation of the tissue constructs. Lastly, a tunable resistance element on the media outflow tubing allowed for control over the degree to which the PDMS membrane returned to its non-deformed baseline at the end of systole. For the NL devices, these elements were tuned such that the membrane was flat at the end of systole, while those of the VO CTCs were tuned to maintain residual strain (in comparison to the device resting length) at the end of systole. Based on imaging measurements, the NL tissue constructs experienced ∼9% strain, calculated as [(end-diastolic length–end-systolic length)/end-diastolic length whereas the VO constructs experienced ∼18%. The tissue constructs were subjected to these hemodynamic parameters for 96 h before being harvested for molecular studies or histology.

All animal procedures and experiments were approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee and conformed with the principles of the National Institutes of Health Guide for the Care and Use of Laboratory Animals. In our study we used eight C57BL/6 male mice (Taconic Biosciences, Hudson, NY). The AVFs were created by connection of the jugular vein to the side of carotid artery, as previously published (Pike et al., 2017; Ingle et al., 2021; Somarathna et al., 2021) (Figure 3). Mice were 10–11 weeks old at the time of AVF creation. Age-matched sham-operated mice served as the control group. For this study, we measured end timepoint cardiac parameters. All mice had baseline 2D echocardiography performed at 10 weeks of age, and at the time of sacrifice at 28 days post-surgery. Hearts were harvested at 28 days after AVF creation in experimental group and corresponding control group. The LVs were dissected and preserved for histology and RNA analysis.

Cardiac function in the C57BL/6 mice were evaluated using a VEVO 2100 micro-ultrasound system with a 24–30 MHz bi-frequency transducer (Visual Sonics, Toronto, Canada) as previously described in detail and published (Ingle et al., 2021). To characterize the LV systolic function we calculated EF and FS, with details as previously published (Ingle et al., 2021). We measured CO to measure the rate of blood flowing from the heart with details as previously published (Ingle et al., 2021).

Several morphometry techniques were employed to stain collagen in the LV, including Picrosirius red staining (PSR) under bright light and polarized light, and combination of Verhoeff’s elastic and Masson’s trichrome staining. The mouse heart tissues and cardiac tissue constructs were fixed in 10% formalin and processed for paraffin embedding. Sections of 5 µm (Leica RM 2135) were applied to charged slides, deparaffinized, and hydrated with distilled water. Slides were stained as previously described (O’Connor and Valle, 1982). The elastic fibers are stained black by Verhoeff’s technique, the muscle and connective tissue stained red and green/blue, respectively, by a modification of Masson’s trichrome, and cell nuclei stain blue-black with iron hematoxylin.

Cardiac tissue constructs and LV tissues from 28D sham and AVF mice were collected to be analyzed for fibrosis. The presence of collagen in the LV sections was evaluated in the interstitial area and perivascular areas. Images were captured using an Olympus IX73 Brightfield microscope. For each sample, 4-10 images were taken, and the collagen stain was analyzed by imaging software (CellSens dimensions (Olympus version 3.1) as mentioned in our previous work (Ingle et al., 2021). In brief, interstitial fibrosis and collagen deposition in the LV sections and cardiac tissue constructs, was quantified by first selecting a threshold range. All images were quantified using this same threshold. The positive staining was then calculated as a percentage of the total image area. For quantifying perivascular fibrosis, a region of interest was traced around the vessel such that the lumen and surrounding tissue were not included in the calculation. The stained area within the region of interest was calculated as a percentage of the region of interest.

Total RNA was isolated using the miRNA isolation kit (Qiagen, cat #74106) according to the protocol of the manufacturers. Samples were treated with DNAse I (Qiagen cat# 79254). RNA was quantitated using Epoch/Take3 spectrophotometer system (BioTek Instruments, Inc.). Complementary DNA was generated with iScript cDNA Synthesis kit (Qiagen). We used StepOnePlus™ Real-Time PCR System to run a Taqman gene expression assay. Taqman expression assay probes for human genes have cross reactivity with corresponding rat genes expressed by H9c2 cells. To prepare the PCR reaction mix for each Taqman gene expression assay and samples on cycling plate: for one sample combine 10 µl of 2x Taqman fast universal PCR master mix, 1 µL Taqman gene expression assay, 5 µL water with 4 µL cDNA mix.

LV was harvested from AVF mice and sham-operated mice at postoperative day 28 and frozen with liquid nitrogen and stored at -80 °C. Isolation of total RNA and RNA sequencing were performed by Discovery Life Sciences (Huntsville, AL). Isolation and preparation of RNA was performed by Discovery Life Sciences standard protocols, and details of quality control assessment and library preparation were previously published (Somarathna et al., 2021). RNA libraries were sequenced on a NovaSeq 6,000 instrument (Illumina, San Diego, CA) at a 150 base pair read length in order to achieve 50 M paired-end reads (100 M total reads).

Reads were quality controlled by FASTQC 0.11.5 (Andrews, 2015). Adaptors were trimmed from reads using cutadapt 1.13.1 (Martin, 2011). Adaptor trimmed reads were aligned to the mm10 reference genome using STAR 2.5.3 (Dobin et al., 2012). Transcript quantification and calculation of differential gene expression were performed using DESeq2 1.18.1 in R 4.1.2 (Love et al., 2014). Ingenuity pathway analysis (IPA) (Krämer et al., 2014) was used to detect canonical pathways associated with differentially expressed genes, the activation or inhibition of upstream regulators and novel pathways and networks of differentially expressed genes.

All data was expressed as means and standard deviation of means. Mann-Whitney U test analysis was used to compare the control and AVF groups for the echocardiographic measurements. Multiple t-test analysis was used to compare the VO and NL conditions for RT-PCR studies. Results for perivascular and interstitial PSR staining were analyzed using unpaired t-test. GraphPad Prism version 9 (La Jolla, CA) was used for statistical analysis and p < 0.05 was considered statistically significant. Statistical tests for differential expression were performed using the built-in method in DESeq2. Genes with a FDR corrected p < 0.05 were considered significant.