Cell-free DNA as a potential biomarker of differentiation and toxicity in cardiac organoids

  1. Brian Silver
  2. Kevin Gerrish
  3. Erik Tokar  Is a corresponding author
  1. Mechanistic Toxicology Branch, DNTP, United States
  2. Molecular Genomics Core, DIR NIEHS, United States

Abstract

Cell-free DNA (cfDNA) present in the bloodstream or other bodily fluids holds potential as a noninvasive diagnostic for early disease detection. However, it remains unclear what cfDNA markers might be produced in response to specific tissue-level events. Organoid systems present a tractable and efficient method for screening cfDNA markers. However, research investigating the release of cfDNA from organoids is limited. Here, we present a scalable method for high-throughput screening of cfDNA from cardiac organoids. We demonstrate that cfDNA is recoverable from cardiac organoids, and that cfDNA release is the highest early in differentiation. Intriguingly, we observed that the fraction of cell-free mitochondrial DNA appeared to decrease as the organoids developed, suggesting a possible signature of cardiac organoid maturation, or other cardiac growth-related tissue-level events. We also observe alterations in the prevalence of specific genomic regions in cardiac organoid-derived cfDNA at different timepoints during growth. In addition, we identify cfDNA markers that were increased upon addition of cardiotoxic drugs, prior to the onset of tissue demise. Together, these results indicate that cardiac organoids may be a useful system towards the identification of candidate predictive cfDNA markers of cardiac tissue development and demise.

Editor's evaluation

This important study presents a comprehensive investigation into cell-free DNA (cfDNA) within the context of 3D cell cultures, offering valuable insights into this emerging research area. Despite being an exploratory study, the findings provide compelling evidence that continued progress holds the potential for the establishment of versatile 3D cell culture cfDNA assays. Such assays could serve as invaluable research and clinical tools, enabling monitoring of organoid growth and development, enhanced characterization of tissue dynamics, and ultimately facilitating the identification of novel biomarkers.

https://doi.org/10.7554/eLife.83532.sa0

Introduction

Biomarkers in blood or bodily fluids are valued for their potential as noninvasive diagnostics for early disease detection (Zukowski et al., 2020; Leal et al., 2020). Biological fluids contain many forms of information. One such parameter is cell-free DNA (cfDNA), which refers to free-floating DNA in the bloodstream or extracellular media (Grabuschnig et al., 2020). Although cfDNA is believed to be generated largely by apoptotic cells (Grabuschnig et al., 2020), this phenomenon is not limited to cells undergoing a death pathway. Active release of nucleic acids encapsulated in extracellular vesicles (Kustanovich et al., 2019) or through mitochondrial channels (Xian et al., 2022) are additional potential routes that may permit the release of cfDNA. Several measures, including concentration, fragment size, sequence, and epigenetic modifications such as differential methylation, histone modification, and nucleosome spacing, can be used to characterize cfDNA and imply information about the cells from which it originated (Zukowski et al., 2020; Lehmann-Werman et al., 2016; Moss et al., 2018). Currently, cfDNA already has some diagnostic uses, including noninvasive prenatal testing for chromosome abnormalities (Willems et al., 2014), fetal gender determination (Jacobsen et al., 2018), and cfDNA-based tests for the detection of mutations and cancer screening (Cisneros-Villanueva et al., 2022; Bronkhorst et al., 2019). Although the clinical utility and accuracy of these tests in the prediction of cancer is uncertain in some cases (Hackshaw et al., 2022), clinical trials are promising and some FDA-approved tests are already available (Cisneros-Villanueva et al., 2022). Still, we have likely yet to unlock the full spectrum of clinical usage for cfDNA in noninvasive diagnostics. It remains unclear which specific events at the cellular level can be detected in cfDNA and what cfDNA profiles signify tissue-level changes during development and differentiation of specific tissues. Broadening our understanding of how cfDNA changes in response to normal tissue development and maintenance is necessary to strengthen recognition of aberrant cfDNA signatures that could reflect deleterious phenotypes such as teratogenesis or toxicity.

Human organoids have many advantages as models of both development and disease, including similarity to human tissues, applications in precision medicine, and high-throughput capabilities (Kim et al., 2020). Organoids are more complex than 2D cell cultures or spheroids, yet permit a more isolated view of specific tissues as compared to in vivo settings or human blood samples, which contain cfDNA from numerous tissue sources (Moss et al., 2018). These properties make human organoids an appealing system to study cfDNA in response to cell differentiation or chemical exposure. Although cfDNA has been successfully isolated from pancreatic organoids (Dantes et al., 2020), whether cfDNA can be recovered from additional organoid types is unclear.

In this study, we used cardiac organoid models to identify properties of cfDNA that were reflective of tissue differentiation and toxicity. Our analyses demonstrate the tractability of cardiac organoid systems to rapidly identify cfDNA markers associated with tissue-level events in a human-relevant setting. Further, we identified specific cfDNA sequences that were elevated prior to major tissue defects initiated by toxicant treatment, suggesting that cardiac organoids have potential value as a tool to explore cfDNA biomarkers of cardiotoxic events.

Results

Cardiac organoids derived from H9 embryonic stem cells exhibit characteristic morphological changes and express markers of differentiation

H9 embryonic stem cells were seeded, coalesced into spheroids, and induced to differentiate towards cardiac lineage (Figure 1A). The organoids began contracting rhythmically at approximately 20–30 beats per minute (bpm) on day 6 and maintained this beating throughout maturation (Figure 1B). We observed that the organoids expressed increased levels of several markers of cardiac differentiation at both the transcript (Troponin T, Nkx2.5) and protein (MEF2C, GATA-6, α-actinin, and Nkx2.5) levels, and decreased transcription of the pluripotency marker Oct 3/4 (Figure 1C–F).

Cardiac organoids derived from H9 embryonic stem cells exhibit characteristic morphological changes and express markers of differentiation.

(A) Brightfield images of cardiac organoids during growth and differentiation. Scale bars represent 100 µm. (B) Quantification of beats per minute across three separate biological replicates for growth day 6 (n = 9 organoids total) and day 9 (n = 7 organoids total). ddPCR of Troponin T (C), Nkx2.5 (D), and Oct3/4 (E) in cardiac organoids at different timepoints during differentiation, normalized to GAPDH. (F) Western blot showing protein expression of MEF2C, GATA-6, α-actinin, and Nkx2.5 relative to GAPDH in cardiac organoids at different timepoints. Graphs show average concentrations + SD. Samples were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant.

Figure 1—source data 1

Cardiac organoids derived from H9 embryonic stem cells exhibit characteristic morphological changes and express markers of differentiation.

https://cdn.elifesciences.org/articles/83532/elife-83532-fig1-data1-v2.zip

Cardiac organoids release cfDNA during their development

Conditioned media was collected from cardiac organoids at several time points during their maturation (Figure 2A). We observed that cfDNA could be extracted consistently during development of both organoid models, and total cfDNA concentration increased as the organoids matured (Figure 2B). To account for changes in cell number during organoid growth, we normalized the concentrations of cfDNA to levels of genomic DNA (gDNA) collected on the same growth days. Normalized cfDNA was the highest early in development (Figure 2C), indicating that cfDNA output may be higher in tissues on growth day 1, which are comprised of less differentiated cells. The recovered cfDNA from mature cardiac organoids on day 9 consisted of a broad distribution of fragments ranging from approximately 200–6000 base pairs (Figure 2D). Fragment distributions were more difficult to resolve at timepoints prior to day 6, likely due to low DNA concentrations and sensitivity limitations (Figure 2—figure supplement 1).

Figure 2 with 1 supplement see all
Cardiac organoids release cfDNA during their development.

(A) Schematic illustrating timepoints of cfDNA collection in cardiac organoids. (B) Shown are cfDNA concentrations taken from n = 3 biological replicates during cardiac organoid differentiation. Data points represent the average of 2–3 technical replicates per biological replicate. (C) Cardiac organoid-derived cfDNA concentrations from panel (B), normalized to gDNA concentrations (multiplied by 100 to facilitate axis readability). (D) Electropherogram showing fragment sizes of cfDNA derived from mature cardiac organoids on day 9. Graphs show average + SD. Samples were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant.

Figure 2—source data 1

Electropherogram showing fragment lengths of cfDNA derived from cardiac organoids at different time points during development.

https://cdn.elifesciences.org/articles/83532/elife-83532-fig2-data1-v2.zip

Cardiac organoids exhibit a time-dependent decrease in cell-free mitochondrial DNA abundance during growth

To further characterize the cfDNA recovered from cardiac organoids, we examined abundance of cell-free mitochondrial DNAcell-free mitochondrial DNA by quantifying the levels of markers located at several regions within the mitochondrial genome (Figure 3A). We observed that cardiac cfDNA levels of the mitochondrial genes ND1, mtCOX2, ND6, and ND4 dropped substantially in a time-dependent manner as the organoids differentiated (Figure 3B), and the markers were approximately equally represented in the cfDNA at most time points (Figure 3C). The tissue-level expression of mitochondrial genes slightly decreased midway through differentiation (Figure 3D), but did not fully account for the time-dependent drop in cell-free mitochondrial DNA. As anticipated, an increase in the mitochondrial protein VDAC/Porin was observed in mature cardiac organoids (Figure 3E and F), which is representative of increased mitochondrial size expected in cardiac tissues (Guo and Pu, 2020).

Cardiac organoids exhibit a time-dependent decrease in cell-free mitochondrial DNA abundance during growth.

(A) Schematic showing the location of markers examined on the mitochondrial genome, adapted from Figure 1 of Uhler and Falkenberg, 2015. (B) Relative abundance of mitochondrial markers ND1, ND4, ND6, and mtCOX2 in cardiac organoid-derived cfDNA, obtained using ddPCR. Shown are the average copies/μl of n = 3 biological replicates, each normalized to the values on day 1, to correct for batch variation. Samples on day 2 were compared to day 1 using a one-sample t-test comparing to a hypothetical value of 1.0. Day 6 samples were compared to day 2 values using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant. (C) Ratio of the values obtained in panel (B) for each marker to mtCOX2. Significant deviations from a ratio of 1.0 indicating unequal marker abundance were determined using a one-sample t-test comparing to hypothetical value of 1.0. Shown are the averages of n = 3 biological replicates, + SD. (D) Expression of mitochondrial DNA markers at the tissue level in cardiac organoids during differentiation obtained using ddPCR, normalized to β-actin. (E) Western blot showing protein expression of the mitochondrial protein VDAC/Porin and loading control vinculin in n = 3 biological replicates of cardiac organoids taken at different timepoints during differentiation. (Vinculin loading control is the same for Figure 4D.) (F) Quantification of VDAC protein expression from the western blot in panel (E), normalized to vinculin. Samples were further normalized to day 1 and compared using a one-sample t-test with hypothetical value of 1.0 to test for significant deviation from day 1. (G) Concentration of cfDNA extracted from media conditioned by cardiac organoids on growth days 2 or 6 after ultracentrifugation (100,000 × g for 1 hr) compared to control (no ultracentrifugation). Abundance of mitochondrial markers ND1 (H) and mtCOX2 (I) in cfDNA derived from ultracentrifuged media conditioned by cardiac organoids during development, obtained using ddPCR. Graphs show average concentrations + SD. Samples were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant.

Figure 3—source data 1

Cardiac organoids exhibit a time-dependent decrease in cell-free mitochondrial DNA abundance during growth.

https://cdn.elifesciences.org/articles/83532/elife-83532-fig3-data1-v2.zip

We hypothesized that mitochondrial fragments within cell debris might be expelled from the organoids as they differentiate, leading to higher cell-free mitochondrial DNA early in the growth process. To investigate this possibility, we examined cell-free mitochondrial DNA abundance in media conditioned by cardiac organoids that had been ultracentrifuged (100,000 × g) for 1 hr, to eliminate small fragments of cellular debris and extracellular vesicles. We observed similar cfDNA concentrations and levels of cell-free mitochondrial DNA as in samples that had not been ultracentrifuged (Figure 3G–I). This indicates that the cell-free mitochondrial DNA recovered from cardiac organoids is not due to cell debris or mitochondrial fragments in the media.

Abundance of specific gDNA sequences is reflective of tissue-level changes in marker expression

Differentiating cardiac organoids show a switch-like increase in protein expression of the transcription factor Nkx2.5 between growth days 2 and 4 (Figure 4A). Intriguingly, we also observed an increase in cfDNA levels of Nkx2.5 during the same time frame using ddPCR (Figure 4B). We therefore wished to examine the prevalence of additional genomic regions within the cfDNA collected during cardiac organoid differentiation. The transcription factor p53, which is widely recognized for its role in cancer progression, has also been observed to regulate many components of cardiac transcription including Nkx2.5 (Mak et al., 2017). We observed that cfDNA levels of genomic p53 regions were increased later in cardiac differentiation (Figure 4C). However, in contrast to Nkx2.5, tissue-level expression of p53 decreased as cardiac organoids differentiated (Figure 4D and E). Although the endodermal marker Sox17 and stem cell marker Oct3/4 were both detectable by ddPCR in cardiac organoid-derived cfDNA, no significant changes with respect to cardiac differentiation were observed (Figure 4—figure supplement 1). Troponin T was not detectable in cardiac organoid-derived cfDNA. Together, these results suggest that specific regions of genomic DNA may be reflective of cardiac differentiation status.

Figure 4 with 1 supplement see all
Abundance of specific gDNA sequences is reflective of tissue-level changes in marker expression.

(A) Immunofluorescence staining of Nkx2.5 in cardiac organoids during differentiation. Scale bars represent 200 μm. (B) Nkx2.5 copies/μl in 1 ng of cardiac organoid-derived cfDNA at different time points during differentiation obtained using ddPCR, taken from n = 3 biological replicates. (C) p53 copies/μl in 1 ng of cardiac organoid-derived cfDNA at different time points during differentiation obtained using ddPCR, taken from n = 3 biological replicates. (D) Western blot showing p53 protein levels in cardiac organoids during differentiation. (Vinculin loading control is the same for Figure 3E.) (E) Quantification of the western blot in panel (D), normalized to Vinculin (n = 3 biological replicates). Graphs show average concentrations + SD. Samples were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant.

Figure 4—source data 1

Copy number of Sox17 or Oct3/4 in cfDNA derived from cardiac organoids on different growth days during development.

https://cdn.elifesciences.org/articles/83532/elife-83532-fig4-data1-v2.zip

Doxorubicin causes severe cardiac organoid malformation in comparison to CPI-203

To determine how or whether mitochondrial and genomic cfDNA markers change in response to toxicity and chemical exposure, we treated cardiac organoids with the known cardiotoxicant doxorubicin (DOX) or CPI 203 (CPI) – an epigenetic bromodomain inhibitor with unclear cardiotoxic potential. The drugs were added to the culture medium on day 4 for 48 hr, and cfDNA was collected at two time points post drug addition (Figure 5A). By day 6, organoids treated with either of these drugs exhibited decreased organoid size and an increased ring of diffuse cells and debris compared to control tissues (Figure 5B–D). By day 9, the tissues treated with DOX exhibited severe toxicity and had nearly dissociated; whereas, CPI-treated tissues demonstrated tissue recovery with a nominal size decrease compared to controls (Figure 5E–G). By day 9, detectable bpm were significantly reduced in DOX-treated organoids in contrast to CPI-treated tissues where bpm were comparable to untreated tissues (Figure 5H). Both drugs resulted in decreased size of mature organoids, with DOX having a more severe effect than CPI (Figure 5I).

Figure 5 with 1 supplement see all
Doxorubicin (DOX) causes severe cardiac organoid malformation in comparison to CPI.

(A) Schematic showing drug treatment and collection times of cfDNA during cardiac organoid growth. Representative images of tissues on day 6: control (B), DOX-treated (C), and CPI-treated (D). Representative images of tissues on day 9: control (E), DOX-treated (F), and CPI-treated (G). Scale bars represent 100 µm. (H) Beats per minute measured on growth day 9 for tissues treated with either DOX or CPI compared to DMSO-treated control. Shown are the average counts for 3–5 tissues each across three independent replicates. (I) Average tissue size of mature cardiac organoids treated with DOX or CPI measured on growth day 9. Shown are the average counts across three independent replicates (3–5 organoids per replicate). (J) ddPCR showing tissue-level expression of Nkx2.5 normalized to TBP on growth day 6 in cardiac organoids treated with DOX or CPI. Graphs show average concentrations + SD. Samples were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant. (Tissue morphology in response to treatment with DOX later in organoid growth is shown in Figure 5—figure supplement 1.)

Figure 5—source data 1

Doxorubicin causes severe cardiac organoid malformation in comparison to CPI-203.

https://cdn.elifesciences.org/articles/83532/elife-83532-fig5-data1-v2.zip

We wished next to know how these drug treatments impacted differentiation of the cardiac organoids. On day 6, levels of the cardiac transcription factor Nkx2.5 were substantially increased in samples treated with DOX (Figure 5J), while CPI-treated organoids did not significantly differ in expression compared to control. Together, these results indicate that treatment with DOX dysregulates cardiac gene expression and causes more severe toxicity in cardiac organoids than CPI in this model, at the concentrations used. Notably, the timing of drug addition greatly impacted the degree of toxicity observed. Treating tissues with DOX later in development on day 7 resulted in less dramatic toxicity and no substantial changes in beating rate or tissue size compared to control. DOX still increased Nkx2.5 expression regardless of treatment time, while the impact of CPI on Nkx2.5 expression was less pronounced (Figure 5—figure supplement 1).

Specific sequences of cfDNA may be predictive of toxicity in cardiac organoids

Our next step in the study investigation was to determine whether specific properties of cfDNA could be predictive of the outcomes we observed in each of these conditions: severe cardiac malformation (DOX), mild effects (CPI), or normal (control). Tissues that had been treated with DOX showed significantly increased levels of the mitochondrial markers ND1 and mtCOX2 on day 6 after 48 hr of drug exposure (Figure 6A and B). Nkx2.5 was also increased at the cfDNA level in DOX-treated samples (Figure 6C), while cfDNA levels of p53 were unchanged by toxicant addition (Figure 6D). The impact of DOX or CPI treatment on the abundance of these markers in cfDNA later in development was more subtle. In addition, the overall concentration of cfDNA was not significantly impacted by these drug treatments either early or late in development (Figure 6—figure supplement 1). Notably, the increases in ND1, mtCOX2, and Nkx2.5 in cfDNA collected from DOX-treated samples on day 6 preceded the detriment of the organoids we observed on day 9. Further, this increase was not observed in CPI-treated samples, which had largely recovered tissue morphology and beating function by day 9. Together, these results suggest that the abundance of specific markers (ND1, mtCOX2, and Nkx2.5) in cfDNA may be predictive of the severity of exposure outcomes in cardiac development.

Figure 6 with 1 supplement see all
Specific cfDNA sequences may be predictive of toxicity in cardiac organoids.

Abundance of ND1 (A), mtCOX2 (B), Nkx2.5 (C), or p53 (D) in 0.5 ng cfDNA collected from cardiac organoids on growth day 6 treated with doxorubicin (DOX) or CPI, obtained using ddPCR. Graphs show average concentrations + SD. Samples were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant. (Concentration of cfDNA and additional analysis of cfDNA sequences upon treatment with DOX later in growth are shown in Figure 6—figure supplement 1.)

Figure 6—source data 1

Specific sequences of cfDNA may be predictive of toxicity in cardiac organoids.

https://cdn.elifesciences.org/articles/83532/elife-83532-fig6-data1-v2.zip

Discussion

Previously, cfDNA had been extracted successfully from pancreatic organoids derived from cancer patients (Dantes et al., 2020). However, whether cfDNA could be recovered from additional organoid types and the utility of these models for drug screening and predictive biomarker identification has been largely unexplored. Our study designed a method for rapid screening of cfDNA from cardiac organoids. We identified that decreases in cell-free mitochondrial DNA levels occur in response to cardiac differentiation. Further, we observed increases in specific cfDNA sequences corresponding to the genes Nkx2.5 and p53 upon cardiac organoid maturation. Our findings indicate that cfDNA concentration alone is not a reliable marker of toxicity. Rather, the prevalence of cell-free mitochondrial DNA and specific genomic regions may be more predictive of deleterious events. Specifically, cfDNA levels of ND1, mtCOX2, and Nkx2.5 in mature cardiac organoids may be potential predictive biomarkers of DOX-induced cardiotoxicity. We saw increases in the abundance of these markers on day 6 in DOX-treated samples, but not CPI-treated samples. By day 9, the organoids treated with DOX early in growth were severely malformed, while the CPI-treated organoids had mostly recovered.

Although our cardiac organoid model does not approach the complexity of cardiac growth in vivo, mapping which combinations of cfDNA markers reflect specific cardiotoxic outcomes would be a valuable step towards implementing cfDNA as a clinical screening tool. In addition, organoid models permit the investigation of cfDNA in response to toxic compounds and chemicals with unclear bioactive potential, which could aid the development of cfDNA as a tool for detection of toxic exposures. Future work is needed to explore the impact of additional compounds and drugs on cardiac cfDNA, and what changes occur in other tissue types. Cell-free mitochondrial DNA output in response to differentiation and toxicity may be an important area of further investigation. The increase in cardiac organoid-derived cell-free mitochondrial DNA observed after treatment with DOX might reflect aberrant cardiomyocyte death. Accordingly, the decrease in cell-free mitochondrial DNA during cardiac organoid differentiation may represent preferential retention of cells with more mature mitochondria. Alternatively, the channels mPTP and VDAC may play a role in increased cfDNA release from stressed mitochondria (Xian et al., 2022). Fittingly, we observed that cell-free mitochondrial DNA markers (ND1, mtCOX2) were increased in organoids exposed to DOX. The exact mechanism of preferential cell-free mitochondrial DNA release by certain cell types is an intriguing area of future study.

In addition, identification of cfDNA sequences of nuclear origin is also critical for our understanding of the cell-free genome. Differential abundance of cell-free genomic DNA may reflect regions of DNA protected from DNases by transcription factor binding or epigenetic modifications (Ulz et al., 2019). The transcription factor p53, which is widely recognized for its role in cancer progression, also has demonstrated involvement in cardiac tissues and interacts with Nkx2.5 (Kojic et al., 2015). Although both of these markers increased at the cfDNA level as the organoids matured, we observed with interest that the overall expression of Nkx2.5 increased whereas that of p53 decreased at the tissue level during maturation. Analysis and identification of multiple combinations of cfDNA markers may ultimately be necessary to effectively distinguish events at the tissue level. The cfDNA we recovered from cardiac organoids consisted of a heterogeneous distribution of fragment lengths that ranged from approximately 200–6000 base pairs. The shorter fragments are potentially indicative of DNA that has been cleaved during apoptosis, while the longer lengths could result from necrotic processes (Aucamp et al., 2018). However, cfDNA can also be expelled actively from healthy cells in the form of extracellular vesicles or nucleoprotein complexes. Further, different DNases may contribute uniquely to patterns of fragmentation (Han et al., 2020). Tracing individual fragments back to the cellular event triggering their release would be a valuable area of future research.

In the bloodstream, multiple cell types contribute to the cfDNA population, with a large fraction released by white blood cells and erythrocyte progenitor cells (Moss et al., 2018). This complexity highlights the need to identify key cfDNA sequences that can be used as clinical markers for specific conditions. Although media conditioned by organoids in culture does not approach the complexity of human blood or bodily fluids, organoids provide a tractable system for identifying potential cfDNA targets in response to known tissue-level events. These cfDNA markers may be valuable targets for further screening in clinical samples. Although cfDNA concentration is known to be elevated in cardiac disease patients (Polina et al., 2020), we did not observe a significant increase in overall cfDNA concentration in response to phenotypic changes. Although cell death is one route of cfDNA release, alternative mechanisms exist through which cfDNA concentration may increase in the absence of cell death, such as association with extracellular vesicles (Kustanovich et al., 2019; Aucamp et al., 2018). In clinical samples, cfDNA concentration can fluctuate substantially between individuals and increase in response to a number of conditions including cancer (Salvi et al., 2016), exercise (Breitbach et al., 2012), aging, and recent surgery (Aucamp et al., 2018). These observations combined with our results indicate that cfDNA concentration alone may not be an optimal parameter for targeted disease diagnosis. Rather, screening for specific cfDNA sequences may be more indicative of certain conditions.

In this study, we have demonstrated that cfDNA can be reliably extracted from cardiac organoid models. Further, we present the tractability of organoid systems for screening cfDNA markers in response to tissue-level events, including differentiation and malformation. This workflow has the potential for rapid screening of cfDNA from toxicity and disease models and could likely be extended to additional organoid systems. Our future efforts include the identification of additional cfDNA sequences that are impacted by differentiation and understanding how different chemicals impact the cfDNA profile of mature and developing cardiac tissues. Ultimately, our approach could contribute to establishing a database of cfDNA markers that map to specific disease processes or toxicants, which would hold promise for early, non-invasive detection of disease.

Study limitations and future directions

Here, cfDNA samples were collected at a small number of timepoints along the development of the organoids. Future research might explore more frequent samplings of cfDNA to obtain a better understanding of how cfDNA changes in time during organoid growth. In addition, it is unclear to what degree cfDNA may be degraded between the time it is released from the tissues to the time of collection. Further, although we attempted to minimize DNA degradation post-collection by limiting freeze–thaw cycles and using good sample storage practices, we cannot rule out that longer fragments of DNA could have broken down into shorter ones over time. In addition, although the Maxwell ccfDNA automated extraction kit is one of the better tools currently available for cfDNA extraction (Sorber et al., 2017), we cannot rule out potential bias towards short fragments. The electropherograms which showed cfDNA size distribution are qualitative and cannot be taken as a definitive statement of fragment profiles, especially since events such as post-collection degradation could be missed through this analysis. Future studies might investigate more closely the molecular dynamics of cfDNA release and the timing of its degradation, and also how degradation rate may vary in culture media versus the bloodstream.

To account for the growing size of the organoids, we used total tissue-level genomic DNA concentration to normalize cfDNA concentrations as the dense nature of the tissues hindered their dissociation and single-cell counts via enzymatic or mechanical means. However, this is only an approximation and may have introduced errors resulting from multinucleated cells, aneuploidy, or cells undergoing division. Future work might seek to optimize methods of normalizing cfDNA concentration to further improve our knowledge of how cfDNA concentration is impacted by tissue-level events. Unlike tissue-level analyses which examine expression of RNA transcripts converted to complementary DNA (cDNA), cell-free DNA (cfDNA) consists of expelled DNA fragments that are present in the extracellular media. It is currently unclear whether certain fragment sequences might be present more consistently than others, so it is therefore difficult to normalize ddPCR analyses of cfDNA copy number to a housekeeping gene. This is a possible source of error in determining the abundance of specific cfDNA sequences. Future studies of normalization and copy number analyses in free nucleic acids would be valuable towards precise quantitative assessments of cfDNA sequences.

In this study, only a handful of cfDNA sequences were assessed. In addition, our study only assessed organoids derived from one cell type (H9 ESCs) in a limited number of batches (n = 3). However, we feel this work has demonstrated that cfDNA can be successfully recovered from cardiac organoids in quantities sufficient for characterization and quantification. This lays the groundwork for further studies of cfDNA from additional organoid systems in response to toxicant treatments, as well as broader global analyses such as microarrays or sequencing.

Materials and methods

Cell culture

Request a detailed protocol

WA09 (H9) human embryonic stem cell line was purchased from WiCell. Authentication of cell lines and mycoplasma was conducted by WiCell. H9 human embryonic stem cells (hESCs) were maintained in mTeSR+ medium (Stemcell Technologies, 100-0276) at 37°C and 5% CO2. Cells were passaged between 60 and 80% confluence using 0.5 mM EDTA (Gibco, 15575) onto plates pre-coated with growth factor-reduced Matrigel (Corning, 354230) for 30 min at 37°C in DMEM/F:12 (Gibco, 11330-032) at a concentration of 1.2 µL/mL cm2. Cultures were passaged a maximum of 10 times provided no morphological changes indicative of differentiation were present.

Cardiac organoid generation

Request a detailed protocol

Cardiac organoids were generated using a previously established protocol (Israeli et al., 2020). Briefly, H9 hESCs were detached using TrypLE (Gibco, A12177-01) and resuspended at a density of 100,000 cells/mL in E8 media (Gibco, A15169-01) + 10 µM Y27632 (Tocris, 129830-38-2). 100 µL of cell suspension was added per well to a 96-well round-bottom ultra-low attachment plate (Thermo Fisher, 174925) and centrifuged for 5 min at 200 × g to coalesce the cells. The day of seeding was defined as day 2 of growth. On day 1, the media was changed by removing 50 µL media from each well and adding 200 µL fresh prewarmed E8 media. On day 0, 166 µL media was removed and replaced with 166 µL RPMI 1640 (Gibco, 11875-093) supplemented with 1x B27 (no insulin; Gibco, A18956-01), CHIR 99021 (4 µM; Selleck, S2924), BMP4 (0.36 pM; BioTechne, 314BP-010/CF), and ActA (0.08 pM; BioTechne, 338-AC-010/CF). On day 1, 166 µL of media was removed and replaced with 166 µL RPMI 1640 supplemented with B27 (no insulin). On day 2, 166 µL of media was removed and replaced with 166 µL RPMI 1640 supplemented with B27 (no insulin) and IWR-1-endo (5 µM; Selleck, S7086), and incubated at 37°C for 48 hr. On day 6, 166 µL of media was removed and replaced with 166 µL of RPMI 1640 supplemented with 1x B27 (with insulin; Gibco, 17504-044). On day 7, 166 µL of media was removed and replaced with 166 µL RPMI 1640 supplemented with B27 (with insulin) and 4 µM CHIR99021, and incubated for 1 hr at 37°C, then the media was again replaced as described above using RPMI 1640 supplemented with B27 (with insulin; no CHIR). The plate was incubated at 37°C for 48 hr, and the cardiac organoids were considered mature on day 9. To examine the effects of chemical addition, doxorubicin (DOX, 1 nM; Sigma, D1515) or CPI-203 (0.5 μM; Tocris, 5331) were applied for 48 hr on either day 4 or 7 of cardiac organoid development (control, 1:1000 DMSO) in culture media.

Nucleic acid extraction

Request a detailed protocol

RNA and gDNA from organoids (Promega, miRNA Tissue Kit AS1460; Tissue DNA Kit AS1610) and cfDNA from media (Promega, ccfDNA Plasma Kit AS1480) were extracted using the Promega Maxwell RSC instrument. RNA quality and purity were assessed using a NanoDrop 2000 spectrophotometer, and DNA concentrations were measured using the Promega QuantiFluor dsDNA System. A minimum of 3–5 organoids were used per biological replicate for RNA and gDNA extractions. RNA was converted to cDNA using the iScript Reverse Transcription Kit (Bio-Rad). CfDNA from cardiac organoids was collected from media conditioned by 10 tissues (2–3 technical replicates per batch, 3 independent batches). Prior to cfDNA extraction, all conditioned media was centrifuged for 10 min at 1600 × g, the top 800 µL supernatant centrifuged 10 min at 16,000 × g, and the top 400 µL collected and stored at –80°C until extraction. Prior to extraction, conditioned media was thawed at 4°C to prevent degradation.

Quantification of cfDNA concentration, fragment analysis, and gene expression

Request a detailed protocol

Nucleic acid concentration was measured using the QuantiFluor ONE dsDNA System (Promega, E4871). Fragment analysis of cfDNA size was performed using the 5200 Fragment Analyzer System, Agilent Inc using the HS Large Fragment 50 kb Kit (Agilent, DNF-464-0500). Gene expression and cfDNA copy number were evaluated using droplet digital PCR (ddPCR; HEX/FAM system) using 0.5 ng or 1 ng of cDNA or cfDNA per reaction. Prior to quantification, cfDNA concentrations were diluted to 0.05 ng/µL and 10 µL added per reaction. ddPCR was performed using 25 µL reaction volumes with ddPCR Supermix for Probes (no dUTP; Bio-Rad) and 1 µL each of the following probes: Troponin T (dHsaCPE5052345, Bio-Rad), Nkx2.5 (dHsaCPE5042098, Bio-Rad), Oct 3/4 (POU5F1, dHsaCPE5191335, Bio-Rad), ND1 (dHsaCPE5029121, Bio-Rad), MT-COX2 (dHsaCPE5192286), p53 (TP53, dHsaCPE5037521, Bio-Rad), SOX17 (dHsaCPE5039713, Bio-Rad), ND6 (dHsaCNS941916401, Bio-Rad), ND4 (dHsaCNS186386931, Bio-Rad), GAPDH (dHsaCPE5031597, Bio-Rad), TPB (dHsaCPE5058363, Bio-Rad), and β-actin (dHsaCPE5190200, Bio-Rad). Droplets were generated using the Automated Droplet Generator (Bio-Rad, CA) and PCR performed using annealing conditions of 60°C for 1 min, 40 cycles. Droplets were read using the QX200 Droplet Reader (Bio-Rad). Thresholding was applied equally across all sample conditions under comparison for each probe.

Protein detection and quantification

Request a detailed protocol

A minimum of five organoids per batch were rinsed 1× with PBS and homogenized in Pierce RIPA (Thermo Scientific, 89900) supplemented with 25× cOmplete Protease Inhibitor, EDTA-free (Roche, 11836170001) and centrifuged for 15 min at 14,000 × g at 4 °C. The supernatant was removed and stored at –80°C until analysis. Protein concentration was calculated using the Pierce BCA Protein Assay (Thermo Scientific, 23225). 3.5 µg protein was incubated for 10 min at 70°C with NuPAGE LDS Sample Buffer (Invitrogen, NP0007) and NuPAGE Sample Reducing Agent (Invitrogen, NP 0009), loaded onto a NuPAGE 4 to 12%, Bis-Tris gel (Invitrogen, NP0336BOX), and run for 35 min at 200 V in 1× NuPAGE MES Run Buffer (Invitrogen, NP0002) plus NuPAGE Antioxidant (Invitrogen, NP0005). The proteins were transferred to nitrocellulose membranes using the iBlot Gel Transfer Device (Thermo Fisher). Protein transfer was visualized using Ponceau S solution (Sigma, P7170), and the membranes were cut into regions containing the proteins of interest. Membranes were blocked for 15 min in EveryBlot Blocking Reagent (Bio-Rad, 120110020) and incubated with primary antibody overnight at 4°C diluted 1:1000 in EveryBlot Blocking Reagent unless otherwise specified: rabbit anti-vinculin (Novus, NBP2-20859), rabbit anti-Nkx2.5 (Cell Signaling, 8792T), rabbit anti-VDAC (Abcam, ab15895), mouse anti-p53 (1:500; Novus, NBP2-29453), rabbit anti-MEF2C (Cell Signaling, 5030), rabbit anti-GATA-6 (Cell Signaling, 5851), rabbit anti-α-actinin (Cell Signaling, 6487), rabbit anti-GAPDH (Bio-Rad, VPA00187). Following primary antibody incubation, membranes were washed 3 × 5 min in 1× Tris-buffered saline (TBS; Bio-Rad, 1706435) plus 0.1% tween-20 (Sigma, P7949) (TBST) and incubated for 45 min at room temperature with secondary antibody diluted 1:5000 in EveryBlot Blocking Reagent (Bio-Rad): goat-anti-rabbit HRP (Novus, NB7160) or goat-anti-mouse HRP (Invitrogen, 32230). Membranes were washed 3 × 15 min in TBST and HRP visualized using SuperSignal West Pico Plus Chemiluminescent Substrate (Thermo Scientific, 34580). Bands were quantified in ImageJ (Schneider et al., 2012) using a rolling ball background subtraction with r = 50, followed by normalization to vinculin.

Immunostaining

Request a detailed protocol

We used a previously described protocol (Matsumoto et al., 2019) for immunostaining and clearing of the organoids. Briefly, the organoids were fixed overnight in 4% paraformaldehyde followed by delipidation by successive overnight incubations in CUBIC-L solution (50%, 100%) at 37°C. The organoids were then blocked in 3% bovine serum albumin and incubated for 2 d in rabbit anti-Nkx2.5 primary antibody diluted 1:200 in blocking solution. Excess primary antibody was removed using PBST (0.2% Triton X-100 diluted in 1× PBS). The organoids were then incubated again for 2 d in secondary antibody (1:200, Alexa 647 goat-anti-rabbit; Invitrogen, A21244) and DAPI (1 µg/mL; Sigma, P9542) diluted in blocking solution. Samples were cleared for imaging in 96-well optical plates (Thermo Fisher, M33089) using CUBIC-R solution as described previously (Matsumoto et al., 2019). Confocal images were obtained on a Zeiss LSM 880 inverted confocal microscope with AiryScan (Jena, Germany).

Statistical analysis

Request a detailed protocol

Data were analyzed using GraphPad Prism version 9.0.0 for Windows, GraphPad Software, San Diego, CA (https://www.graphpad.com/). Averages of sample conditions were compared using an unpaired t-test with Welch’s correction. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant. In instances where data were normalized to the initial timepoint (day 1) to account for batch variation, or ratios were compared, a one-sample t-test was used to compare the average of each normalized condition to a hypothetical value of 1.0, to detect significant deviation from the day 1 starting point. *p<0.05; **p<0.01; ***p<0.001; ns, nonsignificant. All graphs show average + SD unless otherwise stated. Three independent biological replicates (separate batches of organoids cultured consecutively on different days) were performed for each experiment.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting file.

References

    1. Willems PJ
    2. Dierickx H
    3. Vandenakker E
    4. Bekedam D
    5. Segers N
    6. Deboulle K
    7. Vereecken A
    (2014)
    The first 3,000 non-invasive prenatal tests (NIPT) with the harmony test in Belgium and the Netherlands
    Facts, Views & Vision in ObGyn 6:7–12.

Decision letter

  1. Abel Bronkhorst
    Reviewing Editor; German Heart Centre, Technical University Munich, Germany
  2. Didier YR Stainier
    Senior Editor; Max Planck Institute for Heart and Lung Research, Germany
  3. Abel Bronkhorst
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "Cell-free DNA as a potential biomarker of differentiation and toxicity in cardiac organoids" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Abel Bronkhorst as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Didier Stainier as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Please provide a point-by-point response to all reviewer comments.

2) Note that additional experiments have been requested by the reviewers. Please perform these experiments or provide a convincing rebuttal.

Reviewer #1:

CfDNA molecules possess various physicochemical features that correlate with the cellular origin and a wide range of disease indications, making it an ideal biomarker. However, systematic mapping of the information available through cfDNA characterization is complicated by the immense complexity of in vivo systems. While cell culture models represent only a limited view into complex biological systems, they have still proven to be extremely useful in many cases and a growing number of studies indicate their utility in cfDNA analysis.

in vitro research on cfDNA has to date focused mainly on 2D cell cultures. Therefore, this work by Silver et al. is among only a small number of studies that profiled cfDNA in 3D cell cultures and represents one of the most detailed assessments to date. They demonstrated that (i) cfDNA release correlates with differentiation; (ii) cell-free mitochondrial DNA levels correlate with the development of organoids; (iii) the sequence composition of cfDNA fluctuates during growth; and (iv) specific cfDNA features correlates with toxic drug treatment.

Based on these interesting observations the authors conclude that cfDNA profiling may not only serve as a useful tool to monitor the growth and development of organoids but may be leveraged to gain insight into tissue dynamics and potentially identify new biomarkers. The authors acknowledge that their study is only an explorative study and that more research is needed before in vitro data will reveal accurate and practically useful information about the in vivo setting. However, the various issues that may confound the results obtained in their study, and the general limitations of 3D cell models in relation to cfDNA analysis, are not clearly outlined. For example, (i) factors that affect the normalization of cfDNA concentration measurements, (ii) the influence of "natural" cell death on the measurement of drug-induced cell death-related cfDNA features, (iii) the effects of DNA degradation over time, (iv) important dynamic changes in cfDNA properties that are missed due to widely spaced measurement time-points.

1. Page 3, lines 53-54: "Several measures including concentration, fragment size, and sequence can be used to characterize cfDNA, and imply information about the cells from which it originated [1]."

• Several epigenetic features of cfDNA (e.g. fragment end-point motifs, nucleosome spacing, topological features, differentially methylated regions, and histone modifications) enable enhanced cell-of-origin identification. These features are increasingly characterized and could be mentioned here.

2. Page 3, lines 55-57: "Still, the current diagnostic uses of cfDNA are primarily limited to specific applications such as the detection of fetal chromosome abnormalities responsible for Down syndrome [6] and fetal gender determination from maternal blood [7]."

• There are a number of FDA-approved cancer cfDNA assays for routine clinical use. Most of these are companion tests but could be mentioned here, as the clinical use of cfDNA assays is not primarily limited to non-invasive prenatal testing (NIPT).

3. Page 5, lines 98-101: "To account for changes in cell number during organoid growth, we normalized the concentrations of cfDNA to levels of genomic DNA (gDNA) collected on the same growth days. Normalized cfDNA was highest early in development (Figure 2C), indicating that cfDNA output may be higher in tissues comprised of less differentiated cells."

Please expand this section after considering the following points:

• There is no data linking the timing of specific cellular events with the extracellular occurrence of DNA fragments. For example, cell division, differentiation, apoptosis, necrosis, active release, drug-induced death, etc., may be initiated at one point in time, while DNA may only be released into the growth medium at a much later time point. Similarly, a major portion of specific DNA fragments released due to specific reasons may already be degraded at the time of measurement. How is this factored into cfDNA measurements and normalization of measurements?

• DNA content does not necessarily reflect cell number, for example during specific cell division phases or aneuploidy

• Is it possible and would it be better to perform both gDNA analysis and total cell counts?

• Would it be better to perform more frequent cfDNA assessments, e.g., every couple of hours?

4. Page 5, lines 102-103: "The recovered cfDNA consisted of a broad distribution of fragments ranging from approximately 200 to 6000 base pairs (Figure 2D)."

• Some comments here:

• The cfDNA size profile is interesting. The short fragments demonstrate an apoptotic laddering pattern, but there is also a population of long cfDNA fragments, which may indicate an origin from apoptosis. Could the authors elaborate on this?

• A recent publication on the characterization of cfDNA in 2D cell cultures suggested that shorter fragments and longer fragments may originate from the same process and that longer fragments are degraded into shorter fragments over time.

o This paper also shows that the apparently high concentration of longer cfDNA fragments as seen in electropherograms may be an artefact of the method. Could the authors comment on this?

• If it is true that longer fragments also contain important biological information, this information could be missed in some cases, for example:

o The use of extraction kits that are biased toward the capture of short fragments

o DNA sequencing techniques that are limited to short reads

o cfDNA sizing using PCR techniques

• The method used for cfDNA sizing is not discussed in the methods and materials section.

• It would have been interesting to not only see the cfDNA size profile on one day but how it changes over the course of incubation. Perhaps the authors can speculate on what insights this may reveal.

5. Another factor that may complicate the characterization of organoid-derived cfDNA:

• Literature suggests that natural levels of apoptosis /necrosis increase over the course of organoid growth and development. How does one differentiate between cfDNA fragments originating from "natural" vs "induced" cell death? Is this a confounding factor for analysis?

Reviewer #2:

Silver et al. demonstrated that cell-free DNA can be detected from cultures of cardiac organoids. This subject has not been explored and presents a real interest towards the possible use of organoids to detect markers according to the physiological or pathological evolution of the original tissues. So they tried to assess the possibility of cardiotoxicity of a drug from organoid cultures. The writing of the manuscript is of good quality, concise, and the presentation of the results is clear. However, the manuscript has many flaws including many inaccuracies; and authors should address the following major concerns:

1. The analysis of free mitochondrial DNA raises many questions: (i), there is no technical explanation for standardization, calibration, and direct comparison of the quantification of the number of copies of different regions of the mitochondrial genome. The quantification of mitochondrial DNA by the quantitative PCR technique must be determined with caution; (ii) the readers may remain confused regarding the method used to determine DNA concentration; (iii) the reviewers is confused throughout the manuscript in respect to characterize gene expression (RT-PCR) and cfDNA amount (qPCR); (iv), the authors state in the abstract that there is a decrease in the fraction of free mitochondrial DNA during development organoids which is not proven by the results; (v), on what basis can the authors say that this identifies a unique signature of cardiac differentiation if they did not make a comparison with other cell types; (vi), the authors do not comment on why the expression of the four mitochondrial regions is equivalent, which is counterintuitive; (vii), Figure 3C shows standard variations when it looks impossible when showing ratios; and (viii), there is no discussion when comparing amount of cfDNA from both origins while they showed an apparent similar level, especially in regards to the literature comparing for instance cfDNA from mitochondria or nucleus.

2. How is it possible that wild-type DNA, in this case P53, can show a difference in quantity by quantitative PCR analysis during differentiation?

3. No or very few indications specified the number of passages or the number of organoids within the same culture. This is crucial and must be filled in.

Reviewer #3:

The article analyses for the first time the kinetics of different cfDNA species during the development of cardiac organoids generated from H9 human embryogenic stem cells. The stimulation of the cells with Doxorubicin or CPI, as a model of toxic exposure, is a relevant approach to identifying the validity of cfDNA as biomarkers of toxic events. The authors show an increase in cfDNA and a decrease in mtDNA during cardiac organoid differentiation. However, the research relies on a single cell line, and the number of biological replicates with n = 3 is very small. The inclusion of additional two or three replicates is required to validate the finding. Moreover, the inclusion of another selected cell line would strengthen the finding.

The article is well-written and the authors used various methodological approaches to prove their findings. However, with respect to the main message, namely, the increase of specific cfDNA targets in response to DOX treatment, the reviewer has some concerns. Those are related to statistical aspects as well as methodological as described below.

– In figures 1-4 the authors compare paired samples over time. However, in their figure legends, the authors indicate that an unpaired test was used. Does this have any rationale like missing samples? The single point of the samples should be included in the figure.

– In all figure legends it is noted that t-tests with Welch´s correction were used unless otherwise indicated. The authors could indicate if a t-test (homogenous variances between samples) or a welch´s t-test (not normal variances) was used.

– The authors should additionally highlight which of the results are related to cfDNA or cDNA analysis. For example, Figures 4 B and C relate to cfDNA (labelled with cfDNA). Figure 5 J relates to cDNA (not labelled). Figure 6 (not labelled with cfDNA).

– Figure 3 B, please clarify if the relative abundance of mtDNA relates to total cfDNA or one of the reference genes.

– Figure 4 should be restructured. Figure 4B, C, F, and G belong together, and D and E belong together. As long as F and G display cfDNA analysis (should be highlighted). Aftewards the legend text could be reduced relevantly.

– Figure 6: The copy number of the different cfDNA targets relies relatively on the input material (0.5 or 1 ng were included in the ddPCR assay). Especially for Figure 6 and the discussion that ND1 and NKx2.5 increase in response to DOX treatment, but not the total amount of cfDNA and not p53. Can this result be strengthened by the analysis of GAPDH or TBP in the same sample?

– The reviewer feels that 3 more replicates are needed to validate the finding of different cfDNA release in response to DOX. Moreover, a comparison to another suitable and well-selected cell line would be welcome.

https://doi.org/10.7554/eLife.83532.sa1

Author response

Reviewer #1:

1. Page 3, lines 53-54: "Several measures including concentration, fragment size, and sequence can be used to characterize cfDNA, and imply information about the cells from which it originated [1]."

• Several epigenetic features of cfDNA (e.g. fragment end-point motifs, nucleosome spacing, topological features, differentially methylated regions, and histone modifications) enable enhanced cell-of-origin identification. These features are increasingly characterized and could be mentioned here.

The reviewer raises an excellent point. Epigenetic modifications are indeed an important area of cfDNA research in addition to the characterizations we mentioned. We have added this information and the additional supporting references as shown below.

“Several measures including concentration, fragment size, sequence, and epigenetic modifications such as differential methylation, histone modification, and nucleosome spacing can be used to characterize cfDNA,and imply information about the cells from which it originated [1, 6, 7].”

2. Page 3, lines 55-57: "Still, the current diagnostic uses of cfDNA are primarily limited to specific applications such as the detection of fetal chromosome abnormalities responsible for Down syndrome [6] and fetal gender determination from maternal blood [7]."

• There are a number of FDA-approved cancer cfDNA assays for routine clinical use. Most of these are companion tests but could be mentioned here, as the clinical use of cfDNA assays is not primarily limited to non-invasive prenatal testing (NIPT).

We thank the reviewer for raising this point and have revised and expanded this section to include the additional mentioned assays for cancer. In addition, we have reworded these statements to more accurately reflect our view that cfDNA likely has many additional uses in noninvasive testing beyond that which is currently available.

“Currently, cfDNA already has some diagnostic uses including noninvasive prenatal testing for chromosome abnormalities [8], fetal gender determination [9], and cfDNA-based tests for the detection of mutations and cancer screening [10, 11]. Although the clinical utility and accuracy of these tests in the prediction of cancer is uncertain in some cases [12], clinical trials are promising and some FDA-approved tests are already available [10]. Still, we have likely yet to unlock the full spectrum of clinical usage for cfDNA in noninvasive diagnostics.”

3. Page 5, lines 98-101: "To account for changes in cell number during organoid growth, we normalized the concentrations of cfDNA to levels of genomic DNA (gDNA) collected on the same growth days. Normalized cfDNA was highest early in development (Figure 2C), indicating that cfDNA output may be higher in tissues comprised of less differentiated cells."

Please expand this section after considering the following points:

• There is no data linking the timing of specific cellular events with the extracellular occurrence of DNA fragments. For example, cell division, differentiation, apoptosis, necrosis, active release, drug-induced death, etc., may be initiated at one point in time, while DNA may only be released into the growth medium at a much later time point.

We thank the reviewers for bringing up this point, and we agree that our data do not necessarily link cfDNA abundance to cellular differentiation status or specific events, since multiple cell types are present and change dynamically throughout the cardiac organoid development process. Accordingly, we have revised the statement in lines 98-101 to clarify that our cfDNA observations are associated with organoid growth day, and not necessarily specific cellular events. Although an assessment of the precise cellular origin of cfDNA and the impact of specific differentiation events on cfDNA is beyond the scope of this study, we feel additional future research in this area would be of great value. To further address the points discussed by the reviewer, we have added a limitations section in our discussion (page 13) to further clarify the boundaries of our current study and suggest future research topics.

Lines 98-101: “Normalized cfDNA was highest early in development (Figure 2C), indicating that cfDNA output may be higher in tissues on growth day 1, which are comprised of less differentiated cells.”

Similarly, a major portion of specific DNA fragments released due to specific reasons may already be degraded at the time of measurement. How is this factored into cfDNA measurements and normalization of measurements?

The reviewer raises a thoughtful point. Although we attempted to minimize potential degradation by limiting freeze/thaw cycles and keeping conditioned media and cfDNA at 4oC between steps, a molecular description of how cfDNA is degraded between the time of release and collection was not studied here. We feel this would be beyond the scope of our current study as we were more interested in the validation of cardiac organoids as a tool for identifying potential candidate cfDNA biomarkers. Future research might explore the molecular dynamics of cfDNA release and degradation, and how degradation rate may vary in culture media vs the bloodstream. To highlight this important area of future research, we have included the following statements in the limitations section of our discussion.

“… it is unclear to what degree cfDNA may be degraded between the time it is released from the tissues to the time of collection. Future studies might investigate more closely the molecular dynamics of cfDNA release and the timing of its degradation, and also how degradation rate may vary in culture media versus the bloodstream.”

• DNA content does not necessarily reflect cell number, for example during specific cell division phases or aneuploidy

• Is it possible and would it be better to perform both gDNA analysis and total cell counts?

We thank the reviewer for this point. Accounting for cell number using gDNA may indeed have slight inaccuracies due to events such as the reviewer mentions. We previously attempted single cell counts but were unable to dissociate the cells in the dense organoids via mechanical and enzymatic means without damaging them, which hindered obtaining accurate single cell counts. We felt using gDNA content was our best attempt to normalize cfDNA to account for an estimate of growing tissue size. To clarify that this could be a potential source of error, we have included the following statement in the limitations section of our discussion:

“To account for the growing size of the organoids, we used total tissue-level genomic DNA concentration to normalize cfDNA concentrations, as the dense nature of the tissues hindered their dissociation and single-cell counts via enzymatic or mechanical means. However, this is only an approximation, and may have introduced errors resulting from multinucleated cells, aneuploidy, or cells undergoing division. Future work might seek to optimize methods of normalizing cfDNA concentration to further improve our knowledge of how cfDNA concentration is impacted by tissue-level events.”

• Would it be better to perform more frequent cfDNA assessments, e.g., every couple of hours?

Although an interesting question, current cardiac organoid culture methods require precise timing of media changes (24 or 48 hr). It is unclear whether organoids can be successfully generated with more frequent media collections. Additionally, because the organoids are cultured over a 12-day period, taking more frequent timepoints (every couple of hours) would yield a very large number of samples that would exceed our current capabilities in terms of cost and time. Although beyond the scope of our current study, we feel the experiment the reviewer suggests would indeed be a valuable area of future research. To highlight this point, we have added the following statements to our discussion:

“Here, cfDNA samples were collected at a small number of timepoints along the development of the organoids. Future research might explore more frequent samplings of cfDNA, to obtain a better understanding of how cfDNA changes in time during organoid growth.”

4. Page 5, lines 102-103: "The recovered cfDNA consisted of a broad distribution of fragments ranging from approximately 200 to 6000 base pairs (Figure 2D)."

• Some comments here:

• The cfDNA size profile is interesting. The short fragments demonstrate an apoptotic laddering pattern, but there is also a population of long cfDNA fragments, which may indicate an origin from apoptosis. Could the authors elaborate on this?

The reviewer raises a valuable point. Accordingly, we have expanded our discussion to include a statement of possible origins of cfDNA fragment lengths and how these might be produced through different modes of cellular death.

“The cfDNA we recovered from cardiac organoids consisted of a heterogeneous distribution of fragment lengths that ranged from approximately 200 to 6000 base pairs. The shorter fragments are potentially indicative of DNA that has been cleaved during apoptosis, while the longer lengths could result from necrotic processes [19]. However, cfDNA can also be expelled actively from healthy cells in the form of extracellular vesicles or nucleoprotein complexes. Further, different DNases may contribute uniquely to patterns of fragmentation [20]. Tracing individual fragments back to the cellular event triggering their release would be a valuable area of future research.”

• A recent publication on the characterization of cfDNA in 2D cell cultures suggested that shorter fragments and longer fragments may originate from the same process and that longer fragments are degraded into shorter fragments over time.

Although we attempted to minimize degradation by storing DNA at -80oC, limiting freeze/thaw cycles, and keeping DNA/reagents at 4oC during protocols, it is indeed possible that longer fragments could have been degraded into shorter fragments post-collection. To highlight this potential source of error, we have included the following statements in the limitations section of our discussion.

“… although we attempted to minimize DNA degradation post-collection by limiting freeze-thaw cycles and using good sample storage practices, we cannot rule out that longer fragments of DNA could have broken down into shorter ones over time.”

o This paper also shows that the apparently high concentration of longer cfDNA fragments as seen in electropherograms may be an artefact of the method. Could the authors comment on this?

We are unfortunately not completely clear which paper/method the reviewer is referring to. We located this study (doi: 10.2144/btn-2022-0040) which shows that standard culture methods may introduce contaminating cfDNA fragments, resulting from FBS. However, our culture system uses serum-free media so we do not believe this is of concern. In addition, we wish to emphasize that our intention in showing the electropherograms was only to provide a qualitative assessment of fragment lengths, and we agree that definitive conclusions or quantifications cannot be drawn from these. Accordingly, we have included the following statements in our discussion.

“Further, the electropherograms which showed cfDNA size distribution are qualitative, and cannot be taken as a definitive statement of fragment profiles, especially since events such as post-collection degradation could be missed through this analysis.”

• If it is true that longer fragments also contain important biological information, this information could be missed in some cases, for example:

o The use of extraction kits that are biased toward the capture of short fragments

o DNA sequencing techniques that are limited to short reads

o cfDNA sizing using PCR techniques

Although we used standard extraction protocols and made our best effort to control for variability, it is of course possible that bias towards specific fragments could have occurred during the extractions. DNA sequencing or sizing of cfDNA using PCR were not employed in this study, but indeed future studies using such techniques should be aware that longer fragments could potentially be excluded. Accordingly, we have included the following statement in our discussion.

“… although the Maxwell ccfDNA automated extraction kit is one of the better tools currently available for cfDNA extraction [24], we cannot rule out potential bias towards short fragments.”

• The method used for cfDNA sizing is not discussed in the methods and materials section.

We thank the reviewer for pointing this out. We have added the methods for cfDNA size profiling shown below. Also, for clarity we created an additional methods subtitle “Quantification of cfDNA concentration, fragment analysis, and gene expression”.

“Fragment analysis of cfDNA size was performed using the 5200 Fragment Analyzer System, Agilent Inc using the HS Large Fragment 50kb Kit (Agilent, DNF-464-0500).”

• It would have been interesting to not only see the cfDNA size profile on one day but how it changes over the course of incubation. Perhaps the authors can speculate on what insights this may reveal.

We had difficulty obtaining an accurate fragment analysis profile at timepoints prior to day 6. This is likely because total cfDNA concentration was lower at earlier timepoints (Figure 2B). However, with more sensitive assays (ddPCR, QuantiFluor), DNA could be detected and analyzed. Still, we wished to include the results of the cfDNA fragment analysis time course and have added this as Figure 2 —figure supplement 1. We have updated the text in this section of our results to the following.

“The recovered cfDNA from mature cardiac organoids on day 9 consisted of a broad distribution of fragments ranging from approximately 200 to 6000 base pairs (Figure 2D). Fragment distributions were more difficult to resolve at timepoints prior to day 6, likely due to low DNA concentrations and sensitivity limitations (Figure 2 —figure supplement 1).”

5. Another factor that may complicate the characterization of organoid-derived cfDNA:

• Literature suggests that natural levels of apoptosis /necrosis increase over the course of organoid growth and development. How does one differentiate between cfDNA fragments originating from "natural" vs "induced" cell death? Is this a confounding factor for analysis?

We felt our best effort to account for this possibility was to compare cfDNA from treated (DOX) condition to cfDNA from the control condition, which would reflect cell death occurring as a “natural” part of organoid generation. DOX treatment could indeed impact many tissue-level processes including pathways of cell death and differentiation. However, we wished to focus this study on the ability of cardiac organoids to release cfDNA that could potentially be predictive of normal vs abnormal (DOX-treated) outcomes. Future studies of released cfDNA in response to specific modes of cell death would be valuable.

Reviewer #2:

Silver et al. demonstrated that cell-free DNA can be detected from cultures of cardiac organoids. This subject has not been explored and presents a real interest towards the possible use of organoids to detect markers according to the physiological or pathological evolution of the original tissues. So they tried to assess the possibility of cardiotoxicity of a drug from organoid cultures. The writing of the manuscript is of good quality, concise, and the presentation of the results is clear. However, the manuscript has many flaws including many inaccuracies; and authors should address the following major concerns:

1. The analysis of free mitochondrial DNA raises many questions: (i), there is no technical explanation for standardization, calibration, and direct comparison of the quantification of the number of copies of different regions of the mitochondrial genome. The quantification of mitochondrial DNA by the quantitative PCR technique must be determined with caution;

We chose to assess the presence of multiple sequences around the mitochondrial genome (Figure 3A) to determine whether cell-free mitochondrial DNA was present in our samples. Normalization to a housekeeping gene such as β-actin, TBP, or GAPDH is difficult in cfDNA. Unlike tissue-level analyses which examine expression of RNA transcripts converted to complementary DNA (cDNA), cell-free DNA (cfDNA) consists of expelled DNA fragments that are present in the extracellular media. It is currently unclear whether certain fragment sequences might be present more consistently than others. For these reasons, we chose droplet-digital PCR, which delivers more accurate counts and does not necessarily require the rigorous normalization of RT-qPCR. However, we thank the reviewer for raising the point that this could be a possible source of error and have included the following statements in our discussion.

“Unlike tissue-level analyses which examine expression of RNA transcripts converted to complementary DNA (cDNA), cell-free DNA (cfDNA) consists of expelled DNA fragments that are present in the extracellular media. It is currently unclear whether certain fragment sequences might be present more consistently than others, so it is therefore difficult to normalize ddPCR analyses of cfDNA copy number to a housekeeping gene. This is a possible source of error in determining the abundance of specific cfDNA sequences. Future studies of normalization and copy number analyses in free nucleic acids would be valuable towards precise quantitative assessments of cfDNA sequences.”

(ii) the readers may remain confused regarding the method used to determine DNA concentration;

As stated in our Materials and methods section, we determined DNA concentration using the Promega QuantiFluor dsDNA System, which employs a fluorescent double-stranded DNA-binding dye (504nmEx/531nmEm) to enable sensitive quantitation of DNA concentration.

(iii) the reviewers is confused throughout the manuscript in respect to characterize gene expression (RT-PCR) and cfDNA amount (qPCR);

We wish to clarify that we used droplet-digital PCR (ddPCR) for all gene characterizations in this study. To quantify gene expression at the tissue-level, we extracted RNA from the organoids using the Maxwell automated nucleic acid extraction system, then converted the RNA to complementary DNA (cDNA) using the iScript reverse transcription kit, then quantified cDNA using ddPCR with probes specific to genes of choice. In the case of cell-free DNA (cfDNA), we extracted free DNA from conditioned media and used ddPCR to determine copy number.

(iv), the authors state in the abstract that there is a decrease in the fraction of free mitochondrial DNA during development organoids which is not proven by the results;

We thank the reviewer for their careful assessment of our study. However, we respectfully disagree and feel the data indeed show a trending decrease of cell-free mitochondrial DNA as the organoids develop. But to avoid overstating the implications of these data, we have revised our wording in the abstract to the following.

“Intriguingly, we observed that the fraction of cell-free mitochondrial DNA appeared to decrease as the organoids developed, suggesting a possible signature of cardiac organoid maturation, or other cardiac growth-related tissue-level events.”

(v), on what basis can the authors say that this identifies a unique signature of cardiac differentiation if they did not make a comparison with other cell types;

We agree with the reviewer that it is an overstatement to say that these data definitively identify a unique signature of cardiac differentiation. Our intention was to point out the possibility that the observed pattern of decreasing cell-free mitochondrial DNA with growth day could possibly be a signature of cardiac tissue formation. However, many tissue-level events in addition to differentiation (cell death, proliferation) may contribute to cfDNA in our model. Accordingly, we have revised our wording to the following.

“Intriguingly, we observed that the fraction of cell-free mitochondrial DNA appeared to decrease as the organoids developed, suggesting a possible signature of cardiac organoid maturation, or other cardiac growth-related tissue-level events.”

(Figure 3) “Cardiac organoids exhibit a time-dependent decrease in cell-free mitochondrial DNA abundance during growth.”

(vi), the authors do not comment on why the expression of the four mitochondrial regions is equivalent, which is counterintuitive;

We wish to clarify that in our analyses of cell-free DNA (cfDNA) we are not measuring expression (RNA cDNA), rather we used ddPCR to assess the abundance of sequences detectable using probes targeting certain gene regions. As shown in Figure 3A, we are determining the presence of several mitochondrial gene sequences in cfDNA, not transcript expression of mitochondrial genes. We believe that the presence of multiple gene regions within the mitochondrial genome suggests that most of the mitochondrial genome is present in cfDNA, which is interesting as it could indicate preferential retention of cells with more mature mitochondria or increased cfDNA release from stressed mitochondria.

(vii), Figure 3C shows standard variations when it looks impossible when showing ratios; and

In this figure we wished to compare the amounts of the four mitochondrial markers we examined in cell-free DNA, to confirm whether or not they were indeed approximately equally represented in the cell-free DNA. To do this, we took the ratios of each of three markers (ND1, ND4, and ND6) to MT-COX2. If the amounts of ND1 and MT-COX2 were equal for example, the ratio would be 1. To check whether or not this was the case, we performed a one-sample t-test for each marker at each timepoint, comparing to a hypothetical value of 1. We observed that the markers were represented approximately equally in abundance at all timepoints, with respect to MT-COX2. Since there were three biological replicates performed in this experiment, standard deviations would definitely be expected. Although there is variability in these ratios, they are very similar and in most cases do not deviate significantly from a value of 1.

(viii), there is no discussion when comparing amount of cfDNA from both origins while they showed an apparent similar level, especially in regards to the literature comparing for instance cfDNA from mitochondria or nucleus.

In this study we did not make any direct comparisons between total concentrations of nuclear and mitochondrial sources of cfDNA. Care must be taken when evaluating these concentrations as the concentration (copies/µL) of one marker (i.e. NKX2.5, p53), may not necessarily be representative of the total concentration of cfDNA from genomic origin, which can be composed of many different fragments. Our goal was to determine whether cfDNA of both mitochondrial and genomic origin can be detected in cfDNA derived from cardiac organoids, which we feel we were able to successfully demonstrate. We are unfortunately unclear what literature example the author is referring to and hope that we have answered the question appropriately.

2. How is it possible that wild-type DNA, in this case P53, can show a difference in quantity by quantitative PCR analysis during differentiation?

This is what the data have shown us. We wish to clarify that in this experiment, we are not looking at gDNA, nor RNA expression but cell-free DNA (cfDNA), which is very different. CfDNA is released largely from dying cells or enclosed within extracellular vesicles and does not necessarily follow the same dynamics as DNA inside living cells. Epigenetic changes such as histone wrapping or protein-DNA interactions may protect certain regions of DNA from destruction by DNases or degradation. Therefore, changes in the abundance of recovered cfDNA fragment sequences may reflect changing events occurring in the cells from which they originated.

3. No or very few indications specified the number of passages or the number of organoids within the same culture. This is crucial and must be filled in.

This information is provided in our Materials and methods section, as shown below. If the reviewer is referring to the passages of organoids, we wish to clarify that the cardiac organoids grow continuously over the time course and are not individually passaged. In addition, we provide a statement of the number of cardiac organoids cultured per replicate (batch), shown below. To further clarify, we collected conditioned media from appx. 10 organoids per technical replicate, 2 technical replicates per batch, 3 batches total.

“Cells were passaged between 60 and 80% confluence using 0.5 mM EDTA (Gibco, 15575) onto plates pre-coated with growth factor-reduced Matrigel (Corning, 354230) for 30 min at 37oC in DMEM/F:12 (Gibco, 11330 032) at a concentration of 1.2 µL/mLcm2. Cultures were passaged a maximum of 10 times provided no morphological changes indicative of differentiation were present.”

“CfDNA from cardiac organoids was collected from media conditioned by 10 tissues (2-3 technical replicates per batch, 3 independent batches).”

Reviewer #3:

The article analyses for the first time the kinetics of different cfDNA species during the development of cardiac organoids generated from H9 human embryogenic stem cells. The stimulation of the cells with Doxorubicin or CPI, as a model of toxic exposure, is a relevant approach to identifying the validity of cfDNA as biomarkers of toxic events. The authors show an increase in cfDNA and a decrease in mtDNA during cardiac organoid differentiation. However, the research relies on a single cell line, and the number of biological replicates with n = 3 is very small. The inclusion of additional two or three replicates is required to validate the finding. Moreover, the inclusion of another selected cell line would strengthen the finding.

We thank the reviewer for raising this important discussion. Many cell culture studies conventionally use three biological replicates per experiment. However, we agree with the reviewer that many studies could benefit from more replicates, as n=3 is the minimum number needed to evaluate statistical significance, and may not resolve smaller changes between conditions. Still, there is debate in this area as some researchers feel that increasing replicate number can lead to “chasing significance” and bias analysis. Due to the time involved in organoid generation (appx. two weeks per batch) and cost, we decided beforehand to perform three separate batches (biological replicates) and evaluate the data afterwards, to limit our bias/expectations of how the cfDNA data should turn out. We wish to further clarify that at least two technical replicates were collected from each batch, using conditioned media collected from a total of appx. 10 organoids. Each cfDNA sample is then representative of the average released cfDNA from all of these organoids, not just one tissue per batch. Still, we wholeheartedly agree that these data should not be taken as a definitive statement of cardiac cfDNA markers in the body in response to DOX or other conditions. Rather, we feel all our experiments combined show the utility of cardiac organoid systems for revealing candidate cfDNA biomarkers that may be reflective of normal growth vs malformation. We feel that exploring additional cell lines and organoid systems would indeed be a valuable next step, but beyond the scope of the current study. Accordingly, we have added the following discussion.

“… our study only assessed organoids derived from one cell type (H9 ESCs) in a limited number of batches (n=3). However, we feel this work has demonstrated that cfDNA can be successfully recovered from cardiac organoids in quantities sufficient for characterization and quantification. This lays the groundwork for further studies of cfDNA from additional organoid systems in response to toxicant treatments.”

The article is well-written and the authors used various methodological approaches to prove their findings. However, with respect to the main message, namely, the increase of specific cfDNA targets in response to DOX treatment, the reviewer has some concerns. Those are related to statistical aspects as well as methodological as described below.

– In figures 1-4 the authors compare paired samples over time. However, in their figure legends, the authors indicate that an unpaired test was used. Does this have any rationale like missing samples? The single point of the samples should be included in the figure.

Although within the same batch, the samples compared on different growth days are not necessarily from exactly the same tissues on the plate. For example in Figure 1, tissue-level protein expression was examined by collecting tissues on each growth day. For consistency and unbiased sample analysis, we chose to perform an unpaired t-test with Welch’s correction. The graphs show the average of three biological replicates + SD. We have revised the figure legends to state the method of statistical analysis directly after the graph description, for clarity.

– In all figure legends it is noted that t-tests with Welch´s correction were used unless otherwise indicated. The authors could indicate if a t-test (homogenous variances between samples) or a welch´s t-test (not normal variances) was used.

We wish to clarify that all t-tests performed were unpaired t-tests with Welch’s correction. The exception was in two graphs (3B and 3C), where we wished to compare samples normalized to growth day 1 (3B) or ratios of markers (3C). Since we were comparing the significance of the samples to a hypothetical value of one, we could not use the unpaired t-test with Welch’s correction. Rather, we used a one-sample t-test comparing to a hypothetical value of 1.0.

– The authors should additionally highlight which of the results are related to cfDNA or cDNA analysis. For example, Figures 4 B and C relate to cfDNA (labelled with cfDNA). Figure 5 J relates to cDNA (not labelled). Figure 6 (not labelled with cfDNA).

We thank the reviewer for pointing out that this could be unclear. In addition to the label in the figure legend we have added an additional label “tissue-level expression” above graph 5J.

– Figure 3 B, please clarify if the relative abundance of mtDNA relates to total cfDNA or one of the reference genes.

The relative abundance of mtDNA would definitely contribute to total cfDNA concentration in addition to cfDNA from genomic origin. The same amount of cfDNA (0.5 or 1 ng) was added to each ddPCR reaction.

– Figure 4 should be restructured. Figure 4B, C, F, and G belong together, and D and E belong together. As long as F and G display cfDNA analysis (should be highlighted). Aftewards the legend text could be reduced relevantly.

We thank the reviewer for this suggestion. For clarity, we moved Figure 4 panels F and G to their own supplementary figure (Figure 4 —figure supplement 1), labeled “cfDNA”.

– Figure 6: The copy number of the different cfDNA targets relies relatively on the input material (0.5 or 1 ng were included in the ddPCR assay). Especially for Figure 6 and the discussion that ND1 and NKx2.5 increase in response to DOX treatment, but not the total amount of cfDNA and not p53. Can this result be strengthened by the analysis of GAPDH or TBP in the same sample?

We wish to clarify that the same amount of cfDNA was used in each assay (due to limited sample volume, we decreased the amount to 0.5 ng for some experiments, but ensured the same amount was used for each PCR.) Normalization of cfDNA to a housekeeping gene such as GAPDH or TBP is difficult, as released fragmented DNA sequences may not necessarily be consistently present as with studies of gene expression inside the cell. For this reason, we opted to use ddPCR which provides a more accurate count of copy number using a droplet-based approach and does not necessarily require a housekeeping gene like RT-qPCR. To address this point we have added the following discussion.

“Unlike tissue-level analyses which examine expression of RNA transcripts converted to complementary DNA (cDNA), cell-free DNA (cfDNA) consists of expelled DNA fragments that are present in the extracellular media. It is currently unclear whether certain fragment sequences might be present more consistently than others, so it is therefore difficult to normalize ddPCR analyses of cfDNA copy number to a housekeeping gene. This is a possible source of error in determining the abundance of specific cfDNA sequences. Future studies of normalization and copy number analyses in free nucleic acids would be valuable towards precise quantitative assessments of cfDNA sequences.”

– The reviewer feels that 3 more replicates are needed to validate the finding of different cfDNA release in response to DOX. Moreover, a comparison to another suitable and well-selected cell line would be welcome.

Addressed in the first response above.

https://doi.org/10.7554/eLife.83532.sa2

Article and author information

Author details

  1. Brian Silver

    1. Mechanistic Toxicology Branch, DNTP, Research Triangle Park, United States
    2. Molecular Genomics Core, DIR NIEHS, Research Triangle Park, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4480-5009
  2. Kevin Gerrish

    Molecular Genomics Core, DIR NIEHS, Research Triangle Park, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4236-5329
  3. Erik Tokar

    Mechanistic Toxicology Branch, DNTP, Research Triangle Park, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing
    For correspondence
    erik.tokar@nih.gov
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1668-2830

Funding

National Institutes of Health (ES103378-01)

  • Erik Tokar

National Institutes of Health (ES102546)

  • Kevin Gerrish

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the NIEHS Molecular Genomics Core facility, and in particular Wesley Gladwell, for generous assistance with the analysis and setup of Droplet Digital PCR experiments. Additionally, we are very thankful to Dr. Xian Wu for training in cardiac organoid culture. Also, we express our appreciation for the NIEHS Fluorescence Microscopy and Imaging Center for their assistance with fluorescence imaging of the organoids. In addition, we wish to acknowledge and thank Dr. Ian Chen for training and assistance in immunofluorescence staining and tissue-clearing protocols.

Senior Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Reviewing Editor

  1. Abel Bronkhorst, German Heart Centre, Technical University Munich, Germany

Reviewer

  1. Abel Bronkhorst

Version history

  1. Received: September 17, 2022
  2. Accepted: May 31, 2023
  3. Accepted Manuscript published: June 1, 2023 (version 1)
  4. Version of Record published: June 22, 2023 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 736
    Page views
  • 135
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Brian Silver
  2. Kevin Gerrish
  3. Erik Tokar
(2023)
Cell-free DNA as a potential biomarker of differentiation and toxicity in cardiac organoids
eLife 12:e83532.
https://doi.org/10.7554/eLife.83532

Share this article

https://doi.org/10.7554/eLife.83532

Further reading

    1. Biochemistry and Chemical Biology
    2. Developmental Biology
    Sima Stroganov, Talia Harris ... Michal Neeman
    Research Article Updated

    Background:

    Fetal growth restriction (FGR) is a pregnancy complication in which a newborn fails to achieve its growth potential, increasing the risk of perinatal morbidity and mortality. Chronic maternal gestational hypoxia, as well as placental insufficiency are associated with increased FGR incidence; however, the molecular mechanisms underlying FGR remain unknown.

    Methods:

    Pregnant mice were subjected to acute or chronic hypoxia (12.5% O2) resulting in reduced fetal weight. Placenta oxygen transport was assessed by blood oxygenation level dependent (BOLD) contrast magnetic resonance imaging (MRI). The placentae were analyzed via immunohistochemistry and in situ hybridization. Human placentae were selected from FGR and matched controls and analyzed by immunohistochemistry (IHC). Maternal and cord sera were analyzed by mass spectrometry.

    Results:

    We show that murine acute and chronic gestational hypoxia recapitulates FGR phenotype and affects placental structure and morphology. Gestational hypoxia decreased labyrinth area, increased the incidence of red blood cells (RBCs) in the labyrinth while expanding the placental spiral arteries (SpA) diameter. Hypoxic placentae exhibited higher hemoglobin-oxygen affinity compared to the control. Placental abundance of Bisphosphoglycerate mutase (BPGM) was upregulated in the syncytiotrophoblast and spiral artery trophoblast cells (SpA TGCs) in the murine gestational hypoxia groups compared to the control. Hif1α levels were higher in the acute hypoxia group compared to the control. In contrast, human FGR placentae exhibited reduced BPGM levels in the syncytiotrophoblast layer compared to placentae from healthy uncomplicated pregnancies. Levels of 2,3 BPG, the product of BPGM, were lower in cord serum of human FGR placentae compared to control. Polar expression of BPGM was found in both human and mouse placentae syncytiotrophoblast, with higher expression facing the maternal circulation. Moreover, in the murine SpA TGCs expression of BPGM was concentrated exclusively in the apical cell side, in direct proximity to the maternal circulation.

    Conclusions:

    This study suggests a possible involvement of placental BPGM in maternal-fetal oxygen transfer, and in the pathophysiology of FGR.

    Funding:

    This work was supported by the Weizmann Krenter Foundation and the Weizmann – Ichilov (Tel Aviv Sourasky Medical Center) Collaborative Grant in Biomedical Research, by the Minerva Foundation, by the ISF KillCorona grant 3777/19.

    1. Developmental Biology
    2. Neuroscience
    Simon Desiderio, Frederick Schwaller ... Frederic Marmigere
    Research Article

    Touch sensation is primarily encoded by mechanoreceptors, called low-threshold mechanoreceptors (LTMRs), with their cell bodies in the dorsal root ganglia. Because of their great diversity in terms of molecular signature, terminal endings morphology, and electrophysiological properties, mirroring the complexity of tactile experience, LTMRs are a model of choice to study the molecular cues differentially controlling neuronal diversification. While the transcriptional codes that define different LTMR subtypes have been extensively studied, the molecular players that participate in their late maturation and in particular in the striking diversity of their end-organ morphological specialization are largely unknown. Here we identified the TALE homeodomain transcription factor Meis2 as a key regulator of LTMRs target-field innervation in mice. Meis2 is specifically expressed in cutaneous LTMRs, and its expression depends on target-derived signals. While LTMRs lacking Meis2 survived and are normally specified, their end-organ innervations, electrophysiological properties, and transcriptome are differentially and markedly affected, resulting in impaired sensory-evoked behavioral responses. These data establish Meis2 as a major transcriptional regulator controlling the orderly formation of sensory neurons innervating peripheral end organs required for light touch.