Use of stable isotope-tagged thymidine and multi-isotope imaging mass spectrometry (MIMS) for quantification of human cardiomyocyte division

Abstract

Quantification of cellular proliferation in humans is important for understanding biology and responses to injury and disease. However, existing methods require administration of tracers that cannot be ethically administered in humans. We present a protocol for the direct quantification of cellular proliferation in human hearts. The protocol involves administration of non-radioactive, non-toxic stable isotope ¹⁵Nitrogen-enriched thymidine (¹⁵N-thymidine), which is incorporated into DNA during S-phase, in infants with tetralogy of Fallot, a common form of congenital heart disease. Infants with tetralogy of Fallot undergo surgical repair, which requires the removal of pieces of myocardium that would otherwise be discarded. This protocol allows for the quantification of cardiomyocyte proliferation in this discarded tissue. We quantitatively analyzed the incorporation of ¹⁵N-thymidine with multi-isotope imaging spectrometry (MIMS) at a sub-nuclear resolution, which we combined with correlative confocal microscopy to quantify formation of binucleated cardiomyocytes and cardiomyocytes with polyploid nuclei. The entire protocol spans 3–8 months, which is dependent on the timing of surgical repair, and 3–4.5 researcher days. This protocol could be adapted to study cellular proliferation in a variety of human tissues.

Introduction

Quantification of proliferating cells is critical to understanding fundamental mechanisms of tissue development, homeostasis, and response to stress or injury. The capacity for proliferation in the human heart, in particular, has attracted intense research interest due to its central role in the maintenance of systemic homeostasis.

Historically, ³H-thymidine autoradiography was the cornerstone for studies of cell proliferation; however, this method requires administration of radiolabeled substances¹. ³H-thymidine has largely been replaced by the thymidine analog 5-bromo-2’-deoxyuridine (BrdU). BrdU is incorporated into the DNA during S-phase, but distorts the DNA double helix¹. Alternative methods of quantification of proliferating cells utilize cell cycle phase-specific proteins that can be detected by antibodies: examples include topoisomerase II alpha, phosphorylated-histone H3, and Ki-67 (ref. ²). Approaches using cell cycle phase-specific proteins assess an instantaneous fraction of cardiomyocytes in the cell cycle. To convert this data into a rate of new cardiomyocytes produced over a period of time requires making assumptions about the duration of cell cycle phases and does not account for nonproductive mitotic events, i.e., karyokinesis or cytokinesis failure. Additionally, these proteins can degrade during tissue ascertainment, which can underestimate proliferation³.

Clinical methods to quantify cellular proliferation at a gross anatomical, but not cellular level, can be performed using positron emission tomography (PET) and labeling with [18F]-3′-fluoro-3′-deoxythymidine (FLT), a nucleoside-analog⁴. FLT is a poor substrate for DNA polymerase, but a good substrate for cytosolic thymidine kinase, which is up-regulated during S-phase and, as such, accumulates in the cytosol⁴. FLT-PET has been used to image pediatric brain tumors and for other oncologic applications⁵. However, FLT does not show uptake into the myocardium⁶. In conclusion, the currently available clinical tests to assess proliferation are not suitable for the study of cardiomyocyte proliferation, a relatively rare event as compared to oncologic applications.

Proliferation of the cardiomyocytes in the human heart has attracted intense research interest due to its central role in the maintenance of systemic homeostasis coupled with the heart’s limited capacity for self-repair in adults^7,8. It has been challenging to reach a consensus about the capability of mammalian cardiomyocytes to re-enter the cell cycle and divide after birth⁹. Cardiomyocyte mitotic activity in the human heart has been reported through the third trimester, in early infancy, and in young adults^10–13. These results and others suggesting that cardiomyocytes regenerate after injury in neonatal mice and rats raise the tantalizing possibility that heart regeneration might be possible in human infants and children^14–16. However, the age- and context-dependent dynamics of cardiomyocyte cell cycle activity and proliferation in humans are not completely understood.

Cardiomyocytes can be categorized as mononucleated diploid (2N), polyploid (≥2N), or multi-nucleated (two or more nuclei, which can be diploid or polyploid). These phenotypes are generated through different cell cycle mechanisms^17–19. Mononucleated diploid cardiomyocytes are generated by cell division and some of them retain capacity to re-enter the cell cycle and divide. The percentage of mononucleated cardiomyocytes with polyploid nuclei increases in humans during the first 8–15 years after birth and changes in the presence of pathological stressors^18–22. This trend must be distinguished from increases in the number of nuclei: human newborns without heart disease have approximately 30% binucleated cardiomyocytes, and this percentage does not change after birth^10,23.

We recently reported that infants with tetralogy of Fallot (ToF), a common type of congenital heart disease, have a higher proportion of multi-nucleated cardiomyocytes compared to infants without ToF (about 50–60% compared to 20%), indicating a failure of cytokinesis in this patient population¹⁸. In addition to advancing our understanding of post-natal cardiomyocyte proliferation and division, these results raise relevant biological questions with significant clinical implications: (i) how do hemodynamic changes seen in patients with ToF cause a failure of cytokinesis?, (ii) what are the implications for a decrease in cardiomyocyte endowment for adults living with ToF? and (iii) what clinical therapies could be developed to normalize cardiomyocyte cytokinesis in this patient population?

Although the proliferative potential of cardiomyocytes declines after birth, the precise age-dependent dynamics of cardiomyocyte division, including the duration of the post-natal period of cardiomyocyte proliferation, is still controversial⁹. While some studies have found that cardiomyocyte proliferation in humans stops before or at birth, others show continued proliferation in the first year after birth^23,24. Using healthy human heart tissue and immunofluorescence microscopy detection of markers of cell cycle activity, we found evidence of cardiomyocyte cell cycle activity extending through the first 10–20 years of life¹⁰. We also found a stable percentage of mononucleated cardiomyocytes (60%) and an increase in cardiomyocyte nuclear ploidy with age: at less than one year of age 16% of cardiomyocytes had increased nuclear ploidy, which increased to 40% between the ages of 10 and 20 years¹⁰.

We have identified that cardiomyocyte proliferation is impacted by disease state, as there is increased cytokinesis failure in infants with ToF, as identified by an increase in binucleation^16,18. Therefore, new techniques and approaches are needed to dissect the dynamics of cardiomyocyte proliferation as a function of developmental age, disease state, and in response to therapeutic interventions.

In this protocol we present the use of oral ¹⁵N-thymidine in infants with ToF to label cardiomyocytes that undergo S-phase, and couple it with a correlative confocal approach to determine nuclear ploidy. This paired approach allows the quantification of DNA synthesis (¹⁵N-thymidine positive cells), cytokinesis failure (binucleation), and polyploidy (assessed by DNA content in nuclei), which allows for a comprehensive analysis of the dynamics of cardiomyocyte division, karyokinesis, and cytokinesis failure on a cellular level. Tissue resected at the time of surgical repair is analyzed with multi-isotope imaging mass spectrometry (MIMS) to reveal label retention and confocal microscopy to quantify DNA content. The ¹⁵N incorporated into DNA is a stable marker of newly-replicated DNA. Due to the low natural abundance of ¹⁵N, it has a high signal-to-noise ratio²⁵. As the timing of surgical repair is a clinical decision, there is variability in the chase period. Despite this variability, we have identified ¹⁵N-positive cardiomyocytes with chase periods ranging from 2 weeks to 6 months. This is the first approach to directly quantifying cardiomyocyte cell cycle activity in humans.

Application of the method

This protocol can be used to quantify cell cycle activity in any organ from which sufficient tissue can be acquired. We use tissue specimens that are 2–4 mm in diameter. Using tissue samples that are removed as part of clinical care reduces potential risk to the subject. In the human heart, for example, myocardial specimens can be obtained during surgery as part of congenital heart disease repair, at the time of ventricular assist device placement, or heart transplantation. Tissue samples are also obtained at the time of myocardial biopsy as part of rejection surveillance; however, biopsy is not without associated risk of myocardial injury or tricuspid valve damage. In the hematopoietic system, MIMS can be used to determine cellular proliferation by analyzing peripheral blood samples (Fig. 1) or bone marrow aspirates. In oncologic applications, samples can be obtained from resected tumors and lymph nodes²⁶. Skin and muscle samples can be obtained through biopsy. Bowel biopsies are part of the clinical care for patients with inflammatory bowel disease or as part of cancer screening, and this tissue can be used to study crypt cell proliferation²⁷.

Fig. 1 |. MIMS analysis demonstrates proliferation of white blood cells (WBCs).

¹⁵N-thymidine (50 mg/kg/d) was administered orally over 5 d to an infant with tetralogy of Fallot at the age of 3.5 weeks. Peripheral blood was drawn at the age of 7 months. The ¹²C¹⁴N, ³¹P, and ³²S images demonstrate cellular morphology. The hue saturation intensity (HSI) image maps the ratio of ¹⁵N/¹⁴N. The rainbow scale is set from 0.37% (expressed as 0% above natural ratio) to red, where the ratio is twofold above natural ratio (expressed as 100%). White arrows indicate cells that are ¹⁵N-labeled and underwent S-phase during the period of ¹⁵N-thymidine administration. Scale bar, 10 μm.

Depending on the biological question of interest and the proliferative potential of the cell type of interest, there are multiple ways in which this protocol could be optimized. The duration of labeling can be reduced to a short pulse (e.g., single dose) for cells with high proliferative rates or extended for very slow turnover tissues (e.g., weeks to months). While the current protocol involves a single label (¹⁵N-thymidine), the NanoSIMS instrument can be configured to measure several different stable isotope labels in parallel. This allows analysis of different cell cycle labels (e.g., ²H-, ¹³C-, ¹⁸O-thymidine) administered at different time-points prior to tissue acquisition, providing a more dynamic assessment of proliferation events. Alternatively, multiplexed stable isotope imaging can be used to couple cell cycle labeling with other metabolic labels. For example, in murine atherosclerosis, heterogeneous glucose avidity (²H-glucose) was quantified as a function of smooth muscle and foam cell proliferation, establishing a link between proliferation and metabolic function²⁸. Because the application of MIMS to biology is relatively new, applications might require iterative testing prior to broad human deployment. In some instances, labeling protocols can be initially tested in vitro (Fig. 2) or in animal models^7,27. Mouse models are well established, and, due to their small size, pilot studies will not require large quantities of the costly stable isotope label.

Fig. 2 |. MIMS analysis using ex vivo ¹⁵N-thymidine labeling of human fetal myocardium.

Human fetal myocardium was collected at 18 weeks gestation and cultured in the presence of 20 μM ¹⁵N-thymidine. Media was changed every 3 d. After 5 d, the myocardium was fixed and analyzed by MIMS. a, Mosaic ¹⁵N/¹⁴N HSI ratio image in which 50 × 50 μm imaging fields are tiled together. The rainbow scale is set from 0.37% (expressed as 0% above natural ratio) to red, where the ratio is fourfold above natural ratio (expressed as 300%). b, The region contained in the white square was reanalyzed at higher resolution with a 25 × 25 μm imaging field, demonstrating a cluster of ¹⁵N-labeled nuclei. Analysis of 500 cardiomyocytes demonstrated 16% of cardiomyocytes in the sample were ¹⁵N-positive. For both (a) and (b): scale bar = 10 μm.

Another strength of the MIMS approach is the ability to produce high-resolution quantitative mass images at a lateral resolution of down to ~30 nm. As such, a range of subcellular structures are identifiable, such as stereocilia, nucleoli, mitochondria, lysosomes, lipid droplets, and intracellular granules^27,29–32. When imaging of subcellular structures is coupled with the quantification of stable isotope tracers, it is possible to probe cell biology questions related to the cell cycle and a range of metabolic processes in humans at a spatial resolution unattainable by other methods.

MIMS can also be used together with orthogonal imaging modalities. In the current application, we paired MIMS with correlating confocal microscopy of the same cardiomyocyte to determine binucleation and nuclear ploidy. These correlative light microscopy techniques are generally accessible to any laboratory. Depending on the nature of the specific cell biological question, the very thin sections used for MIMS analysis (500 nm), relative to more standard histology, leave abundant vertical depth of the tissue specimen for additional analyses, whether with light microscopy or other modalities. There is precedent for correlative analysis of MIMS with fluorescence microscopy. MIMS analysis was performed in MerCreMer/ZEG mice, an inducible cardiomyocyte-specific expression GFP system, and confirmed that new cardiomyocytes are derived from pre-existing cardiomyocytes⁷. In murine hepatocytes, MIMS and immunofluorescence were combined to determine the impact of YAP oncogene overexpression, visualized with tdTomato, on glucose and glutamine metabolism²⁶. The combination of MIMS with additional imaging techniques would allow researchers to refine their experimental design to match their hypothesis of interest.

Development of the protocol

While the highest profile early applications of the NanoSIMS instrument were largely in non-biological fields (including cosmochemistry, where it was utilized to quantify isotopic compositions of extraterrestrial samples^32,33), it was clear that the technique could be applied to basic research in biological and clinical laboratories. Central to MIMS is a nano-scale secondary ion mass spectrometer (NanoSIMS, CAMECA)^34–36. Using sophisticated ion optics, a standard NanoSIMS instrument probes a sample surface with a primary beam of cesium ions, resulting in the sputtering and ionization of atoms and small polyatomic clusters at the sample surface. The ions that are released from the surface during this process are referred to as secondary ions. An immersion lens collects these ions, which are shaped by ion optics into a secondary ion beam that is directed to a magnetic sector mass spectrometer. Each detector within the mass spectrometer measures the signal from ions of one mass:charge ratio (m/q). In the most recent model (50L), seven detectors are aligned to capture individual ions with seven different m/q. Quantitative mass images are generated, and the intensity of each pixel reflects the underlying ion counts from each of the detectors. The NanoSIMS can be configured as a nano-scale isotope ratio mass spectrometer by tuning two of the detectors to capture two different isotopes of the same element in parallel. When this capability is merged with stable isotope labeling, the NanoSIMS can quantitatively track labels in sub-micron domains (maximum lateral resolution ~30 nm). For example, a molecule enriched in the rare stable isotopic variant of nitrogen (¹⁵N) can be quantified by an increase in the ¹⁵N/¹⁴N ratio above the naturally occurring background ratio of 0.0037.

For both masses, 26 Dalton (Da, ¹²C¹⁴N) and 27 Da (¹²C¹⁵N), there can be other molecular species that must be resolved from the CN molecules in order to obtain accurate data. At mass 26 Da, there can be peaks at ¹³C₂, ¹⁰B¹⁶O, and ¹²C¹³CH. At mass 27 Da, there can be interfering peaks at ¹¹B¹⁶O and ¹³C ¹⁴N. Depending on the amount of ¹⁵N label given (e.g., very high enrichments), these interferences may be small enough as to not significantly impact the accuracy of the data. Note, however, that even though the samples themselves may not contain much boron (B), typical lab glassware contains abundant borosilicates, which can leach into any chemicals used in the sample preparation process and in turn any samples they are applied to; hence, boron oxide contaminants can appear in the tissue.

Using the NanoSIMS 50 prototype instrument housed at the Brigham and Women’s Hospital, Harvard Medical School, the laboratory of Claude Lechene performed early proof-of-concept experiments involving cellular imaging, including measurement of stable isotope tracers in bacterial and mammalian cells^35,37,38. Subsequently, the Lechene laboratory and others demonstrated the power of measuring stable isotope tracers with NanoSIMS in diverse applications, including measurement of nitrogen fixation by individual bacteria^39,40, dopamine distribution within individual vesicles⁴¹, protein turnover in hair cell stereocilia⁴², lipid movement across capillaries³¹, nuclear pore turnover⁴³, and metabolism of atherosclerotic plaques²⁸, among other applications.

The application of MIMS to measure DNA synthesis and cell cycle activity was first demonstrated through direct measurement of halogenated nucleotide analogues³⁵, then extended to stable isotope nucleotide labels with studies of murine stem cell replication in the small intestine²⁷, cardiomyocyte generation^7,44, adipogenesis⁴⁵, and atherosclerotic plaque growth²⁸. The introduction of stable isotope labels to track cell cycle activity was an essential prequel to the application of MIMS to human studies, not just because of the high measurement fidelity gained by tracking labels via their corresponding isotopic ratios, but because of the innocuous nature of stable isotopes and their already extensive use in studies involving even the most vulnerable human populations, such as pregnant women and critically ill pediatric populations²⁵.

The first protocols involving MIMS in humans involved the measurement of cell cycle activity in leukocytes and adipose tissue by the incorporation of ¹⁵N-thymidine administered via continuous intravenous infusion^27,46. In order to develop an approach that would be more practical for studies of human infants with congenital heart disease (CHD), including patients who are not hospitalized, we adapted protocols for parenteral administration of ¹⁵N-thymidine in mice and adults²⁷ to oral administration in patients with ToF¹⁸. This approach achieved easily-detectable labeling in cycling cardiomyocytes using NanoSIMS analytical parameters previously developed for the measurement of ¹⁵N-enriched tracers.

Selection of stable isotope tracer, dosing, and compounding

In order to safely perform quantitative assessment of cell proliferation in infants, instead of a radioactive isotope we selected an enriched stable isotope. Enriched stable isotope tracers are nontoxic and can be used in experiments involving living organisms without any toxicological concerns⁴⁷. There is an extensive precedent for the utilization of stable isotopes as molecular tracers where one or more of the atoms in the parent molecule is enriched with the rare stable isotope^25,48. Nitrogen exists in two stable isotopes, the vast majority (99.63%) as ¹⁴N, and the remaining as the rare ¹⁵N isotope. Administration of ¹⁵N-thymidine has a favorable signal-to-noise ratio once incorporated into DNA due to the low natural abundance of ¹⁵N.

Since the introduction of ¹⁵N-labeled amino acids in research by Schoenheimer in the late 1930s, there have been over 95 human studies encompassing 1,185 study subjects utilizing ¹⁵N (ref. ^25,49). Stable nitrogen isotopes have been used to study catabolism and nitrogen turnover in infants as early as 1969 (ref. ⁵⁰). Notably, the molecular structure of the parent molecule is not altered, nor does enrichment of a molecule with a rare stable isotope introduce toxicity, unlike radioactive isotopes. The principal national regulatory agency in the United States, the Federal Drug Administration (FDA), does not require specific oversight or an investigational new drug (IND) application for administration of stable isotopes in humans as long as the proposed stable isotope labeling protocol meets specific conditions, including (i) the research aim is to obtain basic information regarding tracer metabolism related to human physiology, pathophysiology, or biochemistry; (ii) the research is not intended for immediate therapeutic, diagnostic, or preventive application to the study subject; (iii) the dose is not known to have a pharmacological effect in humans; (iv) relevant quality standards are met; (v) institutional review board (IRB) or ethical review board approval is obtained; and (vi) informed consent regulations are followed. While regulatory requirements differ in other countries, the guidance from the FDA provides a framework with which to consider utilization of stable isotopes for human studies while underscoring their general innocuous nature.

In order to track DNA replication in cardiomyocytes of pediatric patients with ToF, we selected thymidine labeled with ¹⁵N, similar to prior murine heart studies and the initial human MIMS work with peripheral leukocytes and adipose tissue^7,27. Thymidine is synthesized endogenously by de novo nucleotide synthesis or recycled by nucleotide salvage pathways. It is a natural constituent of cells and found in the systemic circulation in micromolar concentrations⁵¹.

Thymidine is considered safe for administration in infants and does not require approval by the FDA in the US for its use at concentrations used in this study. Thymidine was also investigated as a rescue agent for methotrexate toxicity, where intravenous doses of approximately one order of magnitude higher than our protocol were well tolerated⁵².

The first human MIMS studies in adults involved continuous intravenous administration of ¹⁵N-thymidine with total doses of 750 mg and 1,095 mg given over 2 and 3 d, respectively^27,46. In these studies, ¹⁵N-thymidine incorporation into DNA was used to identify proliferation of white blood cells (WBCs) and adipocytes. Preclinical experience in mice demonstrated effective labeling of dividing cells with systemically administered doses down to as low as 0.1 mg/kg and the initial human experience demonstrated effective labeling with a daily intravenous dose of approximately 5–10 mg/kg; however, we selected a higher dose of 50 mg/kg/d for oral administration to infants, assuming that oral bioavailability would be less than 100%. In an infant with ToF who received 50 mg/kg/d ¹⁵N-thymidine for 5 d at 1 month of age, 12% of cardiomyocytes were ¹⁵N-positive¹⁸. In a 4.5-month-old infant who also received 50 mg/kg/d ¹⁵N-thymidine for 5 d, 3.3% of cardiomyocytes were ¹⁵N-positive (unpublished findings from the authors of this protocol). In summary, our findings in these two patients demonstrates that this labeling protocol is effective.

Although potential safety concerns are not related to the parent molecule (thymidine) or its enrichment with stable isotope (¹⁵N), we have considered the potential risk of unintended pathogen introduction as part of the manufacturing process. While the method of production of ¹⁵N-thymidine is proprietary information of Cambridge Isotope Laboratories (CIL), we obtained a product that has been tested for endotoxin and microbial contamination, including Staphylococcus aureus, Pseudomonas aeruginosa, Escherichia coli, Salmonella, aerobic bacteria, yeast, and mold. The product was packaged and sealed in sterile vials in a laminar flow hood. Because thymidine is known to be highly stable, it can be stored at room temperature with standard attention to minimizing light and extreme humidity. Prior to administration, ¹⁵N-thymidine is reconstituted under a laminar flow hood by a research pharmacist. Thymidine is fully soluble in water up to a maximum concentration of ~20 mg/mL.

Design of an ethically appropriate human ¹⁵N-thymidine labeling and myocardial sampling protocol

Before initiating a human research study, the protocol must be approved by the IRB or equivalent body. Local regulatory guidelines may vary; however, most adhere to principles outlined in the declaration of Helsinki, aiming to ensure that the risks to human research subjects are minimized and are reasonable in relation to the anticipated benefit. The application of MIMS to quantify post-natal cardiomyocyte generation is an example of a study with no anticipated direct benefit to the research subject and, as such, there is a particularly strong necessity to minimize risk. This is compounded by application to a pediatric study population, as minors are considered a vulnerable population with heightened requirements to minimize harm.

In order to quantify cardiomyocyte proliferation with MIMS, a sample of myocardium must be obtained. Sampling myocardium is not routine, nor can it be considered minimal risk. Endomyocardial biopsies confer risks, including vessel perforation, effusion, and tricuspid valve damage. Therefore, for pediatric studies in particular, the only ethically feasible MIMS protocol involves obtaining discarded tissue samples whose resection is part of usual clinical care. Patients with CHD offer an opportunity for stable isotope labeling and MIMS examination of the myocardium because (i) many patients undergo open heart surgery, (ii) in most cases the surgery is planned weeks to months in advance and (iii) portions of the myocardium are often removed and discarded as part of the surgery.

ToF is one of the most common forms of CHD, which manifests with a combination of four defects: anterior deviation of the infundibulum, pulmonary stenosis (PS), ventricular septal defect (VSD), and right ventricular hypertrophy. Patients with ToF usually undergo heart repair within the first year of life with a surgical procedure to enlarge the right ventricular outflow tract (RVOT). Enlargement of the RVOT typically involves resection of pieces of right ventricular (RV) myocardium, which would otherwise be discarded. Our analysis of these specimens has demonstrated that patients with ToF have abnormal cardiomyocyte proliferation^16,18. Therefore, ToF patients represent an ideal study population for investigating abnormalities of cardiomyocyte proliferation by stable isotope labeling and MIMS analysis. Aside from minimal risks more generally applicable to clinical research, such as privacy concerns related to the accessing of protected patient information, any procedural risks are attributable to usual clinical care, and additional study-related risks are minimal.

Sample preparation for MIMS analysis

An often-underappreciated aspect of any MIMS experiment is sample processing, which can include application-specific nuances. The current protocol reflects the authors’ cumulative experience measuring ¹⁵N-thymidine in a variety of contexts, including early work conducted on the murine heart⁷. For applications like ours, when correlative histological analyses are anticipated, deciding how to fix the tissue is important. We recommend paraformaldehyde (4% vol/vol), as opposed to other aldehyde and osmium tetroxide-based fixation protocols traditionally used for transmission electron microscopy (TEM), which limit orthogonal histological analyses. Samples are then embedded in plastic resin, as used for TEM. Similar to prior work, we selected a relatively porous and hydrophobic resin (LR white), as opposed to more commonly used epoxies, to facilitate orthogonal analyses of adjacent sections. Samples are sectioned to 0.5 μm, a thickness that allows for alignment of MIMS images with those obtained from adjacent sections with complementary modalities (e.g., fluorescence microscopy). Sections are mounted for MIMS analysis on custom-cut and processed silicon wafers. In contrast to the early experience measuring ¹⁵N-thymidine, we now routinely apply a thin coating of gold to the sample surface, which increases ionization. Any improvement in the ionization yield translates into a gain in throughput, an aspect of particular relevance to analyzing very large areas (swaths) of myocardial tissue sections.

MIMS and correlative histological analyses of tetralogy of Fallot myocardium to quantify cell cycle activity

NanoSIMS analysis of tissue sections to track ¹⁵N-tagged molecules such as thymidine is well established^27,35,45,46. In the current protocol, we applied analytical methods that were refined at the Brigham and Women’s Hospital Center for NanoImaging (formerly the National Resource for Imaging Mass Spectrometry) to assess cardiomyocyte proliferation in infants¹⁸. These protocols are similar to the approach taken by NanoSIMS facilities around the world^53–56. Similar to analysis using other ¹⁵N-tagged tracers, ¹⁵N-thymidine accumulation is measured by tuning two parallel detectors to simultaneously quantify ¹²C¹⁵N and ¹²C¹⁴N isotopes. CN ions are used to capture the ¹⁵N/¹⁴N isotope ratio because native nitrogen has a negligible ionization efficiency. The images derived from ¹²C¹⁴N ions alone reveal histological details, including cell and nuclear borders. Additional subcellular features, such as nucleoli and sarcomeric structures, might also be evident. Further histological details are resolvable with parallel ion images. Of particular relevance to the labeling of replicating genomic DNA, ³¹P images reveal nuclei due to the high phosphorus content of chromatin and ¹⁵N-thymidine labeling correlates with the ³¹P signal⁴⁶. ³²S images reveal sulfur-rich subcellular structures, such as sarcomeres; however, in some instances they also provide even better histological delineation than CN images³⁶. Importantly, each of these images are obtained using ions that are collected at the same time as those required for isotopic ratio measurement. As such, the images used for histological delineation are inherently colocalized with the quantitative isotopic ratio images. It is this instrumental capability that effectively enables high-precision measurements of stable isotope labels in sub-organelle domains.

The translation of a MIMS measurement of DNA replication into a measure of cardiomyocyte proliferation is especially challenging because cardiomyocytes are frequently polyploid and multi-nucleated. Although MIMS provides high-fidelity quantification of cell cycle activity, in isolation it cannot distinguish genomic replication due to polyploidization, multinucleation, or a complete cycle of cell division. Therefore, multimodal interrogation of adjacent sections to characterize ploidy and nucleation is critical for any assessment of cardiomyocytes. To address this concern, we have developed a complementary staining protocol to determine nuclear ploidy status in sections adjacent to those used for MIMS analysis.

Comparison with other methods

The extent of cardiomyocyte proliferation after birth and the ability to induce proliferation as a potential strategy to achieve heart regeneration has been a highly researched topic for over 100 years^24,57–59. In 1960, Linzbach developed a quantitative growth model of the human heart by analyzing cardiomyocyte cross-sectional areas on H&E-stained perpendicular sections of the papillary muscle²⁴. He concluded that cardiomyocyte enlargement would be sufficient to explain post-natal heart growth in humans and predicted that cardiomyocyte proliferation approaches zero before or at birth. This approach required biased sampling and did not directly measure proliferation.

Histological staining of cell cycle markers is a classical approach to quantifying cardiomyocyte cell cycle activity. Mollova et al. identified cycling cardiomyocytes using an antibody against phosphorylated histone H3 (a marker for M-phase)¹⁰. Cardiomyocytes were identified with antibodies specific to sarcomeric proteins and visualized with confocal microscopy¹⁰. Counts of cycling cardiomyocytes were verified with cytometry¹⁰. This study demonstrated that the highest cell cycle activity existed in infants <1 year old and that the frequency decreases with age. However, this methodology does not directly determine karyokinesis or cytokinesis failure, which would result in polyploidy and binucleation, respectively. In addition, converting an instantaneous fraction of M-phase cardiomyocytes into a rate per unit of time requires assumptions including duration of M-phase and corrections for nonproductive mitotic events.

The Frisen group developed models of human cardiomyocyte turnover from carbon-dating of homogenized cardiomyocyte nuclei⁶⁰. In their growth model, cardiomyocyte generation is highest in infants and declines to very low levels in adults⁶⁰. They concluded that existing cardiomyocytes are replaced in growing hearts, that most of this replacement occurred within the first years of life, and that the number of cardiomyocytes did not change²³. Carbon dating of cells was possible due to an atmospheric pulse of carbon-14 in the atmosphere from above-ground nuclear bomb testing and the subsequent chase that ensued with the nuclear test ban treaty. Cardiomyocyte birth dates were calculated from averages of ¹⁴C-measurements from millions of nuclei and inputted into a dynamic population model⁶⁰. Therefore, cell cycle activity was not directly measured, but inferred from the turnover model. This approach also does not directly determine karyokinesis or cytokinesis failure. Additionally, the carbon-14 birth-dating method cannot be easily applied to prospective human studies, as carbon-14 exposure is not a controllable variable.

When the approach of ¹⁵N-thymidine labeling with MIMS is considered in the context of previously established methods, several advantages are evident. Unlike cytometric modeling or immunoprobing of cell cycle markers, the introduction of a prospective and temporally controllable cell cycle label (¹⁵N-thymidine) introduces the unit of time into the readout. In addition, the measurement of ¹⁵N-thymidine labeling is highly precise over a dynamic range of 2–3 orders of magnitude and not encumbered by artifacts due to high background, as can occur with immunostaining²⁷. Although all methods of studying myocardium at the cellular level are limited by the need for tissue acquisition, including MIMS and carbon-14 birth-dating methods, MIMS enables safe, prospective studies of cardiomyocyte cell cycle activity in humans and is therefore uniquely positioned for the assessment of experimental interventions aimed at modifying cell cycle activity.

Experimental design

An illustration of the experimental design is outlined in Fig. 3. The protocol spans the consent of potential research patients, administration of ¹⁵N-thymidine, surgical tissue resection, tissue processing, MIMS, and subsequent data analysis.

Fig. 3 |. Flowchart of the presented protocol to determine cardiomyocyte proliferation and formation of bi- and multinucleated cardiomyocytes and polyploid nuclei.

Infants with tetralogy of Fallot are given ¹⁵N-thymidine by mouth. At the time of surgical repair, resected right ventricle myocardium is collected. The tissue is fixed, embedded, and sectioned. Sections are mounted on silicon chips for analysis with MIMS and on glass slides for confocal microscopy analysis of number of nuclei and DNA content (ploidy analysis). Then, using OpenMIMS (a plug in for ImageJ/Fiji), individual nuclei are identified (regions of interest, ROI) and the incorporation of ¹⁵N-thymidine is quantified. In sections mounted on glass slides, nuclei are stained using Hoechst, photographed, and aligned with the MIMS images to determine bi- and multinucleation and nuclear ploidy.

Regulatory approval

Our study was approved by the University of Pittsburgh School of Medicine IRB with consideration for this particularly vulnerable pediatric study population (STUDY19030250). The processes for regulatory approval will differ according to the institution in which the protocol will be conducted. For example, scientific or statistical review might be required prior to consideration by the IRB, and research pharmacies might have different practices for the formulation of stable isotopes. It is important to check that there are proper quality control measures throughout the supply chain of ¹⁵N-thymidine, i.e., from preparation to ingestion.

Patient recruitment

We identify study patients who are likely to have RV myocardium removed as part of their clinically indicated surgical repair from lists of fetal cardiology clinics, pediatric cardiology clinics, or inpatient settings. We screen for patients with ToF or double outlet right ventricle, tetralogy type, over 35 weeks gestational age and over 2 kg at the time of enrollment. A clinician introduces the study to the patient’s parents. If parents express an interest in learning more about the study, a member of the research team subsequently contacts the family, in person or via phone. We provide the parent(s) with a brochure and a copy of the consent form, and then schedule a follow-up meeting to answer additional questions and obtain written consent. As part of the consent process, thymidine administration and urine collection are reviewed. Details regarding tissue and blood collection at the time of surgery are also reviewed. We emphasize that the extent and timing of myocardial tissue resection is entirely at the clinical discretion of the medical and surgical team; it is independent of the parent(s)’ decision to enroll their child in this study.

Isotope label administration and monitoring

¹⁵N-thymidine is dissolved in sterile water without addition of flavor and dispensed by the research pharmacy. We administer five daily oral doses of ¹⁵N-thymidine (50 mg/kg/d). In order to verify ¹⁵N-thymidine delivery and absorption, we request that the patient’s caregiver collect urine by placing cotton balls in every diaper from the time of first dose until 24 h after the last dose. This circumvents the need for obtaining blood samples to confirm label uptake. Families maintain a dosing log to track timing of dosing, diaper changes, and any additional relevant information, such as emesis after ingestion of a dose.

Tissue collection and processing

The timing of surgical repair of ToF is dictated by each patient’s specific clinical course; however, in most instances of ToF, the surgical repair is a planned procedure. The research team alerts the surgical team of the patient’s participation in the study prior to the surgery. Intraoperatively, if RV myocardium is removed, the research team is contacted to retrieve the specimen and additional discarded tissue specimens that can serve as labeling controls (approximately 3 mL of blood) from the operating room. Deidentified samples are then processed for fixation, embedded in LR white, sectioned, and mounted on glass slides for histological assessment and silicon wafers for MIMS analysis. Figure 4 shows representative images of resected myocardial pieces (Fig. 4a), sections on a silicon chip for MIMS analysis (Fig. 4b), and sections on a glass slide (Fig. 4c). Importantly, the protocols for embedding and sectioning specimens for MIMS analysis are similar to those used for preparation of samples for TEM. Therefore, TEM core facilities typically have the requisite equipment and expertise to prepare the samples. The key deviations from standard TEM include the thickness of the sections (500 nm) and their mounting on silicon chips rather than TEM grids.

Fig. 4 |. Images of myocardial specimens and sections to highlight sample processing.

a, Specimens of freshly resected myocardium in a sample container are kept on ice. The size of the pieces can be estimated by using the diameter of the container (5 cm) as reference. The tissue pieces will be cut into cubes of 1–8 mm³ size before processing. b, After fixation and embedding, 500 nm sections are prepared. One tissue section adhered to a silicon chip in a capsule ready for shipping to the MIMS center is shown. The section is visualized by Toluidine blue staining. c, For every section placed on a silicon chip (b), 10 sections are placed on glass slides for later analysis by confocal microscopy (c). Scale bar, 100 μm.

MIMS/NanoSIMS analysis

We conduct MIMS analyses at the Center for NanoImaging, Brigham and Women’s Hospital, Boston; however, additional centers are available worldwide (https://www.cameca.com/products/sims/nanosims). Deidentified myocardial specimens are embedded in resin, sectioned, and mounted on silicon wafers. The silicon wafers can be shipped at room temperature without special handling. This facilitates ease of collaboration when the site of patient recruitment and surgical repair is geographically distant from the NanoSIMS instrument. We utilize a commercial NanoSIMS 50L instrument for MIMS analysis, which is configured with seven detectors. However, for the purposes of this protocol, only four detectors, as contained in the original NanoSIMS 50 prototype, are needed for the measurement of ³¹P, ³²S, ¹²C¹⁴N, and ¹²C¹⁵N (Fig. 5). The NanoSIMS instrument contains a charged-coupled device (CCD) camera that enables visualization of the sample holder in the analysis chamber; however, analytical deadtime can be reduced by off-line imaging and documentation of coordinates of interest to guide NanoSIMS analyses⁶¹. Therefore, after the silicon chips containing the myocardial sections have been mounted on a sample holder, but prior to their introduction into the NanoSIMS, we use a differential interference contrast (DIC) microscope to identify myocardial regions for analysis, record their coordinates, and capture images. While such imaging and documentation is not absolutely critical, in the case of myocardial sections where the goal is to image swaths of tissue in automated chain analysis mode, doing so directs the NanoSIMS analysis to tissue regions with clearly discernable cardiomyocytes, thereby limiting analysis of dead space. In the absence of a DIC microscope, other imaging modalities (e.g., scanning electron microscopy) can be applied to the section⁶². Alternatively, images captured from an adjacent section mounted on a glass slide and stained with Toluidine blue can also be used to guide analyses.

Fig. 5 |. MIMS.

Schematic depicting principles of the NanoSIMS instrument, which is central to MIMS. The surface of a sample is sputtered with a primary cesium ion beam. A fraction of the sputtered surface atoms and small polyatoms are ionized. These secondary ions are extracted by an immersion objective and directed through a series of ionic lenses and slits, shaping and focusing the secondary ion beam in a plane. A double sector mass spectrometer (“magnet”) separates ions by mass in the focusing plane. A series of seven moveable detectors can be aligned to capture data for seven different masses (Unit: Da) from the same sputtered surface material. A quantitative mass image is derived from the counts of each mass. For the purposes of this protocol, only four of the detectors were used. ¹²C¹⁴N, ³¹P, and ³²S images provide histological detail (four black and white images, bottom left). ¹²C¹⁵N and ¹²C¹⁴N measurements are used for the ¹⁵N/¹⁴N isotope ratio. The isotope ratio can be visually displayed using a hue saturation intensity (HSI) transformation. The bottom right image shows the HSI image from ¹²C¹⁵N and ¹²C¹⁴N measurements shown in the schematic. In this case, the lower bound of the scale (blue) is set at the naturally occurring ¹⁵N/¹⁴N ratio of 0.37%. The upper bound of the scale is set to demonstrate regional differences in enrichment, such as the ¹⁵N-labeled nucleus (white arrows). In this case, the upper bound is set to 0.74% (or 100% above natural background). Importantly, any changes to the scale will affect the color pattern and thus the visual interpretation of the data but the underlying quantitative data remains unmodified. Scale bar, 10 μm.

We first analyze myocardial sections in automated chain analysis mode, where numerous adjacent fields are sequentially imaged to capture much larger tissue swaths than would be contained in a single 50 × 50 μm imaging field. This approach also enables running the instrument during off hours (e.g., nights and weekends). Of note, NanoSIMS is similar to many other imaging modalities in that operating at the highest possible spatial resolution is more time intensive. Therefore, we routinely sacrifice imaging resolution with our first pass chain-imaging to gain throughput. Even though each section is typically 500 nm thick, our protocol for first-pass imaging sputters away only the uppermost fractions of the sample surface, leaving an abundant residual sample for additional higher resolution analysis when needed. Repeat imaging of selected cells can be performed at higher lateral resolution in the event that some feature is obscured with the first pass analysis. One common example where this can be useful is if there is imperfect movement of the stage and adjacent tiles are not completely aligned.

NanoSIMS data files can be viewed and analyzed using different software platforms, including WinImage (CAMECA), L’Image (L. R. Nittler, Carnegie Institution of Washington), or OpenMIMS. WinImage and L’Image are proprietary software, whereas OpenMIMS is a freely available plugin to ImageJ/Fiji, developed and maintained at Brigham and Women’s Hospital/Harvard Medical School, and therefore the platform used in our protocol (https://nano.bwh.harvard.edu/openmims). Irrespective of the software used to visualize the datafiles or the exact visual scale applied to the data, the underlying measurements that form the basis for the images are not affected. For biologists who are new to this type of analysis, OpenMIMS can be the most accessible because it is the only platform that is freely available, it was specifically developed for biological applications, and many biologists are already familiar with the ImageJ/Fiji software.

Identification of ¹⁵N-thymidine-labeled cardiomyocyte nuclei

The NanoSIMS data files are opened and analyzed using the OpenMIMS plugin for ImageJ (Fig. 6). The ³¹P images allow for easy tracing and identification of all nuclei as regions of interest (ROI, Fig. 6a). The resolution of MIMS allows for the identification of cardiomyocyte borders and cellular organelles⁷. Cardiomyocyte nuclei are roughly circular or elongated⁶³ and defined as nuclei present within sarcomere-containing cells. Using the ¹²C¹⁴N and/or ³²S images, each ROI (corresponding to a nucleus) should be categorized as a cardiomyocyte, non-cardiomyocyte, or undetermined (Fig. 6b). If the cell type is not easily identified on the initial analysis, targeted repeat imaging at higher x/y resolution may resolve cellular identity. This might include analysis of ³²S images to highlight sarcomeres (Fig. 7). Indeterminate cells are excluded from the analysis.

Fig. 6 |. Integration of ³¹P, ¹²C¹⁴N, and ¹²C¹⁵N isotope signals to identify DNA synthesis in cardiomyocyte nuclei.

Mosaic (.nrrd) images obtained from MIMS analysis are shown as depicted when using the openMIMS plug-in for ImageJ/Fiji. Regions highlighted by a dashed yellow rectangle are shown at a higher magnification to the right of each panel. a, All nuclei on the ³¹P MIMS image are identified and outined in red by the user to define regions of interest (ROI). b, ROIs are categorized by the user as cardiomyocyte nuclei, non-cardiomyocyte nuclei, or indeterminate nuclei using the ¹²C¹⁴N images (and/or ³²S images, not shown). Far right panel demonstrates the appearance of sarcomeres, which are the defining structural feature of cardiomyocytes. Reanalysis of indeterminant nuclei at higher resolution can clarify cell type. c, The quantitative isotope ratio measurements, which are visually represented by the hue saturation image (HSI), can be extracted for each ROI. Those ROI that demonstrate a ¹⁵N/¹⁴N ratio above natural background indicate nuclei that underwent DNA replication during ¹⁵N-thymidine administration. Scale bar, 10 μm.

Fig. 7 |. Distinguishing cardiomyocyte from non-cardiomyocyte nuclei with additional analysis of ³²S images.

¹²C¹⁴N images demonstrate the cellular and subcellular architecture, especially sarcomeres. ³¹P images highlight the nucleus. ³²S images reveal sulfur-rich sarcomere structures, which provide an additional means for identifying cardiomyocytes. a, A ¹⁵N-thymidine positive nucleus, identified by using the ¹⁵N/¹⁴N HSI image, is identified as a cardiomyocyte nucleus. b, A cluster of cells is depicted in which two nuclei demonstrate increased ¹⁵N labeling (yellow asterisks). Due to a combination of the sectioning orientation and sectioning artifact (arrows), the cells are categorized as indeterminate and excluded from the analysis. Scale bar, 10 μm.

After blinded assignment of ROIs is completed, a hue, saturation, intensity (HSI) ratio image for each ROI is constructed in order to determine which nuclei are at natural abundance (0.37% = unlabeled) or are above natural abundance (typically in the range of 0.45–0.60% with our protocol, Fig. 6c). The quantitative measurements for each ROI can be extracted from the ratio image. In addition, ¹⁵N-thymidine labeling in cardiomyocyte nuclei tends to be more prominent at the periphery of the nucleus, which is similar to chromatin condensation seen by DAPI⁷. These small, distinct points display a higher ¹⁵N/¹⁴N ratio than the mean enrichment of the entire nucleus. These puncta also colocalize with high signal in the ³¹P images, which are a proxy for DAPI staining, due to the high P content of chromatin^7,46.

Identification of cardiomyocyte division, formation of binucleated cardiomyocytes, and formation of polyploid nuclei

Cardiomyocyte division is defined as ¹⁵N-thymidine labeling of a mononucleated diploid cardiomyocyte. In contrast, formation of binucleated cardiomyocytes is defined as two ¹⁵N-thymidine-positive nuclei within one cardiomyocyte, and increased ploidy is defined as a nucleus of >2N in a mono- or binucleated cardiomyocyte. Binucleation is best determined with cardiomyocyte sections on the long axis. In order to determine nuclear ploidy of ¹⁵N-thymidine positive cardiomyocytes, we analyzed the sections adjacent to the MIMS section with confocal microscopy. Four neighboring sections were selected: two above and two below the MIMS section (i.e., assuming the index of the section for MIMS imaging is k, the indices of the sections for ploidy study are k-2, k-1, k+1, and k+2, from top to bottom (Fig. 8a,b). Slides containing the four selected tissue sections are fixed with 3.7% (vol/vol) formaldehyde. The sections are then stained with Hoechst 33342 and imaged with a Nikon A1R confocal microscope using a 40× oil immersion objective with NIS Elements software. For potential future applications to subjects at later developmental stages where the sarcomeric structures are more mature and the cardiomyocytes much larger, analyses of additional sections in the z-axis may be required in order to capture the nucleation state for each cardiomyocyte of interest, similar to approaches used in adult mice⁷.

Fig. 8 |. Alignment of ¹⁵N/¹⁴N image with Hoechst-stained serial sections for quantification of nuclear ploidy.

a, An identified ¹⁵N-thymidine-positive cardiomyocyte nucleus on section k is indicated with a red arrow. The silicon chip used for MIMS imaging (³¹P, ¹⁵N/¹⁴N) is referred to as section k. b, Four adjacent sections (k-2, k-1, k+1, k+2, each 500 nm thick) were stained with Hoechst (blue staining) to detect DNA. The sections above the reference section k are k-1, k-2, and sections below are k+1, k+2. A reference pattern (yellow circle) was selected to assist the image alignment. c, To align the orientation of the reference pattern between section k and sections k±2, the Hoechst images were rotated and flipped. Squares outlined with interrupted yellow lines indicate magnified areas shown in (d). d, The nucleus of interest was outlined on the different sections and the Hoechst fluorescence intensities were measured by Fiji. The sum of all four images represents the total fluorescence of the nucleus. Scale bar, 10 μm.

Because the samples can be placed in various orientations when they are imaged with MIMS and confocal microscopy, the images acquired from neighboring sections need to be accurately aligned. Image alignment is a challenging but critical step. We use Fiji software (https://imagej.net/Fiji) for orienting and zooming in on the images to facilitate alignment. Each section is 500 nm thick; therefore the total distance analyzed is 2,500 nm. At this distance, the shape and size of an individual nucleus can have significant variability; however, the spatial relationship between neighboring nuclei remains relatively stable. Therefore, we match a pattern that contains several nuclei that form a unique and recognizable spatial pattern to align the neighboring sections. We used the ³¹P image to identify the reference pattern, as this is specific for the nucleus due to the phosphorus content in chromatin (Fig. 8b). Next, we search for the reference pattern on the k±1, followed by the k±2 Hoechst images. The orientation of the searching image might need to be rotated and/or flipped to align with the reference pattern (Fig. 8b,c). Using the previously identified ROI that corresponds to a ¹⁵N-positive cardiomyocyte, the nucleus is traced on the Hoechst images (Fig. 8d). The Hoechst fluorescence intensity is measured using Fiji.

These aligned, Hoechst-stained serial sections are used to determine binucleation and nuclear polyploidization. The cardiomyocyte containing a ¹⁵N-positive nucleus is compared in all four neighboring sections to determine whether it also contains a second nucleus. To determine nuclear ploidy, we summed the Hoechst intensity from all four sections in a single ¹⁵N-positive cardiomyocyte nucleus. We defined a diploid (2N) as the Hoechst fluorescence intensity of a ¹⁵N-thymidine negative non-cardiomyocyte nucleus, and it was used as the reference ploidy. The ploidy of ¹⁵N-thymidine-positive cardiomyocyte nuclei was determined by comparing their Hoechst fluorescence intensity with that of the reference ploidy.

Confirmation of adequate labeling

A convenient way to find out whether the label has been taken up by the body is to measure excretion of the stable isotope in the urine. Although it is possible to measure urine samples with MIMS that have been dried on the surface of a silicon chip, higher throughput bulk methods are also practical, such as with an isotope ratio mass spectrometer configured to measure nitrogen isotope ratios^64,65. In our experience, an increase of 800% of ¹⁵N in urine samples is sufficient to indicate labeling in cardiomyocytes (Fig. 9). A lack of administration or decreased absorption would be suggested by low ¹⁵N excretion in the urine. Missed doses or decreased absorption should be considered when interpreting the final MIMS results.

Fig. 9 |. Urinary ¹⁵N/¹⁴N ratio measurement confirms uptake after oral administration of ¹⁵N-thymidine.

During a 5 d oral administration of ¹⁵N-thymidine (50 mg/kg body weight) to an infant with ToF, cotton balls were placed in every diaper to collect urine. Urine from each day was pooled. The ¹⁵N/¹⁴N ratio was analyzed using IRMS. Control samples from the same patient before label administration demonstrated a ¹⁵N/¹⁴N ratio at natural abundance (0.37%, dashed line). Urine collected during ¹⁵N-thymidine labeling demonstrated ¹⁵N/¹⁴N ratios of approximately 3% (800% above natural background).

MIMS analysis of WBCs is another positive control (Fig. 1) since hematopoiesis involves cell proliferation followed by differentiation. Consequently, ¹⁵N-thymidine positive cells are released into circulation where they can remain for weeks to months. There is a lag period before bone marrow-derived cells appear in the peripheral blood; therefore, a lack of ¹⁵N-positive WBCs in circulation during short chase periods does not necessarily indicate that the ¹⁵N-thymidine was not absorbed and incorporated. Additionally, WBCs can also die in circulation with longer chase periods. Therefore, the presence of ¹⁵N-thymidine in WBCs is a reassuring sign that the dose is sufficient; however, its absence does not necessarily indicate an unsuccessful labeling.

Limitations

We recruited patients from the UPMC Children’s Hospital of Pittsburgh, a tertiary referral hospital that cares for a relatively high number of CHD patients. Although many large academic centers have similar pediatric patient populations, access to a large number of patients with a specific type of CHD might not be feasible at all settings.

For ethical reasons, this human research protocol cannot influence the clinical care provided to the research participants; therefore, certain important variables cannot be controlled. For example, it is impossible to control the timing of the right ventricular muscle resection. As such, there can be significant variations in the chase time, i.e., the duration between ¹⁵N-thymidine administration and ToF surgical repair. Additionally, there are challenges with giving the oral label to infants. For example, some infants are more likely to spit up part or all of the dose, and we cannot entirely exclude variations in label absorption through the GI tract.

There are also limitations specifically related to the conduct of NanoSIMS analysis. First, there are a limited number of NanoSIMS instruments in the world, approximately 50 at the time of publication. Not all centers operate as fee-for-service or core facilities. In addition, due to diverse applications in non-biological sciences, not all centers allocate instrument time to biological samples. Second, the operation of a NanoSIMS instrument requires significant training. In the Brigham and Women’s Hospital Center for NanoImaging, where we conducted our study, analyses are usually conducted by a physicist. Therefore, the application of MIMS by a research team with no prior experience requires working with a core facility or partnering with an experienced operator with access to a NanoSIMS instrument. Finally, MIMS analysis is time-consuming. Despite iterative improvements in analytical approaches, analyzing sufficiently large swaths of myocardial tissue for identification of ¹⁵N-thymidine-labeled cardiomyocyte nuclei involves a minimum of 1 d of instrument time per patient specimen at a cost of a few thousand US dollars. Aside from the analytical costs associated with NanoSIMS analysis, stable isotope tracers represent an additional cost, particularly for human studies where the requisite per subject dose is approximately 3 orders of magnitude greater than for murine studies (Table 1). A current question is whether continued growth of MIMS applications will drive improvements to the technology, which should include a decrease in costs.

Table 1 |.

Cost estimates for labeling one study subject (infant)

Item	Approximate cost per unit	Approximate cost for one study subject (3.6 kg)
15N-thymidine (50 mg/kg body weight)	$400/kg body weight	$1,600
Tissue sectioning	NA	$250–$500
MIMS	$1,500–$3,000 per day of instrument use	$4,000–$8,000
Confocal time: binucleation and ploidy	$40–$80 per h instrument time	$500–$1,500
General lab supplies	NA	$200
Total cost estimate for one infant	NA	$6,550–$11,800

Cost was estimated for a single patient, assuming weight of 3.6 kg at time of ¹⁵N-thymidine dosing. The greatest variability in cost should be anticipated to originate from the different charge structures of NanoSIMS facilities and microscopy cores. In addition, the dose and duration of ¹⁵N-thymidine administration is a major cost-driving variable.

In addition to time and costs associated with NanoSIMS analysis, the analysis of adjacent sections with correlative microscopy is also time-intensive. This includes identification of individual nuclei on the images and then alignment of the regions of interest (ROI) from the MIMS analysis and the Hoechst-stained images. With the development of artificial intelligence (AI) approaches, it might become possible to automatically identify nuclei and align ROI between images. This would significantly decrease the requisite time for this protocol and would be particularly useful for larger projects.

Materials

Biologics

• Human blood, collected by venipuncture in tubes with anticoagulant

• Human right ventricular myocardium !CAUTION The study must receive IRB approval before initiation. All study participants (or their parents) must give their written, informed consent. !CAUTION Human samples should be handled according to the standard safety guidelines for noninactivated blood specimens. ▲ CRITICAL Human tissue should be stored and/or transported in a cool box (4–15 °C) and processed within 4 h of collection.

Reagents

• ¹⁵N-thymidine (Cambridge Isotope Laboratories, cat. no. NLM-3901-MPT-PK)

• Paraformaldehyde (PFA), 16% (vol/vol) (Electron Microscopy Sciences, cat. no. 15700)

• Formaldehyde, 37% (vol/vol) (Sigma Aldrich, cat. no. F8775–25ML)

• PBS (sodium chloride, potassium chloride, sodium phosphate dibasic, potassium phosphate monobasic, no calcium, no magnesium, pH 7.4). (Fisher Scientific, cat. no. R582350010F)

• Reagent Alcohol, 99.5% (vol/vol) (Fisher Scientific, cat. no. AC615090040)

• LR White Resin, Medical Grade Catalyzed (Electron Microscopy Sciences, cat. no. 14381)

• Toluidine Blue O and Sodium Borate (Fisher Scientific, cat. no. T161–25)

• Hoechst 33342, 10 mg/mL solution in water (Fisher Scientific, cat. no. H3570)

• Mounting medium: 1% (wt/vol) n-propyl gallate (Fluka, cat. no. 02370) dissolved in glycerol (Cayman, cat. no. 62600)

• Nail polishm (Fisher Scientific, cat. no. NC1849418)

• ACK lysing buffer (Gibco, cat. no. A10492–01)

• Nitrogen gas (Airgas, NI, cat. no. NF300)

• Acetone (Sigma-Aldrich, cat. no. 154598, ACS spectrophotometric grade >99.5%)

• Methanol (Sigma-Aldrich, cat. no. M3641, spectrophotometric grade >99%)

• Isopropanol (Sigma-Aldrich, cat. no. 154970, ACS spectrophotometric grade >99.5%)

• 0.2 μm-filtered ethanol (Sigma-Aldrich, cat. no. 493546 (200 proof) USP/NF)

• Water, ultrapure, spectrophotometric grade (Alfa Aesar, cat. no. 19391)

Equipment

• 600 mL glass beaker (Fisher Scientific, cat. no. 10768962)

• Petri dish (Fisher Scientific, cat. no. FB0875711C)

• Gelatin capsules (Electron Microscopy Sciences, cat. no. 70102)

• 4 mL EDTA tubes (Fisher Scientific, cat. no. 02–687-107)

• Disposable pipettes (Fisher Scientific, cat. no. 13–711-9BM)

• 50 mL Falcon tubes (Fisher Scientific, cat. no. 1443222)

• Sterile surgical blade #22 (Fisher Scientific, cat. no. 50–364-928)

• Sterile forceps (Fine Science Tools, cat. no. 11370–42)

• Eppendorf tubes, 2 mL (Fisher Scientific, cat. no. 50–408-138)

• Silicon wafers (Pure Wafer, cat. no. BCM-6N1.008-.017SSP)

• Superfrost Plus microscope slides (Fisher Scientific, cat. no. 1255015)

• Confocal microscope (Nikon, A1R)

• Ultramicrotome equipped with diamond blade capable of cutting plastic resin embedded samples to 500 nm (e.g., Leica Ultracut 7)

• Gold sputter coater (Electron Microscopy Sciences Peltier Cooled EMS575x Turbo Sputter Coater)

• NanoSIMS 50 or 50L (Cameca)

• DIC scope (Nikon Eclipse E800) equipped with an electronically controlled x-y stage (LEP MAC5000) and a CCD camera (AXIS 221 Network Camera)

• Tin capsules with smooth walls (e.g., Elemental Microanalysis, cat. no. D4037)

• Elemental analyzer-isotope ratio mass spectrometer (e.g., isoprime precisION, Elementar)

Software

• Fiji (https://imagej.net/Fiji)

• OpenMIMS plug-in for ImageJ/Fiji (https://nano.bwh.harvard.edu/openmims)

• NIS-Elements (https://www.microscope.healthcare.nikon.com/products/software/nis-elements/viewer)

• Coordinate transformation program (https://github.com/BWHCNI/scope2MIMS)

Equipment setup

Preparation of silicon chips

1. Dice a 150 mm diameter silicon wafer into 4.95 × 4.95 mm pieces with either a diamond knife or laser cutting machine and separate them into ~75 square silicon chips.

2. Ultrasonicate the chips in a clean 600 mL glass beaker filled with a ~250 mL bath of acetone, fully submerging the chips, for ~30 min while minimizing the chips’ contact with the beaker’s surface.

3. Repeat step 2, in order, for each of the following liquids: methanol, isopropanol, and finally ultrapure water.

4. Once the chips have been thoroughly rinsed with filtered, distilled water, place them in a petri dish and ensure that all have their mirrored (shiny) sides up.

5. With filtered, high-purity nitrogen or argon gas, dry the chips’ surfaces, being careful not to invert or damage them, and place the petri dish for storage.

Procedure

Patient enrollment and ¹⁵N-thymidine administration ● Timing 2 h

• 1Identify potential patients, introduce the research study to each family, review the consent document, and, if families agree to participate, obtain written consent according to local institutional guidelines.
• 2Arrange the preparation of ¹⁵N-thymidine with the research pharmacy. For each patient, acquire five oral doses of 50 mg/kg/d in individual single-dose syringes.
• 3Provide the family with a daily log to record time of dose, any issues surrounding administration (e.g., emesis), and urine output (Supplementary Manual).
• 4Explain to the parents how to place cotton balls in the study patient’s diaper to collect urine. This should start after the first dose is administered. With each diaper change, collect the cotton balls in a specimen bag and store in the freezer. Collect urine for 24 h after the last dose.

Ascertainment of tissue samples and subsequent tissue processing ● Timing 3–4 d; 12–16 researcher hours

• 5
  On the day when the myocardium sample is obtained, assemble the samples required for analysis:

  • De-identify all sample collection containers by replacing the names with a number or code
  • Collect a blood sample. Draw the blood from the patient, collect it in an EDTA tube, and keep it on ice. Bring the tube to the lab and WBCs as described below.
  • Collect the myocardium sample in a container and keep it on ice for tissue processing as described below. Process the tissue within 4 h of collection.
    !CAUTION Human samples should be handled according to the standard safety guidelines for non-inactivated blood specimens.

WBC extraction

• 6Spin the blood at 500g for 15 min at room temperature (approximately 25 °C).
• 7Collect the buffy coat using disposable pipettes; place in a 50 mL Falcon tube.
• 8Lyse the 3 mL of whole blood in 15 mL of red blood cell ACK lysis buffer for 10 min on ice.
• 9Spin 500g for 5 min at room temperature.
• 10Discard the supernatant by gently decanting it.
• 11Vortex gently to loosen the pellet.
• 12Fix in 5 mL of 4% (vol/vol) PFA at 4 °C for 24 h.
• 13Place the fixed WBCs in a clear 1.5 mL polypropylene microcentrifuge tube and centrifuge at 500g for 15 min at room temperature.
• 14Dehydrate the WBC pellet in a graded series of ethanol: 15 min in 30% (vol/vol), 50%, 70%, 90%, and then three times in 100%.
• 15After the infiltration, embed the samples in fresh LR White resin in the bottom of gelatin capsules, allowing for the removal of as much air as possible and cure at 60 °C overnight.

Tissue fixation (myocardium)

• 16Wash the tissue in cold PBS.
• 17Cut the myocardial sample into smaller sections (~1 mm).
• 18Fix the tissue in 1 mL of 4% (vol/vol) PFA at 4 °C for 24 h.
• 19Wash the tissue with cold PBS.
• 20Dehydrate the fixed specimens in a graded series of ethanol, then embed them in LR White Resin, as described above.

Tissue sectioning

• 21Cut the tissue into 500 nm sections (Leica Ultracut 7). Stain sections mounted on glass slides with 0.5% (wt/vol) Toluidine Blue in 1% (wt/vol) sodium borate and quality control the sectioning by inspecting with a light microscope.
• 22Establish and maintain a record of consecutive sections with careful labeling for later analysis of mono/binucleation and ploidy of nuclei.
• 23For the myocardium samples, collect ten 500 nm sections on plus-coated glass slides for immunofluorescent staining, followed by one 500 nm section collected on a silicon chip for MIMS. Repeat this four more times for a minimum of five sets of sections.
• 24For the WBC samples, collect two sets of five 500 nm sections on plus-coated glass slides for immunofluorescent staining, followed by two 500 nm sections, each collected on a silicon chip for MIMS.
• 25Follow the second set of WBC sections with an additional five sections on glass slides.
• 26Label each glass slide and silicon chip in a numbered format, including deidentified patient numbers, tissue specimen numbers, silicon chip numbers, and section numbers. (For example, #1,1,6,11 corresponds to patient #1, tissue piece #1, silicon chip #6, section #11).

NanoSIMS analysis ● Timing 2–3 d per sample

▲ CRITICAL The following is an overview of a typical workflow of what is involved in NanoSIMS analyses⁶⁶. The focus here is on the measurement of N isotopes in order to capture incorporation of ¹⁵N-enriched tracers (e.g., ¹⁵N-thymidine), together with ³¹P and ³²S ions that reveal histological features in parallel. This process is repeated for cardiomyocytes, WBCs, or any additional tissues of interest.

• 27Mount the silicon chips in a typical NanoSIMS holder. Although there are multiple types of holder, accommodating silicon chips of various dimensions, we have found that a holder containing 16 holes (7 mm diameter) is ideal. Take caution to ensure that the chips are flat before fixing the cover plate of the sample holder in place. Look at each chip individually with an optical microscope; if the focus goes in and out as the chip is traversed from side-to-side, it should be remounted.
• 28
  Document regions of interest (ROIs) on each chip for analysis before putting the sample holder into the instrument. This helps to minimize NanoSIMS analysis of dead space and tissue regions that do not contain myocardium, thereby augmenting analytical throughput. We use a differential interference contrast (DIC) optical microscope; however, other imaging methods such as an electron microscope can also be used.

  • Capture images at increasingly high resolutions and, if possible, record the coordinates of these ROIs.
  • Capture images/coordinates to assist with coordinate transformation and navigating the samples with the CCD camera once the holder is in the NanoSIMS instrument. For some fiducial markers, a minimum of three reference points is necessary to be useful for accurate prediction of the ROI positions. Fiducial markers can be made with a diamond knife or a focused ion beam.
• 29Sputter coat a 10–20 nm thick layer of at least 99.999% pure gold onto the surfaces of the silicon chips with the holder oriented such that the samples’ surfaces are facing outward.

  ▲ CRITICAL STEP This coating improves conductivity at the sample surfaces, reducing the possibility of charging effects due to resin embedding. This is important because charging reduces ionization efficiency, thereby reducing analytical throughput, and leads to distorted, blurry ion images. NanoSIMS is sensitive to sample topology, and the gold coat smooths the sample surface, which also improves the quality of the image.

• 30Insert the sample holder into the NanoSIMS and find the previously identified and documented ROIs with the CCD camera and record their coordinates. For analyses of myocardial sections, the ROIs will be swaths of myocardium. While the CCD camera has an adjustable focus, it cannot reach as high a magnification as typical optical cameras or microscopes. This is where previous documentation is helpful. In the Navigator window in the software, save these coordinates one by one, and, if possible, use coordinate transformations to estimate the approximate areas where the ROIs may be located (see Step 33 below).
• 31Start up the primary ion source by sequentially increasing the high voltage, the ionizer, and, finally, the reservoir to achieve a stable primary Cs⁺ ion beam of 25–35 nA in the primary Faraday cup (FcP). The NanoSIMS is designed to run at a high voltage of 8,000 V, an ionizer current of approximately 1.5–2.2 mA (nominally 1.8 mA), and a reservoir current between 0.05 and 0.6 mA (depending on amount of beam needed and age of the source).
• 32Switch from CCD mode into SIMS mode and obtain a secondary electron (SE) image by sputtering any part of the sample, preferably close to a ROI found with the CCD camera. Note that SE and CCD images might not be precisely aligned with each other; an offset between the two can be determined by finding the same feature in both imaging modes and applying a correction factor in the instrument software. Determine the location of as many of the ROIs as possible as measured within the SE image and record these new positions in the Navigator software.
• 33If available, use a coordinate transformation program (some are freely available on the Internet) to calculate the offset between the microscope coordinates obtained in Step 30 and the NanoSIMS electron image coordinates. As mentioned above, only three points are needed (finding more, though, will increase the accuracy of the transformations); locate the same three points in both coordinate systems.
• 34Set up the detectors at the masses required. Our recent work used a NanoSIMS 50L instrument, which contains seven detectors. The analytical approaches described here are also achievable with a NanoSIMS 50 instrument, which contains five detectors. For both instruments, all but one of the detectors (the last) is movable, thereby affording a range of instrumental setups that can be adapted for the particular stable isotopes one wants to measure.
• 35Obtain a high mass-resolving power (MRP; m/Δm) by adjusting the quadrupole lens (Q), the slit lens (LF4), and the Hexapole lens (its voltage and its physical position), in this order. Repeat this iteratively to obtain the highest MRP possible until there is no further improvement with additional adjustments.
  The instrument parameters are as follows:

  • Correctly place an entrance slit (ES) of a maximum width of 30 μm (ES-3; smaller widths are possible using ES-4 and ES-5, thereby further increasing the MRP).
  • Insert an aperture slit (AS) of maximum width 200 μm (AS-2; similar to ES, smaller AS widths are possible, which also can increase the MRP) and maximize the signal going through it by slightly adjusting its physical position.
  • Place an exit slit (ExS) of a maximum width of 50 μm in front of each detector to achieve a MRP of at least ~6,000, which is high enough to separate CN^- ions from other masses.
• 36Determine an appropriate amount of pre-sputtering time necessary to achieve a steady state secondary ion beam signal. Direct the primary ion beam of Cs at the sample with a high beam current of ~1.5 nA (achieved by increasing the primary source lens L1 to ~4.5 kV), as measured in the object Faraday cup (FcO) and using the D1–1 aperture (700 μm diameter). This is the amount of beam that impacts the sample. The time required using a 1.5 nA primary beam on a 10 nm thick Au coated sample is typically 1–3 min. However, pre-sputtering time depends on the primary beam intensity, thickness of the gold coating, and the size of the area the beam is rastered over. Lower beam currents, thicker gold coats, and larger areas will require increasingly longer pre-sputter times, reducing throughput. The method of preparation and type of the sample also affects the ionization efficiency, or time until enough Cs has been implanted in the tissue before a steady-state secondary ion beam signal is achieved.
• 37Insert the D1–3 aperture (200 μm diameter), if it is not already in place, and adjust the E0P (~8,500 V) and the primary octupole lenses to obtain the optimal spatial resolution. With the primary lenses L0 and L1 set to 0 V, obtain a primary beam of roughly 100 nm in diameter. Focusing with the SE images is the easiest way to see how adjusting E0P and octupole lense alters image quality.
• 38Set up the conditions for the acquisition of data. Images are composite sequential scans (termed “planes” or “frames”), each consisting of 256 × 256 pixels, with counting times of ~2,000 μs/pixel. The resultant image stack, consisting of 5–20 frames, can then be summed up to produce one total image, as explained in detail in Steps 41–43. Because the summed image contains multiple frames as it sputters through the samples, it can also act as a depth profile. When analyzing swaths of myocardium (similar for sections of other tissue types), start in an automated chain mode, where a series of adjacent imaging fields are obtained sequentially. These images can then be stitched together in post-processing, as described in detail in Step 41, to provide an overview of a larger area of tissue than can be contained in a single imaging field (e.g., 40–60 μm diameter). In this way, large areas of sizes up to 250,000 μm² can be analyzed in a single pass.
• 39Upon completion of the initial analyses in automated chain mode, leave the sample holder in the NanoSIMS instrument under vacuum until no additional second pass analyses are required. Due to the limited number of slots for sample holders in the instrument, try to quickly process the image files in order to determine whether additional imaging (e.g., at higher spatial resolution) is needed. Once a sample holder is removed from vacuum, areas of the sample surface that have already been analyzed (and implanted with Cs) can become damaged. Therefore, avoid removing samples from the instrument if additional second pass analyses are required on previously measured areas.
• 40After reviewing the initial images obtained from automated mode analyses, certain areas might benefit from analysis at higher spatial resolution. The initial analyses in automated chain mode only sputter away a fraction of the 500 nm-thick tissue section and therefore abundant tissue remains for higher resolution imaging.
• 41To image at higher spatial resolution, reduce the field size to capture just the features of interest (often 10–30 μm in diameter), acquire images of 512 × 512 pixels, and increase L1 to between 7 kV and 8 kV. However, this reduces the primary beam density, leading to a reduction in primary ion yield. This requires longer integration times per pixel (e.g., 10,000 μs/pixel) and translates into longer requisite measurement times and lower throughput. Most analyses do not require this level of spatial resolution, and therefore do not routinely operate at the highest possible spatial resolution with a “first pass” analysis.

Image analysis and identification of ¹⁵N-positive nuclei ● Timing 2–4 h per image

▲ CRITICAL NanoSIMS data files can be viewed and analyzed using different software platforms, including WinImage (CAMECA), L’Image (L. R. Nittler, Carnegie Institution of Washington), or OpenMIMS. WinImage and L’Image are proprietary, whereas OpenMIMS was developed at the National Resource for Imaging Mass Spectrometry (now Center for NanoImaging, Brigham and Women’s Hospital, Boston) and is available at no cost as a plugin to ImageJ. Therefore, this part of the protocol will focus on image viewing and analysis with OpenMIMS.

• 42(Optional) For analysis of swaths of a histological section in chain analysis mode, stitch the tiles together into a single image file. Utilize a script (“mosaic_nrrd”; also freely available online) that enables the merger of stitched tiles into a single “.nrrd” file.
• 43Open a NanoSIMS “.im” or “.nrrd” file by dragging the file into the OpenMIMS interface.
• 44Select the “Stack Editing” tab. Click on CN image (mass 26). Select the “Autotrack” button to align sequentially obtained a stack of imaging planes. Click the “Sum” button for images that will be used for histological identification (mass 26/CN, mass 31/P, mass 32/S).
• 45Prior to selecting regions of interest (ROI), define relevant ROI categories. In the “MIMS ROI Manager” window, create and assign identifiers to individual “Groups” for each category of region of interest (ROI) for which ratio data will be extracted (e.g., cardiomyocyte nuclei or stromalvascular nuclei). Create a new group by right clicking over the “Groups” section. ROIs generated while a given group is selected will be listed in that group and tagged with the group name when the corresponding data is extracted.
• 46Determine which mass image(s) will be utilized for the selection of ROIs. We recommend selecting ROIs and assigning them to categories as defined above prior to construction of ratio images, so as to blind the observer to labeling status. ¹²C¹⁴N images and ³²S images are useful for revealing general histological details, whereas ³¹P images reveal nuclear contours due to the high phosphorus emission from chromatin (Figs. 6 and 7).
• 47
  Build a hue saturation ratio image (HSI) under the “Process” tab in the OpenMIMS interface by selecting the ratio of interest.

  • In the case of experiments involving ¹⁵N-labels (e.g., ¹⁵N-thymidine), select ¹²C¹⁵N/¹²C¹⁴N.
  • Click “Display HSI”.
  • Click “Use Sum” to compress sequentially obtained imaging planes into one compressed image. The software includes a “scale factor,” set at a default of 10,000. This feature is a convenience to enable working with round numbers rather than decimals and can be changed if desired. In the case of ¹⁵N-enrichment, for example, a scale factor of 10,000 results in the ¹⁵N background natural abundance being reported as 37, rather than 0.0037 (or 0.37%).
  • Adjust the upper bound of the scale to visually reveal differences in the isotope ratio, which are indicative of areas of differential labeling.
  • Finally, adjust the RBG maximum and minimum values to the desired intensity.
• 48Please note that any scaling changes modify only the visual representation of the underlying isotope ratio measurements; the underlying data for any given ROI is unaffected.
• 49Identify ROIs that would benefit from higher resolution imaging (i.e., indeterminant cells). Situations where the relationship between a nucleus and cell borders are not clearly discernable or when a nucleus is at the margin of two adjacent imaging fields might require higher resolution imaging. There are certain artifacts that are generally not worth pursuing with higher resolution imaging, for example, if there is a crack or wrinkle in the section (Fig. 7).
• 50In order to extract the isotope ratio data, highlight all of the ROIs in the “ROI manager.” Select the “Tomography” tab in the OpenMIMS interface. Within this tab, select the experiment-specific analytics needed for interpretation of the data. The mean is always critical, and additional metrics can be used for certain applications, such as x and y coordinates or ROI areas. Select the masses of interest. For ¹⁵N-labeling experiments, at a minimum, select “HSI 27/26”. Select “Table” to extract the data. Save the resultant data table, which can then be imported into a spreadsheet program.

Identify mono- and binucleated ¹⁵N-positive cardiomyocytes ● Timing 2 h per sample

▲ CRITICAL Both nuclei of a binucleated cardiomyocyte are usually arranged along the long axis of the cell. We selected cardiomyocytes sectioned along the long axis for quantification of mono- and binucleated cardiomyocytes.

• 51Open the MIMS images.
• 52According to the sarcomere structure, which is best visualized on the ¹⁴N and ³²S images, identify cardiomyocytes and determine whether they are sectioned along the long axis.
• 53Examine the ³¹P image to identify nuclei. Circle each nucleus as an ROI.
• 54Using the ¹²C¹⁴N and ³²S images, determine whether each ROI corresponds to a cardiomyocyte, noncardiomyocyte, or undetermined cell. Cardiomyocytes are defined as having round to oval nuclei in cells containing sarcomeres.
• 55Examine the ¹²C¹⁵N/¹²C¹⁴N ratio HSI images to identify the ¹⁵N-thymidine positive nuclei. Those cells with a background ratio of ¹⁵N/¹⁴N appear blue, while nuclei with an increased ratio have rainbow colored nuclei. Visual representation can be adjusted using the color scale, which does not impact the data. In the exported tomography table (Step 48), confirm that ¹⁵N/¹⁴N ratio is above the natural baseline of 0.37%.
• 56Quantify the ¹⁵N-thymidine positive and negative nuclei in both mono- and binucleated cardiomyocytes. ¹⁵N-thymidine positive cardiomyocyte nuclei are those nuclei (³¹P) within cardiomyocytes (¹²C¹⁴N and ³²S) that have a ¹⁵N/¹⁴N ratio above the natural abundance (0.37%).
• 57Use a long axis section of a cardiomyocyte to determine whether it is binucleated. Binucleated cardiomyocytes contain two nuclei within one cellular membrane. If cardiomyocytes are sectioned off-axis, additional sections in the z axis must be analyzed.

Determine the ploidy of cardiomyocyte nuclei ● Timing 5.5 h

DNA staining (30 min)

• 58Select four (k-1 and k-2 and k+1 and k+2) sections that are immediately adjacent to the section (k) used for MIMS.
• 59Fix the selected sections (k±1 and k±2) by adding 100 μL of 3.7% (vol/vol) formaldehyde solution at room temperature for 12 min.
• 60Remove the formaldehyde solution and carefully wash the sections with PBS three times.
• 61Stain the sections with 100 μL of Hoechst working solution (10 μg/mL in PBS) at room temperature for 5 min.
• 62Remove the Hoechst working solution carefully by pipetting.
• 63Wash the sections carefully with PBS three times.
• 64Add 50 μL of mounting medium (1% (wt/vol) n-propyl gallate dissolved in glycerol) to each section and then cover them with a clean cover glass.
• 65Seal the gap between the cover glass and the glass slide at the edges with nail polish.

Acquisition of the Hoechst images of section k±1 and image alignment (1 h)

• 66Acquire a single image of the whole section k±1 stained by Hoechst by using the “Acquire large image” function of the Nikon Imaging Systems (NIS) Elements software controlling the Nikon A1R confocal microscope.
• 67Open the ³¹P image of the section k in the Fiji software.
• 68Select a group of cell nuclei in a distinct pattern on the ³¹P image of section k. This group of cell nuclei is used as the reference pattern k.
• 69Open the image of sections k-1 in the Fiji software.
• 70Search and identify the reference pattern k on the image of the sections k-1.
• 71Rotate the image of section k-1 to match the orientation of the image of section k.
• 72Align the image of section k+1 by using the reference pattern k as described above.

Acquisition of the Hoechst images of section k±2 and image alignment (1 h)

• 73To align the Hoechst images of section k-2, select a new group of cell nuclei in a distinct pattern on the aligned image of section k-1 as the reference pattern k-1.
• 74Search and identify the reference pattern k-1 on the image of the section k-2.
• 75Rotate the image of section k-2 to match the orientation of the aligned image of section k-1.
• 76To align the Hoechst images of section k+2, select a new group of cell nuclei in a distinct pattern on the aligned image of section k+1 to serve as the reference pattern k+1.
• 77Search and identify the reference pattern k+1 on the image of the section k+2.
• 78Rotate the image of section k+2 to match the orientation of the aligned image of section k+1.

Identify the ¹⁵N-thymidine positive and negative cardiomyocyte nuclei and non-cardiomyocyte nuclei on the aligned Hoechst images (2 h)

• 79Open the ³¹P image on which the ¹⁵N positive nuclei ROI were labeled in the Fiji software.
• 80Open the aligned Hoechst images of section k±1 and k±2 in the Fiji software.
• 81Tile the opened images on the computer screen.
• 82Identify and label the ¹⁵N positive and negative cardiomyocyte nuclei and non-cardiomyocyte nuclei, respectively.

Determine the Hoechst fluorescence intensity of a nuclear section on an aligned Hoechst image (1 h)

• 83Open an aligned Hoechst image on which the ¹⁵N-thymidine positive and negative cardiomyocyte nuclei and non-cardiomyocyte nuclei are labeled using the Fiji software.
• 84
  Determine the unit-area background fluorescence intensity:

  • Select an area where no cells are found and define the region of interest (ROI).
  • Measure the gray value and the area size value of the ROI using the “Measure” function in the Fiji software.
  • Calculate the unit-area fluorescence intensity of the ROI by dividing the gray value with the area size value.
  • Select at least five ROIs as indicated above and determine their unit-area background fluorescence intensity.
  • Average the five unit-area background fluorescence intensities to obtain the mean unit-area background fluorescence intensity of this image.
• 85
  Determine the Hoechst fluorescence intensity of ¹⁵N-thymidine negative non-cardiomyocyte nuclear section:

  • Circle a ¹⁵N-thymidine negative non-cardiomyocyte nuclear section by tracing the nuclear boundary to define the ROI.
  • Measure the gray value and the area value of the ROI.
  • Calculate the total background fluorescence intensity of the ROI by multiplying the area value of the ROI and the mean unit-area background fluorescence intensity of this image.
  • Obtain the Hoechst fluorescence intensity of the nuclear section by subtracting the total background fluorescence intensity from the gray value of the ROI.
• 86
  Determine the Hoechst fluorescence intensity of ¹⁵N-thymidine positive cardiomyocyte nuclear section:

  • Circle a ¹⁵N-thymidine positive cardiomyocyte nuclear section by tracing the nuclear boundary to define the ROI.
  • Measure the gray value and the area value of the ROI.
  • Calculate the total background fluorescence intensity of the ROI by multiplying the area value of the ROI and the mean unit-area background fluorescence intensity of this image.
  • Obtain the Hoechst fluorescence intensity of the nuclear section by subtracting the total background fluorescence intensity from the gray value of the ROI.

Total Hoechst fluorescence intensity

• 87Open all four of the aligned Hoechst images (i.e., k±1 and k±2) in the Fiji software.
• 88Find the nuclear sections from each aligned Hoechst image that belong to the same nucleus.
• 89Obtain the Hoechst fluorescence intensity of the nuclear section on each image.
• 90Calculate the total Hoechst fluorescence intensity by adding together all the values of the Hoechst fluorescence intensity of the nuclear sections from different images.

Reference ploidy

• 91Obtain the total Hoechst fluorescence intensity of at least five ¹⁵N-thymidine negative non-cardiomyocyte nuclei.
• 92Calculate the average of the total Hoechst fluorescence intensity of the ¹⁵N-thymidine negative non-cardiomyocyte nuclei. The average value is the reference ploidy and is defined as 2N.
• 93Calculate the total Hoechst fluorescence intensity of the ¹⁵N-thymidine positive cardiomyocyte nuclei.
• 94
  To obtain the ploidy of a ¹⁵N-thymidine positive mononucleated cardiomyocyte, use the following equation:

  $$\begin{gathered} mononucleatedcardiomyocyteploidy= \\ \frac{totalHoechstfluorescenceintenticyofthecardiomyocytenuclues}{referenceploidy} \end{gathered}$$

Quantification of ¹⁵N in urine using IRMS ● Timing 2 h

▲ CRITICAL Samples can be sent to academic core or commercial stable isotope laboratories that have IRMS capability.

• 95Pipette urine into smooth-walled tin capsule (as opposed to tin foil capsules that can leak). Fill the capsule if sufficient urine is available. Also consider duplicate samples if sufficient urine is available.
• 96Dry samples overnight. Optional: the drying process can be accelerated by placing capsules in an oven at 50–60 °C.
• 97Once the sample is dried, crush the samples using forceps. At this point, the samples can be sent for analysis.
• 98Different EA-IRMS instruments will have differing protocols for tuning and quality control. However, for any analytical run, it is important to include replicates of a control standard at natural abundance (e.g., urea standard).

Troubleshooting

Troubleshooting advice can be found in Table 2.

Table 2 |.

Troubleshooting table

Step	Problem	Possible reason	Solution
2–4	15N-labeling not detected	Inadequate label dose or delivery	Use controls to assess for absorption (urine) and/or incorporation into alternative cell types (WBCs)
		Label is diluted away by ongoing proliferation during label free chase	Increase dose of 15N-thymidine labeling or decrease chase period
27–40	Indeterminate cell type in NanoSIMS images	Inadequate resolution or edge effect at margin of imaging field	Reanalyze with NanoSIMS at higher resolution and with centering of the histologic feature of interest in the imaging field
	Cannot resolve mass peaks for the N isotopes	The MRP is not high enough	Iteratively adjust Q, LF4, HEX and incrementally use smaller entrance and aperture slits
			Insert the smallest exit slit (ExS 3) for the detector monitoring 15N
36	Long pre-sputtering time	Primary ion beam intensity is too low	Adjust the voltage (higher or lower) slightly on the L1 primary lens, in coordination with deflection plates C1x and C1y
	Secondary ion intensity unstable	Sample is not flat and/or smooth	Inspect sample height with a microscope and readjust sample in NanoSIMS holder Apply a thicker gold coat
55	Cardiomyocytes in oblique axis for binucleation quantification		Review additional serial sections in the z-axis (sections are on glass slides) for a second nucleus
66	Over- or undersaturation of photomicrographs		Decrease concentration of Hoechst staining or incubation time
			Adjust the strength of the excitation light, the sensitivity of the emission light detector, and the exposure time

Timing

• Steps 1–4: Patient enrollment and ¹⁵N-thymidine administration: 2 h

• Steps 5–26: Ascertainment of tissue samples and subsequent tissue processing: 12–16 h

• Steps 27–41: NanoSIMS analysis: 2–3 d

• Steps 42–50: Image visualization and identification of ¹⁵N-positive nuclei: 2–4 h

• Steps 51–57: Identify mono- and binucleated ¹⁵N-positive cardiomyocytes: 2 h

• Steps 58–94: Determine the ploidy of cardiomyocyte nuclei: 5.5 h

• Steps 95–98: Quantification of ¹⁵N in urine using IRMS: 2 h

Anticipated results

Analysis of human tissues after a period of ¹⁵N-thymidine administration as described in this protocol will demonstrate the fraction of cells that have completed S-phase of the cell cycle during the labeling period. For most cell types, DNA replication is temporally linked to completion of mitosis and cytokinesis, such that ¹⁵N-thymidine labeling can be considered a proxy for cell division. Therefore, the fraction of labeled cells can be divided by the labeling duration to provide a proliferative rate for the cell type(s) of interest. For cells that exhibit stereotypical features of a specific cell type in MIMS images, no further analysis will be required. Endothelial cells with elongated nuclei and scant cytoplasm lining a blood vessel, or the signet ring appearance of adipocyte nuclei in close approximation to an intracellular dominant lipid droplet, are examples of cells that display such stereotypical features^28,46. In other instances, quantitative mass images themselves provide cell-type specific information. The sulfur-rich granules of small intestinal Paneth cells or neutrophils are examples where the ³²S images are of particular value for identification of cells²⁷.

Although cardiomyocytes also display identifying features, including intracellular sarcomeric structures, they are also amongst a select subset of cell types in which identification of nuclei that have undergone DNA replication is not sufficiently specific for identification of cell division due to the programmed failure of karyokinesis (leading to polyploid nuclei) and cytokinesis (leading to multinucleation)^18,19. In this context, orthogonal analysis of nuclear ploidy and nucleation state, as described in this protocol, are required to distinguish true cellular division events (¹⁵N-thymidine labeled, 2N, mononucleate cells) from binucleation (two ¹⁵N-thymidine labeled nuclei in a single cell) or polyploidization (¹⁵N-thymidine labeled nucleus that is 4N or greater). With measurement of these additional variables, the frequency of cardiomyocyte division events, binucleation, and polyploidization can be determined and each frequency converted to a rate by dividing by the duration of the labeling period. In human hearts, cardiomyocyte division, binucleation, and polyploidization follow distinct temporal patterns^10,23. Consequently, it is anticipated that application of this protocol will show similar temporal patterns. However, these patterns may differ in diseased hearts.

The rate of ¹⁵N-thymidine labeling in tissue is dependent on two key factors: the intrinsic proliferative rate during the labeling period and the duration of both labeling and label-free chase. For rapidly proliferating cells, such as those in the small intestinal crypt or tumor cells, a single labeling pulse will be sufficient to reveal the true proliferative rate. In the murine small intestine, for example, extensive proliferative activity in crypt cells is appreciable 4 h after a single intraperitoneal dose of ¹⁵N-thymidine²⁷. For rapidly proliferating cell types, a period of label-free chase that exceeds several rounds of the cell cycle (e.g., several days) will result in successive replacement of labeled DNA strands and therefore dilution of the label.

The quantitative power of MIMS when applied to ¹⁵N-thymidine labeling is linear over three orders of magnitude and was determined using murine crypt cells²⁷. Therefore, MIMS results can be used to infer the number of cell divisions based on the degree that the label is diluted. Indeed, a prior ¹⁵N-thymidine pulse-chase study of cell turnover in the murine small intestine captured cell division events for two additional rounds of cell division during chase²⁷. Additional cell division events during chase should be detectable based on the instrumental specifications of the NanoSIMS instrument; however, subtle degrees of labeling, as would be observed after several rounds of division during chase, require more time-intensive analytical approaches, including more involved instrumental tuning and more frequent comparisons to unlabeled control tissues and standards, a process successfully applied to other stable isotope studies⁶².

In the labeling protocol described here, standard clinical care dictates the timing of surgery. As such, we have often administered ¹⁵N-thymidine starting before the date for the complete surgical repair was known. This obviously results in variable chase periods, which have ranged from 2 weeks to 6 months. This protocol was designed with the assumption that, even in the event that a subset of labeled cardiomyocytes re-entered the cell cycle during label-free chase, the likelihood of complete dilution of the label is low due to the reported low proliferative capacity of cardiomyocytes and the high signal-to-noise ratio of ¹⁵N. Confirmation of sufficient ¹⁵N-thymidine labeling is suggested after 800% increase of ¹⁵N in the urine. Additional confirmation of sufficient labeling is suggested by the identification of ¹⁵N-positive WBCs. Future studies will take advantage of the multiplexed potential of MIMS in protocols in which two different labels are administered at different timepoints to determine whether a subset of cells is undergoing the type of frequent cycling that could result in complete label dilution.