Differential Growth Rates, Rather Than Compaction, Determine Left Ventricular Wall Formation


 Trabecular muscle makes up most of the ventricular wall of the human embryo. It is presumed that compaction in the foetal period changes ventricular wall morphology by converting weaker trabeculae to stronger compact muscle. Using developmental series covering the embryonic and foetal periods of human, mouse and chicken, we show ventricular morphology is determined by differential rates of positive growth of trabecular and compact layers rather than compaction. In mouse, foetal cardiomyocytes are relative weak, but adult trabecular and compact cardiomyocytes are equally strong. In foetal and adult humans, trabecular and compact myocardium exhibit a similar abundance of immunohistochemically detected vascular, mitochondrial and sarcomeric proteins. Even in human noncompaction, a congenital malformation characterized by excessive trabeculation, the trabecular and compact muscle are similar. In conclusion, trabecular and compact myocardium are equally equipped for force production and their proportions are determined not by compaction, but by differential growth rates.


Introduction
The relative thickness of the trabeculated over the compact myocardial layer in the normal adult left ventricle varies according to a log-normal distribution 1,2 whereby a ratio of > 2.3 3 is thought to be the most sensitive diagnostic criterion for Left Ventricular Non-Compaction (LVNC) cardiomyopathy 4 . This cardiomyopathy is characterised by prominent left ventricular trabeculae and, concomitantly, deep intertrabecular recesses. LVNC has been associated with ventricular dilatation, systolic dysfunction, arrhythmias and embolic events 5,6 . The notion that trabecular myocardium is ill-equipped to maintain good pump function has, therefore, been implied 7 . However, from large cohort studies a relationship between a hypertrabeculated left ventricle and cardiac dysfunction has not come forward 8,9 . The function of the trabecular myocardium has never been investigated directly, also because trabecular myocardium is di cult to model due to its complex 3 dimensional structure 10 .
Like the name implies, LVNC is thought to result from a failure of compaction 11 . Compaction has been described as a process of compression of trabecular myocardium to form part of the compact wall 12 , thereby simultaneously decreasing the amount of trabecular myocardium and increasing the amount of compact myocardium 13 . It has been claimed to occur in the embryonic period (e.g. 14,15 ), the foetal period (e.g. 16,17 ), or even up to birth 18 . Nevertheless, there are no comprehensive quantitative assessments of compact and trabecular layer growth to support the notion of compaction 19 . In fact, only two studies have quanti ed the trabecular layer volume in embryonic development and both report a continuous increase and not a decrease as would be expected if compaction was to occur (see 19 ).
Our aim was to investigate left ventricular free wall development during gestation in order to measure the magnitude and timing of compaction. Because the LVNC diagnosis depends on ratio in layer thickness 3,20 and calculated ventricular volume as proxy for myocardial mass 21 , those were also investigated. We compared the ndings from human to the growth curves we established for mouse and chicken, which are the most-used animal models in cardiac development with compact ventricular walls. Finally, we investigated if there are functional differences between trabecular and compact myocardium.

Methods
All studies were in compliance with the Helsinki Declaration and all protocols were approved by the Medical Ethics Committee of the University of Amsterdam, The Netherlands. The study was carried out in compliance with the ARRIVE guidelines and animal care and experiments conformed to the Directive 2010/63/EU of the European Parliament. All animal work was approved by the Animal Experimental Committee of the Amsterdam University Medical Centers, location AMC, Amsterdam, and was carried out in compliance with the Dutch government guidelines, approved by the Central Committee Animal Experiments.

Human
Single sections from transversely sectioned hearts from 3 to 36 weeks of gestation were used for morphometric analysis (N = 37). The origins of twenty hearts were previously described 22,23 . Eight sections were derived from the Medical Biology Department archives at Amsterdam UMC. These sections had all been immuno uorescently stained for cardiac troponin I 22 or histochemically with haematoxylin and eosin. Nine foetal sections, stained with Masson Trichrome, were provided by Dr. Sánchez-Quintana, and these hearts were obtained with permission of the Bioethical Committee on Human Research of the University of Extremadura, Spain.
Para n sections of previously described aborted foetal hearts with either normal (N = 2) or LVNC morphology (N = 4) 24 , and sections from adult autopsy derived morphologically normal human hearts (N = 6) were obtained from archives of the Pathology Department of Amsterdam UMC, and used for the functional phenotypical analysis of myocardium. Informed consent was obtained for all autopsies from the decedent's next of kin and for the anonymous use of archived autopsy materials a waiver was granted by the Medical Ethics Committee of the University of Amsterdam.
Mouse FVB/NRj wild-type mouse embryos (Mus musculus, Janvier labs, www.janvier-labs.com) of 9.5 to 17.5 gestational days were obtained from a variety of in-house bred litters. Hearts from these mice, and from postnatal mice from day 0 to 23 were isolated and xed with 4% paraformaldehyde in phosphate buffered saline for morphometric analysis (N = 47).
For the mitochondrial activity assay, samples of left ventricular papillary muscle with surrounding trabeculae carneae, compact wall, as well as left atria were collected from freshly killed adult surplus mice (N = 6).
For the contractile force analysis, whole ventricles from foetal mice at 15.5 gestational days (N = 2), as well as separated left ventricular compact and trabecular myocardium from surplus wild-type adult hearts (N = 6) were obtained by ne dissection before being snap frozen in liquid nitrogen.

Chicken
Wild-type chicken embryos (Gallus gallus, Drost BV, Nieuw Loosdrecht, The Netherlands) of 2 to 19 gestational days were obtained from eggs incubated in a humidi ed rocking stove at 38.5°C (N = 28).
Whole embryos or isolated hearts were xed with 4% paraformaldehyde in phosphate buffered saline. Additionally, sections of Bouin xed chicken embryos, previously histochemically stained for troponin bound antibodies (N = 10) or haematoxylin and eosin (N = 6), were included from the Medical Biology Department archives at Amsterdam UMC.

Morphometry
In situ hybridisation After xation, tissues were embedded in para n (Paraplast, Leica) and sectioned at 10 µm thickness with a Leica microtome. Sections at approximately 4 chamber view were then stained with digoxigenin labelled RNA probes against TNNI3 for mouse (1 ng/µl) and chicken (0.1 ng/µl) as previously described 25 .

Morphometric assessments
On images of stained hearts, a wheel with spokes at 30° intervals was projected (Online Fig. 1A). The centre of the wheel was placed at the centre of the left ventricular lumen while the spoke of 0° was placed approximately at the atrioventricular canal, the spoke of 180° was located towards the apex of the heart and the spoke of 330° pointed towards the aorta. The thickness of the lumen, the trabeculated myocardial layer, and compact myocardial layer were measured along the spokes of 0° to 210° with ImageJ 1.51j8 software (National Institutes of Health, USA) (Online Fig. 1B). The border between the myocardial layers was determined on sight by following the free lumen to where the crevasses between the trabeculae were deepest.
The surface areas of the compact and trabeculated myocardium between each spoke from 0° to 210°w ere measured on binary transformed images (Online Fig. 1C) with the polygon selection tool in ImageJ after the free lumen area had been subtracted by projection of an, as large as possible, ellipse in the ventricular cavity that did not cross any myocardial structures. Both the total surface area and the percentage of myocardialised surface area were collected (Online Fig. 1D). In case a spoke did not cross any left ventricular myocardium, for example if the 0° spoke lay in the atrioventricular canal, or when parts of the section were damaged, measurements were resumed at the next suitable spoke. In case no clear distinction between blood and myocardium could be made in the histochemically stained sections, no surface area measurements were taken.
To calculate the and the the following equations were used: The tilde stands for median, so the median of each total ventricular volume, with each measured total spoke length taken as r, was multiplied with the median of each myocardial fraction of surface area. Because the measured surface areas are not fully myocardial but also contain intertrabecular lumen and coronary artery lumina, the surface areas were corrected for the myocardial occupancy: %Myocardium. For the trabecular fraction, the lumen area, as obtained by placement of the ellipse, had been subtracted. respectively (1:100, monoclonal mouse-anti-human, Sigma-Aldrich,), which were visualised with DAB+ (Dako) in an immunoperoxidase stain and faintly counterstained with haematoxylin.
Images of sections were imported to ImageJ and converted to 8 bit (grey scale) images. In the adult samples, staining intensities in the trabecular layer, compact layer and epicardium were calculated respectively as their average grey scale values, which informed us on the relative amount of detected protein in each different tissue area per section. For the foetal specimens, the signal intensity of 5 (epicardium) to 10 (trabecular and compact) areas within a single section were pooled and median corrected per specimen.

Functional analysis
Mitochondrial respiration From 6 adult mice we isolated left ventricular papillary muscle with surrounding trabeculae carneae, compact wall, and left atrium. The mitochondrial function in these tissues were assessed separately as previously described 26 . Tissue

Force generation
To investigate if trabecular and compact myocardium are functionally different, mouse heart samples were subjected to calcium induced contraction force measurements from which maximal active (F max ) and passive force (F pas ) as well as the Hill coe cient (nHill), and EC 50 , were obtained.
Snap-frozen samples were defrosted and ground up in relax solution containing 1 mM Mg 2+ , KCl 145 mM, 2 mM EGTA, 4 mM ATP, 10 mM imidazole (pH 7.0). After a 5 minute incubation in 0.5% triton X-100 in relax solution, single adult cardiomyocytes or small clusters of foetal cardiomyocytes were selected on the basis of a minimal width of 20µm and a maximal width of 61µm, with visible sarcomeric structures to be measured after a force transducer and a piezoelectric motor needle were glued onto the cell with shellac.
Calcium response curves were measured at a sarcomeric length of 2.

Imaging
A Leica CTR5000 light microscope with a DFC450 camera was used to image the histochemical and in situ hybridisation stains. Immuno uorescent stains were imaged using a Leica CTR6000 microscope mounted with a Retiga EXi Fast camera (SN:Q19688, Qimaging). Confocal microscopy was performed with a Leica TCS SP8 SMD mounted on a Leica DMI6000 inverted microscope with objective HC PL APO CS2 63x/1.40.

Statistics
Analyses were performed using GraphPad Prism 8.3.0 (GraphPad Software, LLC). Error bars for the thickness slopes represent 95% con dence intervals. In all other gures error bars represent standard deviations and lines represent means. Growth curves were compared with linear regression analyses and correlation of thickness growth with age was calculated with a Pearson correlation coe cient. Staining signal intensities were compared with Tukey corrected One-Way ANOVA or Mixed Effects Analysis. In case of only 2 groups a paired t-test was performed.

Ventricular wall growth
The human heart grows from 0.07 to 4.7 cm in length between 20 days and 36 weeks of gestation. Early embryonic hearts t in toto into individual trabeculae of the late foetal heart (Fig. 1A). The trabecular layer thickness never decreased as would be expected if compaction took place (Fig. 1B). Dramatic changes to the LVNC-diagnostic ratio, however, did occur in the embryonic period (Fig. 1C). This was due to changes in layer growth rates. From approximately 22-31 gestational days, the trabecular layer increased faster in thickness than the compact layer (16 ± 9 µm/day, F = 21.26, 1 degree of freedom versus 0 ± 1 µm/day, difference p < 0.001, F = 0.64, 1 degree of freedom). This caused a thickness ratio increase to 10.2. This is much higher than the clinical LVNC cut-off value of 2.3. From approximately 32 to 56 gestational days, the growth rates of both layers were not signi cantly different (trabecular myocardium 2 ± 7 µm/day, F = 0.87, 1 degree of freedom, compact myocardium 6 ± 3 µm/day, F = 0.79, 1 degree of freedom, difference p = 0.309). However, the trabecular thickness growth rate did not correlate with age (p = 0.387), while the compact wall did (p = 0.003). This indicates that only the trabecular thickness remained constant, causing the thickness ratio to decrease to below the LVNC threshold. Since the ventricles continuously expand, the compact and trabecular layer volumes have to grow to maintain a stable layer thickness, which is seen most clearly in late embryonic development where the trabecular layer thickness is stable and the volume grows (Fig. 1D), see also In mouse, the most used animal model for studying LVNC, and chicken, the animal model in which the process of compaction of the base of the trabeculae was demonstrated for the right ventricle 28 , the growth curves of the trabeculated and compact myocardium were similar to human. This was also illustrated by the transient peak in layer thickness ratio in the embryo (Fig. 2). No decrease in absolute trabecular layer thickness was seen (Fig. 2B), even if the trabecular layer was labelled to be excessive in younger stages and exceedingly regressed in older stages (Online Fig. 3). Although the post-natal mouse hearts appeared to show a larger variation in trabecular layer thickness between specimens than foetal human and late-gestational chicken, at no point was there a re-increase in thickness ratio (Fig. 2C) or a decrease in trabecular volume (Fig. 2D).
Since the trabecular and compact growth rate in foetal stages is similar, we hypothesised that, despite the difference in gross morphology, trabecular and compact myocardium may be phenotypically similar.

Trabecular and compact myocardial functional phenotype
To test for similarity in myocardial phenotype, we assessed the relative abundance of proteins related to ATP production (mitochondrial protein), force generation (sarcomeric protein) and blood supply (vascular endothelium) in adult human trabecular and compact myocardium. Since sarcomeric proteins occur in tissues with a xed stoichiometry 29 , values on cardiac troponin I (cTnI) are a good indicator of total sarcomeric protein in a cell. The staining signal was not different between trabecular and compact myocardium (p = 0.247, F = 62.24, 2 degrees of freedom), but greater than that of non-myocardial tissue (Fig. 3A). Similarly, there was no difference in the amount of mitochondrial protein COX4 (p = 0.857, F = 36.89, 2 degrees of freedom) (Fig. 3A), mitochondrial surface protein MAB1273 (p = 0.802, F = 63.56, 2 degrees of freedom) (Fig. 3B), and endothelial protein CD31 (p = 0.283, F = 86.67, 2 degrees of freedom) (Fig. 3C) based on signal intensity. The distance between capillaries was also similar indicating similar cardiomyocyte size (17.9 ± 1.3 µm in trabecular versus 18.9 ± 2.9 µm in compact myocardium, p = 0.284, T value = 1.237, 4 degrees of freedom, N = 5, two-tailed paired Student's T-test) and similar microvascular density. Only collagen was slightly more abundant in the trabecular layer, which, however, contained a few collagenous chordae tendineae (p = 0.038, F = 50.05, 2 degrees of freedom, Fig. 3D).

Trabecular and compact myocardial function
Despite the absence of differences in myocardial sarcomeric proteins, mitochondrial proteins and capillary densities, we next tested the hypothesis that the trabecular and compact myocardium have similar mitochondrial and contractile function. Myocardial work requires ATP and is therefore correlated with cellular oxygen consumption 30 . Therefore, we compared mitochondrial respiration in trabecular and compact muscle. In adult mouse hearts, the leak respiration was higher in trabecular than in compact myocardium (p = 0.033) and the left atrial wall (p = 0.027), but the NADH-linked respiration, maximal oxidative phosphorylation capacity and maximal uncoupled respiration did not differ between the two myocardial tissues (p = 0.999, p = 0.989, p = 0.997 respectively), while they did show higher mitochondrial respiration than the left atrial wall (Fig. 4A).
To assess the force generating capacity of trabecular and compact myocardium, we compared force production of isolated cardiomyocytes. There was no difference between adult compact and trabecular cardiomyocytes in maximum force production (F max , p = 0.959), average passive force, (F pas , p = 0.060), or calcium sensitivity (p = 0.075) (Fig. 4B-G). Trabecular myocardium is thought to be more embryoniclike than compact myocardium 31 , but both trabecular and compact adult cardiomyocytes had much greater maximum force production than foetal ventricular myocardium (p < 0.001 and p < 0.001, F = 19.83, 2 degrees of freedom) (Fig. 4C-D).

Trabecular and compact myocardium are similar in non-compaction
Because trabecular and compact myocardial function were similar, we speculated that the trabecular and compact myocardium may be equally affected in non-compaction. We found no signi cant differences between compact and trabecular myocardium regarding the COX4 (F = 51.64, 5 degrees of freedom,  5D). Even though these data do not provide evidence of altered mitochondrial or sarcomeric protein densities in LVNC, we found no indication that only the trabecular myocardium would be less equipped for contractile function in LVNC.

Discussion
Here, we show for the rst time that in the foetal period the left ventricular trabecular and compact myocardium are surprisingly similar in terms of growth, functional phenotype, and function. While this is surprising when premised on hypertrabeculation in LVNC causing heart failure, our ndings are fully consistent with the growing number of cohort studies showing no or only very slight correlations between extent of trabeculations and pump function 4,8,9 .
Differential growth rate determines left ventricular wall development Histological techniques showed continuous growth of both trabecular and compact myocardium during development in human, mouse, and chicken. This indicates that differential growth shapes the left ventricular wall. Differential growth also forms other chambers of the heart and, more generally, embryonic organs, limbs, head, etc. 32,33 . We found no evidence for compaction or a loss of myocardium of the trabecular layer, as would be expected when compaction would occur, though we did see dramatic changes to the LVNC-diagnostic criterium of wall ratio in the embryo. So, differential growth can lead morphometrically to LVNC positivity, in the absence of compaction. LVNC is, therefore, likely a misnomer.
Since there may be distinct phases in trabecular layer thickness growth, natural and pathological variation in the duration of these phases may impact on wall ratio. Because the ratio does not indicate which layer behaves abnormally, patients with LVNC may have either excessive trabeculae or an underdeveloped compact wall [34][35][36] . Also, a longer duration of fast trabecular growth does not necessarily imply inferiority of the trabecular muscle. In fact, in our cases of LVNC, the trabecular and compact layers appeared equally equipped with vasculature, mitochondria and sarcomeric protein.
No functional phenotypical differences between trabecular and compact myocardium Generating contractile force requires a high metabolic rate which has to be supported by the presence of a vast capillary network and a high mitochondrial density. There was no evidence that this was lacking in adult trabecular myocardium and it has long been demonstrated, for example, that the papillary muscles and trabeculae carneae are as vascularized as the neighbouring compact wall 37,38 . Likely, the initial differences between embryonic trabeculae and embryonic compact myocardium regarding functional phenotype disappear in the older heart as is also supported by the absence of an unambiguous marker for trabecular myocardium in the adult heart 24,[39][40][41] .

No functional differences between trabecular and compact myocardium
Our results indicate that adult cardiomyocytes from the trabecular and compact layer are equally equipped for force generation which is supported by in silico genetic and phenotypic correlation assays 42 . The maximal generated forces we described, corresponded to those in other species which have been inferred from systemic blood pressure, wall thickness and ventricular dimensions 43 . Foetal cardiomyocytes were found to be less strong than adult cardiomyocytes, again con rming that adult trabecular myocardium does not have an embryonic identity 24 . That the vast majority of people with hypertrabeculation, diagnosed as LVNC, are asymptomatic is consistent with the trabecular and compact myocardium having a similar contractile force potential 8,44,45 .

Limitations
Because of the morphometric analysis sensitivity it is possible that there is still a small contribution of compaction to the growth of the compact wall. However, our analysis excludes a large contribution, and, therefore, it is unlikely that non-compaction can lead to the extremely trabeculated ventricles observed in some patients with LVNC 20 .
Compaction has been previously described for chicken in the right ventricle 28 and has been suggested to occur in the left ventricle 46 . However, we could not replicate the same steep increase in compact wall thickness around day 10 in chick development as reported previously 46 even when arti cially shifting our trabecular to compact borderlines (Online Fig. 3).

Conclusions
We assessed left ventricular trabecular and compact layer growth in human, mouse and chicken. The left ventricular wall acquires its adult form by differential growth rather than by compaction (conversion) of trabecular to compact myocardium. LVNC is, therefore, likely a misnomer and its diagnosis on the basis of layer thickness ratio alone is uninformative on the overall ventricular morphology and functional state of myocardium. Because we found no difference in trabecular and compact myocardial functional phenotype and function, it appears that it is not trabecular weakness that causes heart failure in some patients with LVNC. Growth of the human heart. Absolute size increase of human hearts between 30 and 130 days showing that the trabeculated layer in foetal stages (red arrow) is much greater than the entire embryonic heart (A). In early chamber development, between 22 and 31 gestational days (P1), the trabecular layer thickness increases more than the compact (B) which is re ected in a thickness ratio that is LVNC criterion positive (C). In late embryogenesis, between 32 and 56 gestational days (P2), the trabecular layer thickness plateaus while the compact layer continues to thicken, this causes a drop in thickness ratio to below LVNC criterion levels. In the foetus (P3), the trabecular and compact layer thickness increase similarly which stabilises the thickness ratio. Calculated trabecular and compact volumes continuously increase in gestation, even in P2 when the thickness ratio drops dramatically (D).

Figure 2
Growth of the mouse and chicken heart. Absolute size increase of mouse hearts between 9.5 days of gestation and 23 days postnatally, and chicken hearts between 2 and 19 days of gestation (A). In both the mouse and chicken embryo the trabecular thickness rapidly increases. Towards the end of foetal development, even continuing till after birth in mouse (dotted line), the trabecular thickness stabilises.
The compact wall thickness continuously grows (B). In the embryo, the maximum peak ratio of median trabecular layer thickness over median compact layer thickness is reached while in the foetus the ratio lowers and stabilises (C). Calculated trabecular and compact volumes continuously increase (D).   . Sarcomeric protein cTnI signal intensity was signi cantly higher in trabecular than in compact myocardium in normal (p=0.0003) and LVNC hearts (p=0.001) and higher than in non-myocardium (all p<0.0001) (D). Statistical analysis: One-way ANOVA, Tukey corrected.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.