Magnetic resonance imaging of the regenerating neonatal mouse heart

We present longitudinal magnetic resonance imaging (MRI) of neonatal mouse hearts during the first three weeks following coronary artery ligation to mimic heart attack. We confirm heart regeneration in individual animals injured on post-natal day 1 (P1) while those injured on P7 show the adult response of fibrosis, scarring and impaired heart performance. We document heart growth and development of the principal functional cardiac parameters, and also remodeling during tissue regeneration as compared to fibrosis when imaging repeatedly up to 21 days after myocardial infarction (MI). We reveal compensatory changes in cardiac function with the restoration of tissue and resolution of injury for the P1 cohort and sustained injury responses for the P7 cohort. This study resolves the controversy surrounding neonatal mouse heart regeneration and establishes a functional platform for live capture of the regenerative process and for the future testing of genetic or therapeutic interventions.


2
We present longitudinal magnetic resonance imaging (MRI) of neonatal mouse hearts during the first three weeks following coronary artery ligation to mimic heart attack. We confirm heart regeneration in individual animals injured on post-natal day 1 (P1) while those injured on P7 show the adult response of fibrosis, scarring and impaired heart performance. We document heart growth and development of the principal functional cardiac parameters, and also remodeling during tissue regeneration as compared to fibrosis when imaging repeatedly up to 21 days after myocardial infarction (MI). We reveal compensatory changes in cardiac function with the restoration of tissue and resolution of injury for the P1 cohort and sustained injury responses for the P7 cohort. This study resolves the controversy surrounding neonatal mouse heart regeneration and establishes a functional platform for live capture of the regenerative process and for the future testing of genetic or therapeutic interventions.
Human patients who suffer a myocardial infarction (MI or "heart attack") are unable to replace the lost or damaged heart muscle (up to 25% of the myocardium) and instead a non-contractile fibrotic scar is laid down at the site of injury to prevent cardiac rupture. The result is compensatory function and pathological remodeling of the surviving myocardium (irreversible chamber dilation and wall thinning) which ultimately leads to congestive heart failure, the major contributor to morbidity and mortality worldwide 1 .
Strategies in regenerative medicine to replace lost myocardium include stimulating resident heart muscle cell (cardiomyocyte) division, direct reprogramming of fibroblasts into cardiomyocytes, and transplantation of cells that may either transdifferentiate into new heart muscle or, through paracrine effects, 3 promote repair and regeneration (reviewed in 2 ). In addition, studies of animal models which can intrinsically regenerate injured myocardium, such as the adult zebrafish 3 , have proven invaluable in identifying cellular and molecular mechanisms underlying this regenerative capability. In 2011, the first evidence of mammalian heart regeneration was reported by Hesham Sadek, Eric Olson and colleagues 4 , Following surgical resection of ~15% of the left ventricle apex of a one-day old (P1) neonatal mouse, the heart fully regenerated by 21-days post-injury, whereas if the procedure was repeated one week later on a post-natal day 7 (P7) mouse heart, fibrosis and scarring ensued, recapitulating the adult wound healing response. The mechanism of regeneration observed was analogous to that described in the adult zebrafish heart 5,6 , in the P1 mouse there was evidence of significant cardiomyocyte turnover and proliferation to restore damaged heart muscle, which was lost by P7.
Since the original study several others have described neonatal myocardial regeneration after resection and following alternative insults, such as cryo-injury 7 , and coronary artery ligation to invoke MI 8 . Others have documented roles for specific cell types such as tissue resident macrophages 9 or implicated important regulatory pathways such as Hippo signaling 10 in neonatal myocardial regeneration. Despite the apparent utility of the various models of neonatal heart injury, and the resulting tissue response, there is some controversy surrounding the extent of regeneration during the first weeks of life. It has been reported that, following apical resection, regeneration did not occur and was replaced by a fibrotic response 11 which persisted long-term (180 days) with extensive cardiac remodelling 12 . The differences between the results of Andersen et al. and those of the original report 4 were not due to genetic background, instead differences in the method of injury (i.e. removing up to 4 40% of the ventricle as opposed to ~15%) were suggested to be incompatible with regeneration 13,14 .
The major issue to-date with all of these studies is that they are based on sacrificing individual animals at specific time-points post-injury followed by histological assessment of the heart. This provides no insight into the extent of the initial injury following surgery, whether the heart was indeed injured from the outset or the regenerative process over time. Moreover, there remains an important question as to how the regenerating mammalian heart copes in terms of functional output and remodeling during the processes of scar resolution and tissue restoration, as is directly relevant to human ischemic heart disease patients subjected to regenerative therapies. To address both of these key issues, we have developed a non-invasive Magnetic Resonance Imaging (MRI) approach that enabled tracking of individual new-born mice from the first day after birth over time, and also after myocardial injury. While MRI is well established for quantifying and monitoring cardiac function in adult mice 15,16 , its longitudinal application in neonatal mice requires significant refinement, and has not yet been reported at the early P1 time point, or following neonatal heart injury.

Growth and development of the neonatal mouse heart and early cardiac function parameters
The non-invasive nature of MRI allowed us to track the growth and developing cardiac function in the same neonate over time starting from P1, with five scans performed over a period of three weeks (day 0 -baseline, days 4, 7, 14 and 21 after 5 injury/sham-operated controls), corresponding pup ages of P1, P5, P8, P15 and P22.
A similar analysis was carried out for our P7 cohort, providing data for P7, P11, P14, P21 and P28. The mouse pup was positioned in a modified MRI cradle for imaging (Supplementary Figure 1) and scanned for 35-45 minutes each time. The values of left ventricular mass to body mass ratio (LVM/BM), heart rate (HR) and ejection fraction (EF) for the sham-operated animals (which serve as our control group) are depicted in Figure 1, along with a body mass (BM) growth chart. BM increased linearly until day 21, after which the rate of increase was higher due to the pups being weaned. With recurrent sampling, we observed an increase in the growth rate of the P7 group vs the P1 group (Fig. 1a). It has been suggested that repeated administration of anesthesia may have a negative effect on body growth 16 , consistent with this the P1 group that received anesthesia and scanning earlier weighed less than the P7 group at three comparable time points (P7-8, P14-15, and P21-22). The mean weights for the P1 group were nearly 2.5g lighter than the P7 at the midpoint, though they appeared to recover with age.
Left-ventricular mass index (i.e. LVM normalized to body mass), increased over time within the P1 group (Fig. 1b), which is contrary to a previous study 15 that reported a decrease in LVM/BM with age. The differences between our findings and those previously reported 16 relate to the stages examined, in that we imaged early time points throughout the first three weeks of life when heart growth rate is maximal, whereas the previous study assessed older-aged animals of up to four months where the relative increase in LVM is reduced. The P7 group did not show this early increase and in fact revealed a steady reduction over time in LVM. 6 HR increased from P1 to P5 (Fig. 1c), likely caused by an increasing anesthesia tolerance of the maturing animal. Thereafter, we observed a decreased HR to a lowpoint at around P15, consistent with the widely recognized inverse relation of HR with age in human fetal, neonatal and infant periods 17 . This data contrasts with Wiesmann and colleagues who reported that HR measurements were similar over time, when measured in later stage neonates 16 .
The EF for the mice scanned at P1 (the earliest time point) was recorded at approximately 72% (Fig. 1d). This reduced to about 65% in 2-4 days, and then continued to reduce to the more adult heart-like value of approximately 60% during the following two weeks 14 . This data may reflect the many changes happening in heart development during the first week after birth, such as a reduction in relative cardiomyocyte proliferation and cell cycle exit 4 and the rapid expansion of the coronary vasculature 18 .

In vivo MRI
Next we analysed end-diastolic and end-systolic MRI cine images of P1 and P7 mice   2d).
There were no significant differences in heart rate (HR) between the injured P1 and P7 versus sham controls over the duration of imaging (Fig. 2e) 2f), consistent with an adult cardiomyocyte hypertrophy response, whilst no significant differences were observed between the P1 injured versus sham controls (Fig. 2f). The previous intact MRI study revealed that LVM/BM decreased with 10 increasing age at later stages 16 but the situation appears more complex during the first three weeks of growth: there was a reduction in LVM/BM immediately after birth during the first week, then an increase for the second week, followed by a further decrease after three weeks. This suggests that around 14 days post-MI (P15) there is a compensatory effect to increase LVM. This is consistent with a report that between P10 and P16 at baseline (no-injury) heart growth exceeds that of the body, and subsequently between P16 and P21, body growth exceeds that of the heart; which in turn has been attributed to a T3-hormone induced proliferative burst of cardiomyocyte numbers during the early preadolescence phase at P14 21 .
These data collectively reveal pathological remodeling of the P7 heart post-MI which was significantly reduced and transient in the P1 heart following injury; consistent with restoration of cardiac output and the increased regenerative capacity of the latter. This was further supported by the different responses in the injured P1 versus P7 hearts at Day 21 post-MI ( Figure 3). This endpoint data also revealed that, while the P1 injured neonate heart recovers the majority of function with similar values for all the measured cardiac parameters as matching sham-injured control hearts, there was some residual remodeling, as indicated by significantly higher EDV (Fig. 3a) that remained even after 21 days.

Left-ventricular function over time following injury
In order to further elucidate the evolution of left ventricular function post-surgery in both age groups, we quantified EDV, ESV, EF and LVM/BM over time in the same mouse ( Figure 4). In P1 mice these functional parameters were preserved in injured hearts relative to sham controls up to 21 days post-MI, with increases over the time course in both MI and sham groups reflecting heart growth ( Figure 1). In contrast, we observed significant differences in functional parameters between injured and sham hearts for the P7 group from day 4 post-injury until day 21. Notably, differences in EDV (Fig. 4a) and ESV (Fig. 4b)  in the P7, indicative of regeneration in the P1 but not the P7 group (Fig. 3g, h and Supp. Fig. 3a, b). The extent of initial scarring at Day 4 ( Supplementary Fig. 3b) and persistent scarring by Day 21 (Fig. 3i) in the P7 injured heart versus P1 was significant in both cases (P < 0.05).
In summary, we present a novel and significant adaptation of MRI to longitudinally assess tissue regeneration and cardiac function in a living mammalian model of heart regeneration. This provides definitive evidence for regeneration of the heart in the P1 mouse following injury and important insight into the output and remodeling of the regenerating heart over time, as directly compared to P7 animals which undergo a default adult wound-healing response. The regenerative capacity of the injured P1 mouse heart may simply be explained by harnessing the intrinsic growth responses that are ongoing following birth, however, the mechanisms of tissue growth and 12 replacement following injury are highly likely to be relevant for targeting in the adult situation.
The neonatal mouse MRI method described can be applied by all groups with small animal MRI capability, and delivers a platform for future testing of therapeutic agents to enhance myocardial regeneration for ultimate extrapolation to the diseased adult human heart. or sourced from the manufacturer (Agilent Technologies).

Methods
Animal preparation for imaging. Pups were housed with the mother until weaning.
Prior to examination by MRI, all pups were simultaneously removed from the mother and placed together in a heated enclosure at 32°C. In order to prepare for cine-MRI, anesthesia was induced in an anesthetic chamber using 4% isoflurane in 100%

Additional information
Competing interests: PRR is co-founder and equity holder in OxStem Cardio, an Oxford University spin-out that seeks to exploit therapeutic strategies stimulating endogenous repair in cardiovascular regenerative medicine. and P7 cohort; significant differences between injury and sham cohorts are indicated as *P < 0.05; **P < 0.01; ***P < 0.001 and ****P < 0.0001; n=5 or 6 for all groups.