Anatomy of the heart with the highest heart rate

Abstract Shrews occupy the lower extreme of the seven orders of magnitude mammals range in size. Their hearts are large relative to body weight and heart rate can exceed a thousand beats a minute. It is not known whether traits typical of mammal hearts scale to these extremes. We assessed the heart of three species of shrew (genus Sorex) following the sequential segmental analysis developed for human hearts. Using micro‐computed tomography, we describe the overall structure and find, in agreement with previous studies, a large and elongate ventricle. The atrial and ventricular septums and the atrioventricular (AV) and arterial valves are typically mammalian. The ventricular walls comprise mostly compact myocardium and especially the right ventricle has few trabeculations on the luminal side. A developmental process of compaction is thought to reduce trabeculations in mammals, but in embryonic shrews the volume of trabeculations increase for every gestational stage, only slower than the compact volume. By expression of Hcn4, we identify a sinus node and an AV conduction axis which is continuous with the ventricular septal crest. Outstanding traits include pulmonary venous sleeve myocardium that reaches farther into the lungs than in any other mammals. Typical proportions of coronary arteries‐to‐aorta do not scale and the shrew coronary arteries are proportionally enormous, presumably to avoid the high resistance to blood flow of narrow vessels. In conclusion, most cardiac traits do scale to the miniscule shrews. The shrew heart, nevertheless, stands out by its relative size, elongation, proportionally large coronary vessels, and extent of pulmonary venous myocardium.


| INTRODUC TI ON
Shrews are eutherian insectivore mammals that can be tiny. Adult Eurasian least shrew (Sorex minutissimus) and Eurasian pygmy shrews (Sorex minutus), which we study in this report, only weigh a few grams.
They occupy therefore the lower extreme of the seven orders of magnitude that mammals range in size. Because not much is known of their cardiac anatomy (Rowlatt, 1990;Vornanen, 1989), it is not clear whether typical traits of mammal hearts scale to such miniscule sizes. Valves and chamber wall thicknesses would be predicted to scale linearly to cavity size according to the law of Laplace (Jensen, 2021;Seymour & Blaylock, 2000) and typical proportions of valves, walls, and cavities might well scale to shrews. In contrast, resistance to blood flow is inversely related to vessel diameter raised to the power of four according to the Hagen-Poiseuille equation and very small arteries and veins in the shrew may impose high resistance.
Blood perfusion constantly must be critical in shrews since with the tiny body size also comes with the highest rates of massspecific metabolism among mammals. This metabolism is supported by the greatest mass-specific cardiac outputs, or volume of blood per gram tissue pumped per minute (Jurgens et al., 1996;Morrison et al., 1959). Whereas in human, it takes 1 min to circulate the entire blood volume, in a shrew it takes a few seconds (Schmidt-Nielsen, 1984). A key component of the prodigious circulation is the incredibly fast heart rate, which may exceed a thousand beats a minute in the smaller shrew species (Jurgens et al., 1996;Morrison et al., 1959;Nagel, 1986;Vornanen, 1992).
In many mammals, a well-developed cardiac conduction system initiates the cardiac impulse in the sinus node. While fast activation in the atriums is ongoing, activation is slowly propagated through the atrioventricular (AV) node and then rapidly spread throughout the ventricles via the His bundle, bundle branches and Purkinje fibers (Davies, 1942;Dobrzynski et al., 2013). Because shrews have extremely high heart rates, their conduction system may be extensive but this has not been investigated. Besides high heart rates, an additional component in achieving great cardiac output is likely a relatively large stroke volume because relative heart mass is substantially greater in shrews than in most other mammals (Bartels et al., 1979;Pucek, 1965;Vornanen, 1989).
Besides the role of scale on the structure of the shrew heart, phylogeny likely has an impact as well. Some traits of mammal hearts are universally shared such as two atriums and two ventricles (Rowlatt, 1990). Also, the valves of the aorta and the pulmonary artery always have three leaflets each and the left AV valve, or the mitral valve, always has two leaflets. Between mammals variation does exist, however, and the right AV valve can be dominated by a single large leaflet, have two leaflets, or have three leaflets such as in human. Other examples of variation is the number of caval veins connecting to the right atrium (RA) which is either two or three and the number of pulmonary veins connecting to the left atrium (LA) which varies even more (Kroneman et al., 2019). In short, across mammals, most structures exhibit some degree of variation, but it is not known where shrew hearts fall on such spectrums of variation. There is no consensus on how to analyze the gross anatomy of a mammal heart albeit a common approach is to describe structures in the same order as blood flows through the heart (Marais & Crole, 2021;Kareinen et al., 2020). One similarly ordered and broadly used approach for human hearts is the sequential segmental analysis which is sufficiently versatile to be applicable to crocodile hearts (Cook et al., 2017). Here we followed this manner of analysis.
Our specimens were caught in the wild in fall traps. By coincidence, three of the trapped shrews were pregnant females. From the embryos of these, we could study the development of the ventricular walls. Early cardiogenesis has been described for the house shrew and demonstrates the presence of highly trabeculated ventricles (Yasui, 1992;Yasui, 1993), while anatomical studies on adult animals of the closely related moles of the genus Talpa suggest the adult ventricle may have few trabeculations only (Rowlatt, 1990). Gestational changes to the extent of ventricular trabeculations have attracted much attention in the context of so-called "noncompaction" cardiomyopathy (Chin et al., 1990;Del Monte-Nieto et al., 2018;D'Silva & Jensen, 2020). Focus has been on a process of "compaction", whereby trabeculated muscle on the luminal side of the ventricle is added to the compact wall (Sedmera et al., 2000). There is very little quantitative evidence, however, for such a process (Faber, D'Silva, et al., 2021;Rychterova, 1971). Instead, there is much stronger quantitative support for the trabecular and compact layers can grow at different rates and when they do, it changes the proportion of trabecular-to-compact myocardium (Blausen et al., 1990;Faber, Hagoort, et al., 2021;Faber, Wüst, et al., 2021). If the ventricular walls of adult shrews have few trabeculations, while the embryonic walls are much trabeculated, shrews may be a good model system to test whether compaction or differential growth rates better characterize gestational changes to the trabeculated layer.
The primary research question of this study is whether typically traits of mammal hearts scale to the miniscule size of shrews.

| Animals
All animals were collected from near Yenisei Ecological Station Briefly, animals were collected from fall traps filled with water containing alcohol and then deskinned and fixed in either 10% or 20% formalin for one day and then kept in 70% ethanol until further use. Of the Eurasian pygmy shrew (S minutus), we used six heart-lung preparations for the description of the formed heart and four embryos from each of three pregnant females for the description of developing hearts (Table 1). The embryos were at three different times of gestation, which when staged on heart morphology, we assessed to correspond to mouse heart development ages embryonic day 10.5, 12.5, and 16.5. The actual gestational ages was probably a few days older since our youngest stage (10.5) had more developed limb buds than the oldest stages described by Yasui (Yasui, 1992;Yasui, 1993), which were approximately 12 days old. Besides specimens of the Eurasian pygmy shrew, we also used heart-lung preparations of formed animals of two Eurasian least shrew (S minutissimus) and two taiga shrew (Sorex isodon; Table 1). No embryos were found for these species.
The animals of the three species were either adult (overwintered 2017-2018 and sexually mature), juvenile (from 2018 and not sexually mature), or, in one case, subadult (from 2018, but sexually mature). The age and sexual maturity was based on wear of teeth and fur, and state of internal organs (Zaitsev et al., 2014).

| Micro-computed tomography
The three heart-lung specimens that were investigated with micro-CT, two S minutus and one S isodon, were first stained for 2 days with Lugol's solution (1.75 g I 2 and 2.50 g KI, both from Fischer Scientific, dissolved in 100 ml deionized water; Metscher, 2009).
The volume of Lugol's solution was at least 10 times that of the tissue and the solution was kept in the dark during the staining.
The specimens were then immobilized by imbedding in an agar solution (1.3 g per 100 ml water). Subsequently, they were scanned at isotropic 10 μm resolution using a Bruker, Skyscan 1272 (Blom et al., 2019). All shown images of micro-CT are of specimen 127 which had the best tissue-lumen contrast of the two S minutus specimens.

| Sectioning and histology
The hearts that were investigated with histology were embedded in paraplast and cut in 10 or 12 μm sections in either the transverse or frontal plane (four-chamber view). Staining was with saturated picro-sirius red in which muscle becomes orange and collagen becomes red following 2 min differentiation in 0.01 M HCl (Jensen et al., 2017). Imaging of the stained slides was done with a Leica DM5000 light microscope.

| Immunohistochemistry
With fluorescent immunohistochemistry, as done previously (Jensen et al., 2017), we detected smooth muscle actin TA B L E 1 Overview of the used specimens. #XXX is the specimen-specific code Scientific, RRID:AB_2536183). In addition, for identification of the sinus node we tested Isl1 (RRID:AB_2126323) and Shox2 (RRID:AB_945451) but we never detected positive nuclei. We did not seek to clarify whether the absence of signal was due to the absence of the protein or deterioration of the epitopes due to suboptimal tissue fixation and preservation. Slides were viewed and photographed with a Leica DM6000B fluorescent microscope. For estimations of the volume of trabecular and compact myocardium in embryos, we stained for cTnI as above on one section of 10 μm thickness per 100 μm (nine sections for specimen 699; eight sections for specimen 422; 14 sections for specimen 630).

| Analyses and statistics
We imported to Amira (version 2020.2, ThermoFisher Scientific) the micro-CT image series (three adult heart-lung specimens, one embryo of each of the three gestational ages) and the histology image series of three embryos of each of the three gestational ages and the images used for estimation of ventricular trabecular and compact tissue volume. Concerning S minutissimus, from the histological section series of the entire adult heart-lung preparation we imported to Amira every 25th section, in total 13 sections for specimen 368 and 14 sections for specimen 639. Structures of interest were then labelled and their volume was derived using the "Materials Statistics" tool. To measure distances of structures of the lungs, we imported images to ImageJ (Abràmoff et al., 2004; version IJ 1.46r) and used the straight line selection tool. We used a one-way ANOVA to test for differences in distance to the lung surface of pulmonary venous myocardium, pulmonary arteries, and alveolar ducts.

| Heart position and orientation
The body mass differed between the three investigated species (Table 1) and so did the size of the heart. The hearts were proportionally large, their tissue volume comprised more than 0.7% of fresh body mass (Table 2) and this percentage would likely have been greater still if fresh hearts had been measured (Vornanen, 1989).
Besides the difference in absolute size of the heart between the species, there were no gross morphological differences between the hearts. The position of the shrew heart is much like in human

| The RA and atrial septum
A very prominent pillar-shaped muscular ridge, the crista terminalis is The main trunk of the pulmonary artery is short and it bifurcates into one branch to each lung ( Figure 5). Just as the right lung is bigger than the left, the right branch of the pulmonary artery is bigger than the left. The right branch lies between the ascending and descending aorta and splits into four branches, one branch for each of the four lobes of the right lung (Figure 5a,b). The smaller solitary left pulmonary artery has an angle of approximately 75° to the main trunk and goes into the left lung which has one lobe only (Figure 5c,d).

| Pulmonary veins
The left-sided atrial body receives two very short stems of pulmonary veins, with the right stem coming from the first two lobes of the right lung and the left stem coming from the third and fourth lobe of the right lung together with the solitary vein of the left lung (Figure 1f). The right stem passes between the entrance of the RCCV and CCV and left stem passes over the coronary sinus to reach the LA.
A remarkable feature of the shrew heart is the extent of the myocardial sleeves of the pulmonary veins (Figure 7). In S minutissimus and S minutus these sleeves reach within 0.2 mm of the lung surface, even in the distal parts of the lungs (Figure 7c,d). Measured as the distance to the lung surface, the pulmonary venous myocardium extends as far as the pulmonary arteries and the terminal parts of the alveolar ducts (Figure 7c,d). Despite the great extent of the pulmonary venous myocardium, it only constitutes approximately 2% of total myocardial volume (Table 2). In S isodon the sleeves were also extensive, but they did not extend as far as the pulmonary arteries and the terminal parts of the alveolar ducts ( Figure 7c).

| Left atrium
The LA is situated dorso-cranially. It has a venous component (body), a trabeculated component (appendage), and a vestibule (Figure 8).
The body has the two orifices of the pulmonary veins. The appendage is proportionally large when compared to the human setting. It is clog-shaped and its junction with the body is relatively narrow (Figure 1f). The pectinate muscles are much more extensive in the left atrial appendage than the right atrial appendage (Figure 8). Even though the LA lacks the muscular bundle like crista terminalis in the RA, the junction between the appendage and body is well defined by the coarse trabeculation in the appendage (Figure 8).

| Left ventricle and aorta
The left AV valve has two leaflets and two papillary muscles (Figure 9).  (Crick et al., 1998). The left ventricular outflow tract is without trabeculations. The septal AV leaflet and fibrous continuity are the part of the outflow tract (Figures 6a and 9a).
There are three cusps to the aortic valve and each of the two coronary arteries arise from its own aortic sinus as in human Consequently, the ventricular surface does not have large coronary vessels and an interventricular sulcus that demarcate the left and RV as, for example, in human and pig (Crick et al., 1998).
The aorta and the pulmonary trunk have a spiral relationship (Figure 9g,h). The aortic arch crosses cranially to the bifurcation of the pulmonary arteries ( Figure 5d). The ascending aorta is located to the right of the trunk of the pulmonary artery. The aortic arch gives rise to only two branches leading cranially. The descending aorta lies between the esophagus and LCCV (Figure 5d).

| Development of ventricular trabeculation
From embryonic development to adulthood, the heart undergoes a great enlargement (Figure 10a). This includes the acquisition of a pronounced elongate shape already in intrauterine development

| The cardiac conduction system
Given the extremely high heart rates of shrews, it could be presumed that their hearts would contain a structurally pronounced conduction system. To investigate whether this was the case, we surveyed three histological series (Figure 11a,b). A sinus node was found in the parietal and ventral junction of the RCCV and the RA (Figure 11c).
It was identifiable by being node-like in appearance, by being relatively rich in collagen and by a low expression of cardiac troponin I ( Figure 11d). In addition, Hcn4 which is a key marker of the conduction system (Boyett et al., 2021), was expressed in a subset of the This myocardium therefore had the appearance of the penetrating bundle of His. Ventrally, this myocardium came into continuity with the myocardium of the ventricular septal crest (Figure 11h). We did not identify structures that resembled the AV node or bundle branches, but many of the sections that could have contained these structures also had artefacts from blood and connective tissue that had come loose during the staining procedures.

| DISCUSS ION
The shrew heart is in most ways a typical mammal heart, despite being at the extreme lower end of the seven orders of magnitude that mammals range in size. Its exceptional traits can be summarized as its large relative size and great elongation, which have been reported before (Vornanen, 1989), and, as we show here, the extreme extent of the pulmonary venous myocardium. Sleeves of myocardium around the pulmonary veins are found in many mammals including human (Rowlatt, 1990). The sleeves are typical confined to the parts that are most proximal to the LA and they are thought to act as throttle valves of pulmonary venous return (Nathan & Gloobe, 1970). In the shrew, however, there is almost no part of the pulmonary venous tree that is without a myocardial sleeve. It is therefore difficult to envision the role of throttle valve to this myocardium. In human, ectopic pacing of the atria often originates from the pulmonary venous myocardium and this setting requires pulmonary vein isolation by catheter-ablation. While electrocardiograms have been reported for shrews (Jurgens et al., 1996;Morrison et al., 1959;Nagel, 1986;Vornanen, 1989), it is not known whether their extensive pulmonary venous myocardium associates with a propensity to develop atrial arrhythmias.

| Differences between species of shrew
The primary aim of this study was to identify the traits that species of shrew share and how these relate to the same traits in other mammals. Our data also reveal, however, some degree of difference between the three investigated species. First, the absolute size of heart is, unsurprisingly, greater in the heavier species, S isodon.
Second, in S isodon the pulmonary venous myocardium appears to extend less far into the lungs than the pulmonary artery, whereas in S minutus and S minutissimus the extent of the two structures was not different. The observations on the lung tissues, however, is based on few specimens only and had the study included more specimens of S isodon and S minutissimus we may have been able to detect more differences. Overall, the hearts of the investigated species appear highly similar, even in their highly unusual traits such as the proportional size and elongation of the heart.

| Traits that shrews share with most mammals
Shrews and most mammals have three caval veins. The human setting of having a regressed left caval vein is less common (Carmona et al., 2018;Jensen, Boukens, et al., 2014;Rowlatt, 1990). Compared to human, the caval vein myocardium is extensive, but many mammals have similarly extensive myocardium, that is, the myocardium extends to the vicinity of the pericardial border (Jensen, Boukens, et al., 2014;Nathan & Gloobe, 1970). The number of pulmonary veins that connect to the LA is more variable in mammals than in other tetrapods (Kroneman et al., 2019) and two veins, as in shrews, is not uncommon (Rowlatt, 1968). The AV valves and tricuspid pulmonary and aortic valves of the shrews were typical of mammals (Rowlatt, 1990). In monotreme and marsupial mammals, the right AV valve can be dominated by the parietal leaflet (Lankester, 1882;Runciman et al., 1992), but in the shrews the septal leaflet was welldeveloped as is common in eutherian mammals (Rowlatt, 1990).
Eutherians are distinct from other mammals by having a second atrial septum which leaves a circular depression called the oval fossa on the right face of the atrial septum Röse, 1890;Rowlatt, 1990). The shrews also have an oval fossa and its dorsocranial rim is provided by a fold in the atrial roof in a manner that much resembles the human setting (Anderson et al., 2014).
Given the extremely high heart rates of shrews, one could presume an unusual cardiac conduction system. The cardiac conduction system has been identified on the basis of histological characters such as weak stain and richness in collagen (Davies, 1942;Ho et al., 2003) and on the presence of the funny current channel, Hcn4 (Boyett et al., 2021;Sizarov et al., 2011).
We show in the shrew the presence of myocardium that is insulated by collagen and that expresses Hcn4 where a mammal sinus node and His bundle would be expected. A lot of our histology was perturbed by loose and displaced connective tissue and blood, and this hampered the identification of the AV conduction axis in particular. We therefore suggest that the absence of a clear identification of an AV node and bundle branches in our data should not be seen as strong evidence for the absence of these structures.
From the data we have, we consider it unlikely that the shrew cardiac conduction system is extensive.

| Unusual traits of the shrew heart
A right ventricular wall and septum almost free of trabeculations is not typical of mammals but it is found in shrews, as we report here, and bats, squirrels, and mustelids (Rowlatt, 1990). In many ungulates, for example, the LV can have few trabeculations only (Jensen et al., 2016;Rowlatt, 1990) and the shrew LV is also somewhat sparse in trabeculation. Across vertebrates it is always so that the ventricles of embryos are highly trabeculated (Jensen et al., 2016), which has also been documented in shrews (Yasui, 1993) and we confirm here. So a proportional change must take place from the embryonic and highly trabeculated setting to the adult setting. Two different processes have been proposed to explain such proportional change.
One is that the trabecular and compact layers grow throughout development, but the growth rate of the two layers may differ periodically which then changes the layer proportions (Faber, Wüst, et al., 2021). In agreement with this view, our data show the compact layer has a greater rate of growth than the trabecular layer and this reduces the proportion of trabecular muscle during gestation. The other proposed process is compaction, whereby trabeculations are removed from the trabecular layer and added to the compact wall (Chin et al., 1990;Rychterova, 1971). No decrement of the trabecular layer thickness has been documented, however (Faber, D'Silva, et al., 2021), even though this is a predicted outcome if compaction takes place. In this light, the shrew ventricles are an interesting test for the hypothesis of compaction, because their right ventricular wall is comparatively very smooth. The developmental data, however, do not support a role of compaction but it does support a role of differential growth rates.

| Unusual traits that may be ascribed to small size
One unusual trait of the shrew hearts was the little amount of fibrofatty tissue that comprised the insulating plane between the atria and ventricles. In the left AV junction, the insulation was so meagre that it was difficult to assess whether the atrial and ventricular myocardium was in fact insulated from each other. Irrespective of the extent of insulation by fibro-fatty tissues, myocardium that slowly propagates the electrical impulse normally occupies the AV junction and this provides a degree of AV insultion (Aanhaanen et al., 2011;Rentschler et al., 2011). While the relative importance of connective tissue and myocardium with slow propagation cannot be deduced from histology, a functional insulation is present since shrews exhibit a delay between atrial and ventricular activation (Jurgens et al., 1996;Morrison et al., 1959;Nagel, 1986;Vornanen, 1992). In larger animals, millimeters of connective tissue separate the atrial and ventricular myocardium (Ho et al., 2003). large impact on resistance. In this light, the proportions of aorta-tocoronary-arteries found in human, for example, may not scale to the small size of shrews, because it would reduce the diameter of the coronary arteries so much as to render them high-resistance vessels.

| Highly unusual traits of the shrew heart
A very elongate ventricle is a highly unusual trait in a mammal and it is found in shrews (Vornanen, 1989) and the closely related moles of the genus Talpa (Rowlatt, 1968). Otherwise, elongate ventricles appear restricted to animals that are highly elongate themselves, such as snakes and caecilian amphibians (de Bakker et al., 2015;Ramaswami, 1944). Shrews and moles are not particularly elongate mammals and the significance of their highly unusual heart shape is not clear. Heart shape may affect function according to the law of Laplace (Vornanen, 1989), but it is not implausible that heart shape is also an outcome of structural restrictions offered by the ribcage, diaphragm and maybe even the lungs. The early developing hearts that were investigated here, and previously (Yasui, 1993), were not much different in shape from developing mouse hearts, whereas the hearts in late gestation were essentially shaped like the adult heart. It is then likely that the shrew heart grows into its unusually long shape, rather than it first acquires a typical mammal fetal shape and then remodels.
Perhaps the most unusual morphological trait of the shrew heart is the extent of the pulmonary venous myocardium. It was previously demonstrated that pulmonary venous myocardium is present in shrews (Endo et al., 1997) and in mice, for example, myocardial sleeves extend three bifurcations up the pulmonary venous tree (Mommersteeg et al., 2007). To the best of our knowledge, however, our report is the first documentation of the extreme extent of the shrew pulmonary venous myocardium which much exceeds that of mouse. The extreme extent aside, the amount of muscle around the pulmonary veins is not much, it is just a few percent of the total cardiac mass. Therefore, even if cardiac muscle is energetically very demanding (Mootha et al., 1997), the pulmonary venous myocardium may not necessarily impose a large metabolic cost and its advantage to organismal performance may not have to be great.
Observations on valves in veins have a long history (Franklin, 1927), but to the best of our knowledge this is the first report of a valve on the myocardial-venous boundary in the CCV.
Whales have a sphincter at the same position (Lillie et al., 2018), but not a valve of leaflets. Even in reptiles, where the caval vein myocardium functions as a chamber and a valve in the CCV would seem advantageous (Jensen et al., 2017;Joyce et al., 2020), there is no valve. Because such valve is rarely sought after and it may easily be obscured by coagulated blood or by its collapse against the vessel wall, a dedicated investigation may be required to establish whether a valve in the CCV is highly unusual among vertebrates.
That the pulmonary and systemic veins contains much myocardium is consistent with them functioning as contractile chambers. The presence of the valve in CCV is also consistent with the intrapericardial caval veins functioning as a chamber, as the valve likely prevents regurgitation during contraction. In reptiles, the caval vein myocardium is electrically activated well before the atria and this aids the filling of the RA (Jensen et al., 2017). If conditions are favorable, the electrical activation of the caval vein myocardium can be detected on the body-surface electrocardiogram as a unique deflection, the SVwave, and which occurs before the P-wave which heralds the activation of the atria (Jensen et al., 2017;Mullen, 1967;Valentinuzzi & Hoff, 1972). While a P-wave has been detected on electrocardiograms of shrews (Jurgens et al., 1996;Morrison et al., 1959;Nagel, 1986;Vornanen, 1992), no study has reported a SV-wave. This is not necessarily evidence against the venous myocardium of shrews being activated before the atrial muscle, since the elicited potentials are likely very small and only a small fraction of electrocardiograms of reptiles, for example, exhibit a SV-wave (Boukens et al., 2019;Mullen, 1967).
Possibly, the systemic and pulmonary venous myocardium in the shrews function as chambers that aid the filling of the right and LA respectively, but we consider it more likely that this myocardium is activated and contracts together with the myocardium of the atria as in other mammals (Jensen, Boukens, et al., 2014).

| CON CLUS ION
In this study, we investigated whether typically traits of mammal hearts scale to the extremely small size of shrews and key anatomical structures, such as valves and septums, do scale. Traits that do not scale may be the proportionally very large coronary arteries and the AV junction insulation, which comprise very little tissue. The pronounced elongation of the ventricle, the extreme extent of the pulmonary venous myocardial sleeves, and the valve in the CCV may set shrew hearts aside from other mammal hearts.

ACK N OWLED G M ENTS
The authors thank Antonina Yu. Alexandrova for the assistance with collecting shrews, Quinn Gunst and Corrie de Gier-de Vries for their assistance with histology and immunohistochemistry, and Jaco Hagoort for his assistance with Amira. The authors have no conflict of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.