This article addresses selected aspects of the roles neural crest cells play in the development of the cardiovascular system. The neural crest is a fascinating structure for several reasons. For the developmental biologist it offers a unique system to address questions about the regulation of cell differentiation. Due to its extensive proliferation and wide range of migration within the embryo, the neural crest is vulnerable to environmental insults and genetic disorders. Many birth defects are neural crest defects. Examples include cleft lip/cleft palate, other craniofacial malformations, and malformations of the cardiac outflow tract. Childhood tumors and familial diseases include neuroblastoma, Hirschsprung's disease, neurofibromatosis, and complex neurocristopathies (Bolande, 1994). In many of these instances mutations in a gene that encodes either a growth factor or growth factor receptor are responsible for the developmental defect, illustrating the fact that growth factor action drives multiple steps in neural crest cell differentiation. This review addresses the progenitor cell composition of the cardiac neural crest and selected aspects of growth factor action in cardiac neural crest cell differentiation.
Whereas the heart itself is of mesodermal origin, components of the cardiac outflow tract are formed by the neural crest, an ectodermal derivative that gives rise to the peripheral nervous system, endocrine cells, melanocytes of the skin and internal organs, and connective tissue, bone, and cartilage of the face and ventral neck, among other tissues. Cardiac neural crest cells participate in the septation of the cardiac outflow tract into aorta and pulmonary artery. The migratory cardiac neural crest consists of stem cells, fate-restricted cells, and cells that are committed to the smooth muscle cell lineage. During their migration within the posterior branchial arches, the developmental potentials of pluripotent neural crest cells become restricted. Conversely, neural crest stem cells persist at many locations, including in the cardiac outflow tract. Many aspects of neural crest cell differentiation are driven by growth factor action. Neurotrophin-3 (NT-3) and its preferred receptor, TrkC, play important roles not only in nervous system development and function, but also in cardiac development as deletion of these genes causes outflow tract malformations. In vitro clonal analysis has shown a premature commitment of cardiac neural crest stem cells in TrkC null mice and a perturbed morphology of the endothelial tube. Norepinephrine transporter (NET) function promotes the differentiation of neural crest stem cells into noradrenergic neurons. Surprisingly, many diverse nonneuronal embryonic tissues, in particular in the cardiovascular system, express NET also. It will be of interest to determine whether norepinephrine transport plays a role also in cardiovascular development. Anat Rec Part A 276A:34–42, 2004. © 2004 Wiley-Liss, Inc.
RESULTS AND DISCUSSION
Formation of the Neural Crest
The neural crest is a transient embryonic structure that gives rise to many cell types and tissues of the adult vertebrate organism. Of particular interest for the purpose of this article are the neural crest cells from the vagal level that migrate through the posterior branchial arches to the cardiac outflow tract, where they participate in its septation into the aorta and pulmonary artery (Kirby et al., 1983). Neural crest cells originate in the neural folds at the dorsal aspect of the developing spinal cord. They delaminate from the neural tube and migrate via different pathways into the embryo (reviewed by Le Douarin and Kalcheim, 1999). The mechanisms that regulate neural crest cell formation and migration are currently being elucidated. They involve interactions between the somatic ectoderm and the neural ectoderm. Wnt6 is a prime candidate for a signaling molecule. Wnt6 is localized in the somatic ectoderm, and its inhibition perturbs neural crest formation (Garcia-Castro et al., 2002). Wnt1 and Wnt3a are involved in dorsal patterning of the neural tube, regulating the expansion of dorsal neural precursors (Ikea et al., 1997; Saint-Jeannet et al., 1997). In Xenopus, Wnt proteins synergize with Noggin, an inhibitor of bone morphogenetic protein 4 (BMP4), to generate the neural crest as assessed by an increase in the expression of slug, an early neural crest cell marker (Saint-Jeannet et al., 1997). In the avian embryo, BMP4 mRNA is homogeneously distributed along the longitudinal extent of the dorsal neural tube, whereas its inhibitor, noggin, is expressed in a rostro-caudally decreasing concentration gradient (Sela-Donenfeld et al., 1999). Noggin represses neural crest formation and delays neural crest emigration (Selleck et al., 1998; Sela-Donenfeld et al., 1999). In the mouse, Noggin is not essential for neural induction, but is required for the subsequent growth and patterning of the neural tube (McMahon et al., 1998). In summary, ectoderm-derived Wnt6 promotes neural crest formation. Furthermore, early BMP action induces expression of other Wnt proteins in the neural tube, which in synergy with BMP antagonists are also involved in neural crest formation. Conversely, late BMP action may be involved in promoting delamination of neural crest cells from the neural tube.
BMPs are also involved in altering stem cell properties. Neural crest stem cells are descendants of neuroepithelial stem cells, which reside in the neural tube. BMP2/4 causes neuroepithelial cells to become p75NTR immunoreactive and to differentiate into neural crest cell derivatives (Mujtaba et al., 1998).
Neural crest cells migrate via different pathways into the embryo and give rise to all neuronal and nonneuronal cells of the autonomic and enteric nervous systems, most primary sensory neurons, to endocrine cells (adrenal medulla, C cells of the thyroid), and to the pigment cells (melanocytes) of the skin and internal organs. Moreover, neural crest cells contribute to the septation of the cardiac outflow tract, and they form the cranial mesenchyme, which gives rise to dermis, bone, and cartilage of the face and ventral neck (reviewed by Le Douarin and Kalcheim, 1999).
Neural Crest Stem Cells
At the onset of migration, the neural crest forms a heterogeneous population of cells with regard to developmental potentials. At one extreme, there are neural crest stem cells that can give rise to most or all known neural crest derivatives. At the other end of the spectrum are various types of progenitor cells that are committed to a particular cell lineage. Furthermore, there are pluripotent neural crest cells that are more restricted in their developmental potentials than stem cells (Sieber-Blum and Cohen, 1980; Baroffio et al., 1988; Bronner-Fraser and Fraser, 1988; Sieber-Blum, 1989a; Ito and Sieber-Blum, 1991; Ito et al., 1993; Ito and Sieber-Blum, 1993).
The existence of pluripotent progenitors was shown by in vitro clonal analysis (Sieber-Blum and Cohen, 1980; Baroffio et al., 1988; Sieber-Blum, 1989; Ito and Sieber-Blum, 1991, 1993; Ito et al., 1993) and by labeling individual neural crest cells in vivo (Bronner-Fraser and Fraser, 1988). In both approaches it was found that an individual neural crest cell can give rise to an array of differentiated progeny, including sympathetic neurons, sensory neurons, nerve supporting cells, pigment cells, smooth muscle cells, chondrocytes, fibroblasts, and possibly other cell types (Ito and Sieber-Blum, 1991). Part of the definition of a stem cell is not only that it can generate one or more types of progeny but also that it is capable of generating new stem cells (self-renewal). The potential for self-renewal of rat neural crest stem cells was shown by serial cloning in vitro (Stemple et al., 1988).
By in vitro colony assay we have determined the relative frequency at the onset of migration of neural crest stem cells and lineage-committed cells between axial levels of somites 1–27 in the quail embryo (Fig. 1.) Neural crest cells emigrate from the neural tube progressively more caudally with advancing embryonic development. Thus by explanting neural tube segments from older embryos, neural crest cells from more caudal levels are obtained. In five sets of experiments we have cultured the last 5–7 segments of neural tube from Hamburger-Hamilton stages (Hamburger and Hamilton, 1951) 10, 11, 13, 14, and 17 to obtain neural crest cells from somitic levels 1–5, 6–11, 12–17, 18–22, and 22–27, respectively (Table 1). Eighteen hours postexplantation, the neural tubes were removed, the neural crest cells resuspended by trypsinization, and colony assays prepared as described (Sieber-Blum, 1999). By morphological criteria and with cell type-specific antibodies, four major types of colony can be identified. Colonies formed by stem cells contain pigment cells and several types of unpigmented cells, such as neurons, Schwann cells, smooth muscle cells, and other cell types (Sieber-Blum and Cohen, 1980; Sieber-Blum, 1989a). Colonies formed by committed melanogenic cells consist of pigment cells only. Cells that are committed to the smooth muscle cell lineage form colonies that consist entirely of smooth muscle actin-immunoreactive cells with mature smooth muscle cell morphology. Fate-restricted cells form colonies that lack pigment cells but contain several other cell types. The portion of committed smooth myogenic cells and committed melanogenic cells changes with different axial levels, whereas the frequency of stem cells is more evenly distributed. Committed smooth muscle progenitor cells form a rostro-caudally decreasing gradient from a maximum of 44.5% at somitic levels 1–5 down to 1.3% at somitic levels 22–27 (Fig. 1, Table 1). Committed melanogenic cells form an inverse, rostro-caudally increasing gradient, starting at 6.1% between somites 1–5 and increasing to 63.9% at somitic levels 22–27 (Table 1). By contrast, the distribution of stem cells is more even, at around 20% of the entire population, except for somitic levels 12–22, where it reaches almost 30% (Fig. 1, Table 1). The portion of colonies formed by fate-restricted cells is inversely correlated to the frequency of lineage-committed cells, suggesting that at least some lineage-committed cells are recruited from the pool of fate-restricted cells, rather than from stem cells. The distribution of committed myogenic and melanogenic cells along the neuraxis suggests that its formation is regulated by mechanisms that act at premigratory stages and that are associated either with gastrulation or with morphogens that are expressed in a similar graded pattern, such as Noggin.
|Neural tube segments (HH stage)||Colonies formed bya|
|Cells committed to smooth muscle cell lineage||Cells committed to melanogenic cell lineage||Stem cells||Fate-restricted cells (total minus SC, committed)|
|Somites 1–5 (stage 10)||44.5 ± 4.2||6.1 ± 1.6||20.1 ± 2.9||29.3|
|Somites 6–11 (stage 11)||5.4 ± 1.4||7.6 ± 1.1||20.2 ± 2.0||66.8|
|Somites 12–17 (stage 13)||0.35 ± 0.4||27.7 ± 5.1||28.8 ± 2.5||43.1|
|Somites 18–22 (stage 14)||1.8 ± 0.4||39.9 ± 1.8||29.7 ± 2.0||28.6|
|somites 22–27 (stage 17)||1.3 ± 0.4||63.9 ± 3.0||20.1 ± 1.3||14.7|
During advanced migration, the developmental potentials of neural crest cells become increasingly more restricted. For instance, cultured cardiac neural crest stem cells that have been cultured at the onset of migration can give rise to at least six phenotypes, smooth muscle cells, fibroblasts, chondrocytes, sensory neurons, autonomic neurons, and pigment cells (Ito and Sieber-Blum, 1991). In contrast, once they have migrated through the posterior branchial arches and have arrived in the cardiac outflow tract, they can no longer generate sensory neurons or melanocytes (Ito and Sieber-Blum, 1993; Sieber-Blum et al., 1993).
Neural crest stem cells persist in target locations (Fig. 2). When somatic ectoderm was explanted from day 3.5 quail embryos, neural crest cells emigrated onto the substratum. They were resuspended by trypsinization and replated at clonal density. At stage 21, 20% of clone-forming cells gave rise to mixed clones that contained pigment cells, neurons, and possibly other cell types, confirming their pluripotentiality (Richardson and Sieber-Blum, 1993; Sieber-Blum et al., 1993). The number of neural crest stem cells within the ectoderm decreased with progressing differentiation and reached 1% at stage 26. Similarly, dorsal root ganglia and sympathetic ganglia contain neural crest stem cells (Duff et al., 1991). At Hamburger-Hamilton stage 29 they constituted 7% and 5%, respectively, of colony-forming cells. Cardiac neural crest stem cells that can give rise to at least six cell types constitute 26% of colony-forming cells at the onset of migration (Ito and Sieber-Blum, 1991). At later stages, when they have arrived in the cardiac outflow tract, their abundance is comparable (23%), but they can no longer generate sensory neurons or pigment cells (Ito and Sieber-Blum, 1993; Sieber-Blum et al., 1993). Another type of progenitor cell within the quail outflow tract is at least tripotent, able to give rise to smooth muscle cells, chondrocytes, and fibroblasts. It is absent in the leading edge of the neural crest cell population that first arrives in the outflow tract at stage 19+/20. Its numbers, however, increase to 31% at stage 22 and to 67% at stage 24. It is thus thought that this type of neural crest progenitor cell is derived from stem cells through fate restriction and that it is relevant for outflow tract septation. More recently, neural crest stem cells have also been identified in the sciatic nerve and gut (Bixby et al., 2002; Kruger et al., 2002).
Neural crest stem cells change their properties in different target locations. Under the same culture conditions, stem cells from dorsal root ganglia generate 10 times more sensory neurons than sympathetic neurons. Conversely, stem cells isolated from sympathetic ganglia of the same embryo form clones that contain approximately 10 times more sympathetic neuroblasts than sensory neurons when cultured under the same culture conditions (Sieber-Blum et al., 1993). This observation indicates that the embryonic microenvironment modulates stem cell properties to increase their probability to generate site-appropriate cell types.
The Cardiac Neural Crest
Cardiac neural crest cells originate from the neural tube segment that extends from the midotic placode to somite 3 axial levels (Kirby et al., 1983). They migrate through the posterior branchial arches to the cardiac outflow tract and the proximal great vessels (Kirby and Waldo, 1990, 1995). Ablation of the cardiac neural crest in chick embryos results in cardiac outflow tract defects that resemble human congenital malformations (Kirby et al., 1983, 1985). They include persistent truncus arteriosus and transposition of the great vessels. In the outflow tract, the cardiac neural crest cells form two opposing spiraling ridges that grow together and eventually merge, dissecting the single outflow tract tube into two vessels, the aorta and the pulmonary artery (Sumida et al., 1989). A distinct population of either neural crest cells or ventrally derived rhombencephalic cells enters the heart at the venous pole. Although their ultimate fate is apoptosis, they may play important roles in the development of the conduction system (Poelmann and Gittenberger-de Groot, 1999). By contrast, the conduction system is not a neural crest derivative, but is derived from cardiac myogenic precursors (Cheng et al., 1999).
The Role of NT-3 and TrkC in Heart Development
Mice that have either a deleted neurotrophin-3 (NT-3) gene or deleted TrkC gene have multiple heart defects. Cardiac outflow tract defects include persistent truncus arteriosus and transposition of the great vessels (Tessarollo et al., 1997). To elucidate the putative neural crest defect in TrkC null mice, we have analyzed cardiac neural crest cell differentiation in wild-type and TrkC null mice (Youn et al., 2003).
NT-3 belongs to a family of neurotrophins that includes nerve growth factor (NGF), neurotrophin-4/5 (NT-4/5), and brain-derived neurotrophic factor (BDNF). They bind to the Trk tyrosine kinase receptors. NGF binds to TrkA, BDNF and NT-4/5 to TrkB, and NT-3 to TrkC. Under certain conditions and cellular contexts, NT-3 can also bind to and activate TrkA and TrkB (Barbacid, 1994). Bound NT-3 activates the TrkC kinase, leading to the recruitment, docking, and phosphorylation of signaling proteins, which activate several parallel signaling pathways that regulate proliferation, neuronal differentiation, and survival (Barbacid, 1994). TrkC null mice have a phenotype that is similar to that of NT-3-deficient mice, confirming that NT-3 acts predominantly through TrkC (Snider, 1994). Alternative splicing of the TrkC gene can generate several types of truncated TrkC isoforms, which lack the kinase domain (Tsoulfas et al., 1993; Valenzuela et al., 1993; Garner and Large, 1994). These noncatalytic receptors, which are well conserved among species, are thought to negatively modulate signaling by the catalytic receptor when the catalytic and truncated isoforms are co-expressed (Palko et al., 1999; Das et al., 2000). However, they may also have limited signaling capacity (Tsoulfas et al., 1993). Both, catalytic and truncated TrkC receptors are absent in the TrkC null mouse line (Tessarollo et al., 1997) that was used in our study (Youn et al., unpublished).
In contrast to other Trk receptors, TrkC is widely expressed in many nonneuronal tissues of the mouse embryo, including muscle, lung, kidney, heart, and vascular smooth muscle cells (Tessarollo et al., 1993; Donovan et al., 1996). The wide expression pattern suggests that NT-3 has pleiotropic functions during embryonic development (Tessarollo et al., 1993), including mitogenic action in cardiac myocytes during cardiac looping and establishment of ventricular trabeculation (Lin et al., 2000).
To elucidate the mechanisms that lead to neural crest-related cardiac outflow tract defects in TrkC null mice, we have performed in vitro clonal analysis of cardiac neural crest cell development. Based on their colony-forming ability we have identified three types of progenitor cells in wild-type mice, stem cells (CNC-SC; Fig. 3A and B), fate-restricted cells (CNC-RC; Fig. 3C), and cells that are committed to the smooth muscle lineage (CNC-smC; Fig. 3D; Youn et al., 2003). CNC-SC can give rise to at least five differentiated phenotypes, including smooth muscle cells, neurons, Schwann cells, chondrocytes, and pigment cells. Serial cloning showed that CNC-SC are capable of self-renewal, confirming their stem cell nature. Fate-restricted CNC-RC cannot generate neurons or pigment cells, whereas rare chondrocytes and Schwann cells were observed. CNC-smC form colonies that consist exclusively of smooth muscle cells, indicating that the colony-forming cell is committed to the smooth muscle cell lineage. The expansion rate, measured as number of cells per colony, differs between the three progenitor cell types with a ratio of approximately CNC-SC:CNC-RC:CNC-smC = 35:10:1. Thus, stem cells proliferate in culture significantly faster than progenitors that are committed to the smooth muscle cell lineage. In the mouse, approximately 77% of early migratory cardiac neural crest cells are committed to the smooth muscle cell lineage (CNC-smC; Youn et al., 2003). CNC-RC form 37% of the population, whereas stem cells comprise approximately 15%. These numbers are similar to those obtained with quail embryos (Fig.1, Table 1). In TrkC null mice, there is a change in progenitor cell composition. The portion of stem cells (CNC-SC) is decreased by half approximately, whereas the fraction of fate-restricted cells (CNC-RC) is increased by an equivalent degree. The subset of cells that are committed to the smooth muscle cell lineage (CNC-smC) is unchanged. The loss of stem cells and equivalent increase of fate-restricted cells suggests that the developmental potentials of stem cells become restricted prematurely. Whereas we have analyzed neural crest cells at the onset of their migration, the stem cell loss is likely to be larger in the neural crest cell population within the outflow tract, as proliferation and differentiation continue simultaneously with migration over a period of several days. Thus the precocious differentiation of stem cells into fate-restricted cells is likely to lead to at least two changes in outflow tract development. First, since the expansion rate of fate-restricted cells is significantly lower than that of stem cells, the overall number of neural crest-derived cells present in the TrkC null outflow tract may be reduced compared to that in the wild type. Second, there is an inverse relationship between migratory capacity and state of differentiation, which may lead to premature cessation of migration. The notion of premature cessation of migration and precocious cell differentiation is supported by the observation of a thickening of the TrkC null outflow tract wall. In addition, the morphology of the endothelial tube in the cardiac outflow tract is disrupted (Youn et al., 2003).
A Role of the Norepinephrine Transporter (NET) in Heart Development?
NET is a channel-like transmembrane protein that is characteristically expressed in adult noradrenergic neurons of the sympathetic nervous system and in the brain stem nucleus, locus ceruleus (Amara, 1995; Galli et al., 1996). In the mature nervous system, its function is the presynaptic reuptake of released norepinephrine in order to terminate excitation of the postsynaptic neuron (Axelrod, 1965; Iversen, 1967; Snyder, 1970). The same NET gene product is expressed in the locus ceruleus of the central nervous system, in sympathetic neurons of the peripheral nervous system, and in the adrenal gland (Amara, 1995).
Unexpectedly, however, NET was found to be expressed also very early and ubiquitously in both avian and mammalian embryos, indicating broader functions of the transporter (Ren et al., 2001, 2003). NET mRNA levels are regulated synergistically by autocrine NT-3, transforming growth factor-β1 (TGF-β1), and fibroblast growth factor-2 (FGF-2) action (Zhang et al., 1997a; Ren et al., 2001). Quail and mouse neural crest cells express NET protein even before onset of their emigration from the neural tube (Ren et al., 2003). NE transport promotes differentiation of neural crest cells (Zhang et al., 1997b) and locus ceruleus cells (Sieber-Blum and Ren, 2000) into noradrenergic neurons. Conversely, drugs that block NE uptake, such as tricyclic antidepressants or cocaine, block noradrenergic differentiation in vitro and in vivo (Sieber-Blum, 1989b; Zhang and Sieber-Blum, 1992; Zhang et al., 1997b; Sieber-Blum and Ren, 2000). While NET may not be functional during early neural crest migration, it is known that neural crest cells are capable of NE transport during advanced migration, when they arrive in the vicinity of the notochord (Rothman et al., 1978). Interestingly, the avian notochord has been shown to synthesize NE and other neurotransmitters before the emergence of any differentiated neurons in the central or peripheral nervous systems (Strudel et al., 1977). Moreover, serum and embryonic tissues contain nanomolar to micromolar amounts of NE and other neurotransmitters (Zhang and Sieber-Blum, 1992), which are thought to originate in the maternal bloodstream and enter the fetal bloodstream by crossing the placenta (Shuey et al., 1992). This observation and studies with adrenergic receptor agonists and antagonists (Zhang et al., 1997b) confirm that NE transport, rather than activation of adrenergic receptors, is responsible for promoting noradrenergic differentiation. These experimental observations have a human correlate. Infants that have been exposed to antidepressants or cocaine prenatally have a higher incidence of sudden infant death syndrome and autonomic disturbances, which appear to be due to a delayed maturation of noradrenergic neurons in the locus ceruleus and in the autonomic nervous system (Kandall et al., 1993; Hill and Tennyson, 1986).
Surprisingly, NET is expressed ubiquitously in nonadrenergic neurons and in nonneuronal cells both in the avian and mammalian embryos, as determined by immunohistochemistry, in situ hybridization, and reverse transcriptase polymerase chain reaction (RT-PCR) (Ren et al., 2001, 2003). NET-positive nonneuronal locations include the notochord, dermamyotome, kidneys, lung buds, intestinal tract, mesenchyme in the limb buds, and the cardiovascular system. In the heart, both the epicardium and the trabeculae express NET protein at high levels. Neural crest cells in the cardiac outflow tract, the endothelium of arteries and veins, as well as endothelial cells that are scattered in the embryonic mesenchyme express NET at high levels (Ren et al., 2001, 2003).
Does NET play a role in heart development? One of the most prominent roles of NE (and epinephrine) is its function in the regulation of cardiovascular physiology. Here, NET not only functions as a scavenger for circulating NE, but under some conditions, such as ischemia or exposure to indirect sympathomimetic drugs like tyramine or amphetamines, reverses and causes outward transport of NE (Paton, 1973; Langeloh et al., 1987; Schomig et al., 1987). Adrenergic receptor-mediated NE action maintains cardiovascular function in response to hypoxic stress in utero possibly through NET-mediated concentration and release of NE in the heart (Portbury and Chikaraishi, 2001). NE increases protein synthesis in adult cardiomyocytes (Pönicke et al., 2001), suggesting a growth factor-like function of the neurotransmitter.
NET and NE transport, as well as other transporters, such as the serotonin transporter (SERT), may impact embryonic morphogenesis. One common denominator of many NET-expressing cells is that they either form an epithelium or are derived from an epithelium (Ren et al., 2003). Lauder and collaborators have shown that serotonin promotes neural crest cell migration during craniofacial development and that inhibition of serotonin transport causes craniofacial malformations (Shuey et al., 1992; Moiseiwitsch and Lauder, 1995). Since many epithelial cells express NET or SERT, and many underlying mesenchymal cells express adrenergic or serotonergic receptors, we have proposed that in epithelia NET and SERT may function as a targeted delivery system of the neurotransmitters (Ren et al., 2003) (Fig. 4). Both neurotransmitters are present in high concentrations in fetal serum (Zhang and Sieber-Blum, 1992). Moreover, norepinephrine and epinephrine may be synthesized by cardiac cells (Ebert and Thompson, 2001). Epithelial cells exposed to blood or other fluids may take up the neurotransmitter and accumulate it. If the amine concentration gradient exceeds a certain threshold, the transporter may reverse and release the neurotransmitter into the subjacent mesenchymal space. Neurotransmitter receptors, which are expressed by many mesenchymal cells (Shuey et al., 1992; Wang et al., 1997), may in turn be activated and affect cellular processes (summarized in Fig. 4). Activation of norepinephrine or serotonin receptors expressed by embryonic cells can affect many growth processes, including cell division (Deng et al., 2000; Dao et al., 2001), apoptosis (Uchida-Oka et al., 2001), and cell differentiation (Fiorica-Howells et al., 2000). It will be of interest to determine if norepinephrine and NET are involved in heart development.
I thank Larissa Wiskowski for her help with the colony assays described in Table 1 and Alex Sieber for his advice on preparing the illustrations.