help button home button Endocrine Society Endocrine Reviews
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hinson, J. P.
Right arrow Articles by Smith, D. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hinson, J. P.
Right arrow Articles by Smith, D. M.
Endocrine Reviews 21 (2): 138-167
Copyright © 2000 by The Endocrine Society

Adrenomedullin, a Multifunctional Regulatory Peptide1

Joy Patricia Hinson, Supriya Kapas and David Michael Smith

Department of Molecular and Cellular Biology (J.P.H.) and Clinical Sciences Research Centre (S.K.), St. Bartholomew’s and the Royal London School of Medicine and Dentistry, Queen Mary and Westfield College, University of London, London E1 4NS; and Department of Metabolic Medicine (D.M.S.), Division of Investigative Sciences, Imperial College School of Medicine, Hammersmith Hospital, London W12 0NN United Kingdom


    Abstract
 Top
 Abstract
 I. Introduction
 II. Synthesis and Secretion...
 III. Receptors and Signal...
 IV. Biological Actions of...
 V. Unresolved Issues and...
 References
 
Since the discovery of adrenomedullin in 1993 several hundred papers have been published regarding the regulation of its secretion and the multiplicity of its actions. It has been shown to be an almost ubiquitous peptide, with the number of tissues and cell types synthesizing adrenomedullin far exceeding those that do not. In Section II of this paper we give a comprehensive review both of tissues and cell lines secreting adrenomedullin and of the mechanisms regulating gene expression. The data on circulating adrenomedullin, obtained with the various assays available, are also reviewed, and the disease states in which plasma adrenomedullin is elevated are listed. In Section III the pharmacology and biochemistry of adrenomedullin binding sites, both specific sites and calcitonin gene-related peptide (CGRP) receptors, are discussed. In particular, the putative adrenomedullin receptor clones and signal transduction pathways are described. In Section IV the various actions of adrenomedullin are discussed: its actions on cellular growth, the cardiovascular system, the central nervous system, and the endocrine system are all considered. Finally, in Section V, we consider some unresolved issues and propose future areas for research.

I. Introduction
II. Synthesis and Secretion of Adrenomedullin
A. Structure and synthesis of adrenomedullin
B. Circulating adrenomedullin: adrenomedullin assays
C. Circulating adrenomedullin in disease
D. Origins of circulating adrenomedullin
E. Metabolic clearance of adrenomedullin
F. Adrenomedullin in other biological fluids
G. Regulation of adrenomedullin gene expression and peptide synthesis in vivo
H. Experimental regulation of adrenomedullin gene expression and peptide synthesis in vitro
III. Receptors and Signal Transduction
A. Do CGRP receptors mediate the effects of adrenomedullin?
B. Are there specific adrenomedullin receptors?
C. The pharmacology of specific adrenomedullin receptors
D. Receptor biochemistry: chemical cross-linking of adrenomedullin and CGRP receptors
E. Receptor biochemistry: molecular characterization of adrenomedullin and CGRP receptors
F. Signal transduction pathways activated by adrenomedullin
IV. Biological Actions Of Adrenomedullin
A. Vascular actions
B. Growth and development
C. Endocrine effects
D. Renal effects
E. Other peripheral effects
F. CNS effects
V. Unresolved Issues and Future Perspectives


    I. Introduction
 Top
 Abstract
 I. Introduction
 II. Synthesis and Secretion...
 III. Receptors and Signal...
 IV. Biological Actions of...
 V. Unresolved Issues and...
 References
 
A NEW regulatory peptide was discovered when a group of scientists in Japan were screening a panel of peptides extracted from a pheochromocytoma. They were looking for biological activity by testing whether the peptides could raise platelet cAMP levels. The scientists found a peptide with this activity, purified and sequenced it, and termed it "adrenomedullin" as it was derived from the adrenal medulla. The first paper on adrenomedullin, published in April 1993, described not only the purification of this peptide, but also its action on blood pressure, and the development of a specific RIA to measure circulating adrenomedullin (1 ). Three months later the gene encoding human adrenomedullin was sequenced (2 ), followed by the rat gene in September (3 ). Within 2 yr plasma adrenomedullin had been measured in a wide range of clinical conditions, the first candidate receptor for adrenomedullin had been identified (4 ), and a new field of endocrine research had begun.

The growth of interest in adrenomedullin has been exponential, with more than 600 papers published in this field to date including a number of reviews (5 6 7 8 9 10 ). This review aims to summarize the present state of our knowledge of adrenomedullin biology, and to focus on issues that are currently unresolved, with an indication of likely areas for future research.


    II. Synthesis and Secretion of Adrenomedullin
 Top
 Abstract
 I. Introduction
 II. Synthesis and Secretion...
 III. Receptors and Signal...
 IV. Biological Actions of...
 V. Unresolved Issues and...
 References
 
A. Structure and synthesis of adrenomedullin
Human adrenomedullin is a 52-amino acid peptide with a single disulfide bridge between residues 16 and 21 and with an amidated tyrosine at the carboxy terminus (1 ). It shows some homology with calcitonin gene-related peptide (CGRP) (1 ) and has therefore been added to the calcitonin/CGRP/amylin peptide family. Rat adrenomedullin has 50 amino acids, with 2 deletions and 6 substitutions compared with the human peptide (3 ). Porcine adrenomedullin is nearly identical to the human peptide, with only a single substitution (Gly for Asn) at position 40 (11 ). The sequences for both canine (12 ) and bovine (13 ) adrenomedullin have also been elucidated, and a comparison of the amino acid sequences of adrenomedullin from different species is shown in Fig. 1Go.



View larger version (23K):
[in this window]
[in a new window]
 
Figure 1. A comparison of the amino acid sequences of adrenomedullin from different species. Residues are shown as single amino acid codes and compared with the human sequence. -, Deleted residue; X, substituted residue; x, assumed residue in incomplete sequence for dog.

 
Adrenomedullin is synthesized as part of a larger precursor molecule, termed preproadrenomedullin. In both rat and human this precursor consists of 185 amino acids (2 3 ), while the porcine precursor has 188 residues (11 ). Preproadrenomedullin contains a 21-amino acid N-terminal signal peptide that immediately precedes a 20-amino acid amidated peptide, designated proadrenomedullin N-terminal 20 peptide or PAMP (2 ). The biological actions of PAMP have been reviewed elsewhere (14 ) and will not be discussed further here. It has also been suggested that a further biologically active peptide, termed adrenotensin, may be a product of the adrenomedullin gene (15 ), but this awaits confirmation.

The gene encoding preproadrenomedullin is termed the adrenomedullin gene and has been mapped and localized to a single locus of chromosome 11 (16 ). The human adrenomedullin gene comprises 4 exons and 3 introns, with TATA, CAAT and GC boxes in the 5'-flanking region (16 ) (Fig. 2Go). There are several binding sites for activator protein-2 (AP-2) and a cAMP-regulated enhancer element (16 ). It has also been found that there are nuclear factor-{kappa}B (NF-{kappa}B) sites on the promoter of the adrenomedullin gene (16 ). The organization and chromosomal localization of the murine adrenomedullin gene have also been elucidated (17 ).



View larger version (23K):
[in this window]
[in a new window]
 
Figure 2. Schematic presentation of the adrenomedullin gene, pre-proadrenomedullin, and biosynthesis of adrenomedullin. The genomic DNA of human adrenomedullin comprises four exons and three introns, and mature adrenomedullin peptide is coded in the fourth exon. Arrows indicate known regulatory sequences in the 5'-noncoding and intron 1 region of the human adrenomedullin gene.

 
The adrenomedullin gene is expressed in a wide range of tissues. The initial report on the distribution of adrenomedullin mRNA suggested that the highest levels of expression were seen in the adrenal medulla, ventricle, kidney, and lung (2 ). Since the discovery that the adrenomedullin gene is more highly expressed in endothelial cells than even in the adrenal medulla (18 ), this peptide has come to be regarded as a secretory product of the vascular endothelium, together with nitric oxide (NO) and endothelin. Clearly, therefore, adrenomedullin expression is seen in all tissues of the body, and comparisons between tissues may simply reflect varying degrees of tissue vascularity. However, evidence from both immunocytochemistry and studies with cultured cell lines reveals that the adrenomedullin gene is expressed by many different cell types, in addition to vascular endothelial cells (Table 1Go). In addition, many tumor cell lines express the adrenomedullin gene or have been shown to synthesize the immunoreactive peptide (58 ). Although adrenomedullin is often described as a ubiquitous regulatory peptide, it is worth noting that certain cell types and cell lines do not appear to express the adrenomedullin gene (Table 2Go). It is widely recognized that adrenomedullin is a product of vascular endothelial cells, but it appears that not all vascular endothelial cells synthesize the peptide. In some tissues, although adrenomedullin immunostaining has been demonstrated in certain cells, the vascular endothelial cells and smooth muscle cells remain unstained. This is particularly noticeable in the rat adrenal gland (25 29 ).


View this table:
[in this window]
[in a new window]
 
Table 1. Cell types that have been found to express the adrenomedullin gene or to synthesize immunoreactive adrenomedullin

 

View this table:
[in this window]
[in a new window]
 
Table 2. Cell lines/cell types/tissues that do not express the adrenomedullin gene or synthesize immunoreactive adrenomedullin

 
The question as to whether adrenomedullin, in common with many other regulatory peptides, is stored by some cell types has been raised. There is evidence that adrenomedullin is stored in secretory granules in the pancreas (54 ), but to date other endocrine tissues have not been investigated at the electron microscope level. When the adrenomedullin content of cultured cells is compared with the peptide concentration in the culture medium, there is no evidence for intracellular storage of the peptide (27 ), and it is thought that adrenomedullin is constitutively secreted (59 ).

B. Circulating adrenomedullin: adrenomedullin assays
After the initial report of picomolar levels of adrenomedullin circulating in plasma (60 ), several research groups have developed in-house assays to measure plasma adrenomedullin levels. In general these assays appear to have been carefully validated, with evidence presented from HPLC analysis to show that immunoreactive adrenomedullin from human plasma coelutes with authentic human adrenomedullin1–52 (60 61 62 ). There is a remarkable consistency between these different methods in terms of the absolute concentrations of adrenomedullin reported in the circulation of healthy controls (see Table 3Go). We can therefore conclude with some certainty that the normal plasma concentration of adrenomedullin is in the range of 1 to 10 pM, with most values between 2 and 3.5 pM. There do not appear to be significant differences between males and females or between different age groups, although to date these questions have not been directly addressed. It has been suggested that the Peninsula assay (Peninsula Laboratories, Inc., Belmont, CA) may overestimate the levels of adrenomedullin (63 ), and although a comparison of the values obtained with the Peninsula assay in different laboratories shows that these tend to be higher than values obtained with other assays (see Table 3Go), these values are still within the range of 1 to 10 pM. The sole exception is the value reported by Hata and co-workers (77 ), which is clearly outside the normal range, although the reason for this is unclear.


View this table:
[in this window]
[in a new window]
 
Table 3. Comparison of plasma adrenomedullin levels measured in healthy control subjects obtained using different immunoassays

 
There is evidence to suggest that the major circulating form of adrenomedullin is a carboxy-terminal glycine-extended peptide, which is converted to mature adrenomedullin by enzymatic amidation (78 ). It is thought that this glycine-extended peptide, termed the intermediate form, is the peptide that is processed from preproadrenomedullin. The plasma concentration of glycine-extended adrenomedullin (iAM) has been reported to be 2.7 ± 0.18 pM, with mature adrenomedullin (mAM) present at a concentration of 0.48 ± 0.05 pM (78 ). It has been observed that in congestive heart failure both iAM and mAM are similarly increased: the ratio of iAM to mAM does not significantly change (79 ). As far as it may be ascertained from the range of values reported, it appears likely that most assays used for the determination of plasma adrenomedullin are measuring either iAM or total adrenomedullin. Unless future studies demonstrate significant divergence between mAM levels and total adrenomedullin, this situation would appear to be satisfactory. One factor that may challenge this assumption is the recent discovery of adrenomedullin serum-binding protein (AMBP-1), which may affect adrenomedullin bioavailability (80 ).

Adrenomedullin has also been measured in rat plasma, using an antiserum that recognizes both rat and human forms of adrenomedullin. Adrenomedullin levels in rat plasma were comparable with those measured in man, at 3.6 ± 0.34 pM (81 ).

C. Circulating adrenomedullin in disease
Adrenomedullin has been measured in a wide range of disease states (see Table 4Go). In many cardiovascular disorders, plasma adrenomedullin is reported to be elevated, possibly suggesting that increased adrenomedullin is part of the homeostasis of blood pressure, released to compensate for elevated blood pressure. The finding that adrenomedullin is lower in preeclampsia compared with uncomplicated pregnancies may suggest that adrenomedullin is involved in the pathogenesis of this disorder (77 ). These findings have been questioned, however, on the basis of the exceptionally high adrenomedullin levels measured in the healthy controls in this study (77 ) (see Table 3Go). A more recent study has reported no difference in plasma adrenomedullin between preeclamptic and normotensive pregnant women, although adrenomedullin concentrations in amniotic fluid were found to be higher in preeclampsia (89 ).


View this table:
[in this window]
[in a new window]
 
Table 4. Disease states associated with elevated plasma adrenomedullin

 
In general, it appears that elevated adrenomedullin is a consequence rather than a cause of the pathology. It is unclear whether systemic increases in adrenomedullin reflect an overflow from local sites of production and action, or whether in certain conditions increased plasma adrenomedullin has a hormonal function causing a general decrease in vascular resistance and a fall in blood pressure. In cirrhosis, a progressive increase in adrenomedullin has been reported with increasing severity of the disease (64 ). A positive correlation has been demonstrated between PRA, aldosterone concentration, and adrenomedullin (64 101 103 ), with a negative correlation between adrenomedullin and glomerular filtration rate (101 ), creatinine clearance, and sodium excretion (103 ), leading to the suggestion that adrenomedullin may have a role in the hemodynamic changes in hepatic cirrhosis that lead to the formation of ascites (103 ). In congestive heart failure, classified according to the New York Heart Association criteria, a progressive increase in plasma adrenomedullin has been reported from classification I to IV (87 ), although the association is not strong enough to provide a diagnostic marker for the different stages of heart disease (107 ). In patients with mitral stenosis, a significant correlation was found between plasma adrenomedullin levels and pulmonary artery pressure (70 ).

Of all the conditions investigated, the greatest increment in plasma adrenomedullin has been observed in septic shock (75 106 ). It appears that adrenomedullin has a key role in the pathophysiology of septic shock. This is the only pathological condition in which plasma levels of adrenomedullin approach the levels required for receptor activation (see Section III below). The plasma levels of adrenomedullin observed in patients with sepsis are likely to be directly responsible for the hypotension characteristic of septic shock, as a correlation has been demonstrated between plasma adrenomedullin concentrations and relaxation of vascular tone in this condition (108 ). One corollary of this is that the actions of adrenomedullin under normal conditions must therefore be autocrine or paracrine in nature.

D. Origins of circulating adrenomedullin
Adrenomedullin is synthesized in most tissues of the body (see Table 1Go). Although the gene encoding adrenomedullin is very highly expressed in the adrenal gland, in both zona glomerulosa and the adrenal medulla (25 109 ), there is considerable evidence against the adrenal as the major source of circulating peptide. Adrenal venous sampling reveals levels of adrenomedullin that are not significantly different from arterial plasma, in contrast to epinephrine and norepinephrine, which show marked trans-adrenal gradients (110 ). Insulin-induced hypoglycemia has been used in an attempt to provoke the release of medullary adrenomedullin, and while plasma epinephrine levels were increased 20-fold, no significant change in circulating adrenomedullin was observed (75 ). Furthermore, in a patient with a pheochromocytoma, no change in plasma adrenomedullin concentration was seen during a hypotensive attack, although both epinephrine and norepinephrine concentrations increased significantly (111 ). As it has been shown that adrenomedullin is cosecreted with catecholamines, at least by bovine chromaffin cells in culture (112 ), these data suggest that the adrenal medulla is unlikely to be a significant source of circulating adrenomedullin.

Selective arterial and venous sampling across various vascular beds (including heart, lungs, kidney, and adrenal) in patients with a variety of cardiovascular pathologies, failed to identify a site of significant adrenomedullin production (111 ). In congestive heart failure, however, significant cardiac secretion of adrenomedullin has been reported (88 ). It has been suggested that the increased plasma adrenomedullin seen in pregnancy (see below) may derive from the placenta, but in the study that measured both venous and arterial umbilical cord plasma, no difference in adrenomedullin concentrations was found, suggesting that no net production or clearance of adrenomedullin occurs in the placenta (113 ). In certain disease states, notably cerebrovascular disease, the reported increase in plasma adrenomedullin concentration is thought to reflect the degree of endothelial cell damage (86 ).

In measuring plasma adrenomedullin concentrations in disease states, it is unclear whether the generally elevated levels reflect increased production or decreased clearance. This question has not been directly addressed. However, in septic shock there is evidence for increased adrenomedullin production by several different cell types (19 34 35 36 57 59 ). In congestive heart failure there is also evidence for increased cardiac production of adrenomedullin (67 ).

E. Metabolic clearance of adrenomedullin
The plasma half-life of adrenomedullin has been reported to be 22.0 ± 1.6 min with a MCR of 27.4 ± 3.6 ml/kg·min and with an apparent volume of distribution of 880 ± 150 ml/kg (74 ). The effects of plasma membrane enzymes on adrenomedullin have been investigated. It appears likely that adrenomedullin is degraded initially by metalloproteases to yield adrenomedullins 8–52, 26–52, and 33–52, followed by an aminopeptidase action to yield adrenomedullins 2–52, 27–52, and 28–52 (114 ). It has been suggested that the lung may be a major site of adrenomedullin clearance in man (111 ).

F. Adrenomedullin in other biological fluids
In addition to peripheral plasma, significant levels of adrenomedullin have been measured in urine (66 115 ), milk (48 ), cerebrospinal fluid (CSF) (116 117 ) saliva (S. Kapas, unpublished data), amniotic fluid (89 ), sweat (50 ), and in umbilical vein blood (89 ). Using healthy subjects, urinary adrenomedullin concentrations have been reported to be approximately 6-fold higher than plasma levels (66 ). The authors suggest that the lack of correlation between urinary and plasma levels argues against the kidney as a major site of adrenomedullin excretion (66 ). However, urinary adrenomedullin is reported to be decreased in various renal disorders, such as IgA nephropathy (99 ), with an increase in plasma peptide levels, possibly suggesting impaired excretion. In healthy subjects, however, the high urinary adrenomedullin concentrations relative to plasma may suggest that the kidney itself is the major source of urinary adrenomedullin (66 ). The concentration of adrenomedullin in CSF is lower than that in plasma, and while plasma adrenomedullin increases in pregnancy, no change in CSF concentration is seen, suggesting independent regulation of adrenomedullin in the two compartments (116 ).

Although the presence of adrenomedullin in murine milk has been demonstrated (48 ), no data are presently available on the concentration, so it is unclear whether it is actively secreted into milk. In urine (66 ) and sweat (50 ) the concentration of adrenomedullin appears to be significantly higher than in plasma, and there is evidence for production of adrenomedullin in both skin and kidney (see Table 1Go).

G. Regulation of adrenomedullin gene expression and peptide synthesis in vivo
The effects of various physiological manipulations on plasma adrenomedullin concentrations have been investigated in both man and other species. Exercise has been reported by some studies (83 ), but not others (118 119 ), to increase plasma adrenomedullin in man, with a correlation between plasma adrenomedullin and blood pressure (83 ). Moving from low to high altitude was also associated with an increase in plasma adrenomedullin probably related to the degree of hypoxia experienced by the subjects (69 ). In dogs, as in man, hemorrhagic shock causes an increase in plasma adrenomedullin (120 ), and endotoxic shock increases adrenomedullin gene expression in blood vessels (12 ). In rat and mouse a consistent finding is that experimentally induced sepsis increases adrenomedullin gene expression and plasma concentration (121 122 123 124 125 ). In rats, a period of fasting causes an increase in adrenomedullin concentration in the gastrointestinal tract (126 ).

Pregnancy is associated with increased circulating adrenomedullin concentrations in both rats (127 ) and women (77 89 116 ). The plasma concentration of adrenomedullin has been reported to increase progressively from the first to third trimester, with a further increase postpartum, although these data have been questioned due to the excessively high concentrations of adrenomedullin found in this study (see Table 3Go) (63 128 ). A more recent study reported that, while plasma adrenomedullin was increased on average 5-fold in pregnant women compared with nonpregnant women, there was no correlation with gestational age, and within 48 h post partum plasma adrenomedullin concentrations had significantly decreased (89 ). An interesting observation is that babies delivered by the vaginal route had significantly higher umbilical cord adrenomedullin concentrations than babies delivered by elective cesarean section (129 ). In rats it is possible to mimic the effects of pregnancy on plasma adrenomedullin concentrations by the administration of a progesterone derivative (127 ), suggesting that the increased adrenomedullin has a role in the cardiovascular changes of pregnancy. It has been shown that adrenomedullin mRNA in the rat uterus is significantly increased in pregnancy (130 ), suggesting that the uterus itself may be the source of plasma adrenomedullin.

The effects of various endocrine manipulations have also been investigated. Hyperthyroid rats were found to have increased plasma adrenomedullin concentrations and also an increased adrenomedullin mRNA level in the lung (131 ). Glucocorticoids are also implicated in the regulation of adrenomedullin: patients with Addison’s disease (primary adrenal insufficiency) had their plasma adrenomedullin levels reduced by glucocorticoid replacement (68 ). However, insulin-induced hypoglycemia, a potent stimulus to glucocorticoid secretion, had no effect on plasma adrenomedullin concentrations (75 ). In rats with septic shock, dexamethasone did not alter plasma adrenomedullin levels, but in control adrenalectomized animals dexamethasone significantly increased both lung mRNA levels and plasma adrenomedullin (122 ). In the rat ventral prostate adrenomedullin expression is highly androgen dependent, with a 25-fold reduction in mRNA after castration, which is fully reversible by androgen administration (132 ).

Adrenomedullin is implicated in the regulation of fluid and electrolyte status (5 ), and it has been shown that adrenomedullin concentrations are reduced by hemodialysis in patients with renal disease (71 ). Altering the renin-angiotensin system by the use of captopril or furosemide was found to have no effect on plasma adrenomedullin in normal subjects (133 ), and an infusion of ACTH was also without effect (133 ). Similarly it has been found that feeding rats a diet either high (4%) or low (0.02%) in salt has no effect on renal adrenomedullin gene expression (40 ). In the Dahl salt-sensitive rat strain, however, those on a high-salt diet had increased plasma and ventricular adrenomedullin by comparison with those on a control diet (134 ). In human subjects, changes in salt intake, either acute or chronic, had no effect on plasma adrenomedullin in either normotensive or hypertensive subjects (135 ).

It has been demonstrated that an infusion of atrial natriuretic peptide increases plasma adrenomedullin levels in healthy control subjects (73 ). In this study blood was taken for adrenomedullin measurement at 30-min intervals for 5 h. During this time there was apparently no change in plasma peptide levels in the control subjects, while the test subjects showed an elevated plasma adrenomedullin concentration for only the 60-min duration of the infusion. A steady-state 4-fold increase was achieved within 20 min of the onset of the infusion, and levels returned to basal within 30 min of the cessation of the infusion (73 ). Two models of pressure overload have also been used to investigate the regulation of adrenomedullin in the rat: hormonally induced overload, using either arginine vasopressin (AVP) or angiotensin II, resulted in an increase in cardiac adrenomedullin mRNA and peptide (136 ). No effect on adrenomedullin expression was seen in the surgical model, however, despite a marked increase in atrial natriuretic peptide (137 ).

From the data outlined above it is difficult to describe the exact mechanisms that regulate adrenomedullin synthesis and secretion in vivo. The question is clouded by the fact that these data were obtained from several different species and using different techniques. However, the major consistent findings are of increased adrenomedullin in two conditions: sepsis and pregnancy. It also appears likely that adrenomedullin is not, in general, subject to regulation by electrolyte balance, although in some conditions of altered blood pressure, adrenomedullin levels appear to change in a manner consistent with the possible role of this peptide in a compensatory mechanism. The data concerning hormonal regulation of adrenomedullin in vivo are, at present, conflicting. It also appears likely that specific regulatory mechanisms may exist in different tissues for the local control of adrenomedullin production.

H. Experimental regulation of adrenomedullin gene expression and peptide synthesis in vitro
The in vitro regulation of adrenomedullin gene transcription and peptide synthesis has been studied in a number of comprehensive papers by Kangawa and co-workers (19 20 22 ), using either rat vascular smooth muscle cells (VSMCs), or rat endothelial cells. Adrenomedullin production by vascular smooth muscle cells is increased by a range of cytokines, growth factors, and hormones, including tumor necrosis factor {alpha} and ß, interleukin-1{alpha} and ß, (19 20 ), dexamethasone, cortisol, aldosterone, retinoic acid, and thyroid hormone (138 139 ). Other hormones and growth factors were found to have little effect, including fibroblast growth factor, epidermal growth factor, platelet-derived growth factor, progesterone, estradiol, and testosterone (19 138 139 ). Other studies on VSMCs have shown that oxidative stress, induced by diethyldithiocarbamate, also increases adrenomedullin production (140 ). The regulation of fibroblast adrenomedullin gene expression is essentially the same as that of VSMCs (57 ), but there are some differences between vascular endothelial cells and VSMCs, notably in their response to thrombin and {gamma}-interferon (59 ). {gamma}-Interferon also increases adrenomedullin expression by rat astrocytes (32 ), while interleukin-1ß, tumor necrosis factor-{alpha}, and dexamethasone stimulate cardiac myocytes to produce adrenomedullin (24 141 ). An interesting observation on human aortic endothelial cells is the finding that shear stress down-regulates adrenomedullin gene expression (142 ).

In mouse and human macrophages, lipopolysaccharide, {gamma}-interferon, tumor necrosis factor {alpha}, retinoic acid, and the phorbol ester phorbol 12-myristate 13-acetate increase adrenomedullin gene transcription and secretion (34 35 36 ). Synergistic effects were found when retinoic acid was added in combination with other effectors (36 ). It has also been shown that the antiestrogen, tamoxifen, induces adrenomedullin synthesis in endometrial macrophages (47 ).

Lipopolysaccharide is a potent stimulus to adrenomedullin secretion by macrophages (34 35 36 ), VSMCs (19 ), fibroblasts (57 ), and endothelial cells (59 ). It is clear that the induction of adrenomedullin transcription and synthesis by lipopolysaccharide and cytokines gives this peptide a significant role in sepsis and inflammatory states. However, more tissue-specific regulatory mechanisms also exist. In rat granulosa cells, for example, adrenomedullin gene expression is decreased by FSH treatment (44 ), and there is evidence that adrenomedullin is differentially regulated in renal mesangial and glomerular epithelial cells (37 ). In general, it appears that cAMP-mediated effects decrease adrenomedullin, while activation of the phospholipase C-protein kinase C pathway stimulates adrenomedullin (37 143 ). Studies on transcriptional regulation of both human and rat adrenomedullin gene suggest that the effects of the cytokines are mediated by the NF-IL-6 regulatory element in the promoter region of the adrenomedullin gene (143 144 ).


    III. Receptors and Signal Transduction
 Top
 Abstract
 I. Introduction
 II. Synthesis and Secretion...
 III. Receptors and Signal...
 IV. Biological Actions of...
 V. Unresolved Issues and...
 References
 
Because the receptors involved in signaling the effects of adrenomedullin have already been reviewed in part (5 6 ), in the present work we have concentrated on exciting recent findings relating to the molecular identity of adrenomedullin and CGRP receptors.

A. Do CGRP receptors mediate the effects of adrenomedullin?
Adrenomedullin receptors have always been closely associated with receptors for the related peptide, CGRP (145 ). CGRP receptors have been classified into two subtypes on the basis of the potency (pA2) of the CGRP receptor antagonist fragment, CGRP8–37 (146 147 ). Some CGRP1 receptors, at least in the rat, are antagonized by the fragment with a pA2 of about 8.0 while CGRP2 receptors require higher concentrations (pA2 ~6.0) (148 149 150 ). CGRP2 receptors have also been characterized by the ability of the CysACM analog ({alpha}-[acetimidomethyl-Cys2, 7]h{alpha}CGRP) to act as an agonist at these receptors but not at CGRP1 receptors (151 ), but this has recently been disputed (152 ).

Initial pharmacology seemed to indicate that the vascular effects of adrenomedullin were directly mediated by a well characterized CGRP1 receptor mechanism (153 154 ). Nuki et al. (155 ) showed that the vasodilator effects of adrenomedullin and CGRP on the rat mesenteric vascular bed (a prototypic CGRP1 preparation) could be blocked by CGRP8–37. Similar effects of CGRP8–37 on CGRP- or adrenomedullin-induced vasodilation were shown in the isolated rat heart preparation (156 ) and in the rat and hamster microvasculature (157 ). Since publication of these studies, a large body of evidence suggests that some adrenomedullin effects can be blocked by CGRP8–37 (130 157 158 159 160 161 162 163 164 ), but here we will consider only the receptors. Adrenomedullin certainly can bind with high affinity to and activate CGRP receptors in SK-N-MC neuroblastoma cells, commonly used as a model of CGRP1 receptors (156 162 ). In the study of Zimmermann et al. (162 ) adrenomedullin was only 7 times weaker in affinity than CGRP (IC50= 0.3 vs. 2 nM). Adrenomedullin has also been shown to compete with 125I-CGRP binding in rat lung and heart membranes (165 ), rat brain (166 ), SK-N-MC (156 162 167 ), L6 myoblasts (163 ), rat spinal cord (168 ), rat aorta (169 ), rat uterus (130 ), guinea-pig vas deferens (170 ), and rat hypothalamus (158 ). These data in general support the idea of high-affinity (low nanomolar) binding of adrenomedullin to all CGRP receptors with an affinity of about one-tenth to one-hundredth that of CGRP itself.

Interestingly, adrenomedullin has a low affinity (IC50=129 nM) for 125I-CGRP binding sites in guinea-pig vas deferens (170 ), a model of CGRP2 receptors, indicating perhaps that adrenomedullin has a lower affinity at these receptors than at CGRP1 receptors. Some caution should be used when interpreting the inhibitory effect of CGRP8–37 on adrenomedullin actions as evidence of CGRP receptor involvement. In some studies very high concentrations of the antagonist are used, which may bind to specific adrenomedullin receptors (e.g., Ref. 56 ; IC50 for CGRP8–37 binding was 214 nM at adrenomedullin receptors in Rat-2 cells, where 10 µM CGRP inhibited cAMP elevation via specific adrenomedullin receptors) and cloud the interpretation. Binding studies showing an affinity of CGRP8–37 for 125I-CGRP sites similar to the concentrations used to inhibit adrenomedullin effects (e.g., Ref. 163 ) are very useful although not always possible in animal experiments. In vivo experiments that show a lack of effect of CGRP8–37 on adrenomedullin effects but antagonism of CGRP effects at the same concentration are convincing evidence for specific adrenomedullin effects (169 171 172 173 174 175 ).

B. Are there specific adrenomedullin receptors?
Thus, CGRP receptors mediate at least some of the effects of adrenomedullin. However, later experiments using 125I-adrenomedullin showed that specific adrenomedullin receptors existed. Eguchi et al. (176 ) demonstrated binding of 125I-rat adrenomedullin to rat VSMCs and that this binding could be competed by rat adrenomedullin [dissociation constant (KD) = 13 nM] and CGRP [ inhibition constant (Ki ) = 300 nM (176 )]. This 23 times greater affinity for adrenomedullin over CGRP would not be expected for a CGRP receptor. Surprisingly, the rat adrenomedullin-mediated stimulation of cAMP levels seen in these VSMCs was inhibited by CGRP8–37, albeit at high concentrations [IC50 = 300 nM (176 )]. In another study on rat VSMCs, human adrenomedullin increased intracellular cAMP with an EC50 of 20 nM compared with 8.5 nM for CGRP with CGRP8–37 blocking the action of adrenomedullin (IC50=93 nM) (177 ). However, 125I-human adrenomedullin binding in these cells (IC50=73 nM) was not inhibited by either CGRP or CGRP8–37 at concentrations up to 10 µM (177 ).

C. The pharmacology of specific adrenomedullin receptors
The data of Ishizaka et al. (177 ) and Eguchi et al. (176 ) indicate the presence of receptors with a higher affinity for adrenomedullin than CGRP, distinguishing these from any known CGRP receptor (148 150 ). Examination of specific 125I-adrenomedullin binding sites in rat tissues (165 ) showed high levels of specific binding in heart, lung, spleen, liver, skeletal muscle (soleus, diaphragm, and gastrocnemius), and spinal cord. CGRP receptors are also abundant in highly vascular tissues such as lung, heart, and spleen (145 150 ). Apart from spinal cord, binding in the central nervous system (CNS) was low in contrast to CGRP binding, which is widespread and abundant in the brain (7 145 ). Binding in adrenal and kidney membranes was low but as binding was measured in membrane preparations from whole tissues/glands, this in no way negates the large bodies of evidence for important roles for adrenomedullin in these tissues acting via highly localized receptors (5 25 178 ).

Binding sites in heart and lung were further characterized. These sites showed saturation dissociation and competition as would be expected of receptor binding sites. Rat and human adrenomedullin competed at both sites with rat adrenomedullin showing the greater affinity (Table 5Go). Competition by CGRP, amylin, and calcitonin was approximately 3 orders of magnitude less than rat adrenomedullin, indicating a high level of specificity. Other reports of 125I-adrenomedullin binding also show a low affinity of CGRP at this site [VSMCs (176 177 ), NG108–15 neuroblastoma-glioma cells (179 ), Swiss 3T3 mouse fibroblasts (57 180 ), rat hypothalamus (158 ), rat spinal cord (168 ), rat blood vessels (169 ), L6 myoblasts (163 ), rat uterus (130 ), human brain (181 ), bovine endothelial cells (182 ), mouse astrocytes (183 ), human oral keratinocytes (184 ), rat-2 fibroblasts (56 ), rabbit kidney glomeruli (174 ), guinea-pig vas deferens (170 ), rat adrenal zona glomerulosa cells (25 ), and human skin cells (50 )]. These results are summarized in Table 5Go. It is clear from the table that high-affinity (mean affinity = 6 nM) 125I-adrenomedullin binding sites can be detected in tissues and cells from a number of species with differing methodologies. These sites all show low affinity for CGRP and, where measured, amylin and calcitonin, and therefore appear highly specific. The human adrenomedullin fragment, adrenomedullin22–52, has been used as a specific adrenomedullin receptor antagonist (164 185 188 189 190 191 192 193 ) in a similar way that CGRP8–37 is used for CGRP1 receptors. In rabbit aortic endothelial cells [Ki for adrenomedullin-stimulated cAMP was 2.6 nM with no effect on CGRP-stimulated cAMP (188 )] and rat cerebral blood vessels [5 µg/kg/min infusion inhibited adrenomedullin-mediated vasodilation (193 )] adrenomedullin22–52 was an effective antagonist. Some specificity was demonstrated by its lack of effect in T47 D cells [calcitonin receptor, 1000 nM (190 )] or L6 myocytes [CGRP receptor, 1000 nM (192 )]. However, in rat mesangial cells (IC50 for inhibition of adrenomedullin-stimulated cAMP: 70 nM compared with 50 nM for CGRP8–37) and human neuroblastoma TGW cells [DNA synthesis stimulated by adrenomedullin (190 )], there was no difference in potency between the effects of adrenomedullin22–52 and CGRP8–37. In rat VSMCs [half-maximal antagonism of adrenomedullin-stimulated cAMP was 4000 nM and Ki for binding 1600 nM (185 )], adrenomedullin22–52 was a very weak antagonist. Worse still, in rat cardiac cells [adrenomedullin-stimulated cAMP (189 )] and the hindlimb vascular bed of the cat [vasodilation — effect of 30 nmol adrenomedullin22–52 (194 )], adrenomedullin22–52 was inactive against adrenomedullin but inhibited CGRP effects. Thus, better antagonists need to be developed. It has been suggested that human adrenomedullin26–52 is a more specific antagonist (195 ). The use of either rat or human adrenomedullin as radioligand appears not to affect the results and the two labels cross-react across species (170 174 177 179 182 185 186 ). We have found that human adrenomedullin is associated with a much higher nonspecific binding than rat adrenomedullin and therefore prefer this radioligand (nonspecific binding was 8% for rat adrenomedullin and 23% for human adrenomedullin in rat lung (A. A. Owji and D. M. Smith, unpublished observation). One obvious conclusion from this binding data is that circulating levels of adrenomedullin, approximately 3–6 pM in man (60 62 ) and 3 pM in rat (81 ), cannot mediate the physiological effects of adrenomedullin by these specific receptors or by CGRP receptors, placing it firmly as a paracrine/autocrine factor.


View this table:
[in this window]
[in a new window]
 
Table 5. [125I]Adrenomedullin binding sites

 
Adrenomedullin seems to have little affinity for receptors for the other two members of the peptide family, calcitonin and amylin. Adrenomedullin has been shown to interact with the calcitonin receptor on human breast cancer T 47D cells on the basis that adrenomedullin-stimulated cAMP in these cells could be inhibited by the calcitonin receptor antagonist, salmon calcitonin8–32 but not adrenomedullin22–52 or CGRP8–37 (190 ). However, very high concentrations of adrenomedullin (EC50=132 nM) were required to stimulate cAMP, questioning the physiological role of such an interaction. In a comprehensive study, Vine et al. (167 ) showed little effect of human adrenomedullin in vivo on typical amylin (inhibition of soleus muscle glycogen synthesis, hyperlactemia) or calcitonin (hypocalcemia, inhibition of gastric emptying) effects. This correlated with weak competition in model amylin (rat nucleus accumbens membranes, Ki = 51 nM) and calcitonin (T 47D cells, Ki = 33 nM) receptor binding sites. Adrenomedullin also showed very low affinity for amylin binding sites (IC50=0.33–18 µM depending on brain region) in an autoradiographic study of rat brain (166 ). Amylin binding sites in rat lung (196 ) have a higher affinity for adrenomedullin than amylin suggesting that these sites may be adrenomedullin binding sites (165 ). Very little receptor autoradiography has been performed using 125I-adrenomedullin (probably because of technical problems with the radioligand). However, binding sites in the rat lung, kidney cortex, and liver have been shown by in vivo autoradiography with most binding localized to endothelial cells of small arteries (197 ). Binding sites in the human adrenal, specifically in the zona glomerulosa and capsular vessels, were demonstrated using in vitro autoradiography with only the vessel sites not competed by CGRP8–37 (198 ).

One aspect of this binding data that remains puzzling is the lack of competition of 125I-adrenomedullin binding by CGRP in tissues and cells that express both adrenomedullin and CGRP binding, since adrenomedullin will bind effectively to 125I-CGRP sites. One explanation is that if the affinity of adrenomedullin for CGRP receptors is at least 10-fold less than for adrenomedullin receptors, then binding of the low concentration of 125I-adrenomedullin radioligand to CGRP receptors may not be apparent in competition studies. Another possible explanation is that the 125I-adrenomedullin is specific for adrenomedullin sites, whereas adrenomedullin binds to CGRP sites as well. This was supported by the lack of competition of nonradioactive iodoadrenomedullin with 125I-CGRP (165 ). This requires further investigation, perhaps involving the development of new adrenomedullin probes such as fluorosceinated/biotinylated adrenomedullin. 3H-adrenomedullin would be effective in tissues where receptor numbers are high, such as lung, but would be limited by its low specific activity. One possible improvement on this would be the use of higher specific activity metabolically labeled 14C-adrenomedullin.

D. Receptor biochemistry: chemical cross-linking of adrenomedullin and CGRP receptors
Chemical cross-linking experiments using 125I-rat adrenomedullin showed relative molecular weights (Mr) for adrenomedullin binding site-ligand complexes (in this section minor bands are shown in italics separated from the major band by a slash mark) in rat VSMCs of 120,000 and 70,000 (176 ) and in rat tissues of 83,000/105,000 and 94,000 (165 ). In neither case were the labeled bands competed by CGRP. In further experiments specific adrenomedullin binding site-ligand complexes were demonstrated with Mrs, of 83,000 in rat 2 fibroblasts (56 ), of 84,000/122,000 in rat spinal cord (199 ) of 76,000 in L6 myoblasts (163 ), and of 89,000/105,000 in rat aorta (169 ). These complexes can be compared with cross-linked 125I-CGRP sites that vary in Mr: 70,000/110,000 in rat skeletal muscle, 55,000/44,000 in rat liver (200 ), 70,000 and 120,000 in porcine ventricle (201 ), 75,000–90,000 in rat spleen (202 ), and 60,000–70,000 in rat VSMCs (203 ). Thus, on average, CGRP binding sites have an Mr of about 70,000 with adrenomedullin binding site about 85,000. This assertion should be treated with some caution as cross-linked bands are often broad, making comparison across studies difficult [e.g., in rat liver Stangl et al. (202 ) found Mr = 74,000 and 68,000 compared with Mr = 55,000 by Chantry et al. (200 )]. Also the same receptor protein may be subject to large variations in size due to differential glycosylation in different tissues and species.

In rat spinal cord and L6 myoblasts, a direct comparison of the cross-linked adrenomedullin and CGRP bands was made (163 199 ). In spinal cord 125I-CGRP complexes showed an Mr= 74,000 and 61,000 compared with Mr = 84,000/122,000 for 125I-adrenomedullin complexes. Deglycosylation of 125I-adrenomedullin complexes in spinal cord, heart, and lung resulted in a number of complexes with the lowest Mr being 52,000, 47,000, and 43,000, respectively (199 ). In L6 myoblasts 125I-CGRP complexes showed an Mr = 82,000 compared with Mr = 76,000 for 125I-adrenomedullin complexes. Thus, on the whole, although there is some evidence for CGRP receptors showing a lower Mr on SDS-PAGE than adrenomedullin receptors, this remains an outstanding question. Also a number of cross-linking studies of both adrenomedullin and CGRP binding sites (165 169 176 199 200 ) show second bands of higher mol wt than the major band, indicating the possibility of a further complex in addition to that of the ligand and binding site. On the whole, these second bands are not large enough to be receptor dimers (204 ).

E. Receptor biochemistry: molecular characterization of adrenomedullin and CGRP receptors
Examination of the pattern of binding of 125I-adrenomedullin in rat tissues led us to reconsider the role of the L1 orphan receptor (205 ) [also known as G10d from rat liver (206 ), a 395-amino acid seven-transmembrane receptor, GenBank accession number L04672], which is expressed in lung, adrenal, heart, and spleen (4 ). When transfected into COS-7 cells, this receptor bound 125I-adrenomedullin (KD = 8.2 nM) and gave adrenomedullin-mediated increases in intracellular cAMP (ED50 = 7 nM) that were only inhibited by high concentrations of CGRP8–37 (Ki = 1 µM) (4 ). The human homolog of this receptor was then cloned but not expressed and found to show 73% similarity by amino acid sequence, which is not high for a species homolog (404 amino acids, EMBL accession number Y13583) (207 ). The identification of both rat and human sequences as adrenomedullin receptors has recently been questioned (208 ). No binding of rat or human 125I-adrenomedullin followed transfection of either sequence into COS-7 cells, despite the presence of mRNA and expression of the protein at the cell surface (208 ). The most closely related receptor to L1 is a dog 7-transmembrane receptor called RDC-1 (49 ) (GenBank accession number X14048). Expression of this receptor in COS-7 cells gave a pharmacology typical of a CGRP1 receptor with CGRP-stimulated cAMP generation (EC50 = 3 nM) potently inhibited by CGRP8–37 (209 ). Adrenomedullin also stimulated cAMP levels with an EC50 of 100 nM, as would be expected of a CGRP1 receptor. Binding studies showed a similar affinity for CGRP and CGRP8–37 (9 and 13 nM respectively) (209 ). RDC-1 mRNA expression is high in vascular tissues such as lung and liver, which express high levels of 125I-CGRP binding (49 ).

An interesting study from Luebke et al. (210 ) showed the expression cloning of a hydrophilic 146-amino acid protein from guinea pig organ of Corti, which conferred CGRP receptor activity on Xenopus oocytes (210 ) (GenBank accession number U50188). This protein, called receptor component protein (RCP), is expressed in human and mouse mainly in testis, with smaller amounts in human in prostate, ovary, small intestine, and spleen (211 ). RCP conferred CGRP (10 nM) effects on oocytes but not calcitonin or amylin (100 nM) effects. Using in situ hybridization in the guinea pig CNS, RCP was shown to be abundant in the cerebellum and hippocampus (212 ). Adrenomedullin effects via RCP have not been tested, but the limited distribution of RCP means it can only account for a small subset of adrenomedullin receptors at best unless other RCPs are yet to be cloned.

The other side of the CGRP/adrenomedullin receptor story relates to another orphan receptor called calcitonin receptor-like receptor (CRLR). This was originally cloned by two groups, Legon and co-workers (213 ) (GenBank accession number X70658) using a PCR strategy amplifying rat hypothalamic mRNA with primers based on the porcine calcitonin and opossum PTH receptors and Chang et al. (214 ), who also described the identification of a CRF receptor (GenBank accession number L27487). The full sequence of the human homolog of CRLR from cerebellum was reported by Fluhmann et al. (215 ) (GenBank accession number U17473). Human CRLR, expressed mainly in lung, heart, and kidney, is a 461-amino acid seven-transmembrane protein with 91% homology to its rat homolog and 51% similarity to the human calcitonin receptor. This receptor expressed in COS-7 cells did not bind any member of the calcitonin family of peptides and was considered an orphan receptor. However, in 1996 Aiyar and co-workers (216 ) showed that hCRLR stably transfected into human embryonic kidney (HEK) 293 cells exhibited the pharmacology of a CGRP1 receptor (CGRP Kd = 19 pM, CGRP8–37 pA2 = 7.57, CysACM-CGRP ineffective up to 1 µM) (216 ). Adrenomedullin showed binding and stimulation of cAMP, albeit weak, in these cells. These results were confirmed using the rCRLR stably transfected into HEK 293 cells (217 ) and later the porcine CRLR as well (218 ). CRLR mRNA is extremely abundant in the rat lung [as is specific 125I-adrenomedullin binding (165 )] and was shown by in situ hybridization studies to be associated with blood vessels (213 ). CRLR protein was also shown by immunocytochemistry to be associated with vascular endothelial cells (217 ). This fits well with a role for adrenomedullin as a pulmonary vasodilator (see Section IV.A) and the presence of adrenomedullin binding on endothelial cells (197 ) but disagrees with the previous localization by in situ PCR to alveolar cells (216 ).

The question now became what was the factor in HEK 293 cells that was not present in COS-7 cells that allowed CGRP receptor expression? The surprising answer was provided by Foord’s group using a Xenopus oocyte/cystic fibrosis transmembrane regulator system, similar to that used to clone RCP, where increases in intracellular cAMP can be detected as chloride currents. They cloned a receptor-activity modifying protein (RAMP-1) of 148 amino acids with a single transmembrane domain that conferred CGRP1 receptor activity (no response to adrenomedullin, amylin, or calcitonin but inhibited by CGRP8–37) to the oocytes (186 ). When transiently cotransfected with hCRLR into HEK 293 cells, RAMP-1 conferred 125I-CGRP binding properties to these cells, but no binding was seen with either RAMP-1 or CRLR alone. The pharmacology of this combination was very similar to that of CGRP receptors in SK-N-MC cells (186 ). The failure of expression of CGRP binding by CRLR in COS-7 cells is therefore probably due to the lack of endogenous RAMP-1. The action of RAMP-1 was shown to involve transport of CRLR to the cell surface. 125I-CGRP could be cross-linked to proteins of Mr = 66,000 (CRLR) and 17,000 (RAMP-1). This Mr of 66,000 compares well with that reported for CGRP receptors in tissues (see previous section). Cross-linking of CGRP to RAMP-1 allows for the intriguing possibility that RAMP-1 forms part of the binding site for CGRP. CRLR is present in HEK 293 cells as a Mr = 58,000 glycosylated receptor, which in the presence of RAMP-1 is further glycosylated to Mr = 66,000, consistent with RAMPs acting to transport CRLR.

Two further members of the RAMP family that did not confer CGRP receptor activity were also identified, but overall the three RAMPs showed 31% identity in amino acids (186 ). The three RAMP mRNAs have widespread and different distributions in human tissues. RAMP-2 and RAMP-3 also facilitated expression of CRLR on the cell surface but as a Mr = 58,000 glycoprotein. Coexpression of CRLR and RAMP-2 in oocytes or HEK 293 cells resulted in a typical specific adrenomedullin receptor pharmacology with no effects of CGRP (186 ). Thus, the RAMP hypothesis offers an extremely interesting explanation of CGRP/adrenomedullin receptor pharmacology, i.e., CRLR/RAMP-1 = CGRP1 and CRLR/RAMP-2 = adrenomedullin.

Since the discovery of RAMPs, some aspects of the hypothesis have been confirmed. Kamitani et al. (219 ) showed that RAMP-2 is expressed in human VSMCs and endothelial cells, and a combination of RAMP-2/CRLR transfected into HeLa EBNA or 293 EBNA cells led to adrenomedullin-stimulated cAMP with no effect of CGRP. Transfection of RAMP-1/CRLR conferred CGRP and adrenomedullin-stimulated increases in cAMP, which differs from the oocyte studies by McLatchie et al. (186 ) but agrees with the binding data for CGRP1 receptors (see above). Muff et al. (188 ) showed that rabbit endothelial cells express RAMP-2 and CRLR and that adrenomedullin stimulated cAMP (EC50=0.18 nM). When these cells are transfected with RAMP-1, they then express CGRP-stimulated cAMP (EC50 = 0.41 nM) which is inhibited by CGRP8–37 (100 nM), indicating that CRLR can be converted from an adrenomedullin-specific receptor to a CGRP receptor by a dominant effect of RAMP-1. This effect was further investigated by the same group using rCRLR and RAMPs expressed in UMR-106 rat osteoblast-like cells and COS-7 cells (187 ). UMR-106 cells transiently transfected with CRLR express 125I-adrenomedullin binding, which was enhanced by RAMP-2 cotransfection. Here 125I-CGRP binding required transfection of RAMP-1 but was unaffected by RAMP-2. Similar results were shown in COS-7 but, as expected, since COS-7 lack RAMPs, adrenomedullin binding required RAMP-2 cotransfection. The amino terminus of the RAMPs has been shown to be the major factor controlling glycosylation and ligand binding using chimeric RAMP-1/2 proteins (220 ). RAMPs 2 and 3 appear indistinguishable in terms of CRLR glycosylation and adrenomedullin binding in this study. Thus, the RAMP hypothesis appears correct but has not yet been totally proven, and some aspects of it need to be investigated. What is the purpose of RAMP-3/CRLR if it yields an identical pharmacology with RAMP-2/CRLR? In rat tissues, adrenomedullin receptors appear by cross-linking to be equal to or larger than CGRP receptors (see previous section), whereas the RAMP hypothesis predicts that CGRP receptors should be larger than adrenomedullin receptors. Do partners other than CRLR exist for RAMP, e.g., the calcitonin receptor (221 )? Do CRLR and RAMP-1/2 account for all adrenomedullin and CGRP binding?

There has been little study of whether the presence of RDC-1, L1, and CRLR correlate with CGRP/adrenomedullin binding. In the rat brain, the distributions of the three putative receptors were compared by in situ hybridization (222 ). RDC-1 mRNA was mainly associated with the dentate gyrus, hippocampal CA3, choroid plexus, and blood vessels. L1 was very weakly expressed except in cells of the pia mater. CRLR was expressed in the caudate putamen and the central and basolateral amygdaloid nuclei. These data match well in some areas but not at all in others with CGRP or adrenomedullin binding in the rat brain (145 181 223 ). None of the three mRNAs was present in spinal cord (222 ) despite high levels of CGRP and adrenomedullin binding (168 ). All three mRNAs were expressed in adult rat heart and neonatal cardiac myocytes, with RDC-1 being most abundant followed by CRLR with L1 being of low abundance (224 ). In rat aortic VSMCs, RDC-1, but not L1 or CRLR, mRNA was detected using specific RNase protection assays (225 ). Binding experiments using 125I-adrenomedullin and fragments of adrenomedullin and CGRP showed two subtypes of adrenomedullin receptor in astrocytes and NG108–15 cells (179 ). Chemical cross-linking of 125I-adrenomedullin binding sites in rat tissues also shows heterogeneity of mol wt (165 168 ). Thus, there are indications that CRLR/RAMP and/or L1/RDC-1 do not account for all CGRP/adrenomedullin binding sites.

F. Signal transduction pathways activated by adrenomedullin
It is now clear from a vast number of studies that the major effect on adrenomedullin-stimulated cells is an elevation of cAMP (25 33 38 39 44 56 58 59 156 162 163 164 174 176 177 179 180 182 183 184 185 189 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 ). This is typical of the calcitonin family of peptides, all of which have been shown to elevate cAMP levels in various tissues and cells (8 150 ). It should not be forgotten that this was the property that was used to discover adrenomedullin (1 ). Also, all of the cloned receptors, regardless of whether they are actually adrenomedullin/CGRP receptors in vivo, are associated with increased cAMP when transfected into cells (4 186 188 209 216 217 ). Thus, the initial mechanism of action of adrenomedullin (and CGRP) is in most cases via G-protein linked receptor activation of Gs, adenylyl cyclase, and protein kinase A (PKA) (163 ). In this section we will examine other possible mechanisms and some consequences of elevated cAMP. Most of the studies on adrenomedullin signaling have been performed using primary cells (especially VSMCs) or cell lines. Since adrenomedullin will bind to both specific adrenomedullin and CGRP receptors and these are often expressed together in cells, the failure to define which receptor is actually mediating the effect is a problem. This can be addressed by use of inhibitors such as CGRP8–37 or adrenomedullin22–52, but these are not especially potent or specific, and better antagonists would greatly advance this work. At present, the only conclusive results are obtained by careful use of antagonists, use of cells expressing only one type of binding, or use of transfected cells. Of course the transfection approach has problems associated with heterologous expression and also the doubt surrounding which receptor should actually be transfected.

The effects of adrenomedullin on calcium signaling mechanisms have been investigated since endothelial NO has been implicated in adrenomedullin-mediated vasodilation (see Section IV.A), and logically this should be the result of an increase in intracellular calcium ([Ca2+]i) activating endothelial cell nitric oxide synthase (NOS). In the bovine aortic endothelial cell-specific 125I-adrenomedullin binding (IC50 = 10 nM) and adrenomedullin-mediated cholera toxin-sensitive increases in cAMP were observed (EC50=0.17 nM) (182 ). Here adrenomedullin also directly increased [Ca2+]i (EC50 = 3 nM) with an initial peak followed by a prolonged increase. The initial effect was blocked by thapsigargin, and the prolonged effect by EGTA and nifedipine. Pretreatment of cells with U-73122, the phospholipase C (PLC) inhibitor, but not its inactive analog U-73343, blocked all calcium responses to adrenomedullin. Similarly, cholera toxin, but not H89 (PKA inhibitor) or pertussis toxin, blocked all effects. As expected for a PLC-mediated effect, adrenomedullin (100 nM) increased the intracellular levels of ITP. Adrenomedullin also increased intracellular cGMP. This very detailed study offers an interesting account of how adrenomedullin might increase NO production and thereby vasodilation. One omission in this study was the effects of CGRP and the action of CGRP8–37 on the adrenomedullin effects. Another problem is that Barker et al. (247 ) were able to show increases in cAMP but unable to show any [Ca2+]i effects in bovine endothelial cells . Adrenomedullin did not affect [Ca2+]i but did increase cAMP in Swiss 3T3 cells that express only specific adrenomedullin receptors and not CGRP receptors (180 ). Similar results were obtained using cultured rat astrocytes (179 ). Adrenomedullin was shown to decrease [Ca2+]i and calcium sensitivity in porcine coronary artery strips, possibly by a direct cAMP-mediated mechanism (248 ). In favor of adrenomedullin increasing [Ca2+]i are results using KG-1C human oligodendroglial cells where adrenomedullin and CGRP both increased cAMP and [Ca2+]i (228 ). Unfortunately the effects of CGRP8–37 or 125I-ligand binding were not investigated. Also, in the perfused rat heart, adrenomedullin enhanced cardiac contractility by a mechanism that was independent of cAMP but involved changes in [Ca2+]i (249 ). In the L6 skeletal muscle cell line, both adrenomedullin and CGRP receptors were present, but increases in intracellular cAMP were mediated only via CGRP receptor binding (163 ).

Reports of effects of adrenomedullin on growth and mitogenesis (see Section IV.B) have led to investigation of the regulation of mitogen-activated protein kinase (MAPK) by adrenomedullin. In rat glomerular mesangial cells, adrenomedullin increased cAMP and PKA but inhibited proliferation (both of quiescent and platelet-derived growth factor (PDGF)-stimulated cells) and MAPK activity (241 ). Also in mesangial cells, adrenomedullin (and other agents that increased cAMP) inhibited endothelin-1 (ET-1)-stimulated MAPK (but not, in this case, basal levels) and MAPK kinase (250 ) and stimulated expression of a MAPK phosphatase (251 ). Chini et al. (230 ) also showed that adrenomedullin reduced PDGF-stimulated MAPK activity and mitogenesis in rat VSMCs, effects that were blocked by the PKA inhibitor, H89. However, in quiescent rat VSMCs, adrenomedullin increased DNA synthesis, cell proliferation, tyrosine phosphorylation, MAPK activity, and expression of the immediate-early gene, c-fos. These effects could be blocked by CGRP8–37 and the tyrosine kinase inhibitor, genistein, but not by cAMP or PKA antagonists, indicating a cAMP-independent effect (252 ). Interestingly, in these VSMCs, adrenomedullin had no effect on [Ca2+]i or ITP. In the Rat-2 fibroblast cell line, which expresses specific adrenomedullin but not CGRP receptors, adrenomedullin stimulated cAMP and inhibited basal and PDGF-stimulated MAPK (56 ).

Adrenomedullin has also been shown to activate other signal transduction mechanisms including K+-ATP channels (253 ) and c-fos expression (229 254 ). Adrenomedullin augmented interleukin-1ß-stimulated NO synthesis in rat VSMCs by a cAMP-dependent mechanism (239 ). Desensitization of adrenomedullin receptors has not been widely investigated, but Iwasaki et al. (227 ) showed that adrenomedullin pretreatment caused a loss of adrenomedullin-stimulated adenylyl cyclase activity in rat aortic VSMCs. Drake et al. (226 ) showed that in SK-N-MC cells, preexposure to CGRP or adrenomedullin desensitized the cells to a subsequent CGRP stimulus, but preexposure to CGRP or adrenomedullin did not affect a subsequent exposure to adrenomedullin. This is interesting since if CRLR is responsible for cAMP stimulation with both CGRP and adrenomedullin in SK-N-MC cells, then they would be expected to give similar desensitization patterns. It seems then that we still need to learn a lot more about adrenomedullin signaling before its mechanisms of action in each of its different roles can be deduced.


    IV. Biological Actions of Adrenomedullin
 Top
 Abstract
 I. Introduction
 II. Synthesis and Secretion...
 III. Receptors and Signal...
 IV. Biological Actions of...
 V. Unresolved Issues and...
 References
 
Although the first paper on adrenomedullin described the cardiovascular effects of this peptide (1 ), it is now known that adrenomedullin is rather more than simply a vasodilator. Adrenomedullin has been shown to have a remarkable range of actions, from regulating cellular growth and differentiation, through modulating hormone secretion, to antimicrobial effects.

A. Vascular actions
In rat, cat, sheep, and man, intravenous infusion of adrenomedullin results in a potent and sustained hypotension (101 174 255 256 257 258 259 260 261 262 ), mainly via NO generation in the vasculature (261 263 264 ) and is comparable to that of CGRP (1 265 ). Initial studies of the hemodynamic effects of human adrenomedullin used anesthetized rats (266 267 ). Acute or chronic administration of adrenomedullin resulted in a significant decrease in total peripheral resistance accompanied by a fall in blood pressure. This is concomitant with a rise in heart rate, cardiac output, and stroke volume (266 267 268 269 ). Similar effects are seen in both conscious (270 ) and hypertensive rats (266 268 ). The hypotensive effect of adrenomedullin on mean arterial pressure in the anesthetized rat is not inhibited by CGRP8–37, suggesting this effect is not mediated via CGRP receptors (169 ). The vascular beds in which adrenomedullin is effective are listed in Table 6Go.


View this table:
[in this window]
[in a new window]
 
Table 6. Vasodilatory actions of adrenomedullin

 
Responses to rat and human synthetic adrenomedullin have been studied in the systemic vasculature as well as regional vascular beds, and there is a marked variation in these responses among species. Adrenomedullin lowered vascular resistance in lung, heart, kidney, and adrenal gland (266 ). In rat mesenteric vascular beds preconstricted with methoxamine, administration of adrenomedullin resulted in a dose-dependent decrease in perfusion pressure and arterial pressure (155 ). This vasodilator response was attenuated by CGRP8–37, suggesting that the CGRP receptor may mediate, at least in part, the vasodilator response to adrenomedullin in the mesenteric vasculature (155 161 265 ). In canine renal arteries, adrenomedullin-induced vasorelaxation occurs in the absence and presence of the endothelium (195 ) and coincides with an increase in cAMP levels. These effects were blocked by CGRP8–37. In contrast, the relaxant effect in dog veins, which is endothelium dependent, is not dependent on cAMP production, NO synthesis, or oxygen free radical generation (195 ). CGRP8–37 has no effect on canine venous tone. Adrenomedullin injection into the human forearm causes a prolonged rise in forearm blood flow (259 ).

Adrenomedullin is vasodilatory in the systemic vascular system of the cat (171 255 287 290 ), an effect that is antagonized by CGRP8–37. However, in the hindlimb vascular bed of the cat, adrenomedullin induces a vasodilatory effect not altered by the CGRP1 receptor antagonist (171 194 275 ). The mechanism by which adrenomedullin reduces vascular resistance in the cat hindlimb circulation is not clear, but it is possible that adrenomedullin may relax vascular smooth muscle by inducing an increase in cAMP levels (176 177 185 291 ). It is also possible that vasodilation induced by adrenomedullin may be mediated via NO release or arachidonic acid metabolism from the endothelium. It has been previously shown that NO mediates responses to adrenomedullin in the renal vascular bed of the dog and the rat pulmonary and hindquarter vascular beds (261 272 278 ). However, in the cat, NOS inhibitors were without effect and, in fact, duration of the vasodilator response to adrenomedullin was significantly increased after administration of rolipram, a type IV phosphodiesterase inhibitor, suggesting a cAMP-mediated mechanism of action (287 290 ).

In pig renal artery smooth muscle strips stimulated by phenylephrine, adrenomedullin caused a fall in tension that was concomitant with a decrease in [Ca2+]i (273 ). Using the intact canine kidney it has been demonstrated that adrenomedullin caused an increase in renal blood flow (RBF) that was attenuated by NOS inhibitors (263 ). Similar effects were seen in isolated perfused rat kidneys (264 274 ) and in rabbits (174 ). In all cases CGRP8–37 was without effect.

The feline model has been used in investigating the physiology and pharmacology of penile erection (292 293 ). Champion and co-workers (287 288 289 ) have used this model to study the effects of adrenomedullin on the erectile response. Adrenomedullin caused significant dose-dependent increases in intracavernous pressure and penile length when injected directly into the corpus cavernosum. Responses to adrenomedullin were comparable to those induced by intracavernous injection of a standard triple-drug combination composed of papaverine, phentolamine, and PGE1. The mechanism of the erectile response to adrenomedullin is unclear but is unlikely to be NO dependent since administration of NG-nitro-L-arginine methyl ester (L-NAME) was without effect.

In the intact cat pulmonary vascular bed, adrenomedullin has no effect on resting arterial pressure; however, in the presence of a thromboxane A2 agonist, adrenomedullin caused a decrease in pulmonary arterial pressure (276 277 ). Using the isolated perfused rat lung preparation, it has been shown that human adrenomedullin causes decreases in preconstricted vascular tone with no effect on resting tone (173 ). Neither CGRP8–37 nor NOS inhibitors had any effect, suggesting that in the lung at least, vasodilation occurs via an NO-independent mechanism. This finding, however, contradicts a study using pulmonary arterial rings whereby NO produced by the vascular endothelium was required for relaxation in response to adrenomedullin (294 ). In conditions of hypoxia, the vasodilatory effect appears to be mediated via PG synthesis rather than NO production (294 ).

PGs have been implicated in another regional vascular system. Adrenomedullin is 16 times more potent than PGI2 as a vasodilator in the uterine circulation. Using nonpregnant, oophorectomized ewes, Friedman et al. (279 ) infused either adrenomedullin (0.01 to 3 µg/min) or PGI2 (0.03 to 10 µg/min) for a period of 5 min. At the doses administered there were no changes in heart rate, cardiac output, or blood pressure;