MMPs (matrix metalloproteinases) play a major role in the pathogenesis of hypertension by altering the extracellular matrix during cardiovascular remodelling. In the present study we show that MMP-2, but not MMP-9, cleaves the vasodilator peptide AM (adrenomedullin). Addition of the AM-binding protein, complement factor H, prevents this cleavage, providing a hitherto unknown mechanism of action for this binding protein. We identified the signature cleavage fragments and found some of them in human urine, suggesting that MMP-2 processing of AM may occur in vivo. Synthetic AM fragments regulated blood pressure in rats. The larger peptides are vasodilators, as is intact AM, whereas intermediate fragments did not affect blood pressure. In contrast, AM(11–22) elicited vasoconstriction. Studies of AM receptor activation in Rat2 cells confirm that the larger AM cleavage peptides activated this receptor, whereas AM(11–22) did not. The present study defines a new mechanism through which MMP-2 may regulate blood pressure by simultaneously eliminating a vasodilator and generating a vasoconstrictor.
Hypertension is a major public health problem due to its high prevalence and increased risk of cardiovascular morbidity and mortality [1,2]. Development of hypertension is accompanied by cardiac and arterial remodelling, a process that eventually compromises organ function. While the precise molecular pathogenesis is not fully understood, recent studies suggested that MMPs (matrix metalloproteinases) contribute to this process by remodelling the extracellular matrix [2–4]. The MMP family includes more than 20 members that share structural domains, but differ in substrate specificity, cellular sources and transcriptional regulation. A common characteristic of these enzymes is their ability to degrade components of the extracellular matrix. This feature has relevance to almost every aspect of mammalian biology and pathophysiology .
Blood pressure is regulated by the interaction between the sympathetic nervous system and a complex panel of vasoactive peptides. Current translational research involves finding specific inhibitors for the enzymes that either activate hypertensive peptides (as in the case of angiotensin-converting enzyme) or degrade hypotensive peptides (as with neutral endopeptidases which cleave natriuretic peptides) [6,7]. AM (adrenomedullin), a potent and long-lasting vasodilator , is becoming increasingly attractive as a potential key mediator of blood pressure homoeostasis. In addition, plasma AM levels are increased in cardiovascular diseases, such as heart failure, hypertension and septic shock, where AM seems to play a protective role . AM is a 52-amino-acid peptide with an internal disulphide bond between amino acids 16 and 21 and with an amidated tyrosine at the carboxy end . We have previously described the existence of a serum-binding protein for AM and characterized it as factor H (complement factor H) . This binding interaction with factor H is able to increase the activity of AM in several experimental models, but so far the molecular mechanism responsible for this enhancing effect is unknown . The functions of AM are mediated through a complex receptor system that requires the presence of a seven-transmembrane domain polypeptide known as CRLR (calcitonin-receptor-like receptor) and the single-transmembrane domain protein, RAMP (receptor-activity-modifying protein) 2 or 3. Receptor activation with nanomolar concentrations of AM results in intracellular elevation of cAMP levels .
Previous analysis of AM degradation by cell membrane extracts indicated a potential role for aminopeptidases and MMPs, with a pattern different from those involved in angiotensin II formation or atrial natriuretic peptide degradation . Considering the importance of AM and the gelatinase subfamily of the MMPs on hypertension, we decided to study whether there is an interaction between both regulatory systems.
In the present study we report on the specific degradation of AM by MMP-2, but not by MMP-9. We also present the protective effect that factor H exerts on the half-life of AM in the presence of MMP-2, thus providing an explanation for the observed increase of AM activity in the presence of its binding protein. Furthermore, we identified some MMP-2-specific digestion products of AM in the urine of normal individuals, indicating that this process occurs in vivo. In addition, one of the peptide products exhibited a delayed vasoconstrictor activity in rats without interacting with CRLR/RAMP2, suggesting that other independent receptor systems may be involved in this activity. Therefore specific MMP-2 inhibitors may be useful in maintaining local AM levels and thus mitigate hypertension.
Synthetic human AM was purchased from Peninsula Laboratories (San Carlos, CA, U.S.A.). Purified human factor H was obtained from Sigma (St. Louis, MO, U.S.A.). Predicted AM fragments were synthesized by Princeton Biomolecules (Langhorne, PA, U.S.A.) with a purity higher than 95%. Human recombinant MMP-2 and MMP-9 were obtained and activated as described previously . At the end of the activation period enzyme samples were diluted and assayed for activity using the thiopeptolide assay, as described previously .
Synthetic AM (240 μg/reaction) was exposed to 11.5 μg of MMP-2 or MMP-9 in low-salt collagenase buffer (50 mM Tris, pH 7.5, 50 mM NaCl and 0.02% Brij) for various periods of time in the presence or absence of 6 mg of factor H. These amounts correspond to an approximate molar ratio of 1:250:250 (enzyme/AM/factor H). The reaction was stopped with EDTA. Digestion reactions were run in 16% Tricine gels (Invitrogen, Carlsbad, CA, U.S.A.) under reducing conditions and the gels were stained with Gel Code Blue Stain Reagent (Pierce, Rockford, IL, U.S.A.).
Digestion reactions were also loaded into an analytical reverse phase R2H HPLC column (5 mm×100 mm; Poros, Applied Biosystems, Foster City, CA, U.S.A.) in a 10–60% acetonitrile gradient over 10 min, and the protein peaks eluted from the column were detected with a wavelength of 230 nm.
The protein peaks identified by HPLC were further characterized by MS. One microlitre of each HPLC fraction was mixed with 1 μl of α-cyano-4-hydroxycinnamic acid. One microlitre of each mixture was applied to a MALDI (matrix-assisted laser-desorption ionization) plate and allowed to air dry. The plate was loaded into a PerSeptive Biosystems Voyage DE mass spectrometer (ABI, Foster City, CA, U.S.A.). The instrument was calibrated with angiotensin I using a two-point calibration, angiotensin I at m/z 1296.7 and the matrix dimer at m/z 379.0. The laser intensity used to observe the peptides was 2721 with an accelerating voltage of 25000 V. The extraction delay time was 50 ns. One hundred laser shots comprised a spectrum. The resulting data were analysed using Voyager Data Explorer (ABI).
Detection of AM fragments in urine
Urine (250 ml) from healthy male volunteers was extracted through C-18 Sep-Pak cartridges (Waters Corporation, Milford, MA, U.S.A.). The eluate was freeze dried, resuspended in 5 ml of 5% acetonitrile in water plus 0.1% trifluoroacetic acid, and fractionated by HPLC in a C-18 preparative column (30 mm ×30 cm; Delta Pack, Waters Corporation) using a 5–60% acetonitrile gradient over 155 min at a flow rate of 15 ml/min. The column had been previously standardized with the synthetic peptides (see Figure 3a). Fractions were freeze dried, resuspended in sample buffer (Invitrogen) and run on 12% Bis/Tris gels (Invitrogen) under reducing conditions. Peptides were transferred on to nitrocellulose filters, and Western blotting was performed with a previously characterized antibody against AM  and a chemiluminescence kit (ECL Plus™ Western Blotting Detection System, Amersham Biosciences, Piscataway, NJ, U.S.A.).
Blood pressure measurements
The effects of AM peptide fragments on blood pressure were measured as described previously . Male 10-week-old Lewis/ssncr rats (SAIC, Frederick, MA, U.S.A.) were anaesthetized with 3% halothane, intubated and maintained with 1% halothane in 70% nitrous oxide/30% oxygen (VMS Anesthesia Machine, Matrix Medical Inc., Orchard Park, NY, U.S.A.) at 82 strokes/min. Rectal temperature was monitored throughout the experiment. A PE50 catheter was placed on to the right femoral artery and arterial blood pressure was recorded through a P23XL transducer (Grass Instruments, Quincy, MA, U.S.A.). Peptides were injected into the right femoral vein through another catheter. To confirm that changes in blood pressure were not an artifact of the anaesthesia, the experiments were repeated in conscious animals. After catheters were placed under anaesthesia, the animals were allowed to recover for 24–48 h before taking blood pressure measurements. All procedures were performed under a protocol approved by the National Institutes of Health.
Measurement of cAMP response
The rat fibroblast cell line Rat2 was obtained from the A.T.C.C. (Manassas, VA, U.S.A.) and kept in RPMI 1640 medium supplemented with 10% fetal calf serum (Invitrogen). Accumulation of intracellular cAMP was measured as described previously . Cells were seeded into 24-well plates at 2×104 cells/well and incubated at 37 °C in 5% CO2/95% air until they reached 80% confluency. Before the assay, cells were incubated for 15 min in TIS medium (RPMI 1640 plus 10 μg/ml transferrin, 10 μg/ml insulin and 50 nM sodium selenite) containing 1% BSA, 1 mg/ml bacitracin and 100 μM isobutylmethylxanthine. Peptides were applied in the same medium for 5 min at the indicated concentrations in a volume of 250 μl. The reaction was terminated by adding an equal volume of ice-cold ethanol. cAMP contents were measured using the Biotrac cAMP radioimmunoassay (Amersham).
AM is a substrate for MMP-2
MMP-2 rapidly cleaves synthetic AM in a time-dependent manner as demonstrated by the progressive appearance of lower-molecular-mass bands in polyacrylamide gels (Figure 1, lanes 1–4). The resistance of AM to degradation by MMP-9 (Figure 1, lanes 9–12) underscores the specificity of MMP-2 degradation, considering that both enzyme preparations (MMP-2 and MMP-9) are able to efficiently digest a variety of common substrates, including gelatin and thiopeptolide (results not shown). Factor H, the serum-binding protein for AM, is not a substrate for either one of the MMPs (results not shown), but addition of factor H completely prevents MMP-2-mediated degradation of AM (Figure 1, lanes 5–8). In contrast, factor H does not interfere with the ability of MMP-2 to degrade thiopeptolide (results not shown), demonstrating that prevention of AM degradation by MMP-2 is dependent on the specific protein–protein interaction between AM and factor H.
MMP-2, but not MMP-9, degraded AM in the absence of factor H
Analysis of these digestion reactions by reverse-phase chromatography reveals a rapid decrease in the area of the peak representing the intact peptide (arrow in Figure 2a) and the concomitant progressive appearance of additional new peaks (Figures 2b–2e). MS analysis identifies the AM fragments generated by MMP-2 digestion. These include AM(8–52), AM(11–52), AM(23–52), AM(29–52), AM(11–28) and AM(11–22). The amino acid patterns at the cleavage sites are compatible with the predicted motifs for MMP-2 targets . Two of the AM cleavage peptides, AM(8–52) and AM(11–52), maintain both the intramolecular disulphide ring structure and the α-amide, two characteristics required for AM receptor activation . Two other fragments, AM(23–52) and AM(29–52), retain only the terminal amide; whereas the remaining two fragments, AM(11–28) and AM(11–22), have the loop but not the carboxy end of the molecule. We synthesized all these peptides to further characterize their biological activities.
MMP-2 digestion of AM generated novel peptide fragments
Predicted fragments of MMP-2-digested AM are found in vivo
To investigate whether some of the fragments obtained by in vitro digestion of AM are also present in a biological fluid, we assayed urine samples from healthy human volunteers for signature AM peptide fragments generated by MMP-2 cleavage. Urine is an abundant source for both MMP-2  and AM . First, we standardized a preparative C-18 HPLC column using the synthetic peptides as markers and identified the fractions in which particular peptides eluted from the column (Figure 3a, arrows). Using the same conditions, we fractionated the equivalent of 250 ml of urine and analysed selected fractions by Western blotting with a well-characterized antibody against AM . Chemiluminescent detection reveals a moiety of approx. 6 kDa (that co-migrates with synthetic AM) in fractions 95 and 96, and smaller fragments in fractions 57, 58, 63 and 64 (Figure 3b). Comparison of these results with the elution profile of the synthetic peptides identifies the low-molecular-mass band observed in fractions 57 and 58 as AM(29–52), the one in fractions 63 and 64 as AM(23–52), and the larger peptide of fractions 95 and 96 as undigested AM. As our detection antibody recognizes the carboxy end of AM , only the fragments containing this region could be detected by Western blot analysis.
Urine from normal volunteers contained products of MMP-2-dependent AM degradation
Peptide fragments exhibit both hypotensive and hypertensive activity
Vasodilatation is the best characterized function of AM [8,9]. To understand the physiological implications of the digestion of AM by MMP-2, we studied the impact of the AM fragments on blood pressure regulation in anaesthetized rats. Untreated animals had a systolic blood pressure of 125±10 mmHg (n=5). Intact AM (Figure 4a) and the peptide fragments containing both the intramolecular disulphide loop and the final tyrosine amide (Figure 4b) induced a deep and long-lasting hypotension (reduction of 55±5 mmHg, n=5). In contrast, AM(23–52), AM(29–52) and AM (11–28) do not have any discernible effect on blood pressure regulation (Figure 4c). Interestingly, the fragment AM(11–22) showed a vasopressor effect (71±24 mmHg over basal levels, n=5, Figure 4d), indicating that MMP-2 degradation of AM not only attenuated the hypotensive effect of AM, but that it also generated a hypertensive fragment. Furthermore, the hypotensive and hypertensive peptides exhibited very different modes of action, with the vasodilator molecules acting almost immediately following injection, and the vasoconstrictor peptide needing 4–5 min before eliciting its effect. This divergence in timing suggests that AM(11–22) may be acting through a receptor system independent of the one used by AM and the larger peptide fragments. To demonstrate that these effects on blood pressure are not influenced by the anaesthesia, the synthetic peptides were also tested in conscious animals. Given the complexity of the technique, only one animal per peptide was used and therefore no quantification is possible, but in all cases the individual values were within the range obtained in anaesthetized animals.
One of the new fragments elevated blood pressure in rats
The vasoconstrictor peptide does not act via a classic AM receptor mechanism
To determine if AM peptide fragments elicit their hypotensive and hypertensive effects via the AM receptor, we examined their cAMP response in fibroblasts. Rat2 is a fibroblast cell line that contains a well-characterized AM receptor and does not bind other members of the AM peptide family, such as CGRP (calcitonin-gene-related peptide) [20,21]. AM(8–52) and AM(11–52) elicited an intracellular elevation of cAMP in Rat2 cells equal to the one induced by the intact AM molecule, whereas the rest of the fragments did not elevate cAMP levels over basal values (Figure 5a). These observations confirmed a previous report that indicated loss of either the disulphide bond or the terminal amide results in inability of receptor binding . We also studied whether the peptides that did not elicit a cAMP response are in fact competitors for AM binding to its receptor. This seemed not to be the case, since increasing concentrations of these peptides did not have any effect on the induction of the cAMP response by AM, as exemplified by AM(11–22) (Figure 5b). This was also compatible with the vasoconstrictor activity of AM(11–22) being mediated through a different receptor system.
Some digestion products were no longer able to activate the AM receptor
In the present study we have shown that AM is rapidly cleaved by MMP-2 and that as a result smaller peptides are sequentially produced. The specificity of the reaction was demonstrated by the fact that another gelatinase, MMP-9, did not affect AM integrity. Interestingly, the AM-binding protein factor H efficiently blocked MMP-2 degradation of AM, while it did not inhibit catabolism of other MMP-2 substrates. Also, factor H was not a substrate for MMP-2. Unique peptide fragments, consistent with MMP-2-mediated degradation of AM, were found in a biological fluid, suggesting that the in vivo catabolism of AM is predicted, at least in part, by MMP-2 activity, although at this moment we cannot exclude that other MMPs may also process AM in vivo. The peptide products that retained the amidated end of the molecule and the internal disulphide loop induced vasodilatation in vivo and elevated cAMP levels in a cell line known to express the specific AM receptor, whereas peptides lacking either of these features did not. The smallest peptide fragment, which contains little more than the internal disulphide loop, elicited a hypertensive response without influencing AM receptor activation. These observations suggest that MMP-2 activity may contribute to the hypertensive phenotype both by reducing the levels of the potent vasodilator AM and by generating a new hypertensive peptide (Figure 6).
Diagram of the sequential degradation of AM into smaller peptides and the physiological implications of the process
In a previous study, Lewis et al.  exposed synthetic human AM to cell membrane preparations from sheep kidney, adrenal and lung tissues. Among the series of peptide fragments generated in this manner, one of the most abundant was AM(8–52), which we have also found as one of the first products resulting from MMP-2-mediated degradation. In the discussion of that article , the authors predicted the need for metalloproteinase and aminopeptidase activities to explain all the fragmentary peptide products observed.
We had previously reported that factor H enhances AM function in several experimental model systems, such as induction of cAMP in Rat2 fibroblasts , growth promotion of breast cancer cell lines  and reduction of insulin secretion by rat isolated pancreatic islets . We also determined that this enhancing effect was not due to changes in the kinetics of AM binding to its receptor . In the present study we have shown that factor H was able to prevent the degradation of AM by MMP-2, therefore defining a mechanism by which the binding protein prolongs the half-life of AM and thus increases its biological effects. This regulatory process may be very relevant in understanding the biology of AM in regulating vascular tone. Current radioimmunoassay protocols used to determine the circulating levels of AM require a preliminary purification step through a C-18 cartridge. An unanticipated consequence of this step is the removal of factor H and the fraction of AM that is bound to it . As a result, these protocols measure only free AM and the reported concentrations are extremely low, varying from 1 to 10 pM . Obviously, these levels are insufficient for receptor activation and have raised doubts about the endocrine effects of AM. On the other hand, the existence of a serum AM-binding protein, which circulates at a high concentration (500 μg/ml)  and protects AM from proteolytic degradation, suggests that the most important pool of AM may be the one that circulates bound to factor H, rather than the free fraction. Additional studies are needed to address this provocative question in disease states.
The finding of a vasoconstrictor peptide as a result of the digestion of AM by MMP-2 was intriguing and is in accordance with a previous study  where different fragments of AM were arbitrarily synthetized and assayed for blood pressure regulation activity in anaesthetized rats. Although the fragment AM(11–22) was not studied in that report , the structurally related peptide acetyl-AM(16–21) showed vasopressor activity. Another similar fragment, AM(11–26), was purified from bovine adrenal medulla, but, although the peptide induced vasopressor activity, the elevation in blood pressure lasted only about 70 s . The involvement of the catecholamine system and α-adrenergic receptors has been suggested as a mechanism to explain the pressor activity of these particular fragments of AM [24,26]. Our observations of the delay between the administration of AM(11–22) and the onset of the vasopressor response, together with the disconnection of AM(11–22) and the AM receptor system, is consistent with the existence of an alternative mechanism responsible for the reported effects. Treatment with α-adrenergic receptor blockers, such as phentolamine or reserpine, may resolve whether AM(11–22) affects this pathway. AM has been previously found in urine and its physiological impact in renal homoeostasis has been extensively studied . Our finding of MMP-2-induced fragments of AM being in the urine suggests that renal physiology may also be modulated by these peptide fragments, either by affecting directly glomerular/tubular function or by altering renal hemodynamics. Further investigation in this field is warranted.
There are precedents where MMP-2 can also cleave other vasoactive substances. For instance, this enzyme digests the vasoconstrictor big endothelin-1, ET-1(1–38), yielding the smaller peptide ET-1(1–32), which is also a vasoconstrictor . Furthermore, CGRP is also degraded by MMP-2, but, in contrast with our observations on AM, the resulting vasoconstriction is just a consequence of the reduction in the levels of the vasodilator CGRP . All these observations suggest that MMP-2 inhibitors may be useful as regulators of high blood pressure.
The ability of MMP-2 to radically change the physiological action of a substrate has been demonstrated before in the case of MCP-3 (monocyte chemoattractant protein-3). Full-length MCP-3 induces chemotaxis of mononuclear inflammatory cells, but the cleavage products act as general chemokine antagonists and dampen inflammation .
MMP-2 is also a neutral endopeptidase and investigators are intensively studying clinical applications for inhibitors of this enzyme family [6,7]. There is evidence that three different neutral endopeptidase inhibitors (candoxatrilat, thiorphan and SCH32615) enhance clinical aspects attributed to AM [30–32]. Whether this enhancement is due to MMP-2 blockade or to other endopeptidases remains to be determined. Specific inhibitors of MMPs are being used to prevent extracellular matrix remodelling in cardiovascular diseases, with encouraging results . It should be interesting to investigate whether AM, a protective agent in cardiovascular disorders , is responsible for some of the beneficial actions generated by MMP inhibitors in cardiovascular disease states. Our observation that MMP-2-mediated digestion of AM generates a vasoconstrictor out of the original vasodilator peptide defines a new mechanism by which MMP-2 further contributes to regulation of vasomotor tone, and suggests that application of MMP-2 inhibitors may regulate blood pressure.