| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Institut National de la Santé et de la Recherche Médicale (INSERM), Unité (U) 808 (D.P., I.F., P.T., P.A.K., V.G.), F-75730 Paris Cedex 15, France; INSERM, U809 (N.B.), F-75730 Paris Cedex 15, France; Université Paris Descartes (D.P., I.F., N.B., P.T., P.A.K., V.G.), Faculté de Médecine René Descartes—Site Necker, F-75015 Paris, France; and Department of Endocrinology and Reproductive Medicine (P.T.), GH Pitié Salpêtrière Hospital, F-75651 Paris Cedex 13, France
Address all correspondence and requests for reprints to: Dr. Vincent Goffin, Institut National de la Santé et de la Recherche Médicale, Unité 808, Faculté de Médecine Necker, 156 rue de Vaugirard, Paris 75015, France. E-mail: goffin{at}necker.fr.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Cryptic antiangiogenic factors are a growing class of negative angiogenesis regulators (9, 10). Although they are genetically or structurally unrelated, they share the fact that they are buried within larger precursors, which themselves do not exert antiangiogenic activity, and are released by the processing of the latter. The pioneering member of this family is 16K PRL, an N-terminal 16-kDa fragment of the pituitary hormone prolactin (PRL), whose antiangiogenic properties were discovered 15 yr ago (11, 12). Soon thereafter, angiostatin (13) and endostatin (14) were identified as antiangiogenic polypeptides released respectively from plasminogen, a circulating protein, and collagen XVIII, a protein of the extracellular matrix (ECM). More recently, other cleavage products of ECM proteins were shown to also exert antiangiogenic properties, including tumstatin, arresten, canstatin, vastatin, or restin (9, 15). Today, the cryptic antiangiogenic family includes more than 20 molecules, the majority of which were identified in the context of human or experimental tumors (9, 10).
The molecular and cellular mechanisms by which 16K PRL, endostatin, and angiostatin exert their antiangiogenic and antitumor properties have been intensively studied and are beginning to be better understood (16, 17, 18). In sharp contrast, very little is known about the mechanisms by which these cryptic factors are generated from precursors in vivo. In vitro studies identified various members of the cathepsin family as candidate proteases able to generate 16K PRL (19), angiostatin (20), and endostatin (21). However, because cathepsins exert their activity at acidic pH, all these studies were performed using acidified cell lysates, cell conditioned media, or tissue homogenates (12, 19, 20, 22), or even by culturing cell monolayers in culture media adjusted to pH 5.5 (21). The physiological relevance of cathepsins in the generation of these cryptic fragments thus awaits the demonstration that these enzymes are also able to process the cognate precursors under physiological conditions. Because cathepsins are lysosomal proteases, one mechanism could obviously involve intracellular proteolysis. With respect to PRL, receptor-mediated internalization of the circulating ligand, followed by proteolysis within lysosomes (or other acidic compartments) and release of the cryptic fragment into the extracellular medium has been proposed as a possible mechanism (23). Alternatively, because cathepsins can be secreted (24, 25), they could also process the precursors outside the cell, providing the acidic conditions required for their activity can be achieved.
In this study, we have challenged the mechanisms of antiangiogenic peptide generation with respect to 16K PRL. The rationale for using PRL as the preferred substrate was based on two criteria. First, in contrast to endostatin and angiostatin, 16K PRL is generated by a unique enzyme, namely cathepsin D (Cath D) (16, 19); therefore, the proteolytic process can be analyzed without any ambiguity regarding its acid pH dependency. Second, the compartment in which PRL processing occurs (intracellular or extracellular) could be unambiguously addressed using tissues from PRL receptor knockout mice (PRLR KO) (26) because we previously showed that PRL internalization cannot occur in non-PRLR-expressing cells (27). Using this unique KO model, we demonstrate that the PRLR is dispensable for Cath D-mediated processing of PRL. The proteolytic activity of the enzyme requires local acid secretion, mainly driven by cell proton extruders of the Na+/H+ exchanger (NHE) family and H+/ATPAse. Although it is usually assumed that cathepsin-mediated generation of cryptic antiangiogenic peptides occurs in the moderately acidic pericellular milieu found in malignant tumors (28), we propose a new mechanism by which secreted acidic proteases could be active in the extracellular milieu to process large precursors into antiangiogenic peptides under physiological conditions.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
). Full-length mouse (m)PRL could be detected only in the circulation of PRLR KO animals, in agreement with the fact that the latter are hyperprolactinemic (26, 29). Otherwise, we could not detect any 16K-like mPRL fragment. Although in some experiments, overloading the gels displayed 14- to 16-kDa bands, these could not be assigned to mPRL-related products because they were also detected in WT sera, whereas full-length mPRL itself was not detected (data not shown). The failure to detect any 16K mPRL fragment in serum suggested either that the latter is not circulating, or that our immunoblot procedures were not appropriate. To address these issues, we analyzed two PRL-producing tissues, namely the pituitary (the major PRL-secreting tissue), and the prostate from transgenic mice overexpressing rat (r)PRL specifically in this tissue (30) (Fig. 1, B and C). Both full-length rPRL and an immunoreactive fragment comigrating with control 16K rPRL (produced by in vitro digestion of recombinant rPRL by Cath D) were strongly detected in the prostates of these transgenic animals, especially in the dorsolateral lobe (Fig. 1B). As expected, neither rPRL nor 16K rPRL were detected in prostates of WT mice used as controls. Not only did these data validate our experimental procedures for detecting 16K-like PRL fragments, but they also suggested that PRL processing could be a local phenomenon because neither rPRL nor 16K rPRL were detected in serum of these animals (data not shown). With respect to the pituitary, immunoreactive bands migrating slightly faster than control 16K rPRL were detected in addition to mPRL (left lanes on Fig. 1C). This is reminiscent of previous reports showing that 14K rPRL fragments can be detected in rat neurohypophisis and hypothalamus (16, 31). Next, to evaluate whether the PRLR is involved, or required for in vivo PRL processing, we investigated the presence of endogenous PRL fragments in pituitaries from PRLR KO mice. As shown on Fig. 1C (right lanes), they were indistinguishable from WT pituitaries with respect to the presence of PRL fragments. Finally, the ability of PRLR-deficient tissues to produce 16K-like fragments was also analyzed in another PRL-secreting tissue, i.e. the placenta. When egg implantation could be successfully rescued in PRLR KO mice by progesterone treatment, as previously reported (32), we showed that limited amounts of an immunoreactive mPRL fragment comigrating with control 16K mPRL were also detected in the placenta of these animals (Fig. 1D).
|
Processing of Exogenous rPRL in the Absence of PRLR
We first analyzed the processing of purified recombinant rPRL injected ip into PRLR KO mice. To avoid cross-reaction with any endogenous PRL-related product, we used biotinylated rPRL (rPRLbiot). Preliminary experiments confirmed that PRL biotinylation did not interfere with the proteolytic efficiency of Cath D in vitro (data not shown). Blood samples were harvested 0–90 min after rPRLbiot injection and were analyzed by blotting using streptavidin-horseradish peroxidase (HRP) (Fig. 2A). Rat PRLbiot was rapidly processed because 16K rPRLbiot was detected 30–60 min after injection. When leupeptin, an inhibitor of serine and cystein-proteases, was preinjected into the animals, no inhibition of PRL processing was observed. Although this might be due in part to the instability of this inhibitor, previous studies have shown that leupeptin maintained its ability to inhibit enzyme activity at least during the first hour after injection (33). In contrast, preinjection of pepstatin A, a well-known inhibitor of Cath D activity, totally abolished PRL processing. To the best of our knowledge, these data provided the first evidence that Cath D is able to process PRL in vivo. In addition, the use of PRLR KO mice demonstrated that this phenomenon did not require PRLR-mediated internalization of PRL, which strongly suggested that Cath D processes rPRL in the extracellular compartment.
|
, lane 1) could be processed in a time-dependent manner into an immunoreactive 16-kDa product (left panel) corresponding to 16K rPRL, and a smaller fragment (7 kDa), that was not visible in most of our experiments. No significant spontaneous degradation was observed after incubation of rPRL for 5 h at 37 C in the absence of tissue (lane 2).
When the same experimental approach was performed using kidneys harvested from PRLR KO mice, rapid rPRL proteolysis into 16K rPRL was also observed (right panel). Because this in vitro model confirmed the conclusions of in vivo studies (Figs. 1, C and D, and 2A), i.e. that PRL processing is independent of the PRLR, it appeared to be suitable for investigating the processing of PRL in the context of living tissues, and it was therefore used throughout the present work along with WT animals.
In Conditioned Medium, Secreted Cath D Processes PRL only at Acidic pH
To confirm that the proteolytic reaction occurs in the extracellular compartment, culture medium conditioned for 3 h in the presence of kidney explants was used to test its ability to process rPRL into 16K rPRL. Results were identical whether tissues were harvested from WT or PRLR KO mice; therefore, only data obtained with the latter are presented on Fig. 3A. When incubation was performed using conditioned medium directly taken from explant cultures, i.e. without adjusting the pH, rPRL proteolysis failed to be detected (Fig. 3, first lane of left panel). Significant rPRL processing occurred when pH of conditioned medium was dropped to values equal to or below 5.5 (Fig. 3A, left panel). This pH dependency of the proteolytic pattern was very similar to that obtained with purified Cath D (Fig. 3A, right panel), the main candidate involved in this process (19). Secretion of Cath D by kidney explants was confirmed by Western blot analysis of conditioned medium (Fig. 3B). Three Cath D-related bands were detected: the inactive precursor (
50 kDa), and mainly the active forms, referred to as mature single chain (43 kDa) and mature large chain (33 kDa), the latter comigrating with commercially available Cath D (Fig. 3B, left lane). When Cath D was immunoprecipitated from conditioned media using specific antibodies (Fig. 3C), the proteolytic activity observed after acidification became associated exclusively with the pellet, whereas it remained associated with the supernatant when anti-Cath D antibody was omitted in control immunodepletion experiments.
|
In Explant Cultures, Secreted Cath D Processes rPRL at Physiological pH
Although we routinely observed slight acidification of culture media after 5 h of explant culture, the minimal values detected (pH 6.8) were not sufficient with the acidic conditions required to detecting Cath D activity (see Fig. 3A). Thus, we suspected that the proteolysis of rPRL observed in kidney explants (Fig. 2B) could be mediated by an enzyme different from Cath D, and active at neutral pH. Thrombin is a protease that was shown to process PRL at pH 7. Intriguingly, a major cleavage product is also a 16-kDa fragment, but it differs from the so-called 16K PRL because it in fact corresponds to the C-terminal sequence of PRL, and it is biologically inert (34). As shown in Fig. 4A, the 16-kDa rPRL fragment generated in kidney explant cultures comigrates with N-terminal 16K rPRL generated by Cath D, whereas thrombin-generated C-terminal 16K rPRL exhibits slower electrophoretic mobility. This confirms that the protease involved in kidney explants was not thrombin, but most likely Cath D. Because activity of an acidic protease under physiological pH is unexpected, the involvement of this protease in rPRL cleavage was further evaluated by other approaches.
|
), which supports that amino acids 145–146 constitute the proteolytic site for the enzyme involved. We further confirmed the involvement of Cath D by using two inhibitors of its activity. When pepstatin A was added to culture medium, rPRL processing by kidney explants was dramatically altered (Fig. 4C, left panel). Because pepstatin A is an inhibitor of all aspartate proteases, it is not strictly speaking specific for Cath D. We confirmed these data by using the ProN peptide, which was previously shown to be a specific inhibitor of Cath D activity (Ref35 ; also Materials and Methods). This peptide inhibited rPRL processing by kidney explants in a dose-dependent manner (Fig. 4C, right), definitely confirming that Cath D is the enzyme involved in extracellular rPRL processing by kidney explants. These data demonstrate that, in a model of kidney explant maintained at physiological pH, Cath D is able to generate the cleaved PRL isoform from which antiangiogenic 16K rPRL is released.
Cath D Activity in Explant Cultures Requires Functional Proton Extruders
Because of the acidic requirements of Cath D activity, we suspected that the slight acidification observed after few hours of explant cultures might be sufficient to allow Cath D to exhibit some activity (Figs. 2B and 4) that, for unknown reasons, could not be observed when using conditioned media (Fig. 3A). To address this hypothesis, the medium used for explant cultures was carefully buffered by addition of HEPES (50 mM). We assessed that the initial pH of culture media (pH 7.4) remained unchanged during the 5 h incubation of tissue explants. Compared with experiments performed in standard medium (pH 6.8–7 after 5 h), rPRL cleavage was reduced by approximately 30% in the presence of high HEPES concentration, as measured by densitometric analysis of Western blots (Fig. 5A). Although this experiment highlighted the pH-sensitivity of Cath D activity in explant cultures, it further confirmed that the enzyme is able to process rPRL at globally neutral pH.
|
) from tissue explants (proteolysis at pH > 6.8, Figs. 2B, 4, and 5A), we investigated the hypothesis that the tissue might provide a local acidic environment compatible with enzymatic activity of secreted Cath D. Acidification of the extracellular space can be achieved by various mechanisms extruding acids outside the cell. One such mechanism involves intrinsic cellular proton extruders controlling intracellular pH homeostasis, mainly represented by NHE and H+/ATPase pumps. Ethylisopropylamiloride (EIPA) and bafilomycin A1 are well-characterized pharmacological inhibitors of NHE and H+/ATPase, respectively (36, 37). As shown in Fig. 5B (top left panel), EIPA dose-dependently inhibited rPRL processing when added to the incubation medium of kidney explants, with maximal (but not total) inhibition observed at 1 µM EIPA. Addition of bafilomycin A1 also inhibited rPRL processing dose-dependently (Fig. 5B, bottom left panel), although the maximal effect (obtained at 1 µM) tended to be less marked than for EIPA. For both inhibitors, identical results were obtained using PRLR KO or WT tissues (one of each is shown), strengthening the assumption that the proteolytic phenomenon observed in this model does not involve hormone internalization. It is noteworthy that no change in global pH of incubation medium was observed in any of the experiments involving the inhibitors. Also, PRL processing by purified Cath D (pH 4) was not affected by addition of EIPA or bafilomycin A1 (Fig. 5B, right panels), which assessed that the effects of both inhibitors did not involve nonspecific interference with intrinsic enzyme activity.
Validation of the Model in Various Tissues
All the data presented in Figs. 1 to 5 converge to the conclusion that rPRL processing into 16K rPRL by kidney explants 1) takes place outside the cell, 2) is mediated by Cath D, 3) does not require experimental acidification, and 4) requires functional proton extruders. We next wanted to evaluate whether this mechanism could also be observed with other tissues. First, we assessed that Cath D was secreted by other tissues, including spleen, liver, brain, lung, heart, and prostate (kidney was added in all these experiments as a control). Although the absolute amount of secreted Cath D differed from one organ to another, the enzyme was detected in culture medium conditioned for 6 h in the presence of all tissues investigated (Fig. 6A). Active forms (mature single chain and mature large chain) largely predominated, although faint amounts of pro-Cath D could be detected in kidney, lung, and heart media. We then performed rPRL proteolysis experiments following procedures similar to those described for kidney explants. Experiments were performed using both WT and PRLR KO mice, with no detectable difference. As shown on Fig. 6B, the proteolytic efficiency was as follows: strong (kidney, spleen), moderate (prostate, heart), moderate-low (brain) and repeatedly nil (liver). We failed to achieve fully reproducible results with lung explants, although cleavage efficiency was at best very low. There was no clear correlation between the amount of secreted Cath D (Fig. 6A) and cleavage efficiency (Fig. 6B). PRL incubated with nonacidified conditioned media (Fig. 6C, top panel) was almost indistinguishable from control, with at best the faint doublet characteristic of poor cleavage observed for spleen and brain explants. In contrast, when conditioned media were acidified (Fig. 6C, bottom panel), rPRL processing was observed for all tissues, which demonstrated that the variable cleavage efficiencies observed in explant experiments (Fig. 6B) did not result from any alteration of the intrinsic properties of Cath D. In many experiments, the proteolytic pattern obtained using prostate explants or conditioned media (acidified) displayed one additional band of approximately 20 kDa, the nature of which was not determined (arrowheads in Figs. 6C and 7B).
|
|
). Both EIPA and bafilomycin A1 led to significant inhibition of rPRL processing by brain and spleen explants, as already observed for kidney (Fig. 7A, top panel). We confirmed that Cath D secretion remained unchanged after 5 h incubation with or without inhibitors (Fig. 7A, bottom panel).
In contrast to these three tissues, proteolysis by heart and prostate explants was not affected by inhibitors of proton extruders (Fig. 7B, top panels), suggesting that other acidification mechanisms and/or other enzymes could be involved. This was investigated using above-mentioned inhibitors. As observed for the kidney (Fig. 4C), the ProN peptide dose-dependently inhibited rPRL processing by the heart and the prostate (Fig. 7B, bottom panels), strengthening a role for Cath D in both tissues. In good agreement, proteolysis of rPRL by prostate explants was pH sensitive because it was markedly inhibited by HEPES buffering (Fig. 7B, top right panel), as observed for the kidney (Fig. 5A). These data strongly suggested that the difference between the heart and the prostate, on the one hand, and the other tissues, on the other hand, mainly resides in the intrinsic mechanisms of acidification, although minor involvement of other enzyme(s) could not be totally ruled out based on our observations.
While this paper was in revision, the group of Clapp showed that matrix metalloproteases (MMPs) were able to process PRL into a 17K fragment (amino acids 1–155) in a model of chondrocytes (38). Although the PRL fragment that we obtained in our experiments was clearly the classical 16K PRL (amino acids 1–145), the involvement of MMPs was nevertheless investigated for three tissues representative of the different categories: kidney (strong cleavage, EIPA sensitive), heart (moderate cleavage, EIPA insensitive), and brain (moderate-low cleavage, EIPA sensitive). MMP inhibition was performed by adding EDTA into incubation media of explants, as earlier described (38). As shown on Fig. 8, rPRL processing by heart explants was not affected by the addition of EDTA, strengthening that the main enzyme involved is Cath D (Fig. 7B). It is of note that secretion of the latter was not modified in the presence of EDTA (Fig. 8, bottom left panel). With respect to the kidney, we wanted to explore whether MMPs could be responsible for the cleavage observed at strictly physiological pH, i.e. when culture medium was buffered by HEPES (as shown in Fig. 5A). Although partial inhibition of rPRL processing was observed at the highest EDTA concentrations that we tested (Fig. 8, arrowheads on middle top panel), this was concomitant to a significant decrease of Cath D secretion (Fig. 8, arrowheads on middle bottom panels). Similar observations were made when kidney explants were incubated in the absence of HEPES (data not shown), as well as with brain explants (right panels). In agreement with the arguments supporting the role of Cath D in PRL processing by these tissues, the inhibitory effect exerted by 5 mM EDTA is probably due more to the reduction of Cath D secretion than to the inhibition of MMP activity.
|
demonstrate that the model of extracellular PRL proteolysis is valid for many tissues, primarily involves secreted Cath D, is pH sensitive, and requires local acidification, which can be achieved by various mechanisms, that nonexhaustively include Na+/H+ exchangers and H+/ATPase. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
5.5) potentially compatible with Cath D activity (31, 45, 46). Alternatively, because PRL processing was shown to also occur in explants of various non-PRL-producing tissues (16, 23, 47), the generation of 16K PRL could involve PRL internalization via its receptor, intracellular processing inside acidic compartments containing Cath D (such as lysosomes), and finally release of the antiangiogenic fragment into the extracellular space. This mechanism remains speculative, however, because, to the best of our knowledge, it has never been specifically addressed. Finally, a third hypothesis has recently emerged. Based on the observation that minimal amounts of 16K-like PRL fragments could be generated by incubating rPRL in conditioned medium of rat mammary gland acini, it was suggested that PRL cleavage could take place in the extracellular milieu and involve secreted Cath D (47). This hypothesis remained very fragile, however, because 1) very partial PRL cleavage could be obtained using this experimental model, as highlighted by the fact that the authors detected the 16- to 16.5-kDa doublet characteristic of incomplete proteolysis at 145–146 site; 2) no mechanism was proposed to explain the very low activity of this acidic enzyme at neutral pH; and 3) intracellular proteolysis could not be definitely ruled out because the experimental model used in that study involved PRLR-expressing tissue shown to internalize PRL. In the present study, we used tissues from PRLR KO mice, in which PRL internalization cannot occur. The use of explants is perhaps open to criticism, because we cannot discard that tissue damage, occurring either during slicing itself or during short term incubation (e.g. hypoxia), may interfere with experimental observations. However, the inhibition of PRL processing observed using inhibitors of H+ extruders supports that these tissues maintained intrinsic control of intracellular/extracellular pH, and that the phenomena that we observed were mainly under active cell control. In addition, the striking differences of proteolytic activity between conditioned media or explant cultures demonstrated that PRL processing was not simply due to nonspecific release of enzymes from damaged cells. Taking advantage of this unique model, we provide the first, unambiguous demonstration that PRL cleavage does not require the receptor, meaning it can occur in the extracellular milieu. Because we confirmed that the protease involved in the generation of 16K PRL was the one expected, i.e. Cath D, our observations again addressed the question raised above regarding the mechanism by which this acidic enzyme could be catalytically active in explant cultures maintained in neutral pH media.
Based on structural comparisons of free Cath D at acidic pH and pepstatin A-bound Cath D at neutral pH, Lee et al. (40) suggested that high affinity substrates of aspartyl-proteases could possibly maintain the enzyme in its active conformation at neutral pH. However, such high affinity binders remain to be identified among natural substrates of these proteases, providing they exist. A more recent study proposed a similar mechanism for cathepsin L, a cysteine-protease also known to function at acidic pH. When this enzyme is associated with p41, an MHC-class II associated invariant chain, it remains active at neutral pH, and this interaction was proposed to be a means to accumulate catalytically active enzyme in the extracellular milieu (48). Although such hypotheses could eventually explain the very modest PRL cleavage occurring at neutral pH in conditioned media (Ref. 47 and Fig. 6C), they are unlikely to account for the strong proteolytic activity observed in some tissue explants (Fig. 6B). We then investigated whether the tissues themselves could provide local acidic environments required for Cath D activity outside the cell.
Local acidification of the extracellular space has been described in very specific cell or tissue contexts, and this was shown to be closely related to highly specialized biological responses or cell behaviors. The best documented cases involve solid tumors, which exhibit a moderately acidic extracellular environment (49). This has been linked to lactic acid production, resulting from hypoxia induced by limited vascularization of growing tumors (50, 51, 52). Extracellular acidification was also shown to involve increased H+ extrusion, mediated by the two major membrane proton extruders: NHE, also referred to as antiporters, and vacuolar-type proton pumps (H+/ATPase) (52, 53). Acidification was demonstrated to be correlated with tumor invasiveness (Ref. 49 and references therein), which suggested increased ECM degradation by acidic proteases (e.g. Cath D and B; Ref. 28), supposed to be activated at low pH. Extracellular acidic microenvironments have also been described to be required for highly specialized functions involving proteases in nontumor contexts. These include bone resorption by osteoclasts (54, 55), acid-dependent hydrolysis of ingested proteins in stomach (56), lipid processing required for maintenance of skin permeability (57), or elastin-degradation by macrophages in inflammatory states (58). Based on the use of knockout models or of well-characterized inhibitors (EIPA, bafilomycin A1, etc.), antiporters, and proton pumps were again identified as the major players in the generation of these acidic microenvironments. Although the involvement of proton extruders and acidic proteases in the abovementioned biological responses have been well established, the functional link between acid extrusion and the catalytic activity of these enzymes has remained speculative. Actually, we are aware of only a single study in which the activity of two low pH-dependent enzymes (cathepsin B and hyaluronidase-2) was directly correlated to NHE activity (53). However, because these observations were made using breast cancer cells, which are known to secrete high amounts of enzymes and to actively acidify the extracellular space (28, 59), their relevance in a physiological context was uncertain. In the present study, we provide evidence that proton extruders also play a critical role in permitting extracellular activity of Cath D under physiological conditions. We propose that proton extrusion generates local acidification of the extracellular milieu compatible with Cath D activity. Although, in various cell models, extracellular space has been documented to achieve only mild acidic pH (between 5.5 and 6.5) (52, 53, 57), enzymes known to exert maximal activity at lower pH are nevertheless able to exhibit detectable activity in these conditions (53, 60). In addition, such pH are compatible with the interaction between the ProN peptide and the active site of the enzyme (60), which is also supported by the inhibitory effect of this peptide reported in this work for kidney, heart and prostate explants (Figs. 4C and 7C). It is thus tempting to speculate that the generation of 16K PRL reported in other studies also using tissue explants was mediated through the same mechanism (16, 23, 47, 61), although none of these reports provided arguments supporting (or infirming) this hypothesis.
NHE antiporters are intrinsic cell components that are known to regulate intracellular pH and cell volume homeostasis (62). Since initial cloning of the first NHE isoform (63), at least eight NHE genes have been identified to date, which encode closely related antiporters expressed either at the plasma membrane (NHE-1 to -5) or in intracellular compartments (other NHE) (64, 65). Although NHE-1 is assumed to be expressed ubiquitously (62), the distribution of other NHE isoforms is more tissue specific (65, 66). NHE expression was also found to be not uniform at the cell membrane of a given tissue. For example, in heart and kidney, NHE were shown to be preferentially located within caveolae (67), which are defined as membrane invaginated smooth vesicles highly enriched in caveolin, cholesterol, and sphingolipids (68). Interestingly, Bourguignon et al. (53) showed that the extracellular activity of cathepsin B and hyaluronidase-2 required the location of active NHE-1 inside caveolae, suggesting that these membrane invaginations could favor local proton concentration. Such delimited microenvironments are reminiscent of the lacunae formed between osteoclasts and bone surface (54), or between macrophages and elastin fibers (58), which both permit proton accumulation. Based on their specific location within caveolae, a new role of NHE in regulating the activity of extracellular acidic enzymes is perhaps emerging. The abundance of caveolae is highly variable among tissues (68), however, so that further studies are obviously required to delineate to which extent these membrane structures direct NHE-dependent activity of secreted acidic enzymes. There are obviously other mechanisms of acid extrusion that should be considered. Based on the inhibitory effect of bafilomycin A1 on PRL processing by brain, spleen, and kidney explants, the involvement of H+/ATPase pumps has also emerged from our study. These pumps were already shown to be involved in the generation of acidic microenvironments at the osteoclast/bone interface (54) and at the macrophage periphery (58). Because vacuolar H+/ATPase are expressed at the plasma membrane of a wide variety of cells (69), it is therefore not surprising that tissues exhibiting sensitivity to EIPA were also affected by the H+/ATPase inhibitor. Acidification mechanisms other than linked to proton extrusion can also be involved, e.g. for the prostate, which is known to secrete large amounts of citric acid (70).
The absence of detectable PRL processing by liver and lung explants remains unexplained. We showed that the two mature forms of Cath D are secreted, as observed for other tissues exhibiting proteolytic activity in explant cultures. Morikawa et al. (20) suggested that carbohydrate moieties could markedly weaken the activity of secreted pro-Cath D under acidic pH. This is likely not the case in our model because acidified conditioned media from lung and liver were as efficient as that from other tissues for processing PRL. NHE deficiency is not likely involved either because these (as well as other) proton extruders are ubiquitously expressed (62, 65, 71). These observations suggest that the inability of some tissues to process PRL is due to parameters other than the expression of Cath D or proton extruders. We could speculate on the role of some structural features, as evoked above for caveolae, or of putative ECM components regulating Cath D activity, which remain to be identified.
Very recently, MMPs secreted by chondrocytes have been shown to generate a 17K PRL fragment exhibiting antiangiogenic properties (38, 72). Although this suggests an important physiological role of these proteases in the control of angiogenesis in cartilage, none of our observations supports a major role for MMPs in other tissues. For example, we failed to observe significant cleavage at neutral pH, and the PRL fragment that we obtained in our experiments was clearly the classical 16K PRL and not the longer fragment (1–155) generated by MMPs. Thus, the mechanism proposed in this work addresses the relevance of secreted Cath D in the potential generation of antiangiogenic factors in a physiological context. The recent publication by Lkhider and colleagues (47) has shown that mammary acini secreted both inactive pro-Cath D and active mature forms of the enzyme, although whether autoactivation occurred before or after secretion remains unclear. Secretion of Cath D was also reported for other cell types, including macrophages (58), fibroblasts (73), keratinocytes (74), or immune cells (25), and indirect evidence based on the ability of acidified serum to process PRL into 16K PRL also suggests that this enzyme is present in the circulation (23). Thus, detection of Cath D in the culture media of all tissues that we investigated is not surprising. Actually, numerous studies have shown that lysosomal proteases are secreted from various cell types, and the picture is emerging that they could originate from either conventional lysosomes (75), or from specific organelles referred to as secretory lysosomes (76). The processing of PRL injected into PRLR KO mice, and the inhibition thereof by pepstatin A is in good agreement with in vitro observations. The functional role of Cath D in physiological context has been evaluated by generating Cath D-deficient mice. Progressive atrophy of intestinal mucosa and destruction of lymphoid tissues were among the main phenotypes of homozygotes, leading to premature death of less than 4-wk-old pups (77). However, the relative roles and relevance of secreted vs. lysosomal enzyme remain unknown. Because our data provided direct evidence that cleaved PRL, the precursor of 16K PRL, can be generated extracellularly in nontumor contexts, the generation of this antiangiogenic peptide could not be restricted to intracellular proteolysis in PRL-producing cells, as long suspected, but could also occur by extracellular processing of circulating PRL in many Cath D-secreting tissues. The fact that PRL fragments could not be detected in serum from PRLR-KO or prostate-specific PRL transgenic mice, whereas they were detected in their pituitaries or prostates, respectively, also suggests that PRL processing could be mainly a local process, with minimal release of cleavage products into the circulation. If this assumption is correct, it is tempting to postulate that 16K PRL could exert its antiangiogenic properties within the tissue in which it was generated.
Interestingly, angiostatin and endostatin are two other well-known antiangiogenic fragments that are also produced by proteolysis of extracellular precursors, plasminogen and collagen XVIII, respectively (10). Despite that these cryptic peptides were identified in the circulation as well as in the ECM of normal subjects/mice (6, 9, 78) the mechanism by which they are generated in vivo remains elusive, although it is widely assumed to occur in the ECM (as reflected by their name matrikines) (9). In vitro evidence has suggested that angiostatin and endostatin can be generated by multiple enzymes, including secreted Cath D and L, respectively (20, 21). Reminiscent of our observations with 16K PRL, acidic conditions (acidified conditioned media or spontaneous acidification of long term high density cultures) were required to see efficient proteolysis of the precursors. Although it is usually extrapolated from this type of observations that the mildly acidic pH environment of tumors could be compatible with the activity of acidic enzymes, the model that we propose suggests that these cathepsins could also participate to the generation of antiangiogenic fragments in nonpathological conditions. If our findings are confirmed by further studies, this could imply that novel antiangiogenic therapies based on the use of cathepsin inhibitors (28, 79) may have the undesirable effect of blocking the generation of cryptic antiangiogenic fragments.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Hormones, Enzymes, Antibodies, and Inhibitors
The expression plasmid for rPRL (pT7L-rPRL) was a kind gift of Dr I. Struman (University of Liège, Liège, Belgium). Recombinant rPRL was produced in bacteria and purified as previously reported for human PRL (80). Mutated rPRL (L146P) was generated by site-directed mutagenesis as previously described (22). Biotinylated rPRL (rPRLbiot) was prepared using the Biotin Labeling kit purchased from Roche (Meylan, France), and we strictly followed the instructions of the manufacturer.
Extractive human liver Cath D, human plasma thrombin and protease inhibitors (pepstatin A, leupeptin) were from Sigma. ProN peptide corresponds to the N-terminal propeptide of pro-Cath D. It binds into the active site of the enzyme at neutral or moderately acidic pH, which results in the inhibition of its catalytic activity (35, 40, 60). It was synthesized by Eurogentec (Sart Tilman, Belgium). EIPA was purchased from Sigma, and bafilomycin A1 was kindly provided by Dr. G. Planelles (INSERM, U806, Necker site, Paris, France). These inhibitors of proton extruders were used at concentrations previously shown to be not toxic (53, 81, 82).
Polyclonal anti-16K rPRL antiserum was previously characterized (31); it preferentially detects 16K PRL, but also cross reacts with full-length PRL. Goat polyclonal antihuman Cath D antibody (sc 6486) were purchased from Santa Cruz (Santa Cruz, CA). Protein-A Sepharose CL-4B was from Amersham Pharmacia Biotech (Orsay, France), and streptavidin-HRP from Sigma. Secondary antirabbit immunoglobulin antibody conjugated to horseradish peroxydase were from Santa Cruz.
Animals
PRLR-deficient (PRLR KO) mice were generated by our group several years ago (26). Colonies were amplified in the animal facility of Necker site (25 C, 12-h light, 12-h dark cycles) on the 129sv genetic background, and animals were fed a pelleted diet ad libitum. The absence of PRLR in KO animals was systematically assessed by PCR as described (26). WT mice were either nontransgenic littermates or balb/c mice. Both males and females were used at different ages, with no detectable influence on the results. Transgenic probasin-rPRL expressing rPRL only in the prostate were described previously (30). All experimental designs and procedures were performed in agreement with the guidelines of animal ethics committee of the Ministère de l’Agriculture.
Tissue Explants
Animals were killed by cervical dislocation. Tissues used for in vitro experiments (kidney, liver, prostate, lung, heart, spleen, and brain) were rapidly harvested after animals were killed, and then tissues were incubated in culture medium with antibiotics. Tissues were carefully cut into small pieces or sliced as previously described (23, 83). These explants were rinsed for 5–10 min in culture medium (two to three changes of medium) under gentle agitation to remove residual blood and tissue/cell debris. Tissue explants were dried on Wathman paper before being weighed, then they were incubated at 37 C/5% CO2 in serum-free culture medium for up to 5 h. The equivalent of one organ (average of 100–200 mg of explants) were aliquoted per well of P24/48 plates in 200 or 400 µl of medium (depending on organ size). The pH of incubation medium was checked at the end of most experiments using a micro-electrode pH meter (IQ 150; Scientific Instruments, London Ontario, Canada).
In Vitro Proteolysis of rPRL
By Tissue Explants.
PRL (5–10 µg/ml) was incubated with tissue explants, in the presence or absence of the following additives, as indicated in the figures: pepstatin A (1 µM), ProN peptide, EIPA, Bafilomycin A1, or HEPES (50 mM). The incubation was routinely performed for 3–5 h. For kinetic studies, aliquots of medium were harvested at the time indicated in the figures.
By Conditioned Medium.
Conditioned medium is defined as culture medium that was incubated for 3–5 h in the presence of the various tissue explants. Proteolysis experiments were performed by incubating rPRL (5 µg/ml) 3 h in conditioned medium, containing or not abovementioned additives. For some experiments, conditioned medium was acidified by incremental dilutions of citrate buffer [50 mM citrate-phosphate/75 mM NaCl (pH 3.2)].
By Purified Enzymes.
For control experiments, 10 µg rPRL were digested at room temperature by purified Cath D in a final volume of 25 µl of citrate buffer, using enzyme/substrate ratio of 1:10 or 1:100 (44). Proteolysis by thrombin was performed in 50 mM Tris-HCl (pH 7.4), using a 1:5 enzyme/substrate (19).
In Vivo Proteolysis of rPRL
Eighty micrograms of rPRLbiot were injected ip into WT or PRLR KO mice (no anesthesia). Blood samples were collected in the eye every 30, 60, and 90 min after injection. For investigations involving protease inhibitors, pepstatin A (50 µg) or leupeptin (50 µg) were injected 30 min before rPRLbiot injection.
Cath D Immunodepletion
One milliliter of conditioned media containing mouse Cath D was incubated overnight at 4 C with 2 µg of goat polyclonal anti-Cath D antibody. Twenty microliters of protein A Sepharose beads were then incubated in the immunoprecipitation mixture for an additional 1 h, and immune complexes were removed by brief centrifugation. Both the pellet and the supernatant were acidified at pH 3 (which dissociates immune complexes from the beads) by addition of citrate buffer, and their proteolytic activity was analyzed by adding rPRL (5 µg/ml).
Western Blot Analysis
Analysis of 16K PRL or mouse Cath D obtained in the various experimental conditions described above was performed using 15–17% reducing SDS-PAGE, respectively. Proteins were transferred onto nitrocellulose membranes and were immunoblotted using anti-16K rPRL antibodies (1/1000 dilution) or antihuman Cath D (1/1000 dilution). Antigen-antibody complexes were detected by enhanced chemiluminescence (ECL, Amersham) using 1/10,000 dilution of antirabbit or antigoat Ig antibody conjugated to HRP. Biotinyliated rPRL was revealed by 1/10,000 dilution of streptavidin-HRP.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Disclosure Statement: the authors have nothing to disclose.
First Published Online September 7, 2006
1 D.P. and I.F. are co-first authors. 
Abbreviations: Cath D, Cathepsin D; ECM, extracellular matrix; EIPA, ethylisopropylamiloride; HRP, horseradish peroxidase; h, human; KO, knockout; m, mouse; MMP; matrix metalloprotease; NHE, Na+/H+ exchanger; PRL, prolactin; 16K PRL, N-terminal 16-kDa fragment of PRL; PRLR, PRL receptor; r, rat; WT, wild type.
Received for publication January 25, 2006. Accepted for publication August 28, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
16K rPRL), which preferentially recognize 16K PRL but also cross-react with full-length, 23K PRL. Samples were harvested from WT, PRLR KO or prostate-specific PRL transgenic mice, as indicated. Each lane corresponds to one different animal, and control 16K PRL was obtained by in vitro Cath D-mediated proteolysis of mouse (mPRL) or rat PRL (rPRL). D, Dorsolateral prostate lobe; V, ventral prostate lobe. See text for description of the results.
