Chromogranin a Restricts Drug Penetration and Limits the Ability of NGR-TNF to Enhance Chemotherapeutic Efficacy

Published OnlineFirst July 28, 2011; DOI: 10.1158/0008-5472.CAN-11-1273

Cancer Therapeutics, Targets, and Chemical Biology Research

Chromogranin A Restricts Drug Penetration and Limits the Ability of NGR-TNF to Enhance Chemotherapeutic Efficacy

Eleonora Dondossola, Anna Maria Gasparri, Barbara Colombo, Angelina Sacchi, Flavio Curnis, and Angelo Corti

Abstract NGR-TNF is a derivative of TNF-a that targets tumor blood vessels and enhances penetration of chemotherapeutic drugs. Because of this property, NGR-TNF is being tested in combination with chemotherapy in various phase II and III clinical trials. Here we report that chromogranin A (CgA), a protein present in variable amounts in the blood of normal subjects and cancer patients, inhibits the synergism of NGR-TNF with doxorubicin and melphalan in mouse models of lymphoma and melanoma. Pathophysiologically relevant levels of circulating CgA blocked NGR-TNF–induced drug penetration by enhancing endothelial barrier function and reducing drug extravasation in tumors. Mechanistic investigations done in endothelial cell monolayers in vitro showed that CgA inhibited phosphorylation of p38 MAP kinase, disassembly of VE-cadherin–dependent adherence junctions, paracellular macromolecule transport, and NGR-TNF–induced drug permeability. In this system, the N-terminal fragment of CgA known as vasostatin-1 also inhibited drug penetration and NGR-TNF synergism. Together, our results suggest that increased levels of circulating CgA and its fragments, as it may occur in certain cancer patients with nonneuroendocrine tumors, may reduce drug delivery to tumor cells particularly as induced by NGR-TNF. Measuring CgA and its fragments may assist the selection of patients that can respond better to NGR-TNF/chemotherapy combinations in clinical trials. Cancer Res; 71(17); 1–10. 2011 AACR.

Introduction vascular damage and inflammatory/immune responses (4, 8). Phase I and II clinical studies with low-dose NGR-TNF, alone The efficacy of chemotherapy in cancer patients is often and in combination with chemotherapy, showed manage- limited by inadequate and uneven diffusion of drugs in tumor able toxicity profile and evidence of disease control, parti- tissues (1). This phenomenon is related to abnormal vascu- cularly in subpopulations of patients with hepatocellular lature and altered composition of tumor tissues, which may carcinoma and pleural mesothelioma (9–14). A randomized cause irregular blood flow, heterogeneous permeability, double-blind phase III study of NGR-TNF (in combination increased interstitial fluid pressure and, consequently, uneven with chemotherapy or supportive care) started in 2010 in drug penetration in tumors (2, 3). patients with malignant pleural mesothelioma (http://www. We have previously shown that a neovasculature-homing clinicaltrialsfeeds.org/clinical-trials/show/NCT01098266). The TNF-a derivative, called NGR-TNF, can be exploited to improve identification of factors that promote tumor resistance to drug penetration in neoplastic tissues (3, 4). NGR-TNF is a targeted-TNF/chemotherapy combination could be, therefore, cytokine–peptide fusion protein, originally developed by our pharmacologically and clinically relevant. group, consisting of TNF fused to the CNGRCG peptide (3). This We have previously shown that chromogranin A (CgA), a peptide is a ligand of a CD13 form overexpressed in the tumor protein released in circulation by many endocrine and neu- neovasculature (5–7). Experiments in mouse models have roendocrine cells, neurons and granulocytes, is an important shown that the administration of ultra-low doses of NGR- modulator of the endothelial barrier function and a potent TNF (picograms) increases the penetration of doxorubicin, inhibitor of TNF-induced vascular leakage in the liver (15, 16). melphalan, cisplatin, gemcitabine, and paclitaxel into solid Structure–activity studies have shown that a bioactive site is tumors (4, 8). NGR-TNF can also induce, at later time points, located in the N-terminal region of CgA (15, 17). A fragment of CgA spanning the N-terminal residues 1–78, called vasostatin- 1 (VS-1), can inhibit a series of effects exerted by TNF on Authors' Affiliation: Division of Molecular Oncology and IIT Network endothelial cells, including phosphorylation of p38 MAP Research Unit of Molecular Neuroscience, San Raffaele Scientific Institute, kinase [p38-MAPK (mitogen-activated protein kinase)], redis- Milan, Italy tribution of VE-cadherin, formation of gaps, paracellular Corresponding Author: Angelo Corti, Division of Molecular Oncology, San Raffaele Scientific Institute, via Olgettina 58, 20132 Milan, Italy. transport of macromolecules (15, 18). VS-1 can also inhibit Phone: 39-02-26434802; Fax: 39-02-26434786; VEGF- and thrombin-induced endothelial cell permeability E-mail: [email protected] (19). Thus, CgA and its N-terminal fragment VS-1 are efficient doi: 10.1158/0008-5472.CAN-11-1273 inhibitors of endothelial cell activation and permeability 2011 American Association for Cancer Research. caused by TNF and by other stimuli.

www.aacrjournals.org OF1

Dondossola et al.

Increased levels of circulating CgA have been detected in VS-1 (3 mg) or CgA (0.2 mgor1mg) 60, 30, and 0 minutes before patients with neuroendocrine tumors, in subpopulations of administration of NGR-TNF (0.1 ng). Two hours later, mice patients with prostate, breast or non–small cell lung cancer were given melphalan or doxorubicin solutions (90 or 80 mg, (NSCLC), and in patients with heart failure, renal failure, respectively). All drugs were administered intraperitoneally hypertension, rheumatoid arthritis, sepsis, hepatocellular (i.p.), diluted with 0.9% sodium chloride, containing 100 mg/mL carcinoma, liver cirrhosis, chronic hepatitis, pancreatitis, endotoxin-free HSA (Farma-Biagini SpA), except for doxoru- inflammatory bowel disease, or atrophic gastritis (20–31). bicin and melphalan, which were diluted in 0.9% sodium Furthermore, increased levels of CgA have been detected also chloride alone. Tumor growth was monitored daily by measur- in patients treated with proton pump inhibitors, a class of drugs ing the tumors with callipers as previously described (4). commonly used to treat acid peptic disorders (32, 33). Thus, Animals were killed before the tumors reached 1.0 to 1.5 cm CgA levels higher than normal values can occur not only in in diameter. patients with neuroendocrine tumors or with tumors that may undergo neuroendocrine differentiation but also in patients In vitro endothelial permeability assay with nonneuroendocrine tumors for a variety of reasons (30). The endothelial permeability assay was done using a trans- Given these premises and considering the protective effects well system (6.5-mm insert, 0.4-mm pore size filters; Corning), of CgA on TNF-induced alteration of the endothelial barrier essentially as previously described, using 40 kDa fluorescein function, we have hypothesized that abnormal levels of cir- isothiocyanate (FITC)–dextran conjugate (1 mg/mL; Sigma), culating CgA might inhibit the synergism between NGR-TNF as a tracer. The FITC–dextran conjugate present in the lower and chemotherapy. To address this hypothesis, we have chamber of the transwell system at various time points was investigated the effect of CgA and VS-1 on the antitumor analyzed using a fluorimeter (excitation, 495 nm; emission, activity of NGR-TNF/chemotherapy against murine lym- 520 nm). The same assays were done using doxorubicin phoma and melanoma, that is, 2 models of nonneuroendo- (200 mg/mL) in place of FITC–dextran (excitation, 471 nm; crine tumors. The rationale for this choice relies on the fact emission, 556 nm). that these models have been largely exploited for preclinical studies on NGR-TNF combined with various chemotherapeu- Results tic drugs and that this drug combination is currently investigated in patients with nonneuroendocrine tumors (3, 4, 8). CgA and VS-1 inhibit the synergistic activity of NGR-TNF We show that both proteins, at levels similar to those found in and chemotherapy in tumor-bearing mice patients, can markedly reduce the response of tumors to NGR- To investigate the effect of CgA and VS-1 on the synergism TNF combined with melphalan or with doxorubicin, by redu- between NGR-TNF and chemotherapeutic drugs, we treated cing the transport and the penetration of these drugs into RMA lymphoma–bearing mice with CgA or VS-1 (1 or 3 mg, neoplastic tissues. respectively, 3 times in 1 hour), then with NGR-TNF (0.1 ng), and finally with doxorubicin (80 mg, 2 hours after NGR-TNF; Materials and Methods Fig. 1A). The combination of doxorubicin with NGR-TNF, but not doxorubicin alone, delayed tumor growth, as previously Drugs and reagents shown (4). This effect was completely inhibited by CgA or VS-1 Melphalan (Alkeran) was obtained from Glaxo-Wellcome; (Fig. 1B). We obtained similar results using a model based on doxorubicin (Adriblastina) was purchased from Pharmacia- B16-F1 melanoma–bearing mice and melphalan. Also in this Upjohn. Murine NGR-TNF (consisting of TNF fused with the case, both CgA and VS-1 inhibited the antitumor effect exerted C-terminus of CNGRCG) were prepared by recombinant DNA by NGR-TNF in combination with melphalan (Fig. 1C). These technology and purified from E. coli cell extracts as described results suggest that CgA can inhibit the synergism between (3). NGR-TNF and chemotherapy and that its N-terminal domain contains a bioactive site. VS-1, CgA, and anti-CgA antibodies Notably, CgA and VS-1 reached concentrations of about Ser-Thr-Ala-CgA1–78 (VS-1), human CgA (hCgA), and mur- 5 nmol/L after 1 to 2 hours, that is, at the time at which ine CgA (mCgA) were prepared by recombinant DNA tech- NGR-TNF is expected to exert its action (Fig. 1D). Notably, nology and expression in E. coli cells as described (34). these levels of CgA are in the range of those observed in certain Anti-human CgA monoclonal antibody (mAb) 5A8 (epitope cancer patients with nonneuroendocrine tumors (31, 37). CgA53-57), B4E11 (epitope CgA68-71, all IgG1k) were produced and characterized as described previously (35, 36). CgA and VS-1 do not inhibit the cytotoxic activity of doxorubicin and melphalan alone and in combination In vivo studies with NGR-TNF Studies on animal models were approved by the Ethical The mechanism of action of CgA and VS-1 was then Committee of the San Raffaele Scientific Institute and done investigated. Neither CgA nor VS-1 affected the cytotoxic according to the prescribed guidelines. C57BL/6 mice (Charles activity of doxorubicin and melphalan, alone and in combina- River Laboratories), weighing 16 to 18 g, were challenged with tion with NGR-TNF, against human umbilical vein endothelial subcutaneous injection in the left flank of 1 105 RMA or cells (HUVEC) and RMA cells (Fig. 2). These results B16-F1 living cells. Six to 10 days later, mice were treated with suggest that the inhibition of the antitumor activity of

OF2 Cancer Res; 71(17) September 1, 2011 Cancer Research

Chromogranin A and NGR-TNF/Chemotherapy

Figure 1. CgA and VS-1 inhibit the synergistic antitumor effects between NGR-TNF and chemotherapeutic drugs. A, tumor-bearing mice were treated, i.p., with CgA (1 or 0.2 mg) or VS-1 (3 mg), NGR-TNF (0.1 ng) and doxorubicin (80 mg) or melphalan (90 mg) at the indicated time. B and C, effect of CgA or VS-1 on the synergistic activity of NGR-TNF and chemotherapy on tumor growth (n ¼ 5 mice per experiment, mean SE). D, serum levels of human CgA and VS-1 as measured by ELISA (mean SE, n ¼ 5 mice per experiment). **, P < 0.01; ***, P < 0.001 by t test (2-tailed).

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 OF3

Dondossola et al.

Figure 2. CgA and VS-1 do not inhibit the cytotoxic activity of NGR-TNF in combination with chemotherapeutic drugs in vitro. A and B, effect of CgA, VS-1, NGR-TNF, and chemotherapeutic drugs on HUVEC cell viability. HUVEC cells were treated for 48 hours with the indicated drugs, alone or in combination. Viable cells were stained with MTT. C and D, the same procedure was applied for RMA cell cytotoxicity assays (mean SE of 2 assays, each conducted in quadruplicate).

NGR-TNF/chemotherapy by CgA or VS-1 was not related to hypothesis that CgA and VS-1 can inhibit the delivery of inhibition of their cytotoxic activity against endothelial or doxorubicin to RMA cells in vivo. tumor cells. More likely, it was related to a decreased drug delivery to tumor cells. Induction of endogenous CgA with omeprazole inhibits the penetration of doxorubicin in RMA CgA and VS-1 inhibit the penetration of doxorubicin tumors induced by NGR-TNF in RMA tumors induced by NGR-TNF To assess the role of endogenous CgA, we investigated the To investigate whether CgA and VS-1 could affect the effect of omeprazole, a drug known to induce the release of delivery of chemotherapeutic drugs to tumor cells in vivo, CgA in circulation in patients, on doxorubicin penetration in we exploited the fact that doxorubicin is a fluorescent com- RMA tumors. Preliminary experiments showed that a single pound and that tumor cell fluorescence is an indication of the administration of omeprazole to mice (1 mg, i.v.) can increase amount of doxorubicin that penetrate tumor cells (4). Tumor- the circulating levels of murine CgA from 0.9 0.08 to bearing mice were treated with doxorubicin and, 2 hours later, 1.48 nmol/L 0.079 (n ¼ 5, mean SE) by ELISA, peaking tumor were excised, disaggregated, and analysed by fluores- 2 hours after injection (Fig. 3B, left). Furthermore, repeated cence-activated cell sorting (FACS). Preadministration of NGR- injection of omeprazole (1 mg/day, for 5 days) induced an TNF to mice (2 hours before doxorubicin) increased the tumor increase in circulating murine CgA, reaching levels of cell fluorescence (i.e., the penetration of doxorubicin in tumor 1.75 nmol/L at day 4. The CgA returned to normal values 4 days cells), as observed previously (4). However, pretreatment with after discontinuation of omeprazole administration. Repeated CgA and VS-1 completely inhibited this effect (Fig. 3A). Notably, administration of omeprazole significantly decreased the VS-1 significantly reduced also the spontaneous penetration of penetration of doxorubicin induced by NGR-TNF in RMA doxorubicin in tumor cells (Fig. 3A). These results support the tumors (Fig. 3B, right). No significant effects were observed

OF4 Cancer Res; 71(17) September 1, 2011 Cancer Research

Chromogranin A and NGR-TNF/Chemotherapy

Figure 3. CgA and VS-1 inhibit the NGR-TNFinduced penetration of doxorubicin in RMA tumor cells in vivo. A, effect of exogenous CgA and VS-1 on the NGR- TNFinduced penetration of doxorubicin in RMA tumor. Mice bearing RMA tumors (diameter 0.5–1 cm) were treated with VS-1 (3 mg), CgA (1 mg), NGR-TNF (0.1 ng), or doxorubicin (360 mg) as indicated. Mice were then killed. Tumors were excised, weighed, disaggregated, and filtered through 70-mm filters. Then, the amount of doxorubicin that penetrated tumor cells was analysed by FACS. Representative FACS analysis and fluorescence positive tumor cells are shown (mean SE, cumulative data of 2 (left) or 3 (right) independent experiments, n ¼ 5–6 mice per group per experiment). B, left: effect of single or repeated administrations of omeprazole (Ome) on CgA serum levels (mean SE; n ¼ 8 mice). Right: effect of omeprazole and anti-CgA mAb 5A8 on the NGR- TNF–induced penetration of doxorubicin in RMA tumors. Mice were injected with omeprazole (1 mg per mouse) and mAb (10 mg per mouse) as indicated. The % of fluorescence-positive cells recovered from RMA tumors is shown (mean SE, n ¼ 6 mice per group). *, P < 0.05; **, P < 0.01; and ***, P < 0.001 by t test.

when omeprazole was administered in combination with mAb CgA and VS-1 do not affect RMA cell membrane 5A8, a neutralizing anti-CgA antibody (Fig. 3B, right), suggest- permeability ing that the effect of omeprazole was mediated by CgA. The observation that CgA and VS-1 can reduce the delivery These results suggest that even a moderate increase of of doxorubicin to RMA tumor cells in vivo prompted us to endogenous CgA is sufficient to significantly reduce the pene- investigate whether these polypeptides could affect the trans- tration of doxorubicin induced by NGR-TNF in tumor cells. port of drugs through RMA cell membranes. To this aim, we

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 OF5

Dondossola et al.

Figure 4. CgA and VS-1 do not affect the penetration of doxorubicin in RMA tumor cells in vitro. RMA cells were incubated with different amounts of CgA, VS-1, or NGR-TNF in complete medium for 1 hour and then with various amounts of doxorubicin (1 hour). Fluorescence intensity of cells (as measured by FACS) is shown (mean SE, cumulative results of 3 independent experiments, each conducted in duplicate).

preincubated RMA cells in vitro with different amounts of NGR-TNF could increase the flux of FITC–dextran from NGR-TNF, CgA, or VS-1 (1 hour) and then with different the upper to the lower chamber of the transwell system. amounts of doxorubicin (1 hour). Finally, we analyzed the CgA (4–20 nmol/L) efficiently inhibited this effect (Fig. 6A, fluorescence intensity of cells by FACS. As expected, increas- top panel). This result supports the concept that this protein ing concentrations of doxorubicin increased the fluorescence can inhibit the paracellular transport of molecules induced by intensity of cells, indicating that doxorubicin could penetrate NGR-TNF through endothelial cells. the cells in a dose-dependent manner under these conditions (Fig. 4, left panel). Neither NGR-TNF nor CgA/VS-1 could affect the fluorescence intensity of doxorubicin-treated cells (Fig. 4, right panels). These results argue against the hypothesis that these polypeptides can affect the transport of doxorubicin through cell membranes.

CgA inhibits the NGR-TNFinduced transport of molecules from the vascular compartment to the tumor cell microenvironment We then investigated whether CgA could inhibit the NGR- TNFinduced transport of drugs from blood to the tumor microenvironment. To this aim, we treated RMA-bearing mice with CgA, alone or in combination with NGR-TNF, followed 2 hours later by Patent Blue VF, a hydrophilic nontoxic dye that does not penetrate into cells and that can be easily quantified. After 5 minutes, the mice were killed and perfused with a saline solution. The amount of Patent Blue VF present in tumor tissue homogenates was then quantified spectropho- tometrically. CgA inhibited the NGR-TNF–induced penetration of Patent Blue VF in tumor tissues (Fig. 5). These results support the hypothesis that CgA can indeed affect the transport of molecules from the vascular compartment to the tumor cell microenvironment induced by NGR-TNF.

CgA and VS-1 inhibit NGR-TNFinduced endothelial barrier alteration The positive effect of TNF on endothelial permeability and vascular leakage is a well-known phenomenon (38). This effect can be efficiently inhibited by CgA and VS-1 (15). Keeping this Figure 5. CgA inhibits the NGR-TNF induced penetration of Patent Blue VF in tumor. RMA tumor–bearing mice (diameter, 0.5–1 cm) were in mind, we tested the hypothesis that NGR-TNF can promote treated with CgA (1 mg), NGR-TNF (0.1 ng), and, 2 hours later, with Patent drug extravasation by affecting the endothelial permeability Blue VF (12.5 mg/mL, 0.1 mL, i.v.). After 5 minutes, mice were killed and and that CgA and VS-1 can inhibit this effect by protecting the perfused. Tumors were excised, homogenized, and centrifuged. The endothelial barrier integrity. To this aim we analyzed, first, the supernatants were mixed with trichloroacetic acid [10% (v/v), final – concentration] and cleared by centrifugation. The absorbance at 405 nm effect of NGR-TNF on the paracellular transport of FITC was then measured using a spectrophotometer (n ¼ 4 mice per group, dextran (a macromolecular tracer) through endothelial box-plots with median, interquartile, and min/max values). *, P < 0.05 cell monolayers cultured in transwell systems. As expected, by t test (2-tailed).

OF6 Cancer Res; 71(17) September 1, 2011 Cancer Research

Chromogranin A and NGR-TNF/Chemotherapy

Figure 6. CgA inhibits the NGR-TNFinduced permeability, gap formation, VE–cadherin delocalization, and p38-MAPK phosphorylation in endothelial cell monolayers in vitro. A, effect of NGR-TNF and CgA on the flux of FITC–dextran (40 kDa) and doxorubicin through endothelial cell monolayers (HUVEC). Experiments were carried out as described in "Materials and Methods" (n ¼ 4, mean SE). *, P < 0.05; **, P < 0.01; and ***, P < 0.001 by t test (2-tailed). B, effect of CgA (2.5 mg/mL) on VE–cadherin delocalization and gap formation (arrows) induced by NGR-TNF (5 ng/mL) in confluent endothelial cells. VE–cadherin was detected by immunofluorescence analysis with a rabbit anti–VE-cadherin antibody and goat anti-rabbit Alexa Fluor 546 conjugate (red). Nuclei were stained with DAPI (blue; representative picture of 2 experiments each conducted in duplicate). Bar, 10 mm. C, effect of CgA on NGR-TNFinduced phosphorylation of p38 MAPK in endothelial cells. Endothelial cells were incubated with NGR-TNF and CgA and analyzed by Western blot with antip38- MAPK and antiphosphorylated p38-MAPK antibodies after 0, 5, and 20 minutes (representative results of 3 experiments).

To assess weather CgA and VS-1 could also inhibit the inhibit the permeabilizing activity of NGR-TNF by preventing transport of drugs induced by NGR-TNF, we carried out a VE-cadherin–dependent adherence junction disassembly. similar experiment using doxorubicin in place of FITC- dextran. Both CgA and VS-1 could inhibit the flux of doxor- CgA inhibits the phosphorylation of p38-MAPK ubicin from the upper chamber to the lower chamber induced induced by NGR-TNF in endothelial cells by NGR-TNF (Fig. 6A, bottom panels). These data support the TNF is known to induce the phosphorylation of p38-MAPK, hypothesis that these polypeptides can affect drug delivery to a signaling mechanism that can alter VE-cadherin expression tumor cells in vivo by inhibiting the alteration of the endothe- and endothelial permeability (39). To assess whether NGR- lial barrier function induced by NGR-TNF. TNF and CgA regulate this signaling pathway in a positive and negative manner, respectively, we analysed the effect of these CgA inhibits the NGR-TNFinduced disassembly proteins on p38-MAPK phosphorylation in HUVEC cells at of VE-cadherin–dependent adherence junctions different time points. Western blot analysis of total and The molecular mechanism underlying the inhibitory effect phosphorylated p38-MAPK showed that NGR-TNF can induce exerted by CgA on the NGR-TNFinduced permeability was the phosphorylation of this enzyme in a time-dependent then investigated. First, we analyzed whether NGR-TNF could manner and that CgA can inhibit this effect in a dose- induce the disassembly of VE-cadherin, a protein critical for dependent manner (Fig. 6C). CgA alone did not affect adherence junctions, in confluent HUVEC cells. Immunofluor- p38-MAPK phosphorylation (data not shown). escence staining of VE-cadherin showed a homogeneous dis- tribution of this molecule at cell–cell contacts (Fig. 6B). Discussion Incubation with NGR-TNF induced the disassembly of VE-cadherin and caused the formation of intercellular gaps. This effect The results of this study show that pretreatment of mice wasinhibitedbyCgA(Fig.6B).TheseresultssuggestthatCgAcan bearing RMA lymphomas or B16-F1 melanomas with doses of

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 OF7

Dondossola et al.

CgA that enhance its circulating levels 4 to 5 times above the vascular compartment of the tumor microenvironment. Thus, normal values completely inhibits the synergism of NGR-TNF NGR-TNF enhances the drug extravasation steps (steps 2 and with doxorubicin and melphalan. This inhibitory activity is 3), whereas CgA can inhibit this effect. Accordingly, the results unlikely related to inhibition of the cytotoxic activity of this of endothelial permeability assays showed that CgA can drug combinations, as no effect was observed in cytotoxicity inhibit the NGR-TNFinduced flux of FITC-dextran through assays with cultured endothelial or tumor cells. More likely, endothelial cell monolayers. Given that FITC-dextran is a good the inhibitory activity of CgA was related to a reduction of the tracer for the paracellular transport of soluble compounds delivery of chemotherapeutic drugs to tumor cells. This through the endothelial barrier, these findings suggest that hypothesis is supported by the results of FACS analysis of NGR-TNF can reduce the barrier function of endothelial cells tumor cells recovered from tumor-bearing mice treated with whereas CgA inhibits this effect (step 2). Notably, CgA could doxorubicin (which is a fluorescent compound) showing that also inhibit the NGR-TNFenhanced flux of doxorubicin the penetration of drugs into neoplastic cells induced by NGR- through endothelial cell monolayers. Thus, whereas the altera- TNF in vivo can be almost completely inhibited by CgA. tion of the endothelial barrier function by NGR-TNF might How does NGR-TNF and CgA increase and decrease, respec- account for the improved delivery of chemotherapeutic drugs tively, the penetration of drugs into tumor cells? Drug delivery to tumor cells, the protective activity exerted by CgA on the to neoplastic cells in solid tumors requires, in principle, at endothelial barrier integrity might account for the inhibition least 4 steps: the drug must (a) enter the tumor blood vessels of drug penetration and therapeutic activity observed in vivo. (step 1), (b) cross the endothelial barrier (step 2), (c) migrate Notably, all the in vitro and in vivo effects observed with CgA through the tumor interstitium (step 3), and (d) cross the were observed also with its N-terminal fragment (VS-1). This membrane of the target cells (step 4; see Fig. 7 for a schematic observation suggests that the biologically active site of CgA is representation). The results obtained with cultured cells show located in its N-terminal domain. A schematic representation that neither NGR-TNF nor CgA can affect the penetration of of these concepts is shown in Figure 7. doxorubicin into RMA cells, arguing against a crucial role of HowdoNGR-TNF,CgA,andVS-1affectendothelialcell these molecules on step 4. The effect on other steps are likely permeability? The results of in vitro studies show that NGR- more important. Interestingly, the results of in vivo assays TNF,similartoTNF,canalterthebarrierfunctionof done with RMA tumor–bearing mice injected with Patent Blue endothelial cell monolayers by causing a redistribution of VF (a hydrophilic dye that does not cross the cell membrane) VE-cadherin, a component of adherence junctions critical show that NGR-TNF can increase the accumulation of dye in for the maintenance of endothelial barrier integrity (40) and the tumor microenvironment and that CgA can inhibit this that both CgA and VS-1 can inhibit this effect. The changes effect. Because tumors were perfused before excision (to of VE-cadherin expression induced by TNF involve remove the dye present in the intravascular compartment), p38-MAPK phosphorylation (39). We observed that also these findings suggest that the dye was located in the extra- NGR-TNF, similar to TNF, can induce p38-MAPK phosphorylation in endothelial cells. Again, CgA could inhibit this effect. Notably, CgA did not inhibit the cytotoxic activity of NGR-TNF in assays done in the presence of actinomycin D, a transcription inhibitor (data not shown). This observation suggests that CgA does not inhibit the interaction of NGR-TNF with TNF membrane receptors. More likely, CgA inhibits the NGR-TNF–induced phosphorylation of p38-MAPK by interfering with downstream signaling path- ways triggered by TNF receptor activation. Are the doses of CgA used in our experimental models physiologically relevant? Administration of 1 mg of CgA, i.p., generated circulating levels of about 3 to 5 nmol/L at the time of NGR-TNF administration in our murine models. This dose was selected to achieve circulating levels similar to those observed in subpopulation of patients with nonneuroendocrine tumors (11, 12, 31, 37). For example, although serum CgA in the blood of normal subjects is about 1 nmol/L (21), increased levels of CgA, up to 4 to 8 nmol/L have been detected in the sera of a subgroup of patients with NSCLC, Figure 7. Hypothetical mechanism of the effects exerted by NGR-TNF despite the lack of neuroendocrine differentiation, with and CgA on the penetration of drugs in tumors. To reach cancer cells important prognostic implication (37). Considering that the drugs must (1) enter tumor vasculature, (2) cross the vessel wall, combination of NGR-TNF and chemotherapy is currently (3) diffuse through the interstitium, and (4) penetrate into the cancer cells. under investigation in patients with NSCLC, hepatocellular NGR-TNF exerts synergistic effects with chemotherapy by altering the – endothelial barrier function and promoting drug penetration in tumors carcinoma, or other nonneuroendocrine tumors (9 14), the (step 2). CgA and VS-1 inhibit NGR-TNF/chemotherapy synergism by results showing inhibitory activity of CgA in animal models of protecting the endothelial barrier integrity. nonneuroendocrine tumors might be clinically relevant.

OF8 Cancer Res; 71(17) September 1, 2011 Cancer Research

Chromogranin A and NGR-TNF/Chemotherapy

A growing body of evidence suggest that administration of circulating CgA to predict the therapeutic response. Interest- proton pump inhibitors, which are drugs commonly used in ingly, in 3 studies carried out on SCLC, NSCLC, and prostate the treatment of acid peptic disorders, can enhance 2 to 3 cancer, the response rates to chemotherapy inversely corre- times, and in certain patients even up to 10 times, the lated with baseline CgA (37, 47, 48). However, another study on circulating levels of CgA, depending on the time of adminis- prostate cancer, a tumor that may undergo neuroendocrine tration (32, 33). These drugs, by decreasing gastric acidity, differentiation, showed higher response rates in patients induce the release of gastrin that, in turn, can induce enter- with high baseline CgA (49). Moreover, in a recent study, a ochromaffin cell-like cell hyperplasia and CgA secretion (41, large proportion of patients with metastatic pancreatic endo- 42). Accordingly, we found that daily treatment of mice with crine carcinomas experienced a major biochemical response omeprazole can increase 2-fold the circulating levels of CgA despite an elevated serum CgA before therapy (50). These and that 4 day discontinuation of therapy is necessary to conflicting results further underline the need of new studies to return to base line. Interestingly, we observed that admini- elucidate the role of CgA in the response to chemotherapy of stration of omeprazole to mice can inhibit the penetration of neuroendocrine tumors or of tumors that may undergo neu- doxorubicin in tumors and that this effect can be inhibited by roendocrine differentiation. administration of a CgA-neutralizing antibody. Thus, a 2-fold In conclusion, the results of this study suggest that CgA can enhancement of CgA levels was sufficient to inhibit the inhibit the NGR-TNF/chemotherapy synergism in murine penetration of doxorubicin induced by NGR-TNF in tumors. models of nonneuroendocrine tumors by inhibiting Considering the wide use of proton pump inhibitors in NGR-TNF–induced changes of vascular permeability and drug patients, also these findings may have important clinical penetration in neoplastic tissues. These results could stimu- implications. late further work aimed at assessing whether circulating CgA This study has a limitation that must be highlighted: and VS-1 levels can predict the response to NGR-TNF/ although our models are suitable to investigate the role of chemotherapy in patients. Furthermore, the finding that circulating CgA in nonneuroendocrine tumors, they are omeprazole-induced CgA can inhibit the penetration of dox- not necessarily suitable to address the role of tumor orubicin in tumor tissues may suggest that discontinuation of derived/locally produced CgA that may occur in patients with administration of proton pump inhibitors to patients and/or neuroendocrine tumors. Indeed, in the latter case the tumor its replacement with H2 receptor antagonists (a class of drugs vasculature is exposed to much higher levels of CgA (possibly that do not increase CgA levels; ref. 33) might increase the >1 mmol/L) compared with those of nonneuroendocrine response rate to this therapy in patients with documented tumors, owing to the fact that the source of CgA is the tumor elevation of circulating CgA or VS-1. itself. Because the dose–response curves of CgA and VS-1 in various biological assays are "bell shaped" with loss of activity Disclosure of Potential Conflicts of Interest at micromolar concentrations (36, 43, 44), higher levels of CgA in the tumor microenvironment could be, paradoxically, less Angelo Corti is the inventor of a patent on NGR-TNF. active than low levels of circulating CgA. Furthermore, different proteolytic processing may occur to CgA released by Grant Support neuroendocrine tumor cells and by normal cells (45). Thus, This work was supported by Associazione Italiana per la Ricerca sul Cancro, further studies with suitable neuroendocrine tumor models Alleanza Contro il Cancro (ACC), and FIRB (Italy). are necessary to address this point. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in Although many studies showed that CgA is a good marker accordance with 18 U.S.C. Section 1734 solely to indicate this fact. for diagnosis and prognosis of neuroendocrine tumors, even for monitoring their responses to chemotherapy (29, 30, 46), Received April 13, 2011; revised July 7, 2011; accepted July 7, 2011; only few works have investigated the capability of baseline published OnlineFirst July 28, 2011.

References 1. Minchinton AI, Tannock IF. Drug penetration in solid tumours. Nat Rev 6. Arap W, Pasqualini R, Ruoslahti E. Cancer treatment by targeted drug Cancer 2006;6:583–92. delivery to tumor vasculature in a mouse model. Science 1998;279: 2. Jain RK. Barriers to drug delivery in solid tumors. Sci Am 1994;271: 377–80. 58–65. 7. Pasqualini R, Koivunen E, Kain R, Lahdenranta J, Sakamoto M, Stryhn 3. Curnis F, Sacchi A, Borgna L, Magni F, Gasparri A, Corti A. Enhance- A, et al. Aminopeptidase N is a receptor for tumor-homing peptides ment of tumor necrosis factor alpha antitumor immunotherapeutic and a target for inhibiting angiogenesis. Cancer Res 2000;60:722–7. properties by targeted delivery to aminopeptidase N (CD13). Nat 8. Sacchi A, Gasparri A, Gallo-Stampino C, Toma S, Curnis F, Corti A. Biotechnol 2000;18:1185–90. Synergistic antitumor activity of cisplatin, paclitaxel, and gemcitabine 4. Curnis F, Sacchi A, Corti A. Improving chemotherapeutic drug pene- with tumor vasculature-targeted tumor necrosis factor-alpha. Clin tration in tumors by vascular targeting and barrier alteration. J Clin Inv Cancer Res 2006;12:175–82. 2002;110:475–82. 9. Gregorc V, Santoro A, Bennicelli E, Punt CJ, Citterio G, Timmer- 5. Curnis F, Arrigoni G, Sacchi A, Fischetti L, Arap W, Pasqualini R, et al. Bonte JN, et al. Phase Ib study of NGR-hTNF, a selective vascular Differential binding of drugs containing the NGR motif to CD13 iso- targeting agent, administered at low doses in combination with forms in tumor vessels, epithelia, and myeloid cells. Cancer Res doxorubicin to patients with advanced solid tumours. Br J Cancer 2002;62:867–74. 2009;101:219–24.

www.aacrjournals.org Cancer Res; 71(17) September 1, 2011 OF9

Dondossola et al.

10. Gregorc V, Zucali PA, Santoro A, Ceresoli GL, Citterio G, De Pas TM, nostic utility and limits. World J Gastroenterol 2005;11:1987– et al. Phase II study of asparagine-glycine-arginine-human tumor 90. necrosis factor alpha, a selective vascular targeting agent, in pre- 32. Giusti M, Sidoti M, Augeri C, Rabitti C, Minuto F. Effect of short-term viously treated patients with malignant pleural mesothelioma. J Clin treatment with low dosages of the proton-pump inhibitor omeprazole on Oncol 2010;28:2604–11. serum chromogranin A levels in man. Eur J Endocrinol 2004;150:299– 11. Gregorc V, Citterio G, Vitali G, Spreafico A, Scifo P, Borri A, et al. 303. Defining the optimal biological dose of NGR-hTNF, a selective vas- 33. Sanduleanu S, Stridsberg M, Jonkers D, Hameeteman W, Biemond I, cular targeting agent, in advanced solid tumours. Eur J Cancer Lundqvist G, et al. Serum gastrin and chromogranin A during medium- 2010;46:198–206. and long-term acid suppressive therapy: a case-control study. Ali- 12. Santoro A, Pressiani T, Citterio G, Rossoni G, Donadoni G, Pozzi F, ment Pharmacol Ther 1999;13:145–53. et al. Activity and safety of NGR-hTNF, a selective vascular-targeting 34. Corti A, Sanchez LP, Gasparri A, Flavio C, Longhi R, Brandazza A, agent, in previously treated patients with advanced hepatocellular et al. Production and structure characterization of recombinant chro- carcinoma. Br J Cancer 2010;103:837–44. mogranin A N-terminal fragments (vasostatins): evidence of dimer- 13. Mammoliti S, Andretta V, Bennicelli E, Caprioni F, Comandini D, monomer equilibria. Eur J Biochem 1997;248:692–9. Fornarini G, et al. Two doses of NGR-hTNF in combination with 35. Corti A, Longhi R, Gasparri A, Chen F, Pelagi M, Siccardi AG. Anti- capecitabine plus oxaliplatin in colorectal cancer patients failing genic regions of human chromogranin A and their topographic rela- standard therapies. Ann Oncol 2011;22:973–8. tionships with structural/functional domains. Eur J Biochem 1996; 14. CortiA,PastorinoF,CurnisF,ArapW,PonzoniM,PasqualiniR.Targeted 235:275–80. drug delivery and penetration into solid tumors. Med Res Rev 2011 36. Ratti S, Curnis F, Longhi R, Colombo B, Gasparri A, Magni F, et al. Feb 1. doi: 10.1002/med.20238. [Epub ahead of print]. Structure-activity relationships of chromogranin A in cell adhesion. 15. Ferrero E, Scabini S, Magni E, Foglieni C, Belloni D, Colombo B, et al. Identification and characterization of an adhesion site for fibroblasts Chromogranin A protects vessels against tumor necrosis factor alpha- and smooth muscle cells. J Biol Chem 2000;275:29257–63. induced vascular leakage. FASEB J 2004;18:554–6. 37. Gregorc V, Spreafico A, Floriani I, Colombo B, Ludovini V, Pistola L, 16. Helle KB, Corti A, Metz-Boutigue MH, Tota B. The endocrine role for et al. Prognostic value of circulating chromogranin A and soluble chromogranin A: a prohormone for peptides with regulatory proper- tumor necrosis factor receptors in advanced nonsmall cell lung ties. Cell Mol Life Sci 2007;64:2863–86. cancer. Cancer 2007;110:845–53. 17. Dondossola E, Gasparri A, Bachi A, Longhi R, Metz-Boutigue MH, 38. Brett J, Gerlach H, Nawroth P, Steinberg S, Godman G, Stern D. Tota B, et al. Role of vasostatin-1 C-terminal region in fibroblast cell Tumor necrosis factor/cachectin increases permeability of endothelial adhesion. Cell Mol Life Sci 2010;67:2107–18. cell monolayers by a mechanism involving regulatory G proteins. J 18. Blois A, Srebro B, Mandala M, Corti A, Helle KB, Serck-Hanssen G. Exp Med 1989;169:1977–91. The chromogranin A peptide vasostatin-I inhibits gap formation and 39. Nwariaku FE, Chang J, Zhu X, Liu Z, Duffy SL, Halaihel NH, et al. The signal transduction mediated by inflammatory agents in cultured role of p38 map kinase in tumor necrosis factor-induced redistribution bovine pulmonary and coronary arterial endothelial cells. Reg Pept of vascular endothelial cadherin and increased endothelial perme- 2006;135:78–84. ability. Shock 2002;18:82–5. 19. Belloni D, Scabini S, Foglieni C, Veschini L, Giazzon A, Colombo B, et al. 40. Hordijk PL, Anthony E, Mul FP, Rientsma R, Oomen LC, Roos D. The vasostatin-I fragment of chromogranin A inhibits VEGF-induced Vascular-endothelial-cadherin modulates endothelial monolayer per- endothelial cell proliferation and migration. FASEB J 2007;21:3052–62. meability. J Cell Sci 1999;112:1915–23. 20. O’Connor DT, Bernstein KN. Radioimmunoassay of chromogranin A 41. Ligumsky M, Lysy J, Siguencia G, Friedlander Y. Effect of long-term, in plasma as a measure of exocytotic sympathoadrenal activity in continuous versus alternate-day omeprazole therapy on serum gas- normal subjects and patients with pheochromocytoma. New Engl J trin in patients treated for reflux esophagitis. J Clin Gastroenterol Med 1984;311:764–70. 2001;33:32–5. 21. Ceconi C, Ferrari R, Bachetti T, Opasich C, Volterrani M, Colombo B, 42. Waldum HL, Brenna E. Personal review: is profound acid inhibition et al. Chromogranin A in heart failure; a novel neurohumoral factor and safe? Aliment Pharmacol Ther 2000;14:15–22. a predictor for mortality. Eur Heart J 2002;23:967–74. 43. Vaingankar SM, Li Y, Biswas N, Gayen J, Choksi S, Rao F, et al. Effects 22. Pieroni M, Corti A, Tota B, Curnis F, Angelone T, Colombo B, et al. of chromogranin A deficiency and excess in vivo: biphasic blood Myocardial production of chromogranin A in human heart: a new pressure and catecholamine responses. J Hypertens 2010; 28:817–25. regulatory peptide of cardiac function. Eur Heart J 2007;28:1117–27. 44. Corti A, Mannarino C, Mazza R, Angelone T, Longhi R, Tota B. 23. Taupenot L, Harper KL, O’Connor DT. The chromogranin-secreto- Chromogranin A N-terminal fragments vasostatin-1 and the synthetic granin family. New Engl J Med 2003;348:1134–49. CGA 7-57 peptide act as cardiostatins on the isolated working frog 24. Corti A, Ferrari R, Ceconi C. Chromogranin A and tumor necrosis heart. Gen Comp Endocrinol 2004;136:217–24. factor alpha in heart failure. Adv Exp Med Biol 2000;482:351–9. 45. Portel-Gomes GM, Grimelius L, Johansson H, Wilander E, Stridsberg 25. Zhang D, Lavaux T, Sapin R, Lavigne T, Castelain V, Aunis D, et al. M. Chromogranin A in human neuroendocrine tumors: an immuno- Serum concentration of chromogranin A at admission: an early bio- histochemical study with region-specific antibodies. Am J Surg Pathol marker of severity in critically ill patients. Ann Med 2009;41:38–44. 2001;25:1261–7. 26. O’Connor DT, Mahata SK, Taupenot L, Mahata M, Livsey Taylor CV, 46. Modlin IM, Gustafsson BI, Moss SF, Pavel M, Tsolakis AV, Kidd M. Kailasam MT, et al. Adv Exp Med Biol 2000;482:377–88. Chromogranin A–biological function and clinical utility in neuro endo- 27. Syversen U, Ramstad H, Gamme K, Qvigstad G, Falkmer S, Waldum crine tumor disease. Ann Surg Oncol 2010;17:2427–43. HL. Clinical significance of elevated serum chromogranin A levels. 47. Ross RW, Beer TM, Jacobus S, Bubley GJ, Taplin ME, Ryan CW, et al. Scand J Gastroenterol 2004;39:969–73. A phase 2 study of carboplatin plus docetaxel in men with metastatic 28. BorchK,StridsbergM,BurmanP,RehfeldJF.BasalchromograninAand hormone-refractory prostate cancer who are refractory to docetaxel. gastrin concentrations in circulation correlate to endocrine cell prolifera- Cancer 2008;112:521–6. tion in type-A gastritis. Scand J Gastroenterol 1997;32:198–202. 48. Johnson PW, Joel SP, Love S, Butcher M, Pandian MR, Squires L, et al. 29. Corti A. Chromogranin A and the tumor microenvironment. Cell Mol Tumour markers for prediction of survival and monitoring of remission Neurobiol 2010;30:1163–70. in small cell lung cancer. Br J Cancer 1993;67:760–6. 30. Lawrence B, Gustafsson BI, Kidd M, Pavel M, Svejda B, Modlin IM. 49. Cabrespine A, Guy L, Gachon F, Cure H, Chollet P, Bay JO. Circulating The clinical relevance of chromogranin A as a biomarker for gastro- chromogranin A and hormone refractory prostate cancer chemother- enteropancreatic neuroendocrine tumors. Endocrinol Metab Clin apy. J Urol 2006;175:1347–52. North Am 2011;40:111–34, viii. 50. Strosberg JR, Fine RL, Choi J, Nasir A, Coppola D, Chen DT, et al. First- 31. Spadaro A, Ajello A, Morace C, Zirilli A, D’Arrigo G, Luigiano C, line chemotherapy with capecitabine and temozolomide in patients with et al. Serum chromogranin-A in hepatocellular carcinoma: diag- metastatic pancreatic endocrinecarcinomas.Cancer 2011;117:268–75.

OF10 Cancer Res; 71(17) September 1, 2011 Cancer Research

Chromogranin A Restricts Drug Penetration and Limits the Ability of NGR-TNF to Enhance Chemotherapeutic Efficacy

Eleonora Dondossola, Anna Maria Gasparri, Barbara Colombo, et al.

Cancer Res Published OnlineFirst July 28, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-11-1273

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2011/08/19/0008-5472.CAN-11-1273. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.