Oncolytic adenovirus modified with somatostatin ... - Semantic Scholar

Gene Therapy (2011) 18, 1052–1062 & 2011 Macmillan Publishers Limited All rights reserved 0969-7128/11 www.nature.com/gt

ORIGINAL ARTICLE

Oncolytic adenovirus modified with somatostatin motifs for selective infection of neuroendocrine tumor cells ¨ berg5, V Giandomenico5 J Leja1, D Yu1, B Nilsson1, L Gedda2, A Zieba1, T Hakkarainen3, G A˚kerstro¨m4, K O 1 and M Essand We have previously described the oncolytic adenovirus, Ad(CgA-E1A-miR122), herein denoted Ad5(CgA-E1A-miR122) that selectively replicates in and kills neuroendocrine cells, including freshly isolated midgut carcinoid cells from liver metastases. Ad5(CgA-E1A-miR122) is based on human adenovirus serotype 5 (Ad5) and infects target cells by binding to the coxsackieadenovirus receptor (CAR) and integrins on the cell surface. Some neuroendocrine tumor (NET) and neuroblastoma cells express low levels of CAR and are therefore poorly transduced by Ad5. However, they often express high levels of somatostatin receptors (SSTRs). Therefore, we introduced cyclic peptides, which contain four amino acids (FWKT) and mimic the binding site for SSTRs in the virus fiber knob. We show that FWKT-modified Ad5 binds to SSTR2 on NET cells and transduces midgut carcinoid cells from liver metastases about 3–4 times better than non-modified Ad5. Moreover, FWKT-modified Ad5 overcomes neutralization in an ex vivo human blood loop model to greater extent than Ad5, indicating that fiber knob modification may prolong the systemic circulation time. We conclude that modification of adenovirus with the FWKT motif may be beneficial for NET therapy. Gene Therapy (2011) 18, 1052–1062; doi:10.1038/gt.2011.54; published online 14 April 2011 Keywords: oncolytic adenovirus; somatostatin; neuroendocrine tumors; carcinoid; neuroblastoma; proximity ligation

INTRODUCTION Neuroendocrine tumors (NETs) are rare cancers with approximately 2/3 of lesions occurring in the gastrointestinal tract and 1/3 in the lower respiratory tract.1 They are by tradition known as carcinoids and divided based on their embryonic origin into foregut (thymus, lung and gastroduodenal mucosa), midgut (jejunum, ileum, appendix and proximal colon) and hindgut (descending colon and rectum). The majority of NET cells express somatostatin receptors (SSTRs), which are G-protein-coupled plasma membrane receptors. There are five SSTRs, denoted as SSTR1 5, which exert various functions. Activation of SSTRs on NET cells results in growth inhibition.2,3 Somatostatin, the natural ligand for SSTRs exists in two forms: one of 14 amino acids (SST-14) and one of 28 amino acids (SST-28). As a result of their short plasma half-life they are not useful clinically and stable synthetic analogues have been developed for clinical applications. The synthetic analogues are generally designed to retain the natural Phe-Trp-Lys-Thr (FWKT) loop of somatostatin, which contains the binding site for SSTRs.4 Octreotide, the first synthetic somatostatin analog available for clinical use, binds with highest affinity to SSTR2, intermediate affinity to SSTR3 and SSTR5, and minimal or no affinity to SSTR1 and SSTR4. Radiolabeled somatostatin analogues have become important diagnostic tool in patients with suspected or recurrent neuroendocrine malignancies.2,5 They are also use as a therapeutic agent against NETs and neuroblastomas.6,7 Oncolytic viruses are emerging therapeutic agents for cancer.8 Results from clinical trials show that they are safe but that the delivery and transduction efficiency of target cells need to be improved in

order to fulfill therapeutic goals. We have previously described the oncolytic adenovirus, Ad(CgA-E1A-miR122) where E1A gene expression is driven by the human chromogranin A (CgA) promoter and further controlled by six tandem repeats of miR122 target sequences in the 3¢UTR of E1A.9,10 This virus selectively replicates in and kills neuroendocrine cells, including freshly isolated midgut carcinoid cells from liver metastases. However, it does not kill normal hepatocytes. Herein the virus is denoted Ad5(CgA-E1A-miR122) to clarify that it is based on human adenovirus serotype 5 (Ad5) and infects cells according to the natural Ad5 tropism. The fiber knob of Ad5 binds to the coxsackie-adenovirus receptor (CAR) on target cells.11 Thereafter, virus pentons at the fiber base interacts with avb3 and avb5 integrins, leading to internalization of the virus, which is followed by partial degradation of the viral particle and transfer of the viral genome into the nucleus of the target cell. Several studies demonstrate loss of CAR expression on tumor cells, and thereby intrinsic resistance of tumor cells to treatment with oncolytic Ad5 viruses.12,13 The adenovirus tropism can be modified through genetic modification of the capsid to target receptors that are highly expressed on tumor cells and thereby avoid CAR deficiency.14,15 In this study, we investigate Ad5 vectors with fibers modifications for their ability to transduce NET cells. We evaluate an Ad5 with fiber from human adenovirus serotype 35 (Ad5f35), which binds to the CD46 receptor16 and an Ad5 with fiber knob from human adenovirus serotype 3 (Ad5fk3), which binds to desmoglein 2 (DSG2).17 In order to selectively target SSTRs and thereby infect NET cells, we incorporate the binding motif of octreotide (FWKT) into the HI loop of the Ad5

1Department of Immunology, Genetics and Pathology, Science for Life Laboratory, Uppsala University, Uppsala, Sweden; 2Department of Radiology, Oncology and Radiation Sciences, Uppsala University, Uppsala, Sweden; 3Cancer Gene Therapy Group, Finnish Institute for Molecular Medicine, University of Helsinki, Helsinki, Finland; 4Department of Surgical Sciences, Uppsala University, Uppsala, Sweden and 5Department of Medical Sciences, Uppsala University, Uppsala, Sweden Correspondence: Professor M Essand, Department of Immunology, Genetics and Pathology, Science Life Laboratory, Uppsala University, SE-75185 Uppsala, Sweden. E-mail: [email protected] Received 21 November 2010; revised 2 March 2011; accepted 3 March 2011; published online 14 April 2011

Oncolytic adenovirus with somatostatin motifs J Leja et al 1053

fiber knob. The FWKT-modified adenovirus shows increased transduction efficiency as well as strong wild-type like replication and killing in freshly isolated carcinoid cells and it is partly detargeted from neutralizing anti-Ad5 antibodies (NAbs) in human blood. RESULTS High expression of SSTRs in primary midgut carcinoid cells Carcinoid cells obtained from patient tumor material as well as NET and neuroblastoma cell lines were evaluated by flow cytometry for expression of cell surface receptors involved in Ad5 attachment (CAR), Ad5f35 attachment (CD46) and Ad5fk3 attachment (DSG2) as well as adenovirus internalization (avb3 and avb5 integrins), (Table 1). The data indicate that it may be advantageous to use fibers from Ad3 and Ad35 for transduction of NET cells. However, CD46 is a complement regulatory protein expressed on all nucleated human cells, which may increase transduction of unwanted cells. DSG2 is a component of the cell–cell adhesion structure, which may cross-talk with the CAR molecule within junctional complexes.18 Similarly to CAR, DSG2 expression has been found in normal epithelial tissues and epithelial cancers. Considering the fact that CAR is often downregulated in tumors, more studies are needed to investigate DSG2 expression pattern in human cancers. SSTR1, SSTR2, SSTR3 and SSTR5 are often highly expressed on NET cells so next we investigated the RNA expression levels on our samples and cell lines by quantitative real-time PCR (Figure 1a). Freshly isolated midgut carcinoid cells (M3livmet, M4livmet and M4mesmet) express high levels of SSTR2 with approximately 5-fold higher expression than foregut carcinoid cells (F1livmet and F2livmet) and 10-fold higher than the NET cell lines BON and CNDT2.5 and the neuroblastoma cell lines SH-SY-5Y and SK-N-BE(2). The normal fibroblast cell line 1064SK and the melanoma cell line mel526 are negative for SSTR2 and included as negative controls. SSTR1 expression is approximately 10-fold lower than SSTR2 in the carcinoid samples while it is not expressed in the cell lines. SSTR3 and SSTR5 are expressed at a

lower degree, especially in the foregut carcinoid samples. Higher expression of SSTR in midgut carcinoid cells than in foregut carcinoid cells and the cell lines was confirmed at the protein levels by the [125I]Tyr11-somatostatin binding assay (Figure 1b). Specific binding is seen for all cells, because the binding is competed by high concentration of non-radiolabeled somatostatin. However, expression of SSTR seems rather low in all cell lines as well as in the foregut carcinoma cell

Table 1 Expression of cell surface molecules involved in adenovirus entry CAR

Integrin avb5 Integrin avb3

CD46

DSG2

Pos (%) MFI Pos (%) MFI Pos (%) MFI Pos (%) MFI Pos (%) MFI Foregut carcinoid

20

10

1

0.4

21

20

75

66

5

7

F1livmet Foregut carcinoid

51

24

3

2

29

24

47

20

32

10

F2livmet Midgut carcinoid

2

14

1

1

2

2

91

37

1

4

M3livmet Midgut carcinoid

12

9

1

1

2

1

68

18

2

3

M4livmet Midgut carcinoid

8

6

2

1

2

1

66

15

1

1

M4mesmet BON

52

87

8 31

1

1

92

129

59

121

CNDT2.5 SK-N-BE(2)

1 15

6 15

4 14 61 83

45 19

11 13

95 26

34 22

83 4

101 8

SH-SY-5Y mel526

60 17

103 5

4 17 11 5

20 89

19 57

1 7

6 2

4 4

7 16

1064SK Hepatocytes

9 63

8 79 26 24 100 15

83 12

26 6

91 63

29 29

1 ND

1 ND

Abbreviations: CAR, coxsackie adenovirus receptor; DSG2, desmoglein 2; livmet, liver metastasis; mesmet, mesentery metastasis; MFI, mean fluorescence intensity; ND, not done; Pos, positive cells.

Figure 1 Expression of SSTR mRNA in NET cells. (a) Total RNA was isolated from primary carcinoid cells (midgut and foregut), cancer cell lines and normal hepatocytes. Thereafter, complementary DNA (cDNA) was synthesized and relative expression was measured by quantitative real-time PCR. Data were evaluated using the 2 DDCT method and normalized to b-actin expression from each sample. Average value of two independent experiments (each sample run in triplicates) is presented with s.d. (b) Primary foregut (F2livmet) and midgut (M3 livmet, M4 livmet) carcinoid cells and NET cell lines were assayed for [125I]Tyr11-somatostatin binding (0.5 and 1 nM in case M4 livmet). A concentration of 1.2 mM somatostatin was used to specifically compete with the binding of [125I]Tyr11somatostatin. Average binding (mean±s.d.) of [125I]Tyr11-somatostatin from duplicate samples is given in c.p.m./105 cells for somatostatin competed (gray bars) and non-competed cells (black bars). Gene Therapy


sample (F2livmet) with background level (blocked) at almost half of the signal (non-blocked). The exception being the midgut carcinoid samples (M3livmet and M4livmet) that shows approximately fivefold higher binding capacity of [125I]-somatostatin than the foregut carcinoid sample and any of the tested cell lines. These data indicate that SSTRs may be valid targets for midgut carcinoids. A recombinant adenoviral vector with FWKT motif in the fiber knob interacts with SSTR2 and specifically transduces SSTR-positive NET and neuroblastoma cells All adenoviral vectors used in this study, and their characteristics, are presented in Table 2. The vectors have surface modifications, which influence their transduction capacity and a titration method based on plaque-forming or fluorescence-forming unit assays would depend on the receptor expression on the cell line used for titration. Instead we chose a new titration method based on quantitative real-time PCR, similar to the method developed by Wold et al.19 We designed primers for the adenoviral E4 orf1 gene to detect intact encapsidated viral genome (evg) of the purified vectors in order to determine their titers. That way a fair comparison can be made between the various vectors. We constructed a recombinant adenoviral vector with cyclic peptides with FWKT motifs in the fiber knob, Ad5fkFWKT(green fluorescent protein (GFP)), and investigated its interaction with SSTR2 using Duolink in situ proximity ligation assay (PLA) technology (Figure 2a). The SSTR2-negative cell line COS7 was transfected with a plasmid encoding SSTR2 with three HA-tags at the N-terminus (3HA-SSTR2). The PLA assay was then applied with an anti-HA antibody and an anti-Ad hexon antibody and analyzed with the BlobFinder software to quantify the signals. PLA signals representing close proximity between adenovirus and surface receptor were counted in 150 cells. Representative images are shown in Figure 2b. Incubation with Ad5fkFWKT(GFP) resulted in 64% of the cells with at least 20 PLA signals per cell and 32% above 40 PLA signals per cell, while Ad5(GFP) gave only 20 and 3% of cells with at least 20 or 40 PLA signals per cell, respectively (Figure 2c). Incubation of COS7 cells, transfected with the control plasmid pEGFP-C2, resulted in PLA signals at background level (as obtained with no virus) both for Ad5fkFWKT(GFP) and Ad5(GFP). Next, we performed PLA on BON cells, which are positive both for SSTR2 and CAR. Primary antibodies were anti-SSTR2 and anti-hexon. Representative images are shown in Figure 2d. Incubation with Ad5fkFWKT(GFP) resulted in an increase in signal over time with 85 and 96% of cells having at least 20 PLA signals per cell after 30 min and 2 h, respectively, while incubation with Ad5(GFP) resulted in PLA signals at background level (as obtained

with no virus), (Figure 2e). We next used primary antibodies against CAR and hexon with representative images shown in Figure 2f. Both Ad5(GFP) and Ad5fkFWKT(GFP) bound to the CAR receptor as demonstrated by 84 and 51% of cells with more than 20 PLA signals per cell, respectively (Figure 2g). The negative control Ad5fk3(GFP) did not yield PLA signals above background level (as obtained with no virus), (Figures 2f and g). Our data show that incorporation of FWKT in the HI loop of the fiber knob of Ad5 yields binding to SSTR2 and partly blocks binding to CAR. In addition, by using the in situ PLA assay, we confirmed specific binding of Ad5f35(GFP) to CD46 and Ad5fk3(GFP) to DSG2 (Supplementary Figures S1a–d). To further examine the role of SSTR2 and CAR in the transduction of FWKT-modified and unmodified adenoviral vectors, we performed blocking analysis using either SST-14 to compete with SSTR2mediated uptake or CAR receptor knockdown (CXADR small interfering RNA) to reduce CAR-mediated uptake. Pre-incubation of BON with SST-14 blocked as expected Ad5fkFWKT(GFP) but not Ad5(GFP) transduction (Supplementary Figure S2a). Transfection of BON with CAR small interfering RNA, resulted in inhibition of Ad5(GFP) transduction in a dose-dependent manner, while transduction of Ad5fkFWKT(GFP) was not affected, (Supplementary Figure S2b). Downregulation of CAR receptor protein after CAR small interfering RNA transfection was confirmed by western blot analysis (Supplementary Figure S2c). An FWKT-modified Ad5 vector is superior to a non-modified vector in transduction of primary carcinoid cells and NET cell lines Ad5fkFWKT(GFP) transduction (50 evg per cell) of midgut carcinoid cells from liver metastases (M4livmet and M3livmet) was 3–4 times more efficient than Ad5(GFP) and Ad5fk3(GFP) transduction and similar to transduction with Ad5f35(GFP). Ad5fkFWKT(GFP) transduction (50 evg per cell) of foregut carcinoid cells from liver metastases (F1livmet and F2livmet) was more efficient than Ad5(GFP), while Ad5f35(GFP) was the most efficient (Figure 3a). Importantly, Ad5fkFWKT(GFP) transduction of normal hepatocytes (500 evg per cell), both with or without blood coagulation factor X (FX), was still less efficient (Figure 3b) than Ad5fkFWKT(GFP) transduction of midgut carcinoid cells (50 evg per cell), (Figure 3a). Ad5fkFWKT(GFP) yielded higher transduction efficacy than Ad5(GFP) in CNDT2.5 and SK-N-BE(2) and similar transduction efficacy in BON and SH-SY-5Y (Figure 3c). As expected, the SSTR-negative melanoma cell line mel526 and normal fibroblasts 1064SK were transduced to a lesser degree by Ad5fkFWKT(GFP) than by Ad5(GFP), (Figure 3d). Importantly, we found that primary carcinoid cells are easier to transduce

Table 2 Characteristics of adenoviruses used in the study Virus

Transgene in E1A

Fiber

Aim

Evg ml–1

FFU ml–1

Evg/FFU

Ad5(GFP)

CMV-GFP

Fiber 5

Target CAR

1.2E+08

4.0E+07

2.9

Ad5f35(GFP) Ad5fk3(GFP)

CMV-GFP CMV-GFP

Fiber 35 Fiber knob 3

Target CD46 Target DSG2

2.9E+08 5.4E+08

6.9E+07 4.7E+07

4.2 11.4

Ad5fkFWKT(GFP)

CMV-GFP

FWKT in Ad5 HI loop

Target SSTR2

3.0E+08

1.5E+07

19.5

— E1A

Fiber 5 Fiber 5

Neg Ctrl virus Oncolytic virus

7.2E+07 2.5E+08

2.7E+07 2.4E+07

2.5 10.6

CgA-E1A-miR122 CgA-E1A-miR122

Fiber 5 FWKT in Ad5

Oncolytic virus for NETs Oncolytic virus for NETs

1.2E+08 4.3E+07

2.0E+07 5.2E+06

6.1 8.2

Ad(Mock) Ad5(wt) Ad5(CgA-E1A-miR22) Ad5fkFWKT(CgA-E1A-miR122)

HI loop Abbreviations: Ad, adenovirus; CAR, coxsackie adenovirus receptor; CgA, chromogranin A promoter; CMV, cytomegalovirus promoter; DSG2, desmoglein 2; Evg, encapsidated viral genomes; FFU, fluorescent-forming unit; fk, fiber knob; GFP, green fluorescent protein; NET, neuroendocrine tumor; SSTR2, somatostatin receptor 2; wt, wild-type.

Gene Therapy


Figure 2 FWKT-modified Ad5 cell entry is mediated by SSTR2 but also CAR. (a) Schematic illustration of the PLA. The technology is capable of detecting proteins in close proximity by using two primary antibodies, which have been raised in different species. In our case one primary antibody is directed at the virus and the other against the receptor. A pair of oligonucleotide-labeled secondary antibodies (PLA probes) is applied to the sample and a signal is generated only when the free ends of the two PLA probes are brought in proximity and are capable of hybridizing. The signal from each detected pair of PLA probes is visualized as an individual fluorescent dot and quantified. (b, d, f) Fluorescence staining of the cells: nuclei—Hoechst (blue); actin - Phalloidin (green); virus-receptor proximity (PLA signal)—Cyanine 3 (red). (b) COS7 cells where transfected with the 3HA-SSTR2 plasmid (SSTR2 with three HA tags at the N-terminus) or with the control plasmid pEGFP-C2. The next day cells were incubated with viruses at 1000 evg per cell for 2 h at 4 1C, followed by PLA applied on fixed, non-permeabilized cells. Primary antibodies were a mouse anti-hexon (adenovirus) and a rabbit anti-HA (SSTR2 receptor). (c) PLA signals showing proximity between virus and SSTR2 receptor were counted in 150 cells from each sample and divided into three groups: 0–20; 20–40; over 40 PLA signals per cell. (d) BON cells were incubated with viruses at 1000 evg per cell for 30 min or 2 h at 4 1C followed by PLA on fixed and permeabilized cells. Primary antibodies were rabbit anti-SSTR2 and mouse anti-hexon. (e) PLA signals showing proximity between virus and SSTR2 receptor were counted in 200 cells from each sample and divided into four groups: 0–20; 20–40; 40–80; over 80 PLA signals per cell. (f) BON cells were incubated with viruses at 1000 evg per cell for 2 h at 4 1C followed by PLA on fixed and permeabilized cells. Primary antibodies were rabbit anti-CAR and mouse anti-hexon. (g) PLA signals showing proximity between virus and CAR receptor were counted in 200 fixed and permeabilized cells and divided into three groups: 0–20; 20–40; over 40. ***Po0.0001; **Po0.001. Gene Therapy


than the NET cell lines. For primary cells 50 evg per cell gives approximately 50% transduction efficiency, while BON and SH-SY-5Y required 100 evg per cell and CNDT2.5 and SK-N-BE(2) 500 evg per cell to reach 50% transduction levels. The FWKT-modified adenovirus exhibits enhanced oncolytic ability We next modified the oncolytic Ad5(CgA-E1A-miR122) virus by including FWKT motifs in the HI loop of the fiber knob to create Ad5fkFWKT(CgA-E1A-miR122). We examined the lytic ability of the FWKT-modified by using an MTS ([3-(4,5-dimethylthiazol-2-yl)5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) cell viability assay on BON, SH-SY-5Y, SK-N-BE(2) and mel526. Primary carcinoid cells are not eligible for the MTS assay because of their low metabolic activity. CNDT2.5 was not included because CgA is not expressed in this cell line, as shown in Supplementary Figure S3, and therefore the CgA promoter is not supported. Cells were transduced at evg ranging from 0.001 to 100. As expected, Ad5wt was most efficient in killing all cell lines (Figure 4a). Ad5fkFWKT(CgA-E1A-miR122) was significantly more efficient than Ad5(CgA-E1A-miR122) in killing NET cells, which is most likely due to increased viral infectivity. Importantly, neither Ad5fkFWKT(CgA-E1A-miR122) nor Ad5(CgAE1A-miR122) killed the negative control cell line mel526. We also evaluated virus replication by quantitative real-time PCR analysis, measuring genomic viral DNA (E4 orf1) copies 48 h after transduction. Replication index was defined as the increase in copy number of progeny viral DNA at 48 h compared with the copy number obtained 2 h after transduction (set to 1). The wild-type Ad5 virus (Ad5wt) replicated in all cell types examined, while AdMock did not, (Figure 4b). Ad5fkFWKT(CgA-E1A-miR122) and Ad5(CgA-E1AmiR122) replication was as efficient as Ad5wt and yielded 200–700 progeny virions both in midgut and foregut carcinoid samples and in the NET and neuroblastoma cell lines (Figure 4b). Importantly, in normal hepatocytes and the negative control cell line mel526 both Ad5fkFWKT(CgA-E1A-miR122) and Ad5(CgA-E1A-miR122) were strongly attenuated while Ad5wt replicated well (Figure 4b). We next wanted to examine if the replication of Ad5fkFWKT(CgA-E1A-miR122) in NET cells resulted in production and release of progeny viruses. The presence of progeny viruses was verified by adding supernatants collected from infected primary carcinoid cells (F2 livmet and M3livmet), BON, SH-SY-5Y and SK-N-BE(2) onto 911 cells followed by staining for adenoviral hexon proteins 24 h later (Supplementary Figure S4a). Supernatant collected from the negative control cell line mel526 did not yield any progeny virus (Supplementary Figure S4a). Furthermore, an MTS cell viability assay using the supernatants on 911 cells confirmed lytic ability of newly produced viruses from NET cells but not from the melanoma cell line (Supplementary Figure S4b).

Figure 3 Transduction efficiency of FWKT-modified Ad5 in primary carcinoid cells, NET cell lines and normal hepatocytes. (a) Freshly isolated foregut and midgut carcinoid cells from mesentery metastases (mesmet) and liver metastases (livmet) were transduced with Ad5(GFP), Ad5f35(GFP), Ad5fk3(GFP) or Ad5fkFWKT(GFP) at 50 evg per cell. (b) Viral vectors were preincubated with or without FX for 30 min in 37 1C and then added to normal human hepatocytes at 500 evg per cell. Average of three experiments is presented (mean±s.d.). (c) The NET cell lines CNDT2.5 and SK-N-BE(2), were transduced at 500 evg per cell, while BON and SH-SY-5Y were transduced at 100 evg per cell. (d) As negative controls, the non-neuroendocrine cell lines mel526 and 1064SK were transduced at 500 evg per cell. Cells were collected 24 h after transduction and GFP-positive cells were counted by flow cytometry. Average of three experiments is presented (mean±s.d.). Gene Therapy

Fiber modification can protect adenovirus from neutralization in whole human blood A high proportion of the human population has been infected with Ad5 and has consequently preformed antibodies against the virus.20,21 This is thought to hamper clinical applications of Ad5-based vaccines and oncolytic Ad5 viruses.22,23 Therefore, we evaluated the recombinant GFP-expressing, fiber-modified Ad5 vectors in a newly developed human blood loop system.24 Viruses were mixed with whole blood or medium and incubated in heparin-coated polyvinyl chloride tubing at 37 1C while rotating. We used blood from six healthy donors (D); four with low ability to neutralize Ad5 (D1, D2, D5 and D6) and two with high neutralizing ability (D3 and D4), to study the degree of neutralization in whole blood by adding the virus/blood sample dilutions to 911 cells and measuring GFP expression the day after.


The titers obtained after 15 min and 6 h of incubation in blood were related to the titer of viruses incubated with medium, which was defined as 1. We found that the infectious capacity of

Ad5fkFWKT(GFP) was significantly higher than Ad5(GFP) after 15 min of incubation in human blood, indicating prolonged circulation in the blood and partial escape from clearance (Figure 5a, upper panel). Also after 6 h there was a trend, although not statistically significant that Ad5fkFWKT(GFP) retained higher infectious capacity than Ad5(GFP) (Figure 5a, lower panel). There was also a trend that Ad5fk3(GFP) retained higher infectious capacity than Ad5(GFP) both after 15 min and 6 h of incubation in human blood (Figure 5a). Next, we examined the proportions of viruses associated with blood cells and plasma. Blood samples in the blood loop were incubated with the different viruses for 15 min where after blood cell and plasma was separated by centrifugation. Viral DNA was isolated from the two fractions and detected by quantitative real-time PCR. As expected, 490% of Ad5(GFP) was found in the cell fraction in all blood samples. In blood samples, especially in the ones with low ability to neutralize Ad5 (D1, D2, D5 and D6), Ad5f35(GFP), Ad5fk3(GFP) and Ad5fkFWKT(GFP) have tendencies to retain in the plasma although the highest proportion is still found in the cell fraction, Figure 5b, indicating that fiber knob-modified Ad5 binds less to CAR on erythrocytes. DISCUSSION We have previously described the neuroendocrine-selective oncolytic adenovirus, Ad5(CgA-E1A-miR122), which effectively kills NET cells, leaving non-cancerous cells such as normal hepatocytes unharmed.9,10 The aim of this study was to combine transcriptional and transductional targeting and to introduce a cyclic peptide (CDCFWKTCFC) in the virus fiber knob to selectively target Ad5 infection to neuroendocrine cells. The peptide sequence was chosen from octreotide, a stable synthetic somatostatin analogue that binds with high affinity to SSTR2, which is abundantly expressed on carcinoid cells. We choose to insert the peptide in the HI loop of the Ad5 fiber knob at the same position where the cyclic peptide (CDCRGDCFC) has successfully been introduced for targeting Ad5 to integrins.25–27 The FWKTmodified viruses were as stable as the non-modified viruses both at 4 and 37 1C. Ad5fkFWKT(CgA-E1A-miR122) showed strong replication and efficient killing of the carcinoid cell line BON and the neuroblastoma cell lines SH-SY-5Y and SK-N-BE(2) than did Ad5(CgA-E1AmiR122). More importantly, we confirmed efficient replication of Ad5fkFWKT(CgA-E1A-miR122) in carcinoid cells from liver metastases. We were unable to use MTS assay to visualize direct cell killing of freshly isolated carcinoid cells. However, we confirmed production of infectious progeny viruses from infected carcinoid cells as a result of viral replication.

Figure 4 The FWKT-modified adenovirus exhibits enhanced oncolytic ability (a) The NET cell line BON, the neuroblastoma cell lines SK-N-BE(2) and SH-SY-5Y and melanoma cell line mel526 were transduced with oncolytic viruses at various evg per cell (0.001–1000). Cell viability was examined 6 days after transduction (for BON after 4 days). Cell viability values were expressed in relation to the viability of untransduced cells. Average enzymatic activities from three experiments each with triplicate samples are shown (mean±s.d.). (b) Foregut and midgut carcinoid cells from liver metastases (livmet) were transduced at 50 evg per cell. Normal human hepatocytes were transduced at 500 evg per cell. BON, SK-N-BE(2), SH-SY5Y and mel526 were transduced at 100 evg per cell. Viral DNA was isolated 2 h and 2 days after transduction. Quantitative real-time PCR was performed to measure E4 orf1 copies. Fold increase in DNA copies at day 2 is expressed in relation to DNA copies detected after 2 h (set as 1). *Po0.05 and **Po0.001. Gene Therapy


Figure 5 FWKT-modified Ad5 partly escapes from neutralization in whole human blood and from association with blood cells. (a) GFP-expressing adenoviral vectors were incubated in the human blood loop system and samples were taken after 15 min and 6 h of incubation when the viral ability to infect 911 cells was examined. Infectivity was measured by detection of GPF expression by flow cytometry. The titers of each Ad vector exposed to blood were related to the titer of the same Ad vector incubated with medium, which was set as 1. (b) GFP-expressing adenoviral vectors were incubated with whole human blood for 15 min. Blood samples were taken, followed by isolation of the plasma and cell fractions by centrifugation. Viral DNA was extracted and quantified by quantitative real-time PCR. The percentage of plasma and cell-bound fractions is presented. *Po0.05.

From the surface expression of SSTRs, as assessed by the [125I]Tyr11SSTR binding assay, we estimate that the expression level of SSTRs on midgut carcinoid cells from liver metastases is approximately five times higher than on the carcinoid cell lines. Therefore, one can predict that Ad5fkFWKT(CgA-E1A-miR122) would kill midgut carcinoid cells from clinical samples even more efficiently than carcinoid cell lines. Furthermore, by using the adenoviral vector Gene Therapy

Ad5fkFWKT(GFP) we could show that primary carcinoid cells required virus concentration of 50 evg per cell to achieve approximately 50% GFP-positive cells, while NET cell lines need to be transduced with 100–500 evg per cell to achieve the same effect. The biggest advantage for Ad5fkFWKT(GFP) over Ad5(GFP) is seen on cells, which are positive for SSTR2 but has no or very low CAR expression, such as CNDT2.5. Ad5fkFWKT(GFP) is approximately equally efficient in all NET and neuroblastoma cells lines, which corresponds well to their SSTR2 expression. We also found that Ad5f35 and Ad5fk3 are better than Ad5 in transducing carcinoid cells isolated from surgery samples. However, Ad5f35 and Ad5fk3 fall short of Ad5fkFWKT in all cases except for the CNDT2.5 cell line. Overall, transduction efficiencies follow what could be expected from the target cell expression of CAR, CD46, DSG2 and SSTRs. We did not have access to freshly isolated neuroblastoma cells but efficient killing of neuroblastoma cell lines indicates that the virus is also potentially efficient for neuroblastomas. Carcinoids form metastases in the liver and oncolytic adenovirus treatment could be applied as intrahepatic or intratumoral injections. Therefore, it is important that virus activity is low in normal hepatic cells to minimize unwanted toxicity. We have previously utilized both transcriptional (promoter) and post-transcriptional (microRNA target sequences) targeting to suppress virus activity in liver cells.10 An alternative approach to suppress virus activity in liver cells is to redirect virus infectivity through capsid or fiber modifications as we now did with the FWKT-modified virus. It is attractive that Ad5fkFWKT(GFP) is less efficient than Ad5(GFP) to transduce normal hepatocytes, both in the presence and absence of blood coagulation factor FX. It has been reported that Ad5 uptake in the mouse liver is mediated by blood factors and does not depend on virus interaction with CAR receptor.28 Efficient transduction of mouse liver cells after intravenous Ad5 injections raised concern about Ad-mediated hepatic toxicity.29 However, studies on Ad5 interaction in human blood indicate that binding to erythrocytes has a greater role than interaction with blood coagulation factors.24,30 Moreover, data from clinical trials strongly indicate that systemic administration of Ad5 do not affect the liver to the same extent as has been observed in mouse models.30,31 Treatment with oncolytic Ads has been well tolerated and more importantly no significant liver toxicity has been reported to date after Ad infusions including numerous injections directly into the hepatic artery.31 We demonstrated that FWKT modification in the fiber retargets adenovirus binding toward SSTR2. This was shown by PLA, which is a modern technology to study protein–protein interactions.32,33 PLA has been used in the past to detect parvovirus and an intracellular bacterium34 and our data indicate that it may become an important tool to investigate and verify binding of native and modified viruses and other pathogen to cell surface molecules. The PLA assay also revealed that insertion of FWKT motif in the HI loop allowed for binding to SSTR2 but did not completely abolish CAR binding. This was confirmed by transduction experiments together with SST competition and CAR knock-down. If complete detargeting from CAR is wanted, other strategies should be applied such as mutagenesis of CAR binding amino acids in the fiber knob11 in conjunction with introducing of the targeting peptide in the HI loop25 C-terminal of the fiber35 or hexon.36 Alternatively the fiber knob14 or fiber37 of Ad5 could be exchanged with an alternative serotype. The double binding of Ad5FWKT(GFP) to SSTR2 and CAR indicates that the adenoviral knob protein possesses a range of structural plasticity.38 This may be beneficial for adenovirus therapy of neuroendocrine malignancies, where somatostatin analog treatment often desensitizes SSTRs and induce loss of cell surface receptors through ligand-induced endocy-


tosis.39 However, longer exposure to the SST ligand and its analogs can restore SSTRs functions and surface levels.39,40 Moreover, the ratio of internalized SSTRs to membranous SSTRs varies from one patient to the other, mainly due to active circulation of SSTRs regulated by various factors.41 We examined the fiber knob-modified Ad5 vectors in the newly developed human blood loop model.24 Ad5(GFP), Ad5f35(GFP), Ad5fk3(GFP) and Ad5fkFWKT(GFP) share the same capsid (hexon, penton, fiber base and shaft). Nevertheless, we observed that Ad5fkFWKT(GFP) has the ability to overcome neutralization to a certain degree after incubation in whole human blood. However, the capacity of human blood to neutralize surface-modified Ad5 is clearly individual because some blood samples goes against the trend. These data are in agreement with previous studies, suggesting that RGD or 5/3 modification of the Ad5 fiber knob can, at least partly, avoid neutralization from preexisting Ad5 NAbs.42,43 Ad5fkFWKT(GFP), Ad5f35(GFP) and Ad5fk3(GFP) show a tendency to remain in the plasma fraction, which may be explained by decreased interaction with CAR and complement receptor 1 on human erythrocytes.30 Beneficial effects of fiber modification could not have been observed in a standard neutralization assay with heat-inactivated plasma, which strongly support the use of the ex vivo blood loop system as a more accurate preclinical model. Further studies are needed to fully understand adenovirus interaction with human blood components in order to design novel approaches to efficiently deliver therapeutic adenovirus to tumor tissues. In conclusion, adenovirus with FWKT motifs in the fiber knob selectively infects SSTR2-positive neuroendocrine cells. The FWKTmodified oncolytic adenovirus Ad5(CgA-E1A-miR122) shows enhanced viral replication and strong wild-type like killing ability of neuroendocrine cells but not of non-neuroendocrine cells. Moreover, partial escape from preexisting NAb indicates ability of prolonged circulation in the blood and possibility for improved anti-tumor activity. Our data suggest that Ad5fkFWKT(CgA-E1A-miR122) is potentially useful as a specific agent for therapy of NET liver metastases. MATERIALS AND METHODS Ethics statement, isolation and culture of primary cells Carcinoid specimens were obtained from patients who underwent liver surgery to remove midgut or foregut carcinoid metastases. A written consent statement was obtained from each patient. The experimental procedures have been approved by the Regional Ethical Committee, Uppsala University Hospital (ref. no. 2009/152). Isolation and culture conditions of carcinoid tumor specimens have been described earlier.10 Human hepatocytes (Lot 281, 285 and 286) were purchased from BD Bioscience (San Jose, CA, USA) and (batch: HEP220507) from Biopredic International (Rennes, France).

Cell lines The human endocrine pancreatic tumor cell line BON (a kind gift from Professors CM Townsend and JC Thompson, Galveston, TX, USA) was cultured in Dulbecco’s modified Eagle’s medium (DMEM) with Glutamax-I and F12 nutrient mixture (Kaighn’s modification) at a ratio of 1:1, supplemented with 10% fetal bovine serum (FBS), 1 mM sodium pyruvate and 1% penicillin–streptomycin (PEST). The cell line CNDT2.5 (a kind gift from Professor LM Ellis, MD Anderson, Houston, TX, USA) was cultured in DMEM/F-12 medium, supplemented with 10% FBS, 1% non-essential amino acids, 1 mM vitamins, 3.2 mM L-glutamine, 1 mM sodium pyruvate, 1% PEST. The human neuroblastoma cell lines SH-SY-5Y and SK-N-BE(2) (a kind gift from Dr F Hedborg, Uppsala University, Uppsala, Sweden) were cultured in minimum essential medium with Earl’s salt and L-glutamine, supplemented with 10% FBS, 1 mM sodium pyruvate, 1% PEST. The melanoma cell line mel526 (ATCC, Manassas, VA, USA) was cultured in Iscove’s modification of

Dulbecco’s medium supplemented with 10% FBS, 1 mM sodium pyruvate, 1% PEST and 3.2 mM L-glutamine. The normal fibroblast cell line 1064SK (ATCC) was cultured in DMEM medium with 10% FBS, 1 mM sodium pyruvate, 1% PEST. The 911 cell line (Crucell, Leiden, The Netherlands) used for virus production was cultured in DMEM supplemented with 10% FBS, 1 mM sodium pyruvate, 1% PEST. All cell culture reagents were from Invitrogen (Carlsbad, CA, USA).

Recombinant adenoviruses The recombinant adenoviruses in this study are E1B-deleted, based on human serotype 5 (Ad5), and constructed using the AdEasy technology. The transfer plasmids for the Ad5 E1 region have been described earlier: pShuttle-i/CgAE1AmiR122 (ref. 10) and pShuttle(CMV-GFP).44 The Ad5 backbone plasmids have been described earlier: pAdEasy1 (ref. 45) and pAdEasy1-5/3-E3+.14 A serotype 5 adenoviral vector with the fiber shaft and knob from serotype 35, Ad5f35(GFP), was a kind gift from Dr Fan (Lund University, Lund, Sweden) and has been described and evaluated before.37,46 New Ad5 backbone plasmids. The two primers Somatostatin.F: 5¢-CGAAGTG TGACTGCTTCTGGAAGACCTGTTTCTG-3¢ and Somatostatin.R: 5¢-CGCAG AAACAGGTCTTCCAGAAGCAGTCACACTT-3¢ were annealed and the product was cloned into the BstBI–ClaI site of pFiber,27 kindly provided by Dr PJ Bosma (University of Amsterdam, Amsterdam, The Netherlands) to create pFiber-FWKT. The primer pair encodes the amino acid sequence: KCDCFWKTCFC. The pFiber plasmid contains the Ad5 fiber with the BstBI–ClaI site located in the sequence encoding the HI loop.27 The adenovirus E3 sequence was subcloned/recombined from pBlue(E3)47 using SpeI and NdeI into SrfI-linearized pAdEasy-Sce,27 kindly provided by PJ Bosma, to create pAdEasy-Sce-E3. The FWKT-modified fiber sequence was then subcloned/ recombined from pFiber-FWKT using EcoRV and HpaI into I-Sce-1-linearized pAdEasy-Sce-E3, to create pAdEasy-Sce-E3-FWKT. Recombinant viruses were produced in 911 cells, purified by CsCl banding and dialyzed against buffer containing 10 mmol l–1 Tris-HCL (pH 8.0), 2 mmol l–1 MgCl2 and 4% sucrose. Infectious viral particles were determined by a fluorescence-forming unit assay47 and evg titers were determined by quantitative real-time PCR using primers for E4 orf1: E4.For 5¢-CATCAGGTT GATTCACATCGG-3¢ and E4.Rev 5¢-AAGCGCTGTATGTTGTTCTG-3¢. A standard curve with threshold cycles (Ct) for 1103–1107 copies of a plasmid containing the cloned PCR product (pCR2.1-E4orf1) was generated. Viralspecific PCR products were measured by the CFX96 real-time detection system (Bio-Rad, Hercules, CA, USA) using iQ SYBR Green supermix (Bio-Rad). The recombinant viruses and wild-type Ad5 (Ad5wt) were stored in aliquots in 80 1C.

Flow cytometry Primary carcinoid cells and cell lines were analyzed by flow cytometry (FACSCalibur, BD Biosciences, San Diego, CA, USA) using the following antibodies: fluorescein isothiocyanate-labeled mouse anti-CD46 (BD Biosciences); mouse anti-integrin avb5 (Chemicon, Temecula, CA, USA), mouse anti-integrin avb3 (Chemicon), mouse anti-CAR (RmcB hybridoma, ATCC), mouse anti-DSG2 (BD Biosciences), secondary fluorescein isothiocyanatelabeled anti-mouse/anti-rabbit. Unspecific antibody bindings were analyzed by staining with isotype-matched fluorescein isothiocyanate-labeled control antibodies (BD Biosciences).

Analysis of SSTRs and CgA RNA expression Total RNA from primary carcinoid cells, normal hepatocytes and cell lines was extracted by using RNeasy Mini Kit (QIAGEN, Hilden, Germany). RNA purity and concentration were determined by using Nanodrop 1000 (Thermo Scientific, Waltham, MA, USA). Total RNA was aliquoted and stored at 80 1C. Reverse transcriptase reactions were performed using 1 mg of total RNA of each sample by SuperScript II Reverse Transcriptase kit (Invitrogen). Complementary DNA (diluted 1:5) was used for quantitative reverse transcriptase-PCR together with the primers: SSTR1.F 5¢-AGTTGGTCTGCGCGAAGATC-3¢; SSTR1.R 5¢-CGCGTCAGCAGCAAAGTG-3¢; SSTR2.F 5¢TCAACGTTTCTTCCGTCTCCAT-3¢; SSTR2.R 5¢- GAGGACCACCACAAA GTCAAACA-3¢; SSTR3.F 5¢-TCTACGTGCTCAACATCGTCAA-3¢; SSTR3.R Gene Therapy

Oncolytic adenovirus with somatostatin motifs J Leja et al 1060 5¢-TGGCACAGCTGTTGGCATAG-3¢; SSTR5.F 5¢-CATCCTCTCCTACGCCA ACAG-3¢; SSTR5.R 5¢-TGGAAGCTCTGGCGGAAGT-3¢; CgA.F 5¢-CCCCA CTGTAGTGCTGAACC-3¢; CgA.R 5¢-GGAGTGCTCCTGTTCTCCC-3¢; bactin.F 5¢-CGAGAAGATGACCCAGATCATG-3¢; b-actin.R 5¢-ACAGCCTGGATAGCAACGTACA-3¢. The PCR products were measured by CFX96 real-time detection system (Bio-Rad) using iQ SYBRGreen supermix (Bio-Rad). The data were evaluated using the 2 DDCT method48 and normalized against b-actin expression (set to 1).

Proximity ligation assay The PLA assay was performed with the Duolink detection kit from Olink Bioscience (Uppsala, Sweden). Cells were cultured in Lab-Tek chambers (Thermo Scientific), washed (5 min, 2phosphate-buffered saline with 2.5 mM EDTA) and incubated with adenoviruses (1000 evg per cell) for 2 h at 4 1C. The cells were washed (5 min, 1phosphate-buffered saline), fixed with ice-cold 3% paraformaldehyde, permeabilized with 0.5% TritionX-100, washed (5 min, 3phosphate-buffered saline) and blocked with 1 blocking solution (Olink Bioscience) for 2 h. Primary antibodies were: mouse monoclonal antiAd-hexon (Chemicon); rabbit polyclonal anti-CAR (Abgent Inc., San Diego, CA, USA); rabbit polyclonal anti-SSTR2,49 a kind gift from F Leu, Verto Institute LLC, The Cancer Institute of New Jersey, New Brunswick, NJ, USA; rabbit monoclonal anti-HA (Cell Signaling, Danvers, MA, USA), mouse monoclonal anti-DSG2 (BD Bioscience); mouse monoclonal anti-CD46 (BD Bioscience). Antibodies were diluted 1:50–1:400 in 1 antibody diluent (Olink Bioscience) and incubated overnight at 4 1C. Secondary probes were diluted 1:10 in 1 antibody diluent (Olink Bioscience) and incubated for 2 h at room temperature. The slides were counterstained with Hoechst 33342 (SigmaAldrich, St Louis, MO, USA) and Phalloidin 488 (Invitrogen) for 15 min at room temperature. After mounting with SlowFade medium (Invitrogen), the slides were examined with a Zeiss AxioPlan2 microscope equipped with a PlanNEOF Zeiss 40/1.32–0.6 oil objective. Images were acquired with an AxioCam MRm camera (Zeiss, Stockholm, Sweden). Fluorescence microscopy signals were quantified with the BlobFinder software,50 a freely available image analysis tool for image cytometry. Cell nuclei and point-source fluorescence signals were defined and quantified by shape and fluorescence intensity according to the standard procedures for this software tool. Settings were kept constant within each experimental series. Background level (unspecific binding of antibodies) was established, for primary antibody and secondary probes on COS7, transfected with pEGFP-C2 or p3HA-SSTR2, and BON cells without incubation with the adenoviruses. Background level was also done also for each individual primary antibodies and secondary probes in order to evaluate the working concentrations.

Receptor-binding assay [125I]Tyr11-somatostatin (NEX389, Perkin Elmer, Waltham, MA, USA), was dissolved in MQ-water and diluted to 0.5 nM (18.5 kBq ml–1) in Quantum263 cell-culture media with L-glutamine (PAA Laboratories GmbH, Pasching, Austria). Cells grown in 12-well plates were rinsed once with culture media and thereafter incubated with 0.5 ml [125I]Tyr11-somatostatin for 2 h at 4 1C. To verify receptor specificity some cells were co-incubated with 1.2 mM somatostatin (Sigma-Aldrich, S9129), for blocking [125I]Tyr11-somatostatin binding to SSTR. The receptor binding was terminated by washing cells once with 5 ml and thereafter twice with 2 ml of cold serum-free culture media. Cells were then detached using 0.5 ml of Trypsin/EDTA for 15 min in 37 1C followed by resuspension in additional 1 ml culture media. Finally, cells were counted (Z2, Coulter Counter, Beckman, CA, USA) and cell-associated radioactivity was measured in an automated gamma counter (1480 Wallac Wizard, Perkin Elmer). Bound [125I]Tyr11-somatostatin per cell was calculated.

Infectivity assay and incubation with blood coagulation FX Cells were transduced in suspension with viruses at 50, 100 or 500 evg per cell. Two hours after transduction, cells were washed and plated in 24-well plates (2105 cells per well). Cells were harvested 2 days after transduction, followed by flow cytometry analysis of GFP-positive cells. Viruses were pre-incubated with or without FX (8 mg ml 1) for 30 min at 37 1C. Adherent normal hepatocytes were then transduced with virus supernatant at a concentration of 500 evg per cell. Two hours after transduction, cells were washed and fresh Gene Therapy

medium was added. Cells were harvested 2 days later, followed by flow cytometry analysis of GFP-positive cells.

Replication assay Primary foregut and midgut carcinoid cells were transduced in suspension with viruses at 50 evg per cell. Cell lines were transduced in suspension with viruses at 100 evg per cell. Two hours after transduction, cells were washed and plated in 24-well plates (2.5105 cells per well). For replication assay, cell were harvested directly (day 0) and 2 days after transduction, followed by viral DNA isolation using the High Pure Viral Nucleic Acid Kit (Roche Applied Sciences, Mannheim, Germany). Viral replication was detected by quantitative real-time PCR with primers for the adenoviral E4 orf1 transcript (section: Recombinant adenoviruses). Gene-specific PCR products were measured by the CFX96 real-time detection system (Bio-Rad) using iQ SYBR Green supermix (Bio-Rad). The data were evaluated using the 2 DDCT method48 using the DNA level of E4 orf1 from day 0 from each sample for normalization.

Progeny virus assay The progeny virus assay was used subsequently to the replication assay to demonstrate that infectious progeny virus are formed and released on virus replication. Three days after transduction, the supernatants were collected and diluted 1:7. Diluted supernatants (100 ml) were added to confluent 911 cells in 96-well plates. Fresh medium was added after 2 h. A standard fluorescenceforming unit assay was performed after 24 h using a primary mouse monoclonal anti-Ad hexon antibody (Chemicon) and a secondary Alexa488-labeled goat polyclonal anti-mouse antibody (Invitrogen) followed by inspection of green cells in a fluorescence microscope (Olympus CK40, Tokyo, Japan) to detect progeny virus that had infected 911 cells. Furthermore, supernatants with progeny virus from the replication assay were collected after 2 days and used to infect 911 cells followed by the MTS cell titer 96 aqueous one solution cell proliferation assay (Promega, Madison, WI, USA) according to the manufacturer’s instructions.

Cell viability assay Cells were transduced in suspension with viruses at concentrations 0.01–1000 evg per cell. Two hours after transduction, the cells were plated in 96-well plates (1104 cells per well). Cell viability was analyzed 4 or 6 days after transduction using the MTS cell titer 96 aqueous one solution cell proliferation assay (Promega) according to the manufacturer’s instructions. Cell viability values were expressed in relation to the viability of untransduced cells. Average enzymatic activities from three experiments each with triplicate samples are shown (mean±s.d.).

Human blood loop model Permit to collect blood from health donor was approved by the Regional Ethical Committee (ref. no. 2008/264). The ex vivo blood loop model for whole human blood has been described earlier.24 Minor changes are that 0.5 U ml ml–1 of hirudin (LEO Pharma Nordic, Ballerup, Denmark) was immediately added to the venous blood in heparinized Falcon tubes. The blood was then transferred to heparin-coated polyvinyl chloride tubing, 21 cm long with an inner diameter of 4 mm (Corline, Uppsala, Sweden). Each loop was filled with 2 ml blood, followed by addition of 4109 evg of viral vector or phosphatebuffered saline as control.

Titration of infectious virus after blood exposure Whole blood (55 ml) from the loop was taken after 15 min and 6 h, diluted from 1:10 to 1:10 000 in DMEM medium and added to confluent 911 cells in 24-well plates. After 2 h of incubation, the plates were washed once with medium, after which fresh medium was added and the incubation continued for 24 h. The cells were harvested and GFP expression was detected by flow cytometry (FACSCalibur) as a measurement of infection. The titers obtained after 15 min and 6 h of incubation were related to the titer of viruses not exposed to human blood, which was set to 1.

Blood cell and plasma association Fractions of 10 ml of whole blood, cell and plasma (separated by centrifugation at 3000g, 20 min, 4 1C) were collected from the blood loops after 15 min. The

Oncolytic adenovirus with somatostatin motifs J Leja et al 1061 volumes were adjusted to 200 ml with DMEM medium and viral DNA was purified from each fraction using the High Pure Viral Nucleic Acid Kit (Roche Applied Sciences). Copy numbers of viral DNA were detected by quantitative reverse transcriptase-PCR with primers for the adenoviral E4 orf1 transcript (section: Recombinant adenoviruses). Gene-specific PCR products were measured by iCycler IQ real-time detection system (Bio-Rad) using iQ SYBR Green supermix (Bio-Rad). To compare copy numbers in plasma and cell fractions, the results were adjusted according to the packed cell volume (hematocrit), which is 46% for men and 42% for women.

Statistical analysis Significant differences of proximity ligation data, where the number of PLA signals per cell were evaluated using Wilcoxon Rank-Sum test, which is a nonparametric test for independent samples (GraphPad Software, San Diego, CA, USA), are denoted in the graphs as ***Po0.0001 and **Po0.001. For the cell viability (MTS assay), the values were found be normally distributed by the Shapiro–Wilk Normality test. Therefore, Student’s t-tests were performed with *Po0.05, **Po0.001. For neutralization (titration of infectious virus after blood exposure) paired Wilcoxon Rank-Sum test was performed with *Po0.05.

CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS We would like to thank Linda Sandin for experimental help and Professor Ulf Landegren for sharing equipment. The Swedish Cancer Society (contracts 08-0582 and 10-0105), the Swedish Research Council (Grant K2008-68X15270-04-3), Gunnar Nilsson’s Cancer Foundation (Grant E50/08) and the Swedish Children Cancer Foundation (PROJ08/006) supported this work. ME is a recipient of the Swedish Cancer Society Senior Investigator Award.

1 Gustafsson BI, Kidd M, Modlin IM. Neuroendocrine tumors of the diffuse neuroendocrine system. Curr Opin Oncol 2008; 20: 1–12. 2 Oberg K, Kvols L, Caplin M, Delle Fave G, de Herder W, Rindi G et al. Consensus report on the use of somatostatin analogs for the management of neuroendocrine tumors of the gastroenteropancreatic system. Ann Oncol 2004; 15: 966–973. 3 Kumar U, Grant M. Somatostatin and somatostatin receptors. Results Probl Cell Differ 2009; 50: 137–184. 4 Wright RM, Gram H, Vattay A, Byme S, Lake P, Dottavio D et al. Binding epitope of somatostatin defined by phage-displayed peptide libraries. Biotechnology (NY) 1995; 13: 165–169. 5 Zou Y, Xiao X, Li Y, Zhou T. Somatostatin analogues inhibit cancer cell proliferation in an SSTR2-dependent manner via both cytostatic and cytotoxic pathways. Oncol Rep 2009; 21: 379–386. 6 Howman-Giles R, Shaw PJ, Uren RF, Chung DK. Neuroblastoma and other neuroendocrine tumors. Semin Nucl Med 2007; 37: 286–302. 7 Pashankar FD, O’orisio MS, Menda Y. MIBG and somatostatin receptor analogs in children: current concepts on diagnostic and therapeutic use. J Nucl Med 2005; 46 (Suppl 1): 55S–61S. 8 Toth K, Dhar D, Wold WS. Oncolytic (replication-competent) adenoviruses as anticancer agents. Expert Opin Biol Ther 2010; 10: 353–368. 9 Leja J, Dzojic H, Gustafson E, Oberg K, Giandomenico V, Essand M et al. A novel chromogranin-A promoter-driven oncolytic adenovirus for midgut carcinoid therapy. Clin Cancer Res 2007; 13: 2455–2462. 10 Leja J, Nilsson B, Yu D, Gustafson E, Akerstro¨m G, Oberg K et al. Double-detargeted oncolytic adenovirus shows replication arrest in liver cells and retains neuroendocrine cell killing ability. PLoS One 2010; 5: e8916. 11 Roelvink PW, Mi Lee G, Einfeld DA, Kovesdi I, Wickham TJ. Identification of a conserved receptor-binding site on the fiber proteins of CAR-recognizing adenoviridae. Science 1999; 286: 1568–1571. 12 Douglas JT, Kim M, Sumerel LA, Carey DE, Curiel DT. Efficient oncolysis by a replicating adenovirus (ad) in vivo is critically dependent on tumor expression of primary ad receptors. Cancer Res 2001; 61: 813–817. 13 Kim M, Zinn KR, Barnett BG, Sumerel LA, Krasnykh V, Curiel DT et al. The therapeutic efficacy of adenoviral vectors for cancer gene therapy is limited by a low level of primary adenovirus receptors on tumour cells. Eur J Cancer 2002; 38: 1917–1926. 14 Kanerva A, Zinn KR, Chaudhuri TR, Lam JT, Suzuki K, Uil TG et al. Enhanced therapeutic efficacy for ovarian cancer with a serotype 3 receptor-targeted oncolytic adenovirus. Mol Ther 2003; 8: 449–458.

15 Murakami M, Ugai H, Belousova N, Pereboev A, Dent P, Fisher PB et al. Chimeric adenoviral vectors incorporating a fiber of human adenovirus 3 efficiently mediate gene transfer into prostate cancer cells. Prostate 2009; 70: 362–376. 16 Wang H, Liaw YC, Stone D, Kalyuzhniy O, Amiraslanov I, Tuve S et al. Identification of CD46 binding sites within the adenovirus serotype 35 fiber knob. J Virol 2007; 81: 12785–12792. 17 Wang H, Li ZY, Liu Y, Persson J, Beyer I, Mo¨ller T et al. Desmoglein 2 is a receptor for adenovirus serotypes 3, 7, 11 and 14. Nat Med 2011; 17: 96–104. 18 Lie PP, Cheng CY, Mruk DD. Crosstalk between desmoglein-2/desmocollin-2/Src kinase and coxsackie and adenovirus receptor/ZO-1 protein complexes, regulates blood-testis barrier dynamics. Int J Biochem Cell Biol 2010; 42: 975–986. 19 Thomas MA, Lichtenstein DL, Krajcsi P, Wold WS. A real-time PCR method to rapidly titer adenovirus stocks. Methods Mol Med 2007; 130: 185–192. 20 Zaiss AK, Machado HB, Herschman HR. The influence of innate and pre-existing immunity on adenovirus therapy. J Cell Biochem 2009; 108: 778–790. 21 Tsai V, Johnson DE, Rahman A, Wen SF, LaFace D, Philopena J et al. Impact of human neutralizing antibodies on antitumor efficacy of an oncolytic adenovirus in a murine model. Clin Cancer Res 2004; 10: 7199–7206. 22 Bangari DS, Mittal SK. Current strategies and future directions for eluding adenoviral vector immunity. Curr Gene Ther 2006; 6: 215–226. 23 Seregin SS, Amalfitano A. Overcoming pre-existing adenovirus immunity by genetic engineering of adenovirus-based vectors. Expert Opin Biol Ther 2009; 9: 1521–1531. 24 Danielsson A, Elgue G, Nilsson BM, Nilsson B, Lambris JD, To¨tterman TH et al. An ex vivo loop system models the toxicity and efficacy of PEGylated and unmodified adenovirus serotype 5 in whole human blood. Gene Therapy 2010; 17: 752–762. 25 Dmitriev I, Krasnykh V, Miller CR, Wang M, Kashentseva E, Mikheeva G et al. An adenovirus vector with genetically modified fibers demonstrates expanded tropism via utilization of a coxsackievirus and adenovirus receptor-independent cell entry mechanism. J Virol 1998; 72: 9706–9713. 26 Mizuguchi H, Koizumi N, Hosono T, Utoguchi N, Watanabe Y, Kay MA et al. A simplified system for constructing recombinant adenoviral vectors containing heterologous peptides in the HI loop of their fiber knob. Gene Therapy 2001; 8: 730–735. 27 Lie ALM, Bakker CT, Wesseling JG, Bosma PJ. AdEasy-based cloning system to generate tropism expanded replicating adenoviruses expressing transgenes late in the viral life cycle. Gene Therapy 2005; 12: 1347–1352. 28 Bradshaw AC, Parker AL, Duffy MR, Coughlan L, van Rooijen N, Ka¨ha¨ri VM et al. Requirements for receptor engagement during infection by adenovirus complexed with blood coagulation factor X. PLoS Pathog 2010; 6: e1001142. 29 Shayakhmetov DM, Gaggar A, Ni S, Li ZY, Lieber A. Adenovirus binding to blood factors results in liver cell infection and hepatotoxicity. J Virol 2005; 79: 7478–7491. 30 Carlisle RC, Di Y, Cerny AM, Sonnen AF, Sim RB, Green NK et al. Human erythrocytes bind and inactivate type 5 adenovirus by presenting Coxsackie virus-adenovirus receptor and complement receptor 1. Blood 2009; 113: 1909–1918. 31 Au T, Thorne S, Korn WM, Sze D, Kirn D, Reid TR. Minimal hepatic toxicity of Onyx015: spatial restriction of coxsackie-adenoviral receptor in normal liver. Cancer Gene Ther 2007; 14: 139–150. 32 Gustafsdottir SM, Schallmeiner E, Fredriksson S, Gullberg M, So¨derberg O, Jarvius M et al. Proximity ligation assays for sensitive and specific protein analyses. Anal Biochem 2005; 345: 2–9. 33 Weibrecht I, Leuchowius KJ, Clausson CM, Conze T, Jarvius M, Howell WM et al. Proximity ligation assays: a recent addition to the proteomics toolbox. Expert Rev Proteomics 2010; 7: 401–409. 34 Gustafsdottir SM, Nordengrahn A, Fredriksson S, Wallgren P, Rivera E, Schallmeiner E et al. Detection of individual microbial pathogens by proximity ligation. Clin Chem 2006; 52: 1152–1160. 35 Hidaka C, Milano E, Leopold PL, Bergelson JM, Hackett NR, Finberg RW et al. CAR-dependent and CAR-independent pathways of adenovirus vector-mediated gene transfer and expression in human fibroblasts. J Clin Invest 1999; 103: 579–587. 36 Vigne E, Mahfouz I, Dedieu JF, Brie A, Perricaudet M, Yeh P et al. RGD inclusion in the hexon monomer provides adenovirus type 5-based vectors with a fiber knob-independent pathway for infection. J Virol 1999; 73: 5156–5161. 37 Nilsson M, Karlsson S, Fan X. Functionally distinct subpopulations of cord blood CD34+ cells are transduced by adenoviral vectors with serotype 5 or 35 tropism. Mol Ther 2004; 9: 377–388. 38 Howitt J, Bewley MC, Graziano V, Flanagan JM, Freimuth P. Structural basis for variation in adenovirus affinity for the cellular coxsackievirus and adenovirus receptor. J Biol Chem 2003; 278: 26208–26215. 39 Reubi JC, Waser B, Cescato R, Gloor B, Stettler C, Christ E et al. Internalized somatostatin receptor subtype 2 in neuroendocrine tumors of octreotide-treated patients. J Clin Endocrinol Metab 2010; 95: 2343–2350. 40 Hua YP, Yin XY, Peng BG, Li SQ, Lai JM, Liang HZ et al. Mechanisms and influence of octreotide-induced regulation of somatostatin receptor 2 on hepatocellular carcinoma. Chemotherapy 2009; 55: 312–320. 41 Korner M, Eltschinger V, Waser B, Schonbrunn A, Reubi JC. Value of immunohistochemistry for somatostatin receptor subtype sst2A in cancer tissues: lessons from the comparison of anti-sst2A antibodies with somatostatin receptor autoradiography. Am J Surg Pathol 2005; 29: 1642–1651. 42 Sarkioja M, Pesonen S, Raki M, Hakkarainen T, Salo J, Ahonen MT et al. Changing the adenovirus fiber for retaining gene delivery efficacy in the presence of neutralizing antibodies. Gene Therapy 2008; 15: 921–929.

Gene Therapy

Oncolytic adenovirus with somatostatin motifs J Leja et al 1062 43 Pesonen S, Nokisalmi P, Escutenaire S, Sa¨rkioja M, Raki M, Cerullo V et al. Prolonged systemic circulation of chimeric oncolytic adenovirus Ad5/3-Cox2 L-D24 in patients with metastatic and refractory solid tumors. Gene Therapy 2010; 17: 892–904. 44 Cheng WS, Kraaij R, Nilsson B, van der Weel L, de Ridder CM, To¨tterman TH et al. A novel TARP-promoter-based adenovirus against hormone-dependent and hormonerefractory prostate cancer. Mol Ther 2004; 10: 355–364. 45 He TC, Zhou S, da Costa LT, Yu J, Kinzler KW, Vogelstein B. A simplified system for generating recombinant adenoviruses. Proc Natl Acad Sci USA 1998; 95: 2509–2514. 46 Nilsson M, Ljungberg J, Richter J, Kiefer T, Magnusson M, Lieber A et al. Development of an adenoviral vector system with adenovirus serotype 35 tropism; efficient transient gene transfer into primary malignant hematopoietic cells. J Gene Med 2004; 6: 631–641.

47 Danielsson A, Dzojic H, Nilsson B, Essand M. Increased therapeutic efficacy of the prostate-specific oncolytic adenovirus Ad[I/PPT-E1A] by reduction of the insulator size and introduction of the full-length E3 region. Cancer Gene Ther 2008; 15: 203–213. 48 Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-delta delta C(T)) method. Methods 2001; 25: 402–408. 49 Leu FP, Nandi M, Niu C. The effect of transforming growth factor beta on human neuroendocrine tumor BON cell proliferation and differentiation is mediated through somatostatin signaling. Mol Cancer Res 2008; 6: 1029–1042. 50 Allalou A, Wahlby C. BlobFinder, a tool for fluorescence microscopy image cytometry. Comput Methods Programs Biomed 2009; 94: 58–65.

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)

Gene Therapy