Gene Therapy (2010) 17, 1234–1243 & 2010 Macmillan Publishers Limited All rights reserved 0969-7128/10 www.nature.com/gt ORIGINAL ARTICLE Efficient inhibition of B- xenografts with a novel recombinant fusion : anti-CD20Fab-LDM

C Xin1,3,SYe1,3, Y Ming1, Z Shenghua2, M Qingfang2, G Hongxing1,SXu1, X Yuanfu1, Z Yuan1, F Dongmei1, L Juanni1, G Yingdai1, J Lianfang2, S Rongguang1, Z Zhenping1, W Jianxiang1,CTao1, Y Chunzheng1, X Dongsheng1 and Z Yongsu2 1State Key Laboratory of Experimental Hematology, Department of Pharmacy, Institute of Hematology & Hospital of Blood Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, PRC and 2Department of Oncology, Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, PRC

Lidamycin (LDM) is a new member of enediyne antitumor respectively, and then anti-CD20Fab-LDM was administered antibiotics family that can be separated and reconstituted. It intravenously with the highest dose of 4 nmol kg À1. The consists of a labile active enediyne chromophore (AE) and a inhibition rates of tumor growth were 90.1 and 85%, which noncovalently bound apoprotein (LDP). LDM is now in phase were saliently superior to those of nontargeted LDM. It is II clinical trials. In this study, we described the antitumor noteworthy that anti-CD20Fab-LDM can inhibit the growth of features of a fusion protein of LDM, anti-CD20Fab-LDM, patient-derived cells, including rituximab-resistant patient- targeted to CD20 expressed by B-lymphoid malignancies. derived cells. Thus, CD20-targeted delivery of LDM is a Especially, LDM was prepared by a novel two-step method specific and potent therapeutic strategy for B-lymphoid including DNA recombination and molecular reconstitution. malignancies. In addition, the two-step approach could serve Anti-CD20Fab-LDM exerted potent cytotoxicity against as a new technology platform for making a series of highly CD20+ B-cell lymphoma cell lines in vitro (IC50: 10–30 pM) potent engineered antibody-based drugs. and in the Raji xenograft model. Two Raji xenografts were Gene Therapy (2010) 17, 1234–1243; doi:10.1038/gt.2010.76; allowed to grow to an initial mass of 80 and 500 mm3, published online 13 May 2010

Keywords: anti-CD20Fab-LDM; B-cell lymphoma; recombinant fusion protein; LDM; CD20

Introduction B-cell non-Hodgkin’s lymphoma have primary refrac- tory disease or develop relapses after treatment with The efficiency of treatments for B-lymphoid malignan- RTX, whereas the efficacy of RTX is limited in B-cell cies remains a seriously unmet medical need. Nowadays, chronic lymphocytic leukemia patients.1 The success of B-lymphoid malignancies are the fifth most common RTX presents clear opportunity for antibody-based malignancy in the world and its incidence is conti- agents in lymphoma, although the limitations of it calls nuously rising. Besides, B-lymphoid malignancies for such agents with different mechanisms. Focused tumors can relapse in B-lymphoid malignancies were delivery of the cytotoxic agent to tumor cells maximizes observed who initially responded to treatment. its antitumor effect and minimizes its normal tissue Rituximab (RTX), the CD20-specific chimeric monoclonal exposure, resulting in an improved therapeutic index. antibody, was approved by the Food and Drug Admin- Moreover, antibody-targeted chemotherapeutic agent is istration in 1997, and markedly improved clinical especially effective against RTX-resistant malignant B responses in B-lymphoid malignancies were observed. cells that currently escape from therapy. Although the Unfortunately, about 50% of patients with aggressive first antibody-targeted strategy has been applied since 1999, it had not been successfully trans- lated into clinical benefit until the recent introduction of Correspondence: Dr X Dongsheng, Department of Pharmacy, Institute of Hematology & Hospital of Blood Diseases, Chinese gemtuzumab ozogamicin, the only antibody-targeted Academy of Medical Sciences & Peking Union Medical College, 288 chemotherapeutic agent approved for clinical use in the Nanjing Road, Tianjin 300020, PRC. United States.2 or Professor Z Yongsu, Department of Oncology, Institute of Gemtuzumab ozogamicin is comprised of a huma- Medicinal Biotechnology, Chinese Academy of Medical Sciences & nized IgG4 anti-CD33 antibody covalently linked to Peking Union Medical College, 1# Tiantan Xili, Beijing 100050, PRC. N-acetyl-g-calicheamicin dimethyl hydrazide, which is a E-mail: [email protected] or [email protected] 3 3These authors contributed equally to this work. member of enediyne antibiotics. Enediyne antibiotics Received 30 November 2009; revised 7 April 2010; accepted 8 April have been focused on their potent antitumor activity 2010; published online 13 May 2010 because of their unique ability to damage the DNA Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1235

Figure 1 Expression and purification of the fusion protein. (a) Schematic representation of the expression plasmids for the fusion protein. G4S, the Gly Gly Gly Gly Ser linker. Note: the drawings are not to the scale. (b and c) Expression and purification of the fusion protein. The fusion protein was expressed in E. coli, purified by affinity chromatography and analyzed by SDS–PAGE and immunoblot analysis. (b) Colloidal staining of a SDS–PAGE gel. Lane 1: fusion protein eluted from the anti-protein G column under nonreducing condition; lane 2: flow through from the anti-protein G column under nonreducing condition; lane 3: periplasmic extract of pCANTAB 5E-CD20Fab-LDP expressing the fusion protein under nonreducing condition; lane 4: fusion protein eluted from the anti-protein G column under reducing condition; lane 5: flow through from the anti-protein G column under reducing condition; lane 6: periplasmic extract of pCANTAB 5E-CD20Fab-LDP expressing the fusion protein under reducing condition. Also shown on left are the positions of molecular weight markers. (c) Immunoblot analysis with an anti-LDP antibody. Lane 1: fusion protein eluted from the anti-protein G column under nonreducing condition; lane 2: flow through from the anti-protein G column under nonreducing condition; lane 3: periplasmic extract of pCANTAB 5E- CD20Fab-LDP expressing the fusion protein under nonreducing condition. Note: the anti-LDP antibody identified only three-dimensional epitope (data under reducing condition not shown). of tumor cells by inducing single-strand and/or recombinant DNA techniques. Second, the energized double-strand breaks through free radical attacks on fusion protein anti-CD20Fab-LDM was prepared by the deoxyribose moieties in DNA. Lidamycin (LDM, also molecule reconstitution, that is, putting the active designated as C1027) is a new member of the enediyne enediyne chromophore into anti-CD20Fab-LDP. Impor- antibiotic family, which was isolated from a Streptomyces tantly, anti-CD20Fab-LDM resulted in a superior anti- globisporus C1027 strain in China. LDM shows extremely tumor activity compared with LDM: this fusion protein potent cytotoxicity, antitumor activity and marked showed extremely potent cytotoxicity to B-cell growth inhibition of transplantable tumors in mice.4–7 lymphoma and marked antitumor activity in vivo.In In terms of the half maximal inhibitory concentration contrast, anti-CD20Fab-LDM lacked systemic toxicity in (IC50) values, the cytotoxicity of LDM was 10 000-fold nude mice. The efficient antitumor activity of anti- more potent than those of mitomycin and .7,8 CD20Fab-LDM made it a treatment option for Unlike calicheamicin, LDM consists of a noncovalently B-lymphoid malignancies. bound apoprotein LDP (10 500 Da), which formed a hydrophobic pocket for protecting the chromophore, and an enediyne chromophore (843 Da), responsible for the Results extremely potent bioactivity, which binds DNA in the minor groove and causes double-strand DNA breaks in to Construction and production of the fusion protein a thiol-dependent manner, leading to apoptosis and cell To achieve the fusion protein, the DNA sequence death.9–11 The apoprotein and chromophore can be encoding for the CH1 fragment of the anti-CD20Fab dissociated and reconstituted, and the biological activity was genetically fused to the LDP gene, which encodes for of the rebuilt molecule is comparable to that of nature the apoprotein of LDM, with a five-amino-acid spacer LDM. Therefore, anti-CD20Fab-LDM can be prepared by a (GGGGS) between the C-terminus of CH1 and the novel two-step method including DNA recombination and N-terminus of LDP, and cloned into the ApaI/SphI molecular reconstitution. It provided a new technology restriction sites of the expression vector (Figure 1a). platform for making a series of highly potent engineered The soluble fusion protein was released from the antibody-based drugs for a variety of . periplasmic space of E. coil strain HB2151 containing We constructed a chimeric anti-CD20Fab0 fragment the expression vector by osmotic shock and purified with from the hybridoma cell line and demonstrated that a protein G tag affinity chromatography. The yield of the Fab fragment could be produced in Escherichia coli the purified fusion protein in the shaker flask was about À in soluble form with good retention of antigen- 4–5 mg l 1 media from an overnight culture. binding activities.12 In this study, we developed a two-step approach to construct a new fusion protein, SDS polyacrylamide gel electrophoresis and western anti-CD20Fab-LDM. First, the fusion protein anti- blot analysis of the fusion protein CD20Fab-LDP that is composed of the apoprotein LDP The fusion protein was analyzed on 12% SDS poly- of LDM and Fab from anti-CD20mAb was prepared by acrylamide gel electrophoresis under nonreducing and

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1236 of the fusion protein had driven their selective accumu- lation in the cores of solid tumors after i.v. injection.

Preparation of energized fusion protein anti-CD20Fab-LDM The energized fusion protein anti-CD20Fab-LDM was prepared by adding AE, the active enediyne chromophore of LDM, to anti-CD20Fab-LDP solution in controlled conditions, followed by separating from free AE by the sephadex G-75 column (Amersham Bioscience, Arlington Heights, IL, USA). As shown in Figure 4, the correct anti- CD20Fab-LDM molecules showed strong absorbance at both 340 and 280 nm, which represents the absorbance of the chromophore and protein moiety, respectively.

CD20-restricted killing activity in vitro Figure 2 CD20-specific binding of the fusion protein as assayed by fluorescence-activated cell sorting analysis. CD20-positive Raji The effects of CD20-specific anti-CD20Fab-LDM on the cells were exposed for 1 h at 4 1C to increasing concentrations of growth of CD20-positive cell lines and patient-derived anti-CD20Fab or anti-CD20Fab-LDP. Binding of the above anti- cells were examined in 72 h in vitro culture assays. LDM bodies to Raji cells were detected by flow cytometry. Ka ¼ was used as a nonbinding control to explore antigen- 9 À1 0.1 Â10 M ; n ¼ 3. nonspecific effects. Anti-CD20Fab was also used to assess the sensitivity of these cell lines to LDM (not shown). Adriamycin (ADR) was as a positive drug reducing condition. In nonreducing condition, the control. As shown in Figure 5, the cytotoxicity of anti- purified fusion protein yielded a single protein band of CD20Fab-LDM was 10 000-fold more potent than that of expected size; however, under the reducing condition, ADR. Especially, CD20-positive cells exhibited greater the band at 62 kD was resolved into two protein bands sensitivity to anti-CD20Fab-LDM than CD20-negative with the calculated molecular weights of 39 and 23 kD, cells, and anti-CD20Fab-LDM exerted a 6- to 10-fold because the disulfide bond was broken (Figure 1b). more potent growth inhibitory effect on those of the Western blot analysis using a mouse anti-LDP antibody CD20-positive cell lines than LDM (Po0.05). Of note, confirmed the existence of the 62-kD fragment, which in most cell lines tested, anti-CD20Fab-mediated killing contained LDP fragment (Figure 1c). These results did not exceed 10% even at the presence of the highest m À1 indicated that the purified expression product was concentration (0.01 mol l ). And antiCD20Fab-LDP anti-CD20Fab-LDP fragment. The fusion protein was also exerted a 5- to 8.9-fold more potent growth inhi- not detected in the flow through fraction, suggesting bitory effect on primary patient-derived tumor B cells high efficiency of the purification process. than LDM (Table 1).

CD20-specific binding in vitro Effects of the fusion protein on growth of Flow cytometry was carried out to confirm the CD20- subcutaneous xenografts specific binding of the fusion protein. Incubation of Anti-CD20Fab-LDM was examined for its efficacy CD20-positive Raji cells with different concentrations against subcutaneous Raji xenografts established in of anti-CD20Fab-LDP resulted in specific binding. nude mice. LDM was used as a nonbinding control. Similar affinity data were obtained with anti-CD20Fab The mice were administered i.v. with 4 nmol kgÀ1 (Figure 2). This further confirmed that fusing an LDP LDM, 2 nmol kgÀ1 LDM, 4 nmol kgÀ1 anti-CD20Fab-LDM, to anti-CD20Fab did not have an obvious impact on 2nmolkgÀ1 anti-CD20Fab-LDM, 4 nmol kgÀ1 anti-CD20Fab the ability of the fusion protein to bind the surface of and phosphate-buffered saline (PBS) (vehicle) (Q1W Â 2). CD20-positive cells. Tumors were allowed to grow to an initial mass of 40–60 mm3 (small size, Figure 6a) or 450–500 mm3 In vivo targeting and accumulation of the fusion protein (large size, Figure 6b) before the initiation of therapy. In vivo distributions of the fusion protein were observed The anti-CD20Fab-LDM inhibited the subcutaneous in nude mice bearing the Raji or K562 xenografts. growth of Raji xenografts in a dose-dependent manner. In vivo fluorescein isothiocyanate (FITC)-labeling of nude Mice treated with LDM at tolerated dose of 4 nmol kgÀ1 mice revealed that fusion protein penetrated into CD20- showed an inhibition rate of 65 (small size) and 62% positive tumors within 1 h after intravenous (i.v.) (large size). The anti-CD20Fab-LDM at the same con- injection (16 nmol per mouse) and then gradually centration suppressed the tumor growth by 90.1 (small accumulated in the tumor cores. Such accumulation size) and 84.5% (large size), which demonstrated spread to the whole tumor area within 3 h after injection, statistically significantly differences (Po0.05) compared but no similar accumulation was noticed in CD20- with those of LDM in each of the studies, regardless of negative tumors. At 8–10 h after injection, anti-CD20Fab the initial tumor mass. Almost no therapeutic benefit and anti-CD20Fab-LDP still remained in the tumors, could be achieved in the group treated with 4 nmol kgÀ1 whereas the labels had been cleared from other areas. In anti-CD20Fab. The same conclusion was shown in contrast, there was hardly any accumulation of LDP Figure 6, the tumor volume of mice treated with LDM (16 nmol per mouse) in tumors (Figure 3). Collectively, at tolerated dose of 4 nmol kgÀ1 was two- to fourfold our data strongly suggested that the targeting specificity more potent than that of anti-CD20Fab-LDM. During the

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1237

Figure 3 Fluorescence images of tumors in nude mice treated with FITC-labeled anti-CD20Fab, anti-CD20Fab-LDP and LDP. Green fluorescence signals marked the location of FITC-labeled antibody. Images showed that within 1 h after injection, anti-CD20Fab and anti- CD20Fab-LDP penetrated into the CD20-positive tumor then gradually accumulated in the core area of the tumor. At 8–10 h after injection, anti-CD20Fab and anti-CD20Fab-LDP still remained in the tumors, whereas the labels had been cleared from other areas. But no similar accumulation was observed in CD20-negative tumors. Images showed that LDP could also penetrated into the whole tumor either CD20-positive or CD20-negative tumors, but hardly any accumulation of LDP was observed in tumors at 8–10 h after injection; n ¼ 5.

Figure 4 (a) Schematic representation of anti-CD20Fab-LDM. Note: the drawings are not to the scale. (b) Separation of the fusion protein anti-CD20Fab-LDP from free AE on a PD-10 column (sephadex G-25; Amersham Bioscience). Elute: 1 Â PBS. (c and d) Structures of C-1027 and the enediyne chromophore. experiment, mice in all groups were observed with no kidney. Nude mice in this study were exposed to total significant adverse reactions, such as lethargy and body irradiation (400 rad) to further suppress their anorexia. residual immune system, so data were toxic than Kunming mice without irradiation. Anti-CD20Fab-LDM lacks toxicity in nude mice As it is shown in Figure 7 and Table 2, anti-CD20Fab- LDM lacked systemic toxicity in nude mice. In contrast, Discussion treatment with LDM (4 nmol kgÀ1) was associated with some toxicity in mice with weight loss; an 8% damages We explored a CD20-specific fusion protein anti- the liver, a 4% damages the spleen and a 4% damages the CD20Fab-LDM. The therapeutic strategy targeting

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1238

Figure 5 Effect of CD20-specific fusion protein on the growth of human B cell lines in vitro. Human B cell lines were cultured individually in the presence of increasing concentrations of anti-CD20Fab-LDM or LDM. After 72 h, viability in each culture was assessed using the MTT method. The half maximal inhibitory concentration (IC50) values were calculated by logistic nonlinear regression and were presented in the table as pM equivalents; n ¼ 3.

Table 1 Effect of CD20-specific fusion protein on the growth of patient-derived cells in vitro

Patient characteristics and specific killing activity of the fusion protein

Patient number Gender Diagnosis CD19+ (%) IC50 (mmol lÀ1) anti-CD20Fab-LDM LDM Rituximab resistant

1 F NHL 30 7.2 Â 10À6 6.1 Â 10À5 Y 2 M NHL 27 2 Â 10À6 1 Â 10À5 n.d 3 F DLBCL 23 3.5 Â 10À6 2 Â 10À5 n.d 4 F DLBCL 15 9.2 Â 10À6 8 Â 10À5 n.d

Abbreviations: DLBCL, diffuse large B-cell lymphoma; f, female; IC50, the half maximal inhibitory concentration; LDM, lidamycin; m, male; n.d., not detected; NHL, non-Hodgkin’s lymphoma; y, yes.

CD20 has been extensively exploited for years with a The anti-CD20Fab-LDM was successfully constructed as number of antibody invented, conjugated or not, an extremely potent and specific antitumor fusion and some of them are already in the first line of clinical protein with tolerable systematic toxicity to eradicate use. Yet the improvements in this area are still in the target malignant B cells. pressing need. Especially, anti-CD20Fab-LDM is innovative and In this study, we developed a new way of CD20- noteworthy in that it simultaneously circumvented targeting therapy by exploiting the anti-CD20Fab-LDM, several mechanisms that were held primarily responsible which exhibited a potent, dose-dependent cytotoxicity for RTX resistance. To our knowledge, it is the first time against CD20+ B-lymphoma cells. that all these valuable features, together with highly Primarily, we systematically assayed the validity of the specific and potent antitumor activity, were artificially fusion protein’s bioactivity. As the potency is concerned, integrated into one single molecule. with the IC50 value of only 10–30 pM, it was 10 000-fold First, anti-CD20Fab-LDM is effective against the more potent than that of ADR. Theoretically, only one effector exhaustion. The therapeutic effect of RTX copy of it is enough to kill a cell.6,13 Anti- depends on the host immune effector system, including CD20Fab-LDM also caused a dose-dependent regression effector cells (natural killer (NK) cell, mediate antibody- of Raji xenografts grown as subcutaneous solid tumors in dependent cell-mediated cytotoxicity (ADCC) effect) and athymic nude mice. It was effective not only in the molecules complement-depend cytotoxicity effect (CDC), initially established Raji xenografts, but also in preexist- continuous administration of RTX and high tumor ing large established ones. The inhibition rates of tumor burden can exhaust host effector function and nullify growth were 90.1 and 84.5%, respectively, with the the effect of RTX. Fischer et al.14 reported that NK-cell- concentration of anti-CD20Fab-LDM up to 4 nmol kgÀ1. mediated ADCC of RTX-opsonized cells promotes Whereas those of non-targeted LDM were only 65 and upregulation of CD107a, a marker of degranulation and 62%, respectively. Further, anti-CD20Fab lacked anti- presumably cell exhaustion. Berdeja et al.15 found that tumor activity at the dose evaluated. The data indicated ADCC may be significantly reduced owing to high that both the targeting moiety and the ‘warhead’ moiety burdens of RTX-opsonized cells. Kennedy et al.16 of the fusion protein served their duty as expected. reported that RTX infusion in chronic lymphocytic

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1239

Figure 6 The antitumor activity of the fusion protein anti-CD20Fab-LDM and LDM in vivo.(a) Growth inhibition and volume ratio of small subcutaneous xenografts in nude mice. Drugs were administrated when tumor mass reached 40–60 mg. (b) Growth inhibition and volume ratio of large subcutaneous xenografts in nude mice. Drugs were administrated when tumor mass reached 450–500 mg; n ¼ 7.

g-radiation is the new safety issue, which can damage nearby progenitor cells in bone marrow, and cause pancytopenia. Sublethal radiation damage to bone mar- row stem cells increases the risk of myelodysplastic syndrome and/or acute myelogenous leukemia.17 Inter- estingly, the anti-CD20Fab-LDM elicit cytotoxicity through the conjugated LDM without any g-radiation, rather than ADCC or CDC effects, which depends on the host effector system, and therefore avoids the effector- exhaustion problem, which hampered the efficacy of RTX. Second, anti-CD20Fab-LDM is effective against ‘shaving of CD20’, which refers to the phenomenon that soon after the clearance of circulating RTX-opsonized B cells, additional malignant B cells enter blood stream from other compartments, and CD20 are removed (‘shaved’) from these cells.16,18,19 This was mediated by Figure 7 Body weight of nude mice treated with anti-CD20Fab- the interaction between the Fc fragment of RTX and Fc-g LDM or LDM; n ¼ 7. receptor I of monocytes and macrophages; data indicate 0 20 that F(ab )2 of RTX effectively avoided the problem, indicating that removal of Fc fragment from the CD20- leukemia patients depleted complement. It seems that targeting antibody is a solution for shaving of CD20. We new strategy independent of host immune system adopted CD20-Fab fragment as the targeting moiety of should avoid this problem. The radiolabelled antibodies, the fusion protein; compared with the whole antibody, Y-90 ibritumomab tiuxetan (Zevalin, Biogen-IDEC, Fab downsized the molecule weight while retaining Cambridge, MA, USA) and I-131 tositumomab (Bexxar, Glaxo the binding specificity, and most importantly, with the Smithkline, Middlesex, British), exert effects dependent Fc fragment deleted, the CD20-Fab-LDM successfully on b-emissions with path lengths of 1–5 mm, and thus avoided the ‘shaving of CD20’ phenomenon, which is avoided the effector exhaustion problem of RTX, but mediated by the Fc–FcR interaction.

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1240 Table 2 Quantitative comparison of tissue toxicity with histopathological sections images

Lesion in spleen a Lesion in liver b Lesion in kidney b (%, x¯±s.d., n ¼ 7) (%, x¯±s.d., n ¼ 7) (%, x¯±s.d., n ¼ 7)

PBS No change No change No change Anti-CD20Fab (4 nmol kgÀ1) No change No change No change Anti-CD20Fab-LDM (2 nmol/kg) No change No change No change Anti-CD20Fab-LDM (4 nmol kgÀ1) No change No change No change LDM (2 nmol kgÀ1) No change No change No change LDM (4 nmol kgÀ1) 4.2±1.7 8.34±2.9 2.6±0.34

Abbreviations: LDM, lidamycin; PBS, phosphate-buffered saline. aTissue injury was measured by the ratio of 4 nmol kgÀ1 LDM group versus PBS group in the total number of nucleated cells in spleen. bTissue injury was measured by the area ratio of lesions versus total on the tissue surface.

Finally, anti-CD20Fab-LDM is effective against BCL-2 tumor then gradually accumulated in the core area of the overexpression. Raised apoptosis threshold by BCL-2 tumor within 3 h. overexpression is considered a major cause of RTX Besides, we innovatively constructed the CD20-Fab resistance. The crosslink of CD20 receptors by RTX can fragment and the apoprotein of LDM into one fusion trigger caspase3-dependent apoptosis signaling, which protein, which was expressed in E. coli solubly. To our can be hampered by the overexpression of anti-apoptotic knowledge, anti-CD20Fab-LDP is the first soluble BCL-2. Combination of antisense BCL-2 oligonucleotides expression of CD20-Fab with prokaryote. Compared can enhance the efficacy of RTX.21 sodium, with eukaryote expression systems, which were used a BCL-2 antisense oligonucleotide, is under phase II trial by many merchandised monoclonal antibodies or other in combination with RTX in patients with recurrent antibody products, the prokaryote expression is far more B-cell non-Hodgkin’s lymphoma,22 which holds out the suitable for massive production, in that the condition promise of combination therapy with anti-CD20 and management is simpler and the cost is much cheaper. BCL-2 expression suppressing method. It is reported that In general, insufficient and costly production of mono- LDM alone is able to suppress the expression of BCL-2,23 clonal antibodies using conventional technology has the same effect obtained by the management of BCL-2 severely hampered the clinical evaluation of these antisense nucleotide. According to our own experiment reagents. Expression of these molecules in E. coil can (data not shown), with conjugated LDM, anti-CD20Fab- improve the clinical application, therefore expressing LDM can efficiently circumvent the upregulated apop- and purifying anti-CD20Fab-LDM seems to be an attrac- tosis threshold mediated by BCL-2 overexpression, tive and feasible approach for experimental and clinical which have been hindering the effect of RTX. And tumor therapy. therefore the conjugated LDM can facilitate the apoptosis In our study, we obtained the energized fusion protein induction effect conducted by the crosslink of CD20 anti-CD20Fab-LDM by molecular reconstitution; the receptors, upon binding of anti-CD20Fab-LDM or RTX. anti-CD20Fab-LDP fusion protein were integrated with With all above-mentioned exceptional antitumor an active enediyne chromophore. The molecule of the application features, anti-CD20Fab-LDM is especially energized fusion protein anti-CD20Fab-LDM contains easy and economical to produce, in the unique way we three parts, including anti-CD20Fab, LDP and AE, and constructed and expressed it. Productivity and feasi- each of them has different roles in exerting the biological bility are also important in the manufacture of anti- function of the whole molecule. The Fab portion of body product, especially in consideration of the future anti-CD20 antibody serves as a tumor-targeting vehicle, industrialization. Immunotoxins containing chemical and the active enediyne AE serves as a ‘warhead’ with conjugates of ligand and toxin were often difficult to highly potent cytotoxicity. In addition, LDP works as produce because the ligand and toxin had to be purified a carrier for AE by offering a hydrophobic pocket. separately, the conjugation procedure is difficult As expected, we successfully integrated the above- with low yield, and the junction between the toxin and mentioned activities into one single energized fusion ligand could be at many different sites (for example, at protein of anti-CD20Fab-LDM. The chomophore released any lysine residue). And percentage of the conjugate from the holoprotein is highly labile, the LDM has to product would have an inappropriate toxin: ligand enter target cell in the form of holoprotein, and bind to ratio, and consequently would suffer from inactivity. In the minor groove of the DNA strands, thus the chromo- contrast to those antibody conjugates linked by chemical phore is released sufficiently close to extract hydrogen reaction, our way preserve the high-order structure from ribose rather than other sources, to break the DNA and activity of the Fab, and easily obtained the perfect strands. Therefore the pre-binding release of the chromo- 1:1 antibody-effector ratio in the final products. Data phore is ineffective, in that way, the specificity of the indicated fusing an LDP to anti-CD20Fab did not have ultracytotoxicity of the conjugated LDM is ensured.24 an obvious impact on the ability of the fusion protein to Obviously, the anti-CD20Fab-LDM is in superior bind to the surface of CD20-positive cells. The affinity advantage to unconjugated LDM in terms of specificity. constant of anti-CD20Fab and anti-CD20Fab-LDP were Anti-CD20Fab-LDM exerted potent cytotoxicity against 0.10eÀ9 MÀ1 and 0.11eÀ9 MÀ1. In vivo images showed CD20+ B-cell lymphoma cell lines (inhibitory concentra- that within 1 h after injection, anti-CD20Fab and anti- tion of 50%: 10–30 pM) and retarded the growth of Raji CD20Fab-LDP both penetrated into the CD20-positive in mice. All these demonstrated that AE was added

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1241 accurately into the hydrophobic pocket of apoprotein Construction of anti-CD20Fab-LDP LDP and indicated that anti-CD20Fab-LDM has super The first antibody heavy chain constant domain (CH1) antitumor activity in contrast with the same concentra- gene of the anti-CD20 antibody and the LDP gene tion of LDM. were cloned from vector pCANTAB 5E-Fcd20Fab0 and Moreover, much of the data addressing CD20-targeted pGEMT-LDP as previously described,12 and then the immunoconjugate of enediyne antibiotic comes from synthetic DNA sequence encoding CH1-LDP was assays with cell lines, and the relevance to activity in generated by Splicing by Overlapping Extension PCR patients remains uncertain. In this report, we demon- (SOE PCR) technology using the CH1 and LDP genes. strated firstly that anti-CD20Fab-LDM can inhibit the The CH1 and LDP sequences were genetically linked proliferation of primary patient-derived malignant B through a flexible peptide linker (GGGGS). Moreover, cells and, its inhibition was even effective in one relapsed restriction enzyme sites ApaI (GGCCCA) and SphI patient after 2-year standard RTX therapy. Hence, (GCATGC) were added to the 50-end and 30-end of this our data provide evidence for the therapeutic potential sequence, yielding a 669-bp DNA fragment designated of anti-CD20Fab-LDM, which may help those RTX- ApaI-CH1 LDP-SphI. After digestion with ApaI and SphI, resistant patients. this fragment was cloned into the vector pCANTAB 5E- In summary, we successfully constructed and Fcd20Fab0, which contained the variable heavy expressed the fusion protein of anti-CD20Fab-LDM, chains (VH), the variable light chains (VL) and the anti- which transported the potent antitumor agent LDM to body light chain constant domain (CL) gene of the anti- the valuable target CD20. The fusion protein exhibited CD20 antibody, yielding a plasmid named pCANTAB satisfying specificity and cytotoxicity both in vitro and 5E-CD20Fab-LDP. in vivo. Its unique structure and feature successfully solved several shortcomings in conventional CD20- Expression and purification of the fusion protein targeting therapeutics, similar to ‘host effector exhaus- This fusion protein was secreted from E. coli strain tion’, ‘shaving of CD20’ and BCL-2-mediated apoptosis HB2151 containing the expression plasmid grown at resistance. The fusion protein was expressed solubly 30 1C in the shaker flask following a procedure with high yields in E. coli at low cost and possessed previously described.25 A periplasmic extract of the highly potent activity. The anti-CD20Fab-LDM provides bacteria was prepared by resuspending the bacteria pellet a promising new solution for anti-CD20 therapies, its in 25 mM Tris (pH 7.5) containing 20% (w/v) sucrose, construction and expression revealed a new technique 200 mM Nacl and 1 mM EDTA, followed by incubation routine for exploring antibody-based and tumor-targeting at 4 1C with gentle shaking for 1 h. After centrifugation fusion protein for cancer therapy. at 12 000 r.p.m. for 15 min, the soluble antibodies were purified from the supernatant by a protein G tag affinity chromatography using the RPAS Purification Module Materials and methods (Amersham Bioscience). The endotoxin level of the Cell lines purified antibody preparations was assayed routinely The B-cell lymphoma line, Raji, Daudi and chronic before their use in any in vitro and in vivo studies. myelogenous leukemia cell line K562 were cultured in RPMI-1640 supplemented with 10% fetal calf serum at SDS polyacrylamide gel electrophoresis and western

37 1C in humidified 5% CO2 atmosphere. blot analysis of the fusion protein The E. coli periplasmic extract and the purified fusion Isolation of primary patient-derived malignant B cells protein preparation were electrophoresed in a 12% Blood were obtained from four patients with B-lymphoid polyacrylamide gel and visualized by staining with malignancies (three newly diagnosed and one relapsed) Coomassie brilliant blue. In addition, an identical gel granting informed consent as approved by the Institute was transferred onto polyvinylidene difluoride (PVDF) of Hematology and Hospital of Blood Diseases, Chinese membranes and subjected to western blot analysis using Academy of Medical Sciences and Peking Union Medical a mouse anti-LDP antibody (3H8) and a rabbit anti- College Institutional Review Board. Mononuclear mouse IgG antibody conjugated with horseradish per- leukocytes were separated with Ficoll from patient oxidase (Bejing Zhongshan Golden Bridge Biotechnology peripheral blood and were analyzed by incubating with Co. Ltd, Beijing, China). All signals were detected by FITC-conjugated anti-CD19 antibody, phycoerythrin using enhanced chemiluminescence (Amersham -conjugated anti-CD20 antibody and FITC-/phycoerythrin- Bioscience). conjugated polyclonal human IgG as a nonbinding control (BD Biosciences, San Jose, CA, USA). Only In vitro CD20-specific binding of the fusion protein CD20-positive cells in the CD19-positive cells accounted Anti-CD20Fab-LDP and anti-CD20Fab were labeled with for more than 95%, CD19-positive cells were sorted by FITC (EZ-labeled FITC protein labeling kit, Pierce BD FACSAriaII Cell-Sorting System u sing a FITC- Biotech, Rockford, IL, USA). CD20-specific bindings of conjugated anti-CD19 antibody. Incubations were carried the fusion protein were analyzed by flow cytometry. out for 0.5 h at room temperature and were followed by Briefly, about 1 Â10 6 Raji lymphoma cells were incu- washing twice with serum-free medium. bated with varying concentrations of FITC-conjugated fusion protein, anti-CD20Fab or polyclonal human IgG Animals as a nonbinding control in a final volume of 0.2 ml. Female nude mice (5–6 weeks old) (BALB/C, nu/nu, Incubation was continued for an additional 60 min on body weight 16–18 g) were obtained from the Vital River ice. After two washes with serum-free medium, the cells Laboratories (Beijing, China). were analyzed by flow cytometry.

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1242 In vivo targeting and accumulation of the fusion protein previous manner, when tumors reached 450–500 mm3 Female, 5- to 6-week-old nude mice were exposed to total (large xenografts). In both experiments, mice were body irradiation (400 rad) to further suppress their continuously monitored for changes in behavior, and residual immune system and facilitate the establishment the body weights and tumor volumes were measured at of xenografts. After 12 h, the mice were injected least every 3 days. Tumors’ mass were calculated with subcutaneously with 2  107 Raji or K562 cells suspended the following formula: V ¼ 0.5a  b2, where a and b are in PBS in the right flank. Mice with staged tumors (0.1 g) the long diameter and its perpendicular short diameter were administered i.v. with FITC-labeled anti-CD20Fab- of the tumor, respectively. The data are presented as LDP, anti-CD20Fab or anti-LDP (16 nmol per mouse).26 the mean s.d. Tumor growth curves were plotted and The mice were imaged using the previous procedure the inhibitory rates of tumor growth were calculated to monitor the distributions of the fusion protein at according to the tumor volume. Volume ratio was different time points.27 calculated according to the following formula: volume ratio (T/C) ¼ Vtreatment/Vcontrol  100%). Student’s t-test Preparation of energized fusion protein was used to determine statistically with significant anti-CD20Fab-LDM differences. Po0.05 was considered significant. LDM (10 mg) with high potent activity were suspended in 5 ml methanol and whisked for 5 min at 4 1C, and then Histopathological examination the mixture was placed at 20 1C for 1 h. AE was obtained All the animals that were killed were dissected and from the supernatant of reaction mixture by centrifuga- examined macroscopically for tumors or signs of gross tion at 16 000 r.p.m. for 20 min at 4 1C. The above tissue or organ abnormalities. Histopathological exam- procedure was repeated once to isolate AE completely. ination of organs was conducted on selected animals in For obtaining energized fusion protein anti-CD20Fab- each group. Tissue sections were stained with hema- LDM, AE in methanol was added to anti-CD20Fab-LDP/ toxylin and eosin and examined by investigators with PBS (10 mmol lÀ1, pH 7.0) with the molecular ratio of previous experience in this model. LEICA (Wetzlar, 5:1 and the volume ratio of 1:50, respectively, and then Germany) QWINV3 image analysis system was used to placed at room temperature for 12 h. Purified anti- determine the area of damage lesions in histopatho- CD20Fab-LDM was prepared finally by separating from logical sections. free AE by using a sephadex G-75 column.

In vitro CD20-restricted killing activity of the fusion Conflict of interest protein CD20-positive Daudi, Raji, Ramos cells or leukocytes The authors declare no conflict of interest. from patients with non-Hodgkin’s lymphoma or ALL (for individual characteristics, see Table 1) cells were seeded at 2 Â 104 per well in a 96-well plate, and CD20- Acknowledgements negative K562 cells were used as a nonbinding control. Then different concentrations of anti-CD20Fab-LDM, This work was supported by grants from Natural Science LDM and ADR were added and incubated for 72 h. Foundation of China (Grant numbers 30873091 and The effects on cell growth were examined by 30971291), Natural Science Foundation of Tianjin (Grant (4, 5-dimethylthiazol-2yl)-2,5-diphenyl tetrazolium bromide number. 05YFGZGX02800), National Science and Tech- (MTT) assay. Briefly, 20 ml of MTT solution (5 mg mlÀ1)was nology Major Project (Grant number 2009ZX09103-720) addedtoeachwellandincubatedat371C for 4 h. The and the National High Technology Research and Devel- supernatant was removed and cells were dissolved in opment Program (‘863’Program) of China (Grant number 100 ml of dimethylsulfoxide, and then monitored with 2006AA02A255). a microplate reader at a wavelength of 546 nm. Survival ratio was calculated according to the following formula: survival ratio ¼ (AtestÀAblank)/(AcontrolÀAblank) Â 100%. References

Assessment of antitumor efficacy against 1 van de Loosdrecht AA, Huijgens PC, Ossenkoppele GJ. Anti- subcutaneous xenografts body-targeted therapy: a paradigm of innovative treatment The first experiment was carried out with 5- to 6-week- strategies in indolent and aggressive B-cell non-Hodgkin’s old female nude mice. The mice were exposed to total lymphoma. Curr Opin Hematol 2004; 11: 419–425. body irradiation (400 rad) to further suppress their 2 Bross PF, Beitz J, Chen G, Chen XH, Duffy E, Kieffer L et al. residual immune system and facilitate the establishment Approval summary: gemtuzumab ozogamicin in relapsed acute myeloid leukemia. Clin Cancer Res 2001; 7: 1490–1496. of xenografts. After 12 h, mice were injected subcuta- 3 Sievers EL, Larson RA, Stadtmauer EA, Estey E, Lowenberg B, neously with 2 Â 107 Raji cells suspended in PBS in the 3 Dombret H et al. Efficacy and safety of gemtuzumab ozogamicin right flank. When tumors reached 40–60 mm (develop- in patients with CD33-positive acute myeloid leukemia in first ing xenografts), animals were randomized into six relapse. J Clin Oncol 2001; 19: 3244–3254. À1 groups, and administered i.v. with 4 nmol kg LDM, 4 Hu JL, Xue YC, Xie MY, Zhang R, Otani T, Minami Y et al. A new À1 À1 2 nmol kg LDM, 4 nmol kg anti-CD20Fab-LDM, macromolecular antitumor antibiotic, C-1027. I. Discovery, taxo- À1 À1 2 nmol kg anti-CD20Fab-LDM, 4 nmol kg anti- nomy of producing organism, fermentation and biological CD20Fab and PBS (vehicle), respectively (Q1W Â 2). To activity. J Antibiot (Tokyo) 1988; 41: 1575–1579. further verify the antitumor efficacy of the fusion 5 Otani T, Minami Y, Marunaka T, Zhang R, Xie MY. A new protein, a second experiment was carried out in the macromolecular antitumor antibiotic, C-1027. II. Isolation

Gene Therapy Anti-CD20Fab-LDM inhibited B-cell lymphoma xenografts C Xin et al 1243 and physico-chemical properties. J Antibiot (Tokyo) 1988; 41: 17 Dillman RO. Infusion reactions associated with the therapeutic 1580–1585. use of monoclonal antibodies in the treatment of malignancy. 6 Zhen YS, Ming XY, Yu B, Otani T, Saito H, Yamada Y. A new Cancer Metastasis Rev 1999; 18: 465–471. macromolecular antitumor antibiotic, C-1027. III. Antitumor 18 Williams ME, Densmore JJ, Pawluczkowycz AW, Beum PV, activity. J Antibiot (Tokyo) 1989; 42: 1294–1298. Kennedy AD, Lindorfer MA et al. Thrice-weekly low-dose 7 Xu YJ, Li DD, Zhen YS. Mode of action of C-1027, a new rituximab decreases CD20 loss via shaving and promotes macromolecular antitumor antibiotic with highly potent cyto- enhanced targeting in chronic lymphocytic leukemia. J Immunol toxicity, on human hepatoma BEL-7402 cells. Cancer Chemother 2006; 177: 7435–7443. Pharmacol 1990; 27: 41–46. 19 Li Y, Williams ME, Cousar JB, Pawluczkowycz AW, Lindorfer 8 Huang YH, Shang BY, Zhen YS. Antitumor efficacy of lidamycin MA, Taylor RP. Rituximab-CD20 complexes are shaved from on hepatoma and active moiety of its molecule. World J Z138 mantle cell lymphoma cells in intravenous and subcuta- Gastroenterol 2005; 11: 3980–3984. neous SCID mouse models. J Immunol 2007; 179: 4263–4271. 9 Xu YJ, Zhen YS, Goldberg IH. C1027 chromophore, a potent 20 Beum PV, Kennedy AD, Williams ME, Lindorfer MA, Taylor RP. new enediyne antitumor antibiotic, induces sequence-specific The shaving reaction: rituximab/CD20 complexes are removed double-strand DNA cleavage. Biochemistry 1994; 33: 5947–5954. from mantle cell lymphoma and chronic lymphocytic leukemia 10 McHugh MM, Gawron LS, Matsui S, Beerman TA. The anti- cells by THP-1 monocytes. J Immunol 2006; 176: 2600–2609. tumor enediyne C-1027 alters progression and induces 21 Smith MR, Jin F, Joshi I. Enhanced efficacy of therapy with chromosomal aberrations and telomere dysfunction. Cancer Res antisense BCL-2 oligonucleotides plus anti-CD20 monoclonal 2005; 65: 5344–5351. antibody in scid mouse/human lymphoma xenografts. Mol 11 Jiang B, Li DD, Zhen YS. Induction of apoptosis by enediyne Cancer Ther 2004; 3: 1693–1699. antitumor antibiotic C1027 in HL-60 human promyelocytic 22 Pro B, Leber B, Smith M, Fayad L, Romaguera J, Hagemeister F leukemia cells. Biochem Biophys Res Commun 1995; 208: 238–244. et al. Phase II multicenter study of oblimersen sodium, a Bcl-2 12 Lai ZZ, Xiong DS, Fan DM, Xu YF, Liu HZ, Peng H et al. Efficacy antisense oligonucleotide, in combination with rituximab in of anti-CD20 chimeric Fab0 fragment on proliferation of B patients with recurrent B-cell non-Hodgkin’s lymphoma. Br J lymphoma cells. Chin Sci Bull 2001; 46:5. Haematol 2008; 143: 355–360. 13 Sugimoto Y, Otani T, Oie S, Wierzba K, Yamada Y. Mechanism of 23 Liu HZ, Li DD, Zhen YS, Shao RG. Potentiation and mechanism action of a new macromolecular antitumor antibiotic, C-1027. of -induced apoptosis by lidamycin in human hepatoma J Antibiot (Tokyo) 1990; 43: 417–421. BEL-7402 cells]. Yao Xue Xue Bao 2003; 38: 250–254. 14 Fischer L, Penack O, Gentilini C, Nogai A, Muessig A, 24 Tanaka T, Fukuda-Ishisaka S, Hirama M, Otani T. Solution Thiel E et al. The anti-lymphoma effect of antibody-mediated structures of C-1027 apoprotein and its complex with the immunotherapy is based on an increased degranulation of aromatized chromophore. J Mol Biol 2001; 309: 267–283. peripheral blood natural killer (NK) cells. Exp Hematol 2006; 34: 25 Zhu Z, Rockwell P, Lu D, Kotanides H, Pytowski B, Hicklin DJ 753–759. et al. Inhibition of vascular endothelial growth factor-induced 15 Berdeja JG, Hess A, Lucas DM, O0Donnell P, Ambinder RF, receptor activation with anti-kinase insert domain-containing Diehl LF et al. Systemic interleukin-2 and adoptive transfer of receptor single-chain antibodies from a phage display library. lymphokine-activated killer cells improves antibody-dependent Cancer Res 1998; 58: 3209–3214. cellular cytotoxicity in patients with relapsed B-cell lymphoma 26 Qiu XQ, Wang H, Cai B, Wang LL, Yue ST. Small antibody treated with rituximab. Clin Cancer Res 2007; 13: 2392–2399. mimetics comprising two complementarity-determining regions 16 Kennedy AD, Beum PV, Solga MD, DiLillo DJ, Lindorfer MA, and a framework region for tumor targeting. Nat Biotechnol 2007; Hess CE et al. Rituximab infusion promotes rapid complement 25: 921–929. depletion and acute CD20 loss in chronic lymphocytic leukemia. 27 Frangioni JV. In vivo near-infrared fluorescence imaging. Curr J Immunol 2004; 172: 3280–3288. Opin Chem Biol 2003; 7: 626–634.

Gene Therapy