G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
Journal of Biotechnology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Journal of Biotechnology journal homepage: www.elsevier.com/locate/jbiotec
In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris
1
2
Q1
3 4 5
Sujatha Gomathinayagam a , Drake Laface b , Nga Rewa Houston-Cummings a , Ruban Mangadu b , Renee Moore a , Ishaan Shandil a , Nathan Sharkey a , Huijuan Li a , Terrance A. Stadheim a , Dongxing Zha a,∗ a
6
b
7
GlycoFi Inc., A Wholly-Owned Subsidiary of Merck & Co Inc., 16 Cavendish Court, Lebanon, NH 03766, United States Biologics Discovery, Palo Alto, Merck Research Laboratories, 901 California Avenue, Palo Alto, CA 94304, United States
8
9 26
a r t i c l e
i n f o
a b s t r a c t
10 11 12 13 14 15
Article history: Received 10 February 2015 Received in revised form 21 April 2015 Accepted 13 May 2015 Available online xxx
16
25
Keywords: Afucosylation Anti-CS1 HuLuc63 ADCC Monoclonal antibodies Multiple myeloma Pichia pastoris N-glycosylation
27
1. Introduction
17 18 19 20 21 22 23 24
Monoclonal antibody (mAb) therapy has been successfully used for the treatment of B-cell lymphomas and is currently extended for the treatment of multiple myeloma (MM). New developments in MM therapeutics have achieved significant survival gains in patients but the disease still remains incurable. Elotuzumab (HuLuc63), an anti-CS1 monoclonal IgG1 antibody, is believed to induce anti-tumor activity and MM cytotoxicity through antibody dependent cellular cytotoxicity (ADCC) and inhibition of MM cell adhesion to bone marrow stromal cells (BMSCs). Modulations of the Fc glycan composition at the N297 site by selective mutations or afucosylation have been explored as strategies to develop bio-better therapeutics with enhanced ADCC activity. Afucosylated therapeutic antibodies with enhanced ADCC activity have been reported to possess greater efficacy in tumor growth inhibition at lower doses when compared to fucosylated therapeutic antibodies. The N-linked glycosylation pathway in Pichia pastoris has been engineered to produce human-like N-linked glycosylation with uniform afucosylated complex type glycans. The purpose of this study was to compare afucosylated anti-CS1 mAb expressed in glycoengineered Pichia pastoris with fucosylated anti-CS1 mAb expressed in mammalian HEK293 cells through in vitro ADCC and in vivo tumor inhibition models. Our results indicate that Fc glycosylation is critical for in vivo efficacy and afucosylated anti-CS1 mAb expressed in glycoengineered Pichia pastoris shows a better in vivo efficacy in tumor regression when compared to fucosylated anti-CS1 mAb expressed in HEK293 cells. Glycoengineered Pichia pastoris could provide an alternative platform for generating homogeneous afucosylated recombinant antibodies where Fc mediated immune effector function is important for efficacy. © 2015 Published by Elsevier B.V.
Q2 28 29 30 31 32 33 34
Multiple myeloma (MM) is characterized by the abnormal expansion of plasma cells within the bone marrow leading to malignancy. Although new developments in the treatment of MM over the past two decades have significantly improved the survival of MM patients (Stewart, 2009), most patients eventually relapse or become refractory to those treatments (Dimopoulos et al., 2007; Richardson, 2005). These challenges illustrate that there exists a
∗ Corresponding author. Current address: The University of Texas, MD Anderson Cancer Center, 1901 East Rd, Unit 1956, Houston, TX 77054-1901, United States. Tel.: +1 713 745 2576. E-mail address: [email protected] (D. Zha).
strong need for the development of new drugs or drug combinations in the management of MM. The onset of monoclonal antibody (mAb) therapies revolutionized the area of B-cell lymphoma treatment. Rituximab, the first approved mAb for cancer therapy has been successfully used for the treatment of B-cell lymphoma (Habermann et al., 2006; Held et al., 2006; Hiddemann et al., 2007). However, when rituximab was used for MM, very minimal response was observed due to the lack of CD20 expression on MM cells in most of the patients (Kapoor et al., 2008). Multiple potential antigen targets are now being evaluated in the context of MM therapy and several of them have successfully reached clinical trials. The potential targets include cell surface receptors, signaling molecules, plasma cell growth factors and cell surface proteins that are highly expressed on the MM cells or on components of the bone marrow microenvironment like bone marrow stromal cells (BMSCs) (de la and Azab, 2013).
http://dx.doi.org/10.1016/j.jbiotec.2015.05.005 0168-1656/© 2015 Published by Elsevier B.V.
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
G Model BIOTEC 7107 1–9 2 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
ARTICLE IN PRESS S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
Anti-CS1 (elotuzumab) (Hsi et al., 2008; Jakubowiak et al., 2012; Lonial and Kaufman, 2012), anti-IL-6 (siltuximab) (Hunsucker et al., 2011; Voorhees et al., 2013), anti-CD38 (daratumumab) (de Weers et al., 2011) and anti-CD138 (BT-062) (Heffner and Lonial, 2003; Ikeda et al., 2009) are the most promising monoclonal antibodies that are under clinical trial. Elotuzumab (HuLuc63) is a humanized IgG1 monoclonal antibody directed against CS1, a cell surface glycoprotein receptor that is highly expressed on MM cells but limited on normal cells (Hsi et al., 2008; Tai et al., 2008). HuLuc63 exhibited potent in vivo and in vitro anti-myeloma activity primarily through antibodydependent cell-mediated cytotoxicity (ADCC) (Tai et al., 2008; Hsi et al., 2008). The ADCC mechanism in anti-tumor therapeutics is believed to be mediated by Natural Killer (NK) cells through the interaction of Fc␥RIIIA on its surface with the Fc binding region (Tai et al., 2008; Hsi et al., 2008; van Rhee et al., 2009). While the antigen specificity of antibodies is determined by the Fab region, the effector functions are determined by the Fc domain. Specifically, the carbohydrate moieties at the conserved N297 site have been shown to have major implications on Fc effector function (Nimmerjahn et al., 2007; Radaev and Sun, 2001; Tao and Morrison, 1989). Several studies have demonstrated that afucosylated IgGs exhibit enhanced ADCC when compared to fucosylated IgGs due to enhanced Fc␥RIIIA binding (Shields et al., 2002; Yamane-Ohnuki et al., 2004; Okazaki et al., 2004; Kanda et al., 2007; Zhang et al., 2011). Antibodies lacking fucose have demonstrated higher therapeutic efficacy at lower doses, compared to fucosylated equivalents (Yamane-Ohnuki et al., 2004; Shinkawa et al., 2003). Crystal structure analysis of afucosylated IgG1s revealed that the enhancement of ADCC is accredited to conformational change in the Fc region (Matsumiya et al., 2007). Antibody engineering to enhance ADCC activity has been a focus of the biopharmaceutical industry. Currently, strategies involving modification of antibodies to enhance Fc␥RIIIA binding either by selective mutations or afucosylation are in progress. The enhancement of ADCC due to afucosylation has been shown to be relevant for all human IgG subclasses (Niwa et al., 2004), scFv-Fc proteins (Natsume et al., 2005, 2006) and Fc-fusion proteins (Shoji-Hosaka et al., 2006). Afucosylated antibodies can be produced by inactivation of the fucosylation pathway, which has been demonstrated by fucosyltransferase (FUT8) knockout in Chinese Hamster Ovary (CHO) (Yamane-Ohnuki et al., 2004; Mori et al., 2004). In vitro enzymatic removal of fucose from N-glycans to create defucosylated forms has also been demonstrated (Yazawa et al., 1986). GlycoMab® has developed an innovative technology to glycoengineer antibodies by introducing a bisecting N-acetylglucosamine (GlcNAc) in the carbohydrate of the Fc region that interferes with the core fucose addition and results in the generation of afucosy˜ et al., 1999). lated antibody (Umana Since the enhancement of ADCC caused by the afucosylated antibody is significantly reduced due to the presence of fucosylated counterparts as they compete for antigen binding, it is essential to have a homogeneous afucosylated antibody in order to achieve maximum in vivo efficacy (Yamane-Ohnuki and Satoh, 2009). The key challenges for the next generation of afucosylated therapeutic antibodies include a robust production process and a homogeneous afucosylated glycoform profile (Yamane-Ohnuki and Satoh, 2009). Glycoengineered Pichia pastoris expression system has the capability to overcome both challenges and generate afucosylated antibodies. In glycoengineered Pichia pastoris, the N-glycosylation pathway has been engineered to eliminate fungal type hyper-mannosylation and produce complex biantennary N-linked glycans (Choi et al., 2003; Hamilton et al., 2006; Hamilton and Gerngross, 2007). Since Pichia pastoris inherently lacks the mechanism to produce GDP-fucose, the glycoengineered strain is capable of producing
antibodies without fucosylation. Li et al. (2006) demonstrated the production of recombinant antibody in glycoengineered Pichia pastoris with homogeneous complex glycoforms without fucose. Subsequently, Potgieter et al. (2009) demonstrated that glycoengineered Pichia pastoris can produce functional monoclonal antibody with a titer greater than 1 g/L in a robust and scalable process while maintaining N-glycan homogeneity. In this study, we compared afucosylated and aglycosylated anti-CS1 mAbs expressed in glycoengineered Pichia pastoris with fucosylated anti-CS1 mAb expressed in HEK 293 cells, to evaluate the effect of fucose on in vitro ADCC and in vivo tumor inhibition in OPM-2 xenograft tumor model. Our results demonstrate that glycan composition plays an important role in the in vivo anti-tumor efficacy. Afucosylated anti-CS1 mAb produced in glycoengineered Pichia pastoris showed greater in vivo anti-tumor efficacy with enhanced in vitro ADCC activity when compared to conventional fucosylated anti-CS1 mAb generated from HEK293 cells.
117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
2. Materials and methods
134
2.1. Strain construction and anti-CS1 antibodies generation
135
Afucosylated and aglycosylated anti-CS1 monoclonal IgG1 antibodies were produced using glycoengineered Pichia pastoris and fucosylated anti-CS1 monoclonal IgG1 antibody was produced in mammalian host HEK293 cells. The amino acid sequence for all forms of anti-CS1 monoclonal antibody are the same as elotuzumab (sequence source – EMBL-EBI database Compound ID: CHEMBL1743010) except to generate a variant of anti-CS1 mAb without Fc receptor function, the conserved N-glycosylation site Asn at position 297 (EU numbering system) was mutated to Ala to avoid N-glycosylation. The variable regions of heavy and light chain of anti-CS1 mAb were codon optimized using Pichia pastoris preferred codon usage, synthesized by GeneArt® and cloned into Pichia antibody expression vector which contains constant regions of heavy and light chains. Heavy and light chain transcription was regulated under methanol inducible promoter AOX1. The final expression vectors were pGLY8040 for afucosylated anti-CS1 mAb and pGLY8041 for aglycosylated anti-CS1 mAb. The expression vectors pGLY8040 and pGLY8041 were transformed into YGLY17108 which is a glycoengineered Pichia pastoris host strain capable of adding terminal galactose on bi-antennary N-linked glycans using linearized plasmids which are able to integrate into TRP2 locus. HEK293 cells produced, fucosylated anti-CS1 mAb was outsourced to Sino Biological Inc. 2.2. Fermentation and purification Fermentation was performed with the following modifications in 15 L glass bioreactors (Applikon, Foster City, CA) using the protocols described earlier (Potgieter et al., 2009). Peptone was replaced with soytone and sorbitol was replaced with maltitol. Methanol induction was carried out in oxygen limited conditions where dissolved oxygen (DO) control was turned-off and agitation speed was fixed at a pre-determined value to achieve a desired oxygen transfer rate. Antibody purification was conducted using the protocols described earlier (Li et al., 2006; Potgieter et al., 2009) with modifications. We employed MabSelectTM resin from GE Healthcare Life Sciences (Cat. No. 17-5199-03) for the initial affinity chromatographic step. Cation Exchange chromatography utilizing ProResTM -S media from Millipore Corporation (Cat. No. 776981729) was used as the polishing step for afucosylated anti-CS1 mAb. For the aglycosylated anti-CS1 mAb, the polishing step was performed using mix-mode cation exchange chromatography CaptoTM MMC media from GE healthcare Life Sciences
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
159
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176
G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx 177 178 179
180 181
(Cat. No.17-5317-03). All anti-CS1 mAb forms were formulated into 3% Mannitol, 50 mM Arginine, and 50 mM Histidine at pH 6.0 ± 0.2, sterile filtered through 0.2 m filter, and stored at 4 ◦ C until use. 2.3. Quadrupole time of flight mass spectrometry (Q-TOF) and SDS-PAGE analysis
190
Purified antibody samples were analyzed using an AccurateMass Q-TOF LC/MS 6520 (Agilent technologies, Santa Clara, CA) as described previously (Choi et al., 2012). In short, the antiCS1 mAb samples were reduced with TCEP at 37 ◦ C for 30 min and analyzed by LC/MS. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was conducted on 4–20% precast Ready Gel® with Tris–HCl (BIO-RAD, Hercules, CA, USA) according to Laemmli method. Protein bands were visualized by Coomassie Brilliant Blue staining and destained in 10% acetic acid (v/v).
191
2.4. N-Glycan normal phase HPLC analysis
182 183 184 185 186 187 188 189
192 193 194 195
196 197
The N-linked glycans from the anti-CS1 mAb samples were released by PNGaseF digest, labeled with 2-aminobenzamide (2AB), and analyzed by high performance liquid chromatography (HPLC) as described previously (Hamilton et al., 2006). 2.5. Binding kinetic constants with CS-1 extracellular domain protein
213
CS-1 extracellular domain (ECD) protein (GenBank: CAB76561.1) with C-terminal Hexa histidine tag was produced in HEK293 cells using human sequence. A Series S CM5 Chip (GE Healthcare) was immobilized via amine coupling kit (GE Healthcare) to >10,000 RU with an anti-human Fc capture antibody kit (GE Healthcare). Anti-CS1 mAbs were captured to ∼25–40RU on the active flowcells (Fc-2, 3, 4) and the reference flowcell (Fc-1) was left blank. Serially diluted human CS1 antigen (ECD), from 80 to 0.625 nM, was injected for 5 min over all of the flowcells and dissociation was monitored for 7 min. Binding data was double referenced by subtracting the reference flowcell signal and a 0 nM CS1 antigen injection. All of the reagents were prepared in 1× HBS-EP+ running buffer (GE Healthcare) and the binding measurements were performed on a Biacore T100 at 25 ◦ C. All data was fitted with 1:1 Binding Model in Biacore T100 Evaluation Software (v2.0.4).
214
2.6. Anti-CS1 antibodies binding to intact cells expressing CS1
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
219
Relative target binding by various forms of anti-CS1 antibodies to CS-1 antigens expressed on human MM cells OPM-2 purchased from DSMZ (Braunschweig, Germany) was conducted by flow cytometry analysis using the method described earlier (Zhang et al., 2011).
220
2.7. Determination of Fc gamma receptor binding
215 216 217 218
221 222 223 224 225 226 227 228 229 230 231 232
A Series S CM5 chip (GE Healthcare) was immobilized via amine coupling (GE Healthcare) with recombinant Protein A/G (Pierce/Thermo Scientific) to ∼4000 RU as a capture surface. Extracellular domain proteins of human Fc␥RIIA (131R and 131H), Fc␥RIIIA (158F and 158 V) and Fc␥RIIB were expressed in CHO-GS production cell line with C-terminal 9 His tag and affinity purified before being used in the assay. Mouse Fc␥RIV produced from NSOderived mouse myeloma cell line was purchased from R&D system (Cat. No. 1974-CD). Testing antibodies were diluted to 10 g/ml and captured to the protein A/G surface for 60 s at 10 l/min serially diluted Fc␥Rs (1600 nM–0 nM) were injected for 60 s at 30 l/min and dissociation was monitored for 300 s. The running/dilution
3
buffer for this assay was 1× HBS-EP+ (GE Healthcare) and the binding measurements were performed at 25 ◦ C on a Biacore 4000 (GE Healthcare). The measured binding was double referenced by subtracting the reference spot signal and a 0 nM Fc␥R injection in each flowcell. Referenced binding data was fitted to the steady state affinity model and equilibrium constants (KD ) were calculated using the Biacore 4000 Evaluation Software. 2.8. Determination of antibody dependent cell-mediated cytotoxicity (ADCC)
233 234 235 236 237 238 239
240 241
Antibody dependent cell-mediated cytotoxicity assay was performed as described previously (Zhang et al., 2011) with modifications. In brief, ADCC activities were assayed with the human MM cells OPM-2 (from DSMZ, Braunschweig, Germany) as target cells. Primary human-enriched NK cells as effector cells were isolated from human PBMC using RosetteSep Human NK cell enrichment cocktail kits (Stemcell Technologies, # 15065). OPM-2 target cells were labeled with 100 Ci 51 Cr and plated at 1.6 × 105 cells/ml, and then the target cells were treated with various anti-CS1 antibody concentrations 10 g/ml, 1 g/ml, 0.1 g/ml, 0.01 g/ml and
Table 1 Study groups and dose of in vivo anti-tumor efficacy in SCID mice. Group
N (mice)
Treatment
Dose
1 2 3 4 5 6 7
10 10 10 10 10 10 10
PBS (diluent) Aglycocylated anti-CS1 mAb, Pichia Afucosylated anti-CS1 mAb, Pichia Fucosylated anti-CS1, Hek293 Aglycosylated anti-CS1 mAb, Pichia Afucosylated anti-CS1 mAb, Pichia Fucosylated anti-CS1, Hek293
SQ 10 mg/kg 10 mg/kg 10 mg/kg 1 mg/kg 1 mg/kg 1 mg/kg
Fig. 1. SDS-PAGE analysis of fucosylated, afucosylated and aglycosylated anti-CS1 monoclonal IgG1 antibodies stained with Coomassie Blue. Lane 1, 5 and 9 are molecular weight markers. Lane 2 and 6 are HEK293 cells produced fucosylated anti-CS1 monoclonal antibody under non-reducing and reducing conditions respectively. Lane 3 and 7 are glycoengineered Pichia pastoris produced afucosylated anti-CS1 monoclonal antibody under non-reducing and reducing conditions respectively. Lane 4 and 8 are glycoengineered Pichia pastoris produced aglycosylated anti-CS1 monoclonal antibody under non-reducing and reducing conditions respectively.
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
242 243 244 245 246 247 248 249 250 251
G Model BIOTEC 7107 1–9 4
ARTICLE IN PRESS S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
Fig. 2. Q-TOF analysis of anti-CS1 monoclonal antibodies under reducing condition. A. Comparison of anti-CS1 mAbs deconvoluted masses under reducing condition after the removal of N-linked glycans by PNGaseF digest. (1) Fucosylated anti-CS1 mAb; (2) afucosylated anti-CS1 mAb; (3) aglycosylated anti-CS1 mAb. Aglycosylated anti-CS1 mAb heavy chain mass differs from the other anti-CS1 mAbs due to N297A mutation in the heavy chain. B. Zoomed Q-TOF analysis of anti-CS1 mAbs heavy chain to show mass variation due to glycosylation difference. (4) Fucosylated anti-CS1 mAb heavy chain; (5) afucosylated anti-CS1 mAb heavy chain; (6) aglycosylated anti-CS1 mAb heavy chain. Lc – Light chain; Hc – Heavy chain; G0 – GlcNAc2 Man3 GlcNAc2 ; G1 – GlcNAc2 Man3 GlcNAc2 Gal; G0F – GlcNAc2 Man3 GlcNAc2 Fuc; G1F – GlcNAc2 Man3 GlcNAc2 GalFuc; G2F – GlcNAc2 Man3 GlcNAc2 GalFuc; Hybrid – GlcNAc2 Man5 GlcNAc or GlcNAc2 Man5 GlcNAcGal; O-Mann – O-linked mannose; GlcNAc – N-acetyl-d-glucosamine; Gal – galactose; Fuc – fucose
252 253 254
255
256 257 258 259 260 261 262 263 264
0.001 g/ml. ADCC function was measured in a 6-h 51 Cr release assay using a 15:1 effector:target ratio. Target cells killing was normalized with no antibody control values.
Tumor volumes were measured at days 0, 3, 7, 9, 14 and 17 after antibody treatment.
265 266
3. Results
267
2.9. In vivo anti-tumor efficacy in OPM-2 xenograft model
3.1. Analytical characterization of anti-CS1 mAbs
268
The anti-tumor activity of various glycosylated anti-CS1 antibodies were compared in OPM-2 xenograft tumor model in SCID mice (SCID mice C.B-Igh-1b/GbmsTac-Prkdcscid -Lystbg ; Jackson Laboratory). OPM-2 cells (1 × 107 ) were implanted by subcutaneous route. After about 20 days, when tumor sizes reached ∼80 mm3 , treatment was initiated with three glycosylation variants of antiCS1 antibodies. Dosage was administered at 1 mg/kg or 10 mg/kg by subcutaneous injection three-times weekly for three weeks (q3wx3). The 7 groups included in this study are listed in Table 1.
Afucosylated and aglycosylated anti-CS1 mAbs expressed in a glycoengineered Pichia pastoris GS5.0 host strain that is capable of adding -1,4 galactose onto bi-antennary N-linked glycans without any fucose addition, and fucosylated anti-CS1 mAb expressed in mammalian host HEK293 cells were subjected to analytical characterization to determine the purity and N-linked glycan distribution. The anti-CS1 mAbs analyzed by SDS-PAGE under non-reducing condition confirmed that they are completely assembled antibodies (>99%) and comparable in purity (Fig. 1).
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
269 270 271 272 273 274 275 276 277
G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
5
Fig. 3. N-linked glycan HPLC analysis of anti-CS1 monoclonal antibodies. N-linked glycan was released from the anti-CS1 mAb by PNGaseF enzyme digest and subsequently quantitated by 2-aminobenzamide (2-AB) labeling and analyzed by HPLC. (A) Aglycosylated anti-CS1 mAb, (B) afucosylated anti-CS1 mAb, (C) fucosylated anti-CS1 mAb.
278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
The Q-TOF analysis of anti-CS1 mAbs was carried out under reducing condition after PNGaseF digestion for comparison (Fig. 2A). The amino acid sequences for all anti-CS1 mAbs were identical to elotuzumab except the aglycosylated anti-CS1 mAb as the conserved N-glycosylation site Asparagine (N) residue at position 297 in the Fc region was mutated to Alanine (A) to preclude N-glycosylation. The light chain masses for all three anti-CS1 mAbs matched to the expected theoretical mass. The mass difference observed on the heavy chain of aglycosylated anti-CS1 mAb when compared to the fucosylated and afucosylated anti-CS1 mAbs confirmed the N297A mutation. Q-TOF analysis was also performed under reducing condition without PNGaseF digestion to compare the N-glycosylation divergence (Fig. 2B). The heavy chain of HEK293 cells produced antiCS1 mAb predominantly showed GlcNAc2 Man3 GlcNAc2 GalFuc (G1F) and GlcNAc2 Man3 GlcNAc2 Fuc (G0F) glycoforms and a lesser extent of GlcNAc2 Man3 GlcNAc2 Gal2 Fuc (G2F). Fucose addition was observed in a majority of glycoforms, regardless of the disparity in the amount of terminal galactose. Glycoengineered Pichia pastoris produced glycosylated anti-CS1 mAb predominantly showed GlcNAc2 Man3 GlcNAc2 (G0) followed by GlcNAc2 Man3 GlcNAc2 Gal (G1) and GlcNAc2 Man5 GlcNAc or
GlcNAc2 Man5 GlcNAcGal (hybrids) containing glycoforms without any fucose addition. The aglycosylated anti-CS1 mAb did not show any addition of N-linked glycans as expected, but O-mannosylation was observed. Results of HPLC analysis of N-linked glycans released from antiCS1 mAbs are given in Fig. 3 and Table 2. The N-linked glycan composition analysis by 2AB labeling reconfirmed that the HEK293 cells produced anti-CS1 mAb was comprised of mainly fucose containing glycoforms (G0F, G1F, G2F) whereas the anti-CS1 mAb produced in glycoengineered Pichia pastoris contained 100% afucosylated glycoforms including G0, G1, G2, Man5 and hybrids. The aglycosylated anti-CS1 mAb did not show any detectable levels of N-linked glycans. 3.2. Anti-CS1 antibodies show similar antigen binding kinetics Antigen binding affinities of anti-CS1 mAbs were compared in OPM-2 (human, peripheral blood, MM) cells expressing antigen CS1 and recombinant human CS1 protein. The anti-CS1 mAbs binding to the OPM-2 cell surface revealed that both anti-CS1 mAbs produced in glycoengineered Pichia pastoris and mammalian HEK 293 cells showed comparable binding (Fig. 4). The binding kinetics of
Table 2 N-linked glycan composition of anti-CS1 mAbs.
Fucosylated anti-CS1 mAb Afucosylated anti-CS1 mAb Aglycosylated anti-CS1 mAb
G0
G0F
G1
G1F
G2
G2F
Man5
Hybrid
ND 61.5 ND
36 ND ND
ND 21.9 ND
47 ND ND
ND 1.2 ND
14 ND ND
1 8.8 ND
2 6.6 ND
N-linked glycans were released from the anti-CS1 mAbs and quantitated by 2-aminobenzamide (2-AB) labeling and analyzed by HPLC. The numbers are percentage reflecting relative proportion of different glycoforms. G0, GlcNAc2Man3GlcNAc2; G1, GlcNAc2Man3GlcNAc2Gal; G2, GlcNAc2Man3GlcNAc2Gal2; G0F, GlcNAc2Man3GlcNAc2Fuc; G1F, GlcNAc2Man3GlcNAc2GalFuc; G2F, GlcNAc2Man3GlcNAc2GalFuc; Man5, GlcNAc2Man5; Hybrid, GlcNAc2Man5GlcNAc or GlcNAc2Man5GlcNAcGal; Man, mannose; GlcNAc, N-acetyl-d-glucosamine; Gal, galactose; Fuc, fucose. ND, not detected.
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
300 301 302 303 304 305 306 307 308 309 310 311 312
313
314 315 316 317 318 319
G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
6
Fig. 4. Flow cytometry analysis of relative target binding by anti-CS1 monoclonal antibodies to CS-1 (antigen) expressed on OPM-2 (human, peripheral blood, multiple myeloma) cells. (A) Isotype control OPM-2 cells. (B) Fucosylated anti-CS1 mAb binding to CS-1 expressed on OPM-2 cells. (C) Afucosylated anti-CS1 mAb binding to CS-1 expressed on OPM-2 cells. (D) Aglycosylated anti-CS1mAb binding to CS-1 expressed on OPM-2 cells.
Table 3 Rate constants and affinities of anti-CS1 mAb for human CS-1 antigen.
Fucosylated anti-CS1 mAb Afucosylated anti-CS1 mAb Aglycosylated anti-CS1 mAb
Ka (1/M s) Association rate
Kd (1/s) Dissociation rate
KD (nM) Dissociation constant
4.17E+05
4.95E−04
1.19
4.11E+05
3.92E−04
0.95
4.34E+05
4.17E−04
0.96
Kinetic rates were determined by 1:1 binding model in BIAcore T100 Evaluation Software (v2.0.4) using at least 5 point concentration range sensograms.
328
anti-CS1 mAbs to human antigen CS1 protein was evaluated on Biacore T100. Kinetic calculations revealed that there was no significant difference between the anti-CS1 mAbs in terms of association (Ka ) and dissociation (Kd ) rates (Table 3). The afucosylated and aglycosylated anti-CS1 mAbs produced in glycoengineered Pichia pastoris showed similar kinetic binding dissociation constants as HEK293 cells produced fucosylated anti-CS1 mAb. Since the Fab region is responsible for antigen binding, the glycosylation variation observed in the Fc region did not alter antigen binding.
329
3.3. Fc receptors binding and its effect on ADCC
320 321 322 323 324 325 326 327
330 331 332 333 334 335 336
Binding affinities of human Fc␥ receptors RIIA (H131/R131), RIIB and RIIIA (F158/V158) including polymorphism were tested for anti-CS1 mAbs (Table 4). Mouse Fc␥ receptor RIV which has high similarity to Fc␥RIIIA was also included in this study. Due to lack of glycosylation at the Fc region, the aglycosylated anti-CS1 mAb completely abolished binding to both variants of human Fc␥RIIA (H/R) and human Fc␥RIIIA (F/V). Aglycosylated anti-CS1 mAb also
Table 4 Comparison of Fc␥Rs binding affinities to anti-CS1 mAb. Fucosylated anti-CS1 mAb KD , nM hFc␥RIIA (H) hFc␥RIIA® hFc␥RIIB hFc␥RIIIA (F) hFc␥RIIIA (V) mFc␥RIV
202 226 937 481 240 68
± ± ± ± ± ±
6 9 23 46 4 7
Afucosylated anti-CS1 mAb KD , nM 299 207 786 20 4.32 25
± ± ± ± ± ±
10 8 20 3 0.6* 4
Aglycosylated anti-CS1 mAb KD , nM
ARTICLE IN PRESS
BIOTEC 7107 1–9
Journal of Biotechnology xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Journal of Biotechnology journal homepage: www.elsevier.com/locate/jbiotec
In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris
1
2
Q1
3 4 5
Sujatha Gomathinayagam a , Drake Laface b , Nga Rewa Houston-Cummings a , Ruban Mangadu b , Renee Moore a , Ishaan Shandil a , Nathan Sharkey a , Huijuan Li a , Terrance A. Stadheim a , Dongxing Zha a,∗ a
6
b
7
GlycoFi Inc., A Wholly-Owned Subsidiary of Merck & Co Inc., 16 Cavendish Court, Lebanon, NH 03766, United States Biologics Discovery, Palo Alto, Merck Research Laboratories, 901 California Avenue, Palo Alto, CA 94304, United States
8
9 26
a r t i c l e
i n f o
a b s t r a c t
10 11 12 13 14 15
Article history: Received 10 February 2015 Received in revised form 21 April 2015 Accepted 13 May 2015 Available online xxx
16
25
Keywords: Afucosylation Anti-CS1 HuLuc63 ADCC Monoclonal antibodies Multiple myeloma Pichia pastoris N-glycosylation
27
1. Introduction
17 18 19 20 21 22 23 24
Monoclonal antibody (mAb) therapy has been successfully used for the treatment of B-cell lymphomas and is currently extended for the treatment of multiple myeloma (MM). New developments in MM therapeutics have achieved significant survival gains in patients but the disease still remains incurable. Elotuzumab (HuLuc63), an anti-CS1 monoclonal IgG1 antibody, is believed to induce anti-tumor activity and MM cytotoxicity through antibody dependent cellular cytotoxicity (ADCC) and inhibition of MM cell adhesion to bone marrow stromal cells (BMSCs). Modulations of the Fc glycan composition at the N297 site by selective mutations or afucosylation have been explored as strategies to develop bio-better therapeutics with enhanced ADCC activity. Afucosylated therapeutic antibodies with enhanced ADCC activity have been reported to possess greater efficacy in tumor growth inhibition at lower doses when compared to fucosylated therapeutic antibodies. The N-linked glycosylation pathway in Pichia pastoris has been engineered to produce human-like N-linked glycosylation with uniform afucosylated complex type glycans. The purpose of this study was to compare afucosylated anti-CS1 mAb expressed in glycoengineered Pichia pastoris with fucosylated anti-CS1 mAb expressed in mammalian HEK293 cells through in vitro ADCC and in vivo tumor inhibition models. Our results indicate that Fc glycosylation is critical for in vivo efficacy and afucosylated anti-CS1 mAb expressed in glycoengineered Pichia pastoris shows a better in vivo efficacy in tumor regression when compared to fucosylated anti-CS1 mAb expressed in HEK293 cells. Glycoengineered Pichia pastoris could provide an alternative platform for generating homogeneous afucosylated recombinant antibodies where Fc mediated immune effector function is important for efficacy. © 2015 Published by Elsevier B.V.
Q2 28 29 30 31 32 33 34
Multiple myeloma (MM) is characterized by the abnormal expansion of plasma cells within the bone marrow leading to malignancy. Although new developments in the treatment of MM over the past two decades have significantly improved the survival of MM patients (Stewart, 2009), most patients eventually relapse or become refractory to those treatments (Dimopoulos et al., 2007; Richardson, 2005). These challenges illustrate that there exists a
∗ Corresponding author. Current address: The University of Texas, MD Anderson Cancer Center, 1901 East Rd, Unit 1956, Houston, TX 77054-1901, United States. Tel.: +1 713 745 2576. E-mail address: [email protected] (D. Zha).
strong need for the development of new drugs or drug combinations in the management of MM. The onset of monoclonal antibody (mAb) therapies revolutionized the area of B-cell lymphoma treatment. Rituximab, the first approved mAb for cancer therapy has been successfully used for the treatment of B-cell lymphoma (Habermann et al., 2006; Held et al., 2006; Hiddemann et al., 2007). However, when rituximab was used for MM, very minimal response was observed due to the lack of CD20 expression on MM cells in most of the patients (Kapoor et al., 2008). Multiple potential antigen targets are now being evaluated in the context of MM therapy and several of them have successfully reached clinical trials. The potential targets include cell surface receptors, signaling molecules, plasma cell growth factors and cell surface proteins that are highly expressed on the MM cells or on components of the bone marrow microenvironment like bone marrow stromal cells (BMSCs) (de la and Azab, 2013).
http://dx.doi.org/10.1016/j.jbiotec.2015.05.005 0168-1656/© 2015 Published by Elsevier B.V.
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
G Model BIOTEC 7107 1–9 2 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
ARTICLE IN PRESS S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
Anti-CS1 (elotuzumab) (Hsi et al., 2008; Jakubowiak et al., 2012; Lonial and Kaufman, 2012), anti-IL-6 (siltuximab) (Hunsucker et al., 2011; Voorhees et al., 2013), anti-CD38 (daratumumab) (de Weers et al., 2011) and anti-CD138 (BT-062) (Heffner and Lonial, 2003; Ikeda et al., 2009) are the most promising monoclonal antibodies that are under clinical trial. Elotuzumab (HuLuc63) is a humanized IgG1 monoclonal antibody directed against CS1, a cell surface glycoprotein receptor that is highly expressed on MM cells but limited on normal cells (Hsi et al., 2008; Tai et al., 2008). HuLuc63 exhibited potent in vivo and in vitro anti-myeloma activity primarily through antibodydependent cell-mediated cytotoxicity (ADCC) (Tai et al., 2008; Hsi et al., 2008). The ADCC mechanism in anti-tumor therapeutics is believed to be mediated by Natural Killer (NK) cells through the interaction of Fc␥RIIIA on its surface with the Fc binding region (Tai et al., 2008; Hsi et al., 2008; van Rhee et al., 2009). While the antigen specificity of antibodies is determined by the Fab region, the effector functions are determined by the Fc domain. Specifically, the carbohydrate moieties at the conserved N297 site have been shown to have major implications on Fc effector function (Nimmerjahn et al., 2007; Radaev and Sun, 2001; Tao and Morrison, 1989). Several studies have demonstrated that afucosylated IgGs exhibit enhanced ADCC when compared to fucosylated IgGs due to enhanced Fc␥RIIIA binding (Shields et al., 2002; Yamane-Ohnuki et al., 2004; Okazaki et al., 2004; Kanda et al., 2007; Zhang et al., 2011). Antibodies lacking fucose have demonstrated higher therapeutic efficacy at lower doses, compared to fucosylated equivalents (Yamane-Ohnuki et al., 2004; Shinkawa et al., 2003). Crystal structure analysis of afucosylated IgG1s revealed that the enhancement of ADCC is accredited to conformational change in the Fc region (Matsumiya et al., 2007). Antibody engineering to enhance ADCC activity has been a focus of the biopharmaceutical industry. Currently, strategies involving modification of antibodies to enhance Fc␥RIIIA binding either by selective mutations or afucosylation are in progress. The enhancement of ADCC due to afucosylation has been shown to be relevant for all human IgG subclasses (Niwa et al., 2004), scFv-Fc proteins (Natsume et al., 2005, 2006) and Fc-fusion proteins (Shoji-Hosaka et al., 2006). Afucosylated antibodies can be produced by inactivation of the fucosylation pathway, which has been demonstrated by fucosyltransferase (FUT8) knockout in Chinese Hamster Ovary (CHO) (Yamane-Ohnuki et al., 2004; Mori et al., 2004). In vitro enzymatic removal of fucose from N-glycans to create defucosylated forms has also been demonstrated (Yazawa et al., 1986). GlycoMab® has developed an innovative technology to glycoengineer antibodies by introducing a bisecting N-acetylglucosamine (GlcNAc) in the carbohydrate of the Fc region that interferes with the core fucose addition and results in the generation of afucosy˜ et al., 1999). lated antibody (Umana Since the enhancement of ADCC caused by the afucosylated antibody is significantly reduced due to the presence of fucosylated counterparts as they compete for antigen binding, it is essential to have a homogeneous afucosylated antibody in order to achieve maximum in vivo efficacy (Yamane-Ohnuki and Satoh, 2009). The key challenges for the next generation of afucosylated therapeutic antibodies include a robust production process and a homogeneous afucosylated glycoform profile (Yamane-Ohnuki and Satoh, 2009). Glycoengineered Pichia pastoris expression system has the capability to overcome both challenges and generate afucosylated antibodies. In glycoengineered Pichia pastoris, the N-glycosylation pathway has been engineered to eliminate fungal type hyper-mannosylation and produce complex biantennary N-linked glycans (Choi et al., 2003; Hamilton et al., 2006; Hamilton and Gerngross, 2007). Since Pichia pastoris inherently lacks the mechanism to produce GDP-fucose, the glycoengineered strain is capable of producing
antibodies without fucosylation. Li et al. (2006) demonstrated the production of recombinant antibody in glycoengineered Pichia pastoris with homogeneous complex glycoforms without fucose. Subsequently, Potgieter et al. (2009) demonstrated that glycoengineered Pichia pastoris can produce functional monoclonal antibody with a titer greater than 1 g/L in a robust and scalable process while maintaining N-glycan homogeneity. In this study, we compared afucosylated and aglycosylated anti-CS1 mAbs expressed in glycoengineered Pichia pastoris with fucosylated anti-CS1 mAb expressed in HEK 293 cells, to evaluate the effect of fucose on in vitro ADCC and in vivo tumor inhibition in OPM-2 xenograft tumor model. Our results demonstrate that glycan composition plays an important role in the in vivo anti-tumor efficacy. Afucosylated anti-CS1 mAb produced in glycoengineered Pichia pastoris showed greater in vivo anti-tumor efficacy with enhanced in vitro ADCC activity when compared to conventional fucosylated anti-CS1 mAb generated from HEK293 cells.
117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
2. Materials and methods
134
2.1. Strain construction and anti-CS1 antibodies generation
135
Afucosylated and aglycosylated anti-CS1 monoclonal IgG1 antibodies were produced using glycoengineered Pichia pastoris and fucosylated anti-CS1 monoclonal IgG1 antibody was produced in mammalian host HEK293 cells. The amino acid sequence for all forms of anti-CS1 monoclonal antibody are the same as elotuzumab (sequence source – EMBL-EBI database Compound ID: CHEMBL1743010) except to generate a variant of anti-CS1 mAb without Fc receptor function, the conserved N-glycosylation site Asn at position 297 (EU numbering system) was mutated to Ala to avoid N-glycosylation. The variable regions of heavy and light chain of anti-CS1 mAb were codon optimized using Pichia pastoris preferred codon usage, synthesized by GeneArt® and cloned into Pichia antibody expression vector which contains constant regions of heavy and light chains. Heavy and light chain transcription was regulated under methanol inducible promoter AOX1. The final expression vectors were pGLY8040 for afucosylated anti-CS1 mAb and pGLY8041 for aglycosylated anti-CS1 mAb. The expression vectors pGLY8040 and pGLY8041 were transformed into YGLY17108 which is a glycoengineered Pichia pastoris host strain capable of adding terminal galactose on bi-antennary N-linked glycans using linearized plasmids which are able to integrate into TRP2 locus. HEK293 cells produced, fucosylated anti-CS1 mAb was outsourced to Sino Biological Inc. 2.2. Fermentation and purification Fermentation was performed with the following modifications in 15 L glass bioreactors (Applikon, Foster City, CA) using the protocols described earlier (Potgieter et al., 2009). Peptone was replaced with soytone and sorbitol was replaced with maltitol. Methanol induction was carried out in oxygen limited conditions where dissolved oxygen (DO) control was turned-off and agitation speed was fixed at a pre-determined value to achieve a desired oxygen transfer rate. Antibody purification was conducted using the protocols described earlier (Li et al., 2006; Potgieter et al., 2009) with modifications. We employed MabSelectTM resin from GE Healthcare Life Sciences (Cat. No. 17-5199-03) for the initial affinity chromatographic step. Cation Exchange chromatography utilizing ProResTM -S media from Millipore Corporation (Cat. No. 776981729) was used as the polishing step for afucosylated anti-CS1 mAb. For the aglycosylated anti-CS1 mAb, the polishing step was performed using mix-mode cation exchange chromatography CaptoTM MMC media from GE healthcare Life Sciences
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
159
160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176
G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx 177 178 179
180 181
(Cat. No.17-5317-03). All anti-CS1 mAb forms were formulated into 3% Mannitol, 50 mM Arginine, and 50 mM Histidine at pH 6.0 ± 0.2, sterile filtered through 0.2 m filter, and stored at 4 ◦ C until use. 2.3. Quadrupole time of flight mass spectrometry (Q-TOF) and SDS-PAGE analysis
190
Purified antibody samples were analyzed using an AccurateMass Q-TOF LC/MS 6520 (Agilent technologies, Santa Clara, CA) as described previously (Choi et al., 2012). In short, the antiCS1 mAb samples were reduced with TCEP at 37 ◦ C for 30 min and analyzed by LC/MS. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was conducted on 4–20% precast Ready Gel® with Tris–HCl (BIO-RAD, Hercules, CA, USA) according to Laemmli method. Protein bands were visualized by Coomassie Brilliant Blue staining and destained in 10% acetic acid (v/v).
191
2.4. N-Glycan normal phase HPLC analysis
182 183 184 185 186 187 188 189
192 193 194 195
196 197
The N-linked glycans from the anti-CS1 mAb samples were released by PNGaseF digest, labeled with 2-aminobenzamide (2AB), and analyzed by high performance liquid chromatography (HPLC) as described previously (Hamilton et al., 2006). 2.5. Binding kinetic constants with CS-1 extracellular domain protein
213
CS-1 extracellular domain (ECD) protein (GenBank: CAB76561.1) with C-terminal Hexa histidine tag was produced in HEK293 cells using human sequence. A Series S CM5 Chip (GE Healthcare) was immobilized via amine coupling kit (GE Healthcare) to >10,000 RU with an anti-human Fc capture antibody kit (GE Healthcare). Anti-CS1 mAbs were captured to ∼25–40RU on the active flowcells (Fc-2, 3, 4) and the reference flowcell (Fc-1) was left blank. Serially diluted human CS1 antigen (ECD), from 80 to 0.625 nM, was injected for 5 min over all of the flowcells and dissociation was monitored for 7 min. Binding data was double referenced by subtracting the reference flowcell signal and a 0 nM CS1 antigen injection. All of the reagents were prepared in 1× HBS-EP+ running buffer (GE Healthcare) and the binding measurements were performed on a Biacore T100 at 25 ◦ C. All data was fitted with 1:1 Binding Model in Biacore T100 Evaluation Software (v2.0.4).
214
2.6. Anti-CS1 antibodies binding to intact cells expressing CS1
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
219
Relative target binding by various forms of anti-CS1 antibodies to CS-1 antigens expressed on human MM cells OPM-2 purchased from DSMZ (Braunschweig, Germany) was conducted by flow cytometry analysis using the method described earlier (Zhang et al., 2011).
220
2.7. Determination of Fc gamma receptor binding
215 216 217 218
221 222 223 224 225 226 227 228 229 230 231 232
A Series S CM5 chip (GE Healthcare) was immobilized via amine coupling (GE Healthcare) with recombinant Protein A/G (Pierce/Thermo Scientific) to ∼4000 RU as a capture surface. Extracellular domain proteins of human Fc␥RIIA (131R and 131H), Fc␥RIIIA (158F and 158 V) and Fc␥RIIB were expressed in CHO-GS production cell line with C-terminal 9 His tag and affinity purified before being used in the assay. Mouse Fc␥RIV produced from NSOderived mouse myeloma cell line was purchased from R&D system (Cat. No. 1974-CD). Testing antibodies were diluted to 10 g/ml and captured to the protein A/G surface for 60 s at 10 l/min serially diluted Fc␥Rs (1600 nM–0 nM) were injected for 60 s at 30 l/min and dissociation was monitored for 300 s. The running/dilution
3
buffer for this assay was 1× HBS-EP+ (GE Healthcare) and the binding measurements were performed at 25 ◦ C on a Biacore 4000 (GE Healthcare). The measured binding was double referenced by subtracting the reference spot signal and a 0 nM Fc␥R injection in each flowcell. Referenced binding data was fitted to the steady state affinity model and equilibrium constants (KD ) were calculated using the Biacore 4000 Evaluation Software. 2.8. Determination of antibody dependent cell-mediated cytotoxicity (ADCC)
233 234 235 236 237 238 239
240 241
Antibody dependent cell-mediated cytotoxicity assay was performed as described previously (Zhang et al., 2011) with modifications. In brief, ADCC activities were assayed with the human MM cells OPM-2 (from DSMZ, Braunschweig, Germany) as target cells. Primary human-enriched NK cells as effector cells were isolated from human PBMC using RosetteSep Human NK cell enrichment cocktail kits (Stemcell Technologies, # 15065). OPM-2 target cells were labeled with 100 Ci 51 Cr and plated at 1.6 × 105 cells/ml, and then the target cells were treated with various anti-CS1 antibody concentrations 10 g/ml, 1 g/ml, 0.1 g/ml, 0.01 g/ml and
Table 1 Study groups and dose of in vivo anti-tumor efficacy in SCID mice. Group
N (mice)
Treatment
Dose
1 2 3 4 5 6 7
10 10 10 10 10 10 10
PBS (diluent) Aglycocylated anti-CS1 mAb, Pichia Afucosylated anti-CS1 mAb, Pichia Fucosylated anti-CS1, Hek293 Aglycosylated anti-CS1 mAb, Pichia Afucosylated anti-CS1 mAb, Pichia Fucosylated anti-CS1, Hek293
SQ 10 mg/kg 10 mg/kg 10 mg/kg 1 mg/kg 1 mg/kg 1 mg/kg
Fig. 1. SDS-PAGE analysis of fucosylated, afucosylated and aglycosylated anti-CS1 monoclonal IgG1 antibodies stained with Coomassie Blue. Lane 1, 5 and 9 are molecular weight markers. Lane 2 and 6 are HEK293 cells produced fucosylated anti-CS1 monoclonal antibody under non-reducing and reducing conditions respectively. Lane 3 and 7 are glycoengineered Pichia pastoris produced afucosylated anti-CS1 monoclonal antibody under non-reducing and reducing conditions respectively. Lane 4 and 8 are glycoengineered Pichia pastoris produced aglycosylated anti-CS1 monoclonal antibody under non-reducing and reducing conditions respectively.
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
242 243 244 245 246 247 248 249 250 251
G Model BIOTEC 7107 1–9 4
ARTICLE IN PRESS S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
Fig. 2. Q-TOF analysis of anti-CS1 monoclonal antibodies under reducing condition. A. Comparison of anti-CS1 mAbs deconvoluted masses under reducing condition after the removal of N-linked glycans by PNGaseF digest. (1) Fucosylated anti-CS1 mAb; (2) afucosylated anti-CS1 mAb; (3) aglycosylated anti-CS1 mAb. Aglycosylated anti-CS1 mAb heavy chain mass differs from the other anti-CS1 mAbs due to N297A mutation in the heavy chain. B. Zoomed Q-TOF analysis of anti-CS1 mAbs heavy chain to show mass variation due to glycosylation difference. (4) Fucosylated anti-CS1 mAb heavy chain; (5) afucosylated anti-CS1 mAb heavy chain; (6) aglycosylated anti-CS1 mAb heavy chain. Lc – Light chain; Hc – Heavy chain; G0 – GlcNAc2 Man3 GlcNAc2 ; G1 – GlcNAc2 Man3 GlcNAc2 Gal; G0F – GlcNAc2 Man3 GlcNAc2 Fuc; G1F – GlcNAc2 Man3 GlcNAc2 GalFuc; G2F – GlcNAc2 Man3 GlcNAc2 GalFuc; Hybrid – GlcNAc2 Man5 GlcNAc or GlcNAc2 Man5 GlcNAcGal; O-Mann – O-linked mannose; GlcNAc – N-acetyl-d-glucosamine; Gal – galactose; Fuc – fucose
252 253 254
255
256 257 258 259 260 261 262 263 264
0.001 g/ml. ADCC function was measured in a 6-h 51 Cr release assay using a 15:1 effector:target ratio. Target cells killing was normalized with no antibody control values.
Tumor volumes were measured at days 0, 3, 7, 9, 14 and 17 after antibody treatment.
265 266
3. Results
267
2.9. In vivo anti-tumor efficacy in OPM-2 xenograft model
3.1. Analytical characterization of anti-CS1 mAbs
268
The anti-tumor activity of various glycosylated anti-CS1 antibodies were compared in OPM-2 xenograft tumor model in SCID mice (SCID mice C.B-Igh-1b/GbmsTac-Prkdcscid -Lystbg ; Jackson Laboratory). OPM-2 cells (1 × 107 ) were implanted by subcutaneous route. After about 20 days, when tumor sizes reached ∼80 mm3 , treatment was initiated with three glycosylation variants of antiCS1 antibodies. Dosage was administered at 1 mg/kg or 10 mg/kg by subcutaneous injection three-times weekly for three weeks (q3wx3). The 7 groups included in this study are listed in Table 1.
Afucosylated and aglycosylated anti-CS1 mAbs expressed in a glycoengineered Pichia pastoris GS5.0 host strain that is capable of adding -1,4 galactose onto bi-antennary N-linked glycans without any fucose addition, and fucosylated anti-CS1 mAb expressed in mammalian host HEK293 cells were subjected to analytical characterization to determine the purity and N-linked glycan distribution. The anti-CS1 mAbs analyzed by SDS-PAGE under non-reducing condition confirmed that they are completely assembled antibodies (>99%) and comparable in purity (Fig. 1).
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
269 270 271 272 273 274 275 276 277
G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
5
Fig. 3. N-linked glycan HPLC analysis of anti-CS1 monoclonal antibodies. N-linked glycan was released from the anti-CS1 mAb by PNGaseF enzyme digest and subsequently quantitated by 2-aminobenzamide (2-AB) labeling and analyzed by HPLC. (A) Aglycosylated anti-CS1 mAb, (B) afucosylated anti-CS1 mAb, (C) fucosylated anti-CS1 mAb.
278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
The Q-TOF analysis of anti-CS1 mAbs was carried out under reducing condition after PNGaseF digestion for comparison (Fig. 2A). The amino acid sequences for all anti-CS1 mAbs were identical to elotuzumab except the aglycosylated anti-CS1 mAb as the conserved N-glycosylation site Asparagine (N) residue at position 297 in the Fc region was mutated to Alanine (A) to preclude N-glycosylation. The light chain masses for all three anti-CS1 mAbs matched to the expected theoretical mass. The mass difference observed on the heavy chain of aglycosylated anti-CS1 mAb when compared to the fucosylated and afucosylated anti-CS1 mAbs confirmed the N297A mutation. Q-TOF analysis was also performed under reducing condition without PNGaseF digestion to compare the N-glycosylation divergence (Fig. 2B). The heavy chain of HEK293 cells produced antiCS1 mAb predominantly showed GlcNAc2 Man3 GlcNAc2 GalFuc (G1F) and GlcNAc2 Man3 GlcNAc2 Fuc (G0F) glycoforms and a lesser extent of GlcNAc2 Man3 GlcNAc2 Gal2 Fuc (G2F). Fucose addition was observed in a majority of glycoforms, regardless of the disparity in the amount of terminal galactose. Glycoengineered Pichia pastoris produced glycosylated anti-CS1 mAb predominantly showed GlcNAc2 Man3 GlcNAc2 (G0) followed by GlcNAc2 Man3 GlcNAc2 Gal (G1) and GlcNAc2 Man5 GlcNAc or
GlcNAc2 Man5 GlcNAcGal (hybrids) containing glycoforms without any fucose addition. The aglycosylated anti-CS1 mAb did not show any addition of N-linked glycans as expected, but O-mannosylation was observed. Results of HPLC analysis of N-linked glycans released from antiCS1 mAbs are given in Fig. 3 and Table 2. The N-linked glycan composition analysis by 2AB labeling reconfirmed that the HEK293 cells produced anti-CS1 mAb was comprised of mainly fucose containing glycoforms (G0F, G1F, G2F) whereas the anti-CS1 mAb produced in glycoengineered Pichia pastoris contained 100% afucosylated glycoforms including G0, G1, G2, Man5 and hybrids. The aglycosylated anti-CS1 mAb did not show any detectable levels of N-linked glycans. 3.2. Anti-CS1 antibodies show similar antigen binding kinetics Antigen binding affinities of anti-CS1 mAbs were compared in OPM-2 (human, peripheral blood, MM) cells expressing antigen CS1 and recombinant human CS1 protein. The anti-CS1 mAbs binding to the OPM-2 cell surface revealed that both anti-CS1 mAbs produced in glycoengineered Pichia pastoris and mammalian HEK 293 cells showed comparable binding (Fig. 4). The binding kinetics of
Table 2 N-linked glycan composition of anti-CS1 mAbs.
Fucosylated anti-CS1 mAb Afucosylated anti-CS1 mAb Aglycosylated anti-CS1 mAb
G0
G0F
G1
G1F
G2
G2F
Man5
Hybrid
ND 61.5 ND
36 ND ND
ND 21.9 ND
47 ND ND
ND 1.2 ND
14 ND ND
1 8.8 ND
2 6.6 ND
N-linked glycans were released from the anti-CS1 mAbs and quantitated by 2-aminobenzamide (2-AB) labeling and analyzed by HPLC. The numbers are percentage reflecting relative proportion of different glycoforms. G0, GlcNAc2Man3GlcNAc2; G1, GlcNAc2Man3GlcNAc2Gal; G2, GlcNAc2Man3GlcNAc2Gal2; G0F, GlcNAc2Man3GlcNAc2Fuc; G1F, GlcNAc2Man3GlcNAc2GalFuc; G2F, GlcNAc2Man3GlcNAc2GalFuc; Man5, GlcNAc2Man5; Hybrid, GlcNAc2Man5GlcNAc or GlcNAc2Man5GlcNAcGal; Man, mannose; GlcNAc, N-acetyl-d-glucosamine; Gal, galactose; Fuc, fucose. ND, not detected.
Please cite this article in press as: Gomathinayagam, S., et al., In vivo anti-tumor efficacy of afucosylated anti-CS1 monoclonal antibody produced in glycoengineered Pichia pastoris. J. Biotechnol. (2015), http://dx.doi.org/10.1016/j.jbiotec.2015.05.005
300 301 302 303 304 305 306 307 308 309 310 311 312
313
314 315 316 317 318 319
G Model
ARTICLE IN PRESS
BIOTEC 7107 1–9
S. Gomathinayagam et al. / Journal of Biotechnology xxx (2015) xxx–xxx
6
Fig. 4. Flow cytometry analysis of relative target binding by anti-CS1 monoclonal antibodies to CS-1 (antigen) expressed on OPM-2 (human, peripheral blood, multiple myeloma) cells. (A) Isotype control OPM-2 cells. (B) Fucosylated anti-CS1 mAb binding to CS-1 expressed on OPM-2 cells. (C) Afucosylated anti-CS1 mAb binding to CS-1 expressed on OPM-2 cells. (D) Aglycosylated anti-CS1mAb binding to CS-1 expressed on OPM-2 cells.
Table 3 Rate constants and affinities of anti-CS1 mAb for human CS-1 antigen.
Fucosylated anti-CS1 mAb Afucosylated anti-CS1 mAb Aglycosylated anti-CS1 mAb
Ka (1/M s) Association rate
Kd (1/s) Dissociation rate
KD (nM) Dissociation constant
4.17E+05
4.95E−04
1.19
4.11E+05
3.92E−04
0.95
4.34E+05
4.17E−04
0.96
Kinetic rates were determined by 1:1 binding model in BIAcore T100 Evaluation Software (v2.0.4) using at least 5 point concentration range sensograms.
328
anti-CS1 mAbs to human antigen CS1 protein was evaluated on Biacore T100. Kinetic calculations revealed that there was no significant difference between the anti-CS1 mAbs in terms of association (Ka ) and dissociation (Kd ) rates (Table 3). The afucosylated and aglycosylated anti-CS1 mAbs produced in glycoengineered Pichia pastoris showed similar kinetic binding dissociation constants as HEK293 cells produced fucosylated anti-CS1 mAb. Since the Fab region is responsible for antigen binding, the glycosylation variation observed in the Fc region did not alter antigen binding.
329
3.3. Fc receptors binding and its effect on ADCC
320 321 322 323 324 325 326 327
330 331 332 333 334 335 336
Binding affinities of human Fc␥ receptors RIIA (H131/R131), RIIB and RIIIA (F158/V158) including polymorphism were tested for anti-CS1 mAbs (Table 4). Mouse Fc␥ receptor RIV which has high similarity to Fc␥RIIIA was also included in this study. Due to lack of glycosylation at the Fc region, the aglycosylated anti-CS1 mAb completely abolished binding to both variants of human Fc␥RIIA (H/R) and human Fc␥RIIIA (F/V). Aglycosylated anti-CS1 mAb also
Table 4 Comparison of Fc␥Rs binding affinities to anti-CS1 mAb. Fucosylated anti-CS1 mAb KD , nM hFc␥RIIA (H) hFc␥RIIA® hFc␥RIIB hFc␥RIIIA (F) hFc␥RIIIA (V) mFc␥RIV
202 226 937 481 240 68
± ± ± ± ± ±
6 9 23 46 4 7
Afucosylated anti-CS1 mAb KD , nM 299 207 786 20 4.32 25
± ± ± ± ± ±
10 8 20 3 0.6* 4
Aglycosylated anti-CS1 mAb KD , nM
