development and preclinical characterization of a humanized … · hamster immunization and splenic...

Cancer Therapy: Preclinical

Development and Preclinical Characterization of aHumanized Antibody Targeting CXCL12

Cuiling Zhong, Jianyong Wang, Bing Li, Hong Xiang, Mark Ultsch, Mary Coons, Terence Wong,Nancy Y. Chiang, Suzy Clark, Robyn Clark, Leah Quintana, Peter Gribling, Eric Suto, Kai Barck,Racquel Corpuz, Jenny Yao, Rashi Takkar, Wyne P. Lee, Lisa A. Damico-Beyer, Richard D. Carano,Camellia Adams, Robert F. Kelley, Weiru Wang, and Napoleone Ferrara

AbstractPurpose: Our goal was to develop a potent humanized antibody against mouse/human CXCL12. This

report summarized its in vitro and in vivo activities.

Experimental Design: Cell surface binding and cell migration assays were used to select neutralizing

hamster antibodies, followed by testing in several animal models. Monoclonal antibody (mAb) 30D8 was

selected for humanization based on its in vitro and in vivo activities.

Results: 30D8, a hamster antibody against mouse and human CXCL12a, CXCL12b, and CXCL12g , wasshown to dose-dependently block CXCL12a binding to CXCR4 and CXCR7, and CXCL12a-induced Jurkat

cell migration in vitro. Inhibition of primary tumor growth and/or metastasis was observed in several

models. 30D8 alone significantly ameliorated arthritis in a mouse collagen-induced arthritis model (CIA).

Combination with a TNF-a antagonist was additive. In addition, 30D8 inhibited 50% of laser-induced

choroidal neovascularization (CNV) in mice. Humanized 30D8 (hu30D8) showed similar in vitro and

in vivo activities as the parental hamster antibody. A crystal structure of the hu30D8 Fab/CXCL12acomplex in combination with mutational analysis revealed a "hot spot" around residues Asn44/Asn45 of

CXCL12a and part of the RFFESH region required for CXCL12a binding to CXCR4 and CXCR7. Finally,

hu30D8 exhibited fast clearance in cynomolgus monkeys but not in rats.

Conclusion: CXCL12 is an attractive target for treatment of cancer and inflammation-related diseases;

hu30D8 is suitable for testing this hypothesis in humans. Clin Cancer Res; 19(16); 4433–45.�2013 AACR.

IntroductionCXCL12, also known as stromal-derived factor 1 (SDF1),

is a CXC chemokine that binds to two G-protein–coupledreceptors, CXCR4 and CXCR7 (reviewed in ref. 1). It parti-cipates in many developmental and physiologic processes,including hematopoiesis and angiogenesis. Gene deletionof CXCL12 or CXCR4 results in embryonic lethality atE18.5, with severe developmental defects affecting centralnervous system, bone marrow, heart, and vascular system(2, 3). CXCR7 inactivation, on the other hand, seems toaffect primarily cardiac development, although defects inneuronal migration have also been noted (4). CXCL12a,the major form present in mice and humans, is highly

conserved, as the mouse and human forms differ onlyin one amino acid (1). CXCL12a is produced by numerouscell types includingmacrophages, lymphocytes, endothelialcells, and fibroblasts. CXCR4 activates multiple down-stream targets, including extracellular signal–regulatedkinase (ERK)1/2, AKT, and small GTPases of the Rho family(5). CXCR7, on the other hand, seems to function at least inpart by sequestering CXCL12 (4, 6).

TheCXCL12/CXCR4/CXCR7pathway has also generatedconsiderable interest as a potential therapeutic target givenits role in tumor growth, survival, and angiogenesis (1, 7).CXCL12 levels have been shown to be important in dictat-ing organ-specific metastasis of several cancers (8). It hasbeen previously reported that CXCL12 autocrine signalingloops, together with TGF-b, initiate and maintain the dif-ferentiation of fibroblasts to myofibroblasts and the con-current mammary tumor-promoting phenotype (9, 10).CXCL12 induces angiogenesis, both in vitro and in vivo,synergistically with VEGF (11). VEGF, in turn, can regulateexpression of CXCR4 in vascular endothelial cells, hemato-poietic stem cells, and promote invasion in an autocrinemanner by regulating CXCR4 on cancer cells (12, 13). Inrecent years, it has become clear that CXCR4þ myeloidcells, recruited from the bone marrow into the tumor

Authors' Affiliation: Genentech, Inc., South San Francisco, California

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Corresponding Author: Napoleone Ferrara, Department of Pathology &Moores Cancer Center, University of California San Diego, 3855 HealthSciences Drive, Mail Code # 0819, La Jolla, CA 92093. Phone: 858-822-6822; Fax: 858-5342157; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-13-0943

�2013 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 4433

on August 29, 2021. © 2013 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst June 28, 2013; DOI: 10.1158/1078-0432.CCR-13-0943

http://clincancerres.aacrjournals.org/

microenvironment, participate in tumor angiogenesis (14,15).Growing evidence also suggests that CXCR7 is involvedin several aspects of tumorigenesis and could become animportant target for new antimetastatic and anticancerdrugs (16). Recent preclinical studies suggest that small-molecule inhibitors that target the tyrosine kinase func-tion of VEGF receptors result in increases in serumCXCL12 levels, which might relate at least in part to theirability to promote a more invasive tumor phenotype,when administered at high doses (17, 18). Induction ofCXCL12 has also been shown to cause development ofresistance and to favor recurrences upon radiation (19).

Here, we describe a high-affinity anti-CXCL12 antibodythat shows antitumor effects inmultiplemouse and humantumor models, as a single agent or in combination withanti-VEGF antibodies. Our data further validate CXCL12 asa target not only in cancer but also in several inflammation/angiogenesis–related disease models.

Materials and MethodsCell lines

Jurkat, EL4, LL/2 (LLC1), 4T1, BxPC3, KMS11, Calu6,A673, andOvcar-3 tumor cellswere from theAmericanTypeCulture Collection (ATCC). HM7 cells were originally fromDr. Young Kim (University of California, San Francisco, SanFrancisco, CA). Jurkat cells were cultured in RPMI mediacontaining 10% FBS (Sigma). 4T1 cells were cultured inIscove’s modified Dulbecco’s medium supplemented with10% FBS. The remaining cell lines were maintained in highglucose Dulbecco’s modified Eagle medium (DMEM; Invi-trogen) containing 10% FBS, with the exception of EL4 cellswhich were cultured in DMEM with 10% horse serum.

Hamster immunization and splenic fusionFemale Armenian hamsters were hyperimmunized

intraperitoneally, twice per week for a total of 12 to 13boosts, with 2 mg/animal recombinant human CXCL12a(R&D Systems). Three days after the final boost, spleno-cytes were fused with cells derived from the murinemyeloma cell line P3�63AgU.1 (ATCC). Hybridomaswere selected using methylcellulose-based hypoxan-

thine–aminopterin–thymidine (HAT) medium (STEM-CELL Technologies). Ten to 14 days after fusion, culturesupernatants were collected and screened by direct ELISAas previously described (20). Positive cell lines weresubcloned four to six times by limiting dilutions. Thosethat kept showing consistent activity were scaled up andpurified by Protein A affinity chromatography. The eluatewas dialyzed against PBS and concentrated. For large-scale antibody purification, HiTrap (Amersham) proteinA column chromatography was used.

Generation of 293-CXCR4 and 293-CXCR7 cells293 cells were cotransfected with CXCR7 in pCMV-XL or

CXCR4 in PRK5 with P8vE-Neo vectors. Cells were culturedin F12:DMEM (50:50) media containing 2 mmol/L L-glu-tamine, 1� GHT (glycine, hypoxanthine, thymidine), and10% FBS in the presence of 400 mg/mLG418 (Calbiochem)with change of media every 2 to 3 days. Expression ofCXCR4 and CXCR7 was confirmed by FACS analysis usingspecific antibodies (MAB173/FAB170 for CXCR4;MAB4227/MAB42273 for CXCR7; R&D systems). All sec-ondary antibodies were obtained from Jackson ImmunoR-esearch Laboratories, Inc..

125I-CXCL12a–binding assayAnti-CXCL12 antibodies were prepared in PBS contain-

ing 2% FBS, pH 7.4 in serial 1:3 dilutions and incubatedwith 1 ng/mL 125I-huCXCL12a (PerkinElmer) in 96-wellpolypropylene plates (Costar) for 1.5 hours. Meanwhile,293-CXCR4 cells or 293-CXCR7 cellswere prepared at 0.8�105 cells per well and incubated with preincubated 125I-huCXCL12a and anti-CXCL12 antibodies for 2 hours. Theincubated cell solutions were transferred to prewet Multi-Screen, BVfilter plate (Millipore). The cell-boundCXCL12awas determined by counting the cpm captured on individ-ual filters. AMD3100 (Sigma) was included as a control andused in the same way as antibodies.

Chemotaxis assaysDuplicate wells of isolated mouse bone marrow or

Jurkat cells (5 � 105 cells), resuspended in Hank’sbalanced salt solution containing 0.2% bovine serumalbumin or in 5% FBS, were put into 24-Transwell insertswith 5-mm pore size. Media alone or media with variouschemokines were added to the lower chamber. After 3hours at 37�C/5% CO2 incubator, cells in the lowerchamber were counted. For 4T1 migration, the top ofthe 24-Transwell plates (8-mm pore size; BD Biosciences)was coated with 10 mg/mL collagen type I from rattail (Upstate) prepared in PBS for 1 hour at 37�C. Atotal of 5 � 104 4T1 cells per well were seeded in 200 mLof serum-free medium on the top and 350 mL of serum-free media containing CXCL12a, with or without vari-ous concentrations of 30D8 were added to the bottomwell. After 16-hour incubation at 37�C/5% CO2, cellsremaining on the top were removed and those migratedthrough the filter were counted under inverted micro-scopy (Zeiss Inc.).

Translational RelevanceThe CXCL12/CXCR4 axis has been shown to play

important roles in tumorigenesis and in a variety ofother pathologic conditions.Here,wedescribe thedevel-opment and characterization of a high-affinity human-ized monoclonal antibody (mAb) specific for CXCL12.The antibody had antitumor activity in multiple xeno-graft and orthotopic tumor models, as single agent or incombination with anti-VEGF antibodies. This antibodyis suitable to test the hypothesis that targeting CXCL12 isa valid strategy to treat cancer and inflammatory diseasesin humans.

Zhong et al.

Clin Cancer Res; 19(16) August 15, 2013 Clinical Cancer Research4434




Rac GTP activity assayJurkat cells stimulated with CXCL12a in the absence or

presence of anti-CXCL12 antibodies were lysed in RIPAbuffer (Sigma) with proteinase (Roche) and phosphataseinhibitor cocktails (Sigma). Lysates were incubated at 4�Cfor 2 hours with GST-human p21-activated kinase PBD-coupled glutathione-sepharose beads (Cytoskeleton), fol-lowed by washing with RIPA buffer. Total and GTP-boundRac1 were detected by Western blotting with an antibodyagainst Rac1 (Cell Signaling Technology, Inc.).

Cell viability assayCells were incubated for various lengths of time in clear

bottom/black wall 96 wells at 37�C in a 5%CO2 incubator.Cell numbers were determined by the CellTiter-Glo Lumi-nescent Cell Viability Assay Kit (Promega). In some cases,proliferating cells and apoptotic cells were visualized byKi67 and caspase-3 staining.

Xenograft modelsSix- to 8-week-old female Balb/c, Balb/c nude mice, and

C57BL/6mice were obtained fromCharles River or JacksonLaboratory. Maintenance of animals and experimental pro-tocols were conducted following federal regulations andwere approved by Institutional Animal Care and Use Com-mittee at Genentech. For EL4, LLC, A673, HM7, Calu6,BxPC3, and KMS11 tumors, 106 cells were washedwith PBSonce before resuspending in growth factor–reduced Matri-gel (BD Biosciences) at a concentration of 107 cells/mL andinjected (100 mL) subcutaneously into the dorsal flank ofmice. Anti-Ragweed, gD:2566, and gD:5237 were used asanti-mouse, -hamster, and humanized antibody controls.Primary tumor volumes were measured once a week usingthe ellipsoid volume formulas (0.5� L�W2, where L is thelength and W is width) and verified at terminal point bytumor weight. All antibodies used in animal studies hadendotoxin level below 0.02 eu/mg. AMD3100 was admin-istered five times/week at a dose of 5 mg/kg, i.p.

Breast cancer orthotopic model4T1 cells were injected into the right fourth mammary fat

pad of female Balb/c mice at the concentration of 2 � 105

cells per 10 mL of PBS using an insulin syringe. Treatmentwas started 2 days after tumor cell implantation. At the endof the study, lungs were perfused with PBS, fixed in 10%formalin, and visible tumors were counted (21). In somecases, authenticity of tumorswithin lungswas confirmed byH&E staining and Bioluminescence Imaging (BLI).

FACS analysisWhole tumors were minced into small pieces, followed

by sequential digestion with Media I for 60 minutes andMedia II for 15minutes, as previously described (22). Singlecells were first incubated at room temperature for 10 min-utes with rat anti-mouse CD16/CD32 mAb (Fc block; BDBiosciences), followed by staining with the following fluor-ophore-conjugated antibodies: anti-CD11b (integrin aMchain, Mac-1, a-chain, M1/70; BD Pharmingen), anti-Ly6G(Gr1, RB6-8C5; eBioscience), anti-CD31 (PECAM-1;

eBioscience), anti-human/mouse CXCR7 (8F11-M16; Bio-Legend), anti-F4/80 (BM8; eBioscience), and anti-CXCR4(CD184; BioLegend). All isotype-control antibodies werefrom R&D Systems. Propidium iodide (Sigma) was used todistinguish viable and dead cells. Data were acquired usingthe FACSCalibur or BDLSR II instruments (BDBiosciences)and analyzed using FlowJo software (TreeStar).

Statistical analysisUnless specifically stated, Student t test was used for all

the statistical analyses, and P � 0.05 was considered to besignificant. Graphs present mean � SD.

ResultsIn vitro characterization of hamster anti-CXCL12aantibodies

Antibodies against both mouse and human CXCL12awere generated in hamsters. Serum titers were assayed fortheir ability to bind mouse and human CXCL12a by directELISA (Supplementary Methods). Five hamster antibodiesshowed different degrees of inhibition of CXCL12a-induced chemotaxis of Jurkat cells, with 30D8 being themost potent (Fig. 1A). MAB310 (R&D Systems) served as apositive control. 30D8 at 2 mg/mL completely blockedchemotaxis of Jurkat cells elicited by 10 ng/mL CXCL12a,with an average IC50 of approximately 0.5 mg/mL (�3nmol/L). MAB310 was about 50-fold less potent. None ofthese antibodies had any effect on migration when testedalone. 30D8 could also bind to CXCL12b and CXCL12g , asassessed by direct ELISA and CXCL12b-induced Jurkat cellmigration (data not shown). Next, we compared the abil-ities of various antibodies to block CXCL12a binding to293-CXCR4 or 293-CXCR7 cells (Fig. 1B and C). Again,30D8 was the most potent in blocking binding to CXCR4,whereas all antibodies showed various degrees of inhibitoryeffect on binding to CXCR7. The inhibition was specific as30D8 did not affect chemotaxis of isolated mouse bonemarrow cells induced by unrelated chemokines such as KC,MIP2, or RANTES (Fig. 1D).

AMD3100 (23), a CXCR4 small-molecule antagonist,showed inhibitory effects in 293-CXCR4 cell-binding assay,with an IC50 of 0.16 nmol/L.

Upon binding to CXCL12a, CXCR4 undergoes dimer-ization and stabilization leading to activation of a numberof downstream signaling pathways including the smallGTPase Rac1, which has an essential role in cytoskeletonremodeling (5). Jurkat cells showed a marked increasein GTP-bound Rac1 when treated with CXCL12a at 100ng/mL for 5 minutes. 30D8, 18E9, or 46H9 inhibitedCXCL12a-induced Rac1 activation in a dose-dependentmanner. In the same assay, anti-CXCR4 antibody MAB170from R&D systems at 200 mg/mL and AMD3100 at 40mg/mL had moderate effects (Fig. 1E).

Inhibition of primary tumor growth by 30D8 alone orin combination with anti-VEGF antibody

The mouse lymphoma model EL4 has been shown to berelatively refractory to anti-VEGF antibody therapy in vivo

A Humanized Antibody Targeting CXCL12

www.aacrjournals.org Clin Cancer Res; 19(16) August 15, 2013 4435




D EGTP-Rac1

Total Rac1

100 ng/mL CXCL12α

200 µg/mL Control Ab

20 µg/mL 30D8

200 µg/mL 30D8

20 µg/mL 46H9

200 µg/mL 46H9

200 µg/mL 18E9

200 µg/mL MAB170

40 µg/mL AMD3100

+– +

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

Mig

rate

d c

ells

/well

mCXCL12α mRANTESmKC mMIP20 20 100 20 100 20 100 20 100 20 100 20 100 20 100 20 100100Chemokines (ng/mL)

Control Ab +

30D8 + + + + + + + +

B

0

20

40

60

80

100

120

0.01 0.1 1 10 100 1,000

18E9

30D8

46H9

MAB310

AMD3100

Perc

enta

ge v

s. negative c

ontr

ol of

no a

ntibody tre

atm

ent

Antibody (nmol/L)

C

0

20

40

60

80

100

120

0.01 0.1 1 10 100 1,000

Antibody (nmol/L)

18E9

30D8

46H9

MAB310

AMD3100

20,000

40,000

60,000

80,000

Mig

rate

d c

ells

/we

ll

100,000

120,000

140,000

160,000

1:3 1:3 1:3 1:3 1:3 1:3

0

10 n

g/m

L h

uC

XC

L12α

2 µ

g/m

L

2 µ

g/m

L

2 µ

g/m

L

2 µ

g/m

L

120 µ

g/m

L

30 µ

g/m

L

IC50

(µg/mL)

Antibodies

Dilution factor

MAB310

25.53±3.15

12C1

6.88±1.60

18E9

0.95±0.22

46H9

0.69±0.13

15C11

0.91±0.27

30D8

0.40±0.15

A

Perc

enta

ge v

s. negative c

ontr

ol of

no a

ntibody tre

atm

ent

Figure 1. Characterization of hamster anti-CXCL12 antibodies. A, dose-dependent inhibition of CXCL12a-induced Jurkat cell migration in the absence orpresence of various concentrations of hamster antibodies. Of note, 10 ng/mL human CXCL12a was preincubated with antibodies in serial 1:3 dilutions atroom temperature for 30 minutes. The highest concentration used for each antibody is shown. IC50 was the average of 3 independent experiments. Data areexpressed as mean � SD. B, relative potencies of various antibodies in inhibiting 125I-CXCL12a binding to 293-CXCR4 or (C) 293-CXCR7 cells. The samebatch of antibodies was used in two binding assays, with AMD3100 as a control. Results shown are representative of 2 independent experiments. Thecell-bound 125I-CXCL12a was determined and the results are expressed as percentage versus negative control of no antibody treatment. D, effect of 30D8 orcontrol hamster antibody gD:2566 (Genentech) at 20 mg/mL on freshly isolated bone marrow cell migration induced by mouse KC, MIP2, and RANTES at 20 or100 ng/mL. Data are expressed as mean � SD. E, effects of 30D8 on CXCL12a-induced Jurkat cell Rac activation. Jurkat cells were treated with 100 ng/mLhuman CXCL12a for 10 minutes in the presence of various concentrations of 30D8, 46H9, 18E9, or control hamster antibody, MAB170 or AMD3100, asindicated. The total andGTP-boundRac1weredetectedbyWestern blotting using an antibodyagainst Rac1. Datawere representative of 3 independent studies.

Zhong et al.





(15, 24, 25). Single-agent 30D8 significantly inhibited EL4tumor growth in C57BL6 mice compared with the controlantibody at a dose as low as 10 mg/kg. A 50% growthinhibitionwas achieved at the dose of 30mg/kg, three timesa week. The magnitude of inhibition was comparable tothat obtained with the anti-VEGF antibody B20.4.1.1 (26)at 10 mg/kg twice weekly (Fig. 2A). A combinationof B20.4.1.1 and 30D8 monoclonal antibodies (mAb)achieved a greater inhibition (�70%). There was a signif-icant difference in tumor size between mice treated solelywith B20.4.1.1 versus those treated with a combination of30D8 and B20.4.1.1 (P < 0.01).A significant inhibition of proliferation and induction of

apoptosis was noted in tumor sections from animals treatedwith 30D8, as assessed by caspase-3 and Ki67 staining (Sup-plementary Fig. S1). There was a 53% and 40% reductionin F4/80þ and CD11bþGr1þ cell infiltration, respectively(P < 0.01) in 30D8-treated mice compared with controlantibody–treated mice. Only a 21% reduction in F4/80þ

cell infiltration was observed in B20.4.1.1-treated mice(Fig. 2B). In addition, an increase in CD11bþGr1þ cellswithin the tumors was observed following treatment withB20.4.1.1. Blood vessel density as assessed by CD31þ cellswas reduced in both 30D8- and B20.4.1.1-treated groups.Next, we evaluated the effects of 30D8on growth of Lewis

lung carcinoma (LLC) cells implanted subcutaneously inC57BL6 mice (ref. 25; Fig. 2C). 30D8, administered on thesame day of tumor cell implantation at the dose of 30 mg/kg, resulted in about 50% inhibition at day 14, which wascomparable with that obtained with B20.4.1.1 at 10mg/kg.Single-agent activity was also observed when 30D8 wasgiven to animals bearing established tumors of approxi-mately 400 mm3 (data not shown). An additive effect (P <0.01) was observed when 30D8 was combined withB20.4.1.1 (Fig. 2C). Interestingly, CXCL12a had a signifi-cant effect on LLC cell survival in vitro under serum-freeconditions. As shown in Fig. 2D, about 50% of the cells inthe control group began to die after 24 hours in culture,whereas cells incubated with as low as 1.6 ng/mL CXCL12asurvived as well as those in 10% serum and continuedto proliferate (Fig. 2D). Such effects were completely neu-tralized in the presence of 100 mg/mL 30D8.

Inhibition of metastasis in a mouse orthotopicbreast cancer modelTo investigate the effects of 30D8 on cancer metastasis,

4T1 breast cancer cells transfected with the luciferase genewere orthotopically inoculated into Balb/c mice (21, 27).Although treatment with 30D8 had no effect on primarytumor growth (Fig. 2E), it resulted in approximately 60%reduction in the number of lung cancer nodules at the doseof 10 mg/kg (Fig. 2F). All lungs in the control groupdevelopedmetastasis, with an average of 14nodules,where-as one third of the mice treated with 30D8 had no visiblelung metastasis and those that did develop lung metastasishad an average of 5 nodules. Combination of 30D8 at 30mg/kg with B20.4.1.1 at 10 mg/kg further enhanced theinhibitory effects comparedwithB20.4.1.1 alone (P<0.05).

The result was confirmed on 7-mm thick lung paraffin-embedded sections stained with hematoxylin and eosin(H&E) to reveal cancer nodules not only on the surface oflung but also within lung tissues (data not shown).

Consistent with the in vivo observations, CXCL12a hadno effect on 4T1 cell proliferation. However, it significantlystimulated 4T1 cell migration at 10 ng/mL (�4.5-foldincrease compared with control) in vitro. Such stimulationcould be completely blocked by 50 mg/mL 30D8 (Supple-mentary Fig. S2).

Effects of 30D8 in collagen-induced arthritis modelCXCL12a is highly expressed in the synovium of patients

with rheumatoid arthritis and has been implicated in therecruitment and accumulation of CD4þ memory T cells, Bcells to the synovium during the induction phase as wellas in the recruitment of myeloid cells and synovial fibro-blasts during disease onset and development (28). Toinvestigate whether 30D8 had any effect in an experimentalmouse collagen-induced arthritis (CIA) model (Supple-mentaryMethods), we first injected 30D8 intoDBA-1Jmiceat day 24 following bovine collagen type II injection (pre-ventive phase). Mouse/hamster chimera 30D8 (ch30D8)was used to avoid immune responses.muTNFRII-IgG2awasused as a positive control (29). Compared with muTNFRII-IgG2a, ch30D8at 30mg/kgwasnot effective duringonset ofthe disease (day 24–35) but could significantly slow downdisease progression at day 35 to 73 (Fig. 3A).

To determine whether ch30D8 can inhibit establishedinflammation, we initiated treatments in cohorts of CIAmice at day 42. ch30D8 monotherapy significantly ame-liorated arthritis compared with the control antibody (Fig.3B). Although treatment with muTNFRII-IgG2a or ch30D8alone could reduce the clinical score on day 83, the com-bination was significantly better than each monotherapy inpreventing the progression of arthritis in mice with milddisease at enrollment (Fig. 3C). Such combination substan-tially reduced bone-erosive changes in CIA mice, resultingin a considerably higher joint cortical bone volume (JCBV)compared with mice treated with either inhibitor alone(ref. 30; Fig. 3D).

Effects of 30D8 on laser-induced choroidalneovascularization in mice

Laser-induced rupture of Bruch’s membrane results inchoroidal neovascularization (CNV) in mice (31). Bonemarrow–derived hematopoietic stem cells are known to beincorporated into sites of retinal and CNV (32). Treatmentwith 30D8 at 30 mg/kg intraperitoneally (i.p.) resultedin a 50% reduction of neovascularization not only inprevention (Fig. 3E) but also in intervention studies, inwhich 30D8 was given 48 hours post-laser (Supplemen-tary Methods and Supplementary Fig. S3A). RepresentativeimagesofCNV in animals treatedwith various antibodies areshown in Supplementary Fig. S3B. In both circumstances,10 mg/kg B20.4.1.1 resulted in more than 90% inhibition.

Infiltration of macrophages/dendritic cells expressingCX3CR1 is known to occur as early as 8 hours after laser






0

20,000

40,000

60,000

80,000

100,000

120,000

140,000

160,000 24 h 48 h

EL4A B

Concentration of human CXCL12α in

serum-free media (ng/mL)

Ce

lls/w

ell

F4/80+

CD11b+/Gr1+

CD31+

0

2

4

6

8

10

12

14

16

18

Control Ab 30 30D8 30 B20 10

% o

f via

ble

ce

lls

LLCC D

Days after tumor cell implantation

0

500

1,000

1,500

2,000

2,500

3,000

3,500 Control Ab

30D8 3.3

30D8 10

30D8 30

Tu

mo

r vo

lum

e (

mm

3)

53 8 10 12

B20

30D8 30+B20

Cell

plated

10% FBS100 μg/mL 30D8 − − − − − +

0 1.6 8 40 2000

1,000

2,000

3,000

0 5 10 15 20

**


Control Ab

30D8

B20

30D8+B20

Tu

mo

r vo

lum

e (

mm

3)

F

Me

t co

un

ts p

er

lun

g

Control Ab 30 30D8 10 30D8 30 B20 10 B20+30D8 30

02468

101214161820

P < 0.05

4T1E

0

400

800

1,200

1,600

2,000

6 12 21 28 35 39

Days after tumor implantation

Tu

mo

r vo

lum

e (

mm

3)

Control Ab

30D8 30

30D8 10

B20

30D8 30+B20

***

**

******

***

8

***

**

******

****

P < 0.05

P < 0.01

** ******

***

**

Figure 2. Effects of 30D8 in various mouse tumor models, alone or in combination with anti-VEGF antibody B20.4.1.1 (B20). All antibodies wereadministered intraperitoneally. A, dose-dependent efficacy of 30D8 in the EL4 model. C57BL6 mice (n ¼ 10/group) bearing EL4 tumors were treatedwith 30D8 at various doses (three times/week), with or without anti-VEGF antibody B20.4.1.1 at 10 mg/kg (twice/week). Hamster antibodygD:2566 (Genentech) served as a control antibody for 30D8. Tumor size was measured at day 3, 5, 8, 10, and 12. B, percentage of F4/80þ,CD11bþ/Gr1þ, and CD31þ cells in EL4 tumors treated with control antibody, 30D8, or B20.4.1.1 at day 10 after tumor cell implantation using FACS.Five tumors per treatment group were used and stained with appropriate antibodies. C, efficacy of 30D8 in the LLC model. C57BL6 mice (n ¼ 10/group) bearing LLC tumor were treated with a control hamster antibody, 30D8 at 30 mg/kg, B20.4.1.1 at 10 mg/kg, or combination of both. Tumor sizewas measured at day 7, 14, and 21. D, effects of various concentrations of CXCL12a on proliferation/survival of LLC cells under serum-free conditionin vitro. In the same study, 100 mg/mL 30D8 was used to show the effect was specific against CXCL12a. E, efficacy of 30D8 on 4T1 primarytumor growth. 4T1 cells were injected orthotopically into the right fourth mammary fat pad of 10 female Balb/c mice. Two days after tumor cellimplantation, animals were treated with either 30D8 (at 10 or 30 mg/kg, i.p.) or B20.4.1.1 (at 10 mg/kg), or a combination of both (30D8 at 30 mg/kg).Tumor size was measured at day 6, 12, 21, 28, 35, and 39. F, effect of 30D8 on 4T1 breast cancer lung metastasis. Data are expressed as mean � SDand are representative of 3 independent studies. ��, P < 0.01; ��, P � 0.05 compared with the control treatment group.

Zhong et al.





A

C

Control Ab mTNFRII-mIgG2a

fusion protein

ch30D8 ch30D8 + mTNFRII-mIgG2a

fusion protein

0

2

4

6

8

10

12

14

16

18

37 43 50 58 64 71 78 83

Control Ab

mTNFRII-IgG2a

ch30D8

mTNFRII-IgG2a+ch30D8

0

2

4

6

8

10

12

14

16

0 2

4

6

8

10

12

14

16

18

21 28 35 42 49 56 62 67

Control Ab

mTNFRII-IgG2a

ch30D8

Clin

ical score

B

31

0

31

5

32

3

33

3

34

1

34

6

35

7

36

2

36

7

37

0

38

3

38

4

39

4

30

9

31

4

32

0

32

1

34

2

35

1

36

4

36

9

37

2

37

4

37

8

37

9

39

2

31

6

32

4

33

4

33

5

33

8

34

0

36

3

36

6

38

8

38

9

39

3

39

5

39

9

30

7

32

7

33

0

33

2

33

7

33

9

34

9

36

1

36

8

37

1

38

2

38

5

39

6

Clin

ical score

Animal ID

Day Day

D

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

Jo

int

co

rtic

al

Bo

ne

vo

lum

e (

mm

3)

Co

ntr

ol A

b

ch

30

D8

ch

30

D8

+ m

TN

FR

II-I

gG

2a

mT

NF

RII

-Ig

G2

a With controlDunnett's

0.05

E

Contr

ol A

b

30D

8

B2

0

CN

V a

rea

0

20,000

40,000

60,000

80,000

100,000

120,000

140,000

160,000

Contr

ol A

b

A

vera

ge flu

ore

scent units/

laser

are

a

30D

8

F

****

** ****

** **

***

***

***

**

0

50,000

100,000

150,000

200,000

250,000

Figure 3. Activity of 30D8 in mouse inflammatory models. A, effects of mouse/hamster chimeric 30D8 (ch30D8) in the CIA mouse model (see SupplementaryMethods) during the induction phase. Clinical scores of mice (n ¼ 15/group) that had been treated with ch30D8 at 30 mg/kg (three times/week) starting atday 24 after primary immunization with bovine collagen type II. Anti-Ragweed served as a negative control and mTNF-aRII-mIgG2a fusion protein was thepositive control. B, ch30D8 results in inhibition of established CIA. Clinical scores of mice treated with antibodies or fusion protein, starting on day 42after primary immunization. C, combination of ch30D8 and mTNF-aRII-mIgG2a significantly prevented disease progression compared with eitherantibody alone. Clinical scores at day 37 (unfilled symbols) and day 83 (solid filled symbols) after primary immunization. D, combination of ch30D8 andmTNF-aRII-mIgG2a significantly increased JCBV in the metatarsal-phalangeal and metacarpophalangeal joints compared with treatment with controlantibody or either antibody alone. Statistical analysis was done by Dunnett's method. E, effects of 30D8 on laser-induced CNV in the mouse (seeSupplementary Methods). Plot represents mean CNV area in animals (n¼ 10/group) treated with control antibody or 30D8 at 30mg/kg 24 hours before laser.Treatmentwas continued for 9 days before collecting retinal pigment epithelium–choroid–sclera complex (RPE) formicroscopic evaluation. Each neovasculararea (mm2) surrounding the optic nerve was quantified by NIH ImageJ 1.37V software. F, inhibition of macrophage/dendritic cell infiltration in laserspots inCX3CR1-GPF transgenicmice (34).Mice (n¼6/group)were treatedwith either control antibody or 30D8 at 30mg/kg24hours before laser (preventionstudies). Quantification of macrophage/dendritic cell infiltration using flat-mounted RPE was made at day 11. Data are expressed as mean � SD andare representative of 3 independent studies. ��, P < 0.01; ��, P � 0.05 compared with the control treatment group.






injury, with a peak at day 2 (33). We used a transgenicmouse model in which one copy of the CX3CR1 gene isreplaced by the GFP reporter gene (34). Compared withnormal eyes, there was an approximately 20-fold increasein macrophage infiltration in eyes with laser injury. Whenanimals were pretreated with 30D8 24 hours before laserinjury, macrophage/dendritic cell infiltration into theinjured sites was significantly reduced compared with thecontrol antibody group (Fig. 3F). Representative images ofinfiltrating GFPþ cells are shown in Supplementary Fig.S3C.

Generation and characterization of humanized 30D830D8 was selected for humanization (Supplementary

Methods) based on its in vitro and in vivo activities. Thealignment of light chain k between hamster 30D8 andhumanized 30D8 (hu30D8) is shown in SupplementaryFig. S4 (35–38). hu30D8 bound not only to CXCL12a, butalso to CXCL12b and CXCL12g in direct ELISA. It specif-ically blocked CXCL12a- and CXCL12b-induced Jurkat cellmigration, with an IC50 of 0.16 mg/mL (�1 nmol/L), whichis comparable with both hamster and mouse/hamster chi-meric counterparts. Because of a high apparent on-rate andnonspecific binding of CXCL12a to the dextran-coatedsensor chip at high concentration, the affinity for bindingto hu30D8 could not be reliably determined using Biacoretechnology. KD was later measured by Biolayer Interferom-etry (Supplementary Methods) as 0.923 and 2.39 nmol/Lfor human and mouse CXCL12, respectively.

hu30D8 induced dose-dependent inhibition of EL4tumor growth in C57BL6 mice at 10 to 50 mg/kg doserange. This effect was comparable with that of its hamstercounterpart (Fig. 5A). When hu30D8 was given in com-bination with B20.4.1.1, an additive effect was seen(Fig. 5B). hu30D8 also elicited inhibitory effects in thehuman Calu6 lung carcinoma, a tumor model highlyresponsive to VEGF blockade (39). Of note, 10 and 30mg/kg hu30D8 were equally efficacious in this modeland resulted in approximately 50% tumor growth inhib-ition, which was statistically significant compared withthe control antibody group (P < 0.05; Fig. 5C). There wasno tumor shrinkage when we combined 30D8 withB20.4.1.1. CD11bþGr1þ, F4/80þ, and CD31þ cells insidethe tumor were examined at week 4, 24 hours afterthe last dosing. Treatment with hu30D8 at 30 mg/kgled to average 45%, 31%, and 36% decreases in tumorF4/80þ, CD11bþGr1þ, and CD31þ cells, respectively, asassessed by fluorescence-activated cell sorting (FACS)analysis (P < 0.05; Fig. 5D).

We also tested the activity of hu30D8 as a single agentin other human xenografts, including the A673, BxPC3,HM7, KMS11, and Ovcar-3 models. In general, hu30D8was less effective in these models compared with Calu6,with 15% to 30% tumor inhibition achieved at the doseof 30 mg/kg. On the basis of TaqMan analysis, expres-sion levels of CXCL12a and its receptors in humantumor cells in vitro did not correlate with efficacy ofhu30D8 in vivo.

Epitope mapping and crystal structure of hu30D8 Fab/human CXCL12a revealed a protein–proteininteraction "hot spot"

We initially sought to identify the epitope in severalhamster antibodies, including 30D8, by alanine scanningof CXCL12a (Supplementary Methods). This analysis indi-cated that Asn44 in the middle region of the molecule is thekey binding epitope for 30D8, whereas Val18, Leu42, andAsn45were also important for 30D8binding.Other hamsterantibodies and the commercial antibody MAB310 seemedto cover a wider region. On the basis of epitope mappingdata, CXCL12a processing at N- and C-termini in bloodshould not affect its binding to 30D8.

To further elucidate the molecular recognition ofCXCL12a by 30D8 and the mechanism of inhibition, wedetermined the crystal structure of hu30D8 Fab in complexwith human CXCL12a (Supplementary Methods and Sup-plementary Fig. S4). The crystallographic asymmetric unitcontains a CXCL12a dimer and 2 Fab molecules eachbinding to one CXCL12a monomer (Fig. 4A). This isconsistent with concentration-dependent dimerization ofCXCL12a (40) as the crystallization media contains 160mmol/LCXCL12a (SupplementaryMethods; ref. 41). It alsoindicates that the antibody does not inhibit CXCL12adimerization.

The hu30D8 paratope involves all six complementarity-determining regions (CDR; Fig. 4B). The specific interac-tions at the two Fab-CXCL12a interfaces are essentiallyidentical. Remarkably, each interface is composed of resi-dues from only one CXCL12a monomer, suggesting thatthe antibody can bind to the monomeric form of CXCL12aas well. Figure 4C depicts the specific interactions at Asn44

and Asn45 of CXCL12a that occupy a groove generated bythe three heavy-chain CDRs of hu30D8, and engage polarinteractions with CDR-H3. The network of interactionsconcentrated in this region seems to contribute to themajority of binding free energy, therefore constitutes abinding "hot spot". The presence of such a hot spot cor-relates well with the mutagenesis results, where alaninesubstitution at Asn44 causes drastic reduction in binding,whereasmutations at other sites hadmoderate to no effects.

The crystal structure revealed key interactions contribut-ing to hu30D80s inhibitory activity. The RFFESH fragmentnearN-terminusofCXCL12a is known tobe responsible forreceptor binding (42). As shown in Fig. 4D, residues Glu15

andHis17 interact with hu30D8 throughmultiple hydrogenbonds. Antibody binding is likely to interfere withCXCL12a activity by blocking RFFESH binding to CXCR4,and likely to CXCR7 as well, because both receptors sharesimilar ligand-binding surfaces for the binding of the syn-thetic ligands (43).

Pharmacokinetics of hu30D8Pharmacokinetic characteristics of 30D8 were first stud-

ied in mice and subsequently in rats and cynomolgusmonkeys (Supplementary Methods). Fast clearances wereobserved following a single dose of 30mg/kg in Balb c/nudemice for both 30D8 (intraperitoneally; Fig. 6A) and

Zhong et al.





hu30D8 (intravenously; Fig. 6E) and in cynomolgus mon-keys for hu30D8 (intravenously; Fig. 6C). A significantdecrease in serum CXCL12a was observed 3 hours afterch30D8 administration in NCr nude mice. This correlatedinversely with the serum 30D8 levels (Fig. 6B). Similarly,serum CXCL12a level was undetectable in HM7-implant-ed nude mice 3 hours after 30D8 treatment, whereas nosignificant change was observed after AMD3100 or thecontrol antibody. In the same study, serum VEGF was notaffected significantly by any of the treatments (Supplemen-tary Fig. S5A). Furthermore, a pharmacokinetic study wasconducted using multiple dosing regimens as the animalsreceived 30D8, three times/week, in all in vivo models.When 30D8 was given intraperitoneally for 24 consecutivedays at 30 mg/kg, three times/week in HM7-implantednude mice, no significant accumulation of the antibodywas observed (Supplementary Fig. S5B).In contrast, hu30D8 had a normal clearance in nude rats

at the same dose level (intravenously; Fig. 6D). KD forhu30D8 against rat CXCL12a was 154 nmol/L, 80-foldhigher than CXCL12a frommouse/human CXCL12amea-sured by Biolayer Interferometry. Lower affinity wasexpected as rat CXCL12a differs from mouse/humanCXCL12a at amino acids Asn44 and Leu63, with the formerbeing the most critical residue for hu30D8 binding.To investigate whether epitope contributed to fast clear-

ance, several antibodies raised in hamsters that target dif-

ferent binding epitopes in CXCL12a were tested. Althoughepitopes were different, all these antibodies had similarbinding affinity toward CXCL12a and showed similarclearance as 30D8 (Supplementary Fig. S6).

Finally, we generated a series of single point mutationsin the heavy chain of hu30D8 around amino acid Asp95,Gln96, or Tyr100 (Supplementary Methods). Among them,one mutation, hu30D8D95A, which removes a key interac-tion with Asn 44 of CXCL12a, showed 1,000-fold loweraffinity to human CXCL12a and much slower clearancein vivo than parental hu30D8 (Fig. 6E). Other mutants suchas hu30D8Y100A had similar affinity toward CXCL12a andsimilar pharmacokinetic profile as the parental antibody.

DiscussionThe CXCL12/CXCR4 signaling axis has been actively

explored as a potential drug target (44). So far, most of thestudies focused on the CXCL12 receptor CXCR4 and onsmall-molecule inhibitors targeting CXCR4 (45), althoughefforts using neutralizing anti-CXCL12 antibodies havebeen also reported (46, 47).

Here, we show efficacy of the anti-CXCL12 antibody30D8 in several tumors as well as inflammatory models asa single agent, despite a relatively fast clearance. In allmodels tested, the efficacious doses of 30D8 were higherthan those of anti-VEGF mAb B20.4.1.1, but in the samerange as those of DC101, a widely used mAb directed

E15

S16

H17 Y100

Y31

N50

N99

D

HC LC

HC LC

CXCL12αN-term

C-term

A

H1

H2

H3

L1

L3

L240s loop

10s-20s loop

Trp 57

B

N44

N45

N22

N99

W57

D95 Y100

R47

N46

S54

K43

E58

W92

G100a

L98

C

Figure 4. Crystal structure ofhu30D8 Fab in complex withhuman CXCL12a A humanCXCL12a dimer is shown asribbon diagram, green andmagenta. Two Fab molecules(shown as surface rendering)and human CXCL12a dimerconstitute the crystallographicasymmetric dimer. Blue, heavychain; orange, light chain. B, aclose-up view of the interface. Inthe front, CXCL12a fragment thatengages direct interactions withhu30D8 are shown in greenribbons. In the back, hu30D8 isshown as surface rendering, heavychain in blue, light chain in orange.The paratope is colored in red. Thebright redpatches comprise atomswithin 4 Å from human CXCL12a;pink patches comprise atomswithin 4–4.5 Å. C, the specificinteractions around the "hot spot"N44 and N45 of human CXCL12a.Color-coding for the carbonatoms: green, human CXCL12a;blue, Fab heavy chain; orange,Fab light chain. Other atoms arecolored by type: blue, nitrogen;red, oxygen. D, the Fab interactioninvolving RFFESH fragment ofhuman CXCL12a. Color-coding isthe same as in C.






against VEGFR2, which is typically administered at 40mg/kg every 3 days in mouse models (27).

mRNA levels of CXCL12a and its receptor in humantumor cells were not useful to predict antitumor activity ofhu30D8, possibly because they were neither representativeof the levels of protein on the cell surface nor indicative ofthe levels of proteins with specific functions. In fact, anumber of CXCR4 isoforms, mutants with varying levelsof glycosylation, have been found on the cell surface, withvarious functions (48, 49).

Humanization of 30D8 allowed us to investigate phar-macokinetics in cynomolgus monkeys. Such antibodieswill also enable safety studies to assess the consequencesof long-term CXCL12 blockade. hu30D8 bound to circu-lating CXCL12a with very high affinity and subsequently

blocked CXCL12a binding (free or heparin-boundforms) to both CXCR4 and CXCR7. The contribution ofCXCR7 in tumor growth/metastasis in each model couldnot be addressed in the current study, due to lack ofspecific inhibitors. AMD3100 was less efficacious in someof our in vivomodels, which might be due to its activity asa CXCR7 ligand with allosteric agonist properties (50).

When 30D8 was used in combination with a TNFRII-IgG2a fusion protein, efficacy could be further improv-ed, suggesting that CXCL12 could become a noveltarget at least in subsets of arthritis patients whofail to respond to anti-TNF-a therapy (28). In the CIAand CNV models, blocking CXCL12 function preventedmacrophage and/or lymphocyte infiltration at an earlyphase, although the current study could not rule out

A

C

0

500

1,000

1,500

2,000

5 7 9 12

Human Ab Control

Hamster Ab Control

hu30D8 50

hu30D8 30

hu30D8 10

30D8 50

30D8 30

30D8 10


**

***

Tu

mo

r vo

lum

e (

mm

3)

EL4

Calu6

0

200

400

600

800

1,000

1,200 Control human Ab

hu30D8 30

hu30D8 10

hu30D8 3.3

B20

B20+hu30D8

D

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

hu30D8Control Ab

F4/80

CD11b/Gr1

CD31

7 10 14 17 21


% o

f via

ble

ce

lls

Tu

mo

r vo

lum

e (

mm

3)

**

***

B

Tu

mo

r vo

lum

e (

mm

3)


0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

5 8 11 13 15 19

Human Ab Control

hu30D8

B20

B20+hu30D8

**

***

**** **

Figure 5. In vivo activity of hu30D8. A, dose-dependent efficacy of hu30D8 in the EL4 tumormodel, which was comparable with that of parental 30D8. C57BL6mice (n ¼ 10/group) bearing EL4 tumor were treated with 30D8 or hu30D8 at various concentrations (intraperitoneally three times/week). In the samestudy, hamster antibody (Ab) gD:2566 and human antibody gD:5237 (Genentech) served as controls for 30D8 and hu30D8, respectively. Tumor size wasmeasured at day 5, 7, 9, and 12. B, additive effects of mAbs hu30D8 (30 mg/kg) and B20.4.1.1 (10 mg/kg) on EL4 tumor growth. C, dose-dependentefficacy of hu30D8 in the human Calu-6 tumor model. Balb c/nude mice (n ¼ 10/group) were treated with hu30D8 at various doses, alone) or with 30 mg/kghu30D8 in combination with 10 mg/kg B20.4.1.1. Tumor size was measured at day 7, 10, 14, 17, and 21. D, percentage of F4/80þ, CD11bþ/Gr1þ, andCD31þ cells in Calu6 tumors treated with the control human antibody or hu30D8 at day 21 after tumor cell implantation using FACS. Data are expressedas means � SD and are representative of 3 independent experiments. ��, P < 0.01; ��, P � 0.05 compared with the control treatment group.

Zhong et al.





the possibility that 30D8 inhibited angiogenesis bydirectly inhibiting endothelial cell migration and sur-vival as well.

Preliminary data indicate a greater efficacy of hu30D8 insyngeneic immunocompetent mice compared with immu-nodeficient mice bearing EL4 tumor (data not shown). If

30 mg/kg; i.v.

0.00

100.00

200.00

300.00

3 7 24

ch

30

D8

(µ

g/m

L)

0.00

1.00

2.00

3.00

Predose 3 7 24

Mo

use

CX

CL

12

α (n

g/m

L)

BA

Sampling time (h)

Sampling time (h)

0.125 µg/mL

0.1

1

10

100

1,000

0 1 2 3 4 5 6 7 8

30 mg/kg; i.p.

Sampling time (d)

Ha

mste

r Ig

G1

(µ

g/m

L)

10

100

1,000

0 2 4 6 8 10 12 14

Sampling time (d)

hu

30

D8

(µ

g/m

L)

DC

0.01

0.1

1

10

100

1,000

10,000

0 2 4 6 8 10

30 mg/kg; i.v.

Sampling time (d)

0.063 µg/mL hu

30

D8

(µ

g/m

L)

Hu

ma

n I

gG

1 (

µg

/mL

)

127.5 µg/mL

E

0.1

1

10

100

1,000

0.01 0.042 0.167 0.333 1 3 7

30D8

30D8D95A

30D8Y100A

Sampling time (d)

Figure 6. Pharmacokinetic profiles of 30D8/hu30D8 in mice, rats, and cynomolgus monkeys. A, 30D8 serum concentration–time profiles after a singleintraperitoneal administration in Balb/c nude mice at 30 mg/kg (n ¼ 3/time point with a total of n ¼ 9/group). Serum was collected at various time points,up to 7 days after 30D8 administration. Total hamster IgG1 levels were measured by a generic hamster antibody ELISA. B, free mouse/hamster chimera30D8 (ch30D8) level and free mouse CXCL12a levels in serum at various time points after a single intraperitoneal dosing at 30 mg/kg in Ncr nude mice(n ¼ 5/group). After 24 hours, the free CXCL12a levels were comparable with those present in the predosing samples. C, hu30D8 serum concentration–timeprofiles in nude rats after a single intravenous administration at 30mg/kg (n¼ 3/group). Serawere collected for up to 14 days. D, hu30D8 e serum concentration–time profiles in cynomolgusmonkeys after a single intravenous administration at 30mg/kg (n¼ 5/group). Sera were collected for up to 10 days. hu30D8 serumconcentration showed in (C) and (D) was measured by a specific ELISA using human CXCL12a as a coating agent. E, serum concentration-time profilesafter single intravenous dose of hu30D8 and two hu30D8 mutants at 10 mg/kg in Balb c/nude mice (n ¼ 3/time point with total of n ¼ 12/group). Serum wascollected up to 7 days. Antibody levels were measured by a generic human immunoglobulin G (IgG) ELISA. Data are expressed as mean � SD.






CD8þ T lymphocyte–mediated immune responses wereimportant for antitumor activity, then immunodeficientmice would not be expected to fully reveal the efficacy ofhu30D8.

In the present study, we observed fast clearance of30D8 in mice and cynomolgus monkeys. On the basisof data using hu30D8 mutants, fast clearance seemed tocorrelate well with affinity of the antibody towardCXCL12a. A preliminary screen using a library of 1,000human secreted proteins indicated that hu30D8 does notnonspecifically bind to other secreted proteins (Unpub-lished Data), arguing against the possibility that nonspe-cific binding is a reason for the fast clearance. Likewise,hu30D8 was classified as "low risk" for fast clearance inan assay of nonspecific binding shown to be useful foridentifying antibodies likely to show fast clearance incynomolgus monkeys (51). Notably, the crystal structureof hu30D8/human CXCL12a complex revealed thathu30D8 only partially blocks heparin-binding sites onCXCL12a, raising the possibility that target-mediatedclearance could be related to heparin binding byCXCL12a. Therefore, balancing pharmacokinetic prop-erties and binding affinity seems to be a requirement foroptimal efficacy of the antibody. Further studies areneeded to unravel the mechanism(s) of fast clearance.

In conclusion, hu30D8 is a suitable tool to test thehypothesis that targeting CXCL12 is a valid strategy to treatcancer and inflammatory diseases in humans. The avail-ability of predictive biomarkers to select patientsmost likelyresponsive to anti-CXCL12 therapy would be a major steptoward the initiation of such human studies.

Disclosure of Potential Conflicts of InterestJ. Wang is an employee and shareholder of Genentech/Roche. H.

Xiang is employed as Scientist in Roche and has ownership interest(including patents) in the same. L.A. Damico-Beyer is employed asSenior Scientist in Genentech and has ownership interest (includingpatents) in Roche. R.D. Carano and R.F. Kelley have ownership interest(including patents) in Roche. N. Ferrara is a former employee of Gen-

entech/Roche. No potential conflicts of interest were disclosed by theother authors.

Authors' ContributionsConception and design: C. Zhong, J. Wang, B. Li, H. Xiang, R. Corpuz,R. Takkar, W.P. Lee, R.F. Kelley, N. FerraraDevelopment ofmethodology:C. Zhong, B. Li, H. Xiang, R. Corpuz, J. Yao,R. Takkar, W.P. Lee, L.A. Damico-Beyer, R.D. Carano, C. Adams, W. WangAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.):C. Zhong,M.Ultsch,M. Coons, T.Wong, S. Clark,R. Clark, L. Quintana, P. Gribling, E. Suto, J. Yao, W.P. Lee, W. WangAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): C. Zhong, B. Li, H. Xiang, L. Quintana,K. Barck, J. Yao, R. Takkar, W.P. Lee, L.A. Damico-Beyer, R.D. Carano, R.F.Kelley, W. WangWriting, review, and/or revision of the manuscript: C. Zhong, J. Wang,B. Li, H. Xiang, M. Ultsch, T. Wong, S. Clark, R. Corpuz, L.A. Damico-Beyer,R.F. Kelley, W. Wang, N. FerraraAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructingdatabases):C. Zhong, B. Li,H. Xiang,M.Ultsch,T. Wong, J. Yao, W. WangStudy supervision: C. Zhong, H. Xiang, W.P. Lee, N. FerraraCloning the variable heavy and light chain sequence from the hybrid-oma cell line using 50 RACE: N.Y. ChiangConfirmation of the correct sequence cloned and reformatting intoexpression vectors for mammalian cells: N.Y. Chiang

AcknowledgmentsThe authors thank A. Bruce for graphic artwork and I. Hotzel for

sequence alignment; M. Kowanetz for help with the 4T1 model; and M.Dennis and X. Qu for reading the article. The authors also thank severalpeople who provided experimental support or participated in scientificdiscussions: S. William, G. Zhuang, J. Marik, D. Kallop, Y. Xia, G. Meng,W.L. Wong, K. Katschke, F. Peale, P. Luan, C. Quan, J. Lee, Y. Gan, C.Brown, C. Spiess, L. Gonzales Jr, S. Ranjani Ramani, A. Chuntharapai, R.Vandlen, P. Hass, M. Nagel, M. Schweiger, J. Zavala-Solorio, F. Chu, R.Liu, S. Krycia. We acknowledge the use of synchrotron X-ray sources atStanford Synchrotron Radiation Lightsource, supported by the Depart-ment of Energy’s Office of Science and Office of Biological and Envi-ronmental Research, and by the National Institutes of Health. The crystalstructure of hu30D8 Fab/CXCL12a complex has been deposited intoProtein Data Bank, accession code 4LMQ.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received April 4, 2013; revised June 3, 2013; accepted June 5, 2013;published OnlineFirst June 28, 2013.

References1. Kryczek I, Wei S, Keller E, Liu R, ZouW. Stroma-derived factor (SDF-1/

CXCL12) and human tumor pathogenesis. Am J Physiol Cell Physiol2007;292:C987–95.

2. Nagasawa T, Hirota S, Tachibana K, Takakura N, Nishikawa S, Kita-mura Y, et al. Defects of B-cell lymphopoiesis and bone-marrowmyelopoiesis inmice lacking theCXCchemokinePBSF/SDF-1.Nature1996;382:635–8.

3. Tachibana K, Hirota S, Iizasa H, Yoshida H, Kawabata K, Kataoka Y,et al. The chemokine receptor CXCR4 is essential for vascularization ofthe gastrointestinal tract. Nature 1998;393:591–4.

4. Sierro F, Biben C, Martinez-Munoz L, Mellado M, Ransohoff RM, Li M,et al. Disrupted cardiac development but normal hematopoiesis inmice deficient in the second CXCL12/SDF-1 receptor, CXCR7. ProcNatl Acad Sci U S A 2007;104:14759–64.

5. Williams DA, Zheng Y, Cancelas JA. Rho GTPases and regulation ofhematopoietic stem cell localization. Methods Enzymol 2008;439:365–93.

6. Levoye A, Balabanian K, Baleux F, Bachelerie F, Lagane B. CXCR7heterodimerizes with CXCR4 and regulates CXCL12-mediated Gprotein signaling. Blood 2009;113:6085–93.

7. Lerman OZ, Greives MR, Singh SP, Thanik VD, Chang CC, Seiser N,et al. Low-dose radiation augments vasculogenesis signaling throughHIF-1-dependent and -independent SDF-1 induction. Blood 2010;116:3669–76.

8. Du R, Lu KV, Petritsch C, Liu P, Ganss R, Passegue E, et al. HIF1alphainduces the recruitment of bone marrow-derived vascular modulatorycells to regulate tumor angiogenesis and invasion. Cancer Cell2008;13:206–20.

9. Kojima Y, Acar A, Eaton EN, Mellody KT, Scheel C, Ben-Porath I,et al. Autocrine TGF-beta and stromal cell-derived factor-1 (SDF-1) signaling drives the evolution of tumor-promoting mammarystromal myofibroblasts. Proc Natl Acad Sci U S A 2010;107:20009–14.

10. Orimo A, Gupta PB, Sgroi DC, Arenzana-Seisdedos F, Delaunay T,Naeem R, et al. Stromal fibroblasts present in invasive human breastcarcinomaspromote tumorgrowthandangiogenesis throughelevatedSDF-1/CXCL12 secretion. Cell 2005;121:335–48.

11. Stratman AN, Davis MJ, Davis GE. VEGF and FGF prime vascular tubemorphogenesis and sprouting directed by hematopoietic stem cellcytokines. Blood 2011;117:3709–19.

Zhong et al.





12. Bachelder RE, Wendt MA, Mercurio AM. Vascular endothelial growthfactor promotes breast carcinoma invasion in an autocrine manner byregulating the chemokine receptor CXCR4. Cancer Res 2002;62:7203–6.

13. Petit I, Jin D, Rafii S. The SDF-1-CXCR4 signaling pathway: amolecular hub modulating neo-angiogenesis. Trends Immunol2007;28:299–307.

14. Jin DK, Shido K, Kopp HG, Petit I, Shmelkov SV, Young LM, et al.Cytokine-mediated deployment of SDF-1 induces revascularizationthrough recruitment of CXCR4(þ) hemangiocytes. Nat Med 2006;12:557–67.

15. Ferrara N. Role of myeloid cells in vascular endothelial growth factor-independent tumor angiogenesis.CurrOpinHematol 2010;17:219–24.

16. Maksym RB, Tarnowski M, Grymula K, Tarnowska J, Wysoczynski M,Liu R, et al. The role of stromal-derived factor-1–CXCR7 axis indevelopment and cancer. Eur J Pharmacol 2009;625:31–40.

17. Ebos JM, Lee CR, Cruz-Munoz W, Bjarnason GA, Christensen JG,Kerbel RS. Accelerated metastasis after short-term treatment with apotent inhibitor of tumor angiogenesis. Cancer Cell 2009;15:232–9.

18. Chung AS, Kowanetz M, Wu X, Zhuang G, Ngu H, Finkle D, et al.Differential drug class-specific metastatic effects following treatmentwith a panel of angiogenesis inhibitors. J Pathol 2012;227:404–16.

19. Kioi M, Vogel H, Schultz G, Hoffman RM, Harsh GR, Brown JM.Inhibition of vasculogenesis, but not angiogenesis, prevents the recur-rence of glioblastoma after irradiation in mice. J Clin Invest 2010;120:694–705.

20. Baker KN, Rendall MH, Patel A, Boyd P, Hoare M, Freedman RB, et al.Rapid monitoring of recombinant protein products: a comparison ofcurrent technologies. Trends Biotechnol 2002;20:149–56.

21. Kowanetz M, Wu X, Lee J, Tan M, Hagenbeek T, Qu X, et al. Gran-ulocyte-colony stimulating factor promotes lung metastasis throughmobilization of Ly6GþLy6Cþ granulocytes. Proc Natl Acad Sci U S A2010;107:21248–55.

22. Liu BY, Soloviev I, Chang P, Lee J, Huang X, Zhong C, et al. Stromalcell–derived factor-1/CXCL12 contributes to MMTV-Wnt1 tumorgrowth involving Gr1þCD11bþ cells. PLoS ONE 2010;5:e8611.

23. De Clercq E. The AMD3100 story: the path to the discovery of a stemcell mobilizer (Mozobil). Biochem Pharmacol 2009;77:1655–64.

24. Crawford Y, Kasman I, Yu L, Zhong C,Wu X,Modrusan Z, et al. PDGF-C mediates the angiogenic and tumorigenic properties of fibroblastsassociatedwith tumors refractory to anti-VEGF treatment. Cancer Cell2009;15:21–34.

25. Shojaei F, Wu X, Zhong C, Yu L, Liang XH, Yao J, et al. Bv8 regulatesmyeloid-cell–dependent tumour angiogenesis. Nature 2007;450:825–31.

26. LiangWC,Wu X, Peale FV, Lee CV, Meng YG, Gutierrez J, et al. Cross-species vegf-blocking antibodies completely inhibit the growth ofhuman tumor xenografts and measure the contribution of stromalvegf. J Biol Chem 2006;281:951–61.

27. Prewett M, Huber J, Li Y, Santiago A, O'Connor W, King K, et al.Antivascular endothelial growth factor receptor (fetal liver kinase 1)monoclonal antibody inhibits tumor angiogenesis. Cancer Res1999;59:5209–18.

28. Iwamoto T, OkamotoH, ToyamaY,Momohara S.Molecular aspects ofrheumatoid arthritis: chemokines in the joints of patients. FEBS J2008;275:4448–55.

29. Katschke KJ Jr, Helmy KY, SteffekM, Xi H, Yin J, LeeWP, et al. A novelinhibitor of the alternative pathway of complement reverses inflam-mation and bone destruction in experimental arthritis. J Exp Med2007;204:1319–25.

30. Barck KH, Lee WP, Diehl LJ, Ross J, Gribling P, Zhang Y, et al.Quantification of cortical bone loss and repair for therapeuticevaluation in collagen-induced arthritis, by micro-computed tomog-raphy and automated image analysis. Arthritis Rheum 2004;50:3377–86.

31. CampaC,Kasman I, YeW,LeeWP, FuhG, FerraraN. Effects of an anti-VEGF-A monoclonal antibody on laser-induced choroidal neovascu-

larization in mice: optimizing methods to quantify vascular changes.Invest Ophthalmol Vis Sci 2008;49:1178–83.

32. Sengupta N, Caballero S, Mames RN, Butler JM, Scott EW, Grant MB.The role of adult bone marrow-derived stem cells in choroidal neo-vascularization. Invest Ophthalmol Vis Sci 2003;44:4908–13.

33. Eter N, Engel DR, Meyer L, Helb HM, Roth F, Maurer J, et al. In vivovisualization of dendritic cells, macrophages, and microglial cellsresponding to laser-induced damage in the fundus of the eye. InvestOphthalmol Vis Sci 2008;49:3649–58.

34. Jung S, Aliberti J, Graemmel P, Sunshine MJ, Kreutzberg GW, Sher A,et al. Analysis of fractalkine receptor CX(3)CR1 function by targeteddeletion and green fluorescent protein reporter gene insertion.Mol CellBiol 2000;20:4106–14.

35. Carter P, Kelley RF, Rodrigues ML, Snedecor B, Covarrubias M,Velligan MD, et al. High level Escherichia coli expression and produc-tion of a bivalent humanized antibody fragment. Biotechnology 1992;10:163–7.

36. Foote J, Winter G. Antibody framework residues affecting the confor-mation of the hypervariable loops. J Mol Biol 1992;224:487–99.

37. Wu TT, Kabat EA. An analysis of the sequences of the variable regionsof Bence Jones proteins and myeloma light chains and their implica-tions for antibody complementarity. J Exp Med 1970;132:211–50.

38. Al-Lazikani B, Lesk AM, Chothia C. Standard conformations for thecanonical structures of immunoglobulins. JMol Biol 1997;273:927–48.

39. Tejada ML, Yu L, Dong J, Jung K, Meng G, Peale FV, et al. Tumor-driven paracrine platelet-derived growth factor receptor alpha signal-ing is a key determinant of stromal cell recruitment in amodel of humanlung carcinoma. Clin Cancer Res 2006;12:2676–88.

40. Ryu EK, Kim TG, Kwon TH, Jung ID, Ryu D, Park YM, et al. Crystalstructure of recombinant human stromal cell-derived factor-1alpha.Proteins 2007;67:1193–7.

41. Murphy JW, Cho Y, Sachpatzidis A, Fan C, Hodsdon ME, Lolis E.Structural and functional basis of CXCL12 (stromal cell-derived factor-1 alpha) binding to heparin. J Biol Chem 2007;282:10018–27.

42. Crump MP, Gong JH, Loetscher P, Rajarathnam K, Amara A, Are-nzana-Seisdedos F, et al. Solution structure and basis for functionalactivity of stromal cell-derived factor-1; dissociation of CXCR4 acti-vation from binding and inhibition of HIV-1. EMBO J 1997;16:6996–7007.

43. Gravel S,MaloufC, Boulais PE,BerchicheYA,Oishi S, Fujii N, et al. Thepeptidomimetic CXCR4 antagonist TC14012 recruits beta-arrestin toCXCR7: roles of receptor domains. J Biol Chem 2010;285:37939–43.

44. Duda DG, Kozin SV, Kirkpatrick ND, Xu L, Fukumura D, Jain RK.CXCL12 (SDF1alpha)-CXCR4/CXCR7 pathway inhibition: an emerg-ing sensitizer for anticancer therapies? Clin Cancer Res 2011;17:2074–80.

45. Peled A, Wald O, Burger J. Development of novel CXCR4-basedtherapeutics. Expert Opin Investig Drugs 2012;21:341–53.

46. Phillips RJ, Burdick MD, Lutz M, Belperio JA, Keane MP, Strieter RM.The stromal derived factor-1/CXCL12-CXC chemokine receptor 4biological axis in non–small cell lung cancer metastases. Am J RespirCrit Care Med 2003;167:1676–86.

47. Zhuo W, Jia L, Song N, Lu XA, Ding Y, Wang X, et al. The CXCL12–CXCR4 chemokine pathway: a novel axis regulates lymphangiogen-esis. Clin Cancer Res 2012;18:5387–98.

48. Sloane AJ, Raso V, Dimitrov DS, Xiao X, Deo S,Muljadi N, et al. Markedstructural and functional heterogeneity in CXCR4: separation of HIV-1and SDF-1alpha responses. Immunol Cell Biol 2005;83:129–43.

49. Baribaud F, Edwards TG,SharronM,Brelot A,HevekerN, PriceK, et al.Antigenically distinct conformations of CXCR4. J Virol 2001;75:8957–67.

50. Kalatskaya I, Berchiche YA, Gravel S, Limberg BJ, Rosenbaum JS,Heveker N. AMD3100 is a CXCR7 ligand with allosteric agonistproperties. Mol Pharmacol 2009;75:1240–7.

51. Hotzel I, Theil F-P, Bernstein LJ, Prabhu S, Deng R, Quintana L, et al. Astrategy for risk mitigation of antibodies with fast clearance. MAbs2012;4:753–60.






2013;19:4433-4445. Published OnlineFirst June 28, 2013.Clin Cancer Res Cuiling Zhong, Jianyong Wang, Bing Li, et al. Antibody Targeting CXCL12Development and Preclinical Characterization of a Humanized

Updated version

10.1158/1078-0432.CCR-13-0943doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2013/06/27/1078-0432.CCR-13-0943.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/19/16/4433.full#ref-list-1

This article cites 51 articles, 23 of which you can access for free at:

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

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/19/16/4433To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-13-0943

http://clincancerres.aacrjournals.org/content/suppl/2013/06/27/1078-0432.CCR-13-0943.DC1

http://clincancerres.aacrjournals.org/content/19/16/4433.full#ref-list-1

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/19/16/4433


development and preclinical characterization of a humanized … · hamster immunization and splenic...

Documents