Báo cáo khoa học: Differential binding of human immunoagents and Herceptin to the ErbB2 receptor ppt

Abbreviations EDIA, Erbicin-derived immunoagent; ErbB2-ECD, extracellular domain of ErbB2 receptor; ERB-hcAb, human compact antibody against ErbB2 receptor; ERB-hRNase, human anti-ErbB2

Differential binding of human immunoagents and

Herceptin to the ErbB2 receptor

Fulvia Troise1,*, Valeria Cafaro1,*, Concetta Giancola2, Giuseppe D’Alessio1and

Claudia De Lorenzo1

1 Dipartimento di Biologia Strutturale e Funzionale, Universita` di Napoli Federico II, Italy

2 Dipartimento di Chimica, Universita` di Napoli Federico II, Italy

ErbB2 (HER2⁄ neu) is a proto-oncogene of the erbB

family of tyrosine kinase receptors [1] It encodes a

185 kDa transmembrane protein, which comprises an

extracellular domain (ECD) and an intracellular

tyro-sine kinase activity Although no natural ligand has

been identified for this receptor, it has been

ascer-tained that its overexpression is associated with

various carcinomas, in particular with human breast

cancer [2] As ErbB2 overexpression is involved in the

progression of the malignancy, and is a sign of a poor

prognosis, ErbB2 is a valid target of therapeutic inter-vention

However, when ErbB2 is overexpressed, not all of the ErbB2-ECD protein is embedded in the membrane

of malignant cells; a fraction of ErbB2-ECD is proteo-lytically removed from the receptor [3] and shed as a soluble protein in the sera of breast cancer patients [4] Herceptin [5], a humanized anti-ErbB2 IgG1, has been proven to be an essential tool in the immunother-apy of breast carcinoma However, some

ErbB2-posi-Keywords

binding affinity; ErbB2; herceptin;

immunoRNase; immunotherapy

Correspondence

C De Lorenzo, Dipartimento di Biologia

Strutturale e Funzionale, Universita` di Napoli

Federico II, Via Cinthia, 80126 Naples, Italy

Fax: +39081679159

Tel: +39081679158

E-mail: cladelor@unina.it

*These authors contributed equally to this

work

(Received 9 June 2008, revised 29 July

2008, accepted 1 August 2008)

doi:10.1111/j.1742-4658.2008.06625.x

Overexpression of the ErbB2 receptor is associated with the progression of breast cancer, and is a sign of a poor prognosis Herceptin, a humanized antibody directed to the ErbB2 receptor, has been proven to be effective in the immunotherapy of breast cancer However, it can result in cardiotoxicity, and a large fraction of breast cancer patients are resistant to Herceptin treat-ment We have engineered three novel, fully human, anti-ErbB2 immuno-agents: Erbicin, a human single-chain antibody fragment; ERB-hRNase, a human immunoRNase composed of Erbicin fused to a human RNase; ERB-hcAb, a human ‘compact’ antibody in which two Erbicin molecules are fused to the Fc fragment of a human IgG1 Both ERB-hRNase and ERB-hcAb strongly inhibit the growth of ErbB2-positive cells in vivo The interactions of the Erbicin-derived immunoagents and Herceptin with the extracellular domain of ErbB2 (ErbB2-ECD) were investigated for the first time by three different methods Erbicin-derived immunoagents bind soluble extracellular domain with a lower affinity than that measured for the native antigen on tumour cells Herceptin, by contrast, shows a higher affinity for soluble ErbB2-ECD Accordingly, ErbB2-ECD abolished the in vitro anti-tumour activity of Herceptin, with no effect on that of Erbicin-derived immu-noagents These results suggest that the fraction of immunoagent neutralized

by free extracellular domain shed into the bloodstream is much higher for Herceptin than for Erbicin-derived immunoagents, which therefore may be used at lower therapeutic doses than those employed for Herceptin

Abbreviations

EDIA, Erbicin-derived immunoagent; ErbB2-ECD, extracellular domain of ErbB2 receptor; ERB-hcAb, human compact antibody against ErbB2 receptor; ERB-hRNase, human anti-ErbB2 immunoRNase with Erbicin fused to human pancreatic RNase; ITC, isothermal titration

calorimetry; RU, response unit; scFv, single-chain antibody fragment; SPR, surface plasmon resonance.

tive carcinomas are resistant to the inhibitory effect on

growth of Herceptin [6], and, in other patients, the

resistance of malignant cells is induced at a later stage

in treatment [7] Furthermore, it has been found that

Herceptin can lead to cardiotoxicity in a significant

fraction of treated patients [8,9] An alternative

approach to the use of Herceptin in immunotherapy

has been promoted, based on the administration of

Herceptin combined with other antibodies directed to

the ErbB2 receptor [10,11] A prerequisite for this

strategy is that the latter antibodies are directed to

epi-topes on ErbB2-ECD different from that recognized

by Herceptin

Based on these considerations, we instituted a search

for novel immunoagents directed to epitopes different

from that recognized by Herceptin, with no cardiotoxic

side-effects and able to fulfil the therapeutic need of

Herceptin-unresponsive patients This led us to the

production of a novel, fully human, anti-ErbB2

single-chain antibody fragment (scFv), isolated from a large

phage display library through a double selection

strat-egy performed on live cells This scFv, named Erbicin

[12], specifically binds to ErbB2-positive cells, inhibits

receptor autophosphorylation and is internalized by

target cells Erbicin was used to construct human

anti-ErbB2 immunoagents by two different strategies The

first was based on Erbicin fused to an RNase, i.e a

pro-toxin, as RNase becomes toxic only when Erbicin

promotes its internalization in target cells An

immun-oRNase, denoted as ERB-hRNase

(Erbicin-human-RNase), was produced by the fusion of Erbicin to

human pancreatic RNase [13]

The second strategy aimed to produce a therapeutic

reagent with an increased half-life, prolonged tumour

retention and an ability to recruit host effector

func-tions Erbicin was thus fused to the Fc region from a

human IgG1 to obtain an immunoglobulin-like

anti-body version [14,15] The engineered antianti-body was

denoted as ERB-hcAb (human anti-ErbB2-compact

antibody) because of its ‘compact’ size (100 kDa)

com-pared with the full size (155 kDa) of a natural IgG

Both Erbicin-derived immunoagents (EDIAs) were

found to selectively and strongly inhibit the growth of

ErbB2-positive cells, both in vitro and in vivo

How-ever, to define and implement the antitumour potential

of Erbicin and EDIAs, we deemed it essential to study

their interaction with ErbB2 To determine and

evalu-ate quantitatively their affinity for ErbB2, we used

recombinant ErbB2-ECD as a homogeneous, soluble

antigen With this aim, three different analytical

meth-ods were employed: ELISA, surface plasmon

reso-nance (SPR) and isothermal titration calorimetry

(ITC) The results obtained with Erbicin and EDIAs

were compared with the results obtained with Hercep-tin Furthermore, we determined and compared the affinity values of Herceptin and EDIAs for the free ECD structured within the whole receptor molecule, natively inserted into the cell membrane, with the val-ues measured using isolated ECD

We found that EDIAs bound soluble ECD with an affinity lower than that of Herceptin However, the novel EDIAs bound ErbB2 exposed on the cell surface with a higher affinity than that of Herceptin [13,14] These results indicate that the fraction of immunoagent neutralized by free ECD shed into the bloodstream, and hence lost to immunotherapy, could be much higher for Herceptin than for the novel immunoagents

Results

Production and characterization of ErbB2-ECD The cDNA coding for ErbB2-ECD was stably trans-fected in the 293 cell line The encoded protein was expressed as a secretion product in the culture med-ium, as revealed by western blotting (Fig 1A) and immunoprecipitation analyses performed (see Experi-mental procedures) with ERB-hcAb or Herceptin as anti-ErbB2 agent (see Fig 1B) The final yield of ErbB2-ECD, purified by affinity chromatography (see Experimental procedures), was 12 mgÆL)1 of medium The protein was analysed by SDS-PAGE, followed

by Coomassie staining and western blotting with Her-ceptin or ERB-hcAb (Fig 1C) Its molecular size was about 80 kDa, as expected

Analysis by ELISA of the interactions of EDIAs and Herceptin with soluble ErbB2-ECD

ELISA sandwich assays were performed to determine the ability of Erbicin and EDIAs to recognize soluble ErbB2-ECD Herceptin fixed on the microplate was used to capture ErbB2-ECD, which, in turn, could inter-act with the anti-ErbB2 immunoagents The affinity of ERB-hcAb or Herceptin for ErbB2-ECD was measured

by ELISA on ECD directly coated to the wells

The results are given in Table 1 as apparent binding constants, measured from the binding curves as the concentrations corresponding to half-maximal sat-uration

The values obtained (50 nm for Erbicin, 30 nm for ERB-hRNase and 7 nm for ERB-hcAb) were found to

be higher than those obtained with ErbB2 embedded

in ErbB2-positive cells [13,14] Thus, these data indi-cate that the immunoagents have a higher affinity for ErbB2-ECD when it is inserted in the cell membrane

Interestingly, the lower binding affinity of EDIAs

for soluble ErbB2-ECD was not shared by Herceptin,

which displayed a high affinity for soluble ErbB2-ECD

(0.1 nm), about 50-fold higher than that determined when Herceptin was tested with ErbB2-ECD expressed

on live cells (see Table 1) These findings can be explained by the fact that parent Erbicin was selected from a phage library using ErbB2-ECD inserted into ErbB2-positive cells, whereas, for the isolation of Herceptin, free, soluble ECD was used [16]

ELISA sandwich assays with Herceptin as a capture agent have been performed to confirm that Erbicin and the novel immunoconjugates recognize,

on ErbB2-ECD, an epitope different from that selected

by Herceptin, as reported previously [17]

However, this type of assay was carried out for Erbicin and the immunoRNase only, as the peroxi-dase-conjugated anti-His IgG1 capable of revealing scFv and Erb-hRNase is unaffected by the presence of Herceptin; it was not performed with Erb-hcAb, as the anti-human secondary IgG serum fraction, used for its detection, could not discriminate between Erb-hcAb and Herceptin Thus, for Erb-hcAb and Herceptin, the assays were performed on ECD directly immobilized

on the plate

We then tested whether soluble ErbB2-ECD could affect the binding of anti-ErbB2 immunoagents to ErbB2-positive cells by performing ELISA with ERB-hcAb or Herceptin in the absence or presence of free ECD Each antibody was tested at increasing concentrations, with soluble ECD added either in equimolar amounts or in a 10-fold molar excess to the number of receptor molecules on the cell membrane [18] As a control, parallel assays were carried out in the absence of ErbB2-ECD

As shown in Fig 2A, the binding curves obtained for ERB-hcAb in the absence or presence of soluble ECD were found to be superimposable This finding suggests that the binding ability of ERB-hcAb to ErbB2-positive cells is unaffected by the presence of soluble ECD In contrast, the binding of Herceptin to ErbB2-positive cells (Fig 2B) was strongly reduced by ECD used at a 1 : 1 ratio with the receptor number, and fully inhibited with a 10-fold molar excess of ECD These results, in line with those described above

on the high affinity of Herceptin for soluble ECD, indicate that, for Herceptin, there is a favourable com-petition of soluble ErbB2-ECD over ECD on the cell membrane, whereas there is no detectable competition

in the case of ERB-hcAb

Effects of soluble ErbB2-ECD on the cytotoxicity

of ERB-hcAb and Herceptin

On the basis of the results discussed above, the antitumour effects of ERB-hcAb and Herceptin on

1

A

B

C

ERB-hcAb (100 kDa)

1

Herceptin (155 kDa)

2

80 kDa

ECD (80 kDa)

ECD (80 kDa)

80 kDa

2

2

Fig 1 Detection of ErbB2-ECD expression (A) Western blotting

analyses from conditioned medium of transfected 293 cells, with

Herceptin as primary antibody followed by horseradish

peroxidase-conjugated anti-human (Fc-specific) IgG serum fraction Lane 1,

negative control (medium from non-transfected 293 cells); lanes

2–4, conditioned medium produced by various selected clones (B)

Immunoprecipitation analyses of ErbB2-ECD from 293 cell

condi-tioned medium with ERB-hcAb (lane 1) or Herceptin (lane 2) (C)

SDS-PAGE analyses of purified ErbB2-ECD Lane 1, molecular

weight standards; lane 2, ErbB2-ECD eluted from immunoaffinity

chromatography stained with Coomassie blue; lanes 3 and 4,

wes-tern blot analyses of the sample in lane 2 using ERB-hcAb and

Her-ceptin, respectively, as anti-ErbB2-ECD immunoagents.

Table 1 Relative affinity of Erbicin, EDIAs and Herceptin for

solu-ble ErbB2-ECD, as measured by ELISAs Previous data [13,14]

obtained with ErbB2-positive cells are also shown.

K D (apparent) (n M )

ErbB2-ECD

ErbB2-positive cells

ErbB2-positive cells were tested in the absence or

pres-ence of soluble ErbB2-ECD Antibodies were

incu-bated with soluble ECD, added at a concentration of

20 nm (eight-fold molar excess over antibodies), which

was chosen on the basis of ELISA conditions in which

Herceptin binding to ErbB2-positive cells was fully

inhibited (Fig 2B)

As shown in Fig 2C, ERB-hcAb inhibited the growth of SKBR3 cells similarly in the absence or presence of soluble ErbB2-ECD In contrast, soluble ErbB2-ECD completely abolished the antitumour activity of Herceptin

These results indicate that, in the presence of soluble ECD, ERB-hcAb preserves its cytotoxic power on ErbB2-positive cells, whereas Herceptin does not exert cytotoxic activity because of its high affinity for solu-ble ECD ECD is capasolu-ble of neutralizing antibody binding to the cells, in agreement with previously reported data [19]

Analyses by SPR of the interactions of EDIAs and Herceptin with ErbB2-ECD

To compare the binding properties of Erbicin, EDIAs and Herceptin with ErbB2-ECD using a direct meth-odology based on physicochemical principles, SPR analyses were carried out The experimental system consisted of ErbB2-ECD (a monovalent ligand) cova-lently immobilized on the chip surface, with monova-lent (Erbicin or ERB-hRNase) or bivamonova-lent (Herceptin

or ERB-hcAb) analytes injected and flowing over the sensor chip

The kinetic constants for monovalent Erbicin and ERB-hRNase were obtained by fitting the curves with

a 1 : 1 interaction model Similar binding curves were recorded for these immunoagents (see Fig 3A,B), with almost identical association rate constants, but slightly different dissociation rate constants (see Table 2) Erbicin, with a kd value of 6.16· 10)3s)1, dissociated from ErbB2-ECD 1.5 times faster than did ERB-hRNase (kd= 4.12· 10)3s)1) This indicated a higher stability for the ERB-hRNase⁄ ErbB2-ECD complex with respect to the Erbicin⁄ ErbB2-ECD com-plex, with equilibrium KDvalues of 46.7 and 27.2 nm, respectively The significant difference in KD values could be clearly ascribed to the lower dissociation rate constant measured for the ERB-hRNase⁄ ErbB2-ECD complex It should be noted that the data were in very good agreement with those reported above from the ELISA experiments (see Table 1)

The possibility was considered that the higher stabil-ity of the ERB-hRNase⁄ ErbB2-ECD complex might

be caused by aspecific electrostatic interactions between the positively charged RNase linked in the immunoconjugate and the negatively charged carb-oxymethyl-dextran matrix of the SPR chip Thus, the kinetic analyses of the ERB-hRNase⁄ ErbB2-ECD complex were repeated in the presence of soluble carboxymethyl-dextran as an added quencher How-ever, identical constants were measured for the

1.5

2

A

B

C

0

0.5

1

2

2.5

Protein concentration (n M )

0

0.5

1

1.5

40

Protein concentration (n M )

10

20

30

0

Control

Herceptin Herceptin + ECD

ERB-hcAb ERB-hcAb + ECD

Fig 2 Effects of soluble ErbB2-ECD on the binding and

cytotoxic-ity of ERB-hcAb and Herceptin Binding curves of ERB-hcAb (A)

and Herceptin (B) to SKBR3 cells obtained by ELISA performed in

the absence (open symbols) or presence (filled symbols) of soluble

ECD Soluble ECD was added at a ratio of 1 : 1 (filled squares) or

10 : 1 (filled circles) to the number of receptor molecules on the

cell membrane (C) Antitumour activity of ERB-hcAb and Herceptin

on SKBR3 cells determined in the absence or presence of soluble

ECD.

ERB-hRNase⁄ ErbB2-ECD complex in the presence or

absence of soluble carboxymethyl-dextran (Table 2)

This indicates that the higher stability of the

ERB-hRNase⁄ ErbB2-ECD complex is not caused by simple coulombic interactions with the non-immune moiety, but by its specific structural features which

120

60

80

40

0

40

20

0

0 100 200 300 400 500

160

80

A B

C D

E F

60

0 100 200 300 400 500

120

40

80

0

40

20

0

0 100 200 300 400 500

6

100

120

0 100 200 300 400 500

450

550

300

350

400

0

2

4

0 20 40 60 80 100 120

Req (RU)

Req

/ERB-hcAb (RU/n

0

20

40

60

80

Req

-50

50

150

250

350

200 250 300 350

Req (RU)

Req

/Herceptin (RU/n

0

50

100

150

200

250

Req

0 50 100 150 200 250 300 350 400

[ERB-hcAb] (n M )

0 50 100 150 200 250 300 350 400

[Herceptin] (n M )

Fig 3 Determination by SPR of the binding between anti-ErbB2 immunoagents and ErbB2-ECD Representative sensorgrams (jagged grey lines) recorded for Erbicin (A), ERB-hRNase (B), ERB-hcAb (C) and Herceptin (D) Smooth black lines represent the global fits of the sensor-grams to a 1 : 1 bimolecular interaction model Erbicin was passed across the surface (500 RU of ErbB2-ECD) at concentrations of 10.9–

350 n M (A) and ERB-hRNase at concentrations of 8.4–269 n M (B) Soluble ErbB2-ECD was passed over ERB-hcAb (density, 202 ± 4 RU) or Herceptin (density, 1151 ± 5 RU), each captured by Protein A, at concentrations of 14.6–470 n M (C) and 22.7–728 n M (D) Binding isotherms

of ERB-hcAb (E) and Herceptin (F) to immobilized ErbB2-ECD (500 RU) The equilibrium binding data (Req) were measured directly on the sensorgrams obtained by subsequent injections of analytes, and represent the mean of two determinations The analysed concentrations were 7.6–356.6 n M for ERB-hcAb (E) and 0.5–361 n M for Herceptin (F) The equilibrium binding data were fitted to a two-site binding hyper-bola (R 2 = 0.994 and R 2 = 0.999 for ERB-hcAb and Herceptin, respectively) The insets show the Scatchard analysis of the binding data The calculated constants from these plots (KD1and KD2) are listed in Table 2.

Table 2 Affinity and rate constants for ErbB2-ECD ⁄ ligand interactions determined by SPR.

k a ( M )1Æs)1)a

ERB-hRNase (1.50 ± 0.18) · 10 5

Herceptin (7.25 ± 2.41) · 10 3

a The reported constants are average values obtained from three independent analyses using different biosensors, sample preparations and ligand densities on the flow cell surfaces The equilibrium dissociation constants (KD) were calculated from the relationship: KD= kd⁄ k a

b Equilibrium dissociation constants for the 1 : 1 complexes, calculated from the Scatchard plot analyses c Apparent affinity constants for the bivalent complexes, calculated from the Scatchard plot analyses d Reported data were measured in the presence of soluble carboxym-ethyl-dextran added to the sample at a final concentration of 5 mgÆmL)1.

lead to tighter binding in the complex, a result also

obtained for the other Erbicin-derived

immunoconju-gate, ERB-hcAb (see below)

We then studied, by equilibrium SPR analyses, the

interactions of bivalent Herceptin and ERB-hcAb,

each endowed with two identical antigen binding sites,

with immobilized ErbB2-ECD In this experimental

approach, the analytes were expected to have the

potential to bind either to a single site or bivalently

The ratio between the monovalent and bivalent

complexes would be dependent on the relative

concentrations of ligand and analyte flowing over the

surface

Equilibrium binding responses (Req) were determined

directly at increasing analyte concentrations The

anal-yses of the equilibrium binding data were carried out

using Scatchard plots with the binding models

described by Junghans [20] to count receptors and

other cell surface molecules with bivalent analytes

(IgG) The binding isotherms and corresponding

Scat-chard plots are shown in Fig 3 Biphasic ScatScat-chard

plots were obtained, which are consistent with the

bivalent binding model

The constants calculated from these plots, listed in

Table 2, highlight the different binding behaviour as a

function of analyte concentration At higher antibody

concentrations (about 45–360 nm for ERB-hcAb and

100–360 nm for Herceptin), the KD1 constants were 31

and 8.9 nm for ERB-hcAb and Herceptin, respectively

The maximum binding responses (Bmax1), which are

proportional to the moles of bound antibody, were

116 and 336 response units (RU) for ERB-hcAb and

Herceptin, respectively At lower analyte

concentra-tions (about 8–45 nm for ERB-hcAb and 0.5–10 nm

for Herceptin), the KD2 constants were 5.6 and 0.1 nm

for ERB-hcAb and Herceptin, respectively, and the

maximum binding responses (Bmax2) were 76 and

298 RU for ERB-hcAb and Herceptin, respectively

These data can be interpreted by surmising that, at

low concentrations, the antibody can simultaneously

bind two receptor molecules This reflects the high

affinity of the antibody for the receptor, as marked by

a low KD2 constant It should be noted (see Table 2)

that the KD2values are virtually identical to the

appar-ent binding constants determined by ELISA (7 and

0.1 nm for ERB-hcAb and Herceptin, respectively,

compared with 5.6 and 0.1 nm, respectively) Indeed,

the antibody concentration range explored by ELISA

was very similar to that examined by SPR

At high antibody concentrations, the crowding of

antibody molecules on the immobilized ErbB2-ECD

renders it difficult to obtain simultaneous binding of

antibody to two receptor molecules, and the binding is

mainly monovalent; this is reflected in the low affinity with a higher KD1constant

The analysis of the maximum binding values of the bivalent analytes, calculated from the Scatchard plots

at low and high analyte concentrations (see below), supports these hypotheses If bivalent binding is achieved, the maximum binding response (Bmax2) should be one-half of the maximum binding response expected for monovalent binding (Bmax1) For ERB-hcAb, Bmax2 (76 RU) was indeed about one-half

of the Bmax1 value (116 RU), in good agreement with the above-mentioned hypothesis

Furthermore, as a control experiment, a Scatchard plot analysis of the equilibrium binding data of ERB-hRNase was carried out (data not shown) This immunoagent is a monovalent analyte with a molec-ular weight of 46 kDa, about half of the ERB-hcAb molecular weight (100 kDa), and showed binding to immobilized ECD in a 1 : 1 ratio

In this case, the expected Bmaxvalue should be simi-lar to Bmax2 determined for ERB-hcAb bivalent bind-ing The calculated Bmax value for ERB-hRNase was found to be 84 RU, very similar to the value calcu-lated for the bivalent binding of ERB-hcAb (Bmax2= 76 RU), in line with the hypothesis formu-lated above The KD value calculated from the Scat-chard plot for ERB-hRNase was 24 nm (Table 2), in very good agreement with that determined by SPR kinetic experiments (27.2 nm)

However, the Bmax2value for the bivalent binding of Herceptin (298 RU) was higher than the expected one-half value of Bmax1(1⁄ 2Bmax1= 168 RU) This differ-ence could be ascribed to the lack of equilibrium response data at very low Herceptin concentrations (< 0.5 nm), required to accurately define the Scat-chard plot for bivalent binding

These results, in line with those predicted using the binding models described by Junghans [20], indicate that bivalent binding typically dominates over mono-valent binding up to very high antibody concentra-tions

To further test this hypothesis, different ECD densi-ties were immobilized on the chip surface for analysis

of the antibody affinity at equilibrium However, when

a low ECD density (260–340 RU) was used, it was not possible to record the equilibrium responses at anti-body concentrations close to the KD value; when a higher ECD density (1500–1800 RU) was used, equi-librium responses were recorded, but the maximum antibody binding values (900–1300 RU) were too high

to be reliable

It may be of interest that the equilibrium binding analyses carried out by SPR may be applied to

deter-mine both affinity constants: the equilibrium

dissocia-tion constant (KD1) for the 1 : 1 complex, an intrinsic

property of the binding site, and the apparent affinity

constant (KD2) for the bivalent complex, dependent on

steric features

It has been reported that, in mammary carcinomas,

in which ErbB2 is overexpressed, the antitumour

action of Herceptin can be neutralized in part by

bind-ing to soluble ErbB2-ECD, which is proteolytically

cleaved and shed into the patients’ sera [19] Given

the interest in the binding properties of anti-ErbB2

immunoagents (ERB-hcAb and Herceptin) to soluble

ErbB2-ECD, a different SPR system was implemented

by performing assays on the antibodies trapped on the

sensor chip (see Experimental procedures), with soluble

ErbB2-ECD freely passed over the chip

The kinetic constants for association (ka) and

disso-ciation (kd) were determined Figure 3C,D shows the

binding curves used to determine ka and kd (see

Table 2) for the complexes of ERB-hcAb or Herceptin

with ErbB2-ECD These analyses highlight the

differ-ent kinetic behaviour for the two antibodies Herceptin

binds ErbB2-ECD with a relatively low value of ka

(7.25· 103m)1Æs)1), about three-fold lower than the

value determined for the ERB-hcAb⁄ ErbB2-ECD

com-plex (1.77· 104m)1Æs)1) With regard to the

dissocia-tion step, the Herceptin⁄ ErbB2-ECD complex was

found to be much more stable, with a kd value of

6.5· 10)5s)1, about one order of magnitude lower

than the kd value of the ERB-hcAb⁄ ErbB2-ECD

complex (4.35· 10)4s)1) The calculated equilibrium

dissociation constants (KD) for Herceptin⁄ ErbB2-ECD

and ERB-hcAb⁄ ErbB2-ECD 1 : 1 complexes were

9.4 and 24.7 nm per binding site, respectively

Interestingly, these KD values were very similar to the

KD1values determined by equilibrium SPR analyses at

high antibody concentrations, when only a single

antigen binding site interacts with ErbB2-ECD (see

Table 2)

The kinetic constants for the association (ka) and

dissociation (kd) phases, determined by SPR, are

clo-sely correlated with the bivalent affinity enhancements

(avidities) relative to the monovalent interactions

(intrinsic affinities)

The avidity is usually considered as a measure of the

resistance of antibody⁄ antigen complexes to dissociation

after dilution [21] As the stability of immunocomplexes

is mainly the result of a low dissociation rate constant

value (kd), it could be predicted that the increase in

affinity of bivalent complexes in comparison with

mono-valent complexes should be inversely proportional to

the dissociation rate constant value (kd) Therefore, the

Herceptin⁄ ErbB2-ECD complex with a dissociation rate

constant (kd) of 6.5 · 10)5s)1, about one order of mag-nitude lower than the kd value of the ERB-hcAb⁄ ErbB2-ECD complex (4.35· 10)4s)1), is expected to exhibit a larger increase in affinity when bivalent binding is allowed This is in line with the findings that the apparent affinity constant for the Herceptin⁄ ErbB2-ECD bivalent complex (KD2= 0.1 nm) is 100-fold lower than the equilibrium dissociation constant (KD1= 9.6 nm) for the monovalent complex, whereas the apparent affinity constant for the ERB-hcAb⁄ ErbB2-ECD bivalent complex (KD2= 5.6 nm) is only five-fold lower than the equilibrium dissociation con-stant (KD1= 24.7 nm) for the monovalent complex Together, these data confirm once again that Herceptin binds to soluble ErbB2-ECD with a higher affinity than does ERB-hcAb, so that, in vivo, the frac-tion of Herceptin strongly sequestered into this immu-nocomplex may not be available for interactions with cell-embedded ErbB2

Analyses by ITC of the interactions of soluble EDIAs and Herceptin with soluble ErbB2-ECD Given the intriguing results obtained by studying the binding of free, soluble ErbB2-ECD to anti-ErbB2 immunoagents, we studied these interactions by ITC, another analytical tool firmly based on physicochemical principles Using this methodology, a ligand is gradually titrated against a macromolecule, thus evolving or tak-ing up measurable heat In the ITC experimental set-up, both the macromolecules, in our case ErbB2-ECD and the immunoagents, are free in solution

Figure 4 shows the results of calorimetric titrations for the interactions of the immunoagents (Erbicin, EDIAs, Herceptin) with ErbB2-ECD Exothermic heat pulses were observed after each injection of immuno-agent into the ErbB2-ECD solution (see insets in Fig 4) The integration of the heat produced per injec-tion as a funcinjec-tion of time, and conversion to per mole

of immunoagents, gave the corresponding binding iso-therms (see Fig 4) The data, plotted as a function of the molar ratio, indicate a binding stoichiometry of

1 : 1 for ERB-hRNase and Erbicin and of 0.5 : 1 for ERB-hcAb and Herceptin, i.e each identical antigen binding site binds one molecule of ErbB2-ECD The binding constants (Kb; converted for comparison to

KD) and enthalpy changes (DbH), obtained by stan-dard equations, led to the thermodynamic parameters summarized in Table 3 Their inspection reveals KD values close to those obtained with ELISA and SPR (see Tables 1, 2) for the binding of monovalent Erbicin and ERB-hRNase With regard to the complexes with bivalent ERB-hcAb and Herceptin, the KDvalues were

higher than those obtained with ELISA, but close to

those determined by SPR, based on either the kinetic

constants or equilibrium measurements analysed using

the Scatchard equation In particular, they were close

to the KD1 values calculated at high ligand

concentra-tions when monovalent binding prevails (see Table 2)

As shown by the DbH values in Table 3, binding is

driven by a favourable binding enthalpy, but opposed

by an unfavourable binding entropy change, DbS

It is of interest that, of the studied systems, Erbicin

shows the lowest affinity for ErbB2-ECD, in

agree-ment with ELISA and SPR results, and the interaction

is characterized by the lowest enthalpy change

This indicates a lower number of non-covalent

interactions on binding of EDIAs (ERB-hcAb and

ERB-hRNase), as also revealed by ELISA and SPR

(see above) However, the unfavourable DbS value recorded for these interactions indicates a greater dec-rease in conformational stability on complex formation When an attempt was made to study the interactions

of the immunoagents with live cells, i.e with ErbB2 inserted on the cell membrane, it was verified that the experimental system could, in principle, be used, with stoichiometric values identical to those obtained with ErbB2-ECD and immunoagents in solution However, surprising results were found: the DbH values were about 100-fold higher than those measured with solu-ble receptor and ligand, and the binding constants were about 1000-fold higher

To verify whether the very high DbH and affinity constants could be related to events occurring on inter-nalization of the immunoagents, the experiments were

–100

–50

0

A B

C D

0 2000 4000 6000 8000 10 000

0

1

2

3

0.0 0.5 1.0 1.5 2.0 2.5 3.0 –350

–300

–250

–200

–150

Time (s)

–1 )

0.15 0.25

–60 –30

0

2000 4000 6000 8000 10 000 –0.15

–0.05 0.00 0.10

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 –150

–120 –90

Time (s)

–1 )

–1 ) –100

–80

–60

–40

–20

0

0.6 0.9 1.2 1.5

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 –180

–160

–140

–120

[ERB-hcAb]/[ECD]

0

0 2000 4000 6000 8000 10 000 –0.3

0.0 0.3

Time (s)

–120 –100 –80 –60 –40 –20

0.3 0.6 0.9 1.2 1.5

–1 )

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 –160

–140

[Herceptin]/[ECD]

0 2000 4000 6000 8000 10 000 –0.3

0.0

Time (s)

Fig 4 Determination by ITC of the binding interactions between anti-ErbB2 immunoagents and ErbB2-ECD (ECD) Binding isotherms are shown for the titration of ErbB2-ECD with Erbicin (A), ERB-hRNase (B), ERB-hcAb (C) and Herceptin (D) The raw data are shown in the insets.

Table 3 Thermodynamic parameters of Erbicin, EDIAs and Herceptin for soluble ErbB2-ECD by ITC assays.

n K b ⁄ 10 7

( M )1) K

D (n M ) D b H (kJÆmol)1) TD b S (kJÆmol)1) D b G (kJÆmol)1)

repeated using, as immunoagent, an anti-ErbB2

mono-clonal which was not internalized (anti-ErbB2 N28), or

by testing cells poisoned to inhibit endocytosis Very

similar values were obtained This indicates that the

surprising values are not a result of the internalization

process

As an alternative, we concluded that the interactions

of anti-ErbB2 immunoagents with ErbB2 on live cells

could not be interpreted as simple ligand⁄ receptor

interactions It has been anticipated in recent reports

that ligand binding to cell receptors may trigger higher

order events in the membrane of targeted cells In

these events, directly stimulated receptors and other

seemingly unrelated receptors and effectors are

engaged in the formation of complex networks and

receptor mosaics [22], and may induce membrane

bending and remodelling [23] An ErbB signalling

net-work was proposed [24] after an analysis at the

sys-tems level, with ErbB2 as an amplifier of the network

[25] Thus antibody binding, which mimics ligand

binding, may set off events beyond binding, which are

demonstrated by a higher binding affinity and a high

heat production

Discussion

The novel antitumour immunoagents Erbicin,

ERB-hcAb and ERB-hRNase have most, if not all, the

features that make an immunoagent a valid, precious

tool for anticancer immunotherapy: (a) they are all of

human origin, which strongly decreases, if not

elimi-nates, the risks of an immune response; (b) they are

directed to a cell receptor, such as ErbB2, which is

minimally present in non-malignant cells, but

overex-pressed in many carcinomas, especially in breast cancer

cells; (c) they selectively kill ErbB2-positive cells, both

in vitro and in vivo; (d) their size, smaller than that of

immunoglobulins, should favour penetration in solid

tumours; however, in the case of hcAb and

ERB-hRNase, it should also allow for a prolonged half-life

in the bloodstream

Binding to a cell-embedded tumour-associated

anti-gen is the first key step in the mechanism of

antitu-mour immunoagents Thus, we directed our attention

to studying the binding properties of the novel EDIAs,

as well as Herceptin, an immunoagent successfully

employed in the therapy of breast cancer

Further-more, the availability of soluble ErbB2-ECD enabled

us to describe, for the first time, the binding of these

immunoagents to the isolated, free ECD of ErbB2 In

addition, for the first time, the binding study was

con-ducted not only using a semiquantitative methodology,

such as that based on ELISA, previously used to

measure Herceptin binding [14], but also using quanti-tative methods based on physicochemical principles, such as SPR and ITC

The main results of this study can be summarized as follows

1 For the first time, extensive and conclusive infor-mation is reported on the relative affinity and binding kinetics of the EDIAs and Herceptin for soluble or cell-linked ErbB2 2 The results were validated by the use of three independent methodologies, ELISA, SPR and ITC, which gave coherent results 3 The binding

of Erbicin to ErbB2-ECD was found to be enhanced and stabilized by the linking of Erbicin scFv to either

an RNase or the Fc antibody fragment, as in ERB-hRNase and ERB-hcAb, respectively This was revealed by the higher binding affinity of the Erbicin immunoconjugates with respect to that of free Erbicin scFv 4 The novel EDIAs display a binding affinity towards soluble ErbB2-ECD which is lower than that measured for ECD embedded in the membrane of ErbB2-positive cells Herceptin, by contrast, shows a higher affinity for soluble ErbB2-ECD Furthermore, binding of ERB-hcAb to cancer cells and its antitu-mour activity are not affected by soluble ECD, whereas the same properties of Herceptin are strongly inhibited As soluble ECD is proteolytically released from the surface of ErbB2-overexpressing cancer cells, and is detected in the serum of patients with advanced breast cancer, a fraction of Herceptin is neutralized in these patients by serum ECD, and hence its cell-direc-ted antitumour action is reduced [19] It has been reported that free, soluble ECD can induce resistance

to the growth inhibitory activity of ErbB2 anti-bodies [26], can neutralize their activity and affect their pharmacokinetics, thus leading to resistance to immu-notherapy [27] However, the use of immunoagents with a low affinity for soluble ECD, such as the Erbi-cin-based immunoagents, will allow for lower thera-peutic doses to be used compared with those needed for Herceptin-based therapy 5 A binding study car-ried out by ITC on anti-ErbB2 immunoagents tested directly on live cells revealed that the association of the immunoagents with the receptor inserted into live cells cannot be interpreted as a simple ligand⁄ receptor interaction Antibody binding, just like ligand binding, triggers higher order events which engage other mem-brane receptors and effectors in the formation of com-plex networks and receptor mosaics These data strongly imply that ITC on live cells and high-affinity antibodies to ErbB2, a critical receptor for which no specific ligand has yet been found, could be employed

in a systems biology approach to unravel the physio-logical significance of the cell receptor

Experimental procedures

Cell lines and antibodies

The 293 cell line (human embryonic kidney) was cultured

in DMEM (Gibco Life Technologies, Paisley, UK) The

SKBR3 cell line (human breast cancer) was cultured in

RPMI 1640 (Gibco Life Technologies) The media were

supplemented with 10% heat-inactivated fetal bovine

serum, 50 UÆmL)1 penicillin and 50 lgÆmL)1 streptomycin

(all from Gibco Life Technologies) All the cell lines were

obtained from the American Type Culture Collection and

cultured at 37C in a 5% CO2atmosphere

The antibodies used were as follows: Herceptin

(Genen-tech, South San Francisco, CA, USA); horseradish

peroxi-dase-conjugated anti-His IgG1 (Qiagen, Valencia, CA,

USA); horseradish peroxidase-conjugated goat anti-human

affinity-isolated IgG1 (Fc-specific; Sigma, St Louis, MO,

USA) Erbicin, ERB-hRNase and ERB-hcAb were

pre-pared as described previously [12–14] The anti-ErbB2 N28

monoclonal was a generous gift from Michael Sela

(Weiz-man Institute of Science, Rehovot, Israel)

Production of ErbB2-ECD

ErbB2-ECD, the extracellular domain of ErbB2 (residues

1–624), was stably expressed and secreted by the 293 cell

line The culture medium of 293 cells, before and after

transfection, was analysed: (a) by 8% SDS-PAGE and

wes-tern blotting with Herceptin followed by horseradish

perox-idase-conjugated anti-human (Fc-specific) IgG serum

fraction; (b) by immunoprecipitation assays carried out by

the incubation of 10 mL aliquots of 293 cell conditioned

medium with 10 lgÆmL)1 of Herceptin or ERB-hcAb in

NaCl⁄ Pifor 3 h at 4C The immune complexes were then

collected by adsorption to protein A-Agarose (Sigma) for

1 h at 4C After washing with NaCl ⁄ Pi, the proteins were

released by boiling in loading buffer [28], and run using 8%

SDS-PAGE, followed by immunoblotting assays as

described above

Protein purification

ErbB2-ECD, secreted by transfected 293 cells, was purified

from the culture medium by immunoaffinity

chromatogra-phy with the AKTA Purifier system (GE Healthcare,

Amer-sham Bioscience AB, Uppsala, Sweden) The affinity

column was prepared by coupling 8 mg of Herceptin to

1.5 g of CNBr-activated Sepharose 4B Fast Flow (GE

Healthcare) The antibody was immobilized to agarose via

a secondary amine according to the manufacturer’s

instruc-tions The resulting 4 mL column was loaded with 10 mL

of 10-fold concentrated conditioned medium, washed with

three volumes of 10 mm Tris⁄ HCl, pH 7.4 and eluted with

50 mm glycine pH 3.0 containing 1 m NaCl The collected

fractions were immediately neutralized with a 1 : 10 volume

of 1 m Tris⁄ HCl pH 8.0

The purity of the preparation was evaluated by 8% SDS-PAGE, followed by Coomassie staining or western blotting analyses with either Herceptin or ERB-hcAb as primary antibody, followed by horseradish peroxidase-conjugated anti-human IgG1 (Fc-specific) mAb

ELISA

The affinity of Erbicin or ERB-hRNase for soluble ErbB2-ECD was measured by an ELISA sandwich assay

A 96-well plate was coated with 5 lgÆmL)1of Herceptin in NaCl⁄ Pi (Sigma), kept overnight at 4C and blocked for 1 h at 37C with 5% BSA (Sigma) in NaCl ⁄ Pi To the plate, rinsed with NaCl⁄ Pi, a solution of purified ErbB2-ECD in NaCl⁄ Pi(5 lgÆmL)1) was added After 1 h

at room temperature, the plate was washed, and increasing concentrations of purified ERB-hRNase or Erbicin (50–

500 nm) were added in ELISA buffer (NaCl⁄ PI–BSA 1%)

in triplicate wells, and incubated for 2 h at room tempera-ture with a blank control of NaCl⁄ Pi After rinsing with NaCl⁄ Pi, an anti-His horseradish peroxidase-conjugated IgG1 was added in ELISA buffer After 1 h at room temperature, the plate was rinsed with NaCl⁄ Pi, and bound immunoagents were detected using 3,3¢,5,5¢-tetramethyl-benzidine as a substrate (Sigma)

The product was measured at 450 nm using a microplate reader (Multilabel Counter Victor 3, Perkin Elmer, Cologno Monzese, Italy)

The affinity of ERB-hcAb or Herceptin antibodies for ErbB2-ECD was measured as follows A 96-well plate was coated with 5 lgÆmL)1of purified ECD in NaCl⁄ Piand left overnight at 4C After blocking as above, increasing concentrations of ERB-hcAb (10–60 nm) or Herceptin (0.1–10 nm) were added in ELISA buffer for 2 h at room temperature The plate was rinsed with NaCl⁄ Pi and an anti-human (Fc-specific) horseradish peroxidase-conjugated IgG serum fraction was incubated for 1 h, and detected as described above The reported affinity values are the means

of at least three determinations (standard deviation,£ 5%) ELISAs with ErbB2-positive cells were carried out

on SKBR3 cells as described previously [14] ERB-hcAb (1–16 nm) or Herceptin (1–8 nm) was tested in the presence

or absence of soluble ErbB2-ECD, added either in equimo-lar amounts or in a 10-fold moequimo-lar excess to the ErbB2 receptor number on SKBR3 cells [18]

Cytotoxicity assays

ErbB2-positive cells were treated as described previously [14] with ERB-hcAb or Herceptin at concentrations of 2.5 nm in the absence or presence of soluble ErbB2-ECD (20 nm) Cell growth inhibition was reported as the percent-age of cell survival reduction induced by the treatment with