Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (21)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Catimel, B.
Right arrow Articles by Nice, E. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Catimel, B.
Right arrow Articles by Nice, E. C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Glycobiology Pages 927-938  

Direct immobilization of gangliosides onto gold-carboxymethyldextran sensor surfaces by hydrophobic interaction: applications to antibody characterization
Introduction
Results
Discussion
Materials and methods
Abbreviations
References

Direct immobilization of gangliosides onto gold-carboxymethyldextran sensor surfaces by hydrophobic interaction: applications to antibody characterization

Direct immobilization of gangliosides onto gold-carboxymethyldextran sensor surfaces by hydrophobic interaction: applications to antibody characterization

B.Catimel, A.M.Scott1, F.T.Lee1, N.Hanai2, G.Ritter3, S.Welt3, L.J.Old3, A.W.Burgess, E.C.Nice4

Ludwig Institute for Cancer Research, Melbourne, Australia, 1Tumour Targeting Program, Ludwig Institute for Cancer Research and Austin Hospital, Heidelberg, Victoria, Australia and 2Tokyo Research Laboratories, Kyowa Hakko Kogyo Co. Ltd., Japan, and 3Ludwig Institute for Cancer Research, New York, NY 10021, USA

Received on January 15, 1998; revised on March 26, 1998; accepted on March 26, 1998

We describe a novel immobilization technique to investigate interactions between immobilized gangliosides (GD3, GM1, and GM2) and their respective antibodies, antibody fragments, or binding partners using an optical biosensor. Immobilization was performed by direct injection onto a carboxymethyldextran sensor chip and did not require derivatization of the sensor surface or the ganglioside. The ganglioside appeared to bind to the sensor surface by hydrophobic interaction, leaving the carbohydrate epitope available for antibody or, in the case of GM1, cholera toxin binding. The carboxyl group of the dextran chains on the sensor surface did not appear to be involved in the immobilization as evidenced by equivalent levels of immobilization following conversion of the carboxyl groups into acyl amino esters, but rather the dextran layer provided a hydrophilic coverage of the sensor chip which was essential to prevent nonspecific binding. This technique gave better reactivity and specificity for anti-ganglioside monoclonal antibodies (anti-GD3: KM871, KM641, R24; and anti-GM2: KM966) than immobilization by hydrophobic interaction onto a gold sensor surface or photoactivated cross-linking onto carboxymethydextran. This rapid immobilization procedure has facilitated detailed kinetic analysis of ganglioside/antibody interactions, with the surface remaining viable for a large number of cycles (>125). Kinetic constants were determined from the biosensor data using linear regression, nonlinear least squares and equilibrium analysis. The values of kd, ka, and KA obtained by nonlinear analysis (KA KM871 = 1.05, KM641 = 1.66, R24 = 0.14, and KM966 = 0.65 × 107 M-1 ) were essentially independent of concentration and showed good agreement with data obtained by other analytical methods.

Key words: biosensor/gangliosides/kinetic analysis/ protein-protein interactions/surface plasmon resonance

Introduction

Gangliosides are neuraminic acid containing glycosphingolipids present in the plasma membrane of all animal cells, composed of a sialyl-oligosaccharide glycosidically linked to ceramide (Wiegandt, 1985). The lipophilic ceramide portion is anchored in the lipid bilayer and the hydrophilic carbohydrate region is exposed to the outer environment of the cells. While their physiological function is largely unclear, altered ganglioside biosynthesis has been observed during malignant transformation in many tumors (Hakomori, 1985). The gangliosides GD3, GD2, and GM2 expressed in malignant melanoma and other types of cancer appear to be attractive targets for specific immunotherapy with monoclonal antibodies and vaccines (Oettgen and Old, 1991; Ritter and Livingstone 1991). GD3 is of particular interest because it is abundantly expressed on the cell surface of >90% of melanomas (Pukel et al., 1982; Scott and Cebon 1997).

Clinical trials with a murine anti-GD3 monoclonal antibody R24 have been conducted in patients with melanoma, and have demonstrated responses in a small number of patients (Houghton et al., 1985; Welt et al., 1987; Balch et al., 1993). Repeated treatment has been limited by immune responses (HAMA) to the murine antibody; however, a chimeric version of R24 has been constructed, which retains immune effector function but has a lower level of binding to GD3 than its mouse counterpart (Chapman et al., 1994). A novel murine anti-GD3 antibody KM641 has also been developed, and has been shown to have potent immune effector function (complement-dependent cytotoxicity (CDC) and antibody-dependent cellular-cytotoxicity (ADCC)) against GD3 expressing melanoma cells (Ohta et al., 1993). A chimeric humanized version of KM641 (KM871) has been produced (Shitara et al., 1993, 1994), which has more potent immune effector function than KM641, and induces inhibition of melanoma xenograft growth. KM871 is currently being developed for clinical trials in patients with metastatic melanoma.

As part of the development of antibodies for immunotherapy, it is important to determine their binding characteristics with the appropriate antigen. Measurement of antibody binding affinity is a key element in the process development of constructs for clinical trials, in view of the potential changes in affinity induced by purification and conjugation procedures. In addition, antibody engineering strategies that preferentially select constructs of varying molecular sizes with optimal rapid on rates and slow off rates to antigen targets are dependent on an accurate technique for determining these parameters. However, the evaluation of binding kinetics for anti-GD3 antibodies (R24, KM641, KM871) with conventional assays such as solid phase assays or cell binding assays using radiolabeled antibody has been problematic due to increased binding of anti-GD3 antibodies to GD3 positive cells in the presence of excess unlabeled anti-GD3 antibody (Chapman et al. 1990). The mechanism of this effect is unclear and has been suggested to be due to homophilic binding of IgG variable domains (Chapman et al. 1990), although the low affinity of anti-GD3 antibodies may also contribute to these observations. The inability to perform accurate affinity assays with these standard methods has led to the search for more sophisticated methods of affinity analysis.

Table I. Evaluation of GD3 sensor surfaces using alternative immobilization strategies
Sensor chip Method GD3 immobilized (ng/mm2) Reactivity (RU)a Specificityb Molar binding activityc
KM871 KM641 KM871 KM641 KM871 KM641
Chip CM5 Photoactivated
crosslinking		 1.8 1200 1344 ND ND 0.008 0.009
Chip J1 Hydrophobic 0.97 9780 9092 1500 3072 0.10 0.07
Chip CM5 Hydrophobic 0.08 946 1292 34 79 0.13 0.17
    0.18 1823 ND 55 ND 0.12 ND
Derivatized chip CM5 Hydrophobic 0.07 898 ND 40 ND 0.14 ND
a100 µg/ml IgG injected.bBinding to blank channel (RU)cMaterials and methods.Sensor chip J (prototype), unmodified gold surface. Sensor chip CM5, carboxymethyldextran surface. Derivatized sensor chip CM5, carboxymethyldextran surface blocked with ethanolamine using NHS/EDC chemistry. ND, Not determined.

Recent advances in biosensor technology (Malmqvist, 1993) have led to the development of an optical biosensor (BIAcore) using the detection principal of surface plasmon resonance (Liedberg et al., 1993) capable of measuring binding interactions in real time as well as allowing detailed analysis of complex reaction kinetics (Karlsson et al., 1991; O'Shannessy et al., 1993). These analyses yield both rate and affinity constants which are not readily attainable by classical techniques. Binding studies are performed by covalently attaching one interactant (monoclonal antibody (mAb), protein, peptide or DNA) onto the sensor (Löfas and Johnsson, 1990) and flowing reagents of interest over this surface.

The study of lipid-protein interactions using surface plasmon resonance detection following the immobilization of lipid-binding proteins has been reported recently (End et al., 1993; Machesky et al., 1994; James et al., 1995; Rajaram and Sawyer, 1996). However, immobilization chemistries designed specifically for proteins and peptides cannot be readily transferred to direct lipid immobilization. Alternative methods are therefore required. We present herein a technique for direct ganglioside attachment to the conventional CM5 carboxymethyldextran sensor chip, forming a stable surface, without either prior derivatization of the ganglioside or incorporation of the ganglioside into liposomes. We have analyzed the kinetics of the interaction between immobilized gangliosides and related monoclonal antibodies and antibody fragments or other binding partners.

Results

Immobilization strategies

In an attempt to immobilize the ganglioside by hydrophobic interaction, GD3 (1 mg/ml in ethanol, diluted 1/3 (v/v) in HBS buffer) was directly injected at 2 µl/min over a gold sensor surface (Pioneer chip J1, partial hydrophobic character). The resultant sensorgram indicated that GD3 bound onto the gold surface (a response of 810RU corresponding to ~0.97ng/mm2; Table I); this assumes a refractive index increment of 0.15 for gangliosides, based on a refractive index increment of 0.15 for carbohydrates and 0.14 for lipids (Armstrong et al., 1947), rather than the more generally quoted refractive index increment of 0.18 for proteins (Karlsson et al., 1991). This surface gave good reactivity resulting in a molar binding activity (Johnsson et al., 1991; Catimel et al., 1997) of 0.10 with anti-GD3 KM871 IgG and 0.07 with KM641 IgG (Table I). However, considerable nonspecific binding was observed on the control channel (Table I), suggesting incomplete, or ineffectual, blocking of residual hydrophobic sites by BSA. Furthermore, in the case of antibodies such as KM641, showing strong non-specific interaction with the blank channel, virtually no dissociation of antibody from the immobilized GD3 was observed during the dissociation phase (apparent kd = 8.8 × 10-5 s-1, data not shown). A similar apparent dissociation rate was observed on the control channel, suggesting that nonspecific interaction with the gold surface was occurring (results not shown). Complete antibody dissociation and regeneration of the surface at the end of each cycle could only be achieved using 10 mM sodium hydroxide, but this was also found to partially remove the immobilized GD3 from the gold surface and thereby modify the level of immobilization.

Covalent immobilization of GD3 by photoactivated cross-linkage onto the CM5 sensor chip resulted in relatively high ganglioside immobilization (1500RU, equivalent to 1.8 ng/mm2, Table I) but gave poor molar binding activity with KM871 and KM641 monoclonal antibodies (0.008 and 0.009, respectively; Table I).

We next investigated whether the CM5 sensor chip could be used for direct ganglioside immobilization. Injection of GD3 (60 µl at a flow rate of 5 µl/min, Figure 1A, signal a1) onto a nonactivated carboxymethyldextran sensor chip, followed by extensive washing with 10 mM NaOH (typically four or five washes of 20 µl at a flow rate of 5 µl/min) until a stable baseline was obtained, resulted in a low level of GD3 immobilized (78RU, Figure 1). Injection of KM871 IgG (100 µg/ml, 30 µl at a flow rate of 5 µl/min) resulted in a response of 989RU (Figure 1A, signal a2). A molar binding activity of 0.12 was calculated. At the end of the injection phase, when buffer alone flowed over the sensor surface, dissociation was apparent. Complete dissociation of the antibody/GD3 interaction and surface regeneration could be achieved using 10 mM NaOH. Reinjection of the KM871 IgG resulted in a similar signal (1006RU, Figure 1A, signal a3), demonstrating that the regeneration conditions did not affect the ganglioside surface. However, injection of the detergent n-octyl [beta]-d glucopyranoside (80 mM, 50 µl at 5 µl/min) (Figure 1A, signal a4) removed ~31% of the immobilized GD3 with a concomitant reduction in binding when rechallenged with the KM871 antibody (from 1006RU to 629RU, Figure 1A, signal a5). A second detergent wash removed a further 16% of immobilized GD3 (data not shown). Injection of 20% (v/v) n-propanol/water removed immobilized GD3 in a similar manner and also resulted in reduced KM871 binding on rechallenge with the same sample (data not shown). Taken together these results are compatible with a hydrophobic mechanism for the immobilization of the ganglioside onto sensor chip CM5. After further washes with n-octyl [beta]-d glucopyranoside, the level of KM871 binding was reduced to 430RU (Figure 1B, signal b1). However, reinjection of GD3 (50 µl, 1 mg/ml in 100% ethanol diluted 1/3 v/v in HBS buffer) at a flow rate of 5 µl/min (Figure 1B, signal b2) resulted in further immobilization of GD3 (86RU immobilized). This new surface gave a proportionally increased signal with KM871 IgG (1239 RU, Figure 1B, signal b3). Thus, levels of immobilization could be readily modulated. All levels of immobilization tested (30-160 RU) showed significant reactivity resulting in higher molar binding activities for anti-GD3 antibodies (0.13 for KM871 IgG, 0.17 for KM641 IgG, Table I) compared with GD3 immobilized onto the gold sensor surface of chip J1 (Table I). Furthermore, virtually no nonspecific binding was observed on the corresponding blank channel and a blocking step was not necessary. The molar binding activity was independent of surface loading (Table I). Low surface loadings were used for kinetic experiments to minimize mass transport effects (Glaser, 1993).


Figure 1. (A) Hydrophobic immobilization of GD3 onto CM5 sensor chip. a1: Injection of 60 µl GD3 (1 mg/ml stock solution in ethanol, diluted 1/3 in HBS buffer) at a flow rate of 5 µl/min. The maximum signal (off scale) was 32165RU. R: Injection of 20 µl of 10 mM NaOH (dissociation antibody/GD3 interaction). a2: Injection of 30 µl KM871 IgG (100 µg/ml, response 989RU). a3: Injection of 30 µl KM871 IgG (100 µg/ml, response 1006RU). a4: Injection of 50 µl of 80 mM n-octyl [beta]-d glucopyranoside (the maximum signal was 18760RU). a5: Injection of KM871 IgG (100 µg/ml) giving a response of 629RU, demonstrating removal of GD3 from the sensor surface by n-octyl [beta]-d glucopyranoside (cf. a2, a3). (B) Reimmobilization of GD3 following desorption with n-octyl [beta]-d glucopyranoside. b1: Injection of 30 µl KM871 IgG (100 µg/ml, flow rate 5 µl/min) following four detergent washes (response 430 RU). R: Injection of 20 µl of 10mM NaOH. b2: Injection of 50 µl GD3 (1 mg/ml stock solution in ethanol, diluted 1/3 in HBS buffer). The maximum signal (off scale) was 32300RU. b3: Injection of 30 µl KM871 IgG (100 µg/ml, response 1239RU). (C) Stability of the GD3 surface to repetitive cycles of injection/desorption. c1-50: Fifty repetitive injections of 30 µl KM871 IgG (100 µg/ml) were performed over GD3 ganglioside immobilized onto CM5 sensor chip by hydrophobic interaction (120RU immobilized). R: Following each cycle of binding and dissociation, the surface was regenerated by injection of 20 µl of 10 mM NaOH at a flow rate of 20 µl/min.

A similar level of immobilization (0.07 ng/mm2) and molar binding activity (0.14 for KM871 IgG) was obtained when GD3 was bound to a CM5 sensor chip which had been reacted previously with ethanolamine using NHS/EDC chemistry (Table I). This result suggests that the carboxymethyldextran sites were not involved in ganglioside immobilization.

Following the initial washing of the sensor surface with 10 mM NaOH, no further GD3 dissociation from the CM5 chip occurred during the 10 mM NaOH regeneration step, and no surface denaturation was evident as shown by equivalent signals on 50 consecutive cycles of antibody injection (667 nM) followed by surface regeneration (Figure 1C, signals c1-c50). These injections gave an average signal of 1520 RU with a standard deviation of 8.5RU, demonstrating the stability of the surface to both sodium hydroxide regeneration and continued use. Indeed, the same surface has been used for more than 125 consecutive runs with virtually no change in reactivity.

Characterization of ganglioside solutions

To confirm that the gangliosides were still soluble as monomers in the ethanol/saline (1:2 v/v) solution used for immobilization, samples were analyzed by micropreparative size exclusion HPLC using a Superdex Peptide column (separation range 100-7000 Mr for peptides, 160-6000 Mr for polyethylene glycols), using the ethanol/saline solution as the mobile phase. Eluant fractions were analyzed using the biosensor with immobilized antibody, or BCT in the case of GM1, to detect eluting gangliosides. Gangliosides eluted close to the column volume in freshly prepared solutions (retention time 18-20 min, Vc = 20.5 min), suggesting that monomeric forms of the ganglioside were used for immobilization. By comparison, a sample which had been stored frozen in the ethanol/saline (1:2 v/v) solution showed significant material eluting close to the void volume (9 min).

Surface specificity

GD3 immobilized surface. To evaluate the specificity of the GD3 immobilized surface (CM5 sensor chip), a panel of anti-GD3 monoclonal antibodies and nonrelated control monoclonal antibodies were injected over either GM1, GM2 or GD3 immobilized by hydrophobic interaction, a nonderivatized blank channel or a blank channel blocked with ethanolamine using the NHS/EDC chemistry (Table II). The anti-GD3 antibodies (KM871 IgG and Fab[prime], KM641 IgG and R24 IgG) were found to bind specifically to immobilized GD3, but not significantly to immobilized GM1 or control channels (Table II). The anti-GM2 KM966 IgG as well as other control IgGs (A33 IgG, anti-GM-CSF IgG) did not bind significantly to immobilized GD3, GM1, or to the blank channels.

Table II. Binding specificity of anti-ganglioside antibodies to gangliosides immobilized onto CM5 sensor chip by hydrophobic interaction
Sensor surface   GM1 (RU) GM2 (RU) GD3 (RU) Blank (RU) Mock-derivatized (RU)
Monoclonal antibodies Subclass  
Chimeric KM871 IgG IgG1/IgG3 8.5 ± 1.2 5.7 ± 0.5 1770 ± 4.5 3 ± 0.5 10 ± 2
Murine KM641 IgG IgG3 230 ± 26 47 ±1.5 2569 ± 35 144 ± 22 126 ± 41
Murine R24 IgG IgG3 76 ± 28 15 ± 2.5 1609 ± 16 99 ± 4 74 ± 3
Chimeric KM966 IgG IgG1/IgM 1 ± 0.2 1348 ± 8.4 23 ± 1.3 13 ± 0.5 10 ± 2.5
Humanized A33 IgG IgG1 269 ± 13 42 ± 5.5 257 ± 26 139 ± 15 88 ± 6
Murine A33 IgG IgG2a 2 ± 0.5 2 ± 0.4 2 ± 0.3 3 ± 0.7 1 ± 0.4
Murine anti-GM-CSF IgG IgG1 6 ± 1.9 12 ± 1.2 12 ± 3.3 12 ± 1.2 11 ± 1.2
Chimeric KM871 Fab[prime] IgG1/IgG3 7 ± 0.6 2 ± 1.2 699 ± 13 4 ± 0.3 4 ± 1
Humanized A33 Fab[prime] IgG1 3.5 ± 0.5 1.3 ± 0.7 1 ± 0.5 3 ± 0.8 1 ± 0.5
Anti-GD3 mAbs (KM871, KM641, R24), anti-GM2 mAb (KM966) and control mAbs (anti-colonic antigen A33, anti-granulocyte macrophage colony stimulating factor (GM-CSF)) were injected (30 µl, 150 µg/ml) over immobilized GM1, GM2, GD3, blank channel (carboxymethyl dextran) or mock-derivatized channel (ethanolamine immobilized using NHS/EDC). Specific binding of anti-GD3 and GM2 monoclonal antibodies is highlighted. Four experiments were used to calculate mean and SD.

Table III. Inhibition of KM871 IgG binding to immobilized GD3 by competition with GM1 and GD3
Ganglioside added (µg) % inhibition by GM1 % inhibition by GD3
1 7 41
2 16 62
3 19 74
4 22 85
5 20 86
KM871 IgG (100 µl, 100 µg/ml in HBS buffer) was incubated for 30 min with various concentration of ganglioside GM1 or GD3 or with buffer alone. After incubation, an aliquot (30 µl) was injected at a flow rate of 5 µl/min over immobilized GD3. The percentage inhibition was calculated by comparison to the response obtained with KM871 IgG incubated with buffer alone.

To further demonstrate the specificity of the GD3/anti-GD3 mAb interaction, competition experiments were also performed in which KM871 IgG (1 µg) was incubated with GM1 or GD3 ganglioside prior to injection (Table III). The interaction between KM871 IgG and immobilized GD3 was inhibited in a dose dependent manner by GD3, although some apparent inhibition by GM1 ganglioside was also noted. Competition with GD3 (1, 2, and 3 µg) inhibited the binding by 41%, 62%, and 74%, respectively (Table III). By contrast, competition with similar amounts of GM1 resulted in inhibition of 7%, 16%, and 19% (Table III). Competition with higher ganglioside levels (4 and 5 µg) showed a maximal inhibition by GD3 of ~85% (Table III). In comparison, incubation with similar amounts of GM1 gave a maximal inhibition of 20-22% (Table III).

GM2 immobilized surface. The responsiveness of the GM2 surface, immobilized by hydrophobic interaction onto the CM5 sensor chip, was also tested against the panel of monoclonal antibodies (Table II). Significant binding was only observed with the corresponding anti-GM2 mAb, KM966.


Figure 2. The binding of cholera toxin B chain (BCT) to immobilized gangliosides. (A) The gangliosides GM1, GM2, and GD3 were immobilized by hydrophobic interaction on parallel channels on a CM5 sensor chip as described in Materials and methods. The fourth channel was an underivatized blank. BCT (30 µl, 50 µg/ml) was injected over the surfaces at a flow rate of 5 µl/min. (B) Scatchard analysis of the interaction between BCT and GM1 immobilized by hydrophobic interaction. Data were obtained from the equilibrium binding responses from two independent experiments using concentrations of GM1 from 150-1620 nM.

GM1 immobilized surface. As shown above (Table II), the GM1 surface was not recognized by anti-GM2 and anti-GD3 antibodies. To confirm that the surface was viable, it was tested with cholera toxin B chain (BCT), a paradigm for protein-sugar interactions (Kuziemko et al., 1996). Varying concentrations (1620-150 nM) of BCT were injected over GM1 immobilized onto the CM5 surface by hydrophobic interaction, or a nonderivatized blank channel. BCT was shown to bind specifically to immobilized GM1 but not significantly to GM2, GD3, or the control channel (Figure 2A).


Kinetic analysis of the interaction between GD3 and anti-GD3 mAbs

Having demonstrated both the specificity and reactivity of sensor surfaces on which ganglioside GD3 had been immobilized by hydrophobic interaction, we analyzed the binding kinetics of a panel of anti-GD3 monoclonal antibodies. Varying concentrations of HPLC purified KM641 IgG (667-67 nM, Figure 3A), KM871 IgG (667-67 nM, Figure 3B), R24 IgG (667-67 nM, Figure 3C), KM871 F(ab)[prime]2 (667-133 nM, Figure 3D), and KM871 Fab[prime](2 µM to 31.2 nM, Figure 3E) were injected at a continuous flow of 5 µl/min over the sensor surface. Using the fluidics of the BIAcore 2000, the sample was also injected sequentially over a parallel channel with immobilized GM1 which served as a control to distinguish both refractive index and nonspecific binding events. The very low nonspecific binding observed when KM871 Fab[prime] was injected over the GM1 control channel is illustrated in Figure 3F. Similar low nonspecific binding was observed with the other antibodies. The signal obtained on the control channels was subtracted electronically from the corresponding binding curves obtained with the GD3 channel prior to kinetic analysis.


Figure 3. Biosensor analysis of the interaction between anti-GD3 ganglioside antibodies (R24, KM641, and KM871) and immobilized GD3. Varying concentrations of KM641 IgG (667-67 nM) (Figure 3A), KM871 IgG (667-67 nM) (Figure 3B), R24 IgG (667-67 nM) (Figure 3C), KM871 F(ab)[prime]2(667-67nM) (Figure 3D), and KM871 Fab[prime] (2000-37.5 nM) (Figure 3E) were injected (30 µl) at a flow rate of 5 µl/min over the GD3 ganglioside which had been immobilized onto the CM5 sensor surface via hydrophobic interaction, as described in Materials and methods. The sensorgrams shown have been subtracted with the corresponding signal obtained when the same sample was passed over a control channel generated by immobilization of GM1 onto sensor chip CM5 via hydrophobic interaction. The sensorgrams obtained with varying concentrations of KM871 Fab[prime] (2000-31.2 nM) over the control channel are shown in Figure 3F.

It has been shown recently that the method of analysis can influence interpretation of complex binding kinetics (Morton et al., 1995; Nice et al., 1996; Schuck and Minton, 1996). The apparent association (ka) and dissociation rate constants (kd) and affinity constant, KA, were therefore determined from the binding curves following linearization of the primary data (LR) (Karlsson et al., 1991, 1992; Nice et al., 1996), nonlinear least squares analysis (NLLS) (O'Shannessy et al., 1993), and analysis at equilibrium (Karlsson et al., 1991; Nice et al., 1996) (Table IV).

Linear analysis (LR). The ka and kd were determined following linear transformation of the primary data shown in Figure 3. To determine regions of the curve where kinetic data could be readily extracted, and to avoid areas which might be influenced by mass transport effects (Glaser, 1993), rebinding (Panayotou et al., 1993; Payne et al., 1993) or complex multi-order kinetics (Karlsson et al., 1994; Schuck and Minton, 1996), regions were selected from the sensorgrams which were linear with respect to plots of ln(abs(dR/dt)) versus t (Kuziemko et al., 1996; Nice et al., 1996). Values of ka = 4.41 × 105 M-1 s-1 and kd = 6.9 × 10-2 s-1 were found for KM871 IgG (Figure 4A). However, the kd value obtained from LR analysis of the association phase, which is obtained from the intercept of the y-axis, can be prone to error, particularly when the kd values are low (Chaiken et al., 1992) since kd is usually small in relation to kaC, and slight errors in the slope will have large effects on the measured intercept. The kd was therefore also determined at each individual concentration by analysis of the dissociation phase data (Karlsson et al., 1991; Nice et al., 1996). Values ranging from 3.03 × 10-2 s-1 to 3.40 × 10-2 s-1 (average 3.2 × 10-2 s-1, Table IV) were obtained for KM871 IgG (667-67 nM). The corresponding values of ka and kd obtained by linear analysis for R24, KM641, and KM871 antibodies and the calculated values obtained for the equilibrium constant, KA (= ka/kd), are listed in Table IV.

Equilibrium analysis (EQ). The KA was also derived directly from analysis of equilibrium binding data. The plot of Req/nC versus Req for KM871 IgG is shown in Figure 4B, indicating a KA of 4.76 × 106 M-1. The KA values obtained by equilibrium analysis for the other antibodies tested are shown in Table IV.

Nonlinear least squares analysis (NLLS). The apparent binding constants were derived using a double exponential form of the rate equation (O'Shannessy et al., 1993) applied to the total association or dissociation phase. This type of analysis results in two values for the dissociation phase (kd1and kd2) and also for the association phase (ka1 and ka2). However, the relative amplitudes obtained from this analysis suggested than one set of values, corresponding to ka1 and kd1, predominated. Similar results were obtained at higher flow rates (up to 50 µl/min), suggesting that mass transfer effects were minimal. A single exponential fit of the binding data over regions which were linear for plots of ln(abs(dR/dt)) versus t (i.e., those regions which had been used for LR analysis) gave similar values to those calculated for the predominant interaction (data not shown). Examples of NLLS fitting for the interaction between immobilized GD3 and KM871 IgG at 667 nM using a double exponential function are shown in Figure 4C (dissociation phase fitting) and Figure 4D (association phase fitting), in which the values of kd1 and kd2 obtained from the dissociation phase analysis (Figure 4C) have been used to constrain the analysis (O'Shannessy et al., 1993). The ka1 and kd1 values were determined at each concentration tested, and the average values reported in Table IV. The kd1 values (0.04-0.05 s-1, average 0.045 s-1, Table IV) were found to be similar over the concentration range tested. For ka1, the values obtained were 4.68 × 105 M-1 s-1 to 4.92 × 105M-1s-1 (average 4.75×105M-1s-1, Table IV). This resulted in a principal affinity constant, KA1, of 1.05×107M-1 (Table IV).


Figure 4. Kinetic analysis of the interaction between KM871 IgG and GD3 immobilized via hydrophobic interaction onto CM5 sensor chip. (A) Linear regression analysis. The ka is obtained from the slope of the plot of ks against the concentration of mAb injected. n is the valency (= 2 for IgG). The coefficient of correlation, R, is indicated. (B) Scatchard analysis. The KA is obtained from the slope of the plot of Req/nC versus Req. (C) Nonlinear least squares analysis of the dissociation phase of the interaction between KM871 IgG (667 nM) and immobilized GD3. Data were analyzed using the double exponential form of the rate equation (O'Shannessy et al., 1993). The correlation, [chi]2, between the experimental (solid line) and fitted data (circles), is indicated. (D) Nonlinear least squares analysis of the association phase of the interaction between KM871 IgG (667 nM) and immobilized GD3. Data were analyzed using the double exponential form of the rate equation (O'Shannessy et al., 1993). kd1and kd2 were determined initially by analysis of the dissociation phase of the sensorgram (Fig. 4C) and used to constrain the analysis of the association phase (O'Shannessy et al., 1993). The correlation, [chi]2, between the experimental (solid line) and fitted data (circles), is indicated.

Table IV. Kinetic analysis of the interaction of anti-GD3 monoclonal antibodies with GD3 immobilized by hydrophobic interaction onto sensor chip CM5
Antibody Analysis ka1 × 10-5 (M-1 s-1) kd1 × 102 (s-1) KA1 × 10-7 (M-1) ka2 × 10-5(M-1 s-1) kd2 × 103 (s-1) KA2 × 10-7 (M-1)
KM871 IgG LR 4.41 3.2 1.37 - - -
EQ - - 0.48 - - -
NLLS 4.75 4.5 1.05 0.25 1.99 1.25
KM641 IgG LR 2.65 1.8 1.45 - - -
EQ - - 2.50 - - -
NLLS 3.15 1.9 1.66 0.22 1.65 1.30
R24 IgG LR 0.42 2.4 0.18 - - -
EQ - - 0.20 - - -
NLLS 0.39 2.8 0.14 0.04 4.1 0.10
KM871 F(ab)[prime]2 LR 4.62 7.2 0.64 - - -
EQ - - 0.43 - - -
NLLS 5.57 7.1 0.78 0.41 5.86 0.70
KM871 Fab'[prime] LR 1.98 16 0.12 - - -
EQ     0.24      
NLLS 3.37 27 0.12 1.43 21 0.68
KM966 IgG LR 2.06 3.95 0.52      
EQ     0.37      
NLLS 2.58 3.95 0.65 0.32 1.70 1.90
LR, Linear regression analysis; EQ, Scatchard analysis of equilibrium binding; NLLS, nonlinear least squares regression analysis. KA1 (bold) is the predominant association constant for the interaction indicated from the relative amplitudes obtained by nonlinear least squares regression analysis using a double exponential form of the rate equation (O'Shannessy et al., 1993). KA2 routinely accounted for less than 10% of the total interaction. All the antibodies were against GD3 except the anti-GM2 antibody, KM966.

Similar analysis for KM641 IgG, the murine counterpart of chimeric KM871, showed a predominant dissociation rate kd1 = 0.019 s-1 and association rate ka1 = 3.15 × 105 M-1 s-1 (Table IV), resulting in a KA1 = 1.66 × 107M-1 (Table IV). For murine R24 IgG, the calculated kd1 and ka1 were 0.028 s-1 and 0.39 × 105M-1 s1, respectively. The slower association rate resulted in a lower KA1 = 0.14 × 107 M-1 (Table IV) .


The kinetics of binding of KM871 antibody fragments (Fab)[prime]2 and Fab[prime] to immobilized GD3 were also investigated (Table IV). The KM871 F(ab)[prime]2 and Fab[prime] fragments displayed a lower KA1 (0.78 × 107 M-1 and 0.12 × 107 M-1, respectively) compared to the corresponding KM871 IgG (1.05 × 107 M-1) due in the main to the faster dissociation rate constants (0.071 s-1 and 0.270 s-1, respectively, compared to 0.045 s-1 for IgG).

Kinetic analysis of the interaction between GM2 and anti-GM2 mAb, KM966

Similar methods were used to analyze the interactions between anti-GM2 mAb KM966 and the ganglioside GM2 immobilized by hydrophobic interaction. Good correlation was observed between LR, NLLS, and EQ data (Table IV). Again the values were concentration independent and were not affected by increasing flow rate, suggesting that mass transfer effects were minimal.

Kinetic analysis of the interaction between GM1 and BCT

The association and rate constants for the interaction between GM1 and BCT were investigated using the methods described above. For these calculations it was assumed that BCT was a pentameric complex with a valency of 5 and a Mr of 55,000 (Gill, 1976). Equilibrium analysis (Figure 2B) indicated a KA = 5.3 × 106 M-1. A maximum binding of 2060RU was indicated from the intercept on the x-axis. However, interrogation of the sensorgrams using the plot of ln(abs(dR/dt) versus t indicated that the reaction kinetics were complex. LR and NLLS analysis of the regions which were linear with respect to ln(abs(dR/dt)) versus t suggested a ka of 9.7 × 104 M-1 s-1 and a kd of 2.6 × 10-4 s-1 giving a KA of 3.8 × 108 M-1. By comparison, NLLS of the total dissociation phase using the double exponential function suggested that there was an initial fast off rate (1.3 × 10-2 s-1) which accounted for less than 3% of the dissociation while the remainder of the curve was fitted by a kd of 5.8 × 10-5 s-1. This latter figure resulted in an apparent KA1 of 1.5 × 109 M-1.

Discussion

We have investigated the kinetic interaction of purified anti-ganglioside mAbs with their ganglioside antigens. The ganglioside, rather than the mAb, was immobilized onto the sensor surface because: (1) the smaller Mr analyte is immobilized which facilitates detection; (2) gangliosides form micelles in aqueous solution which preclude molecular mass assignment for kinetic analysis; (3) phase transfer effects associated with micelles or liposomes are avoided.

The carbohydrate moiety of the ganglioside, which projects from the bilayer lipid membrane of tumor cells, is the target for anti-ganglioside monoclonal antibody binding (Ritter and Livingstone, 1991). Ideally, the ganglioside should be immobilized onto the sensor surface via the lipid portion of the molecule, leaving the carbohydrate residues accessible for antibody binding, in an orientation analogous to that found in the cell membrane (Hakomori, 1985; Wiegandt, 1985). Immobilization of GD3 onto the sensor surface via carbohydrate using aldehyde coupling (Löfas et al., 1995), as would be anticipated, completely abolished the reactivity toward KM871 IgG (results not shown).

Lipid vesicles have been previously immobilized onto sensor chip surfaces using thiol coupling (Lang et al., 1992, 1994; Erdelen et al., 1994), biotin/streptavidin interaction (Masson et al., 1994; Stachowiak et al., 1996), covalent immobilization using a photoactive cross-linker (Stein and Gerish, 1996) or capture attachment using immobilized monoclonal antibodies (MacKenzie et al., 1997). Hydrophobic interaction has also been used to immobilize lipids onto a carboxymethydextran coated chip derivatized with alkylamine (Stein and Gerish, 1996) or onto a gold surface coated with a hydrophobic self-assembled monolayer of octadecanethiol (Plant et al., 1995), although this surface appeared to have only limited stability. While earlier studies were performed using stripped and modified carboxymethyldextran surfaces (Kuziemko et al., 1996; Stein and Gerish, 1996) a hydrophobic sensor chip composed of long chain alkanethiol groups linked to the gold sensor surface is now commercially available (HPA, Pharmacia Biosensor). This chip has been recently used in the study of the interaction between gangliosides and cholera toxin (Kuziemko et al., 1996). Gangliosides were incorporated in lipids (e.g., POPC: 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) prior to immobilization onto sensor chip HPA or a gold surface (Kuziemko et al., 1996).

Alternative strategies which require minimum manipulation to directly attach the GD3 molecule to the sensor surface, in a hydrophobic manner, via the lipid portion of the molecule have been investigated. These were compared with covalent attachment using photoactivated crosslinking which resulted in relatively high levels of ganglioside immobilization but low specific molar binding activity toward anti-GD3 monoclonal antibodies (Table I). This may be due to nonselective reaction of aryl nitrene groups in the formation of covalent bonds that may induce multiple orientations of the GD3 molecule onto the sensor surface and hence reduce the accessibility of the carbohydrate binding domains. Furthermore, to achieve this immobilization, which was performed manually outside the biosensor, GD3 was immobilized over the whole sensor surface. A different sensor chip was required for each surface, negating the advantage of the BIAcore 2000 to perform simultaneous analysis with the same sample using up to four channels and including appropriate controls.

In our initial investigation on hydrophobic immobilization of GD3, the experiment was performed in a manner analogous to trace enrichment in HPLC, without the requirement of incorporating the ganglioside molecule into liposomes. A gold sensor surface (Pioneer Chip J1) was chosen because of reduced hydrophobic character compared to sensor chip HPA in the hope that, whilst it would be compatible with hydrophobic attachment, subsequent nonspecific binding would be reduced. However, complete blockage of the remaining hydrophobic sites following GD3 immobilization was found to be problematic.

While optimizing preconcentration conditions for the covalent immobilization of GD3 onto the CM5 sensor chip using photoactivated crosslinking, it was observed that the GD3 ganglioside was binding to the CM5 surface in the absence of immobilized cross-linker. We therefore decided to investigate GD3 binding onto the CM5 chip following the protocol used for GD3 immobilization onto the gold surface. The molar binding activity observed with anti-GD3 antibodies was greater than for GD3 immobilized by photoactivated cross-linkage (Table I), suggesting that the ganglioside was immobilized with the carbohydrate epitopes accessible to the antibody.

This GD3 surface was shown to be stable to the regeneration conditions chosen (10 mM NaOH, pH 11.5) and was suitable for multiple analyses (>125). However, immobilized GD3 could be removed by a pulse of detergent (80 mM n-octyl [beta]-d glucopyranoside) (Figure 1B) or 20% n-propanol, suggesting the hydrophobic character of the immobilization. Electrostatic interactions were unlikely to occur because the immobilization of gangliosides onto the sensor surface was performed at pH 7.4, under which conditions the carboxyl groups of the dextran chains are negatively charged. At this pH, the carboxyl groups would be fully ionized and electrostatic repulsion between dextran and the sialic acid groups of the ganglioside (negative charge at pH 7.4) would be maximum. The lack of charge based interactions was supported by the observation that high concentrations of NaCl (up to 2 M) or MgCl2 (4 M) were ineffectual in desorbing bound ganglioside. The fact that the carbohydrate epitopes were available for antibody binding further reinforced the concept that immobilization was by hydrophobic interaction of the non polar ceramide chain of the ganglioside molecule and not via the polar heads of the molecule composed of sialy-oligosaccharides. Furthermore, the level of GD3 immobilization was similar when GD3 was sequentially injected over a nonderivatized channel or a channel in which the carboxyl groups had been transformed into acyl amino esters using NHS/EDC chemistry followed by reaction with ethanolamine. This chemistry has been shown to convert ~30-45% of the carboxyl groups (Cullen et al., 1987/8) and would have significantly reduced the level of GD3 immobilized if these groups were directly involved in a GD3/dextran interaction.

A positive characteristic of hydrophobic immobilization on the CM5 sensor chip was that, in marked contrast to the gold surface, negligible levels of nonspecific binding were observed obviating the requirement for a blocking step. The carboxymethyldextran layer seemed to provide a hydrophilic coverage of the sensor chip which minimized nonspecific binding. The ganglioside probably binds to discrete hydrophobic sites on the sensor surface, either available sites on the dextran/alkylthiol layer or the gold surface itself (Löfas and Johnsson, 1990). This is in some ways analogous to the stable hydrophobic binding of gangliosides reported with some chromatographic supports. Ganglioside GM3 has been bound in a stable hydrophobic manner to octyl-Sepharose (low level binding, removal by organic solvent) and used as an affinity matrix (Hirabayashi et al., 1983). Gangliosides have also been found to bind via hydrophobic and electrostatic interaction (via the carboxyl of the sialyl groups) onto positively charged DEAE-Sepharose (high level binding, removal using salt containing buffers; Rodriguez and Cumar, 1990). This DEAE-Sepharose-ganglioside complex was used for the immunopurification of anti-ganglioside antibodies (Rodriguez and Cumar, 1990).

Surface specificity was demonstrated using a panel of monoclonal antibodies and BCT, for which GM1 is the major receptor (MacKenzie et al., 1997). Competition studies, performed on immobilized GD3, also indicated specific binding, although some competition was observed with GM1 (up to 22%). Additionally the maximum inhibition by GD3 was only ~85% (Table III). Failure to totally compete binding to ganglioside surfaces has been noted previously (Kuziemko et al., 1996), which was suggested to be due, in part, to incorporation of soluble ganglioside onto the gold hydrophobic sensor surface used in these experiments. This could also have been a factor in our studies since we noted that a GD3 surface (initially negative for BCT: see Figure 2) gave a positive response to BCT after solution competition studies with soluble GM1 had been performed (data not shown). The fact that antibody does not bind directly to immobilized GM1 (see Table II), but appears to be inhibited by GM1 in solution, could also be due to the fact that immobilization of the hydrophobic ceramide head on the sensor surface renders it unavailable for binding, while in solution hydrophobic interactions are possible.

The stable hydrophobic immobilization of GD3 onto the CM5 sensor chip allowed us to study in detail the interaction between anti-GD3 and GM2 monoclonal antibodies and the immobilized ganglioside antigen. Purified antibody and antibody fragments were analyzed and repurified by micropreparative SEC prior to analysis to ensure homogeneity, which is essential for successful kinetic analysis (Nice et al., 1994, 1996). Kinetic constants were extracted from the sensorgram curves by linear transformation of the primary data (LR), by nonlinear least squares regression analysis (NLLS) and equilibrium binding. The value of the association and dissociation rate constants obtained by LR and NLLS were in good agreement for all antibodies analyzed, suggesting the appropriateness of the models used. The kd1 and ka1, corresponding to the predominant signal determined by NLLS (as indicated by the relative amplitudes of the signals), using a double exponential function (O'Shannessy et al., 1993) were similar over the range of concentrations tested and showed no significant dependence on flow rate (up to 50 µl/min), suggesting that rebinding and mass transport effects were not significant. Indeed, mass transport effects are possibly reduced by the relatively low levels of immobilization used (Glaser, 1993) which can readily be controlled using this hydrophobic method. The affinity constants, which were also calculated from equilibrium binding data, were in good agreement with those obtained by LR and NLLS.

The affinity constants obtained by biosensor analysis were slightly lower than values previously obtained by enzyme linked immunosorbent assay for R24 IgG (2 × 107 M-1) (Dippold and Bernhard, 1992), although for KM641 they were similar to those obtained for by Scatchard analysis of antibody binding to a tumor cell line (5 × 107 M-1; Ohta et al., 1993). In the comparative studies using biosensor technology described herein, the KM641 antibody was found to have a higher affinity than R24 IgG. The kinetic analyses suggested that this difference in affinity was mainly due to the slower association rate constant observed for the R24 IgG. In previous studies, the kinetic analysis of the interaction between the anti-ganglioside antibody R24 and GD3 has been further complicated by the reported homophilic binding (Chapman et al., 1990). This problem, observed in both solid phase assays and binding studies using labeled antibody, could also explain the slight differences in the affinity constants derived using the different techniques. We were unable to demonstrate homophilic binding on the sensor surface when anti-ganglioside antibodies were injected over the corresponding antibody immobilized using several alternative chemistries, or captured by immobilized ganglioside (unpublished observations).

It was of interest to study the effect of chimerization of KM641 IgG on apparent binding rates. The KM871 chimeric IgG was shown to have only marginally lower affinity than the murine antibody due to a faster dissociation constant. The kd was considerably faster for the KM871 F(ab)[prime]2 and the KM871 Fab[prime] fragments as is evident from an inspection of the sensorgrams (cf. Figure 3A,B and Table IV). The large increase in the observed kd for the Fab[prime] fragment was probably due to monovalent binding compared to predominantly bivalent binding for the F(ab)[prime]2. The effect of valency on the dissociation and association rate constants was demonstrated recently by biosensor analysis of the binding of multimeric constructs of scFv fragments of an antibody against the tumor-associated carbohydrate antigen Lewis Y (Rheinnecker et al., 1996).

By contrast, kinetic analysis of the interaction between the pentameric cholera toxin B subunit and its ganglioside receptor, GM1, was not facile. In particular, different methods of analysis yielded conflicting results, with values of KA ranging from 5.2 × 106 M-1 as calculated by EQ to 4.9 × 108 M-1 as calculated from the ratio of the rate constants obtained by NLLS using a double exponential function. Interrogation of the primary data using a plot of ln(abs(dR/dt)) versus t was complex, suggesting that multiple interactions were operative. Similar problems have been noted previously for this interaction using biosensor surfaces based on liposome capture techniques (MacKenzie et al., 1997) or self-assembly chemistry (Kuziemko et al., 1996). Our data are most similar to those of MacKenzie et al. (1997) in that we find the expected surface specificity for the interaction between GM1, GM2, and GD3 with BCT, with a KA between 5.2 × 106 M-1 and 4.9 × 108 M-1. By contrast Kuziemko et al. (1996) observed considerable cross reactivity with other gangliosides and a high apparent KA of 2.2 × 1011 M-1. High-sensitivity isothermal titration calorimetry (Masserini et al., 1992) gave a KA of 7 × 107 M-1, similar to the values obtained in this study. The pentavalent nature of BCT undoubtedly complicates the analysis of the association and dissociation rate constants. It has been suggested (Kalinin et al., 1995) that, for multivalent interactions, the more correct analysis of functional affinity is obtained by thermodynamic characterization of equilibrium binding.

In conclusion, the BIAcore 2000 biosensor, which uses the optical detection principle of surface plasmon resonance, has been used to study the interaction between gangliosides and their binding partners (monoclonal antibodies or cholera toxin), following direct hydrophobic immobilization of the ganglioside onto a carboxymethyldextran sensor surface. The technique of immobilization is simple, rapid, and reproducible and results in an extremely stable surface. Unlike covalently modified surfaces, the surface can be readily reconditioned and reused. This technique will enable further studies on a range of ganglioside/antibody interactions and the evaluation of the effect of chimerization or humanization on the relative affinity of monoclonal antibodies prior to the start of clinical studies. This technique should also be applicable for hydrophobic immobilization of other glycosphingolipids, or other amphipathic molecules (e.g., glycerophospholipids, fatty acids, bile acids) in order to perform biosensor studies.

Materials and methods

Biosensor studies

Analyses were performed using a BIAcore 2000 biosensor (Pharmacia Biosensor, Uppsala, Sweden). Carboxymethyldextran coated sensor chip, CM5 (Research grade) and the amine coupling reagents (N-ethyl-N[prime]-dimethylaminopropyl-carbodiimide (EDC), hydroxy-succinimide (NHS) and ethanolamine) were obtained from Pharmacia Biosensor. Sensor chip J (a prototype gold surface with partial hydrophobic character) was a kind gift from Dr. S. Löfas, Pharmacia Biosensor. The gangliosides GM1, GM2, and GD3 were from ALEXIS corporation, Lãufelfingen Switzerland. Cholera toxin B subunit (BCT) was from Calbiochem, La Jolla, CA. Monoclonal antibodies and antibody fragments were repurified to ensure homogeneity, immediately prior to use in kinetic studies, by micropreparative HPLC (Nice, 1990) using a Superose 12 HR 3.2/30 size-exclusion column equilibrated in HBS buffer at a flow rate of 100 µl/min and connected to a SMART µHPLC system(Pharmacia Biotech, Uppsala, Sweden). Protein concentrations were determined by ultra-violet absorption at 280 nm using A280 (1 mg/ml) of 1.46 for IgG, 1.53 for Fab and 1.48 for F(ab)[prime]2 (Andrew and Titus, 1991).

HPLC characterization of ganglioside solutions

Gangliosides GD3, GM2, and GM1 (1mg/ml in ethanol) were diluted 1/3 in HBS buffer and injected (100 µl sample) onto a micropreparative Superdex Peptide column (HR 3.2/30) equilibrated in HBS/ethanol (1/2 v/v) and connected to the SMART system. Elution was performed at a flow rate of 100 µl/min, and fractions (100 µl) were collected automatically. Fractions were dried using a SpeedVac concentrator (Savant instruments Inc., Farmingdale, NY) and resuspended in 100 µl HBS for biosensor analysis. GD3, GM2, and GM1 in the eluting fractions were detected by using immobilized anti-GD3 KM871 IgG, anti-GM2 KM966 IgG, or BCT, respectively.

Ganglioside immobilization

Ganglioside GD3 was dissolved in ethanol (1 mg/ml) while gangliosides GM1 and GM2 were dissolved (1 mg/ml) in 90% ethanol, 10% methanol (v/v). Dissolved gangliosides were stored at -20°C prior to use. For direct immobilization onto the gold (chip J) or carboxymethyldextran (chip CM5) surfaces, gangliosides were diluted 1/3 (v/v) in HBS buffer and injected (60 µl) at a flow rate of 5 µl/min over the unmodified surface. To facilitate retention of gangliosides by the sensor surface, the nonionic detergent Tween 20, which is generally used in HBS buffer to prevent nonspecific interaction, was omitted from the HBS buffer used in these experiments. After immobilization on chip J, remaining hydrophobic sites were blocked using BSA (60 µl, 1 mg/ml). Blocking was not found necessary with the CM5 surface.

GD3 was also covalently immobilized onto the CM5 sensor chip by photoactivated crosslinkage (Stein and Gerish, 1996). Diaminoethane was first immobilized onto the sensor surface using NHS/EDC chemistry as described previously (Johnsson et al., 1991). The sensor chip was then removed from the instrument and the immobilization performed directly on the sensor surface in the dark. The heterobifunctional (NHS-ester and photoactivated) cross-linker SAND (Pierce, Rockford, IL) was added (50 µl, 5 mM in HBS buffer) and incubated for 15 min in order to achieve linkage via a NHS-ester onto the immobilized amino group. After washing with HBS buffer, the well of the sensor chip was filled with 50 µl of GD3 (1 mg/ml in ethanol, diluted 1/3 (v/v) in detergent free HBS buffer) and exposed to 5 bursts from a camera flashlight. After further washing with HBS buffer, the sensor chip was reinstalled in the BIAcore and 60 µl ethanolamine (1 M, pH 9.0) was injected over the surface at a flow rate of 2 µl/min in order to block remaining reactive groups.

Biosensor binding assays

Purified antibodies and antibody fragments were diluted in HBS buffer prior to analysis. Samples (30 µl) were injected over the sensor surface at a flow rate of 5 µl/min. Following completion of the injection phase, dissociation was monitored in HBS buffer for 360 s at the same flow rate. Residual bound antibody was desorbed, and the surface regenerated between injections, using 20 µl of 10 mM NaOH. This treatment did not denature the GD3 as shown by equivalent signals on reinjection of an antibody containing sample. Desorption of the GD3 with 80 mM octylglucoside or 20% n-propanol were used to demonstrate the hydrophobic character of the GD3 immobilization to the CM5 sensor chip.

To enable a comparison of the surface reactivity resulting from the different immobilization chemistries, the molar binding activities were calculated from the following equation: molar binding activity (Johnsson et al., 1991; Catimel et al., 1997) = (antibody response)(antigen Mr)/(immobilized antigen response)(antibody Mr). The GD3 molecular weight used in this calculation was 1545 Da.

Surface specificity

To evaluate the specificity of the GD3 immobilized surface, anti-GD3 mAbs (KM871 IgG (IgG1/IgG3) (Shitara et al., 1993, 1994) and Fab[prime] fragment, KM641 IgG (IgG3) (Ohta et al., 1993), R24 IgG (IgG3) (Welt et al., 1987)), anti-GM2 mAb (KM966 IgG (IgG1/IgM) (Nakamura et al., 1994)), and nonrelated control mAbs (murine and humanized anti-colonic antigen A33 IgG (IgG2a and IgG1, respectively) and Fab[prime] fragment (Catimel et al., 1997) and anti-GM-CSF IgG (IgG1) (Nice et al., 1990)) were injected (30 µl at 100 µg/ml) over either GM1 or GD3 immobilized by hydrophobic interaction onto the CM5 sensor chip, a nonderivatized blank channel or a blank channel blocked with ethanolamine using the NHS/EDC chemistry.

To further demonstrate the specificity of the GD3/anti-GD3 mAb interaction, solution competition experiments were also performed in which KM871 IgG (1 µg) was incubated with GM1 or GD3 ganglioside (1-5 µg) for 30 min at 25°C prior to injection.

The surface availability of the GM1 ganglioside was also confirmed by binding to BCT. The reactivity of GM1, immobilized to the CM5 surface by hydrophobic interaction, was characterized by injecting varying concentrations (30 µl, 1600-150 nM) of BCT over the ganglioside surface. Desorption was achieved using 4 M MgCl2 (Kuziemko et al., 1996).

Kinetic analysis of biosensor data

The biosensor curves obtained following injection of monoclonal antibodies or antibody fragments over immobilized GD3 or GM2 were subtracted with the control signal obtained with the antibody injected over immobilized GM1 prior to kinetic analysis.

The apparent association rate (ka) and dissociation rate (kd) constants were calculated using both linear transformation of the primary data (Karlsson et al., 1991, 1992) (LR) and nonlinear least squares regression (NLLS) (O'Shannessy et al., 1993) as we have described previously (Nice et al., 1996; Catimel at al, 1997). A valency of 2 has been assumed for IgG and F(ab)[prime]2 and 1 for the Fab[prime] fragment. A valency of 5 was assumed for BCT (Gill, 1976).

The affinity constant, KA, was also determined by equilibrium binding analysis (Karlsson et al., 1991; Nice et al., 1996). KA is obtained from the slope of the graph obtained by plotting the biosensor data in Scatchard format (Req/nC versus Req where Req is the biosensor response at equilibrium, n is the valency, and C is the concentration).

Data were analyzed using the BIA Evaluation kinetics evaluation package supplied by Pharmacia Biosensor. The goodness of fit between experimental data and fitted curves was estimated for linear fitting routines from the coefficient of correlation, R2, and for nonlinear least squares fitting by chi-squared analysis using the equation:

where rf is the fitted value at a given point, rx is the experimental value at the same point, n is the number of data points and p is the number of degrees of freedom.

Abbreviations

BCT, cholera toxin B chain; BSA, bovine serum albumin; EDC: N-ethyl-N[prime]-dimethylaminopropyl-carbodiimide; GM-CSF: granulocyte-macrophage colony stimulating factor; HBS, 10 mM HEPES buffer (pH 7.4) containing 3.4 mM EDTA and 150 mM NaCl; HPLC, high performance liquid chromatography; NHS, N-hydroxysuccinimide; SAND, sulfosuccinimidyl 2-[m-azido-o-nitrobenzamido]-ethyl-1,3[prime]-dithipropionate.Gangliosides are abbreviated as proposed by Svennerholm (1963): GM2, II3NeuAc-Gg3Cer; GM1, II3NeuAc-Gg3Cer; GD3, II3(NeuAc)2-LacCer.

References

Andrew ,S.M. and Titus,J.A. (1991) Purification and fragmentation of antibodies. In Coligan,J.E., Kruisbeek,A.M., Marguiles,D.H., Shevach,E.M. and Strober,W. (eds.), Current Protocols in Immunology, Vol. 2. John Wiley and Sons, New York, pp 2.7.1-2.7.12.

Armstrong ,S.H.,Jr., Budka,M.J.E., Morrison,K.C. and Hasson,M. (1947) Preparation and properties of serum and plasma proteins. XII. The refractive properties of the proteins of human plasma and certain purified fractions. J. Am. Chem. Soc., 69, 1747-1753.

Balch C.M. , Houghton,A.N. and Peters,L.J. (1993) Cutaneous Melanoma. In DeVita,V., Hellman,S. and Rosenberg,S.A. (eds.), Cancer: Principles and Practice of Oncology. J. B. Lippincott, Philadelphia, pp. 1612-1661.

Catimel ,B., Nerrie,M., Lee.,F.T., Scott,A.M., Ritter,G., Burgess,A.W. and Nice,E.C. (1997) Kinetic analysis of the interaction between the monoclonal antibody A33 and its colonic epithelial antigen by the use of an optical biosensor. A comparison of immobilization strategies. J Chromatogr. A., 776, 15-30. MEDLINE Abstract

Chaiken ,I., Rose,S. and Karlsson,R. (1992) Analysis of macromolecular interactions using immobilized ligand. Anal. Biochem., 201, 197-210. MEDLINE Abstract

Chapman ,P.B., Yuasa,H. and Houghton,A.N. (1990) Homophilic binding of mouse monoclonal antibodies against GD3 ganglioside. J. Immunol., 145, 891-898. MEDLINE Abstract

Chapman ,P.B., Gillies,S.D., Houghton,A.N. and Reilly,R.M. (1994) Mapping effector functions of a monoclonal antibody to GD3 by characterization of a mouse-human chimeric antibody. Cancer Immunol. Immunother., 39, 198-204. MEDLINE Abstract

Cullen ,D.C., Brown,R.G.W. and Lowe,C.R. (1987/8) Detection of immuno-complex formation via surface plasmon resonance on gold-coated diffraction gratings. Biosensors, 3, 211-225. MEDLINE Abstract

Dippold ,W. and Bernhard,H. (1992) Immunorecognition of ganglioside epitopes: correlation between affinity and cytotoxicity of gangliosides antibodies. Eur. J. Cancer, 28A, 1605-1610. MEDLINE Abstract

End ,P., Gout,I., Fry,M., Panayotou,G., Dhand,R., Yonezawa,K., Kasuga,M. and Waterfield,M.D. (1993) A biosensor approach to probe the structure and function of the p85 alpha subunit of the phosphatidylinositol 3-kinase complex. J. Biol. Chem., 268, 10066-10075. MEDLINE Abstract

Erdelen ,C., Haussling,L., Naumann,R., Ringsdorf,H., Wolf,H., Yang,J., Liley,M., Spinke,J. and Knoll,W. (1994) Self-assembled disulfide-functionalised amphiphilic copolymers on gold. Langmuir, 10, 1246-1250.

Gill ,D.M. (1976) The arrangement of subunits in cholera toxin. Biochemistry, 15, 1242-1248. MEDLINE Abstract

Glaser ,R.W. (1993) Antigen-antibody binding and mass transport by convection and diffusion to a surface: a two-dimensional computer model of binding and dissociation kinetics. Anal. Biochem., 213, 152-161. MEDLINE Abstract

Hakomori ,S.I. (1985) Aberrant glycosylation in cancer cell membranes as focused on glycolipids: overview and perspectives. Cancer Res., 45, 2405-2414. MEDLINE Abstract

Hirabayashi ,Y., Suzuki,T., Susuki,Y., Taki,T., Matsumoto,M., Higashi,H. and Kato,S. (1983) A new method for the purification of anti-glycosphingolipid antibody. J. Biol. Chem., 94, 327-330.

Houghton ,A.N., Mintzer,D., Cordon-Cardo,C., Welt,S., Fliegel,B., Vadhan,S., Carswell,E., Melamed,M.R., Oettgen,H.F. and Old,L.J. (1985) Mouse monoclonal IgG3 antibody detecting GD3 ganglioside: A phase I trail with malignant melanoma. Proc. Natl. Acad. Sci. USA, 82, 1242-1246. MEDLINE Abstract

James ,S.R., Paterson,A., Harden,T.K. and Dowes,P. (1995) Kinetic analysis of phospholipase C beta isoforms using phospholipid-detergent mixed micelles. Evidence for interfacial catalysis involving distinct micelle binding and catalytic steps. J. Biol. Chem., 270, 11872-11881. MEDLINE Abstract

Johnsson ,B., Löfas,S. and Lindquist,G. (1991) Immobilization of proteins to a carboxymethyldextran-modified gold surface for biospecific interaction analysis in surface plasmon resonance sensors. Anal. Biochem., 198, 268-277. MEDLINE Abstract

Kalinin ,N.L., Ward,L.D. and Winzor,D.J. (1995) Effects of solute multivalence on the evaluation of binding constants by biosensor technology: studies with concanavalin A and interleukin-6 as partitioning proteins. Anal. Biochem., 228, 238-244. MEDLINE Abstract

Karlsson ,R., Michaelsson,A. and Mattson,L. (1991) Kinetic analysis of monoclonal antibody-antigen interaction with a new biosensor based analytical system. J. Immunol. Methods, 145, 229-240. MEDLINE Abstract

Karlsson ,R., Altschuh,D. and Van Regenmortel,M.H.V. (1992) Measurement of antibody affinity. In Van Regenmortel,M.H.V. (ed.), Structure of Antigens, Vol. 1. CRC Press, Boca Raton, FL, pp. 127-148.

Karlsson ,R., Roos,H., Fagerstam,L. and Persson,B. (1994) Kinetic and concentration analysis using BIA technology. Methods: A Companion to Methods in Enzymology, 6, 99-110.

Kuziemko ,G.M., Stroh,M. and Stevens,R.C. (1996) Cholera toxin binding affinity and specificity for gangliosides determined by surface plasmon resonance. Biochemistry, 35, 6375-6384. MEDLINE Abstract

Lang ,H., Duschl,C. and Vogel,H. (1994) A new class of thiolipids for the attachment of lipid bilayers on gold surfaces. Langmuir, 10, 197-210.

Lang ,H., Duschl,C., Gratzel,M. and Vogel,H. (1992) Self assembly of thiolipid molecular layers on gold surfaces. Optical and electrochemical characterisation. Thin Solid Films, 210/211, 818-821.

Liedberg ,B., Lundstom,I. and Stenberg,E. (1993) Principles of biosensing with an extended coupling matrix and surface plasmon resonance. Sensors Actuators, 11, 63-72.

Löfas ,S. and Johnsson,B. (1990) A novel hydrogel matrix on gold surfaces in surface plasmon resonance sensors for fast and efficient covalent immobilization of ligands. J. Chem. Soc. Chem. Commun., 21, 1526-1528.

Löfas ,S., Johnsson,B., Edström,A., Hansson,A., Lindquist,G., Müller-Hillgren,R.M. and Stigh,L. (1995) Methods for site controlled coupling to carboxymethyldextran surfaces in surface plasmon resonance sensors. Biosensors Bioelectronics, 10, 813-822.

Machesky ,L.M., Cole,N.B., Moss,B. and Pollard,T.D. (1994) Vaccinia virus expresses a novel profilin with a higher affinity for polyphosphoinositides than actin. Biochemistry, 33, 10815-10824. MEDLINE Abstract

MacKenzie ,C.R., Hirama,T., Lee,K.K., Altman,E. and Young,M.N. (1997) Quantitative analysis of bacterial toxin affinity and specificity for glycolipid receptor by surface plasmon resonance. J. Biol. Chem., 272, 5533-5538. MEDLINE Abstract

Malmqvist ,M. (1993) Biospecific interaction analysis using biosensor technology. Nature, 361, 186-187. MEDLINE Abstract

Masserini ,M., Freire,E., Palestini,P., Calappi,E. and Tettamanti,G. (1992) Fuc-GM1 ganglioside mimics the receptor function of GM1 for cholera toxin. Biochemistry, 31, 2422-2426. MEDLINE Abstract

Masson ,L., Mazza,A. and Brousseau,R. (1994) Stable immobilization of lipid vesicles for kinetic studies using surface plasmon resonance. Anal. Biochem., 218, 405-412. MEDLINE Abstract

Morton ,T.A., Myszka,D.G. and Chaiken,I.M. (1995) Interpreting complex binding kinetics from optical biosensors: a comparison of analysis by linearization, the integrated rate equation and numerical integration. Anal. Biochem., 227, 176-185. MEDLINE Abstract

Nakamura ,K., Koike,M., Shitara,K., Kuwana,Y., Kiuragi,K., Igarashi,S., Hasegawa,M. and Hanai,N. (1994) Chimeric anti-ganglioside GM2 antibody with antitumor activity. Cancer Res., 54, 1511-1516. MEDLINE Abstract

Nice ,E.C., Dempsey,P., Layton,J., Morstyn,G., Cui,D.-F., Simpson,R.J., Fabri,L. and Burgess,A. (1990) Human granulocyte-macrophage colony stimulating-factor (hGM-CSF): identification of a binding site for a neutralazing antibody. Growth Factors, 3, 159-169. MEDLINE Abstract

Nice ,E.C. (1990) Micropreparative LC of proteins. Nature, 348, 462-463.

Nice ,E.C., Lackmann,M., Smyth,F., Fabri,L. and Burgess,A.W. (1994) Synergies between high-performance liquid chromatography and an optical biosensor. J. Chromatogr., 660, 169-180.

Nice ,E.C., McInerney,T.L. and Jackson,D.J. (1996) Analysis of the interaction between a synthetic peptide of influenza virus hemagglutinin and monoclonal antibodies using an optical biosensor. Mol. Immunol., 33, 659-670. MEDLINE Abstract

Oettgen ,H.F. and Old,L.J. (1991) The history of cancer immunotherapy. In DeVita,V., Hellman,S. and Rosenberg,S.A., (eds.), Biologic Therapy of Cancer. J. B. Lippincott, Philadelphia, pp. 87-119.

Ohta ,S., Honda,A., Tokutake,Y., Yoshida,H. and Hanai, N. (1993) Anti-tumor effects of a novel monoclonal antibody with high binding affinity to ganglioside GD3. Cancer Immunol. Immunother., 36, 260-266. MEDLINE Abstract

O'Shannessy ,D.J., Brigham-Burke,M. and Peck,K. (1992) Immobilization chemistries suitable for use in the BIAcore surface plasmon resonance detector. Anal. Biochem., 205, 132-136. MEDLINE Abstract

O'Shannessy ,D.J., Brigham-Burke,M., Soneson,K.K., Hensley,P. and Brooks,I. (1993) Determination of rate and equilibrium binding constants for macromolecular interactions using surface plasmon resonance: use of nonlinear least squares analysis. Anal. Biochem., 212, 457-468. MEDLINE Abstract

Panayotou ,G., Gish,G., End,P., Truong,O., Gout,I., Dhand,R., Fry,M.J., Hiles,I., Pawson,T. and Waterfield,M.D. (1993) Interactions between SH2 domains and tyrosine-phosphorylated platelet-derived growth factor beta-receptor sequences: analysis of kinetic parameters by a novel biosensor-based approach. Mol. Cell. Biol., 13, 3567-3576. MEDLINE Abstract

Payne ,G., Shoelson,S.E., Gish,G.D., Pawson,T. and Walsh,C.T. (1993) Kinetics of p56lck and p60src Src homology 2 domain binding to tyrosine-phosphorylated peptides determined by a competition assay or surface plasmon resonance. Proc. Natl. Acad. Sci. USA, 90, 4902-4906. MEDLINE Abstract

Plant ,A.L., Brigham-Burke,M., Petrella,E.C. and O'Shanessy,D.J. (1995) Phospholipid/alkanethiol bilayers for cell-surface receptor studies by surface plasmon resonance. Anal Biochem., 226, 342-348. MEDLINE Abstract

Pukel ,C.S., Lloyd,K.O., Travassos,L.R., Dippold,W.G., Oettgen,H.F. and Old,L.J. (1982) GD3, a prominent ganglioside of human melanoma. Detection and characterisation by mouse monoclonal antibody. J. Exp. Med., 155, 1133-1147. MEDLINE Abstract

Rajaram ,O.V. and Sawyer,W.H. (1996) Characterisation of lipid-protein interaction using a surface plasmon resonance biosensor. Biochem. Mol. Biol. Int., 39, 31-39. MEDLINE Abstract

Rheinnecker ,M., Hardt,C., Ilag,L.L., Kufer,P., Gruber,R., Hoess,A., Lupas,A., Rottenberger,C., Pluckthun,A. and Pack,P. (1996) Multivalent antibody fragments with high functional affinity for a tumor-associated carbohydrate antigen. J. Immunol., 157, 2989-2997. MEDLINE Abstract

Ritter ,G. and Livingston,P.O. (1991) Ganglioside antigens expressed by human cancer cells. Semin. Cancer Biol., 2, 401-409. MEDLINE Abstract

Rodriguez ,P.E.. and Cumar,F.A. (1990) Gangliosides noncovalently bound to DEAE-Sephadex: application to purification of anti-ganglioside antibodies. Anal. Biochem., 188, 48-52. MEDLINE Abstract

Schuck ,P. and Minton,A.P. (1996) Kinetic analysis of biosensor data: elementary test for self-consistency. Trends Biochem. Sci., 21, 458-460. MEDLINE Abstract

Scott ,A.M. and Cebon,J. (1997) Clinical promise of tumour immunology. Lancet, 349 (suppl II), 19-22.

Shitara ,K., Kuwana,Y., Nakamura,K., Tokutake,Y., Ohta,S., Miyaji,H., Hasegawa,M. and Hanai,N. (1993) A mouse/human chimeric antiganglioside GD3 antibody with enhanced antitumor activities. Cancer. Immunol. Immunother., 36, 373-380. MEDLINE Abstract

Shitara ,K., Nakamura,K., Tokutake-Tanaka,Y., Fukushima,M. and Hanai,N. (1994) A new vector for the high level expression of chimeric antibodies in myeloma cells. J. Immunol. Methods, 167, 271-278. MEDLINE Abstract

Stachowiak ,O., Dolder,M. and Wallimann,T. (1996) Membrane-binding and lipid vesicle cross-linking kinetics of the mitochondrial creatine kinase octamer. Biochemistry, 35, 15522-15528. MEDLINE Abstract

Stein ,T. and Gerish,G. (1996) Orientated binding of a lipid anchored cell adhesion protein onto a biosensor surface using hydrophobic immobilization and photoactive cross-linking. Anal. Biochem., 237, 252-259. MEDLINE Abstract

Svennerholm ,L. (1963) Chromatographic separation of human brain gangliosides. J. Neurochem.,10, 613-623.

Welt ,S., Carswell,E.A., Vogel,C.W., Oettgen,H.F. and Old,L.J. (1987) Immune and nonimmune effector functions of IgG3 mouse monoclonal antibody R24 detecting the disialoganglioside GD3 on the surface of melanoma cells. Clinical Immunol. Immunopathol., 45, 214-229.

Wiegandt ,H. (1985) Gangliosides. In Neuberger,A. and van Deenen,L.L.M., (eds.), Glycolipids. Elsevier, Amsterdam, pp. 199-260.

4To whom correspondence should be addressed at: Ludwig Institute for Cancer Research, PO Box 2008, Royal Melbourne Hospital, Victoria, 3050, Australia

This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 22 Jul 1998
Copyright©Oxford University Press, 1998.
Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J. Immunol.Home page
J. Hu, X. Huang, C.-C. Ling, D. R. Bundle, and N.-K. V. Cheung
Reducing Epitope Spread during Affinity Maturation of an Anti-Ganglioside GD2 Antibody
J. Immunol., November 1, 2009; 183(9): 5748 - 5755.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
M. Flores-Diaz, A. Alape-Giron, G. Clark, B. Catimel, Y. Hirabayashi, E. Nice, J.-M. Gutierrez, R. Titball, and M. Thelestam
A Cellular Deficiency of Gangliosides Causes Hypersensitivity to Clostridium perfringens Phospholipase C
J. Biol. Chem., July 22, 2005; 280(29): 26680 - 26689.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
H. Ideo, A. Seko, and K. Yamashita
Galectin-4 Binds to Sulfated Glycosphingolipids and Carcinoembryonic Antigen in Patches on the Cell Surface of Human Colon Adenocarcinoma Cells
J. Biol. Chem., February 11, 2005; 280(6): 4730 - 4737.
[Abstract] [Full Text] [PDF]

Home page
 GlycobiologyHome page
H. Ideo, A. Seko, I. Ishizuka, and K. Yamashita
The N-terminal carbohydrate recognition domain of galectin-8 recognizes specific glycosphingolipids with high affinity
Glycobiology, October 1, 2003; 13(10): 713 - 723.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
C. M. House, I. J. Frew, H.-L. Huang, G. Wiche, N. Traficante, E. Nice, B. Catimel, and D. D. L. Bowtell
A binding motif for Siah ubiquitin ligase
PNAS, March 18, 2003; 100(6): 3101 - 3106.
[Abstract] [Full Text] [PDF]

Home page
 JCOHome page
A. M. Scott, F.-T. Lee, W. Hopkins, J. S. Cebon, J. M. Wheatley, Z. Liu, F. E. Smyth, C. Murone, S. Sturrock, D. MacGregor, et al.
Specific Targeting, Biodistribution, and Lack of Immunogenicity of Chimeric Anti-GD3 Monoclonal Antibody KM871 in Patients With Metastatic Melanoma: Results of a Phase I Trial
J. Clin. Oncol., October 1, 2001; 19(19): 3976 - 3987.
[Abstract] [Full Text]

Home page
 Cancer Res.Home page
F.-T. Lee, A. Rigopoulos, C. Hall, K. Clarke, S. H. Cody, F. E. Smyth, Z. Liu, M. W. Brechbiel, N. Hanai, E. C. Nice, et al.
Specific Localization, Gamma Camera Imaging, and Intracellular Trafficking of Radiolabelled Chimeric Anti-GD3 Ganglioside Monoclonal Antibody KM871 in SK-MEL-28 Melanoma Xenografts
Cancer Res., June 1, 2001; 61(11): 4474 - 4482.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (21)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Catimel, B.
Right arrow Articles by Nice, E. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Catimel, B.
Right arrow Articles by Nice, E. C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?