Limits...
Comparing the efficacy and neuroinflammatory potential of three anti-abeta antibodies.

Fuller JP, Stavenhagen JB, Christensen S, Kartberg F, Glennie MJ, Teeling JL - Acta Neuropathol. (2015)

Bottom Line: In this study, we directly compared 3 clinical candidates in the same pre-clinical model, with the same effector function, for their ability to clear plaques and induce inflammation in the brain.This translated into 3D6 effectively clearing plaques and reducing the levels of insoluble Aβ, whilst chGantenerumab and mC2 did not. 3D6 caused a significant increase in the levels of pro-inflammatory cytokines IL-1β and TNFα, and an associated increase in microglial expression of CD11B and CD68. chGantenerumab increased pro-inflammatory cytokines and microglial activation, but minimal changes in CD68, as an indicator of phagocytosis.Injection of mC2 did not cause any significant inflammatory changes.

View Article: PubMed Central - PubMed

Affiliation: Centre for Biological Sciences, University of Southampton, Southampton, UK. jpf1g11@soton.ac.uk.

ABSTRACT
Immunotherapy is a promising strategy for the treatment of Alzheimer's disease (AD). Antibodies directed against Amyloid Beta (Aβ) are able to successfully clear plaques and reverse cognitive deficits in mouse models. Excitement towards this approach has been tempered by high profile failures in the clinic, one key issue has been the development of inflammatory side effects in the brain (ARIAs). New antibodies are entering the clinic for Alzheimer's disease; therefore, it is important to learn all we can from the current generation. In this study, we directly compared 3 clinical candidates in the same pre-clinical model, with the same effector function, for their ability to clear plaques and induce inflammation in the brain. We produced murine versions of the antibodies: Bapineuzumab (3D6), Crenezumab (mC2) and Gantenerumab (chGantenerumab) with an IgG2a constant region. 18-month transgenic APP mice (Tg2576) were injected bilaterally into the hippocampus with 2 µg of each antibody or control. After 7 days, the mice tissue was analysed for clearance of plaques and neuroinflammation by histology and biochemical analysis. 3D6 was the best binder to plaques and in vitro, whilst mC2 bound the least strongly. This translated into 3D6 effectively clearing plaques and reducing the levels of insoluble Aβ, whilst chGantenerumab and mC2 did not. 3D6 caused a significant increase in the levels of pro-inflammatory cytokines IL-1β and TNFα, and an associated increase in microglial expression of CD11B and CD68. chGantenerumab increased pro-inflammatory cytokines and microglial activation, but minimal changes in CD68, as an indicator of phagocytosis. Injection of mC2 did not cause any significant inflammatory changes. Our results demonstrate that the ability of an antibody to clear plaques and induce inflammation is dependent on the epitope and affinity of the antibody.

No MeSH data available.


Related in: MedlinePlus

In vitro binding and effector function of 3D6 chGantenerumab and mC2. a Purified recombinant antibodies were separated by reduced and non-reduced polyacrylamide electrophoresis. Non-reduced antibodies run as a single band, reduced antibodies break into heavy and light chain fragments (50, 25 kDa respectively). b Binding of anti-Aβ antibodies to immobilised Aβ 1–40. 3D6 and chGantenerumab both bind with relatively high affinity (EC50 3D6 = 0.17 pM; EC50 chGantenerumab = 0.34 pM) mC2 bound with lower relative affinity (EC50 mC2 = 17.4 pM). c Binding of 3D6, chGantenerumab and mC2 to Tg2576 (APP) and wild-type (WT) brain sections. d TNFα levels produced by RAW264.7 cells in response to immobilised Aβ antibodies or IgG1 control. Data analysed by one-way ANOVA and Tukey post hoc test, and expressed as mean pg/ml supernatant ± standard deviation (SD, n = 6). Data is representative of 3 independent experiments (****p < 0.0001). e Binding of antibodies to formalin-fixed tissue from Tg2576 mice, with and without formic acid antigenic retrieval. f Binding of antibodies to human AD brain tissue, with and without formic acid antigen retrieval
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4612324&req=5

Fig1: In vitro binding and effector function of 3D6 chGantenerumab and mC2. a Purified recombinant antibodies were separated by reduced and non-reduced polyacrylamide electrophoresis. Non-reduced antibodies run as a single band, reduced antibodies break into heavy and light chain fragments (50, 25 kDa respectively). b Binding of anti-Aβ antibodies to immobilised Aβ 1–40. 3D6 and chGantenerumab both bind with relatively high affinity (EC50 3D6 = 0.17 pM; EC50 chGantenerumab = 0.34 pM) mC2 bound with lower relative affinity (EC50 mC2 = 17.4 pM). c Binding of 3D6, chGantenerumab and mC2 to Tg2576 (APP) and wild-type (WT) brain sections. d TNFα levels produced by RAW264.7 cells in response to immobilised Aβ antibodies or IgG1 control. Data analysed by one-way ANOVA and Tukey post hoc test, and expressed as mean pg/ml supernatant ± standard deviation (SD, n = 6). Data is representative of 3 independent experiments (****p < 0.0001). e Binding of antibodies to formalin-fixed tissue from Tg2576 mice, with and without formic acid antigenic retrieval. f Binding of antibodies to human AD brain tissue, with and without formic acid antigen retrieval

Mentions: First, the specificity and relative affinity to Aβ were tested by measuring binding to immobilised Aβ 1–40 peptide. Figure 1b shows that both 3D6 and bind to recombinant peptide with relative high affinity (EC50 3D6 = 0.17 pM; EC50 chGantenerumab = 0.34 pM), however, 100-fold higher levels of mC2 were required to reach half maximal binding (EC50 mC2 = 17.4 pM), suggesting significantly lower affinity to immobilised Aβ. Antibodies were then tested for binding to Aβ plaques in brain sections from Tg2576 mice, and to better mimic in vivo binding conditions tissue sections were not subjected to any antigen retrieval before immuno-staining. 3D6 bound plaques in tissue obtained from Tg2576, while no binding was observed in wild-type mice. mC2 also bound but fewer plaques were labelled (Fig. 1c). chGantenerumab labelled plaques, however, it also appeared to bind to neurons in both Tg2576 and wild-type mice. All antibodies were produced as IgG2a isotype, using the same constant region, and therefore should all have the same ability to bind and activate FcγRs. We then tested the ability of these antibodies to bind to plaques in Tg2576 tissue sections with and without formic acid antigenic retrieval (Fig. 1e). Formic acid treatment breaks down aggregated Aβ into more soluble species. We found that 3D6 was able to bind to plaques without any antigen retrieval, but mC2 and chGantenerumab could not. After formic acid treatment, mC2 labelled plaques very well and chGantenerumab labelled them faintly. The conformation of Aβ in Tg2576 mice and human AD cases may be different, and this has previously been reported to affect target engagement of anti-Aβ antibodies [28]. To characterise the ability of antibodies to bind to Aβ from human cases, brain sections from AD cases were stained with and without formic acid antigenic retrieval (Fig. 1f). The results were comparable to Tg2576 tissue, with 3D6 able to bind plaques without antigenic retrieval, but staining improved after formic acid treatment. The antibody mC2 bound poorly without formic acid but labelled plaques well after treatment. chGantenerumab again showed background staining, but plaque binding was evident after formic acid treatment. Using an FcγR crosslinking assay, we tested the ability of each antibody to activate macrophages in vitro. All IgG2a anti-Aβ antibodies stimulate secretion of the cytokine TNFα compared to cell only controls, while mouse IgG1, a subclass which has lower affinity for activating FcγR, fails to induce TNFα secretion (Fig. 1d). In summary, 3D6 binds plaques and recombinant Aβ 1–40 peptide with high affinity and specificity, mC2 selectively binds Aβ but with lower affinity, while chGantenerumab binds Aβ 1–40 peptide with high affinity, in situ binding indicates non-specific binding to neurons. All antibodies have the comparable ability to activate macrophages through FcγRs.Fig. 1


Comparing the efficacy and neuroinflammatory potential of three anti-abeta antibodies.

Fuller JP, Stavenhagen JB, Christensen S, Kartberg F, Glennie MJ, Teeling JL - Acta Neuropathol. (2015)

In vitro binding and effector function of 3D6 chGantenerumab and mC2. a Purified recombinant antibodies were separated by reduced and non-reduced polyacrylamide electrophoresis. Non-reduced antibodies run as a single band, reduced antibodies break into heavy and light chain fragments (50, 25 kDa respectively). b Binding of anti-Aβ antibodies to immobilised Aβ 1–40. 3D6 and chGantenerumab both bind with relatively high affinity (EC50 3D6 = 0.17 pM; EC50 chGantenerumab = 0.34 pM) mC2 bound with lower relative affinity (EC50 mC2 = 17.4 pM). c Binding of 3D6, chGantenerumab and mC2 to Tg2576 (APP) and wild-type (WT) brain sections. d TNFα levels produced by RAW264.7 cells in response to immobilised Aβ antibodies or IgG1 control. Data analysed by one-way ANOVA and Tukey post hoc test, and expressed as mean pg/ml supernatant ± standard deviation (SD, n = 6). Data is representative of 3 independent experiments (****p < 0.0001). e Binding of antibodies to formalin-fixed tissue from Tg2576 mice, with and without formic acid antigenic retrieval. f Binding of antibodies to human AD brain tissue, with and without formic acid antigen retrieval
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4612324&req=5

Fig1: In vitro binding and effector function of 3D6 chGantenerumab and mC2. a Purified recombinant antibodies were separated by reduced and non-reduced polyacrylamide electrophoresis. Non-reduced antibodies run as a single band, reduced antibodies break into heavy and light chain fragments (50, 25 kDa respectively). b Binding of anti-Aβ antibodies to immobilised Aβ 1–40. 3D6 and chGantenerumab both bind with relatively high affinity (EC50 3D6 = 0.17 pM; EC50 chGantenerumab = 0.34 pM) mC2 bound with lower relative affinity (EC50 mC2 = 17.4 pM). c Binding of 3D6, chGantenerumab and mC2 to Tg2576 (APP) and wild-type (WT) brain sections. d TNFα levels produced by RAW264.7 cells in response to immobilised Aβ antibodies or IgG1 control. Data analysed by one-way ANOVA and Tukey post hoc test, and expressed as mean pg/ml supernatant ± standard deviation (SD, n = 6). Data is representative of 3 independent experiments (****p < 0.0001). e Binding of antibodies to formalin-fixed tissue from Tg2576 mice, with and without formic acid antigenic retrieval. f Binding of antibodies to human AD brain tissue, with and without formic acid antigen retrieval
Mentions: First, the specificity and relative affinity to Aβ were tested by measuring binding to immobilised Aβ 1–40 peptide. Figure 1b shows that both 3D6 and bind to recombinant peptide with relative high affinity (EC50 3D6 = 0.17 pM; EC50 chGantenerumab = 0.34 pM), however, 100-fold higher levels of mC2 were required to reach half maximal binding (EC50 mC2 = 17.4 pM), suggesting significantly lower affinity to immobilised Aβ. Antibodies were then tested for binding to Aβ plaques in brain sections from Tg2576 mice, and to better mimic in vivo binding conditions tissue sections were not subjected to any antigen retrieval before immuno-staining. 3D6 bound plaques in tissue obtained from Tg2576, while no binding was observed in wild-type mice. mC2 also bound but fewer plaques were labelled (Fig. 1c). chGantenerumab labelled plaques, however, it also appeared to bind to neurons in both Tg2576 and wild-type mice. All antibodies were produced as IgG2a isotype, using the same constant region, and therefore should all have the same ability to bind and activate FcγRs. We then tested the ability of these antibodies to bind to plaques in Tg2576 tissue sections with and without formic acid antigenic retrieval (Fig. 1e). Formic acid treatment breaks down aggregated Aβ into more soluble species. We found that 3D6 was able to bind to plaques without any antigen retrieval, but mC2 and chGantenerumab could not. After formic acid treatment, mC2 labelled plaques very well and chGantenerumab labelled them faintly. The conformation of Aβ in Tg2576 mice and human AD cases may be different, and this has previously been reported to affect target engagement of anti-Aβ antibodies [28]. To characterise the ability of antibodies to bind to Aβ from human cases, brain sections from AD cases were stained with and without formic acid antigenic retrieval (Fig. 1f). The results were comparable to Tg2576 tissue, with 3D6 able to bind plaques without antigenic retrieval, but staining improved after formic acid treatment. The antibody mC2 bound poorly without formic acid but labelled plaques well after treatment. chGantenerumab again showed background staining, but plaque binding was evident after formic acid treatment. Using an FcγR crosslinking assay, we tested the ability of each antibody to activate macrophages in vitro. All IgG2a anti-Aβ antibodies stimulate secretion of the cytokine TNFα compared to cell only controls, while mouse IgG1, a subclass which has lower affinity for activating FcγR, fails to induce TNFα secretion (Fig. 1d). In summary, 3D6 binds plaques and recombinant Aβ 1–40 peptide with high affinity and specificity, mC2 selectively binds Aβ but with lower affinity, while chGantenerumab binds Aβ 1–40 peptide with high affinity, in situ binding indicates non-specific binding to neurons. All antibodies have the comparable ability to activate macrophages through FcγRs.Fig. 1

Bottom Line: In this study, we directly compared 3 clinical candidates in the same pre-clinical model, with the same effector function, for their ability to clear plaques and induce inflammation in the brain.This translated into 3D6 effectively clearing plaques and reducing the levels of insoluble Aβ, whilst chGantenerumab and mC2 did not. 3D6 caused a significant increase in the levels of pro-inflammatory cytokines IL-1β and TNFα, and an associated increase in microglial expression of CD11B and CD68. chGantenerumab increased pro-inflammatory cytokines and microglial activation, but minimal changes in CD68, as an indicator of phagocytosis.Injection of mC2 did not cause any significant inflammatory changes.

View Article: PubMed Central - PubMed

Affiliation: Centre for Biological Sciences, University of Southampton, Southampton, UK. jpf1g11@soton.ac.uk.

ABSTRACT
Immunotherapy is a promising strategy for the treatment of Alzheimer's disease (AD). Antibodies directed against Amyloid Beta (Aβ) are able to successfully clear plaques and reverse cognitive deficits in mouse models. Excitement towards this approach has been tempered by high profile failures in the clinic, one key issue has been the development of inflammatory side effects in the brain (ARIAs). New antibodies are entering the clinic for Alzheimer's disease; therefore, it is important to learn all we can from the current generation. In this study, we directly compared 3 clinical candidates in the same pre-clinical model, with the same effector function, for their ability to clear plaques and induce inflammation in the brain. We produced murine versions of the antibodies: Bapineuzumab (3D6), Crenezumab (mC2) and Gantenerumab (chGantenerumab) with an IgG2a constant region. 18-month transgenic APP mice (Tg2576) were injected bilaterally into the hippocampus with 2 µg of each antibody or control. After 7 days, the mice tissue was analysed for clearance of plaques and neuroinflammation by histology and biochemical analysis. 3D6 was the best binder to plaques and in vitro, whilst mC2 bound the least strongly. This translated into 3D6 effectively clearing plaques and reducing the levels of insoluble Aβ, whilst chGantenerumab and mC2 did not. 3D6 caused a significant increase in the levels of pro-inflammatory cytokines IL-1β and TNFα, and an associated increase in microglial expression of CD11B and CD68. chGantenerumab increased pro-inflammatory cytokines and microglial activation, but minimal changes in CD68, as an indicator of phagocytosis. Injection of mC2 did not cause any significant inflammatory changes. Our results demonstrate that the ability of an antibody to clear plaques and induce inflammation is dependent on the epitope and affinity of the antibody.

No MeSH data available.


Related in: MedlinePlus