Science 2 December 2005:
Vol. 310. no. 5753, pp. 1442 - 1443
DOI: 10.1126/science.1122009
Prev | Table of Contents | Next
Perspectives
IMMUNOLOGY:
Tipping the Scales Toward More Effective Antibodies
Jenny M. Woof*
Antibodies are lifesavers par excellence. Not only do these vital proteins of the immune system armory protect millions of people across the globe after being elicited in vaccination programs, but also they serve increasingly as potent therapeutics in the clinic. Monoclonal antibodies, homogeneous antibody preparations specific for single antigens that have long been heralded as magic bullets, are finally fulfilling their promise. Indeed, monoclonal antibodies directed against targets such as cancer cells represent a major focus of the biopharmaceutical industry. But antibodies come in many different classes and subclasses, so how do we decide which category of antibody is most suitable for a particular clinical application? Mouse models of diseases can provide important insights, and a report by Nimmerjahn and Ravetch on page 1510 of this issue (1) raises the possibility of predicting the in vivo efficacy of individual immunoglobulin G (IgG) subclasses in treating particular cancers and infections.
IgG, the major immunoglobulin class in serum, exists as four structurally distinct subclasses in humans and mice. All IgG subclasses recognize antigens on foreign cells but have markedly different abilities to trigger immune mechanisms to eliminate these foreign targets. The latter processes rely on interaction of the Fc region of the antibody with effector molecules such as complement in the serum or Fc receptors (Fc:Rs) expressed on a variety of immune cells. Fc:Rs also come in different classes (2). In humans, there are three types: Fc:RI, Fc:RII, and Fc:RIII. Fc:RII is further subdivided into Fc:RIIA, Fc:RIIB, and Fc:RIIC. Although the correspondence is not absolute, mice also have Fc:RI, Fc:RII, and Fc:RIII, along with a recently discovered additional class, Fc:RIV, which is absent in humans (3). Mice have only the Fc:RIIB form of Fc:RII.
Activation versus inhibition. With a high activatory to inhibitory (A/I) ratio, IgG2a (red) antibody binds antigen (dark green) on a target cell surface and preferentially binds to activatory receptor Fc:RIV (blue) rather than to inhibitory receptor Fc:RIIB (yellow), and immune cell activation results. In contrast, an IgG2b antibody (orange), with a lower A/I ratio, will bind both Fc:RIV and Fc:RIIB and activation is dampened.
CREDIT: K. SUTLIFF/SCIENCE
In terms of function, Fc:Rs fall into two camps--those that activate a cellular response, and those that block it. Activatory receptors usually associate with a transmembrane signaling component that carries an immunoreceptor tyrosine-based activation motif (ITAM), or in some cases the activatory receptor possesses its own ITAM. In contrast, the single inhibitory receptor Fc:RIIB carries a cytoplasmic immunoreceptor tyrosine-based inhibitory motif (ITIM). When one or more Fc:Rs on an Fc:R-positive cell simultaneously bind IgG presented in an aggregated form (e.g., on the surface of a foreign cell), cellular responses are triggered. Engagement of activatory receptors initiates intracellular kinase cascades, culminating in responses such as phagocytosis and release of inflammatory mediators. On the other hand, engagement of both activatory and inhibitory receptors by antigen results in recruitment of intracellular phosphatases that effectively terminate any activation. Because activatory and inhibitory receptors are frequently coexpressed on immune cells such as macrophages and monocytes, the final nature of the cellular response reflects a finely tuned balance between activatory and inhibitory signaling.
How can we better understand the rules of engagement of Fc;Rs so as to predict the cellular outcome? The situation is complicated by the fact that each IgG subclass displays different affinities for the various types of Fc;Rs. Nimmerjahn and Ravetch's starting point was to determine the affinities of each IgG subclass for soluble forms of different Fc;Rs, and then divide the affinity for the relevant activating receptor (either Fc;RIII or Fc;RIV, depending on IgG subclass) by that for the inhibitory receptor (Fc;RIIB). This calculation yields an activatory to inhibitory (A/I) ratio. The A/I ratios of the mouse IgG subclasses differed markedly, with values of 69 for the IgG2a subclass, 7 for IgG2b, and 0.1 for IgG1. IgG3 showed no detectable binding to the receptors tested, so no A/I ratio was assigned.
They subsequently used two mouse model systems to test whether these A/I ratios could predict in vivo outcomes. In one model, IgG subclasses with identical tumor antigen specificity were assessed for the ability to clear lung metastases in mice injected with tumor cells. In the other, matched integrin-specific IgG subclasses were tested for performance in clearing integrin-bearing platelets. In both cases there was a very good correlation between high A/I ratio and biological efficacy. Further experiments with mice deficient in Fc;Rs or complement components, or in which the activatory Fc;RIV was blocked by a specific monoclonal antibody, indicated that, in these antigen models at least, the in vivo activity of the antibodies depended on activatory Fc;Rs and not complement. Eliminating expression of the inhibitory receptor Fc;RIIB in mice by genetic knockout had the greatest impact on the in vivo activity of the IgG subclasses with low A/I ratios. Finally, the investigators modulated the A/I ratios of the IgG subclasses by altering their glycosylation profiles. Consistent with other studies (4), antibodies deficient in fucose had increased affinities for the various Fc;Rs, and therefore had different A/I ratios. Again, a correlation between A/I ratio and efficacy was noted. For example, the A/I of IgG2b increased from 7 to 20 upon defucosylation, which translated into considerably enhanced in vivo activity. Overall, the tests show that the A/I ratio of a particular IgG subclass is a good predictor of efficacy in a rodent model. However, for general applicability, it might be desirable to adjust the A/I ratios by factoring in the expression levels of each Fc receptor.
How might this translate to the human system? Can we select IgG subclasses for optimal in vivo effect? It's not clear yet, but important differences between the mouse and human systems may foreshadow difficulties. First, there is no direct correspondence between mouse and human IgG subclasses. Second, unlike mice, humans possess activatory versions of Fc;RII (Fc;RIIA and C) that share the same IgG subclass binding affinities as the inhibitory Fc;RIIB, thereby complicating A/I ratios. Third, receptor polymorphisms may obscure the picture. For example, an Fc;RIIA polymorphism renders only some individuals able to bind human IgG2 by this receptor. Finally, in the human system, complement is suggested to contribute to antibody-mediated elimination of certain targets (5, 6). However, for antibody therapeutics, the prospect of predicting suitable antibody isotypes is appealing, especially for companies eager to minimize development costs. One thing is certain: More and more monoclonal antibodies are headed for the clinic, and a means of accurately predicting antibody performance at an early stage is a goal worth pursuing.
References
F. Nimmerjahn, J. V. Ravetch, Science 310, [1510] (2005).
F. Nimmerjahn et al., Immunity 23, 41 (2005). [Medline]
M. Daeron, Annu. Rev. Immunol. 15, 203 (1997). [Medline]
R. L. Shields et al., J. Biol.Chem. 277, 26733 (2002). [Medline]
N. Di Gaetano et al., J. Immunol. 171, 1581 (2003). [Medline]
J.Teeling et al., Blood 104, 1793 (2004). [Medline]
Vol. 310. no. 5753, pp. 1442 - 1443
DOI: 10.1126/science.1122009
Prev | Table of Contents | Next
Perspectives
IMMUNOLOGY:
Tipping the Scales Toward More Effective Antibodies
Jenny M. Woof*
Antibodies are lifesavers par excellence. Not only do these vital proteins of the immune system armory protect millions of people across the globe after being elicited in vaccination programs, but also they serve increasingly as potent therapeutics in the clinic. Monoclonal antibodies, homogeneous antibody preparations specific for single antigens that have long been heralded as magic bullets, are finally fulfilling their promise. Indeed, monoclonal antibodies directed against targets such as cancer cells represent a major focus of the biopharmaceutical industry. But antibodies come in many different classes and subclasses, so how do we decide which category of antibody is most suitable for a particular clinical application? Mouse models of diseases can provide important insights, and a report by Nimmerjahn and Ravetch on page 1510 of this issue (1) raises the possibility of predicting the in vivo efficacy of individual immunoglobulin G (IgG) subclasses in treating particular cancers and infections.
IgG, the major immunoglobulin class in serum, exists as four structurally distinct subclasses in humans and mice. All IgG subclasses recognize antigens on foreign cells but have markedly different abilities to trigger immune mechanisms to eliminate these foreign targets. The latter processes rely on interaction of the Fc region of the antibody with effector molecules such as complement in the serum or Fc receptors (Fc:Rs) expressed on a variety of immune cells. Fc:Rs also come in different classes (2). In humans, there are three types: Fc:RI, Fc:RII, and Fc:RIII. Fc:RII is further subdivided into Fc:RIIA, Fc:RIIB, and Fc:RIIC. Although the correspondence is not absolute, mice also have Fc:RI, Fc:RII, and Fc:RIII, along with a recently discovered additional class, Fc:RIV, which is absent in humans (3). Mice have only the Fc:RIIB form of Fc:RII.
Activation versus inhibition. With a high activatory to inhibitory (A/I) ratio, IgG2a (red) antibody binds antigen (dark green) on a target cell surface and preferentially binds to activatory receptor Fc:RIV (blue) rather than to inhibitory receptor Fc:RIIB (yellow), and immune cell activation results. In contrast, an IgG2b antibody (orange), with a lower A/I ratio, will bind both Fc:RIV and Fc:RIIB and activation is dampened.
CREDIT: K. SUTLIFF/SCIENCE
In terms of function, Fc:Rs fall into two camps--those that activate a cellular response, and those that block it. Activatory receptors usually associate with a transmembrane signaling component that carries an immunoreceptor tyrosine-based activation motif (ITAM), or in some cases the activatory receptor possesses its own ITAM. In contrast, the single inhibitory receptor Fc:RIIB carries a cytoplasmic immunoreceptor tyrosine-based inhibitory motif (ITIM). When one or more Fc:Rs on an Fc:R-positive cell simultaneously bind IgG presented in an aggregated form (e.g., on the surface of a foreign cell), cellular responses are triggered. Engagement of activatory receptors initiates intracellular kinase cascades, culminating in responses such as phagocytosis and release of inflammatory mediators. On the other hand, engagement of both activatory and inhibitory receptors by antigen results in recruitment of intracellular phosphatases that effectively terminate any activation. Because activatory and inhibitory receptors are frequently coexpressed on immune cells such as macrophages and monocytes, the final nature of the cellular response reflects a finely tuned balance between activatory and inhibitory signaling.
How can we better understand the rules of engagement of Fc;Rs so as to predict the cellular outcome? The situation is complicated by the fact that each IgG subclass displays different affinities for the various types of Fc;Rs. Nimmerjahn and Ravetch's starting point was to determine the affinities of each IgG subclass for soluble forms of different Fc;Rs, and then divide the affinity for the relevant activating receptor (either Fc;RIII or Fc;RIV, depending on IgG subclass) by that for the inhibitory receptor (Fc;RIIB). This calculation yields an activatory to inhibitory (A/I) ratio. The A/I ratios of the mouse IgG subclasses differed markedly, with values of 69 for the IgG2a subclass, 7 for IgG2b, and 0.1 for IgG1. IgG3 showed no detectable binding to the receptors tested, so no A/I ratio was assigned.
They subsequently used two mouse model systems to test whether these A/I ratios could predict in vivo outcomes. In one model, IgG subclasses with identical tumor antigen specificity were assessed for the ability to clear lung metastases in mice injected with tumor cells. In the other, matched integrin-specific IgG subclasses were tested for performance in clearing integrin-bearing platelets. In both cases there was a very good correlation between high A/I ratio and biological efficacy. Further experiments with mice deficient in Fc;Rs or complement components, or in which the activatory Fc;RIV was blocked by a specific monoclonal antibody, indicated that, in these antigen models at least, the in vivo activity of the antibodies depended on activatory Fc;Rs and not complement. Eliminating expression of the inhibitory receptor Fc;RIIB in mice by genetic knockout had the greatest impact on the in vivo activity of the IgG subclasses with low A/I ratios. Finally, the investigators modulated the A/I ratios of the IgG subclasses by altering their glycosylation profiles. Consistent with other studies (4), antibodies deficient in fucose had increased affinities for the various Fc;Rs, and therefore had different A/I ratios. Again, a correlation between A/I ratio and efficacy was noted. For example, the A/I of IgG2b increased from 7 to 20 upon defucosylation, which translated into considerably enhanced in vivo activity. Overall, the tests show that the A/I ratio of a particular IgG subclass is a good predictor of efficacy in a rodent model. However, for general applicability, it might be desirable to adjust the A/I ratios by factoring in the expression levels of each Fc receptor.
How might this translate to the human system? Can we select IgG subclasses for optimal in vivo effect? It's not clear yet, but important differences between the mouse and human systems may foreshadow difficulties. First, there is no direct correspondence between mouse and human IgG subclasses. Second, unlike mice, humans possess activatory versions of Fc;RII (Fc;RIIA and C) that share the same IgG subclass binding affinities as the inhibitory Fc;RIIB, thereby complicating A/I ratios. Third, receptor polymorphisms may obscure the picture. For example, an Fc;RIIA polymorphism renders only some individuals able to bind human IgG2 by this receptor. Finally, in the human system, complement is suggested to contribute to antibody-mediated elimination of certain targets (5, 6). However, for antibody therapeutics, the prospect of predicting suitable antibody isotypes is appealing, especially for companies eager to minimize development costs. One thing is certain: More and more monoclonal antibodies are headed for the clinic, and a means of accurately predicting antibody performance at an early stage is a goal worth pursuing.
References
F. Nimmerjahn, J. V. Ravetch, Science 310, [1510] (2005).
F. Nimmerjahn et al., Immunity 23, 41 (2005). [Medline]
M. Daeron, Annu. Rev. Immunol. 15, 203 (1997). [Medline]
R. L. Shields et al., J. Biol.Chem. 277, 26733 (2002). [Medline]
N. Di Gaetano et al., J. Immunol. 171, 1581 (2003). [Medline]
J.Teeling et al., Blood 104, 1793 (2004). [Medline]
