On a key postulate of T-cell receptor restrictive function: the V-gene loci act as a single pool encoding recognition of the polymorphic alleles of the species major histocompatibility complex
Summary
The proposition that single Vα or Vβ gene segments specify the recognition of the allele-specific determinants expressed on the major histocompatibility complex-encoded restricting elements of the species has as its consequence a totally different picture of the functioning of the T-cell receptor. This commentary justifies this assumption and outlines some of its most important consequences.
When restrictive recognition of antigen was first identified it became clear that there were two classes of model for the interaction between the T-cell antigen receptor (TCR) and the major histocompatibility complex (MHC)-encoded restricting element (R).1 These two classes were loosely referred to as the ‘altered self’ and ‘dual recognitive’ models. The ‘altered self’ model as originally developed by Matzinger and Bevan2 can be juxtaposed to the ‘dual recognitive’ model as it was developed by Langman and Cohn.3–5 The ‘altered self’ models in their modern form are based on the assumption that the peptide (P) bound to the restricting element (R) forms a new antigenic determinant (Q) recognized by a TCR that has a single combining site (anti-Q), analogous to the B-cell antigen-receptor (BCR). There is a single anti-Q repertoire under this model that is somatically generated, large and random. The discrimination between positive and negative selection as well as between restrictive and allo-recognition is dependent on quantitative differences in the occupancy of the TCR anti-Q site by Q. For this reason I like to refer to this class of model as the ‘single recognitive–multiple receptor’ model to be contrasted with the ‘dual recognitive’ class of model which I like to refer to as a ‘multirecognitive–single receptor’ model.6 The limitations of the ‘altered self’ category of model have been discussed7 and need no elaboration here. Essentially they fail because accounting for allele-specific recognition is either ad hoc or metaphorical and all explanations of alloreactivity can be reduced to a denial of restrictive reactivity, the phenomenon we are trying to explain.7
The dual recognitive model as we formulated it viewed restrictive recognition of antigen by the TCR as dependent upon its specification of two distinctly different repertoires. One is germline-selected and encoded to recognize the polymorphic allele-specific determinants (a) on the MHC-encoded restricting elements (R) of the species (i.e. the mating pool). This is referred to as the anti-R repertoire. The other is a somatically generated, large, random repertoire recognizing peptide (P) bound to the restricting element (R) as a PR complex. This is referred to as the anti-P repertoire. This commentary deals largely with the germline-selected and encoded anti-R repertoire.
Although our dual recognitive model as it was originally formulated3–5 proved wrong (see discussion in ref. 8) there was one postulate that survived unscathed and was carried over into the Tritope Model,8,9 the modern version of a multirecognitive–single receptor model. This postulate is that:
the V-gene segments, Vα and Vβ, form a single gene pool encoding recognition of the polymorphic allele-specific determinants (a) expressed on the MHC-encoded restricting elements (R) of the species.
Why reject the assumption that the two domains/subunits of R interact to form a single allele-specific determinant seen by the anti-Q repertoire of the BCR-like TCR?
First, there is no way that a somatic mechanism can select from a random repertoire the recognition of the alleles of the species, in this case, to distinguish an allele-specific from a common or shared determinant on R, because both are Self to the individual. The problem cannot be dealt with by poetics like, the TCR is ‘intrinsically biased’ or ‘skewed’ towards recognition of MHC, or has a ‘predilection’, ‘affinity’, ‘innate preference’ or ‘obsession’ for MHC. The recognition of the alleles of R is either germline-selected and encoded or it is not. If one accepts that it is germline-selected and encoded then ‘altered self’ models are in need of serious tweaking.
Second, let us consider how the class II MHC-encoded restricting elements (RII) made up of complementing α and β subunits might evolve. If the expression of the allele-specific determinant were dependent on the interaction between subunits, then a mutation in one subunit that affected the determinant would simultaneously change the specificity of every R-element of the species that used the mutant subunit. This would make all of them functionless because the TCR could not be germline-selected to recognize every member of the family of new alleles created by a single mutation. Put differently, a TCR that recognizes an allele-specific determinant formed by interacting subunits cannot track mutational changes in them. Analogously, a TCR that uses a combining site made up of interacting subunits cannot track by germline mutation new alleles and retain recognition of the existent alleles of R. This is the ‘why’ of the postulate that RII-elements made up of freely complementing α and β subunits express the allele-specific determinants on each subunit independently.
Why assume that single V-domains of the TCR, Vα or Vβ, interact with the allele-specific determinants (a) to initiate function (positive selection, restrictive and allogeneic reactivity)?
The polymorphic alleles of R are, in the end, defined by the TCR interacting either via restrictive or allo-recognition, both of which, one might note, define the same set of alleles. It is known that the TCR has a fixed docking mode on R. Vα always engages the West (W) domain (α2 of RI; β1 of RII) and Vβ, the East (E) domain (α1 of RI; α1 of RII) of R (reviewed in refs 8,10,11). These two domains/subunits interact to form the groove that binds peptide (P). The allele-specific determinants (a) are situated on the West and East domains/subunits separated by the bound peptide.
At the experimental level, an allele-specific determinant (a) formed by the interaction of subunits would be expected to be recognized by an evolutionarily selected unique VαVβ pair and this is clearly not what is observed (see item 3 below).
To restate the postulate, the R-element, RI or RII, displays an allele-specific determinant (a) on each of its domains/subunits. Each determinant (a) is recognized by either the Vα or the Vβ domain of the TCR.
Given this postulate, the following conclusions that comprise what we call the Tritope Model8,9 can be derived, largely by logic.
- 1
Each R-element displays, on average, two allele-specific determinants, aW and aE.
- 2
A given TCR is positively selected to recognize either aW or aE on the thymic R.
- 3
If the Vα recognizing a given aW is positively selected, then the family of Vβs recognizing aE, with which the Vα is complemented, determines the alloreactive repertoire of that Vα-group. If the Vβ recognizing a given aE is positively selected, then the family of Vα recognizing aW with which that Vβ is complemented, determines the alloreactive repertoire of that Vβ-group.
- 4
A fixed docking mode, Vα on the West domain/subunit and Vβ on the East domain/subunit, requires two signalling orientations, one for restrictive, the other for allo-reactivity. One signalling orientation is initiated by interaction with aW; the other by interaction with aE. Every TCR that is positively selected for its restrictive specificity encoded by one subunit entrains an alloreactivity encoded by the other subunit. This is why restrictive and allo-recognition define the same set of alleles; the Vα and Vβ gene-segments act as a single pool, functioning in both restrictive reactivity and alloreactivity dependent on how the TCR is assembled.8
- 5
The total number of functional polymorphic determinants (a) on the restricting elements (R) of the species cannot be greater than the number of V-gene segments that encode their recognition. If, for mouse, there are roughly 80 Vα and 20 Vβ functional gene segments, then there are ≤ 100 allele-specific determinants (polymorphs) evolutionarily selected to be functional in restrictive recognition. These are partitioned between RI and RII and the W and E domains/subunits of each class.
- 6
Two signalling orientations responsible for restrictive reactivity and alloreactivity, respectively, require that the TCR be born in one or the other of two distinct conformations that, upon interaction with ligand, convert to a common intermediate conform that delivers a signal (Signal 1) to the T cell.
- 7
The signalling orientation used for restrictive recognition requires engagement of the anti-peptide site for delivery of Signal 1, whereas the opposite signalling orientation used for alloreactivity does not. Alloreactivity (as well as positive selection) is peptide unspecific (does not require engagement of the antipeptide site). Restrictive reactivity is peptide-specific, requiring engagement of the antipeptide site to trigger Signal 1. The signals for positive selection, negative selection and alloreactivity are qualitatively distinct.12,13 This is in sharp contrast to the standard ‘altered self’ model where the signals are distinguished quantitatively (level of occupancy by Q).
In sum, the simple, almost unavoidable, assumption that each V-gene segment, Vα or Vβ, specifies recognition of an allele-specific determinant (a) on the R-elements of the species gives us a totally different, testable and validly competing view of TCR behaviour.6,8,9 Whether or not the Tritope Model survives over time, the popular ‘altered self’ class of model is in need of intensive care.
Acknowledgements
This paper was written while Melvin Cohn was a visiting scholar at the Gulbenkian Science Institute, Oeiras, Portugal, from May to September 2006. The fellowship support from the Ministério Da Ciência e Da Tecnologia is gratefully acknowledged, as is the encouragement and criticism of the Director, Professor. Antonio Coutinho. This work was made possible by Grant Number RR07716 from the National Center For Research Resources (NCRR), a component of the National Institutes of Health (NIH). This paper's content is solely the responsibility of the author and does not represent the official views of NCRR or NIH.
