The role of naïve T cell precursor frequency and recruitment in dictating immune response magnitude

Abstract

Recent advances in technology have lead to the realization that the populations of naïve T cells specific for different foreign peptide:MHC (p:MHC) ligands vary in size. This variability is due in part to the fact that certain peptides contain amino acids that engage in particularly favorable interactions with TCRs. In addition, deletion of clones with cross-reactivity for self p:MHC ligands may reduce the size of some naïve populations. In many cases, the magnitude of the immune response to individual p:MHC epitopes correlates with the size of the corresponding naïve populations. However, this simple relationship may be complicated by variability in the efficiency of T cell recruitment into the immune response. The knowledge that naïve population size can predict immune response magnitude may create opportunities for production of more effective subunit vaccines.

It is well established that the magnitude of the primary T cell response is influenced by the amount and duration of presentation of the relevant peptide:MHC (p:MHC) ligands by APC in secondary lymphoid organs (1-4). Peptides that are derived from abundant proteins, or are processed efficiently, or bind strongly or stably to MHC are more likely to be displayed in larger amounts and for longer periods of time on APC than peptides that lack these properties (5, 6). Abundantly presented p:MHC complexes will then trigger more intense TCR signaling in cognate T cells than less abundant complexes, which will promote effector and memory cell formation. It is also possible, however, that responses to certain p:MHC ligands are strong because the cognate naïve T cell population is larger than average.

The latter possibility has been difficult to test because the frequency of T cells specific for individual p:MHC ligands is so low (7). This infrequency is a direct consequence of the vast number of αβ TCRs that can be produced by random joining of Tcra and Tcrb V (D) J segments and non-templated N-region additions (8). Thus, within the vast pool of TCRs displayed by individual cells in the naïve T cell repertoire, only a few are likely by chance to have high affinity for any individual p:MHC ligand. We now know that the frequency of such cells is at most about 100 cells per million naive T cells (Tables 1 and 2) (9). This low frequency together with the stringent activation requirements of naïve T cells explains why conventional 96 well plate proliferation assays containing ~10⁶ T cells per well are incapable of detecting p:MHC-specific T cell populations in individuals who were not previously immunized (10).

Table 1.

Foreign p:MHC-specific naïve T cell frequencies in mice determined by tetramer-based cell enrichment

Source	Epitope	Total cells per mouse1	Cells per million naive CD4+ or CD8+ T cells2	Reference
MHCI
Ovalbumin	OVA-257:Kb	600	36	(50)
		170	10	(51)
		130	8	(39)
		70	4	(44)
LCMV	gp-33:Db	449	27	(40)
		287	17	(39)
	NP-396:Db	151	9	(39)
		117	7	(40)
	L-2062:Kb	90	5	(40)
	NP-205:Kb	57	3	(40)
	gp-118:Kb	43	3	(40)
	L-156:Kb	24	1	(40)
	L-338:Db	15	1	(40)
MCMV	M45:Db	1500	89	(52)
		603	36	(39)
	M57:Kb	900	54	(52)
	m139:Kb	300	18	(52)
	IE3:Kb	90	5	(52)
	M38:Kb	80	5	(52)
Influenza virus	NS2-114:Kb	282	17	(45)
	PB1-F2-62:Db	225	13	(45)
	PA-224:Db	120	7	(39)
		68	4	(45)
		64	4	(36)
	NP-366:Db	36	2	(45)
		18	1	(36)
Vaccinia virus	B8R:Kb	1400	83	(50)
		1100	65	(52)
		1070	64	(51)
	B8-20:Kb	320	19	(44)
	F2-26:Ld	150	9	(44)
	A47-138:Kb	120	7	(44)
	K3-6:Db	40	2	(44)
	RSV M2-82:Kd	419	25	(46)
	M-187:Db	370	22	(46)
HSV	1gB:Kb	490	29	(51)
VSV	N:Kb	166	10	(39)
MHCII
2W peptide	2W:I-Ab	384	15	(23)
		307	12	(53)
		300	12	(54)
		225	9	(27)
		190	8	(7)
		183	7	(24)
L. monocytogenes	LLO-190:I-Ab	80	3	(25)
MHV	M133:I-Ab	62	2	(55)
S. typhimurium	FliC-427:I-Ab	32	1	(26)
		27	1	(24)
		24	1	(23)
		20	0.8	(7)
Ovalbumin	OVA-329:I-Ab	43	2	(23)

¹Numbers represent total cells detected in pooled spleen and macroscopic lymph nodes.

²Frequencies calculated from total cell numbers reported divided by approximate total numbers of naive T cells in pooled spleen and lymph nodes (1.7 × 10⁷ CD8⁺ T cells or 2.5 × 10⁷ CD4⁺ T cells).

Table 2.

Foreign p:MHC-specific naïve T cell frequencies in humans determined by tetramer- based cell enrichment

Source	Epitope	Frequency per million naive CD4+ or CD8+ T cells1	Reference
MHCI
HCV	NS3-1073:A2	60	(56)
	NS3-1406:A2	29	(42)
		11	(57)
	Core-132:A2	1	(57)
		1	(42)
	NS5B-2594:A2	1	(42)
HIV	Gag-77:A2	6	(56)
		4	(57)
CMV	pp65-495:A2	6	(56)
		1	(57)
MHCII
B. anthracis	PA-713:DR1	10	(32)
	PA-401:DR1	2	(32)
	PA-505:DR1	0.2	(32)

¹In cases where data was only reported as a proportion of total CD4⁺ or CD8⁺ T cells, frequencies were calculated assuming that 50% of these cells were naive phenotype.

Sallusto and colleagues recently devised a clever “T cell library” approach to solve this problem (11). These investigators cultured 384,000 naïve human CD4⁺ T cells in 192 wells at 2,000 cells/well with the phytohemagglutinin mitogen, allogeneic blood cells, and IL-2. These conditions led to 1,000-fold expansion of all the naïve cells in each well and converted them into hardy effector cells that could be restimulated with a foreign antigen plus autologous blood cells as APC. If 1 of the 192 original wells contained cells that proliferated in response to the foreign antigen, then it could be deduced that the frequency of naïve T cells specific for p:MHCII ligands derived from the antigen was 1/384,000. Using this approach, it was determined that naïve CD4⁺ T cells specific for p:MHCII ligands derived from keyhole limpet hemocyanin exist at frequencies of 10-70 cells per million naive CD4⁺ T cells, and for Bacillus anthracis protective antigen at 10-26 cells per million. If there are 10 different p:MHCII epitopes derived from keyhole limpet hemocyanin and each epitope is recognized by a naïve population of the same size, then each naïve population would exist at a frequency of 1-7 cells per million naive CD4⁺ T cells. An advantage of this technique is that it does not require knowledge of the subject’s MHC molecules and yields a total frequency that is the sum of the frequencies of the populations specific for all the relevant p:MHCII epitopes from the protein. A disadvantage of the technique is that it does not reveal the frequency of T cells specific for individual p:MHCII epitopes.

Recently, however, the combined use of fluorochrome-labeled p:MHC tetramers and magnetic particle-based cell enrichment has solved this problem. Fluorochrome-labeled p:MHC tetramers bind to the TCRs on specific T cells, marking them for detection (12). The difficulty, however, has been the limited capacity of flow cytometers to analyze only about 10⁶ cells at a time, while the rare naïve T cells of interest are mixed in with ~2 × 10⁸ total nucleated cells from the secondary lymphoid organs of a mouse or 100 ml of human peripheral blood. This problem was solved by staining all of the cells in the secondary lymphoid organs with a p:MHC tetramer and then with magnetic beads coupled to antibodies specific for the fluorochrome component of the tetramer (13-22). The sample could then be passed over a magnetic column to capture all of the tetramer-bound cells plus about 10⁶ contaminants. The total number of cells in this bound fraction was therefore small enough that it could be analyzed in its entirety by flow cytometry. The tetramer-bound cells could be distinguished from contaminants by staining with a mixture of fluorochrome-labeled antibodies specific for T cell-and non-T cell-specific surface proteins.

This approach was used to identify naïve CD4⁺ T cells specific for different foreign p:MHCII ligands in C57BL/6 (B6) mice (7, 23-27). The tetramers used in these studies contained the I-A^b MHCII molecule bound to the 2W variant of peptide 52-68 from the I-E MHCII alpha chain (28), peptide 427-441 from the FliC protein of Salmonella typhimurium (29), peptide 329-337 from chicken ovalbumin (OVA) (30), or peptide 190-201 from the listeriolysin O protein of Listeria monocytogenes (31). The 2W:I-A^b- and LLO190-201:I-A^b-binding CD4⁺ T cell populations consisted of about 260 (7, 23-25, 27) and 80 (25) cells per mouse, respectively, while the FliC:I-A^b-binding (7, 23, 24, 26) and OVA:I-A^b-binding (23) populations contained about 20 and 40 cells per mouse. Assuming that mice contain 2.5 × 10⁷ CD4⁺ T cells with a naive phenotype (9), the frequencies of naïve T cells specific for these individual p:MHCII complexes ranged from 0.8-10 cells per million naive CD4⁺ T cells.

Recently, Kwok and colleagues (32) used p:MHCII tetramer-based cell enrichment to enumerate human naïve CD4⁺ T cells specific for 3 different B. anthracis protective antigen peptide:HLA-DR1 epitopes in unvaccinated individuals. The frequencies of these naïve populations were 0.2, 2, and10 cells per million naïve CD4⁺ T cells (Table 2). This range of frequencies is remarkably similar to the 0.8-10 cells per million range reported for mouse T cells specific for individual p:MHCII epitopes determined by the same method (Tables 1 and 2). This congruence is surprising because the TCR diversity of humans has been estimated to be 10 times greater than that of mice (33, 34),

The fact that Kwok et al. (32) and Geiger et al. (11) both studied B. anthracis protective antigen allows for a comparison between the p:MHCII tetramer-based cell enrichment and T cell library methods. The sum of the frequencies of the three peptide:HLA-DR1-specific naïve populations measured by p:MHCII tetramer-based cell enrichment was ~12 cells per million naïve CD4⁺ cells. The collective frequency of T cells specific for all the p:MHCII epitopes in the protein measured by the T cell library approach was 10-26 cells per million. The similarity of the frequencies derived from p:MHCII tetramer staining and the T cell library approach that relies on antigen-driven T cell proliferation suggests that the former method may not greatly underestimate the number of T cells that are capable of responding to antigen as recently suggested (35).

p:MHCI tetramer-based cell enrichment has also been performed by several laboratories to enumerate the pre-immune frequencies for many different foreign p:MHCI-specific CD8⁺ T cell populations (Tables 1 and 2). The number of naïve phenotype CD8⁺ T cells in B6 mice ranged from 15 Lymphocytic choriomeningitis virus (LCMV) L_338-346:D^b-specific cells to 1,100 Vaccinia virus (VACV) B8R:K^b-specific cells per mouse (Table 1). These numbers translate to frequencies of 1-89 cells per million naive CD8⁺ T cells, assuming a total of 2 × 10⁷ CD8⁺ naïve T cells per mouse (9). Again, a similar range of frequencies has been reported for human CD8⁺ T cells (Table 2).

An important conclusion from these studies is that naïve T cell populations specific for different p:MHC vary in size in a predictable fashion in individuals that express the same MHC molecules. Our analysis of relatively small and large naïve CD4⁺ T cell populations specific for different peptides bound to the same MHCII molecule revealed that the smaller population underwent more clonal deletion on cross-reactive self p:MHCII ligands than the larger population (24). Similarly, Day et al. (36) found that the reduction in the number of influenza acid polymerase 224–233:H-2D^b-specific CD8⁺ naïve T cells in mice that expressed H-2D^b and H-2K^k compared to mice that expressed H-2D^b alone was explained in part by the deletion of influenza acid polymerase 224–233:H-2D^b-specific cells due to cross-reactivity with H-2K^k molecules. An abnormally low amount of deletion of T cells that cross-react on self p:MHC may explain the extraordinary capacity of CD8⁺ T cells specific for HIV peptide:HLA-B57 complexes to control infection (37). HLA-B57 molecules bind a less diverse set of self peptides than other MHCI molecules, and thus may mediate less extensive negative selection of the CD8⁺ T cell repertoire. As a result, foreign peptide:HLA-B57-specific T cell populations may be larger and more promiscuous than average, thereby providing more effective immune responses to HIV.

In our studies, we also found that a peptide recognized by a relatively large naive cell population contained tryptophan residues as TCR contacts, which when changed to other amino acids reduced the size of the population (23). Thus, foreign p:MHCII ligands that tend to be recognized by large naive populations may have chemical properties that allow favorable interactions with many different TCRs. The finding of Turner and colleagues that CD8⁺ T cell populations with more TCR diversity recognize peptides containing TCR contact amino acids with large side chains than peptides with small side chains (38) is consistent with this possibility.

Variation in naïve T cell population size is of more than academic interest because it can determine the magnitude of the T cell response. Several studies demonstrated that naïve T cells expand in proportion to their starting frequency following exposure to the relevant p:MHC ligand (7, 39). Work by Sette and colleagues (40) showed that variation in naïve T cell population size is a partial explanation for the phenomenon of immunodominance. Lymphocytic choriomeningitis virus (LCMV) infection of B6 mice activates CD8⁺ T cells specific for one of at least 28 different pMHCI combinations. However, three of these p:MHCI ligands account for about one third of the total response. This dominance was not completely explained by MHCI binding affinity because several of the other peptides bound to MHCI in the same affinity range as the dominant peptides. Rather, the best predictor of response magnitude was the size of the naïve T cell population. The populations specific for the dominant peptides were about 10 times larger than those specific for the less dominant peptides. This correlation between naive population size and immunodominance was also seen in studies of influenza-infected humanized HLA-transgenic mice (41) or hepatitis C virus-specific human peripheral T cells (42), and B. anthracis-vaccinated humans (32). Therefore, although antigen abundance, efficiency of peptide generation by antigen processing, MHC binding affinity, and stability of p:MHC complexes certainly influence immunodominance (5, 6), naïve T cell population size is also an important factor. Large naïve T cell populations are advantageous to the host because they can generate a fixed number of microbicidal effector cells more quickly than smaller populations. For example, it took 4 days to generate 10,000 effector cells from 200 naïve phenotype 2W:I-A^b-specific CD4⁺ T cells, but 8 days from 20 naïve phenotype FliC:I-A^b-specific cells (7). Large naïve T cell populations may be particularly important to aged individuals because small populations are susceptible to extinction as the total pre-immune repertoire contracts during aging (43).

While the correlation between naive T cell population size and immune response magnitude has been substantiated by several studies (7, 32, 39-42, 44), this relationship was not observed in some recent comparisons of p:MHCI-specific CD8⁺ T cell populations during virus infection (45, 46). A disconnect between naïve population size and response magnitude could occur if only a fraction of the population is recruited into the response. This situation could result if the number of APC displaying the relevant p:MHC ligand is very low such that some T cells by chance interact with a p:MHC⁺ APC while other cells in the populations do not. Alternatively, low numbers of p:MHC ligands per APC could result in the response of only those T cells in the naïve population with highest affinity TCRs. In either case, a large naïve population would produce a smaller than expected response.

Schumacher and colleagues introduced unique molecular tags into each cell of a monoclonal naive CD8⁺ T cell population and then tracked the efficiency of naïve T cell recruitment into the primary response (47). Following adoptive transfer, it was found that essentially all of the cells in the population were efficiently recruited into the primary response regardless of the dose of antigen or type of infection. Therefore, naïve CD8⁺ T cells with an identical affinity for a p:MHCI ligand were recruited into the response in an all or none fashion. This finding raises the possibility that at a very low dose of antigen, no members of a large T cell population would respond, while at a slightly higher antigen dose, all of the members of the population would respond. To imagine how this situation could result in a disconnect between naïve T cell population size and primary response magnitude, consider two p:MHC complexes, pA:MHC and pB:MHC, derived from the same virus and recognized by naïve T cell populations containing 1,000 or 100 members, respectively. If the amount of pA:MHC complexes produced during infection is very small and the number of pB:MHC complexes is large, then it is possible that the 1,000 pA:MHC-specific cells will remain naïve while the 100 pB:MHC-specific cells will proliferate to produce 100,000 progeny.

Differential T cell proliferation following initial recruitment could also produce a lack of correlation between naïve T cell population size and primary response magnitude. This contention is supported by work from Zehn and colleagues who used altered peptide ligands with variations in TCR binding affinity for a monoclonal TCR to model what would happen to clones within a polyclonal p:MHCI-specific CD8⁺ T cell population with varying affinities for the p:MHCI complex in question (48). The monoclonal T cells began to proliferate in response to each of the peptides. However, the high-affinity peptides induced larger expanded T cell populations by inducing longer periods of proliferation following initial activation. These findings are largely consistent with TCR repertoire analyses showing that TCR affinity-based immunodominance patterns are not apparent immediately after the first few rounds of cell division, but emerge later on in the primary immune response (49). Taken together, these studies indicate that the initial recruitment of naive T cells from a pMHC-specific population into a primary immune response is fairly uniform across all TCR affinities, but sustained proliferation and eventual effector cell population size depends on TCR signal strength. Again consider two p:MHC complexes, pA:MHC and pB:MHC derived from the same virus and recognized by naïve T cell populations containing 1,000 or 100 members, respectively. If the pB:MHC-specific population contains a greater proportion of high affinity clones than the pA:MHC-specific population, then despite the initial response from all clones from both populations, the 100 naive cells from the pB:MHC-specific population will eventually undergo more extensive proliferation than the 1,000 naive cells from the pA:MHC-specific population, resulting in 100,000 versus 10,000 effector progeny and creating a disconnect between naïve population size and response magnitude. In contrast, if the two populations are represented by similar proportions of high-affinity clones and they see similar levels of p:MHC complexes, then they will proliferate largely in proportion to their starting numbers, thereby creating a situation where the larger pA:MHC-specific naïve population will produce more effector cells than the smaller pB:MHC-specific population. Under these conditions, naïve T cell population size will be the major determinant of primary immune response magnitude.

CONCLUSION

We speculate that an understanding of the rules that govern the size of naive T cell populations could lead to bioinformatic methods for assessing a person’s immune potential. Advances in microbial genomics, HLA typing, and peptide-HLA binding prediction algorithms suggest that it may soon be possible to scan all the peptides from a microbe that will bind to any of a person’s MHC molecules. In addition, some of the studies mentioned above hint that it will eventually be possible to predict the number of naive T cells specific for each MHC-binding peptide based on the nature of the TCR contacts in the peptide. Together, this information could be used to predict the magnitude of a person’s T cell response to all epitopes from a given microbe. This knowledge could inform the composition of subunit vaccines and identify people who are particularly susceptible to certain infections.