Divergent memory responses driven by adenoviral vectors are impacted by epitope competition

Abstract Adenoviral vectors induce robust epitope‐specific CD8+ T cell responses. Within the repertoire of responses generated both conventional memory evolution and the phenomenon of memory inflation are seen. The rules governing which epitopes inflate are not fully known, but may include a role for both antigen processing and competition. To investigate this, we looked at memory generated from vectors targeting the Gp33‐41 (KAVYNFATC/K9C) epitope from the gp of lymphocytic choriomeningitis virus (LCMV) in mice. This well‐described epitope has both the Gp33‐41 and Gp34‐41 epitopes embedded within it. Vaccination with a full‐length gp or a minigene Ad‐Gp33/K9C vector‐induced conventional memory responses against the immunodominant Gp33/K9C epitope but a strong inflationary response against the Gp34/A8C epitope. These responses showed sustained in vivo function, with complete protection against LCMV infectious challenge. Given the unexpected competition between epitopes seen in the minigene model, we further tested epitope competition using the full‐length Ad‐LacZ (β‐galactosidase) model. Generation of an Ad‐LacZ vector with a single amino acid disruption of the inflationary β‐gal96‐103/D8V epitope transformed the β‐gal497‐504/I8V epitope from conventional to inflationary memory. This work collectively demonstrates the importance of epitope competition within adenoviral vector inserts and is of relevance to future studies using adenoviral vectored immunogens.


Introduction
T cell memory inflation is well described in the context of murine CMV (MCMV) infection [1]. Certain epitope-specific CD8 + T cell populations are noted to expand after an initial viral infection epitopes have been identified, β-gal 96-103 /D8V (DAPIYTNV), which inflates, and β-gal 497 -504 /I8V (ICPMYARV), which demonstrates conventional memory. The Ad-LacZ model has been shown to replicate what is seen in MCMV memory inflation [6,7].
Memory inflation has been previously described to depend upon the antigen-processing context of the epitope. Inflationary epitopes tend to be immunoproteasome (specifically, low-molecular mass protein-7/LMP7)-independent [6,8]. We reasoned that adenoviral vectors expressing epitopes as "minigenes" (requiring no further processing) may allow conversion of conventional/non-inflationary epitopes to inflationary ones, with previous experiments validating this concept. Both I8V (Ad-LacZ) and M45 (MCMV) are examples of immunoproteasome (LMP7)dependent epitopes that induce conventional responses when processed from their natural context but that can drive strong inflationary responses if the antigenic context is modified [9].
To look at this further, we developed a lymphocytic choriomeningitis virus (LCMV) minigene model. This includes a full-length gp vector Ad-Gp, and minigene vectors; Ad-Gp33 (K9C:KAVYNFATC) and Ad-Gp34 (A8C:AVYNFATC). We were surprised to find that the CD8 + T-cell responses to all vectors were dominated by an inflationary population directed against the shorter Gp34/A8C epitope, compared to a conventional memory response against the Gp33/K9C epitope, suggesting a role for competition (with both epitopes showing LMP7-independence [10]). To assess this further, we directly tested the role of competition between epitopes within the Ad-LacZ model. We designed a fulllength Ad-LacZ vector with the β-gal 96-103 /D8V epitope disrupted (Ad-disrupted D8V (LacZ)) using a single amino acid mutation in an anchor residue. This vector induced a dramatically altered, inflationary response from the β-gal 497-504 /I8V epitope, where normally a conventional response is seen. This work has implications for models of memory inflation and, translationally, immunodominance in response to adenoviral vector immunogens.

Results and discussion
Ad-Gp and Ad-Gp33 vectors generate an inflationary response, but only against the Gp34/A8C epitope Initially we analyzed CD8 + T cell responses in mice immunized with the full-length LCMV Gp (Ad-Gp) vector. We tracked both Gp33 (KAVYNFATC/K9C) and Gp34 (AVYNFATC/A8C) tetramer + CD8 + T cell responses in blood, following intravenous immunization into C57BL/6 mice. The Gp33/K9C tetramer + populations displayed a conventional/non-inflationary kinetic. In contrast, the Ad-Gp immunization induced a robust Gp34/A8C tetramerspecific CD8 + T cell response, with a sustained inflationary kinetic (Supporting Information Fig. 1).
Based upon previous minigene work, we had expected the Ad-Gp33 minigene to generate an inflationary response to the Gp33 epitope. However, the results following immunization showed the same divergent responses as seen in the Ad-Gp construct. Gp33/K9C epitope-specific CD8 + T cells showed conventional con-traction and emergence of central memory pools without enrichment in peripheral organs. However, Gp34/A8C epitope-specific CD8 + T cells showed memory inflation with a sustained peripheral kinetic over time accompanied by an effector memory phenotype and enrichment in organs (Fig. 1a-c). Overall these data demonstrate that two divergent memory responses can be generated from the same peptide immunogen in parallel, even when processing requirements are bypassed.
It was originally noted in studies of a recombinant MCMV expressing LCMV Gp33 at the N terminus of IE2 that inflation was seen against Gp34, and no sustained response observed against Gp33 [11]. In the setting of MCMV, there are numerous other endogenous epitopes that may be competing and additionally immune evasion genes that could influence peptide loading. However, in the current experiments, peptide processing is bypassed and only two epitopes are present. Furthermore, responses to both epitopes are equally well primed by the full-length and the minigene vectors, but only the Gp34 response is sustained. The existence of two epitopes within Gp33-41 has long been recognized, as is the immunodominance of these epitopes seen in natural infection [12]. The importance of ER aminopeptidase 1 (ERAP1) and its function of trimming MHC class I-presented peptides in vivo and the role of this in immunodominance are also well recognized [13].
Other factors that may influence this outcome of divergent memory responses include TCR avidity of the responses, naïve precursor frequencies, and peptide stability. The discrete nature of the differences seen in the populations suggests a major difference in sustained peptide-MHC availability rather than small quantitative differences that may impact on immunodominance. Finally, the MHC may be of relevance, where the Gp34 epitope is restricted by H-2Kb, compared to that of the Gp33 epitope, which is restricted by H-2Db [14].
Thus, the simplest explanation for our data, since we know that long-term antigen availability drives memory inflation, is that the two different antigen-specific T-cell populations compete for peptide-MHC signals on APCs during the memory phase. This would fit with a model whereby peptides are cross-presented on dendritic cells during priming, allowing presentation of many epitopes but directly presented during the memory phase, at which point competition for a more limited resource at the surface of a nonhematopoetic long-lived cell may become critical. In other words, even if both epitopes are independently presented on different MHC molecules, the Gp34-specific T cell pool can outcompete the Gp33 pool in vivo in the critical niche for antigen re-encounter.

Ad-Gp and Ad-Gp33 vectors generate a protective response
We then went on to look at whether these responses would provide protection against challenge. Mice immunized with the Ad-Gp or the Ad-Gp33 minigene vector received an LCMV challenge at day 50. Figure 1d shows the quantification of virus (measured by www.eji-journal.eu plaque assay) present in full-length Ad-Gp or Ad-Gp33 minigeneimmunized mice in the spleen (i) and inguinal lymph nodes (ii) at day 4 after LCMV WE challenge, compared to naïve controls. No virus could be detected in neither the Ad-Gp nor the Ad-Gp33 minigene immunized mice at this time-point. The system demonstrates that minigene vectors can afford full protection, directly comparable to that of full-length proteins. A number of studies have previously demonstrated protection, including those with the adjuvant effect of linkage to the invariant chain [15,16] and those with full-length gp inserts [17,18]. We had predicted that our full-length Ad-Gp vector would recapitulate this robust protection, and here, we show that the minimal Gp33 epitope was similarly sufficient.

An Ad-Gp34 minigene vector further enhances the inflationary response over the Ad-Gp33 minigene
The Ad-Gp34 minigene was later synthesized, to assess Gp34/A8C tetramer-specific responses and to gauge the maximum level of response. The kinetic of inflationary cells in blood against the Gp34/A8C epitope revealed that inflation was significantly enhanced when compared to Ad-Gp33 (Fig. 2a). Again, the inflationary kinetic was sustained, as demonstrated by both the phenotypic markers (Fig. 2b), enrichment within organs (Fig. 2c), and functionality (Supporting Information Fig. 2).

An Ad-disrupted D8V (LacZ) vector allows for inflation of I8V specific CD8 + T cell responses
Having shown the potential impact of competition between epitopes on inflation using the Gp33 and Gp34 minigene vectors, we were prompted to revisit the role of antigen context and competition in longer length inserts. We generated an Ad-LacZ vector with the inflationary D8V epitope disrupted (nonfunctioning) through modification of an anchor residue. Figure 3a(i) shows the absence of D8V-tetramer specific responses in the Ad-Disrupted D8V (LacZ) vector, while Figure 3a(ii) shows the inflationary kinetic in blood from I8V-tetramer specific responses from this same vector. The features of these cells are inflationary, demonstrated through the inflationary phenotype (Fig. 3b), high distribution in peripheral organs (Fig. 3c), and the functionality (Supporting Information  Fig. 3). This striking result suggests that in the absence of interepitope competition even naturally processed non-inflationary epitopes can promote inflation.

Concluding remarks
Overall, this work suggests a hierarchy of factors involved in deciding which epitopes are likely to drive inflation from within the context of the larger protein. These factors may include antigen context, processing (including immunoproteasome independency), binding affinity, MHC, and possibly further associated fac-tors from the vector (adenoviral or MCMV) itself. Mutation of a single amino acid within a protein antigen can have a huge impact on the memory behavior of a T cell response to an entirely independent peptide. These data are of relevance to the development of adenoviral vectored vaccines, in consideration of the chosen peptide immunogen.

Lymphocytic choriomeningitis virus
LCMV WE (obtained from R.M. Zinkernagel) was propagated on mouse L929 fibroblast cells at a low MOI and was quantified as described [19]. Mice were infected i.v. with 200 plaque-forming units (pfu) of LCMV WE . Organs (spleens and inguinal lymph nodes) were then harvested at day 4-post infection for plaque assays.

LCMV plaque assays
LCMV plaque assays were performed using spleen and inguinal LN samples harvested from infected mice. Plaque assays were performed using permissive MC57 cell lines, and following methods as previously described [19][20][21].