The CD4 T-cell response to infectious pathogens is important for optimal CD8 CTL responses and for helping microbe-specific antibody responses. These effects have been proven for vaccinia and protection from poxvirus infection (Bennett et al., 1998
; Janssen et al., 2003
; Shedlock and Shen, 2003
; Ashton-Rickardt, 2004
; Bevan, 2004
; Janssen et al., 2005
; Castellino and Germain, 2006
; Castellino et al., 2006
; Ekkens et al., 2007
). Priming of naive CD4 T-cells is accomplished by specialized APC such as monocytes, macrophages, and dendritic cells. These cells have specific phagocytic and protein-degrading machinery and constitutively express HLA class II molecules that can present peptides to T-cell receptors. In humans, the protein products of the polymorphic HLA DRB1 locus and of the less variable DRB3, DRB4, and DRB5 loci form a heterodimer with the invariant HLA DRA1 protein. Not all individuals have DRB3, DRB4, or DRB5 genes. Therefore, the complement of unique DR molecules per person ranges from one (in the instance of homozygosity at DRB1 and the absence of DRB3, DRB4, and DRB5) to four. There is a single DQB locus per chromosome, but both the HLA DQA1 and DQB1 loci are polymorphic. Various combinations of DQA1 and DQB1 proteins can form stable heterodimers. In some individuals, chimeras between the proteins encoded within haplotypes are possible, leading to a maximum complement of four distinct DQ heterodimers per person (Kwok et al., 1993
). In addition to DR and DP molecules, DP heterodimers are encoded by the polymorphic HLA DPB1 and DPA1 loci such that up to four productive pairings are possible per person.
Recently, it has been shown that replication-competent vaccinia modulates cell surface HLA class II levels and antigen presentation to CD4 T-cells in various experimental systems (Li et al., 2005
). It is not known if these effects differentially influence distinct HLA class II molecules. Poxvirus vectors, both replication-competent and –incompetent in human cells, that express heterologous antigens are being increasingly used for infectious disease and malignancy indications to boost specific immunity. These vectors might differ in modulation of class II antigen presentation and thus HLA class II-restricted T-cell responses. The reason that HLA DR appears to dominate the response in terms of HLA locus restriction for CD4 T-cells remains unknown but could be related to cell surface density of HLA molecules or the nature of the APC responsible for priming and amplifying the CD4 response.
Using multi-cytokine ICC, we found that the majority of vaccinia-specific CD4 T-cells in direct ex vivo assays were DR-restricted. Polyclonal responder lines contained components of DP- and DQ-restricted responder cells, but DR-restricted responses were dominant at the bulk and clonal levels. By studying one subject in detail, we determined that multiple DRB gene products were involved in his response.
Because DQ-restricted responses were numerically rare and exclusively restricted by HLA DQB1*0602 in heterozygous persons, we focused attention on determining the fine specificity of DQB1*0602-restricted responses. HLA DQB1*0602 has been associated with several autoimmune diseases (Gersuk and Nepom, 2009
; Miyagawa et al., 2009
), and the database of DQB1*0602-restricted peptides is small, such that the DQ*0602 binding motif is not well defined. We used a panel of each predicted vaccinia ORF, expressed in vitro using linked transcription and translation, in a one-step assay configuration to determine the fine specificity of the minority DQ-restricted clones. The vaccinia ORFeome panel, previously used to interrogate polyclonal responders (Jing et al., 2008
), was readily able to decode clone specificity to the ORF level. This approach required almost 6 plates of 3
H thymidine incorporation assays per T-cell clone, but was able to interrogate in a single step the entire predicted proteome.
The proteome set used was based on the vaccinia strain WR and may be missing a few proteins encoded by the Dryvax™
vaccine used to immunize the test subject. Dryvax™
is genetically related to strain Copenhagen. Sequence comparisons (available at http://www.poxvirus.org/vaccinia_orthologs.asp
) indicate T-cell antigenic genes such as Copenhagen B22R that do not have homologs present in our protein set. We did not detect any clones reactive with whole vaccinia that failed to react with our ORFeome set, so we do not believe omission of these few ORFs has significantly influenced our results.
Our study of a small subset of DR-restricted clones also showed that many different vaccinia ORFs were recognized. To accomplish this, we performed a more economical, higher-throughput 3H proliferation screen, again using the ORFeome protein panel but this time as pools. The proteins were strongly biologically active at final dilutions of up to 1:12 000, and in pools of up to 24 proteins (). In each case, the candidate antigenic ORF, identified from the pattern of positive pools in a matrix array, was confirmed as antigenic when tested as a single protein.
The pool process used only 38 wells of 3H thymidine per clone in the primary screen followed by a few more wells for confirmation. We still had to expand the T-cell clones of interest after they passed initial screening, which consumed time and resources. In ongoing work we are downscaling and speeding up proliferation assays at the screening, ORFeome interrogation, and confirmatory stages to allow us to complete the assignment of T-cell clone specificity from the initial T-cell clone.
Our methods provide one pathway to rapid assignment of CD4 T-cell reactivity to the ORF level. Once reactive proteins are determined, additional approaches are required to define peptide epitopes. We chose to use “brute force” testing of overlapping peptides in antigenic regions, in some cases preceded by truncation analyses. Alternative and complementary methods can independently discover peptide epitopes, and some are also suitable once an antigenic ORF or sub-region has been defined. In the field of orthopoxvirus research, computational algorithms have been applied to discover CD4 epitopes (Moutaftsi et al., 2007
). Recently, a consensus approach using several algorithms proved very efficient for a specific HLA class II allele, leading to a high ratio of “hits” to peptides tested (Calvo-Calle et al., 2007
). Mass spectroscopy of peptides eluted from antigen-loaded APC has also been successful for vaccinia (Johnson et al., 2005
; Strug et al., 2008
). The method discussed in this report is similar to algorithm and mass spectroscopy-driven approaches in that specific HLA loci and alleles of interest can be targeted, but may offer efficiencies of scale and cost.
The methods reported may be applicable in other situations in which CD4 T-cells are of biological interest. Indeed, the entire predicted ORFeome for P. falciparum
and F. tularensis
has been expressed using the same platform reported here, and CD4 T-cell responses are an important component of the immune response to these organisms (Sundaresh et al., 2006
; Doolan et al., 2008
). In the fields of cancer and autoimmunity, it may be rational to screen blood- or tissue-derived T-cell clones for reactivity with whole tumor or inflamed tissue, and then investigate fine specificity with a collection of suitable host organism cDNAs such as those available from several commercial and non-profit agencies. Several human autoimmune diseases are linked with specific HLA class II alleles. The CD4 clone screening pathway that is outlined in the present report in which yes/no reactivity to whole antigen plus self APC and candidate HLA restriction are determined simultaneously can be used to speed up isolation of CD4 clones restricted by the disease-linked allele.
We recognize that CD8 T-cell responses are vital in many infectious and non-infectious conditions and that the E. coli
-derived ORFeome panel described in this report will be difficult to apply to CD8 T-cells given the differences in antigen processing and presentation between T-cell subsets. We have previously reported genomic DNA library approaches to CD8 antigen discovery (Koelle, 2003
; Jing et al., 2005
), and are currently shifting to a virtual library of full-length ORFs. An outstanding challenge is to determine an efficient, inexpensive method to transducer suitable, abundant APC such as autologous, EBV-transformed B-cells with such an ORF collection in a high-throughput, non-toxic fashion for CD8 antigen discovery.
The present report contributes to the collection of known vaccinia epitopes. Knowledge concerning vaccinia antigens and epitopes recognized by vaccinia-specific CD4 T-cells is accumulating at an accelerating pace. On-line databases emphasizing T-cell epitopes in pathogens of biodefense concern (Vita et al., 2008
) and review articles (Kennedy and Poland, 2007
) capture the complexity of this field. Several fundamentally different approaches have been used, with ours having been termed T-cell driven (Yewdell, 2006
). T-cell lines or clones reactive with whole vaccinia are paired with genomic libraries or ORF sets to resolve fine specificity. Using this approach, we previously reported peptide epitopes in ORFs A3L (Jing et al., 2007
; Jing et al., 2008
), L4R, and F11L (Jing et al., 2008
) and in A7L, A33R, A4R, C10L, E4L, H2R, H3L, and L1R (Jing et al., 2007
), including examples of restriction by HLA DR, DP, and DQ molecules. In common with Tang et al
. (Tang et al., 2006
), we also specifically targeted vaccinia proteins that are known IMV neutralizing antibody targets, and found a discrete, novel epitope in ORF A27L. At the ORF level, we previously showed that more than 50% of known vaccinia ORFs are CD4 antigens within a relatively small set (11) of subjects. Overall, we can anticipate an almost overwhelming complexity of CD4 responses both within individuals and within the population.