Cytomegalovirus (CMV) infection remains a significant cause of morbidity and mortality in young children. We have previously shown that CD8+ T cell responses to CMV pp65 or IE1 protein were readily detectable in children with congenital or postnatal CMV infection. Here, we have further characterized the evolution of the peptide specificity of these responses in 7 infants <6 months of age at the start of the study. Thirteen pp65 and 15 IE1 peptides (median, 5 peptides/infant) were targeted, and most (61%) represented sequences not previously reported. Peptide specificity remained stable or broadened over time despite the clearance of CMV viremia. Loss of peptide recognition was not observed. Responses with the highest functional peptide avidity were not necessarily detected earliest. These data provide additional evidence that young infants can generate diverse CMV-specific CD8+ T cell responses but show that early responses may exhibit relatively focused peptide specificity and lower peptide avidity.
Human cytomegalovirus (CMV) infection remains the most common congenital infection worldwide . Infants who acquire CMV in utero during maternal primary infection or reactivation may develop severe neurologic sequelae that result in long-term disability . The virologic and/or host factors that contribute to protection from severe CMV disease are not definitively known. Strategies for the prevention of CMV infection and transmission, including the development of a vaccine, may decrease the morbidity and mortality of this infection and form the basis for lifetime immunity .
CMV-specific CD8+ T cell responses have been associated with the control of viral replication and protection from severe disease in murine  and adult human [4, 5] CMV infection. In studies of adults with chronic infection, the CMV proteins pp65 (UL83) and immediate-early 1 (IE1; UL123) have been identified as major targets of CD8+ T cells [6–8], and an increasing number of viral epitopes are being defined [9, 10]. Moreover, early responses may play a particularly important role in the control of viremia, disease outcome, and induction of CD8+ T cell memory [11–15].
Although we [16–18] and others [19–23] have shown that the human fetus and infant are capable of generating virus-specific T cell responses, the age-related properties of cell-mediated immunity in young children are poorly understood . Viral infection in utero and in young infants has been associated with delayed onset, lower frequency, or impaired function of virus-specific T cell responses [18, 21, 25, 26]. In contrast, other studies have reported similar frequency, differentiation pattern, or effector function of these responses in children and adults with primary viral infection [16, 19, 20, 22]. Taken together, these studies highlight our limited understanding of protective virus-specific cell-mediated immunity in early life. More-complete knowledge of infant CMV-specific CD8+ T cell responses may contribute to the development of effective prevention and treatment strategies for CMV infection.
Our previous work showed that CD8+ T cell responses to the CMV proteins pp65 and IE1 were commonly detectable in children with congenital or postnatal CMV infection . In the present study, we examine CD8+ T cell recognition of pp65 and IE1 peptides over the course of primary CMV infection in infants and examine the role played by functional peptide avidity in the evolution of these responses.
Subjects, Materials, and Methods
Study population. Sufficient cells from 7 CMV-infected infants (4 congenital and 3 postnatal) were available for study. Three CMV-uninfected infants served as control subjects.CMV infection was diagnosed as described elsewhere . Because the mechanism of the variation in disease severity that is often observed in congenital versus postnatal CMV infection is unknown and may or may not be immune mediated, infants were included in the study regardless of clinical symptoms or timing of infection. The present study was approved by the human subjects committees at participating institutions, and written, informed consent was obtained from a parent or legal guardian.
Generation and maintenance of B and CD8+ T cell lines. Peripheral-blood mononuclear cells (PBMCs) and B lymphoblastoid cell lines (BLCLs) were processed as described elsewhere . CD8+ T cell lines were generated by incubating PBMCs in R10 medium  with anti-CD3, 4b (final concentration, 2.5 μg/mL ), interleukin-2 (50 U/mL), and L-glutamine (4 mmol/L) and were stimulated with irradiated allogeneic PBMCs and anti-CD3 every 2–3 weeks for a maximum of 6 weeks. Cell lines were >80% CD8+ T cells, as determined by flow cytometry.
Molecular HLA typing. Molecular class I HLA typing was performed using BLCLs according to the manufacturer's instructions (GenoVision).
CMV peptides. Peptides (20 aa [Mimotopes] or 9–11 aa [QCB]) were stored at −80°C at a working concentration of 1 μμ/mL in dimethyl sulfoxide.
Detection of CMV DNA in the peripheral blood. CMV load was determined as described elsewhere  or as follows. Crude lysates of whole-blood-cell pellets were prepared using extraction reagent containing 10 mmol/L Tris-KCl buffer, 0.1% SDS, and 0.01% proteinase K. Total DNA was ethanol precipitated and resuspended in DNAse- and RNAse-free water. Polymerase chain reaction (PCR; 20 μL) was performed using the CMV UL54 Primer/Hybridization Probes Kit (Roche Diagnostics) in a Roche LightCycler Real-Time PCR thermocycler, in accordance with the manufacturer's protocol. The quantity of CMV DNA per sample was calculated from the standard curve of serially diluted control plasmids (106–102 copies) containing a cloned fragment of the UL54 gene (CMV UL54 Template Set; Roche Diagnostics). CCR5 copy number was used to measure the number of PBMC equivalents per sample and was determined as described elsewhere  using SYBR Green in the Roche LightCycler. The specificity of the CCR5 amplicons was confirmed by melting-curve analysis. CMV load is expressed as the number of copies of CMV DNA per 105 PBMCs.
Enzyme-linked immunospot (ELISpot) assay for CMV peptide mapping. CD8+ T cell lines were incubated with peptide (final concentration, 10 μg/mL) in an interferon (IFN)—γ ELISpot assay as described elsewhere . Pools of peptides (20 aa overlapping by 10 aa) spanning CMV pp65 (56 peptides/15 pools) or IE1 (49 peptides/14 pools) protein were used for the initial mapping studies. When 2 pools containing the same 20mer peptide elicited IFN-γ secretion, the peptide was then screened for a response in at least 1 separate ELISpot assay. Responses first detected at later time points were confirmed as undetectable at earlier time points. Truncations (11 aa overlapping by 3 aa) spanning each 20mer peptide were used for the fine mapping studies, as identification of optimal 8–10-aa epitopes was not feasible because of limited cells. Results are expressed as the number of spot-forming cells per 106 T cells.
ELISpot assay for functional peptide avidity. To avoid selection for high-avidity responses by peptide stimulation in culture, non—antigen-specific CD8+ T cell lines were used to determine functional peptide avidity. In the ELISpot assay, cells were incubated with serial 5-fold dilutions of peptide (starting concentration, 10 μg/mL), and the effective concentration that elicited 50% of the maximal spot-forming cell frequency (EC50) was calculated using the Sigmoid Fit tool in Origin software (version 6.0; OriginLab Corporation).
HLA class I restriction analysis. CD8+ T cell lines were incubated with 51Cr-labeled BLCLs (autologous or each of 4 single-HLA-matched allogeneic) at an effector to target cell ratio of 100:1 or with CMV peptide (final concentration, 10 μg/mL) in a 4-h chromium-release assay, as described elsewhere [17, 25]. Controls included spontaneous (51Cr-labeled BLCL and medium or peptide) and maximal (51Cr-labeled BLCL and Renex solution) chromium release in sextuplet wells. Specific lysis >10% above that in medium wells with a background of <30% was considered to be positive.
Antigen specificity of PBMCs compared with that of CD8+ T cell lines. Because limited cells were available for study, CD8+ T cell lines were used for peptide mapping studies. Previous reports  and our present study using cells from CMV-seropositive adults demonstrate that these lines reflect the antigen specificity of PBMCs (figure 1).
Breadth of CMV peptide—specific CD8+ T cell responses in infants. Having previously shown that CMV pp65- and IE1- specific CD8+ T cell responses were commonly detectable , we sought to more fully characterize the evolution of pp65 and IE1 peptide recognition by CMV-specific CD8+ T cells over time in infants with primary CMV infection. Table 1 shows the characteristics of the study cohort. Responses were not detected in CMV-uninfected infants (data not shown).
Peptide nomenclature, sequences, and HLA restriction for CMV peptides recognized by the cohort of 7 infants are summarized in table 2. Overall, CD8+ T cell responses to 28 CMV peptides were detected, including 13 pp65 and 15 IE1 peptides. Recognition of truncated 11mer sequences could not be determined for four 20mer peptides (5, 24, 60 for infant P103, and 63), despite repeated confirmation of CD8+ T cells in lines and additional screening with peptides representing predicted optimal epitopes. The majority of CMV peptides (17/28 [61%]) identified in the study cohort represented previously unreported targets of CMV-specific CD8+ T cells. Eleven (39%) of 28 peptides have been defined as CMV epitopes in adults with chronic infection [6–9, 29–32], but responses to these peptides were not detected in all infants expressing the appropriate HLA allele. For example, responses to peptides NV, TM, QF, and VL (table 2) were detected in only 2 of 3 HLA-A2, 1 of 2 HLAB7, 3 of 5 HLA-A24, and 1 of 3 HLA-A2 infants, respectively.
HLA restriction was defined or confirmed as outlined in table 2, including previously unreported IE1 peptides 23-B (B*5501) and 25-D (A*2402) and pp65 peptide 60-C (A*0201). Although epitopes within IE1 peptides 9-D and 31-D have been reported [6, 9], in our present study, responses to these peptides were restricted by HLA-A*0222 and -A*2402, respectively. For the remaining peptides, HLA restriction was inferred by comparison to epitopes previously reported in adults, epitope prediction models, and/or recognition of peptides by 2 infants sharing a single common HLA allele. In these cases, all infants expressed the relevant reported or predicted restricting HLA allele for the peptide recognized, except infant W401, who recognized IE1 peptide 20-B but did not have the predicted (B*08) or reported (B*07 ) HLA alleles.
Several infants recognizing the same CMV peptide may have used different restricting HLA alleles. IE1 peptide 20-D was recognized by infants W401 and P103, who expressed the predicted HLA-B*44  or reported HLA-B*08 [6, 9] restricting alleles, respectively. Although these infants share HLA-A*2402, this allele is not known to restrict the 20-D sequence. Similarly, IE1 peptide 31-D was recognized by infants B101 and M101. The restricting allele was A*2402 for infant B101 but could not be identified for infant M101. Although HLA-A*2402, which was shared by both infants, could be the restricting allele, infant M101 responded to peptide 31-D (EFCRVLCCYVL) and, with lower frequency, to overlapping peptide 31-E (RVLCCYVLEET) (data not shown). This pattern suggests that infant M101 recognized the epitope CRVLCCYVL, which has been previously shown to be restricted by HLA-B*07  and which was also expressed by this infant.
CMV peptide—specific CD8+ T cell responses broaden over time. Our laboratory  and others [13, 14] have previously demonstrated that the antigen specificity of memory CD8+ T cells detected during the chronic phase of persistent viral infection does not necessarily reflect the repertoire generated during acute infection. We therefore sought to test the hypothesis that CD8+ T cell recognition of pp65 and IE1 peptides changes over the course of primary CMV infection in infants. As shown in figure 2, all infants recognized at least 2 CMV pp65 or IE1 peptides by 12 months of age. The number of peptides recognized by each infant increased (5/7 infants) or remained stable (2/7 infants) over time, and loss of peptide recognition was not observed. Among the 5 infants, initial responses detected at <5 months of age were directed at pp65 in 2 infants, at IE1 in 2 infants, and at both proteins in 1 infant. The other 2 infants (P104 and W103) recognized the same peptides at both time points tested, but lack of available cells prevented evaluation of earlier trends in peptide recognition. Regions of pp65 or IE1 targeted by these responses were dispersed throughout the full protein sequences (figure 2). Of note, the breadth of responses continued to increase after resolution of detectable CMV DNA in the peripheral blood (figure 3). In this small cohort, differences in peptide specificity between infants with congenital or postnatal infection and between those with symptomatic or asymptomatic infection were not observed.
CMV-specific CD8+ T cell functional peptide avidity. The relative sensitivity of T cell populations to antigenic stimulation may be a primary mechanism by which certain viral peptides are recognized before others during the course of viral infection [35, 36]. However, the T cell receptor (TCR) displays distinct properties in neonates compared with adults , and the role played by CD8+ T cell functional avidity in the generation and maintenance of antiviral T cell responses in infants is not well understood. We hypothesized that, for individual infants, the functional avidity of CMV-specific CD8+ T cell responses would be highest for peptides recognized earliest.
Figure 4 shows representative EC50 calculations for pp65 peptide 77-C, recognized by infant W405 at 5 and 12 months of age, and table 3 summarizes EC50 values for CMV peptides targeted by 5 infants. EC50 did not vary by number of input cells (data not shown). Variability in functional avidity was observed for the same peptide recognized by different infants and for multiple peptides recognized by individual infants. For example, CD8+ T cell responses to pp65 peptide 84-B were detected in congenitally infected infants W405 and P103, but the peptide avidity was at least 100-fold higher in infant P103. In contrast, infants P104 and B101 recognized pp65 peptide 68-D with similar avidity. Peptides showing the highest avidity with EC50 <2 ng/mL included those previously defined in adults (31-D, VL, 84-B, and NV [6, 7, 29, 30]) but also the novel sequence 25-D for infant B101. Responses detected at the earliest time points did not necessarily have the highest avidity. For example, for infant P103, the response to IE1 peptide 20-D (11 months) was detected before pp65 peptide 84-B (24 months) but had a lower functional avidity.
We first examined the CMV peptide sequences recognized by the cohort of infants. Most peptides (61%) represented novel sequences not previously reported as targets of CMV-specific CD8+ T cells. Of note, published CMV epitopes have been mapped primarily in healthy adults with chronic CMV infection [6–10]. In a recent study of chronically infected adult volunteers, Sylwester et al.  published a comprehensive analysis of CD4+ and CD8+ T cell responses to CMV gene products using overlapping 15mer peptides spanning the entire viral genome, but this study did not include serial blood sampling or fine epitope mapping. Because of the difficulty in identifying immunocompetent adults with primary CMV infection, direct comparison of peptides recognized by CMV-specific CD8+ T cells in our infant cohort versus that in an equivalent adult control group was not possible in our study. Moreover, to our knowledge, no published studies characterizing the fine specificity of these responses during primary CMV infection in immunocompetent adults are available to provide historical control data.
We then compared CMV peptide recognition at multiple time points over the course of infection for each infant. CMV-specific CD8+ T cell responses broadened in most infants, as shown by an increasing number of peptides recognized over time. In the mouse model, CD8+ T cell responses detected during acute murine CMV infection typically become more focused with respect to the number of gene products recognized, although the frequency of these protein-specific responses may expand or contract with time [38, 39]. In contrast, longitudinal adult human studies of acute viral infection have reported dynamic changes in CD8+ T cell recognition of viral proteins, primarily attributed to differential viral gene expression during lytic versus latent stages of infection [12, 14] or to genetic mutation during prolonged viral replication [11, 13, 40]. Frequencies of CMV peptide—specific CD8+ T cells have been shown to decrease after primary CMV infection , but again, to our knowledge, no published longitudinal studies have examined changes in the breadth of CMV peptide recognition over time using non—HLA-limited epitope mapping assays similar to our work. Although Sylwester et al.  have suggested that analysis of the total CMV-specific T cell response could be accomplished using 15 × 106 PBMCs, this number of cells is rarely available for studies involving infants. Such limited cells necessitated our approach of using CD8+ T cell lines and examining responses to pp65 and IE1 only, and we have therefore likely underestimated the true diversity of the CMV epitope—specific CD8+ T cell responses in the cohort of infants. Although the small study population did not provide sufficient power to demonstrate statistical correlation, no apparent relationship was observed between the diversity of CMV-specific CD8+ T responses and the severity of CMV disease.
The mechanisms that control the evolution of CD8+ T cell peptide specificity in persistent viral infections are not definitively known, but ongoing exposure to viral antigen is likely an important factor [36, 42]. Although expression of particular genes only during latency is not characteristic of CMV, viral transcripts from immediate-early genes are most commonly detected in latently infected cells  and may represent a source of persistent viral antigen leading to expansion of IE1- specific CD8+ T cells over time [3, 4, 16, 37, 44]. However, our data show broadening of peptide-specific responses to both immediate-early (IE1) and late (pp65) CMV gene products over many months despite the resolution of CMV viremia. These findings suggest that productive infection at sites other than the peripheral blood may contribute to the broadening of CMV peptide—specific CD8+ T cell responses. Prolonged viral replication and shedding in tissues after primary CMV infection in early life has been demonstrated in murine  and human  studies. Tissue-specific antigen recognition, perhaps associated with differential T cell homing or the effects of viral immune-evasion mechanisms, may thereby lead to later expansion of CD8+ T cell clones not present during early disseminated infection [39, 47, 48]. Other features of cell-mediated immunity in early life that may contribute to focused CMV-specific CD8+ T cell responses that later broaden with age include deficient CMV-specific CD4+ T cell responses, inefficient antigen presentation or TCR triggering, and lack of heterologous viral infections [23, 24, 49]. Alternatively, responses first detected at later time points may have been present earlier but were not trafficking through the peripheral blood or were not detected by ELISpot assay in our present study because of low frequencies, lack of IFN-γ secretion, or inadequate expansion in T cell culture.
Sequential peptide recognition by CMV-specific CD8+ T cells may also be influenced by functional peptide avidity, a global measure of the ability of T cells to be stimulated by antigen. Neonatal T cells have distinct properties that may affect the generation of antiviral cell—mediated immune responses, including a higher threshold of responsiveness to cytokines and a lower density of cell-surface TCR complexes . We therefore hypothesized that, for individual infants, peptides recognized earlier would have higher avidity than those recognized later, thus providing a possible explanation for preferential CD8+ T cell recognition of certain CMV peptides before others. Our data showed that, although EC50 values varied overall, responses detected earliest did not necessarily have the highest avidity, as predicted. This finding suggests that functional avidity did not play a major role in determining the order in which CD8+ T cells recognized CMV peptides in some infants and is consistent with the findings of other studies of acute viral infection [14, 40]. The heterogeneity of functional avidity observed in our present study also suggests that non—antigenspecific stimulation of CD8+ T cell lines did not simply select for high-avidity clones.
In summary, young infants with primary CMV infection generated CD8+ T cell responses directed at a broad range of peptides spanning pp65 and IE1. Most of these peptides have not been previously reported as targets of CD8+ T cells in adults with chronic CMV infection. Peptide specificity of detectable CMV-specific CD8+ T cell responses broadened over time in most infants despite the resolution of CMV viremia, suggesting persistent antigen exposure at other sites. The functional avidity of these responses varied and did not appear to determine the order of peptide recognition.
We are grateful for the assistance of Wanda DePasquale and Margaret McManus in preparing the manuscript and of Linda Lambrecht in supervising the transport and storage of clinical specimens. We also thank the Women and Infants Transmission Study for providing clinical specimens.