JDMC is employed by Miltenyi Biotech, and declares this as a possible conflict of interest.
Dr H. B. Gaspar, Molecular Immunology Unit, Institute of Child Health, University College London, 30 Guilford Street, London WC1N 1EH, UK. E-mail: firstname.lastname@example.org
Adenoviral infections represent a major cause of morbidity and mortality following haematopoietic stem cell transplantation. Current anti-viral agents are virostatic and it is evident that elimination of adenovirus (ADV) infection is only achieved by recovery of cellular immunity. Using an interferon-gamma (IFN-γ) secretion and capture assay to isolate ADV-specific T cells, followed by a 2 week expansion and restimulation protocol, we generated ADV T cells that may be used for cellular immunotherapy. In contrast to virus-specific T cells for cytomegalovirus or Epstein-Barr virus, the ADV response was dominated by CD4+ T cells and the majority of captured cells exhibited an effector/memory immunophenotype. Highly specific antigen responses were demonstrated by intracellular IFN-γ expression and cytotoxicity assays when the expanded cells underwent restimulation with ADV-pulsed target cells. Although T cells were initially generated in response to ADV species C, the expanded populations also showed strong activity against ADV species B, suggesting cross-reactivity across ADV species; a finding that has important clinical consequences in the paediatric setting, where the majority of infections are caused by ADV type B and C. The protocols can be readily translated to generate ADV-specific T cells suitable for clinical use and offer an effective immunotherapeutic strategy to control ADV infection.
Haematopoietic stem cell transplantation (HSCT) represents the only curative option for certain haematological malignancies and inborn errors. One of the major factors to influence survival outcome is viral infection. Recipient/donor human leucocyte antigen (HLA) disparity often leads to depletion of T cells from the donor graft, which results in a period of profound immunocompromise and renders the recipient susceptible to viral and opportunistic infections. Cytomegalovirus (CMV), Epstein-Barr virus (EBV) and Adenovirus (ADV) represent the most common infections post-transplant and are major causes of morbidity and mortality.
Cellular therapies, used in combination with antiviral agents and monoclonal antibody therapies, have led to highly effective strategies to manage CMV- and EBV-related disease. Current approaches to generate antigen-specific cytotoxic T cells (CTLs) require prolonged culture of antigen-presenting cells [Epstein-Barr virus-transformed lymphoblastoid cell lines (EBV-LCLs) or monocyte-derived dendritic cells (DCs)] for antigen-specific T cell stimulation (Heslop et al, 1996; Peggs et al, 2001; Einsele et al, 2002a) and a number of studies have shown efficacy in preventing and treating infection using this approach (Heslop et al, 1996; Einsele et al, 2002b; Peggs et al, 2003). Immunodominant peptide epitopes have been used to stimulate specific T cell populations (Kleihauer et al, 2001; Vannucchi et al, 2001; Einsele et al, 2002a) but this approach is HLA-restricted and not applicable to all transplant recipients. In most cases, these strategies are time consuming, labour intensive and the cells generated are mostly used prophylactically for high-risk groups and are unnecessary for a large number of subjects. A more cost-effective approach would be to generate ADV-specific CTLs rapidly for individuals in whom high viral loads are detected during the highest risk period early after HSCT.
Recently, technology has been developed that enables the capture of T cells that secrete interferon gamma (IFN-γ) in response to antigen stimulation (Brosterhus et al, 1999; Becker et al, 2001; Bissinger et al, 2002). This technique can be used to isolate antigen-specific T cells and protocols for generation of CMV-specific cells have recently been reported (Rauser et al, 2004). The feasibility of using this approach has also been investigated for capturing ADV-specific T cells, and the present study established a clinically relevant protocol and demonstrated that ADV-specific T cells are highly specific, show species cross reactivity, but reduced alloreactivity. We also show that, unlike the cellular response against CMV and EBV, cytotoxicity against ADV is mediated predominantly by CD4+ T cells.
Isolation of ADV SpC virus-specific cells
Adenovirus lysate was derived from ADV SpC-infected human embryonic lung fibroblasts (HELF), and control lysate (CL) was derived from uninfected HELF cells. Lysate protein was prepared by two freeze/thaw cycles, clarified by centrifugation, and inactivated by boiling. Primary human T cells were isolated from healthy ADV sero-positive adult volunteers by Ficoll/Paque (Pharmacia, Uppsala, Sweden) density gradient separation and resuspended at a cell density of 107/ml in complete culture medium X-vivo 15 (Cambrex, Verviers, Belgium)/5% human AB serum (Sigma, Poole, UK). Cells were stimulated for 16 h with 100 μg/ml of ADV spC lysate or CL at 37°C, 5% CO2.
After overnight stimulation, IFN-γ secreting cells were selected using the IFN-γ secretion assay, cell enrichment and detection kit (Miltenyi Biotec, Bisley, UK). In brief, cells were labelled with a bi-specific monoclonal antibody, specific for IFN-γ and CD45. Cells were diluted in RPMI medium/1% AB serum and incubated at 37°C for the IFN-γ capture period of 45 min. The IFN-γ positive secreting cells were then selected using anti-IFN-γ microbeads and Miltenyi Mini-MACS column. The negative fraction was retained and irradiated (2300 rads) for use as feeder cells.
Expansion protocol of antigen-specific cells
Eluted virus-specific T cells were resuspended at a cell density of 105/ml in X-Vivo/5% AB serum and 50 IU interleukin 2 (IL-2)/ml and kept in culture for 14 d. Feeder cells were added at day 0 at a ratio of 100:1 and day 7 at 10:1. Control cultures were generated using CL stimulated cells, stained as described for the ADV-specific cells, and then divided in two fractions – one half placed directly in culture and the remainder irradiated and used as feeder cells as described above.
Characterisation of cell phenotypes
Cells were resuspended in fluorescence-activated cell sorting (FACS) washing buffer (0·5% bovine serum albumin (BSA; Sigma), phosphate-buffered saline (PBS; Invitrogen, Paisley, UK), 0·2 mmol/l EDTA (Sigma) and incubated for 10 min at room temperature with saturating amounts of antibody (5 μl), then washed and fixed (FacsFix; Becton Dickinson, BD, Oxford, UK). Flow cytometric analysis was performed on 100 000 gated events per sample (10 000 events for DCs) on a Becton Dickinson FACS Calibur and analysed using cellquest software (Becton Dickinson). Cells were stained for CD4, CD8, CD3, CD16+56, CD27, CD19, CD33, CD45RO, CD95 (all BD Bioscience) and CCR7 (RD Systems, Abingdon, UK). Data were collected for seven experiments using six donors.
Cells were surface-stained with anti-CD4, anti-CD8. After one wash with FACS washing buffer, cells were permeabilised with CellPerm (BD Bioscience) for 20 min at room temperature, washed with CytoPerm (BD Bioscience), stained with anti-perforin-fluorescein isothiocyanate (FITC) or a FITC isotype matched control antibody (BD Bioscience), washed and fixed. Flow cytometric acquisition and analysis was performed as described above.
Dendritic cells stimulations
Peripheral blood mononuclear cells (PBMC) separated by density gradient separation on Lymphoprep (Axis-Shield, Oslo, Norway). CD14 positive cells were enriched by magnetic bead selection according to the manufacturers instructions (Miltenyi Biotech) and DCs derived by 6-d culture in X-vivo 15 (Cambrex)/3% human AB serum (Sigma) supplemented with 100 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF; Insight Biotechnology, Wembley, UK) and 25 ng/ml IL-4 (PeproTech, London, UK).
Autologous DCs were pulsed for 6–8 h with CL, ADV SpC lysate, ADV SpB lysate or CMV antigen (Dade Behring, Marburg, Germany). At the end of this incubation period, 12·5 μg/ml Poly (I:C) (Sigma) and 1 μg/ml PGE2 (Sigma) were added and cells were incubated overnight.
Virus-specific T cells were added to pulsed DCs at the indicated concentrations and co-cultured overnight at 37°C. Virus-specific reactivation of the expanded T cells was assessed by intracellular IFN-γ expression and specific cytotoxicity by PKH26 assay.
Intracellular IFN-γ staining
Adenovirus-specific T cells were incubated overnight with pulsed DCs, at a 10:1 ratio, in 96-well plates at 37°C. Cells were then harvested, washed and incubated for 20 min on ice and stained with anti-CD8 phycoerythrin (PE) and anti-CD4 allophycocyanin (APC). Cells were permeabilised by adding perm/fix solution (BD) and stained with saturating amounts of IFN-γ FITC antibody (BD). Flow cytometric analysis was performed for 10 000 events per sample on a Becton Dickinson FACS Calibur and analysed using cellquest software.
T cells from HLA-A1 donors, seropositive for ADV, were stained with ADV-specific tetramers (a kind gift from M. Cobbold, CRUK, Birmingham, UK) before and after stimulation with ADC SpC and expansion for 14 d. Co-staining with anti-CD8 FITC was used to identify virus-specific populations.
PKH26 Red fluorescent cell linker kit FACS-based cytotoxicity assay
Pulsed DCs were loaded with PKH26 dye according to the manufacturers’ instructions (Sigma). Target cells (PKH26-loaded pulsed autologous DCs) were incubated with effectors (ADV-specific T cells) at effector:target ratios of 10:1 for 16 h. Cells were then washed and labelled with Annexin V (BD) and 7-AAD (Pharmigen, San Diego, CA, USA). The cytotoxicity assay was performed three times for four different donors. Flow cytometric analysis was performed for 100 000 PKH26 gated events per sample on a Becton Dickinson FACS Calibur and analysed using cellquest software.
Mixed lymphocyte reactions
Responder T cells and stimulator cells were resuspended at 106/ml in X-vivo 15/5% AB serum. Stimulator cells were irradiated (2300 rads) and 100 μl were plated in triplicate in a 96-U bottomed well plate. Autologous or allogeneic responder cells were added in an equal volume and the cells were cultured for 5 d at 37°C and 5%CO2, before a 16-h pulse with 18·5 kBq /well 3H-thymidine (Amersham Bioscience, Little Chalfont, UK). Plates were harvested onto a filtermat using a Wallac 96 well plate harvester. Radioactive incorporation was measured using a Wallac counter.
Where indicated, statistical significance was calculated using a paired student t-test. P-values < 0·05 were considered significant.
Isolation and expansion of ADV-specific T cells results in generation of predominantly CD4+ effector memory cells.
Stimulated PBMC from six ADV-positive donors showed variability in IFN-γ secretion. In accordance with previous studies, a consistently increased response was seen in CD4+ compared with CD8+ cells. Following the IFN-γ secretion and capture assay, the numbers of eluted virus-specific cells ranged from 1 to 7 × 105cells. After the 2 weeks of culture, a 1·5–2 log expansion was seen with cell numbers ranging from 7·8 × 106 to 6 × 107 (Fig 1A).
To determine the phenotype of the final culture population, standard immunophenotype staining was performed. Control and ADV-stimulated cultured cells from individual donors were stained. At day 0, the mean number of cells staining for CD4+ were 43% compared with 22·4% of CD8+ cells. By day 14, the control culture consisted of a mixed population of CD4+ and CD8+ cells with a mean of 29% CD4+ and 27% CD8+ cells, whereas in the ADV-stimulated cultures, the mean proportion of CD4+ cells was 75% with a mean CD8+ proportion of 4%. In a number of donors the percentage of CD4+ cells in the final culture reached over 90% (Fig 1B). Further characterisation of CD4+ cells shows that more than 90% of CD4+ cells displayed a CCR7−CD45RO+ effector/memory phenotype (Fig 1C) whereas the control culture showed a mixed population of CD4+ subtypes with the majority being CCR7+ CD45RO− naive cells.
Despite the dominance of CD4+ T cells, we were able to detect populations of ADV-specific CD8+ T cells using tetramer staining. T cells from a HLA-A1, ADV-seropositive donor were stained following stimulation with ADV SpC and at 14 d expansion. Fig 1D shows that after this period, ADV-specific CD8 T cells were readily detectable.
ADV SpC T cells show cross-reactivity against ADV SpB
We next determined the specificity and cross-reactivity of the expanded ADV SpC-specific T cells against ADV SpB. Following the 2-week expansion culture, the ADV T cells were restimulated with pulsed autologous DCs and assayed for intracellular IFN-γ expression. When restimulated with ADV SpC pulsed autologous DCs, intracellular IFN-γ production was detected in over 20% of the total ADV T cell culture. When restimulated with autologous DCs alone, CL or CMV-pulsed autologous DCs; negligible IFN-γ production was seen. However, 15% of cells restimulated with ADV SpB-pulsed autologous DCs expressed IFN-γ, suggesting that the cells exhibit significant ADV cross-reactivity (Fig 2A). By gating on CD4+ cells, a similar pattern of IFN-γ expression was observed (Fig 2B). Up to 60% of CD4+ cells secreted IFN-γ, following ADV SpC stimulation and approximately 20% following ADV SpB stimulation, but the response to control stimulation remained very low, ranging from 1·6% to 3% of CD4+ cells.
ADV-specific T cells exhibit ADV-specific cytotoxicity
Cytoxicity of the ADV-specific T cells against pulsed DC targets was demonstrated using a PKH26 dye-based cytotoxicity assay using flow cytometry (see Methods). Unpulsed autologous DCs, autologous DCs pulsed with ADV SpC, CMV or CL remained viable. Autologous DCs pulsed with CL or CMV and incubated with the ADV T cells at a 10:1 effector-target ratio, also remained highly viable. However, when autologous DCs pulsed with ADV SpC were incubated with the ADV-specific T cells, there was significant cytotoxicity with killing of over 60% (Fig 3A). These results clearly demonstrate that the ADV T cells specifically kill adenoviral targets. High levels of cytotoxicity were also seen against ADV SpB-pulsed DC targets (Fig 3B), a result that confirmed the ADV species cross-reactivity previously seen by intracellular IFN-γ secretion (Fig 2A).
CD4+ ADV T cells exhibit ADVspecific cytotoxicity
We showed that the ADV-specific T cells were predominantly CD4+ cells with a minor CD8+ population. To determine whether ADV-specific cytotoxicity was mediated by the CD4+ cells rather than by the CD8+ minority, CD4+ cells were positively selected from the whole T cell culture by magnetic bead isolation. Purities of greater than 98% were obtained (data not shown). CD4+ cytotoxicity was compared with the cytotoxicity of the whole T cell culture (Fig 3B). Cytotoxicity assays were set up as previously described for both populations in parallel under the same conditions. CD4+ purified cells and the whole T cell population showed similar levels of cytotoxicity (over 80%) after incubation with ADV SpC-pulsed autologous DCs. In addition a similar level of killing was seen after incubation with ADV SpB-pulsed autologous DCs. These data demonstrate that the CD4+ population within the ADV-specific T cells is capable of ADV-specific cytotoxicity and also exhibits species cross-reactivity.
ADV T cell lines demonstrate reduced levels of alloreactivity
Adoptive transfer of virus-specific T cells into allogeneic HSCT recipients may induce graft-versus-host disease. We therefore compared (with five donors) the alloreactive capacity of the generated ADV-specific T cells with fresh PBMC from the original donor in mixed lymphocyte culture (MLC). Unmanipulated donor PBMC (A) showed high levels of reactivity against irradiated allogeneic (B*) PBMC (AB*, Fig 4). The cultured ADV T cells (C) showed significantly decreased alloreactive capacity against the same stimulator cells (CB*). Both fresh PBMC and expanded T cells showed similar low levels of proliferation against irradiated autologous stimulators (data not shown). The results demonstrate that the 2-week period of culture and expansion results in a significantly reduced alloreactive potential of the ADV-specific T cell lines when compared with unmanipulated PBMC from the same donor.
ADV T cell lines express high levels of perforin
The mechanism of killing in the ADV T cell lines was then investigated. The two most likely mechanisms are perforin secretion by the cytotoxic T cell population or interaction of Fas/Fas L on DCs and T cells respectively. Analysis of perforin expression by intracellular FACs analysis in day 14 cultures showed high levels of perforin in CD4+ and CD8+ gated cells (Fig 5A). In contrast, PBMC showed no perforin expression in CD4+ cells and only small amounts in CD8+ cells, suggesting that the period of ADV stimulation leads to upregulation of perforin expression in the T cell population. For Fas/Fas L-mediated killing, we would expect to see up regulation of Fas on DCs following ADV infection. However, analysis of unpulsed, CL-pulsed and ADV C-pulsed DCs did not show any significant levels of Fas expression (Fig 5B). These data suggest that ADV T cell line cytolytic activity is mediated through a perforin-dependent pathway.
Adenoviral infection represents one of the major infective complications following HSCT, especially in the paediatric setting. In disseminated disease and in patients with high viral loads, the prognosis is poor. Current treatment options are limited and the drug therapies available are not of proven benefit, and may cause significant complications. The efficacy of Ribavirin has not been conclusively shown and some studies suggest that it has little effect in preventing increase in ADV viral load in HSCT recipients (Hromas et al, 1994; Chakrabarti et al, 1999; Lankester et al, 2004). Cidofovir may offer some benefit in certain patients but can be associated with severe nephrotoxicity (Vandercam et al, 1999; Ljungman et al, 2003). Furthermore, resolution of viraemia is often temporally related to the re-emergence of immunity following withdrawal of immunosuppression (Kampmann et al, 2005). A number of lines of evidence now suggest that a T cell immune response is critical for control and prevention not only of ADV viraemia but also of specific ADV infections, such as haemorrhagic cystitis (Childs et al, 1998; Miyamoto et al, 1998). Such observations have encouraged the development of strategies to generate ADV-specific T cells.
We report a rapid and efficient method for the generation of ADVspecific T cells that can be readily translated to a clinical grade protocol. In contrast to the prolonged culture of whole PBMC populations with ADV stimulators to generate ADV-specific cells, this protocol relies on initial stimulation of cells with ADV antigen and isolation of IFN-γ secreting cells. A further 2-week expansion allows the generation of highly specific ADV T cells that show significant ADV cytotoxicity, species cross-reactivity and low levels of alloreactivity. Starting with 100 ml of peripheral blood, this two-week protocol allows a two log expansion of selected cells, with a final average culture population of 1 × 107 virus-specific cells. This would enable infusions of upto 105 ADV-specific cells/kg for most adults and larger amounts for paediatric recipients. Larger or repeat blood collections could be obtained if a greater number of specific T cells were needed. A major advantage of this protocol is the short 14-d culture period, which allows generation of cells in response to first detection of virus during routine, twice weekly, surveillance using PCR analysis of blood, stool and urine. We would advocate starting patients on Cidofovir if two consecutive surveillance samples detect adenoviraemia, and initiating the process of ADV-specific CTL generation. This is a reactive approach, circumventing the need to generate cells for prophylactic use in all patients at risk, and is therefore less labour intensive and has a more favourable cost:benefit profile.
The use of adoptive T cell therapy following HSCT for control of viral infection has gained significant momentum and trials outlining the use of CMV- and EBV-specific T cells have previously been documented (Heslop et al, 1996; Peggs et al, 2003). Feuchtinger et al (2006), have recently reported clinical experience with ADV T cells following HSCT. In this pilot study, ADVspecific cells were selected by interferon-γ capture and infused without further expansion in vitro. Three of nine patients cleared ADV following T cell transfer, but three further patients showed no specific T cell response and a further rise in ADV load. Our protocol involving 2 weeks of culture will generate a larger number of cells for infusion. The culture period is also likely to generate a more specific and less alloreactive population for infusion.
The final T cell culture following the 2 week in vitro expansion is highly skewed towards CD4+ cells. This may have been determined by the specific culture conditions of our protocol, which may have minimised CD8+ cell expansion, but a number of other studies have also shown that the cellular response to ADV is mediated predominantly by CD4+ T cells (Sester et al, 2002; Heemskerk et al, 2003; Heemskerk et al, 2005). The functional role of the CD4+ population is demonstrated by their effector/memory surface phenotype and by the specific reactivity to ADV restimulation. The CD4+ cytotoxicity demonstrated in these studies following CD4+ cell separation is one of only a few examples of anti-viral CD4+ cytotoxicity (Sun et al, 2002) and seems to be more prevalent for ADV than for other viruses. The mechanism of cytotoxicity is most likely to be dependent on perforin-mediated lysis, and we have seen high levels of CD4+ perforin expression, and others have documented inhibition of CD4+ cell lysis through use of perforin inhibitors (M. Schilham, personal communication). The possibility also exists that these CD4+ cells, despite their phenotype and in vitro function, may serve in vivo to provide help for the smaller number of CD8+ cells and it is very likely that both the CD4+ and CD8+ population have important cytotoxic activity. In support of this, we were able to detect expansions of virus-specific CD8 T cells using tetramer staining. Alternatively, the CD4+ T cells may be mediating direct killing of targets, and we have shown that T cells enriched for CD4 expression harbour cytotoxic potential. Establishing the effectors of the in vivo response can really only be determined in clinical studies where expansion of virus-specific CD4+ and CD8+ cells may be followed over a period of time.
The cross-reactive nature of the response was effectively demonstrated in our studies where, although the T cells were generated against ADV serotype 5 (a species C virus), an equivalent IFN-γ and cytotoxic response was also seen against ADV species B. This has clinical importance since, although species C viruses appear to have the worst prognosis, especially if they disseminate, species A and B viruses can also cause significant clinical sequelae. The adenoviral targets for the final T cell population are yet to be determined. It is likely, however, that components of the hexon protein will be a major focus for the cellular response. Leen et al (2004a,b) isolated CD8+ ADV CTL clones from seropositive individuals and have identified five novel class I restricted epitopes all located in conserved regions of the capsid hexon protein. Other studies have shown CD4+ responses to peptides derived from the hexon protein (L. Veltrop-Duits and M. Schilham, personal communication), although the specific class II restricted epitopes have not been identified. The hexon protein is highly conserved across ADV sub genera. If indeed, CD4+ and CD8+ immunodominant peptides lie with this region, such homology would also explain the cross-reactivity of the cellular response.
Infusion of donor-derived PBMC always carries the risk of graft-versus-host disease (GvHD). The present study showed that our 2-week culture process significantly reduced alloreactivity against allogeneic stimulators in comparison with fresh PBMC from the same donor. Though alloreactive potential was greatly reduced in MLC, there may still be a risk of GvHD. However, such risks are likely to be minimal. Unmanipulated donor lymphocyte infusions of up to 105 cells/kg in the matched unrelated donor setting, or 5 × 104 cells/kg in the mismatched setting, have a very low incidence of GvHD. Similar numbers of ADV-specific T cells numbers would probably be sufficient to provide effective anti-viral immunity, as in vivo expansion of cells is likely to occur in a similar manner to that seen following adoptive immunotherapy for CMV (Peggs et al, 2003; Cobbold et al, 2005).
The translation of this protocol to clinical grade scale-up is achievable. The majority of reagents required for IFN-γ secretion, capture and expansion are now available as clinical grade reagents and the CliniMACs system can be used to select IFN-γ secreting cells on a clinical scale. However, it will be necessary to develop an appropriate clinical grade lysate or ADV antigen for T cell stimulation. Clinical studies of CMV-specific CTLs using this IFN-γ approach are already underway, and if encouraging results are seen, it is likely that ADV-specific T cell therapy using this approach will be possible in the near future.
We would like to thank the Children's Research Foundation who have generously funded I.C. W.Q is the recipient of Leukaemia Research Fund Clinician Scientist Fellowship. MC was supported by an M.S. McLeod fellowship from the Women's and Children's Hospital, Adelaide, Australia.