Successful isolation and expansion of CMV-reactive T cells from G-CSF mobilized donors that retain a strong cytotoxic effector function


Correspondence: Mark W. Lowdell, Department of Haematology, University College London Medical School, University College London, London NW3 2PF, UK.



Cytomegalovirus (CMV) infections post-haematopoietic stem cell transplantation (HSCT) can be effectively controlled through the adoptive transfer of donor-derived CMV-specific T cells (CMV-T). Current strategies involve a second leukapheresis collection from the original donor to manufacture CMV-T, which is often not possible in the unrelated donor setting. To overcome these limitations we have investigated the use of a small aliquot of the original granulocyte-colony stimulating factor (G-CSF) mobilized HSCT graft to manufacture CMV-T. We explored the T cell response to CMVpp65 peptide stimulation in G-CSF mobilized peripheral blood mononuclear cells (PBMC) and subsequently examined isolation of CMV-T based on the activation markers CD154 and CD25. CD25+ enriched CMV-T from G-CSF mobilized PBMC contained a higher proportion of FoxP3 expression than non-mobilized PBMC and showed superior suppression of T cell proliferation. Expanded CMV-T enriched through CD154 were CD4+ and CD8+, demonstrated a high specificity for CMV, secreted cytotoxic effector molecules and lysed CMVpp65 peptide-loaded phytohaemagglutinin-stimulated blasts. These data provide the first known evidence that CMV-T can be effectively manufactured from G-CSF mobilized PBMC and that they share the same characteristics as CMV-T isolated in an identical manner from conventional non-mobilized PBMC. This provides a novel strategy for adoptive immunotherapy that abrogates the need for successive donation.

Infectious complications arising from the reactivation of human cytomegalovirus (CMV) following allogeneic haematopoietic stem cell transplantation (HSCT) continue to be a significant cause of morbidity and mortality (Chakrabarti et al, 2002; Gooley et al, 2010) with the incidence of reactivation being reported to be as high as 70% (Reusser et al, 1991). Reduction in the incidence of CMV disease has been achieved through the pre-emptive and prophylactic use of anti-viral drugs (Goodrich et al, 1993; Einsele et al, 1995). However, their use has not been effective in improving overall survival and they are also associated with late-onset CMV disease.

Several strategies have been employed to manufacture CMV-specific T lymphocytes (CMV-T) for clinical use in the treatment of CMV disease in patients at a high risk of reactivation (Riddell et al, 1992; Walter et al, 1995; Einsele et al, 2002; Peggs et al, 2003, 2011; Cobbold et al, 2005; Leen et al, 2006; Micklethwaite et al, 2007; Mackinnon et al, 2008). These include generation of T cell lines in short-term culture and direct selection of γ-interferon (IFN-γ) secreting cells in response to CMVpp65 stimulation. Antigen-specific T cells have also been identified through a number of T cell surface markers that are up-regulated after activation, including CD25, CD69, CD137 and CD154 (Gallot et al, 2001; Chattopadhyay et al, 2005; Frentsch et al, 2005; Wolfl et al, 2007; Watanabe et al, 2008; Wehler et al, 2008). CD154 is predominantly expressed on activated CD4+ T cells and has been previously reported as a promising candidate for identification and expansion of CMV-T due to its high specificity and sensitivity (Frentsch et al, 2005). The generation of antigen-specific T cells against multiple targets from fresh peripheral blood in short-term culture using CD154 expression has also recently been described (Khanna et al, 2011). The interleukin-2 α (IL-2α) receptor, CD25, is also a promising candidate for targeting the isolation of anti-viral T cells and its application in the generation of CMV-T has previously been demonstrated (Gallot et al, 2001). Furthermore, clinical grade CD25 antibodies are freely available and have been effectively used with steady-state leukapheresis samples in large scale manufacture procedures (Powell et al, 2005). Identification of antigen-specific T cells through CD25 expression can be confounded by the similar CD25 expression pattern exhibited by regulatory T cells (Tregs). What remains uncertain in this setting is whether isolating CMV-T based on CD25 activation-induced expression, preferentially favours selection of antigen-specific Tregs with suppressive properties rather than one of an effector anti-viral response, which would be a prerequisite in this setting.

To date, models for generating CMV-T have focused primarily on using peripheral blood mononuclear cells (PBMC) collected by leukapheresis from the original HSCT donor. In the case of related donors, procuring a second apheresate for CMV-T generation is inconvenient to the donor and is associated with some level of pain and discomfort. Obtaining a second apheresate from unrelated donors has proven more difficult, either due to donor refusal, registry refusal or simply scheduling difficulties. The prospect of manufacturing antigen-specific T cells from an aliquot of the original HSCT obtained by leukapheresis after mobilization by recombinant human granulocyte-colony stimulating factor (G-CSF) as an alternative PBMC source is attractive but has yet to be investigated. However, the effective demonstration of γδ T cell isolation from G-CSF mobilized donors that synthesize immunomodulatory cytokines, such as IFN-γ and tumour necrosis factor α (TNF-α), and retain a strong tumoricidal activity (Otto et al, 2005) illustrates the potential for identification and isolation of anti-viral T cells from the same starting material.

Granulocyte-colony stimulating factor mobilization is associated with a significant increase in the numbers of circulating lymphocytes (Sica et al, 1996; Rutella et al, 1997; Hartung et al, 1999). Murine and human studies have suggested that G-CSF mobilization inhibits type 1 cytokine production by T cells, through inhibition of secretion at a single cell level as well as reducing the fraction of cytokine-secreting cells in the periphery; arguing against the use of these cells for adoptive immunotherapy (Pan et al, 1999; Arpinati et al, 2000; Tayebi et al, 2001). In contrast, it has been reported that use of donor lymphocyte infusions (DLI) obtained following G-CSF mobilization for the treatment of relapse in advanced myeloid malignancy compares favourably with other treatment approaches (Levine et al, 2002), suggesting maintenance of a graft-versus-leukaemia (GvL) effect mediated by functional T cells within the lymphocyte population. What remains unclear is the degree of functionality in terms of the anti-viral response from within the T cell compartment of G-CSF mobilized PBMC and how these cells compare against conventional non-mobilized T cells that are traditionally used for adoptive transfer of CMV immunity post-allogeneic HSCT.


Blood donors and cell preparation

Fresh blood samples were obtained from leukapheresis collections from both G-CSF mobilized and non-mobilized CMV-seropositive healthy donors, after a 3–5 h collection on the COBE Spectra apheresis system (Caridian BCT, Lakewood, CO, USA). Informed consent was obtained in accordance with the Declaration of Helsinki and studies were approved by a national research ethics review board. PBMC were isolated from all donor samples after density gradient separation using Lymphoprep (Axis Shield Diagnostics, Dundee, UK) and subsequently cultured in RPMI 1640 medium supplemented with 1% antibiotic (both Life Technologies, Paisley, UK) and 10% heat inactivated human AB serum (Biosera, Ringmer, UK) at a concentration of 1 × 107/ml. PBMC were stimulated for up to 24 h in flat-bottomed six-well culture plates (NUNC, Roskilde, Denmark) with 1 μg/peptide/ml of CMVpp65 peptide pool spanning the entire CMVpp65 protein (CMVpp65 Peptivator; Miltenyi Biotec, Bergisch Gladbach, Germany) and incubated at 37°C/5% CO2. For CD154 experiments, cultures were stimulated in the presence of 1 μg/ml anti-CD40 antibody (BioLegend, San Diego, CA, USA). In some experiments, freshly isolated PBMC were cryopreserved at a 1:1 ratio with human serum albumin (HSA) 4·5% (Bio Products Laboratory Ltd, Elstree, UK) containing 20% dimethylsulphoxide (WakChemie, Steinbach, Germany) according to standard protocols, for use in future experiments as well as a source of feeder cells.

Flow cytometric analysis

Flow cytometry experiments consisted of four to six colour panels, where a minimum of 50 000 CD3+ events were acquired after gating of viable lymphocytes using forward scatter (FSC) and sideways scatter (SSC) signals on a FACScan flow cytometer (Cytek, London, UK) and data analysed using FlowJo version 7.6 (TreeStar Inc., Ashland, OR, USA). For control staining of cytokines and activation markers we used phycoerythrin (PE)-conjugated mouse antibodies of matching isotype and supplier of the specific antibodies. Cells were stained for 15 min in the dark, washed in 2 ml of Hanks Balanced Salt Solution for 5 min and resuspended in 200 μl of FACS Flow (BD Biosciences, Franklin Lakes, NJ, USA) before acquisition.

Cytokine analysis by cytometric bead array

Supernatants were collected from CMVpp65-stimulated and unstimulated control cultures at 16–24 h from both mobilized and non-mobilized donors and stored at −80°C. Analysis by Cytometric Bead Array Kit (BD Biosciences) was used to quantify the level of IL-2, IL-4, IL-5, IL-10, IFN-γ and TNF-α. Cytokine bead standards were used where the concentration of recombinant protein in each serial dilution allowed for the construction of a standard curve using the mean fluorescence intensity (MFI) from each bead standard. Cytokine analyses of supernatants from CMVpp65-stimulated and unstimulated PBMC were performed on a FACS Aria flow cytometer (BD Biosciences) and a minimum of 20 000 events collected. Analysis of acquired data was performed using fcap array Software version 1.0.1 (Soft Flow Hungary Ltd, Pecs, Hungary). The concentration of cytokine was calculated using the standard curve and expressed as a net value (pg/ml) after subtraction of the unstimulated control.

Time course assay of activation marker kinetics

Peripheral blood mononuclear cells isolated from mobilized and non-mobilized donors were stimulated in 96-well plates at a concentration of 1 × 107/ml for 24 h with either CMVpp65 Peptivator or 1 μg/ml of Staphylococcal Enterotoxin B (SEB; Sigma-Aldrich, Gillingham, UK) or left untouched. Samples were taken at 1, 4, 6, 16 and 24 h and stained with allophycocyanin (APC)-conjugated anti-CD3, fluorescein isothiocyanate (FITC)-conjugated anti-CD4, perdinin chlorophyll (PerCP)-conjugated anti-CD8 and either PE-conjugated anti-CD154, anti-CD25, anti-CD69 or anti-CD137 (all BD Bioscience). The percentage of activation marker up-regulation was calculated as net expression in the CD3+ population after subtraction of the unstimulated control. In some experiments PBMC were stimulated for 4 and 6 h in the presence and absence of 1 μg/ml of purified anti-CD40 antibody (BioLegend) and samples stained as described but with only CD154-PE or Mouse IgG1-PE antibodies. As a control, PBMC were left untouched and incubated with anti-CD40 alone.

Isolation of antigen-specific T cells

For the isolation of antigen-specific T cells following CMVpp65 stimulation, cells were either stained with PE-conjugated anti-CD25 after 16 h or PE-conjugated anti-CD154 after 6 h (both BD Bioscience); these times had been derived from previous optimization experiments. Labelling was performed for 20 min using 10 μl of antibody per 107 cells in 100 μl of CliniMACS buffer. After a 20-min incubation with PE-conjugated microbeads (20 μl/107 cells) in 80 μl of CliniMACS buffer the cell suspension was enriched using MS columns on a MiniMACS (all Miltenyi Biotec). All incubation steps were performed at 4–8°C in the dark. Antigen-specific T cells were also isolated using the IFN-γ secretion assay according to the manufacturer's recommendation (Miltenyi Biotec) and isolation was identical to that of CD154 and CD25 separation. Enriched cells were stained with CD3-APC, either CD154 PE, IFN-γ PE or CD25 PE, CD69-APC Cy7, CD8 PerCP and CD4 FITC to determine the phenotype. All antibodies were purchased from BD Biosciences except IFN-γ PE and CD25 PE (Miltenyi Biotec).

Identification of T regulatory cells by FoxP3 Staining

1 × 106 PBMC pre-stimulation, CMVpp65-stimulated and CD25-positive fractions were stained with CD3-APC, CD25-PE, CD4-APC Cy7 and CD8-PerCP, then fixed and permeabilised using the FoxP3 staining kit (BD Bioscience) before staining with either FoxP3-Alexa Fluor 488 monoclonal antibody or IgG isotype control. Samples were acquired on the FACScan flow cytometer with a minimum of 50 000 CD4+ events recorded.

Carboxyfluorecein diacetate succinimidyl ester (CFSE)-based suppression assay

The CFSE-based suppression assay was used to assess the suppressive capacity of CMV-T isolated through the activation marker CD25 after a 16-h stimulation with CMVpp65 peptides and compared directly against the CD25-negative fraction. PBMC were labelled with 1 μmol/l of CFSE (CellTrace CFSE; Life Technologies) and cultured in round-bottomed 96-well plates (NUNC) with a total of 2 × 104 CFSE-labelled PBMC/well. Depending on the number of effector cells available, CD25+ CMV-T and CD25-PBMC were added to the cultures at a ratio of 1:1 (2 × 104 cells/well) and 1:2 (4 × 104 cells/well) of responders to effectors. Cell cultures were stimulated with 1 μg/ml of SEB (Sigma-Aldrich). In control experiments CFSE-labelled and unlabelled PBMC were cultured alone. Cultures were incubated for 5 d at 37°C/5% CO2 before harvesting and cells stained with CD3-APC and CD69-APC Cy7 antibodies (both BD Biosciences). Samples were acquired on the FACScan flow cytometer and a minimum of 10 000 CD3+ CFSE+ events recorded. The frequency of suppression was analysed by gating on the CD3+ CFSE+ population and the percentage of undivided cells (CFSEhigh) determined using the CFSE-labelled and unlabelled PBMC control samples.

Expansion of antigen-specific T cell lines

After isolation of CD154+ CMV-T, up to 0·25 × 106 cells were cultured in the presence of 12·5 × 106 (50:1) γ-irradiated (30 Gy) autologous PBMC to act as feeder cells in 24-well plates with RPMI 1640 medium containing 10% human AB serum, 1% antibiotic and supplemented with 10 ng/ml of IL-7 and IL-15 (CellGenix, Freiburg, Germany). Culture medium was replenished every 2–3 d and cells split when necessary. Cells were expanded up to a maximum of 23 d before harvest.

Re-stimulation of expanded antigen-specific T cell lines

Expanded cells were restimulated for a period of 5–6 h with either CMVpp65 Peptivator-loaded autologous PBMC or untouched autologous PBMC as a control, all labelled with 1 μmol/l CFSE (Sigma-Aldrich) at a ratio of 5:1 at a concentration of 1 × 107/ml in 48-well plates. For analysis of intracellular cytokines, cells were incubated in the presence of anti-CD28 antibody (BD Bioscience) and 1 μg/ml of Brefeldin A (Sigma-Aldrich) added after 2 h. Cells were fixed and permeabilized using Intrastain (DakoCytomation, Ely, UK) according to the manufacturer's instructions and stained with CD154 APC, CD4 PerCP either PE-conjugated anti-IL-2, anti-TNF-α, anti-Granzyme B or anti IFN-γ and CD69 APC Cy7 monoclonal antibodies (all BD Biosciences). For surface staining, cells were incubated in the presence of anti-CD40 antibody and then stained for 15 minwith CD4 FITC, CD154 PE, CD8 PerCP, CD3 APC and CD69 APC Cy7 monoclonal antibodies (all BD Biosciences).

Cytotoxicity assay

Autologous PBMC were stimulated with 3 μg/ml of phytohaemagglutinin (PHA; Sigma) for 24 h and then 20 u/ml of IL-2 (Miltenyi Biotec) for a further 72 h at a concentration of 1 × 106/ml in RPMI 1640 with 10% AB serum, in 24-well flat-bottomed plates (NUNC). PHA blasts were then used as target cells in the killing assay and loaded with CMVpp65 Peptivator, or left untouched. Loaded target cells were labelled with Calcein-AM (Life Technologies) at a concentration of 10 μmol/l and incubated for 1 h at 37°C. After four washes in complete medium, cells were adjusted to 7 × 104/ml and added to effector cells at E:T ratios ranging from 20:1 to 0·5:1, in triplicate, in U-bottomed 96-well plates (NUNC). Triplicate wells were also set up to measure spontaneous release (target cells only), maximal release (target cells plus 2% Triton X-100) and medium alone. After incubation at 37°C/5% CO2 for 4 h, 100 μl of supernatant was harvested and transferred into new plates. Samples were measured using a BMG FLUOstar Galaxy microplate fluorescence spectrophotometer (MTX Lab Systems Inc., Vienna, VA, USA) (excitation filter: 485 ± 9 nm: band-pass filter: 530 ± 9 nm). Data were expressed as arbitrary fluorescent units (AFU) and percent lysis was calculated using the formula [(test release − spontaneous release/maximal release − spontaneous release) × 100].

Statistical analysis

Analyses were conducted using graphpad prism version 4.0 (Graph Pad Software Inc., La Jolla, CA, USA). The nonparametric Mann–Whitney test was used to determine the statistical significance between G-CSF mobilized and non-mobilized PBMC and a Paired t-test for analysing the effect of CD40 blocking on CD154 expression. An unpaired t-test was used for comparison of FoxP3 expression, which followed a normal distribution. Statistical significance was considered to be achieved when P was <0·05.


Cytokine profile and IFN-γ enrichment of CMVpp65-stimulated G-CSF mobilized PBMC

The repertoire of cytokines secreted by G-CSF mobilized PBMC was analysed to determine whether there was equivalence with non-mobilized PBMC. PBMC from CMV+ healthy individuals were stimulated with CMVpp65 overlapping peptides spanning the entire CMVpp65 protein, in 16-h cultures. After 16 h, aliquots of supernatant were taken and assayed for the release of IL-2, TNF, IFN-γ, IL-10, IL-4 and IL-5 (Fig. 1A, B). No significant difference was observed between G-CSF mobilized and non-mobilized PBMC in terms of the effector cytokines IL-2, TNF and IFN-γ, but a significant decrease in IL-10 secretion from G-CSF mobilized PBMC (= 0·002) was detected together with a reduction in secretion of IL-4 and IL-5. Interestingly, we also observed a significant decrease in TNF-α secretion from unstimulated G-CSF mobilized PBMC when compared to non-mobilized PBMC (= 0·026). This decrease was also seen in IFN-γ and IL-10 secretion from unstimulated cultures.

Figure 1.

Functional profile, identification and isolation of CMVpp65-specific T cells in unpaired G-CSF mobilized and non-mobilized donors. Quantitative assessment of IL-2, TNF-α, IFN-γ, IL-10, IL-4 and IL-5 in the supernatant of cultures after 16 h of CMVpp65 stimulation was performed by cytometric bead array in six G-CSF mobilized (A) and six non-mobilized (B) donors. ND indicates where cytokines were not detectable. Identification and isolation of IFN-γ secreting CMV-specific T cells was performed with PBMC samples fromnon-mobilized (n = 6) and G-CSF mobilized (n = 6) leukapheresis stimulated for 16 h with CMVpp65 peptides or left untouched (nil peptide) and the frequency of IFN-γ secreting cells analysed amongst CD3+ T cells (C). IFN-γ secreting cells were isolated using magnetic cell sorting and the purity and yield of IFN-γ+ cells determined from within the CD3+ population (D). G-CSF mobilized (G-CSF Mob; n = 5) and non-mobilized (Non Mob; n = 5) PBMC were also stimulated for 24 h and samples taken at 1, 6, 16 and 24 h for analysis of CD25, CD69, CD137 and CD154 expression amongst CD3+ T cells (E). Bars represent net expression in the CD3+ population after subtraction of the negative control (unstimulated).

Next, the question of whether CMV-specific T cells could be isolated from G-CSF mobilized PBMC based on IFN-γ secretion was investigated, as this system has been used previously for the manufacture of CMV-specific T cells from non-mobilized PBMC and their clinical efficacy has been successfully demonstrated (Peggs et al, 2011). Cells secreting IFN-γ in response to CMVpp65 stimulation were captured using IFN-γ specific antibodies and selected using magnetic beads. IFN-γ was measured before and after magnetic enrichment to assess purity and yield between mobilized and non-mobilized PBMC. IFN-γ secretion was reduced in G-CSF mobilized PBMC (Fig. 1C) and furthermore, both purity and yield (Fig. 1D) were also negatively affected. Analysis of the MFI of IFN-γ amongst CD3+ T cells after CMVpp65 stimulation showed no significant difference between G-CSF mobilized (mean, 39·26 ± 4·78) and non-mobilized (mean, 46·62 ± 12·06) PBMC, suggesting that the intensity of IFN-γ secretion in CMVpp65 responding cells is not unduly affected by G-CSF mobilization. The ratio of CD4+ to CD8+ IFN-γ secreting cells appeared to be unchanged in G-CSF mobilized PBMC. In summary, PBMC from G-CSF mobilized PBMC are capable of secreting IFN-γ and other effector cytokines at a level similar to non-mobilized PBMC, but isolation and detection after CMVpp65 stimulation on a per cell basis appeared to be impaired. These results are in line with previously published data suggesting that G-CSF mobilization impairs the potential for IFN-γ production at a single cell level (Tayebi et al, 2001), which could result in reduced detection of CMV-T using this method and ultimately in the efficiency of magnetic bead enrichment.

Assessment of activation marker expression after CMVpp65 stimulation in G-CSF mobilized PBMC

The kinetics of activation-induced CD25, CD69, CD154 and CD137 expression on CMVpp65-specific T cells in G-CSF mobilized PBMC was investigated to determine both the optimal target and the optimal time for maximal expression compared to non-mobilized PBMC. Prior to CMVpp65 stimulation, baseline expression of activation markers amongst CD3+ T cells was compared between G-CSF mobilized and non-mobilized PBMC with no significant differences observed. PBMC were stimulated over a 24-h period with CMVpp65 peptides; PBMC were removed from cultures at 1, 6, 16 and 24 h, and analysed for surface expression of activation markers by flow cytometry (Fig. 1E). Antigen-dependent expression of CD25 was maximal at 16 h in G-CSF mobilized PBMC (mean, 3·19% ± 1·43) and was elevated when compared to non-mobilized PBMC (mean, 1·73% ± 0·63), with no significant difference in MFI observed. The peak intensity of CD69 expression was stabilized at 6–24 h with an elevation in expression observed in G-CSF mobilized PBMC compared to non-mobilized PBMC but with equivalence in MFI. CD69 expression in G-CSF mobilized PBMC peaked at 16 h (mean, 1·32% ± 0·51), compared to 6 h in non-mobilized PBMC (mean, 0·33% ± 0·21). CD137 reached peak expression at 24 h after CMVpp65 stimulation and was comparable in intensity between G-CSF mobilized (mean, 0·88% ± 0·45) and non-mobilized (mean, 0·56% ± 0·16) PBMC. CD154 expression in G-CSF mobilized PBMC was maximal at 6 h (mean, 0·95% ± 0·44) and elevated when compared to non-mobilized PBMC (mean, 0·16% ± 0·15). Both CD137 and CD154, at optimal times of expression after CMVpp65 stimulation, also showed similar MFI when comparing G-CSF mobilized and non-mobilized PBMC.

Previously published data have demonstrated that CD154 is a suitable marker for the detection and isolation of CMV-specific T cells (Chattopadhyay et al, 2005; Frentsch et al, 2005; Khanna et al, 2011) from PBMC in the presence of CD40 blockade. CD154 on activated CD4+ T cells binds CD40 on B cells, providing co-stimulator signals for B cell activation (Brines & Klaus, 1993). Prevention of CD40-CD154 ligation enables the preservation of CD154 expression at the cell surface (Frentsch et al, 2005). The effect of G-CSF mobilization on this was unknown. PBMC were stimulated with either SEB or CMVpp65 peptides for 4–6 h in the presence or absence of CD40-specific antibody, and then analysed for CD154 expression amongst the CD4+ T cell population (Fig. 2A). Low background CD154 expression in resting CD4+ T cells was comparable between G-CSF mobilized PBMC (mean, 0·30% ± 0·14) and non-mobilized PBMC (mean, 0·22% ± 0·06). Mean CD154 expression in the presence of CD40-specific antibody at the optimal time point of 6 h, showed no significant difference between G-CSF mobilized PBMC (1·86% ± 0·59) and non-mobilized PBMC (1·22% ± 0·13) but was in fact elevated in the G-CSF mobilized donor setting (Fig. 2B, C), without any non-specific activation-induced CD154 expression.

Figure 2.

Direct comparison of CD154 surface expression at 4 and 6 h between a G-CSF mobilized and non-mobilized donor after CD40 blocking. (A) PBMC were stimulated with either CMVpp65 peptides or SEB in the presence or absence of CD40-specific antibody (1 μg/ml) and with CD40-specific antibody alone. Cells are gated on CD3+ CD4+ T cells. (B, C) Comparison of CD154 expression in non-mobilized (n = 5) and G-CSF mobilized (n = 5) donors at 4 and 6 h. *< 0·05, **< 0·01, ***< 0·001, paired t-test.

Enrichment of CMV-T from G-CSF Mobilized and Non-Mobilized PBMC

On the basis of expression kinetics after CMVpp65 stimulation, both CD25 and CD154 activation markers were investigated for isolating CMV-T from G-CSF mobilized PBMC. A single enrichment step of CMV-T using CD25 or CD154 antibodies from G-CSF mobilized and non-mobilized PBMC was performed using magnetic-bead based cell separation in five healthy CMV+ donors (Table 1). No significant difference in the purity of CD154+ CMV-T between G-CSF mobilized (mean, 48·9% ± 12·8) and non-mobilized (mean, 64·7% ± 14·28) PBMC was observed and this was also the case with CD25 enrichment (G-CSF mobilized: mean, 89·4% ± 2·15; non-mobilized: mean, 80·8% ± 9·80). Figure 3A, B illustrates the enrichment of CMV-T from a G-CSF mobilized and non-mobilized donor through both CD154 and CD25 activation-dependent expression. Purity in the enriched fractions from G-CSF mobilized donors was, however, significantly increased when using CD25 selection of CMV-T compared to CD154 selection (< 0·05), as shown in Table 1. The majority of CD25+ and CD154+ fractions from both G-CSF mobilized and non-mobilized PBMC were CD4+ with a small population of CD8+ T cells. Analysis of the efficiency of isolation also demonstrated no significant difference in the yield of CD154+ CMV-T between G-CSF mobilized (mean, 30·8% ± 12·54) and non-mobilized (mean, 28·4% ± 7·24) PBMC. The yield of CMV-T after CD25+ enrichment in G-CSF mobilized PBMC (mean, 26·12% ± 8·83) was comparable to that of CD154, although we observed a reduction in the non-mobilized setting (mean, 11·3% ± 6·88).

Figure 3.

Isolation of CMV-specific T cells through CD154 and CD25 activation-induced expression. (A, B) PBMC isolated from non-mobilized and G-CSF mobilized leukapheresis were stimulated with CMVpp65 peptides for either 6 h for CD154 isolation or 16 h for CD25 isolation and antigen expression was measured amongst CD3+ lymphocytes before stimulation after stimulation and after magnetic enrichment. (C) Evaluation of T regulatory phenotype was assessed by FoxP3 expression in CMV-specific T cells identified and isolated through CD25 expression in G-CSF mobilized (n = 5) and non-mobilized (n = 5) donors. Fra, fraction. (D) Dose-dependent suppression of T cell proliferation was assessed by CD25-positive and CD25-negative fractions after sorting. CFSE-labelled PBMC were cultured at two ratios in the presence of SEB for 5 d. Unlabelled, PBMC alone and PBMC with SEB were cultured as experimental controls.

Table 1. Comparison of CMV-specific T cell enrichment through CD25 and CD154 activation-induced expression in G-CSF mobilized and non-mobilized PBMC
 Non-mobilizedG-CSF mobilized
  1. The mean and range of yield and purity of five CD154+ and five CD25+ enrichments in both G-CSF mobilized and non-mobilized PBMC after stimulation with CMVpp65 peptides. The mean purity is further dissected to show CD4+ and CD8+ frequencies within the CD3+ populations of positively enriched fractions.

CD25 Yield11·3% (0·1–34·75)26·12% (3·3–52·7)
% CD25+ CD3+ (Purity)82·5 (43·0–95·9)89·9 (85·1–96·5)
% CD25+ CD4+ (% of CD3+)75·9 (91·8)87·7 (97·5)
% CD25+ CD8+ (% of CD3+)6·6 (8·2)2·2 (2·5)
CD154 Yield28·4% (5·9–44·5)30·8% (4·0–48·2)
% CD25+ CD3+ (Purity)64·7 (10·6–91·2)48·9 (9·6–78·7)
% CD25+ CD4+ (% of CD3+)59·9 (92·6)45·3 (92·6)
% CD25+ CD8+ (% of CD3+)4·8 (7·4)3·6 (7·4)

To determine whether CD25 was the optimal target for enrichment of CMV-T from G-CSF mobilized PBMC we explored the frequency of Treg enrichment by staining for the transcription factor FoxP3. The frequency of CD25+ FoxP3+ cells within the CD4+ compartment was equivalent between G-CSF mobilized (mean, 3·67% ± 0·62) and non-mobilized (mean, 3·16% ± 1·09) PBMC after CMVpp65 stimulation. However, upon CD25+ enrichment we observed a significant increase (≤ 0·01) in the frequency of FoxP3+ Tregs in G-CSF mobilized (mean, 64·24% ± 4·88) compared to non-mobilized (mean, 33·84% ± 7·17), as illustrated in Fig. 3C. We next looked at the ability of CD25+ CMV-T enriched from G-CSF mobilized PBMC to suppress T cell proliferation in a CFSE-based suppression assay. We directly compared the suppressive capacity of CD25-negative fractions with CD25-positive fraction, after CMVpp65 stimulation and magnetic bead cell sorting, on autologous CFSE-labelled PBMC in a 5-d culture. Figure 3D shows a typical result from a G-CSF mobilized donor and demonstrates the highly suppressive function of CD25+ CMV-T with >90% suppression of CD3+ T proliferation compared to c. 25% observed with the CD25-negative fraction at responder to suppressor ratios of 1:1 and 1:2.

Having shown the suppressive capacity of CD25+ enriched from G-CSF mobilized PBMC, we investigated CD154 as a primary target for generating CMV-T. CD154-positive fractions were subsequently expanded in short-term culture to determine in vitro proliferation and CMV specificity of isolated cells in G-CSF mobilized PBMC and compared directly against non-mobilized PBMC. We also looked at the co-expression of CD25 and CD69 in CD154+ CMV-T pre- and post-enrichment to rule out the possible selection of Tregs. As shown in the representative FACS plots from a G-CSF mobilized donor (Fig. 4), co-expression of CD25 was minimal (c. 3% of CD154+) at 6 h whereas the majority of CD154+ cells co-expressed CD69 (>99%).

Figure 4.

Co-expression of activation markers in CD154 CMV-T in G-CSF mobilized PBMC. After 6 h of CMVpp65 peptide stimulation and subsequent CD154 isolation by magnetic bead enrichment, CD3+ T cells were analysed for CD154, CD25 and CD69 co-expression for phenotypic assessment.

Re-stimulation of in-vitro expanded antigen-specific from G-CSF mobilized PBMC

CMV-specific T cells isolated on the basis of CD154 expression were cultured for up to 23 d in complete medium containing IL-7 and IL-15 in the presence of autologous irradiated feeder cells. CD154+ responder populations from G-CSF mobilized PBMC (n = 6) did not differ in their proliferative capacity (90·4-fold expansion; range 2–186) compared to cells from non-mobilized PBMC (n = 5) (104·4-fold expansion; range 18–196) (Fig. 5A, B). Expanded cells were >90% CD3+ and were predominantly CD4+ (mean, 98%) in all cultures, although a small fraction of expanded cells expressed CD8 (mean, 2%). All cultures showed high specificity for CMVpp65, determined by up regulation of CD154+ and CD69+ expression upon re-challenge with autologous CMVpp65-loaded PBMC in contrast to re-challenge with autologous PBMC alone, where low to undetectable levels of CD154 expression were observed (Fig. 5C). The mean CD154+ CD69+ expression upon re-challenge in cells expanded from G-CSF mobilized PBMC was 83·3% (±5·04) compared to 60·9% (±15·53) in non-mobilized (Fig. 5D). In selected experiments, expanded cells were re-challenged with CMV IE-1 peptides and no CD154 activation was observed, confirming specificity to the CMVpp65 protein against which they were selected.

Figure 5.

Expansion of CD154+ CMV-T in short-term culture and qualitative and quantitative assessment after antigen re-challenge. 0·25 × 106 CD154+ T cells were expanded in culture for up to 21 d in the presence of IL-7, IL-15 and 12·5 × 106 irradiated autologous feeder cells in G-CSF mobilized (A) and non-mobilized (B) PBMC. Expanded CD154+ cells were subsequently co-cultured with autologous PBMC loaded with CMVpp65 peptides or PBMC alone for 6 h in the presence of CD40-specific antibody. Representative FACS plots show expression of CD154 vs. CD69 among CD3+ T cells after re-challenge in one G-CSF mobilized and one non-mobilized donor (C). CD154+ expanded cells from G-CSF mobilized (n = 6) and non-mobilized (n = 5) PBMC were analysed for CD154 and CD69 after re-challenge are summarized (D). Bars represent net expression after subtraction of the negative control (nil peptide). Expanded CD154+ CMV-T were also stimulated as previously described with CMVpp65 peptides in the presence of Brefeldin A and CD28-specific antibody. Histograms illustrate cells analysed for IL-2, TNF-α, IFN-γ and Granzyme B among CD3+ T cells in a representative G-CSF mobilized and non-mobilized donor (E) with solid lines representing matched isotype controls. (F) Combined assessment of IL-2, TNF-α, IFN-γ and Granzyme B after re-challenge in G-CSF mobilized (n = 6) and non-mobilized (n = 5) PBMC are summarized. Data are presented with standard error of the mean.

Expanded cells synthesized and secreted IL-2, TNF-α, IFN-γ and no significant difference was observed between cultured cells from G-CSF mobilized and non-mobilized donors (Fig. 5E, F) Expanded cells were also highly positive for Granzyme B both prior to re-challenge and after re-challenge with CMVpp65 peptides, indicating that cells possessed the effector molecules necessary for cytotoxic activity. In experiments where expanded cells were unstimulated or incubated with CMV IE-1 peptides, minimal cytokine secretion was observed.

Cytotoxic activity of expanded cells

Cytotoxic activity was evaluated using autologous PHA blasts loaded with CMVpp65 peptides and labelled with Calcein-AM dye as targets. CD154+ expanded cells from G-CSF mobilized PBMC effectively lysed CMVpp65 loaded targets at all E:T ratios in a dose-dependent manner (Fig. 6A) and this was consistent with lysis by CD154+ CMV-T expanded from non-mobilized PBMC (Fig. 6B).

Figure 6.

CD154+ CMV-specific T cells isolated and expanded from G-CSF mobilized and non-mobilized PBMC effectively kill target cells. Specific lysis of autologous PHA blasts loaded with CMVpp65 peptides at E:T ratios from 20:1 to 0·5:1 determined using fluorescent dye Calcein-AM cytotoxicity assay in CD154 CMV-T expanded from (A) G-CSF mobilized and (B) non-mobilized PBMC. Graphs show the percentage of calcein released from target cells after normalization in the supernatants of cultures measured by fluorescence.


Generation of CMV-T from freshly isolated, non-GCSF mobilized blood or apheresates has been demonstrated using a number of methods, resulting in products containing CD4+ T cells, CD8+ T cells and a mixture of both for adoptive transfer (Peggs, 2009). More recently, a method has been described for the manufacture of multipathogen-specific T cells based on the activation-dependent expression of CD154 (Khanna et al, 2011). In this study we describe the detection, isolation and expansion of CMV-T through CD154 expression in G-CSF mobilized PBMC using fresh and, in some experiments, cryopreserved PBMC. Our data demonstrate that following a short-term stimulation period using commercially available antigens and magnetic enrichment, CMV-T can be expanded over 21 d without the need for repeated CMVpp65 stimulation. Expanded cells showed functional specificity against CMVpp65, illustrated by high levels of IFN-γ, TNF-α and IL-2 secretion, were highly specific for the lytic granule Granzyme B, associated with cytotoxic effector function and were capable of specifically lysing CMV targets in a cytotoxicity assay. This system has the potential to abrogate the need for a second leukapheresis from the original HSCT donor for manufacture of CMV-T, which offers huge advantages in the unrelated donor setting. It also supports a concept of planned cryopreservation of excess PBMC from mobilized apheresates for subsequent generation of anti-CMV T cells if required.

G-CSF mobilization was not associated with a significant decrease in the secretion of inflammatory or anti-viral cytokines in response to CMVpp65 stimulation or an increase in the T-helper cell type 2-associated cytokines, IL-4 and IL-5. The finding that IL-10 secretion was decreased in G-CSF mobilized PBMC, both in CMV-stimulated and unstimulated cultures is an interesting one, as it has often been proposed that G-CSF augments the generation of IL-10 producing Tregs (Morris et al, 2004).

However, upon identification and isolation of CMV-T using the IFN-γ catch method, the total number of IFN-γ secreting CD3+ T cells decreased, which was reflected in a decrease in both purity and yield. Although we did not observe any significant difference in the intensity of IFN-γ staining after CMVpp65 stimulation, the reduction in purity and yield after IFN-γ enrichment is most likely to be a consequence of a reduction in the frequency of IFN-γ secreting cells. It remains unclear whether the use of individual CMVpp65 peptides based upon donor human leucocyte antigen type or the use of CMV IE-1 peptides would result in a more efficient system for the identification and isolation of CMV-T (Slezak et al, 2007; Gratama et al, 2008; Zandvliet et al, 2010), as the kinetics for IFN-γ secretion in G-CSF mobilized PBMC may differ between epitopes at different peptide concentrations.

Analysis of T cell activation markers after CMV peptide stimulation in G-CSF mobilized PBMC suggested that targeting CMV-T through up-regulation of surface markers was a potential route to isolation. We observed no significant differences in the level of expression of CD25, CD69, CD137 and CD154 in G-CSF mobilized PBMC when comparing directly to unpaired non-mobilized PBMC although isolation of CD25+ cells from G-CSF mobilized PBMC produced a product enriched with Treg cells, which were capable of strongly suppressing T cell proliferation. The proportion of cells with a CD4+CD25+FoxP3+ phenotype isolated after CMVpp65 stimulation from non-mobilized PBMC was consistent with previous studies (Lugthart et al, 2012). Although it has been suggested that the co-infusion of Tregs with anti-viral T cells can be advantageous in terms of graft-versus-host disease prevention, the same cannot be said for CMV-T enriched through CD25 expression where the Treg population is significantly higher, as seen in the G-CSF mobilized setting (mean 64·24% ± 4·88).

The short period of stimulation needed for CD154-activated expression together with the low to undetectable levels of background CD154 expression made this antigen an optimal target. Furthermore, we also demonstrate that CD154 expressing cells do not co-express CD25, ruling out the potential for enrichment of Tregs. In line with published results (Chattopadhyay et al, 2005; Frentsch et al, 2005), we have shown that after CMVpp65 peptide stimulation, CD154 expression peaked at 6 h and was preserved at the cell surface through the use of a CD40-specific antibody, which successfully prevented ligation of CD154. Furthermore, our data in G-CSF mobilized PBMC supported this finding, allowing the effective identification of CMV-T. We also observed a trend towards greater expression of CD154 in G-CSF mobilized PBMC (mean, 1·86% ± 0·59) compared to non-mobilized PBMC (mean, 1·22% ± 0·13). Upon isolation we have demonstrated equivalence in CD154 purity between G-CSF mobilized and non-mobilized PBMC and have been able to effectively expand these cells in short-term culture. Expanded cells were highly specific for CMVpp65 determined by CD154 staining and co-expressed CD69. We have also demonstrated, through the use of intracellular cytokine staining, that expanded cells secreted high levels of the inflammatory cytokine IFN-γ as well as IL-2 and TNF-α to a lesser extent, illustrating functionality. Expanded cells contained high levels of Granzyme B, indicating a possible mechanism for cytotoxicity through the process of directed exocytosis resulting in apoptosis. The cytokine profile of expanded cells together with the identification of the lytic granule Granzyme B are in keeping with a population of anti-viral CD4+ cytotoxic T cells characterized in previous studies (Appay et al, 2002) and reminiscent of antiviral CD8+ T cells.

Furthermore we have demonstrated using a Calcein-AM cytotoxicity assay (Neri et al, 2001) that CD154+ CMV-T expanded from G-CSF mobilized PBMC can successfully lyse CMVpp65 peptide-pulsed target cells, illustrating their cytotoxicity. The memory phenotype of CD154+ CMV-T expanded from G-CSF mobilized PBMC were predominantly effector memory and, to a lesser extent, central memory determined by staining for CD45RA and CCR7 by flow cytometry (data not shown). The effector memory phenotype is consistent with previous findings that CCR7 is down regulated during prolonged culture (Zandvliet et al, 2009).

The effective isolation and expansion of CMV-T from G-CSF mobilized PBMC offers the potential for a two-step approach to manufacturing an adoptive immunotherapy in this setting, particularly given the added flexibility of using cryopreserved PBMC as starting material. Current clinical protocols for transfer of CMV-T have used doses starting at 1 × 104 CD3+ cells/kg rising to 3 × 104 CD3+ cells/kg and have shown successful reconstitution of CMV immunity (Peggs et al, 2011). Generating small doses for adoptive transfer lends itself to using an aliquot of the HSCT graft; where dependent upon the quality of the graft only a small aliquot might be attainable. Equally, the demonstration in this study that PBMC from G-CSF mobilized samples can also be expanded following selection, from small numbers of CD154+ CMV-T, with functional and cytotoxic capabilities illustrate that short-term culture is also an option given the size of the aliquot available. One key issue and a potential limitation of the system used in this study can be viewed as the generation of a predominantly CD4+ T cell population in comparison to systems used in other studies that generate either a mixture of CD4+ and CD8+ (Peggs et al, 2011) or CD8+ alone (Cobbold et al, 2005). Whether the adoptive transfer of a predominantly CD4+ CMV-T product from G-CSF mobilized PBMC can control CMV disease cannot be answered without patient studies, but previous clinical trials using both CD4+ T cell clones (Perruccio et al, 2005) and CD4+ T cell lines (Einsele et al, 2002) established from non-mobilized PBMC have demonstrated effective control of CMV disease.

Our results illustrate the similarities in identification and isolation of CMV-T from both G-CSF mobilized and non-mobilized PBMC when targeting CD154 activation induced expression as a method for manufacture. We also highlight the limits of using G-CSF mobilized PBMC for CMV-T generation when the IFN-γ secretion assay is utilized. However it must be noted that a limitation of this study was the failure to procure paired donor samples before and after G-CSF mobilization for subsequent comparison of the CMV response. The use of paired samples could have allowed for stronger statistical analysis to be made when analysing the effect of G-CSF mobilization given the knowledge of high donor variability that exists, in terms of the CMV immune response.

In summary, we have demonstrated a working protocol for the identification, isolation and expansion of functionally active CMV-T using CD154 expression, from G-CSF mobilized PBMC and that they have the same characteristics as CMV-T isolated in the same manner from non-mobilized PBMC. This is equivalent to published data in non-mobilized PBMC (Frentsch et al, 2005; Khanna et al, 2011) but is the first time this has been demonstrated in G-CSF mobilized PBMC. These results represent a feasible approach to manufacturing anti-viral immunotherapies directed against a range of pathogens that does not require successive donations, alleviating the many problems incurred with procuring cells from unrelated donors. Questions remain as to whether these cells are capable of conferring protection from CMV in vivo in recipients of HSCT, raising the possibility of future clinical trials in the area. Nevertheless, to our knowledge, this is an entirely novel finding and the first report to show the feasibility of identifying, isolating and expanding CMV-T from G-CSF mobilized donors and that these cells are functional when re-challenged.


E.S. and this work were supported by a UK Government Technology Strategy Board translational award. K.N. is an employee of Cell Medica and an honorary fellow at UCL. Gavin Holmes (Cell Medica) and staff in the Paul O'Gorman Laboratory of Cellular Therapeutics (The Royal Free London NHS Foundation Trust) provided aliquots of donor blood for this work. Simon Thomas (Cell Medica) helped with the methodology of the cytotoxicity assay.

Author contributions

E.S. performed experiments and analysed data; M.L. and K.N. helped define experimental strategy and analysed data; E.S. wrote the manuscript with M.L.; K.N and S.M. contributed to the drafting of the manuscript.

Conflict of interest disclosures

S.M., M.L. and K.N. are shareholders in Cell Medica, a clinical-stage cellular therapeutics company. E.S. has no conflicts.