Ex vivo T lymphocyte expansion for retroviral transduction

Experimental Hematology 28 (2000) 1137–1146

Ex vivo T lymphocyte expansion for retroviral transduction: Influence of serum-free media on variations in cell expansion rates and lymphocyte subset distribution Stefan Carlensa,d, Mari Gilljamb, Mats Remberger , Johan Aschana,b, Birger Christenssonc,d, and M. Sirac Dilberb d

a Centre for Allogeneic Stem Cell Transplantation, and Departments of Hematology, cPathology, and dClinical Immunology, Huddinge University Hospital, Huddinge, Sweden

b

(Received 21 October 1999; revised 23 May 2000; accepted 21 June 2000)

Objective. In the setting of allogeneic stem cell transplantation, suicide gene-manipulated donor T cells that can be selectively inactivated in vivo would potentially allow optimal control of the GVL (graft-vs-leukemia)/GVHD (graft-vs-host disease) balance. Retroviral T-cell transduction requires ex vivo cell expansion, which is often achieved by IL-2 and anti-CD3 stimulation. Traditionally, culture media for cell expansion are supplemented with fetal bovine serum (FBS) or human serum. While these sera promote cell growth and viability, they contain uncharacterized elements that may yield inconsistent results from batch to batch. Cell expansion in serum-free media would therefore be preferable. Materials and Methods. We compared T-cell expansion rates in three commercially available serum-free culture media (X-VIVO 15, AIM-V, and Cellgro SCGM), with or without the addition of human serum (HS, 5%). We also aimed to evaluate how the in vitro expansion affected the composition of the various T-cell subsets. Buffy-coats from four healthy donors were expanded for 21 days. The media were compared to standard RPMI 1640 medium, supplemented with HS (5%) or FBS (10%). For retroviral transductions, the LN vector carrying the neomycin- resistance gene was used in four additional donors. Results. In our hands, X-VIVO 15 gave the highest rate of serum-free expansion (a median of 79-fold expansion, range 20–117). For serum-free expansion, activation with OKT3 for 21 days gave slightly higher expansion rates than a 5-day course (however, without statistical significance). When serum was added, this discrepancy was not seen. Cytokine analysis (IFN-␥, IL10, and IL-4) showed a distinct type1 cytokine pattern with elevated IFN-␥ levels during the whole period of culture. Flow cytometric analyses showed substantial inter-media, but also some inter-donor, variability in T-cell subset compositions. Transduction of cells with the LN vector and G418 selection resulted in a 14-fold increase (range 3–18) for serum-free X-VIVO 15 based cultures. Cell phenotypes remained unchanged by the transduction procedure as compared to nontransduced cells. Conclusion. Among the tested serum-free media, X-VIVO 15 has shown to best support the in vitro expansion of T cells, resulting in equal percentages of CD4⫹ and CD8⫹ cells. These cells can easily be transduced and selected. There seem to be no significant benefits, regarding absolute cell numbers or T-cell subset compositions, with OKT3-stimulation for more than five days. The addition of low levels of HS increases the consistencies in the cell expansion rates for all media. © 2000 International Society for Experimental Hematology. Published by Elsevier Science Inc. Keywords: T lymphocyte—Serum-free medium—Retroviral transduction

Offprint requests to: Stefan Carlens, M.D., Dept. of Clinical Immunology, F79, Huddinge University Hospital, S-141 86 Huddinge, Sweden; E-mail: [email protected]

0301-472X/00 $–see front matter. Copyright © 2000 International Society for Experimental Hematology. Published by Elsevier Science Inc. PII S0301-472X(00)0 0 5 2 6 - 9

1138

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

Introduction Recurrent leukemia remains one of the major obstacles following allogeneic stem cell transplantation [1,2]. T lymphocytes of donor origin play an important role by promoting engraftment and by mediating the so-called graft-vs-leukemia effect (GVL) [3–5]. However, donor T cells, activated by host histocompatibility antigens, are also the main effector cells of graft-vs-host disease (GVHD) [6,7]. Alloreactive T cells are normally controlled by posttransplant immunosuppressive drugs. T-cell depletion of the graft is an alternative approach but has drawbacks in the form of increased graft rejection and higher rates of leukemia relapse [8,9]. When donor lymphocytes are administered to patients, as treatment for a leukemic relapse (DLI, donor lymphocyte infusion), the chance of both inducing a remission and eliciting GVHD is correlated to the number of given CD3⫹ T cells [10]. It is further believed that the T-cell threshold level for inducing a GVL effect is lower than for triggering GVHD. For these reasons, escalating dose regimens are increasingly used, rather than giving single large cell doses [11]. Achievement of the maximal GVL effect while preventing the development of severe GVHD is a main goal. By introducing a so-called suicide gene into the donor T cells, a specific prodrug, when administered to the patient, can be metabolized by the gene-modified cells to become toxic and eliminate the same cells selectively [12]. Theoretically, several suicide genes could be used for this purpose. Most clinical trials so far have used the gene encoding the herpes simplex thymidine kinase (HSVtk) enzyme. Studies have also shown that these T cells retain their anti-tumor effects following gene modification [13–15]. Ex vivo transduction of target cells with a retroviral vector, containing HSVtk and a dominant selectable marker such as the neomycin resistance gene, requires activated mitotic T cells. For clinical purposes, T-cell activation may be achieved by anti-CD3 stimulation (such as OKT3) in the presence of recombinant interleukin-2 (IL-2) [16,17]. However, other mitogens, like phytohemagglutinin (PHA), have been used [13,18]. The selective survival of gene-transduced cells is mediated by resistance to the toxic neomycin analogue, G418. To achieve sufficient cell expansion rates for transduction, supplementation of culture media with fetal bovine serum (FBS) or human serum (HS) is usually required. However, all sera contain poorly characterized substances, including growth factors, antibodies, and other potentially immunologically active substances [19,20]. The composition can vary slightly from batch to batch, and this may give inconsistent results. For these reasons, T-cell expansion and transduction under serum-free conditions is preferable. The T-cell subset composition shifts during IL-2/ OKT3-mediated expansion, usually with a predominant expansion of CD8⫹ cells [18,21], and often with a large interdonor variability [15]. While IL-2 mainly stimulates CD8⫹ cells and NK cells, OKT3 helps to enhance the expansion of

CD4⫹ cells [22]. The transduction procedure in itself does not seem to affect the T-cell subset composition [15,23]. Many studies have addressed the problem of separating the GVL effect from GVHD. However, the exact composition of cell subsets mediating these effects has yet to be determined. In CML patients, allo-transplants and donor lymphocyte infusions depleted of CD8⫹ cells have been shown to reduce the incidence of GVHD, while preserving the graft-vs-leukemia (GVL) effect [24,25]. Various murine models have shown a reduction in the incidence of GVHD through CD4 and/or CD8 depletion, in some cases with a preserved GVL effect [26,27]. In the clinical setting, CD4 depletion alone did not prevent severe GVHD [28]. The rapid onset of a severe acute GVHD, initially described as a cytokine storm [29], has been characterized as a type1 cytokine response [30], comprising increasing levels of predominantly IL-2 and IFN-␥. The type2 cytokine response, comprising release of IL-4 and IL-10, is thought to be more preventive in the development of acute GVHD. In preparation for clinical trials involving the use of suicide gene transduced donor T cells, we aimed to optimize the conditions for in vitro T-cell expansion. We compared the rates of cell expansion in three defined serum-free media (Cellgro, AIM-V, and X-VIVO 15), with or without the addition of human serum (HS, 5%), to that obtained with the standard RPMI 1640 medium, supplemented with either HS (5%) or FBS (10%). We also investigated whether a prolonged course of OKT3 stimulation could improve the T-cell expansion rates significantly, as compared to the standard period of five days, and how the phenotypic properties are affected by these various culture conditions. Since different patterns of cytokine release could have different impact on the development of GVHD, when cells are administered to the patient, we compared cytokine levels in the culturing media. Transduction with the retroviral LN vector, carrying the neomycin resistance gene, was done to prove that the cells in our cultures could be efficiently transduced and selected.

Materials and methods Cell culture media and reagents The serum-free culture media purchased and used for this study comprised AIM-V (Gibco, Grand Island, NY), Cellgro SCGM (Cellgro, Boehringer Ingelheim, Heidelberg, Germany), and X-VIVO 15 (Biowhittaker, Walksville, MD). These media were used with or without the addition of 5% HS (Sigma, St. Louis, MO). Cell culture in standard RPMI 1640 (Gibco, Paisley, UK), supplemented with 10% FBS (Gibco) or 5% HS (Sigma), was done for comparison. All cells were cultured without the addition of antibiotics. The PG13 producer cell line was cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco). G418 used for selection of the transduced cells was produced by Gibco. IL-2, with a minimum concentration of 1 ⫻ 107 IU/mg, was purchased from Peprotech (London, UK). We used recombinant,

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

murine, anti-human CD3 antibodies (Orthoclone OKT3) as manufactured by Ortho Biotech Inc. (Raritan, NJ). Cell cultures and quantification Buffy-coat cells were obtained from healthy blood-bank donors (n ⫽ 4) on the day before starting the study. On day 0, peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation, using Ficoll (Nycomed Pharma AS, Oslo, Norway). They were washed in phosphate-buffered saline (PBS, Gibco), their viability being assessed by a trypan blue dye exclusion assay, and they were plated onto six-well dishes (Falcon by Becton-Dickinson, Meytan Cedex, France) with 200,000 cells/well. IL-2 was used throughout the whole culturing period in a final concentration of 500 IU/mL, a concentration commonly used by others for T-cell expansion [15,23]. OKT3 was supplemented to a final concentration of 10 ng/mL. For each serum-free medium, cell culturing was performed in four different serum/OKT3 combinations, to define optimal conditions for T-cell activation: 1) serum-free ⫹ 5 days of OKT3; 2) serum-free ⫹ 21 days of OKT3; 3) serum supplement (5% HS) ⫹ 5 days of OKT3; 4) serum supplement (5% HS) ⫹ 21 days of OKT3. For every donor triplicate wells were cultured for each one of these combinations. For cell culturing in RPMI, two additional combinations were also evaluated comprising: 1) serum supplement (10% FBS) ⫹ 5 days of OKT3 and 2) serum supplement (10% FBS) ⫹ 21 days of OKT3. Absolute cell numbers were assessed by the cell Coulter technique (Coulter Multisizer II, Coulter Electronics Ltd., Luton, UK), on days 3, 5–6, 10–11, 15– 16, and 21. Viability was analyzed with the trypan blue dye exclusion assay at each time point, except on day 3. The medium was exchanged on day 5–6 and on day 13–14. In addition to this, fresh media and IL-2 were added regularly throughout the culturing period. To prevent contact inhibition of cell growth, the cells were transferred into T25 flasks (TPP, Trasadingen, Switzerland) on day 10–11. Cytokine assays Supernatant was collected from each well after 5 hours of culturing and on days 5–6, 10, 15–16, and 21 for cytokine analyses. We performed analyses for gamma-interferon (IFN-␥), IL-10, and IL-4 production by using a quantitative sandwich enzyme immunoassay technique (Quantikine kit, R&D Systems Europe, Oxon, UK). The lowest detectable levels were 2.0 pg/mL for IFN-␥ and 3.0 pg/mL for IL-4 and IL-10. Samples were diluted 1:10 before analyzing IFN-␥ and IL-10 because of the low sample volumes and expected high levels. Analyses of lymphocyte subsets and activation molecules A subset of the cultures was prepared for flow cytometric phenotypic analysis. Three-color fluorescent analysis was performed according to standard procedures. In short, 105 cells/tube were mixed with appropriate concentrations of directly fluorochrome-conjugated monoclonal antibodies to CD45/14 and antigens on T cells (CD2, CD3, CD4, CD5, CD8), B cells (CD19), and NK cells (CD56, CD16) and molecules associated with cell activation (IL-2 receptor; CD25), T cell co-receptor (CD28), CD69, and HLA-DR. Antibodies to detect CD45RA⫹ and CD45RO⫹ T cells were used to define “naive” and “memory” T-cell subsets, respectively. All antibodies were obtained from Becton-Dickinson (BD, Mountain View, CA). After the addition of the primary antibody and incubation for 15 minutes at room temperature, cells were washed in phosphate-buffered saline, pending analysis.

1139

Propidium iodide (PI) staining was used for viability analysis. Data acquisition was made on a FACScan (BD), using Cellquest software (BD), both for acquisition and analysis. In each sample, 3000 cells were acquired in the analysis region of viable cells, using log-amplified fluorescence and linearly amplified side and forward scatter signals. All samples were analyzed by setting appropriate SSC/FSC gates around the lymphocyte population, using back-gating on CD45⫹/ CD14⫺, PI⫺ cells. Consistency of analysis parameters was ascertained by calibrating the flow cytometer with Calibrite beads and the AutoComp software, both from BD. Retrovirus-containing supernatant The retroviral vector LN, carrying the neomycin phosphotransferase II gene, and the PG13 GALV producer cell line have been described previously [31,32]. PG13-LN (ATCC CRL-10685) was first grown in DMEM supplemented with 1% L-glutamine and 10% (vol./vol.) heat inactivated FBS. When cells were subconfluent, they were washed three times with PBS and then a serum-free medium (Cellgro) was added for 24 hours. Finally the supernatant was collected, filtered through a 0.45 ␮m filter, and frozen at ⫺70⬚C. The supernatant had a titer of 4 ⫻ 105 G418-resistant colony-forming units per mL measured on HeLa cells. Supernatant containing viral particles were negative for helper virus when target HeLa cells were tested by PCR for GALV envelope sequences. Transduction and selection procedure To prove that the cells, cultured under the conditions previously described, could be targeted by a retroviral vector, PBMCs from four donors were transduced with the LN vector. The transduction and selection studies were performed only under the culturing conditions that initially gave a consistent and sufficient T-cell expansion. All transductions were carried out by replacing the media with supernatant at a multiplicity of infection (MOI) of 3, in the presence of 4 ␮g/ mL polybrene (Sigma) and 500 IU/mL IL-2 via centrifugation at 1000 ⫻ g at room temperature for two hours on day 4 and 5. After centrifugation, the supernatant was replaced with the T-cell medium (serum-free medium with IL-2). Forty-eight hours after the second transduction, the cells were resuspended in fresh serum-free media containing 1 mg/mL of G418 and 500 IU/mL IL-2 for 7 days (Day 7–14). The cells were then expanded without G418 in order to get sufficient numbers until Day 21. As a control, PBMCs of the same donor were grown and manipulated similarly but were not transduced. These control cells were cultured with or without G418. At the end, cells were counted and assessed for lymphocyte subset composition by FACS. The transduction efficiencies were estimated at day 11 by comparing the number of G418 resistant cells to that of transduced unselected cells, using trypan blue dye exclusion assay. Statistical comparisons The Mann-Whitney u-test was used for comparing median cell expansion rates in between cultures. Differences were considered significant when p ⱕ 0.05.

Results Cell expansion rates Median cell expansion rates for each medium/serum/OKT3 combination are depicted in Figures 1 and 2 for serum-free

1140

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

and serum-containing cultures, respectively. As seen in Figure 1, median expansion rates under strict serum-free conditions varied substantially between media, ranging from no expansion up to over 100-fold. In our hands, X-VIVO 15 yielded the most consistent and efficient cell expansion. The median cell expansion for 21 days of culture in serum-free media was 79-fold (range 20–117 times) for X-VIVO 15. For AIM-V and Cellgro, the corresponding median cell expansions were ninefold (range 0,1–26 times) and eightfold (range 1–58 times). In serum-free cultures, activation with OKT3 for 21 days frequently resulted in a minor increase in expansion rates, as compared to a 5-day course (Fig. 1), although without statistical significance. However, in cultures supplemented with serum (Fig. 2), this discrepancy was not evident. The variations in expansion rates for the serumcontaining cultures were less pronounced than in the serumfree cultures. Serum supplementation with 5% HS yielded a median of 267-fold (range 124–399), 231-fold (range 151– 358), and 242-fold (range 125–277) expansions for XVIVO 15, AIM-V, and Cellgro, respectively, after 21 days of culture. RPMI, serving as a reference medium, did not support cell expansion without the addition of serum. Supplementing RPMI with 5% HS in our hands gave a more consistent cell expansion than with 10% FBS, reaching a median of 154-fold expansion (range 103–196), as compared to 33-fold (range 1–199) for FBS. Cytokine production No cytokine production was detected in samples taken after five hours of culture. For the rest of the culture period, a

predominant type1 cytokine pattern was seen, with increased levels of IFN-␥ in most media combinations. Between day 5 and day 21, the median levels of IFN-␥ in serum-free cultures were 872 (range 110–10,000) pg/mL, as compared to 516 (range 416–955) pg/mL for the serum supplemented cultures. Cellgro was to a large extent responsible for this discrepancy, reaching a median level of 6837 (range 1047–10,000) pg/mL in serum-free cultures. IL-10 production could be detected only sporadically in a few percent of the samples taken, ranging from 30 to 620 pg/mL. No IL-4 production was found in any of the cultures, thus indicating concentrations below 3.0 pg/mL.

Transduction and G418 selection Exposure of primary T cells to the retroviral vector-containing supernatant followed by G418 selection resulted in G418-resistant cells, whereas control nontransduced cells did not survive the selection process. Since no major differences in cell expansion rates were seen when comparing the length of OKT3 stimulation, a five-day course of OKT3 was used in these experiments. Figure 3 depicts the fold expansion for transduced and G418 selected cells. The cultures supplemented with 5% HS yielded a median of 194-fold (range 49–419), 218-fold (range 34–390), 178-fold (range 29–401), and 180-fold (range 57–330) expansions for X-VIVO 15, AIM-V, Cellgro, and RPMI, respectively. X-VIVO 15 without serum reached a median of 14-fold expansion (range 3– 18). Transduction efficiencies were similar for all serum containing cultures, ranging between 30% and 40%. For se-

Figure 1. Median serum-free cell expansion for four donors. Each value given represents the median of 12 wells with identical culturing conditions. IL-2 (500 IU/mL) was present throughout the culturing period (5d OKT3 ⫽ OKT3 stimulation for the initial 5 days (10 ng/mL), 21d OKT3 ⫽ OKT3 stimulation for 21 days (10 ng/mL)).

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

1141

Figure 2. Median cell expansion for four donors with the supplement of 5% HS or 10% FBS. Each value given represents the median of 12 wells with identical culturing conditions. IL-2 (500 IU/mL) was present throughout the culturing period (5d OKT3 ⫽ OKT3 stimulation for the initial 5 days (10 ng/mL), 21d OKT3 ⫽ OKT3 stimulation for 21 days (10 ng/mL)).

rum-free X-VIVO 15 cultures the transduction efficiencies were lower, ranging from 10% to 15%. Phenotypical analysis of nontransduced and transduced cells Results of the flow cytometric analysis of nontransduced cells are summarized in Table 1 and Figure 4. Cultured cells

showed a rapid decrease in B cells and monocytes early in the culture period to levels lower than 1% of the cells. Initially, CD3⫺CD56⫹ NK cells also decreased, making CD3⫹ T cells the entirely predominating cell population, regularly constituting more than 95% of the cells on days 5 and 10. In most media combinations, the CD3⫹ T cells heavily pre-

Figure 3. Median cell expansion for transduced and G418 selected cells (4 donors). The cells were stimulated with OKT3 (10 ng/mL) for 5 days, and IL-2 (500 IU/mL) was present throughout the culturing period. Retroviral transduction with the LN vector was performed on day 4 and 5. This was followed by selection with G418 (1 mg/mL) for 7 days (day 7–14).

1142

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

Table 1. Flow cytometric analysis of viability, CD3⫹, CD4⫹, CD8⫹, CD3⫺, and CD56⫹ cells. Tha values presented are percentages (median (range)). Culture mediuma 0-Sample AIM-V

X-VIVO 15

X-VIVO 15 (21d OKT3) Cellgro

AIM-V, 5% HS

X-VIVO 15, 5% HS

X-VIVO 15, 5% HS (21d OKT3) Cellgro 5% HS

RPMI 5% HS

RPMI 10% FBS

Dayb

Viabilityc

% CD3⫹

% CD4⫹

% CD8⫹

% CD3⫺ CD56⫹

0 5 10 21 5 10 21 5 10 21 5 10 21 5 10 21 5 10 21 5 10 21 5 10 21 5 10 21 5 10 21

92 (89–99) 67 (36–80) 83 (75–93) 57 (10–71) 65 (50–78) 83 (68–86) 79 (52–94) 64 (50–80) 77 (63–90) 67 (40–93) 67 (56–78) 58 (36–75) 51 (14–89) 93 (79–97) 94 (90–98) 94 (93–97) 88 (77–97) 91 (80–98) 91 (89–95) 87 (77–98) 94 (69–97) 90 (86–92) 88 (74–97) 91 (85–96) 89 (85–92) 81 (64–96) 90 (76–98) 93 (90–98) 86 (51–96) 86 (55–93) 88 (6–98)

50 (36–67) 69 (67–79) 95 (88–99) 95 (82–99) 73 (67–84) 92 (82–98) 91 (13–92) 70 (65–86) 93 (81–98) 89 (51–96) 66 (58–79) 57 (35–79) 26 (3–48) 84 (80–96) 92 (78–97) 87 (47–98) 79 (71–94) 91 (60–97) 83 (24–98) 80 (70–95) 93 (58–97) 85 (42–98) 82 (64–94) 86 (54–96) 41 (6–92) 81 (74–91) 94 (77–97) 86 (45–96) 82 (68–84) 96 (86–97) 96 (44–96)

32 (22–43) 48 (37–59) 67 (50–85) 65 (27–70) 49 (35–57) 66 (45–71) 45 (4–60) 49 (33–55) 60 (42–71) 37 (6–63) 51 (36–63) 36 (10–47) 21 (1–37) 57 (36–67) 56 (46–71) 26 (17–33) 53 (39–71) 57 (32–75) 35 (8–39) 51 (36–73) 55 (30–72) 31 (12–48) 59 (43–75) 61 (31–74) 18 (3–23) 52 (51–69) 54 (41–78) 25 (17–54) 49 (45–64) 44 (38–55) 12 (6–15)

16 (6–34) 18 (14–18) 25 (14–35) 28 (26–50) 21 (18–25) 23 (16–35) 40 (10–47) 20 (17–26) 24 (17–44) 45 (30–65) 17 (12–22) 22 (15–29) 7 (2–19) 21 (15–30) 25 (15–46) 57 (30–62) 16 (10–25) 24 (11–46) 44 (17–55) 16 (14–34) 25 (11–48) 41 (30–64) 15 (12–22) 18 (11–32) 20 (3–56) 20 (13–29) 26 (10–56) 47 (24–63) 25 (13–26) 40 (24–52) 68 (34–78)

24 (5–38) 16 (5–25) 4 (1–9) 2 (1–13) 15 (4–22) 4 (2–13) 6 (4–84) 12 (4–25) 4 (1–14) 7 (3–47) 13 (2–32) 35 (16–51) 47 (37–96) 6 (1–13) 6 (1–20) 4 (0–14) 7 (2–20) 5 (2–33) 16 (1–76) 7 (2–22) 5 (2–34) 14 (1–56) 5 (2–20) 10 (3–38) 56 (7–92) 11 (3–16) 3 (1–6) 12 (3–50) 9 (2–15) 3 (1–10) 3 (1–55)

a

0-Sample designates Ficoll-separated leukocytes. All cells were stimulated with OKT3 for the initial 5 days of culture. RPMI without serum did not yield enough cells for analysis and is therefore not presented. A prolonged course of OKT3 (21 days) was assessed by flow cytometry for X-VIVO 15 only, indicated as 21d OKT3. b Samples were harvested on days 0, 5–6, 10–11, and 21. For simplicity, day 5–6 is presented as 5 and day 10–11 is presented as 10. c Propidium iodide staining was used for the viability analysis. The values indicate the percentage viable cells.

dominated the cell population throughout the 21-day culturing period. CD4⫹ T cells were commoner in the early culture period, whereas the increase in expansion of CD8⫹ T cells in most media was seen in day 10 and day 21 cultures. In the late culture period, not only CD8⫹ T cells became more frequent, but also CD3⫹ T cells expressing CD56 (data not shown). This was found in most media types but appeared to be more pronounced in the Cellgro medium. In Cellgro also, the CD3⫺/CD56⫹ NK cells were more frequent in the late cultures. However, the expansion rate was more limited, unless human serum was added (Table 1 and Fig. 4). Activation antigens, such as CD25 and HLA-DR, were induced in the early phase of the cultures, followed by a gradual decrease in the case of CD25. However, the kinetics of the CD25 and HLA-DR up-regulation appeared to differ slightly between donors and also between T-cell subsets. In one donor, the peak percentage of both CD25- and HLADR- expressing CD4⫹ T cells was seen on day 5, irrespec-

tive of media type, whereas in another, the peak was on day 10. Interestingly, in both donors the peak expression of CD25 on CD8⫹ T cells was seen on day 5, whereas HLADR expression peaked on day 10 (data not shown). The co-receptor CD28 remained highly expressed on CD4⫹ T cells over day 10, irrespective of media type. On day 21, CD4⫹ T cells in some of the RPMI cultures showed lower CD28 expression (⬍65%). CD8⫹ T cells had initially a lower and more variable expression of CD28. Compared to the frequency at culture onset, an increased proportion of the CD8⫹ T cells were CD28⫹ on day 10. In some of the X-VIVO 15 cultures, more than 80% of the CD8⫹ T cells were CD28⫹. On day 21, a gradual decrease in CD28 expression was seen also on CD8⫹ T cells (data not shown). The frequency of cells expressing CD45RA, associated with “naive” T cells, was higher on day 5, in both CD4⫹ and CD8⫹ T cells, than in the starting cell populations. On day 10, the expression of CD45RA dramatically dropped. Interestingly, on day 21, both the CD4⫹ and the CD8⫹ T-cell

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

1143

Figure 4. Absolute cell numbers (⫻ 106) for CD4⫹, CD8⫹, CD3⫺CD56⫹, and CD19⫹ cells. The left column presents serum-free cultures, and the right column presents cultures supplemented with 5% HS or 10% FBS. Each value given represents the median of 12 wells with identical culturing conditions. In addition to IL-2 (500 IU/mL), the cells were stimulated with OKT3 (10 ng/mL) for 5 days.

1144

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

populations from one donor increased the frequency of CD45RA⫹ cells again (data not shown). For transduced cells, the selection process itself did not shift the T-cell subsets significantly. For example, the percentages (median (range)) of cells expressing CD3⫹, CD4⫹, CD8⫹, and CD3⫺CD56⫹ were 97 (95–99), 36 (21–59), 50 (29–57), and 2 (0–4), respectively, for transduced and selected cells, cultured in serum-free X-VIVO 15 medium. The corresponding values for transduced and unselected cells, cultured under the same conditions, were 91 (81–99), 31 (23–76), 51 (18–59), and 3 (0–10), respectively. More importantly, when comparing the transduced cells with the nontransduced cells, no significant differences in cell subsets were seen. For example, the percentages (median(range)) of cells expressing CD3⫹, CD4⫹, CD8⫹, and CD3⫺CD56⫹ were 92 (84–96), 32 (15–41), 55 (44–63), and 5 (1–12), respectively, for transduced and unselected cells cultured in X-VIVO 15 supplemented with HS. The corresponding values for nontransduced unselected cells, cultured under the same conditions, were 89 (82–97), 44 (10– 56), 42 (33–67), and 5 (1–9), respectively.

Discussion The generation of suicide gene transduced T-lymphocytes in sufficient amounts for clinical use requires controlled in vitro cell expansion, preferably according to GMP standards. This may be achieved by defining all culture conditions where, as far as possible, chemically defined reagents and materials are used. Therefore, a serum-free approach to this process would be preferable. We have studied the expansion of T cells in three commercially available serum-free media, to establish an optimal protocol for retroviral suicide gene transduction. Lymphocytes cultured for 21 days have been assessed with regard to expansion rates, cytokine production, and cell subsets. Transduction with the retroviral LN vector, carrying the neomycin resistance gene, has been used to prove that cells could be efficiently transduced and selected. It is evident that cultures supplemented with serum give a more consistent and effective expansion of the cells than serum-free cultures. All cultures supplemented with HS, regardless of culturing media, supported T-cell expansion well (Fig. 2). RPMI, supplemented with FBS, gave inconsistent cell expansions, as compared to RPMI cultures supplemented with HS. If strictly serum-free conditions are required, X-VIVO 15 is, in our hands, an option, giving a median of 79-fold cell expansion. X-VIVO 15 was clearly superior to the other two serum-free media tested in this study (AIM-V and Cellgro SCGM), which did not reach a median cell expansion greater than ninefold. We assessed the effect of prolonged OKT3 stimulation— that is, 21 days instead of 5 days. In many cases of serumfree expansion, activation with OKT3 for 21 days gave a

very minor increase in expansion rates, as compared to a 5-day course. However, in cultures supplemented with serum, this difference was not evident. The addition of OKT3 to IL-2, has been shown to promote preferentially the expansion of CD4⫹ T cells [22]. In our study, many media combinations showed a peak in CD4⫹ cells around day 10. A continuous stimulation with OKT3, assessed by flow cytometry only for X-VIVO 15 cultures, did not prevent a decline in this cell subset toward the end of the expansion period (Table 1). The optimal CD4/CD8 ratio for obtaining a maximal GVL effect, together with minimal GVHD, is, however, controversial. Moreover, it is likely that this may vary for different types of leukemias, and depending on whether the T cells are administered at the time of transplant or as a so-called donor leukocyte infusion (DLI), at a later time point. In CML patients, allo-transplants and donor lymphocyte infusions depleted of CD8⫹ cells have been shown to reduce the incidence of GVHD, while preserving the GVL effect [24,25]. However, since these CD8-depleted grafts still contain a small number of CD8⫹ cells, and these cells may expand in vivo, one cannot be certain that CD8⫹ cells have no importance for the GVL effect. Since there may be inter-donor variations in the expanded T-cell subsets, simultaneous and separate expansions of CD4⫹ and CD8⫹ cells have been suggested for clinical practice [15]. Until more data become available on the clinical effects of the various T-cell subsets, we believe that it is favorable to have a mixture of both CD4⫹ and CD8⫹ T cells in the expanded population. In our hands, this was achieved quite consistently with the X-VIVO 15 medium. Further T-cell subset analysis by flow cytometry indicated that, in addition to inter-individual variation, there might be a complex population dynamic between phenotypically distinguishable subsets. Thus CD45RA expression dropped on day 10 and then reappeared in one patient. CD28 expression on CD8⫹ T cells also appeared to change over time and to be more frequent in X-VIVO–based cultures. Whether these phenotypical changes relate to loss and reappearance of T-cell subsets or maturation-related phenotypic changes on continuously present T cells remains to be determined. The relative enrichment of CD8⫹ T cells over time may influence the expected in vivo effect. Retroviral transduction of the cells with the LN vector and subsequent G418 selection did not seem to cause any significant shifts in the different T-cell subsets. The final yield of transduced and selected cells at day 21 was high in this study. We believe that one reason for this is that the control cells (transduced and unselected) expand to a higher cell density, as compared to the cells that are under G418 selection. Due to contact inhibition by cell-cell interactions this may temporarily lead to a relative decrease in the mitotic rate of the control cells. On the other hand, as the G418-resistant cells grow out, these will initially be less dense, allowing them to temporarily expand at a relatively higher rate. Therefore the G418 selected cultures are, to

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

some degree, able to catch up with the unselected cultures, with respect to the final cell yield. Serum-free Cellgro medium was used for the production of the retrovirus-containing supernatant, as it gave the best titers, and since X-VIVO 15 did not support the growth of our murine producer cell line (PG13). However, since the transduction procedure was short (2 hours) we believe that the media itself did not affect the cells during this time period. From an experimental point of view, the starting number of cells in the culture and the purity of the Ficoll-separated product seemed to be of importance for the ability of the cells to expand in vitro. For simplicity, we attempted to start a cell expansion series with 30,000 cells/well. However, these cells did not expand in any of the media combinations. Initial flow-cytometric analysis gave no abnormal findings that could be donor-related (data not shown). We therefore assume that this was merely an effect of the cells being too diluted at the start of expansion. We also noticed, in another expansion series, that cells from one donor did not expand, in spite of an adequate starting cell number (200,000 cells/ well). In this case, the initial FACS analysis showed a very heterogeneous cell population, probably due to a poor Ficoll separation of the T cells (data not shown). Looking at the cytokine content in the cell supernatants, the levels varied substantially between the two donors and the different media combinations. A type1 cytokine pattern with increased levels of IFN-␥ was seen for virtually all media combinations (median 872, range 110–10,000 pg/mL). In general, serum-free cultures achieved higher median IFN-␥ levels than did serum-supplemented cultures. Cellgro was responsible for this discrepancy to a large extent, reaching a median level of 6837 (range 1047–10,000) pg/mL. The reason for this may be that Cellgro tended to promote NK-cell expansion to a higher extent than did the other culturing media (Table 1). No IL-4 production was detected, and IL-10 production was seen only sporadically in low levels. In conclusion, we have found that serum-free X-VIVO 15 gives the best support to the expansion of T cells in vitro. Culturing for 21 days gave a median of 79-fold cell expansion, with approximately equal percentages of CD4⫹ and CD8⫹ cells. There seem to be no significant benefits, regarding absolute cell numbers or T-cell subset compositions, with OKT3-stimulation for more than 5 days. These cells can easily be transduced and selected without significantly shifting the various T-cell subsets. For clinical large-scale expansions, the addition of low levels of HS will increase the consistencies in the cell expansion rates. This should preferentially be done with serum from the T cell donor. Acknowledgments We thank Birgitta Stellan, Kristina Friberg, and Hernán ConchaQuesada for skillful technical assistance. This study was supported by grants from the Swedish Child Cancer Fund, the Swedish Tobias Foundation, the Swedish Cancer Fund, the Swedish Society for Medicine, the Clas Groschinskys

1145

Minnesfond (Sweden), the Swedish Society for Medical Research, Henning & Ida Perssons Forskningstiftelse, and the Swedish Medical Research Council.

References 1. Giralt SA, Champlin RE (1994) Leukemia relapse after allogeneic bone marrow transplantation: a review. Blood 84:3603 2. Carlens S, Ringden O, Aschan J, et al. (1998) Risk factors in bone marrow transplant recipients with leukaemia. Increased relapse risk in patients treated with ciprofloxacin for gut decontamination. Clin Transpl 12:84 3. Weiden PL, Flournoy N, Thomas ED, et al. (1979) Antileukemic effect of graft-versus-host disease in human recipients of allogeneicmarrow grafts. N Engl J Med 300:1068 4. Horowitz MM, Gale RP, Sondel PM, et al. (1990) Graft-versus-leukemia reactions after bone marrow transplantation. Blood 75:555 5. Ringden O, Labopin M, Gluckman E, et al., Acute Leukemia Working Party of the European Group for Blood and Marrow Transplantation. (1996) Graft-versus-leukemia effect in allogeneic marrow transplant recipients with acute leukemia is maintained using cyclosporin A combined with methotrexate as prophylaxis. Bone Marrow Transplant 18:921 6. Ferrara JL, Deeg HJ (1991) Graft-versus-host disease: a review. N Engl J Med 324:667 7. Klingebiel T, Schlegel PG (1998) GVHD: overview on pathophysiology, incidence, clinical and biological features. Bone Marrow Transplant 21:S45 8. Goldman JM, Gale RP, Horowitz MM, et al. (1988) Bone marrow transplantation for chronic myelogenous leukemia in chronic phase. Increased risk for relapse associated with T-cell depletion. Ann Intern Med 108:806 9. Aschan J, Ringden O, Sundberg B, Gahrton G, Ljungman P, Winiarski J (1991) Methotrexate combined with cyclosporin A decreases graftversus-host disease, but increases leukemic relapse compared to monotherapy. Bone Marrow Transplant 7:113 10. Dazzi F, Szydlo RM, Goldman JM (1999) Donor lymphocyte infusions for relapse of chronic myeloid leukemia after allogeneic stem cell transplant: where we now stand. Exp Hematol 27:1477 11. Dazzi F, Szydlo RM, Craddock C, et al. (2000) Comparison of singledose and escalating-dose regimens of donor lymphocyte infusion for relapse after allografting for chronic myeloid leukemia. Blood 95:67 12. Tiberghien P, Reynolds CW, Keller J, et al. (1994) Ganciclovir treatment of herpes simplex thymidine kinase-transduced primary T lymphocytes: an approach for specific in vivo donor T-cell depletion after bone marrow transplantation? Blood 84:1333 13. Bonini C, Ferrari G, Verzeletti S, et al. (1997) HSV-TK gene transfer into donor lymphocytes for control of allogeneic graft-versus-leukemia. Science 276:1719 14. Contassot E, Murphy W, Angonin R, et al. (1998) In vivo alloreactive potential of ex vivo-expanded primary T lymphocytes. Transplantation 65:1365 15. Marini FC, III, Kornblau SM (1999) Production and culture of HSVtk transduced suicidal lymphocytes induces variable changes in the lymphocyte subset composition. Bone Marrow Transplant 23:907 16. Tiberghien P, Cahn JY, Brion A, et al. (1997) Use of donor T-lymphocytes expressing herpes-simplex thymidine kinase in allogeneic bone marrow transplantation: a phase I-II study. Hum Gene Ther 8:615 17. Link CJ, Jr, Burt RK, Traynor AE, et al. (1998) Adoptive immunotherapy for leukemia: donor lymphocytes transduced with the herpes simplex thymidine kinase gene for remission induction. HGTRI 0103. Hum Gene Ther 9:115 18. Morecki S, Gelfand Y, Levi S, et al. (1997) Activated long-term peripheral blood cultures as preparation for adoptive alloreactive cell therapy in cancer patients. J Hematother 6:115 19. Gersten MJ, Cohn M (1987) T cell responses studied in a basal serumfree medium. Methods Enzymol 150:129

1146

S. Carlens et al./Experimental Hematology 28 (2000) 1137–1146

20. Shive W, Pinkerton F, Humphreys J, Johnson MM, Hamilton WG, Matthews KS (1986) Development of a chemically defined serum- and protein-free medium for growth of human peripheral lymphocytes. Proc Natl Acad Sci U S A 83:9 21. Munshi NC, Govindarajan R, Drake R, et al. (1997) Thymidine kinase (TK) gene-transduced human lymphocytes can be highly purified, remain fully functional, and are killed efficiently with ganciclovir. Blood 89:1334 22. Yoshino I, Yano T, Miyamoto M, Sugimachi K, Kimura G, Nomoto K (1993) Phenotypic and functional modulation of interleukin-2-activated peripheral blood mononuclear cells by anti-CD3 and anti-CD28 antibody. Lymphokine Cytokine Res 12:191 23. Contassot E, Ferrand C, Certoux JM, et al. (1998) Retrovirus-mediated transfer of the herpes simplex type I thymidine kinase gene in alloreactive T lymphocytes. Hum Gene Ther 9:73 24. Champlin R, Ho W, Gajewski J, et al. (1990) Selective depletion of CD8⫹ T lymphocytes for prevention of graft-versus-host disease after allogeneic bone marrow transplantation. Blood 76:418 25. Giralt S, Hester J, Huh Y, et al. (1995) CD8-depleted donor lymphocyte infusion as treatment for relapsed chronic myelogenous leukemia after allogeneic bone marrow transplantation. Blood 86:4337 26. Truitt RL, Atasoylu AA (1991) Contribution of CD4⫹ and CD8⫹ T

27.

28.

29. 30.

31.

32.

cells to graft-versus-host disease and graft-versus-leukemia reactivity after transplantation of MHC-compatible bone marrow. Bone Marrow Transplant 8:51 Sykes M, Abraham VS, Harty MW, Pearson DA (1993) IL-2 reduces graft-versus-host disease and preserves a graft-versus-leukemia effect by selectively inhibiting CD4⫹ T cell activity. Journal of Immunology 150:197 Nagler A, Condiotti R, Nabet C, et al. (1998) Selective CD4⫹ T-cell depletion does not prevent graft-versus-host disease. Transplantation 66:138 Ferrara JL (1995) Cytokine inhibitors and graft-versus-host disease. Ann N Y Acad Sci 770:227 Rus V, Svetic A, Nguyen P, Gause WC, Via CS (1995) Kinetics of Th1 and Th2 cytokine production during the early course of acute and chronic murine graft-versus-host disease. Regulatory role of donor CD8⫹ T cells. J Immunol 155:2396 Miller AD, Buttimore C (1986) Redesign of retrovirus packaging cell lines to avoid recombination leading to helper virus production. Mol Cell Biol 6:2895 Miller AD, Garcia JV, von Suhr N, Lynch CM, Wilson C, Eiden MV (1991) Construction and properties of retrovirus packaging cells based on gibbon ape leukemia virus. J Virol 65:2220