- Email: [email protected]
The COSTIM bioassay: a novel potency test for dendritic cells
Journal of Immunological Methods 285 (2004) 293 – 299 www.elsevier.com/locate/jim
Protocol
The COSTIM bioassay: a novel potency test for dendritic cells Gopi Shankar *, Robert Bader, Patricia A. Lodge Northwest Biotherapeutics, Inc., 22322-20th Avenue S.E., Bothell, WA 98021, USA Received 8 December 2003; accepted 15 December 2003
Abstract The utility of dendritic cells (DCs) in experimental immunotherapy has driven significant advances in the manufacture of these cells. They are increasingly prepared in vitro for use in clinical trials of human disease, particularly cancer. Thus, it has become imperative that, in concert with other quality control measures, a potency test be employed for lot (batch)-release testing of DC products, both in preclinical studies and human clinical trials. The mixed lymphocyte reaction (MLR) assay has served as a ‘gold standard’ for evaluating the functional ability of antigen presenting cells. Alternatively, some researchers also employ immunophenotyping, a test unrelated to cellular function, as a potency-determining test. We have developed a novel method named the ‘COSTIM bioassay’, which, as we describe in this paper, is suitable for quality control or lot-release testing. In this method T-cells are stimulated with a sub-optimal amount of anti-CD3 antibody, such that they remain unable to proliferate unless a source of co-stimulation (accessory cells, such as DC) is added to the culture. Thus, the COSTIM bioassay is a functional test that selectively measures co-stimulatory activity, or functional potency. This method takes less than 2 days for completion and assures better quality control than the MLR. D 2004 Elsevier B.V. All rights reserved. Keywords: Dendritic cells; Potency; Quality control; Co-stimulation; Bioassay
1. Type of research Potency testing is critical for lot (batch)-release of a therapeutic product. From a regulatory perspective, it is defined as ‘‘the specific ability or capacity of the product, as indicated by appropriate laboratory tests or by adequately controlled clinical data obtained through administration of the product in the manner intended, to effect a given result’’ (U.S. Code of Federal Regulations, 21 CFR 600.3s). In other words, potency is essentially a measure of the potential effi-
* Corresponding author. Centocor, Inc., 200 Great Valley Pkwy, Malvern, PA 19355, USA. Tel.: +1-610-407-8922. E-mail address: [email protected] (G. Shankar). 0022-1759/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2003.12.008
cacy of a vaccine or drug substance. In the case of cellbased biotherapeutics, due to the complexity of the myriad reactions of the immune system to the product, it is not clear that a single activity (such as cytokine secretion or surface marker expression) will eventually correlate with clinical efficacy. Thus, a test of one of the expected mechanisms of action of the therapeutic product serves as a potency test. The case of dendritic cell (DC)-based vaccines in particular is unique since DC lots (or batches) are patient-specific, unlike currently marketed biotherapeutic and pharmaceutic drugs. Leukocytes (generally, peripheral blood derived mononuclear cells) from a patient are obtained, cultured ex vivo, often exposed to a target antigen, and then infused back into the same patient with the expectation that a target-specific im-
294
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
mune response will be stimulated in the patient’s body resulting in a therapeutic effect. The requisite mechanism of action can be described in terms of the classic ‘two-signal hypothesis’: Signal one is composed of the MHC – antigen complex and signal two is composed of co-stimulatory signals. For antigen-loaded DC products, the ideal potency test would evaluate the level of target antigen-specific T-cell-stimulatory activity of the DCs in vitro. This could comprise a bioassay measuring activation of the patient’s own T-cells by antigenloaded autologous DCs. This approach is fraught with deficiencies: (U.S. Code of Federal Regulations) as such, it would have to be assumed that the patient’s cells would respond under the experimental conditions being employed, (Monji et al., 2000) the detection of an antigen-specific T-cell response is near-impossible using primary T-cells (peripheral blood derived T-cells), warranting the in vitro production of antigen-specific secondary or tertiary cell lines that could give a detectable response in such an assay, (Steinman and Witmer, 1978) false positive or negative T-cell responses could result from the in vitro manipulations, (Van Voorhis et al., 1983) the strategy of making secondary or tertiary cell lines consumes too many of the patient’s valuable peripheral blood cells, and is time-consuming (at least 1 –2 months) and expensive, without any assurance that the assay would eventually succeed. Therefore, the quandary is that the mechanism of action, antigen-specific T-cell stimulation, cannot be tested for autologous DC-based vaccines. To circumvent this, the allostimulatory capacity of DCs has been generally accepted as a potency determinant and tested via the mixed lymphocyte reaction (MLR) assay. The MLR has served as the ‘gold standard’ for testing the functional ability of antigen presenting cells (Steinman and Witmer, 1978; Van Voorhis et al., 1983). However, MLR assays take several days to complete and may be, quite arguably, non-representative of the activity of dendritic cells in context of their stimulatory interactions with autologous antigen-specific T-cells. MHC antigens constitute a powerful signal in the MLR; the co-stimulation of alloreactive T-cells by DCs in the MLR amplifies the extent of histoincompatibility between the DCs and T-cells, making it difficult to independently assess the impact of the co-stimulatory capacity of the DCs versus the selective presentation of strong alloanti-
gens. From the quality control perspective, the MLR is also a rather uncontrolled system since the stimulating alloantigens will vary vastly between the batches of T-cells (responder cells) and batches of DCs (stimulator cells). For example, when low MLR data is obtained for a sample, it is unclear whether this was a result of truly impotent DCs, or due to a significant histocompatibility match between the DCs and T-cells. Alternatively, immunophenotyping by flow cytometry has been used as a measure of DC potency (Monji et al., 2000), but is obviously inaccurate because it does not measure cellular function. A highly pure but non-viable or apoptotic (and thus obviously impotent) DC preparation would erroneously pass such a potency test. The ‘COSTIM bioassay’ described here selectively measures co-stimulatory activity but not antigen processing and presentation. We believe that this bioassay is more relevant to the function of DCs in a vaccine product, is more accurate, reliable, and efficient than currently used DC potency tests, and is also suitable for lot-release quality control testing.
2. Time required
It is strongly advised that a large T-cell batch be prepared in advance and convenient aliquots cryopreserved. It is impractical to make individual T-cell preparations for each run of the bioassay as it adds several hours and is also a source of variation in the assay. The time for purifying T-cells is thus not specified here. Similarly, the time for preparing DCs is also not specified here since the methods of manufacture and the times vary significantly.
The COSTIM assay set-up time is V 3 h, involving:
Thawing T-cells and DCs, washing, counting, and re-suspension: 1 – 2 h Culture set-up: V 1 h
The total culture period is 44 h, involving:
Initial culture period, prior to the addition of tritiated thymidine: 26 h Preparing working stock of tritiated thymidine and addition to culture wells: V 10 min
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
Final culture period: 18 h Harvesting cells and scintillation counting: V 1 h, but times may vary based on the equipment used. Data manipulation and results analysis: V 15 min.
295
4. Detailed procedure
3. Materials 3.1. Cells and reagents
Dendritic cells Allogeneic T-cells AIM-Vk Culture Medium (Life Technologies, Rockville, MD) Anti-Human CD3 Antibody, 1 mg/ml, Azidefree/low endotoxin (BD Pharmingen, San Diego, CA) Tritiated (3H) thymidine, 1 mCi/ml, 74.0 GBq/ mmol (Perkin Elmer Life Sciences, Boston, MA)
3.2. Supplies
Ice Hemacytometer Trypan blue (Sigma, St. Louis, MO) 96-well U-bottom plates (VWR International, West Chester, PA) 15-ml centrifuge tubes with screw caps (VWR International) 70% Isopropanol
3.3. Equipment
Sterile, laminar flow, biological safety cabinet Incubator, 37 F 2 jC, 5% CO2, humidifying ( z 70% relative humidity) Microscope for cell counting Centrifuge Water bath, 37 F 2 jC Pipettor Single-channel micropipettors, 0.5 –10, 20 – 200, and 100– 1000 Al Multi(12)-channel micropipettor, 20 – 200 Al FilterMatek 96-channel culture plate harvester (Packard Instruments, Meriden, CT), or equivalent TopCountk NXT scintillation counter (Packard Instruments), or equivalent
4.1. Preparation of cells (i) It is advised that one or more T-cell batches be prepared well in advance and convenient aliquots cryopreserved. T-cells may be enriched or purified by any one of several methods. We have employed biomagnetic separation using antiHLA-DR monoclonal antibody-conjugated paramagnetic beads (Dynal, Lake Success, NY) to deplete accessory cells such as monocytes, DCs, and B-cells, applying the manufacturer’s recommended protocol. This typically resulted in a suspension comprising of 80 –90% T-cells (the remainder being NK cells, which were found to be inert in this assay). The cells are cryopreserved in 90% autologous serum and 10% dimethyl sulfoxide (DMSO) until further use. Irrespective of the T-cell enrichment method used, the purity of the preparation(s) should be determined by flow cytometry, and their general functional ability qualified by PHA-induced mitogenesis or MLR prior to use in the COSTIM bioassay. Similarly, the preparation of DCs is beyond the scope of this paper since the methods of manufacture vary significantly. (ii) Thaw the cryopreserved T-cells and DCs in a 37 F 2 jC water bath. Immerse the cryovials in a water bath just long enough to thaw completely, but no longer than 2 –3 min to ensure optimal post-thaw viability. (iii) Immediately after thawing, resuspend T-cells and DCs in separate 15-ml centrifuge tubes containing 10 – 12 ml of warm AIM-V medium (add the cells to the media slowly to ensure good viability). (iv) Centrifuge the tubes at 400 g, at 1– 8 jC, for 10 F 2 min to sediment the cells. (v) After centrifugation, decant or aspirate supernatant completely, and resuspend the cells in AIM-V medium, and perform cell counting and viability determination by Trypan blue exclusion using the hemacytometer and microscope. Alternatively, automated cell counting/viability may be performed. (vi) Using AIM-V medium, resuspend the DCs at 1 105 viable cells/ml, and T-cells at 1 106
296
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
viable cells/ml. Split each cell suspension by half into separate tubes, and store on ice. Label the four individual tubes as follows: (a) T-cells (b) DCs (c) T-cells with anti-CD3 (d) DCs with anti-CD3 (vii) Prepare a working solution of anti-CD3 monoclonal antibody: Add 2.0 Al of 1.0 mg/ ml anti-CD3 antibody [stock] into 1 ml of AIM-V medium, resulting in a 2.0 Ag/ml working solution (be cautious when dispensing the 2.0 Al of anti-CD3; immerse only the tip of the pipette tip into the meniscus of the media to minimize dilution errors). (viii) To tubes labeled ‘‘with anti-CD3’’, add 5 Al of anti-CD3 antibody working solution per milliliter of cell suspension, resulting in a 0.01 Ag/ml antibody concentration within those cell suspensions.
(ii) Dispense 50 Al of the above working solution per culture well (i.e., 0.5 ACi per well). (iii) Incubate the plate for 18 h in a 37 F 2 jC, 5% CO2, humidifying incubator. 4.4. Harvest and scintillation counting (i) Using the 96-channel culture plate harvester, collect the cells onto a fiberglass-backed filter paper. In this process, wash the plate wells, with filter in place, five times with distilled water and subsequently once with 70% isopropanol. This lyses the cells, retaining cellular debris and nucleic acids on the filter. (ii) Completely dry the filter by placing it for 20 min under a heat lamp (alternatively, the filter may be air-dried at room temperature, but this can take several hours). (iii) Following the scintillation counter manufacturer’s instructions, determine the counts per minute (cpm) of the incorporated radioactivity in the individual culture wells.
4.2. Culture set-up 4.5. Data manipulation and results analysis (i) In a 96-well culture plate, co-culture the following three combinations of DCs and T-cells in triplicate wells, in a total volume of 200 Al per well: (a) 100 Al of T-cells + 100 Al of DCs (Background MLR control group) (b) 100 Al of T-cells + 100 Al of DCs with anti-CD3 (COSTIM group) (c) 100 Al of T-cells with anti-CD3 + 100 Al of AIMV medium (T-cell background proliferation control)
(i) Calculate the average cpm result of triplicate wells. The scintillation counter is programmable to perform this calculation. The average cpm value is sufficient for data analysis. (ii) Assure that the background T-cell proliferation control (T-cells with anti-CD3 only) does not proliferate over 2000 cpm. This system suitability
As a result, there will be 1 105 T-cells and/or 1 104 DCs, with or without 0.005 Ag/ml antiCD3 antibody in the wells. (ii) Incubate the plate for 26 h in a 37 F 2 jC, 5% CO2, humidifying incubator. 4.3. Addition of tritiated (3H) thymidine to culture wells (i) A few minutes prior to the end of the 26h incubation, prepare a 10 ACi/ml working solution of tritiated (3H) thymidine in AIM-V medium.
Fig. 1. Typical results of the COSTIM assay are shown. Due to the short (44 h) culture in this assay, the alloantigenic MLR response is insignificant.
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
297
Fig. 2. Different numbers of dendritic cells, monocytes, and B-cells (stimulator cells) were plated in this assay with a constant number (1 105 per well) of responder cells T-cells. The COSTIM assay delineated the magnitude of difference in the co-stimulatory ability of accessory cells.
Fig. 3. (A) Three lots of dendritic cells were placed in culture with T-cells and anti-CD3 antibody for 44 h (COSTIM assay) or with T-cells alone for 6 days (MLR assay). Results of the two assays are shown. (B) Six different lots of T-cells were tested against three lots of dendritic cells in the COSTIM assay to gain an estimate of method robustness.
298
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
control does not exceed 2000 cpm if the T-cell batch was previously qualified properly (see Section 4.1, step i). Proliferation higher than 2000 cpm by the T-cells alone cultured with anti-CD3 indicates that there is accessory cell activity present in the T-cell preparation, and that the COSTIM assay results may not be reliable. (iii) Calculate DC potency by using the average cpm and the following equation: Potency (cpm) = COSTIM cpm (T-cells + DCs + anti-CD3) Background MLR cpm(T-cells + DCs).
5. Results The COSTIM assay largely precludes alloantigenic MLR responses (Fig. 1) because of the short (44 h) culture period. This method for potency determination is not limited to DCs; other accessory cells such as monocytes (Fig. 2) may also be tested, possibly with minor adjustments to the assay. For instance, the results in Fig. 2 show that a greater number of monocytes per well (compared to DCs) is needed for a detectable proliferative response by the 1 105 T-cells in the well. We routinely use this test for DC potency, employing a ratio of 1:10 for DCs to T-cells. To demonstrate the advantage of the COSTIM bioassay over the MLR bioassay, the experiment depicted in Fig. 3A was performed. Three lots of DCs (from three separate PBMC donors) were prepared under identical manufacturing conditions and tested by the COSTIM method versus the traditional 6-daylong MLR assay. As shown, the inter-lot variation for DCs was much lower for COSTIM (CV, 10.8%; S.D., 5225 cpm) versus the MLR (CV, 28.7%; S.D., 29721 cpm). Thus, the COSTIM assay is a more robust method than the MLR. The major variable that we anticipated to have a profound effect on assay performance was the responder T-cell preparations. To understand the variation between T-cell lots, the performance of 6 different allogeneic T-cell lots (all prepared identically) in the COSTIM bioassay was compared against three lots of DCs (Fig. 3B). The inter-lot variation for DCs in this experiment was also low (CV, 9.0%; S.D., 5036), whereas the inter-lot variation for T-cells was much higher (CV, 30.6%; S.D., 18618). These results indicate that different Tcell lots vary widely in their proliferative abilities but
within a given T-cell lot assay variation is markedly low.
6. Discussion This protocol describes a simple method for testing the potency of accessory cells such as DCs. Unlike the MLR, it effectively separates co-stimulatory activity from antigen processing and presentation. The response of T-cells in this assay is almost completely dependent upon the co-stimulatory molecules CD54, CD80, and CD86 on the surface of DCs (not shown). The COSTIM bioassay is more functionally relevant to DCs, is more accurate, robust, and efficient than the MLR, and is also well suited for lot-release quality control testing of DC-based vaccines. For a ‘naked’ DC product (i.e., when antigen is not loaded into DCs), the COSTIM assay alone should suffice as a lot-release test. Perhaps an antigen-uptake activity assay could be used in conjunction with it. In the case of antigen-loaded DC products, a test that determines the presence of loaded antigen in the cells may be combined with the COSTIM test for lot-release purposes. We have used the latter approach successfully for our product, DCVaxk-Prostate (prostate specific membrane antigen-loaded DCs), with acceptance by our product reviewers at the U.S. Food and Drug Administration (FDA). 6.1. Troubleshooting We have found that most assay failures relate to the cellular concentration and viability of the cells. Since the level of proliferative response in this bioassay is directly proportional to the numbers and viability of Tcells and DCs in culture, an inadvertent miscalculation of cell concentration, or estimation of Trypan blue excluding cells (viable cells) by an inexperienced analyst can lead to great variation in assay performance. It is critical that the viability be determined accurately. The use of an automated viability counter leads to improved precision in our experience. We also recommend that a well-characterized lot of DCs be utilized as a positive control or reference lot that is included in every run. The qualification of a responder T-cell lot must be performed carefully, and specifications for the acceptance of a T-cell lot could be set in
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
place to reduce variations in assay performance. This qualification should involve purity, determined by flow cytometry, and general T-cell functional ability qualified by mitogenesis (MLR or PHA-induced) prior to use in the COSTIM bioassay. By monitoring the performance our system suitability controls, we have also found that the stability of tritiated thymidine is generally unpredictable after 3 months, leading to lower-than-normal results for the positive control and false-negative sample results. It is recommended that a lot of anti-CD3 antibody be frozen in appropriate aliquots to preserve stability. 6.2. Alternative and support protocols The COSTIM assay can be performed with either autologous or allogeneic T-cells resulting in nearly identical data. We have confirmed this with experiments comparing the proliferative responses of autologous T-cells with allogeneic T-cells and have found that they were statistically indifferent (not shown). The DCs, as described in this protocol, need not be irradiated since DC do not proliferate. We have confirmed this using irradiated (20 – 30 Gy) and non-irradiated DCs and found that they were statistically indifferent (not shown). Instead of measuring T-cell proliferation, activation of T-cells by the DCs may also be evaluated by immunophenotypic measurement of CD69 and CD25 on the T-cell surface. Since these markers may be evaluated within hours of activation, this approach can offer quicker analysis of potency for products with short expiration dating. Yet, another alternative approach (depending on the targeted mechanism of action of the DC vaccine) is to collect the supernatant culture medium from COSTIM cultures and determine the level of a T-cell cytokine. We have evaluated the level of IFN-g in our cultures, which correlated extremely well with the proliferation data. However, the proliferation read-out for this assay is logistically the simplest, most efficient, and economical. 7. Quick procedure (i) Prepare and qualify responder T-cells in advance. (ii) Thaw the cryopreserved T-cells and DCs, wash, and count.
299
(iii) Resuspend DCs at 1 105 viable cells/ml, and T-cells at 1 106 viable cells/ml. (iv) To one half of each cell suspension, add antiCD3 monoclonal antibody to a final concentration of 0.01 Ag/ml. (v) In a 96-well culture plate, co-culture the following three combinations of DCs and Tcells in triplicate wells, in a total volume of 200 Al per well: (a) 100 Al of T-cells + 100 Al of DCs (Background MLR control group) (b) 100 Al of T-cells + 100 Al of DCs with anti-CD3 (COSTIM group) (c) 100 Al of T-cells with anti-CD3 + 100 Al of AIMV medium (T-cell background proliferation control) (vi) Incubate the plate for 26 h in a 37 jC F 2 jC, 5% CO2, humidifying incubator. (vii) To the culture wells, add 50 Al of 10 ACi/ml of tritiated (3H) thymidine. (viii) Incubate the plate for 18 h in a 37 F 2 jC, 5% CO2, humidifying incubator. (ix) Harvest the cells onto a filter and measure the incorporated radioactivity by scintillation counting. (x) Using the average result of the triplicate wells, calculate DC potency by using the average cpm and the following equation: Potency (cpm) = COSTIM cpm (T-cells + DCs + anti-CD3) Background MLR cpm(T-cells + DCs).
References U.S. Code of Federal Regulations. 21 CFR 600.3(s). U.S. Government Printing Office, Washington, DC, USA. Monji, T., Smits-Sana, B., Qian, L., Stark, L., Therond, J., Shiomoto, C., Peshwa, M.V., 2000. Development of assay for assessment of dendritic cell product potency. International Society for Hematotherapy and Graft Engineering (ISHAGE) 2000, San Diego, CA, Abstract 139. Steinman, R.M., Witmer, M.D., 1978. Lymphoid dendritic cells are potent stimulators of the primary mixed leukocyte reaction in mice. Proc. Natl. Acad. Sci. U. S. A. 75, 5132 – 5136. Van Voorhis, W.C., Valinsky, J., Hoffman, E., Luban, J., Hair, L.S., Steinman, R.M., 1983. Relative efficacy of human monocytes and dendritic cells as accessory cells for T cell replication. J. Exp. Med. 158, 174 – 191.
Protocol
The COSTIM bioassay: a novel potency test for dendritic cells Gopi Shankar *, Robert Bader, Patricia A. Lodge Northwest Biotherapeutics, Inc., 22322-20th Avenue S.E., Bothell, WA 98021, USA Received 8 December 2003; accepted 15 December 2003
Abstract The utility of dendritic cells (DCs) in experimental immunotherapy has driven significant advances in the manufacture of these cells. They are increasingly prepared in vitro for use in clinical trials of human disease, particularly cancer. Thus, it has become imperative that, in concert with other quality control measures, a potency test be employed for lot (batch)-release testing of DC products, both in preclinical studies and human clinical trials. The mixed lymphocyte reaction (MLR) assay has served as a ‘gold standard’ for evaluating the functional ability of antigen presenting cells. Alternatively, some researchers also employ immunophenotyping, a test unrelated to cellular function, as a potency-determining test. We have developed a novel method named the ‘COSTIM bioassay’, which, as we describe in this paper, is suitable for quality control or lot-release testing. In this method T-cells are stimulated with a sub-optimal amount of anti-CD3 antibody, such that they remain unable to proliferate unless a source of co-stimulation (accessory cells, such as DC) is added to the culture. Thus, the COSTIM bioassay is a functional test that selectively measures co-stimulatory activity, or functional potency. This method takes less than 2 days for completion and assures better quality control than the MLR. D 2004 Elsevier B.V. All rights reserved. Keywords: Dendritic cells; Potency; Quality control; Co-stimulation; Bioassay
1. Type of research Potency testing is critical for lot (batch)-release of a therapeutic product. From a regulatory perspective, it is defined as ‘‘the specific ability or capacity of the product, as indicated by appropriate laboratory tests or by adequately controlled clinical data obtained through administration of the product in the manner intended, to effect a given result’’ (U.S. Code of Federal Regulations, 21 CFR 600.3s). In other words, potency is essentially a measure of the potential effi-
* Corresponding author. Centocor, Inc., 200 Great Valley Pkwy, Malvern, PA 19355, USA. Tel.: +1-610-407-8922. E-mail address: [email protected] (G. Shankar). 0022-1759/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2003.12.008
cacy of a vaccine or drug substance. In the case of cellbased biotherapeutics, due to the complexity of the myriad reactions of the immune system to the product, it is not clear that a single activity (such as cytokine secretion or surface marker expression) will eventually correlate with clinical efficacy. Thus, a test of one of the expected mechanisms of action of the therapeutic product serves as a potency test. The case of dendritic cell (DC)-based vaccines in particular is unique since DC lots (or batches) are patient-specific, unlike currently marketed biotherapeutic and pharmaceutic drugs. Leukocytes (generally, peripheral blood derived mononuclear cells) from a patient are obtained, cultured ex vivo, often exposed to a target antigen, and then infused back into the same patient with the expectation that a target-specific im-
294
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
mune response will be stimulated in the patient’s body resulting in a therapeutic effect. The requisite mechanism of action can be described in terms of the classic ‘two-signal hypothesis’: Signal one is composed of the MHC – antigen complex and signal two is composed of co-stimulatory signals. For antigen-loaded DC products, the ideal potency test would evaluate the level of target antigen-specific T-cell-stimulatory activity of the DCs in vitro. This could comprise a bioassay measuring activation of the patient’s own T-cells by antigenloaded autologous DCs. This approach is fraught with deficiencies: (U.S. Code of Federal Regulations) as such, it would have to be assumed that the patient’s cells would respond under the experimental conditions being employed, (Monji et al., 2000) the detection of an antigen-specific T-cell response is near-impossible using primary T-cells (peripheral blood derived T-cells), warranting the in vitro production of antigen-specific secondary or tertiary cell lines that could give a detectable response in such an assay, (Steinman and Witmer, 1978) false positive or negative T-cell responses could result from the in vitro manipulations, (Van Voorhis et al., 1983) the strategy of making secondary or tertiary cell lines consumes too many of the patient’s valuable peripheral blood cells, and is time-consuming (at least 1 –2 months) and expensive, without any assurance that the assay would eventually succeed. Therefore, the quandary is that the mechanism of action, antigen-specific T-cell stimulation, cannot be tested for autologous DC-based vaccines. To circumvent this, the allostimulatory capacity of DCs has been generally accepted as a potency determinant and tested via the mixed lymphocyte reaction (MLR) assay. The MLR has served as the ‘gold standard’ for testing the functional ability of antigen presenting cells (Steinman and Witmer, 1978; Van Voorhis et al., 1983). However, MLR assays take several days to complete and may be, quite arguably, non-representative of the activity of dendritic cells in context of their stimulatory interactions with autologous antigen-specific T-cells. MHC antigens constitute a powerful signal in the MLR; the co-stimulation of alloreactive T-cells by DCs in the MLR amplifies the extent of histoincompatibility between the DCs and T-cells, making it difficult to independently assess the impact of the co-stimulatory capacity of the DCs versus the selective presentation of strong alloanti-
gens. From the quality control perspective, the MLR is also a rather uncontrolled system since the stimulating alloantigens will vary vastly between the batches of T-cells (responder cells) and batches of DCs (stimulator cells). For example, when low MLR data is obtained for a sample, it is unclear whether this was a result of truly impotent DCs, or due to a significant histocompatibility match between the DCs and T-cells. Alternatively, immunophenotyping by flow cytometry has been used as a measure of DC potency (Monji et al., 2000), but is obviously inaccurate because it does not measure cellular function. A highly pure but non-viable or apoptotic (and thus obviously impotent) DC preparation would erroneously pass such a potency test. The ‘COSTIM bioassay’ described here selectively measures co-stimulatory activity but not antigen processing and presentation. We believe that this bioassay is more relevant to the function of DCs in a vaccine product, is more accurate, reliable, and efficient than currently used DC potency tests, and is also suitable for lot-release quality control testing.
2. Time required
It is strongly advised that a large T-cell batch be prepared in advance and convenient aliquots cryopreserved. It is impractical to make individual T-cell preparations for each run of the bioassay as it adds several hours and is also a source of variation in the assay. The time for purifying T-cells is thus not specified here. Similarly, the time for preparing DCs is also not specified here since the methods of manufacture and the times vary significantly.
The COSTIM assay set-up time is V 3 h, involving:
Thawing T-cells and DCs, washing, counting, and re-suspension: 1 – 2 h Culture set-up: V 1 h
The total culture period is 44 h, involving:
Initial culture period, prior to the addition of tritiated thymidine: 26 h Preparing working stock of tritiated thymidine and addition to culture wells: V 10 min
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
Final culture period: 18 h Harvesting cells and scintillation counting: V 1 h, but times may vary based on the equipment used. Data manipulation and results analysis: V 15 min.
295
4. Detailed procedure
3. Materials 3.1. Cells and reagents
Dendritic cells Allogeneic T-cells AIM-Vk Culture Medium (Life Technologies, Rockville, MD) Anti-Human CD3 Antibody, 1 mg/ml, Azidefree/low endotoxin (BD Pharmingen, San Diego, CA) Tritiated (3H) thymidine, 1 mCi/ml, 74.0 GBq/ mmol (Perkin Elmer Life Sciences, Boston, MA)
3.2. Supplies
Ice Hemacytometer Trypan blue (Sigma, St. Louis, MO) 96-well U-bottom plates (VWR International, West Chester, PA) 15-ml centrifuge tubes with screw caps (VWR International) 70% Isopropanol
3.3. Equipment
Sterile, laminar flow, biological safety cabinet Incubator, 37 F 2 jC, 5% CO2, humidifying ( z 70% relative humidity) Microscope for cell counting Centrifuge Water bath, 37 F 2 jC Pipettor Single-channel micropipettors, 0.5 –10, 20 – 200, and 100– 1000 Al Multi(12)-channel micropipettor, 20 – 200 Al FilterMatek 96-channel culture plate harvester (Packard Instruments, Meriden, CT), or equivalent TopCountk NXT scintillation counter (Packard Instruments), or equivalent
4.1. Preparation of cells (i) It is advised that one or more T-cell batches be prepared well in advance and convenient aliquots cryopreserved. T-cells may be enriched or purified by any one of several methods. We have employed biomagnetic separation using antiHLA-DR monoclonal antibody-conjugated paramagnetic beads (Dynal, Lake Success, NY) to deplete accessory cells such as monocytes, DCs, and B-cells, applying the manufacturer’s recommended protocol. This typically resulted in a suspension comprising of 80 –90% T-cells (the remainder being NK cells, which were found to be inert in this assay). The cells are cryopreserved in 90% autologous serum and 10% dimethyl sulfoxide (DMSO) until further use. Irrespective of the T-cell enrichment method used, the purity of the preparation(s) should be determined by flow cytometry, and their general functional ability qualified by PHA-induced mitogenesis or MLR prior to use in the COSTIM bioassay. Similarly, the preparation of DCs is beyond the scope of this paper since the methods of manufacture vary significantly. (ii) Thaw the cryopreserved T-cells and DCs in a 37 F 2 jC water bath. Immerse the cryovials in a water bath just long enough to thaw completely, but no longer than 2 –3 min to ensure optimal post-thaw viability. (iii) Immediately after thawing, resuspend T-cells and DCs in separate 15-ml centrifuge tubes containing 10 – 12 ml of warm AIM-V medium (add the cells to the media slowly to ensure good viability). (iv) Centrifuge the tubes at 400 g, at 1– 8 jC, for 10 F 2 min to sediment the cells. (v) After centrifugation, decant or aspirate supernatant completely, and resuspend the cells in AIM-V medium, and perform cell counting and viability determination by Trypan blue exclusion using the hemacytometer and microscope. Alternatively, automated cell counting/viability may be performed. (vi) Using AIM-V medium, resuspend the DCs at 1 105 viable cells/ml, and T-cells at 1 106
296
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
viable cells/ml. Split each cell suspension by half into separate tubes, and store on ice. Label the four individual tubes as follows: (a) T-cells (b) DCs (c) T-cells with anti-CD3 (d) DCs with anti-CD3 (vii) Prepare a working solution of anti-CD3 monoclonal antibody: Add 2.0 Al of 1.0 mg/ ml anti-CD3 antibody [stock] into 1 ml of AIM-V medium, resulting in a 2.0 Ag/ml working solution (be cautious when dispensing the 2.0 Al of anti-CD3; immerse only the tip of the pipette tip into the meniscus of the media to minimize dilution errors). (viii) To tubes labeled ‘‘with anti-CD3’’, add 5 Al of anti-CD3 antibody working solution per milliliter of cell suspension, resulting in a 0.01 Ag/ml antibody concentration within those cell suspensions.
(ii) Dispense 50 Al of the above working solution per culture well (i.e., 0.5 ACi per well). (iii) Incubate the plate for 18 h in a 37 F 2 jC, 5% CO2, humidifying incubator. 4.4. Harvest and scintillation counting (i) Using the 96-channel culture plate harvester, collect the cells onto a fiberglass-backed filter paper. In this process, wash the plate wells, with filter in place, five times with distilled water and subsequently once with 70% isopropanol. This lyses the cells, retaining cellular debris and nucleic acids on the filter. (ii) Completely dry the filter by placing it for 20 min under a heat lamp (alternatively, the filter may be air-dried at room temperature, but this can take several hours). (iii) Following the scintillation counter manufacturer’s instructions, determine the counts per minute (cpm) of the incorporated radioactivity in the individual culture wells.
4.2. Culture set-up 4.5. Data manipulation and results analysis (i) In a 96-well culture plate, co-culture the following three combinations of DCs and T-cells in triplicate wells, in a total volume of 200 Al per well: (a) 100 Al of T-cells + 100 Al of DCs (Background MLR control group) (b) 100 Al of T-cells + 100 Al of DCs with anti-CD3 (COSTIM group) (c) 100 Al of T-cells with anti-CD3 + 100 Al of AIMV medium (T-cell background proliferation control)
(i) Calculate the average cpm result of triplicate wells. The scintillation counter is programmable to perform this calculation. The average cpm value is sufficient for data analysis. (ii) Assure that the background T-cell proliferation control (T-cells with anti-CD3 only) does not proliferate over 2000 cpm. This system suitability
As a result, there will be 1 105 T-cells and/or 1 104 DCs, with or without 0.005 Ag/ml antiCD3 antibody in the wells. (ii) Incubate the plate for 26 h in a 37 F 2 jC, 5% CO2, humidifying incubator. 4.3. Addition of tritiated (3H) thymidine to culture wells (i) A few minutes prior to the end of the 26h incubation, prepare a 10 ACi/ml working solution of tritiated (3H) thymidine in AIM-V medium.
Fig. 1. Typical results of the COSTIM assay are shown. Due to the short (44 h) culture in this assay, the alloantigenic MLR response is insignificant.
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
297
Fig. 2. Different numbers of dendritic cells, monocytes, and B-cells (stimulator cells) were plated in this assay with a constant number (1 105 per well) of responder cells T-cells. The COSTIM assay delineated the magnitude of difference in the co-stimulatory ability of accessory cells.
Fig. 3. (A) Three lots of dendritic cells were placed in culture with T-cells and anti-CD3 antibody for 44 h (COSTIM assay) or with T-cells alone for 6 days (MLR assay). Results of the two assays are shown. (B) Six different lots of T-cells were tested against three lots of dendritic cells in the COSTIM assay to gain an estimate of method robustness.
298
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
control does not exceed 2000 cpm if the T-cell batch was previously qualified properly (see Section 4.1, step i). Proliferation higher than 2000 cpm by the T-cells alone cultured with anti-CD3 indicates that there is accessory cell activity present in the T-cell preparation, and that the COSTIM assay results may not be reliable. (iii) Calculate DC potency by using the average cpm and the following equation: Potency (cpm) = COSTIM cpm (T-cells + DCs + anti-CD3) Background MLR cpm(T-cells + DCs).
5. Results The COSTIM assay largely precludes alloantigenic MLR responses (Fig. 1) because of the short (44 h) culture period. This method for potency determination is not limited to DCs; other accessory cells such as monocytes (Fig. 2) may also be tested, possibly with minor adjustments to the assay. For instance, the results in Fig. 2 show that a greater number of monocytes per well (compared to DCs) is needed for a detectable proliferative response by the 1 105 T-cells in the well. We routinely use this test for DC potency, employing a ratio of 1:10 for DCs to T-cells. To demonstrate the advantage of the COSTIM bioassay over the MLR bioassay, the experiment depicted in Fig. 3A was performed. Three lots of DCs (from three separate PBMC donors) were prepared under identical manufacturing conditions and tested by the COSTIM method versus the traditional 6-daylong MLR assay. As shown, the inter-lot variation for DCs was much lower for COSTIM (CV, 10.8%; S.D., 5225 cpm) versus the MLR (CV, 28.7%; S.D., 29721 cpm). Thus, the COSTIM assay is a more robust method than the MLR. The major variable that we anticipated to have a profound effect on assay performance was the responder T-cell preparations. To understand the variation between T-cell lots, the performance of 6 different allogeneic T-cell lots (all prepared identically) in the COSTIM bioassay was compared against three lots of DCs (Fig. 3B). The inter-lot variation for DCs in this experiment was also low (CV, 9.0%; S.D., 5036), whereas the inter-lot variation for T-cells was much higher (CV, 30.6%; S.D., 18618). These results indicate that different Tcell lots vary widely in their proliferative abilities but
within a given T-cell lot assay variation is markedly low.
6. Discussion This protocol describes a simple method for testing the potency of accessory cells such as DCs. Unlike the MLR, it effectively separates co-stimulatory activity from antigen processing and presentation. The response of T-cells in this assay is almost completely dependent upon the co-stimulatory molecules CD54, CD80, and CD86 on the surface of DCs (not shown). The COSTIM bioassay is more functionally relevant to DCs, is more accurate, robust, and efficient than the MLR, and is also well suited for lot-release quality control testing of DC-based vaccines. For a ‘naked’ DC product (i.e., when antigen is not loaded into DCs), the COSTIM assay alone should suffice as a lot-release test. Perhaps an antigen-uptake activity assay could be used in conjunction with it. In the case of antigen-loaded DC products, a test that determines the presence of loaded antigen in the cells may be combined with the COSTIM test for lot-release purposes. We have used the latter approach successfully for our product, DCVaxk-Prostate (prostate specific membrane antigen-loaded DCs), with acceptance by our product reviewers at the U.S. Food and Drug Administration (FDA). 6.1. Troubleshooting We have found that most assay failures relate to the cellular concentration and viability of the cells. Since the level of proliferative response in this bioassay is directly proportional to the numbers and viability of Tcells and DCs in culture, an inadvertent miscalculation of cell concentration, or estimation of Trypan blue excluding cells (viable cells) by an inexperienced analyst can lead to great variation in assay performance. It is critical that the viability be determined accurately. The use of an automated viability counter leads to improved precision in our experience. We also recommend that a well-characterized lot of DCs be utilized as a positive control or reference lot that is included in every run. The qualification of a responder T-cell lot must be performed carefully, and specifications for the acceptance of a T-cell lot could be set in
G. Shankar et al. / Journal of Immunological Methods 285 (2004) 293–299
place to reduce variations in assay performance. This qualification should involve purity, determined by flow cytometry, and general T-cell functional ability qualified by mitogenesis (MLR or PHA-induced) prior to use in the COSTIM bioassay. By monitoring the performance our system suitability controls, we have also found that the stability of tritiated thymidine is generally unpredictable after 3 months, leading to lower-than-normal results for the positive control and false-negative sample results. It is recommended that a lot of anti-CD3 antibody be frozen in appropriate aliquots to preserve stability. 6.2. Alternative and support protocols The COSTIM assay can be performed with either autologous or allogeneic T-cells resulting in nearly identical data. We have confirmed this with experiments comparing the proliferative responses of autologous T-cells with allogeneic T-cells and have found that they were statistically indifferent (not shown). The DCs, as described in this protocol, need not be irradiated since DC do not proliferate. We have confirmed this using irradiated (20 – 30 Gy) and non-irradiated DCs and found that they were statistically indifferent (not shown). Instead of measuring T-cell proliferation, activation of T-cells by the DCs may also be evaluated by immunophenotypic measurement of CD69 and CD25 on the T-cell surface. Since these markers may be evaluated within hours of activation, this approach can offer quicker analysis of potency for products with short expiration dating. Yet, another alternative approach (depending on the targeted mechanism of action of the DC vaccine) is to collect the supernatant culture medium from COSTIM cultures and determine the level of a T-cell cytokine. We have evaluated the level of IFN-g in our cultures, which correlated extremely well with the proliferation data. However, the proliferation read-out for this assay is logistically the simplest, most efficient, and economical. 7. Quick procedure (i) Prepare and qualify responder T-cells in advance. (ii) Thaw the cryopreserved T-cells and DCs, wash, and count.
299
(iii) Resuspend DCs at 1 105 viable cells/ml, and T-cells at 1 106 viable cells/ml. (iv) To one half of each cell suspension, add antiCD3 monoclonal antibody to a final concentration of 0.01 Ag/ml. (v) In a 96-well culture plate, co-culture the following three combinations of DCs and Tcells in triplicate wells, in a total volume of 200 Al per well: (a) 100 Al of T-cells + 100 Al of DCs (Background MLR control group) (b) 100 Al of T-cells + 100 Al of DCs with anti-CD3 (COSTIM group) (c) 100 Al of T-cells with anti-CD3 + 100 Al of AIMV medium (T-cell background proliferation control) (vi) Incubate the plate for 26 h in a 37 jC F 2 jC, 5% CO2, humidifying incubator. (vii) To the culture wells, add 50 Al of 10 ACi/ml of tritiated (3H) thymidine. (viii) Incubate the plate for 18 h in a 37 F 2 jC, 5% CO2, humidifying incubator. (ix) Harvest the cells onto a filter and measure the incorporated radioactivity by scintillation counting. (x) Using the average result of the triplicate wells, calculate DC potency by using the average cpm and the following equation: Potency (cpm) = COSTIM cpm (T-cells + DCs + anti-CD3) Background MLR cpm(T-cells + DCs).
References U.S. Code of Federal Regulations. 21 CFR 600.3(s). U.S. Government Printing Office, Washington, DC, USA. Monji, T., Smits-Sana, B., Qian, L., Stark, L., Therond, J., Shiomoto, C., Peshwa, M.V., 2000. Development of assay for assessment of dendritic cell product potency. International Society for Hematotherapy and Graft Engineering (ISHAGE) 2000, San Diego, CA, Abstract 139. Steinman, R.M., Witmer, M.D., 1978. Lymphoid dendritic cells are potent stimulators of the primary mixed leukocyte reaction in mice. Proc. Natl. Acad. Sci. U. S. A. 75, 5132 – 5136. Van Voorhis, W.C., Valinsky, J., Hoffman, E., Luban, J., Hair, L.S., Steinman, R.M., 1983. Relative efficacy of human monocytes and dendritic cells as accessory cells for T cell replication. J. Exp. Med. 158, 174 – 191.