ORIGINAL ARTICLE
BIOPRESERVATION AND BIOBANKING Volume 13, Number X, 2015 ª Mary Ann Liebert, Inc. DOI: 10.1089/bio.2014.0054
Method Validation for Automated Isolation of Viable Peripheral Blood Mononuclear Cells Gael Hamot, Wim Ammerlaan, Conny Mathay, Olga Kofanova, and Fay Betsou
Background: This article is part of a series of publications providing formal method validation for biospecimen processing in the context of accreditation in laboratories and biobanks. We report the optimization and validation for fitness-for-purpose of automated and manual protocols for isolating peripheral blood mononuclear cells (PBMCs) from whole blood, and compare the two methods. Methods: The manual method was optimized for whole blood centrifugation speed, gradient type (Ficoll, Leucosep, CPT), and freezing method (Mr Frosty, Controlled Rate Freezing). Various parameters of the automated protocol using a CPT gradient on a Tecan liquid handler were optimized. Optimal protocols were validated in parallel for reproducibility and robustness. Optimization and validation were assessed in terms of cell yield, viability, recovery, white blood cell (WBC) subpopulation distribution, gene expression, and lymphoblastoid cell line (LCL) transformation. Results: An initial centrifugation of whole blood at 2000 g was considered optimal for further processing, allowing isolation of plasma and PBMCs from a single sample. The three gradients gave similar outcomes in terms of cell yield, viability, and WBC subpopulation distribution. Ficoll showed some advantages and was selected for further evaluations. Optimization of the automated protocol script using a CPT gradient gave 61% cell recovery. No significant differences in quality, quantity, and WBC subpopulation distribution were seen between the two freezing methods, and Mr. Frosty was selected. The manual and automated protocols were reproducible in terms of quantity, recovery, viability, WBC subpopulation distribution, gene expression, and LCL transformation. Most (75%–100%) of the 13 robustness parameters were accepted for both methods with an 8 h pre-centrifugation delay versus 38%–85% after 24 h. Differences identified between the automated and manual methods were not considered consequential. Conclusions: We validated the first fully automated method for isolating viable PBMCs, including RNA analysis and generation of LCLs. We recommend processing within 8 h of blood collection.
lations are increasingly being used for identification of clinically relevant biomarkers. Examples include lymphocyte subset-specific gene expression signatures in cancer6 or autoimmune diseases,7 lymphocyte subset-specific miRNA signatures in multiple sclerosis,8 or T cell subset-specific flow cytometric signatures in Parkinson’s disease.9 Viable PBMCs are sorted for specific subpopulations, via immunomagnetic sorting to obtain purified monocyte and lymphocyte populations,10 and are used for functional studies,11 immunophenotyping,12 establishment of lymphoblastoid cell lines (LCL),13 and purification of CD34 + cells.14 Functional assays for monitoring immune responses are often used in vaccine development. Different white blood cell (WBC) types can be identified on the basis of their expression of CD antigens. CD45 is the most widely expressed antigen and is found on all WBCs. CD3, CD14, and CD15 are characteristic of T lymphocytes,
Introduction
W
e present here the fourth in a series of articles on the optimization and validation of biospecimen processing methods for downstream applications.1–3 For biobanks, this is an important aspect of quality assurance, particularly in the context of automated handling, and is a normative requirement for biobank accreditation.4,5 Optimization and validation are based on fitness-for-purpose, collection and processing efficiency, reproducibility, and robustness. Peripheral blood mononuclear cells (PBMCs), blood cells with a rounded nucleus, include a diverse population of cells participating in the body’s immune defense, notably lymphocytes (T cells, B cells, and natural killer cells), monocytes, and dendritic cells. PBMCs can be used for a broad range of downstream applications. Sorted PBMC subpopuIntegrated BioBank of Luxemburg (IBBL), Luxembourg.
1
2
monocytes, and granulocytes, respectively. WBC types have specific gene expression signatures15 and methylation profiles,16 and the use of specific subpopulations may increase resolution in the quest for identification of clinicallyrelevant biomarkers. Transformation of peripheral B lymphocytes by Epstein Barr virus (EBV) is the method of choice for generating LCLs.17 LCLs serve a variety of purposes including antibody generation,18 antigen presentation for immunologic assays and cancer vaccines,19,20 and also express genes from a wide range of metabolic pathways specific to individuals from whom the cells are derived, making them an important source of material for genetic and functional research.21 In the context of downstream applications, high cell yields, subpopulation composition, and cell viability are the most relevant quality attributes. Best Practices22 and Standard Operating Procedures22–24 for processing biobank samples have been published, but formal method validation is lacking. It has been shown that anticoagulant type, time from venipuncture to PBMC isolation and cryopreservation, and isolation method may impact the performance of T cells in downstream immunological assays.25,26 The gold standard of PBMC isolation is density gradient centrifugation, allowing separation of PBMCs from granulocytes, platelets, and red blood cells (RBCs). A number of separation media are available, of which Ficoll, a neutral, high molecular weight, hydrophilic polysaccharide, is the most widely used. Upon centrifugation, PBMCs remain at the less dense, upper interface. Separation can be done manually or using prefilled tubes, such as vacuum-driven BD Vacutainer Cell Preparation Tubes (CPT) or Greiner Bio-One Leucosep tubes, in which blood is separated from the density medium with a porous membrane. The current study was designed to identify the optimal protocol for extracting PBMCs from peripheral blood in collection tubes supplemented with anticoagulant. Processing was optimized in terms of initial whole blood centrifugation conditions, choice of gradient, and PBMC freezing method. In light of the labor-intensive nature of gradient separation (a major source of inter-operator variability) alongside the advantages of automating biobanking sample processing (allowing high throughput and reducing processing time), we established an automated workflow, and compared automated and manual methods. Optimal protocols were validated with respect to fitness-for-purpose for downstream applications, based on reproducibility and robustness for pre-centrifugation delay. Parameters evaluated included cell counts, viability, and recovery, WBC subpopulation distribution, mRNA and miRNA relative quantity, and the establishment of LCLs.
Materials and Methods Study design The study was performed in two steps. In the initial optimization step, four parameters were evaluated; 1) centrifugation speed of whole blood (1000 g, 2000 g, 4000 g) prior to gradient separation, evaluated in terms of cell viability and WBC subpopulation distribution; 2) gradient type (Ficoll, prefilled Leucosep tubes, prefilled CPT tubes), evaluated in terms of PBMC yield (total counts), viability, and WBC subpopulation distribution; 3) freezing method (Controlled Rate
HAMOT ET AL.
Freezer [CRF], Mr Frosty [MRF]) evaluated in terms of cell yield (total counts), viability, WBC subpopulation distribution, and establishment of LCLs; and 4) automated method approach to streamline the procedure and improve operational efficiency and reproducibility, evaluated in terms of recovery. Evaluation of centrifugation, gradient, and automated versus manual optimization parameters was performed in freshly extracted PBMCs (without a freeze-thaw cycle). Optimized automated and manual protocols were then validated in parallel. Reproducibility and robustness (4 hours vs. 8 and 24 hours pre-centrifugation delay) were evaluated according to yield (total counts and CBC), viability, recovery, WBC subpopulation distribution, gene expression, and establishment of LCLs. The reproducibility acceptance criterion was a CV < 20% between samples from a given donor for all assays. The robustness acceptance criteria were assay outcomes at 8 and 24 h within 15% of the 4-h outcome, or that all samples used to establish lymphoblastoid cell lines had the same transformability. An ongoing stability study with annual re-testing to evaluate the effect of long-term storage on fitness-forpurpose is planned. Only baseline results are included in this report (comparison of automated vs. manual processing). Validation assays (reproducibility and robustness) were performed after PBMC isolation and cryopreservation, with the exception of CBC counts that were performed in whole blood and in freshly isolated PBMCs, prior to freezing. Two vials were frozen; one was used for cell counts, viability, and WBC subpopulation analyses, and the other for EBV transformation and RNA extraction. Finally, variation between the automated and manual extraction methods was determined according to fitnessfor-purpose in terms of PBMC yield (evaluated by CBC), recovery, viability, WBC subpopulation distribution, gene expression analysis, and establishment of LCLs. Peripheral blood samples were collected from healthy volunteers who provided written informed consent (CNER approval ##201107/02). An overview of the workflow is shown in Figure 1.
PBMC isolation protocol Blood collection. For centrifugation optimization, three 10mL blood samples were collected from one donor in K2E Vacutainers containing K2EDTA (BD #367525). For gradient optimization, three donors each provided two blood samples; 10 mL in a K2E Vacutainer containing EDTA (BD #367525) for Ficoll and Leucosep, and 4 mL in a 4-mL CPT Vacutainer containing sodium citrate (BD #362760). For freezing optimization, PBMCs isolated from the gradient optimization were used. For automation optimization, one donor provided four blood samples in 8-mL CPT Vacutainer tubes (BD #362761). For reproducibility and stability assessments, one donor provided three 10-mL blood samples in K2E Vacutainers (diluted 1:1 with PBS), and three 8-mL samples in CPT Vacutainers for each experiment. For robustness, three donors each provided three 10-mL blood samples in K2E Vacutainers (diluted 1:1 with PBS) and three 8-mL samples in CPT Vacutainers, collected within 30 min of each other. Variation between the automated and manual extraction methods was evaluated using results from evaluations for the reproducibility, robustness and stability.
VALIDATION OF PBMC ISOLATION
3
FIG. 1. Overview of peripheral blood mononuclear cell (PBMC) optimization and validation.
All samples (except those collected for pre-centrifugation delay) were processed within 4 h of collection and maintained at room temperature (18–24C). Blood processing and gradient separation. For centrifugation optimization, tubes were centrifuged at 1000 g, 2000 g, or 4000 g for 20 min, medium brake. Plasma was removed and the pellet was resuspended in an equal volume of PBS, loaded on a Ficoll-Paque PLUS gradient (density 1.077 – 0.001; General Electric; #17-1440-02) in a 15-mL tube (2:1), then centrifuged for 30 min at 400 g, soft brake. The interphase was collected and transferred to a 15-mL tube containing 10 mL PBS then centrifuged at 200 g for 10 min, soft break. The pellet was suspended in 1 mL PBS and divided between two tubes. For gradient optimization, three gradient systems were compared: 1) Ficoll; 8 mL diluted blood (4 mL whole blood, collected in K2E Vacutainer, plus 4 mL PBS) was loaded on a Ficoll-Paque PLUS gradient, centrifuged as described above, and the interphase was transferred to a 15-mL tube. 2) Leucosep; 8 mL diluted blood (4 mL whole blood, collected in K2E Vaculatainer, plus 4 mL PBS) was added to a Leucosep tube (prefilled with Leucosep separation medium; Greiner Bio-one; #163288) and centrifuged for 10 min at 1000 g, medium brake. The top layer was removed to 5 mm above the interphase, then all liquid down to the porous barrier was transferred to a 15-mL tube. 3) CPT; 4-mL blood sample collected directly in a 4-mL tube was inverted 10 times then centrifuged for 20 min at 1800 g, medium brake. Approximately half the plasma was aspirated and the
cell layer was collected and transferred to a 15-mL tube. 15mL tubes from the three methods were topped up to 10 mL with PBS and centrifuged for 10 min, 200 g, soft brake. Supernatant was removed and pellets were resuspended in 5 mL PBS. For reproducibility, robustness and stability, samples were manually processed using the Ficoll gradient. Cell pellets were slowly suspended in pre-chilled Cryostor CS10 cryopreservation medium (BioLife solutions; #210102) according to the manufacturer’s instructions at a concentration of 5–10 x 106 cells/mL and frozen in 1-mL aliquots. Automated PBMC extraction protocol. An automated protocol for PBMC and plasma extraction was implemented with a Tecan Freedom EVO automated liquid handler with EVOware standard software (v2.5 Service Pack 3 patch 1) using a Ficoll gradient in CPT tubes and gradient centrifugation for 20 min at 1800 g, high brake. The in-house protocol was optimized to improve recovery and reproducibility in terms of; 1) detection of the liquid height in the tube after the first centrifugation, 2) calculation of the separation gel height based on the whole volume detection, 3) aspiration of a maximum volume of the PBMC layer without touching the gel, 4) resuspension of the cell pellet after centrifugation in 11 mL PBS, and 5) aspiration of the maximal amount of resuspended pellet to a cryotube. PBMC recovery was evaluated as the total number of PBMCs in various fractions obtained during processing, compared to the total number in the volume loaded. Acceptance was based on a 50% PBMC recovery rate in the final cryotube.
4
The optimized script was used for reproducibility, robustness, and baseline stability experiments,. Pellets were slowly resuspended in pre-chilled 1.5 mL Cryostor CS10 in FluidX cryotubes (#65-7502) according to the manufacturer’s instructions then frozen. Freeze-thawing. For optimization, PBMCs pellets were resuspended in 6 mL freezing medium (RPMI1640 [Invitrogen, #72400-021] supplemented with 20% FBS HI [Invitrogen #26140] and 10% DMSO [Sigma; #D2438]), and aliquoted into six cryovials [FluidX; #65-7502] 1 mL/ vial. Two freezing methods were evaluated using three vials per method. 1) Controlled Rate Freezer (CRF); vials were placed in an IceCube 14S Controlled Rate Freezer (Sy-Lab) using IceCube Software v1.21a, running a program of 1C/ min cooling until - 80C, then quickly transferred to liquid nitrogen; 2) Mr. Frosty (MRF); vials were placed in a Mr. Frosty Freezing Container (Nalgene; #5100-0001) filled with isopropanol, cooled slowly overnight to - 80C, then transferred to liquid nitrogen. Vials were thawed rapidly (2 min at 37C) 6 months later. For validation of reproducibility, robustness and baseline stability, the Mr. Frosty method was used. PBMCs were frozen in cryovials (two aliquots per sample for reproducibility and robustness, 11 aliquots for stability) with 1 mL Cryostor CS10 per vial, placed in a Mr. Frosty Freezing Container filled with isopropanol precooled at 4C, left for 4 h at - 80C, then transferred to the liquid nitrogen vapor phase. Vials were thawed rapidly (2 min at 37C), then washed with 9 mL RPMI1640 supplemented with 10% FBS. Cells were centrifuged at 300 g for 5 min at room temperature, supernatant was removed and the pellet was resuspended in 1 mL RPMI-10 for evaluation assays.
Evaluation assays Complete blood counts (CBCs). A 100-mL sample of whole blood was counted on an ABX Micros CRP 200 hematology counter (Horiba), according to the manufacturer’s instructions. Total PBMC was calculated by multiplying the sum of lymphocytes and monocytes per mL times the blood volume used. PBMC cell count by CASY. PBMCs were quantified on a CASY Cell Counter and Analyzer (Roche Applied Science) using an in-house PBMC counting protocol (unpublished validation data). Counted samples were diluted to be within the reliable counting range. For reproducibility, recovery was evaluated by 1) comparing the total PBMC counts (CASY) after isolation relative to total PBMCs (ABX lymphocyte plus monocyte counts) in whole blood, and 2) comparing total PBMC counts (CASY) before and after freezing. Total PBMC yield was calculated by counted events/mL, times the dilution, times the total volume. Cell viability. Cell viability was assessed by flow cytometry analysis using the FITC Annexin V Apoptosis Detection Kit 1 (BD Pharmingen; #556547) according to the manufacturer’s instructions on a BD FACSVerse with FACSuite v 1.0.4.2650 software. In this article, single Annexin V-stained cells are reported as ‘‘early apoptotic’’, and double Annexin V / propidium iodide-stained cells are reported as ‘‘late apoptotic’’ cells. WBC subpopulation distribution. WBC distribution was assessed by flow cytometry using multiplex marking of
HAMOT ET AL.
PBMCs with a panel of fluorescently labeled monoclonal antibodies (CD45*FITC, BD #555482; CD3*PerCP, Miltenyi #130-094-965; CD14*PE, BD #345785; CD15* PECy7, BD #560827). Stained samples were analyzed on a BD FACSVerse using FACSuite v 1.0.4.2650 software. Establishment of lymphoblastoid cell lines. Approximately 2– 5 · 106 PBMCs were exposed to EBV (supernatant from B95-8 cell line, ATCC) for 60 to 90 min at 37C. Cells were then washed and resuspended in RPMI-20 supplemented with 1 mg/mL cyclosporine A (Sigma #C1832). Cells (*2 x 106/mL) were plated in 96-well plates and cultured at 37C, 5% CO2, in a humidified environment. Transformation was evaluated after 3 weeks in terms of clump formation score. RNA extraction and spectrophotometry. Total RNA, including miRNA, was extracted from 2–5 · 106 PBMCs using the miRNeasy mini kit (QIAgen #217004) on a QIAcube then quantified by spectrophotometry on a Synergy MX (Biotek). To be in the range for downstream PCR applications, eluted RNA samples were concentrated using a CentriVap (Labconco), resuspended in 15 mL PCR grade water and quantified again. mRNA and miRNA qRT-PCR. mRNA (hDUSP1, hJUN) and miRNA (RNU24, miR16) were quantified using commercial TaqMan assays (Life Technologies) according to the manufacturer’s instructions using commercial probes(Life Technologies). qPCR reactions were run on the ABI 7500 Fast Real Time PCR System (Life Technologies) and results were scored in Cts.
Statistical analyses Mean, standard deviation (SD) and CV% were calculated using Microsoft Excel 2010. Optimization outcomes and variability between automated and manual methods were compared using paired two-tailed t-tests. Significance was calculated using Sigma Plot v.12.0 (Systat Software) with a 5% significance threshold using Tukey and Dunn tests.
Results Optimization of PBMC isolation protocols Centrifugation conditions. Initially, processing of whole blood was optimized in terms of centrifugation speed to determine the feasibility of simultaneously isolating plasma and PBMCs from a single sample, based on our previously validated method for plasma isolation.1 The majority of PBMCs isolated from whole blood were viable, and similar proportions were isolated with all three speeds following Ficoll gradient separation; 91.7% at 1000 g, 88.1% at 2000 g, and 94.0% at 4000 g (CV 3.3%). Late apoptotic cells accounted for £ 10% of isolated PBMCs, and early and necrotic PBMCs for < 2%. The proportions of CD45 + , CD3 + , and CD14 + cells were similar with each of the three speeds (CV 1.1%, 5.2% and 6.6%, respectively; Supplementary Table S1; supplementary material is available in the online article at www.liebertpub.com/bio). A high 32.2% CV seen for CD15 + cells was probably due to the low proportions of these cells detected ( < 3% of cells). Gradient separation. Three gradient separation methods were compared in terms of cell yield, viability and WBC subpopulation distribution (Table 1).Cell counts varied between the three methods, notably for two of the three donors. The Ficoll method consistently gave the highest cell
VALIDATION OF PBMC ISOLATION
5
Table 1. Comparison of Manual Gradient Separation Methods of Peripheral Blood Mononuclear Cells (PBMCs) in Terms of Yield, Viability, and White Blood Cell (WBC) Composition Ficoll (EDTA)
Leucosep (EDTA)
CPT (citrate)
Meana (SD)
CV%
7.05 9.60 10.15
6.71 8.94 8.40
6.12 6.37 5.54
6.6 (0.47) 8.3 (1.7) 8.0 (2.3)
7.1% 20.6% 29.0%
PBMC counts (CASY) Donor 1 Total PBMCs ( · 106)b Donor 2 Total PBMCs ( · 106)b Donor 3 Total PBMCs ( · 106)b Viability (FACSVerse) Donor 1 Viable cells (%) Donor 2 Viable cells (%) Donor 3 Viable cells (%) WBC composition Donor 1 CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) Donor 2 CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) Donor 3 CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) a
92.9 92.4 94.2
92.1 92.7 92.4
90.7 80.8 85.8
91.9 (1.1) 88.6 (6.8) 90.8 (4.4)
93.8 63.7 15.7 1.0 97.1 62.3 15.9 1.3 91.9 59.6 12.1 3.9
99.2 63.8 16.9 0.5 98.4 65.3 16.3 0.3 91.7 64.8 11.9 0.4
98.8 62.8 17.4 0.6 99.4 63.0 16.5 0.4 88.9 66.0 11.9 0.5
97.3 63.4 16.7 0.7 98.3 63.5 16.2 0.7 90.8 63.4 11.9 1.6
(3.0) (0.6) (0.9) (0.3) (1.2) (1.6) (0.3) (0.5) (1.6) (3.4) (0.1) (2.0)
1.2% 7.7% 4.9% 3.1% 0.9% 5.1% 41.2% 1.2% 2.5% 1.9% 77.8% 1.8% 5.4% 1.0% 123.4%
Mean of the three methods per donor; bTotal PBMCs = Counts for 1 mL · total volume (5 mL).
yields, followed by Leucosep which gave from 5% to 17% fewer cells, while the CPT method gave from 13% to 50% fewer cells than the Ficoll method. In terms of cell viability, CPT gave 9% to 13% fewer viable cells than either the Ficoll or Leucosep method for two donors and was approximately equivalent for the third, with CV% ranging from 1.2% to 7.7%. Very little difference was seen between the three methods in terms of WBC subpopulation distribution, with < 5.5% CV for leukocytes (CD45 + ), lymphocytes (CD3 + ) and monocytes (CD14 + ). PBMCs isolated with Leucosep and CPT had lower granulocyte contamination (CD15 + ) compared to Ficoll. The CV% was much higher with CD15 + , likely due to the very small percent positive cells ( < 4%). Ficoll was selected as the gradient for manual PBMC isolation for subsequent reproducibility and robustness evaluations.
Automation. CPT, which is compatible for use with the Tecan automated liquid handling system, was used for gradient separation. Recovery of total PBMCs was evaluated in various fractions obtained with the automated method and compared with whole blood counts for two successive optimization protocols (Table 2). After the first (non-optimized) script, a 15% yield was obtained in the final aliquots. The greatest sources of cell loss were the volume left above the gel after PBMC aspiration (to avoid touching interphase) and the volume left in the mixing tube after aspiration. A first script optimization was performed taking these aspects into account, which resulted in a 42% yield; however, this was still below the required 50% recovery threshold. Further modifications were implemented (second optimization) to improve the yield, moving the liquid aspiration to a lower level and using a larger volume
Table 2. Optimization of Automated Peripheral Blood Mononuclear Cell (PBMC) Isolation According to Two Scripts (Tecan Liquid Handler) in Terms of PBMC Yield Initial scripta
Fraction Whole bloodd Volume above gel after PBMC aspiration Washing gel with 1 ml PBS Volume left in mixing tube after aspiration Final cryotube aliquot
Optimization 1b
Optimization 2c
Volume (mL)
Total PBMCs (106)
% total PBMC
Volume (mL)
Total PBMCs (106)
% total PBMC
Volume (mL)
Total PBMCs (106)
% total PBMC
9 0.7
18 4.62
100% 26%
9 0.35
18 2.17
100% 12%
9 nd
11.2 nd
100% nd
1
1.4
8%
1
2.3
13%
nd
nd
nd
0.3
4.23
24%
0
0
0
0
0
0%
1.5
2.7
15%
1.6
7.52
42%
1.6
6.75
60.9%
a Initial (non-optimized) protocol script; bFirst revised protocol script; cSecond revised protocol script; dBlood diluted with 2.5 mL citrate; nd, not determined.
6
HAMOT ET AL.
pre-aspiration mix. This final revised script gave a 61% yield and was used for validation. Freeze-thawing. The effect of two freezing methods, the Controlled Rate Freezer and Mr. Frosty, on cell yield, viability, WBC subpopulation distribution, and biological activity was compared for three donors. No significant differences between the two methods were seen for total PBMC or WBC subpopulation distribution for any of the three donors (Table 3). Some fluctuations were apparent in subpopulation distribution, with Mr. Frosty favoring lymphocytes and Controlled Rate Freezer favoring monocytes (data not shown). A significant difference between the two methods in terms of the number of viable cells was identified in one of the three donors, however overall, post-thaw viability was high (approximately 80%). WBC subpopulation distribution was not significantly different between the two freezing methods for any of the donors. Finally, PBMCs from all three donors were 100% compatible with EBV transformation for both freezing methods. The Mr. Frosty method was applied for subsequent reproducibility and robustness evaluations.
parameters with a CV > 20% (RBC and platelet counts, and PBMC pre-freeze/thaw recovery). For each processing method (manual and automatic), both mRNA (hDUSP1 and hJUN) and miRNA (RNU24 and miR16) C + values showed good reproducibility with CVs < 20% (CV ranged from 0.97% to 7.77%). Both methods were reproducible for the functional assay of LCL transformation, with 100% transformation in all cases. Overall, both methods were reproducible according to the acceptance criterion of CV < 20%.
Reproducibility of manual and automated PBMC isolation protocols
Variation between automated and manual extraction methods
The manual Ficoll (EDTA) and the automated CPT (citrate) protocols were validated for reproducibility in terms of cell quantity and quality, functionality, and gene expression, using an acceptance criterion of a CV < 20% between three samples from the same donor (Table 4). With the exception of CD15 + , all parameters assayed in PBMCs extracted using the manual Ficoll method were within the 20% acceptance criterion, with most parameters having a CV of approximately 5% or less. CD15 + is a granulocyte marker which should not be present in the PBMC interphase. Increased CD15 + cells indicate granulocyte contamination in the Ficoll layer. CV% variability was greater for PBMCs isolated using the Tecan automated liquid handler, with several parameters having a CV > 15% including three
Variation between the CPT automated and the EDTA/ Ficoll manual methods was determined for each of the nine samples collected per donor for reproducibility, 4 h robustness and baseline stability (three replicates each), and evaluated according to the two isolation methods. Initially, a comparison of CBC counts in whole blood collected in citric acid CPT Vacutainers and EDTA K2E Vacutainers was performed. Blood samples collected in the automated CPT tubes had approximately 17% fewer WBCs and platelets compared to Ficoll tubes, while lymphocyte, monocyte, granulocyte, and RBC counts were approximately the same with both collection tubes (Table 6). Although PBMC recovery in freshly isolated samples compared to whole blood was higher with the manual
Robustness of manual and automated PBMC collection protocols Cell quality, quantity, and functional parameters were evaluated for robustness by comparing an 8 h and a 24 h precentrifugation delay to a 4 h delay. Most parameters were accepted (i.e., were within 15% of the values at 4 h) for all donors with both methods for the 8h pre-centrifugation delay (Table 5). However when the pre-centrifugation delay was increased to 24 h, acceptance varied between donors and methods, ranging from 38% to 85%.
Table 3. Optimization of Peripheral Blood Mononuclear Cell (PBMC) Freezing Methods in Terms of Yield, Viability, and White Blood Cell (WBC) Subpopulation Distribution Donor 1a Mean (SD) MRF
P valueb
2.04 (0.10)
0.238
(2.45) 83.16 (0.72)
0.128
(0.12) 99.10 (0.10) (2.48) 52.67 (0.32) (0.84) 22.60 (0.46) (0.35) 0.47 (0.06)
0.635 0.216 0.093 0.336
CRF PBMC counts (CASY) Total PBMCs 1.86 ( · 106)c Viability (FACSVerse) Healthy 78.57 cells (%) WBC Composition CD45 + (%) 99.03 CD3 + (%) 50.17 CD14 + (%) 24.87 CD15 + (%) 0.70
Donor 2a
(0.26)
Mean (SD)
Donor 3a Mean (SD)
MRF
P valueb
CRF
2.52 (0.05)
0.542
2.23 (0.05)
83.18 (0.42) 86.19 (1.02)
0.036
77.88 (4.73) 82.51 (1.33) 0.242
0.423 0.086 0.286 0.370
99.37 49.53 19.83 2.20
CRF 2.58 (0.15)
99.63 51.923 23.73 0.60
(0.06) 99.53 (0.12) (0.92) 54.33 (0.40) (1.29) 22.40 (0.66) (0.10) 0.77 (0.21)
MRF
P value
2.18 (0.16) 0.614
(0.25) 99.20 (0.10) 0.199 (4.17) 51.87 (1.36) 0.540 (1.10) 19.93 0.64 0.930 (0.80) 2.10 (0.36) 0.845
CRF, Controlled Rate Freezer; MRF, Mr Frosty. Bold, p < 0.05. a For each donor, cells were pooled from the three manual gradient methods; bTwo-tailed paired t-test; cTotal PBMCs = Counts for 1 mL · total volume (1 mL); single reading.
VALIDATION OF PBMC ISOLATION
7
Table 4. Reproducibility of Peripheral Blood Mononuclear Cells (PBMCs) According to Acceptance Criteria for Cell Yield, Viability, White Blood Cell (WBC) Subpopulation Composition, and Ribonucleic Acid (RNA) Targeted Gene Expression Automated (CPT) Mean (SD)a Whole Blood CBC counts
Freshly isolated PBMCs Cell counts (CASY)b Frozen isolated PBMCs Cell counts (CASY)c Cell viability, FACSVerse
WBC distribution
RT-qPCR
LCL transformation
WBC (106/mL) Lymphocytes (106/ml) Lymphocytes (%) Monocytes (106/mL) Monocytes (%) Granulocytes (106/ml) Granulocytes (%) RBC (109/mL) Platelets (106/mL) Total cells (106/mL) PBMC Recoveryb (%) Total cells (106/mL) PBMC Recoveryc (%) Necrosis (%) Late apoptosis (%) Healthy cells (%) Early apoptosis (%) CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct) 100%
5.2 1.2 23.5 0.2 4.9 3.8 71.6 5.1 157.3
Acceptance criteria CV (%)
(0.36) (0.17) (2.27) (0) (0.36) (0.2) (2.58) (1.05) (53.05)
6.9 14.4 9.6 0 7.4 5.3 3.6 20.5 33.7
4.71 (0.94) 60.9 (15.41) 3.30 83.9 1.09 13.17 75.03 10.71 98.6 63.8 14.7 1.6 20.02 26.34 23.63 20.96
(0.59) (3.4) (0.22) (1.50) (1.15) (0.29) (0.35) (1.45) (1.46) (0.26) (1.56) (1.30) (0.23) (0.25) +
Manual (Ficoll) Mean (SD)a 6.3 1.43 24.0 0.3 5.4 4.6 70.6 5.3 192.7
Acceptance criteria CV (%)
(0.1) (0.06) (0.62) (0) (0.23) (0.06) (0.8) (0.12) (9.87)
1.6 4.0 2.6 0 4.3 1.3 1.1 2.3 5.1
19.9 25.3
1.05 (0.08) 72.1 (6.01)
7.2 8.3
18.0 4.1 19.7 11.4 1.5 2.7 0.4 2.3 10.0 15.6 7.8 5.0 1.0 1.2 +
3.90 (0.23) 64.8 (6.3) 0.86 (0.13) 14.22 (0.35) 78.71 (0.55) 6.21 (0.28) 99.2 (0.09) 58.8 (0.58) 17.2 (0.47) 2.40 (1.36) 18.48 (0.88) 24.14 (0.69) 23.54 (0.63) 20.20 (1.09) +
5.8 9.7 15.0 2.5 0.7 4.5 0.1 1.0 2.7 57.4 4.8 2.9 2.7 5.4 +
Bold indicates assays for which CV% was > 20%. a Mean of three samples from one donor; bTotal PBMC counts (CASY) after isolation relative to total PBMCs (ABX lymphocyte plus monocyte counts) in whole blood; cTotal PBMC counts (CASY) after freezing relative to before.
method compared to the automated method (69.7% vs. 60.9%; Table 7), this finding was not statistically significant and was balanced out by the freeze/thaw cycle, with the automated method giving higher recovery after a passage in liquid nitrogen (87.4% vs. 65.4%). Significantly higher viability was found with the manual extraction after one freeze-thaw cycle (79.9% versus 74.0%; p = 0.01). All data taken together, the final post-thaw yield in terms of recovery of viable PBMCs is 36.4% for the manual method and 39.4% for the automated method. Significantly different proportions of both CD3 + and CD14 + populations were present in PBMCs isolated by the automated method compared to the manual method ( p = 0.03 and 0.02, respectively); however, the CV% of each method were within the reproducibility acceptance criterion. From identical RNA starting quantities, a statistically significant difference was seen in hJUN mRNA quantification, with the automated method yielding significantly less hJUN than the manual method ( p = 0.015) but not in hDUSP, while no statistically significant differences were found for miRNA target quantification. This variation may be due to hJUN expression being sensitive to the citrate anticoagulant in the CPT samples. There was no difference between the
two extraction methods in terms of transformability of LCLs.
Discussion The value of automated isolation of PBMCs over manual Ficoll separation was first reported for cell therapy purposes in 2008,27 confirming significantly higher recovery of clinical grade PBMCs using a Sepax system, while also eliminating inter-user differences. Here we report for the first time automated isolation of PBMCs for research purposes using a current widely available liquid handler. The initial optimization step demonstrated the feasibility of simultaneously obtaining plasma and PBMCs from a single whole blood sample with Ficoll manual separation in terms of viable cells and WBC subpopulation distribution. The 2000 g speed was selected as optimal on the basis of a previous report for validation of plasma isolation,1 and is the recommended centrifugation speed for obtaining both plasma and viable PBMCs suitable for downstream research applications from a single blood sample. Of the three manual gradient methods, Ficoll and Leucosep gave higher PBMC yields than CPT. Besides different
8
HAMOT ET AL.
Table 5. Robustness of Peripheral Blood Mononuclear Cell (PBMC) Protocol with 8 H and 24 H Pre-centrifugation Delay, According to Cell Quantity, Quality, and Biological Function, Post Freeze-Thaw Donor 1
8 hour delay CBC counts
Cell viability (FACSVerse) WBC distribution RT-qPCR
Donor 2
Donor 3
Automated (CPT)
Manual (Ficoll)
Automated (CPT)
Manual (Ficoll)
Automated (CPT)
Manual (Ficoll)
acc rej acc acc acc acc rej acc acc acc acc acc acc 85
acc acc acc acc acc acc acc rej acc acc acc acc acc 92
rej acc acc acc acc acc rej acc acc acc acc acc acc 85
rej rej acc rej acc acc acc acc acc acc acc acc acc 77
acc acc acc acc acc acc acc rej acc acc acc acc acc 92
acc acc acc acc acc acc acc acc acc acc acc acc acc 100
rej rej acc rej rej acc rej rej acc acc acc acc acc 54
acc rej acc acc rej acc rej rej acc acc acc acc acc 69
rej rej rej rej rej rej rej rej acc acc acc acc acc 38
rej rej acc rej rej rej rej rej acc acc acc acc acc 46
acc acc acc rej acc acc rej acc acc acc acc acc acc 85
acc acc acc rej rej acc rej rej acc acc acc acc acc 69
Total cells (/ml) WBC (106/ml) Lymphocytes (%) Monocytes (%) Viable cells (%) CD3 + (%) CD14 + (%) CD15 + (%) hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct)
LCL transformation 8 hour % accepted parameters 24 hour delay CBC counts ABX) Total cells (/ml) WBC (106/ml) Lymphocytes (%) Monocytes (%) Cell viability (FACSVerse) Viable cells (%) WBC distribution CD3 + (%) CD14 + (%) CD15 + (%) RT-qPCR hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct) LCL transformation 24 hour %accepted parameters
Acc, < 15% of 4 h value; Rej, ‡ 15% variation from 4 h value.
anticoagulant and processing methods, this may be attributed to a potentially smaller starting volume of blood due to CPT tubes allowing a maximum 4-mL sample, leading to potentially smaller collected volumes, whereas the other two methods guaranteed a 4-mL starting volume. Both Ficoll
Table 6. Comparison of Complete Blood Cell (CBC) Counts in Isolated Peripheral Blood Mononuclear Cell (PBMC) with Automated Extraction (Citric Acid) Versus Manual Extraction (EDTA) Ratio of Automated/ Manual (%) Mean (SD)a CBC counts White blood cell (106/mL) Lymphocytes (%) Monocytes (%) Granulocytes (%) Red blood cell (109/mL) Platelets (106/mL)
82.6 (8.5) 99.1 96.1 101.2 96.3 83.2
(11.1) (13.4) (6.4) (17.1) (23.2)
a Mean (SD) percent ratio comparing automated outcome versus manual outcome for each parameter from nine samples (reproducibility, robustness and stability-related samples) for each isolation method.
and Leucosep outperformed CPT in terms of pre-freeze PBMC viability. Nonetheless, the CV across the three gradients was less than 20% for viability and all WBC subpopulations other than granulocytes. CPT and Leucosep both gave less granulocyte contamination than Ficoll. Published data comparing gradient methods are conflicting, with Schlenke et al. reporting higher PBMC yields with CPT compared to Ficoll but lower viability,28 while Nilsson et al. showed the inverse in one institution and equivalence in another institution.29 Ruitenberg also reported equivalence in terms of viability and recovery.30 In our study, with the exception of yield, the choice of gradient did not appear to influence the outcome, and all three separation methods were considered fit-for-purpose given the high yields. With freezing PBMCs for long-term storage being the main purpose of this procedure, from a fit-for-purpose perspective, both the CPT and Ficoll methods are appropriate. Ficoll was selected for further evaluations of the manual method, as the most widely implemented method and providing the highest yields. Although mainly used post-cryopreservation, it has been reported that viability of Ficoll-derived PBMCs from cryopreserved samples is significantly lower than in freshly isolated samples.30 Nonetheless, the significance of this difference appears to be of little consequence in our experimental
VALIDATION OF PBMC ISOLATION
9
Table 7. Variation Between Automated and Manual Peripheral Blood Mononuclear Cell (PBMC) Isolation Methods, According to Cell Quantity, Immunophenotype, Gene Expression, and Lymphoblastoid Cell Line (LCL)-Compatibility Parameters Automated (Citrate) a
Freshly isolated PBMCs Cell counts (CASY) Frozen isolated PBMCs Cell counts (CASY) Cell viability (FACSVerse) White blood cell distribution RT-qPCR
LCL transformation
Manual (EDTA/Ficoll)
Mean (SD)
CV (%)
Mean (SD)a
CV (%)
P value
PBMC Recoveryb (%)
60.9 (13.5)
22.3
69.7 (5.7)
8.2
0.21
PBMC Recoveryc (%) Healthy cells (%) CD3 + (%) CD14 + (%) CD15 + (%) hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct) 100%
87.4 74.0 60.8 15.3 1.4 19.9 25.8 20.8 23.5
9.0 4.6 8.2 9.2 43.8 6.7 5.6 5.2 1.7
65.4 79.9 57.9 17.7 1.3 18.8 24.1 20.6 23.6
22.6 3.0 10.4 12.1 94.9 5.1 3.1 4.3 1.2
BIOPRESERVATION AND BIOBANKING Volume 13, Number X, 2015 ª Mary Ann Liebert, Inc. DOI: 10.1089/bio.2014.0054
Method Validation for Automated Isolation of Viable Peripheral Blood Mononuclear Cells Gael Hamot, Wim Ammerlaan, Conny Mathay, Olga Kofanova, and Fay Betsou
Background: This article is part of a series of publications providing formal method validation for biospecimen processing in the context of accreditation in laboratories and biobanks. We report the optimization and validation for fitness-for-purpose of automated and manual protocols for isolating peripheral blood mononuclear cells (PBMCs) from whole blood, and compare the two methods. Methods: The manual method was optimized for whole blood centrifugation speed, gradient type (Ficoll, Leucosep, CPT), and freezing method (Mr Frosty, Controlled Rate Freezing). Various parameters of the automated protocol using a CPT gradient on a Tecan liquid handler were optimized. Optimal protocols were validated in parallel for reproducibility and robustness. Optimization and validation were assessed in terms of cell yield, viability, recovery, white blood cell (WBC) subpopulation distribution, gene expression, and lymphoblastoid cell line (LCL) transformation. Results: An initial centrifugation of whole blood at 2000 g was considered optimal for further processing, allowing isolation of plasma and PBMCs from a single sample. The three gradients gave similar outcomes in terms of cell yield, viability, and WBC subpopulation distribution. Ficoll showed some advantages and was selected for further evaluations. Optimization of the automated protocol script using a CPT gradient gave 61% cell recovery. No significant differences in quality, quantity, and WBC subpopulation distribution were seen between the two freezing methods, and Mr. Frosty was selected. The manual and automated protocols were reproducible in terms of quantity, recovery, viability, WBC subpopulation distribution, gene expression, and LCL transformation. Most (75%–100%) of the 13 robustness parameters were accepted for both methods with an 8 h pre-centrifugation delay versus 38%–85% after 24 h. Differences identified between the automated and manual methods were not considered consequential. Conclusions: We validated the first fully automated method for isolating viable PBMCs, including RNA analysis and generation of LCLs. We recommend processing within 8 h of blood collection.
lations are increasingly being used for identification of clinically relevant biomarkers. Examples include lymphocyte subset-specific gene expression signatures in cancer6 or autoimmune diseases,7 lymphocyte subset-specific miRNA signatures in multiple sclerosis,8 or T cell subset-specific flow cytometric signatures in Parkinson’s disease.9 Viable PBMCs are sorted for specific subpopulations, via immunomagnetic sorting to obtain purified monocyte and lymphocyte populations,10 and are used for functional studies,11 immunophenotyping,12 establishment of lymphoblastoid cell lines (LCL),13 and purification of CD34 + cells.14 Functional assays for monitoring immune responses are often used in vaccine development. Different white blood cell (WBC) types can be identified on the basis of their expression of CD antigens. CD45 is the most widely expressed antigen and is found on all WBCs. CD3, CD14, and CD15 are characteristic of T lymphocytes,
Introduction
W
e present here the fourth in a series of articles on the optimization and validation of biospecimen processing methods for downstream applications.1–3 For biobanks, this is an important aspect of quality assurance, particularly in the context of automated handling, and is a normative requirement for biobank accreditation.4,5 Optimization and validation are based on fitness-for-purpose, collection and processing efficiency, reproducibility, and robustness. Peripheral blood mononuclear cells (PBMCs), blood cells with a rounded nucleus, include a diverse population of cells participating in the body’s immune defense, notably lymphocytes (T cells, B cells, and natural killer cells), monocytes, and dendritic cells. PBMCs can be used for a broad range of downstream applications. Sorted PBMC subpopuIntegrated BioBank of Luxemburg (IBBL), Luxembourg.
1
2
monocytes, and granulocytes, respectively. WBC types have specific gene expression signatures15 and methylation profiles,16 and the use of specific subpopulations may increase resolution in the quest for identification of clinicallyrelevant biomarkers. Transformation of peripheral B lymphocytes by Epstein Barr virus (EBV) is the method of choice for generating LCLs.17 LCLs serve a variety of purposes including antibody generation,18 antigen presentation for immunologic assays and cancer vaccines,19,20 and also express genes from a wide range of metabolic pathways specific to individuals from whom the cells are derived, making them an important source of material for genetic and functional research.21 In the context of downstream applications, high cell yields, subpopulation composition, and cell viability are the most relevant quality attributes. Best Practices22 and Standard Operating Procedures22–24 for processing biobank samples have been published, but formal method validation is lacking. It has been shown that anticoagulant type, time from venipuncture to PBMC isolation and cryopreservation, and isolation method may impact the performance of T cells in downstream immunological assays.25,26 The gold standard of PBMC isolation is density gradient centrifugation, allowing separation of PBMCs from granulocytes, platelets, and red blood cells (RBCs). A number of separation media are available, of which Ficoll, a neutral, high molecular weight, hydrophilic polysaccharide, is the most widely used. Upon centrifugation, PBMCs remain at the less dense, upper interface. Separation can be done manually or using prefilled tubes, such as vacuum-driven BD Vacutainer Cell Preparation Tubes (CPT) or Greiner Bio-One Leucosep tubes, in which blood is separated from the density medium with a porous membrane. The current study was designed to identify the optimal protocol for extracting PBMCs from peripheral blood in collection tubes supplemented with anticoagulant. Processing was optimized in terms of initial whole blood centrifugation conditions, choice of gradient, and PBMC freezing method. In light of the labor-intensive nature of gradient separation (a major source of inter-operator variability) alongside the advantages of automating biobanking sample processing (allowing high throughput and reducing processing time), we established an automated workflow, and compared automated and manual methods. Optimal protocols were validated with respect to fitness-for-purpose for downstream applications, based on reproducibility and robustness for pre-centrifugation delay. Parameters evaluated included cell counts, viability, and recovery, WBC subpopulation distribution, mRNA and miRNA relative quantity, and the establishment of LCLs.
Materials and Methods Study design The study was performed in two steps. In the initial optimization step, four parameters were evaluated; 1) centrifugation speed of whole blood (1000 g, 2000 g, 4000 g) prior to gradient separation, evaluated in terms of cell viability and WBC subpopulation distribution; 2) gradient type (Ficoll, prefilled Leucosep tubes, prefilled CPT tubes), evaluated in terms of PBMC yield (total counts), viability, and WBC subpopulation distribution; 3) freezing method (Controlled Rate
HAMOT ET AL.
Freezer [CRF], Mr Frosty [MRF]) evaluated in terms of cell yield (total counts), viability, WBC subpopulation distribution, and establishment of LCLs; and 4) automated method approach to streamline the procedure and improve operational efficiency and reproducibility, evaluated in terms of recovery. Evaluation of centrifugation, gradient, and automated versus manual optimization parameters was performed in freshly extracted PBMCs (without a freeze-thaw cycle). Optimized automated and manual protocols were then validated in parallel. Reproducibility and robustness (4 hours vs. 8 and 24 hours pre-centrifugation delay) were evaluated according to yield (total counts and CBC), viability, recovery, WBC subpopulation distribution, gene expression, and establishment of LCLs. The reproducibility acceptance criterion was a CV < 20% between samples from a given donor for all assays. The robustness acceptance criteria were assay outcomes at 8 and 24 h within 15% of the 4-h outcome, or that all samples used to establish lymphoblastoid cell lines had the same transformability. An ongoing stability study with annual re-testing to evaluate the effect of long-term storage on fitness-forpurpose is planned. Only baseline results are included in this report (comparison of automated vs. manual processing). Validation assays (reproducibility and robustness) were performed after PBMC isolation and cryopreservation, with the exception of CBC counts that were performed in whole blood and in freshly isolated PBMCs, prior to freezing. Two vials were frozen; one was used for cell counts, viability, and WBC subpopulation analyses, and the other for EBV transformation and RNA extraction. Finally, variation between the automated and manual extraction methods was determined according to fitnessfor-purpose in terms of PBMC yield (evaluated by CBC), recovery, viability, WBC subpopulation distribution, gene expression analysis, and establishment of LCLs. Peripheral blood samples were collected from healthy volunteers who provided written informed consent (CNER approval ##201107/02). An overview of the workflow is shown in Figure 1.
PBMC isolation protocol Blood collection. For centrifugation optimization, three 10mL blood samples were collected from one donor in K2E Vacutainers containing K2EDTA (BD #367525). For gradient optimization, three donors each provided two blood samples; 10 mL in a K2E Vacutainer containing EDTA (BD #367525) for Ficoll and Leucosep, and 4 mL in a 4-mL CPT Vacutainer containing sodium citrate (BD #362760). For freezing optimization, PBMCs isolated from the gradient optimization were used. For automation optimization, one donor provided four blood samples in 8-mL CPT Vacutainer tubes (BD #362761). For reproducibility and stability assessments, one donor provided three 10-mL blood samples in K2E Vacutainers (diluted 1:1 with PBS), and three 8-mL samples in CPT Vacutainers for each experiment. For robustness, three donors each provided three 10-mL blood samples in K2E Vacutainers (diluted 1:1 with PBS) and three 8-mL samples in CPT Vacutainers, collected within 30 min of each other. Variation between the automated and manual extraction methods was evaluated using results from evaluations for the reproducibility, robustness and stability.
VALIDATION OF PBMC ISOLATION
3
FIG. 1. Overview of peripheral blood mononuclear cell (PBMC) optimization and validation.
All samples (except those collected for pre-centrifugation delay) were processed within 4 h of collection and maintained at room temperature (18–24C). Blood processing and gradient separation. For centrifugation optimization, tubes were centrifuged at 1000 g, 2000 g, or 4000 g for 20 min, medium brake. Plasma was removed and the pellet was resuspended in an equal volume of PBS, loaded on a Ficoll-Paque PLUS gradient (density 1.077 – 0.001; General Electric; #17-1440-02) in a 15-mL tube (2:1), then centrifuged for 30 min at 400 g, soft brake. The interphase was collected and transferred to a 15-mL tube containing 10 mL PBS then centrifuged at 200 g for 10 min, soft break. The pellet was suspended in 1 mL PBS and divided between two tubes. For gradient optimization, three gradient systems were compared: 1) Ficoll; 8 mL diluted blood (4 mL whole blood, collected in K2E Vacutainer, plus 4 mL PBS) was loaded on a Ficoll-Paque PLUS gradient, centrifuged as described above, and the interphase was transferred to a 15-mL tube. 2) Leucosep; 8 mL diluted blood (4 mL whole blood, collected in K2E Vaculatainer, plus 4 mL PBS) was added to a Leucosep tube (prefilled with Leucosep separation medium; Greiner Bio-one; #163288) and centrifuged for 10 min at 1000 g, medium brake. The top layer was removed to 5 mm above the interphase, then all liquid down to the porous barrier was transferred to a 15-mL tube. 3) CPT; 4-mL blood sample collected directly in a 4-mL tube was inverted 10 times then centrifuged for 20 min at 1800 g, medium brake. Approximately half the plasma was aspirated and the
cell layer was collected and transferred to a 15-mL tube. 15mL tubes from the three methods were topped up to 10 mL with PBS and centrifuged for 10 min, 200 g, soft brake. Supernatant was removed and pellets were resuspended in 5 mL PBS. For reproducibility, robustness and stability, samples were manually processed using the Ficoll gradient. Cell pellets were slowly suspended in pre-chilled Cryostor CS10 cryopreservation medium (BioLife solutions; #210102) according to the manufacturer’s instructions at a concentration of 5–10 x 106 cells/mL and frozen in 1-mL aliquots. Automated PBMC extraction protocol. An automated protocol for PBMC and plasma extraction was implemented with a Tecan Freedom EVO automated liquid handler with EVOware standard software (v2.5 Service Pack 3 patch 1) using a Ficoll gradient in CPT tubes and gradient centrifugation for 20 min at 1800 g, high brake. The in-house protocol was optimized to improve recovery and reproducibility in terms of; 1) detection of the liquid height in the tube after the first centrifugation, 2) calculation of the separation gel height based on the whole volume detection, 3) aspiration of a maximum volume of the PBMC layer without touching the gel, 4) resuspension of the cell pellet after centrifugation in 11 mL PBS, and 5) aspiration of the maximal amount of resuspended pellet to a cryotube. PBMC recovery was evaluated as the total number of PBMCs in various fractions obtained during processing, compared to the total number in the volume loaded. Acceptance was based on a 50% PBMC recovery rate in the final cryotube.
4
The optimized script was used for reproducibility, robustness, and baseline stability experiments,. Pellets were slowly resuspended in pre-chilled 1.5 mL Cryostor CS10 in FluidX cryotubes (#65-7502) according to the manufacturer’s instructions then frozen. Freeze-thawing. For optimization, PBMCs pellets were resuspended in 6 mL freezing medium (RPMI1640 [Invitrogen, #72400-021] supplemented with 20% FBS HI [Invitrogen #26140] and 10% DMSO [Sigma; #D2438]), and aliquoted into six cryovials [FluidX; #65-7502] 1 mL/ vial. Two freezing methods were evaluated using three vials per method. 1) Controlled Rate Freezer (CRF); vials were placed in an IceCube 14S Controlled Rate Freezer (Sy-Lab) using IceCube Software v1.21a, running a program of 1C/ min cooling until - 80C, then quickly transferred to liquid nitrogen; 2) Mr. Frosty (MRF); vials were placed in a Mr. Frosty Freezing Container (Nalgene; #5100-0001) filled with isopropanol, cooled slowly overnight to - 80C, then transferred to liquid nitrogen. Vials were thawed rapidly (2 min at 37C) 6 months later. For validation of reproducibility, robustness and baseline stability, the Mr. Frosty method was used. PBMCs were frozen in cryovials (two aliquots per sample for reproducibility and robustness, 11 aliquots for stability) with 1 mL Cryostor CS10 per vial, placed in a Mr. Frosty Freezing Container filled with isopropanol precooled at 4C, left for 4 h at - 80C, then transferred to the liquid nitrogen vapor phase. Vials were thawed rapidly (2 min at 37C), then washed with 9 mL RPMI1640 supplemented with 10% FBS. Cells were centrifuged at 300 g for 5 min at room temperature, supernatant was removed and the pellet was resuspended in 1 mL RPMI-10 for evaluation assays.
Evaluation assays Complete blood counts (CBCs). A 100-mL sample of whole blood was counted on an ABX Micros CRP 200 hematology counter (Horiba), according to the manufacturer’s instructions. Total PBMC was calculated by multiplying the sum of lymphocytes and monocytes per mL times the blood volume used. PBMC cell count by CASY. PBMCs were quantified on a CASY Cell Counter and Analyzer (Roche Applied Science) using an in-house PBMC counting protocol (unpublished validation data). Counted samples were diluted to be within the reliable counting range. For reproducibility, recovery was evaluated by 1) comparing the total PBMC counts (CASY) after isolation relative to total PBMCs (ABX lymphocyte plus monocyte counts) in whole blood, and 2) comparing total PBMC counts (CASY) before and after freezing. Total PBMC yield was calculated by counted events/mL, times the dilution, times the total volume. Cell viability. Cell viability was assessed by flow cytometry analysis using the FITC Annexin V Apoptosis Detection Kit 1 (BD Pharmingen; #556547) according to the manufacturer’s instructions on a BD FACSVerse with FACSuite v 1.0.4.2650 software. In this article, single Annexin V-stained cells are reported as ‘‘early apoptotic’’, and double Annexin V / propidium iodide-stained cells are reported as ‘‘late apoptotic’’ cells. WBC subpopulation distribution. WBC distribution was assessed by flow cytometry using multiplex marking of
HAMOT ET AL.
PBMCs with a panel of fluorescently labeled monoclonal antibodies (CD45*FITC, BD #555482; CD3*PerCP, Miltenyi #130-094-965; CD14*PE, BD #345785; CD15* PECy7, BD #560827). Stained samples were analyzed on a BD FACSVerse using FACSuite v 1.0.4.2650 software. Establishment of lymphoblastoid cell lines. Approximately 2– 5 · 106 PBMCs were exposed to EBV (supernatant from B95-8 cell line, ATCC) for 60 to 90 min at 37C. Cells were then washed and resuspended in RPMI-20 supplemented with 1 mg/mL cyclosporine A (Sigma #C1832). Cells (*2 x 106/mL) were plated in 96-well plates and cultured at 37C, 5% CO2, in a humidified environment. Transformation was evaluated after 3 weeks in terms of clump formation score. RNA extraction and spectrophotometry. Total RNA, including miRNA, was extracted from 2–5 · 106 PBMCs using the miRNeasy mini kit (QIAgen #217004) on a QIAcube then quantified by spectrophotometry on a Synergy MX (Biotek). To be in the range for downstream PCR applications, eluted RNA samples were concentrated using a CentriVap (Labconco), resuspended in 15 mL PCR grade water and quantified again. mRNA and miRNA qRT-PCR. mRNA (hDUSP1, hJUN) and miRNA (RNU24, miR16) were quantified using commercial TaqMan assays (Life Technologies) according to the manufacturer’s instructions using commercial probes(Life Technologies). qPCR reactions were run on the ABI 7500 Fast Real Time PCR System (Life Technologies) and results were scored in Cts.
Statistical analyses Mean, standard deviation (SD) and CV% were calculated using Microsoft Excel 2010. Optimization outcomes and variability between automated and manual methods were compared using paired two-tailed t-tests. Significance was calculated using Sigma Plot v.12.0 (Systat Software) with a 5% significance threshold using Tukey and Dunn tests.
Results Optimization of PBMC isolation protocols Centrifugation conditions. Initially, processing of whole blood was optimized in terms of centrifugation speed to determine the feasibility of simultaneously isolating plasma and PBMCs from a single sample, based on our previously validated method for plasma isolation.1 The majority of PBMCs isolated from whole blood were viable, and similar proportions were isolated with all three speeds following Ficoll gradient separation; 91.7% at 1000 g, 88.1% at 2000 g, and 94.0% at 4000 g (CV 3.3%). Late apoptotic cells accounted for £ 10% of isolated PBMCs, and early and necrotic PBMCs for < 2%. The proportions of CD45 + , CD3 + , and CD14 + cells were similar with each of the three speeds (CV 1.1%, 5.2% and 6.6%, respectively; Supplementary Table S1; supplementary material is available in the online article at www.liebertpub.com/bio). A high 32.2% CV seen for CD15 + cells was probably due to the low proportions of these cells detected ( < 3% of cells). Gradient separation. Three gradient separation methods were compared in terms of cell yield, viability and WBC subpopulation distribution (Table 1).Cell counts varied between the three methods, notably for two of the three donors. The Ficoll method consistently gave the highest cell
VALIDATION OF PBMC ISOLATION
5
Table 1. Comparison of Manual Gradient Separation Methods of Peripheral Blood Mononuclear Cells (PBMCs) in Terms of Yield, Viability, and White Blood Cell (WBC) Composition Ficoll (EDTA)
Leucosep (EDTA)
CPT (citrate)
Meana (SD)
CV%
7.05 9.60 10.15
6.71 8.94 8.40
6.12 6.37 5.54
6.6 (0.47) 8.3 (1.7) 8.0 (2.3)
7.1% 20.6% 29.0%
PBMC counts (CASY) Donor 1 Total PBMCs ( · 106)b Donor 2 Total PBMCs ( · 106)b Donor 3 Total PBMCs ( · 106)b Viability (FACSVerse) Donor 1 Viable cells (%) Donor 2 Viable cells (%) Donor 3 Viable cells (%) WBC composition Donor 1 CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) Donor 2 CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) Donor 3 CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) a
92.9 92.4 94.2
92.1 92.7 92.4
90.7 80.8 85.8
91.9 (1.1) 88.6 (6.8) 90.8 (4.4)
93.8 63.7 15.7 1.0 97.1 62.3 15.9 1.3 91.9 59.6 12.1 3.9
99.2 63.8 16.9 0.5 98.4 65.3 16.3 0.3 91.7 64.8 11.9 0.4
98.8 62.8 17.4 0.6 99.4 63.0 16.5 0.4 88.9 66.0 11.9 0.5
97.3 63.4 16.7 0.7 98.3 63.5 16.2 0.7 90.8 63.4 11.9 1.6
(3.0) (0.6) (0.9) (0.3) (1.2) (1.6) (0.3) (0.5) (1.6) (3.4) (0.1) (2.0)
1.2% 7.7% 4.9% 3.1% 0.9% 5.1% 41.2% 1.2% 2.5% 1.9% 77.8% 1.8% 5.4% 1.0% 123.4%
Mean of the three methods per donor; bTotal PBMCs = Counts for 1 mL · total volume (5 mL).
yields, followed by Leucosep which gave from 5% to 17% fewer cells, while the CPT method gave from 13% to 50% fewer cells than the Ficoll method. In terms of cell viability, CPT gave 9% to 13% fewer viable cells than either the Ficoll or Leucosep method for two donors and was approximately equivalent for the third, with CV% ranging from 1.2% to 7.7%. Very little difference was seen between the three methods in terms of WBC subpopulation distribution, with < 5.5% CV for leukocytes (CD45 + ), lymphocytes (CD3 + ) and monocytes (CD14 + ). PBMCs isolated with Leucosep and CPT had lower granulocyte contamination (CD15 + ) compared to Ficoll. The CV% was much higher with CD15 + , likely due to the very small percent positive cells ( < 4%). Ficoll was selected as the gradient for manual PBMC isolation for subsequent reproducibility and robustness evaluations.
Automation. CPT, which is compatible for use with the Tecan automated liquid handling system, was used for gradient separation. Recovery of total PBMCs was evaluated in various fractions obtained with the automated method and compared with whole blood counts for two successive optimization protocols (Table 2). After the first (non-optimized) script, a 15% yield was obtained in the final aliquots. The greatest sources of cell loss were the volume left above the gel after PBMC aspiration (to avoid touching interphase) and the volume left in the mixing tube after aspiration. A first script optimization was performed taking these aspects into account, which resulted in a 42% yield; however, this was still below the required 50% recovery threshold. Further modifications were implemented (second optimization) to improve the yield, moving the liquid aspiration to a lower level and using a larger volume
Table 2. Optimization of Automated Peripheral Blood Mononuclear Cell (PBMC) Isolation According to Two Scripts (Tecan Liquid Handler) in Terms of PBMC Yield Initial scripta
Fraction Whole bloodd Volume above gel after PBMC aspiration Washing gel with 1 ml PBS Volume left in mixing tube after aspiration Final cryotube aliquot
Optimization 1b
Optimization 2c
Volume (mL)
Total PBMCs (106)
% total PBMC
Volume (mL)
Total PBMCs (106)
% total PBMC
Volume (mL)
Total PBMCs (106)
% total PBMC
9 0.7
18 4.62
100% 26%
9 0.35
18 2.17
100% 12%
9 nd
11.2 nd
100% nd
1
1.4
8%
1
2.3
13%
nd
nd
nd
0.3
4.23
24%
0
0
0
0
0
0%
1.5
2.7
15%
1.6
7.52
42%
1.6
6.75
60.9%
a Initial (non-optimized) protocol script; bFirst revised protocol script; cSecond revised protocol script; dBlood diluted with 2.5 mL citrate; nd, not determined.
6
HAMOT ET AL.
pre-aspiration mix. This final revised script gave a 61% yield and was used for validation. Freeze-thawing. The effect of two freezing methods, the Controlled Rate Freezer and Mr. Frosty, on cell yield, viability, WBC subpopulation distribution, and biological activity was compared for three donors. No significant differences between the two methods were seen for total PBMC or WBC subpopulation distribution for any of the three donors (Table 3). Some fluctuations were apparent in subpopulation distribution, with Mr. Frosty favoring lymphocytes and Controlled Rate Freezer favoring monocytes (data not shown). A significant difference between the two methods in terms of the number of viable cells was identified in one of the three donors, however overall, post-thaw viability was high (approximately 80%). WBC subpopulation distribution was not significantly different between the two freezing methods for any of the donors. Finally, PBMCs from all three donors were 100% compatible with EBV transformation for both freezing methods. The Mr. Frosty method was applied for subsequent reproducibility and robustness evaluations.
parameters with a CV > 20% (RBC and platelet counts, and PBMC pre-freeze/thaw recovery). For each processing method (manual and automatic), both mRNA (hDUSP1 and hJUN) and miRNA (RNU24 and miR16) C + values showed good reproducibility with CVs < 20% (CV ranged from 0.97% to 7.77%). Both methods were reproducible for the functional assay of LCL transformation, with 100% transformation in all cases. Overall, both methods were reproducible according to the acceptance criterion of CV < 20%.
Reproducibility of manual and automated PBMC isolation protocols
Variation between automated and manual extraction methods
The manual Ficoll (EDTA) and the automated CPT (citrate) protocols were validated for reproducibility in terms of cell quantity and quality, functionality, and gene expression, using an acceptance criterion of a CV < 20% between three samples from the same donor (Table 4). With the exception of CD15 + , all parameters assayed in PBMCs extracted using the manual Ficoll method were within the 20% acceptance criterion, with most parameters having a CV of approximately 5% or less. CD15 + is a granulocyte marker which should not be present in the PBMC interphase. Increased CD15 + cells indicate granulocyte contamination in the Ficoll layer. CV% variability was greater for PBMCs isolated using the Tecan automated liquid handler, with several parameters having a CV > 15% including three
Variation between the CPT automated and the EDTA/ Ficoll manual methods was determined for each of the nine samples collected per donor for reproducibility, 4 h robustness and baseline stability (three replicates each), and evaluated according to the two isolation methods. Initially, a comparison of CBC counts in whole blood collected in citric acid CPT Vacutainers and EDTA K2E Vacutainers was performed. Blood samples collected in the automated CPT tubes had approximately 17% fewer WBCs and platelets compared to Ficoll tubes, while lymphocyte, monocyte, granulocyte, and RBC counts were approximately the same with both collection tubes (Table 6). Although PBMC recovery in freshly isolated samples compared to whole blood was higher with the manual
Robustness of manual and automated PBMC collection protocols Cell quality, quantity, and functional parameters were evaluated for robustness by comparing an 8 h and a 24 h precentrifugation delay to a 4 h delay. Most parameters were accepted (i.e., were within 15% of the values at 4 h) for all donors with both methods for the 8h pre-centrifugation delay (Table 5). However when the pre-centrifugation delay was increased to 24 h, acceptance varied between donors and methods, ranging from 38% to 85%.
Table 3. Optimization of Peripheral Blood Mononuclear Cell (PBMC) Freezing Methods in Terms of Yield, Viability, and White Blood Cell (WBC) Subpopulation Distribution Donor 1a Mean (SD) MRF
P valueb
2.04 (0.10)
0.238
(2.45) 83.16 (0.72)
0.128
(0.12) 99.10 (0.10) (2.48) 52.67 (0.32) (0.84) 22.60 (0.46) (0.35) 0.47 (0.06)
0.635 0.216 0.093 0.336
CRF PBMC counts (CASY) Total PBMCs 1.86 ( · 106)c Viability (FACSVerse) Healthy 78.57 cells (%) WBC Composition CD45 + (%) 99.03 CD3 + (%) 50.17 CD14 + (%) 24.87 CD15 + (%) 0.70
Donor 2a
(0.26)
Mean (SD)
Donor 3a Mean (SD)
MRF
P valueb
CRF
2.52 (0.05)
0.542
2.23 (0.05)
83.18 (0.42) 86.19 (1.02)
0.036
77.88 (4.73) 82.51 (1.33) 0.242
0.423 0.086 0.286 0.370
99.37 49.53 19.83 2.20
CRF 2.58 (0.15)
99.63 51.923 23.73 0.60
(0.06) 99.53 (0.12) (0.92) 54.33 (0.40) (1.29) 22.40 (0.66) (0.10) 0.77 (0.21)
MRF
P value
2.18 (0.16) 0.614
(0.25) 99.20 (0.10) 0.199 (4.17) 51.87 (1.36) 0.540 (1.10) 19.93 0.64 0.930 (0.80) 2.10 (0.36) 0.845
CRF, Controlled Rate Freezer; MRF, Mr Frosty. Bold, p < 0.05. a For each donor, cells were pooled from the three manual gradient methods; bTwo-tailed paired t-test; cTotal PBMCs = Counts for 1 mL · total volume (1 mL); single reading.
VALIDATION OF PBMC ISOLATION
7
Table 4. Reproducibility of Peripheral Blood Mononuclear Cells (PBMCs) According to Acceptance Criteria for Cell Yield, Viability, White Blood Cell (WBC) Subpopulation Composition, and Ribonucleic Acid (RNA) Targeted Gene Expression Automated (CPT) Mean (SD)a Whole Blood CBC counts
Freshly isolated PBMCs Cell counts (CASY)b Frozen isolated PBMCs Cell counts (CASY)c Cell viability, FACSVerse
WBC distribution
RT-qPCR
LCL transformation
WBC (106/mL) Lymphocytes (106/ml) Lymphocytes (%) Monocytes (106/mL) Monocytes (%) Granulocytes (106/ml) Granulocytes (%) RBC (109/mL) Platelets (106/mL) Total cells (106/mL) PBMC Recoveryb (%) Total cells (106/mL) PBMC Recoveryc (%) Necrosis (%) Late apoptosis (%) Healthy cells (%) Early apoptosis (%) CD45 + (%) CD3 + (%) CD14 + (%) CD15 + (%) hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct) 100%
5.2 1.2 23.5 0.2 4.9 3.8 71.6 5.1 157.3
Acceptance criteria CV (%)
(0.36) (0.17) (2.27) (0) (0.36) (0.2) (2.58) (1.05) (53.05)
6.9 14.4 9.6 0 7.4 5.3 3.6 20.5 33.7
4.71 (0.94) 60.9 (15.41) 3.30 83.9 1.09 13.17 75.03 10.71 98.6 63.8 14.7 1.6 20.02 26.34 23.63 20.96
(0.59) (3.4) (0.22) (1.50) (1.15) (0.29) (0.35) (1.45) (1.46) (0.26) (1.56) (1.30) (0.23) (0.25) +
Manual (Ficoll) Mean (SD)a 6.3 1.43 24.0 0.3 5.4 4.6 70.6 5.3 192.7
Acceptance criteria CV (%)
(0.1) (0.06) (0.62) (0) (0.23) (0.06) (0.8) (0.12) (9.87)
1.6 4.0 2.6 0 4.3 1.3 1.1 2.3 5.1
19.9 25.3
1.05 (0.08) 72.1 (6.01)
7.2 8.3
18.0 4.1 19.7 11.4 1.5 2.7 0.4 2.3 10.0 15.6 7.8 5.0 1.0 1.2 +
3.90 (0.23) 64.8 (6.3) 0.86 (0.13) 14.22 (0.35) 78.71 (0.55) 6.21 (0.28) 99.2 (0.09) 58.8 (0.58) 17.2 (0.47) 2.40 (1.36) 18.48 (0.88) 24.14 (0.69) 23.54 (0.63) 20.20 (1.09) +
5.8 9.7 15.0 2.5 0.7 4.5 0.1 1.0 2.7 57.4 4.8 2.9 2.7 5.4 +
Bold indicates assays for which CV% was > 20%. a Mean of three samples from one donor; bTotal PBMC counts (CASY) after isolation relative to total PBMCs (ABX lymphocyte plus monocyte counts) in whole blood; cTotal PBMC counts (CASY) after freezing relative to before.
method compared to the automated method (69.7% vs. 60.9%; Table 7), this finding was not statistically significant and was balanced out by the freeze/thaw cycle, with the automated method giving higher recovery after a passage in liquid nitrogen (87.4% vs. 65.4%). Significantly higher viability was found with the manual extraction after one freeze-thaw cycle (79.9% versus 74.0%; p = 0.01). All data taken together, the final post-thaw yield in terms of recovery of viable PBMCs is 36.4% for the manual method and 39.4% for the automated method. Significantly different proportions of both CD3 + and CD14 + populations were present in PBMCs isolated by the automated method compared to the manual method ( p = 0.03 and 0.02, respectively); however, the CV% of each method were within the reproducibility acceptance criterion. From identical RNA starting quantities, a statistically significant difference was seen in hJUN mRNA quantification, with the automated method yielding significantly less hJUN than the manual method ( p = 0.015) but not in hDUSP, while no statistically significant differences were found for miRNA target quantification. This variation may be due to hJUN expression being sensitive to the citrate anticoagulant in the CPT samples. There was no difference between the
two extraction methods in terms of transformability of LCLs.
Discussion The value of automated isolation of PBMCs over manual Ficoll separation was first reported for cell therapy purposes in 2008,27 confirming significantly higher recovery of clinical grade PBMCs using a Sepax system, while also eliminating inter-user differences. Here we report for the first time automated isolation of PBMCs for research purposes using a current widely available liquid handler. The initial optimization step demonstrated the feasibility of simultaneously obtaining plasma and PBMCs from a single whole blood sample with Ficoll manual separation in terms of viable cells and WBC subpopulation distribution. The 2000 g speed was selected as optimal on the basis of a previous report for validation of plasma isolation,1 and is the recommended centrifugation speed for obtaining both plasma and viable PBMCs suitable for downstream research applications from a single blood sample. Of the three manual gradient methods, Ficoll and Leucosep gave higher PBMC yields than CPT. Besides different
8
HAMOT ET AL.
Table 5. Robustness of Peripheral Blood Mononuclear Cell (PBMC) Protocol with 8 H and 24 H Pre-centrifugation Delay, According to Cell Quantity, Quality, and Biological Function, Post Freeze-Thaw Donor 1
8 hour delay CBC counts
Cell viability (FACSVerse) WBC distribution RT-qPCR
Donor 2
Donor 3
Automated (CPT)
Manual (Ficoll)
Automated (CPT)
Manual (Ficoll)
Automated (CPT)
Manual (Ficoll)
acc rej acc acc acc acc rej acc acc acc acc acc acc 85
acc acc acc acc acc acc acc rej acc acc acc acc acc 92
rej acc acc acc acc acc rej acc acc acc acc acc acc 85
rej rej acc rej acc acc acc acc acc acc acc acc acc 77
acc acc acc acc acc acc acc rej acc acc acc acc acc 92
acc acc acc acc acc acc acc acc acc acc acc acc acc 100
rej rej acc rej rej acc rej rej acc acc acc acc acc 54
acc rej acc acc rej acc rej rej acc acc acc acc acc 69
rej rej rej rej rej rej rej rej acc acc acc acc acc 38
rej rej acc rej rej rej rej rej acc acc acc acc acc 46
acc acc acc rej acc acc rej acc acc acc acc acc acc 85
acc acc acc rej rej acc rej rej acc acc acc acc acc 69
Total cells (/ml) WBC (106/ml) Lymphocytes (%) Monocytes (%) Viable cells (%) CD3 + (%) CD14 + (%) CD15 + (%) hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct)
LCL transformation 8 hour % accepted parameters 24 hour delay CBC counts ABX) Total cells (/ml) WBC (106/ml) Lymphocytes (%) Monocytes (%) Cell viability (FACSVerse) Viable cells (%) WBC distribution CD3 + (%) CD14 + (%) CD15 + (%) RT-qPCR hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct) LCL transformation 24 hour %accepted parameters
Acc, < 15% of 4 h value; Rej, ‡ 15% variation from 4 h value.
anticoagulant and processing methods, this may be attributed to a potentially smaller starting volume of blood due to CPT tubes allowing a maximum 4-mL sample, leading to potentially smaller collected volumes, whereas the other two methods guaranteed a 4-mL starting volume. Both Ficoll
Table 6. Comparison of Complete Blood Cell (CBC) Counts in Isolated Peripheral Blood Mononuclear Cell (PBMC) with Automated Extraction (Citric Acid) Versus Manual Extraction (EDTA) Ratio of Automated/ Manual (%) Mean (SD)a CBC counts White blood cell (106/mL) Lymphocytes (%) Monocytes (%) Granulocytes (%) Red blood cell (109/mL) Platelets (106/mL)
82.6 (8.5) 99.1 96.1 101.2 96.3 83.2
(11.1) (13.4) (6.4) (17.1) (23.2)
a Mean (SD) percent ratio comparing automated outcome versus manual outcome for each parameter from nine samples (reproducibility, robustness and stability-related samples) for each isolation method.
and Leucosep outperformed CPT in terms of pre-freeze PBMC viability. Nonetheless, the CV across the three gradients was less than 20% for viability and all WBC subpopulations other than granulocytes. CPT and Leucosep both gave less granulocyte contamination than Ficoll. Published data comparing gradient methods are conflicting, with Schlenke et al. reporting higher PBMC yields with CPT compared to Ficoll but lower viability,28 while Nilsson et al. showed the inverse in one institution and equivalence in another institution.29 Ruitenberg also reported equivalence in terms of viability and recovery.30 In our study, with the exception of yield, the choice of gradient did not appear to influence the outcome, and all three separation methods were considered fit-for-purpose given the high yields. With freezing PBMCs for long-term storage being the main purpose of this procedure, from a fit-for-purpose perspective, both the CPT and Ficoll methods are appropriate. Ficoll was selected for further evaluations of the manual method, as the most widely implemented method and providing the highest yields. Although mainly used post-cryopreservation, it has been reported that viability of Ficoll-derived PBMCs from cryopreserved samples is significantly lower than in freshly isolated samples.30 Nonetheless, the significance of this difference appears to be of little consequence in our experimental
VALIDATION OF PBMC ISOLATION
9
Table 7. Variation Between Automated and Manual Peripheral Blood Mononuclear Cell (PBMC) Isolation Methods, According to Cell Quantity, Immunophenotype, Gene Expression, and Lymphoblastoid Cell Line (LCL)-Compatibility Parameters Automated (Citrate) a
Freshly isolated PBMCs Cell counts (CASY) Frozen isolated PBMCs Cell counts (CASY) Cell viability (FACSVerse) White blood cell distribution RT-qPCR
LCL transformation
Manual (EDTA/Ficoll)
Mean (SD)
CV (%)
Mean (SD)a
CV (%)
P value
PBMC Recoveryb (%)
60.9 (13.5)
22.3
69.7 (5.7)
8.2
0.21
PBMC Recoveryc (%) Healthy cells (%) CD3 + (%) CD14 + (%) CD15 + (%) hDUSP1 (Ct) hJUN (Ct) RNU24 (Ct) miR16 (Ct) 100%
87.4 74.0 60.8 15.3 1.4 19.9 25.8 20.8 23.5
9.0 4.6 8.2 9.2 43.8 6.7 5.6 5.2 1.7
65.4 79.9 57.9 17.7 1.3 18.8 24.1 20.6 23.6
22.6 3.0 10.4 12.1 94.9 5.1 3.1 4.3 1.2
