Journal of Immunological Methods 385 (2012) 79–89

Contents lists available at SciVerse ScienceDirect

Journal of Immunological Methods

j ourna l homepage: www.e lsev ie r .com/ locate / j im

Research paper

Optimal cellular preservation for high dimensional flow cytometric analysisof multicentre trials

Amanda A.P. Ng a,⁎, Bernett T.K. Lee a, Timothy S.Y. Teo a, Michael Poidinger a, John E. Connolly a,b

a Singapore Immunology Network (SIgN), Agency for Science, Technology and Research (A*STAR), 8A Biomedical Grove, Immunos Building, Level 3,Biopolis 138648, Singaporeb Institute of Biomedical Studies, Baylor University, Waco, TX 76798-7348, USA

a r t i c l e i n f o

Abbreviations: BCT, Blood collection tube;PFA, Paraperipheral blood mononuclear cells;PBS, phosphate buffbuffer;DC, dendritic cells;mDC, myeloid dendritic cedendritic cells;Treg, regulatory T cells;MFI, median fluonatural killer cells;NMG, NK, monocytes and granulocyte⁎ Corresponding author. Tel.: +65 64070609; fax: +

E-mail address: Amanda_Ng@immunol.a-star.edu.

0022-1759/$ – see front matter © 2012 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.jim.2012.08.010

a b s t r a c t

Article history:Received 27 July 2012Received in revised form 17 August 2012Accepted 17 August 2012Available online 24 August 2012

High dimensional flow cytometry is best served by centralized facilities. However, thedifficulties around sample processing, storage and shipment make large scale internationalstudies impractical. We therefore sought to identify optimized fixation procedures which fullyleverage the analytical capability of high dimensional flow cytometry without the need forcomplex cell processing or a sustained cold chain. Whole blood staining procedure wasemployed to investigate the applicability of fixatives including Cyto-Chex® Blood Collectiontube (Streck), Transfix® (Cytomark), 1% and 4% paraformaldehyde to centralized analysis offield trial samples. Samples were subjected to environmental conditions which mimic fieldstudies, without refrigerated shipment and analyzed across 10 days, based on cell count andmarker expression. This study showed that Cyto-Chex® demonstrated the least variability inabsolute cell count relative to samples analyzed directly from donors in the absence of fixation.Transfix® was better at preserving the marker expression among all fixatives. However,Transfix® caused marked increased cell membrane permeabilization and was detrimental tointracellular marker identification. Paraformaldehyde fixation, at either 1% or 4% concentra-tions, was unfavorable for cell preservation under the conditions tested and thus notrecommended. Using these data, we have created an online interactive tool which enablesresearchers to evaluate the impact of different fixatives on their panel of interest. In this study,we have identified Cyto-Chex® as the optimal cellular preservative for high dimensional flowcytometry in large scale studies for shipped whole blood samples, even in the absence of asustained cold chain.

© 2012 Elsevier B.V. All rights reserved.

Keywords:Cellular preservationImmunophenotypingWhole bloodCell countMarker expressionFlow cytometry

1. Introduction

Flow cytometry is a powerful tool that allows the analysisof cellular phenotype and function in the immune system(Maecker et al., 2012; Baumgarth and Roederer, 2000). Withhigh dimensionality and wide dynamic range capability, this

formaldehyde;PBMCs,ered saline;SB, staininglls;pDC, plasmacytoidrescence intensity;NK,s;TB, T cells and B cells65 64642056.

sg (A.A.P. Ng).

ll rights reserved.

technique permits single cell characterization of variousimmune subsets and detection of intracellular phosphopro-teins (Wu et al., 2010), transcription factors (Crellin et al.,2007) and cytokines (Foster et al., 2007). It has become animportant tool widely used in clinical settings such as for thediagnosis of leukemia (Stetler-Stevenson and Tembhare,2011), immunological disorders (Puxeddu et al., 2012) andmonitoring of drug therapy (Galligan et al., 2009).

Due to its complexity and sensitivity, high dimensional flowcytometry is best served by centralized facilities. This allowsbatching of samples which minimizes site variability, time andoverall cost. However, the use of centralized facilities repre-sents a challenge when designing multicentre clinical studies.

80 A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

The difficulties around sample processing, storage and ship-ment often result in surface marker degradation making largescalemulticentre studies impractical. This is particularly true ininternational pivotal trials involving developing countries,where the infrastructure for cell processing and logistics of asustainable cold chain are largely non-existent.

Various methods have been developed to addresscompromised sample integrity upon shipment (Pinto et al.,2005). The most commonly used reagent is paraformaldehyde(PFA), which cross links cellular proteins (Lanier and Warner,1981). Stabilization reagents were initially used to developbiological standards to evaluate the performance of cytometers(Barnett et al., 1996). Companies such as Streck and Cytomarkhave developed stabilizing reagents called Cyto-Chex® (StreckLaboratories, La Vista, NE) and Transfix® (United KingdomNational External Quality Assessment Service [UK NEQAS]) forcellular preservation. These stabilizing reagents are available invacutainer form which allows direct blood draw and cellularpreservation up to a week at room temperature (Canonico et al.,2004; Davis et al., 2011; Warrino et al., 2005).

In laboratory immunological studies, the most commonlyacquired human samples for flow cytometry are peripheralblood mononuclear cells (PBMCs) which require gradientseparation of cells from whole blood (Fuss et al., 2009). Thismanipulation has an impact onmarker expression (Nieto et al.,2012; Zhou et al., 2012) which results in the loss of specificcellular populations (Renzi and Ginns, 1987; Romeu et al.,1992; Tamul et al., 1995). Furthermore, clinical sites may havevarying levels of infrastructure to support proper cellularseparation and storage. In many clinical studies, whole bloodisolationmethods have replaced density gradient separation ofPBMCs as a preparative technique for immunophenotyping(Romeu et al., 1992; De Paoli et al., 1984).Whole blood analysisresults in decreased sample manipulation and reduced bloodvolume. Consequently, there are lower risks associated withsample handling and an increase in information from volumelimited samples (De Paoli et al., 1984).

With the clear advantages of whole blood collection, thereis a need to identify optimized fixation procedures. An idealprocedure should permit us to fully utilize high dimensionalflow analytical capability without the need for complex cellprocessing or sustained refrigeration.

In this study, we investigated the impact of cellularpreservation by different fixatives at various time points withand without shipment. The data demonstrated that somemethods of fixation were better at preserving cell count whileothers were better at preserving marker expression. An onlineinteractive tool has been created to visualize the impact ofdifferent fixatives on individual markers on different cellularpopulations. This data will be useful for identifying an optimalmethod of preservation depending on the analytical require-ment needed by multicentre clinical studies.

2. Materials and methods

2.1. Subjects

Informed consent was obtained from the donors accordingto NUS-IRB 09‐256. Blood was drawn into 10 ml EDTA (K2)vacutainers (Becton Dickinson) and Cyto-Chex® blood collec-tion tubes (BCT) (Streck) from 4 individuals, aged between 24

and 36 years old, equally distributed between both sexes andall of Asian ancestry. No subjects had taken anymedication 6 hprior to blood collection.

2.2. Reagents

Transfix® reagent (United Kingdom National ExternalQuality Assessment Service [UK NEQAS]), Cyto-Chex® BloodCollection tubes (Streck Laboratories, La Vista, NE) and PFA(Electron Microscopy services) diluted in phosphate bufferedsaline (PBS) (Gibco) were used as fixatives. ACK lysis buffer(Lonza) was used for red blood cells lysis. The monoclonalantibodies used are stated in Table 1. AccuCount Blankcounting beads (Spherotech) were used for enumeratingcell count.

2.3. Storage conditions

Unfixed samples, Transfix® and Cyto-Chex® treatedsamples were shipped at room temperature through courierservice to a destination within Singapore and were sentback to the lab within 4 days. These samples were kept atroom temperature till Day 10. For non-shipped samples,unfixed and PFA treated samples were stored at 4 °C whileCyto-Chex® and Transfix® treated samples were kept atroom temperature for 10 days.

2.4. Cellular fixation

Cyto-Chex® treated samples were directly fixed in thevacutainer by inverting the vacutainer 10 times. For Trans-fix® treated samples, 2 ml of Transfix® was added to 10 mlof blood as recommended by the manufacturer. Sampleswere fixed for 2 h at room temperature before staining at Day0. In order to fix samples with PFA, red blood cells were lysedat room temperature for 8 min with rotation (17 rpm) usingACK lysis buffer (Lonza) in a 1/1 volume ratio. PBS was thenadded and cells were centrifuged for 2 min, 1000 g, at roomtemperature. Cells were fixed with 2 ml of 1% PFA/PBS or 4%PFA/PBS for 20 min on ice. Fetal bovine serum (8 ml)(HyClone Laboratories Inc., South Logan, Utah) was addedto neutralize the cross linking effect of PFA.

2.5. Sample preparation

1.5 ml of unfixed and Cyto-Chex® treated samples and1.8 ml of Transfix® treated samples (to account for thedilution factor of Transfix®) were aliquoted. Red blood cellsin these samples were then lysed based on the protocolmentioned above. 1.5 ml of PFA treated samples wasaliquoted into a new tube and PBS was added. All sampleswere centrifuged and resuspended with 1.5 ml stainingbuffer (SB) (2% FCS, 2 mM EDTA [Gibco], 1.5% HEPES[Gibco], PBS [pH7.4]). 200 μl of cell suspension was seededinto a 96 well v-bottom plate (Corning) for staining withpanels for T and B cells (TB), natural killer cells (NK),monocytes and granulocytes (NMG) and dendritic cells(DC) (shown in Table 1). The remaining cell suspensionwas used for staining regulatory T cells (Treg).

Table 1Antibodies staining panel.

Flurophore TB panel Treg panel DC panel NMG panel

FITC CD45RA (HI100, Biolegend) CD4 (RPA-T4, Biolegend) BDCA3 (AD5-14H12, Macs Miltenyi) CD66 (B1.1, BD Pharmingen)PE CD45RO (UCHL1, Biolegend) CD127 (HIL-7R-M21, BD

Pharmingen)BDCA2 (AC144, Macs Miltenyi) CD14 (M5E2, Biolegend)

ECD CD19 (J3.119, Beckman Coulter) CD45RA (2H4DH11LDB9,Beckman Coulter)

CD62L (DREG56, Beckman Coulter)

PECy5 7AAD viability solution (Ebioscience)PECy7 CD4 (RPA-T4, Biolegend) CD25 (M-A251, BD Pharmingen) CD86 (Fun1(2331)BD Pharmingen) CD11b (ICRF44, BD Pharmingen)APC CD8 (RPA-T8, BD Pharmingen) CD49D (9F10, BD Pharmingen) CD11c (S-HCL-3, BD Pharmingen) CD16 (3G8, Invitrogen)APC-Cy7 HLADR (L243, BD Bioscience) HLADR (L243, BD Bioscience) HLADR (L243, BD Bioscience)Alexa Fluro 700 CD3 (UCHT1, BD Pharmingen) CD3 (UCHT1, BD Pharmingen) CD80 (L307.4, BD Pharmingen) CD56 (B159, BD Pharmingen)Pacific Blue CD86 (FUN1(2331), BD Horizon) FOXP3 (PCH101, Ebioscience) BDCA1 (L161, Biolegend) NKp46 (900, Biolegend)Pacific Orange CD45 (HI30, Invitrogen) CD45 (HI30, Invitrogen) CD45 (HI30, Invitrogen)Qdot655 CD20 (2H7, Ebioscience) CD123 (6H6, Ebioscience)

81A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

2.6. Surface staining

Samples were centrifuged and supernatant were aspirat-ed with a multichannel pipette. Antibody cocktails wereprepared at an optimized concentration and diluted with SB.50 μl of antibody cocktail (excluding FOXP3 for Treg panel)was added to each respective well for each panel. Sampleswere incubated on ice in the dark for 30 min. 100 μl of SB wasadded and samples were centrifuged. Supernatant was aspirat-ed and cells were washed with 150 μl of SB. Samples wereresuspended with 100 μl of SB.

2.7. Intracellular staining for Treg population

After surface staining, cells were centrifuged and re-suspended with 100 μl of diluted fixation/permeabilizationconcentrate (Ebioscience) according to manufacturer's proto-col. Cells were fixed on ice in the dark for 30 min and washedwith 100 μl of SB. Supernatant was aspirated and 100 μl ofpermeabilization buffer (Ebioscience) was added. Cells wereincubated for 10 min on ice in the dark and supernatant wasaspirated after centrifugation. 3 μl of FOXP3 antibodies wasadded directly. Cells were incubated on ice in the dark for30 min followed by washing with 100 μl of permeabilizationbuffer. Washing step was repeated with 200 μl of perme-abilization buffer. Samples were then centrifuged and re-suspended with 100 μl of SB.

2.8. Absolute cell counting

AccuCount Blank counting beads (Spherotech) were usedfor absolute cell enumeration according to the manufacturer'sinstructions. Beads were sonicated for 3 min, after which 50 μlwas added into 1.2 ml propylene tubes (Corning). 100 μl ofsamples was then transferred into these tubes.

2.9. Sample acquisition

Acquisition by flow cytometry was performed with theLSRFortessa™ cell analyzer (Becton Dickinson) using BDFACSDiva Version 6.1.3. The machine was calibrated withCytometer Setup and Tracking beads (BD Biosciences) tocheck for laser delay and voltage settings so as to providereproducible day-to-day cytometer performance for increaseddata consistency over time. In addition, SPHERO™ Rainbow

Calibration Particles (Spherotech) were used before sampleacquisition to check for laser and machine performance. Foreach sample, the whole volume was acquired.

2.10. Data and statistical analysis

Samples were analyzed with FlowJo 7.6.1. Hierarchicalclusteringwas used to cluster the unfixed population cell counts.To identify cell count with similar profiles, Pearson correlationwas used for distancemeasurement and Unweighted Pair GroupMethodwithArithmeticMean (UPGMA)wasused for clustering.The cell counts for unfixed samples were then presented as apercentage scale where the maximum cell count is denoted as100% and lowest as 0%.

To determine if treatments for each measurement(cell count and marker median fluorescence intensity[MFI]) differed significantly, one-way repeated measuresANOVA were used and P values b0.05 were consideredstatistically significant. Results with P values >0.05 meant thatthe measurement was not significantly affected by time orfixation. Repeated measures ANOVA were carried out usinggroups defined by treatment and days. Multiple testing correc-tions were then done using the Benjamini–Hochberg method.

To better understand the effects of timeonmarkerMFI of theunfixed, standardized Euclidean distances [(original value−mean)/standard deviation] were computed for the marker MFIvalues using day 0 of the unfixed treatment as the referencevalue. It measures the sum of absolute difference of the distancebetween all possible datasets and Day 0 unfixed. Given that thisis an absolute value of difference, it does not indicate ifmeasurements increased or decreased. The longer the distance,the bigger the difference compared toDay 0 unfixed. A graphicalillustration for Euclidean distance calculation is stated in Fig. 1.Hence, standardized Euclidean distances allow comparisonsacross cell populations and measurement types (cell counts ormarker MFI).

For comparing various treatments against the unfixed thecell count or marker MFI of a particular treatment and cellpopulation was compared to the same cell population cellcount or marker MFI of the unfixed or data obtained fromdays 4, 7, and 10. The comparison was then scored a 1 if thesummation of the standardized Euclidean distances was lessthan the unfixed and 0 otherwise (Supplementary Table 1).The scores were then counted and expressed as a percentageof the times the treatment for the particular cell count or

Fig. 1. Graphical illustration on Euclidean distance analysis. A, B, C and Drefer to the difference in distance between the data points of a particulartreatment with reference to unfixed Day 0 on a particular day for eachrespective donor. The Euclidean distance is the sum of these differences indistance between the two data points in each donor. The bigger theEuclidean distance, the more different the data (marker MFI or cell count) iswith respect to Day 0 unfixed.

82 A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

marker MFI in the cell population was better than the unfixedover all comparisons made. This summarization allowed for ahigher level interpretation of the results. Day 0 data wasomitted since day 0 unfixed was used as the reference for theEuclidean distance computations.

The data was analyzed using both Accelrys Pipeline Pilot(www.accelrys.com) and R statistical language. The comput-ed results were imported into TIBCO Spotfire and visualiza-tions were generated.

3. Results

3.1. Impact of 4 °C storage in the absence of fixation on absolutecell count

Any form of chemical modification may interfere withantibody antigen interaction. Therefore, we first set out toaccess the impact of simple 4 °C storage on our ability todetect cellular subsets over a period of 10 days. Flow cytometryanalysis was performed at Days 0, 4, 7 and 10. Followingidentification of immune cellular subsets with their lineagemarkers (Supplementary Fig. 1), absolute cell count wasenumerated with counting beads. The data was analyzed byclustering cell types with similar kinetic profile of cell count,resulting in four distinct clusters which range from the least tothe most stable cell profile (Fig. 2a).

The least stable cell types for cell count preservation wereBDCA3+ mDC, granulocytes and pDC (Fig. 2a[I]). This is notsurprising as BDCA3+ mDC and pDC are among the mostrare cell types in our analysis. Additionally, granulocytes areknown to be particularly sensitive to manipulations.Populations that were slightly more stable in cell countinclude the T and B lymphocytes (Fig. 2a[II]). The B cellspopulation displayed a slightly different trend where thepopulation stabilized at Day 4–7 and then dropped at Day 10.Populations including BDCA1+ mDC, CD14+CD16+ mono-cytes, CD14loCD16+monocytes, CD56+NKp46+ NK cells andTreg showed a more gradual decrease over time (Fig. 2a[III]).

The most stable populations were CD14+ monocytes,CD14+CD16− monocytes, CD56+ NK cells and CD56+CD16+

NK cells (Fig. 2a [IV]). In general, there was a decrease in cellcount in all populations and their subsets (Fig. 2a).

3.2. Impact of 4 °C storage in the absence of fixation onmarker expression

Given the observations that our ability to recover cellularpopulation was subset dependent, we set out to investigatethe impact of 4 °C storage on the marker MFI expressed onspecific cellular populations. Based on repeated ANOVA analysiswith multiple testing corrections, the results showed that of the117measurements, only 4were not significant,meaning that themajority of themeasurementswere affected by time and fixation(Supplementary Fig. 2).

Fig. 2b demonstrates the impact of time on individualmarkers for each cell population. Standardized Euclideandistances were used to quantify the absolute difference withina population at different time points in relation to Day 0. CD45and 7AAD were expressed in all populations. Generally, basedon the Euclidean distance of CD45 expression analyzed acrossthe TB, DC and NMG panels, we determined that this markerwas less stable than 7AAD with time of storage at 4 °C,particularly in B cells, BDCA1 and BDCA3 mDCs, CD14loCD16+

monocytes and CD4 T cells. A higher 7AAD expression wasobserved in Treg and mDC populations among the others dueto an increase in cell death.

All markers were different compared to Day 0; the mostdramatic difference was observed in CD62L especially in themonocytes, granulocytes and CD56+NKp46+ populations.This may be due to the ability of CD62L to be rapidly shedfrom the surface of the cell (Feuerecker et al., 2012). Twoother activation markers were also studied; CD86 was mostaffected in BDCA3+ population while CD80 levels weredifferent on Day 7 in mDC and pDC. CD11b and CD16 markerswere impacted specifically in the granulocyte population. Asexpected, long term storage at 4 °C had an effect on markerexpression. Unlike the impact on cell count, however, overallindividual marker expression was largely well maintained.

3.3. Impact of fixation on absolute cell count

Wenext evaluated the impact of fixation on cellularmarkerexpression. Cells were fixed in Cyto-Chex®, Transfix® or PFAfor 10 days and compared to unfixed 4 °C samples at Days 0, 4,7, and 10. Given that Transfix® causes cell membranepermeabilization (Canonico et al., 2004), our data demonstrat-ed that all cellular populations under Transfix® treatmentbecame 7AAD+ from Day 4 onwards. Thus, the gating forTransfix® treated samples on cell count was determined byexpanding the analysis to CD45+ population regardless of7AAD fluorescence.

To determine if the variation was greater in fixed thanunfixed, the comparison between fixatives and unfixed sampleswere made based on the summation of Euclidean distancebetween Days 4, 7 and 10. An example of the scoring method isindicated in Supplementary Table 1.

As seen in Fig. 3a, the best preservative was Cyto-Chex®as it maintained cell count in most populations better thanunfixed. Transfix® was not as good in preserving cell counts

Fig. 2. Impact of time on cell count and median fluorescence intensity (MFI) of individual marker on each population under unfixed 4 °C storage condition.a) Hierarchical clustering based on Pearson correlation distance matrix was used to profile the cell count for unfixed population. Data was presentedaccording to kinetic cell profile, where the Y axis shows the cell counts expressed in a percentage scale where the lowest cell count for the particulartreatment is denoted as 0% and the highest as 100%. Each cell population is depicted as a line in the figure with time at 4 °C storage on the X axis. (I) Theleast stable cell types were defined by a sharp drop of cell count from Day 0 to Day 4 and the levels continued to decrease gradually. (II) The slightly morestable cell type would be populations that showed a less drastic drop from Day 0 to Day 4 and a gradual decrease in cell numbers over time. (III) The morestable populations were those that decrease in a more gradual manner over time. (IV) The most stable populations belonged to populations that do nothave a decrease in cell count within the first 4 days and decrease gradually over the next week. b) Data presented indicates the distance of the particularmarker MFI compared to the marker MFI at Day 0. It was calculated based on Euclidean distance score between unfixed Day 0 and the respective day asdescribed in Materials and methods. The gradient in color represents the distance of the data from Day 0; the bigger the difference, the stronger the color.All markers that were expressed in different population are shown here.

83A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

84 A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

as compared to Cyto-Chex® but it still performed bettercompared to unfixed samples in DC and CD8 T cellpopulations. Transfix® was the best overall in preservingthe BDCA1 population. PFA treatments were the mostdetrimental to cell count as they did not preserve anypopulation better than unfixed. In summary, these datademonstrated that Cyto-Chex® outperformed both Trans-fix® and PFA fixations with regard to preserving cellcount.

Fig. 3. Impact of fixatives on cell count, identity of population and individual markerwas used to compare against unfixed treatment. The scoring method was illustratunfixed, the score is denoted as 1 (indicated as green), otherwise it is given 0 (indicthe reference point for calculating Euclidean distance. b) Using the same scoring medefines the population and normalized by the number of markers. This would indidentity of the population. c) In order to study the effects of fixation on individual msum of counts was computed from all populations expressing the particular markpercentage of time a particular fixative is better in preserving the specific markerdifferent time points (Day 0, 4, 7, 10) (X-axis) for all treatments. Standard error w

3.4. Impact of fixation on marker expressions

We next set out to investigate the impact of fixations onmarker expression. As shown in the gating strategy (Supple-mentary Fig. 1), the identity of a population is defined by aset of markers. The data represented in Fig. 3b indicate thepercentage of time when the fixative was better in preservingthe identity of the population than unfixed (>50%). It can beseen that Transfix® was better at preserving the identity of

MFI of all population. a) Summation of Euclidean distance in each treatmened in Supplementary Table 1. For treatments which performed better thanated as red). Day 0 was not used for comparison as unfixed Day 0 was used asthod as Fig. 3a, the sum of scores was computed from the set of markers thaicate the percentage of time a particular fixative is better in preserving thearker expression, the same scoring method employed in Fig. 3a was used. Theer and normalized by the number of populations. This would indicate thein all populations expressing it. d) 7AAD MFI values (Y-axis) plotted againsas plotted with the mean value.

t

t

t

85A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

the following populations based on their respective set ofmarkers: CD4 T cells, BDCA1 mDC, BDCA3 mDC, pDC, CD56NK and granulocytes, whereas Cyto-Chex® was better atpreserving these populations: BDCA1 DC, BDCA3 DC, pDC andCD56 NK cells. None of the PFA preparations was better inpreserving the overall MFI of any specific population.

Fig. 3c shows the ability to detect a given marker in all ofthe populations which expresses it. When we compared allfixatives to unfixed samples as described in materials andmethods, clear performance patterns emerged. Overall, Trans-fix® was the best fixative among all fixatives in terms of MFIpreservation. This was reflected particularly in BDCA3, CD11b,CD123, CD19, CD4, CD45, CD45RA, CD56, CD66, CD80 and CD86staining. Cyto-Chex® performance in marker expressions wasrelatively robust, with preservation of BDCA2, BDCA3, CD123,CD127, CD25, CD3, CD45, CD45RA, CD56 and CD80. However,although Transfix® was the better fixative for preservingmarker expression, it displayed detrimental effects on cellmembrane integrity. Fig. 3d illustrates that Transfix® treatedsamples displayed elevated 7AAD levels in all cell populations,which is amarker of increased cell membrane permeabilization.Among all the different fixatives, PFA demonstrated the worstperformance in preserving marker expression. 1% PFA wasfavorable in preserving CD123, CD80 and FOXP3 while 4% PFAwas better in preserving BDCA3, CD127 and CD80. In the wholelist of markers that were studied, CD8 was the most sensitivetowards any fixation (Supplementary Fig. 3). Transfix® was themost detrimental to this marker compared to all fixationprocedures employed. This highlights the fact that optimalperformance is dependent on the markers of interest althoughoverall Transfix was the best in MFI preservation.

3.5. Impact of fixation on cell count in shipped samples

We observed that certain fixatives were superior at preserv-ing cell count while others were better at marker preservation.However, all these procedures were carried out in a laboratorysetting. To analyze samples from field studies, the preparationproceduremustminimize any effects of shipment. In addition, inmany large scale international studies, a reliable cold chain isdifficult to establish, so the procedure must also preservesamples in the absence of refrigeration. Thus, we studied theimpact of ambient temperature shipment on the two bestperforming fixatives, Cyto-Chex® and Transfix®. We added acondition where unfixed sample was shipped at room temper-ature to examine if room temperature storagewithout fixative isdetrimental. All samples were shipped under field condition atambient temperature (~31 °C) on Day 0 to a destination withinSingapore, shipped back to the laboratory at Day 3 (~23 °C) andstored at room temperature (~23 °C) for the remaining study. Itcan be seen from Fig. 4a that shipment did not have a majorimpact on cell count with either fixation procedures. Bothfixation procedures out performed shipped unfixed samples incell count preservation.

3.6. Impact of fixation on marker preservation in shipped samples

Since shipment did not have a significant impact on cellcount, we went on to compare the impact of shipment on cellpopulation identity to non-shipped samples. The percentageof time where the specific treatment was better than unfixed

in each cell population is depicted in Fig. 4b. Samples whichwere fixed with Transfix® and then shipped reduced theability to identify CD4 T cells and CD14 monocytes, with theeffect on the latter more pronounced. Cyto-Chex® treatedsamples impaired our ability to identify BDCA3 DC, pDC andgranulocytes. The effect on granulocytes was more pro-nounced as compared to the other populations wherechanges were less than 15%. Shipment of samples withoutprior fixation was detrimental to the ability to detect specificcellular populations.

Fig. 4c depicts the effects on individual markers in allpopulations expressing it. When compared to non-shippedsamples, shipping after Cyto-Chex® fixation had an effect onBDCA2 and BDCA3 markers, and shipping after Transfix®fixation affected CD4. Shipment of unfixed samples had anadverse effect on most markers, though there were some thatwere not affected (BDCA1, BDCA3, CD8 andCD80). In summary,shipment did not have a major impact on marker expression inmost cell populations for Transfix® and Cyto-Chex® preservedsamples as compared to similarly preserved non-shippedsamples.

As a general conclusion, we have shown that Cyto-Chex® isbetter at preserving cell count and Transfix® is better atpreserving marker MFI (Fig. 5a) among all fixatives. However,marker expression is best preserved in the absence of bothshipment and fixation (Fig. 5b). As the selection of the optimalfixationmethod is dependent on the condition of the study, thetime point and markers of interest, we created an online toolfor users to best interrogate the dataset according to theirparticular study design (link: http://public.tableausoftware.com/views/fixation_treatment/MainPage?:embed=y).

4. Discussions

Flow cytometry has long been the tool of choice for theanalysis of cells in the immune system. The diversity of reagents,applications and the ability to dissect cellular phenotype andfunction in ways that is impossible with other assays made itarguably the most powerful technology available for probingimmune phenotype.

It is important to limit those study parameters whichintroduce data variability and error due to compromised sampleintegrity. A study done by Bergeron et al. (2002) demonstratedthat storing whole blood at room temperature for 2 days has animpact on the light scatter properties and an alternative gatingstrategy that based exclusively on cell-lineage specific markerwas required. Another study (Schumacher and Burkhead, 2000)suggested that peripheral blood and umbilical cord blood heldup to 2 days displayed decreased B cell numbers and a reductionin marker intensity. Given that not all laboratories have thecapacity to receive and analyzed blood samples within aday, the problem of sample degradation raises the needfor a fixative that preserves sample integrity, even whenshipped under suboptimal conditions.

There had been several studies investigating the effects offixatives on cellular analysis. A study conducted by Canonicoand co workers had shown that Transfix® was supreme inpreserving scatter characteristics, immunophenotyping andabsolute cell count in the lymphocyte population. Other studies(Warrino et al., 2005; Davis et al., 2011) demonstrated thatCyto-Chex® was ideal in maintaining the immunological

Fig. 4. Effects of shipment on cell count, identity of population and individual marker MFI of all populations. Cyto-Chex-shipped refers to shipped samples fixedwith Cyto-Chex. Transfix-shipped refers to shipped samples fixed with Transfix. Unfix-shipped refers to shipped unfixed samples. a) To determine if the variationof dataset was bigger in fixed than unfixed samples, the comparison between fixatives and unfixed were made based on the summation of Euclidean distancebetween Days 4 and 10. The same scoring method was used as Fig. 3a. The score is denoted as 1 (indicated as green) for treatments which performed better thanunfixed. Otherwise, it is given 0 (indicated as red). b) The same scoring method employed in Fig. 4b was used. The sum of counts was computed and normalizedby the number of markers defining a given population. This would indicate the percentage of time a particular fixative is better in preserving the identity ofspecific cell population. c) In order to study the effects of fixation on individual marker expression, the same scoring method employed in Fig. 4c was used. Thesum of counts was computed and normalized by the number of populations expressing the marker. This would indicate the percentage of time a particularfixative is better in preserving specific marker in all populations expressing it.

86 A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

Fig. 5. Summary on the impact of fixatives on cell count and shipment. Cyto-Chex‐shipped refers to shipped samples fixed with Cyto-Chex. Transfix-shippedrefers to shipped samples fixed with Transfix. Unfixed-shipped refers to shipped unfixed samples. a) Based on all markers or all cell counts in all populations, thesum of counts was computed from all conditions and normalized by the number of comparisons made. This would indicate the percentage of time a particularfixative is better in preservation compared to unfixed condition. b) The sum of counts for unfixed was computed from marker MFI in all populations andnormalized by the number of comparisons made. Unfixed treatment was compared to the respective fixatives indicated on the X-axis and the percentage of timeunfixed outperform in preservation compared to respective fixatives was indicated.

87A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

markers for the white blood cells in whole blood sample. Thesestudies were in contrast to work done by Plate et al. (2009),which illustrated that neither Transfix® nor Cyto-Chex® BCTwas useful in preserving CD4+ quantification. In all of thesestudies, less than 5-color fluorophore staining was employed.Although useful, such a panel cannot be said to leverage theanalytical capability of modern flow cytometers which utilizeFRET paired and Qdot nanocrystals reagents. Furthermore, giventhat cellular populations are defined by coordinated expressionof several markers and the effect on any onemarker couldmeana loss of the population in the analysis. Therefore, we set out toanalyze fixation procedures which utilize whole bloodwith highdimensional flow in an environment that mimics field studies.

Weexamined the effects of commercially available fixativessuch as Transfix® (Cytomark) and Cyto-Chex® (Streck) on cellcount and MFI. Sample processing is simple as these reagentscome in either vacutainer which allows direct draw or liquidform which is added directly into the whole blood. In addition,these fixatives permit storage at room temperature accordingto the manufacturer's protocol. From the data we concludedthat Cyto-Chex® was generally the best in preserving cellnumbers. This result was supported by study from Warrino etal. which showed that Cyto-Chex® displayed good correlationbetween 6 h and 7 days on absolute cell count of someleukocytes in normal and HIV infected patients. However, dueto the constraints of the 4-color staining panel, the number of

88 A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

cellular subsets identified was restricted. Transfix®, displayedbetter cell count preservation in specific populations (BDCA1+,BDCA3+, CD8 T cells, pDC) and this result corresponded to thestudy conducted by Canonico et al. which showed that absolutecell count of lymphocytes was better preserved compared tomonocytes and granulocytes. However, the gating strategy ofthis group was solely based on side scatter and CD45 markerwhich do not allow the dissection of cellular subsets. Next, westudied the impact of these fixatives onmarkerMFI and our datashows that Transfix® was the better preservative compared toCyto-Chex® for all extracellular and intracellular markersstudied. However, Transfix® caused increased membranepermeabilization as reflected by increased 7AAD staining. Thisresult was also shown in the study done by Canonico et al.where they illustrated the effects through propidium iodidestaining. Surprisingly, unfixed samples stored at 4 °C was thebest in retaining marker expression hence indicating that anyform of fixation could potentially have an impact on conforma-tional shape of the epitopes resulting in decreased antibodybinding.

Besides commercial fixatives, the commonly used fixative(PFA) was employed in this study. In this study, we performedred blood cell lysis prior to PFA fixation and excess amount ofFBS was added to neutralize the cross linking effects of PFAafter incubation. PFA treatments consistently yielded poorresults in terms of cell count and marker MFI, thus thispreservation method was not optimal. PFA fixation beforestaining was chosen due to the ease of transporting items asstained samples might degrade faster if the delay betweensample staining and acquisition is too long. In addition, PFAwas not directly added into the whole blood as it resulted infixation of red blood cellswhichwas not possible to lyze at latertime points (data not shown). Although these factorsmay havecontributed to the poor performance of PFA in this study,multicolor staining or red blood cells lysis at clinical sites priorto fixation and shipment is in most cases impractical.

The numbers of studies analyzing the impact of shipmenton fluctuating temperatures on flow cytometry analysis islimited and were mainly done on controlled temperature tomimic shipment conditions (Antonenas et al., 2006; Ekong etal., 1993). Studies conducted by Ahmann et al. (2008) showedthat tissue sample shipment on dry icewithin USA did not havea drastic impact on plasma cell isolation, viability and RNAintegrity. Olson et al. (2011) had investigated the effects ofovernight ambient temperature shipment at different seasonsand samples subjected to 40 °C for more than 8 h showedreduced viability. In Olson et al. study, PBMCs were isolated foranalysis and 4 color flow staining panel was employed.Furthermore, the duration of shipment was overnight withsample analysis was done the following day. Over the last5 years, multi-plex (>12 color) flow cytometry has becomethe mainstay, with international shipping becoming essentialin multicentre studies. In this situation, samples will require3 days for shipment excluding custom clearance delays. Thus,we went on to explore the effects of shipment in our study toassess if shipment could have an impact on cell count and themarker MFI; this shipment effect would mimic field siteslocated within the tropical region. The results demonstratedthat both Transfix® and Cyto-Chex® fixatives were stable inthe presence of shipment and displayed the same trend as thenon-shipped samples across a period of 10 days.

Since populations of interest depend on the study designand the scientific question asked we have developed anonline interactive tool to allow scientists to explore andidentify the preservative of choice that is most applicable totheir study.

5. Conclusions

To our knowledge, this is the first study of its kind analyzingwhole blood fixation procedures for high dimensional flowcytometry under condition amenable to multicentre trials inthe absence of a sustained cold chain. Our data demonstratedthat the appropriate choice of fixative depends on the markersof interest. However, the study reveals general conclusions onmulti-dimensional flow analysis. We have determined that ifan investigator intends to immunophenotype prominentpopulations based on lineage and activation markerswhere cell numbers are not a critical issue, cold storagein the absence of fixative is recommended. In the absenceof available refrigeration, Transfix® demonstrates supe-rior performance for the analysis of surface markersexpression but not intracellular markers. Overall, basedon the studies presented here, we would recommendCyto-Chex®, as the preservative proved effective at retainingcell numbers, extracellular and intracellular surface markersfollowing ambient temperature shipment.

Supplementary data to this article can be found online athttp://dx.doi.org/10.1016/j.jim.2012.08.010.

Acknowledgments

The study was supported by BMRC/EDB grant (IAF 311006,Clinical Immunomonitoring; JC) and A*STAR core funds. Theauthors would like to thank Dr. Anna-Marie Fairhurst for herguidance; Nivashini Kaliaperumal, Wu Xue Lin, Lin Yufang andDr. Richard Hopkins for their valuable contributions on themanuscript and their support rendered in the experimentalwork.

References

Ahmann, G.J., Chng, W.J., Henderson, K.J., Price-Troska, T.L., DeGoey, R.W.,Timm, M.M., Dispenzieri, A., Greipp, P.R., Sable-Hunt, A., Bergsagel, L.,Fonseca, R., 2008. Effect of tissue shipping on plasma cell isolation,viability, and RNA integrity in the context of a centralized goodlaboratory practice-certified tissue banking facility. Cancer Epidemiol.Biomarkers Prev. 17 (3), 666.

Antonenas, V., Garvin, F., Webb, M., Sartor, M., Bradstock, K.F., Gottlieb, D.,2006. Fresh PBSC harvests, but not BM, show temperature-related loss ofCD34 viability during storage and transport. Cytotherapy 8 (2), 158.

Barnett, D., Granger, V.P., Storie, I., Wilson, G.A., Reilly, J.T., 1996. Evaluationof a novel stable whole blood quality control material for lymphocytesubset analysis: results from the UK NEQAS immune monitoringscheme. Cytometry Commun. Clin. Cytom. 26, 216.

Baumgarth, N., Roederer, M., 2000. A practical approach to multicolor flowcytometry for immunophenotyping. J. Immunol. Methods 243 (1–2), 77.

Bergeron, M., Nicholson, J.K.A., Phaneuf, S., Ding, T., Soucy, N., Badley, A.D.,Hawley Foss, N.C., Mandy, F., 2002. Selection of lymphocyte gatingprotocol has an impact on the level of reliability of T cell subsets in agingspecimens. Cytometry Clin. Cytom. 50, 53.

Canonico, B., Zamai, L., Burattini, S., Grangerd, V., Mannelloc, F., Gobbi, P.,Felici, C., Falcieri, E., Reilly, J.T., Barnett, D., Papa, S., 2004. Evaluation ofleukocyte stabilisation in Transfix®-treated blood samples by flowcytometry and transmission electron microscopy. J. Immunol. Methods295, 67.

89A.A.P. Ng et al. / Journal of Immunological Methods 385 (2012) 79–89

Crellin, N.k, Garcia, R.V., Levings, M.K., 2007. Flow cytometry-based methodsfor studying signaling in human CD4+CD25+FOXP3+ T regulatorycells. J. Immunol. Methods 324 (1–2), 92.

Davis, C., Wu, X., Li, W., Fan, H., Reddy, M., 2011. Stability of immunophenotypicmarkers in fixed peripheral blood for extended analysis using flowcytometry. J. Immunol. Methods 363, 158.

De Paoli, P., Reitano, M., Battistin, S., Castiglia, C., Santini, G., 1984. Enumeration ofhuman lymphocyte subsets bymonoclonal antibodies and flow cytometry: acomparative study using whole blood or mononuclear cells separated bydensity gradient centrifugation. J. Immunol. Methods 72, 349.

Ekong, T., Kupek, E., Hill, A., Clark, C., Davies, A., Pinching, A., 1993. Technicalinfluences on immunophenotyping by flow cytometry. The effect of timeand temperature of storage on the viability of lymphocyte subsets.J. Immunol. Methods 164 (2), 263 (15).

Feuerecker, M., Kaufmann, I., Salam, A.P., Choukèr, 2012. Effects of cryopreser-vation with polyethylene glycol on the expression of CD11B and CD62L onthe surface of polymorphonuclear leukocytes. CryoLetters 33 (2), 150 (10).

Foster, B., Prussin, C., Liu, F., Whitmire, J.K., Whitton, J.L., 2007. Detection ofintracellular cytokines by flow cytometry. Curr. Protoc. Immunol. (Chapter6: Unit 6.24).

Fuss, I.J., Kanof, M.E., Smith, P.D., Zola, H., 2009. Isolation of wholemononuclear cells from peripheral blood and cord blood. Curr. Protoc.Immunol. (Chapter 7: Unit7.1).

Galligan, C.L., Siebert, J.C., Siminovitch, K.A., Keystone, E.C., Bykerk, V., Perez,O.D., Fish, E.N., 2009. Multiparameter phospho-flow analysis of lympho-cytes in early rheumatoid arthritis: implications for diagnosis andmonitoring drug therapy. PLoS One 4 (8), e6703.

Lanier, L.L., Warner, N.L., 1981. Paraformaldehyde fixation of hematopoieticcells for quantitative flow cytometry (FACS) analysis. J. Immunol. Methods47, 25.

Maecker, H.T., McCoy, J.P., Nussenblatt, R., 2012. Standardizing immuno-phenotyping for the Human Immunology Project. Nat. Rev. Immunol. 12(3), 191.

Nieto, J.C., Canto, E., Zamora, C., Ortiz, M.A., Juarez, C., Vidal, S., 2012.Selective loss of chemokine receptor expression on leukocytes after cellisolation. PLoS One 7 (3), e31297.

Olson, W.C., Smolkin, M.E., Farris, E.M., Fink, R.J., Czarkowski, A.R., Fink, J.H.,Chianese-Bullock, K.A., Slingluff Jr., C.L., 2011. Shipping blood to a centrallaboratory in multicenter clinical trials: effect of ambient temperature onspecimen temperature, and effects of temperature onmononuclear cell yield,viability and immunologic function. J. Transl. Med. 9, 26 (2011 Mar 8).

Pinto, L.A., Trivett, M.T., Wallace, D., Higgins, J., Baseler, M., Terabe, M.,Belvakov, I.M., Berzofsky, J.A., Hildesheim, A., 2005. Fixation andcryopreservation of whole blood and isolated mononuclear cells:influence of different procedures on lymphocyte subset analysis byflow cytometry. Cytometry B Clin. Cytom. 63B, 47.

Plate, M.M., Louzao, R., Steele, P.M., Greengrass, V., Morris, L.M., Lewis, J.,Barnett, D., Warrino, D., Hearps, A.C., Denny, T., Crowe, S.M., 2009.Evaluation of the blood stabilizers TransFix and Cyto-Chex BCT for low-cost CD4 T-cell methodologies. Viral Immunol. 22 (5), 329.

Puxeddu, I., Bongiorni, F., Chimenti, D., Bombardieri, S., Moretta, A., Bottino,C., Migliorini, P., 2012. Cell surface expression of activating receptors andco-receptors on peripheral blood NK cells in systemic autoimmunediseases. Scand. J. Rheumatol. 41 (4), 298–304.

Renzi, P., Ginns, L.C., 1987. Analysis of T cell subsets in normal adults.Comparison of whole blood lysis technique to Ficoll–Hypaque separa-tion by flow cytometry. J. Immmunol. Methods 98 (1), 53.

Romeu, M.A., Master, M., Gonzalez, L., Valls, A., Verdaguer, J., Corominas, M.,Bas, J., Massip, E., Buendia, E., 1992. Lymphocyte immunophenotypingby flow cytometry in normal adults. Comparison of fresh whole bloodlysis technique, Ficoll-Paque separation and cryopreservation. J. Immunol.Methods 154, 7.

Schumacher, M.J., Burkhead, T., 2000. Stability of fresh and preserved fetaland adult lymphocyte cell surface markers. J. Clin. Lab. Anal. 14, 320.

Stetler-Stevenson, M., Tembhare, P.R., 2011. Diagnosis of hairy cell leukemiaby flow cytometry. Leuk. Lymphoma 52 (Suppl. 2), 11.

Tamul, K.R., Schmitz, J.L., Kane, K., Folds, J.D., 1995. Comparison of the effectsof Ficoll–Hypaque separation and whole blood lysis on results of immuno-phenotypic analysis of blood and bonemarrow samples frompatients withhematologic malignancies. Clin. Diagn. Lab. Immunol. 2 (3), 337.

Warrino, D.E., DeGennaro, L.J., Hanson, M., Swindells, S., Pirruccello, S.J.,Ryan, W.L., 2005. Stabilization of white blood cells and immunologicmarkers for extended analysis using flow cytometry. J. Immunol.Methods 305, 107.

Wu, S., Jin, L., Vence, L., Radvanvi, L.G., 2010. Development and application of‘phosphoflow’ as a tool for immunomonitoring. Expert Rev. Vaccines 9(6), 631.

Zhou, L., Somasundaram, R., Nederhof, R.F., Dijkstra, G., Faber, K.N.,Peppelenbosch, M.P., Fuhler, G.M., 2012. Human granulocyte andmonocyte isolation procedures: impact on functional studies. Clin.Vaccine Immunol. (Epub ahead of print).