Introduction

Historically, the investigation of immunodeficiency diseaseshas centred on in vitro proliferation assays. It was noted earlythat various mitogenic substances would induce T or B lym-phocytes to proliferate and that the diagnosis of certainimmunodeficiency diseases could be supported by an absentor sluggish response. The usefulness of these assays has to acertain extent waned with the advent of flow cytometry toreliably and quickly quantify the relative and absolutenumber of the lymphocyte subsets in peripheral blood. Onthis basis, primary immunodeficiency diseases of the specific

immune system can be divided into those manifest by predominantly B cell deficits, T cell deficits or a combineddeficit.1 Nevertheless these measures yield no useful infor-mation about the ability of cells to respond when exposed totheir target antigen. Assessment of T cell function requiresmore complex assays measuring various aspects of T cellbiology, typically their ability to divide or produce cytokinesin response to mitogen or antigen.

The classical means of measuring the T cell proliferativeresponse to such stimuli has been via their uptake of [3H]-thymidine during the final hours of a 3 to 5 day culture. Thisassay is rather cumbersome and labour-intensive, requiringhandling of radioisotopes, scintillation fluid and expensivecounting equipment. Interpretation of results is also prob-lematic; a low level of [3H]-thymidine uptake could beexplained either by a depressed T cell count, impaired cellfunction or a poor batch of [3H]-thymidine. The latter can be overcome by matching patient samples with appropriate

Immunology and Cell Biology (1999) 77, 559–564

Special Feature

Carboxyfluorescein succinimidyl ester-based proliferative assays forassessment of T cell function in the diagnostic laboratory

DA FULCHER and SWJ WONG

Department of Immunopathology, Institute of Clinical Pathology and Medical Research, Westmead Hospital,Westmead, New South Wales, Australia

Abstract Immune deficiency diseases are often accompanied by abnormalities in one or both arms of the specific immune system. Impairment can often be detected as a decrease in the number of T or B lymphocytes ortheir products in the circulation, but questions are often asked as to the functional capabilities of T lymphocytes inpatients with recurrent infections. Function of T cells has traditionally been measured by their uptake of [3H]-thymidine following stimulation with antigen or mitogen in vitro. However, the ability of carboxyfluorescein succinimidyl ester (CFSE) to label lymphocytes intracellularly and track their mitotic activity by progressive two-fold reduction in fluorescence intensity prompted an alternative methodology based on flow cytometry, an approachwhich has the advantage of allowing specific gating on particular T cell subsets and simultaneous assessment of activation markers. This method was therefore evaluated for T cell responses to mitogen and antigen. Phytohaemagglutinin-induced blast transformation of CFSE-labelled T cells was reflected by an increase inforward and orthogonal light scatter and a progressive two-fold decrease in CFSE fluorescence intensity. Thesechanges allowed the derivation of various measures of mitotic activity, which correlated well with [3H]-thymidineuptake. Patients with T cell functional deficiencies showed impairment in their responses by both assays, whereasthe CFSE-based assay demonstrated that impaired blastogenesis was not simply due to depressed T cell numbers. Concomitant measurement of the activation markers CD69 and CD25 showed that CD69 was rapidly expressed onnon-mitotic cells and that this expression was progressively diluted with subsequent rounds of cell division. In con-trast, CD25 expression was unaffected by cell cycle, but was expressed in proportion to the PHA dose. Antigen-specific responsiveness to Candida was also assessed using a CFSE-based assay. Initial gating on the relativelyminor population of T cells that underwent blast transformation demonstrated progressive twofold dilutions ofCFSE intensity in responsive cells. These normal Candida responses, found in patients who had recovered fromCandida infection, contrasted with those who had not been infected with Candida or who had chronic recurrentinfection, in whom neither blast transformation nor significant mitosis could be detected. Again, there was goodcorrelation with [3H]-thymidine uptake. The CFSE-based assays are equivalent to traditional measures of mitogen-and antigen-specific T cell responsiveness in the diagnostic laboratory and have significant advantages in terms ofdecreased labour intensiveness, avoidance of radioactivity, the ability to gate on a specific population of lympho-cytes and the concomitant measurement of activation markers.

Key words: [3H]-thymidine, blastogenesis, carboxyfluorescein succinimidyl ester, CD25, CD69, T lymphocytes.

Correspondence: Dr DA Fulcher, Immunopathology, Institute of Clinical Pathology and Medical Research, Westmead Hospital,Westmead, NSW 2145, Australia. Email: <davidf@westgate.wh.usyd.edu.au>

Received 23 August 1999; accepted 23 August 1999.

controls, but this is often difficult, particularly whenanalysing hospital patients or children. Some investigatorsrecommend analysing 10 or so controls and cryopreservingcells from the median three responders.2 Newer, simplermethods may address some of these shortcomings.

The use of carboxyfluorescein succinimidyl ester (CFSE)has generated widespread interest for tracking cells in vivoand also tracing their mitotic activity.3 This fluorescent dyestains intracellular proteins and generates a fluorescentsignal that progressively halves with each mitosis. Reductionin fluorescence intensity can be quantified by flow cytometry(FCM) and an algorithm used to evaluate the extent of blasto-genesis. Furthermore, cultured cells can be stained forexpression of other cell surface markers to define lineage oractivation state.

The present paper summarizes our experience with aCFSE-based assay for the assessment of mitogenic and anti-genic T cell proliferation, using phytohaemagglutinin (PHA)and Candida antigen as candidate mitogenic and antigenicstimuli, respectively, in comparison to the traditional [3H]-thymidine-based blastogenesis assay.

Phytohaemagglutinin blastogenesis assay

Carboxyfluorescein succinimidyl ester-based assessmentof blastogenesis

The traditional blastogenesis assay requires culture of periph-eral blood mononuclear cells over a 3-day period with increas-ing doses of PHA, a [3H]-thymidine pulse being added for thefinal 6 h. The CFSE-based assay is essentially identical to this,

DA Fulcher and S Wong560

Figure 1 The 3 day blastogenic response of carboxyfluorescein diacetate succinimidyl ester (CFSE)-labelled peripheral blood mono-nuclear cells to stimulation with PHA at a dose of 0 µg/mL (upper three panels), 1 µg/mL (middle three panels) and 5 µg/mL (lower threepanels). The CFSE-based blastogenesis methodology has been described in detail.4 Briefly, peripheral blood lymphocytes were labelledwith CFSE by incubation at 37°C at a cell density of 5 × 107 cells/mL with 7.5 µmol/L CFSE in Roswell Park Memorial Institute media(RPMI) without added protein. After 10 min, cells were washed in RPMI supplemented with 10% AB serum at 4°C and then cultured at1 × 105/200 µL well. Cells were stimulated with phytohaemagglutinin (Murex Biotech Ltd, England, Cat HA16) at 0, 1 and 5 µg/mL thenharvested after 3 days. Cells were then stained for CD3 (anti-CD3-PECy5, Immunotech, Marseille, France) and either CD69 (anti-CD69-PE, Becton Dickinson, San Jose, CA, USA) or CD25 (IL-2Rα) (anti-CD25-PE, Becton Dickinson). All plots were gated on CD3-posi-tive cells (histograms not shown), while the histograms of (b) and (c) were also gated to include both resting lymphocytes (R1) and blasts(R2). (a) Alterations in light scatter characteristics; (b) expression of CD25 on CD3+ cells in culture; (c) progressive two-fold dilutionsof CFSE that accompanied mitotic cell division. Application of an analysis algorithm (see text) results in a division index of 1.64 (1 µg/mLPHA) and 3.31 (5 µg/mL PHA) divisions per added T cell.

with the addition of an initial labelling stage and the omissionof the [3H]-thymidine pulse. Blast transformation of T cellswas represented by increased forward and orthogonal scatterof light, a response that was proportional to the dose of PHA(Fig. 1a). Their activated state was accompanied by up-regulation of the IL-2Rα (Fig. 1b) and by cellular divisionrepresented by progressive halving of CFSE intensity(Fig. 1c). Even after 3 days, however, not all cells had enteredcell cycle, particularly at the lower stimulation dose.

Measuring mitotic activity with CFSE

The mitotic activity of the T cell population can be estimatedbased on the number of cell divisions, expressed as a pro-

portion of the entire starting T cell population. This algorithmassumes that the presence of two cells in a lower CFSE fluo-rescence gate arose from the mitosis of a single cell of thenext highest fluorescence intensity. A simple formula (seeAngulo and Fulcher4) yields the ‘division index’, a value of 1representing one mitosis for every cell added to culture. Thisanalysis can be directed to a specific population of cellsdefined by both light scatter characteristics and cell surfaceantigen expression.

In a typical analysis of PHA-stimulated cells, light scatterprofiles of CD3-positive cells are used to calculate the per-centage of T cells that have moved from the resting (Fig. 1,R1) to the blast (Fig. 1, R2) population (‘percentage blasttransformation’, %BT). Histograms displaying CFSE andactivation marker expression are gated by light scatter toinclude all viable T lymphocytes (i.e. both resting and blastcells), and the CFSE histogram is then used to determine thedivision index (DI). Both DI and %BT values are likely tounderestimate overall mitotic activity, because cells that diein culture will not be included and cell divisions that followdilution of CFSE intensity to autofluorescence levels cannotbe evaluated.

Correlation with [3H]-thymidine incorporation

To determine the validity of CFSE-based measures of blasto-genic activity, DI and %BT values were compared with [3H]-thymidine incorporation in 11 normal controls alongwith two patients with T cell deficiency. The latter includeda patient with systemic lupus erythematosis, multiple oppor-tunistic infections and marked CD4 lymphopenia, persistentafter withdrawal of all immunosuppressive medication, and a6-month old boy with di George syndrome. Given the greatercell numbers required for flow cytometry, CFSE analysis wasrestricted to three PHA concentrations (0, 1 and 5 µg/mL),while [3H]-thymidine incorporation was measured at 0, 1,2.5, 5, 10 and 25 µg/mL and the peak uptake determined.

All 11 control subjects demonstrated normal blastogenicresponses based on incorporation of [3H]-thymidine, while the

Proliferative assays of T cell function 561

Figure 2 The PHA response, as determined by uptake of [3H]-thymidine, in a control subject (r) compared with the patientwith systemic lupus erythematosus and CD4 lymphopenia (d).Peripheral blood mononuclear cells were cultured with doses ofPHA as indicated and their uptake of [3H]-thymidine over the last6 h of culture determined. Stimulation indices, representing meanc.p.m. in wells cultured with PHA divided by the mean of thosecultured without PHA, are plotted against the PHA dose.

Figure 3 The PHA responses, as determined by carboxyfluorescein diacetate succinimidyl ester (CFSE), in a control subject and thepatient with systemic lupus erythematosus and CD4 lymphopenia. Gating strategy is identical to the corresponding plots in Fig. 1.

two patients with T cell deficiency had ‘flat’ responses (seeFig. 2). The CFSE-based assay also reflected these differ-ences. T cells from control subjects moved into the blast gateand divided whereas T cells from the patients with T cell deficiency did not respond in either fashion (Fig. 3). Further-more, CFSE-derived values correlated well with the peak[3H]-thymidine uptake, with good discrimination betweenresponders and non-responders in this small cohort of sub-jects (Fig. 4). Quantitative analysis demonstrated that thetwo patients with T cell dysfunction showed reduced divisionindices (Fig. 4a,b) and blast transformation percentages(Fig. 4c,d), which corresponded with low uptake of radio-nucleotide. The inability of radionucleotide-based assays todiscriminate between poor T cell function as opposed todecreased T cell numbers was overcome by the CFSE-based

assay, which was able to determine the mitotic activity of Tcells even when they constituted the minority of the culturedpopulation.

Activation markers and cell cycle

T-cell responsiveness to mitogen can also be analysed according to the expression of a number of activationmarkers. CD69 is expressed soon after activation, peaking at24 h, while expression of CD25 and CD71 is maximal afterseveral days’ stimulation.5 In the 13 subjects analysed above,the CD69 response did not differ between the two patientswith T cell immunodeficiency and control subjects(Fig. 4e,f). This may have been due to harvesting the cells ata time well past that of maximal expression, a drawback that

DA Fulcher and S Wong562

Figure 4 Correlation between1 µg/mL (a, c, e) and 5 µg/mL (b, d, f) PHA responses as deter-mined by [3H]-thymidine uptake(maximal c.p.m. obtained at anyPHA dose) and the carboxyfluo-rescein diacetate succinimidylester (CFSE)-derived values divi-sion index, percentage blast trans-formation and increase in CD69-positive cells. The two patientswith T cell immunodeficiency aredepicted by filled circles.

may be overcome using other activation markers that peak atlater timepoints.

Staining for T-cell activation markers as well as with CFSEafforded a unique opportunity to examine the relationshipbetween their expression and the cell cycle. After 3 days inculture, undivided cells demonstrated high expression ofCD69, similar at both PHA doses examined, which thendeclined progressively with each cell division (Fig. 5). Thisdemonstrated that the designation of CD69 as an early T-cellactivation marker is an over-simplification, because expres-sion was only ‘early’ in terms of progression through the cellcycle rather than time in culture. One explanation for theseobservations was that synthesis and expression of CD69occurred soon after activation, at a stage preceding cell divi-sion, and that the subsequent decline occurred as a result of amitotic dilutional effect. By contrast, expression of CD25 wasdirectly proportional to the mitogenic stimulus and bore noapparent relationship to cell cycle (Fig. 5), suggesting activesynthesis throughout the culture period irrespective of mitosis.

Antigen-specific assays

Phytohaemagglutinin induces almost all T cells to enter cellcycle, hence the kinetic activity of stimulated cells provedrelatively easy to measure. In the case of antigenic T-cellstimulation, it was expected that only a small minority of thecultured T cells would respond to any given antigen and

therefore that mitotic activity may be difficult to detect byflow cytometry above the background of unstimulated cells.Indeed, this proved to be the case when the entire populationof CFSE-labelled T cells from Candida-exposed women werecultured with Candida antigen (data not shown). However, bydirecting analysis to cells which migrated into the ‘blast’ lightscatter gate, the typical CFSE dilutional pattern was found.Thus, subjects who had not been infected with Candida, orhad chronic candidiasis, showed no blast transformation andminimal cell division, whereas T cells from Candida-respon-sive controls divided normally (Fig. 6). Similar to the PHAanalysis, there was good correlation between CFSE-derivedindices (DI, %BT and the derived product of these twovalues) and [3H]-thymidine incorporation in the 5-dayculture.4 It should be noted that in occasional non-responders,a significant number of CD3+ cells were sometimes found inthe blast scatter gate but these T cells were non-mitotic,emphasizing the need to measure mitotic activity in conjunc-tion with alterations in cell size. This prompted the mutlipli-cation of the two values (DI and %BT) to generate the‘weighted division index’, which combined these parametersinto a single derived value.

Expression of CD69 was also analysed in this antigen-specific assay by gating on the entire T cell population inculture. In contrast to the PHA studies, there did appear to bedifferences in the two patient populations, with Candidaresponders demonstrating an increased proportion of CD69+

T cells in comparison to non-responders.4 However, there was

Proliferative assays of T cell function 563

Figure 5 Relationship betweencell cycle and expression of T cellactivation markers CD69 (a, b)and CD25 (c, d) at two doses ofPHA. Dot plots are gated by CD3-positivity and light scatter, thelatter to include both resting lym-phocytes and blasts (see legend toFig. 1).

significant overlap between the two groups and differenceswere less useful than mitotic values.

Our interest has focused on Candida blastogenesis, due toa clinical interest in patients with recurrent Candida infec-tions, although the same principles should be generallyapplicable to the measurement of T-cell responsiveness toother antigens.

Conclusion

The close relationship between traditional measures of T-cellresponsiveness and CFSE-derived values has supported achange in our diagnostic laboratory away from cumbersomeradioisotope-based assays towards newer CFSE-based assaysemploying flow cytometry. The latter are: (i) less labourintensive; (ii) cheaper to perform; (iii) able to focus analysison particular lymphocyte subsets; and (iv) able to measureexpression of activation markers directly and in terms of cellcycle. Their utility in the assessment of patients with sus-pected immunodeficiency disorders is the subject of ongoingevaluation.

Acknowledgements

We would like to thank Rudy Angulo for his help in performing the Candida proliferation assays, and the labora-tory ‘volunteers’ who donated blood for the normal controls.

References

1 Report of a WHO Scientific Group. Primary immunodeficiencydiseases. Clin. Exp. Immunol. 1997; 109 (Suppl. 1): S1–28.

2 Maluish AE, Strong DM. Lymphocyte proliferation. In: RoseNR, Friedman HF, Fahey JL (eds). Manual of Clinical Labora-tory Immunology. Washington, DC: American Society forMicrobiology, 1986; 274–81.

3 Lyons AB, Parish CR. Determination of lymphocyte division byflow cytometry. J. Immunol. Meth. 1994; 171: 131–7.

4 Angulo R, Fulcher DA. Measurement of Candida-specific blas-togenesis – Comparison of carboxyfluorescein succinimidylester-labelling of T-cells, thymidine incorporation and CD69expression. Comm. Clin. Cytometry 1998; 34: 143–51.

5 Biselli R, Matricardi PM, D’Amelio R, Fattorossi A. Multipara-metric flow cytometric analysis of the kinetics of surface mole-cule expression after polyclonal activation of human peripheralblood T lymphocytes. Scand. J. Immunol. 1992; 35: 439–47.

DA Fulcher and S Wong564

Figure 6 Candida-specific T-cell blastogenesis in a typical responder (left panels) and a non-responder (right panels). Carboxyfluores-cein diacetate succinimidyl ester (CFSE)-labelled peripheral blood mononuclear cells were cultured with Candida antigen (Bayer) at 0, 0.78 and 12.5 mg/mL and harvested after 5 days. Dot-plots are gated on CD3-positive cells, while CFSE histograms are gated on CD3-positive blasts (R2).