Induction of CD4 T cell proliferation and in vitro Th1-like cytokine responses to measles virus

Abstract

Mechanisms that lead to induction of life-long immunity to measles virus (MV) are poorly understood. In the present study, we have assessed the activation, proliferation and cytokine secreting function of peripheral blood T cells from MV immune individuals. Expression of cell blastogenesis markers, such as increased forward light scatter and CD38 expression, peaked 5–7 days after infection of peripheral blood mononuclear cells (PBMC) with the live attenuated Edmonston strain of MV. Subset analysis revealed that both CD3– and CD3+ cells expressed activation markers but that the CD3+ T cells predominated late in the culture period corresponding to maximal proliferation and cell recovery. The majority of CD3+ T cells consisted of CD4+CD8– cells. IFN-γ and IL-4 production similarly showed optimal production late in culture. Depletion of CD4 cells prior to culture and MV stimulation completely abrogated both IFN-γ and IL-4 production, whereas depletion of CD8 cells did not diminish production, suggesting that CD4+CD8– T cells were principally involved in production of these cytokines. Finally, optimal IFN-γ production was elicited at high MV doses and IL-4 at much lower doses. These results suggest that among MV immune individuals, in vitro responses to measles are dominated by CD4+ T cells that, depending on antigen dose, primarily produce a Th1-like and, to a lesser extent, a Th1/Th2-mixed pattern of cytokine release.

Introduction

Although the number of measles cases worldwide has dropped dramatically since the widespread use of live, attenuated vaccines, this infectious disease continues to be a major health problem in developing world countries, contributing to an estimated 800 000 deaths/year [1]. Among developed countries, even with vaccine efficacy rates as high as 95–97%, scattered outbreaks still occur due to imported cases, the extreme contagiousness of the virus, and vaccine failure. In developing countries, vaccine efficacy is not as high, even in communities with high rates of vaccine administration [1]. Consequently, to eradicate measles, it is likely that a new vaccine will be necessary [1], and the rational design of a new generation vaccine will depend on a detailed understanding of the immune response to measles virus (MV).

One striking feature of measles infection is that while the virus induces an acute and profound state of immunosuppression lasting several weeks or months, most individuals recover and enjoy life-long immunity to MV [2–4]. This protection involves the induction of cellular and humoral immunity [5], although the detailed mechanisms involved are not well understood.

It is well established in many immune responses that distinct subsets of T cells, Th1 and Th2, play important roles in the induction of cell-mediated and antibody responses, respectively. Studies to assess cytokine production during measles infection suggested that although cytokines typical of Th1 cells could be identified during early phases of the infection, the Th2 signature cytokine, IL-4, predominated in the serum as the infection progressed [6]. Similarly, in vitro mitogen responses were skewed towards IL-4 production from subjects during infection or following administration of measles vaccine [7]. These findings, as well as those showing MV-induced depression of the Th1 polarizing cytokine, IL-12, have led to the hypothesis that MV ultimately skews T cells towards Th2 responses [8] and implies that new vaccines should be engineered to induce stronger Th1 responses. In contrast, other studies report that IFN-γ, the signature Th1 effector cytokine, was produced in vitro at higher levels shortly after infant vaccination [9–11], suggesting that measles vaccine does, in fact, induce Th1-like immunity. Further, ex-vivo studies on immune adults, including our own studies, have identified both IFN-γ secreting populations and IL-4 secreting cells [12,13]. Thus, more work is needed to clarify the nature and persistence of immunity that is induced in response to MV, how such immunity is induced in the face of ongoing immune suppression, and the signals involved in regulating these processes. In the absence of a convenient animal model, we have opted in the present study to define further the in vitro response to MV of peripheral blood mononuclear cells (PBMC) from adult immune individuals. We focus on induction of activation markers, proliferation, and quantification of secreted IFN-γ and IL-4 in response to MV during a seven-day in vitro culture period. This approach addresses features not measured by the aforementioned single cell approaches, which cannot assess proliferation or distinguish between populations with similar precursor frequencies but different rates of cytokine secretion.

Materials and methods

Subjects and peripheral blood mononuclear cell (PBMC) isolation

A total of 15 subjects were entered into this study. The primary consideration in subject selection was potential availability of large or repeated cell preparations for detailed analysis. Twelve individuals were anonymous adult blood donors reporting to the Mayo Clinic blood bank and can safely be assumed to be measles immune given the known high rate of immunity in the USA from measles-mumps-rubella vaccination or infection. The three other subjects were consenting laboratory members of the Mayo Vaccine Research Group, two of which had a history of childhood infection, and one of whom had been previously vaccinated. PBMC were isolated by standard Ficoll Hypaque (Amersham Pharmacia Biotech, Piscataway, NJ, USA) centrifugation [14]. Typically 400–500 million PBMC were obtained from buffy coats (blood bank donors), and 50–100 million cells were obtained from 50 cc of whole blood of lab donors. Cells were cryopreserved in liquid nitrogen in aliquots of 10–50 million/vial in 10% dimethylsulfoxide (Sigma, St. Louis, MO, USA), 10% fetal calf serum (FCS, Hyclone, Logan, UT, USA) in RPMI media (Celex, St. Paul, MN, USA) before use.

Cell thawing after cryopreservation

Cryopreserved cells were thawed by a modification of a previously published procedure [15]. Cells were quick-thawed and diluted over 5 min with a 10-fold excess of 10% FCS in RPMI containing 1 µg/ml bovine DNAse (Sigma) prewarmed to room temperature. After centrifugation at 500 g for 10 min, the supernatant was discarded and the pellet gently resuspended, avoiding pipeting. A 10 ml volume of FCS/DNAse containing media was again added and the cells incubated at 37°C, with occasional mixing. After 20 min, the cells were chilled on ice, and then centrifuged at 4°C. The cell pellet was resuspended in chilled RPMI containing 0·2% bovine serum albumin (BSA) until use.

Measles virus, culture media and cell culture

Edmonston B vaccine strain measles virus was prepared as an infected Vero cell lysate [13]. A control uninfected Vero lysate was prepared identically. Multiplicity of infection (MOI) was based on the viral titre of the stock preparation as determined by standard assays [16]. Briefly, serial 10 fold dilutions of MV were added in replicates of 12 to Vero cell monolayers in 96 well microtiter plates (50 000 cells/well) in the presence of 5% FCS. After 2 h, the supernatants were replaced with fresh serum containing medium. After 96 h, the cells were assessed morphologically under the light microspope for the presence of Vero cell syncytia as evidence of MV infection. Viral titres were based on the dose at which 50% of the wells were positive for syncytia. PBMC were cultured in RPMI supplemented with sodium pyruvate (Gibco, Grand Island, NY, USA), penicillin/streptomycin (Sigma) and 5% normal human serum (NHS) (Irvine Scientific, Santa Ana, CA, USA) at 200 000 cells/well in triplicate wells of 96 well microtitre plates for all assays except those in which monoclonal anti-IL-4 receptor antibodies (anti-IL-4R) were added [17]. In such assays, 400 000 PBMC were added together with 2 µg/ml of anti-IL-4R (R & D Systems, Minneapolis, MN, USA). MV or control Vero lysate was added at indicated MOI (the Vero control lysate at an equivalent dilution), and the cultures were incubated at 37°C in 5% CO₂ for various times. The indicated MOI were based on viral tissue culture infectious dose (TCID₅₀) titres of the stock preparation determined by a standard assay using Vero host cells [16]. In some experiments, alloreactive responses were obtained by coculture of 200 000 γ-irradiated (4000 rad) allogeneic or control autologous PBMC with 200 000 autologous nonirradiated PBMC. The degree of human leucocyte antigen (HLA) mismatch between stimulator and responder PBMC was not assessed. For flow cytometric analyses, replicate wells were pooled prior to staining. Cell supernatants were collected and stored at − 70°C before assay for IFN-γ or IL-4. Proliferation was assessed by pulsing cells with 1 µCi of tritiated thymidine (Perkin-Elmer, Boston, MA, USA) for 18 h before harvest.

Flow cytometry

Cells were harvested from cultures, washed with ice cold phosphate buffered saline (PBS) containing 0·2% BSA and 2 m m ethylenediaminetetracid acid (EDTA), and resuspended at cell concentrations not exceeding 4 million/ml. Staining was performed in 15 ml polypropylene tubes (Falcon, BD Labware, Franklin Lake, NJ, USA). Cell suspensions (25 µl) were added to a staining mixture consisting of either anti-CD38-allophycocyanin (APC), anti-CD3-cychrome (Cy), and anti-CD14-phycoerythrin (PE) (cocktail 1); or anti-CD38-APC, anti-CD3-Cy, anti-CD8-PE, and anti-CD4-fluoroscein isothiocyanate (FITC) (cocktail 2), using 5 µl of each monoclonal antibody (mAb). Control staining combinations included mixtures of IgG₁ conjugated to each of the above fluorochromes, or IgG_2a-APC. All antibodies were purchased from BD Pharmingen (San Diego, CA, USA). After 20 min at 4°C in the dark, the cells were washed once with an excess volume of PBS/BSA/EDTA, and the remaining cell pellet was vigorously resuspended, fixed with 0·4 ml of 1% paraformaldehyde, transferred to BSA precoated 4 ml polystyrene tubes (Falcon, BD Labware, Franklin Lake, NJ, USA), and stored at 4°C in the dark before use. Flow cytometry was performed with a FACSCalibur, and data acquisition and analysis were performed with CELLQUEST software (Becton Dickinson). Cell blasts were identified as cells with increased forward light scatter (FSC) and expression of CD38 molecules. Cells of the monocyte/macrophage lineage were excluded by gating on CD14- cells (staining cocktail 1), or gating on cells with reduced perpendicular light scatter (staining cocktail 2). CD4+CD8– and CD4–CD8+ cells were identified after gating on CD3+ cell blasts. Non-specific staining by isotype controls were always negative (<0·1%) among nonmonocyte/macrophage populations. Where applicable, standard errors for cell frequencies were determined by considering the known number of acquired events and applications of the binomial probability distribution (see Fig. 1a, insert).

Fig. 1.

Time course of proliferation and blastogenesis to measles virus. PBMC from four subjects were thawed and cultured with MV or control lysate for 3, 5, and 7 days, or for 7 days supplemented with IL-2 (5 U/ml) added on day 5. (a) At the indicated days, triplicate cultures were pooled, washed, and stained with fluorochrome-conjugated antibodies as described in Materials and Methods, and cells were analysed by FACSCalibur. The insert illustrates the fraction of blasts (high FSC, high CD38 density) at day 7 for each subject. The data for each subject at each time point was recalculated as percentage of day 7 blasts, and the mean and standard deviation of the recalculated data shown for four subjects as a function of day of culture. (b) Samples from (a) were further analysed and the percentage of cells with bright CD38 and high forward scatter determined among CD3+ and CD3– subpopulations. MV-specific CD38+ blasts were calculated by subtracting control lysate stimulated from MV-stimulated blasts. The data shown represent the mean and standard deviation from the four subjects. (c) At the indicated days, parallel cultures were pulsed with 1 µCi tritiated thymidine for 18 h before harvest. The insert depicts thymidine incorporation (Thy uptake) at day 5 for each subject. The maximal response for each subject was identified and data for each time point recalculated as percentage maximal response. The mean and standard deviation of the four subjects’ recalculated responses are shown as a function of time in culture.

Cytokine production and assay

IFN-γ and IL-4 levels in cell culture supernatants were determined by ELISA using commercially available kits (BD Pharmingen) according to the manufacturer's instructions. The levels of sensitivity for each assay were 5 pg/ml. IFN-γ ELISPOT assays were performed according to manufacturer's instructions (R & D Systems). Briefly, 200 000 PBMC were treated with or without MV (MOI of 1·0) in the presence or absence of 5% NHS in RPMI + 0·2% BSA. After 8 h, NHS was adjusted to 5% concentration in all wells, and cultures continued for 24 h prior to measurement of IFN-γ spot forming cells. MV-specific spots were calculated as the difference between antigen stimulated and unstimulated wells. Standard deviations were calculated as the square root of the sum of the variances of antigen stimulated and control wells.

Cell depletion

Cryopreserved PBMC were thawed and depleted of CD4 and CD8 subsets by pretreatment with anti-CD4 or anti-CD8 coated magnetic beads according to the manufacturer's suggestions (Dynal, Oslo, Norway) with the following minor modifications. Anti-CD4– and anti-CD8-coated beads were added to 10 million cells at a bead-cell ratio of 7·0 and 3·5, respectively, in 1 ml volume of PBS/2%FCS. After depletion, cell populations were analysed by flow cytometry to monitor purity of cell fractionation, added at a fixed dose of 200 000 cells/well, and cultured with or without MV as described above.

Results

Expression of activation markers and proliferation in response to MV

Some cell surface molecules, such as CD38, are induced following activation of lymphocytes, and expression of this marker correlates with proliferative responses [18]. In addition it is well accepted that blastogenesis is accompanied by increases in cell size reflected in enhanced forward light scatter by flow cytometry. By these criteria we monitored cell activation by different subsets of cells. Figure 1a depicts the fraction of activated cells in response to MV or control Vero lysate stimulated cultures at days 3, 5, and 7 in PBMC from four representative subjects. Activated cells were readily detectable by days 5–7 and clearly were more frequent in cultures stimulated with MV, comprising as many as 30% of the entire recovered cells. After exclusion of CD14+ cells of the monocyte/macrophage lineage, the appearance of both T cells and non-T cells were apparent in these cultures (Fig. 1b) with T cells clearly the predominant activated cell on day 7. Parallel cultures illustrated that the T cell growth factor IL-2 boosted responses to higher levels.

Figure 1c illustrates parallel proliferative responses, as defined by tritiated thymidine incorporation, as a function of culture duration. In the presence of MV, proliferation is minimally detectable at day 3 but increases dramatically by days 5–7. The responses of control cultures were negligible for three of the four subjects at all time points tested and were significant for one subject but only at day 7. Collectively, these results show that both T cell activation and proliferation to MV is readily detectable at days 5–7 during culture.

T cell subsets among MV activated blasts

Expression of CD4 and CD8 molecules among CD3+ blasts from IL-2-supplemented day 7 cultures of 13 subjects are shown in Table 1. Activated CD4+ T cell blasts were identified in cultures with MV at concentrations substantially higher than control cultures. In contrast, activated CD8+ T cell blasts (although present in MV-elicited cultures at higher levels than control cultures) were at levels much lower than CD4+ blasts. This was true in all subjects tested. This selective activation of CD4+ T cells was related to MV, because both activated CD4+ and CD8+ T cells were observed among PBMC cultured with irradiated allogeneic PBMC as an antigen stimulus (Table 1).

Table 1.

Preferential induction of CD4+ blasts by measles virus.

	Median CD38+ Cell Concentration × 10−4/ml (range)*
	CD4+CD3+	CD8+CD3+	%CD4+CD3+†
Measles	6·8 (0·9–29·4)	0·1 (0·0–1·1)	98·9 (86·6–99·7)
Control	0·1 (0·4–1·8)	0·0 (0·0–0·3)	86·3 (62·0–100·0)
Difference‡	6·7 (0·8–27·6)	0·1 (0·0–0·8)¶	99·4 (87·0–101·1)
Alloantigen§	3·5 (0·3–7·6)	2·4 (0·3–4·0)	57·0 (35·4–72·9)
Control	0·1 (0·0–0·1)	0·0 (0·0–0·0)	86·8 (50·0–100·0)
Difference	3·5 (0·3–7·5)	2·4 (0·3–4·0)**	56·8 (34·7–72·9)††

^*PBMC from 12 donors were cultured in the presence or absence of measles virus (MOI 0·2), and IL-2 was added on day 5. After an additional 2 days of culture, cells from triplicate wells were harvested, pooled and stained with a cocktail of fluorochrome-conjugated antibodies as described in Materials and Methods. The concentration/ml of CD4+CD8– T cell blasts and CD4–CD8+ T cell blasts for each subject were determined by multiplying the fraction of each subset by the viable cell concentration in the culture well prior to harvest, and the data expressed as the median (and range) of all individuals (columns 1 and 2).

^†The percentage of CD4+ T cell blasts (column 3) was calculated for each subject as column 1 divided by the sum of columns 1 and 2 and the median (and range) of the 12 subjects depicted.

^‡For each individual, the cell concentration (columns 1 and 2), or percentage CD4+ CD3+ (column 3) of control stimulated cultures was subtracted from that of measles stimulated cultures, and the results expressed as median (and range) of all subjects.

^§Allogeneic responses from 4 of the donors were initiated by culture of PBMC with irradiated allogeneic PBMC (alloantigen) or irradiated autologous PBMC (control) as described in the Methods section. Four separate stimulator-responder combinations were used. IL-2 was added on day 5, and the cells assayed on day 7 as described above for MV stimulated cultures.

^¶P < 0·0001 (Wilcoxon rank sum test, comparison CD4 blasts versus CD8 blasts)

^**P = 0·56 (Wilcoxon rank sum test, comparison CD4 blasts versus CD8 blasts)

^††P =0·004 (Wilcoxon rank sum test, comparison percentage CD4 blasts, MV stimulated versus alloantigen stimulated)

IFN-γ and IL-4 production

We next ascertained cytokine production by PBMC in response to MV. Figure 2 depicts the accumulation of IFN-γ in the supernatants at different days of culture. Analogous to proliferation and expression of T cell activation markers, IFN-γ accumulated late (days 5–7) in the culture period. Separate experiments suggest that cytokine levels plateaued by day 8, but we did not assess IFN-γ supernatant levels subsequent to that time point (unpublished observations). The appearance of IL-4 in culture supernatants from 1 individual, LD1, was detectable at day 5 or later (35–37 pg/ml in response to MV, 7–12 pg/ml in control stimulated wells). However, in this study, none of the other 14 individuals that were tested in this manner expressed detectable IL-4 in response to MV. Presuming that low levels of secreted IL-4 were being consumed via IL-4 receptors and thus precluded detection, we improved the assay sensitivity by supplementing cultures with anti-IL-4 receptor antibodies [19], as described elsewhere [17]. Using the receptor blockade assay, PBMC from 3 of 13 study subjects consistently produced IL-4.

Fig. 2.

Time course of IFN-γ production. PBMC from four donors were cultured in the presence of measles virus (MOI 0·1) or control Vero cell lysate. After 1, 3, 5 and 7 days replicate culture supernatants were harvested and assayed by ELISA as described in the Methods section. The insert depicts IFN-γ production at day 5 for each subject. The data for each subject at each time point was recalculated as percentage of day 7 IFN-γ. The mean and standard deviation of the four subjects’ recalculated responses are shown as a function of time in culture.

Cytokine production by CD4– or CD8-depleted PBMC

To determine the cell types involved in the production of IFN-γ and IL-4, we compared the responses of unselected PBMC to those that had been first depleted of CD4 or CD8 subsets, respectively (Table 2). Among a total of nine separate donors, the mean percentage of CD14+ cells among unselected, CD4 depleted and CD8 depleted cells was 20·5 & 3·8, 14·9 & 4·5, and 24·0 & 4·4, respectively; the corresponding mean percentage of CD4+ CD8– cells was 45·4 & 8·5, 1·1 & 1·4, and 57·7 & 10·9, respectively, and the corresponding mean percentage of CD4–CD8+ cells was 15·7 & 7·9, 32·3 & 18·2, and 0·6 & 0·8, respectively. Depletion of CD4 positive cells completely abrogated the production of IFN-γ. In contrast, depletion of CD8 positive cells slightly enhanced IFN-γ production, presumably by enriching CD4+CD8– and CD14+ cells. These results indicate that MV stimulated IFN-γ production is completely dependent upon, and most likely produced by, CD4+ T cells. Similarly, IL-4 production was completely inhibited by depletion of CD4 cells prior to culture, whereas depletion of CD8 cells had no inhibitory effect on cytokine production (Table 2).

Table 2.

IFN-γ and IL-4 production by negatively selected subpopulations of PBMC.

	IFN-γ		IL-4
Population*	Median pg/ml† (range)	% of unselected‡	Median pg/ml§ (range)	% of unselected¶
Unselected	570 (25–1707)	100	54 (23–76)	100
CD4 Depleted	″0 (0–194)	″1 (0–12)	″;0 (0–0)	0 (0–0)
CD8 Depleted	1327 (129–2165)	191 (101–1002)	60 (42–62)	90 (82–99)

^*PBMC from 7 donors (IFN-γ) or 6 donors (IL-4) were depleted of CD4+ cells or CD8+ cells by the appropriate antibody coated magnetic beads as described in the Methods section. The remaining cells were then cultured in replicate wells in the presence or absence of measles virus (MOI = 0·1). Cultures to be tested for IL-4 included antibodies to IL-4R.

^†IFN-γ from pooled supernatants from day 5 cultures was determined by ELISA. Shown are the levels produced in the presence MV subtracted by levels produced in the absence of MV.

^‡For each individual, the fraction of IFN-γ produced by depleted subsets relative to unselected PBMC was calculated and the median and range for all subjects shown.

^§IL-4 from pooled supernatants from day 6 cultures was determined by ELISA. The levels from four individuals produced in the presence MV are shown subtracted by the levels produced in the absence of MV. The remaining two subjects produced detectable MV specific IL-4 only after depletion of CD8 cells (21 and 12 pg/ml, respectively); no MV specific IL-4 was detected by unfractionated or CD4 depleted PBMC.

^¶For each of three individuals, the fraction of IL-4 produced by depleted subsets relative to unselected PBMC was calculated and the median and range for all subjects shown.

Dose–response optima

We compared the MV dose optima among multiple subjects for IFN-γ (n = 15) and IL-4 (n = 13). MV-specific IFN-γ responses were optimally produced at a MOI of 0·2, with a median of 1670 pg/ml (range 25–8886 pg/ml). At MOI of 0·04, 0·008 and 0·0016, the corresponding medians were 504 (range 23–954 pg/ml), 77 (0–467 pg/ml), and 47 (0–318 pg/ml), respectively. The percentage of individuals secreting greater than 10 pg/ml MV-specific IFN-γ at an MOI of 0·2, 0·04, 0·008 and 0·0016, was 15·4, 23·1, 46·2, 15·4, respectively. IL-4 production was considerably less than that of IFN-γ but appeared across a broad range of doses, from MOI of 0·2 (median 4, range 0–89 pg/ml) to 0·0016 (median 7, range 0–30 pg/ml). The percentage of individuals secreting greater than 10 pg/ml MV-specific IL-4 at an MOI of 0·2, 0·04, 0·008 and 0·0016, was 100, 84·6, 76·9 and 62·5, respectively. Because many of the subjects produced only borderline or undetectable levels of IL-4, we selected three subjects who produced the highest IL-4 levels and directly compared MV dose optima for IL-4 and IFN-γ among these subjects. As shown in Fig. 3, IFN-γ production required a high MOI > 0·2, whereas levels had decreased to minimal levels at MOI doses < 0·04. In contrast, IL-4 could be elicited at a much broader range of MOI (as low as 0·008). At higher MOI, production appeared mildly suppressed. These results suggest that distinct dose–response optima exist for production of IFN-γ and IL-4.

Fig. 3.

Measles virus optima for IFN-γ and IL-4 production. PBMC from three donors were cultured and assayed for IFN-γ and IL-4 production as described in Materials and Methods. The maximal MV specific response (cytokine production by MV minus that of control Vero extract) for each subject was identified and data for each dose recalculated as percentage maximal response. The mean and standard deviation of the three subjects’ recalculated responses are shown as a function of MV dose. The insert shows representative cytokine production for cultures stimulated with MV at MOI = 0·04, and parallel Vero control extracts.

Effect of infection by MV in serum free media

The aforementioned experiments were performed by addition of MV to PBMC in the presence of NHS. Due to neutralizing antibodies, this NHS preparation may have limited the infectivity of the MV added, thus altering T cell responses, in particular those of CD8 T cells. To address this possibility, we performed several experiments, illustrated in Fig. 4. First, a standard viral titration assay using Vero indicator cell syncytia formation as an index of infectivity [16] was performed, and it was observed that in the presence of the 5% NHS stock, infectious virus was still present although reduced in titre by approximately 10 fold. However, this inhibitory effect of NHS was only observed when it was present within the initial 2 h, presumably by interfering with viral binding and internalization but not subsequent viral spreading by syncytia formation. Secondly, we compared the responses of PBMC populations infected for 2 h in serum free media (to maximize viral binding and internalization) followed by the addition of NHS (final concentration 5%) with those of PBMC populations infected and cultured in the presence of 5% NHS throughout. As shown in Fig. 4b, at day 6, the levels of MV specific IFN-γ among cultures elicited with serum-free MV pulsing did not differ appreciably from that induced in the continuous presence of serum. Similarly, flow cytometric analysis, shown in Fig. 4c, indicated that CD4 T cell blasts were induced with MV infection in either the presence or absence of NHS, and in neither instance were appreciable numbers of CD8 blasts identified. These findings are consistent with a mechanism in which partially neutralizing effects of NHS are more difficult to observe in longer term cultures during which viral spread can occur. This mechanism would predict that T cell responses measured after short-term cultures would be much more sensitive to the neutralizing effects of NHS. Because we were unable to detect secreted IFN-γ after short-term cultures (Fig. 2), to address this possibility we relied on the ELISPOT technique, which is more sensitive to low levels of net cytokine secretion. As shown in Fig. 4d, the effects of NHS during initial culture were clearly seen. Serum-free MV pulsing substantially increased the frequencies of IFN-γ secreting cells observed at 30 h of culture above that of cultures infected in the continuous presence of NHS. In addition, responses of unfractionated PBMC were completely abrogated by prior depletion of CD4+ T cells, but not by depletion of CD8 T cells, again arguing that under these culture conditions CD4+ T cells appear to be the predominant cell secreting IFN-γ.

Fig. 4.

Influence of NHS on infectivity and induction of T cells by MV. (a) MV titrations on Vero cells were performed in the presence or absence of 5% NHS during the first 2 h and/or last 96 h of culture as described in Materials and Methods. (b) PBMC from 5 donors were resuspended in RPMI medium containing 0·2% BSA with or without the addition of 5% NHS in the presence or absence of MV at the indicated MOI. After two hours, NHS was added to microwells to bring the final concentration to 5% in all microwells. Supernatants were harvested after 5 days, assayed for IFN-γ, and the results displayed as described in Fig. 2. (c) PBMC from two donors were similarly infected with MV at the indicated MOI under serum free (– NHS) or serum containing (+ NHS) conditions for 2 h, washed and then cultured in microwells for 6 days in the presence of 5% NHS. Cells were then harvested and analysed by flow cytometry as described in Table 1. (d) PBMC from 3 donors were isolated and treated with magnetic beads as described in Materials and Methods to prepare CD4–, CD8– or unselected populations. Cells were added in duplicate microwells of an IFN-γ-specific ELISPOT plate and infected with MV in the presence (+ serum) or absence (– serum) of NHS for 8 h, at which time NHS was added to a final concentration of 5% and the cells assayed after an additional 24 h culture.

Discussion

A number of approaches have been utilized to study in vitro proliferation and cytokine production among measles-exposed individuals. Most commonly, in vitro responses of PBMC from infected or recently vaccinated individuals to polyclonal mitogens have been employed. These approaches have been especially useful in revealing immunosuppressive features of MV [20–22], but because mitogens activate cells irrespective of antigen specificity, this approach clearly has limitations in the study of MV specific responses in vitro. More recently, single cell approaches have been utilized and these have the advantage of sensitivity and the ability to estimate MV-specific precursor frequencies of cytokine secreting cells, but these approaches do not assess proliferation, and cannot distinguish between populations with similar precursor frequencies but different rates of cytokine production. Proliferative responses and cytokine production have been reported in the past, but these studies have not attempted, as we have here, to assess the relative contribution of different PBMC subsets in the overall response. In addition, we have utilized a flow cytometric approach to assess expression of activation markers of multiple PBMC subsets after stimulation with MV. For the purpose of obtaining large quantities of PBMC for multiple detailed assays, we utilized primarily anonymous blood donors. Although this is very convenient for pilot studies, an obvious limitation of this approach is that the measles exposure history (infection or vaccination) is unknown. Hence, these findings require validation with better defined subjects. This study thus complements previous studies and provides a useful framework within which to define mechanisms of measles-specific induction of T cells in future studies.

We demonstrate that T cell activation, proliferation and cytokine production were readily detected at days 5–7 in our cultures. In part, this presumably reflects low precursor frequencies of responding cells, which require expansion before numbers, and activity can be detected. However, the late appearance of cytokines may also reflect the requirements for additional in vitro maturation steps. Whereas polarized, irreversibly differentiated Th1 or Th2 memory cells are able to produce high levels of signature cytokines within 24 h of antigen exposure, less polarized or committed T helper populations, including so-called ‘Th0’, ‘Thp’ or ‘central’ memory cells, may require several days of additional in vitro maturation [23–27]. Such memory cells are among the most long-lived, and we would expect their predominance in adults with a remote history of measles virus exposure. This might also explain the late appearance of IFN-γ and IL-4 in vitro.

In this study, some subjects produced both IFN-γ and IL-4 in response to MV whereas others produced only IFN-γ. Within subjects’ PBMC that produced both cytokines, maximal IFN-γ production in vitro always greatly exceeded that of IL-4 (Fig. 3). In a separate study, we have obtained similar results on a much larger cohort of subjects who received remote measles-mumps-rubella vaccinations (manuscript in preparation). This Th1-like or mixed Th1/Th2-like pattern, although consistent with some recent studies [10], appears to contradict earlier studies predicting that measles infection or vaccination resulted in Th2 polarization [6,7]. However, the earlier studies were quite different in design, and the conclusions were based primarily on levels of serum cytokines during primary exposure to measles and in vitro responses of PBMC in response to polyclonal mitogens rather than MV, as we have done here. Moreover, Th1 signature cytokines IFN-γ and IL-2 were detected in the serum at early time points during infection [6]. Methodological differences aside, the disparities between the in vivo and in vitro approaches can be resolved with the hypothesis that some polarization may occur during infection or vaccination, but such effector cells wane with time and the persistent long-term memory that emerges is composed of relatively uncommitted memory cells. These cells may give rise to both Th1- and/or Th2-like effectors upon additional in vivo or in vitro challenges. This hypothesis would also be consistent with in vivo observations that extreme and persistent cytokine polarizations more likely occur in scenarios in which individuals are exposed to chronic rather than acute antigen exposures [28].

Th1 and Th2 cytokine responses have been shown in a number of experimental models to have different antigen dose optima (as reviewed by Constant and Bottomly [29]). In many studies, especially in models with soluble proteins as an antigen source, Th2-like responses have been elicited at low doses and Th1-like responses at higher doses, and our results in this study are consistent with this view [29]. Whether different MV doses ultimately lead to different concentrations of T cell epitopes, or alternatively affect cytokine production by selectively infecting functionally distinct antigen presenting cells [30–33] or other potentially influential cells [34,35] is not clear. In addition, the dose–response optima may have been influenced by cross-regulation. For example, IFN-γ levels elicited at high MV doses may have inhibited the generation of IL-4 production [28]. Regardless of the underlying mechanism(s), this finding illustrates how selection of conditions for in vitro assays can influence outcome. Moreover, it illustrates the potential importance of understanding viral dose-immune response relationships in vivo, when vaccination is geared to eliciting Th1 and/or Th2 immunity.

Activation, proliferation, and cytokine production were associated predominantly with the CD4 rather than the CD8 subpopulation of T cells. Although MV-elicited CD8+ blasts were identified, they were present at much lower frequencies than CD4+ blasts. Furthermore, IFN-γ and IL-4 production was completely inhibited by depletion of CD4 but not CD8 T cells. Finally, we have assayed PBMC stimulated by MV in long-term cultures followed by restimulation with MV, and such analysis has indicated that >98% of intracytoplasmic IFN-γ detectable by flow cytometry is synthesized by CD4+ T cells [36]. We consider it unlikely that the failure to induce CD8 cells under these conditions was strictly related to the presence of MV neutralizing antibodies in NHS that might minimize infection of APC and hence interfere with processing and presentation of endogenous class I restricted peptides. MV infections performed in the absence of NHS induced similar responses to those performed in the continuous presence of NHS, presumably because viral spreading could occur during these relatively long 5–7 day cultures. Nonetheless, the culture conditions used in this study may have been suboptimal for the induction of CD8 responses. Other investigations have demonstrated CD8+ responses that have been elicited with infected EBV-transformed cell lines as APC or MV-infected PBMC subsequently treated with UV irradiation to minimize suppressive effects of MV [12,37–41]. This implies greater sensitivity of CD8 T cells to immune suppression or ongoing MV infection and satisfactorily explains the difficulty in eliciting CD8 in vitro, but an analogous in vivo mechanism would likewise constrain the generation of CD8 memory. The emergence and apparent persistence of CD8 memory hence implicates additional but unidentified mechanisms. Alternatively, it is possible that at least some measles-specific memory CD8+ T cells do not persist in vivo as readily as memory CD4+ T cells. Consistent with this possibility, a recent study [42] has shown that an HLA-A2 restricted immunodominant MV peptide reacted with as many as 15% of peripheral blood CD8 T cells from measles-infected individuals in the acute phase, but reactivity was virtually absent within peripheral blood T cells isolated two to eight months thereafter [43].

In conclusion, we describe in vitro activation, proliferation, IFN-γ and IL-4 production of PBMC in response to MV and show that these responses are mediated primarily by CD4+ T cells with a Th1-like phenotype and to a lesser extent by CD4+ T cells with either a Th2 or a Th0 phenotype. We further demonstrate that generation of IFN-γ and IL-4 secreting cells can be modulated in vitro, and presumably in vivo, by antigen dose. This study is a useful framework for future studies on mechanisms of T cell induction to measles virus.