Analysis of cyt0kine gene expression in stimulated T cells of small children by semi-quantitative PCR

Only limited amounts of peripheral blood samples can be obtained from small children. Therefore, a polymerase chain reaction (PCR) aided analysis of cytokine gene expression by PBMC or T cells is a valuable tool. We present a combination of procedures to obtain an accurate estimation of the expression of the cytokines IL-4 and IFN-γ. This can be performed on T cells purified from blood samples of up to 5 ml in volume from children aged 0–4 years with allergic asthma and atopic dermatitis. This procedure includes multiple sampling of PCR products to determine the linear phase of the PCR; inter-experiment correction using a helper T-cell clone, expressing both IL-4 and IFN-γ; interpatient correction by comparing the expression of a housekeeping gene (HPRT); and finally the development of specific software to analyse densitometric data obtained by scanning photographs of agarose gels, separating PCR products. In this way it is possible to study cytokine gene expression from a very small amount of material.


Introduction
Allergic diseases such as allergic asthma and atopic dermatitis are generally characterized by increased IgE levels in the serum. > Cytokines play an important role in the regulation of IgE 4 synthesis. IL-4 is an inducer of IgE nthesis, 5 while IFN-T suppresses IgE synthesis.' Other cytokines play an additional role in the regulation of IgE. IL-5 has an enhancing effect on the stimulatory activity of IL-4 7 and IL-10 has a down-modulating effect on the production of IFN-T. 8   is a recently discovered cytokine, produced by different T-cell subsets, with IgE inducing activities comparable with IL-4 9,0 The presence of IL-4 and IFNq, has been demonstrated in cultured peripheral blood mononuclear cells (PBMC) from healthy individuals, atopic subjects and patients with the hyper-IgE syndrome. Many studies have analysed cytokine production profiles of cultured and stimulated PBMC of allergic patients by ELISA, 12'13 or production of cytokines by cloned T cells. 14 '5 However, only a few studies have described cytokine mRNA expression. 6'7 Ehlers et al. studied cytokine expression in neonatal T cells, compared with adult T cells, by PCR analysis. These authors showed that cord blood T cells are able, upon stimulation in vitro, to transcribe IL-2 mRNA. However, their capacity to transcribe mRNA for IL-3, IL-4, IL-5, IL-6, IFN-T and GM-CSF was mark-edly reduced in comparison with that in adults. 6 Our main interest is the role of cytokines in the development of allergy during infancy, since the immune system of children is generally considered to differ from the system in adults. 9 The number of T cells is relatively low and IFN-T production by T-cells is decreased, in comparison with IFN-T production in adults. 8 '9 We chose to analyse cytokine gene expression in young children (0-4 years), who are developing allergic asthma or atopic dermatitis, in order to study the pathogenesis of these diseases. We aim to correlate the mRNA expression levels in PBMC and T cells with the cytokine production profile and clinical manifestations (manuscript in preparation).
To make such studies feasible, T cells were purified from peripheral blood. 2 The amount of peripheral blood obtained from young children is usually small, and the number of T cells that can be purified is therefore limited. Moreover, cytokine gene expression is usually transient and cytokine mRNA levels generally occur at low abundance. Therefore, such analysis requires sensitive procedures. 2 The polymerase chain reaction (PER) 22'23 is a sensitive method for the detection of gene expression. Here we describe optimizations of the PCR method to analyse cytokine gene expression occurring at low abundance and from a limited number of cells, in a semi-quantitative way. 24 '25 Purification and stimulation of T cells.. PBMC from a maximum of 4 ml heparinized blood of children, were purified by density centrifugation on Ficoll-Hypaque (Pharmacia, Uppsala, Sweden). 2 The children ranged from 0-4 years of age and were healthy or with allergic asthma. These PBMC were incubated (30 min, 4C) with monoclonal antibodies specific for monocytes (My4, 100 l.tl ( (Dynal). The purity of the T-cell fraction was between 85% and 95%, as determined by FACS analysis, after staining with CD3specific antibodies (Leu-4-FITC, 50 l.tl (2.5 l.tg/ml) per 5 x 105 pelleted cells, Becton Dickinson).
The contribution of natural killer cells was always less than 3.5%.
T cells (1 x 106 cells/ml) were cultured for 16-18h in Yssel's medium 27 containing 1% human serum with and without the addition of 4bromo-calcium-ionophore (A23187, final concentration 500 ng/ml, Sigma, St. Louis, MO) and TPA (phorbol-12-myristate-13-acetate, final concentration 1 ng/ml, Sigma) at 37C, 59/o CO2. RNA isolation and eDNA reaction: After spinning down the T cells, RNA was isolated from the cells by the RNAzol B (Cinna-Biotecx Laboratories 28 Inc., Houston, TX) method. Briefly, per 1 x 10 pelleted cells 200 tl RNAzol (minimum 300 I1) was added as well as chloroform (10 l.tl per 100 !1 RNAzol B). After vigorous shaking and centrifugation, the aqueous phase was collected and an equal volume of phenol:chloroform (1:1) was added. After mixing and centrifugation, an equal volume of chloroform was added to the aqueous phase. Following centrifugation, an equal volume of isopropanol was added to the aqueous phase. After incubation at 4C for at least 2 h, the RNA was pelleted by centrifugation and washed with 20 !1 70% ethanol per 1 x 10 cells (minimum 50 btl). The RNA was resuspended in 10 l.tl water and the OD26o/OD280 were determined by spectrophotometry (Ultrospec III, Pharmacia-l_XB). Twenty l.tg glycogen (Boehringer Mannheim, Germany) was added during the phenol:chloroform extraction as a carrier. cDNA synthesis 29 was performed starting with 1/.tg RNA, after heating for 10 min at 65C. We developed the HPRT (human hypoxanthine phosphoribosyl transferase) primerset from the genomic structure, by selecting primers on exon 2 and 6. In this way, the product including the intron sequences will be too large (3.5 kB) to be amplified by PCR. The IL-4 and IFN-7 primersets were kindly donated by Dr R. de Waal Malefijt (DNAX Research Institute, Palo Alto, CA).
It was verified by Southern blot analysis that both primersets did not amplify genomic DNA (data not shown).
The total reaction volume was 100 /.tl. As a negative control 50 ng RNA in 10 btl water, which had not undergone the cDNA reaction, was used was measured on all 20 spots, from which the from six patient samples per PCR. One sample mean value was calculated by integrating the of the cDNA mixture was also used as a negative values. Two rectangles of the same size were control, drawn automatically directly next to the first row The PCR protocol started with an adaptation of rectangles. In this way the background was of the hot-start 33 to prevent aspecific primer measured on two subsequent positions of the annealing. After preparing the PCR mixture on gel. After extrapolation, a correction could be ice, the reaction tubes were placed for 3 min in made for a possible drift in the background the PCR-block (DNA Thermal Cycler, Perkinstaining of the gel. The mean background value Elmer Cetus), which was preheated at 94C. Subwas subtracted from the mean band value. The sequently, the first 25 PCR cycles (HPRT) were resulting value will be subsequently called the started: denaturation at 94C for 0.5 min, anneal-'scan value'. ing at 55C for 0.5 min and extension at 72C for 1 min. The linear phases of HPRT, IL-4 and IFN-3, were found during a different range of PCR cycles (data not shown). To cover these linear phases, the initial number of PCR cycles was 30 for IL-4 and 20 for IFN-3,. After 25 PCR cycles (HPRT) the block was cooled to 4C, 10 l.tl of PCR product was removed from the reaction tubes and the tubes were taken out of the machine. Before re-introducing the reaction tubes, a hot-start was performed again, but now only for 1.5 min, to prevent too much loss of activity of the Taq enzyme. The half-life of Taq enzyme at 94C is 35 min. 34 After five additional PCR cycles, 10 lal was collected again. This was repeated three times. The five time-point samples were analysed on agarose gel and subsequent Southern transfer. 35 The Southern transfer was performed to determine the specificity of the different primer sets and to ensure that no genomic DNA was amplified.
For analysis of patient samples, the procedure was ended by taking a photograph of the gel.  Probes to the inner region of the amplification target were end-labelled with 32ATP (see Reference 37) and hybridization was carried out. 8 The following probes were used: HPRT mRNA appeared to be present in both samples in equal amounts. It is possible to quantify these products by scanning the photographs and assigning values to the intensities of the different bands. To this end, the photograph of the agarose gel ( Fig. 1) was scanned with a scanner and the image was analysed by the described software. The corresponding scan values are shown in Table 1. These results also reflect the differential expression shown in Fig. 1, but now in absolute numbers.
For example, the scan values of the IFN-7 mRNA signal after 40 PCR cycles in stimulated T cells vs To this end, one large batch of cDNA of the B21 T-cell clone was used in the PCR for HPRT, IL-4 and IFN-y. In Fig. 2 the mean scan values are shown of B21 cDNA run in 3 independent PCR reactions, while the PCR products were analysed Standardization ofdifferences in cytokine mRNA production between different patients: HPRT was used as a housekeeping gene to standardize the cytokine mRNA expression of each patient, enabling analysis of inter-patient variation within the PCR reaction. However, for obvious reasons it was not possible to perform all the PCR reactions at the same time. Therefore, an interexperiment standard was considered necessary.
To this end the B21 T-cell clone was used. This clone produced, among others, the cytokines IL-4 and IFN-y. From a large number of B21 cells, RNA was isolated and transcribed into cDNA. Aliquots of this batch were used for all PCR experiments. Fig. 2 shows the kinetics of the PCR reaction of B21 cDNA. HPRT and IFN-y reached a plateau relatively early (HPRT after 35 PCR cycles and IFN-y after 40), while the linear phase of IL-4 was longer (plateau only after 45-50 PCR cycles). The variation in B21 scan values, obtained from independent PCR runs and analysed on different gels, was relatively small (less than 10%), resulting in a generally small interexperiment variation.
Analysis of the scan values: To optimize the reproducibility of data, photographs were processed in a similar way. Special attention was paid to obtaining gel pictures of the same intensity, contrast and paper quality. The reproducibility of the data generated with the scanner and the computer software were also examined. On a photograph displaying amplified PCR products one lane was scanned ten times. The results are depicted in Table 2. The variation in the data collected was always less than 10%.
As described in 'linearity of the PCR', comparisons have to be made in the linear phase of the reaction. To correct for possible fluctuations, mean scan values were calculated from ,the two PCR cycle times in the linear phase of the PCR. This phase was determined for each gene studied.
Correction factors were calculated for B21 mRNA. The mean correction factor for B21 HPRT was 0.78 (_--+ 0.12), for B21 It-4 0.75 (_+ 0.14) and for B21 IFN-y 0.74 (+ 0.14). The relatively small standard deviations show that the repro-  One lane of an 1.2% agarose gel with 75 Ig/500 ml ethidium bromide was scanned ten times. Arithmetic mean and S.D. were calculated for three bands with different levels of intensity.
ducibility of the PCR of B21 cDNA was high. For this reason, only small inter-experiment corrections (mean 0.77 -t-0.14) were generally necessary. The mean correction factor for HPRT mRNA expression of 20 samples was 0.81 _--t-0.21. In case the correction factors for B21 and HPRT were smaller than 0.5, or greater than 1.5, we considered the data not to be reliable because of possible overcorrection. The results of such experiments (in the case of B21) or samples (in the case of HPRT) were excluded from further analysis.
Application of the technique to patient samples: T cells were purified from peripheral blood of healthy and asthmatic children. The T cells were cultured overnight in medium alone or stimulated with a Ca-ionophore and TPA. After RNA isolation and cDNA synthesis, PCR was performed for HPRT, IL-4 and IFN-/ during the described number of PCR cycle times. In Fig. 3 a representative example is shown of one healthy child and one asthmatic child. The shapes of the curves for HPRT were comparable, both for T cells cultured in medium and for stimulated T cells. All curves reached a similar plateau level (1766 -t-74). In both the healthy control (lower left) and the asthmatic child (lower right) there was a marked difference in IFN-/ mRNA expression between stimulated and unstimulated T cells. In stimulated T cells from the healthy control, the plateau phase of IFN-g PCR was reached earlier (35 PCR cycles) than in the stimulated T cells of the asthmatic child (40 PCR cycles), indicating a higher expression of IFN-2 mRNA. In the unstimulated condition, no IFN-y expression was found in the T cells from the asthmatic child, whereas after 35 and 40 PCR cycles IFN-y expression could be detected in T cells from the healthy control. The linear range of the PCR as assessed in B21 cells, was also applicable to HPRT and cytokine gene expression in patient samples of stimulated and unstimulated T cells. In Table 3 the correction method is applied on the scan values of the healthy and asthmatic children presented in Fig. 3. First, the interexperiment variation was determined by calculating the B21 correction factors for HPRT and IFN-7. As shown in Table 3, the inter-experiment correction factor for HPRT was similar for the healthy and the asthmatic child samples (0.76 vs  patient correction, the HPRT correction factors during a limited number of cycles, after which were minimal, as shown in Table 4. The end the amount of product reaches a plateau. 24 There values for IFN-/ could be calculated with these are a number of factors that contribute to this correction factors ( Table 4). The qualitative dif-plateau phenomenon, including substrate saturaferences as observed in Fig. 3 were also reflected tion of enzyme, product strand reannealing and in these end values, but now in a semi-quanti-incomplete product strand separation. 25 The tative way (1673.69 vs 444.68). This analysis linear phase of the PCR has to be determined for allows comparison of cytokine expression each gene to be analysed, because the primers between different cell populations of larger used to amplify their products differ in G/C patient groups, versus A/T content and thus in annealing temper.atures.

Discussion
Preliminary experiments indicated that the described approach could also be applied to the Our studies have shown that semi-quantitative analysis of PCR products, that are detected by analysis of cytokine gene expression is possible autoradiography. This way the sensitivity of the in small peripheral blood samples. This was analysis was further increased. Therefore, the achieved by adapting the RT-PCR method and algorithm developed here can analyse films of combining it with an easy and reliable way of Southern blots, as was found necessary for IL-5 analysing the amplified product. For small and IL-10. numbers of cells, other methods of studying Visual analysis of the photographed gels cytokine gene expression, such as Northern blot enabled us to make qualitative comparisons analysis, cannot be employed because of the low between different patient samples. However, the amount of isolated RNA. PCR enables analysis of appearance of a positive signal after a certain genexpression in a highly sensitive and specific number of cycles did not necessarily reflect the way.
original mRNA concentration of the relevant Several adaptations of the PCR procedure were gene. Two different samples could reach a necessary to permit the required analysis. The detectable PCR signal after the same number of analysis of the expression of a housekeeping PCR cycles, while the intensity of the bands was gene is generally considered essential to correct different and another plateau level was reached. for inter-patient variation. 42 The requirements for Therefore, quantitative analysis of the bands was a suitable housekeeping gene are a stable considered necessary. This was achieved by using expression, a similar detection range as the scanning in combination with newly developed expression of the genes to be studied and no software for data analysis.
amplification of genomic DNA. 4

A HPRT primer
Another necessary step in the standardization set was developed, which meets the above cri-of the procedure was the use of B21, a Th0 teria as verified by Southern blot analysis and clone, which produces upon activation all cytohybridization. DNAse treatment, 44 often used kines relevant for this study. 36 B21 cDNA gave when RNA is contaminated with genomic DNA, reproducible data for HPRT, IL-4 and IFN-7 resulted in a significant loss of RNA and could expression in different experiments. The intertherefore not be used in our studies, experiment correction was thus generally small Determining the linear phase of the PCR was (0.77 4-0.14). This correction had to be deterimportant because at the plateau phase the mined for every individual cytokine. The variation quantified amount of amplified product is no of HPRT from different patient samples was in longer proportional to the starting amount of the same range (0.81 _+ 0.21) as for B21, there- 24 25 target molecules.
The PCR is linear only fore a similar inter-patient correction was neces-sary. When the variation is too large, over-correction can occur. We consider a correction factor smaller than 0.5 or larger than 1.5 too extreme to detect the cytokine response in the sample reliably. The analysis developed here is a reliable and simple alternative to other methods based on the use of radioactivity, like phosphor imaging. 45 The variation found in cytokine expression within a patient group of, for example, healthy children of the same age is relatively high (Koning et al., manuscript in preparation). The accuracy of the semi-quantitative method described here has sufficient discernment in relation to this extent of variation.
Recently, true quantitative methods have been developed. The method developed by Gilliland et a/. 46 uses an internal standard with the same primer requirement, but differing in the size of PCR product. However, the accuracy of quantification may be affected by sequence differences between the DNAs used for the standard and the sample. 4 When studying the intrinsic capacity of cells to express a particular cytokine gene, it is necessary to analyse highly purified cell populations. However, many studies examine the significance of cytokines in patients by analysing the total PBMC fraction only. 11'13 When endogenous cytokine gene expression is to be studied in purified T cells, it is important to note that the T cells should not be stimulated during purification. Indeed, negative selection during purification of T cells was found to be adequate to avoid stimulation. It is also important that for some cytokines, e.g., IFN-7, cells need to be stimulated to obtain detectable expression levels. In our study, we used a polyclonal stimulus, Ca-ionophore and TPA, permitting the simultaneous detection of a broad range of cytokines. For such studies allergen specific stimulation can also be applied (manuscript in preparation). We conclude that it is possible to accurately detect differential cytokine gene expression in T cells isolated from blood of healthy controls and asthmatic children. This can be achieved by using a RT-PCR with differential cycle times for different cytokines and analysing the results with a scanner and specially designed computer software. Cytokine gene expression can thus be studied in a very limited amount of material.