Independent and Epigenetic Regulation of the Interleukin-4 Alleles in CD4+ T Cells

See allHide authors and affiliations

Science  28 Aug 1998:
Vol. 281, Issue 5381, pp. 1352-1354
DOI: 10.1126/science.281.5381.1352


How an individual effector T cell acquires a particular cytokine expression pattern from many possible patterns remains unclear. CD4+ T cells from F1 mice, which allowed assignment of the parental origin of interleukin-4 (IL-4) transcripts, were divided into clones that expressed IL-4 biallelically or monoallelically from either allele. The allelic pattern was transmitted as a stable epigenetic trait. Regulation of cytokine expression by a mechanism that treats each allele independently suggests a probabilistic process by which a diverse repertoire of combinatorially assorted cytokine gene expression patterns could be generated among the clonally related daughters of a single precursor cell.

Descriptions of T helper (TH) cell types that express cytokine patterns distinct from the classic TH1 and TH2 subsets are not readily explained by current models of T cell differentiation from naı̈ve to cytokine-expressing effector cells (1). The unusual cytokine patterns appear as if generated by combinatorial assortment of probabilistically expressed genes (2,3). We hypothesized that a probabilistically regulated gene would have two chances to be expressed in diploid cells and that, if the two alleles were regulated independently, a mixture of cells that used either one or both alleles should exist within a population expressing the gene.

Although the IL-4 gene was nonpolymorphic among a number of traditional Mus musculus inbred strains, a polymorphism in exon 1 allowed discrimination of the IL-4 cDNA of inbred strains from the CAST/Ei strain, by differential sensitivity to the restriction enzyme Bsg I (Fig. 1A) (4). CD4+ T cells from (129 × CAST/Ei)F1 hybrid mice were stimulated in vitro under conditions that favored the generation of IL-4–expressing effector cells (5). Even under such conditions, the frequency of IL-4–expressing cells is less than 5% (3). We used a limiting-dilution approach to screen for monoallelic IL-4 gene expression (6), a strategy used to demonstrate monoallelic expression among olfactory receptor genes (7).

Figure 1

Monoallelic IL-4 gene expression in single-cell populations. (A) Strategy for detecting the origin of the IL-4 transcript. Primers from exons 1 and 2 of the IL-4 gene were used to amplify reverse-transcribed cDNA. The resulting product was further amplified with a nested primer that spanned the exon 1–exon 2 splice junction. The resulting PCR products were discriminated by their sensitivities to the endonucleases Bsg I (CAST/Ei gene product sensitive) and Sau 3AI (all gene products sensitive) as resolved after agarose gel electrophoresis. (B) Limiting-dilution analysis of stimulated CD4+ T cells purified from (129 × CAST/Ei)F1 mice. Pools 3B88 and 3B97 were amplified and the resultant product resolved without (Uncut) or with (Bsg I or Sau 3AI) digestion with the indicated restriction endonucleases in the three consecutive lanes as indicated. Amplicons were detected in 6/16 (3B88) or 7/14 (3B97) separate experiments, of which four and five digestions are shown. 1B1 represents cDNA from undiluted material from the F1 cells demonstrating the capacity to discriminate both alleles in mixed populations. The rightmost three samples were run on a separate gel, accounting for the increases in mobility of the restriction fragments.

Under these conditions of limiting template, the semi-nested polymerase chain reaction (PCR) approach was, on average, capable of detecting IL-4 transcripts 30% of the time from a repeatedly screened cDNA aliquot. Consequently, the PCR assay was done multiple times on each of the nine samples that revealed the presence of a single allele (8). As demonstrated by differential sensitivity to Bsg I digestion, three samples screened repeatedly revealed only the CAST/Ei allele, whereas six revealed only the BALB/c allele [Fig. 1B and (9)]. These data were consistent with monoallelic expression and prompted further studies with cloned cells.

Analysis of cloned cells allows the direct investigation of gene expression in individual cells and thus avoids the statistical imprecisions of the limiting-dilution approach. We examined a panel of 30 CD4+ T cell clones generated from (BALB/c × CAST/Ei)F1 hybrid mice by stimulation with allogeneic H-2b cells in the presence of recombinant IL-4 and IL-12 monoclonal antibody (mAb), conditions that favor the establishment of IL-4–producing clones (10). Of these alloreactive clones, 25 expressed IL-4, and 12 could be expanded and maintained long-term. Seventeen hours after activation of resting clones with immobilized mAbs to the T cell receptor (TCR) and CD28 (11), RNA was isolated and screened for the parental origin of the IL-4 transcripts. Twelve (48%) of the 25 clones revealed monoallelic expression (eight BALB/c, four CAST/Ei) and 13 revealed biallelic expression [Fig. 2A and (9)]. The twofold bias in favor of monoallelic expression of the BALB/c rather than the CAST/Ei allele was not statistically significant. These data confirmed the suggestion from the limiting-template analysis that monoallelic IL-4 expression does occur. Because some clones expressing IL-4 from either one or both alleles were obtained from a single animal [experiment A (10)], the data suggested a process distinct from parental imprinting wherein the same parental allele is used by almost all expressing cells of the individual. Control experiments, in which reconstitutions of discrete mRNA or cDNA pools were used, confirmed that the reverse transcriptase (RT)–PCR assay could discriminate at least a 16- to 32-fold difference in transcript abundance (Fig. 2B). Simple sequence length polymorphisms on either side of the IL-4 gene were used to confirm that both parental chromosomes were maintained in the clones that expressed single alleles(12).

Figure 2

Monoallelic IL-4 expression in CD4+ T cell clones. (A) Four CD4+ T cell clones derived from (BALB/c × CAST/Ei)F1 mice were analyzed as described in the legend to Fig. 1. Groupings of three consecutive lanes represent IL-4 amplification products resolved in the absence or presence of the indicated restriction endonucleases. (B) Reconstruction experiments to assess fidelity of the RT-PCR amplification procedure. Either mRNA (Mix RNA) or reverse-transcribed cDNA (Mix cDNA) derived from the stable monoallelic IL-4–expressing clones 3G6 (CAST/Ei allele) and 1D5 (BALB/c) was mixed at the indicated ratios and used to template the RT or PCR assays. The resulting IL-4 amplification products were analyzed with or without the designated restriction enzymes before resolution on agarose gels.

The detection of a single IL-4 allele from the RNA pooled from more than 104 F1 CD4+ T cells suggested that most (and perhaps all) of the cells in a given clone expressed the same allelic pattern. To examine this issue, we reanalyzed 12 clones, representing all three possible allelic expression patterns, by sequential sampling over an extended period of continuous growth (Table 1) (11). The allelic expression pattern for each clone remained constant. Thus, as early as 23 days after primary stimulation, the allelic expression pattern had become fixed as a heritable, epigenetic trait.

Table 1

Stability of IL-4 allelic expression patterns over time in individual clones.

View this table:

The experiments here suggest that IL-4 belongs to a category of genes whose allelic expression pattern is random and established developmentally late (13). By analogy with some other members of this category, which belong to gene families (7,14), we speculate that other cytokine genes also might be probabilistically expressed. The IL-2 gene was shown to be expressed monoallelically in CD4+ T cells (15). A relatively small number of individual cells were analyzed at an early time after activation, leaving open the possibility that biallelic expression for IL-2 may also occur. Indeed, expression of LY49 NK cell receptors was first reported to be allelically excluded and only later shown to be biallelic in some cells, consistent with a probabilistic expression mechanism (16). We have preliminary evidence that granulocyte-macrophage–colony-stimulating factor can also be expressed from one or both alleles in CD4+ T cells (9). Thus, at least some cytokine genes are regulated by a mechanism that treats the alleles independently.

Why might cytokines be expressed in a probabilistic manner? Such a mechanism would allow combinatorial assortment of distinct cytokine genes among the clonal progeny of individual precursor THcells sharing the same antigen receptor specificity. Once the probabilistic gene-activation mechanism ceases, the cytokine expression pattern among the clonally related but phenotypically diverse daughter cells would become fixed as a heritable epigenetic trait. Depending on microenvironmental signals like IL-12 or IL-4, selective growth or death could act to influence the prevalence of cells that express distinct cytokine patterns, presumably in a manner promoting successful resolution of different immunological challenges. Specific cytokines would thus act to support the survival and proliferation of committed cells rather than to mediate effector commitment, consistent with the role of cytokine growth factors in lineage commitment of hematopoietic precursors (17). This process may underlie a fundamental strategy by which the immune system ensures that a diverse repertoire of cytokine-producing effector cells can be generated from limited numbers of antigen-specific precursors, thus allowing selection of the appropriate immune response to any given pathogen.

  • * To whom correspondence should be addressed at UCSF, Box 0654, C-443, 521 Parnassus Avenue, San Francisco, CA 94143–0654, USA. E-mail: locksley{at}


View Abstract

Navigate This Article