Identification of a Coordinate Regulator of Interleukins 4, 13, and 5 by Cross-Species Sequence Comparisons

See allHide authors and affiliations

Science  07 Apr 2000:
Vol. 288, Issue 5463, pp. 136-140
DOI: 10.1126/science.288.5463.136


Long-range regulatory elements are difficult to discover experimentally; however, they tend to be conserved among mammals, suggesting that cross-species sequence comparisons should identify them. To search for regulatory sequences, we examined about 1 megabase of orthologous human and mouse sequences for conserved noncoding elements with greater than or equal to 70% identity over at least 100 base pairs. Ninety noncoding sequences meeting these criteria were discovered, and the analysis of 15 of these elements found that about 70% were conserved across mammals. Characterization of the largest element in yeast artificial chromosome transgenic mice revealed it to be a coordinate regulator of three genes, interleukin-4,interleukin-13, and interleukin-5, spread over 120 kilobases.

Computational methods for recognizing coding sequences in genomic DNA are capable of detecting most genes; however, identifying regulatory elements in the 95% of the genome composed of noncoding sequences is currently a substantial challenge. Regulatory sequences tend to be highly conserved among mammals (1), suggesting comparative sequence analysis as a strategy for their identification (2). Extensive studies focusing on understanding the regulation of three biomedically important cytokines [interleukin-4(IL-4), IL-13, and IL-5] clustered at human 5q31 have indicated that these genes are coordinately coactivated in type 2 T helper (TH2) cells (3). Transgenic mice propagating human 5q31 yeast artificial chromosomes (YACs) appropriately regulate the humanIL-4, IL-13, and IL-5 transgenes in murine TH2 cells, independent of the site of integration (4). These data indicate that the elements regulating the coordinate tissue and temporal expression of these cytokines are conserved in humans and mice. A comparative sequence-based approach was employed to discover distant regulatory sequences involved in controlling the complex expression patterns of the 5q31 cytokines.

Database searches combined with GenScan predictions identified 23 putative genes (including IL-4, IL-13, andIL-5) in an ∼1-Mb human 5q31 region and determined that the order and orientation of all but one of these genes are conserved in the murine chromosome 11 orthologous region (Fig. 1) (5–7). Comparative analysis of these human and mouse sequences (8) focused on discovering distant regulatory sequences in the region, based on the observation that these elements are typically composed of long sequences [≥100 base pairs (bp) in length] that are highly conserved among mammals (≥70% identity) (1). A total of 245 conserved elements fitting these criteria was identified, of which 155 overlapped with coding sequences (defined as sequences present in mature RNA transcripts) (9), and the remaining 90 were defined as noncoding (Figs. 1 and 2) (10). Of the 90 conserved noncoding elements, 46% were in introns, 9% were within 1 kb of the 5′ and 3′ ends of an identified transcript, and 45% were in intergenic regions >1 kb from any known gene. Many of the noncoding elements were found in clusters, such as in the intergenic region between organic cation transporter 1(OCTN1) and P4–hydroxylase alpha(II), suggesting that they may be working cooperatively as a functional unit (Fig. 1). One of the conserved noncoding sequences (CNSs), CNS-7, located in the intergenic region between granulocyte-macrophage colony-stimulating factor(GM-CSF) and IL-3, had previously been identified experimentally as an enhancer controlling the coregulation of these two cytokines (11). This finding supports the choice of the criteria used in this sequence-based approach to identify biologically relevant noncoding sequences.

Figure 1

Physical map of the 1-Mb human 5q31 region. The locations of the 23 genes in the interval (7) (gray boxes), their direction of transcription (gray vertical arrows), and the locations of all CNSs (red horizontal arrows) are shown. A number next to a horizontal arrow indicates that more than one CNS was found at that location. The lengths and percent identities of the human and mouse alignments of the 15 noncoding sequences investigated in this study are given. The 12 elements determined to be single copy in the human genome are in bold. Noncoding sequences conserved in other vertebrates (Table 1) are marked with an asterisk. The locations of the human YAC (A94G6) and P1 clones (H23 and H24) (vertical bars) used in this study are shown.

Figure 2

Percent identity plot displaying locations for CNS elements in the IL-4 through IL-13region. Conserved sequences are plotted in relation to their coordinates in the human genome (horizontal axes), and their percent identities are indicated on the vertical axis (9). Sequences defined as coding (blue) (8) and noncoding (≥100 bp and ≥70%) (red) are indicated. A horizontal arrow indicates the direction of transcription for each gene. The two DNase I HSs that overlap with CNS-1 are shown as vertical arrows. k, kilobases.

Fifteen of the CNSs were assessed for their presence in other vertebrates and their copy number in the human genome. Degenerate primer pairs of the elements were used to amplify genomic DNA of other vertebrates, and the resulting products were sequenced (12). Ten of the elements were highly conserved in at least two mammals in addition to humans and mice (Fig. 1 and Table 1). Twelve elements appeared unique in the human genome, as determined by low-stringency Southern blot hybridizations (13). Of the noncoding sequences examined, ∼70% are conserved across mammals and unique in the human genome, features commonly noted in experimentally identified distant regulatory elements (1).

Table 1

Cross-species sequence analysis of 15 conserved noncoding elements. Human and mouse sequence alignments of each element were inspected, and primer pairs were chosen from the most conserved regions. Hence, for each CNS, the size of the PCR amplified product(s) is smaller than the size of the element indicated in Fig. 1. Numbers in parentheses are percent identity. Dashes indicate that PCR did not amplify a product.

View this table:

The largest conserved noncoding sequence, CNS-1 (401 bp), which is located in the intergenic region (∼13 kb) between IL-4and IL-13, was chosen for in-depth functional analysis on the basis of several features that suggest it might be a distant regulatory element. CNS-1 demonstrates a high degree of conservation across mammals (∼80% identity in mice, humans, cows, dogs, and rabbits) (Table 1), contrasting sharply with the relatively low conservation observed in the coding regions of the flanking genes,IL-4 and IL-13 (∼50% identity between humans and mice) (Fig. 2). This element is single copy in the human genome and has been conserved during evolution not only with regards to sequence but also to genomic location [mapped in dogs, baboons, humans, and mice to the IL-4 through IL-13 intergenic region (14)]. Binding sites for transcription factors known to regulate the expression of IL-4 andIL-13 were not found in CNS-1 (15); however, we determined that CNS-1 overlaps with two (TH2 cell specific) of the eight deoxyribonuclease (DNase) I hypersensitive sites (HSs) previously localized in the IL-4 throughIL-13 region (16) (Fig. 2).

The biological properties of CNS-1 were characterized through the creation and analysis of multiple lines of mice bearing a 450-kb human YAC transgene (Fig. 1) either containing or lacking the CNS-1 element. To reduce the uncertainties of comparing different founder lines of transgenic mice, we inserted loxP sites (17) into the YAC transgene flanking the CNS-1 element and introduced this modified YAC into the genome of mice (three separate founder lines were created). To delete the CNS-1 element, we bred the human YAC transgenic mice from each of the founder lines with mice expressing a Cre recombinase transgene (18). This resulted in the generation of two lines of transgenic mice derived from each founder line, one with a CNS-1–containing YAC transgene (CNS-1wt) and one with a YAC transgene in which the CNS-1 element had been deleted (CNS-1del). The deletion of CNS-1 was the only rearrangement in the human IL-4 throughIL-13 region detected by Southern blot analysis (19). YAC copy number was the same in the paired CNS-1wt and CNS-1del transgenic animals as determined by both Southern blot analysis (19) and fluorescent in situ hybridization (FISH) (Fig. 3) (20). These findings suggest that the paired CNS-1wt and CNS-1deltransgenic mice are genetically identical in every aspect except for the presence or absence of the CNS-1 element.

Figure 3

YAC copy number determined by FISH. Interphase nuclei of spleen cells isolated from paired (top) CNS-1wt and (bottom) CNS-1del YAC transgenic mice (line 1) hybridized with two P1 probes [H23 (green) and H24 (red)] (Fig. 1) as described (19) are shown. Similar hybridization patterns were obtained for line 3.

To assess whether CNS-1 affects expression of the human 5q31 cytokines, we isolated highly purified naı̈ve CD4+ T cells from the spleen and lymph nodes of paired CNS-1wt and CNS-1del transgenic mice and stimulated the cells in vitro under conditions that favor development of either TH1 or TH2 cells (21, 22). Neither theIL-4 nor IL-13 human transgenes were expressed during TH1 differentiation in either the CNS-1wt or CNS-1del transgenic cells, suggesting that CNS-1 was not required for the repression of these cytokines in TH1 cells. When T cells were stimulated to promote TH2 differentiation, the CNS-1del transgenic cells developed less than half as many human IL-4–producing cells and less than a third as many human IL-13–producing cells as the CNS-1wt transgenic mice did (Fig. 4, A and B) (22). In the TH2 cells that expressed human IL-4 and IL-13, however, the amounts produced per cell, assessed by the mean fluorescence intensity, were the same in the paired human CNS-1wt and CNS-1del YAC transgenics. The observation that CNS-1 influences the number of TH2 cells expressing IL-4 and IL-13 but does not influence their levels of expression per cell suggests that this element does not act as a classical enhancer but rather appears to be involved in modulating chromatin structure. Production of human IL-5 was also significantly reduced in the CNS-1del transgenic cells as compared with the paired CNS-1wt transgenic cells (Fig. 4C). The lack of suitable antibodies for flow cytometry detection of human IL-5 precluded us from distinguishing whether the reduced expression of human IL-5 was due to a decrease in the number of cells producing it or due to a decrease in the amounts produced by individual TH2 cells. All three paired CNS-1wt and CNS-1del YAC transgenic lines were examined, and the absence of CNS-1 consistently resulted in a decrease in the number of TH2 cells expressing human IL-4 and IL-13.

Figure 4

Expression of human and mouse cytokines in paired CNS-1wt and CNS-1del transgenics and control FVB mice. Naı̈ve CD4+ T cells were stimulated with TH2 conditions and analyzed on days 2, 3, 4, and 7 for expression of (A) human IL-4, (B) human IL-13, and (D) mouse IL-4 by intracellular cytokine detection with fluorescent monoclonal antibodies (21,22). (C) Human IL-5 and (E) mouse IL-13 protein levels were determined by ELISA with supernatants of activated T cells collected at either 3 or 7 days (following restimulation) after stimulation of naı̈ve cells (21,22). Bars represent means and standard errors of the means.

Transcript levels of human IL-4, IL-13, andIL-5, as well as the two noninterleukin genes closest to CNS-1, kinesin family member 3A (KIF3A) andRAD50, were quantified in the paired CNS-1wt and CNS-1del transgenic mice (23). The mRNA levels of the three human cytokines were reduced in CNS-1delTH2 cells, reflecting the differences observed in the protein levels and indicating that CNS-1 acts through its effect on the transcriptional activity of these genes. Expression levels of humanKIF3A and RAD50 were essentially the same in the brains, hearts, kidneys, livers, and isolated TH2 cells of the paired CNS-1wt and CNS-1deltransgenic mice. RAD50 is a large gene (spanning 87 kb) located between CNS-1 and IL-5 (Fig. 1). The fact that the CNS-1del transgenic mice produce substantially less humanIL-5, but unaltered amounts of RAD50, suggests that CNS-1 acts over a large genomic interval to specifically affect the expression of TH2 cell–specific cytokines.

The initial production of murine IL-4 and IL-13 was significantly less in CNS-1wt transgenic TH2 cells, whereas at later time points, CNS-1wt and CNS-1del transgenic TH2 cells expressed comparable amounts of these cytokines (Fig. 4, D and E). Human IL-4 and IL-13 do not modulate murine TH2 cell development in vitro, as determined with CD4+ T cells from nontransgenic mice (24). These data indicate that the initial reduction in murine IL-4 and IL-13 production in the CNS-1wt transgenic cells is probably not due to the expression of the human IL-4 and IL-13 transgenes, suggesting that a competitive interaction may exist between the human CNS-1 element and a trans-acting murine factor(s). The observation that, at subsequent time points, the murine cytokines are expressed at equivalent levels in CNS-1wt and CNS-1del transgenic cells may be due either to alterations in the levels of such putative trans factor(s) with increasing cell division or to preferential expansion or survival of murine IL-4– and IL-13–producing cells.

Our study illustrates the utility of comparative sequence analysis in the identification of distant regulatory elements. Of the 15 human and mouse conserved noncoding elements we examined, most are also present in other mammals. These data, in combination with analysis of the orthologous human 5q31 interval in dogs revealing a strikingly similar pattern of CNS elements (25), suggest that most of the human and mouse CNS elements we identified have been actively conserved because of a biological function. CNS-1 appears to be involved in gene activation by modulating chromatin structure, a function for which there are no standard in vitro assays, suggesting that this element would have been difficult, if not impossible, to identify with traditional experimental methods. The use of a 450-kb YAC transgene containing eight human genes, each potentially serving as “a reporter,” allowed us to determine that CNS-1 regulates some, but not all, of the genes in a large genomic interval (120 kb) on human 5q31. These regulatory features of CNS-1 reveal the complexity of long-range regulatory elements and the power of comparative biology in discovering and deciphering the properties of such elements.

  • * To whom correspondence should be addressed. E-mail: kafrazer{at} (K.A.F.) and emrubin{at} (E.M.R.)


View Abstract

Navigate This Article