Skip to main content
  • Loading metrics

Distinct Functional Constraints Partition Sequence Conservation in a cis-Regulatory Element

  • Antoine Barrière ,

    Contributed equally to this work with: Antoine Barrière, Kacy L. Gordon

    Affiliation Department of Ecology and Evolution and Institute for Genomics and Systems Biology, Chicago, Illinois, United States of America

  • Kacy L. Gordon ,

    Contributed equally to this work with: Antoine Barrière, Kacy L. Gordon

    Affiliation Department of Organismal Biology and Anatomy, The University of Chicago, Chicago, Illinois, United States of America

  • Ilya Ruvinsky

    Affiliations Department of Ecology and Evolution and Institute for Genomics and Systems Biology, Chicago, Illinois, United States of America, Department of Organismal Biology and Anatomy, The University of Chicago, Chicago, Illinois, United States of America


Different functional constraints contribute to different evolutionary rates across genomes. To understand why some sequences evolve faster than others in a single cis-regulatory locus, we investigated function and evolutionary dynamics of the promoter of the Caenorhabditis elegans unc-47 gene. We found that this promoter consists of two distinct domains. The proximal promoter is conserved and is largely sufficient to direct appropriate spatial expression. The distal promoter displays little if any conservation between several closely related nematodes. Despite this divergence, sequences from all species confer robustness of expression, arguing that this function does not require substantial sequence conservation. We showed that even unrelated sequences have the ability to promote robust expression. A prominent feature shared by all of these robustness-promoting sequences is an AT-enriched nucleotide composition consistent with nucleosome depletion. Because general sequence composition can be maintained despite sequence turnover, our results explain how different functional constraints can lead to vastly disparate rates of sequence divergence within a promoter.

Author Summary

Comparison between genome sequences of different species is a powerful tool in modern biology because important features are maintained by natural selection and are therefore conserved. However, some important sequences within genomes evolve considerably faster than others. One possible explanation is that they encode little or no function. Alternatively, they may evolve under different constraints that permit sequence turnover while maintaining function. Here we report that the promoter of the unc-47 gene of C. elegans contains two discrete elements. One has a highly conserved sequence that determines the spatial expression pattern. Another shows no sequence conservation, but it makes expression of the gene robust, that is, consistent between individuals and resilient to environmental challenges. Remarkably, multiple unrelated sequences are capable of promoting robust expression. Nucleotide composition of these sequences suggests that open chromatin may play a role in conferring robustness of gene expression. Because general sequence composition and therefore expression robustness can be maintained despite sequence turnover, our results offer an explanation of how rapidly diverging promoter elements can nevertheless remain functionally conserved.


The advent of genome sequencing introduced the practice of searching for regulatory elements in evolutionarily conserved regions [1][5]. However, functional elements are by no means strictly confined to regions of high primary sequence conservation [6][9]. In fact, cis-regulatory elements can retain functionality over great evolutionary distances despite sharing little or no identifiable sequence similarity, and can correctly drive reporter gene expression when placed in a distantly related species [10][12].

Two questions arise from these observations. First, how do different functional constraints account for different degrees of sequence conservation? Whereas the relationship between function and sequence conservation in not well understood in general, this problem is particularly acute for cis-elements [13]. A major obstacle is that we do not have a cis-regulatory code akin to that for protein-coding sequences. For example, even within conserved cis-regulatory elements there are interspersed nonconserved sequences that seem to be important for their function [12], [14][18]. In other cases, cis-regulatory architecture can be cryptically conserved despite sequence divergence [19][21]. In yet other promoters, not even the architecture appears to be conserved [22], [23].

Second, since gene expression is increasingly considered to be a quantitative trait for which populations vary [24], [25], functional comparisons of regulatory elements ought to be made with quantitative measurements across populations of individuals [26] or cells [27]. Only then can expression patterns be compared in terms of how much they differ, and how intrinsically variable they are.

Variation of gene expression can take many forms, for instance the number of cells expressing a gene or the amount of transcript made in individual cells. Despite variation, gene expression, like many other biological processes, exhibits substantial robustness, that is, resilience to perturbations by genetic and environmental challenges [28][31]. Robustness of expression, much like pattern of expression, is encoded in regulatory elements [32], [33]. One way of encoding robustness in cis is with redundant or “shadow” enhancers [34]. The loss of one “shadow” enhancer does not substantially perturb gene expression, unless the organism is challenged by genetic or environmental stresses [35], [36]. Another documented mechanism that confers robustness in cis is the presence of miRNA target sites in 3′ UTR [37], [38].

Our goal is to understand the relationship between function and sequence evolution in a single cis-element. We studied a promoter of the Caenorhabditis elegans unc-47 gene, which drives a simple, easily quantifiable expression pattern. This promoter contains regions of high and low sequence conservation when compared to orthologs from four closely related [39] Caenorhabditis nematodes. We quantified functional similarities and differences of these promoters to infer the constraints that gave rise to the observed patterns of sequence evolution.


Conserved cis-elements recapitulate qualitative aspects of the expression pattern

We first tested the hypothesis that an evolutionarily conserved expression pattern results from evolutionarily conserved regulatory sequences alone. In C. elegans, the unc-47 gene is expressed in all 26 GABAergic neurons, including 19 D-type neurons of the ventral nerve cord and the postanal cell DVB (Figure 1A) [40]. We selected these cells because they are easy to recognize due to a characteristic morphology, and they reside close to the body surface, thus easing the quantification of expression. The endogenous pattern of unc-47 is recapitulated when a reporter construct containing a 1.2 kb sequence immediately 5′ of the gene (we refer to it as a full-length promoter as it extends to the locus of the upstream gene) is used to drive green fluorescent protein (GFP) in C. elegans (Figure 1B). A construct containing a promoter of the same length of the C. briggsae unc-47 ortholog is expressed in a qualitatively indistinguishable pattern in C. briggsae (Figure 1C). Indeed the C. briggsae promoter drives expression in the same neurons even in C. elegans (see below). These results suggest that expression patterns of unc-47 orthologs have been conserved since their common ancestor and that the information required for driving proper expression is contained within ∼1.2 kb promoters upstream of the genes.

Figure 1. Conserved portions of unc-47 promoters direct spatially correct, but not robust, expression.

(A) A schematic depiction of GABAergic neurons in C. elegans. (B) Expression pattern of C. elegans promoter unc-47::GFP in C. elegans. (C) Expression pattern of C. briggsae promoter unc-47::GFP in C. briggsae. Photographs of worms are composites of multiple exposures of the same individual that capture the full complement of D-type neurons, in all focal planes, expressing GFP. (D) Vista plot of primary sequence conservation in the promoter region of unc-47 from C. briggsae, C. remanei and C. brenneri aligned to C. elegans. Window size = 20 bp, threshold = 70%. (E) Schematic depiction of full-length and proximal promoters. Consistency of GFP expression in D-type neurons for full-length and proximal promoters from C. elegans (F) and C. briggsae (G). For both C. elegans and C. briggsae, the average number of cells expressing GFP is lower for the proximal promoter compared to full-length (Wilcoxon test, p = 3.2×10−3 and p = 3.4×10−8, respectively). (H) Distribution of ratios of GFP expression intensity in DVB relative to D-type neurons for the full-length and proximal promoters at 20°C. Each strain is represented by 100 animals. The two distributions are significantly different (Wilcoxon test, p = 1.8×10−5; Kolmogorov-Smironov test, p = 1.1×10−8).

Because expression patterns of nematode unc-47 orthologs are conserved, we investigated whether expression is mediated solely by conserved cis-regulatory elements. We aligned the C. briggsae sequence along with those of two other close relatives C. brenneri and C. remanei, to the C. elegans unc-47 promoter. As reported previously [41], sequence conservation in this promoter is heavily biased to the most proximal ∼250 bp (Figure 1D, Figure S1, Table S1). We carried out extensive analyses which showed that little sequence conservation can be found distal to the ∼250 bp boundary (Figure S2, Table S1). This does not exclude the possibility that there exist short and conserved motifs in the distal promoter; they are simply below our level of detection. Some may exist and even be functional; nonetheless, the rates of sequence divergence are profoundly different between the proximal and the distal portions of this promoter.

If the conserved expression patterns result from solely the conserved portions of the cis-regulatory elements, then the proximal promoters of both C. elegans and C. briggsae should be sufficient to recapitulate the entire pattern. We therefore compared functions, in C. elegans, of both full-length and proximal promoters (Figure 1E) derived from C. elegans and C. briggsae. Strains bearing each of these four constructs exhibited qualitatively similar patterns of expression. However, we noticed that the proximal C. briggsae promoter was not robust – it drove both weak and inconsistent expression. In contrast, a robust promoter would express strongly and consistently, as do the full-length promoters. We next quantified and compared expression patterns driven by these promoters.

Regulatory elements lacking sequence conservation are required for robust gene expression

The expression patterns driven by the proximal and full-length promoters from both species were qualitatively correct, that is, all cells that were expected to show reporter gene expression were GFP-positive in at least some of the examined animals. To obtain a precise measure of variability, we counted the number of D-type neurons that were expressing GFP in 200 individuals bearing each construct. We examined animals from multiple independent strains for each construct and found that overall inter-strain variance was modest for all constructs (data not shown). We conducted the counts in a blinded fashion to exclude the possibility of unconscious experimenter bias (see Methods). The results of these counts address the first aspect of robustness – consistency of expression pattern. We found that the full-length C. elegans promoter drove somewhat more consistent pattern than the proximal promoter (Figure 1F; Wilcoxon test, p = 3.2×10−3), and that the full-length promoters of C. elegans and C. briggsae were indistinguishable (p = 0.7). The C. briggsae proximal promoter was not expressed as consistently in D-type neurons as the full-length promoter (Figure 1G; Wilcoxon test, p = 3.4×10−8).

In a parallel approach we quantified the intensity of GFP fluorescence in DVB and D-type neurons. This allowed us to assess the second aspect of robust expression – consistency of relative expression levels from one cell type to another within an individual. Expression levels in D-type neurons and DVB were relatively similar in animals carrying the full-length promoter (note the mean ratio of one and a tight, normal scatter, Figure 1H). In contrast, individuals with the proximal promoter exhibited a significant increase in variance (Ansari-Bradley test, p = 1.6×10−3), despite a lower relative expression in DVB (Wilcoxon test, p = 1.8×10−5). We thus concluded that the C. briggsae proximal promoter directs less robust expression than the full-length promoter.

To ensure that the apparent decrease in robustness of the proximal promoter was not an artifact of using extrachromosomal arrays, we generated transgenic strains in which single-copy full-length or proximal promoters were integrated into the same genomic location. Whereas the absolute levels of expression were considerably lower for all integrated strains (20–400 fold), the shorter promoter was weaker than the full-length (4–6 fold) and significantly less consistent in its expression (Figure S3; Wilcoxon test, p = 1.9×10−10). Thus the shorter promoter was weaker and less consistently expressed regardless of whether it was tested as an integrated or extrachromosomal transgene. This concordance allowed us to utilize extrachromosomal transgenes for the remainder of this study, because integrated strains showed weak expression that was at the limit of detection.

It is formally possible that our observation of the decreased robustness of the proximal promoter compared to the full-length version was due to a peculiar nature of the C. briggsae regulatory sequence. We therefore tested orthologous cis-regulatory sequences of two additional species, C. brenneri and C. remanei, in C. elegans. Their full-length promoters drove GFP in a strong and consistent pattern, statistically indistinguishable from those of C. elegans and C. briggsae orthologs (Figure 2A, 2B). Both proximal promoters, truncated at the orthologous position at the boundary of conserved sequences around 250 bp, directed weaker (Figure 2C, 2D) and less robust (Figure 2E, 2F; Wilcoxon test, C. brenneri p = 1.2×10−13 and C. remanei p = 1.3×10−13) expression in D-type neurons. Expression of the proximal promoters was also less consistent in the tail neuron DVB (Figure 2G, 2H).

Figure 2. Promoters of C. brenneri and C. remanei unc-47 require the distal sequences for consistent expression.

GFP expression driven by full-length C. brenneri (A) and C. remanei (B) and proximal C. brenneri (C) and C. remanei (D) unc-47 promoters in a C. elegans host. As in Figure 1F, 1G, 200 individuals bearing each transgene were counted, and the percentages of those individuals with the indicated number of D-type neurons expressing GFP is shown (E–F). (G–H) Presence/absence of GFP expression in the cell DVB in the same 200 individuals for each of the four promoters. Photographs of worms are composites of multiple exposures of the same individual that capture the full complement of D-type neurons, in all focal planes, expressing GFP.

Our results suggest that the cis-regulatory elements of unc-47 from the four examined nematodes have similar architectural properties – the proximal, highly conserved promoter is sufficient to deliver the qualitatively correct expression pattern, whereas the distal, nonconserved portion is required for consistent expression. It is important to note that this distal sequence is not alone sufficient to direct any expression in D-type neurons or DVB [41]. It therefore contributes to robustness via a mechanism different from that of recently described “shadow” enhancers [35], [36], each of which is sufficient to drive expression independently. Furthermore “shadow” enhancers are conserved, whereas the distal promoter of unc-47 is not.

Distal nonconserved promoter sequences are required to confer environmental robustness

Distal promoters were required for stronger and more consistent expression, even when worms were reared under constant and nearly optimal growth conditions (20°C). We tested whether these sequences could also buffer against environmental challenges. We compared GFP expression levels directed by the full-length and proximal promoters in worms reared at a high temperature of 26°C and a low of 15°C. We measured the intensity of GFP-fluorescence in D-type neurons and DVB and observed several trends. First, expression levels driven by the full-length C. elegans promoter (Figure 3A) were more consistent than those driven by the proximal promoter (Figure 3B) at both the 26°C (Kolmogorov-Smirnov test, p = 2.9×10−5) and 15°C (p = 1.2×10−6). Second, the full-length promoter was comparably consistent in its expression at 26°C and 15°C (Figure 3A, Table S2). In contrast, consistency of expression of the proximal promoter differed dramatically between the two temperatures (Figure 3B, Table S2).

Figure 3. Only full-length promoters of C. elegans and C. briggsae unc-47 can direct robust expression.

Distribution of fluorescence intensity driven by C. elegans full-length (A) and proximal (B), and C. briggsae full-length (C) and proximal (D) promoters in C. elegans at two temperatures (red for 26°C and blue for 15°C). For each individual, the log intensity in D-type neurons is plotted against the log intensity in DVB. Individuals that did not show any fluorescence in DVB were excluded from analysis. Data for additional strains are given in Figure S4. Superimposed on each graph is a schematic of the construct used: a straight line represents C. elegans promoter of unc-47, a wavy line represents C. briggsae promoter of unc-47. The gray vertical bar indicates the 5′ boundary of the proximal promoter.

Similar results were observed for the C. briggsae promoters. The full-length promoter (Figure 3C) directed more consistent expression than the proximal promoter (Figure 3D) at both temperatures (Kolmogorov-Smirnov test, at 26°C p = 1.2×10−2, at 15°C p = 2.2×10−14). Temperature had a minor effect on the consistency of expression of the full-length promoter, but a more substantial effect on the proximal promoter (Table S2). We repeated measurements for multiple independent strains carrying full-length and proximal promoters from C. elegans and C. briggsae and observed concordant results (Figure S4, Table S2).

We concluded that full-length promoters are more robust to temperature stress, regardless of their species of origin (compare Figure 3A and 3C). Proximal promoters, primarily composed of conserved sequences, were significantly less robust, particularly after the cold treatment (Figure 3B and 3D). These results indicate that a robustness-conferring function is encoded in distal promoters in both species, and is thus conserved despite the lack of detectable sequence conservation.

Distinct sequences in distal promoters can contribute to robust expression

We dissected the distal promoters to determine which of their components were necessary for robust expression. The proximal promoters contain all of the densely arranged blocks of sequence conservation. Additionally, a pair of short motifs (8 and 6 bp) that is shared by all four examined nematodes is located approximately 50 bp distal to the boundary of greatest conservation (1). We considered the distal extent of these motifs to be the absolute boundary of the evolutionarily conserved promoter sequence, because in the remaining distal promoter there were no sequences longer than 10 bp that were shared by all four species. We tested a promoter encompassing all of this “extended conservation” for the ability to drive robust expression. It performed intermediately in terms of consistency of expression between the full-length C. briggsae promoter (Figure 4A; Wilcoxon test, p = 1.4×10−2) and the proximal promoter alone (p = 4.0×10−3). We next examined intensity of GFP expression in the D-type neurons and DVB in animals reared under temperature stress. At 15°C, although not at 26°C, this promoter produced more variable expression than the full-length C. briggsae promoter (compare Figure 4B and Figure 3C; Kolmogorov-Smirnov test, p = 5.2×10−4), but significantly less variable expression than the proximal promoter (compare Figure 4B and Figure 3D; p = 7.7×10−5). Therefore the two conserved motifs and the sequences that surround them contribute to, but do not entirely account for the robustness of the longer promoter.

Figure 4. Different components of the distal promoter sequences regulate consistent expression and expression levels under temperature stress.

(A) Percentage of 200 individuals expressing GFP in the indicated number of D-type neurons under control of C. briggsae promoter with extended conservation, shown in solid black bars compared to C. briggsae full-length (black hashed bars) and proximal (gray hashed bars) promoters. (B) Intensity of GFP expression in D-type neurons and the cell DVB for animals bearing the extended conservation promoter reared at 26°C (red) or 15°C (blue). (C) Percentage of 200 individuals expressing GFP in the indicated number of D-type neurons under control of chimeric promoter fusion of C. elegans distal unc-47 promoter sequence and C. briggsae proximal promoter. For comparison, distributions for C. briggsae full-length and proximal promoters are shown in black and gray hashes, respectively. (D) Intensity of GFP in D-type neurons and the cell DVB for animals bearing the chimeric promoter reared at 26°C (red) or 15°C (blue). The chimeric promoter drives robust expression under temperature stress. (E) Percentage of 200 individuals expressing GFP in indicated number of D-type neurons from a chimeric promoter composed of distal C. briggsae unc-15 sequence and the C. briggsae unc-47 proximal promoter (black bars). For comparison, C. briggsae unc-47 full-length and proximal promoters are shown in black and gray hashed bars, respectively. The unc-15/unc-47 chimera is indistinguishable from the C. briggsae full-length promoter (Wilcoxon test p = 0.37), and it is significantly more consistent than the proximal promoter (Wilcoxon test p = 1.3×10−5). (F) Robustness of unc-15/unc-47 chimeric promoter under temperature stress.

Our results suggest that, despite substantial sequence divergence, distal promoters of C. elegans and C. briggsae unc-47 confer robust expression to their respective proximal promoters (Figure 1F, 1G, Figure 3). To test whether distal promoters confer robustness in a species-specific manner, we asked whether the distal promoter of C. elegans could restore robust expression when fused to the proximal promoter of C. briggsae. We reasoned that if the distal and proximal sequence function as a unit and make up a single cis-regulatory element, the distal part of which has diverged considerably in its sequence, we should expect a chimeric construct not to rescue robustness. If, on the other hand, the proximal, highly conserved promoter and the distal promoter are two distinct functional units, they should be modular.

The C. elegans-distal-C. briggsae-proximal chimeric unc-47 promoter drove expression with a consistency intermediate between the full-length and proximal promoters in terms of cell number (Figure 4C; Wilcoxon test, different from C. briggsae full-length p = 8.0×10−3; different from C. briggsae proximal p = 5.6×10−3). However, at both 15°C and 26°C this promoter was no more variable than the full-length C. briggsae construct (Figure 4D; Kolmogorov-Smirnov test, at 26°C p = 0.6, at 15°C p = 0.1), constituting a significant rescue of robustness relative to the proximal promoter alone (Kolmogorov-Smirnov test, at 26°C p = 4.6×10−4, at 15°C p = 1.2×10−10). Because much, although perhaps not all, of the robustness of expression can be rescued by this chimeric construct, we conclude that the proximal and distal sequences encode distinct and separable regulatory functions. Multiple chimeric and “extended conservation” constructs were consistent with these results (Figure S5, Table S2).

The robustness function of the distal element must have much less stringent sequence requirements than the proximal promoter, because distal sequences have diverged considerably but maintain this function. We next tested whether another genomic fragment lacking detectable sequence similarity to the distal unc-47 sequences could confer robustness of expression. We selected an approximately 1.3 kb fragment upstream of unc-15 because it does not share significant similarity with the C. briggsae unc-47 distal promoter (Figure S2). Furthermore, unc-15 encodes a paramyosin ortholog that is expressed in muscles [42], and thus is not expressed in any of the same cells as unc-47. The overall length of this sequence is comparable, however, and it is also an intergenic sequence as poorly conserved between C. elegans and C. briggsae as is the distal portion of the unc-47 promoter (data not shown).

We were surprised to find that the chimeric promoter containing this distal C. briggsae unc-15 sequence fused to the proximal C. briggsae unc-47 promoter displayed robust expression as consistent as the full-length C. briggsae unc-47 promoter in terms of cell number (Figure 4E; Wilcoxon test, p = 0.37). We observed markedly improved consistency of the expression pattern over the C. briggsae proximal promoter alone (Figure 4F; difference from proximal promoter, Wilcoxon test, p = 1.3×10−5). At 26°C this promoter drove as consistent expression as the full-length C. briggsae promoter (Kolmogorov-Smirnov test, at 26°C p = 0.1; compare Figure 4F, Figure S5 and Figure 3C). Whereas at 15°C, it was less consistent than the full-length C. briggsae promoter (Kolmogorov-Smirnov test, at 15°C p = 2.4×10−4), it was significantly more consistent than the proximal promoter at both temperatures (compare Figure 4F, Figure S5 and Figure 3D; Kolmogorov-Smirnov test, at 15°C p = 4.1×10−7, at 26°C p = 1.4×10−5).

Next, we tested whether another non-conserved intergenic sequence, from upstream of the C. briggsae promoter of gene unc-25 could rescue robustness of the proximal C. briggsae promoter of unc-47. Unlike unc-15, unc-25 is co-expressed with unc-47 [43], yet it shares no detectable sequence similarity within promoter elements (data not shown). It did indeed show substantially increased robustness of expression, comparable to the full-length promoter (Figure S6; indistinguishable from C. briggsae full-length Wilcoxon test, p = 0.4; different from C. briggsae proximal Wilcoxon test, p = 1.3×10−5). These results show that unrelated intergenic sequences are capable of conferring robust expression on a proximal promoter that directs the pattern.

Sequences that confer robust expression are AT-enriched

To understand why such different sequences were able to restore robustness of expression of the proximal C. briggsae unc-47 promoter, we examined them for general features they might have in common. Specifically, we calculated nucleotide frequencies in the distal unc-47, unc-15 and unc-25 promoters, and compared them to those of the 1.1 kb of vector DNA sequence that lies distal to all of the inserted promoters. Since this vector sequence, when it lies directly upstream of the proximal promoter, is not able to confer robustness, we sought out features that are shared by distal promoters but not the vector sequence.

Dinucleotide frequencies differ dramatically between distal unc-47, unc-15 and unc-25 promoter sequences and the upstream vector sequence. There is systematic enrichment for two dinucleotide classes, relative to the vector sequence, and a depletion of two other dinucleotide classes (Figure 5A). While there are between-sequence enrichment differences, the overall biases towards the AA/TT dinucleotides and away from the GC/CG dinucleotides is consistent among all sequences that confer robustness.

Figure 5. Promoter regions that confer robustness are enriched for nucleosome-depleted sequences.

(A) Enrichment/depletion of dinucleotides in the distal promoters of unc-15, unc-25 and unc-47 genes relative to the sequence of the pPD95.75 vector (log scale). Percentage of 200 individuals expressing GFP in indicated number of D-type neurons from a chimeric promoter composed of AT-rich sequence from the C. elegans daf-25 locus (B), AT-poor sequence from the C. elegans let-2 locus (C), AT-rich sequence from the D. melanogaster ChAT locus (D), and AT-poor sequence from the D. melanogaster CG8394.2 locus (E), fused upstream of the C. briggsae unc-47 proximal promoter. For comparison, C. briggsae unc-47 full-length and proximal promoters are shown in black and gray hashed bars, respectively.

This analysis suggests a simple hypothesis, namely that AT-enriched sequences (more specifically those enriched for AA/TT dinucleotides) should promote robust expression, whereas sequences depleted for these dinucleotides and enriched for GC/CG pairs (and to some extent CC/GG pairs) should not. To test this prediction, we subdivided the genome of C. elegans into 1 kb fragments, matching in size the previously tested distal sequences, and computed the extent of their AT-enrichment. A sequence located downstream of the daf-25 locus is enriched for AA/TT dinucleotides to an extent similar to distal promoters of unc-47, unc-15 and unc-25. This 1 kb fragment, when placed upstream of the proximal promoter of C. briggsae unc-47, was able to confer robustness similarly to the distal unc-47 promoter (Figure 5B; indistinguishable from C. briggsae full-length Wilcoxon test, p = 0.09; different from C. briggsae proximal Wilcoxon test, p = 1.9×10−5). In contrast, a 1 kb AT-depleted sequence from the let-2 locus was unable to rescue robustness (Figure 5C; different from C. briggsae full-length Wilcoxon test, p = 1.1×10−5; indistinguishable from C. briggsae proximal Wilcoxon test, p = 0.14). Furthermore, the construct containing the daf-25 sequence drove a more consistent expression than the one containing the let-2 sequence (Wilcoxon test, p = 4×10−3).

To ensure that the ability to rescue expression robustness is not restricted to AT-enriched sequences from nematode genomes, we tested whether sequences from distantly related species can perform this function. We segmented the genome of D. melanogaster into 1 kb fragments and selected one AT-enriched and one AT-depleted sequence using the same criteria as were applied to the fragments from the C. elegans genome. As predicted, a construct carrying the AT-enriched sequence drove substantially more robust expression than the proximal promoter alone (Figure 5D; indistinguishable from C. briggsae full-length Wilcoxon test, p = 0.2; different from C. briggsae proximal Wilcoxon test, p = 1.5×10−5). A construct carrying the AT-depleted sequence was no more robust than the proximal promoter alone (Figure 5E; different from C. briggsae full-length Wilcoxon test, p = 3.7×10−15; indistinguishable from C. briggsae proximal Wilcoxon test, p = 0.04).

Together these results suggest three important conclusions. First, AT-enrichment of a sequence can predict its ability to confer robustness of expression. Second, because two different AT-depleted sequences were not able to improve consistency of transgene expression, it is unlikely that robustness results from simply separating the proximal promoter from unknown repressive effects of the vector sequence. Sequence composition must play a critical role. Third, because multiple unrelated nematode sequences and an AT-enriched Drosophila sequence conferred robust expression, it is unlikely that short, gene- or species-specific motifs play a major role in improving consistency of expression. Our data imply that the mechanism responsible for conferring expression robustness relies on the overall nucleotide composition of promoters rather then on specific sequence motifs.


Our results suggest that promoters of Caenorhabditis unc-47 orthologs are organized into two domains that are markedly distinct in functions and evolutionary dynamics. Whereas proximal promoters are highly conserved and are sufficient to direct the appropriate spatial expression pattern, the distal sequences diverge rapidly and their primary function is to confer robustness of expression. The distal sequences within promoters of unc-47 are not capable of directing expression patterns on their own [41] and must therefore confer robustness via a mechanism distinct from redundant and evolutionarily conserved “shadow” enhancers [35], [36].

The shared nucleotide composition (Figure 5A) of the four sequences that promote robust expression – distal promoters of C. elegans and C. briggsae unc-47 as well as upstream regions of two unrelated genes, unc-15 and unc-25 – hints at a potential mechanism of action. Overall sequence composition plays a large role in establishing chromatin states throughout the genome [44]. In particular, AT-rich sequences tend to be associated with nucleosome-poor regions, although multiple factors determine whether DNA is bound to nucleosomes. Recent studies suggest that sequence-composition codes that displace nucleosomes may be common in active metazoan promoters [45], [46]. Intriguingly, the genomic sequence precisely corresponding to the distal, nonconserved portion of the C. elegans unc-47 promoter is depleted of nucleosomes [47] (Figure S7).

Trinucleotide frequencies are a better predictor of nucleosome positioning than dinucleotides [47]. The robustness-conferring sequences are two-fold enriched for trinucleotides that are preferentially found in nucleosome-depleted regions of the C. elegans genome, far more so than the conserved proximal promoters (Figure S7). Nucleosome occupancy can differ even in evolutionarily conserved promoters [46], [48], [49], still similar levels of enrichment for nucleosome-depleted trinucleotides were seen in the distal unc-47 promoters of C. brenneri and C. remanei (Figure S7). All sequences that confer robustness bear a signature consistent with nucleosome depletion, and the C. elegans sequences were shown to be depleted of nucleosomes (Figure S7). The AT-poor let-2 locus, on the other hand, is enriched for nucleosomes, and other sequences which are unable to improve consistency of expression, show a trinucleotide signature of nucleosome enrichment (Figure S7). We therefore hypothesize that open chromatin may promote robust expression.

We favor the hypothesis that the robustness function is executed by configuring chromatin in an accessible state for other factors to bind the promoter sequence. This hypothesis is consistent with the finding that variability of gene expression may be encoded in nucleosome-positioning sequences [50], and that chromatin regulators may contribute to environmental canalization [51]. Whether this mechanism of robustness arises as a byproduct of other forces that shape nucleotide composition of intergenic sequences, or whether it is directly selected upon, it has been conserved at the unc-47 locus.

We propose a simple scenario to account for the different evolutionary rates between the distal and proximal portions of the unc-47 promoter. The proximal promoter is responsible for directing the expression pattern because it contains numerous transcription factor binding sites. It appears that in the context of the proximal promoter most substitutions are deleterious and thus it evolves relatively slowly. The distal promoter, on the other hand, evolves at a considerably faster rate. Noting that the ability to confer robustness is conserved between distal promoters of unc-47 orthologs, we infer that it is maintained by selection that does not require maintenance of specific sequence identity. Indeed, unrelated sequences from the C. elegans unc-15, unc-25, and daf-25 loci and even an AT-rich sequence from D. melanogaster can rescue robustness of expression. Thus the distal promoters appear to be under a simpler constraint – they are only required to maintain a certain nucleotide composition, for instance that which is consistent with nucleosome depletion, to confer robustness of gene expression. Sequences that satisfy this requirement are quite degenerate, so the element tolerates a relatively high rate of sequence turnover, while retaining functional conservation. This hypothesis is consistent with a report of selection on sequence composition that encodes nucleosome organization in yeast [52]. We consider the distal promoter of the unc-47 gene to be an example of a weakly constrained functional sequence [53]. Such low constraint allows developmental systems drift [54], in which conserved molecular functions are mediated by divergent genetic systems.


Constructs and strains

To generate reporter constructs, promoter sequences were PCR amplified from genomic DNA and cloned upstream of GFP into pPD95.75. In all cases, reverse primers overlapped the start codon of the unc-47 ortholog. Prior to injections, constructs were sequenced to ensure accuracy. Precise boundaries of full-length, extended conservation and proximal constructs are given in Figure S1. To generate strains carrying extrachromosomal arrays, we injected a mixture (5 ng/µL promoter::GFP plasmid, 5 ng/µL pha-1 rescue construct, 100 ng/µL salmon sperm DNA) into C. elegans pha-1 (e2123) strain [55]. Transformants were selected at 25°C. The C. briggsae strains carrying Cbr promoter unc-47::GFP were produced by injecting a mixture (5 ng/µL promoter::GFP plasmid and 100 ng/µL salmon sperm DNA) into AF16 strain. Single copy integrated strains were generated following an established protocol [56]. Copy number of inserts was verified through quantitative PCR of GFP (normalized to genomic unc-47).

Counting the number of expressing cells

Mixed-stage populations of C. elegans carrying transgenes were grown at 20°C with abundant food and young adult- or L4-stage worms were selected. These were immobilized on agar slides with 100 mM NaN3 in M9 buffer. The slides were examined on a Leica DM5000B compound microscope under 400× magnification. Each worm was positioned such that the ventral nerve cord with its D-type neurons could be seen clearly, and the number of cell bodies expressing GFP were counted manually. Worms without any visible GFP expression were assumed to have lost the transgene. For each construct studied, multiple independent transgenic lines were generated, and final counts of 100–200 individuals (see figure legends/text for details) were derived from a mixture of these lines (inter-line variance is generally low). To mitigate against experimenter bias census counts were taken in a blinded fashion. Individual strains were coded by one investigator to obscure their identity. Another investigator then examined 100 individuals of each of these strains. Once all counting was finished, strain identities were revealed and data were analyzed.

Fluorescence measurements and temperature stress experiments

Intensity of GFP expression in individual cells was measured on a Leica DM5000B compound scope fitted with a Qimaging Retiga2000 camera. Images of cells were outlined in imageJ, average intensity was measured and the background subtracted. Multiple strains carrying the same transgene were examined throughout and tested for concordance.

For integrated strains we used 125 ms exposure, 100% excitation. Pictures of 7 cells (DD1, VD1, VD2, DD3, VD6, VD13, DVB) were taken. For each strain and treatment (15°C, 20°C, 26°C) 25 L4-staged worms were measured. For temperature stress experiments (these were conducted on strains carrying extrachromosomal arrays) worms were reared at 15°C or 26°C for at least two generations. Then 50 L4 individuals were mounted for each treatment and strain and intensity of GFP was measured (125 ms exposure, variable excitation) for D-type neurons (average values recorded) and DVB.

Statistical analysis

All statistical analyses were performed in R. In all cases, the logarithm of measured GFP intensity was used. Wilcoxon test was used to assess consistency of the number of cells expressing different constructs. To assess the amount of scatter in fluorescence measurements (data reported in Figure 3, Figure 4, Figures S4 and S5, and in Table S2), we computed geometric distances between all data points for a particular strain/treatment and the mean of that strain/treatment. To test whether distributions of distances derived in such a way were significantly different for different strains/treatments, we conducted Kolmogorov-Smirnov tests. We used Ansari-Bradley test to determine whether the relative DVB fluorescence was more variable for proximal compared to full-length promoters.

Supporting Information

Figure S1.

Annotated alignment of unc-47 upstream sequences from four nematode species.


Figure S2.

Sequence conservation of unc-47 is biased to the proximal promoter.


Figure S3.

Integrated and extrachromosomal transgenes show corresponding effect of promoter length on pattern consistency.


Figure S4.

Multiple independent full-length and proximal strains are consistent in their robustness.


Figure S5.

Multiple independent chimeric and deleted strains are consistent in their robustness.


Figure S6.

The C. briggsae unc-25 distal promoter confers robustness of expression pattern.


Figure S7.

Nucleosome occupancy in robustness-conferring sequences.


Table S1.

Conservation of proximal promoters and distal sequence among four nematode species.


Table S2.

Comparisons of variation of all strains at 15°C and 26°C.



We are grateful to Stefano Alessina and Kevin Bullaughey for advice and to Kelsie Eichel and Paul Wang for technical help. We thank Yoav Gilad, Martin Kreitman, John Reinitz, and Kevin White for critical reading of the manuscript and helpful suggestions.

Author Contributions

Conceived and designed the experiments: AB KLG IR. Performed the experiments: AB KLG. Analyzed the data: AB KLG IR. Wrote the paper: AB KLG IR.


  1. 1. Xie X, Mikkelsen TS, Gnirke A, Lindblad-Toh K, Kellis M, et al. (2007) Systematic discovery of regulatory motifs in conserved regions of the human genome, including thousands of CTCF insulator sites. Proc Natl Acad Sci U S A 104(17): 7145–7150.
  2. 2. Bejerano G, Pheasant M, Makunin I, Stephen S, Kent WJ, et al. (2004) Ultraconserved elements in the human genome. Science 304(5675): 1321–1325.
  3. 3. Boffelli D, Nobrega MA, Rubin EM (2004) Comparative genomics at the vertebrate extremes. Nature Reviews Genetics 5(6): 456–465.
  4. 4. Pennacchio LA, Rubin EM (2001) Genomic strategies to identify mammalian regulatory sequences. Nature Reviews Genetics 2(2): 100–109.
  5. 5. Aparicio S, Morrison A, Gould A, Gilthorpe J, Chaudhuri C, et al. (1995) Detecting conserved regulatory elements with the model genome of the japanese puffer fish, fugu rubripes. Proc Natl Acad Sci U S A 92(5): 1684–1688.
  6. 6. Blow MJ, McCulley DJ, Li Z, Zhang T, Akiyama JA, et al. (2010) ChIP-seq identification of weakly conserved heart enhancers. Nat Genet 42(9): 806-U107.
  7. 7. Ritter DI, Li Q, Kostka D, Pollard KS, Guo S, et al. (2010) The importance of being cis: Evolution of orthologous fish and mammalian enhancer activity. Mol Biol Evol 27(10): 2322–2332.
  8. 8. McGaughey DM, Vinton RM, Huynh J, Al-Saif A, Beer MA, et al. (2008) Metrics of sequence constraint overlook regulatory sequences in an exhaustive analysis at phox2b. Genome Res 18(2): 252–260.
  9. 9. Margulies EH, Cooper GM, Asimenos G, Thomas DJ, Dewey CN, et al. (2007) Analyses of deep mammalian sequence alignments and constraint predictions for 1% of the human genome. Genome Res 17(6): 760–774.
  10. 10. Hare EE, Peterson BK, Iyer VN, Meier R, Eisen MB (2008) Sepsid even-skipped enhancers are functionally conserved in drosophila despite lack of sequence conservation. PLoS Genet 4: e1000106.
  11. 11. Fisher S, Grice EA, Vinton RM, Bessling SL, McCallion AS (2006) Conservation of RET regulatory function from human to zebrafish without sequence similarity. Science 312(5771): 276–279.
  12. 12. Romano LA, Wray GA (2003) Conservation of Endo16 expression in sea urchins despite evolutionary divergence in both cis and trans-acting components of transcriptional regulation. Development 130(17): 4187–4199.
  13. 13. Wray GA, Hahn MW, Abouheif E, Balhoff JP, Pizer M, et al. (2003) The evolution of transcriptional regulation in eukaryotes. Mol Biol Evol 20(9): 1377–1419.
  14. 14. Swanson CI, Evans NC, Barolo S (2010) Structural rules and complex regulatory circuitry constrain expression of a notch- and EGFR-regulated eye enhancer. Developmental Cell 18(3): 359–370.
  15. 15. Nokes EB, Van Der Linden AM, Winslow C, Mukhopadhyay S, Ma K, et al. (2009) Cis-regulatory mechanisms of gene expression in an olfactory neuron type in caenorhabditis elegans. Developmental Dynamics 238(12): 3080–3092.
  16. 16. Crocker J, Tamori Y, Erives A (2008) Evolution acts on enhancer organization to fine-tune gradient threshold readouts. PLoS Biol 6: e263.
  17. 17. Johnson DS, Davidson B, Brown CD, Smith WC, Sidow A (2004) Noncoding regulatory sequences of gona exhibit strong correspondence between evolutionary constraint and functional importance. Genome Res 14(12): 2448–2456.
  18. 18. Small S, Arnosti DN, Levine M (1993) Spacing ensures autonomous expression of different stripe enhancers in the even-skipped promoter. Development 119(3): 767–772.
  19. 19. Brown CD, Johnson DS, Sidow A (2007) Functional architecture and evolution of transcriptional elements that drive gene coexpression. Science 317(5844): 1557–1560.
  20. 20. Papatsenko D, Levine M (2007) A rationale for the enhanceosome and other evolutionarily constrained enhancers. Current Biology 17(22): R955–R957.
  21. 21. Erives A, Levine M (2004) Coordinate enhancers share common organizational features in the drosophila genome. Proc Natl Acad Sci U S A 101(11): 3851–3856.
  22. 22. Weirauch MT, Hughes TR (2010) Conserved expression without conserved regulatory sequence: The more things change, the more they stay the same. Trends in Genetics 26(2): 66–74.
  23. 23. Arnosti DN, Kulkarni MM (2005) Transcriptional enhancers: Intelligent enhanceosomes or flexible billboards? J Cell Biochem 94(5): 890–898.
  24. 24. Bradley RK, Li X, Trapnell C, Davidson S, Pachter L, et al. (2010) Binding site turnover produces pervasive quantitative changes in transcription factor binding between closely related drosophila species. PLoS Biol 8: e1000343.
  25. 25. Gilad Y, Rifkin SA, Pritchard JK (2008) Revealing the architecture of gene regulation: The promise of eQTL studies. Trends in Genetics 24(8): 408–415.
  26. 26. Price AL, Patterson N, Hancks DC, Myers S, Reich D, et al. (2008) Effects of cis and trans genetic ancestry on gene expression in african americans. PLoS Genet 4: e1000294.
  27. 27. Raj A, Rifkin SA, Andersen E, van Oudenaarden A (2010) Variability in gene expression underlies incomplete penetrance. Nature 463(7283): 913-U84.
  28. 28. Masel J, Siegal ML (2009) Robustness: Mechanisms and consequences. Trends in Genetics 25(9): 395–403.
  29. 29. Braendle C, Felix M (2008) Plasticity and errors of a robust developmental system in different environments. Developmental Cell 15(5): 714–724.
  30. 30. Felix M, Wagner A (2008) Robustness and evolution: Concepts, insights and challenges from a developmental model system. Heredity 100(2): 132–140.
  31. 31. Siegal ML, Bergman A (2002) Waddington's canalization revisited: Developmental stability and evolution. Proc Natl Acad Sci U S A 99(16): 10528–10532.
  32. 32. Manu , Surkova S, Spirov AV, Gursky VV, Janssens H, et al. (2009) Canalization of gene expression in the drosophila blastoderm by gap gene cross regulation. PLoS Biol 7: e1000049.
  33. 33. Szoellosi GJ, Derenyi I (2009) Congruent evolution of genetic and environmental robustness in micro-RNA. Mol Biol Evol 26(4): 867–874.
  34. 34. Hong J, Hendrix DA, Levine MS (2008) Shadow enhancers as a source of evolutionary novelty. Science 321(5894): 1314–1314.
  35. 35. Frankel N, Davis GK, Vargas D, Wang S, Payre F, et al. (2010) Phenotypic robustness conferred by apparently redundant transcriptional enhancers. Nature 466(7305): 490-U8.
  36. 36. Perry MW, Boettiger AN, Bothma JP, Levine M (2010) Shadow enhancers foster robustness of drosophila gastrulation. Current Biology 20(17): 1562–1567.
  37. 37. Li X, Cassidy JJ, Reinke CA, Fischboeck S, Carthew RW (2009) A MicroRNA imparts robustness against environmental fluctuation during development. Cell 137(2): 273–282.
  38. 38. Herranz H, Cohen SM (2010) MicroRNAs and gene regulatory networks: Managing the impact of noise in biological systems. Genes Dev 24(13): 1339–1344.
  39. 39. Kiontke K, Gavin NP, Raynes Y, Roehrig C, Piano F, et al. (2004) Caenorhabditis phylogeny predicts convergence of hermaphroditism and extensive intron loss. Proc Natl Acad Sci U S A 101(24): 9003–9008.
  40. 40. McIntire SL, Reimer RJ, Schuske K, Edwards RH, Jorgensen EM (1997) Identification and characterization of the vesicular GABA transporter. Nature 389(6653): 870–876.
  41. 41. Ruvinsky I, Ruvkun G (2003) Functional tests of enhancer conservation between distantly related species. Development 130(21): 5133–5142.
  42. 42. Kagawa H, Gengyo K, Mclachlan AD, Brenner S, Karn J (1989) Paramyosin gene (unc-15) of caenorhabditis-elegans - molecular-cloning, nucleotide-sequence and models for thick filament structure. J Mol Biol 207(2): 311–333.
  43. 43. Eastman C, Horvitz HR, Jin YS (1999) Coordinated transcriptional regulation of the unc-25 glutamic acid decarboxylase and the unc-47 GABA vesicular transporter by the caenorhabditis elegans UNC-30 homeodomain protein. Journal of Neuroscience 19(15): 6225–6234.
  44. 44. Segal E, Widom J (2009) What controls nucleosome positions? Trends in Genetics 25(8): 335–343.
  45. 45. Guertin MJ, Lis JT (2010) Chromatin landscape dictates HSF binding to target DNA elements. PLoS Genet 6: e1001114.
  46. 46. Khoueiry P, Rothbacher U, Ohtsuka Y, Daian F, Frangulian E, et al. (2010) A cis-regulatory signature in ascidians and flies, independent of transcription factor binding sites. Current Biology 20(9): 792–802.
  47. 47. Valouev A, Ichikawa J, Tonthat T, Stuart J, Ranade S, et al. (2008) A high-resolution, nucleosome position map of C. elegans reveals a lack of universal sequence-dictated positioning. Genome Res 18(7): 1051–1063.
  48. 48. Tirosh I, Sigal N, Barkai N (2010) Divergence of nucleosome positioning between two closely related yeast species: Genetic basis and functional consequences. Molecular Systems Biology 6: 365.
  49. 49. Tsankov AM, Thompson DA, Socha A, Regev A, Rando OJ (2010) The role of nucleosome positioning in the evolution of gene regulation. PLoS Biol 8: e1000414.
  50. 50. Choi JK, Kim Y (2009) Intrinsic variability of gene expression encoded in nucleosome positioning sequences. Nat Genet 41(4): 498–503.
  51. 51. Gibert J, Karch F, Schloetterer C (2011) Segregating variation in the polycomb group gene cramped alters the effect of temperature on multiple traits. PLoS Genet 7: e1001280.
  52. 52. Kenigsberg E, Bar A, Segal E, Tanay A (2010) Widespread compensatory evolution conserves DNA-encoded nucleosome organization in yeast. PLoS Comput Biol 6: e1001039.
  53. 53. Cooper GM, Brown CD (2008) Qualifying the relationship between sequence conservation and molecular function. Genome Res 18(2): 201–205.
  54. 54. True JR, Haag ES (2001) Developmental system drift and flexibility in evolutionary trajectories. Evol Dev 3(2): 109–119.
  55. 55. Granato M, Schnabel H, Schnabel R (1994) Pha-1, a selectable marker for gene-transfer in C-elegans. Nucleic Acids Res 22(9): 1762–1763.
  56. 56. Frokjaer-Jensen C, Davis MW, Hopkins CE, Newman BJ, Thummel JM, et al. (2008) Single-copy insertion of transgenes in caenorhabditis elegans. Nat Genet 40(11): 1375–1383.