Research Article

Evolution of Shape by Multiple Regulatory Changes to a Growth Gene

See allHide authors and affiliations

Science  24 Feb 2012:
Vol. 335, Issue 6071, pp. 943-947
DOI: 10.1126/science.1215193


The genetic changes responsible for morphological differences between species are largely unidentified. Such changes can involve modifications of growth that are relevant to understanding evolution, development, and disease. We identified a gene that induces male-specific wing size and shape differences between Nasonia wasp species. Fine-scale mapping and in situ hybridization reveal that changes in at least three regions (two strictly in noncoding sequence) around the gene unpaired-like (upd-like) cause changes in spatial and temporal expression of upd-like in the developing wing and corresponding changes in wing width. Upd-like shows homology to the Drosophila unpaired gene, a well-studied signaling protein that regulates cell proliferation and differentiation. Our results indicate how multiple changes in the regulation of upd-like are involved in microevolution of morphological and sex-specific differences between species.

The diversity of animal size and shape has fascinated scientists for centuries (1, 2). Changes in morphology, such as the beaks of birds, toes of lizards, and hands of humans, underpin diversification and adaptation into different ecological niches (35). Yet the genes and genetic changes responsible for animal shape differences between species remain poorly understood (68). Genetic studies have revealed some genes and pathways required for growth and regulation of size and shape during development, many of which are also involved in human disease such as cancer and diabetes (9). These include signaling genes that regulate organ-specific patterning and growth [e.g., Wnt, Hippo, and Janus kinase–signal transducers and activators of transcription (JAK/STAT)], nutritional and hormonal regulators of body size (such as insulin signaling) and effector genes of cell growth and proliferation (cell cycle, apoptosis, and protein synthesis) (9, 10). However, it is not yet clear whether evolutionary changes in these genes or others underlie size and shape differences between animal species (7).

The parasitoid jewel wasp Nasonia is emerging as a model system for investigating the genetics of species differences in development and morphology. The Nasonia genus consists of four closely related species, each of which has evolved a distinct male wing size (8, 1113). The greater than twofold difference in male wing size between N. vitripennis and N. giraulti (Fig. 1A) provides a tool for investigating the evolution of growth regulation. We isolated the two largest effect quantitative trait loci (QTLs) for this male wing size difference (11, 14), wing-size1 (ws1) (14) and widerwing (wdw) (11), which increase wing size through different mechanisms. The giraulti allele of ws1 increases overall male wing size relative to N. vitripennis. The genetic change underlying ws1 has been mapped to a 13.5-kb noncoding interval containing the 5′-untranslated region of the sex-determining factor doublesex (8). Changes in doublesex expression level in the developing wing are correlated with the wing size difference. doublesex is a major sex-determining factor in animals, found from humans to worms, which provides the downstream signal of somatic sex (15) and which also can have a role in growth regulation (8, 16).

Fig. 1

Wdw maps to the upd-like locus and affects its expression. (A) Male N. vitripennis and N. giraulti have diverged in wing size. (B) Genes in the 200-kb region around wdw in N. vitripennis, including a Maverick transposable element that is absent in N. giraulti and does not appear to affect wing size (18). (C) Recombinant genotypes (left) and phenotypes (right) map the full wdw effect to a 115-kb region. The wing width of haploid males of two genotypes of interest was measured using crosses (fig. S1) (17) between the original wdw introgression (solid orange bar) and one of the following: (i) N. vitripennis, (ii) an introgression recombinant on the left, (iii) an introgression recombinant on the right, and (iv) an introgression recombinant on both sides (wdw-135k). White, N. vitripennis genotype; orange, N. giraulti genotype; gray, genotype between markers is unknown. Dark blue bars and * show statistically significant difference in wing width between genotypes, corrected for multiple testing (Bonferroni); light blue bars, no significant difference. DFDen, Denominator degrees of freedom from restricted maximum likelihood analysis of variance (ANOVA). Error bars show standard error. (D) RNA in situ hybridization reveals that the wdw introgression (wdw-135k line) has broader upd-like expression than N. vitripennis, consistent with the larger wing size. No expression pattern is seen for neighboring genes NV21595 and hmm1064384. NV10313 is not expressed in wing (18). Arrowheads denote boundaries of upd-like expression. Quantification of the size of the expression domain is in fig. S3. Scale bars, 200 μm.

Positional cloning and identification of wdw. Widerwing (wdw) alters male-specific wing shape, as well as wing size, by changing cell numbers differentially across the width of the male wing, but with no effect on female wing size (11). The male wing size phenotype was initially mapped to a 1-Mb region by introgressing chromosome segments from N. giraulti into N. vitripennis. Recombination mapping using generated recessive lethal mutations linked to the introgression (fig. S1) (17) mapped wdw further to 115 kb (Fig. 1). During this process, we broke the region into intervals differentially affecting male wing width, which we made into inbred lines for further analysis (18). We also created a 135-kb introgression (wdw-135k line) of N. giraulti sequence in N. vitripennis, resulting in males with wide wings that are not significantly different in wing width from the original wdw introgression line (Fig. 1).

Only one gene, NV21594, occurs within the 115-kb wdw region (Fig. 1) (18), which suggests that it could be the gene behind the wing size difference. However, it is also possible that genes adjacent to the 115-kb region are responsible, as their expression patterns could be encoded in cis-regulatory elements within this region. We tested the expression of NV21594 and adjacent genes NV10313, NV21595, and hmm1064384 in wings using reverse transcription polymerase chain reaction. We detected expression of all but NV10313 in male wings (18). We further investigated whether NV21594 or adjacent genes have spatial expression patterns in developing wings by using RNA in situ hybridization (17). Whereas NV21595 and hmm1064384 do not show a spatial pattern of expression (Fig. 1) (18), NV21594 shows a clear spatial expression pattern that differs between N. vitripennis and the N. giraulti introgression. Specifically, it is expressed in the tip of the N. vitripennis prepupal male wing but is expressed much more broadly along the wing edge in the giraulti allele of wdw in a vitripennis background (wdwgV) and also in N. giraulti male wings (Fig. 1 and figs. S3 and S4) (18). Thus, the correlation between expression pattern and wing size suggests NV21594 as the likely candidate for wdw.

Through searches of other arthropod genomes, we have identified genes with homology to NV21594 in the holometabolous insect orders Hymenoptera, Coleoptera, and Diptera (fig. S5) (18), which include the Drosophila melanogaster gene unpaired (upd). The upd candidates from these orders share five conserved peptide motifs, including an unusual short motif (WXNPCG) that is diagnostic for upd candidate protein genes in Nasonia, Apis, Tribolium, and D. melanogaster (table S1) (18). We therefore refer to NV21594 as unpaired-like (upd-like). The upd candidates occur in clusters with two paralogous genes (upd-like2 and upd-like3) in each of these taxa, as is also found in D. melanogaster. These proteins are also highly alpha-helical (fig. S5), which is typical of cytokines (19). Upd in Drosophila regulates cell proliferation as a ligand of the JAK-STAT pathway and is thus a reasonable candidate for a wing-growth gene. Upd is thought to play a similar role in Drosophila as the mammalian cytokine interleukin-6 (IL-6), which is involved in a number of human diseases (19).

The wdwg introgression both increases wing width (by 25%) and expands the length of the upd-like expression domain along the wing edge (by 159%) relative to N. vitripennis (fig. S3). It may be that wdwg introgression could be causing the observed expansion in upd-like expression, but it is also possible that this expansion is simply a by-product of the larger wing. We thus examined upd-like expression using the large winged ws1g QTL, which increases wing width by 30% (8). Although ws1g and wdwg have wings similarly wider than N. vitripennis, upd-like expression in ws1g is vitripennis-like (fig. S4). Specifically, expression is absent from the posterior wing margin and significantly contracted relative to wdwg, with a 54% smaller expression edge length (fig. S3). These results suggest that the wdw region is a driver of both upd-like expression and male wing size. In females, the wdwg introgression does not affect wing size (11), and upd-like is expressed in a broad curve in both introgression and N. vitripennis female wings (fig. S6). This indicates that upd-like is not a male-specific gene, but rather that its expression is regulated in a male-specific fashion. The dependence of both male wing size and expression pattern on the wdw allele strongly indicates that upd-like is the gene that mediates the wing size difference.

Multiple noncoding intervals around upd-like contribute to the wing size difference. Several recombinants within the 115-kb wdw region had intermediate wing sizes, which indicated that it is likely that multiple functional sequence changes contribute to the full wdw phenotype. Experimental crosses between recombinant lines that differ in giraulti sequence within the 115-kb region around upd-like (fig. S1) identified three subregions within the 115-kb wdw region that differentially contribute to wing size (Fig. 2). The middle subregion (wdw-B) contains both noncoding differences and a few coding changes in the first exon of upd-like. However, two of the three subregions, wdw-A and wdw-C, contain exclusively noncoding DNA. We created a wdw-A recombinant line (wdw-A-50k) that contains only 50 kb of giraulti noncoding sequence spanning the wdw-A subregion. These wdw-A-50k males have significantly larger wings than N. vitripennis (Fig. 2), which suggests that noncoding changes in the region are sufficient to alter wing size, regardless of whether the upd-like protein coding sequence comes from N. giraulti or N. vitripennis.

Fig. 2

Multiple intervals within the wdw region affect wing size. The wdw phenotype splits into at least three subregions (light dotted lines), each with significant effects on wing size. Heavy dotted lines mark sequence differences between three pairs of adjacent recombinant lines that best define these subregions. Also shown are two recombinants (wdw-A-50k and LB3-125, also used in Figs. 3 and 4) where recombination has isolated the wdw-A and wdw-C subregions.

The three wdw subregions appear to each regulate distinct spatial parts of the wing width difference. We measured changes in width along the wing anterior-posterior (AP) axis in our recombinant lines (Figs. 3 and 4 and fig. S7) and documented that the full wdwg introgression increases wing width in each sector along the AP axis, with a particularly strong effect in the anterior-central sector (Fig. 3). The three subregions subdivide this pattern. Wdw-A only affects wing size in the anterior half of the wing and appears to contribute primarily to the size of the anterior-central sector (Fig. 3 and fig. S7). In contrast, wdw-B and wdw-C have broader effects on wing width, and their specific effects appear to depend on the genotype at other subregions. Specifically, whereas the spatial effects of wdw-A are similar whether we compare recombinants from the left side or from the right side of the region, the spatial effects of wdw-B and wdw-C show significant differences (fig. S7) across all sectors or modestly spatially restricted effects, depending on the direction of the comparison. This suggests either that the subregions interact epistatically to produce the full wing size difference or that there are additional modifier QTLs, most likely located between wdw-B and C. For subsequent analysis of wdw-C, we focused on a recombinant line (LB3-125) that contains giraulti sequence only at wdw-C. This line has a fairly uniform effect, a 7.4 ± 0.2 μm to 11.0 ± 0.9 μm (7 to 15%) increase in each wing sector relative to N. vitripennis (Fig. 4).

Fig. 3

Anterior-specific effects of the wdw-A subregion. (A) The original wdw introgression affects the size of sectors across the width of the adult wing, whereas the wdw-A subregion only affects the anterior-central sector. The width of four A-P sectors was measured for the genotype pairs indicated on the left (17). (i) Original wdw introgression compared with N. vitripennis; (ii) wdw-A-50k compared with N. vitripennis. Dark blue bars and * show statistically significant difference in sector width between genotypes, corrected for multiple testing (Bonferroni). Error bars show standard error. (B) Diagram of the adult wing sectors used in part A and Fig. 4A. (C) Illustration of Nasonia wing imaginal disc, which develops in a similar orientation to the adult wing. (D) Adult wings reveal anterior-specific expansion in wdw-A. Dashed line marks position of median vein (inferred A-P boundary). Dotted lines mark wing edges. (E) wdw-A also regulates anterior-specific expression of upd-like in prepupae. Dashed line marks wing tip (A-P boundary, fig. S3). Arrowheads denote boundaries of upd-like expression. Scale bars, 200 μm.

Fig. 4

Temporal regulation of upd-like by wdw-C. (A) A recombinant line containing the giraulti allele of wdw-C (LB3-125) is wider than N. vitripennis in each of four sectors across the adult wing. (B) Expression of upd-like before and after pupation reveals a temporal shift in wdw-C. In N. vitripennis, upd-like expression in the tip of the wing is visible only in gray larvae and prepupae. In pupal wings, expression is only apparent in the proximal posterior wing (arrow). In wdw-135k, expression is similar to N. vitripennis in gray larvae but expands apically in prepupae and is active later into pupation. In wdw-C (LB3-125), expression is vitripennis-like in the prepupa and giraulti-like in the pupa, which suggests that this subregion controls temporal regulation of upd-like. Control gene wingless is expressed at all stages (fig. S8).

Wdw-A causes an increase in wing width in the anterior half of the wing. It also causes an anterior-specific change in upd-like expression, which is broad and giraulti-like in the anterior of wdw-A (wdw-A-50k line) prepupal wings and yet contracted and vitripennis-like in the posterior (Fig. 3 and fig. S3). We verified that this was a shift in the anterior expression pattern and not a shift in the A-P compartment boundary by costaining with an antibody against the posterior compartment–specific protein Engrailed (fig. S3).

In contrast to wdw-A’s spatial effects, wdw-C (LB3-125 recombinant line) increases wing width across each A-P axis region (Fig. 4). To our surprise, upd-like expression in prepupal wings is vitripennis-like in this line (Fig. 4), despite the increase in adult wing size. An alternative mechanism that could produce a change in wing size is a temporal expansion of upd-like expression. Indeed, a broad giraulti-like expression pattern emerges in wdw-C pupal wings at a stage where expression in the wing tip is absent in N. vitripennis (Fig. 4) (18). Together, these observations suggest that wdw-C is a late-acting regulatory element with a broad effect.

We inferred the effects of wdw-B even though we did not genetically isolate it as cleanly as wdw-A and wdw-C. Upd-like has a broad expression pattern in recombinants bearing giraulti alleles at wdw-B+C (LB3-208 line), not significantly different from the full introgression (figs. S3 and S9). These data suggest that this subregion controls the early expression pattern, whereas wdw-C controls the late expression pattern. This expression pattern also suggests that both wdw-A and wdw-B activate upd-like expression in anterior cells, perhaps in an overlapping pattern.

Interspecific patterns of unpaired-like expression and wing size. Male wing size and upd-like expression were compared in three Nasonia and two large-winged outgroup species, the closely related Trichomalopsis sarcophagae and more distant Muscidifurax raptorellus (Fig. 5). Small male wings appear to be derived, occurring in both N. vitripennis and N. longicornis, even though N. longicornis is more closely related to N. giraulti (12, 20). The reduced male wing width in N. longicornis appears to be due, in part, to evolution at the wdw locus: introgression of the wdwl region into N. vitripennis (wdwlV) increases adult wing size by 11%, intermediate between N. vitripennis and wdwgV (11). Contraction of the upd-like expression pattern has occurred in both N. vitripennis and N. longicornis males (Fig. 5). Furthermore, the N. longicornis upd-like expression pattern change is also caused by evolution at wdw, as the wdwlV introgression has a contracted expression pattern relative to wdwgV (Fig. 5 and fig. S3).

Fig. 5

Evolution of wing size and unpaired-like expression. (A) Adult wings and upd-like expression in prepupal wings from Nasonia males and outgroup species. Both N. vitripennis and N. longicornis males have a contracted expression pattern. The species cladogram is adapted from (29). (B) A N. longicornis introgression of upd-like into N. vitripennis also shows a contracted expression pattern.

Although the smaller wing size and upd-like expression in N. vitripennis and N. longicornis could be the result of a hybridization event between the species followed by introgression of the wdw region, the data do not support this. Sequence divergence in the three subregions is consistent with the species tree; there is 1.2% divergence between N. longicornis and N. giraulti, less than the respective 2.4% and 2.8% between these species and N. vitripennis (18). Furthermore, analysis of phylogenetic tree topology with a sliding window across the wdw region, as well as examination of parsimony-informative sites, revealed no evidence of sequence introgression between N. vitripennis and N. longicornis (18). These results indicate that regulatory evolution of upd-like probably occurred in two separate Nasonia lineages, either by parallel reductions in N. vitripennis and N. longicornis or by reduction in an ancestral Nasonia lineage followed by expansion in the N. giraulti lineage. This observation plus the role of three subregions that we have identified between N. vitripennis and N. giraulti suggests that unpaired-like is a hotspot for wing size evolution in Nasonia.

Discussion. This study determined the genetic basis of wdw, a major component of the male-specific wing size difference between N. vitripennis and N. giraulti. Our data indicate that upd-like causes the spatial changes in cell proliferation and growth within the wing and that upd-like is a hotspot for size and shape evolution in Nasonia. Unpaired (upd) in D. melanogaster is a ligand for the JAK/STAT pathway (21). This pathway’s role in cell proliferation makes it a plausible target of morphological evolution, though it is not yet known whether upd-like mediates its effects in Nasonia through JAK/STAT or other pathways. Nasonia upd-like appears to be particularly susceptible to wing growth–altering changes, which indicates that the gene might have a specialized role in specifying organ size in the growth gene network. Considering that many size and shape differences between animals are due to differences in cell numbers (9), upd-like genes could be hotspots of size and shape evolution in other species as well.

Using a phenotype-based positional cloning approach, we have identified two different major QTL genes (wdw and ws1) that are responsible for sex-specific differences in wing development between closely related Nasonia species. From what is known about fly wing development (2224), the alternative candidate gene approach would not have predicted the role of either upd or dsx in Nasonia wing size. Nevertheless, both genes appear to be homologs of functionally conserved developmental regulatory genes, which is consistent with the hypothesis that core developmental genes tend to be involved in morphological evolution (25, 26). It has also been argued that noncoding cis-regulatory changes could play a central role in developmental differences between species (27, 28). Our findings support this view and further implicate growth-regulating genes in organ-specific size and shape evolution.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S9

Tables S1 to S3

References (3032)

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Supporting text is available as supporting material on Science Online.
  3. Acknowledgments: We thank L. Gu and J. Sysol for volunteer assistance with mapping; M. Rosenberg for sharing the Nasonia wingless plasmid; and H. Jasper, J. D. Lambert, M. Welte, D. Stern, D. Presgraves, D. Wheeler, R. Edwards, A. Avery, and M. Clark for advice and technical assistance. This work was supported by NSF Doctoral Dissertation Improvement Grant DEB-0910017 to D.W.L. and NIH grants 5R01 GM070026-04 and 5R24 GM084917-04 to J.H.W. Sequences are deposited at GenBank with accessions JQ082366 to JQ082369.
View Abstract

Navigate This Article