Insights into Hox protein function from a large scale combinatorial analysis of protein domains

Abstract

Protein function is encoded within protein sequence and protein domains. However, how protein domains cooperate within a protein to modulate overall activity and how this impacts functional diversification at the molecular and organism levels remains largely unaddressed. Focusing on three domains of the central class Drosophila Hox transcription factor AbdominalA (AbdA), we used combinatorial domain mutations and most known AbdA developmental functions as biological readouts to investigate how protein domains collectively shape protein activity. The results uncover redundancy, interactivity, and multifunctionality of protein domains as salient features underlying overall AbdA protein activity, providing means to apprehend functional diversity and accounting for the robustness of Hox-controlled developmental programs. Importantly, the results highlight context-dependency in protein domain usage and interaction, allowing major modifications in domains to be tolerated without general functional loss. The non-pleoitropic effect of domain mutation suggests that protein modification may contribute more broadly to molecular changes underlying morphological diversification during evolution, so far thought to rely largely on modification in gene cis-regulatory sequences.

Figure 1. Combinatorial analyses of AbdA protein domains using a variety of biological readouts.

A. AbdA protein variants generated. Crosses indicate domain mutations. B. Biological readouts used in this study.

Figure 2. Dispensability of HX, TD, and UA protein domains for somatic muscle specification.

Somatic muscles are visualized by Nautilus (Nau, green) expression (upper panels). Expression of AbdA (red) in the thorax using the 24B-Gal4 driver induces abdominal specific muscle pattern (white arrows) in thoracic segments (red arrows) (middle panels). The effect of the AbdA^HX,UA variant is illustrated (lower panels). Right panels are magnifications of boxed areas. Graphs (% of remaining activities compared to the wild type AbdA protein (WT) following domain mutations) using the boxplot representation on the right summarize quantitative analyses (see Text S1 and Figure S1 for full illustration). A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity).

Figure 3. Single protein domain requirement for NB5–6 lineage truncation.

Neuroblast 5–6 are visualized using the ladybird early lbe(K)-Gal4 driver expressing nuclear GFP (green; upper panels). Dotted rectangles highlight differences in cell number of the T2/3 thoracic NB5–6 lineage. Expression of AbdA (red) in NB5–6, through the lbe(K)-Gal4 driver, leads to thoracic lineage truncation (middle panels). The effect of AbdA^UA is illustrated in the lower panel. Graphs (% of remaining activities compared to the wild type AbdA protein (WT) following domain mutations) using the boxplot representation on the right summarize quantitative analysis (see Text S1). A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity).

Figure 4. Protein domain mutations inducing neomorphic activities.

A. Localized PS7 dpp expression (green) in the visceral mesoderm (in situ hybridization, white arrow) relies on posterior repression by AbdA (upper panels). Ubiquitous mesodermal (24B-Gal4 driven) expression of AbdA (red) represses dpp expression in PS7 (red arrow; middle panels) B. Localized PS8 expression of wg (green) in the visceral mesoderm (in situ hybridization, white arrow) relies on activation by AbdA. Ubiquitous AbdA in the mesoderm (24B-Gal4) induces anterior ectopic wg expression (red arrow; middle panels). In A and B, the effect of the AbdA^HX,UA variant on dpp (A) or wg (B) expression is illustrated (lower panels). Right panels are magnifications of the boxed areas. Graphs (% of remaining activities compared to the wild type AbdA protein (WT) following domain mutations) using the boxplot representation on the right summarize quantitative analyses (see Text S1 and Figure S2 (dpp) and S3 (wg) for full illustration). A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity). Neomorphic activities, ie qualitative changes in protein activity, are depicted in red.

Figure 5. Additive contribution of protein domains.

Oenocytes ventrally located in the abdomen are visualized by β-gal (green) driven by the seven-up (svp) promoter (white arrows, upper panel). Expression of AbdA (red) in the thorax through the arm-Gal4 driver induces ectopic oenocytes in the thorax (red arrows, middle panels). The effect of the AbdA^HX/UA variants is illustrated (lower panels). Boxed areas highlight thoracic segments. Right panels are magnifications of the boxed areas. Graphs (% of remaining activities compared to the wild type AbdA protein (WT) following domain mutations) using the boxplot representation on the right summarize quantitative analyses (see Text S1 and Figure S4 for full illustration). A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity).

Figure 6. Functional redundancy of the HX, TD, and UA protein domains.

A. Tracheal branches are visualized by GFP-driven through the breathless btl-Gal4 driver (green). The cerebral branch (white arrowhead) forms only in T2 as a result of abdominal repression mediated by AbdA (upper panels). Expression of AbdA (red) in thoracic segments through the btl-Gal4 driver suppresses cerebral branch formation (red star, middle panels). The effect of the AbdA^TD/UA variants is illustrated (arrow, lower panels). Right panels are magnifications of boxed thoracic areas. B. Double immunostaining for AbdA (red) and Doc1 (green) in wild type embryo (upper panel). Magnifications of thoracic segments T2/T3 and abdominal segments A1–3 (middle and lower panels). Expression of AbdA in thoracic segments through the 24B-Gal4 driver promotes a six cell lineage state, with the two anterior most cells expressing Doc1 (middle panels). The effect of the AbdA^HX/UA variants is illustrated (lower panels). Graphs in A and B (% of remaining activities compared to the wild type AbdA protein (WT) following domain mutations) using the boxplot representation summarize quantitative analyses (see Text S1 and Figure S5 (cerebral branch) and S6 (heart lineage) for full illustration). A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity). C. Abdominal hemi-segments in the cardiac tube are composed of six cardiac cells, labeled in blue by Mef2. The two most posterior cells express Doc1 (green). The thoracic hemi-segments lack the anterior Doc1 positive cells. This distinction between thoracic and abdominal segments is not affected following maternal and zygotic loss of exd.

Figure 7. Mutually suppressive interaction of protein domains.

A. Thoracic restricted expression of Dll (white arrows) followed by Dll enhancer driven β-gal (green) results from repression by AbdA (red) in the abdomen (upper panel). Ubiquitous AbdA expression driven by arm-Gal4 represses Dll thoracic expression (red arrows, middle panels). The effect of the AbdA^HX/UA variants is illustrated (lower panels). Right panels are magnification of boxed thoracic areas. B. Increased thoracic Antp expression (green, white arrows) results from AbdA (red) repression in the abdomen (upper panels). Ubiquitous AbdA expression driven by arm-Gal4 represses Antp expression in the thorax (red arrows, middle panels). The effect of the AbdA^HX/UA variants is illustrated (lower panels). Right panels are magnification of boxed thoracic areas. C. Abdominal segments are characterized by refringent denticles organized in a trapezoidal shape in segments A2 but not A1, while T2/T3 thoracic segments harbors thinner denticles (left panel). Upon AbdA thoracic expression driven by arm-Gal4, the first abdominal segment A1 and thoracic segments acquire abdominal features, including abdominal type of denticles, trapezoidal organization of denticles and suppression of a T1 specific feature (white arrow), the “beard” (middle panel). Full or intermediate transformations were observed for AbdA variants (see Text S1 for quantifying criteria). The effect of the AbdA^HX/TD variants is illustrated (right panel). Weak A1 (wA1) stands for a transformation of thoracic denticles toward abdominal type of denticles, with an organization typical of A1, but with only a partial suppression of the beard in T1 (arrow). D. Snapshots from movies illustrating locomotion in wild type larvae (left panels), or in larvae expressing ubiquitously AbdA (middle panels) or AbdA^HX/UA variant (right panels) driven by the arm-Gal4 driver. White boxed areas show the progression of a peristaltic waves in the abdomen. The red boxed area shows an ectopic peristaltic wave in the thorax following ectopic AbdA expression in the thorax. Graphs in A–D (% of remaining activities compared to the wild type AbdA protein (WT) following domain mutations) using the boxplot representation summarize quantitative analyses (see Text S1 and Figure S7 (Dll), S8 (Antp), and S9 (A2 epidermal morphology) for full illustration, and Figure S10 for data on larval locomotion experiments. A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity).

Figure 8. Hierarchical clustering of AbdA domain requirements.

Hierarchical clustering of domain contribution reveals two functional modules and a predominant role for the UA domain. A graded color-coded bar above the graphs illustrates the level of protein activity, ranging from light green (full activity) to black (no activity). Neomorphic activity is depicted in red. Asterisks indicate Exd dependent biological readouts. Jacknife values are indicated for each node.

Figure 9. Phylogeny of the HX, TD, and UbdA protein domains.

Abbreviations are as follows: B: bilaterians; P: protostomes; D: deuterostomes; L: lophotrochozoa; E:ecdysozoa; Cy, cycloneuralia. PA: pan-arthopods; O: onycophores; T: tardigrades; Ch: chelicerates; M; myriapods; Cr: crustaceans; H:hexapods. Sequence alignment spanning the HX, TD and UA domains are shown for representatives of the four main arthropod branches (Tc: Tribolium castaneum; Mr: Myrmica rubra; Ag: Anophela gambiae; Dm: Drosophila melanogaster) and for a representative of deuterostomes (Mus musculus, Mm) and lophotrochozoa (Hirudo medicinalis, Hme).