|Home | About | Journals | Submit | Contact Us | Français|
The Bicoid morphogen evolved approximately 150 MYA from a Hox3 duplication and is only found in higher dipterans. A major difference between dipteran species, however, is the size of the embryo, which varies up to 5 fold. Although the expression of developmental factors scale with egg length, it remains unknown how this scaling is achieved. To test whether scaling is accounted for by the properties of Bicoid, we expressed eGFP fused to the coding region of bicoid from three dipteran species in transgenic Drosophila embryos using the Drosophila bicoid cis-regulatory and mRNA localization sequences. In such embryos, we find that Lucilia sericata and Calliphora vicina Bicoid produce gradients very similar to the endogenous Drosophila gradient and much shorter than what they would have produced in their own respective species. The common shape of the Drosophila, Lucilia and Calliphora Bicoid gradients appears to be a conserved feature of the Bicoid protein. Surprisingly, despite their similar distributions, we find that Bicoid from Lucilia and Calliphora do not rescue Drosophila bicoid mutants, suggesting that that Bicoid proteins have evolved species-specific functional amino acid differences. We also found that maternal expression and anteriorly localization of proteins other than Bcd does not necessarily give rise to a gradient; eGFP produced a uniform protein distribution. However, a shallow gradient was observed using eGFP-NLS, suggesting nuclear localization may be necessary but not sufficient for gradient formation.
The determination of body size is a fundamental problem in developmental biology (Calder, 1984; McMahon and Bonner, 1983; Peters, 1983). Size affects the developmental dynamics of all aspects of biological structure and function, and one approach to further our understanding of these dynamics is through the problem of scaling (Schmidt-Nielsen, 1983; West et al., 1997). In particular, how scaling of the spatial patterns in segmented, multicellular organisms of different sizes (but with similar genetic background) is achieved during embryonic development. In well studied systems such as Drosophila melanogaster, and in other insects with long germband embryos, the segments develop almost simultaneously from periodic patterns of gene expression, which are activated by earlier gradients of proteins such as Bicoid (Bcd) that span the developing embryo (Peel et al., 2005). Although, Bcd has emerged as a paradigm for morphogen gradients, how it scales during embryogenesis is not fully understood (Gregor et al., 2005; Lott et al., 2007; Gregor et al., 2007a).
Maternally transcribed bcd mRNA is localized to the anterior pole of the Drosophila embryo and upon translation the protein forms an anterior to posterior gradient in the syncytial blastoderm (Berleth et al., 1988; Driever and Nüsslein-Volhard, 1988a; Frohnhöfer and Nüsslein-Volhard, 1986). Bcd encodes a transcription factor and it is thought to act as a classic morphogen through the concentration dependent activation and repression of target genes such as hunchback along the anterior-posterior (A-P) axis (Driever and Nüsslein-Volhard, 1988b; Driever and Nüsslein-Volhard, 1989; Struhl et al., 1989).
bcd is only found in higher dipterans and functional studies indicate that its role is conserved between distantly related members of this taxon, such as Megaselia abdita, Drosophila and the calyptratae species, Musca domestica, Calliphora vicina and Lucilia sericata, reflecting the overall similarity of their embryogenesis (Heming, 2003; McGregor, 2005; Shaw et al., 2001; Stauber et al., 1999; Stauber et al., 2002; Stauber et al., 2000). However, the regulatory interactions between Bcd and the enhancers of its target genes appear to have co-evolved in various dipterans (Hancock et al., 1999; McGregor et al., 2001; Shaw et al., 2002; Wratten et al., 2006) and so Bcd proteins from different flies may not be functionally equivalent (Schröder and Sander, 1993; Shaw et al., 2002).
A major difference between dipterans relevant to understanding the scaling of the Bcd gradient is the size of their embryos. The length of dipteran embryos has evolved rapidly within and between species and can vary by up to 5 fold, from 300 μM in Drosophila busckii to 1500 μM in Calliphora vicina. Despite this difference in size, the Bcd gradient and the expression of segmentation genes scale with embryo length (EL) in dipteran embryos (Gregor et al., 2005; Lott et al., 2007; Sommer and Tautz, 1991). However, the mechanism through which this scaling is achieved is not known. We have recently shown that a simple model based on the gradient being formed by a balance of localized synthesis, diffusion, and spatially uniform degradation cannot explain the scaling properties of the gradients observed in different species, unless the protein lifetimes of each species correlate with embryo size (Gregor et al., 2005).
Here we investigated if the scaling of the Bcd gradient in species with different ELs is dependent on particular properties of the Bcd protein from these species. To do this we compared the gradients produced by Bcd from Drosophila, Calliphora and Lucilia as eGFP fusion proteins in transgenic Drosophila embryos. We found that Calliphora and Lucilia Bcd formed Drosophila Bcd shaped gradients in Drosophila embryos, which suggests the Bcd proteins from these species all have the very similar properties with respect to gradient formation. However, Bcd from Lucilia and Calliphora was unable to rescue Drosophila bcd mutants supporting the hypothesis that interactions involving Bcd have co-evolved in these flies. We also tested whether eGFP or eGFP-NLS proteins could form a Bcd-like gradient when expressed maternally and localized to the anterior pole using the Bcd regulatory sequences. While eGFP protein was found almost uniformly throughout the embryo, eGFP-NLS was able to produce a shallow gradient different from that produced by Bcd protein itself. The eGFP-NLS profile suggests that nuclear localization may be necessary but is not sufficient for normal gradient formation.
To clone the complete coding regions of the Calliphora and Lucilia bcd genes, 5’ and 3’ RACE was performed using primers based on the previously known sequences and cDNA isolated from adult females (Schröder and Sander, 1993; Shaw et al., 2001). The coding regions of Calliphora and Lucilia bcd, in each case immediately followed by a stop codon, were amplified with primers containing SphI sites and subsequently cloned in frame with eGFP in the modified version of pNBGA1 (Barolo et al., 2000; Gregor et al., 2007a; Hazelrigg et al., 1998), which had a section of Drosophila bcd removed by BmgB1 and BglII digestion and re-ligation. The fragment containing the remaining Drosophila bcd genomic region (including all regulatory and localization sequences) with eGFP and either Calliphora or Lucilia bcd inserted was then subcloned into pCaSpeR4 (Thummel and Pirrota, 1992) using BamHI and either KpnI or XhoI respectively (Figure S1). Similarly, eGFP in pNBGA1 was replaced with eGFP followed by a stop codon or eGFP-NLS (amplified from pStinger (Barolo et al., 2000)) followed by a stop codon. The bcd genomic region including these insertions was then subcloned into pCaSpeR4 using BamHI and EcoRI (Fig. S1).
A fragment of 2.5 kb from directly upstream of the Calliphora hb transcription start site, which contains all 9 characterized Bcd-binding sites and a 240 bp sub-fragment containing 8 of these sites were amplified by PCR from Calliphora genomic DNA (McGregor et al., 2001). Primer sequences are available on request. After sequencing of these fragments, enhancer fragment-lacZ fusion constructs were made by subcloning each fragment into pCaSpeR-lacZ.
Standard P-element transformation was used to introduce the constructs into yw or w118 strains of D. melanogaster (Spradling and Rubin, 1982). At least three balanced independent lines were established for each construct.
All embryos were collected at 25 °C, heat fixed, and protein expression visualized using rat anti-Bcd and rabbit anti-Kr antibodies (gifts from J. Reinitz) (Kosman et al., 1998), and mouse anti-Paired/Pax (gift from N. Patel). Secondary antibodies used were conjugated with Alexa-488, Alexa-546 and Toto3 (Molecular Probes), respectively.
Expression of eGFP-Bcd fusion, eGFP-STOP, eGFP-NLS and lacZ reporter constructs was detected with in situ hybridization using anti-sense eGFP and lacZ DIG labeled probes respectively (Tautz and Pfeifle, 1989).
For experiments with flies expressing Drosophila bcd fused to eGFP we used a stock with an insertion on the X chromosome. For substitution of endogenous bcd we conducted the mutant crosses of eGFP-bcd with bcdE1,pp/TM3,Sb to generate egfp-bcd;bcdE1,pp.
To determine whether the eGFP-Bcd transgenes derived from Calliphora and Lucilia rescued the female sterility of the Drosophila null mutant bcdE1, each transgene was introduced as a single copy into a homozygous bcdE1 background. Embryos derived from such females were scored for hatching and segmentation patterns of cuticles from unhatched embryos were analyzed after preparation in Hoyers-lactic acid. As a control for this experiment we tested three independently isolated lines containing Drosophila eGFP-Bcd, as well as five new lines derived by mobilizing the original X chromosome insertion. All but one of the Drosophila lines rescued the bcd sterility. To determine whether the foreign transgenes could drive expression of the Calliphora hb-lacZ reporter constructs, females homozygous for the eGFP Calliphora or the eGFP Lucilia Bcd transgene, or heterozygous for the eGFP Lucilia Bcd transgene and the CyO, hb-lacZ balancer, were mated to males carrying the Calliphora hb-lacZ reporter construct and lacZ expression examined by in situ hybridization.
Stained embryos were imaged with standard epi-fluorescence microscopy, using Nikon 10x air (Lucilia and Calliphora) and 20x air (Drosophila) objectives. Embryos were mounted in AquaPolymount (Poly-sciences, Inc.). Embryos were placed under a cover slip and the image focal plane of the flattened embryo was chosen at the top surface and at the mid-sagittal plane (Figs. 1, ,2,2, and S2 show a projection of both images).
Live imaging was performed on embryos from transgenic lines expressing eGFP using a custom-built two-photon microscope as previously described (Gregor et al., 2007a). Images were taken with a Zeiss 25x (NA 0.8) oil/water-immersion objective and an excitation wavelength of 900-920 nm. Average laser power at the specimen was 15-35 mW. Embryos were imaged at late cell cycle 14 (when nuclear eGFP-NLS concentration was highest); three high-resolution images (512 × 512 nm pixels, with 16 bits and at 6.4 μs per pixel) were taken along the A-P axis (focused at the mid-sagittal plane) at magnified zoom and then stitched together in software (MATLAB, MathWorks, Natick, MA); each image is an average of 6 sequentially acquired frames (Gregor et al., 2007b).
Nuclear eGFP profiles were extracted as we previously described (Gregor et al., 2007b). Nuclear centers were hand selected and the average nuclear fluorescence intensity was computed over a circular window of fixed size.
To analyze the profile of the eGFP-NLS construct, we solved numerically the discretized 1D diffusion equation on a grid with reflecting boundary conditions. Grid points were separated by a linear distance of 1/100 of the embryo length. The initial condition was uniform zero concentration throughout the embryo. A nonlinear fitting routine (Levinberg-Marquard within Matlab, MathWorks, Natick, MA) was used to adjust the diffusion coefficient to best match the nuclear fluorescence profile. For the fit, we chose a single free parameter, the diffusion constant, and we allowed a normalization parameter. Background fluorescence was chosen to be zero at the posterior pole for an adjacently imaged Bcd-GFP expressing embryo. The fit was forced to reproduce the posterior fluorescence level of the eGFP-NLS profile after a simulation time of 200 min (roughly the age of the embryos when imaged).
Although, the embryos of Calliphora and Lucilia are up to three times longer and three times wider than those of Drosophila, in all three species the cellularizing blastoderm consists of approximately 6000 nuclei and early development is very similar. As previously shown for Lucilia and Drosophila (Gregor et al., 2005), the embryos of all three species exhibit the same overall proportionally scaled pattern, such that the size of expression domains and the distances between them are approximately three times higher in the larger embryos. However, the Calliphora and Lucilia embryos do show a slight anterior-ward shift from the position expected for perfect scaling. To quantify this shift we measured the location of Paired stripe 1, which occurs at 31±1% EL (mean ± s.d.; N=17 embryos) in Drosophila and marks the future position of the cephalic furrow/head fold (Fig. 1). In Lucilia, this stripe is found at 25±1% EL (mean ± s.d.; N=9) from the anterior pole and in Calliphora at 27±2% EL (mean ± s.d.; N=17). Consistent with a small head region in the larger dipteran embryos, the anterior margin of the Krüppel (Kr) expression domain (which corresponds roughly to the beginning of the thorax) is shifted to ~42% EL in Lucilia from ~50% observed in Drosophila (Fig. S2). Expressed in absolute distance, however, the shifts we have found represent relatively subtle deviations from the remarkable scaling of pattern between species whose ELs differ by three fold.
To follow the timing of nuclear replications and cytoplasmic movements in larger dipteran embryos we examined movies in which embryos from Lucilia and Calliphora had been injected anteriorly with fluorescently labeled dextran. These embryos had been previously used to extract diffusion constants in the different species (Gregor et al., 2005). Here the injected dextran serves to highlight cytoplasmic movements and nuclear divisions (Fig. S3).
Despite the difference in size, both large and small embryos show remarkably similar periodic 9 to 13 min falls in dextran intensity, as a result of cytoplasmic turbulence during mitosis. This indicates that the duration of interphase and mitosis (the underlying cell cycle) is the same in large and small embryos. Mitosis pauses in all examined species after 13 rounds of division at cycle 14A, and in embryos fixed during cycle 14 and stained with DAPI there are about 100 rows of nuclei along the A-P axis of all species. Because the embryos are 3 times longer, the average internuclear distance in Lucilia and Calliphora (~24 μm) is three times longer than in Drosophila. The nuclear diameter in the larger species is also about three times larger, corresponding to a 27-fold increase in nuclear volume (Gregor, 2005). Note that genome sizes of Calliphora and Lucilia are approximately 5 times larger than the Drosophila genome (Ullerich and Schottke, 2006).
In all three species, syncytial nuclei are uniformly spaced along the A-P axis when the nuclei first migrate into the cortex. By early cycle 14, densities in the future head region of the embryo are lower - a broadened spacing known in Drosophila to correlate with anteriorly localized Bcd activity (Blankenship and Wieschaus, 2001). We previously showed that Bcd has a similar distribution in Drosophila and Lucilia embryos (Gregor et al., 2005). To characterize the distribution of Bcd protein in Calliphora, we stained early embryos from this species with various antibodies raised against Drosophila Bcd. The only reliably reproducible staining was obtained with a rat polyclonal serum made with the C-terminal region of the Drosophila Bcd protein (Kosman et al., 1998) and even with this antiserum the staining was significantly weaker in the larger flies than that observed in Drosophila (Fig. 2). The high background in Calliphora embryos made it difficult to determine accurately how far the gradient extended into the embryo. The minimal estimate for the posterior extent was 30% EL (the point at which the signal to noise ratio was such that staining intensity was clearly distinct from background intensity level, as estimated at the posterior end of the stained embryo), which when translated into absolute distance indicates that Calliphora Bcd, like Lucilia Bcd (Gregor et al., 2005), moves significantly farther in Calliphora eggs than the corresponding protein does in Drosophila (Fig. 2). Our results show that when distributions are expressed as percent EL, a comparable gradient of Bcd is seen in dipteran embryos that vary in size by 3-fold. This means that on average Bcd molecules have to travel 3 times further in embryos of some species compared to others.
One explanation for how Bcd travels further in larger embryos is that the Bcd proteins from such species have properties that make them more stable than Bcd proteins from species with smaller embryos. In order to test this hypothesis we first cloned the full-length bcd coding regions from Lucilia and Calliphora.
Previous studies showed that the bcd homeodomains of calyptrate species (Musca, Lucilia and Calliphora) have 4 amino acid differences from Drosophila and that Musca and Calliphora each have single additional differences (Schröder and Sander, 1993; Shaw et al., 2001). In addition, the PEST domain, a proline, glutamate, serine and threonine rich domain, thought to be involved in protein turnover (Rechsteiner and Rogers, 1996), has 69% sequence identity between the calyptrate species and Drosophila. The N-terminal self-inhibition is conserved between these species, but in the eIF4E/d4EHP binding domain, which is involved in blocking caudal translation, a key amino acid in Drosophila (L73) is not conserved (Cho et al., 2005; Niessing et al., 2002; Zhao et al., 2003) (Fig. S4). The C-terminal regions are more diverged although there are blocks of conservation in the Acidic domain, a transcriptional activation domain (Fig. S4) (Janody et al., 2001).
We then used the Calliphora and Lucilia bcd coding region sequences to make constructs that expressed Bcd from each species as eGFP fusion proteins (Cv-eGFP-Bcd and Ls-eGFP-Bcd respectively) (Fig. S1). To ensure that they would be properly expressed during oogenesis and localized to the anterior region of the Drosophila embryo we used the Drosophila bcd regulatory regions (Hazelrigg et al., 1998; Gregor et al., 2007a). We previously showed that a similar Drosophila construct (hereafter called Drosophila eGFP-Bcd) can fully rescue bcd mutants (Gregor et al., 2007a), and therefore this was used as a control.
If the broadened expression of Bcd in Lucilia and Calliphora reflects an autonomous feature of the Bcd protein in these species, we would expect the proteins to form a Lucilia or Calliphora scale of gradient in Drosophila eggs. This did not appear to be the case (Fig. 2). Instead, each made similar gradients to that of the endogenous nuclear Drosophila Bcd gradient, which suggests gradient formation in different sizes of embryos is not dependent on species-specific properties of Bcd proteins.
Previous cytoplasm transfer experiments suggested that Bcd from Lucilia and Calliphora was unable to rescue Drosophila bcd mutants (Schröder and Sander, 1993). This seems somewhat surprising given the conserved functional domains of the Lucilia and Calliphora Bcd proteins and the Drosophila-like profile of the gradients they form in Drosophila embryos. Therefore, we then tested the activity of the Ls-eGFP-Bcd and Cv-eGFP-Bcd in Drosophila bcd mutants, again using Drosophila eGFP-Bcd as a control.
Although the activity levels of the transgenic Drosophila eGFP-Bcd lines varied with insertion site, 7 of the 8 tested lines rescued the sterility of homozygous null (bcdE1) females, yielding normal patterned larvae that developed to adult stages (not shown). In contrast, none of the lines of Ls-eGFP-Bcd (7) or Cv-eGFP-Bcd (4) tested rescued embryos to hatching larvae. Cuticle preparation showed some thoracic and head structures in all the Lucilia lines and in the strongest cases rescued embryos had up to two thoracic segments and tiny residual mouth hooks, structures never observed in embryos from bcdE1 mothers without the transgene (Fig. 3a,c). The four Calliphora lines showed little sign of rescue even at the cuticle level (Fig. 3b).
To determine if these differences in the ability of Drosophila eGFP-Bcd, Ls-eGFP-Bcd and Cv-eGFP-Bcd to rescue Drosophila bcd mutant embryos was due to differences in the expression levels of these constructs, we performed in situ hybridizations using an eGFP probe. We found that embryos from Drosophila eGFP-Bcd, Ls-eGFP-Bcd and Cv-eGFP-Bcd lines all gave strong anteriorly localized expression, although Ls-eGFP-Bcd gave slightly weaker expression than the other two constructs in the lines tested (Fig. 4). We conclude that the differences in rescue are due to functional differences in the Bcd proteins from each species rather than variation in expression levels. These transgenic results corroborate the results of the earlier cytoplasm transfer experiments between these species (Schröder and Sander, 1993).
The lack of activity of Calliphora Bcd in Drosophila embryos suggests that it contains amino acid changes that prevent it from regulating the expression of the normal Drosophila Bcd target genes. Therefore we asked whether the Cv-eGFP-Bcd could interact with the enhancers of a presumed Calliphora target gene in Drosophila embryos, using two reporter constructs of the Calliphora hunchback (hb) enhancer fused to lacZ. One construct, from the Calliphora hb transcription start site to 2.5 kb upstream containing all 9 characterized Bcd-binding sites, and a second 240 bp construct containing only the cluster of 8 Bcd-binding sites (McGregor et al., 2001). When introduced into Drosophila embryos that contain only Drosophila Bcd, both transgenes were only weakly expressed, indicating that the Calliphora hb enhancer is not completely functional in Drosophila embryos or fully activated by the endogenous Drosophila Bcd protein (Fig. 3d). We then tested whether the expression could be activated in Drosophila embryos containing Ls-eGFP-Bcd or Cv-eGFP-Bcd. In neither case did either construct appear to alter levels of lacZ expression or affect the position of the expression domains (Fig. 3e,f).
It is possible that any localized RNA would produce a stable protein gradient identical to the ones observed above when translated and allowed to diffuse from the anterior end of the embryo. To test this possibility, we introduced a stop codon between the last codon of the N-terminal eGFP and the ATG of Drosophila bcd in the construct we had previously used to make a fusion protein identical to endogenous Bcd (Gregor et al., 2007a). Hence the anteriorly localized mRNA from this new construct would only give rise to eGFP (Fig. S1). Three transgenic lines expressing eGFP-STOP were obtained and all lines showed fluorescence at very low, apparently uniform, levels along the entire A-P axis of the embryo (Fig. S5). This was not the result of mis-localization as the transcripts were anteriorly localized in all three lines (Fig. S6). It is, however, possible that the shallowness results from the weakness of the protein expression level, which is barely above the background fluorescence of these transgenic lines. In that respect our result remains inconclusive.
To test whether nuclear localization might allow eGFP to form a stable gradient similar to Bcd we made a second construct in which a nuclear localization signal (NLS) coding sequence was inserted in frame at the end of the eGFP sequence. A stop codon was included between eGFP-NLS and the first codon of bcd (Fig. S1). The attachment of the NLS to eGFP appeared to increase the levels of the protein and its nuclear localization made it easier to detect. Again the eGFP-NLS transcripts were anteriorly localized in all lines tested (Fig. S6). In contrast to the uniform levels of protein observed with eGFP alone, embryos expressing eGFP-NLS did produce a shallow gradient extending the entire length of the embryo (Fig. 5).
However, unlike eGFP-Bcd, the level of eGFP-NLS continued to rise during late cleavage cycles and persisted until gastrulation. This apparently greater stability allowed significant accumulation of the eGFP-NLS protein at the posterior pole (Fig. 5). Although this distribution differs from the Bcd gradients in normal developing embryos, both the stability and accumulation of the eGFP-NLS protein and its shallower distribution are observed when Drosophila eGFP-Bcd is expressed in unfertilized eggs that lack nuclei (Gregor et al., 2007a). Moreover, the shape of the observed eGFP-NLS gradient fits very closely the shape of a gradient obtained from a numerical solution of the diffusion equation (see Materials & Methods and Gregor et al., 2005) with a diffusion constant of 6±1 μm2/s. In this simulation we omitted protein degradation, which leads both to a more shallow final gradient shape and to an overall rise of protein concentration within the embryo. Note that a diffusion constant of 6±1 μm2/s would be large enough to establish a Bicoid gradient within the available developmental time by a passive diffusion-based mechanism (Gregor et al., 2007a).
These combined experiments argue that while nuclear localization per se is not completely sufficient for any anteriorly localized RNA to form a Bcd-like gradient, the presence of nuclei is required.
Despite differences of up to 5 fold in embryo length, the expression patterns of segmentation genes appear to scale with embryo length in higher dipteran embryos (this work; Gregor et al., 2005; Lott et al., 2007; Sommer and Tautz, 1991). However we did observe a slight anterior-ward shift in expression patterns in the larger embryos of Calliphora and Lucilia compared with Drosophila. Indeed there is evidence that scaling need not be perfect and that other mechanisms can compensate. While the perturbation of normal scaling in Drosophila embryos from females containing up to 6 copies of Bcd resulted in a posterior shift in patterning and an expanded head region, these embryos developed into apparently normal larvae and flies (Berleth et al., 1988; Frohnhöfer and Nüsslein-Volhard, 1986). It has been shown that this is due to a fate map repair system involving increased cell death in expanded regions (Namba et al., 1997). This suggests that some variation in the scaling of embryonic pattern can be permitted and this may explain how differences in scaling with respect to embryo length evolved in sibling species of the D. melanogaster complex (Lott et al., 2007).
It has been shown that the scaling of patterning with embryo length is controlled by maternal factors, which must have evolved in concert with embryo length in dipteran embryos (Gregor et al., 205; Lott et al., 2007). Therefore we tested if Bcd from large embryos could travel further along the A-P axis than Bcd from smaller embryos. However, we found that the Calliphora and Lucilia Bcd proteins actually adopted the profile of the Drosophila Bcd gradient in Drosophila embryos. Although we cannot rule out the possibility that the protein degradation machinery is less potent in Calliphora and Lucilia, our results indicate that some feature of the embryo in which the protein finds itself controls its distribution and not species-specific Bcd stability. This also suggests that the coordinated evolution of embryo length and scaled patterning is independent of the Bcd protein at least between distantly related dipterans.
We found that while Drosophila eGFP-Bcd was able to fully rescue Drosophila bcd mutant embryos, Ls-eGFP-Bcd and Cv-eGFP-Bcd gave little or no rescue of bcd mutants or shifts in morphological markers in normal embryos, supporting the results of cytoplasm transfer experiments by Schröder and Sander (1993). These results are somewhat surprising given the gradients made by Ls-eGFP-Bcd and Cv-eGFP-Bcd in Drosophila embryos and the conservation of functional protein domains. Indeed, it has been shown that even a Drosophila bcd transgene expressing only the first 246 amino acids can rescue bcd mutants (Schaeffer et al., 1999). However, previous in vivo and in vitro experiments of binding affinity and transcriptional output indicated that the protein-DNA interactions between Bcd and Bcd-binding sites in the enhancers of target genes have co-evolved in dipterans (McGregor et al., 2001; Shaw et al., 2002).
We also found that Calliphora Bcd protein failed to activate the Calliphora hb target in Drosophila embryos (although it is possible that our reporter constructs may not contain the complete Calliphora hb enhancer and/or activation of this enhancer requires other Calliphora transcription factors such as Hb (Simpson-Brose et al., 1994)). This suggests that the protein-protein interactions between Bcd and transcription co-factors have also co-evolved as has been described between other systems (Ruvinsky and Ruvkun, 2003). Therefore, while foreign Bcd proteins can produce distributions characteristic of Drosophila Bcd when introduced into Drosophila embryos, productive interactions may no longer be possible in the Drosophila cytoplasm.
Experiments by Crauk and Dostatni (2005) indirectly demonstrated that the synthetic transcription factor GAL4-3GCN4 was able to make an anterior-posterior gradient when anteriorly localized in Drosophila embryos. While this protein is able to localize to the nucleus and activate transcription, it has no obvious similarities to Bcd at the amino acid level. This suggests that while transcriptional activity per se is not required for gradient formation, as demonstrated by our Cv-eGFP-Bcd construct, transcriptional activation domains in combination with nuclear localization (see below) are required. Bcd has at least 4 transcriptional activation domains, and while it has been shown that there is some redundancy in their function (Fig. 3; Schaeffer et al., 1999) each of these domains could be involved in gradient formation.
Here we also tested whether any anteriorly localized source generates a gradient and the effect of nuclear localization on gradients. We found a remarkable absence of any anterior to posterior graded protein distribution in embryos that express a localized source of eGFP at the anterior pole. However, if these anteriorly emanated eGFP molecules are tagged with a NLS we do see a gradient resembling the Bcd gradient seen in unfertilized eggs (Gregor et al., 2007a): it peaks at the anterior pole, falls off more shallowly than a Bcd gradient in fertilized eggs, and it rises along the entire A-P axis over time. This result suggests that nuclear localization sequences are needed for gradient formation, but are not sufficient to recover the shape of the wild type Bcd gradient. Some other Bcd-specific property, such as degradation rate, perhaps mediated by its transcriptional activation and/or PEST domains, seems to be relevant to distinguish its shape from gradients of both unfertilized eggs and eGFP-NLS expressing embryos.
Bcd-fusion protein expression constructs. All constructs in this study are based on pCas-eGFP-bcd, which contains a 6.5 kb genomic region of Drosophila bcd with eGFP inserted in frame with the first codon of bcd (Gregor et al., 2007a; Hazelrigg et al., 1998). Drosophila bcd exons are shown as red boxes, introns by diagonal lines and upstream and downstream genomic regions by black horizontal lines. The positions of restriction sites are indicated, as are stop codons introduced in to constructs in this study. Restriction sites removed in the synthesis of constructs are shown with a line through them. Boxes representing regions coding for eGFP (green), NLS (yellow), Lucilia bcd (purple) and Calliphora bcd (blue) in pCas-eGFP-STOP, pCas-eGFPNLS-STOP, pCas-eGFPLuciliabcd, pCas-eGFPCalliphorabcd are indicated.
Krüppel expression in Lucilia and Drosophila embryos. Optical mid-sagittal sections (a, c and e) and surface views (b, d, and f) of Kr protein localization in Lucilia (a-d) and Drosophila (e and f) embryos at blastoderm stages.
Intensity profiles of fluorescently-labeled dextran molecules, injected at the anterior pole of dipteran eggs, as a function of time after injection. The intensity is extracted from a time series of images of the injected egg with the focal plane at the top surface. For each time point, average pixel intensities over a circular window of the size of a nucleus, located at half the embryo length along the A-P axis, are reported. The top and bottom panels show time courses for a Drosophila and a Calliphora injection, respectively. Three arrowheads in each plot indicate-from left to right-mitosis 11, 12 and 13. Due to cytoplasmic turbulence during mitosis, the intensity gradient is uniformly translated anteriorly at the onset of mitosis and posteriorly at the end of mitosis, which generates the prominent dips in the time course profiles indicated by the arrowheads. The time between adjacent arrowheads indicates the length of cell cycles 12 and 13. In the case of Drosophila (upper panel) the times are ~660 and ~770 seconds and in the case of Calliphora (lower panel) ~670 and ~780 seconds, for cell cycles 12 and 13, respectively.
Alignment of the Drosophila, Musca, Lucilia and Calliphora Bicoid proteins. The homeodomain is highlighted in red. The self-inhibition domain (SID) is highlighted in green. The eIF4E/d4EHP binding domain is highlighted in purple. The PEST domain (highlighted in black) contributes to Ubiquitin dependent protein degradation. The glutamine (Q)-rich and C-terminal acidic transcriptional activation domains are both shown in yellow and the alanine (A)-rich transcriptional repression domain is shown in orange. Conserved amino acids are indicated by asterisks and amino acids with similar properties with colons and dots. Alignment was made using ClustalW (Thompson et al., 1994). Sequence accession numbers X07870, AJ297853, AJ297855 and AJ297856).
Comparison of fluorescence intensity profiles of eGFP-STOP expressing embryos and wild type (Oregon R) embryos. 15 min old embryos were imaged for 3.5 h using two-photon microscopy. Images were taken every minute. Gradients were extracted by sliding a window (in software) along the edge of the embryo and by reporting the average intensity in the window as a function of position along the A-P axis (normalized by egg length L). Time is color coded in the figure, see color-bar at right-hand side. eGFP-STOP expressing embryos (a) show a flat gradient with increasing intensity over time. Decreased intensity in the mid-section (20-80% egg length) is due to optical property changes of the embryo in this region versus the pole regions (see also Gregor et al. 2007a). Wild type embryos were only visible at all at elevated laser power and when the gain of the microscopes PMT was put to maximal position (b). They also show a flat intensity profile, but the intensity does not rise over time as in the eGFP-STOP expressing embryos.
Expression of eGFP-STOP and eGFP-NLS construct transcripts in transgenic Drosophila embryos. eGFP-STOP (a). eGFP-NLS (b). All embryos are at the syncytial blastoderm stage and are shown dorsal up and anterior to the left.
We thank W. Bialek, W. Damen, N. Patel, J. Reinitz, R. Samanta, D. Stern, and D. W. Tank. This work was partially supported by the MRSEC Program of the NSF under Award Number DMR-0213706, and by NIH grants P50 GM071508 and R01 GM077599.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.