Expression cloning screening of a unique and full-length set of cDNA clones is an efficient method for identifying genes involved in Xenopus neurogenesis

Mechanisms of Development 122 (2005) 289–306 www.elsevier.com/locate/modo

Expression cloning screening of a unique and full-length set of cDNA clones is an efficient method for identifying genes involved in Xenopus neurogenesis Jana Voigtb, Jun-An Chenc, Mike Gilchrista, Enrique Amayac, Nancy Papalopulub,* a

Wellcome Trust/Cancer Research UK Gurdon Institute, Tennis Court Road, Cambridge CB2 1QR, UK b Department of Anatomy, University of Cambridge, Downing Site, Cambridge CB2 3DY, UK c Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK Received 28 October 2004; accepted 4 November 2004

Abstract Functional screens, where a large numbers of cDNA clones are assayed for certain biological activity, are a useful tool in elucidating gene function. In Xenopus, gain of function screens are performed by pool screening, whereby RNA transcribed in vitro from groups of cDNA clones, ranging from thousands to a hundred, are injected into early embryos. Once an activity is detected in a pool, the active clone is identified by sib-selection. Such screens are intrinsically biased towards potent genes, whose RNA is active at low quantities. To improve the sensitivity and efficiency of a gain of function screen we have bioinformatically processed an arrayed and EST sequenced set of 100,000 gastrula and neurula cDNA clones, to create a unique and full-length set of approximately 2500 clones. Reducing the redundancy and excluding truncated clones from the starting clone set reduced the total number of clones to be screened, in turn allowing us to reduce the pool size to just eight clones per pool. We report that the efficiency of screening this clone set is five-fold higher compared to a redundant set derived from the same libraries. We have screened 960 cDNA clones from this set, for genes that are involved in neurogenesis. We describe the overexpression phenotypes of 18 single clones, the majority of which show a previously uncharacterised phenotype and some of which are completely novel. In situ hybridisation analysis shows that a large number of these genes are specifically expressed in neural tissue. These results demonstrate the effectiveness of a unique full-length set of cDNA clones for uncovering players in a developmental pathway. q 2004 Published by Elsevier Ireland Ltd. Keywords: Gain of function; Screen; Xenopus; Neurogenesis; Rab32; XER81; Frizzled 10B; Cullin1

1. Introduction Functional screens are valuable tools for identifying genes involved in specific developmental processes. In screens performed on a large scale, genes with similar functions can be grouped together, regulatory pathways suggested and interesting candidates selected for further study. Indeed, large-scale functional screens, involving either gain or loss of function, have been successfully carried out in a number of experimental model organisms. * Corresponding author. Address: Wellcome Trust/Cancer Research UK Gurdon Institute, Tennis Court Road, Cambridge CB2 1QR, UK. Tel.: C44 1223 334126; fax: C44 1223 334089. E-mail address: [email protected] (N. Papalopulu). 0925-4773/$ - see front matter q 2004 Published by Elsevier Ireland Ltd. doi:10.1016/j.mod.2004.11.002

Loss of function screens have been carried out in Drosophila, zebrafish and, more recently, in mouse and provided us with a wealth of information on developmental processes (Haffter et al., 1996; Nusslein-Volhard and Wieschaus, 1980; Driever et al., 1996; Kile et al., 2003; Mitchell et al., 2001; Skarnes et al., 2004). In recent times, the availability of genome sequence data and ORF prediction has allowed more direct, gene based, loss of function screens. For example, RNAi has been used in C. elegans and in Drosophila cell lines to knock down predicted genes on a genome wide basis (Boutros et al., 2004; Fraser et al., 2000; Gonczy et al., 2000; Kamath et al., 2003; Lum et al., 2003). On a smaller scale, antisense oligo (morpholino-MO) based knock-down screens have been carried out in Ciona and Xenopus and should also be

290

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

feasible in zebrafish (Chen et al., 2004a; Kenwrick et al., 2004; Stemple, 2004; Yamada et al., 2003). However, there are limitations to their effectiveness in uncovering gene function. Many genes do not show a loss of function phenotype, either because their phenotypes are too subtle to pick up in a large scale screen or because redundancy compensates for their loss of function (Gu et al., 2003). For example, in a MO screen in Xenopus 15–23% of genes had a detectable phenotype (Kenwrick et al., 2004), which is similar to the percentage found in a large-scale RNAi screen in C. elegans (10–14%; Fraser et al., 2000; Kamath et al., 2003) and in a MO screen in Ciona (20%; Yamada et al., 2003). Gain of function screens provide a complementary approach to loss of function screens in identifying gene function. In Drosophila, gain of function screens have been carried out by crossing lines that contain randomly inserted P elements with lines that carry the overexpression driver Gal4, under a tissue specific promoter (Rørth et al., 1998). However, P-element insertion events are not actually random, thus many areas of the genome might never be targeted by this approach. Moreover, genes are overexpressed, whether they are expressed during development or not, so the outcome may not be physiologically relevant. In vertebrates gain of function screens have been mainly carried out in Xenopus, an organism with a long history as a model system for the study of gene function in development. In Xenopus, gain of function experiments are based on injecting RNA made from cDNA libraries of a developmental stage or tissue of interest. Therefore, genes that are being tested are transcribed during development and all parts of the genome that are transcribed in the stage or tissue of interest are represented, albeit with different frequency. As in loss of function screens, in gain of function screens many genes do not show a phenotype in the assay employed. Therefore, in order to be able to test a large number of cDNA clones in a short period of time the cDNAs are screened in pools. In vitro transcribed RNA from these pools are injected into embryos and once pools that cause phenotypes are found, are sib-selected by sub-dividing into smaller pools, until a single clone that causes the phenotype is identified (Lustig et al., 1997). In this way, several developmentally important molecules have been identified, such as noggin (Smith and Harland, 1992), Wnt-8 (Smith and Harland, 1991), b-catenin (Domingos et al., 2001), gremlin (Hsu et al., 1998), Xnr3 (Smith et al., 1995); (Glinka et al., 1996), Siamois (Lemaire et al., 1995), Xombi (Lustig et al., 1996), Twin (Laurent et al., 1997), Dickkopf-1 (Glinka et al., 1998), geminin (Kroll et al., 1998), Sizzled (Salic et al., 1997), Wise (Itasaki et al., 2003), Xath2 (Taelman et al., 2001), Sox17a/b (Zorn et al., 1999) and Mix.1 (Mead et al., 1996). Although this method has clearly led to the identification of important players in development, there are some significant limitations. First, there is the issue of the sensitivity of phenotype detection. When these gain of function screens were first developed, pool sizes of

thousands of clones per pool were used (e.g. Smith and Harland, 1991; Lemaire et al., 1995; Mead et al., 1996). Since only a maximum of 5–10 ng of RNA can be injected into a X. laevis embryo without causing unspecific toxic effects, screening very large pools usually leads to the identification of only those genes that cause a phenotype when expressed in very small quantities. In order to make the screens more sensitive, pool sizes have been gradually reduced over time (e.g. Smith et al., 1995; Lustig et al., 1996; Salic et al., 1997; Hsu et al., 1998; Kroll et al., 1998, see Lustig et al., 1997). The smallest pool size so far published is 96 clones per pool (Grammer et al., 2000). However, the smaller the pool size, the more pools have to be screened to achieve the same gene coverage, which makes it a lot more time-consuming. A second drawback of this method is the redundancy of the utilised cDNA libraries, thus, abundant clones are screened over and over again. Many of these abundant clones represent housekeeping genes, such as eF1a, which may have no function in specific developmental assays or contaminants such as ribosomal RNA. Moreover, there is a risk that potent genes can swamp any other activity in numerous pools. Finally, a large number of clones contained in an ‘unprocessed’ cDNA library are truncated clones. In oligo dT primed libraries these usually lack the initiation of translation at the 5 0 end. Although some truncated clones are active, as is the case for a truncated version of b-catenin (Funayama et al., 1995; Domingos et al., 2001) many are likely to be inactive and are screened unnecessarily. Here we present the results of a modified, small-pool, gain of function screen in Xenopus that was designed to overcome these limitations. To increase the efficiency and sensitivity of the small-pool screen we made use for the first time of a unique full-length (FL) cDNA clone set, derived from gastrula and neurula stage Xenopus tropicalis cDNA libraries. Unlike normal cDNA libraries this unique fulllength cDNA library contains only clones that are likely to be full length and each is represented by only one copy. Since less clones had to be screened we were able to reduce the pool size to eight clones per pool. We compare a gain of function screen carried out with this modified library (8 clones pool size) with a redundant small pool screen (96 clone pool size) of a redundant library, similar to those published before (Grammer et al., 2000), and we find that our modification made the screen a lot more sensitive and efficient. The modified screen we describe here was designed to discover genes involved in neurogenesis, a multistep process during which neuroepithelial cells make decisions between differentiation and continuation of a progenitor state. It is known that a genetic cascade is involved in controlling neurogenesis, which consists of neural inducers, pre-patterning, proneural and neurogenic genes (reviewed in Bertrand et al., 2002; Munoz-Sanjuan and Brivanlou, 2002). However, many gaps remain in our understanding of how cell fate decisions are regulated during neurogenesis

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

291

Fig. 1. Comparison of the phenotype penetrance between the redundant set and the unique full-length set screen. One hundred and six pools of a redundant neurula stage cDNA library (pool size 96 clonesZ10.176 cDNAs) and 120 pools of the unique full-length (FL) gastrula/neurula cDNA library (pool size eight clonesZ960 unique and full-length cDNAs) were screened by overexpressing z5 ng of in vitro transcribed RNA in one cell of two-cell stage embryos. Embryos were analysed at neurula and tadpole stages for changes in morphology and in the neural differentiation marker N-tubulin. Only phenotypes with a penetrance (percentage of affected embryos in a batch) of 40% or more were scored as ‘real’ but analysis for lower penetrance is also shown. In the redundant set screen z10% (nZ11) of the screened pools showed a phenotype (R40% penetrance) whereas this was raised to 50% (nZ60) in the FL set screen (green star). At a higher penetrance (R80%) there is also a big difference in phenotype recovery between the redundant and FL set screen (3 vs. 28%). Thus the FL set screen was five-fold more sensitive in detecting gain-of-function phenotypes compared to an unmodified cDNA library.

and this functional screen was employed to increase our knowledge of genes that are involved in this process. We present the overexpression phenotypes of the single clones that were identified in this screen together with their expression patterns at some key developmental stages. Many of the single clones identified are novel or show a novel function when overexpressed. Moreover a large number of them are specifically expressed in neural tissue. These results demonstrate the effectiveness of the unique full length set as a resource for uncovering players in a developmental pathway.

2. Results 2.1. Identification of genes by overexpressing RNA from a redundant neurula stage cDNA library We screened 10,176 clones of a neurula stage X. tropicalis cDNA library by injecting 4–5 ng of pools of 96 RNAs in one half of two-cell stage X. laevis embryos. Ninety six clones per pool is a size substantially lower than used traditionally (Smith and Harland, 1991) and is the same pool size as in a recently published screen (Grammer et al., 2000). Screen analysis for phenotypes was carried out at neurula (open neural plate) and tadpole stages of development by the expression of the neuronal marker N-tubulin and morphology. N-tubulin marks the final stage of primary neurogenesis, thus any effect on neurogenesis should result in an effect on N-tubulin expression. In order to be able to distinguish phenotypes from background levels of malformations, phenotypes were

only scored if they had a penetrance of a least 40% (i.e. percentage of affected embryos). We found that only 10% (nZ11) of the pools screened (nZ106) gave rise to a phenotype that was above the 40% penetrance threshold, based on N-tubulin and morphology (Fig. 1). This is similar to what has previously been reported (Grammer et al., 2000, 12%). We reasoned that this is likely to be due to the low amount of each RNA present in the pool (a maximum of approx. 50 pg). To test this hypothesis we injected RNA for a known gene involved in neurogenesis, Xiro3, at different concentrations. While at 1 ng the penetrance of the phenotype (suppression of neurogenesis) was 90%, as previously described (Bellefroid et al., 1998), at 50 pg (the equivalent amount in a pool size of 96) the penetrance was only 18% and 12% when Xiro3 was mixed in a 96-clone pool (see Table 1). Therefore, Xiro3 would have been missed in this screen. By contrast, Gremlin, for example, which was isolated in a 1000 clone pool screen, gives penetrance of 93% in concentrations as low as 50 pg (Hsu et al., 1998). Other genes cloned in previous screens are active at even lower amounts, in the range of 1–10 pg (Smith and Harland, 1991; Laurent et al., 1997). It is not clear what causes such variability in the dose required for each gene, although the combinatorial action of genes is certainly a factor, as has been shown for X-Wnt11 and Xnr-3, where a combination of the two genes dramatically reduces the amount required for each single gene (Glinka et al., 1996). These results indicate that carrying out a gain of function screen in this format is predominantly leading to the isolation of potent genes, which are more likely to have been already described.

292

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

Table 1 The effect of quantity of injected RNA on phenotype penetrance

approximately five-fold higher than previously achieved with the 96-clone per pool redundant set screen, described above (Fig. 1) or in (Grammer et al., 2000). 2.4. Unique full-length set screen summary

In order to distinguish phenotypes from background malformations only phenotypes with a penetrance (number of affected embryos in a batch) of at least 40% are scored. Xiro3 RNA, whose overexpression leads to a reduction of N-tubulin at neurula stage (Bellefroid et al., 1998), was chosen as a representative gene to test if its activity can be detected in a 96 clones/ pool screen. Thus a range of doses of Xiro3 has been injected into one cell of two-cell stage Xenopus laevis embryos and the phenotype penetrance was scored. Five nanograms of an RNA pool containing Xiro3 RNA were also injected as a pool control. The approximate concentration of any individual RNA in a 96-clone pool is approximately 50 pg (due to a maximum of 5 ng that can be injected into embryos). At 50 pg (highlighted in red) or in the control pool, the phenotype penetrance of Xiro3 was below the scoring threshold.

2.2. A unique full-length cDNA library allows the reduction of the pool size One way to increase the amount of injected RNA per clone is to inject the same total amount of RNA but decrease the pool size. To achieve the same overall clone coverage the number of pools to be screened would increase dramatically. Recently, 55,000 clones from each of the gastrula and neurula libraries have been sequenced in a large-scale X. tropicalis EST project (http://www.sanger.ac.uk/ Projects/X_tropicalis). It was possible to use this sequence information in order to create a unique and full-length cDNA clone set by removing multiple as well as truncated presumably non-functional copies of cDNAs from the collection of sequenced clones (see materials and methods). In this way, approximately 100,000 cDNA clones from the gastrula and neurula libraries were ‘condensed’ into a set of 2304 unique and full-length clones. This effectively reduced the number of cDNAs to be screened, which in turn made it feasible to reduce the pool size and concomitantly increase the amount of RNA of each clone in a pool. 2.3. The unique full-length set makes the screen more sensitive We screened 960 clones of the unique full-length clone set in pools of eight clones (120 pools). Approximately 500 pg of RNA were injected per individual clone. As for the previous redundant clone set screen we analysed for phenotypes at neurula and tadpole stages using morphological changes and changes in the in situ pattern of N-tubulin. Interestingly, we now found that 50% (nZ120) of the pools gave phenotypes of 40% penetrance or higher. This is

The phenotypes caused by the overexpression of the pools fell into nine major categories: gastrulation defects, reduction of N-tubulin, ectopic tissue outgrowth, tissue loss, embryonic lethality, secondary axes/displacement of neural markers, ectopic N-tubulin, ectodermal blisters and anteriorisation (see Fig. 2). Most pools gave rise to complex phenotypes that fall into more than one of these categories, therefore the percentages sum up to more than 100%. This complexity might be due to the combined effects of different RNAs within a single pool. Since we wanted to identify novel players in neurogenesis, we were particularly interested in pools that alter the pattern of N-tubulin expression. Since neural tissue is derived from the ectoderm we also selected some pools with interesting ectodermal defects for further analysis. We selected 19 out of the 60 pools of R40% penetrance for phenotypes, which indicate a potential involvement in neural development. Seventeen out of the 19 pools (89.5%) showed a reproducible phenotype upon repeat injections. In order to identify single clones we took advantage of the sequence information that is available for each clone in a combination of candidate gene approach and injection of single clones. In the cases where the activity within a pool matched previously published observations for a gene present in this pool, we only injected the candidate gene. If the pool phenotype was reproduced, the activity was assigned to this clone. We picked candidate genes for five of the 17 reproducible pools and 4/5 matched the phenotype of the pool. For the remaining 13 pools (12 plus the nonmatching pool from the candidate approach), single clones were injected individually. With such a small pool size, the previously used step of sib-selection in gradually smaller pools was not necessary. 2.5. Description of the activities that have been identified Using both approaches we were able to identify the active clones for 15 of the 17 pools and from these, we identified 18 different genes with activity. Seven (39%) of the identified genes have been previously described, which indicates that our screen is able to identify known players during neurogenesis. Two (11%) have been cloned in Xenopus before but showed a novel function, 6 (33%) were novel in Xenopus (i.e. no NCBI database entry—other than predicted genes—exists for Xenopus), and three (17%) were completely novel (no entry in NCBI database, only predicted genes). The identified genes fell into the following categories: reduction or ectopic N-tubulin at neurula stage, secondary axis, anteriorisation, ectopic tissue outgrowth,

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

293

Fig. 2. Phenotypic categories found when screening the unique FL set pools. Sixty pools of the unique full-length gastrula/neurula stage cDNA library gave rise to an overexpression phenotype (at least 40% penetrance). Nine major categories (x-axis) of phenotypes, for which representative images are shown below each bar, were observed and scored. Phenotypes were classified as secondary axes (rather than ectopic tissue outgrowth) if the ectopic tissue shows axial organisation, like secondary heads (at tadpole stage) or complete secondary neural tubes (at neurula stage). Ectopic tissue outgrowth comprises tumour-like growths, protrusions (which lack clear axial structures) and other abnormal thickenings observed. Blisters are oedema-like appearances, which do not contain tissue. Phenotypes such as loss/shedding of cells and or limited cell death were grouped into tissue loss. In contrast, embryos that showed global death are classified in death. Since most of the phenotypes observed were complex, many of the pools were included in more than one category (thus the sum of the percentages is greater than 100%). Of particular interest were phenotypes that affected the neural differentiation marker N-tubulin (stars), since they might contain an activity involved in the neurogenesis pathway. (Purple staining in images: N-tubulin, light blue staining: lineage tracer LacZ).

epidermal defects, gastrulation defects and cell lethal (Table 2 and Fig. 3). 2.6. Reduction/inhibition of N-tubulin 2.6.1. Neurula stage The most common phenotype observed, was a reduction or inhibition of N-tubulin (Table 2, Fig. 3a,b). This was shown by 12 genes, three of which are novel, in that a BLAST search results only in predicted ORFs. The remaining nine are previously published genes although, with the exception of Zic3 (Nakata et al., 1997), their effect on neurogenesis has not been previously described. The reduction of N-tubulin seen here could be due to interference with an early step in neural development, namely neural induction, or a later failure of the neuroepithelial cells to differentiate. To distinguish these possibilities we examined the expression of the early neural marker of proliferating cells, Sox3 (Penzel et al., 1997; Hardcastle and Papalopulu, 2000) (data summarised in Table 2, column 3). Only two of the genes, Sox2 and Myocyte enhancer factor related 2D (SL1) reduced Sox3. For SL1 this is likely to indicate a general interference with neural development. SL1 is expressed in the somitic mesoderm (Fig. 4b; Chambers et al., 1992) and is thought

to lie downstream of myoD and Myf5 (Chambers et al., 1994). The suppression of Sox3 by Sox2 was somewhat surprising since Sox2 is a later neural marker than Sox3 (Rex et al., 1997). We suggest that suppression of Sox3 by Sox2 may reflect a negative transcriptional control of later acting genes (Sox2) on the earlier ones (Sox3) in the neurogenic cascade. Sox2 may also suppress later acting genes, reflected in the suppression of N-tubulin. It may be that such gene downregulation in both directions of the neurogenic pathway is an important step in the progression of cells through different stages of differentiation. A previous study has shown that a dominant negative form of Sox2 (dnSox2) inhibits early neural development in general (Kishi et al., 2000). Reduction of N-tubulin could also indicate an increase in proliferation of the ectoderm at the expense of differentiation. Indeed, five genes that caused a reduction of N-tubulin, (strabismus, Zic3, Frizzled 10B, cullin 1, XER81), caused a concomitant increase of Sox3, which marks proliferating neuroepithelial cells (Table 2, Fig. 3a,b). Four of these genes (designated with asterisks in Table 2) also showed a lateral displacement of N-tubulin. The most dramatic increase in Sox3 expression was shown by XER81, an ets-type transcription factor in the FGF pathway (Chen et al., 1999) (shown in Fig. 5A). Our observation

294

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

Table 2 Phenotype categories of the single clones identified in the screen Gene I. Reduction of N-tubulin at neurula stage 1.2B (TGas002c02) X-dll3 3.1G (TGas026h11) Transcription factor IIF, RAP74 subunit 3.2B (TGas021d05) Novel 3.4D (TGas023a22) Novel 3.5H* (TGas027l08) Strabismus 3.9B* (TGas021i09) Zic3 3.12E* (TGas025k08) Frizzled 10B 4.6D (TGas032i14) Novel 4.6H* (TGas038c21) Cullin 1 19.11A (TNeu063l17) Sox2 22.6D (TNeu107l17) Myocyte enhancer factor related 2D (SL1) 22.9F (TNeu110d19) XER 81

Sox 3 at neurula stage

N-tubulin at tadpole stage

No change No change

No change No change

No change No change C C C No change C K K(Subtle)

No change K C C C No change C C No change

C

C

II. Ectopic N-tubulin at neurula stage 3.12E (TGas025k08)

Frizzled 10B

C

C

III. Secondary axis 3.12E (TGas025k08) 4.6H (TGas038c21)

Frizzled 10B Cullin 1

C C

C C

IV. Anteriorisation 2.8B (TGas012p20)

Casein kinase 1 delta

C

No change

Serine threonine kinase 17A Zic3 Novel Cullin 1 Sox2 Myocyte enhancer factor related 2D (SL1) XER 81

No change C No change C K K(Subtle)

No change C No change C C No change

C

C

Rab 32 Serine/threonine kinase 17A Novel

No change No change No change

No change No change No change

X-dll3 Serine/threonine kinase 17A Novel Novel Strabismus Frizzled 10B Sox2 Myocyte enhancer factor related 2D (SL1)

No change No change No change No change C C K K(Subtle)

No change No change No change K C C C No change

PKA, subunit beta SFRS 5 SFRS 7

K K K

K K K

V. Ectopic tissue outgrowth 1.4A (TGas001g08) 3.9B (TGas021i09) 4.6D (TGas032i14) 4.6H (TGas038c21) 19.11A (TNeu063l17) 22.6D (TNeu107l17) 22.9F (TNeu110d19) VI. Epidermal defects 1.2H (TGas009a06) 1.4A (TGas001g08) 4.6D (TGas032i14) VII. Gastrulation defect 1.2B (TGas002c02) 1.4A (TGas001g08) 3.2B (TGas021d05) 3.4D (TGas023a22) 3.5H (TGas027l08) 3.12E (TGas025k08) 19.11A (TNeu063l17) 22.6D (TNeu107l17) VIII. Embryonic lethal/tissue loss 2.11E (TGas016h07) 3.1C (TGas021n15) 19.3F (TNeu070m22)

The phenotypes of the single clones identified in the unique full-length set screen were grouped into eight categories (I–VIII.) Where embryo phenotypes fell into more than one category the clone was listed repeatedly. Neurula stage embryos were additionally analysed for changes of the proliferating neural tissue marker Sox3 and tadpoles were analysed for N-tubulin expression (alongside morphology analysis), all of which is summarised in column 3 and 4, respectively (C: marker expanded, K: marker reduced, ‘no change’: marker not affected). Asterisks in column 1 indicate genes that in addition to a reduction of N-tubulin also showed a displacement of the stripes (see Fig. 4b). The clones were coded as follows; bold: novel gene with no BLAST hit other than predicted genes, bold italic: novel gene in Xenopus; italic: cloned in Xenopus but novel function; normal: known gene in Xenopus with previous functional analysis carried out. The clone names (TNeu/Gas.) were given by the Sanger Xenopus tropicalis EST project and refer to the position of the clones in the original arrayed library. The gene name given to the clones is based on the highest BLAST hit that roughly aligns to the start methionine of the clone. Sequences, clusters and BLAST alignments are displayed on the website: http://informatics.gurdon.cam.ac.uk/online/xt-fl-db.htm. The other clone name (1.2B, etc.) refers to their position on the full-length set plate.

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

295

Fig. 3. Overexpression phenotypes of the single clones identified in the screen. (aCb) Approximately 400 pg in vitro transcribed RNA together with 200 pg beta-galactosidase RNA as a lineage tracer (light blue staining) were injected animally into one cell of two cell stage Xenopus laevis embryo. Embryos were analysed at neurula and tadpole stages for changes in the neural differentiation marker N-tubulin (purple) and morphology. Unless indicated otherwise, neurula stage embryos are viewed dorsally with anterior to the top and tadpoles are viewed laterally with anterior to the left. A transverse section is shown for clone 1.2B at tadpole stage.

296

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

Fig. 3 (continued)

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

complements previous findings where dorsal XER81 overexpression disrupts eye development (Chen et al., 1999). Moreover, FGF has been implicated in neural induction in Xenopus and chicken (e.g. Launay et al., 1996; Pera et al.,

297

2003; Streit et al., 2000; Wilson et al., 2000) as well as the maintenance of neural progenitors cells (in chicken) (Diez del Corral et al., 2002; Mathis et al., 2001). In Xenopus, a role in enhancing ectopic neurogenesis has also been

Fig. 4. (a,b) Expression patterns of the single clones identified in the unique full-length set screen. Xenopus tropicalis gastrula, neurula and tailbud stage embryos were in situ hybridised with antisense probes to the single clones identified in the unique full-length set screen. Gastrula stages embryos are shown vegetally with dorsal to the top (unless stated otherwise). At neurula stages an anterior and dorsal view is presented. Tailbud embryos are oriented with anterior to the left and dorsal to the top unless indicated otherwise. In come cases the wholemount images are supplemented by neurula and tailbud stage dissected embryos. If no specific staining was detected no image is shown.

298

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

Fig. 4 (continued)

suggested (Hardcastle et al., 2000). The phenotype of XER81 described here, is consistent with XER81 mediatingat least in part-the proposed role of FGF in enhancing the formation of undifferentiating, proliferating neuroepithelium, but further experiments are needed to verify this hypothesis.

2.6.2. Tadpole stage We also analysed the expression of N-tubulin at tadpole stages (right column Table 2, Fig. 3a,b). We found that many of the genes that suppressed N-tubulin early on, caused ectopic N-tubulin at the tadpole stage. Interestingly, in almost all of these cases (5/6) (Strabismus, Zic3, Frizzled

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

299

Fig. 5. Additional detail on the overexpression of 4 selected genes and Rab32 expression. (a) XER 81 RNA injections expand X-Sox3 (purple) on the injected side. (bCc) Frizzled10B RNA injections enhance the medial (motorneuron) stripe of N-tubulin expression (white arrowhead, in wholemount (B) and transverse (C) view), in addition to the ectopic expression of N-tubulin observed in the floorplate (shown in Fig. 4b). (dashed line-midline, l-lateral, i-intermediate, m-medial). (dCe) 1 mm sections of Rab32 (1 ng) injected (E) and control Xenopus laevis (D) embryos. (D) In wild type embryos the melanosomes (darkly pigmented vesicles, arrow) are localised predominantly under the apical surface, but some can be found deeper within the cells (arrowhead). In contrast, when Rab32 was overexpressed the melanosomes were clustered tightly together and localised under the apical surface (arrows). (f) Transverse section (30 mm) of a bleached Xenopus tropicalis tailbud embryo in situ hybridised with an antisense probe to Rab32. Rab 32 is expressed in the retinal pigment epithelium (purple).

10B, Cullin 1, XER81), the expression of Sox3 was expanded at neurula stage. Thus, there appears to be a correlation between suppression of early neurogenesis and ectopic proliferation of neural tissue at neurula stage on the one hand, and ectopic neuronal differentiation at tadpole stage on the other. The genes that show this correlation may function in expanding the population of neural progenitor cells but their effect may be transient. We suggest that once the RNA concentration diminishes at the tadpole stage, the ectopic neural tissue is no longer repressed from differentiating, hence ectopic N-tubulin appears at the tadpole stage. Surprisingly, we observed reduced N-tubulin at tadpole stage only in one case, that of the novel gene TGas023a22 (3.2B). Since this gene had no effect on Sox3, we expect it to function quite late in the neurogenesis cascade. Finally, some of the clones that showed a reduction of N-tubulin at neurula stage did not show any effect on N-tubulin later (Xdll3, Transcription factor IIF, TGas023a22 3.2B and SL1) (Table 2). This could be due to a transient effect of the RNA or less interestingly, a developmental delay on the injected side. It could also be that a subtle reduction might be difficult to detect at the tadpole stage.

2.7. Ectopic N-tubulin at neurula stage Only one of the genes—Frizzled 10B—showed ectopic N-tubulin at neurula stage. This gene is of particular interest since it preferentially enhanced neurogenesis along the medial, motorneuron, stripe (Figs. 3b, 5B,C). The lateral (sensory neuron) stripe was not enhanced and in some cases it was suppressed, hence it is also classified under reduction of N-tubulin. Occasionally, Frizzled 10B overexpression resulted in N-tubulin expression along the floor plate (arrowhead in Fig. 3b). In these cases, the ‘floor plate’ was expanded, causing a lateral shift of the N-tubulin stripes (Table 2, asterisk). The expression of Frizzled 10B is exclusively neural (Fig. 4b), consistent with previous reports (Moriwaki et al., 2000; Wheeler and Hoppler, 1999). The function of Xenopus Frizzled 10B has not been previously studied but it is likely to act as a Wnt receptor, similar to other members of the frizzled family. This clone would be interesting to investigate, as this is the first case of a gene that enhances specifically the primary motorneuron population when overexpressed in Xenopus embryos. The frequency of ectopic N-tubulin in this screen was lower than we might have expected (see Fig. 2). However,

300

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

we note that many published genes that result in ectopic N-tubulin when overexpressed (such as X-ngnr-1 and neuroD (Ma et al., 1996; Lee et al., 1995) were not found in the EST database of the cDNA libraries that we have used. These genes are expressed at the neural plate stage but we speculate that their abundance is still relatively low. As many such genes continue to be expressed later in development, we suspect that later stage libraries might have been more fruitful in uncovering neurogenesisinducing genes. Indeed, later stage libraries have since been added to the EST project (http://www.sanger.ac.uk/ Projects/X_tropicalis/). 2.8. Secondary axis and anteriorisation Two genes, Frizzled 10B and cullin 1 gave rise to secondary axes while one gene, casein kinase (CK) 1 delta gave rise to anteriorised embryos at neural plate and tadpole stages, with an associated increase in Sox3 expression (Fig. 3a,b, Table 2). The phenotype of CK 1 delta, a kinase in the Wnt pathway, is slightly different than previously described (anteriorisation versus secondary axis; (Sakanaka et al., 1999; Peters et al., 1999; McKay et al., 2001) however, this is likely to be due to differences in the site of injection (animally, in our case). Cullin 1 is a ubiquitin ligase (reviewed in Tyers and Jorgensen, 2000) and is currently under investigation in our lab (JV and NP, in preparation). 2.9. Ectopic tissue outgrowth Ectopic tissue outgrowth was observed at the tadpole stage and is quite common as it was observed for seven clones (Table 2, Fig. 3a,b). In all cases, clones that showed ectopic tissue outgrowth also fell into some other category. Six out of the seven genes that showed ectopic tissue outgrowth at the tadpole stage, 4.6D (TGas032i14), Sox2, SL1, XER81, Zic3 and cullin 1 also showed a reduction in N-tubulin at neurula stage, suggesting a role in increasing proliferation at the expense of differentiation. In the case of XER81, Zic3 and cullin 1, such role was further supported by an increase in Sox3 at neurula stages. In the case of cullin 1, the ectopic tissue outgrowths appeared distinct from the secondary axis and are likely to represent separate effects. It is likely that not all tissue outgrowths represent effects on proliferation. For example, in some embryos the effect of one clone, TGas001g08, which shows similarity to the human serine/threonine kinase 17A (Sanjo et al., 1998), appeared more like an ectodermal defect than a tissue outgrowth. Consistent with a change in ectodermal architecture rather than proliferation, this clone did not cause changes in Sox3 and N-tubulin expression. 2.10. Epidermal defects The distinction between epidermal outgrowths and other epidermal defects was not always clear and two clones were

classified in both categories (see also above). Of particular interest in the epidermal defect category is the third clone, Rab32, encoding a small molecular weight Rab GTPase of the ras superfamily (Table 2, Fig. 3a,b). Overexpression of this clone caused a concentration of pigmentation into a spot like structure in each ectodermal cell (Fig. 3a). In sections, it was clear that Rab32 overexpression caused a tight clustering of the pigment organelles (melanosomes), which are normally evenly spread under the apical surface (Fig. 5D,E). In pigmented Xenopus cells, melanosomes are transported within the cell along microtubules and actin filaments (Gross et al., 2002). In apico-basally polarised early blastomeres (Chalmers et al., 2003), they have a clear enrichment on the apical side (Fig. 5D). Rab32 would be interesting to investigate further as it may be involved in directional melanosome transport, localisation or dispersion. The expression of this gene was not detected in early stages of development, perhaps indicating a ubiquitous low level of expression. However, Rab32 was strikingly upregulated in the pigmented epithelium of the retina at the tadpole stage, consistent with a role in melanocyte function (Fig. 5F). In cell lines, murine Rab32 co-localises with melanosomal proteins (Cohen-Solal et al., 2003) while the human protein is thought to act as a tether for cAMPdependent protein kinase (PKA; Alto et al., 2002). 2.11. Gastrulation defect Approximately 60% of all the pools showed some form of gastrulation defect (Fig. 2). When deciding which pools to follow up we excluded all the samples that showed only a gastrulation defect. Therefore, clones that cause gastrulation defect in the sib-selected list, also had some neural defect (Table 2). It is interesting to note here, that the morphogenetic defects of Strabismus and Frizzled 10B at the neural plate stage are remarkably similar (Fig. 3b). These two genes also have common elements in their effects on Sox3 and N-tubulin. However, Frizzled 10B has additional effects in inducing secondary axis formation and increasing specifically the number of primary motorneurons (Fig. 5B,C), as discussed above, which were not observed with Strabismus. This is what one would expect if Frizzled 10B, acting as a Wnt receptor, activated both the canonical and the non-canonical, PCP, Wnt pathway. Strabismus on the other hand, is known to act only on the PCP pathway (Park and Moon, 2002), which we believe explains why its overexpression phenotype is a subset of the Frizzled10B phenotype. The high incidence of gastrulation defects could mean that gastrulation is easily disrupted and/or that the concentration of some RNAs was too high. Alternatively, it is possible that our clone set was enriched for genes involved in gastrulation, as it derived from both gastrula and neurula stage libraries. As we were primarily interested in neurogenesis, we did not further study gastrulation defects and we made no attempt to distinguish between effects on

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

the morphogenetic movements of the mesoderm or of the ectoderm. 2.12. Embryonic lethal/tissue loss In this category, there are three clones, which were originally selected because they caused suppression of N-tubulin (Table 2, Fig. 3a,b). Upon further investigation, we concluded that this phenotype is due to cell loss, possibly due to excessive cell death. At the moment it is not clear whether this is a specific apoptotic effect and further investigation is needed to clarify this point. However, we are presenting these clones because they have specific expression patterns (Fig. 4a,b, Table 3), they are novel in Xenopus and their function in development is not known. The beta-subunit of PKA is specifically expressed in the notochord while two serine/arginine regulators of alternative RNA splicing, SFRS 5 and 7, are specifically expressed in neural tissue (Fig. 4a,b, Table 3).

301

2.13. Expression pattern of single clones In addition to the overexpression phenotype we present the expression pattern for all of the identified clones (Fig. 4a,b, Table 3) in Xenopus tropicalis embryos. It was striking that 16/18 clones were expressed in neural tissue, which suggests that our screening strategy was successful in specifically identifying potential players in neural development.

3. Discussion 3.1. Functional screens in Xenopus We have created and used a unique full-length clone set from gastrula and neurula stage cDNA libraries in a smallpool gain of function screen for genes involved in neurogenesis. Eliminating redundancy and truncated clones from the original cDNA clone set allowed us to reduce

Table 3 Expression analysis of the single clones identified in the screen Clone ID

Gene

Gastrula

Neurula

Tailbud

Reference

TGas002c02 1.2B

X-dll3/Dlx5

Ubiquitous

Rab32

None detected

TGas001g08 1.4A

Ubiquitous

None detected

TGas012p20 2.8B

Serine/Threonine Kinase 17A Casein Kinase 1 delta

Anterior neural, cement gland, otic vesicle Retina pigment epithelium Eyes, pronephros

Papalopulu and Kintner (1993)

TGas009a06 1.2H

Anterior neural, cement gland, neural folds None detected

Ubiquitous

TGas016h07 2.11E TGas021n15 3.1C

PKA, subunit SFRS 5

Ubiquitous Ubiquitous

TGas026h11 3.1G

None detected

TGas021d05 3.2B

Transcription factor IIF, RAP 74 subunit Novel

Dorsal

Anterior neural fold, low level spinal cord Neural enriched Anterior neural stripe, spinal cord Anterior neural enriched, spinal cord Pan-neural, low level

TGas023a22 3.4D TGas027l08 3.5H

Novel Strabismus

None detected Dorsal

None detectable Anterior neural stripes, spinal cord

TGas021i09 3.9B TGas025k08 3.12E

Zic3 Frizzled 10B

Dorsal None detected

Lateral neural folds Anterior neural stripes and lateral neural folds

Neural tube Neural tube

TGas032i14 4.6D

Novel

Dorsal

Eye

TGas038c21 4.6H

Cullin I

Ubiquitous

TNeu070m22 19.3F

SFRS 7

Ubiquitous

TNeu063l17 19.11A TNeu107l17 22.6D TNeu110d19 22.9F

Sox2 SL-1 XER81

Dorsal None detected Blastopore

Neural enriched, low level Anterior neural enriched? Anterior neural, spinal cord Pan neural,placodes Somites Complex anterior neural, spinal cord, blastopore remnant

Spinal cord Eye, notochord Eye, branchial arches, brain Branchial arches and eye Eye, hindbrain, branchial arches Cement gland Neural tube

Eye, neural tube, branchial arches Eye, neural tube, branchial arches Neural tube, eye Tail somites, heart Mid-hindbrain stripe, branchial arches, tailbud

Darken et al. (2002) and Park and Moon (2002) Nakata et al. (1997) Moriwaki et al. (2000) and Wheeler and Hoppler (1999)

Mizuseki et al. (1998) Chambers et al. (1992) Chen et al. (1999) and Munchberg and Steinbeisser (1999)

Table describing the expression pattern of the identified clones as determined by whole-mount in situ hybridisation. Only two genes (SL-1 and 3.4D) are not expressed in neural ectoderm. References are added where the expression pattern has been analysed previously in Xenopus.

302

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

the pool size to eight clones per pool without unreasonably increasing the amount of time needed to perform the screen. In total, 2304 FL clones were derived starting from 100,000 cDNA clones and of those 960 were screened in this work. Overall, 18 genes have been identified out of this 960 clone set versus 37 out of a recent 55,000 redundant clone set (Grammer et al., 2000) a 30 fold improvement in the return of the screen. This is in fact an underestimate, if one takes into account that we only followed up pools that showed an effect on neural and/or ectodermal development. Of the 18 genes, three were entirely novel, in that they had no similarity to sequences in the sequence (NCBI) database The sensitivity and efficiency of this screen was also higher, with 50% of the screened pools showing a phenotype of 40% penetrance or higher, a five-fold improvement from the best return previously described from an unprocessed set of clones (Grammer et al., 2000). The cDNA clones used here were derived from a X. tropicalis EST project, but the overexpression was carried out in X. laevis because the eggs are larger and more robust. Once candidate genes were identified, the expression analysis was carried out in X. tropicalis. Functional screens have been performed across more distant species, for example between Ciona or mouse and Xenopus (Davis and Smith, 2002; Baker and Harland, 1996). We expect that X. laevis and X. tropicalis are evolutionarily close enough for the results to be fully transferable between the two organisms. Indeed, a recent study examined the expression pattern of several genes in X. laevis and X. tropicalis and no significant differences were found (Khokha et al., 2002). We suggest that these two amphibian systems can be used in parallel, allowing one to exploit the advantages of both. 3.2. Identification of genes The scale of our screen was modest and therefore did not allow us to reach global conclusions about gene networks controlling neurogenesis. At this scale, such a screen is most useful in identifying single genes that can be followed up by more detailed functional analysis. Indeed, we have identified several genes, such as Frizzled 10B, cullin 1, XER81 and Rab32, which warrant further investigation, given our research interests. In addition, even this small scale analysis allowed some interesting correlations to emerge, such as the increase of Sox3/suppression of differentiation at neurula stage and the development of ectodermal ‘growths’ at the tadpole stage (see Section 2). 3.3. Future directions One of the great strengths of a functional screen is that it is highly adaptable to address specific questions (see Chen et al., 2005). Thus, one area of future improvements would be the development of diverse and imaginative assays. The Rab32 phenotype of melanosome aggregation within each cell, shows that such assays need not be limited to

alterations in molecular markers or gross morphology but can be extended to a cellular level. As the sequence of the clones is known a priori, one could pre-select the gene set in such a way that only a certain class of molecules, for example, bHLH-motif containing genes, are screened. Finally, since beginning this work, many more ESTs have become available and the bioinformatic tools for FL prediction have been refined. As a result a larger unique FL set of 7000 clones, from three developmental stages, egg, neurula, gastrula and tadpole has been created (Gilchrist et al., 2004) and is available as a resource for the community (http://informatics.gurdon.cam.ac.uk/online/xt-fl-db.htm). These screens should provide yet another step on the way to assigning function to the large number of genes uncovered by the efforts in genome and cDNA sequencing.

4. Experimental procedures 4.1. Creation and arraying of the cDNA libraries The neurula (TNeu) and gastrula (TGas) stage Xenopus tropicalis cDNA libraries were constructed as described (D’Souza et al., 2003; Gilchrist et al., 2004). Approximately 55,000 clones of each cDNA library were randomly picked, arrayed and sequenced from the 5 0 end, as part of a large scale X. tropicalis EST project (http://www.sanger.ac.uk/ Projects/X_tropicalis/). The individual clones are referred to by their position in the original arrayed library (e.g. TGas002c02). The sequence of each clone can be obtained by using this name, either at the above website or at the Xenopus tropicalis full-length EST database (http://informatics.gurdon.cam.ac.uk/online/xt-fl-db.htm). Furthermore, in the latter website, the sequences for only the clone set used in this work can be accessed by choosing ‘Xt2:initial full length project’. The cDNA clones can be obtained from the MRC Geneservice: (http://www.hgmp. mrc.ac.uk/geneservice/reagents/products/descriptions/XtropEST.shtml). 4.2. Bioinformatics and re-arraying of the unique full-length set 95,219 5 0 EST sequences from the TNeu and the TGas library underwent the clustering process, which has been described in detail elsewhere (Gilchrist et al., 2004). Briefly, ESTs were clustered together that were 99% similar over a length of 100 nucleotides and aligned accordingly. We found that 74,683 ESTs fell into 11,078 clusters of two or more clones, whereas 20,536 clones remained as singletons (ESTs that fail to align with any other ESTs in the collection). In total, this led to 31,614 unique transcripts— the unique clone set. For the purposes of the gain of function screen we wished to extract the clones that are likely to contain the translation

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

initiation site, referred to here as the full-length (FL) clones. An assumption was made that clones which contain the 5 0 start of translation also contain the 3 0 end, as the libraries were constructed with oligo dT priming. In order to identify full-length clones the following criteria were applied to all clusters of the unique clone set. First, a start codon preceded by an in frame stop codon must be present in at least 2 and at least 50% of the EST clones at the same position within a cluster to avoid misidentification of the start codon due to sequencing error. The 5 0 stop codon indicates that translation cannot start before the designated start codon Second, the ORF should be at least 210 nucleotides long (See Fig. 6). Unlike a more recently generated and more extensive fulllength clone set (Gilchrist et al., 2004) the criteria that we used to define the clone set used here did not include similarities in the database revealed by BLAST searches (except for the singletons, as described below). This was introduced to avoid biasing the selection towards genes that have been already described in the literature. This approach identified 1988 clusters, which are very likely to contain a full-length clone. To avoid potential negative translational control from sequences in the 5 0 UTR, the clone with the shortest 5 0 UTR was automatically chosen as the representative for its gene. These representatives were then re-arrayed onto 384 well plates. As we were using 384-well plates for the re-arraying, six plates were required, leaving 316 empty wells. These were allocated to singletons, which were selected using slightly different criteria than the clusters. Singletons were prioritised according to the length of their putative ORFs and those with long ORFs and a start codon preceded by an in frame stop were selected. In single read ESTs, the position of the putative initiator ATG cannot be verified by comparison to other ESTs. Therefore, BLAST information was used to confirm that the proposed initiator ATG corresponded to the start

303

methionine of a matching protein in the NCBI protein database. In this way, 316 singletons were included in the full-length unique set of clones, eventually consisting of six 384-well plates or 2304 clones. The full-length cluster representative and the selected singletons were rearrayed onto 384-well plates at the MRC-Geneservice, UK. From these, a 96-well plate working copy was created for the gain-of-function screen. Since commencing this work, many more ESTs, including 3 0 sequences, have become available and the informatics programs have incorporated more sophisticated features and methods (Gilchrist et al., 2004). When comparing this initial full-length set against later data (Gilchrist et al., 2004) we found that approximately 75% of the clones from FL clusters described here were confirmed as ‘almost certainly’ full length (not shown). 4.3. Creation of mRNA pools for the functional screen Clones were grown separately in 96-well plates and the cultures from each plate were pooled together, plasmid DNA was extracted and RNA was in vitro transcribed. These pools were used for the 96 clones/pool screen. To create the pools used for the unique full-length set the cultures were grown as above, but pools were created from the 12 columns (containing eight clones each) of each 96-well plate. The individual clones are referred to by their position in the original arrayed library (http://www.sanger.ac.uk/ Projects/X_tropicalis/). 4.4. Embryo culture and RNA injections Eggs were collected from hormone-induced adult Xenopus laevis and in vitro fertilised with standard methods.

Fig. 6. Creation of a unique full-length set. (a) The determination of the full-length representative in an EST cluster. 5 0 sequenced ESTs from the X. tropicalis neurula and gastrula stage cDNA libraries that are 99% similar over a length of 100 nucleotides were aligned and clustered. Open reading frames of at least 210 base pair were identified. In order to ascertain that clusters contain full length EST clones in at least 2 and at least 50% of the EST clones at the same position a start codon (green) must be preceded by an in frame stop codon (red). To avoid 5 0 UTR dependent negative translational control an EST with the shortest 5 0 UTR (arrow) was chosen as a full-length representative of the cluster/gene and re-arrayed into the unique and full-length set. (b) Criteria for selecting singletons as full-length. EST clones that did not align with any other clones were classified as singletons. Only singletons that contained a long ORF with a start codon preceded by a stop codon were considered for the full-length set. However, since the 50% rule (see above) cannot be applied in this case, Blast information was also used to determine if a singleton is full length. Only if the translational start of a matching protein in the database matched the start codon of the singleton, the latter was included in the full-length set.

304

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

Five ng of pool RNA were injected animally into one side of two-cell stage embryos. In the unique full-length set screen 200 pg of b-galactosidase (LacZ) RNA were co-injected as a lineage tracer. Embryos were grown to the desired stage and fixed in MEMFA for 1 h. LacZ was developed for approximately 30 min as previously described (Bourguignon et al., 1998). Embryos were stored in ethanol until further processing. Phenotypes were identified by comparing the injected area (one half of the embryo or the LacZ stained tissue) with the uninjected area and with wild type embryos. Abnormalities were scored as a phenotype if at least 40% of the embryos within a batch showed the defect. 4.5. Analysis of molecular marker expression Eggs were collected from hormone-induced adult Xenopus tropicalis and in vitro fertilised as previously described (Khokha et al., 2002). In situ hybridisation was carried out in X. tropicalis as previously described (Khokha et al., 2002) see also Harland (1991). In the 96 clones/pool screen neurula stage embryos were assayed for alterations in N-tubulin expression (Chitnis and Kintner, 1995; Oschwald et al., 1991). Embryos in the unique full length set screen were analysed for N-tubulin and Sox3 (Penzel et al., 1997). The in situ probes were digoxigenin-11-UTP labelled antisense mRNA. The chromogenic reaction was either carried out with Nitro blue tetrazolium/5-bromo-4-chloro-3indolyl-phosphate (NBT/BCIP) or with 5-bromo-6-chloro3-indolyl-phosphate (Magenta phos). 4.6. Sib-selection Some pools were selected for further analysis. Before sib-selection the injections were repeated. If the phenotype was reproducible the single clone(s) within the pools were identified. In the case of the 96 clones/pool screen sibselection was carried out as described previously (Grammer et al., 2000; Lustig et al., 1997). For the unique full-length set screen all eight clones from the selected pool were injected individually and assayed, which directly leads to the identification of one or more responsible activities from the analysed pool.

Acknowledgements We thank Andy Chalmers, Eva Asscher and David Welchman for critical comments on the manuscript. We thank our colleagues at the Wellcome Trust Sanger Institute, Jane Rogers, Jennifer Ashurst, Michael Croning, Elizabeth Huckle and Ruth Taylor for collaborating with us on the X. tropicalis EST project. JV is a student of the Wellcome Trust 4-year PhD Programme in Developmental Biology at the University of Cambridge. NP and EA are Wellcome Trust Senior Research Fellows. This work was supported by funds from the Wellcome Trust. Ethical experimentation.

All research presented here complied with Home Office regulations based on the animals (scientific procedures) act 1986.

References Alto, N.M., Soderling, J., Scott, J.D., 2002. Rab32 is an A-kinase anchoring protein and participates in mitochondrial dynamics. J. Cell. Biol. 158, 659–668. Baker, J.C., Harland, R.M., 1996. A novel mesoderm inducer, Madr2, functions in the activin signal transduction pathway. Genes Dev. 10, 1880–1889. Bellefroid, E.J., Kobbe, A., Gruss, P., Pieler, T., Gurdon, J.B., Papalopulu, N., 1998. Xiro3 encodes a Xenopus homolog of the Drosophila Iroquois genes and functions in neural specification. EMBO J. 17, 191–203. Bertrand, N., Castro, D.S., Guillemot, F., 2002. Proneural genes and the specification of neural cell types. Nat. Rev. Neurosci. 3, 517–530. Bourguignon, C., Li, J., Papalopulu, N., 1998. XBF-1, a winged helix transcription factor with dual activity, has a role in positioning neurogenesis in Xenopus competent ectoderm. Development 125, 4889–4900. Boutros, M., Kiger, A.A., Armknecht, S., Kerr, K., Hild, M., Koch, B., et al., 2004. Genome-wide RNAi analysis of growth and viability in Drosophila cells. Science 303, 832–835. Chambers, A.E., Kotecha, S., Towers, N., Mohun, T.J., 1992. Musclespecific expression of SRF-related genes in the early embryo of Xenopus laevis. EMBO J. 11, 4981–4991. Chambers, A.E., Logan, M., Kotecha, S., Towers, N., Sparrow, D., Mohun, T.J., 1994. The RSRF/MEF2 protein SL1 regulates cardiac muscle-specific transcription of a myosin light-chain gene in Xenopus embryos. Genes Dev. 8, 1324–1334. Chalmers, A.D., Strauss, B., Papalopulu, N., 2003. Oriented cell divisions asymmetrically segregate aPKC and generate cell fate diversity in the early Xenopus embryo. Development 130, 2657–2668. Chen, Y., Hollemann, T., Grunz, H., Pieler, T., 1999. Characterization of the Ets-type protein ER81 in Xenopus embryos. Mech. Dev. 80, 67–76. Chen, E., Hermanson, S., Ekker, S.C., 2004a. Syndecan-2 is essential for angiogenic sprouting during zebrafish development. Blood 103, 1710– 1719. Chen, J.-A., Voigt, J., Gilchrist, M., Papalopulu, N., Amaya, E., 2005. Identification of novel genes affecting mesoderm formation and morphogenesis through an enhanced large scale functional screen in Xenopus. Mech. Dev. 122, 307–331. Chitnis, A., Kintner, C., 1995. Neural induction and neurogenesis in amphibian embryos. Perspect Dev. Neurobiol. 3, 3–15. Cohen-Solal, K.A., Sood, R., Marin, Y., Crespo-Carbone, S.M., Sinsimer, D., Martino, J.J., et al., 2003. Identification and characterization of mouse Rab32 by mRNA and protein expression analysis. Biochim. Biophys. Acta 1651, 68–75. Darken, R.S., Scola, A.M., Rakeman, A.S., Das, G., Mlodzik, M., Wilson, P.A., 2002. The planar polarity gene strabismus regulates convergent extension movements in Xenopus. EMBO J. 21, 976–985. Davis, S.W., Smith, W.C., 2002. Expression cloning in ascidians: isolation of a novel member of the asctacin protease family. Dev. Genes Evol. 212, 81–86. Diez del Corral, R., Breitkreuz, D.N., Storey, K.G., 2002. Onset of neuronal differentiation is regulated by paraxial mesoderm and requires attenuation of FGF signalling. Development 129, 1681–1691. Domingos, P.M., Itasaki, N., Jones, C.M., Mercurio, S., Sargent, M.G., Smith, J.C., et al., 2001. The Wnt/beta-catenin pathway posteriorizes neural tissue in Xenopus by an indirect mechanism requiring FGF signalling. Dev. Biol. 239, 148–160.

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306 Driever, W., Solnica-Krezel, L., Schier, A.F., Neuhauss, S.C., Malicki, J., Stemple, D.L., et al., 1996. A genetic screen for mutations affecting embryogenesis in zebrafish. Development 123, 37–46. D’Souza, A., Lee, M., Taverner, N., Mason, J., Carruthers, S., Smith, J.C., et al., 2003. Molecular components of the endoderm specification pathway in Xenopus tropicalis. Dev. Dyn. 226, 118–127. Fraser, A.G., Kamath, R.S., Zipperlen, P., Martinez-Campos, M., Sohrmann, M., Ahringer, J., 2000. Functional genomic analysis of C. elegans chromosome I by systematic RNA interference. Nature 408, 325–330. Funayama, N., Fagotto, F., McCrea, P., Gumbiner, B.M., 1995. Embryonic axis induction by the armadillo repeat domain of beta-catenin: evidence for intracellular signaling. J. Cell. Biol. 128, 959–968. Gilchrist, M., Zorn, A.M., Voigt, J., Smith, J.C., Papalopulu, N., Amaya, E., 2004. Defining a large set of full lentgh clones from a Xenopus tropicalis EST project. Dev. Biol. 271, 498–516. Glinka, A., Delius, H., Blumenstock, C., Niehrs, C., 1996. Combinatorial signalling by Xwnt-11 and Xnr3 in the organizer epithelium. Mech. Dev. 60, 221–231. Glinka, A., Wu, W., Delius, H., Monaghan, A.P., Blumenstock, C., Niehrs, C., 1998. Dickkopf-1 is a member of a new family of secreted proteins and functions in head induction. Nature 391, 357–362. Gonczy, P., Echeverri, C., Oegema, K., Coulson, A., Jones, S.J., Copley, R.R., et al., 2000. Functional genomic analysis of cell division in C. elegans using RNAi of genes on chromosome III. Nature 408, 331–336. Grammer, T.C., Liu, K.J., Mariani, F.V., Harland, R.M., 2000. Use of largescale expression cloning screens in the Xenopus laevis tadpole to identify gene function. Dev. Biol. 228, 197–210. Gross, S.P., Tuma, M.C., Deacon, S.W., Serpinskaya, A.S., Reilein, A.R., Gelfand, V.I., 2002. Interactions and regulation of molecular motors in Xenopus melanophores. J. Cell. Biol. 156, 855–865. Gu, Z., Steinmetz, L.M., Gu, X., Scharfe, C., Davis, R.W., Li, W.H., 2003. Role of duplicate genes in genetic robustness against null mutations. Nature 421, 63–66. Haffter, P., Granato, M., Brand, M., Mullins, M.C., Hammerschmidt, M., Kane, D.A., et al., 1996. The identification of genes with unique and essential functions in the development of the zebrafish, Danio rerio. Development 123, 1–36. Hardcastle, Z., Papalopulu, N., 2000. Distinct effects of XBF-1 in regulating the cell cycle inhibitor p27(XIC1) and imparting a neural fate. Development 127, 1303–1314. Hardcastle, Z., Chalmers, A.D., Papalopulu, N., 2000. FGF-8 stimulates neuronal differentiation through FGFR-4a and interferes with mesoderm induction in Xenopus embryos. Curr. Biol. 10, 1511–1514. Harland, R.M., 1991. In situ hybridization: an improved whole-mount method for Xenopus embryos. Methods Cell. Biol. 36, 685–695. Hsu, D.R., Economides, A.N., Wang, X., Eimon, P.M., Harland, R.M., 1998. The Xenopus dorsalizing factor Gremlin identifies a novel family of secreted proteins that antagonize BMP activities. Mol. Cell 1, 673– 683. Itasaki, N., Jones, C.M., Mercurio, S., Rowe, A., Domingos, P.M., Smith, J.C., et al., 2003. Wise, a context-dependent activator and inhibitor of Wnt signalling. Development 130, 4295–4305. Kamath, R.S., Fraser, A.G., Dong, Y., Poulin, G., Durbin, R., Gotta, M., et al., 2003. Systematic functional analysis of the Caenorhabditis elegans genome using RNAi. Nature 421, 231–237. Kenwrick, S., Amaya, E., Papalopulu, N., 2004. Pilot morpholino screen in Xenopus tropicalis identifies a novel gene involved in head development. Dev. Dyn. 229, 289–299. Khokha, M.K., Chung, C., Bustamante, E.L., Gaw, L.W., Trott, K.A., Yeh, J., et al., 2002. Techniques and probes for the study of Xenopus tropicalis development. Dev. Dyn. 225, 499–510. Kile, B.T., Hentges, K.E., Clark, A.T., Nakamura, H., Salinger, A.P., Liu, B., et al., 2003. Functional genetic analysis of mouse chromosome 11. Nature 425, 81–86.

305

Kishi, M., Mizuseki, K., Sasai, N., Yamazaki, H., Shiota, K., Nakanishi, S., et al., 2000. Requirement of Sox2-mediated signaling for differentiation of early Xenopus neuroectoderm. Development 127, 791–800. Kroll, K.L., Salic, A.N., Evans, L.M., Kirschner, M.W., 1998. Geminin, a neuralizing molecule that demarcates the future neural plate at the onset of gastrulation. Development 125, 3247–3258. Launay, C., Fromentoux, V., Shi, D.L., Boucaut, J.C., 1996. A truncated FGF receptor blocks neural induction by endogenous Xenopus inducers. Development 122, 869–880. Laurent, M.N., Blitz, I.L., Hashimoto, C., Rothbacher, U., Cho, K.W., 1997. The Xenopus homeobox gene twin mediates Wnt induction of goosecoid in establishment of Spemann’s organizer. Development 124, 4905–4916. Lee, J.E., Hollenberg, S.M., Snider, L., Turner, D.L., Lipnick, N., Weintraub, H., 1995. Conversion of Xenopus ectoderm into neurons by NeuroD, a basic helix–loop–helix protein. Science 268, 836–844. Lemaire, P., Garrett, N., Gurdon, J.B., 1995. Expression cloning of Siamois, a Xenopus homeobox gene expressed in dorsal-vegetal cells of blastulae and able to induce a complete secondary axis. Cell 81, 85–94. Lum, L., Yao, S., Mozer, B., Rovescalli, A., von Kessler, D., Nirenberg, M., et al., 2003. Identification of Hedgehog pathway components by RNAi in Drosophila cultured cells. Science 299, 2039–2045. Lustig, K.D., Kroll, K.L., Sun, E.E., Kirschner, M.W., 1996. Expression cloning of a Xenopus T-related gene (Xombi) involved in mesodermal patterning and blastopore lip formation. Development 122, 4001–4012. Lustig, K.D., Stukenberg, P.T., McGarry, T.J., King, R.W., Cryns, V.L., Mead, P.E., et al., 1997. Small pool expression screening: identification of genes involved in cell cycle control, apoptosis, and early development. Methods Enzymol. 283, 83–99. Ma, Q., Kintner, C., Anderson, D.J., 1996. Identification of neurogenin, a vertebrate neuronal determination gene. Cell 87, 43–52. Mathis, L., Kulesa, P.M., Fraser, S.E., 2001. FGF receptor signalling is required to maintain neural progenitors during Hensen’s node progression. Nat. Cell. Biol. 3, 559–566. McKay, R.M., Peters, J.M., Graff, J.M., 2001. The casein kinase I family in Wnt signaling. Dev. Biol. 235, 388–396. Mead, P.E., Brivanlou, I.H., Kelley, C.M., Zon, L.I., 1996. BMP-4responsive regulation of dorsal-ventral patterning by the homeobox protein Mix.1. Nature 382, 357–360. Mitchell, K.J., Pinson, K.I., Kelly, O.G., Brennan, J., Zupicich, J., Scherz, P., et al., 2001. Functional analysis of secreted and transmembrane proteins critical to mouse development. Nat. Genet. 28, 241–249. Mizuseki, K., Kishi, M., Matsui, M., Nakanishi, S., Sasai, Y., 1998. Xenopus Zic-related-1 and Sox-2, two factors induced by chordin, have distinct activities in the initiation of neural induction. Development 125, 579–587. Moriwaki, J., Kajita, E., Kirikoshi, H., Koike, J., Sagara, N., Yasuhiko, Y., et al., 2000. Isolation of Xenopus frizzled-10A and frizzled-10B genomic clones and their expression in adult tissues and embryos. Biochem. Biophys. Res. Commun. 278, 377–384. Munchberg, S.R., Steinbeisser, H., 1999. The Xenopus Ets transcription factor XER81 is a target of the FGF signaling pathway. Mech. Dev. 80, 53–65. Munoz-Sanjuan, I., Brivanlou, A.H., 2002. Neural induction, the default model and embryonic stem cells. Nat. Rev. Neurosci. 3, 271–280. Nakata, K., Nagai, T., Aruga, J., Mikoshiba, K., 1997. Xenopus Zic3, a primary regulator both in neural and neural crest development. Proc. Natl Acad. Sci. USA 94, 11980–11985. Nusslein-Volhard, C., Wieschaus, E., 1980. Mutations affecting segment number and polarity in Drosophila. Nature 287, 795–801. Oschwald, R., Richter, K., Grunz, H., 1991. Localization of a nervous system-specific class II beta-tubulin gene in Xenopus laevis embryos by whole-mount in situ hybridization. Int. J. Dev. Biol. 35, 399–405. Papalopulu, N., Kintner, C., 1993. Xenopus Distal-less related homeobox genes are expressed in the developing forebrain and are induced by planar signals. Development 117, 961–975.

306

J. Voigt et al. / Mechanisms of Development 122 (2005) 289–306

Park, M., Moon, R.T., 2002. The planar cell-polarity gene stbm regulates cell behaviour and cell fate in vertebrate embryos. Nat. Cell. Biol. 4, 20–25. Penzel, R., Oschwald, R., Chen, Y., Tacke, L., Grunz, H., 1997. Characterization and early embryonic expression of a neural specific transcription factor xSOX3 in Xenopus laevis. Int. J. Dev. Biol. 41, 667– 677. Pera, E.M., Ikeda, A., Eivers, E., de Robertis, E.M., 2003. Integration of IGF, FGF, and anti-BMP signals via Smad1 phosphorylation in neural induction. Genes Dev. 17, 3023–3028. Peters, J.M., McKay, R.M., McKay, J.P., Graff, J.M., 1999. Casein kinase I transduces Wnt signals. Nature 401, 345–350. Rex, M., Orme, A., Uwanogho, D., Tointon, K., Wigmore, P.M., Sharpe, P.T., et al., 1997. Dynamic expression of chicken Sox2 and Sox3 genes in ectoderm induced to form neural tissue. Dev. Dyn. 209, 323–332. Rørth, P., Szabo, K., Bailey, A., Laverty, T., Rehm, J., Rubin, G.M., et al., 1998. Systematic gain-of-function genetics in Drosophila. Development 125, 1049–1057. Sakanaka, C., Leong, P., Xu, L., Harrison, S.D., Williams, L.T., 1999. Casein kinase iepsilon in the wnt pathway: regulation of beta-catenin function. Proc. Natl Acad. Sci. USA 96, 12548–12552. Salic, A.N., Kroll, K.L., Evans, L.M., Kirschner, M.W., 1997. Sizzled: a secreted Xwnt8 antagonist expressed in the ventral marginal zone of Xenopus embryos. Development 124, 4739–4748. Sanjo, H., Kawai, T., Akira, S., 1998. DRAKs, novel serine/threonine kinases related to death-associated protein kinase that trigger apoptosis. J. Biol. Chem. 273, 29066–29071. Skarnes, W.C., von Melchner, H., Wurst, W., Hicks, G., Nord, A.S., Cox, T., et al., 2004. A public gene trap resource for mouse functional genomics. Nat. Genet. 36, 543–544.

Smith, W.C., Harland, R.M., 1991. Injected Xwnt-8 RNA acts early in Xenopus embryos to promote formation of a vegetal dorsalizing center. Cell 67, 753–765. Smith, W.C., Harland, R.M., 1992. Expression cloning of noggin, a new dorsalizing factor localized to the Spemann organizer in Xenopus embryos. Cell 70, 829–840. Smith, W.C., McKendry, R., Ribisi Jr., S., Harland, R.M., 1995. A nodalrelated gene defines a physical and functional domain within the Spemann organizer. Cell 82, 37–46. Stemple, D.L., 2004. TILLING-a high-throughput harvest for functional genomics. Nat. Rev. Genet. 5, 145–150. Streit, A., Berliner, A.J., Papanayotou, C., Sirulnik, A., Stern, C.D., 2000. Initiation of neural induction by FGF signalling before gastrulation. Nature 406, 74–78. Taelman, V., Opdecamp, K., Avalosse, B., Ryan, K., Bellefroid, E.J., 2001. Xath2, a bHLH gene expressed during a late transition stage of neurogenesis in the forebrain of Xenopus embryos. Mech. Dev. 101, 199–202. Tyers, M., Jorgensen, P., 2000. Proteolysis and the cell cycle: with this RING I do thee destroy. Curr. Opin. Genet. Dev. 10, 54–64. Wheeler, G.N., Hoppler, S., 1999. Two novel Xenopus frizzled genes expressed in developing heart and brain. Mech. Dev. 86, 203–207. Wilson, S.I., Graziano, E., Harland, R., Jessell, T.M., Edlund, T., 2000. An early requirement for FGF signalling in the acquisition of neural cell fate in the chick embryo. Curr. Biol. 10, 421–429. Yamada, L., Shoguchi, E., Wada, S., Kobayashi, K., Mochizuki, Y., Satou, Y., et al., 2003. Morpholino-based gene knockdown screen of novel genes with developmental function in Ciona intestinalis. Development 130, 6485–6495. Zorn, A.M., Barish, G.D., Williams, B.O., Lavender, P., Klymkowsky, M.W., Varmus, H.E., 1999. Regulation of Wnt signaling by Sox proteins: XSox17 alpha/beta and XSox3 physically interact with beta-catenin. Mol. Cell 4, 487–498.