Gustavo A. Cerda and Murray Hargrave contributed equally to this work.
RNA profiling of FAC-sorted neurons from the developing zebrafish spinal cord
Article first published online: 18 DEC 2008
Copyright © 2008 Wiley-Liss, Inc.
Volume 238, Issue 1, pages 150–161, January 2009
How to Cite
Cerda, G. A., Hargrave, M. and Lewis, K. E. (2009), RNA profiling of FAC-sorted neurons from the developing zebrafish spinal cord. Dev. Dyn., 238: 150–161. doi: 10.1002/dvdy.21818
- Issue published online: 18 DEC 2008
- Article first published online: 18 DEC 2008
- Manuscript Accepted: 27 OCT 2008
- Medical Research Council. Grant Number: G0600877
- Royal Society University Research Fellowship
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
In this report, we describe a successful protocol for isolating and expression-profiling live fluorescent-protein-labelled neurons from zebrafish embryos. As a proof-of-principle for this method, we FAC-sorted and RNA-profiled GFP-labelled spinal CiA interneurons and compared the expression profile of these cells to those of post-mitotic spinal neurons in general and to all trunk cells. We show that RNA of sufficient quality and quantity to uncover both expected and novel transcription profiles via Affymetrix microarray analysis can be extracted from 5,700 to 20,000 FAC-sorted cells. As part of this study, we also further confirm the genetic homology of mammalian and zebrafish V1 interneurons, by demonstrating that zebrafish V1 cells (CiAs) express genes that encode for the transcription factors Lhx1a and Lhx5. This protocol for dissociating, sorting and RNA-profiling neurons from organogenesis-stage zebrafish embryos should also be applicable to other developing organs and tissues and potentially other model organisms. Developmental Dynamics 238:150–161, 2009. © 2008 Wiley-Liss, Inc.
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
The observation, probing, and isolation of embryonic neurons that have developed in their proper context has been greatly facilitated by transgenic methods for labelling these cells with fluorescent proteins. Zebrafish embryos are a particularly powerful model system for these sorts of studies as they develop outside the mother and are optically transparent, enabling us to easily observe fluorescently-labelled cells in live embryos (e.g., Finley et al.,2001; Zou et al.,2006; Zhang and Rodaway,2007; Detrich,2008; Higashijima,2008). In addition, several effective methods now exist for creating both transient and stable transgenic zebrafish embryos (e.g., Higashijima et al.,1997; Koster and Fraser,2001; Shin et al.,2003; Grabher et al.,2004; Kawakami,2005; Kwan et al.,2007; Villefranc et al.,2007) and it is possible to routinely obtain large numbers of these embryos due to the fecundity of zebrafish and the ease of using these technologies. All of these factors mean that powerful experiments investigating the development and properties of specific neurons should now be feasible. For example, if a particular population of neurons can be successfully isolated, it should be possible to determine the complete transcriptome of these cells through expression-profiling. However, RNA profiling requires the isolation of relatively pure populations of cells. FAC-sorting of fluorescently-labelled neurons is one method for obtaining such pure populations. While a few different tissues have now been FAC-sorted from zebrafish embryos (e.g., Covassin et al.,2006; Takizawa et al.,2007; Fan et al.,2008), to our knowledge there is no published account of expression profiling of FAC-sorted zebrafish neurons. Therefore, we decided to determine whether we could obtain RNA from FAC-sorted GFP-labelled zebrafish neurons of sufficient quality and quantity for expression-profiling. We also wanted to test that this procedure (with its associated cellular dissociation and axotomy) does not change the expression profile of neurons to such an extent, that it precludes the identification of genes normally expressed during the development of these cells.
In this report, we concentrate on isolating and expression-profiling both post-mitotic neurons in general and Circumferential Ascending interneurons (CiAs) as a proof-of-principle for FAC-sorting and RNA-profiling fluorescent-protein-labelled zebrafish neurons. CiAs are one of the most characterised classes of zebrafish spinal interneurons and they share many characteristics and are probably functionally homologous with mammalian V1 cells (Saueressig et al.,1999; Higashijima et al.,2004; Sapir et al.,2004; Alvarez et al.,2005). In both zebrafish and mouse, CiAs/V1 cells are the only spinal cells that express the transcription factor Engrailed-1 (Eng1b in zebrafish, En1 in mouse; Saueressig et al.,1999; Higashijima et al.,2004) and, along with other more dorsal spinal interneurons, CiAs, like V1 cells, also express Pax2 (Burrill et al.,1997; Batista and Lewis,2008). In addition, mammalian V1 cells express Lhx1 and Lhx5, but it is not yet known whether CiAs also express these transcription factors (Sheng et al.,1997; Gross et al.,2002; Pillai et al.,2007).
We have previously shown that CiAs express GFP in Tg(pax2a:GFP) embryos (Picker et al.,2002; Batista and Lewis,2008). This transgenic line labels a subset of Pax2-expressing spinal cord interneurons, the vast majority of which, at 24 hours post fertilization (hpf), are CiAs (Batista and Lewis,2008). Therefore, we reasoned that we could use this line to FAC-sort GFP-labelled CiAs and test whether we could obtain RNA from these cells of sufficient quality and quantity for expression-profiling. As a further proof-of-principle for FAC-sorting and RNA-profiling fluorescent-protein-labelled zebrafish neurons and as a comparison population for the Tg(pax2a:GFP) samples, we decided to FAC-sort EGFP-labelled neurons from Tg(elav13:EGFP) embryos, in which at least most post-mitotic neurons express EGFP (Park et al.,2000; Gribble et al.,2007). Finally, as a control and further comparison population, we also FAC-sorted all trunk cells.
Our results suggest that we were able to successfully isolate all three populations of cells without significantly changing their expression profiles. When we analysed the expression profiles of control genes that we expected to be expressed in particular populations, their expression was as predicted. In addition, when we analysed the top 26 transcripts that were both differentially expressed in at least one of these three different cell populations and where the expression of the associated gene has been documented by in situ hybridization, our expression-profiling results were consistent with this previously published expression data.
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
As a proof-of-principle for FAC-sorting and RNA-profiling fluorescent-protein-labelled zebrafish neurons, we decided to isolate and compare GFP-labelled neurons from Tg(pax2a:GFP) embryos (Picker et al.,2002) and EGFP-labelled neurons from Tg(elav13:EGFP) embryos (Park et al.,2000). In addition, as a control comparison population, we isolated samples of all trunk cells. We envisaged that isolating and expression-profiling these three nested populations of cells would allow our later comparisons and data analysis to distinguish both genes up-regulated in post-mitotic neurons relative to trunk cells in general and genes up-regulated in CiAs relative to all post-mitotic neurons. For the trunk cell samples, two samples were collected from Tg(pax2a:GFP) and two from Tg(elav13:EGFP) non-transgenic sibling embryos, to enable us to confirm that none of our results were due to differences in the genetic background of these two zebrafish lines.
After examining a variety of different developmental stages of Tg(pax2a:GFP) and Tg(elav13:EGFP) embryos, we decided to isolate cells from embryos at 27 hpf (prim stage 10–11), in the region of the trunk that lies above the yolk extension (Fig. 1A). In this region of the spinal cord at this stage, Tg(elav13:EGFP) embryos have EGFP-expressing neurons throughout the dorsal-ventral extent of the spinal cord; there are approximately 100 GFP-labelled spinal neurons in Tg(pax2a:GFP) embryos and the vast majority of these GFP-expressing neurons are CiAs (Fig. 1A; Batista and Lewis,2008). In addition, this was a relatively easy region of the embryo to dissect quickly, enabling us to collect sufficient material for a FAC-sort in a 1- to 2-hr period and, hence, to minimise the time between embryo dissection and RNA preparation (for details, see Experimental Procedures section).
At the chosen stage (prim stage 10–11) Tg(pax2a:GFP) embryos also express GFP in ventral mesoderm (at least some of which is probably pronephros; Fig. 1A; Krauss et al.,1991; Majumdar et al.,2000), blood cells (Supp. Fig. 1, which is available online), and occasional cells that we judged to be vascular tissue (data not shown). Therefore, for the cases where we were collecting GFP-expressing cells from Tg(pax2a:GFP) embryos, we dissected away as much of the pronephros area as we could.
In all cases, up to 185 trunks were dissociated in the same tube, using the Worthington Papain Dissociation system, to produce one sample for FAC sorting (for more details see Experimental Procedures section). We then FAC-sorted GFP-positive cells from Tg(pax2a:GFP) or Tg(elavl3:EGFP) embryos and all trunk cells from Tg(elavl3:EGFP) or Tg(pax2a:GFP) non-transgenic siblings (Fig. 1; for details see Experimental Procedures section). The three different populations [all trunk cells as a control; GFP-positive cells from Tg(elavl3:EGFP) trunks and GFP-positive cells from Tg(pax2:GFP) trunks] were all treated in as identical a manner as possible. In each case, single cells were FAC-sorted and collected in groups of up to 20,000 cells in 100 μl of Ambion RNAaqueous-Micro lysis buffer. Finally, we extracted RNA using an Ambion RNAaqueous-Micro kit (for more details see Experimental Procedures section).
To assess RNA quality, we examined RNA electropherograms generated by running RNA samples on Agilent RNA pico chips on an Agilent 2100 Bioanalyzer. We judged RNA quality primarily on whether samples produced electropherograms with large and well-defined ribosomal RNA peaks and otherwise low signal (Supp. Fig. 2; see also Agilent 2100 Bioanalyzer Application Compendium, Publication Number: 5989-3542EN, at http://www.chem.agilent.com/scripts/library.asp). Analysis software associated with the RNA pico chips also generates an RNA Integrity Number (RIN) score for each RNA sample (Schroeder et al.,2006). However, as the software failed to generate this score for some samples that otherwise appeared to be of acceptable quality (compare Table 1 and Supp. Fig. 2), we chose to use the RIN score as a secondary indicator of RNA quality. We found that RNA samples extracted from 20,000 cells was consistently of good quality whereas the quality of RNA samples extracted from less than 10,000 cells was more variable (some samples were of high quality and some samples were of poorer quality; Supp. Fig. 2).
|Sample||Number of cells||Total RNA concentration [ng/μl]||RIN|
To measure the RNA concentration in our samples, we used nucleic acid-binding fluorescent dye technology (Jones et al.,1998). We found that of 23 samples of acceptable quality (judged as described above), one had an unacceptably low RNA concentration while the remainder had a concentration of 3.8 ng/μl or more and an average concentration of 16.4 ng/μl. From these, we chose 15 samples to RNA-profile. These consisted of four samples of all trunk cells, five samples of GFP-positive cells from Tg(elav13:EGFP) embryos, and six samples of GFP-positive cells from Tg(pax2a:GFP) embryos (see Table 1). The RNA concentration of these samples ranged from 3.8–25.4 ng/μl and their RIN values ranged from 7.7 to 10 (see Table 1 for concentration values; electropherograms for each of these samples are shown in Supp. Fig. 2).
Expression-Profiling Using Affymetrix Microarrays
To determine whether our RNA extracted from FAC-sorted cells was of sufficient quality and quantity to generate accurate expression profiles, we sent samples to the Genomics Core Facility at EMBL, Heidelberg, for two-cycle amplification, labelling, and hybridisation to Affymetrix GeneChip Zebrafish Genome Array microarray chips.
We normalised primary intensity values from the microarrays with the Robust Multi-chip Algorithm (RMA) as this has previously been shown to perform well with Affymetrix microarray data (Irizarry et al.,2003). As an initial assessment of our cell sorting and expression profiling, we examined the expression profiles of transcripts known to be predominantly or exclusively expressed in one of the three experimental groups. As previously discussed, both pax2 and eng1b are expressed by CiAs, which are the majority of the GFP-expressing spinal cells in Tg(pax2:GFP) embryos. Therefore, we expected to see a relatively high level of expression of these genes in the Tg(pax2a:GFP) GFP-positive samples compared to the other samples and, indeed, this is what we observed (Fig. 2A). As the neurons in the Tg(pax2a:GFP) GFP-positive samples are a subset of those in the Tg(elav13:EGFP) GFP-positive samples, a lower level of expression is also seen in the latter. The lowest level of expression is seen in the trunk samples even though this group includes all cells from the other two groups and both pax2a and eng1b are expressed in non-neuronal trunk tissues (pax2a in pronephros and eng1b in muscle pioneers; Hatta et al.,1991; Krauss et al.,1991; Majumdar et al.,2000; Higashijima et al.,2004). However, this probably reflects the fact that the expression domains of these genes only represent a relatively small fraction of the trunk (Fig. 2 shows relative expression levels of each gene, where values have been transformed to a mean of zero and standard deviation of one, rather than absolute expression levels; www.gepas.org; Montaner et al.,2006).
Just as eng1b spinal cord expression is specific to CiA interneurons, in a few additional cases genes that encode for other transcription factors have been identified as being expressed by particular morphologically-identified zebrafish spinal interneurons. These include: vsx1, which is expressed by CiD interneurons that form just ventral to CiAs (Batista et al.,2008; Kimura et al.,2008); gata3, which is expressed by VeLD interneurons that also form just ventral to CiAs and by KA neurons that form in the most ventral part of the spinal cord (Batista et al.,2008); and tbx16 (also called spadetail), which is expressed by more dorsally located DoLA interneurons (Tamme et al.,2002).
As we would expect, the relative expression levels of these three genes are consistent with them being expressed by spinal neurons other than those labelled by the pax2a:GFP transgene (Fig. 2B). Their relative expression levels are highest in the Tg(elav13:EGFP) and lowest in the Tg(pax2a:GFP) GFP-positive samples.
We also examined the expression profiles of two genes expressed in the developing myotome: the transcription factor gene myf5 and the structural protein gene dystrophin (dmd; Bolanos-Jimenez et al.,2001; Coutelle et al.,2001; Thisse et al.,2004; www.zfin.org). We chose these as the myotome is one of the major, non-neural tissues in the trunk at 27 hpf and these genes have well-studied roles in this tissue. Again, we saw excellent correspondence between documented expression of these genes and the relative expression profiles from the microarray data (Fig. 2C).
As our chosen control genes seemed to be expressed as we would expect, we then subjected the microarray data to statistical tests to detect transcripts likely to be differentially expressed in at least one of the experimental groups. We used both a test that adjusts variance by empirical Bayes (eBayes) smoothing (Lönnstedt and Speed,2002; Smyth,2004) and an ANOVA test and we compared the results of each test.
To determine whether we could identify appropriately-expressed developmental genes in our three-different types of FAC-sorted cells, we examined the top 40 ranked probe sets from the eBayes-based test. All of these probe sets had an eBayes-based test value above 16 for the log odds (lods or B statistic; Lönnstedt and Speed,2002) of differential expression for at least one of the pair-wise comparisons between the experimental groups. This equates to odds of approximately 8.8 million:1 that these genes are differentially expressed in at least one of the pair-wise comparisons of our experimental groups (Supp. Table 1). All of these probe sets were also highly ranked by the ANOVA test (Fig. 3 and Supp. Table 1).
We first examined transcripts that were expressed at higher-than-average levels in the Tg(pax2a:GFP) GFP-positive samples. In the top 40 eBayes-ranked transcripts, we found three transcripts with relatively high expression in this experimental group and low expression in the other two experimental groups (Fig. 3A,B). The first transcript, GFP (targeted by two probe sets: AFFX-Dr-AF292560-1_s_at and AFFX-DrM62653-1_at), showed evidence of differential expression between the Tg(pax2a:GFP) GFP-positive samples and the other two experimental groups (but not between the Tg(elavl3:EGFP) EGFP-positive samples and the trunk samples) (Fig. 3A and Supp. Table 1). This is consistent with the GFP transcript only being expressed in GFP-positive cells sorted from Tg(pax2a:GFP) embryos. (The EGFP transcript present in Tg(elavl3:EGFP) embryos is not recognized by these probe sets and the trunk samples from the Tg(pax2a:GFP) line were collected from non-transgenic siblings.)
The other two genes, eng1b and lhx5, showed evidence of differential expression between all three experimental groups (Fig. 3B and Supp. Table 1). Expression was highest in the Tg(pax2a:GFP) GFP-positive samples, lower in the Tg(elavl3:EGFP) GFP-positive samples, and lowest in the trunk samples (Fig. 3B). As previously discussed, eng1b is exclusively expressed by CiA interneurons in spinal cord at 27 hpf (Higashijima et al.,2004) and the major fraction of GFP-positive cells sorted from the Tg(pax2a:GFP) line should be CiAs. lhx5 is expressed in several post-mitotic cells in the medial-dorsal region of zebrafish spinal cord, but the exact interneurons that express this gene have not yet been identified (Toyama et al.,1995). In amniotes, Lhx5 is expressed in several different populations of spinal interneurons including V1 cells, the amniote equivalent of CiAs, and more dorsal Pax2-expressing interneurons (Sheng et al.,1997; Gross et al.,2002; Pillai et al.,2007), so it was not surprising that our analysis suggested that this gene is expressed by CiAs. However, to further confirm this result, we performed double in situ hybridisation between lhx5 and eng1b and demonstrated that all CiAs express lhx5 (n = 110 cells in 6 embryos; Fig. 4A).
The expression profiles of four other transcripts, lhx1a, slc32a1 (previously called viaat), zgc:73142, and LOC100008236 showed evidence of differential expression between both of the GFP-positive samples and the general trunk samples, but not between the Tg(pax2a:GFP) and the Tg(elavl3:EGFP) GFP-positive samples (Fig. 3C and Supp. Table 1). Therefore, these genes are probably expressed both by neurons labelled by the pax2a:GFP transgene (the vast majority of which are CiAs) and other post-mitotic neurons.
In zebrafish, lhx1a, like lhx5, is expressed in a several post-mitotic cells in the medial-dorsal region of zebrafish spinal cord, but the exact interneurons that express this gene have not yet been identified (Toyama and Dawid,1997; Thisse et al.,2004; www.zfin.org). However, like Lhx5, in amniotes, Lhx1 is expressed in several populations of spinal interneurons, including V1 cells and more dorsal Pax2-expressing interneurons (Sheng et al.,1997; Gross et al.,2002; Pillai et al.,2007), so we had predicted that our analysis might identify this gene as being expressed by CiAs. Interestingly, the lack of differential expression between the Tg(pax2a:GFP) and the Tg(elavl3:EGFP) GFP-positive samples suggests that either not all CiAs express lhx1a at 27 hpf, or that the expression of lhx1a in at least some CiAs is weaker than the expression of this gene in other spinal neurons. To distinguish between these possibilities, and to confirm that lhx1a is indeed expressed by CiAs (as opposed to the minority of Tg(pax2a:GFP) GFP-positive neurons that are not CiAs), we performed double in situ hybridisation between lhx1a and eng1b. Our results show that all CiAs express lhx1a (n = 108 cells in 6 embryos), but that the expression of lhx1a in CiAs is often weaker than its expression in more dorsal cells (Fig. 4B).
slc32a1 (also known as vesicular inhibitory amino acid transporter or viaat) is a known marker of inhibitory neurons (McIntire et al.,1997; Sagne et al.,1997) and it is likely to be expressed in a number of spinal neuron types, including CiAs, the majority of which are GABAergic and/or glycinergic (Higashijima et al.,2004; Batista and Lewis,2008). Therefore, we would expect to find relatively high levels of expression of this gene in both the Tg(pax2a:GFP) and the Tg(elavl3:EGFP) GFP-positive samples. However, as for lhx1a, the lack of differential expression between the Tg(pax2a:GFP) and the Tg(elavl3:EGFP) GFP-positive samples suggests that either not all CiAs express slc32a1 at 27 hpf, or that the expression of slc32a1 in at least some CiAs is weaker than the expression of this gene in other spinal neurons. This is not surprising as the FAC-sorted CiAs should have been at a variety of different developmental stages and only the developmentally “older” neurons will have started to express neurotransmitter markers (as CiAs form continuously over the first few days of development; Higashijima et al.,2004).
Of the two remaining transcripts in this category, zgc:73142 is broadly expressed throughout at least most of the spinal cord including the CiA domain (Thisse et al.,2004; www.zfin.org), which is consistent with its expression profile in our experiments, while the expression pattern of LOC100008236 is currently unknown.
In addition, there were five transcripts that, according to the expression profile diagram generated by the T-rex utility at the Gene Expression Profile Analysis Suite (GEPAS; Fig. 3D), appeared to have a relatively high expression in the Tg(elavl3:EGFP) GFP-positive samples and intermediate levels of expression in the Tg(pax2a:GFP) GFP-positive samples (Fig. 3D). However, for all five of these, the eBayes-based test B statistic (log odds) for differential expression between the Tg(elavl3:EGFP) and the Tg(pax2a:GFP) GFP-positive samples was below 2.5 (which is an odds ratio of 12:1) (Supp. Table 1). Three of these, tusc3, wu:fc39d03, and hcn2 had B statistics ranging from 1.42 (an odds ratio of 4:1) to 2.44 (an odds ratio of 11.5:1). Therefore, these may be expressed at a higher level in the Tg(elavl3:EGFP) GFP-positive samples than the Tg(pax2a:GFP) GFP-positive samples. However, according to this test, the other two transcripts, ndrg4 and coro1b, showed no statistical difference between their expression in the Tg(elavl3:EGFP) and the Tg(pax2a:GFP) GFP-positive samples. Therefore, ndrg4 and coro1b and, potentially, tusc3, wu:fc39d03, and hcn2 have similar expression-profiles to the genes in Figure 3C that are described above. This suggests that these genes are expressed by at least some neurons labelled by the pax2a:GFP transgene and some other post-mitotic neurons. Consistent with this, nrdg4 is expressed broadly in the medial region of the spinal cord in a region that overlaps where Tg(pax2a:GFP) GFP-positive neurons are found (Qu et al.,2008) and tusc3 and coro1b appear to be broadly expressed throughout at least most of the spinal cord (Thisse et al.,2004; www.zfin.org). The expression of wu:fc39d03 and hcn2 is currently unknown.
The last group of transcripts with higher-than-average expression in the Tg(pax2a:GFP) GFP-positive samples were six transcripts that had relatively high expression in both the Tg(pax2a:GFP) GFP-positive samples and the trunk samples (Fig. 3E). These transcripts all show differential expression between the Tg(elavl3:EGFP) GFP-positive samples and both the Tg(pax2a:GFP) GFP-positive samples and the trunk samples but not between the Tg(pax2a:GFP) GFP-positive samples and the trunk samples. The relatively high expression in the Tg(pax2a:GFP) GFP-positive samples and yet relatively low expression in Tg(elavl3:EGFP) GFP-positive samples suggests that these transcripts are not expressed in neurons, but are expressed in non-neuronal GFP-positive cells that were present in our Tg(pax2a:GFP) dissected samples. Consistent with this idea, all of these genes have previously documented expression in either the haematopoietic system or the developing trunk vasculature (Brownlie et al.,2003; Paw et al.,2003; Thisse et al.,2004; http://www.zfin.org), both sites of GFP expression in the Tg(pax2a:GFP) line (Supp. Fig. 1 and data not shown). While we attempted to dissect away the pronephros in the Tg(pax2a:GFP) embryos, at this stage of development some haematopoietic and vasculature cells are found more dorsally in the trunk, explaining why these genes are expressed at above-average levels in our FAC-sorted Tg(pax2a:GFP) GFP-positive cell samples.
We next turned our attention to those transcripts in the top 40 that appeared to have relatively high expression in the Tg(elavl3:EGFP) GFP-positive samples with clear differential expression between these samples and the other two experimental groups (Fig. 3F; Supp. Table 1). This relative expression profile suggests that these 6 transcripts are expressed in populations of spinal neurons that do not express the pax2a:GFP transgene. While the majority of these transcripts do not have documented expression patterns, the expression domains of gata3 and sst1 are consistent with this hypothesis. gata3 is, as discussed earlier, expressed in VeLD and KA spinal neurons (Batista et al.,2008) and sst1 is expressed in spinal motor neurons (Devos et al.,2002). Neither of these neuronal sub-types express pax2a (Batista and Lewis,2008). gata3 is also expressed in the posterior pronephros and proctodeum (Thisse et al.,2004; www.zfin.org; Pyati et al.,2006; Wingert et al.,2007) and this may account for the slightly variable, relatively low-level expression of this gene in the trunk samples.
The largest group of the top 40 ranked transcripts were those with highest relative expression in the trunk samples. This is, perhaps, unsurprising given the complex, multi-tissue origin of the RNA in these samples. The majority of transcripts within this group were those with relatively high expression in the trunk samples only (Fig. 3G). All 13 of these transcripts had eBayes-based test B statistics greater than 14.5 (an odds ratio of 1.9 million:1) for differential expression between the trunk samples and both the Tg(elavl3:EGFP) and the Tg(pax2a:GFP) GFP-positive samples (Supp. Table 1). While the expression of some of these genes is still unknown, all of the genes that have been characterised are expressed relatively highly in non-neuronal tissues located in the trunk region of the zebrafish embryo and are either not expressed or are only expressed at basal or background levels in the spinal cord (Ahmed et al.,2004; Thisse et al.,2004; Oishi et al.,2006; Shu et al.,2007).
Finally, two of the transcripts with the highest relative expression in the trunk samples were also expressed at intermediate levels in the Tg(elavl3:EGFP) GFP-positive samples (Fig. 3H). Both of these transcripts had eBayes-based test B statistics greater than 5.7 (an odds ratio of approximately 300:1) for differential expression between all three experimental groups (Supp. Table 1). The expression pattern of one of these transcripts is unknown; the other, zgc:92287, is expressed in myotome (Thisse et al.,2004; www.zfin.org), but it is not clear from the images whether there is any spinal cord expression.
Interestingly, in all of our analyses, we observed no obvious consistent differences between the expression-profiling data obtained from RNA samples that were prepared from different numbers of cells or that had different starting concentrations of RNA. This is even the case for pax2a sample number 5 (prepared from 20,000 cells, with a starting concentration of 8.6 ng/μl) and pax2a sample number 4 (prepared from 5,700 cells, with a starting concentration of 3.8 ng/μl) (see Table 1).
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
In this report, we describe for the first time a protocol for FAC-sorting GFP-expressing zebrafish neurons for RNA-profiling. As a proof-of-principle for this method, we FAC-sorted and expression-profiled a particular class of spinal interneurons (CiAs), all spinal neurons and all trunk cells. The number of cells collected varied from 5,700 to 20,000 and the concentration of the resulting RNA ranged from 3.8 to 25.4 ng/μl. Despite this variability, our analysis of the top 40 probe sets, ranked by an eBayes-based test for identifying differentially expressed transcripts, demonstrates that for all of our chosen samples our protocol produced RNA of sufficient quality and quantity for subsequent expression-profiling, in this case two-cycle amplification and hybridisation to Affymetrix microarray chips. We were motivated to perform this proof-of-principle experiment and analysis because we were concerned that dissecting trunks and then dissociating and FAC-sorting neurons, which presumably results in many of the neurons being axotomized, might change the fates of these cells and, hence, their expression profiles. Given that we chose a more complicated dissection of Tg(pax2a:GFP) embryos to remove the pronephros region and that this limited the number of dissections we could perform in a reasonable time period, we were also uncertain whether we would be able to collect sufficient CiAs for subsequent expression-profiling. While we have only analysed a subset of the genes identified by the expression-profiling of our different experimental samples, our data show that we can identify appropriately-expressed developmental genes in our samples of FAC-sorted neurons. When we analysed the top 40 probe sets that were differentially expressed in at least one of these three different cell populations, we found that for the 26 probe sets where the expression of the associated gene has been documented by in situ hybridization, our expression-profiling results were consistent with this previously published expression data. Notably, this includes genes both known and predicted to be expressed by the neurons that we isolated.
While this report concentrates on a particular class of spinal interneurons, CiAs, as a proof-of-principle for this method, the high relative expression of genes such as gata3, vsx1, tbx16, and sst in the Tg(elavl3:EGFP) GFP-positive samples suggests that other classes of spinal neurons can be successfully isolated and expression-profiled in this manner. Given that this protocol was successful for neurons, it should also be applicable to other fluorescently-labelled tissues in zebrafish embryos and potentially to fluorescently-labelled cells in other animals. In addition, while we chose to RNA-profile our samples using Affymetrix microarray chips, this protocol could just as easily be used to generate RNA for expression-profiling using other microarrays or alternative methods such as massive parallel DNA sequencing using either the SOLiD or the Solexa platforms (e.g., Cloonan et al.,2008; Mortazavi et al.,2008; Shendure,2008).
Our results also demonstrate that it is possible to use comparisons between different cell populations to sort out genes with particular expression profiles, including genes that are likely to be expressed in contaminating tissues (for example, in our case, haematopoietic cells). We primarily included the trunk samples to help us to identify genes that were specifically up-regulated in neurons compared to all trunk cells, but it also enabled us to distinguish between genes expressed by neuronal and non-neuronal GFP-positive cells FAC-sorted from Tg(pax2a:GFP) embryos.
Finally, our results also further demonstrate the genetic homology of zebrafish CiAs and mammalian V1 cells as we identify both lhx1a and lhx5 as being expressed by CiAs. This data adds to an increasing body of evidence suggesting that at least in the ventral spinal cord, fundamental processes of interneuron specification are highly conserved across vertebrates and that we can use the simpler zebrafish spinal cord to elucidate general principles of vertebrate spinal cord development and function (Higashijima et al.,2004; Goulding and Pfaff,2005; Kimura et al.,2006,2008; Lewis,2006; Batista et al.,2008; Batista and Lewis,2008).
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
Our protocol for dissociating and FAC sorting GFP-labelled cells is similar to that used previously to FAC-sort zebrafish germ cells and endothelial cells (Covassin et al.,2006; Takizawa et al.,2007; Fan et al.,2008), with the major differences that we used Papain to dissociate our tissue samples rather than Trypsin and we sorted directly into RNA extraction lysis buffer. Full details are given below.
Zebrafish (Danio rerio) were maintained on a 14-hr light/10-hr dark cycle at 28.5°C and embryos were obtained from natural, grouped spawnings of identified heterozygous or homozygous Tg(pax2a:GFP) (Picker et al.,2002) or Tg(elavl3:EGFP) (Park et al.,2000) adult fish. Embryos were allowed to develop normally and were staged according to prim staging (Kimmel et al.,1995).
Transgenic embryos were examined to determine the extent of GFP expression using a Zeiss Axio Imager M1 microscope. Images of GFP expression were taken with either a Zeiss Axio Imager M1 microscope or a Leica TS SP2 confocal microscope. In the latter case, projections of multiple optical sections were made with ImageJ software (Abramoff et al.,2004). Further image processing was performed with Adobe Photoshop CS software (Adobe, Inc).
Spinal Cord Dissections
For the GFP-positive samples, we dissected trunks from Tg(pax2a:GFP) and Tg(elavl3:EGFP) embryos at 27 hpf. For the trunk samples, we dissected trunks from Tg(pax2a:GFP) or Tg(elavl3:EGFP) non-transgenic sibling embryos at 27 hpf. The chorions were removed and the embryos were placed in a 90-mm Petri dish containing zebrafish Ringer's at 4°C (116 mM NaCl, 2.9 mM KCl, 1.8 mM CaCl2, and 5 mM HEPES, pH 7.2). We removed the yolks mechanically by gently pipetting the embryos up and down using a Gilson p-1000 pipette in groups of 50 embryos. Subsequently, using forceps and 30G needles, we dissected trunks from the yolk extension region; the pronephros region of Tg(pax2a:GFP) trunks was also removed. The trunks were collected on ice in Eppendorf tubes containing zebrafish Ringer's solution in batches (where each batch was then dissociated as one sample). For the Tg(elavl3:EGFP) GFP-positive samples, we usually collected batches of about 100 trunks; for the trunk control samples, we collected batches of about 30 trunks; and for the Tg(pax2a:GFP) GFP-positive samples we collected as many trunks as we could in a 2-hr period (batch sizes ranged from 50–185 trunks). Tubes with non-dissected but dissociated wild type and GFP-positive embryos were also collected to calibrate the FACS machine.
Cell dissociations were performed using the Worthington Papain Dissociation system, as Papain has been shown to be superior to other enzymes such as Trypsin and Collagenase for dissociating neural tissue (Huettner and Baughman,1986) and it is now used routinely for dissociating neural cells, particularly for creating neuroblast and neural stem cell cultures (e.g., Shihabuddin,2008). We performed the dissociations as per the instructions supplied with the kit, with the following modifications. We incubated the embryos in the dissociation reagents for 30 min at room temperature, followed by a mechanical dissociation step where we gently pipetted the tissue up and down three to five times (until the tissue was dissociated) using a Gilson p-200 pipette. Before the Papain inactivation with DNAse and ovomucoid, the cells were centrifuged at 300g for 5 min at room temperature and finally resuspended in 500 μl of L-15 media. Cell suspensions were then filtered through a 30μm nylon mesh (BD Falcon™ 5-ml Polystyrene tubes with a Cell Strainer Cap 30-μm pore size) to remove any remaining undissociated tissue before proceeding immediately to FAC sorting.
FAC Sorting and RNA Extraction
All cell samples (even the non-transgenic trunk samples) were sorted using a Beckman-Coulter 3 laser MoFlo sorting cytometer at the Flow Cytometry Facility, Department of Pathology, University of Cambridge (Cambridge, UK). We stained the cell suspensions with propidium iodide (PI) before FAC sorting. The first step in the FAC sorting procedure was to characterize a homogeneous cell population using Forward Scatter (FSC) and Side Scatter (SSC), measurements of cell size and granularity, respectively (R1 area in Fig. 1B). Subsequently, we used FSC and pulse width to select single cells (R2 area in Fig. 1B). We then selected the live, single-cell population by excluding PI-positive events (dead cells). On average, about 1% of the selected single cells (cells present both in R1 and R2) were excluded by this method (R4 area in Fig. 1B; see also bar chart in Fig. 1B). The parameters to select GFP-positive cell populations were calibrated using dissociated cells from un-dissected wild-type and GFP-positive embryos. The cut-off for collecting GFP-positive cells was identified by comparing the GFP fluorescence levels of Tg(elavl3:EGFP) or Tg(pax2a:GFP) cells versus wildtype cells (Fig. 1C). Finally, the live single GFP-positive cells from Tg(elavl3:EGFP) or Tg(pax2a:GFP) trunks (Fig. 1; events that are common between the R1, R2, R3, and R5 areas) or all live single cells from the trunk sample trunks (Fig. 1; events that are common between the R1, R2, and R3 areas), were selected and collected in a tube containing 100 μl of RNA extraction lysis buffer from the Ambion RNaqueous-Micro RNA extraction kit. The precise number of cells collected for each sample is provided in Table 1. Each collected cell was sorted in a drop of about 10 pl of PBS. Therefore, the largest sample size of 20,000 cells consisted of about 20 μl of PBS+cells, making a total volume of about 120 μl in the collection tube. For the Tg(pax2a:GFP) GFP-positive samples, we collected all of the FAC-sorted cells as a single sample. The only exception was the batch of 185 trunks, where we stopped collecting at 20,000 cells. For the other samples, we collected batches of 20,000 cells. (For example, from one cell dissociation of about 100 Tg(elavl3:EGFP) trunks, we could obtain 3–4 separate batches of 20,000 GFP-positive cells.) We performed the RNA extraction using an Ambion RNAaqueous-Micro Kit according to the supplied instructions but without the optional DNAse step. The RNA was eluted in 20 μl of elution solution (two 10-μl elution volumes subsequently combined).
RNA Quality Assessment
To determine the quality of RNA extracted from each batch of FAC-sorted cells, 1 μl of each of the purified RNA samples was run on an Agilent RNA 6000 pico chip using an Agilent RNA 6000 pico kit and an Agilent 2100 Bioanalyser. The pico chip assay was performed according to the manufacturer's instructions. The quality of the RNA was assessed primarily via the profile of the electropherogram and secondarily by the RNA integrity number (RIN) generated by the Bioanalyser software. As the RNA pico chips allow only an estimation of RNA concentration, we quantified the amount of RNA extracted by using a RiboGreen RNA Quantification Kit (Invitrogen) to fluorescently label the RNA and an Mx3000P qPCR machine (Stratagene) to quantify the fluorescence. In summary, we generated a triplicate series of RNA samples of known amount (0, 0.1, 0.2, 0.5, 1, 2, 3, 4, 5, and 10 ng) from the ribosomal RNA standard included in the RiboGreen kit and then generated a standard curve of fluorescence versus RNA quantity. The quantity of RNA in 0.5-μl aliquots of RiboGreen-labelled RNA from each experimental sample was then inferred from this standard curve. For cases where more than one RNA sample was obtained from the same cell dissociation, only the best sample was used for subsequent expression profiling.
RNA Amplification and Microarray Chip Hybridisation
We chose to expression-profile our samples using Affymetrix GeneChips as Affymetrix is a commonly used microarray platform that is often available through commercial services. In addition, Affymetrix GeneChips offer a technical advantage over several of the other platforms as they have 11 probes per gene, which makes technical replicates and dye swaps unnecessary. In our case, the amplification and hybridization of RNA samples was performed at GeneCore (Genomics Core Facility), EMBL, Heidelberg, Germany. RNA samples extracted from FAC-sorted cells were subjected to two rounds of amplification and labelled using Affymetrix Two-Cycle Target Labelling and Control Reagents according to the manufacturer's instructions. Affymetrix GeneChip spike-in poly-adenylated prokaryotic controls were included as amplification controls. Amplified RNA was hybridized to Affymetrix GeneChip Zebrafish Genome Array microarrays according to the manufacturer's instructions and Affymetrix GeneChip Hybridization Control RNAs were included.
Microarray Data Treatment and Analysis
We performed initial quality control assessments of the expression data from the microarrays using Affymetrix Expression Console software. We examined a number of metrics including background, noise, probe cell intensity, scaling factor, and percent of probe sets called present or absent, all of which suggested that our hybridisations were successful and the microarray data sets were reasonably consistent. The expression profiles shown in this report (Figs. 2 and 3) were generated by the T-rex utility at the Gene Expression Profile Analysis Suite (GEPAS; www.gepas.org; Montaner et al.,2006). They depict normalised expression data from each gene transformed to a mean of zero and standard deviation of one with a colour scale of +1 standard deviation (red) to −1 standard deviation (blue). Any expression values originally further than ±1 standard deviation from the mean are shown as ±1 standard deviation (www.gepas.org; Montaner et al.,2006). The normalised expression levels were calculated from probe-level data using the implementation of the Robust Multichip Algorithm (RMA; Irizarry et al.,2003) at GEPAS.
To reveal differentially expressed transcripts, we employed two statistical tests and compared the results from each. First, we used the affy Bioconductor package (Gautier et al.,2004; Gentleman et al.,2004) within the R statistical computing environment (R_Development_Core_Team,2005) to perform RMA expression level calculations and normalization (using the justRMA function and default values of quantile normalization and RMA background correction version 2) and then we used the eBayes function within the limma Bioconductor package (Smyth,2004) to detect differential expression between experimental groups. The default value of 0.1 was used for the proportion of expected differentially expressed genes and the default values of 0.1 and 4 were used for the standard deviation limits of log2 fold changes for differentially expressed genes. The default Benjamini & Hochberg (BH) method was used for P value adjustment. Potential differentially expressed genes were ranked by the associated F-statistic generated by the F-test of the three pair-wise comparisons as part of the eBayes function. Second, we used the multiclass ANOVA test (using the T-rex utility) at GEPAS on RMA normalised data and potential differentially expressed genes were ranked by their estimated P values.
Double In Situ Hybridisation
Embryos were fixed in 4% paraformaldehyde at 27 hpf and double in situ hybridisation was performed as previously described (Batista et al.,2008; Batista and Lewis,2008), with the exception that the eng1b probe was detected with Roche Anti-Fluorescein-AP FAB fragments and Fast Red (Sigma). Antibody incubations were performed simultaneously, followed by Fast Red staining and then Tyramide staining using Invitrogen Tyramide kit no. 12 (488). Probes used were eng1b (1.3 KB encompassing the ORF, a kind gift from Drs. Kikuchi and Westerfield at the University of Oregon; see also Batista and Lewis,2008), lhx1 (Toyama and Dawid,1997), and lhx5 (Toyama et al.,1995). Embryos were analysed and pictures were taken using a Leica TS SP2 confocal microscope. Projections of multiple optical sections were made with ImageJ software (Abramoff et al.,2004). Further image processing was performed with Adobe Photoshop CS software (Adobe, Inc). For both lhx1a + eng1b and lhx5 + eng1b double stainings, at least 6 embryos were analysed in detail. In all cases, all of the eng1b-expressing cells also expressed lhx1a and lhx5 (n = 108 for lhx1a and n = 110 for lhx5; n values are the total number of cells counted on one side of the spinal cord above the yolk extension in 6 different embryos).
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
Funding was provided by a Royal Society University Research Fellowship and a Medical Research Council Pilot Grant (G0600877) to K.E.L. We are grateful to Rick Livesey, Robert Knight, Vladimir Benes, and Matthew Clark for interesting and helpful discussions about microarray experiments; Krishna Zivraj and Christine Holt for helpful discussions about RNA extraction protocols; James Smith and Rick Livesey for help with RNA quality assessment; Isabelle Latorre, Paulina Kolasinska, David Rivers, and Julie Ahringer for help with RNA quantification; the organisers of the Analysis of Microarray Data course from Simon Tavaré's Computational Biology Group at the CRI for help with statistical analysis; Debbie Goode for help in determining Tg(pax2a:GFP) expression in blood cells; Manuel Batista for help with dissection and double labelling experiments; and Nigel Miller at the Flow Cytometry Facility, Department of Pathology, University of Cambridge, Cambridge, for his expert FAC-sorting. RNA amplification, labelling, and hybridisation was performed by Tomi Ivacevic and Vladimir Benes at the Genomics Core Facility at EMBL, Heidelberg. We also thank Michael Brand and Catherina Becker for providing us with Tg(pax2a:GFP) fish, Lucia Poggi for providing us with Tg(elavl3:EGFP) fish, and Adrian McNabb and Manuel Batista for help in maintaining fish lines.
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
- 2004. Image processing with ImageJ. Biophotonics Intl 11: 36.42. , , .
- 2004. Biochemical and molecular characterization of galectins from zebrafish (Danio rerio): notochord-specific expression of a prototype galectin during early embryogenesis. Glycobiology 14: 219–232. , , , .
- 2005. Postnatal phenotype and localization of spinal cord V1 derived interneurons. J Comp Neurol 493: 177–192. , , , , , , , .
- 2008. Pax2/8 act redundantly to specify inhibitory fates of multiple spinal interneurons. Dev Biol 323: 88–97. , .
- 2008. Zebrafish V2 cells develop into excitatory CiD and Delta/Notch dependent inhibitory VeLD interneurons. Dev Biol 322: 263–275. , , .
- 2001. Dystrophin and Dp71, two products of the DMD gene, show a different pattern of expression during embryonic development in zebrafish. Mech Dev 102: 239–241. , , , , , .
- 2003. Characterization of embryonic globin genes of the zebrafish. Dev Biol 255: 48–61. , , , , , , , , , , , , .
- 1997. PAX2 is expressed in multiple spinal cord interneurons, including a population of EN1+ interneurons that require PAX6 for their development. Development 124: 4493–4503. , , , .
- 2008. Stem cell transcriptome profiling via massive-scale mRNA sequencing. Nat Methods 5: 613–619. , , , , , , , , , , , , , , , , , , .
- 2001. Hedgehog signalling is required for maintenance of myf5 and myoD expression and timely terminal differentiation in zebrafish adaxial myogenesis. Dev Biol 236: 136–150. , , , , , .
- 2006. Global analysis of hematopoietic and vascular endothelial gene expression by tissue specific microarray profiling in zebrafish. Dev Biol 299: 551–562. , , , , , .
- 2008. Fluorescent proteins in zebrafish cell and developmental biology. Methods Cell Biol 85: 219–241. .
- 2002. Differential expression of two somatostatin genes during zebrafish embryonic development. Mech Dev 115: 133–137. , , , , , , .
- 2008. Zebrafish primordial germ cell cultures derived from vasa::RFP transgenic embryos. Stem Cells Dev 17: 585–597. , , , , .
- 2001. Three-color imaging using fluorescent proteins in living zebrafish embryos. Biotechniques 31: 66–70, 72. , , .
- 2004. affy—analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20: 307–315. , , , .
- 2004. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5: R80. , , , , , , , , , , , , , , , , , , , , , , , , .
- 2005. Development of circuits that generate simple rhythmic behaviors in vertebrates. Curr Opin Neurobiol 15: 14–20. , .
- 2004. Highly efficient zebrafish transgenesis mediated by the meganuclease I-SceI. Methods Cell Biol 77: 381–401. , , .
- 2007. Regulation and function of Dbx genes in the zebrafish spinal cord. Dev Dyn 236: 3472–3483. , , .
- 2002. Lbx1 specifies somatosensory association interneurons in the dorsal spinal cord. Neuron 34: 535–549. , , .
- 1991. Diversity of expression of engrailed-like antigens in zebrafish. Development 112: 821–832. , , , .
- 2008. Transgenic zebrafish expressing fluorescent proteins in central nervous system neurons. Dev Growth Differ 50: 407–413. .
- 1997. High-frequency generation of transgenic zebrafish which reliably express GFP in whole muscles or the whole body by using promoters of zebrafish origin. Dev Biol 192: 289–299. , , , , .
- 2004. Engrailed-1 expression marks a primitive class of inhibitory spinal interneuron. J Neurosci 24: 5827–5839. , , , .
- 1986. Primary culture of identified neurons from the visual cortex of postnatal rats. J Neurosci 6: 3044–3060. , .
- 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4: 249–264. , , , , , , .
- 1998. RNA quantitation by fluorescence-based solution assay: RiboGreen reagent characterization. Anal Biochem 265: 368–374. , , , .
- 2005. Transposon tools and methods in zebrafish. Dev Dyn 234: 244–254. .
- 1995. Stages of embryonic development of the zebrafish. Dev. Dyn. 203: 253–310. , , , , .
- 2006. alx, a zebrafish homolog of Chx10, marks ipsilateral descending excitatory interneurons that participate in the regulation of spinal locomotor circuits. J Neurosci 26: 5684–5697. , , .
- 2008. V2a and V2b neurons are generated by the final divisions of pair-producing progenitors in the zebrafish spinal cord. Development 135: 3001–3005. , , .
- 2001. Tracing transgene expression in living zebrafish embryos. Dev Biol 233: 329–346. , .
- 1991. Expression of the zebrafish paired box gene pax [zf-b] during early neurogenesis. Development 113: 1193–1206. , , , .
- 2007. The Tol2kit: a multisite gateway-based construction kit for Tol2 transposon transgenesis constructs. Dev Dyn 236: 3088–3099. , , , , , , , , , .
- 2006. How do genes regulate simple behaviours? Understanding how different neurons in the vertebrate spinal cord are genetically specified. Phil Trans R Soc Lond B Biol Sci 361: 45–66. .
- 2002. Replicated microarray data. Statistica Sinica 12: 31–46. , .
- 2000. Zebrafish no isthmus reveals a role for pax2.1 in tubule differentiation and patterning events in the pronephric primordia. Development 127: 2089–2098. , , , .
- 1997. Identification and characterization of the vesicular GABA transporter. Nature 389: 870–876. , , , , .
- 2006. Next station in microarray data analysis: GEPAS. Nucleic Acids Res 34: W486–491. , , , , , , , , , , , , , , .
- 2008. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat Methods 5: 621–628. , , , , .
- 2006. Regulation of primary cilia formation and left-right patterning in zebrafish by a noncanonical Wnt signaling mediator, duboraya. Nat Genet 38: 1316–1322. , , , , .
- 2000. Analysis of upstream elements in the HuC promoter leads to the establishment of transgenic zebrafish with fluorescent neurons. Dev Biol 227: 279–293. , , , , , , , , , , , , .
- 2003. Cell-specific mitotic defect and dyserythropoiesis associated with erythroid band 3 deficiency. Nat Genet 34: 59–64. , , , , , , , , , , , , , , , , , , , , , , , , .
- 2002. A novel positive transcriptional feedback loop in midbrain-hindbrain boundary development is revealed through analysis of the zebrafish pax2.1 promoter in transgenic lines. Development 129: 3227–3239. , , , , .
- 2007. Lhx1 and Lhx5 maintain the inhibitory-neurotransmitter status of interneurons in the dorsal spinal cord. Development 134: 357–366. , , , , .
- 2006. Sustained Bmp signaling is essential for cloaca development in zebrafish. Development 133: 2275–2284. , , , , .
- 2008. Ndrg4 is required for normal myocyte proliferation during early cardiac development in zebrafish. Dev Biol 317: 486–496. , , , , , , , .
- R_Development_Core_Team. 2005. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing.
- 1997. Cloning of a functional vesicular GABA and glycine transporter by screening of genome databases. FEBS Lett 417: 177–183. , , , , , , .
- 2004. Pax6 and engrailed 1 regulate two distinct aspects of renshaw cell development. J Neurosci 24: 1255–1264. , , , , , , , , .
- 1999. Engrailed-1 and netrin-1 regulate axon pathfinding by association interneurons that project to motor neurons. Development 126: 4201–4212. , , .
- 2006. The RIN: an RNA integrity number for assigning integrity values to RNA measurements. BMC Mol Biol 7: 3. , , , , , , , , , .
- 2008. The beginning of the end for microarrays? Nat Methods 5: 585–587. .
- 1997. Expression of murine Lhx5 suggests a role in specifying the forebrain. Dev Dyn 208: 266–277. , , , , , , .
- 2008. Adult rodent spinal cord-derived neural stem cells: isolation and characterization. Methods Mol Biol 438: 55–66. .
- 2003. Neural cell fate analysis in zebrafish using olig2 BAC transgenics. Methods Cell Sci 25: 7–14. , , , , .
- 2007. Na,K-ATPase alpha2 and Ncx4a regulate zebrafish left-right patterning. Development 134: 1921–1930. , , , , , .
- 2004. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 3: Article3. .
- 2007. Expression analysis of two Eomesodermin homologues in zebrafish lymphoid tissues and cells. Mol Immunol 44: 2324–2331. , , , , .
- 2002. The identity and distribution of neural cells expressing the mesodermal determinant spadetail. BMC Dev Biol 2: 9. , , , .
- 2004. Spatial and temporal expression of the zebrafish genome by large-scale in situ hybridization screening. Methods Cell Biol 77: 505–519. , , , , , , , , .
- 1995. The LIM class homeobox gene lim5: Implied role in CNS patterning in Xenopus and Zebrafish. Dev. Biol. 170: 583–593. , , , , , .
- 1997. lim6, a novel LIM homeobox gene in the zebrafish: comparison of its expression pattern with lim1. Dev Dyn 209: 406–417. , .
- 2007. Gateway compatible vectors for analysis of gene function in the zebrafish. Dev Dyn 236: 3077–3087. , , .
- 2007. The cdx genes and retinoic acid control the positioning and segmentation of the zebrafish pronephros. PLoS Genet 3: 1922–1938. , , , , , , , , , , .
- 2007. SCL-GFP transgenic zebrafish: in vivo imaging of blood and endothelial development and identification of the initial site of definitive hematopoiesis. Dev Biol 307: 179–194. , .
- 2006. The Fugu tyrp1 promoter directs specific GFP expression in zebrafish: tools to study the RPE and the neural crest-derived melanophores. Pigment Cell Res 19: 615–627. , , , , .
- Top of page
- EXPERIMENTAL PROCEDURES
- Supporting Information
Additional Supporting Information may be found in the online version of this article.
|dvdy_21818_sm_SuppFigsS1.tif||404K||Supp. Fig. 1. GFP expression in circulating blood cells in Tg(pax2a:GFP) embryos. Fluorescent microscopy images of a live, prim-11 stage (27 hpf) Tg(pax2a:GFP) embryo showing expression of GFP in circulating blood cells (arrows). The embryo was anaesthetised with tricaine and images were taken at consecutive time points (in order A, B, C, D) approximately one second apart. GFP-labelled blood cells can be seen moving posteriorly through the dorsal aorta and returning anteriorly through the posterior cardinal vein. The embryo is oriented with anterior to the left and dorsal up. Scale bar = 50 μm.|
|dvdy_21818_sm_SuppFigsS2.tif||1287K||Supp. Fig. 2. Electropherograms of RNA samples. Electropherograms produced by running RNA extracted from FAC-sorted cells on Aligent RNA pico chips. A: Electropherograms of all 15 samples chosen for Affymetrix microarray-based RNA profiling (see Table 1) are shown. B: Also shown are 3 RNA samples extracted from small numbers of cells (2,500, 3,800, and 5,000 cells, respectively) illustrating poorer quality RNA obtained from these cell samples. RIN numbers are also shown for these samples. RIN values for the first 15 samples are given in Table 1.|
|dvdy_21818_sm_SuppTableS1.doc||281K||Supplementary Data Table 1. Tabulated output from eBayes-based and ANOVA tests for differential expression, Part I|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.