SEARCH

SEARCH BY CITATION

Keywords:

  • siRNA;
  • nervous system;
  • chick embryo;
  • neuropilin;
  • semaphorin;
  • axon guidance

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE

The chick embryo is widely used for the study of vertebrate development, but a general, reliable loss-of-function strategy for the analysis of gene function is currently not available. By using small inhibitory hairpin RNA (siRNA) molecules generated by the mouse U6 promoter, we have applied an RNA interference approach to achieve quantitative knockdown of the neuropilin-1 (Nrp-1) receptor in chick embryos. Functional knockdown was evident in the abolition of Sema3A-induced growth cone collapse in Nrp-1-siRNA but not Nrp-2-siRNA–expressing dorsal root ganglion (DRG) neurons. Two nervous system defects in Nrp-1 mutant mice were phenocopied in embryos treated with Nrp-1 siRNA. First, DRG axons prematurely entered the dorsal horn and projected inappropriately. Second, targeted early migrating neural crest cells destined for the sympathetic chain arrested ectopically within ventral spinal nerve roots. Localized knockdown induced by specific siRNA constructs will allow rapid functional analysis of genes regulating chick neural development whilst circumventing embryonic lethal effects often associated with global gene knockout in the mouse. Developmental Dynamics 230:299–308, 2004. © 2004 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE

The chick has become one of the most intensively studied vertebrate organisms in developmental biology, owing to the accessibility of its embryo to experimental manipulation and close evolutionary distance of birds to mammals. However, research on the chicken embryo has benefited less from advances in molecular biology and genetics than studies involving other vertebrates. For instance, a reliable method for targeted genetic ablation, a common strategy in mouse and zebrafish studies, currently is not available for the chick. Also, a reverse genetics approach involving knockdown with morpholino antisense nucleotides, which has proven to be very powerful in zebrafish (Nasevicius and Ekker, 2000), has, with a few exceptions (Kos et al., 2001; Granata and Quaderi, 2003; Sugiyama and Nakamura, 2003), to date not found wide application in the chick.

However, recent technical developments promise to rapidly transform this situation. First, by the adaptation of the plasmid electroporation technique to the chick embryo (Itasaki et al., 1999; Momose et al., 1999), introduction of plasmids rapidly expressing foreign cDNAs into the developing chick embryo has become routine in many laboratories, greatly facilitating gain-of-function studies (for review see Brown et al., 2003). Second, molecular resources, like expressed sequence tag (EST) databases, have increased exponentially for the chick (e.g., Boardman et al., 2002). Finally, a very important breakthrough for developmental biology research has been the discovery of the phenomenon of RNA interference, where double-stranded RNA (dsRNA) induces specific silencing of gene expression (for reviews, see Agami, 2002; Shi, 2003). When applied to the developing chick embryo, this approach has the potential to deliver a loss-of-function tool comparable to that available for the mouse and zebrafish. The low cost and great speed at which these studies could be done in the chick might favor the use of this animal over the mouse in large systematic screens of gene function. Encouragingly, recent studies indicate that RNA interference might be adopted to enable loss-of-function screens in the chick (Katahira and Nakamura, 2003; Kawakami et al., 2003; Pekarik et al., 2003).

In the present study, we adopted an RNA interference approach by using small inhibitory hairpin RNA (siRNA) molecules (Sui et al., 2002) to achieve quantitative knock down of the neuropilin-1 (Nrp-1) receptor for class 3 semaphorins in the developing chick embryo. To evaluate its efficacy and specificity in ovo, we have focussed on the development of the primary afferent projection system that connects developing dorsal root ganglion (DRG) neurons to the spinal cord. Thus, a body of evidence implicates Sema3A and its main receptor on neurons, Nrp1, in the guidance of primary sensory afferents to their synaptic targets in the central nervous system by regulating the so-called “waiting period.” First, the expression pattern of Sema3A in the dorsal spinal cord (Puschel et al., 1996; Shepherd et al., 1996) suggests it may initially confine incoming Nrp-1–expressing afferents within the dorsal funiculus, delaying the extension of afferent collateral branches until synaptic targets in the spinal grey have matured (Lee et al., 1988; Ozaki and Snider, 1997). Second, Sema3A exogenously expressed in the embryonic dorsal horn prolongs the duration of the waiting period (Fu et al., 2000; Pasterkamp et al., 2000). Third, in Sema3A null mutant mice (Behar et al., 1996), the normal pattern of sensory afferent pathfinding within the dorsal horn is disrupted. Finally, in mice expressing Nrp-1 mutated in the Semaphorin binding site, but not in the VEGF binding site, to rescue the lethal cardiovascular phenotype, premature entry and extensive misprojection of TrkA-positive afferents in the spinal grey was observed (Gu et al., 2003).

Here, by using a vector system involving the mouse U6 promoter, we demonstrate that siRNA molecules targeted at chick Nrp-1 (cNrp-1) abolish Sema3A-induced collapse of DRG growth cones. Consequently, DRG axons are insensitive to the repellent effects of Sema3A in the dorsal horn and enter the spinal grey 2 days prematurely. In addition, ectopic sympathetic neurons were located in ventral nerve roots, a phenotype previously described in Nrp-1 null mice (Kawasaki et al., 2002). These phenotypes were not observed in chick embryos targeted with control vectors, or vectors expressing cNrp-2–specific siRNAs. Thus, this approach represents a promising step toward a general loss-of-function strategy in the chick opening up the possibility of screening for unknown genes regulating amniote development.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE

Specific Knockdown of Neuropilin Expression by siRNA Generated From the U6 Promoter in Rodent and Avian Cells

We generated vectors coexpressing green fluorescent protein (GFP; from the chick β-actin promoter) and small double-stranded hairpin RNA (from the mouse U6 promoter) specifically targeted at chick Nrp-1 or -2 (see Fig. 1). Cotransfection assays with Chinese hamster ovary (CHO) involving these vectors and tagged chick neuropilins (Fig. 2A) demonstrated the efficacy and specificity of Nrp knockdown. Thus, siRNA synthesized from pCA-β-EGFPm5-siNP1 strongly reduced expression of myc-tagged cNrp-1, but did not affect expression of V5-tagged cNrp-2. Conversely, siRNA produced from both pCA-β-EGFPm5-siNP2A and -siNP2B induced robust knockdown of V5-tagged cNrp-2 levels, but left expression of myc-tagged cNrp-1 unaffected. Expression levels of tagged neuropilins were the same in cells receiving pCA-β-EGFPm5 constructs, expressing nonspecific siRNA, pCA-β-EGFPm5 alone, or no other DNA (results not shown).

thumbnail image

Figure 1. Vectors directing coexpression of small inhibitory hairpin RNA hairpins and green fluorescent protein (GFP). Location of critical sequences on pBS/U6-siNP, pCA-β-EGFPm5/U6-siNP, and RCAS(BP)-B-EGFPm5/U6-siNP vectors. The U6 promoter and downstream hairpin sequences were cut from pBS/U6-siNP and ligated into pCA-β-EGFPm5 (just upstream of the β-actin promoter) and the retroviral vector RCAS(BP)-B-EGFPm5 (between the EGFP cDNA and the 3′ LTR), as detailed in the Experimental Procedures section. The relative sizes of the different elements are not to scale. The unshaded segments contain sequences required for replication and selection of the vectors.

Download figure to PowerPoint

thumbnail image

Figure 2. Specific knockdown of chick neuropilin (cNrp) expression by small inhibitory hairpin RNAs (siRNAs) from the mouse U6 promoter in both a mammalian and avian cell background. A: In a rodent (Chinese hamster ovary) cell line (left column), siRNAs designed to target cNrp-1 prevents expression of myc-tagged cNrp-1, whereas expression of V5-tagged cNrp-2 is unaffected. Conversely, two different siRNAs targeted at cNrp-2 efficiently block expression of V5-tagged cNrp-2, but does not stop expression of myc-tagged cNrp-1 (middle and right columns). B: In an avian (quail QT6) cell line, expression of myc-tagged cNrp-1 (left column) is as efficiently blocked by cNrp-1–specific siRNA (right column) as in the CHO cells. Vectors coexpressing green fluorescent protein (GFP) and cNrp-specific siRNA (pCA-β-EGFPm5/U6-siNP-1 or -2) were cotransfected with mammalian expression vectors expressing myc-tagged cNrp-1 or V5-tagged cNrp-2. At 2 days posttransfection, cells were fixed and stained with myc or V5 antibodies.). Scale bars = 25 μm in A, 20 μm in B.

Download figure to PowerPoint

Because differences have been reported in the efficiency with which U6 genes of different species are transcribed in different species backgrounds (Simmen et al., 1992), we next tested whether the mouse U6 promoter was able to induce siRNA-mediated knockdown of cNrp-1 in avian cells. The results shown in Figure 2B indicate that when pCA-β-EGFPm5-siNP1 was cotransfected with pAG-NT-myc-cNrp-1 into quail QT6 cells, expression of the myc-tagged Nrp-1 was as efficiently reduced by the Nrp-1–specific siRNA as in the CHO cells. From this finding, we conclude that it is possible to direct synthesis of siRNA hairpins from the mouse U6 promoter in an avian background.

Functional Knockdown In Ovo of Nrp-1 in DRG Neurons

To test the efficacy and specificity of siRNA-mediated Nrp-1 knockdown in ovo, we prepared retroviral particles coexpressing cNrp-specific siRNA and GFP with the aim to target dorsal root ganglion (DRG) neurons. DRG neurons express Nrp-1 and respond to Sema3A by collapse of their growth cones and growth arrest and repulsion of their axons (Luo et al., 1993; Eickholt et al., 1997; Kolodkin et al., 1997). The 2- to 2.5-day-old chick embryos were injected in the lumbar neural tube with RCAS virus particles, containing the coding sequences for GFP and U6-hairpin cassettes, generating cNrp-1–specific siRNA hairpins. As controls, embryos were injected with virus expressing GFP alone, or GFP and cNrp-2–specific siRNA (directed at site B). Figure 3A shows robust expression of GFP in a 7-day-old embryo (i.e., 5 days after injection with RCAS-EGFPm5-siNP1 virus). GFP labeling is particularly prominent in the dorsal root ganglia. Ganglia were dissected, cultured overnight, and exposed to increasing doses of Sema3A-Fc. Strikingly, as illustrated in Figure 3B, GFP-labeled growth cones in DRG cultures obtained from embryos infected with cNrp-1–specific siRNA expressing virus were fully intact at Sema3A concentrations, inducing extensive growth cone collapse in control DRG cultures. Figure 3C shows the quantitative analysis of several independent growth cone collapse assays on GFP-labeled neurites. The data indicate that the growth cones of DRG neurons that were infected with the virus expressing cNrp-1–specific siRNA were largely refractory to Sema3A-induced collapse. Growth cones in cultures of DRG infected with cNrp-2–specific siRNA-expressing virus collapsed normally, like growth cones from ganglia expressing only GFP (or uninfected ganglia; not shown). Growth cones from uninfected neurons (i.e., those not expressing GFP) in cNrp-1 siRNA-treated cultures collapsed normally (not shown). This finding emphasizes the importance of coexpression of the GFP marker with the siRNA, allowing the identification and analysis of the siRNA-targeted cells in a wild-type background. Because the Sema3A response in the GFP-expressing growth cones from the embryos targeted with cNrp-1 siRNA-expressing virus was almost reduced to control levels these data also indicate that in these embryos all DRG neurons that express GFP also will express the specific siRNA hairpins.

thumbnail image

Figure 3. Inhibition of Sema3a induced growth cone collapse after in ovo infection of chick embryo dorsal root ganglion (DRG) neurons with chick neuropilin-1 (cNrp-1) small inhibitory hairpin RNA (siRNA) -expressing retrovirus. A: Replication-competent chick retroviral (RCAS) particles, expressing green fluorescent protein (GFP) and cNrp-specific siRNA were injected into the neural tube of 2- to 2.5-day-old chick embryos. Five days later, green fluorescent protein (GFP) -labeled DRG (white arrows indicate position of GFP-labeled DRGs) were dissected from infected embryos and cultured as explants. After overnight culture, the explants were exposed for 30 min to increasing volumes of supernatant taken from cells secreting Sema3A-Fc fusion protein. B: Growth cones in DRG explant cultures derived from embryos treated with virus expressing cNrp-1–specific siRNA are refractory to Sema3A-induced collapse (left panel). By contrast, growth cones in cultures derived from embryos that received the control GFP virus collapse normally on exposure to Sema3A (right panel). C: Quantitative analysis of collapse assays performed on DRG explant cultures derived from embryos infected with RCAS virus expressing GFP only (GFP control), GFP and cNrp-1–specific siRNA (Nrp-1 siRNA), or GFP and cNrp-2–specific siRNA (Nrp-2 siRNA). Only growth cones expressing GFP were included in the analyses. The data are the mean of three different experiments (error bars indicate SEM). DRG neuron growth cones of embryos treated with virus expressing cNrp-1–specific siRNA fail to collapse in the presence of the applied concentrations of Sema3A-Fc. By contrast, DRG neuron growth cones expressing cNrp-2–specific siRNA or GFP alone displayed a typical dose-dependent sensitivity to Sema3A-Fc. Scale bars = 200 μm in A, 20 μm in B.

Download figure to PowerPoint

Premature Entry and Misprojection of DRG Axons Into the Spinal Cord After Nrp-1 Knockdown

When primary sensory afferents enter the spinal cord, they initially bifurcate and extend rostrocaudally within the dorsal funiculus for several days, the “waiting period,” before their collaterals branch into the spinal grey. When we introduced Nrp-1-specific siRNA into the neural tube of chick embryos, either by using retroviral particles or plasmid electroporation, we did not notice conspicuous effects on survival beyond embryonic day (E) 10, when most sensory afferents have reached their spinal cord targets in untreated embryos.

When we analyzed the central sensory afferent projections in embryos treated with Nrp-1 siRNA-expressing retrovirus, we consistently found that a small proportion of axons entered the spinal grey prematurely (n = 4, not shown). However, this was much more pronounced in embryos in which plasmids expressing Nrp-1–specific siRNA were delivered by electroporation. As early as Hamburger and Hamilton stage 28 (E5.5, i.e., 2 days before sensory afferents first penetrate the spinal grey in normal embryos), we observed many sensory afferents precociously entering the dorsal horn of embryos treated with Nrp-1 siRNA (Fig. 4; n = 5). The extent of premature entry is most apparent in horizontal sections taken at the level of the dorsal funiculus (Fig. 4B,C). Many neurofilament (NF)-positive afferents project toward the central canal and then continue to extend rostrocaudally along the inner wall of the lumen. The projection pattern revealed in transverse sections (Fig. 4G,H) suggests that large numbers of afferents continued to extend into the dorsal grey in the same orientation they adopted on entering spinal cord territory at the dorsal root entry zone. This finding suggests that misprojected axons were insensitive to an early repellent signal and possibly as a result of failing to initially bifurcate within the dorsal funiculus, extended unimpeded. No premature entry was seen at the contralateral site of the spinal cord of embryos electroporated with Nrp-1 siRNA-expressing plasmid (Fig. 4B,C) or in embryos electroporated with plasmids expressing cNrp-2–specific siRNA (directed against either site A [n = 3] or B [n = 5]; Fig. 4E,J) or GFP alone (n = 6; Fig. 4D,I).

thumbnail image

Figure 4. Expression of chick neuropilin-1 (cNrp-1)-specific small inhibitory hairpin RNA (siRNA) in dorsal root ganglia (DRGs) leads to premature entry of primary sensory afferents in the spinal cord and misprojection toward the lumen. A–E: Confocal photomicrographs of longitudinal Vibratome sections (flattened Z-series of 30 images taken at 1-μm intervals) of the lumbar spinal cord at the level of the dorsal root entry zone (DREZ; as illustrated in A) from embryonic day (E) 6 chicks electroporated in the neural tube at E2.5 with plasmid vectors coexpressing cNrp-1 or -2 (site A) -specific siRNA and green fluorescent protein (GFP), or GFP alone. Embryos were analyzed 3 days after electroporation (Hamburger and Hamilton stage 28). In all panels, the electroporated side is on the right, as indicated by the GFP expression (green) in B. In sections of embryos treated with Nrp-1 siRNA (B,C), neurofilament (NF) immunostaining (red) shows that large numbers of primary afferents prematurely enter the spinal grey on the electroporated but not on the contralateral side. Most of these axons appear to project straight toward the midline (white arrows), and some then make a right angle turn along the luminal inner surface. D,E: No axons can be seen entering the spinal grey at the electroporated side of embryos expressing GFP alone (D) or GFP and cNrp-2–specific siRNA (E). F–J: Confocal photomicrographs of transverse Vibratome sections (flattened Z series of 30 images taken at 1-μm intervals). Only the electroporated halves are shown, as illustrated in F. In the sections obtained from embryos treated with cNrp-1–specific siRNA G,H: large numbers of DRG axons are seen to have prematurely entered the spinal grey at the DREZ and subsequently project toward the lumen (white arrows). I,J: In contrast, the afferent projections of embryos electroporated with control GFP plasmid (I) or plasmid expressing cNrp-2–specific siRNA (J) remain confined exclusively within the dorsal funiculus (DF) at this stage. DRG, dorsal root ganglion; DR, dorsal roots; DF, dorsal funiculus. Scale bars = 200 μm in B–E; 100 μm in G–J.

Download figure to PowerPoint

Immunostaining with TrkA and TrkC antibodies, specific markers for nociceptive and proprioceptive sensory afferents, respectively, indicated that the central projections of both subpopulations of afferents were equally affected in embryos treated with cNrp-1–specific siRNA (Fig. 5), both entering the spinal grey prematurely.

thumbnail image

Figure 5. Chick neuropilin-1 (cNrp-1)-specific gene knockdown leads to premature entry and misprojection of both nociceptive and proprioceptive sensory afferents in the spinal cord. A–F: Confocal photomicrographs of transverse Vibratome sections at the level of the dorsal root entry zone of Hamburger and Hamilton stage 28 (embryonic day [E] 5.5) embryo spinal cords, electroporated with plasmid expressing green fluorescent protein (GFP) and cNrp-1–specific small inhibitory hairpin RNA (siRNA) at E2.5 (see legend to Fig. 4). Sections were double immunostained with antibodies against neurofilament (NF; A,D; blue label) and TrkA (A,C; red label in A, white in C), marking nociceptive afferents or TrkC (D,F; red label in D, while in F) marking proprioceptive afferents. Images show the dorsal spinal cord plus incoming dorsal roots (DR) at the electroporated side of the embryo. The anti-Trk staining indicates that central projections from both major subclasses of dorsal root ganglia afferents are affected by Nrp-1 knockdown and enter the dorsal spinal grey prematurely and, in addition, project inappropriately within the dorsal horn. DH, dorsal horn; DF, dorsal funiculus. Scale bars = 100 μm.

Download figure to PowerPoint

From this finding, we conclude that the axon guidance defects we describe are caused by knockdown of cNrp-1 expression by cNrp-1–specific siRNA and constitute a phenocopy of one of the observed effects of genetic ablation of Nrp-1 function in the mouse (Gu et al., 2003).

Neural Crest Cells Targeted With Nrp-1–Specific siRNA Fail to Incorporate Into Sympathetic Ganglia

In addition to a central role in many axon guidance events Nrp-1 and Sema3A have also been implicated in regulating cell migration in the nervous system, including that of sympathetic neuroblast precursors during the assembly of the sympathetic chain (Kawasaki et al., 2002). When we analyzed the sympathetic ganglia of embryos treated with Nrp-1–specific siRNA, no gross abnormalities in sympathetic chain organization were immediately apparent. However, we reasoned that, if only a small proportion of the neural crest precursors that contribute progeny to the sympathetic chain were targeted by in ovo-delivered vectors, the majority of the cells that are not targeted will go on to form normal sympathetic ganglia. This process was confirmed when we examined in detail the incorporation of GFP-labeled cells in sympathetic chain structures after electroporation with GFP-expressing plasmid. Sympathetic neurons were identified by staining with antibodies against tyrosine hydroxylase (TH) and islet-1/2. Figure 6A,D shows that a relatively small proportion of sympathetic neurons express the GFP label in electroporated embryos (n = 3). The same was seen in embryos electroporated with plasmid expressing GFP and Nrp-2–specific siRNA (n = 3; Fig. 6B,E). However, in embryos electroporated with plasmid expressing Nrp-1–specific siRNA, no GFP-positive neurons were detected within sympathetic ganglia (n = 4; Fig. 6C,F), although in the same embryos, many GFP-positive cells were seen in the DRG (also derived from neural crest; Fig. 6C,G). Intriguingly, GFP-positive, ectopic neurons were observed appearing to migrate away from the sympathetic chain down the ventral root (Fig. 6F, insets).

thumbnail image

Figure 6. Tyrosine hydroxylase (TH) -expressing sympathetic neuron precursors targeted in ovo with chick neuropilin-1 (cNrp-1)-specific small inhibitory hairpin RNA (siRNA) fail to integrate into the sympathetic chain. A–G: Confocal photomicrographs of Vibratome sections of Hamburger and Hamilton stage 28 (embryonic day [E] 5.5) embryo lumbar spinal cord, electroporated at E2.5 with control green fluorescent protein (GFP), cNrp-1, or cNrp-2 siRNA plasmids (see legend to Fig. 4), cut in the transverse (A–C) or longitudinal (D–G) plane, double immunostained with antibodies to Islet1-2 (red), or tyrosine hydroxylase (TH, blue). Sympathetic neurons expressing both Islet1-2 and TH appear purple. Insets show higher magnifications of the sympathetic ganglia boxed areas within the same panels. For reasons of clarity in the insets, only the islet1-2 (red) and GFP (green) staining is shown. Delaminating neural crest cells electroporated at E2.5 with a plasmid expressing cNrp-1–specific siRNA, as indicated by the GFP label, fail to incorporate at all into the sympathetic chain (see insets in C,F,G) compared with those electroporated with GFP plasmid (insets in E, D) or cNrp-2–specific siRNA (insets in B,E). Note that significant numbers of the misrouted sympathetic neuroblasts targeted with cNrp-1–specific siRNA become ectopically positioned in ventral spinal nerve roots (F, right inset). By contrast primary sensory neuronal precursors within delaminating neural crest targeted with cNrp-1–specific siRNA do become successfully integrated into the DRG (low-power views in C,G). NB: The section shown in G was taken from the same embryo as the section shown in F, but at a more dorsal level. Scale bars = 200 μm in A (applies to A–C), in D (applies to D–G).

Download figure to PowerPoint

These results, therefore, represent another chick siRNA phenocopy of genetic ablation of Nrp-1 in the mouse, this time involving the perturbed migration of neuronal soma, rather than their axons. Furthermore, these data emphasize the utility of the GFP marker in detecting a mutant phenotype in a wild-type background.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE

Vectors Generating GFP and siRNA Hairpins Are Promising Tools for In Ovo Gene Silencing Studies

By using vector systems generating small hairpin RNA molecules and the GFP marker, we have knocked down expression of Nrp-1 in the peripheral nervous system of the developing chick embryo. GFP-expressing DRG growth cones from embryos targeted with cNrp-1–specific siRNA hairpins resisted Sema3A-induced collapse. Consequently, Sema3A in the dorsal horn did not arrest incoming sensory afferents, and these axons entered the spinal grey prematurely. Furthermore, delaminating neural crest cells targeted with cNrp-1–specific siRNA failed to incorporate into the sympathetic chain. Thus, by using siRNA in the chick, we have mimicked two well-documented consequences of genetic ablation of Nrp-1 function in the mouse, involving straying axons, as well as misrouted neuronal somata (Kawasaki et al., 2002; Gu et al., 2003). These results may serve as a guide as to how to use an RNAi approach to investigate axon guidance and gangliogenesis in the developing chick embryo.

A vector-based delivery system for inhibitory RNA molecules, involving plasmid electroporation and retroviral particles, has the potential to enable gene silencing during the entire period of in ovo development and possibly even after hatching. In this respect, it would be superior to an approach published by Pekarik et al. (2003), involving electroporation of in vitro transcribed long dsRNA or approaches involving synthetic siRNA or morpholino antisense molecules. The electroporated dsRNA molecules or morpholinos are likely diluted rapidly after each cell division in the growing embryo, which would guarantee only a short-lived, transient inhibitory effect.

Retroviral Delivery Vs. Plasmid Electroporation

We have expressed GFP and siRNA in chick embryos from retroviral particles or plasmid vectors, the latter introduced by electroporation. These two methods for delivery of the inhibitory RNA provide a large time window for gene silencing during chick development. Retroviral delivery induces persistent expression, but with a slow onset. Electroporated plasmids induce rapid but transient expression. We used retroviral delivery to maximize the number of targeted DRG neurons, used for the collapse assays (Fig. 3). The DRG central projection phenotype (Figs. 4, 5) and the sympathetic phenotype (Fig. 6) likely require reduction of Nrp-1 expression relatively early in development (E3–E5). For this reason, these phenotypes were more robust when the siRNA was introduced by plasmid electroporation.

Another advantage of plasmid electroporation is that the site and timing of delivery can be fairly accurately targeted. The locally confined knockdown of Nrp-1 that we induced allowed the embryos to survive and gave us the opportunity to observe a phenotype that cannot be seen in Nrp-1 null mice, because of early embryonic lethality (Kituskawa et al., 1997; Gu et al., 2003). However, we were unable to target neural crest and its progeny without also targeting the dorsal horn. Another layer of specificity in the targeting of siRNA-mediated knockdown would be required to investigate for instance the role of Nrp-1 in dorsal horn neurons, as opposed to its function on the growth cones of incoming DRG axons. This investigation might be achieved using a conditionally active “floxed” U6-hairpin construct, activated by Cre under control of a tissue-specific promoter.

The retroviral vectors we used to deliver the inhibitory siRNA molecules were replication-competent. Although expression, as indicated by the GFP marker, was initially limited to cells derived from delaminating neural crest and the lining of the central canal, at later stages (E7 and beyond) because of secondary infections, it became widespread in proliferating cells, particularly those of the developing myotome. Whereas this may be irrelevant or even desirable for some research purposes, a more limited, controlled expression pattern could be achieved using replication-deficient retrovirus. These vectors would have the additional advantage that they could accommodate the tissue-specific regulatory elements alluded to above, enabling a range of refined expression patterns.

Nontarget RNAi Effects

Although initially only applicable in plants and organisms such as Drosophila and Caenorhabditis elegans, it was found recently that the interferon-mediated nonspecific block in translation, normally triggered when double-stranded RNA (dsRNA) is introduced in mammalian cells, could be avoided if the sequences used were small (∼21 bp; Elbashir et al., 2001).

It is not known whether interferon-mediated nonspecific silencing occurs in developing chick embryos as a response to dsRNA. That it was not reported by Pekarik et al. (2003) suggests it may not happen, as in undifferentiated mouse embryonic stem cells (Yang et al., 2001) and differentiated mouse neuroblastoma cells (Gan et al., 2002). That the methods we describe here involve small dsRNA molecules should not make investigators, using them in functional screens of unknown genes, complacent about nontarget effects. In contrast to earlier microarray studies (Chi et al., 2003; Semizarov et al., 2003) that showed highly specific gene silencing by siRNA molecules, later studies have demonstrated that some short hairpin RNA molecules may still activate the interferon pathway in mammalian cells (Bridge et al., 2003; Sledz et al., 2003). However, this does not lead to global nonspecific silencing as observed with long dsRNA. Nonetheless, the fact that a subset of siRNA molecules might alter the overall expression profile in a nonspecific manner means that a novel phenotype of an unknown gene cannot be accepted on the evidence of induction by a single siRNA sequence alone. The specificity of a novel phenotype can be corroborated up by induction by at least two different siRNAs, targeting different sites on the same mRNA, and lack of induction by nontarget-related siRNAs. Recently developed polymerase chain reaction (PCR)-based methods for the rapid generation and selection of RNA polymerase III hairpin cassettes (Castanotto et al., 2002; Gou et al., 2003) should greatly facilitate these control experiments.

An effective, specific method for silencing gene expression is a formidable asset to the already impressive arsenal of techniques available for the study of developmental processes in the chick. Such a method will provide a cheap, rapid alternative to mouse knockout studies for a functional screen of unknown genes in vertebrate development. The methods we present here constitute a major step toward this goal.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE

Molecular Biology

Enzymes for molecular biology were purchased from Roche or New England Biolabs. Oligonucleotides were obtained from MWG. For knockdown of cNrp-1, a vector generating double-stranded hairpin RNAs specific for cNrp-1 from the mouse U6 promoter was constructed in two steps. The hairpin RNAs were designed to target the sequence gggccaattcaagaccacaca, corresponding to nucleotides 2080–2100 of cNrp-1 cDNA. First, oligonucleotides ggccaattcaagaccacacaa and agctttgtgtggtcttgaattggcc were annealed and cloned into pBS/U6 (Sui et al., 2002; a gift from Dr Yang Shi), after cutting the vector with ApaI and blunting the ends with T4 polymerase followed by digestion with HindIII. Next, oligonucleotides agctttgtgtggtcttgaattggcccttttt and aaaaagggccaattcaagaccacacaa were annealed and ligated into the product of the previous cloning step after digestion of the vector with HindIII and SmaI. The resulting construct is named pBS/U6-siNP1. Two hairpin vectors targeting chick Nrp-2 were constructed. One targets the sequence gggagactgcaagtatgactg, corresponding to nucleotides 615–636 of cNrp-2 cDNA (site A). The other targets the sequence gggcaactctgaaccaagtcc (nucleotides 2022–2042; site B). These constructs were prepared in one cloning step. To prepare pBS/U6-siNP2A and pBS/U6-siNP2B, oligonucleotides ggcaactctgaaccaagtccctcgagggacttggttcagagttgccctttttg and aattcaaaaagggcaactctgaaccaagtccctcgagggacttggttcagagttgcc (site A) and ggagactgcaagtatgactgctcgagcagtcatacttgcagtctccctttttg and aattcaaaaagggagactgcaagtatgactgctcgagcagtcatacttgcagtctcc (site B) were annealed. pBS/U6 was digested with ApaI, blunt ended with T4 polymerase and then digested with EcoRI, before ligating the annealed oligonucleotides. Inserts were confirmed by sequence analysis. BLAST searches of the targeted cNrp-1 and -2 sequences against general and chick EST databases (Boardman et al., 2002), revealed no significant homology to cDNAs other than cNrp1 and -2, respectively.

To prepare GFP expression plasmids, coexpressing siRNA hairpins, the U6 promoter and hairpin sequences were inserted into pCA-β-EGFPm5 (Yaneza et al., 2002; Vermeren et al., 2003; a gift of Dr Jon Gilthorpe), in the unique SalI site just upstream of the chick β-actin promoter. EGFPm5 contains the additional mutations V163A/I167T/S175G (Siemering et al., 1996) in the mammalian-optimized codon sequence of EGFP (F64L/S65T) and is several-fold brighter than EGFP. pBS/U6-siNP constructs were digested with BamHI, releasing fragments approximately 400-bp long, containing the U6 and hairpin sequences. These fragments were blunted and ligated into pCA-β-EGFPm5 after SalI digestion and blunt ending of the vector. The resulting constructs are called pCA-β-EGFPm5-siNP1, -siNP2A, or -siNP2B.

To prepare chick retroviral vectors coexpressing GFP and siRNA hairpins, the pBS/U6-siNP constructs were digested with SpeI and the resulting ∼400-bp U6-hairpin fragments were inserted in the unique SpeI site of RCAS(BP)B-EGFPm5 (Petropoulos and Hughes, 1991; a gift of Dr Jon Gilthorpe), just upstream of the 3′ LTR.

A full-length chick Nrp-2, tagged at the C-terminus with the V5 peptide, was constructed by first amplifying a 3′ fragment of chick Nrp-2 (cloned in pCR2.1, a gift from Dr. Gera Neufeld, Technion, Haifa, Israel) with high fidelity KOD Hot Start Polymerase (Novagen) using the primers gtcaaaggcgtcattatccag (corresponding to nucleotides 1492–1513 of the Nrp-2 cDNA) and tgcctcggagcagcacttctg (complementary to nucleotides 2737–2757) and cloning it into pCDNA3.1/V5-His-Topo (Invitrogen). A 5′ fragment of Nrp-2 was amplified by using cDNA corresponding to the cNrp-2B splice variant (a gift from Dr. Yuji Watanabe, Tohoku University, Sendai, Japan) as a template and primers cgcgaaggatggatacgtttc (corresponding to the 5′ UTR of cNrp-2) and tgttgccttcaaagagcttgg (complimentary to nucleotides 1637–1657). The 5′ Nrp-2 fragment was linked to the 3′ fragment in pDNA3.1/V5 by using a unique EcoRI site in the cNrp-2 cDNA. First, the V5-tagged 3′ construct was digested with HindIII and the ends were blunted with T4 polymerase. Then, both the 5′ PCR fragment and the construct containing the V5-tagged 3′ fragment were digested with EcoRI and joined by ligation. pAG-NT-myc-cNrp-1–expressing myc-tagged cNrp-1 (Renzi et al., 1999) was a gift from Dr. Jonathan Raper (University of Pennsylvania, Philadelphia).

Virus and Cell Culture

Primary chick embryo fibroblasts (CEF) and quail QT6 cells were grown in DMEM, supplemented with 10% foetal calf serum (FCS) and 2% chick serum. CHO cells were grown in DMEM, supplemented with 5% FCS. Retroviral particles were prepared by transfection of RCAS plasmids into CEF. Briefly, 8 μg of DNA was mixed with 24 μl of Fugene (Roche) and added to a 20–40% confluent 90-mm dish of CEF. Cells were split after 2 days and transferred to four 90-mm dishes. Virus containing supernatants were collected from 3–5 days after transfection. Viral particles were concentrated by ultracentrifugation for 2 hr at 20,000 rpm in a SW28 rotor, resuspended in culture medium, and stored at −80°C. Virus titres were determined on CEF by quantifying the number of GFP-expressing cells. Titres ranged from 1–5 × 108 infectious units/ml, for control virus, expressing only GFP, as well as for virus expressing both GFP and siRNA. DRG growth cone collapse assays were performed essentially as described previously (Eickholt et al., 2002).

siRNA Transfection Into Cultured Cells

Plasmids coexpressing GFP and cNRP-specific siRNA (pCA-β-EGFP/U6-siNP-1 or -2) were premixed with neuropilin expression plasmids at a ratio of 10:1 and transfected into CHO or QT6 cells, grown on eight-well LabTek chamber slides (Nalge Nunc International), using Fugene (Roche), according to the manufacturer's instructions. Forty-eight hours after transfection, cells were fixed with 4% paraformaldehyde and stained with rabbit anti–c-myc (Santa Cruz) or mouse anti-V5 (Sigma), followed by Cy3-conjugated secondary antibodies (Jackson Laboratories). Slides were sealed with glass coverslips by using Mowiol (Calbiochem; Heimer and Taylor, 1974). Images were recorded at constant exposure times on a Zeiss Axioskop Fluorescence microscope, connected to a cooled CCD Spot camera (Diagnostic Instruments).

Virus Injection and Embryo Electroporation

Before injection into embryos, Polybrene (hexadimethrine bromide, Sigma) was added to the virus suspension to 10 μg/ml in addition to Fast Green FCF at 0.2%. Eggs were windowed and bathed in Ringer's solution. Virus was injected in the lumen of the neural tube of HH stage 15–18 embryos at the lumbar level by using a picospritzer (WPI). For the younger embryos, a solution of 5% India ink (Pelikan Fount) was injected under the blastodisc to help visualize the neural tube.

Plasmid DNA for electroporation studies was purified by equilibrium centrifugation on two consecutive cesium chloride gradients. DNA was resuspended in TE buffer at 2 mg/ml. A total of 0.2% Fast Green FCF was added to aid monitoring injection into the neural tube. DNA was injected in the neural tube, and 0.25-mm silver electrodes were positioned at lumbar level, perpendicularly to the axis of the embryo. Five square pulses of 25 V were administered by using an Intracept TSS10 pulser (IntraCell).

Immunohistochemistry and Microscopy

Embryos were fixed in 4% paraformaldehyde, embedded in 3% agar, and sectioned by using a Leica VT1000S Vibratome. Seventy-five-micrometer-thick sections were collected in 24-well culture dishes (Nunc) and blocked for 1 hr in antibody blocking buffer (phosphate buffered saline/10% sheep serum/0.1%Triton X-100). Sections were incubated overnight with the indicated antibodies diluted in antibody blocking buffer. The following antibodies were used: mouse monoclonal anti-islet1,2 (39.5D5; developed by Dr. Thomas Jessell was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242), rabbit anti–cTrk-A and anti–cTrk-C (a gift from Dr. Frances Lefcort, Montana State University, Bozeman, MT), rabbit anti-neurofilament (NF1991; Chemicon), mouse anti-neurofilament (RMO; Zymed), rabbit anti-tyrosine hydroxylase (Pel Freez Biologicals). Sections were washed extensively before overnight incubation with secondary antibodies (Jackson ImmunoResearch and Molecular Probes) diluted in antibody blocking buffer. After extensive washing, sections were transferred to glass slides and sealed under glass coverslips that had been coated with Mowiol. Sections were scanned on an Olympus Fluoview confocal laser scanning microscope connected to an Olympus AX70 research microscope. Images were processed by using the software supplied by the manufacturer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE

We thank Yang Shi for the pBS/U6 plasmid, Jon Raper for the pAG-NT-myc-cNrp-1 plasmid, Jon Gilthorpe for the pCA-β-EGFPm5 and RCAS(BP)-B-EGFPm5 plasmids, Gera Neufeld and Yuji Watanabe for chick Nrp-2 cDNAs, Frances Lefcort and Louis Reichardt for chick-specific Trk antibodies, Thomas Jessell for the islet-1,2 antibody. We also thank Anthony Graham and Andrew Lumsden for comments on the manuscript.

REFERENCE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCE