Electroporation chambers enable reproducible and efficient electroporation
Efficient and reproducible electroporation relies primarily on the precision of the injection of DNA. To control injection accuracy, stage 21–35/36 embryos must be held in the desired position and submerged in a drop of medium, as they easily deform and are highly sensitive to drying. Previously published electroporation procedures could not be used because they do not permit accurate orientation of the embryos [38
] nor take into account the soft-tissue vulnerability or morphology of the targeted stages [12
]. Therefore, we developed electroporation chambers tailored individually to the size and morphology of embryos from stages 21 to 35/36. The basic design of the chambers consists of two channels carved perpendicular to one another in Sylgard in the shape of a cross (Figure ). The embryo is held in the longitudinal channel while the electrodes are placed in the transverse channel. The size and geometry of the longitudinal channel was optimized for each embryonic stage to provide a "snug fit" for the embryo and full immersion in medium. The position of the transverse channel insures reproducible placement of the electrodes along the anterior-posterior axis of the embryo and its depth controls the amount of electrode surface in contact with the medium, and thus the dorso-ventral extent of the embryo exposed to the electric field. The length of the transverse channel is designed so that when electrodes are placed at each end, electroporation efficiency is maximized while damage to the embryo is minimized. In addition, the spacing and immobilization of the electrodes in the transverse channel enable accurate positioning prior to injection. This allows the electric field to be applied immediately after DNA injection which is critical for minimizing diffusion and backflow of the injected solution through the opening made by the capillary [19
Figure 1 Efficient DNA transfection of stage 26–28 Xenopus embryos. a: Schematic representation of the experimental setup. Embryos were placed in the main channel of the electroporation chamber, while the electrode tips (0.5 mm wide) were positioned in (more ...)
Electroporation leads to efficient transfection in Xenopus
To determine the optimal conditions for Xenopus
electroporation, voltage, frequency and duration of electrical pulses were systematically varied using the experimental set-up illustrated in Figure (embryos were kept at 18°C). 93% of the embryos (n = 34) injected at stage 26–28 into the third ventricle with a solution containing 1 μg/μl of green fluorescent protein (GFP) encoding plasmids exhibited bright GFP expression 12 h after being exposed to 8 square-pulses of 20 V 50 ms applied every second (20 V/50 ms/1 s/8 x) (Figure ). Efficient transfection required a high conductivity electroporation medium as the success rate dropped 2.5-fold (n = 26) when 0.1× M
aline (MBS) was used instead of 1× MBS or 1× M
inger's (MMR) (Figure and ). The high conductance of the medium surrounding the low conducting embryo could enhance electroporation by preventing the decrease of electric field inside the embryo as shown on cellular spheroids [47
]. As summarized in Figure (blue), a series of 4 or more 15–20 V pulses with a duration of 25–100 ms each led to >60% electroporation success rate. Electroporation efficiency consistently increased in proportion to pulse number, voltage and duration (when below 200 ms). A decrease in voltage or pulse duration could be partially compensated for by increasing the number of pulses.
To further characterize the efficiency of electroporation, nucleus-targeted GFP (nls-GFP) was transfected to quantify the fraction of GFP-expressing versus non-expressing cells on transverse brain sections counterstained with a nuclear stain (DAPI). 48 h after electroporation (20 V/50 ms/1 s/8 x), the average fraction of cells expressing GFP per section was 47.1 ± 2.5% in the transfected region (n = 47 sections 6 embryos; Figure and ). Transfected cells were scattered along 50–70% of the dorso-ventral axis and throughout the whole neuroepithelium. At this stage, the brain comprises a proliferative region adjacent to the ventricle lumen (ventricular zone) surrounded by layers of migrating and differentiating neurons (mantle zone). 48 h post-electroporation, transfected cells were present in both regions. GFP-expressing cells residing in the superficial third of the brain, populated by differentiated neurons, represented 31.8 ± 1.5% of total labeled cells (n = 32). However, the fraction of transfected cells in the superficial half of the brain increased with pulse duration (see additional file 1
a & b). To check the morphology of transfected cells and further characterize their cell types, we transfected membrane-targeted GFP or RFP (GAP-GFP and -RFP). As expected, the GFP signal was found from the ventricle to the neuropil (Figure ). Cells lining the ventricle could be seen extending radial process towards the pia, typical of dividing cells (Figure ). Transfected neurons appeared to differentiate normally as they expressed the neuronal marker acetylated tubulin, and sent long processes into the neuropil (Figure ). Furthermore, several axon tracts could be recognized in a whole-mount view of the brain (Figure ).
Figure 2 Cell types and morphology of the transfected cells. a: Membrane-tethered GFP (GAP-GFP) delineated the processes of transfected neurons including the axons (the ventricle and neuropil are outlined in white). The arrow indicates a bundle of axons travelling (more ...)
Finally, we assayed the potential adverse side effects of electroporation. Electroporation did not increase either the embryo death rate or the occurrence of morphological abnormalities, provided the pulse voltage remained under 25 V and the pulse duration under 100 ms (Figure red). The anatomy of the embryos and their brains appeared normal on transverse sections at all time points after electroporation tested (Figure ). Some pyknotic nuclei were observed in highly transfected embryos in the first 24 h post-electroporation. Therefore, TUNEL staining was used to assess cell death on sections. 24 h after exposure to 20 V/50 ms/1 s/8 x and 18 V/25 ms/1 s/10 x, the average number of TUNEL positive cells per section was 5.21 ± 0.35 (n = 48) and 2.8 ± 0.48 (n = 28) respectively (see additional file 1
g). This compares favorably with an average of 3.5 ± 0.27 cells/section (n = 63) in control embryos and indicates that the electric pulses are relatively harmless per se
. However, other parameters such as DNA purity, embryo quality, manipulation and injection are critical for minimizing cell death. Thus, electroporation of ventricular injected DNA led to efficient transfection of both the dividing ventricular region and differentiated neurons without increasing cell death or affecting their morphological differentiation.
Electroporation at different stages produces rapid and long-lasting transgene expression
Neuronal differentiation and initial establishment of the major axonal projections progress rapidly from stage 20 to 40 and, during this period, the neuroepithelium undergoes major reorganization with post-mitotic neurons migrating away from the ventricular surface to the superficial layers. To achieve fine temporal resolution, the electroporation procedure should be similarly efficient across different time-windows. Therefore, we compared the efficiency of the electroporation protocol over different stages.
Embryos were electroporated following intraventricular pCS2GFP-DNA injection at stages ranging from 21 to 35/36 in chambers specially adapted to their morphology. External inspection of embryos under the fluorescent strereomicroscope showed that the fraction of embryos exhibiting bright GFP expression 12 h after electroporation was >70%, regardless of the stage at which the electroporation was performed. Despite these similar levels at a gross level, analysis of the GFP positive cell fraction (nls-GFP) on transverse sections revealed that a sharp decrease in efficiency occurs at stage 32 (Figure ). In addition, the GFP-expressing cells from late (stage 32) electroporations were distributed unevenly in the neuroepithelium. First, there was a marked decline in the number of transfected cells in the ventral brain. Indeed, the dorsal shift of the GFP-expression center of mass (relative to the DAPI nuclear marker) significantly increased 1.5-fold between stages 28 and 32. Secondly, 48 h after electroporation, fewer GFP-positive cells can be found in the superficial region of the neuroepithelium closest to the pia (Figure ). Several factors may contribute to the observed changes. As the brain develops, the ventricle lumen expands and post-mitotic cells migrate away from the ventricular surface to differentiate in superficial layers. Consequently, many cells are distant to the injection site, making them less likely to be transfected. In addition, a larger ventricle means a lower intraventricular concentration of injected DNA, which will restrict the transfection to cells lining the ventricle and decrease the electroporation efficiency overall. In agreement with this, doubling the injection volume enhances the electroporation success rate by 1.25 (assessed as in figure ; n = 12). However, if the local DNA concentration was the only factor involved, a stage-dependent distribution of GFP positive cells would be expected shortly after electroporation. In fact, 12 h after electroporation the fraction of GFP positive cells located in the superficial half of the brain was similar in embryos electroporated at stage 28 and 32 (24.4 ± 1.6, n = 36 and 21.9 ± 2.1, n = 20 sections respectively). The delayed onset of the stage-dependent difference in deep-superficial distribution suggests that it results at least partly from developmental changes in patterns of cell proliferation and migration that occur after electroporation.
Figure 3 Electroporation of stage 21–35/36 embryos leads to rapid expression of transgenes. a: Electroporation efficiency decreased with increasing embryonic stage. Percentages of nls-GFP positive cells 12 h after transfection at stage 26, 28 or 32 (n (more ...)
In order to gain access to the cells situated close to the pia, we delivered DNA to the pial surface by injecting under the skin epidermis instead of intraventricularly (stage 29/30 embryos). Subcutaneous injections, followed immediately by electroporation, efficiently and selectively transfected cells in superficial layers of the brain (Figure ).
Overall, GFP expression in embryos electroporated between stages 21 to 35/36 displayed similar kinetics. In whole embryos, the GFP signal can first be detected 5–6 h after electroporation. This signal progressively intensifies and spreads over the subsequent 36 h, and remains high for several days. Nuclear GFP was used to quantify the GFP expression at different time points on transverse sections of embryos electroporated at stage 29/30. A progressive increase in both the fraction and the average intensity of the GFP positive cells was observed between 6 and 48 h post-electroporation (Figure ). The sigmoid shape of GFP kinetics likely reflects the requirement for a progressive accumulation of the GFP signal in the transfected cells to reach the detection threshold, combined with proliferation of transfected progenitors. Interestingly, 6 h after intraventricular injection/electroporation (20 V/50 ms/1 s/8 x) of stage 29/30 embryos, 36.7 ± 3.2% of the GFP positive cells were found in the superficial half of the brain, suggesting that electroporation efficiently targets both proliferating and differentiated cells. The wide range of stages amenable to electroporation, combined with the quick onset of transgene expression, demonstrates that this technique provides precise temporal control.
Controlling spatial targeting to study axon guidance: the retino-tectal projection
The spatial selectivity allowed by electroporation has proven useful for axon guidance studies [17
]. Thus, using the well-characterized retinotectal projection system, we next asked if our procedure could enable selective transfection of retinal ganglion cells (RGCs) and/or the regions through which their axons travel. Normally, axons from RGCs exit the eye, travel along the optic nerve to enter the brain at the ventral diencephalon. After crossing the midline at the optic chiasm, they extend dorsally through the optic tract in the diencephalon before turning caudally to reach the optic tectum, where they arborize and form synapses.
Since only the region lying between the two electrodes is efficiently electroporated, different areas can be selectively electroporated by sliding the embryo forward or backward in the main channel to expose the rostral or caudal part of the head (Figure and ). This configuration gives rise to large transfected areas, extending rostro-caudally over 568 ± 40.5 μm (n = 13). Taking advantage of the insulating property of Sylgard, the electroporated region can be restricted by narrowing the transverse channel (158 ± 17.6 × 89 ± 8.7 μm, n = 14), making specific electroporation of the embryonic tectum or diencephalon feasible (Figure ). In addition, the relative orientation of embryos to the electrodes can be changed to drive DNA towards different regions. Using modified chambers, the ventral-most regions of the brain, which are usually difficult to electroporate, can be targeted, allowing electroporation of the optic chiasm region (Figure and ).
Figure 4 Using electroporation to study retino-tectal projections in vivo: a-b: Regions of the brain can be differentially targeted by sliding the embryo in the main channel (compare upper and lower panels in a). When the caudal part of the head was exposed, most (more ...)
To specifically electroporate the eye, embryos were placed belly up so that the eye, but not the brain, was aligned with the electrodes (Figure ). This avoided non-targeted electroporation of the brain, which is important as the lumen of the eye vesicle, where the DNA injection is made, communicates directly with the brain ventricles at early stages (Figure insert and k). Eye specific electroporation was successfully performed over a range of stages from 22 to 35/36 without affecting eye development (Figure , see additional file 1e
e and 1f
f [electroporated at stage 24, 28 and 32 respectively]). The success rate, evaluated 12 h after electroporation, was over 80%. 48 h after electroporation, all layers of the retina were transfected and cellular morphology within the retina appeared normal (Figure and ). Interestingly, when eyes were electroporated at stage 32, GFP-positive cells were widely distributed in early stage 33/34 retina only 6 h after transfection (see additional file 1c
c and 1d
d). Eye-targeted electroporation yielded high levels of transgene co-expression when pCS2GFP and pCS2GAP-RFP plasmids were injected in a 1:1 ratio. 95.4 ± 1.2% of the GFP positive cells were RFP positive and 81.8 ± 3.1% of the RFP expressing cells were also GFP positive (n = 313 and n = 366 cells respectively from 10 sections). More importantly, the co-electroporation efficiency remained high even if different types of plasmids were mixed (Figure ). Co-electroporation enables multiple perturbations as well as easy monitoring of the transfected cells. Indeed, RGC axons can be easily analyzed both in transverse sections (Figure ) and in the whole brain (Figure ) after GAP-GFP electroporation. In addition, GAP-GFP transfection enables powerful time-lapse analysis of extending retinal axons and growth cone dynamics to be performed in vivo
(Figure ) [52
]. Fixed sample analysis as well as live monitoring of GAP-GFP expressing RGC axons show that electroporation does not perturb axonal growth, navigation or branching. Thus, electroporation is suitable for manipulating and monitoring RGC axons at stages when lipofection has proven to be difficult.
Finally, we asked if both the eye and the pathway where retinal axons grow (e.g. optic tract, optic tectum) could be manipulated separately within the same embryo. We electroporated eyes at stage 24 with GAP-RFP and brains 8 h later at stage 30 with GAP-GFP. We found that dual-electroporation produced specific expression both within the eye and pathway. Importantly, dual electroporation did not decrease embryo viability or cause abnormal development (n = 21), and did not affect brain anatomy (see additional file 2e
e). Furthermore, eye-specific electroporation can be combined with either large or area-specific brain electroporation (Figure ). Similarly, brain electroporation was successfully performed on eye-lipofected embryos (Figure ). Thus, dual-electroporation provides a way to separately control transgene expression in the retinal axons versus the substrate pathway enabling in vivo
analysis of axon-target and axon-pathway interactions.
Figure 5 Both retinal projection neurons and their substrate pathway can be manipulated separately in the same embryo. a-d: Eye-targeted electroporation can be combined with brain electroporation. a: A dorsal view of an embryo doubly transfected. Retinal axons (more ...)
Targeted loading of antisense morpholinos by electroporation
In addition to DNA transfection, intracellular delivery of antisense morpholinos (MOs) was tested. MOs are an effective tool to knock-down protein expression in Xenopus
through blastomere injection [53
] but their inability to be taken up through the plasma membrane has limited their use at late stages. Standard MOs are uncharged and, therefore, cannot be electroporated. However, MOs can be fluorescently tagged for visualization and, fortuitously, the tag introduces a charge making them amenable to electroporation [19
]. Electroporation of MOs in chick has been used to study nervous system development, and single cell electroporation in the brain of late-stage Xenopus
embryos has been reported [12
Lissamine-tagged MOs were electroporated into both the brains and the eyes of stage 22 to 33/34 embryos with a success rate of over 80% with all four settings used (20 V/50 ms/1 s/8 x, 18 V/50 ms/1 s/10 x, 18 V/25 ms/1 s/10 × or 15 V/50 ms/1 s/10 x). In transverse sections, MO-loaded cells were evenly distributed throughout the width of the electroporated side of neuroepithelium, and along most of its dorso-ventral axis (Figure and ). With eye-targeted electroporation, MO-positive cells were found in all of the cellular layers of the retina (Figure and ). In all conditions tested, MO fluorescence was still detectable 48 h after electroporation.
Figure 6 Introducing Morpholinos into young Xenopus tadpoles by electroporation and in vitro approaches. a-d: Frontal sections of embryos 24 h after electroporation with lissamine-tagged MO. Large numbers of cells can be loaded with MO in both the brain (a) and (more ...)
Plasmids were also successfully co-electroporated with negatively charged MOs (both fluorescein- and special delivery lissamine-tagged) (Figure ). For example, 48 h after eye-targeted electroporation (GFP 0.7 μg/μl [0.26 pmol/μl], MO 0.25 mM), 94.4 ± 1.4% of the GFP expressing cells were MO positive (n = 597 cells from 20 sections from 3 embryos). However, in this condition the MO positive domain was slightly larger than the GFP-expressing domain (Figure ), resulting in 65.4 ± 2.3 MO-loaded cells expressing the GFP (n = 863 cells). Finally, as GFP accumulates progressively the degree of co-localization will change slightly with time and should be taken into consideration.
At a cellular level, electroporated MOs seem to diffuse evenly around the cytosol and into the nucleus. Although short neuronal processes could be visualized with the MO fluorescence, longer processes were usually so weakly labeled that they were difficult to follow. However, co-transfection with GAP-GFP highlighted axons emanating from the electroporated RGCs, making monitoring of axons from MO-loaded cells feasible (Figure ).
Finally, we took advantage of the high co-electroporation efficiency to test the ability of the loaded MO to down-regulate translation of its target mRNA. A pCS2 plasmid encoding GFP was electroporated with or without a MO designed to block GFP expression [55
]. Using this MO directed against the pCS2+ and cytoplasmic GFP sequences flanking the GFP start codon, we observed a 4-fold decrease of the GFP signal both in the eye and the brain when the MO against GFP was co-electroporated with pCS2GFP (n = 24, n = 30 respectively, data not shown). However, this decrease could be due to biased electroporation and/or nonspecific effects of the anti-GFP MO (such as cell death or general translation inhibition). To rule out such problems, the experiments were performed on the targetable (pCS2) and non-targetable GFP (pEGFP) in the presence of a control tagged MO to assess the electroporation efficiency. Co-electroporation of the GFP MO only decreased the expression of the targetable plasmid (Figure ). The decrease of the fluorescent ratio of the GFP over GAP-RFP signal in the GFP MO electroporated eyes further supports a specific effect of the MO on the GFP (Figure ). In conclusion, our electroporation procedure enables efficient loading of MOs without impairing their activity in vivo
. This suggests that, similar to chick, controlled spatio-temporal MO knock-down approaches could be achieved by electroporation in early tadpole Xenopus
embryos. Furthermore, electroporation allows sequential modifications of gene function when used in combination with other techniques such as lipofection (Figure and ).
In vivo electroporation provides source of transfected/MO-loaded neurons for in vitro studies
The embryonic Xenopus
brain is extremely small, posing challenges for obtaining a sufficiently large number of cells to perform dissociated cell electroporation protocols [56
] and alternative transfection methods have low efficiencies [58
]. For example, MO uptake by Xenopus
retinal cultures is inefficient even when specific transmembrane trafficking molecules, such as Endo-Porter (GeneTools) are used (data not shown). Thus, most Xenopus
transfected or MO-loaded cells used in culture have been obtained from embryos injected at early blastomere stages [59
]. However, premature death or abnormalities of injected embryos limit the spectrum of MOs or constructs that can be used to analyze later events in vitro
. Therefore, we cultured explants or dissociated cells from different parts of brains electroporated with GAP-GFP DNA and/or fluorescently tagged control MOs (fore-, mid- or hind-brain). As shown in Figure , both MO-loaded and DNA transfected cells can be successfully cultured and up to 40% of the cultured cells showed expression. Moreover, the positive-expressing explants and dissociated cells were readily detected, even at low magnifications suggesting that intracellular levels of the MO and the DNA were high. In culture, MOs could be readily seen in axons and growth cones (Figure ), and could still be detected after 2 days in vitro
. This makes in vivo
electroporation a potent source of transfected cells for in vitro