Application of Impermeable Barriers Combined with Candidate Factor Soaked Beads to Study Inductive Signals in the Chick

The chick embryo provides a superb vertebrate model that can be used to dissect developmental questions in a direct way. Its accessibility and robustness following surgical intervention are key experimental strengths. Mica plates were the first barriers used to prevent chick limb bud initiation1. Protocols that use aluminum foil as an impermeable barrier to wing bud or leg bud induction and or initiation are described. We combine this technique with bead placement lateral to the barrier to exogenously supply candidate endogenous factors that have been blocked by the barrier. The results are analyzed using in situ hybridization of subsequent gene expression. Our main focus is on the role of retinoic acid signaling in the induction and later initiation of the chick embryo fore and hindlimb. We use BMS 493 (an inverse agonist of retinoic acid receptors (RAR)) soaked beads implanted in the lateral plate mesoderm (LPM) to mimic the effect of a barrier placed between the somites (a source of retinoic acid (RA)) and the LPM from which limb buds grow. Modified versions of these protocols could also be used to address other questions on the origin and timing of inductive cues. Provided the region of the chick embryo is accessible at the relevant developmental stage, a barrier could be placed between the two tissues and consequent changes in development studied. Examples may be found in the developing brain, axis extension and in organ development, such as liver or kidney induction.


Introduction
Classical embryologists traditionally employed physical techniques to interrogate the mechanisms controlling embryonic development. In the 1960s and 1970s investigators developed techniques that used impermeable barriers inserted between tissues of the developing embryo to demonstrate the importance of inductive signals during embryogenesis. Our interest is in vertebrate limb development and in particular what events precede limb bud outgrowth. For example, insertion of a barrier can prevent limb bud outgrowth from occurring on one side of the chick embryo while allowing it to proceed normally on the contralateral, unoperated side. Materials used to make these barriers varied; for example, mica plates 1 and tantalum foil 2 . These barriers effectively separate the lateral plate mesoderm (LPM), from which the limb bud forms, from the somites formed from the paraxial mesoderm. Surgical grafting techniques demonstrate that close association with the somites is essential if limb bud initiation is to occur in the LPM 34 . In the 80s and 90s developmental biologists discovered diffusible signaling molecules that are essential in controlling developmental processes. Beads soaked with these signaling molecules can be implanted in specific regions of the embryo and produce alterations to embryonic development. For example, retinoic acid (RA) taken up by ion exchange beads is released over a 24-hour period when implanted into a chick embryo and can produce mirror-image duplication of the digits 5 . Heparin acrylic beads containing FGF protein can initiate limb bud outgrowth from the interlimb LPM 6 .
More recently mouse and fish genetics have taken the study of vertebrate limb initiation into a new era (reviewed in 7 ). There is good evidence that RA present in the somites prior to limb bud initiation is the essential diffusible signal required by the LPM to initiate limb budding. We wanted to exploit the accessibility and experimental robustness of the chick embryo to dissect the effect of barrier implantation on gene expression in the LPM around the time of limb bud initiation. A novel adaptation of our method is to combine a barrier that blocks signals from axial tissues with bead placement lateral to the barrier to exogenously supply candidate endogenous factors that have been blocked by the barrier. The following protocols are those developed to tease out the mechanisms of limb bud induction and initiation.
Studying limb bud initiation means using early stage embryos. Many researchers window chicken eggs by first removing some of the albumin through the air sac with a hypodermic needle. The embryo, which lies on top of the yolk, will then lie lower in the egg. This is an advantage for some techniques, such as electroporation, but can be a disadvantage if surgical manipulation of the embryo is desired. We find having the embryo as close to the opening in the egg as possible is an advantage. 5. Use the curved scissors to enlarge the hole in the shell to a circle with a diameter of approximately 1 cm. Remove the piece of excised shell and cover the hole in the egg with a piece of clear tape approximately 5 cm square. To easily remove this tape, press the tape to stick it only at the lip of the hole, leaving the remaining tape free. The free tape can then be grasped to unseal the egg.  1. Barrier inserted at stage 13 to prevent wing bud initiation.
1. Take a stage 13 embryo from the incubator and place the egg on an egg rest under a binocular microscope set to 3.2X magnification or higher if preferred. Remove the upper layer of clear tape.
1. Using a steel micro knife, make an incision through the vitelline membrane and lateral plate mesoderm (LPM) adjacent to somites 15 to 20 ( Figure 1A) on the right side of the embryo (somites have not all formed at this stage). Cut adjacent to the last four somites formed and continue the cut two somite lengths caudally. 2. Pick up the barrier (0.7-1 mm wide) with No. 5 microforceps at the end protruding from the Petri dish and twist the hand to insert the free end of the barrier into the incision. The hinge shaped barrier lies flat on the vitelline membrane above the LPM with half of it protruding downwards through the incision distal to the somites. As soon as possible seal up the egg with clear tape and return it to the incubator. NOTE: The micro knife and forceps become sticky on contact with the vitelline membrane. It is important to wipe them clean with 70% ethanol before attempting a subsequent operation. This applies to all subsequent operation descriptions and to bead implantations. 3. If a bead is to be inserted in conjunction with a barrier; firstly, take the bead in No. 5 microforceps and insert it into the cut face of the LPM on the distal side of the incision and then insert the barrier into the incision proximal to the bead ( Figure 1B). Seal the egg and return it promptly to the incubator.
2. Barrier inserted at stage 15 to prevent leg bud initiation. NOTE: An example of the result of this operation on Tbox 4 transcription factor (Tbx4) expression is shown in Figure 2D and 2E.
1. Take a stage 15 embryo from the incubator and place the egg on an egg rest under a binocular microscope set to 3.2X magnification or higher if preferred. Remove the upper layer of clear tape.
NOTE: Barriers are removed because they are sharp and can damage embryos as they go through the in situ hybridization protocol, especially if several embryos are stained at once. 1. Following fixation, place the embryo in a dish of PBS under a binocular microscope. 2. Using two pairs of No. 5 microforceps, place the left pair closely either side of the barrier in the operated side of the embryo. Grasp the barrier with the right pair and gently pull the barrier from the embryo using the left pair to keep the embryo down and to prevent any tissue from adhering to the barrier as it is removed. 4. Carry out WISH essentially as described in 11 .

Representative Results
The protocol detailed above describes the methods employed in Nishimoto et al., 2015. The paper studies limb bud induction and initiation in both the fore and hindlimb of the chick embryo. Results following the insertion of impermeable barriers to prevent the outgrowth of the forelimb bud have been published previously 1, 2, 9 but there has been only one study describing barriers preventing leg bud initiation  (Figure 2D and E). We conclude that Tbx4 expression alone is not sufficient to initiate hindlimb outgrowth from the LPM. Other studies in the chick have suggested that Tbx gene expression is sufficient to initiate forelimb bud formation 12,13 . Figure 2C' is included to illustrate how a barrier looks imbedded in the leg region of the embryo post fixation. In the majority of cases barriers were removed prior to in situ hybridization (protocol 4.3).
RA from the Somites Is Essential in the LPM before Fgf10 Induces Limb Bud Outgrowth: Figure 3A is a schematic representing the addition of an RA soaked bead distal to a barrier inserted at stage 15 (protocol 3.2.1.3). Subsequent to this operation limb bud outgrowth (denoted by an asterisk) is observed on the operated right side and Tbx4, Fgf10 and Fgf8 are expressed in the bud as they are in the control left side (Figure 3B, C and D, Table 2). Thus addition of RA to the LPM overcomes the effect of the barrier and limb bud initiation proceeds. We conclude that in normal development RA from the somites is essential in the LPM before limb bud outgrowth can be initiated. The resulting hindlimb buds are generally smaller than those on the control side. This may be due to the RA dose in the LPM being not equivalent to the wild type situation. If the RA dose is too high the Apical Ectodermal Ridge (AER) can be shortened and smaller buds would result 14,15 .
To confirm that RA (from the somites) in the LPM is required for normal limb bud initiation an inverse agonist of RAR, BMS 493, was used. Beads are implanted into the leg region LPM at stage 15 (see protocol 3.2.1.4) (Figure 3E). Fgf10 expression is downregulated following application of BMS 493 beads in the LPM, resulting in smaller hindlimb buds compared to the control side ( Figure 3F; Table 3). Control DMSO beads do not cause these defects (Figures 3G and 3H; Table 3). The defects observed following BMS 493 application are milder than those induced by a barrier operation; i.e., Fgf10 is still expressed and limb buds are formed. This is likely to be because the effects of BMS 493 are restricted locally around the beads and may not be able to antagonize all the RA produced by axial tissues.
Early Axial Signals Specify the LPM Cells that Later Express Tbx4: We tested when the LPM acquires its ability to express Tbx4 in the prospective leg-forming region. The earliest stage at which a barrier can be inserted that will subsequently block leg bud outgrowth, is stage 10. We are the first to develop a protocol to insert barriers at stages earlier than stage 12. Barriers inserted between the paraxial mesoderm and the presumptive leg LPM at stage 10 (see protocol 3.4) block leg bud formation and the expression of Tbx4 in the leg-forming LPM (Figures 2F and 2G, Table 2), suggesting that a signal from axial tissues at stage 10 is required for later expression of Tbx4 in hindlimb LPM. Compare this result with Figures 2D and E where later barrier insertion allows Tbx4 expression to be established. Furthermore, we tested whether this axial signal could be RA by observing whether the inverse agonist of RAR reduces Tbx4 expression. A BMS 493 bead placed in the hindlimb LPM at stage 10 (see protocol 3.4.1.3) downregulates Tbx4 expression (Figures 2H and 2I), whereas control beads soaked in DMSO do not affect Tbx4 expression (Figures 2J and 2K). Together, these results support a model that an RA signal from axial tissues regulates limb induction by positively regulating Tbx4 in the hindlimb LPM 7 .
Post-operative Death: Tables 1, 2 and 3 give details of experimental outcomes including the number of chick embryos that died post operation and before harvesting. Some deaths are to be expected after microsurgery of this nature. The numbers dying can be kept to a minimum by making sure that instruments are clean, the hole in the egg is the minimum size required, that the embryos are left without the protection of the sticky tape seal for a minimum time and, in conjunction with this last point that the operation is carried out as quickly as possible. None of these operations are very time consuming but operations involving beads and barriers take a little more time. Despite these measures, operations carried out at the earliest stages and those in the wing region are more likely to result in death. There are large blood vessels around the limb-forming regions and accidental rupture of these vessels may lead to death due to blood loss. We find that eggs opened at stage 8 or 9, whose embryos have not been operated on, will often die by the next day. Therefore, the only solution to these problems is to carry out a high number of experiments.    Chick embryos fixed at stages 17-19.

Discussion
We describe the use of impermeable barriers to prevent limb bud formation in the chick embryo. This technique has a number of critical steps.
Following preliminary experiments, we found that the hinge shaped barriers used in 9, 10 remain in place in the embryo far better than straight barriers. The distal flat side of the barrier sticks to the vitelline membrane and keeps the barrier in place (Figure 2C'). Careful preparation of hinge shaped barriers from aluminum foil as described in protocol step 1.6 is essential to the success of the operation. A second important step is to ensure that eggs are not left out of the incubator for too long following windowing and staging prior to operating on the embryo. In our experience, embryos survive an operation better if they are at or near incubation temperature when the operation is carried out. Operations should therefore be carried out as quickly as possible and the eggs resealed promptly to ensure good survival rates. Many researchers add a drop of antibiotics to the embryo following microsurgery before returning the eggs to the incubator. The presence of the liquid prevented the barriers from sticking well in position. Addition of antibiotic could also dislodge beads after placement. Therefore not using antibiotics, swift operations and good cleanliness with instruments is recommended to avoid infection. Wiping instruments with 70% ethanol following every stage of an operation is essential to prevent instruments from becoming sticky from contact with the vitelline membrane which can cause both barriers or beads to adhere to forceps (see protocol step 3.1.2.2). Both varieties of beads used could fall out of place following the operation. When the barrier position was checked the next day it was not possible to see through the barrier to ascertain whether the bead remained in place or not. In cases where the bead had fallen out wing bud outgrowth was not rescued ( Table 1). Avoiding transferring liquid with the bead when implanting them means that beads stick more easily to the cut face of the LPM (see protocol 2.3.2.3). Finally ensuring that embryos are fixed immediately after harvesting is critical to good subsequent gene expression analysis (see protocol step 4.2).
This protocol modifies previously published experiments that used barriers to prevent limb bud initiation, defines specific optimum widths for barriers and introduces combining barrier and bead placement followed by gene expression analysis. We found using conventional aluminum foil, which is cheap and readily available in all labs, produces outcomes equal to those that previously used tantalum foil. The foil can be left in the embryo when carrying out in situ hybridization (Figure 2C') but we usually remove it if more than one embryo is being processed to prevent the sharp edges of the foil damaging embryos. Pilot studies demonstrated that the width of leg region barriers is particularly important. Initially the barriers trialed were too short and leg bud outgrowth was not prevented. 1.2-1.3 mm is the optimum width for a barrier to block leg bud outgrowth completely. The narrowest wing barriers that one can place accurately to prevent bud outgrowth give the best survival rates. This can be as narrow as 0.5-0.7 mm. However, barriers that are a little wider are more likely to result in complete blocking of wing bud outgrowth because they allow for a little inaccuracy of placement (we used 0.7-1 mm). Avoidance of blood vessels when making the incision is critical for wing bud level barriers. Semi-permeable barriers have previously been used to bisect the chick limb bud. MacCabe and Parker, 1976 and Summerbell, 1979 both used filters 0.45 µm and 0.8 µm pore size, respectively 16, 17. They reported that diffusible signals could pass through the pores in the filters and that results gained using semi-permeable barriers were midway between a limb bud with no barrier present and one with an impermeable tantalum foil barrier. This protocol could be adapted to use semi-permeable barriers of varying pore size to differentiate candidate signals based on size or to modify the dynamics of signal diffusion.
This protocol has various limitations mainly associated with the model organism itself. Such techniques could be used to probe further questions in limb development and also other questions of vertebrate embryogenesis. A proviso is that the organ/area in question should be developing when the chick embryo is accessible to intervention. The presence of large blood vessels in the area where a barrier needs to be placed makes barrier placement later than stage 13 or 14 technically challenging. If large blood vessels are cut during barrier placement this can cause embryo mortality. Our investigation of when the transcription factors Tbx5 and Tbx4 are first induced in the LPM was limited by the earliest stage at which we could place a barrier that would later end up opposite either the wing or leg bud respectively. It could be illuminating to attempt to block Tbx gene induction at earlier stages, closer to gastrulation. The earliest a barrier could be placed in the presumptive wing region successfully was stage 8 and in the leg region stage 10. This protocol is probably not suitable for the study of gene expression beyond limb bud stages because subsequent rapid growth would displace the barrier and results would relate only to the initial position of the barrier.
Having related the limitations of the chick embryo in this context, its advantages over other vertebrate model organisms are many. The chick embryo is a very suitable animal model for this type of physical intervention in a vertebrate embryo. The embryos are relatively large and are easily accessible through the eggshell. One can return to the embryo for further experimentation (e.g., the removal of a bead 18 or even a barrier) or to check on progress by simply removing the tape window. There is possible scope to include live imaging or time lapse imaging. We describe a method of windowing chicken eggs that is particularly suited to operating on early stage embryos. In the case of limb studies it is very important and helpful that there is a contralateral control limb for every experiment that serves as an ideal internal control for each biological replicate.
The combination of using impermeable barriers to prevent limb outgrowth and beads soaked in signaling molecules has not been reported before. Our analysis of mRNA expression following such interventions is also novel. These techniques enabled us to dissect the difficult problem of when limb specific transcription factors Tbx5 and Tbx4 are induced in the LPM and also to show that later, prior to limb bud initiation, the presence of these Tbx genes in the LPM is not sufficient for the activation of Fgf10 transcription and the start of limb bud outgrowth. Figures 2G,  2D and 2E illustrate these findings well. The absence of Tbx4 expression following the insertion of a barrier at stage 10 in Figure 2G is in stark contrast to its presence in 2D and 2E following barrier insertion at stage 15, after the induction of Tbx4. The implantation of RA soaked beads facilitated the rescue of limb bud initiation following barrier insertion and so points to RA from the somites being essential in the LPM before limb bud outgrowth can begin. Such techniques could be used to solve further questions in limb development but also other questions of vertebrate embryogenesis. Areas of the brain, eye and trunk are likely to be suitable.

Disclosures
The authors declare they have no competing financial interests.