- Top of page
- RESULTS AND DISCUSSION
- EXPERIMENTAL PROCEDURES
- ChIP PROTOCOL
Chromatin immunoprecipitation (ChIP) is a powerful method for analyzing the interaction of regulatory proteins with genomic loci, but has been difficult to apply to studies on early embryos due to the limiting amount of genomic material in these samples. Here, we present a comprehensive technique for performing ChIP on blastula and gastrula stage Xenopus embryos. We also describe methods for optimizing crosslinking and chromatin shearing, verifying antibody specificity, maximizing PCR sensitivity, and quantifying PCR results, allowing for the use of as few as 50 early blastula stage embryos (approximately 5×104 cells) per experimental condition. Finally, we demonstrate the predicted binding of endogenous β-catenin to the nodal-related 6 promoter, binding of tagged Fast-1/FoxH1 to the goosecoid promoter, and binding of tagged Tcf3 to the siamois and nodal-related 6 promoters as examples of the potential application of ChIP to embryological investigations. Developmental Dynamics 238:1422–1432, 2009. © 2009 Wiley-Liss, Inc.
- Top of page
- RESULTS AND DISCUSSION
- EXPERIMENTAL PROCEDURES
- ChIP PROTOCOL
Chromatin immunoprecipitation (ChIP) has emerged as an invaluable tool for the study of the mechanisms of transcriptional control and chromatin dynamics. ChIP allows an investigator to determine whether a genomic locus is occupied by chromatin-bound factors such as transcription factors, chromatin remodeling complexes, and modified histones. The most widespread ChIP procedure uses formaldehyde-crosslinked, sheared chromatin from 106 to 107 cells as the input material for an immunoprecipitation (IP), which is followed by several rounds of washing, crosslink reversal, and DNA purification. Because the genomic DNA is sheared to an average size of 1,000 base pairs or less, the IP results in the purification of discrete genomic DNA fragments that associate with the antigen of interest. Thereafter, the purified DNA from the experimental IP is queried for enrichment relative to a control IP, either by PCR—for small numbers of target genes—or by one of several genome-wide analysis methods (microarray, high-throughput sequencing, library screening). Thus, ChIP represents a powerful method for investigating in vivo protein-DNA interactions.
For molecular embryologists, however, the typical ChIP protocol poses a number of challenges. Embryos represent heterogeneous populations of cells containing limiting amounts of genomic material. In addition, fractionation of embryos under the denaturing conditions commonly used in ChIP releases a large amount of non-chromatin-associated proteins, such as yolk, that complicate sample preparation and can increase nonspecific background. Consequently, molecular embryologists have been slow to adopt ChIP as a routine assay, especially for the analysis of early embryos. However, ChIP analysis of early embryos could help forge new frontiers in developmental biology. Whether by providing for the enhanced analysis of transcription factor function in gene regulatory networks, or by investigating the function of histone modifications and how their patterns unfold during embryogenesis, numerous new avenues of investigation will require the establishment of a ChIP protocol amenable to embryonic tissues.
Our goal was to develop a ChIP procedure that would be sensitive enough to detect transcription factor occupancy at promoters in cleavage-stage Xenopus laevis embryos (stage 7.5 to 8: approximately 1×103 cells per embryo). In addition, we wanted the protocol to be amenable to typical embryological manipulations, such as microinjection, and therefore optimized the protocol to use as few as 50 embryos per sample from these early stages. Finally, we wanted to circumvent several foreseeable problems with sample preparation by minimizing non-chromatin proteins in the ChIP samples, optimizing the crosslinking and sonication steps, and optimizing DNA extraction and PCR conditions to maximize sensitivity. While this protocol was in development, a number of ChIP experiments on Xenopus laevis embryos were reported (Jallow et al.,2004; Kim et al.,2004; Morgan et al.,2004; Messenger et al.,2005; Park et al.,2005; Ng and Gurdon,2008), representing four different protocols. While all of these protocols are similar in principle to ours, we have concentrated on maximizing the sensitivity of this procedure for the analysis of earlier stages of development with a small number of embryos. Here, we present our optimized protocol in detail, with some examples of its implementation. By demonstrating the basic method, and describing how the protocol was optimized, we aim to facilitate the adoption of ChIP as a routine assay in the Xenopus embryological laboratory.
While our method is similar to now-standard ChIP protocols in use with other model systems (Kuo and Allis,1999), some critical differences may be particular to the Xenopus system. First, we report optimized fixation and sonication techniques that yield chromatin crosslinked and sheared enough to detect transcription factor occupancy at promoters with at least 1,000-bp resolution. Second, because the standard 1% SDS lysis buffer used during sonication tended in our hands to produce low-quality chromatin from early embryos, we used a low-SDS (0.1%) radio-immunoprecipitation assay (RIPA) buffer. RIPA buffer produces high-quality sheared chromatin samples while reducing yolk solubilization, thus limiting background by preventing protein precipitation. Doubling the number of washes further reduces background. Finally, we have optimized DNA purification and PCR conditions to allow for the reliable detection of as little as 30 copies of a target sequence per reaction, facilitating the use of as few as 5×104 cells in the starting sample, an improvement of two orders of magnitude from the typical ChIP protocol. We also demonstrate approaches for quantitative PCR and statistical analysis of ChIP results, which can offer several advantages over endpoint PCR detection strategies for detecting differences between samples. Interestingly, several protocols for ChIP using either cultured cells, early mouse embryos, or tissue biopsies have been described that use as few as 1×102 cells (O'Neill et al.,2006; Acevedo et al.,2007; Dahl and Collas,2008), suggesting that, with modification, the sensitivity of this procedure could be enhanced even further. The protocol we present here is well suited to the Xenopus embryologist: it facilitates the use of microinjected embryos and explanted tissues by decreasing the amount of genomic material required to obtain meaningful data. Thus, we present this work with the hope that it will help advance the use of ChIP in embryological experiments and lead to new avenues of research in developmental biology.
RESULTS AND DISCUSSION
- Top of page
- RESULTS AND DISCUSSION
- EXPERIMENTAL PROCEDURES
- ChIP PROTOCOL
In this section, we discuss the critical parameters for optimization and validation as well as general guidelines for ChIP in Xenopus. A step-by-step protocol follows in the Experimental Procedures section. The most critical parameter in our experience is the method used to generate the sheared chromatin sample. Several factors need to be considered: extent of crosslinking, duration/strength of sonication, and yield. In general, the greater the extent of crosslinking, the greater the amount of sonication that will be required to generate ideal (<1,000 bp) fragments. However, prolonged crosslinking will render chromatin impervious to fragmentation (Orlando et al.,1997), and over-sonication will result in reduced overall yield of genomic DNA. In addition, the target antigens should be considered. For example, nucleosomes can be immunoprecipitated with sheared genomic DNA without crosslinking (O'Neill and Turner,2003), while transcription factors and secondarily-associated protein complexes may require extended crosslinking times (or different crosslinking reagents) to achieve sufficient co-immunoprecipitation of genomic DNA (Zeng et al.,2006).
These factors should also be considered when customizing this protocol to particular applications. The following procedure was used to optimize crosslinking and shearing for blastula and gastrula stage Xenopus embryos. All sonication steps were performed using a Branson Sonifier 250 equipped with a ¼ inch microtip horn, set at 20% output. Other makes of sonicators will have different efficiencies, making these optimization steps even more critical.
To optimize crosslinking time, we performed a fixation time course on gastrula stage (stage 10) embryos. Using this stage ensures that enough DNA will be recovered for analysis by agarose gel electrophoresis. We collected embryos fixed in 1% formaldehyde/PBS for 15, 30, 45, and 60 min, as well as control, non-fixed embryos. Samples were prepared according to the ChIP Day 1 protocol up to step 8 (see Experimental Procedures section). We sonicated the samples minimally using conditions (3× 20 sec, 20% output, 20% duty cycle) that would solubilize the genomic DNA from the insoluble pellet, effectively shearing native DNA, but would minimally shear crosslinked DNA. Post- sonication supernatants were adjusted to 1% SDS, 10 mM EDTA prior to crosslink reversal and DNA purification, as described in the Experimental Procedures section, except that following RNase treatment, DNA was ethanol precipitated, resuspended in 50 μl H2O, and analyzed by agarose gel electrophoresis.
While genomic DNA from embryos fixed for 15 min showed no resistance to shearing as compared to control, DNA from embryos fixed for as little as 30 min showed evidence of crosslinking, indicated by the detection of slower-migrating, sonication-resistant DNA (Fig. 1A). Subsequently, the extent of crosslinking was increased incrementally until the 60-min time point. Although 60-min fixation times have been reported (Orlando et al.,1997), typical ChIP protocols performed on yeast and cultured cell systems fix samples for 10 to 15 min (Luo et al.,1998; Kuo and Allis,1999), with similar amounts of formaldehyde in PBS or culture medium. This result suggests that the kinetics of nucleoprotein crosslinking by formaldehyde is different in the case of the Xenopus embryo, perhaps due to a greater non-chromatin protein to DNA ratio compared to other systems. Therefore, it will also be important to optimize crosslinking times for later embryonic stages when this ratio begins to approach typical somatic levels.
Figure 1. Crosslinking and sonication optimization. A: Minimally sonicated, gastrula stage (stage 10) DNA from a 1% formaldehyde/1×PBS time course was resolved by 2% agarose gel electrophoresis. Onset of crosslinking is observed by the recovery of sonication-resistant, high molecular weight DNA (>1 kb). B: Sixty-minute crosslinked chromatin from gastrula stage embryos was fully sonicated for one to four rounds of 20 sec each and resolved by 1.2% agarose gel electrophoresis. Three rounds of sonication balances optimal yield (90%) with maximal shearing (<1 kb) of the crosslinked chromatin.
Download figure to PowerPoint
General Sonication Guidelines
Perform sonication on ice to prevent overheating. Place the sample to be sonicated in a beaker filled with ice, and hold this under the sonicator horn during shearing.
Sonicate in short bursts. Twenty-second rounds of sonication prevent sample overheating. Let the samples rest for at least 1 min before the next round.
Center the horn in the sample and avoid contact with the walls of the microcentrifuge tube. This will improve reproducibility.
Avoid sample foaming, which happens when the tip of the horn draws air into the sample because it was brought too close to the surface.
To determine whether we could generate sufficiently small average DNA fragment size with a longer fixation time, we crosslinked gastrula stage (stage 10) embryos for 60 min and performed one to four rounds of full-strength sonication (20 sec each, 20% output, 100% duty cycle) and repeated the crosslink reversal and DNA purification as described below (see Experimental Procedures section). One round of sonication generated a majority of <1,000-bp fragments, but we also noted a population of high molecular weight DNA that was reduced with each successive round of sonication (Fig. 1B). By four rounds of sonication, the high molecular weight DNA was virtually undetectable, but overall DNA yield was also reduced by 25%. Therefore, we concluded that three 20-sec rounds of sonication balance optimal yield (90%) with average fragment length (<1,000 bp). The efficiency and specificity of this method of crosslinking and shearing was further validated as described below.
Chromatin Immunoprecipitation in Blastula Stage Embryos
We tested the ChIP protocol by scanning for occupancy of the transcription factor β-catenin within a 2.5-kb upstream portion of the Xenopus nodal-related 6a (Xnr6) locus that contains several predicted Tcf/Lef binding sites (Fig. 2A). Wnt/β-catenin pathway activity is required for Xnr6 expression (Takahashi et al.,2000; Rex et al.,2002; Xanthos et al.,2002; Yang et al.,2002), so we predicted that some of these sites would be occupied by β-catenin in blastula stage embryos. In addition, we performed ChIP for a euchromatin marker, di-acetylated [K9/K14] histone H3 (AcH3), reasoning that an active locus should be positive for AcH3 (Roh et al.,2005). A portion of the related Xnr1 locus that is not predicted to bind β-catenin was analyzed as an additional control (Fig. 2A).
Figure 2. Chromatin immunoprecipitation in blastula-stage Xenopus embryos. Schematic representations of the Xnr6a and Xnr1 genomic loci (A) demonstrate the locations of predicted Tcf/Lef consensus sequences relative to the ChIP PCR amplicons. B: These primer sets amplify standard genomic DNA by PCR with similar efficiencies, with a limit of detection at approximately 30 haploid genomes. C: Blastula-stage embryos (Stage 9) were processed for ChIP using either anti-acetylated histone H3, or anti-β-catenin antisera (with the corresponding negative controls). The PCR products were visualized by 3% agarose gel electrophoresis and ethidium bromide staining. Co-immunoprecipitation of associated genomic DNA is observed when the PCR signal is greater for the experimental IP (AcH3, β-catenin) than in the control IP (IgG, Serum). Note that the Xnr1 (-221) product occasionally amplifies as a doublet (seen in C): this represents genetic variation at this locus within our colony.
Download figure to PowerPoint
PCR conditions were optimized to ensure detection of immunoprecipitated DNA present in limiting quantities. To maximize detection, we performed nested PCR, using two rounds of 20 cycles each to amplify target sequences from genomic DNA standards. Nested PCR has the twofold advantage of replenishing the polymerase, primers, and dNTPs available for amplification while increasing priming specificity by using a second, internal primer set for the second round of amplification. By this method, we are able to detect PCR products from as little as 100 pg of genomic DNA, corresponding to approximately 30 haploid genomes (Fig. 2B). Radiolabeling the second, inner PCR reaction further increases sensitivity (not shown).
We performed ChIP on blastula stage embryos with anti-AcH3 and anti-β-catenin (Fig. 2C). All loci tested were associated with AcH3, while only the Xnr6 (−118) and (−2,349) amplicons were bound by β-catenin. Notably, an intermediate amplicon (−1,280) was negative for β-catenin binding. Therefore, the fixation and sonication method in this ChIP protocol sufficiently fragments chromatin, and has at least a ∼1,000-bp resolution. We expect that this approach will be useful for scanning intergenic regions for transcription factor binding sites and occupancy of modified histones. Additionally, these results confirm an expected result, namely that Xnr6 is a direct target of the Wnt/β-catenin signaling pathway.
Controls for Antibody Specificity
Several control experiments confirmed the specificity of our β-catenin antibody under the conditions used for ChIP. To demonstrate antibody specificity, we both confirmed that the antibody could be competed by the immunizing peptide and tested that depletion of β-catenin would reduce the amount of co-immunoprecipitated chromatin. When possible, this latter approach is a powerful method for antibody validation, as it will reveal off-target antibody recognition that could be overlooked by peptide competition alone. First, we optimized conditions for peptide competition of antibody-antigen binding using Western blotting of protein recovered from immunoprecipitated chromatin. The β-catenin antibody was raised against the 145 N-terminal amino acids of the Xenopus laevis β-catenin protein. We therefore used a 6xHis-tagged peptide corresponding to the immunizing peptide to compete for β-catenin binding in the ChIP assays, either by pre-incubation of the peptide with the antibody prior to the addition of sheared chromatin (Fig. 3A, lane 2) or by addition of the peptide to the sheared chromatin before the antibody (Fig. 3A, lane 3). For this experiment, the ChIP protocol was followed through Day 2, step 9, whereupon samples were analyzed by 8% SDS-PAGE followed by Western blotting (see Experimental Procedures section). Both methods of peptide competition reduced immunoprecipitated β-catenin in the ChIP samples, confirming the specificity of the antiserum. Notably, a 1/5th embryo equivalent was loaded in the input lane, while 2 embryo equivalents were loaded in the IP lanes, but similar band intensities for β-catenin are observed by Western blot. This indicates that following ChIP for β-catenin, as little as 10% of the available antigen is recovered, although these conditions effectively deplete lysates of antigen under native conditions (not shown).
Figure 3. Controls for antibody specificity. A: The specificity of the β-catenin antiserum was confirmed by performing ChIP on blastula-stage (stage 9) embryos and competing with an excess of immunizing peptide (lane 2: pre-incubation of antibody with peptide; lane 3, addition of peptide directly to sheared chromatin sample). Following IP, samples were processed as described in the text and separated by 8% SDS-PAGE. Immunoprecipitated β-catenin was analyzed by a standard Western blot using the β-catenin antiserum. B: Peptide competition of the β-catenin ChIP and (C) knockdown of β-catenin by morpholino injection (20 ng/cell at the 2-cell stage) confirms the specificity of the interaction between β-catenin and the Xnr6 promoter. The competitions in B and C were performed by pre-incubation of the peptide with the antiserum for 1 hr before addition to the sheared chromatin sample.
Download figure to PowerPoint
Additionally, we tested whether the immunizing peptide would compete for co-immunoprecipitation of the Xnr6 genomic locus. Indeed, when the β-catenin antibody is competed with the immunizing peptide, only background levels of Xnr6 (−118) are co-immunoprecipitated, thus confirming the specificity of this interaction (Fig. 3B). Likewise, only background levels of signal were detected with a negative control locus, Myosin Light Chain 2 (Xmlc2) (Park et al.,2005). Finally, in Figure 3C, we demonstrate that knockdown of β-catenin by microinjection of a morpholino oligonucleotide (Heasman et al.,2000) also results in a loss of β-catenin binding to the Xnr6 locus, comparable to the reduction seen by peptide competition. This latter result is notable, insofar as the experiment was performed with a single set of microinjected embryos (100 total, plus 100 non-injected controls), demonstrating that this approach is amenable to typical embryological manipulations in common use within the Xenopus community. These observations validate the specificity of the ChIP protocol and the observation that β-catenin binds to the Xnr6 genomic locus in blastula stage embryos.
Quantitative PCR Analysis
To determine whether ChIP samples produced by this protocol would be amenable to quantitative PCR analysis, we designed an experiment to evaluate the binding of Fast-1 (FoxH1) to the endogenous Goosecoid promoter. Goosecoid is an organizer gene with a well-defined promoter region responsive to both Wnt and Nodal signals (Watabe et al.,1995). During early embryogenesis, the Nodal pathway signals through the DNA-bound effector Fast-1 (Shen,2007).
Embryos were injected at the one-cell stage with mRNA encoding myc-tagged Fast-1 (250 pg) alone or in combination with Xnr1 mRNA (50 pg), a Xenopus Nodal-related gene. Embryos were collected at the mid-gastrula stage (stage 10.5) and processed according to the ChIP protocol, using a polyclonal anti-myc antibody to immunoprecipitate myc-Fast-1 containing complexes, followed by QPCR. As a control for non-specific binding of the antibody, uninjected embryo samples were analyzed in parallel. As a negative control, binding of myc-Fast1 to the Ef1α coding region, which is not expected to bind Fast-1, was also examined.
As shown in Figure 4, myc-Fast1 binds to the endogenous Goosecoid promoter, and not to Ef1α. While the signal from the Goosecoid promoter is high, the background from Ef1α is low, indicating that the binding of myc-Fast1 to the Goosecoid promoter is quite robust, as predicted. Similar results were obtained with injection of as little as 25 pg of myc-Fast-1 mRNA (data not shown). From these results, we conclude that myc-Fast-1 indeed binds to the endogenous Goosecoid promoter and that QPCR provides a sensitive and quantitative method for analyzing ChIP samples obtained using this protocol.
Figure 4. Quantitative PCR analysis of ChIP. Quantitative PCR for the Goosecoid promoter (gray bars) and Ef1α (white bars) normalized to uninjected embryo control (A) or quantified as a percentage of input DNA (B). Graphs represent average relative quantification for four independent experiments. An average of 45 one-cell embryos were injected with myc-Fast1 (250 pg) alone or in combination with Xnr1 (50 pg) and harvested at gastrula stage (stage 10.5) for ChIP analysis according to this protocol. QPCR was performed using SYBR green and relative quantification was performed using the ΔΔC(t) method. Error bars shown represent standard error.
Download figure to PowerPoint
Figure 4 presents two approaches to data normalization. Figure 4A represents the fold enrichment when experimental samples are normalized to uninjected control samples and Figure 4B shows the quantity of each immunoprecipitated target sequence as a percentage of total input DNA. In this experiment, PCR amplification was linear over a range of 5% (the highest amount tested) to 0.01% input material. By comparison to genomic DNA standards, this corresponds to a range between 65 and 0.1 ng genomic DNA (data not shown). Thus, in these experiments, single round QPCR is as sensitive as the nested PCR shown in Figure 2B. As such, we have not investigated whether QPCR is amenable to a nested PCR approach, but it is conceivable that a limited, initial (conventional) PCR amplification could enhance the sensitivity of a subsequent QPCR analysis.
Finally, β-catenin binds to chromatin indirectly through interaction with Tcf/Lef family members, raising the possibility that prolonged crosslinking (60 min) is only required for indirect binding proteins, whereas proteins that bind directly to DNA may be crosslinked more efficiently. We therefore evaluated the effect of crosslinking time on the recovery of Tcf3 at target gene promoters by ChIP. Tcf3 is a member of the Tcf/Lef family of transcription factors that is predicted to bind directly the promoters of β-catenin target genes during early embryogenesis (Molenaar et al.,1996). Embryos were injected with 25 pg Myc-tagged Tcf3 per blastomere at the two-cell stage and fixed at stage 10 for 15, 30, 45, or 60 min. Following ChIP, samples were analyzed by QPCR for recovery of promoter sequences for the β-catenin target genes Siamois and Xnr6. Tcf3 has been shown previously to bind the Siamois promoter in gastrula stage embryos, and thus serves as a positive control in this experiment (Park et al.,2005). Recovery of Xmlc2 was also measured as a negative control for Tcf3 binding. Figure 5 demonstrates that recovery of Tcf3-bound promoter sequences is maximal with 60 min of crosslinking time, and that crosslinking for 30 min or less leads to little or no recovery of target gene promoter sequences. Notably, crosslinking for up to 60 min does not result in a large increase of non-specific enrichment for the negative control locus. In addition, a 45-min crosslinking time is also sufficient to recover bound promoters (albeit less efficiently), with the advantage of slightly less signal from the negative control ChIP. These results underscore the observation in Figure 1A that extended crosslinking times are required to perform ChIP for DNA-bound transcription factors in early Xenopus embryos.
Figure 5. Effect of crosslinking times on ChIP. Anti-Myc-Tag ChIP was performed on sets of 37 gastrula stage (stage 10), non-injected or Myc-Tcf3 injected (50 pg) embryos, which were crosslinked for 15, 30, 45, or 60 min, as indicated. Samples were analyzed by QPCR for enrichment of Xmlc2 (-118), Xnr6 (-118), and Siamois (-303), represented as % Input recovery. Crosslinking times between 45 and 60 min are required for the specific enrichment of the promoters for both Siamois (black bars) and Xnr6 (hatched bars) by Myc-Tcf3 ChIP compared to the Xmlc2 promoter (gray bars). All non-injected (negative control) ChIPs yielded a comparatively low signal and are represented as white bars. Recovery of input material was similar for each time point.
Download figure to PowerPoint
In addition to the factors described above, several additional considerations should be made when designing a Xenopus ChIP experiment. Particularly, since embryos represent a heterogeneous population of cells, certain protein:DNA interactions may occur in only a small fraction of the experimental sample. This will result in an overall reduction in the number of available binding events for analysis. In cases where a minimal amount of starting material is used, this heterogeneity could lead to “false negative” results. Several approaches are available to circumvent this problem: increasing the amount of starting material, increasing the amount of material in the PCR analysis, and enriching for the subpopulation of cell types with the predicted protein:DNA interaction. Xenopus embryos are particularly well suited for this latter approach. For example, using explanted tissues containing the lineage of interest can enrich for particular cell types. Alternatively, molecular techniques for expanding cell lineages can be used to troubleshoot negative results. Finally, in certain cases (and also when a suitable antibody is unavailable), it may be possible to overexpress the DNA binding factor of interest throughout the embryo to increase the number of protein:DNA binding events available for analysis. However, care must be taken in the interpretation of overexpression experiments, particularly due to the difficulty in controlling the spatio-temporal expression of injected mRNAs. In this case, it may be possible to refine further overexpression approaches by driving tissue-specific expression of the DNA-binding factor of interest, either by plasmid DNA injection or transgenesis.
We have reported a ChIP assay protocol amenable to early Xenopus embryo research, demonstrated examples of its implementation, and outlined methods to optimize several aspects of the protocol. We strongly encourage repeating the optimization experiments shown in Figure 1 when different conditions or embryo stages are used, as we find this is the most critical aspect of the protocol in our hands. We expect that ChIP will become a useful tool within the Xenopus community, allowing for the identification of direct protein–DNA interactions, which, in the past, have been indirectly surmised, usually by plasmid-based reporter gene assays or by the combination of overexpressed, hormone-inducible chimeras and cycloheximide treatment. In addition, this technique will allow for many unanswered questions in developmental biology to be addressed, including the systematic analysis of gene regulatory networks and the investigation of the function of chromatin modifications during early vertebrate embryogenesis.