CUT&Tag detects high H3K4me3 levels over gene promoters in caudal fin with strong reproducibility.
To establish baseline H3K4me3 patterns in adult fins, we performed CUT&Tag on cells harvested from 3 biological replicates of uninjured fins (Fig. 1A). For each replicate, we pooled cells dissociated from 6 uninjured fins, and each pool was divided in half for use in IgG control and H3K4me3 measurements. Similar to prior studies (5), high H3K4me3 levels were detected over gene promoter regions (Fig. 1B). After peak calling (see methods), we identified nearly 49-thousand sites of H3K4me3 enrichment and found there to be a high degree of correlation between replicates (Fig. 1C, S1A), demonstrating great consistency and reproducibility of this technique. Additionally, we observed a high degree of concordance in total CUT&Tag enrichment for H3K4me3 surrounding gene transcription start sites (TSS) (Figure S1B & S1C). These initial results demonstrate CUT&Tag to be reliable and consistent application for the study of epigenetic marks within the heterogeneous mixture of cells that constitute the zebrafish caudal fin (24).
Measurements of H3K4me3 by CUT&Tag are consistent with prior ChIP-Seq results.
We next compared enrichment of H3K4me3 detected by CUT&Tag with published enrichment measurements acquired by ChIP-Seq. Relative to ChIP-Seq, our CUT&Tag approach detected much higher promoter enrichment scores (RPKM – see methods), demonstrating the improved enrichment signal (as measured by RPKM) (Fig. 2A & 2B). To investigate whether CUT&Tag and ChIP-Seq measurements were similar at enriched loci, we merged replicates, ranked normalized signal independently across promoters or peak regions (to overcome method-specific enrichment differences), and then assessed overall correlations. Measurements at gene promoters were highly correlated when comparing between H3K4me3 CUT&Tag and ChIP-Seq (R = 0.72) (Fig. 2C, 2D, S2A). H3K4me3 CUT&Tag also exhibited high correlation (R = 0.83) with H3K27ac, an another histone modification known to be enriched at actively transcribed genes (25, 26). The observed correlation at promoters was much higher than at peak regions (R = 0.48) or at randomly generated background regions (Fig. 2C), which were uncorrelated (Figure S2B). Overall, these results demonstrate a high degree of consistency across replicates for each method, especially in the context of gene promoters (Fig. 2D).
Changes in H3K4me3 localization occur during early stages of caudal fin regeneration.
Tissue regeneration is achieved by differential expression of a substantial number of genes. To assess regeneration-associated changes in gene promoters, we next applied our CUT&Tag approach to regenerating fin tissues. We collected caudal fins at 2 dpa, a timepoint encompassing blastema formation, which is an essential event of fin regeneration (8), performed CUT&Tag against H3K4me3, and then intersected peaks identified independently for each timepoint. Comparison of H3K4me3 enriched peaks for uninjured (0 dpa) and regenerating (2 dpa) fins identified 29,152 shared peaks present in both samples (Fig. 3A & 3B). Peaks defined as “Common” had consistently elevated H3K4me3 levels across all timepoints and replicates. Peaks defined as “Uninjured” specific had higher H3K4me3 levels across all replicates of 0 dpa, as compared with 2 dpa samples, and peaks defined as “Regeneration” specific had higher H3K4me3 levels across all replicates of 2 dpa samples, as compared with 0 dpa (Fig. 3C & 3D). Interestingly, we found that common and uninjured specific loci were largely associated with binding motifs for FOX and KLF transcription factors, which are well known to have roles in embryonic development (27, 28). Loci classified as regeneration specific were largely associated with motifs for FOS transcription factor, a major component of AP-1 factor which play roles broadly in regenerative context, including zebrafish fins (Figure S3A) (29, 30).
To assess biological pathways associated with H3K4me3 enrichment, we performed the gene ontology (GO) analysis (Fig. 3E) (31). Common peaks tended to reside in close proximity to promoters of genes involved in cell metabolism (Fig. 3F). While uninjured specific peaks generally lacked associations, regeneration specific peaks were associated with embryonic development, morphogenesis, and differentiation (Fig. 3F). For instance, promoters for igfbp6b and lepb were enriched for H3K4me3 in 2 dpa samples. Interestingly, lepb is highly regulated upon fin amputation in zebrafish, and homologs to igfbp6 are known to be important for regeneration in other systems (10, 32). Additional examples include several genes previously described to have putative roles in fin regeneration (33–36) (Figure S3C). Overall, these analyses provide initial insight into the H3K4me3 changes that occur during zebrafish fin regeneration and highlight locations in the genome where epigenetic alterations occur.
Changes H3K4me3 levels correspond with moderate changes in chromatin accessibility.
Active gene promoters are often characterized by high levels of H3K4me3 and elevated chromatin accessibility (37, 38), leading us to explore whether changes in chromatin accessibility during the fin regeneration may accompany the observed H3K4me3 changes. To investigate this, we compared enrichment for H3K4me3 at 0 dpa and 2 dpa with previously published chromatin accessibility measurements at 0 dpa and 1 dpa obtained from ATAC-Seq analysis (7, 39). Initial comparisons of H3K4me3 enrichment at gene promoters (Fig. 4B & S4A) indicated a considerable amount of correlation between CUT&Tag and ATAC-Seq signal (Fig. 4A, 4B, S4A), analogous to associations observed in other biological systems (37, 38). We next utilized the previously classified H3K4me3 peaks regions to investigate similar changes in chromatin accessibility, relying on the aforementioned “common” peaks, as well as uninjured specific and regeneration specific loci. As anticipated, regions which gained H3K4me3 between 0 dpa and 2 dpa (classified as regeneration specific peaks) also become significantly more accessible between 0 dpa and 1 dpa (Fig. 4C). Accordingly, regions with lost H3K4me3 during regeneration (classified as uninjured specific) tended to become less accessible (p = 0.072). These results indicate that the majority of already accessible loci (including promoters) remain accessible during fin regeneration, and regions which gain H3K4me3 experienced a moderate but statistically significant increase in chromatin accessibility during regeneration.
H3K4me3 accumulates during fin regeneration over regions which possessed H3K4me3 in embryos.
Development-related GO terms are enriched in regeneration status samples (Fig. 3E), leading us to hypothesize that changes in H3K4me3 localization during fin regeneration might embody a “return” to embryonic chromatin patterns. To compare regeneration and development samples, we sought embryonic timepoint matching those of 2 dpa regenerating fins. Key transcription factors for appendage development and regeneration include the Msx family of homeodomain-containing transcription factors (40, 41). Upon fin amputation, msx1b (msxB) is strongly induced in blastema at 2 dpa (40, 41). A previous study reported that msx1b is transiently expressed in embryonic fin folds as msx1b transcript is uniformly detectable in caudal fin folds at 24 hours post-fertilization (hpf) but restricted to the distal cells at 36 hpf (40, 41). Given the strong and uniform expression pattern of msx1b at 24 hpf in caudal fin folds, we chose 24 hpf caudal fin fold as representative fin samples for development.
We amputated fin folds of ~ 200 embryos at 24 hpf and performed CUT&Tag with IgG and H3K4me3 antibodies. Despite performing measurements on drastically different staged samples, we observed remarkably similar H3K4me3 enrichment patterns at gene promoters in the 24hpf embryonic fin folds compared with regenerating caudal fins (Fig. 5A & S5C). Furthermore, correlation values resulting from comparisons of development and uninjured or regenerating caudal fin samples were only slightly lower (R = 0.82 and R = 0.86, respectively) than values obtained from comparisons between fin timepoints (Fig. 3A, R = 0.92), indicating that H3K4me3 patterns at gene promoters were not drastically different among sample types. This was not the case when we compared H3K4me3 patterns across peaks, which included many intergenic regions. Correlation between development and uninjured or regenerating fin samples was quite modest (R = 0.38 and 0.41, respectively) (Fig. 5A – right), indicating more substantial differences between tissues.
To explore these differences further, we partitioned peak regions with respect to enrichment for each sample type, enabling us to classify peaks as “shared”, when enrichment occurred across all sample types, or “specific”, when enrichment occurred specifically in development, uninjured, or regeneration samples (Fig. 5B & 5C). Remarkably, 35% of regions which acquired H3K4me3 during fin regeneration (5,055 peaks out of 14,369) also possessed H3K4me3 in development (24 hpf embryo samples), as compared with only 24% of regions that lost H3K4me3 (1,793 peaks out of 7,573). In further support of maintained H3K4me3 enrichment over genic loci (as in Figs. 3A & 5A), a relatively large portion of “shared” peaks occurred within gene promoters (21% of peaks). Whereas uninjured- and regeneration-specific peaks tended to occur more frequently over intergenic regions (Fig. 5D). GO analysis revealed that shared peaks were associated with “housekeeping” genes, loci possessing H3K4me3 in both regenerative fins (2 dpa) and in 24 hpf embryos were associated with developmental genes, and no significant ontology terms were identified for H3K4me3 peaks that were lost during fin regeneration (possessing H3K4me3 at 0 dpa but not at 2 dpa) (Fig. 5E). These results support a mechanism in which accumulation of H3K4me3 occurs during caudal fin regeneration over regions which previously possessed H3K4me3 at the earlier developmental timepoints (24hpf), including many developmentally regulated gene promoters.
Changes H3K4me3 levels at gene promoters are accompanied by gene expression changes.
As noted, high H3K4me3 levels are indicative of gene activation, and loss of H3K4me3 leads to gene expression reduction (38). We therefore investigated whether the observed CUT&Tag H3K4me3 changes during fin regeneration associated with altered gene expression patterns. For this analysis, we first categorized gene promoters based on changes in H3K4me3 levels between 0 dpa and 2 dpa (see methods). Promoters were categorized in a manner similar to our parsing of peak regions, classifying loci as common, uninjured-specific, and regeneration-specific (Figure S6A). In agreement with our prior measurements, chromatin accessibility levels remained mostly stable over promoters during regeneration, and we observed modest but statistically significant increases at 1 dpa for promoters which gained H3K4me3 (regeneration-specific) (Fig. 6A – red profiles & S6B). Changes in RNA transcript levels also followed a pattern highly similar to the observed changes in H3K4me3. Promoters which gained H3K4me3 had higher levels of RNA at 1 dpa compared with 0 dpa, and promoters which lost H3K4me3 experienced a decrease in RNA transcript levels over this same period (Fig. 6A – grey profiles & S6B). Additionally, promoters which acquired H3K4me3 during regeneration also exhibited higher levels of H3K4me3 and a greater abundance of RNA transcripts within 24hpf embryonic fin folds, as compared with promoters that lost H3K4me3 (Fig. 6A – brown and green profiles, respectively & S6C).
To confirm these results, we next parse promoters based on changes in RNA transcript levels, or changes in chromatin accessibility, and then assessed H3K4me3 patterns. For these measurements we again classified promoters using a strategy similar to the one we previously described for H3K4me3 (see methods). Interestingly, H3K4me3 levels increased at promoters which become more accessible, and decreased at loci which lost accessibility (Fig. 6B). In the context of gene expression, we observed a significant increase in H3K4me3 levels at genes which became more transcriptionally active during regeneration (from 0 dpa to 1 dpa) and H3K4me3 significantly decreased at gene promoters which underwent silencing (Fig. 6B). As in our comparisons with 24hpf embryonic fin folds, GO analysis revealed that promoters which maintained or experienced a decrease in H3K4me3 levels were associated with metabolism and housekeeping processes, whereas gene promoters which gained H3K4me3 associated with the developmental processes and establishment of embryonic morphology (Fig. 6C), such as kat7a and hoxc11a (43, 44). Examples of genes which acquire H3K4me3 during early fin regeneration post amputation and embryonic fin development included shha (45, 46) and foxm1 (47), and examples of genes associated with fin fold-specific H3K4me3 included tal1 (48) and sgk2a (49) (Fig. 6D) (24, 50–53).