Transcription-independent TFIIIC-bound sites cluster near heterochromatin boundaries within lamina-associated domains in C. elegans

Background Chromatin organization is central to precise control of gene expression. In various eukaryotic species, domains of pervasive cis-chromatin interactions demarcate functional domains of the genomes. In nematode Caenorhabditis elegans, however, pervasive chromatin contact domains are limited to the dosage-compensated sex chromosome, leaving the principle of C. elegans chromatin organization unclear. Transcription factor III C (TFIIIC) is a basal transcription factor complex for RNA polymerase III, and is implicated in chromatin organization. TFIIIC binding without RNA polymerase III co-occupancy, referred to as extra-TFIIIC binding, has been implicated in insulating active and inactive chromatin domains in yeasts, flies, and mammalian cells. Whether extra-TFIIIC sites are present and contribute to chromatin organization in C. elegans remains unknown. Results We identified 504 TFIIIC-bound sites absent of RNA polymerase III and TATA-binding protein co-occupancy characteristic of extra-TFIIIC sites in C. elegans embryos. Extra-TFIIIC sites constituted half of all identified TFIIIC binding sites in the genome. Extra-TFIIIC sites formed dense clusters in cis. The clusters of extra-TFIIIC sites were highly over-represented within the distal arm domains of the autosomes that presented a high level of heterochromatin-associated histone H3K9 trimethylation (H3K9me3). Furthermore, extra-TFIIIC clusters were embedded in the lamina-associated domains. Despite the heterochromatin environment of extra-TFIIIC sites, the individual clusters of extra-TFIIIC sites were devoid of and resided near the individual H3K9me3-marked regions. Conclusion Clusters of extra-TFIIIC sites were pervasive in the arm domains of C. elegans autosomes, near the outer boundaries of H3K9me3-marked regions. Given the reported activity of extra-TFIIIC sites in heterochromatin insulation in yeasts, our observation raised the possibility that TFIIIC may also demarcate heterochromatin in C. elegans.

The genome of nematode Caenorhabditis elegans has served as a model to study chromatin organization [3,[16][17][18]. The highest level of chromatin organization in C. elegans is the chromatin feature that distinguishes between the X chromosome and the autosomes. The X chromosome in C. elegans hermaphrodites is organized into large self-interacting domains that have some features shared with topologically associated domains (TADs) seen in other metazoan genomes [19][20][21][22]. The five autosomes, however, lack robust self-interacting domains [22]. Instead, each autosome can be subdivided into three, megabase-wide domains, the left arm, the right arm, and the center [23]. The center domains display a low recombination rate [24,25], a high density of essential genes [26], and low heterochromatin-associated histone modifications [16,27]. The autosome arms are rich in repetitive elements [23] and heterochromatinassociated histone modifications [16,27], and are associated with the nuclear membrane [18,28,29]. Within these generally euchromatic centers and heterochromatic arms lie kilobase-wide regions of various chromatin states including transcriptionally active and inactive regions [3,4]. While condensins, a highly conserved class of architectural proteins [30], define the boundaries of TAD-like self-interacting domains in the X chromosome [22], the contribution of condensins to autosomal chromatin organization is unclear [14,31]. Furthermore, CTCF, another conserved architectural protein central to defining TAD boundaries in vertebrates, is thought to be lost during the C. elegans evolution [32]. How chromatin domains and chromatin states in the C. elegans autosomes are demarcated remains an area of active investigation [4,28,33,34].
The transcription factor IIIC complex (TFIIIC) is a general transcription factor required for recruitment of the RNA polymerase III (Pol III) machinery to diverse classes of small non-coding RNA genes [35]. TFIIIC has also been implicated in chromatin insulation [36,37]. TFIIIC binds DNA sequence elements called the Box-A and Box-B motifs [35]. When participating in Pol IIIdependent transcription, TFIIIC binding to Box-A and Box-B motifs results in recruitment of transcription factor IIIB complex (TFIIIB), including TATA-binding protein (TBP), which then recruits Pol III [35,38]. By mechanisms that remain unknown, however, TFIIIC is also known to bind DNA without further recruitment of TBP and Pol III [37]. These so-called "extra-TFIIIC sites", or "ETC", have been identified in various organisms including yeast [39,40], fly [41], mouse [42], and human [43]. In Saccharomyces cerevisiae and Schizosaccharomyces pombe, extra-TFIIIC sites exhibit chromatin boundary functions both as heterochromatin barriers and insulators to gene activation [40,44,45]. In addition, extra-TFIIIC sites in these yeast species have been observed at the nuclear periphery, suggesting a contribution to spatial organization of chromosomes [40,46]. In fly, mouse, and human genomes, extra-TFIIIC sites were found in close proximity to architectural proteins including CTCF, condensin, and cohesin [41][42][43]47]. These studies collectively suggest a conserved role for extra-TFIIIC sites in chromatin insulation and chromosome organization. However, whether extra-TFIIIC sites exist in the C. elegans genome is unknown.
In this study, we unveiled extra-TFIIIC sites in the C. elegans genome. Extra-TFIIIC sites were highly over-represented within a subset of autosome arms that presented a high level of heterochromatin-associated histone H3K9 trimethylation (H3K9me3). Extra-TFIIIC sites formed dense clusters in cis and were embedded in the laminaassociated domains. Despite the heterochromatin environment of extra-TFIIIC sites, the individual clusters of extra-TFIIIC sites were devoid of and resided near the boundaries of H3K9me3-marked regions. Our study thus raised the possibility that, like extra-TFIIIC sites in other organisms, C. elegans extra-TFIIIC sites may have a role in demarcating chromatin domains.

Half of C. elegans TFIIIC binding sites lack Pol III co-occupancy
The TFIIIC complex is a general transcription factor required for the assembly of the RNA polymerase III (Pol III) machinery at small non-coding RNA genes such as tRNA genes (Fig. 1a). Extra-TFIIIC sites are TFI-IIC-bound sites lacking Pol III co-occupancy, and are implicated in insulating genomic domains and spatially organizing chromosomes [48]. To determine whether the C. elegans genome includes extra-TFIIIC sites, we analyzed the ChIP-seq data published in our previous study [49] for TFIIIC subunits, TFTC-3 (human TFIIIC63/ GTF3C3 ortholog) and TFTC-5 (human TFIIIC102/ GTF3C5 ortholog) (Fig. 1b); the Pol III catalytic subunit RPC-1 (human POLR3A ortholog, "Pol III" hereafter); and the TFIIIB component TBP-1 (human TBP ortholog, "TBP" hereafter) in mixed-stage embryos of the wild-type N2 strain C. elegans. We identified 1029 high-confidence TFIIIC-bound sites exhibiting strong and consistent enrichment for both TFTC-3 and TFTC-5 (Fig. 1c). tRNA genes were strongly enriched for TFTC-3, TFTC-5, Pol III, and TBP as expected (Fig. 1d). We also observed numerous TFIIIC-bound sites with low or no Pol III and TBP enrichment (Fig. 1d). Of the 1029 TFI-IIC-bound sites (Additional file 1), we identified 504 sites (49%) with no or very low Pol III and TBP enrichment (Fig. 1e, f ), which we referred to as extra-TFIIIC sites, following the nomenclature in literature [39][40][41][42][43]. We also identified 525 TFIIIC-bound sites with strong Pol III enrichment (51%), which we referred to as Pol III-bound TFIIIC sites (Fig. 1e, f ).
The lack of Pol III and TBP binding in extra-TFIIIC sites may represent a premature Pol III preinitiation complex assembled at Pol III-transcribed non-coding RNA genes [50]. Alternatively, C. elegans extra-TFIIIC sites could be unrelated to Pol III transcription and similar to extra-TFIIIC sites reported in other organisms. To distinguish these two possibilities, we examined the presence of transcription start sites (TSSs) of non-coding RNA genes near TFIIIC-bound sites. We observed that only 4% of extra-TFIIIC sites (20/504) harbored TSSs of non-coding RNA genes within 100 bp of the TFIIICbound site center (Fig. 1g). In contrast, almost all Pol III-bound TFIIIC sites (464 of 525 sites, 88%) harbored TSSs of non-coding RNA genes within 100 bp, and the vast majority of these genes encoded tRNAs (376 sites, 72%) or snoRNAs (52 sites, 10%) (Fig. 1g) as expected [35]. Thus, extra-TFIIIC sites in the C. elegans genome are unlikely to participate in local Pol III-dependent transcription, a characteristic behavior of extra-TFIIIC sites reported in other organisms [37].

C. elegans extra-TFIIIC sites possess strong Box-A and Box-B motifs
The TFIIIC complex binds the Box-A and Box-B DNA motifs [35] (Fig. 1a). However, the majority of extra-TFI-IIC sites in yeast and human possess only the Box-B motif [39,40,43]. To determine whether C. elegans extra-TFI-IIC sites contain Box-A and Box-B motifs, we performed de novo DNA motif analyses at extra-TFIIIC sites. Almost all of the 504 extra-TFIIIC sites in C. elegans harbored both the Box-A and Box-B motifs (90% with Box-A, E = 1.9 × 10 −1568 ; 87% with Box-B, E = 1.5 × 10 −1602 ; Fig. 1h). The pervasiveness of these motifs in extra-TFI-IIC sites was comparable to that in Pol III-bound TFIIIC binding sites (94% with Box-A, E = 1.1 × 10 −589 ; and 92% with Box-B, E = 3.5 × 10 −1308 ) (Fig. 1h).
Because the Box-A and Box-B motifs constitute the gene-internal promoter for tRNA genes in eukaryotic genomes [35] (Fig. 1a), we hypothesized that extra-TFI-IIC sites correspond to genetic elements similar to tRNA genes. In C. elegans, a class of interspersed repetitive elements called CeRep3 has been suspected as tRNA pseudogenes [51]. To determine whether extra-TFIIIC sites coincide with repetitive elements, we surveyed the overlap between extra-TFIIIC sites and all annotated repetitive elements (Fig. 1i). Strikingly, 44.6% (225 sites) of extra-TFIIIC sites overlapped repetitive elements (permutation-based empirical P < 0.001), and almost all of the overlapped elements (95.1%; 214 sites) were the CeRep3 class of repetitive elements (Fig. 1i). In contrast, although significant, only 8.2% (43 sites) of Pol III-bound TFIIIC sites overlapped repetitive elements of any class (permutation-based empirical P < 0.001). Therefore, unlike extra-TFIIIC sites in yeast and humans, C. elegans extra-TFIIIC sites harbored both the Box-A and Box-B motifs; furthermore, a large fraction of these sites corresponded to a class of putative tRNA pseudogenes CeRep3.

C. elegans extra-TFIIIC sites are densely clustered in the distal arms of autosomes
The lack of robust association with local regulatory features of Pol II and Pol III transcription led us to hypothesize that extra-TFIIIC sites were related to large-scale organization of chromosomes as in the case of yeasts [40,53]. To test this hypothesis, we examined the distribution of extra-TFIIIC sites in the genome. We observed that extra-TFIIIC sites were highly overrepresented in chromosome V (195 of the 504 sites, 39%; permutation-based empirical P < 0.001), but strongly under-represented in the X chromosome (18 of the 504 sites, 4%; permutationbased empirical P < 0.001; Fig. 3a). In contrast, Pol IIIbound TFIIIC sites were highly over-represented in the X chromosome (202 of the 525 sites, 38%; permutationbased empirical P < 0.001) consistent with tRNA gene overrepresentation in the X chromosome [23].
C. elegans autosomes can be subdivided into three domains of similar size (left arm, center, and right arm) based on repetitive element abundance, recombination rates, and chromatin organization [17,23,25]. We found that most extra-TFIIIC sites were located in autosome arms (486 of the 504 sites, 96%; Fig. 3b). In addition, extra-TFIIIC sites were overrepresented in only one of each autosome's two arms that harbors the meiotic pairing centers [54] (overrepresented in the right arm of chromosome I; left arm of chromosome II; left arm of chromosome III; left arm of chromosome IV; right arm of chromosome V; permutation-based empirical P < 0.001; Fig. 3c). Furthermore, within autosomal arms, extra-TFI-IIC sites were locally densely clustered (Fig. 3d), with a median interval between neighboring extra-TFIIIC sites of 1207 bp (Mann-Whitney U test vs. within-arm permutations P = 2 × 10 −16 ; Fig. 3e). Among the autosome arms, the chromosome V right arm contained the largest number of extra-TFIIIC sites (188 sites) with extensive clusters (Fig. 3b, d, e). Thus, C. elegans extra-TFIIIC sites were fundamentally different from Pol III-bound TFIIIC sites in their genomic distribution and highly concentrated at specific locations within autosomal arm domains.

Extra-TFIIIC sites Extra-TFIIIC sites
Chr I  (Fig. 4e). Our analysis thus revealed that C. elegans extra-TFIIIC sites were located close to, but not overlapped with, H3K9me2 and H3K9me3-enriched regions within autosomal arm domains.

C. elegans extra-TFIIIC sites are located within nuclear membrane-associated domains
In S. pombe and S. cerevisiae, extra-TFIIIC sites are localized at the nuclear periphery and thought to regulate spatial organization of chromosomes [40,46]. In C. elegans, Pol III-transcribed genes including tRNA genes are associated with the nuclear pore component NPP-13 [49], similar to tRNA genes in S. pombe associated with the nuclear pores [57] (Fig. 5a). We hypothesized that extra-TFIIIC sites are associated with NPP-13, given the similarity of extra-TFIIIC sites to Pol III-transcribed genes. To test this hypothesis, we compared the locations of extra-TFIIIC sites with those of NPP-13-bound sites identified in mixed-stage embryos [49]. We observed that only 6 of the 504 extra-TFIIIC sites (1.2%) overlapped NPP-13-bound sites (Fig. 5b), in contrast to a large fraction of Pol III-bound TFIIIC sites (215 sites, 41%) that overlapped NPP-13-bound sites (permutation-based P < 0.001; Fig. 5b). Thus, extra-TFIIIC sites were not likely to be associated with the nuclear pore. Another mode of chromatin-nuclear envelope interactions in C. elegans is mediated by nuclear membraneanchored, lamin-associated protein LEM-2 [28] (Fig. 5a). LEM-2 associates with the large genomic regions called "LEM-2 subdomains" that occupy 82% of the autosome arms [28]. Between LEM-2 subdomains are non-LEM-2-associated "gap" regions of various sizes [28]. We therefore investigated the locations of extra-TFIIIC sites with respect to those of LEM-2 subdomains and gaps (Fig. 5c). Strikingly, 441 of the 504 extra-TFIIIC sites (88%) were located within LEM-2 subdomains, demonstrating overrepresentation in a statistical test that accounted for the overrepresentation of extra-TFIIIC sites within autosome arms (permutation-based P < 0.001, permutation performed within chromosomal center/arm domains) (Fig. 5c, d). In contrast, Pol III-bound TFIIIC sites were underrepresented within LEM-2 associated domains (permutation-based P < 0.001, permutation performed within chromosomal domains). Together, our results suggest that C. elegans extra-TFIIIC sites are localized at the nuclear periphery and intersperse H3K9me3-marked heterochromatin regions (Fig. 5e).

Discussion
In this paper, we identified genomic sites bound by transcription factor III C (TFIIIC), the basal transcription factor for RNA polymerase III (Pol III), that do not participate in Pol III-dependent transcription in C. elegans. Similar TFIIIC-bound sites devoid of Pol IIIdependent transcription, termed extra-TFIIIC sites, have been reported in yeast [39,40], fly [41], mouse [42], and human [43]. Our data demonstrated that half of all TFI-IIC-bound sites in C. elegans embryos lack Pol III binding, TBP binding, and nearby non-coding RNA genes, revealing pervasive extra-TFIIIC sites in the C. elegans genome.
Previous studies have suggested that extra-TFIIIC sites act as genomic insulators by blocking enhancer activity or heterochromatin spreading [40,56,58], or mediating three-dimensional genome interactions [41,58]. Some of the genomic and chromatin features of C. elegans extra-TFIIIC sites reported in this paper resemble characteristics of extra-TFIIIC sites participating in chromatin insulation. First, C. elegans extra-TFIIIC were densely clustered in cis, similar to clusters of TFIIIC-bound sites capable of insulating heterochromatin and enhancer activities in S. pombe and human cells [40,58]. Second, C. elegans extra-TFIIIC sites were located close to, but not within, H3K9me3-marked regions, similar to the observation that some extra-TFIIIC sites are located at the boundaries of heterochromatin [40,56,58]. Third, C. elegans extra-TFIIIC sites coincided with genomic regions known to be associated with nuclear membrane protein LEM-2 [28], similar to yeast extra-TFIIIC sites localized to the nuclear periphery [40,46]. These observations raise the possibility that C. elegans extra-TFIIIC sites may act as a chromatin insulator at the nuclear periphery.
There are also differences between C. elegans extra-TFIIIC sites and those in other organisms. First, C. elegans extra-TFIIIC sites possess both the Box-A and Box-B motifs, unlike yeast and human extra-TFIIIC sites that only possess the Box-B motif [39,43]. Second, C. elegans extra-TFIIIC sites were neither located near gene promoters nor associated with chromatin features of Pol II-dependent regulatory regions, unlike human, fly, and mouse extra-TFIIIC sites that are located near regulatory elements for RNA polymerase II (Pol II) transcription [41][42][43]. Third, C. elegans extra-TFIIIC sites are unrelated to CTCF binding, unlike human and mouse extra-TFIIIC sites that are located near CTCF binding sites [42,43], because the C. elegans genome does not encode CTCF [32]. Our study could thus offer an opportunity for The molecular mechanisms underlying the chromatin organization of C. elegans autosomes remain poorly understood. Unlike the X chromosome, which is organized into large self-interacting domains that have some features shared with topological associated domains (TADs) reported in other organisms, the autosomes do not present strong and pervasive self-interacting domains [22]. The condensin binding sites that could create boundaries between self-interacting domains in the X chromosome did not do so in the autosomes [14]. Several mechanisms for autosome chromatin organization have been proposed. These mechanisms include the antagonism between H3K36 methyltransferase MES-4 and H3K27 methyltransferase RPC2 that defines active versus repressed chromatin boundaries [4,34]; small chromatin loops emanating from the nuclear periphery that allow active transcription within heterochromatin domains [28]; and active retention of histone acetylase to euchromatin that prevents heterochromatin relocalization [33]. Our observation that extra-TFIIIC sites are highly overrepresented in autosome arms and cluster in cis near the boundaries of H3K9me3-marked regions warrant future investigation of whether TFIIIC proteins participate in chromatin organization in C. elegans autosomes.
How strategies to demarcate chromatin domains have evolved in eukaryotes remain unclear. In vertebrates, CTCF has a central role in defining TAD boundaries and is essential for development [11,59,60]. In D. melanogaster, CTCF is essential but does not appear to define TAD boundaries, and instead acts as a barrier insulator [61][62][63]. In the non-bilaterian metazoans, some bilaterian animals (such as C. elegans), plants, and fungi, CTCF orthologs are absent [32,64]. In contrast to CTCF, the TFIIIC proteins are conserved across eukaryotes [65] and extra-TFIIIC sites have been reported in human, [43], mouse [42], fly [41], C. elegans (this study), and yeast [39,40]. Whether extra-TFIIIC is an evolutionary conserved mechanism for demarcating chromatin domains in eukaryotes, including those lacking a CTCF ortholog, will be an interesting subject of future studies.

Conclusions
We identified TFIIIC-bound sites that do not participate in RNA polymerase III-dependent transcription in the C. elegans genome. These "extra-TFIIIC" sites were highly over-represented in the arm domains of the autosomes interacting with the nuclear lamina. Extra-TFIIIC sites formed dense clusters in cis near the outer boundaries of individual H3K9me3-marked heterochromatin regions. These genomic features of C. elegans extra-TFIIIC sites resemble extra-TFIIIC sites reported in other organisms that have activities in insulating heterochromatin. Our study warrants future investigation of whether TFIIIC proteins participate in heterochromatin insulation in C. elegans.

Reference genome
The ce10 reference sequence was used throughout. The chromosomal domains (left arm, center, and the right arm) defined by recombination rates [24] were used.

Gene annotation
The genomic coordinates and the types of C. elegans transcripts were downloaded from the WS264 annotation of WormMine. The WS264 genomic coordinates were transformed to the ce10 genomic coordinates using the liftOver function (version 343) with the default mapping parameter using the ce11ToCe10.over.chain chain file downloaded from the UCSC genome browser.

TFIIIC site definition
MACS2 [66] (version 2.1.0) identified 1658 TFTC-3-enriched sites. Of those, sites that had the TFTC-3 fold-enrichment (FE) score greater than 5, harbored TFTC-5-binding sites within 100 bp, and were located in the nuclear chromosomes were considered "highconfidence" TFIIIC-bound sites (1029 TFIIIC sites). The "center" of each TFIIIC site was defined by the position of the base with the largest TFTC-3 FE score. Of the 1029 TFIIIC sites, those with the maximum Pol III (RPC-1) FE greater than 20 within ± 250 bp of the site center were defined as "Pol III-bound TFIIIC" sites (525) and the remaining sites were defined as "extra-TFIIIC" sites (504). The genomic coordinates for Pol III-bound TFIIIC sites and extra-TFIIIC sites are listed in Additional file 1.

Heatmap
For the heatmaps around TFIIIC sites, a set of 20-bp windows with a 10-bp offset that covered a 2-kb region centered around the center of TFIIIC sites was generated for each site. For each window, the mean of fold-enrichment score was computed from replicatecombined input-normalized fold enrichment bedgraph files. The signals were visualized using the ggplot2's geom_raster function (version 2.2.1) in R.

Non-coding RNA genes
The genomic location and classification of non-coding RNA genes was described in "Gene annotation" section. For each TFIIIC site extended ± 100 bp from the site center, whether the region contained the transcription start site (TSS) of non-coding RNA genes was assessed using the Bedtools intersect function [67] (version 2.26.0).

DNA motif analysis
To find DNA motifs de novo, 150-bp sequences centered around the center of the TFIIIC-bound sites were analyzed by MEME (version 4.11.3) [68] with the following parameters: minimum motif size, 6 bp; maximum motif size, 12 bp; and the expected motif occurrence of zero or one per sequence (-mod zoops) and with the 1st-order Markov model (i.e., the dinucleotide frequency) derived from the ce10 genome sequence as the background.

Genomic intersection and permutation
Unless otherwise noted, the overlap between the 1-bp center of each of the TFIIIC sites and genomic features of interest (with size ≥ 1 bp) was assessed using the Bedtools intersect function [67] (version 2.26.0). To estimate the probability of observing the overlap frequency by chance given the frequency, location, and size of the features of interest and TFIIIC sites, the TFIIIC sites were permutated using the Bedtools shuffle function (version 2.26.0). The TFIIIC sites were shuffled across the genome, or within the chromosomes, or within chromosomal domains in which they reside, as described in each analysis section. After each permutation, the permutated set of TFIIIC sites were assessed for the overlap with the features of interest. This permutation was repeated 2000 times to assess the frequency at which the number of intersections for the permutated set of the TFIIIC sites was greater or less than the number of intersections for the observed TFIIIC sites. If none of the 2000 permutations resulted in the number of overlaps greater or less than the observed number of overlaps, the observed degree of overlaps was considered overrepresented or underrepresented, respectively, with the empirical P value cutoff of 0.001. The mean number of overlaps after 2000 permutations was computed for visualization.

Repetitive element analysis
The ce10 genomic coordinate and classification of repetitive elements, compiled as the "RepeatMasker" feature, were downloaded from the UCSC genome browser. The intersection between repetitive elements and TFIIIC sites was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within the chromosomes in which they resided.

Protein-coding gene distance
The genomic location of protein-coding genes was described in "Gene annotation" section. For each TFIIIC site, the absolute distance between the center of the TFI-IIC site and the closest TSS of a protein-coding gene was obtained using the Bedtools closest function [67] (version 2.26.0). To assess the probability of observing such distance distribution by chance given the frequency and location of the TSSs and TFIIIC sites, the TFIIIC sites were permutated once using the Bedtools shuffle function (version 2.26.0) such that the TFIIIC sites were shuffled within the chromosomes, and the distance between the permutated TFIIIC sites and closest protein-coding gene TSS was obtained. Mann-Whitney U test, provided by the wilcox.test function in R, was used to assess the difference of the distribution of the TFIIIC-TSS distances between groups.

TFIIIC sites in exons, introns, and 3′-UTRs
The genomic coordinates of exons, introns, and 3′-UTRs were extracted from WS264 gff3 file downloaded from wormbase. Exons, introns, and 3′-UTRs that belong to protein-coding genes (i.e., transcript type is "coding_ transcript") were processed. The genomic coordinates were transformed to the ce10 genomic coordinates as described above. The intersection between these features and TFIIIC sites was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within the chromosomal domains (see "Reference genome").

Chromatin state analysis
The chromatin state annotations and annotations of "active", "regulated", and "border" domains are reported previously [4]. The intersection between chromatin state annotations and TFIIIC sites was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within the chromosomal domains (see "Reference genome") to account for the difference of the chromatin state representation among different chromosomal domains.

Chromosomal distribution of TFIIIC sites
The number of TFIIIC sites in each chromosome was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed across the genome. The number of TFI-IIC sites in each chromosomal domain (see "Reference genome") was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within chromosomes.

Extra-TFIIIC site interval
For each chromosomal domain, the genomic distance between every pair of two neighboring extra-TFIIIC sites (center-to-center distance) was computed in R. To estimate the degree of closeness between extra-TFIIIC sites only explained by the frequency of extra-TFIIIC sites within chromosomal domains, the extra-TFIIIC sites were permutated 2000 times within chromosomal domains as described in "Genomic intersection and permutation" section. In each of the 2000 permutations, the genomic distance between two neighboring permutated extra-TFIIIC sites was computed, and the mean of the distances was computed. The distribution of the 2000 means (by 2000 permutations) was compared with the distribution of observed distribution of TFIIIC interval sizes by Mann-Whitney U test.

Analysis of H3K9me2 and H3K9me3 regions
To define H3K9me2-enriched and H3K9me3-enriched regions, the genome was segmented into 1-kb windows, and the mean fold-enrichment score of H3K9me2 and H3K9me3 (see ChIP-seq dataset) was computed for each window using the Bedtools map function (version 2.26.0). Windows with the mean fold-enrichment score greater than 2.5 (1.5× standard deviation above mean for H3K9me2; and 1.3× standard deviation above mean for H3K9me3) were considered enriched for H3K9me2 or H3K9me3 and merged if located without a gap. This yielded 2967 H3K9me2-enriched regions (mean size, 2.1 kb) and 1331 H3K9me3-enriched regions (mean size, 4.9 kb).
The intersection between H3K9me2-enriched or H3K9me3-enriched regions and TFIIIC sites was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within the chromosomal domains.
For each TFIIIC site, the absolute distance between the center of the TFIIIC site and the closest H3K9me2enriched and H3K9me3-enriched regions was obtained using the Bedtools closest function [67] (version 2.26.0). To assess the probability of observing such distance distribution by chance given the frequency, location, and size of H3K9me2-enriched and H3K9me3-enriched regions and TFIIIC sites, the TFIIIC sites were permutated once using the Bedtools shuffle function (version 2.26.0). For the analysis of the distance to H3K9me2enriched regions, this permutation was performed within the chromosomal domains. For the analysis of the distance to H3K9me3-enriched regions, permutation was performed with the chromosomal domains, but excluding the H3K9me3-enriched regions themselves because the TFIIIC sites were strongly underrepresented in the H3K9me3-enriched regions.

TFIIIC sites in NPP-13-binding sites
The 223 NPP-13-binding sites identified in mixed-stage N2-stage embryos are previously reported [49]. The genomic coordinates were converted to the ce10 genomic coordinates using the UCSCtools liftOver function (version 343). The intersection between NPP-13-binding sites (± 500 bp of NPP-13 binding site center) and TFIIIC sites was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within the chromosomal domains.

TFIIIC sites in LEM-2 subdomain and gaps
The LEM-2 subdomains and gaps between LEM-2 subdomains identified in mixed-stage N2-stage embryos are previously reported [28]. The genomic coordinates were converted to the ce10 genomic coordinates using the UCSCtools liftOver function (version 343). The intersection between LEM-2 subdomains or gaps of variable size classes and TFIIIC sites was assessed as described in "Genomic intersection and permutation" section. The permutation of TFIIIC sites was performed within the chromosomal domains.