N2- and HI-specific probes selectively mark their respective chromosomes

To distinguish maternally vs paternally derived chromosomes within a single nucleus, we focused on F1 hybrid offspring from crosses between divergent C. elegans strains (Fig. 1A). We chose N2, the commonly used laboratory strain, and the related HI as crossing partners for four reasons: (1) the two strains can interbreed; (2) they have been extensively characterized at the sequence level [32,33,34,35]; (3) HI is one of the most divergent C. elegans isolates from N2, with over 170,000 SNPs between the two strains as well as many insertions and deletions, which supplied regions across the two genomes to design strain-specific chromosome marking probes [32, 36,37,38]; (4) the two strains show a high level of synteny and are similar enough in sequence that a set of common chromosome tracing probes could be used to trace both N2 and HI chromosomes, keeping reagent costs low.

Fig. 1

Haplotype-resolved chromosome tracing in C. elegans. A Schematic of crossing experiments. B Schematic of probe design strategy. C Location of whole-chromosome V (ChrV) tracing library with 21 regions and N2 and HI libraries interspersed along ChrV. D Schematic representation of the imaging workflow using an automated imaging system (see “Methods”), and probe design used for the indicated steps. E DNA Fish on embryos from N2 and HI, respectively, using the whole ChrV library, N2 ChrV, and HI ChrV-marking libraries. Note that the N2 ChrV libraries cause a small punctual background staining in HI. Scale bar, 5µm. F DNA FISH on embryos derived from crosses between N2 hermaphrodites and HI males, using the whole ChrV library, N2 ChrV, and HI ChrV-marking libraries. The haplotypes are well distinguishable. Scale bar, 5µm. G Overlay of ChrV territory signal with HI marker (left) and N2 marker (right) in the embryo used in F. Scale bar, 5µm

In conventional chromosome tracing of C. elegans embryos, fluorescently labeled primary DNA FISH probes are hybridized to defined regions along a chromosome, such as the 22 regions along ChrV [26]. Due to a fluorophore present on each whole-chromosome tracing probe, when imaged en masse, the probes reveal the chromosome territory (Fig. 1D, panel 1). Next, fluorescent region-specific probes are hybridized sequentially to readout tails on the primary probes, to visualize individual locations along ChrV (Fig. 1D, panel 2). This method allows one to determine the structure of individual chromosomes in 3D by a molecular connect-the-dots approach.

We modified the chromosome tracing method in two ways for our strain-specific approach. First, in addition to the common whole-chromosome tracing probes, we hybridized the strain-specific N2 and HI probes to the F1 hybrid embryos. In contrast to the common whole-chromosome tracing probes, we generated the strain-specific probes without a fluorophore, but with tails to bind two secondary oligos. The same binding sites for two secondary oligos were present on all probes for N2, and for a different secondary oligo binding site on all probes for HI. This design provided a means to label the strain territory by on-microscope hybridization using the N2-specific secondary oligos and the HI-specific secondary oligos equipped with distinct fluorophores, during image acquisition. We note that restricting fluorophores to the secondary probes gave flexibility regarding the choice of fluorophore for each experiment and lowered the cost of probe synthesis. Second, some regions along ChrV had few possible strain-marking probes due to a lack of strain-specific insertions, and therefore we decided to increase the signal by equipping each N2- and HI-specific probe with two binding sites for secondary oligos. We imaged the N2 and HI markers at the end of the chromosome tracing experiment (Fig. 1D, panel 3). We note, however, that since fluorescent labeling of the strain markers relies on on-stage hybridization during image acquisition, the strain markers could be imaged any time after primary probe imaging.

We made use of previously annotated insertions and deletions within the N2 and HI genomes [32] to design probes that were specific for one strain and that, together, could distinguish HI and N2 chromosome territories (Fig. 1B). We focused on insertions larger than 1000 nucleotides (nts), which would accommodate a minimum of 33 potential 30-mer probes and provided a strong signal to noise ratio for DNA FISH. Across both genomes, each chromosome harbored a varying number of inserted regions >1000 nts, from a high of 172 on ChrV to a low of 37 on the X chromosome (Additional File 1: Figure S1A), where N2 contained more insertions of >1000 nts across all chromosomes compared to HI. The total length of inserted sequences >1000 nts was the highest for ChrV (Additional File 1: Figure S1B), and we therefore decided to focus on ChrV to illustrate the proof of concept. We designed suitable strain-marking probes to N2 and HI, using a previously described probe design method [26, 39] (see “Methods”). This approach resulted in 5577 probes for N2 ChrV and 1831 for HI ChrV, where each probe was unique to one locus and predicted to bind exclusively to one genome. The strain-marking probes could hybridize along the length of the respective chromosome and were interspersed with the shared tracing probes for ChrV (Fig. 1C & Additional File 1: Figure S1C).

To test the specificity of the N2 and HI probes and assure compatibility of the new probe sets with the shared tracing probe set, we hybridized all three probe sets to fixed homozygous N2 and HI (CB4856) embryos. As expected, the shared tracing probe set for ChrV enabled visualization of ChrV both in N2 and HI homozygous animals (Fig. 1E). HI-specific probes marked the HI chromosome without detecting a signal from N2. The N2 probes detected the N2 chromosome robustly. In addition, we observed a faint signal from the HI strain using the N2 strain probes (Fig. 1E). This signal likely corresponds to a small region within the HI genome that was not included in the Thompson HI genome, which we used for our probe design, but which was present in a later HI genome release, as revealed by a BLAST search [32, 38] (see “Methods”). Nevertheless, we found that this low signal did not interfere with image segmentation and chromosome classification, in subsequent experiments (see below). Future libraries could remove these sequences.

To assess if the strain-specific probe sets perform well with heterozygous embryos, we mated N2 mothers with HI fathers. The two chromosomes were clearly visible, and N2 and HI probes marked their respective chromosome, allowing us to determine the parent of origin for each chromosome (Fig. 1F). Similarly, when mating HI mothers with N2 fathers, haplotypes were clearly distinguishable (Additional File 1: Figure S1D). As expected, the N2 and HI signals overlapped partially with the shared ChrV territory signal (Fig. 1G), which reflects the overlap between the shared and strain-specific probe sets along ChrV. Together, it is clear which chromosome comes from which strain (Fig. 1F). We conclude that the N2 and HI probe sets can distinguish ChrV derived from N2 or HI.

N2 and HI chromosomes form barbells, but HI is more compact

N2 and HI are both C. elegans, but they harbor sequence differences, some of which are predicted to affect chromatin architecture [32, 38, 40]. For example, HI lacks the germline RNAi component ppw-1 [41] and the sperm-expressed, selfish genetic element peel-1 [42]. N2 and HI also differ in certain multi-gene families such as BATH factors (BTB/POZ + MATH domains) and nuclear hormone receptors [32]. The effects of all but ppw-1 are unknown. Since experimentally introduced siRNAs are sufficient to induce chromatin compaction, it was possible that the lack of germline RNAi in HI strains might influence chromosome conformation [41, 43]. We therefore wanted to investigate if these two strains exhibited similar genome structures by comparing chromosomes from N2 or HI homozygous embryos.

One method to assess a chromosome’s compaction and 3D path in nuclear space is to compare the genomic (2D) and spatial (3D) pairwise distance measurements using power-law fitting [23, 26, 44]. Power-law fitting takes into account the polymer nature of the chromosome (where regions close in genomic distance are expected to be close in spatial distance, and vice versa). We can derive metrics from the fit of the raw data to describe the polymer step size (compaction) and scaling exponent (3D path). The step size is the average distance between the points along the chromosome. A smaller step size indicates a more compact chromosome. The scaling exponent of a random-walk polymer is expected to be 0.5, while an ideal highly crumpled (fractal globule) polymer where all spatial distances are proportional to genomic distances has a scaling exponent of 0.3. Our prior studies found that C. elegans chromosomes had a scaling exponent of ~0.2, which reflects more intermixing between points with large genomic distances [26].

We examined homozygous N2 or HI embryos that were at or below the 40-cell stage (Additional File 1: Figure S2A). We found that the newly derived N2 conformation was virtually identical to the previously published average N2 configuration for ChrV [26], with a comparable step size (1.030 vs. 1.037) and scaling exponent (0.198 vs. 0.193) when fitted to a power-law function (Fig. 2A). The nascent B compartments (or “chromosome arms” at TADs 1-7 and 18-22) exhibited long-range folding, whereas the nascent A compartment, encompassing TADs 8-17, was more extended (Fig. 2B). This result revealed the reproducibility of chromosome tracing in C. elegans, with little batch-to-batch variation between independent studies.

Fig. 2

Highly similar N2 and HI ChrV organization. A Power-law fits of mean pairwise distances in µm versus genomic distance in Megabases (Mb) for N2 data generated in this study (black) and previously (red) [26]. s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. B ChrV mean distance matrix for N2 (left) and HI (right) in early embryos (2–40 cells), colored by distance (in μm). N = 6072 (N2) & N = 5905 (HI). C Power-law fits of mean pairwise distance for N2 data (magenta) and HI data (green). s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. D Normalized spatial distances for N2 date (left) and HI data (right). Data is plotted as observed over expected. The expected spatial distance is determined by the fit to the power-law function shown in C. Red marks regions that are closer together than expected from the fit and blue regions that are further away. The right panel shows the significant mean changes. E Changes between normalized spatial distances of N2 and HI data (left) and p-value of the spatial distance changes. Red marks regions that are closer together than expected from the fit and blue regions that are further away. F Mean pairwise distances in µm of subpopulations of chromosome conformations for N2 and HI traces as determined by unsupervised clustering. N2 and HI traces were pooled as outlined in H. G Distribution of traces within clusters from F shows both strains have similar cluster frequencies. Chi-square statistic was 29.94, p-value = 5.03 × 10−6 and Cramér’s V = 0.08. H Strategy for co-clustering to determine statistical similarity. Axis numbering in B–F represents positions across the chromosome as shown in Figure 1C

The HI chromosome showed similar overall properties to N2, with compacted chromosome arms and an extended center (Fig. 2B). HI ChrV was slightly more compact than N2, as revealed by its smaller step size in power-law fitting (1.015 for HI vs. 1.030 for N2) (Fig. 2C); however, these differences were not statistically significant. N2 and HI also had differences in their scaling coefficients (0.198 for N2 and 0.187 for HI), which means that increasing genomic distances produced lower growth of spatial distances in HI compared to N2. These differences between N2 and HI may reflect the smaller size of the HI genome, which is ~2 Mb shorter than the N2 genome or approximately 2% of the total [32]. In addition, the relative size difference between chromosomes from HI and N2 is largest for ChrV, with HI ChrV being 741kb shorter than the N2 ChrV, representing a difference of 3.5% [32]. The differences in scaling coefficient and step size between N2 and HI is unlikely to reflect differences in the size of interphase nuclei. When we measured the diameters of nuclei of embryos between 4 and 8 cell stages, contrary to what we might expect from the differences in step size and scaling coefficient, N2 nuclei were slightly smaller on average compared to HI (5µm compared to 5.15µm; Additional File 1: Figure S2B).

Power-law fitting not only reveals the folding properties of chromosomes, but also serves as a means to normalize the spatial distance measurements by taking into consideration the polymer nature of the chromosomes [23, 26, 44]. Normalization of the N2 and HI spatial distances revealed general similarities in folding complexities between HI and N2 (Fig. 2D). For example, the distances within the left and right arms were smaller than expected by the power-law function for both N2 and HI, suggesting these regions were highly folded. In addition, the distances between the center and the right arm were larger than expected, consistent with a barbell configuration for both strains, by population average analysis [26]. Despite the overall similarity between both strains, they showed some differences in pairwise distances, as revealed by the normalized significant changes between N2 and HI (Fig. 2E).

Previously, we observed that individual chromosomes in vivo assumed conformations that were distinct from that of the population average [26]. The prevalent folding patterns can be revealed by unbiased clustering, which we undertook here. To do so, we pooled N2 and HI, adjusted for the difference in chromosome sizes and performed clustering on the mix (Fig. 2H, “Methods”). Co-clustering of pooled N2 and HI traces ensured consistency and comparability between the data for statistical purposes. We used a chi-square test for independence to compare the distribution of traces into clusters. Chi-square statistics assess the association between the type of chromosome (N2 vs HI) and clusters, while Cramer’s V indicates the strength of the association while accounting for the influence of large sample size.

We found similar clusters of traces for each strain (Fig. 2F). These consisted of chromosomes with one arm highly folded, the other arm folded, chromosomes with both arms folded or distended chromosomes with little folding. These resemble the configurations seen previously for homozygous N2 [25]. The distribution between clusters was comparable, even though it was statistically significant (p-value = 5.03 × 10−6), likely because of the very large number of traces analyzed. However, the magnitude of this association was very small, as measured by Cramér’s V (0.08 where 0 represents independence and 1 complete association). We conclude that N2 and HI ChrV show an overall similar structure to each other at the Megabase scale.

A new pipeline to analyze N2:HI hybrids

Our next goal was to examine N2 and HI chromosomes after interbreeding. First, we developed and implemented a new image segmentation and tracing pipeline. Like previous chromosome tracing [26], we applied watershed segmentation on the nuclear signal (DAPI staining) to restrict the definition of chromosome territories to the nuclear volumes and remove any background signal from outside the nuclei (Fig. 3A, step 1). Watershed segmentation is a method of image processing that automatically separates individual elements in the foreground (e.g., nuclei) from the background (e.g., cytoplasm). We implemented watershed segmentation using MATLAB as performed previously [26] (https://www.mathworks.com/help/images/marker-controlled-watershed-segmentation.html).

Fig. 3

HI paternal chromosomes decompact when subjected to the N2 maternal environment. A Example of Z-projections of raw data collected during haploid-specific chromosome tracing and schematic representation of image segmentation, tracing and sorting of traces. Scale bar, 5µm. B ChrV mean distance matrix for traces derived from hybrid embryos (2–40cells) from crosses between N2 hermaphrodites and HI males, N2 maternal traces (left) and HI paternal traces (right), colored by distance (in μm). N = 1384 (N2m) & N = 1066 (HIp). Axis numbering represents positions across the chromosome as shown in Figure 1C. C Power-law fits of mean pairwise distance for N2m data (magenta) and HIp data (green). s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. D Normalized mean spatial distances for N2m and HIp traces shown as observed over expected spatial distances as determined by the power-law fit in C. Red denotes regions that are closer together than expected from the fit and blue regions that are further away. E Differences in normalized mean spatial distances between N2m and HIp traces (left) and p-value (right). Red denotes regions that are closer together than expected from the fit and blue regions that are further away. F Power-law fits of mean pairwise distance for N2m data (magenta) and N2 homozygous data (black). s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. G Power-law fits of mean pairwise distance for HIp data (green) and HI homozygous data (black). s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. H Differences of mean pairwise distances in µm (left), and p-value (right) of N2m and N2 homozygous (top) and HIp and HI homozygous (bottom). Red marks regions that are closer together than expected from the fit and blue regions that are further away. Axis numbering in B, D, E, and H represents positions across the chromosome as shown in Fig. 1C for all panels

Next, watershed segmentation was applied to images of chromosome territories. This step defined the volumes in which chromosomes were traced using a nearest neighbor approach (Fig. 3A, step 2; [26]). The underlying assumption of this approach is that region “n” along the chromosome connects to the closest focus in 3D space that was detected for region “n+1,” and not a more distant “n+1” focus, which we assume belongs to another chromosome. This approach agrees well with simulated polymer models [45]; however, there is no absolute way to assess what the “real” chromosome path is, and therefore it is not possible to calculate an error rate.

The strain-marking territories were segmented for N2 and HI, and the resulting volumes overlayed with the traces generated in the previous step (Fig. 3A, step 3). Since the primary ChrV probes did not overlap perfectly with the strain-marking probes (Fig. 1G), we classified traces into N2 or HI based on whether the majority of regions of traces were located within or closest to a strain-marking volume for N2 or HI. Therefore, we calculated the smallest Euclidian distance for each region within a trace to the boundary points of the strain-marking volumes for N2 and HI. If the region was based on this calculation closer to N2, it was classified as N2 and vice versa for HI. To account for wrongfully segmented and traced chromosomes in the high-throughput analysis, we added several drop-out criteria for traces, which were excluded from further analysis: (i) the presence of more than 4 traces in one nucleus or (ii) the presence of more than 2 traces per strain per nucleus. These situations are biologically impossible in a wild-type setting and likely reflect over segmentation that has split a territory (chromosome) into two.

The paternal chromosome mimics the maternal conformation in N2m x HIp hybrids

We used our new tracing pipeline on embryos derived from crosses between N2 hermaphrodites (N2m) and HI males (HIp). The overall conformation of ChrV in these hybrids agreed well with the homozygous conformations, with compacted arms and more open centers (Fig. 3B). Power-law fitting revealed that both chromosomes showed very similar genomic distances vs. spatial distance relationships with each other, and these resembled N2 homozygotes (Fig. 3C,F,G). Only a minority of pairwise distance changes were significant between N2 maternal (N2m) and HI paternal (HIp) chromosomes, indicating that they were highly similar overall (Fig. 3E).

Despite the similarities, power-law fitting revealed that the N2mand HIp chromosomes differed slightly in step size (1.05 vs. 1.07) and scaling exponent (0.198 vs. 0.186). The scaling exponents were unchanged for HI and N2 chromosomes with respect to their homozygous conformations, but the step size for HIp chromosomes increased, from 1.015 in homozygotes to 1.072 in the N2m background, whereas it remained virtually unchanged for N2m compared to N2 homozygotes. Thus, the HIp chromosome became more similar to the N2 chromosome with regard to step size. Closer examination of mean pairwise distance changes of N2m and HIp chromosomes (compared to homozygous chromosomes from un-crossed embryos), revealed that substantially more regions changed in HI than in N2. Almost all these regions decompacted in HIp chromosomes (red, Fig. 3H). Statistically, N2 maternal (N2m) showed no statistical difference with N2 homozygotes, while the step size of HI paternal (HIp) was statistically different from HI homozygotes. This result reveals that the HIp chromosome decompacts when subjected to the N2m environment and implies that the paternal chromosome was influenced by the maternal environment.

To test the influence of the maternal HI (HIm) environment, we performed the reciprocal cross between HI hermaphrodites and N2 males. Again, we found that the overall conformation agreed well with HI and N2 homozygotes, with compacted arms and more open centers (Fig. 4A, C). Power-law fitting showed differences between N2 paternal (N2p) and HIm chromosomes in the scaling exponent (0.200 vs. 0.164) but the N2 value resembled the homozygous N2. As before, only a minority of N2p and HIm chromosomes’ pairwise distance changes were significant (Fig. 4E). This result confirms that HI and N2 chromosomes behave similarly within hybrids.

Fig. 4

N2 chromosomes influence HI chromosomes in trans. A ChrV mean distance matrices for traces derived from hybrid embryos (2–40cells) from crosses between HI hermaphrodites and N2 males, HI maternal traces (left), and N2 paternal traces (right), colored by distance (in μm). N = 1125 (HIm) & N = 1254 (N2p). Axis numbering represents positions across the chromosome as shown in Figure 1C. B Power-law fits of mean pairwise distance for N2p data (magenta) and HIm data (green). s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. C Normalized mean spatial distances for N2p and HIm traces shown as observed over expected spatial distances as determined by the power-law fit in B. Red denotes regions that are closer together than expected from the fit and blue regions that are further away. D Power-law fits of mean pairwise distance for N2p data (magenta) and N2 homozygous data (black). s = scaling exponent of spatial distance, a = step size. Values in brackets reflect the 95% confidence intervals of the fit. P-values were acquired from a linear regression analysis of log-transformed data. E Differences in normalized mean spatial distances between N2p and HIm traces (left) and p-value (right). Red marks regions that are closer together than expected from the fit and blue regions that are further away. F Power-law fitting of mean pairwise distance for HIm data (green) and HI homozygous data (black). s = scaling exponent of spatial distance, a = step size. Values in brackets are 95% confidence intervals of the fit. P-values are acquired from a linear regression analysis of log-transformed data. G Differences of mean pairwise distances in µm (left), and significant mean changes (right) of N2p and N2 homozygous (top) and HIm and HI homozygous (bottom). Red marks regions that are closer together than expected from the fit and blue regions that are further away. Axis numbering in A, C, E, and G represents positions across the chromosome as shown in Fig. 1C for all panels. Axis numbering represents positions across the chromosome as depicted in Fig. 1C

Despite the similarities, the values indicate that the paternal (N2) chromosome decompacted compared to the homozygous (N2) chromosome as indicated by the increased step size (1.07 vs. 1.03; Fig. 4D). The HIm chromosome changed in two ways compared to the HI homozygous chromosome, first by a decreased scaling exponent (from 0.19 to 0.16) and second by an increased step size, and the result was highly significant (1.015 to 1.131; Fig. 4F). This result suggests that each chromosome was subtly changed in this particular hybrid, with the N2 influencing HI. When we compared mean pairwise distance changes of HIm and N2p chromosomes with their homozygous counterparts, we detected significant changes for both chromosomes. Consistent with the values from power-law fitting, most regions became decompacted (Fig. 4G).

Taken together these data suggest that paternal chromosomes are influenced by the maternal environment in both crosses. Since we also find that HIm chromosomes change in the presence of N2p, we hypothesize that the N2 chromosomes also influence HI chromosomes in trans, while N2 chromosome structure seems to be more resistant to influences by the HI chromosome.

Cluster analysis reveals a new domain in HIm x N2p crosses

We performed unbiased cluster analysis on the mated strains. To enable statistical comparisons, we pooled the chromosome traces together and examined their clustering behavior for all crossed animals (HIm, N2p, N2m, HIp). In this experiment, we detected the prevalent folding patterns seen before, but the enrichment varied depending on the cross. The most noticeable difference was observed for the HIm x N2p cross, where Cluster 5 contained a small, compact structure or domain located approximately 6–12 Mb along the chromosome, with sharp boundaries relative to neighboring sequences. This cluster constituted 6% of the total in the HIm x N2p cross, but <1% in the reciprocal N2m x HIp cross, and was not observed in clusters of the original N2 or HI homozygous strains even when finely resolved to 11 clusters (Figure 2, data not shown, [26]). We calculated the standardized residuals for HIm and N2p as 3.70 and 4.74 respectively, indicating a strong enrichment (Fig. 5C). Conversely, the values for N2m and HIp were −4.70 and −3.55 indicating a deficiency of this configuration (Fig. 5C). Thus, crosses can engender or enrich for new chromosome configurations. The HIm x N2p cross also exhibited a reduction in Cluster 4, with large-scale looping along the right arm. The most prevalent cluster was an extended chromosome, seen in all four chromosomes (N2m, N2p, HIm, HIp). These data suggest that while the cross of HIp males with N2m mothers preserves the expected categories, the reciprocal cross altered large-scale folding along the chromosome. Furthermore, the differences in clustering between crosses is in line with increased differences seen in mean pairwise distance changes in crosses of HI mothers with N2 males (Figure 4).

Fig. 5

HIm and N2p chromosomes show altered large-scale folding. A Matrices of mean pairwise distances in µm for subpopulations of chromosome conformations for crosses as determined by unsupervised clustering. Traces from both crosses were pooled and analyzed together as described in Figure 2H. Missing data are marked in gray. Axis numbering represents positions across the chromosome as shown in Figure 1C. B Distribution of traces within clusters from A shows all crossed strains have similar cluster proportions, except for cluster 5 which is enriched in HIm and N2p traces. Chi-square statistic was 138.209, p-value= 0 and Cramér’s V=0.120. C Standardized residual for chi-square testing to reveal conformations that are enriched (red) or deficient (green) relative to expectation

Homologous chromosomes do not align

Transcriptional regulation depends on cis-regulatory sequences that are typically adjacent to target promoters [27, 46]. In Drosophila and other Dipterans, homologous chromosomes are paired in interphase somatic cells, allowing for interchromosomal interactions between enhancers and promoters, a phenomenon termed transvection [47]. Sequences that promote transvection stabilize the association between alleles, leading to a significant proportion of aligned homologs within 300–400nm of one another (e.g., 20% for gypsy) [48]. Methods to study physical interactions