You are on page 1of 22

Article

Reprogramming of Meiotic Chromatin Architecture


during Spermatogenesis
Graphical Abstract Authors
Yao Wang, Hanben Wang,
Yu Zhang, ..., Qinghua Shi, Xin Wu,
Wei Xie

Correspondence
xinwu@njmu.edu.cn (X.W.),
xiewei121@tsinghua.edu.cn (W.X.)

In Brief
Sexual reproduction is one of the most
fascinating phenomena in biology. During
spermatogenesis, spermatogonia give
rise to spermatozoa through a series of
highly orchestrated developmental and
meiotic events. By applying sisHi-C to
rhesus monkey and mouse models, Wang
et al. elucidated remarkable chromatin
architecture reprogramming during
spermatogenesis and meiosis.

Highlights
d Hi-C analysis of meiotic chromatin architecture during
spermatogenesis

d TADs and compartments A and B dissolve and then re-


appear during spermatogenesis

d Pachytene chromatin has highly refined transcription-


correlated compartments

d Inactive X chromosome during MSCI shows unique


chromatin architecture

Wang et al., 2019, Molecular Cell 73, 547–561


February 7, 2019 ª 2018 Elsevier Inc.
https://doi.org/10.1016/j.molcel.2018.11.019
Molecular Cell

Article

Reprogramming of Meiotic Chromatin


Architecture during Spermatogenesis
Yao Wang,1,5 Hanben Wang,2,5 Yu Zhang,1,5 Zhenhai Du,1,5 Wei Si,3,5 Suixing Fan,4 Dongdong Qin,2 Mei Wang,2
Yanchao Duan,3 Lufan Li,2 Yuying Jiao,4 Yuanyuan Li,1 Qiujun Wang,1 Qinghua Shi,4 Xin Wu,2,* and Wei Xie1,6,*
1Center for Stem Cell Biology and Regenerative Medicine, MOE Key Laboratory of Bioinformatics, School of Life Sciences, THU-PKU Center

for Life Science, Tsinghua University, Beijing 100084, China


2State Key Laboratory of Reproductive Medicine (SKLRM), Nanjing Medical University, Nanjing, Jiangsu 210029, China
3Yunnan Key Laboratory of Primate Biomedical Research, Institute of Primate Translational Medicine, Kunming University of Science and

Technology, Kunming 650500, China


4The First Affiliated Hospital of USTC, USTC-SJH Joint Center for Human Reproduction and Genetics, Hefei National Laboratory for Physical

Sciences at Microscale, The CAS Key Laboratory of Innate Immunity and Chronic Diseases, School of Life Sciences, CAS Center for
Excellence in Molecular Cell Science, Collaborative Innovation Center of Genetics and Development, University of Science and Technology of
China, Hefei 230027, China
5These authors contributed equally
6Lead Contact

*Correspondence: xinwu@njmu.edu.cn (X.W.), xiewei121@tsinghua.edu.cn (W.X.)


https://doi.org/10.1016/j.molcel.2018.11.019

SUMMARY gene expression and cell division (Bickmore, 2013; Gibcus and
Dekker, 2013; Gorkin et al., 2014). For example, heterochromatin
Chromatin organization undergoes drastic reconfi- is preferentially located in close proximity to the lamina of the nu-
guration during gametogenesis. However, the mo- clear membrane, while euchromatin is often positioned in the nu-
lecular reprogramming of three-dimensional chro- clear inner space (Amendola and van Steensel, 2014). At finer
matin structure in this process remains poorly scales, cis-regulatory elements such as enhancers can activate
understood for mammals, including primates. Here, promoters over long distances (Smallwood and Ren, 2013). The
development of the chromosome conformation capture tech-
we examined three-dimensional chromatin architec-
nique (3C) and its derivatives allow the detection of chromatin in-
ture during spermatogenesis in rhesus monkey using
teractions in three dimensions, which revealed key principles for
low-input Hi-C. Interestingly, we found that topolog- genome packaging (Gibcus and Dekker, 2013). For instance,
ically associating domains (TADs) undergo dissolu- contact domains were widely found in the genomes of many spe-
tion and reestablishment in spermatogenesis. Strik- cies as large self-interacting regions (Rowley and Corces, 2016),
ingly, pachytene spermatocytes, where synapsis including topologically associating domains (TADs) (Dixon et al.,
occurs, are strongly depleted for TADs despite their 2012; Lieberman-Aiden et al., 2009; Nora et al., 2012; Rao et al.,
active transcription state but uniquely show highly 2014; Sexton et al., 2012). TADs are believed to constrain the
refined local compartments that alternate between functions of regulatory elements within the domains (Rao et al.,
transcribing and non-transcribing regions (refined- 2014), and disruption of TAD boundaries is associated with aber-
A/B). Importantly, such chromatin organization is rant gene expression (Flavahan et al., 2016; Lupiáñez et al.,
2015; Valton and Dekker, 2016). TAD boundaries are preferen-
conserved in mouse, where it remains largely intact
tially occupied by CTCF, which can provide anchors for cohe-
upon transcription inhibition. Instead, it is attenuated
sin-mediated loop extrusion (Dekker et al., 2013; Dekker and
in mutant spermatocytes, where the synaptonemal Mirny, 2016). This leads to strong interactions between two
complex failed to be established. Intriguingly, this is anchoring CTCF sites, forming chromatin loops (Rao et al.,
accompanied by the restoration of TADs, suggesting 2014). On the other hand, chromatin is highly compartmental-
that the synaptonemal complex may restrict TADs ized, as regions with similar chromatin states preferentially
and promote local compartments. Thus, these data interact with each other (Lieberman-Aiden et al., 2009; Rao
revealed extensive reprogramming of higher-order et al., 2014). Two major compartments, A and B, are frequently
meiotic chromatin architecture during mammalian observed, which largely correspond to gene-dense regions
gametogenesis. and gene deserts, respectively. Notably, both TADs and chro-
matin compartments are restricted to interphase and are not de-
tected in mitosis. Instead, mitotic chromatin appears to adopt a
configuration with linearly compressed arrays of loops folded at
INTRODUCTION presumably random positions (Gibcus et al., 2018; Nagano et al.,
2017; Naumova et al., 2013).
In eukaryotes, the proper three-dimensional folding of chromatin Chromatin undergoes drastic reprogramming during meiosis,
fiber is crucial for DNA-based biological processes, including including spermatogenesis (Tang et al., 2016; Trasler, 2009;

Molecular Cell 73, 547–561, February 7, 2019 ª 2018 Elsevier Inc. 547
Zamudio et al., 2008). Homologous chromosome pairing and sequenced 1.6 billion to 2.7 billion Hi-C reads, resulting in 511
recombination occur at the meiotic prophase stage, followed million to 913 million pairwise chromatin contacts (Table S1).
by segregation of homologous chromosomes in meiosis I and The interaction intensities across the genome are highly repro-
cell division to produce haploid spermatids in meiosis II (Handel ducible between replicates and between our fibroblast and
and Schimenti, 2010). Later, the majority of histones are first re- those previously published (Darrow et al., 2016) (Figures S1C–
placed by transition proteins and then by protamine when round S1E). Consistent with the finding in mouse (Battulin et al.,
spermatids and elongated spermatids form (Rathke et al., 2014; 2016), sperm shows the strongest long-range interactions
Sassone-corsi, 2002). The prophase can be further separated (>10 Mb) (Figure 1B, left, for each stage). This is also confirmed
into leptotene, zygotene, pachytene, and diplotene substages by P(s) analysis (Figure 1C), which measures the probability of in-
(Cobb and Handel, 1998). The homologous chromosomes start teractions between bin pairs within defined distances across the
to align at the leptotene stage, and the synapsis is initiated at genome (Naumova et al., 2013). Such long-range interactions
the zygotene stage. The synaptonemal complex forms at the decreased at other stages, with PAC showing the least long-
pachytene stage, before the chromosomes undergo desynapsis range interactions (Figures 1C and S1D). In addition, PAC has
at the diplotene stage. Defects in meiosis often result in aneu- relatively lower contact probability below 1 Mb but higher con-
ploidy and embryonic lethality (Handel and Schimenti, 2010). tact probability between 1 and 10 Mb (described later). It was re-
Notably, starting at the zygotene-to-pachytene transition, the ported that metaphase chromatin interactions can be modeled
sex chromosomes form a spatially segregated chromatin as polymers using a P(s)  s0.5 curve (Gibcus et al., 2018; Nau-
territory called the XY body or sex body (Handel, 2004). The mova et al., 2013). Strikingly, we found that the PAC chromatin
X and Y chromosomes show homologous synapsis only in the P(s) curve is very similar to P(s)  s0.5 up to 6 Mb, after which
small pseudoautosomal region (PAR). The unsynapsed regions the interactions decreased sharply (Figures 1C and S1D). Such
of X and Y chromosomes are subjected to male-specific tran- an interaction insulation boundary was proposed to reflect the
scription silencing called meiotic sex chromosome inactivation interaction unit in a linearly organized array of consecutive loops
(MSCI) at the pachytene stage (Turner, 2007). Silencing of the for metaphase chromatin (Gibcus et al., 2018; Naumova et al.,
sex chromosomes is then partially alleviated during spermiogen- 2013). By contrast, the chromatin interactions for all other stages
esis (Namekawa et al., 2006; Wang et al., 2018). are largely similar to P(s)  s1. Hence, these data demonstrated
Recent studies have characterized the dynamics of transcrip- global reorganization of chromatin structure during spermato-
tomes, chromatin accessibility, histone modifications, and DNA genesis and a unique configuration of pachytene chromosomes
methylomes during mouse and/or human spermatogenesis at the molecular level, when chromosomes are condensed and
(Chen et al., 2018; Green et al., 2018; Guo et al., 2017; Hammoud synapsed during meiosis.
et al., 2014; Lesch et al., 2016; Maezawa et al., 2018; Stévant
et al., 2018; Wang et al., 2018). However, the three-dimensional Dynamic TADs and Chromatin Compartmentalization
(3D) chromatin structure and its reprogramming in this critical during Primate Spermatogenesis
process remain to be elucidated, and even less is known for pri- Next, we sought to examine the presence of TADs at different
mates. Here, using sisHi-C, which we developed recently (Du stages. Our analysis revealed strong TADs in fibroblast and
et al., 2017), we investigated chromatin architecture in sper- mature sperm (Figure 1B). Surprisingly, TADs appear to be
matogenic cells in rhesus monkey and mouse. Our results not weak in SPA and are even more severely attenuated in PAC (Fig-
only demonstrated dynamic reprogramming of chromatin folding ure 1B, right, for each stage). The strengths of TADs (with TAD
during male germline development but also revealed unique positions defined in fibroblast; similar results were obtained by
high-order meiotic chromatin organization for the synaptonemal using TAD boundaries identified at other stages, data not shown)
complex that is largely conserved from rodent to primate. become stronger in RS and continue to increase in sperm (Fig-
ures 1B and 2A–2C). The majority of fibroblast TAD boundaries
RESULTS (78%) are also shared by sperm (Figure S2A), as reported in
mouse (Battulin et al., 2016; Rowley and Corces, 2016). Fibro-
Global Reorganization of High-Order Chromatin blast and sperm also display strong chromatin compartmental
Structure during Rhesus Monkey Spermatogenesis interactions (Lieberman-Aiden et al., 2009). By contrast, SPA
To examine the 3D chromatin structure in spermatogenesis, we and PAC appear to show weak plaid patterns (Figures 1B and
applied a low-input Hi-C method (sisHi-C) (Du et al., 2017) to S2B). Nevertheless, we were able to call locations of chromatin
spermatogenic cells in rhesus monkey. We collected spermato- compartments through principal-component analysis (PCA) (Lie-
gonia (SPA), pachytene spermatocyte (PAC), round spermatid berman-Aiden et al., 2009) at all stages. These identified com-
(RS), and mature sperm using STA-PUT, a widely used velocity partments showed expected differences in gene density and
sedimentation approach to purify spermatogenic cells (Bryant H3K4me3 enrichment (Lesch et al., 2016) between A and B for
et al., 2013, 2015; Hur et al., 2016; Korhonen et al., 2015; Liu each stage (Figure S2C). However, it is worth noting that PAC ap-
et al., 2015; Luense et al., 2016) (Figures 1A and S1A; see pears to show fine-scale compartments in local regions (Fig-
STAR Methods). We confirmed the identities of these cells by ure 1B, zoomed-in views) (discussed later). As the synaptonemal
immunofluorescence staining and morphology (Figures S1A complex forms at the pachytene stage to allow the synapse of
and S1B; see legend for details). We also derived fibroblast homologous chromosomes and meiotic recombination, such
from testis as a control (see STAR Methods). Next, we performed unique chromatin Hi-C interactions may be closely related to
sisHi-C for a total of 11 billion reads. For each stage, we this structure (see below and Discussion). Taken together, these

548 Molecular Cell 73, 547–561, February 7, 2019


A Figure 1. Reprogramming of Chromatin Or-
ganization in Primate Spermatogenesis
Spermatogenesis process of rhesus monkey
(A) Schematic of the spermatogenesis process of
rhesus monkey.
(B) Heatmaps showing the normalized Hi-C inter-
action frequencies (100 kb bin, chr12). Zoomed-in
views (40 kb bin) are also shown (pooled data from
two or three biological replicates).
Pachytene Round
(C) The chromatin contact probabilities relative
Fibroblast Spermatogonia Spermatocyte Spermatid Sperm to genomic distance (P[s] curves) are shown.
(SPA) (PAC) (RS) P(s)  s0.5 and P(s)  s1 represent the predicted
2n 2n 4n n n
mitotic and fractal globule states, respectively.
See also Figure S1 and Table S1.

Interaction heatmap (chr12)

Fibroblast of the interaction heatmap and between


SPA
B (Mb) (Mb)
105 105
the two ends of the chromosome, which
were confirmed in replicates (Figures 3A
and S3B, arrows). This is also supported
by the P(s) curve analysis, as the X chro-
mosome at the PAC stage showed much
0 0 stronger distal interactions beyond 10 Mb
0 105 60 70 0 105 60 70
than autosomes (Figure S3C, arrow). We
PAC
(Mb)
RS speculated that such pattern may be
(Mb)
105 105 linked to a potential mixture of twisted or
looped chromatin structures for X chromo-
some in the XY body, as previously re-
ported (Solari, 1980) (Figure S3B, right).
Notably, the chromatin organization asso-
0 0 ciated with MSCI is distinct from that for
0 105 60 70 0 105 60 70
female random XCI, where the X chromo-
Sperm some displays bipartite megadomains
(Mb) P(s) curves (chr2)
105 C separated at the locus of DXZ4 (Figure 3A,
-5
10
s-0.5
‘‘Fibroblast XaXi,’’ arrow) (Darrow et al.,
Contact probability

10
-6 s-1 2016; Deng et al., 2015; Giorgetti et al.,
10
-7 2016; Nora et al., 2012; Rao et al., 2014).
Fibroblast Sperm
This further corroborates the notion that
0 10
-8
0 105 60 70 SPA major differences exist between male
-9 PAC RS
Interaction frequency
10 5 6 7 8 MSCI and female random XCI. For
2X10 10 10 10
0 10 Genomic distance (bp) example, Xist, a key regulator of XCI (Lee,
2005; Minajigi et al., 2015), is dispensable
in MSCI (Turner et al., 2002). Thus, these
data revealed stepwise dissolution and re-establishment of data revealed fundamental differences between female XCI
TADs, as well as dynamic chromatin compartmentalization dur- and MSCI not only in their key regulators but also in the 3D chro-
ing monkey spermatogenesis. matin conformation.

Chromatin Organization of MSCI Pachytene Chromosomes Show a Unique Chromatin


As the X chromosome undergoes MSCI during male gametogen- Configuration
esis (Turner, 2007), we sought to determine the dynamics of Interestingly, a comparison of pachytene X chromosome and au-
the X chromosome architecture during spermatogenesis. The tosomes revealed clear differences. Strong local compartments
inactivation of X chromosome in PAC was verified by RNA are visible on autosomes but not on the X chromosome (compare
sequencing (RNA-seq), with the majority of X-linked genes still Figures 1B and 3B, zoomed-in views). These data raise an inter-
being repressed at the RS stage (Figure S3A; see STAR esting question if this is related to the difference of transcription
Methods) similarly as in human (Wang et al., 2018). TADs and state between autosomes and the X chromosome (discussed
chromatin compartmentalization on the X chromosome are later). Such unique local chromatin organization on pachytene au-
strong in sperm and fibroblast but are much weaker in SPA tosomes is also distinct from those at other stages (see zoomed-in
and RS and are nearly depleted in PAC (Figures 3A–3C). Intrigu- heatmaps in Figure 1B) and mitotic chromatin (Gibcus et al.,
ingly, a close examination of X chromosome in PAC revealed 2018; Nagano et al., 2017; Naumova et al., 2013). Specifically,
specific chromatin interactions along the perpendicular diagonal pachytene autosomes show a refined local plaid pattern in the

Molecular Cell 73, 547–561, February 7, 2019 549


A
B
TAD dynamics during spermatogenesis Fibroblast SPA

chr18:36,334,984-44,264,276

Fibroblast
Fibroblast

2XTAD

TAD
TAD

Fibroblast
PAC RS

Fibroblast
2XTAD

TAD
SPA

Sperm
Fibroblast
TAD
2XTAD
PAC

Fibroblast
2XTAD

TAD
Normalized interaction
frequency

0 2

RS Fibroblast
TAD
2XTAD

TAD insulation
C
0
PAC
Sperm RS
SPA
Insulation score

-0.2
Sperm
-0.4
Fibroblast

Interaction frequency -0.6

Low High
-0.8
Fibroblast TAD
2XTAD

Figure 2. TAD Dynamics during Spermatogenesis


(A) WashU Epigenome Browser (Zhou et al., 2013) views (40 kb bin) show chromatin interaction frequencies of cells at different spermatogenic stages and
fibroblast.
(B) Heatmaps showing the normalized average interaction frequencies for all topologically associating domains (TADs) (defined in fibroblast) as well as their
nearby regions (±0.5 TAD length) in fibroblast and different spermatogenic cells (replicates pooled, n = 2 or 3).
(C) The average insulation scores (IS) of fibroblast and cells at different spermatogenic stages at TADs (defined in fibroblast) and nearby regions are shown.
See also Figures S1 and S2.

interaction heatmap, which resembles that of compartmental in- regions, we refined the PCA to identify high-resolution compart-
teractions between conventional compartments A and B but ments in each 10 Mb region (see STAR Methods). Principal
with smaller sizes (Figure 4A). Interestingly, such pattern in PAC component 1 (PC1) from such fine-scale PCA (termed ‘‘local
seems to precisely match transcription activity in the interaction PC1’’ and ‘‘local PCA,’’ respectively) successfully captured com-
heatmap (Figure 4A). An initial analysis showed that such fine- partments that match chromatin structure observed in the heat-
scale compartment-like pattern could not be captured by the con- maps (Figures 4A and S3D–S3E, ‘‘Local PC1’’). As expected, local
ventional PCA (Lieberman-Aiden et al., 2009) (Figures 4A and S3D, PCA identified a larger number of compartments in PAC chromatin
‘‘PC1’’), which considers the whole chromosome for compartment with generally smaller sizes compared with conventional compart-
identification. As these interactions are highly centered near local ments in fibroblast (Figure S4A). These compartments correlate

550 Molecular Cell 73, 547–561, February 7, 2019


A

Chromatin interaction (chrX)


Fibroblast XaXi
Fibroblast XaY SPA PAC RS Sperm
(Darrow et al.)
PC1 A
B
150 150 150 150 150 150
DXZ4

0 0 0 0 0 0
Centromere 150 150 150 150 150 150
(predicted)

1
Insulation
score

-2
30 40 30 40 30 40 30 40 30 40 30 40
Interaction frequency (Mb)
0 10
C
Fibroblast
2XTAD

TAD

Fibroblast Fibroblast Fibroblast Fibroblast Fibroblast Fibroblast


TAD TAD TAD TAD TAD TAD
2XTAD 2XTAD 2XTAD 2XTAD 2XTAD 2XTAD

Normalized interaction frequency


0 2

Figure 3. Chromatin Organization of MSCI


(A) Interaction frequency heatmaps (100 kb bin) of the X chromosome in different cell types. The corresponding chromatin compartments (PC1 values; see STAR
Methods) are shown. Black arrow, boundary between the megadomains on the X chromosome of female fibroblast near DXZ4 (Darrow et al., 2016). Blue arrows
show strong interactions perpendicular to the diagonal in the heatmaps. Green box, predicted centromere (as no annotations are available). XaXi, active chrX +
inactive chrX (female); XaY, active chrX + chrY(male).
(B) Top: zoomed-in views (40 kb bin) of interaction frequency heatmaps. Bottom: insulation score of the zoomed-in views of X chromosome (replicates pooled,
n = 2 or 3).
(C) Heatmaps showing the normalized average interaction frequencies for all TADs (defined in fibroblast) on the X chromosome as well as their nearby regions
(±0.5 TAD length) in different cell types.
See also Figure S3.

well with gene expression (Figure S4B). Although we could also S4B). In contrast, this is not observed when using whole-chromo-
identify smaller compartments by applying local PCA to other some PCA (Figure 4B). To distinguish such refined compartments
stages, they do not appear to precisely match the RNA-seq data in PAC from conventional compartments A and B, we termed
(Figure 4A). In fact, PAC shows the largest difference of transcrip- these compartments ‘‘refined compartments A and B’’ (or
tion activities between A and B among all stages when applying ‘‘refined-A/B’’). We observed strong interactions between
local PC1 analysis, indicating a much closer relationship between refined-A and other refined-A regions, and the same is true for
refined local compartments and transcription (Figures 4B and refined-B (Figure 4A, arrow, and Figures S3D and S3E). In

Molecular Cell 73, 547–561, February 7, 2019 551


A
Chromatin interaction vs. transcription (chr6)
PC1 A
B
RNA D
(Mb) Local PC1 A Fibroblast
160 B
LT

HT
Fibroblast
LT

LT HT LT
0 160 130 140 (Mb)
A
PC1 B
RNA
(Mb) A
Local PC1 PAC
160 B

LT

HT
PAC
LT

LT HT LT
0 160 130 140 (Mb)
PC1 A
B
RNA
(Mb) A
160
Local PC1
B
RS

LT

HT
RS
LT

LT HT LT
0 160 130 140 (Mb)
Normalized interaction
Interaction frequency Correlation frequency
0 10 -0.6 0.6 -0.2 0.2

B Transcription in A vs. B
PC1 Local PC1 C
Enriched Interaction
0 1 2 3 4 5 6

RNA-seq/PRO-Seq
Gene expression in A/B

in A heatmap
Strong interaction
Transcription

Folding LT1
(FPKM)

HT1
LT1HT1 LT2 LT3 HT2 LT4
Strong interaction LT2
Enriched
in B LT3 HT2 LT4
SP t

C
PA A

ob S
SP t
PA A
C
S
s

s
br R

R
la

la
ob
br
Fi

Fi

Figure 4. Pachytene Chromosomes Show a Unique Chromatin Configuration


(A) Left: interaction frequency heatmaps (100 kb bin) for fibroblast, pachytene spermatocyte (PAC), and round spermatid (RS). Middle: zoomed-in views (40 kb
bin) are also shown. Right: correlation heatmaps showing correlations between any two region pairs within 10 Mb for their intra-chromosomal interaction fre-
quency patterns (100 kb bin) (pooled data from two biological replicates). The UCSC Genome Browser views of RNA-seq and chromatin compartments (PC1 and
local PC1 values; see STAR Methods) are also shown. Black arrow shows an example of interaction between two refined-A regions.
(B) Ratio of average gene expression levels in compartment A versus that in compartment B identified by PC1 (red) or local PC1 (blue) in different cell types.
Boxplots show variations among different chromosomes (X chromosome excluded). Boxplots show the median (center line), the 25th and 75th percentiles
(bottom and top edges, respectively), and 1.5 times the interquartile range (whiskers).
(legend continued on next page)

552 Molecular Cell 73, 547–561, February 7, 2019


contrast, few interactions were found between refined-A and tion of monkey is practically difficult, we first sought to determine
refined-B regions. To further independently validate such findings if such structure is conserved in mouse. Therefore, we collected
without relying on the PCA, we sought to directly identify loci with PAC cells from mouse (see STAR Methods) and performed Hi-C
high transcription levels (HT regions), flanked by regions with low analysis, with 3.4 billion total sequencing reads and 0.66 billion
transcription levels (LT regions) (see STAR Methods). By doing chromatin contacts (Table S1). Similar to those of rhesus mon-
so, a quantitative analysis showed preferential interactions be- key, we found that mouse PAC cells have dramatically dimin-
tween HT-HT and LT-LT regions but not HT-LT regions, leading ished TADs (Figures S5A and S5B), and the chromatin
to sharp boundaries between HT and LT regions in the heatmap interaction follows the P(s)  s0.5 curve, similar to that of meta-
for PAC (Figures 4C and 4D). In contrast, such patterns were not phase II (MII) oocyte (Du et al., 2017) (Figure S5C). Importantly,
apparent for fibroblast or RS, where chromatin interactions be- mouse PAC again shows a refined plaid pattern at a local scale
tween HT regions tend to ‘‘spread’’ across HT-LT boundaries (Fig- (Figure S5A) with the HT and LT regions spatially segregated
ure 4D). Thus, these data indicate that the spatial compartmental- (Figure S5D). The X chromosome in mouse PAC again lacks
ization of pachytene chromatin has an unusually high correlation the finer scale compartmentalization pattern but shows interac-
with gene expression. tions along the perpendicular diagonal of the interaction heat-
Interestingly, our analysis did not reveal clear plaid pattern on map similar to that in monkey (Figure S6A, arrow). It is worth
the X chromosome (Figure S4C). Notably, although the simulta- noting that in mouse PAC, refined-A/B appear to be somewhat
neous absence of transcription and refined-A/B on the X chromo- less prevalent than those in rhesus monkey on the basis of a vi-
some further corroborates their correlation in PAC, curiously, we sual examination (data not shown), and we also did not observe
did not observe a stripy plaid pattern near several well-known strong interactions between the two chromosome ends (with the
MSCI escaped genes (such as Akap4; Adams et al., 2018; Miki current 0.66 billion contact reads). Taken together, these data
et al., 2002) (data not shown). These data indicate a unique mech- demonstrated that the 3D chromatin architecture of pachytene
anism underlying chromatin architecture associated with MSCI. chromosomes is largely conserved between mouse and primate.
Notably, transcription-correlated compartmental domains Next, we sought to explore the possible causality between
were reported in Drosophila and other species using high-depth transcription and chromatin organization in PAC. We inhibited
Hi-C datasets (Rowley et al., 2017). In mammals, such domains transcription in purified mouse PAC cells with a-amanitin for
are often obscured by CTCF and cohesin-mediated loop do- 11 hr (before cells underwent apoptosis). We confirmed the ab-
mains (Rao et al., 2017; Schwarzer et al., 2017). Depleting cohe- sent or significantly reduced nascent transcription and RNA po-
sin or the cohesin-loading factor NIPBL resulted in the loss of loop lymerase (Pol) II proteins using immunofluorescence (as the
domains and enhanced fine-scale compartments that are corre- extended treatment of a-amanitin can induce degradation of
lated with distinct chromatin states (Rao et al., 2017; Schwarzer RNA Pol II; Nguyen et al., 1996) (Figure S5E). Surprisingly, Hi-C
et al., 2017). Consistent with the depletion of TADs, loop domains analysis showed that the chromatin organization of PAC largely
are also largely undetectable (with the existing 0.9 billion pairwise remained unchanged after the a-amanitin treatment, as shown
contacts) in PAC as shown by average plots (Figure S4D). By by analyses of interaction heatmaps, TADs, P(s) curve, and
applying the same HT-LT analysis to the published ultra-deep da- HT-LT segregation (Figures S5A–S5D). These results suggest
tasets (Bonev et al., 2017; Rao et al., 2017), we confirmed that HT/ that at least the maintenance of refined-A/B is likely independent
LT regions are poorly segregated in wild-type (WT) mouse and of transcription. However, we could not exclude the possibility
human cells. The spatial segregation becomes clearly enhanced that the impact of transcription inhibition on chromatin architec-
in cohesin-depleted cells, albeit at a weaker level despite its ture is not significant enough to be detected, or the residual RNA
much higher sequencing depth compared with PAC (Figure S4E). Pol (or other lingering transcription factors) may help maintain
PAC also has visibly stronger interactions between LT regions the pachytene chromatin configuration.
compared with those from cohesin-depleted cells (compare Fig-
ures 4D and S4E). Finally, the P(s) curve analysis showed that only Mutants with Defective Synaptonemal Complex Showed
PAC chromatin, but not those in fly and cohesin-depleted cells, Attenuation of Refined-A/B and Restoration of
can be well characterized by the s0.5 curve (Figure S4F) (dis- Conventional Chromatin Structure
cussed later). Taken together, these data demonstrated that We then asked what could be the alternative contributing factors
pachytene chromatin shows a unique configuration with highly that underline the refined-A/B chromatin structure in PAC.
refined transcription-correlated compartments. One possibility is that this may be mechanistically associated
with the loss of TADs and loop domains (Rao et al., 2017;
3D Structure of Pachytene Chromatin Is Largely Schwarzer et al., 2017). CTCF is expressed in spermatogenesis
Conserved in Mouse (Figures S6B and S6C) but is surprisingly dispensable for the
We then asked how such transcription-correlated chromatin maintenance of synaptonemal structure or homologous recom-
structure of PAC is possibly formed. Because genetic manipula- bination (Hernández-Hernández et al., 2016). Several cohesin

(C) Scheme model showing the analysis principle for interaction frequency of regions for HT regions (500 kb regions with transcription levels above genome
average and at least 3-fold more than nearby LT regions) and LT regions (with transcription levels below genome average and less than one-third of nearby HT
regions) across the genome. PRO-seq dataset was obtained from a previous study (Rao et al., 2017).
(D) Heatmaps showing interaction frequency for HT regions or LT regions at different stages (pooled data from two or three biological replicates, 25 kb bin).
See also Figures S3 and S4.

Molecular Cell 73, 547–561, February 7, 2019 553


components are dynamically regulated between PAC and cells, followed by Sycp2-KO cells and then Top6bl-KO ZYG-L
somatic cells, including Smc1b and Stag3 (Figures S6B and cells (Figures 5, 6A, and 6B), which is consistent with the notion
S6C). Nevertheless, as cohesin is also essential for the formation that zygotene is closer to pachytene than leptotene. This is
of the synaptonemal complex (Biswas et al., 2013; Llano et al., confirmed by a hierarchical clustering for the overall chromatin
2012), it is difficult to precisely dissect its functions in loop do- structures (Figure 6C) and the P(s) analysis (Figure 6D). Remark-
mains and synapsis. Alternatively, the synaptonemal complex ably, P(s) analysis revealed that the ‘‘interaction insulation
may inhibit the formation of TADs, perhaps through immobilizing boundary’’ at 6 Mb is compromised in Sycp2-KO cells and is
chromatin and restraining it from loop extrusion. In order to test entirely lost in Top6bl-KO LEP-L mutant cells (Figure 6D, arrow).
this hypothesis, we obtained spermatocytes from mouse defi- This is accompanied by increased long-distance interactions
cient for Sycp2, a core component of the synaptonemal complex (>10 Mb), which is in line with the disruption of the linearly orga-
(Yang et al., 2006). These mutant spermatocytes fail to form nized synaptonemal complex. Finally, we found that TAD
lateral elements (LE), leading to developmental arrest at the strengths increased on the X chromosome in all three mutant
zygotene-like stage, the stage immediately preceding pachytene cells (Figure S7A). This is consistent with the fact that XY body
(Yang et al., 2006). We then performed Hi-C analysis for the has not formed at the leptotene stage yet and is either defective
mutant cells (gH2AX positive; see STAR Methods). Surprisingly, or absent for mouse deficient in meiotic DSB formation or synap-
we observed attenuated refined-A/B compartments in these tonemal complex components (Bellani et al., 2005; Mahadevaiah
cells, accompanied by the reappearance of TADs and increased et al., 2008; Robert et al., 2016; Yang et al., 2006). Notably, the
conventional compartmentalization, albeit at weaker levels fact that both Sycp2 and Top6bl mutants displayed consistent
compared with those in mouse embryonic stem cell (mESC) defects indicates that perhaps the meiosis stages (such as
or fibroblast (Figure 5). These data demonstrate that the pachy- leptotene, zygotene, and pachytene), rather than SYCP2 or
tene chromatin organization is compromised in Sycp2-mutant TOPOVIBL, are more relevant to the observed chromatin struc-
spermatocytes. ture dynamics. Nevertheless, these data are still compatible
It remains to be determined if such defects come from the with the possibility that the PAC chromatin architecture features
absence of SYCP2 protein or the developmental arrest at the revealed by Hi-C analyses requires an intact synaptonemal com-
zygotene-like stage (Yang et al., 2006). The exact stage of these plex, which is not fully established until the pachytene stage.
cells is difficult to be precisely determined, as stage markers Such complex likely restricts conventional TADs and promotes
such as SYCP2 and SYCP3 are either absent or fail to localize the formation of refined compartments.
to axial chromosomes (Yang et al., 2006). Therefore, we
collected spermatocytes from a second mouse mutant line defi- DISCUSSION
cient for TOPOVIBL (TOP6BL), a topoisomerase VI B-like protein
that interacts with SPO11 to regulate double-strand break for- Chromatin undergoes drastic reprogramming during meiosis.
mation during spermatogenesis (Robert et al., 2016; Vrielynck However, little is known about the molecular basis for the reprog-
et al., 2016). Top6bl knockout results in unsynapsed homolo- ramming of higher-order chromatin structure in this process.
gous chromosomes, and the mutant cells were arrested at Here, we examined dynamics of the 3D meiotic chromatin archi-
various stages before pachytene (Robert et al., 2016). We tecture in rhesus monkey, in which we found that TADs undergo
collected leptotene-like (LEP-L) and zygotene-like (ZYG-L) dissolution and re-establishment during spermatogenesis. This
mutant cells using STA-PUT (STAR Methods) and confirmed is accompanied by the emergence of a unique chromatin
their identities on the basis of SYCP3 and gH2AX staining (Fig- configuration as the synaptonemal complex forms, featured by
ures S6D). Their identities were further confirmed by expression highly refined, transcription-correlated chromatin compart-
analysis of stage-specific genes defined previously (Fallahi et al., ments. Such structure is recapitulated in mouse, where its pres-
2010) (Figure S6E). Note that the developmental arrest of LEP-L ence is partially compromised in mutant spermatocytes with
and ZYG-L spermatocytes provided an opportunity to allow us to defective synaptonemal complex (Figure 7). These data revealed
collect them in relatively large quantities using STA-PUT. By highly dynamic chromatin architecture reorganization during
contrast, WT cells of these stages exist in minor fractions in pop- meiosis that is conserved from rodent to primate.
ulations and are difficult to collect with high purity. Interestingly, Transcription-correlated chromatin structure (compartment
we observed similar alterations of chromatin architecture in domains or sub-compartments) was previously reported in
Top6bl-knockout (KO) LEP-L and ZYG-L spermatocytes as Drosophila and other species (Rowley et al., 2017). In mammals,
Sycp2-KO mutants, with attenuated refined-A/B accompanied similar compartmental domains become evident in the absence
by increased strengths in TAD and conventional chromatin of cohesin (Rao et al., 2017; Schwarzer et al., 2017). However,
compartmentalization (Figure 5, middle). The differential heat- even compared with that of cohesin-depleted cells, the tran-
maps between mutant spermatocytes and WT PAC also scription-correlated chromatin segregation is much stronger in
confirmed that the conventional compartments A and B (Figure 5, PAC. One key question is how such meiotic chromatin organiza-
left differential heatmaps) and TADs (Figure 5, right differential tion is formed. One possibility is that certain key architectural
heatmaps) are clearly enhanced. By contrast, refined-A/B plaid proteins for TADs or loops may be absent or not functional. How-
patterns become much weaker (Figure 5, right differential heat- ever, the fact that TADs are present in Sycp2 and Top6bl mutants
maps). This is further confirmed using global quantitative ana- indicates that such architectural factors exist at least shortly
lyses (Figures 6A and 6B). Compared with chromatin of PAC, before the synaptonemal complex appears. The second possi-
the structure changes are most evident in Top6bl-KO LEP-L bility is that TADs and loop domains may be incompatible with

554 Molecular Cell 73, 547–561, February 7, 2019


Mouse chromatin interaction (chr2)

PC1 RNA
175 Local PC1

WT PAC

0 175 (Mb)

Sycp2 KO - WT PAC
Sycp2 KO - WT PAC

PC1

Sycp2 KO
Spermatocyte

0 175 (Mb)
PC1
ZYG-L - WT PAC

ZYG-L - WT PAC
Top6bl KO
ZYG-L

0 175 (Mb)
PC1
LEP-L - WT PAC

LEP-L - WT PAC
Top6bl KO
LEP-L

0 175 (Mb)
PC1
Fibro - WT PAC
Fibro - WT PAC

Fibroblast
(Darrow et al.)

0 175 (Mb)
0 175 55 70 55 70 (Mb)
Normalized interaction Normalized interaction
frequency Interaction frequency frequency
-3 3 0 10 -0.05 0.05

Figure 5. Mutant Spermatocytes with Defective Synaptonemal Complex Showed Attenuation of Refined-A/B and Restoration of Conven-
tional Chromatin Structure
Left: differential (mutant or fibroblast; WT) interaction frequency heatmaps of chromosome 2 (100 kb bin). Middle: interaction frequency heatmaps (100 kb bin)
and zoomed views (40 kb bin) in mouse WT PAC, Sycp2-knockout (KO) spermatocytes, Top6bl-KO zygotene-like (ZYG-L) spermatocytes, Top6bl-KO leptotene-
like (LEP-L) spermatocytes and fibroblast (Darrow et al., 2016). The PC1 values and local PC1 values of the regions and RNA-seq enrichment are also shown in
UCSC Genome Browser tracks. Right: differential (mutant or fibroblast; WT) interaction frequency heatmaps of zoomed views (40 kb bin). The local PC1 tracks of
WT PAC are shown above each zoomed-in differential heatmap as references. To make all samples in Figure 5 comparable, we randomly sampled 0.15 billion
total contacts from WT cells to match those from the mutant cells.
See also Figures S5 and S6.

Molecular Cell 73, 547–561, February 7, 2019 555


A

Sycp2 KO Top6bl KO
WT PAC Spermatocyte ZYG-L LEP-L Fibroblast
Fibroblast
2XTAD

TAD

Fibroblast Fibroblast Fibroblast Fibroblast Fibroblast


TAD TAD TAD TAD TAD
2XTAD 2XTAD 2XTAD 2XTAD 2XTAD

Normalized interaction frequency


0 2

B Conventional compartment segregation C Interaction frequency clustering (chr1)


Highly
segregated WT PAC
Segregation strength

1.8

PAC (α-amanitin)
Top6bl KO ZYG-L

Sycp2 KO spermatocyte
1.2

Top6bl KO LEP-L
Poorly
p6 p2 C

Fi LE L
ob -L
st

mESC
To bl KO
bl G-
To yc PA

segregated
la
br P
p6 ZY
S T
W

Fibroblast

Mouse P(s) curves (chr1)


D -5
10
s-0.5
s-1
-6
Contact probability

10

-7
10
Fibroblast
-8
10 Sycp2 KO Spermatocyte LEP-L
ZYG-L
WT PAC MII oocyte
-9
10
5 6 7 8
2X10 10 10 10
Genomic distance(bp)

Figure 6. Quantitative Analyses of Chromatin Architecture in Mutant Spermatocytes with Defective Synaptonemal Complex
(A) Heatmaps showing the normalized average interaction frequencies for all TADs (defined in fibroblast; Darrow et al., 2016) as well as their nearby regions
(±0.5 TAD length) in mouse WT PAC, Sycp2-KO spermatocytes, Top6bl-KO ZYG-L spermatocytes, Top6bl-KO LEP-L spermatocytes, and fibroblast (Darrow
et al., 2016).
(B) Boxplots showing segregation levels for conventional compartments, which were calculated on the basis of regions enriched for the CpG island clusters (CGI
forests, representing promoter-rich regions and compartment A) and regions depleted for CpG island (CGI prairie, representing promoter-poor regions and
compartment B) (Liu et al., 2018). See STAR Methods for details. Segregation strengths were calculated as (AA + BB)/(AB).

(legend continued on next page)

556 Molecular Cell 73, 547–561, February 7, 2019


Figure 7. Schematic Model for 3D Chromatin
Organization during Spermatogenesis
A schematic model showing the reprogramming of
chromatin organization in primate spermatogen-
esis. TADs undergo dissolvement and re-estab-
lishment during spermatogenesis. Chromatin
organization of PAC is characterized by a finer
scale transcription-correlated compartmentaliza-
tion pattern (refined-A/B) and attenuated compart-
mental interactions from conventional compart-
ments A and B. In PAC synaptonemal complex (SC)
may restrict TADs, leading to a strong correspon-
dence between genome compartmentalization and
transcription activities.
See also Figure S7.

CTCF cohesin-mediated loop extrusion


(Bonev and Cavalli, 2016). CTCF have
been shown to largely fall off mitotic chro-
matin (Oomen et al., 2018), and loops in
mitotic chromatin likely extrude from the
central axis in a random manner without
the synaptonemal complex, which forms symmetric structure defined positions (Gibcus et al., 2018; Naumova et al., 2013),
between homologous chromosomes, presumably by anchoring leading to the absence of TADs. In PAC, chromatin is anchored
chromatin at defined loci (Cobb and Handel, 1998). As a result, to the synaptonemal complex at defined locations. Yet this may
chromatin fiber may not freely undergo cohesin and CTCF-medi- also immobilize chromatin and prevent loop extrusion and for-
ated extrusion. This is supported by the partial restoration of mation of TADs. Finally, differences also exist between these
TADs in Top6bl and Sycp2 mutants. Finally, an intriguing possi- two states. For example, pachytene chromosomes are in a
bility is that the synaptonemal complex may directly correspond more relaxed state with active transcription, unlike mitotic chro-
to refined-A/B structure (Figure S7B). In fact, previous studies matin, which is quiescent (Chelysheva et al., 1990; Gottesfeld
showed that Pol II and H3K4me3 are enriched in the loops ex- and Forbes, 1997). Consistently, PAC has refined-A/B compart-
tending laterally from synaptonemal complexes (Kaufmann ments, which are absent in mitotic chromatin. Thus, the PAC
et al., 2012; Khalil et al., 2004; Prakash et al., 2015), although it structure may render a balance between the needs of chromatin
remains to be determined whether the sizes of loops match the compaction and transcription.
fine-scale refined-A/B compartments.
Notably, one possible clue for the mechanisms underlying the Perspective
unique pachytene chromatin organization may come from its Our study raised several key questions that await further
comparison with mitotic chromatin. Both states exhibit the investigations. First, how exactly the synaptonemal complex
depletion or attenuation of TADs and compartments A and B. contributes to the absence of TADs and emergence of transcrip-
They also share very similar P(s) curves. Both states were sug- tion-associated refined-A/B remains elusive. Future studies are
gested to adopt a configuration featuring linearly compressed warranted to test the aforementioned possibilities. Second,
arrays of loops along a scaffold axis (Gibcus et al., 2018; Nau- what is the relationship between transcription and chromatin or-
mova et al., 2013; Handel and Schimenti, 2010). Such chromatin ganization? PAC provided one of the rare examples for which
condensation is essential for chromosome pairing (for meiosis) extensive transcription occurs with diminished TADs, similarly
and segregation (for both meiosis and mitosis). These data raise as that in early mouse embryos and cohesin-depleted cells (Du
the possibility that these two events (chromatin compaction and et al., 2017; Hug et al., 2017; Ke et al., 2017; Rao et al., 2017;
depletion of compartments and TADs) are linked. Perhaps chro- Schwarzer et al., 2017). Thus, the relationship of transcription
matin compaction along a scaffold axis restricts the free diffu- and chromatin organization is clearly complex and context-
sion of chromatin to form compartments A and B (presumably dependent, and future studies are needed to delineate their
through phase separation; Nuebler et al., 2018). The loss of causal relationship in PAC. Finally, it would be interesting to
TADs in pachytene and mitotic chromatin, however, may be explore the detailed structure of synaptonemal complex using
through different mechanisms. TAD boundaries are primarily inter-chromosomal interactions between homologous chromo-
anchored at CTCF sites, and its formation is presumably through somes. Although our mouse PAC data were obtained using

(C) Hierarchical clustering on the basis of chromatin interaction frequency of different cell types.
(D) P(s) curves are shown for different spermatocytes and fibroblast (Darrow et al., 2016). P(s)  s0.5 and P(s)  s1 represent the predicted mitotic and fractal
globule states, respectively. Arrow, interaction insulation boundary.
See also Figures S5–S7.

Molecular Cell 73, 547–561, February 7, 2019 557


crosses of SNP-containing strains, we found the interaction fre- purity verification and initial development of STA-UP velocity sedimentation
quencies between homologous chromosomes are extremely protocols for monkey cells. We thank Drs. Kehkooi Kee, Jianyang Zeng, and
Qinghua Tao for discussion; Xuechen Zhu for helping with various experi-
low (0.66% of total read pairs) compared with intra-chromo-
ments; and Jing He for discussion. We thank Dr. Yiqin Gao for discussion, Sirui
somal interactions. Therefore, the finer scale structure of synap- Liu for helping with the forest-prairie identification in the mouse genome, and
tonemal complex awaits future investigations. Taken together, Yaqiang Cao (Dr. Jing-dong Jackie Han’s lab) for discussion with loop identi-
our study not only revealed reprogramming of chromatin archi- fication. We appreciate the members of the Xie laboratory and the Wu labora-
tecture during spermatogenesis but also shed light on the tory for comments during preparation of the manuscript. This work was sup-
conserved fundamental principles governing 3D genome in ported by the National Key R&D Program of China (grant 2016YFC0900300
to W.X. and grant 2018YFC1003302 to X.W., 2016YFC1000600 to Q.S.), the
mammals (Figure 7). In addition, these data may pave the way
National Basic Research Program of China (grant 2015CB856201 to W.X.),
for future investigations to link chromatin architecture to male
the National Natural Science Foundation of China (grants 31422031,
infertility research. 31725018, and 31830047 to W.X.; grants 31571537 and 31872844 to X.W.;
grant 31660346 to W.S.; and grant 31630050 to Q.S.), the THU-PKU Center
STAR+METHODS for Life Sciences (W.X.), and the Strategic Priority Research Program of the
Chinese Academy of Sciences (grant XDB19000000 to Q.S.). W.X. is Howard
Detailed methods are provided in the online version of this paper Hughes Medical Institute (HHMI) International Research Scholar.

and include the following:


AUTHOR CONTRIBUTIONS
d KEY RESOURCES TABLE Y.W., H.W., Y.Z., Z.D., X.W., and W.X. conceived and designed the experi-
d ANIMALS ments. Y.W. and Z.D. conducted the sisHi-C experiments. Y.Z. and Z.D. per-
d SPERMATOGENIC CELL SEPARATION formed bioinformatics analysis. H.W., W.S., D.Q., M.W., and L.L. developed
d SPERMATOGENIC CELL PURIFICATION and optimized the STA-UP velocity sedimentation protocols for spermato-
d IMMUNOSTAINING genic cell purification in monkey and mouse testes. H.W., W.S., and Y.D. per-
formed the transportation and surgery of monkeys and advised on critical
d a-AMANITIN TREATMENT
knowledge of primate reproduction. H.W., S.F., and Y.J. collected the mutant
d SPERMATOCYTE MICRO-SPREADING AND IMMUNO- PAC. Y.L. and Q.W. performed NGS sequencing. Q.S., X.W., and W.X. super-
STAINING vised experiments or data analyses. Y.W., Y.Z., Z.D., H.W., X.W., and W.X.
d SISHI-C LIBRARY GENERATION AND SEQUENCING wrote the manuscript with contributions from other authors.
d RNA-SEQ LIBRARY PREPARATION AND SEQUENCING
d DATA ANALYSIS DECLARATION OF INTERESTS
B Hi-C data mapping The authors declare no competing financial interests.
B RNA-seq data processing
B Reproducibility of sisHi-C data Received: January 30, 2018
B Analysis of TADs Revised: September 17, 2018
B Average interaction heatmap of TADs Accepted: November 15, 2018
Published: February 7, 2019
B Identification of conventional chromatin compartments
and Refined-A/B REFERENCES
B Hi-C interaction heatmap, differential interaction heat-
map, conventional correlation heatmap and Refined- Adams, S.R., Maezawa, S., Alavattam, K.G., Abe, H., Sakashita, A., Shroder,
A/B correlation heatmap M., Broering, T.J., Sroga Rios, J., Thomas, M.A., Lin, X., et al. (2018). RNF8 and
B Analysis of gene density and H3K4me3 signal in com- SCML2 cooperate to regulate ubiquitination and H3K27 acetylation for escape
partments A/B gene activation on the sex chromosomes. PLoS Genet. 14, e1007233.

B Hierarchical clustering analysis Akdemir, K.C., and Chin, L. (2015). HiCPlotter integrates genomic data with
interaction matrices. Genome Biol. 16, 198.
B Analysis of interactions between HT/LT regions
Amendola, M., and van Steensel, B. (2014). Mechanisms and dynamics of nu-
B Loops analysis
clear lamina-genome interactions. Curr. Opin. Cell Biol. 28, 61–68.
B The P(s) analysis
Battulin, N., Fishman, V.S., Mazur, A.M., Pomaznoy, M., Khabarova, A.A.,
B Conventional compartment segregation strength
Afonnikov, D.A., Prokhortchouk, E.B., and Serov, O.L. (2016). Erratum to: com-
parison of the three-dimensional organization of sperm and fibroblast ge-
DATA AND SOFTWARE AVAILABILITY nomes using the Hi-C approach. Genome Biol. 17, 6.
Bellani, M.A., Romanienko, P.J., Cairatti, D.A., and Camerini-Otero, R.D.
The accession number for the data reported in this paper is GEO: GSE109344.
(2005). SPO11 is required for sex-body formation, and Spo11 heterozygosity
rescues the prophase arrest of Atm-/- spermatocytes. J. Cell Sci. 118,
SUPPLEMENTAL INFORMATION 3233–3245.
Bickmore, W.A. (2013). The spatial organization of the human genome. Annu.
Supplemental Information includes seven figures and one table and can be
Rev. Genomics Hum. Genet. 14, 67–84.
found with this article online at https://doi.org/10.1016/j.molcel.2018.11.019.
Biswas, U., Wetzker, C., Lange, J., Christodoulou, E.G., Seifert, M., Beyer, A.,
ACKNOWLEDGMENTS and Jessberger, R. (2013). Meiotic cohesin SMC1b provides prophase I
centromeric cohesion and is required for multiple synapsis-associated func-
We are grateful to Dr. P. Jeremy Wang and Yan Fang for kindly sharing the tions. PLoS Genet. 9, e1003985.
Sycp2-KO mouse and the members of the Dr. Dong Zhang and Dr. Ke Zheng Bonev, B., and Cavalli, G. (2016). Organization and function of the 3D genome.
laboratory at SKLRM for their valuable comments and technical help with the Nat. Rev. Genet. 17, 661–678.

558 Molecular Cell 73, 547–561, February 7, 2019


Bonev, B., Mendelson Cohen, N., Szabo, Q., Fritsch, L., Papadopoulos, G.L., Gibcus, J.H., Samejima, K., Goloborodko, A., Samejima, I., Naumova, N.,
Lubling, Y., Xu, X., Lv, X., Hugnot, J.P., Tanay, A., and Cavalli, G. (2017). Nuebler, J., Kanemaki, M.T., Xie, L., Paulson, J.R., Earnshaw, W.C., et al.
Multiscale 3D genome rewiring during mouse neural development. Cell 171, (2018). A pathway for mitotic chromosome formation. Science 359, eaao6135.
557–572.e24.
Giorgetti, L., Lajoie, B.R., Carter, A.C., Attia, M., Zhan, Y., Xu, J., Chen, C.J.,
Bryant, J.M., Meyer-Ficca, M.L., Dang, V.M., Berger, S.L., and Meyer, R.G. Kaplan, N., Chang, H.Y., Heard, E., and Dekker, J. (2016). Structural organiza-
(2013). Separation of spermatogenic cell types using STA-PUT velocity sedi- tion of the inactive X chromosome in the mouse. Nature 535, 575–579.
mentation. J. Vis. Exp. (80), e50648.
Gorkin, D.U., Leung, D., and Ren, B. (2014). The 3D genome in transcriptional
Bryant, J.M., Donahue, G., Wang, X., Meyer-Ficca, M., Luense, L.J., Weller, regulation and pluripotency. Cell Stem Cell 14, 762–775.
A.H., Bartolomei, M.S., Blobel, G.A., Meyer, R.G., Garcia, B.A., and Berger,
Gottesfeld, J.M., and Forbes, D.J. (1997). Mitotic repression of the transcrip-
S.L. (2015). Characterization of BRD4 during mammalian postmeiotic sperm
tional machinery. Trends Biochem. Sci. 22, 197–202.
development. Mol. Cell. Biol. 35, 1433–1448.
Chelysheva, L.A., Soloveĭ, I.V., Rodionov, A.V., Iakovlev, A.F., and Gaginskaia, Green, C.D., Ma, Q., Manske, G.L., Shami, A.N., Zheng, X., Marini, S., Moritz,
E.R. (1990). [The lampbrush chromosomes of the chicken. Cytological maps of L., Sultan, C., Gurczynski, S.J., Moore, B.B., et al. (2018). A comprehensive
the macrobivalents]. Tsitologiia 32, 303–316. roadmap of murine spermatogenesis defined by single-cell RNA-seq. Dev.
Cell 46, 651–667.e10.
Chen, Y., Zheng, Y., Gao, Y., Lin, Z., Yang, S., Wang, T., Wang, Q., Xie, N.,
Hua, R., Liu, M., et al. (2018). Single-cell RNA-seq uncovers dynamic pro- Guo, J., Grow, E.J., Yi, C., Mlcochova, H., Maher, G.J., Lindskog, C., Murphy,
cesses and critical regulators in mouse spermatogenesis. Cell Res. 28, P.J., Wike, C.L., Carrell, D.T., Goriely, A., et al. (2017). Chromatin and single-
879–896. cell RNA-seq profiling reveal dynamic signaling and metabolic transitions dur-
ing human spermatogonial stem cell development. Cell Stem Cell 21,
Cobb, J., and Handel, M.A. (1998). Dynamics of meiotic prophase I during
533–546.e6.
spermatogenesis: from pairing to division. Semin. Cell Dev. Biol. 9, 445–450.
Hammoud, S.S., Low, D.H., Yi, C., Carrell, D.T., Guccione, E., and Cairns, B.R.
Crane, E., Bian, Q., McCord, R.P., Lajoie, B.R., Wheeler, B.S., Ralston, E.J.,
(2014). Chromatin and transcription transitions of mammalian adult germline
Uzawa, S., Dekker, J., and Meyer, B.J. (2015). Condensin-driven remodelling
stem cells and spermatogenesis. Cell Stem Cell 15, 239–253.
of X chromosome topology during dosage compensation. Nature 523,
240–244. Handel, M.A. (2004). The XY body: a specialized meiotic chromatin domain.
Darrow, E.M., Huntley, M.H., Dudchenko, O., Stamenova, E.K., Durand, N.C., Exp. Cell Res. 296, 57–63.
Sun, Z., Huang, S.C., Sanborn, A.L., Machol, I., Shamim, M., et al. (2016). Handel, M.A., and Schimenti, J.C. (2010). Genetics of mammalian meiosis:
Deletion of DXZ4 on the human inactive X chromosome alters higher-order regulation, dynamics and impact on fertility. Nat. Rev. Genet. 11, 124–136.
genome architecture. Proc. Natl. Acad. Sci. U S A 113, E4504–E4512.
Hernández-Hernández, A., Lilienthal, I., Fukuda, N., Galjart, N., and Höög, C.
de Hoon, M.J.L., Imoto, S., Nolan, J., and Miyano, S. (2004). Open source clus- (2016). CTCF contributes in a critical way to spermatogenesis and male
tering software. Bioinformatics 20, 1453–1454. fertility. Sci. Rep. 6, 28355.
Dekker, J., and Mirny, L. (2016). The 3D genome as moderator of chromosomal Hug, C.B., Grimaldi, A.G., Kruse, K., and Vaquerizas, J.M. (2017). Chromatin
communication. Cell 164, 1110–1121. architecture emerges during zygotic genome activation independent of tran-
Dekker, J., Marti-Renom, M.A., and Mirny, L.A. (2013). Exploring the three- scription. Cell 169, 216–228.e19.
dimensional organization of genomes: interpreting chromatin interaction
Hur, S.K., Freschi, A., Ideraabdullah, F., Thorvaldsen, J.L., Luense, L.J.,
data. Nat. Rev. Genet. 14, 390–403.
Weller, A.H., Berger, S.L., Cerrato, F., Riccio, A., and Bartolomei, M.S.
Deng, X., Ma, W., Ramani, V., Hill, A., Yang, F., Ay, F., Berletch, J.B., Blau, (2016). Humanized H19/Igf2 locus reveals diverged imprinting mechanism be-
C.A., Shendure, J., Duan, Z., et al. (2015). Bipartite structure of the inactive tween mouse and human and reflects Silver-Russell syndrome phenotypes.
mouse X chromosome. Genome Biol. 16, 152. Proc. Natl. Acad. Sci. U S A 113, 10938–10943.
Dixon, J.R., Selvaraj, S., Yue, F., Kim, A., Li, Y., Shen, Y., Hu, M., Liu, J.S., and Imakaev, M., Fudenberg, G., McCord, R.P., Naumova, N., Goloborodko, A.,
Ren, B. (2012). Topological domains in mammalian genomes identified by Lajoie, B.R., Dekker, J., and Mirny, L.A. (2012). Iterative correction of Hi-C
analysis of chromatin interactions. Nature 485, 376–380. data reveals hallmarks of chromosome organization. Nat. Methods 9,
Dixon, J.R., Jung, I., Selvaraj, S., Shen, Y., Antosiewicz-Bourget, J.E., Lee, 999–1003.
A.Y., Ye, Z., Kim, A., Rajagopal, N., Xie, W., et al. (2015). Chromatin architec-
Jabbari, K., and Bernardi, G. (2017). An isochore framework underlies chro-
ture reorganization during stem cell differentiation. Nature 518, 331–336.
matin architecture. PLoS ONE 12, e0168023.
Du, Z., Zheng, H., Huang, B., Ma, R., Wu, J., Zhang, X., He, J., Xiang, Y., Wang,
Kaufmann, R., Cremer, C., and Gall, J.G. (2012). Superresolution imaging of
Q., Li, Y., et al. (2017). Allelic reprogramming of 3D chromatin architecture dur-
transcription units on newt lampbrush chromosomes. Chromosome Res. 20,
ing early mammalian development. Nature 547, 232–235.
1009–1015.
Durand, N.C., Robinson, J.T., Shamim, M.S., Machol, I., Mesirov, J.P., Lander,
Ke, Y., Xu, Y., Chen, X., Feng, S., Liu, Z., Sun, Y., Yao, X., Li, F., Zhu, W., Gao,
E.S., and Aiden, E.L. (2016a). Juicebox provides a visualization system for Hi-C
L., et al. (2017). 3D chromatin structures of mature gametes and structural re-
contact maps with unlimited zoom. Cell Syst. 3, 99–101.
programming during mammalian embryogenesis. Cell 170, 367–381.e20.
Durand, N.C., Shamim, M.S., Machol, I., Rao, S.S., Huntley, M.H., Lander,
E.S., and Aiden, E.L. (2016b). Juicer provides a one-click system for analyzing Khalil, A.M., Boyar, F.Z., and Driscoll, D.J. (2004). Dynamic histone modifica-
loop-resolution Hi-C experiments. Cell Syst. 3, 95–98. tions mark sex chromosome inactivation and reactivation during mammalian
spermatogenesis. Proc. Natl. Acad. Sci. USA 101, 16583–16587.
Fallahi, M., Getun, I.V., Wu, Z.K., and Bois, P.R.J. (2010). A global expression
switch marks pachytene initiation during mouse male meiosis. Genes (Basel) Korhonen, H.M., Yadav, R.P., Da Ros, M., Chalmel, F., Zimmermann, C.,
1, 469–483. Toppari, J., Nef, S., and Kotaja, N. (2015). DICER regulates the formation
and maintenance of cell-cell junctions in the mouse seminiferous epithelium.
Flavahan, W.A., Drier, Y., Liau, B.B., Gillespie, S.M., Venteicher, A.S.,
Biol. Reprod. 93, 139.
Stemmer-Rachamimov, A.O., Suvà, M.L., and Bernstein, B.E. (2016).
Insulator dysfunction and oncogene activation in IDH mutant gliomas. Langmead, B., and Salzberg, S.L. (2012). Fast gapped-read alignment with
Nature 529, 110–114. Bowtie 2. Nature Methods 9, 357–359.
Gibcus, J.H., and Dekker, J. (2013). The hierarchy of the 3D genome. Mol. Cell Lee, J.T. (2005). Regulation of X-chromosome counting by Tsix and Xite se-
49, 773–782. quences. Science 309, 768–771.

Molecular Cell 73, 547–561, February 7, 2019 559


Lesch, B.J., Silber, S.J., Mccarrey, J.R., and Page, D.C. (2016). Parallel evolu- Peters, A.H., Plug, A.W., van Vugt, M.J., and de Boer, P. (1997). A drying-down
tion of male germline epigenetic poising and somatic development in animals. technique for the spreading of mammalian meiocytes from the male and fe-
Nat. Genet. 48, 888–894. male germline. Chromosome Res. 5, 66–68.
Lieberman-Aiden, E., van Berkum, N.L., Williams, L., Imakaev, M., Ragoczy, Picelli, S., Faridani, O.R., Björklund, A.K., Winberg, G., Sagasser, S., and
T., Telling, A., Amit, I., Lajoie, B.R., Sabo, P.J., Dorschner, M.O., et al. Sandberg, R. (2014). Full-length RNA-seq from single cells using Smart-
(2009). Comprehensive mapping of long-range interactions reveals folding seq2. Nat. Protoc. 9, 171–181.
principles of the human genome. Science 326, 289–293. Prakash, K., Fournier, D., Redl, S., Best, G., Borsos, M., Tiwari, V.K.,
Liu, Y., Niu, M., Yao, C., Hai, Y., Yuan, Q., Liu, Y., Guo, Y., Li, Z., and He, Z. Tachibana-Konwalski, K., Ketting, R.F., Parekh, S.H., Cremer, C., and Birk,
(2015). Fractionation of human spermatogenic cells using STA-PUT gravity U.J. (2015). Superresolution imaging reveals structurally distinct periodic
sedimentation and their miRNA profiling. Sci. Rep. 5, 8084. patterns of chromatin along pachytene chromosomes. Proc. Natl. Acad. Sci.
Liu, S., Zhang, L., Quan, H., Tian, H., Meng, L., Yang, L., Feng, H., and Gao, U S A 112, 14635–14640.
Y.Q. (2018). From 1D sequence to 3D chromatin dynamics and cellular func- Rao, S.S., Huntley, M.H., Durand, N.C., Stamenova, E.K., Bochkov, I.D.,
tions: a phase separation perspective. Nucleic Acids Res. 46, 9367–9383. Robinson, J.T., Sanborn, A.L., Machol, I., Omer, A.D., Lander, E.S., and
Aiden, E.L. (2014). A 3D map of the human genome at kilobase resolution re-
Llano, E., Herrán, Y., Garcı́a-Tuñón, I., Gutiérrez-Caballero, C., de Álava, E.,
veals principles of chromatin looping. Cell 159, 1665–1680.
Barbero, J.L., Schimenti, J., de Rooij, D.G., Sánchez-Martı́n, M., and
Pendás, A.M. (2012). Meiotic cohesin complexes are essential for the forma- Rao, S.S.P., Huang, S.C., Glenn St Hilaire, B., Engreitz, J.M., Perez, E.M.,
tion of the axial element in mice. J. Cell Biol. 197, 877–885. Kieffer-Kwon, K.R., Sanborn, A.L., Johnstone, S.E., Bascom, G.D., Bochkov,
I.D., et al. (2017). Cohesin loss eliminates all loop domains. Cell 171, 305–
Luense, L.J., Wang, X., Schon, S.B., Weller, A.H., Lin Shiao, E., Bryant, J.M.,
320.e24.
Bartolomei, M.S., Coutifaris, C., Garcia, B.A., and Berger, S.L. (2016).
Comprehensive analysis of histone post-translational modifications in mouse Rathke, C., Baarends, W.M., Awe, S., and Renkawitz-Pohl, R. (2014).
and human male germ cells. Epigenetics Chromatin 9, 24. Chromatin dynamics during spermiogenesis. Biochim. Biophys. Acta 1839,
155–168.
Lupiáñez, D.G., Kraft, K., Heinrich, V., Krawitz, P., Brancati, F., Klopocki, E.,
Horn, D., Kayserili, H., Opitz, J.M., Laxova, R., et al. (2015). Disruptions of to- Robert, T., Nore, A., Brun, C., Maffre, C., Crimi, B., Bourbon, H.M., and de
pological chromatin domains cause pathogenic rewiring of gene-enhancer in- Massy, B. (2016). The TopoVIB-like protein family is required for meiotic
teractions. Cell 161, 1012–1025. DNA double-strand break formation. Science 351, 943–949.

Maezawa, S., Yukawa, M., Alavattam, K.G., Barski, A., and Namekawa, S.H. Rowley, M.J., and Corces, V.G. (2016). The three-dimensional genome: princi-
(2018). Dynamic reorganization of open chromatin underlies diverse transcrip- ples and roles of long-distance interactions. Curr. Opin. Cell Biol. 40, 8–14.
tomes during spermatogenesis. Nucleic Acids Res. 46, 593–608. Rowley, M.J., Nichols, M.H., Lyu, X., Ando-Kuri, M., Rivera, I.S.M., Hermetz,
Mahadevaiah, S.K., Bourc’his, D., de Rooij, D.G., Bestor, T.H., Turner, J.M.A., K., Wang, P., Ruan, Y., and Corces, V.G. (2017). Evolutionarily conserved prin-
and Burgoyne, P.S. (2008). Extensive meiotic asynapsis in mice antagonises ciples predict 3D chromatin organization. Mol. Cell 67, 837–852.e7.
meiotic silencing of unsynapsed chromatin and consequently disrupts meiotic Saldanha, A.J. (2004). Java Treeview—extensible visualization of microarray
sex chromosome inactivation. J. Cell Biol. 182, 263–276. data. Bioinformatics 20, 3246–3248.
Miki, K., Willis, W.D., Brown, P.R., Goulding, E.H., Fulcher, K.D., and Eddy, Salle, S., Sun, F., and Handel, M.A. (2009). Isolation and Short-Term Culture of
E.M. (2002). Targeted disruption of the Akap4 gene causes defects in sperm Mouse Spermatocytes for Analysis of Meiosis (Humana Press).
flagellum and motility. Dev. Biol. 248, 331–342. Sassone-Corsi, P. (2002). Unique chromatin remodeling and transcriptional
Minajigi, A., Froberg, J., Wei, C., Sunwoo, H., Kesner, B., Colognori, D., regulation in spermatogenesis. Science 296, 2176–2178.
Lessing, D., Payer, B., Boukhali, M., Haas, W., and Lee, J.T. (2015). Schwarzer, W., Abdennur, N., Goloborodko, A., Pekowska, A., Fudenberg, G.,
Chromosomes. A comprehensive Xist interactome reveals cohesin repulsion Loe-Mie, Y., Fonseca, N.A., Huber, W., H Haering, C., Mirny, L., and Spitz, F.
and an RNA-directed chromosome conformation. Science 349, aab2276. (2017). Two independent modes of chromatin organization revealed by cohe-
Nagano, T., Lubling, Y., Várnai, C., Dudley, C., Leung, W., Baran, Y., sin removal. Nature 551, 51–56.
Mendelson Cohen, N., Wingett, S., Fraser, P., and Tanay, A. (2017). Cell-cycle Servant, N., Varoquaux, N., Lajoie, B.R., Viara, E., Chen, C.J., Vert, J.P., Heard,
dynamics of chromosomal organization at single-cell resolution. Nature E., Dekker, J., and Barillot, E. (2015). HiC-Pro: an optimized and flexible pipe-
547, 61–67. line for Hi-C data processing. Genome Biol. 16, 259.
Namekawa, S.H., Park, P.J., Zhang, L.F., Shima, J.E., McCarrey, J.R., Sexton, T., Yaffe, E., Kenigsberg, E., Bantignies, F., Leblanc, B., Hoichman,
Griswold, M.D., and Lee, J.T. (2006). Postmeiotic sex chromatin in the male M., Parrinello, H., Tanay, A., and Cavalli, G. (2012). Three-dimensional folding
germline of mice. Curr. Biol. 16, 660–667. and functional organization principles of the Drosophila genome. Cell 148,
Naumova, N., Imakaev, M., Fudenberg, G., Zhan, Y., Lajoie, B.R., Mirny, L.A., 458–472.
and Dekker, J. (2013). Organization of the mitotic chromosome. Science 342, Smallwood, A., and Ren, B. (2013). Genome organization and long-range regu-
948–953. lation of gene expression by enhancers. Curr. Opin. Cell Biol. 25, 387–394.
Nguyen, V.T., Giannoni, F., Dubois, M.-F., Seo, S.-J., Vigneron, M., Kédinger, Solari, A.J. (1980). Synaptosomal complexes and associated structures in mi-
C., and Bensaude, O. (1996). In vivo degradation of RNA polymerase II largest crospread human spermatocytes. Chromosoma 81, 315–337.
subunit triggered by a-amanitin. Nucleic Acids Res. 24, 2924–2929. Stévant, I., Neirijnck, Y., Borel, C., Escoffier, J., Smith, L.B., Antonarakis, S.E.,
Nora, E.P., Lajoie, B.R., Schulz, E.G., Giorgetti, L., Okamoto, I., Servant, N., Dermitzakis, E.T., and Nef, S. (2018). Deciphering cell lineage specification
Piolot, T., van Berkum, N.L., Meisig, J., Sedat, J., et al. (2012). Spatial partition- during male sex determination with single-cell RNA sequencing. Cell Rep.
ing of the regulatory landscape of the X-inactivation centre. Nature 485, 22, 1589–1599.
381–385. Tang, W.W., Kobayashi, T., Irie, N., Dietmann, S., and Surani, M.A. (2016).
Nuebler, J., Fudenberg, G., Imakaev, M., Abdennur, N., and Mirny, L.A. (2018). Specification and epigenetic programming of the human germ line. Nat. Rev.
Chromatin organization by an interplay of loop extrusion and compartmental Genet. 17, 585–600.
segregation. Proc. Natl. Acad. Sci. U S A 115, E6697–E6706. Trapnell, C., Roberts, A., Goff, L., Pertea, G., Kim, D., Kelley, D.R., Pimentel,
Oomen, M.E., Hansen, A., Liu, Y., Darzacq, X., and Dekker, J. (2018). CTCF H., Salzberg, S.L., Rinn, J.L., and Pachter, L. (2012). Differential gene and tran-
sites display cell cycle dependent dynamics in factor binding and nucleosome script expression analysis of RNA-seq experiments with TopHat and Cufflinks.
positioning. bioRxiv. https://doi.org/10.1101/365866. Nat. Protoc. 7, 562–578.

560 Molecular Cell 73, 547–561, February 7, 2019


Trasler, J.M. (2009). Epigenetics in spermatogenesis. Mol. Cell. Endocrinol. cell fate transition during human spermatogenesis. Cell Stem Cell 23,
306, 33–36. 599–614.e4.
Turner, J.M.A. (2007). Meiotic sex chromosome inactivation. Development Yang, F., De La Fuente, R., Leu, N.A., Baumann, C., McLaughlin, K.J., and
134, 1823–1831. Wang, P.J. (2006). Mouse SYCP2 is required for synaptonemal complex as-
Turner, J.M., Mahadevaiah, S.K., Elliott, D.J., Garchon, H.J., Pehrson, J.R., sembly and chromosomal synapsis during male meiosis. J. Cell Biol. 173,
Jaenisch, R., and Burgoyne, P.S. (2002). Meiotic sex chromosome inactivation 497–507.
in male mice with targeted disruptions of Xist. J. Cell Sci. 115, 4097–4105. Yang, Q., Zhang, D., Leng, M., Yang, L., Zhong, L., Cooke, H.J., and Shi, Q.
Valton, A.L., and Dekker, J. (2016). TAD disruption as oncogenic driver. Curr. (2011). Synapsis and meiotic recombination in male Chinese muntjac
Opin. Genet. Dev. 36, 34–40. (Muntiacus reevesi). PLoS ONE 6, e19255.
Vrielynck, N., Chambon, A., Vezon, D., Pereira, L., Chelysheva, L., De Muyt, A., Zamudio, N.M., Chong, S., and O’Bryan, M.K. (2008). Epigenetic regulation in
Mézard, C., Mayer, C., and Grelon, M. (2016). A DNA topoisomerase VI-like male germ cells. Reproduction 136, 131–146.
complex initiates meiotic recombination. Science 351, 939–943. Zhou, X., Lowdon, R.F., Li, D., Lawson, H.A., Madden, P.A.F., Costello, J.F.,
Wang, M., Liu, X., Chang, G., Chen, Y., An, G., Yan, L., Gao, S., Xu, Y., Cui, Y., and Wang, T. (2013). Exploring long-range genome interactions using the
Dong, J., et al. (2018). Single-cell RNA sequencing analysis reveals sequential WashU Epigenome Browser. Nat. Methods 10, 375–376.

Molecular Cell 73, 547–561, February 7, 2019 561


STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER


Antibodies
PGP9.5 Abd serotec Cat # 7863-0504; RRID: AB_2210505
gH2AX (mouse) Merckmillipore Cat # 16-202A; RRID: AB_568825
gH2AX (monkey) Abcam Cat # ab11174; RRID: AB_297813
PNA Sigma Cat # L7381
RNA pol II Active motif Cat # 61667; RRID: AB_2687513
SYCP3 Abcam Cat # ab97672; RRID: AB_10678841
gH2AX Novus Biologicals Cat # NB100-384; RRID: AB_350295
Alexa Fluor 488 Goat anti-Mouse IgG Molecular Probes Cat # A-21121; RRID: AB_141514
Alexa Fluor 555 Donkey anti-Rabbit IgG Molecular Probes Cat # A-31572; RRID: AB_162543
Chemicals, Peptides, and Recombinant Proteins
a-amanitin Sigma Cat # A2263
Critical Commercial Assays
Cell-Light EU Apollo567 In Vitro RIBOBIO Co, China Cat # C10316-1
Imaging Kit
Deposited Data
Raw and analyzed data This paper GEO: GSE109344
Imaging Data (Mendeley) This paper https://doi.org/10.17632/m87bv8xyb7.1
Monkey fibroblast Hi-C Deletion of DXZ4 on the human inactive GSM1847539
X chromosome alters higher-order genome
architecture
Mouse fibroblast Hi-C Deletion of DXZ4 on the human inactive GSM1847533; GSM1847534
X chromosome alters higher-order genome
architecture
PAC H3K4me3 Parallel evolution of male germline epigenetic GSM1673979; GSM1673987
poising and somatic development in animals
RS H3K4me3 Parallel evolution of male germline epigenetic GSM1673983; GSM1673991
poising and somatic development in animals
mESC Hi-C Multiscale 3D Genome Rewiring during Mouse GSM2533818; GSM2533819;
Neural Development GSM2533820; GSM2533821
HCT116 wild type Hi-C Cohesin Loss Eliminates All Loop Domains GSE104333
HCT116 RAD21 degradation Hi-C Cohesin Loss Eliminates All Loop Domains GSE104333
Drosophila Hi-C Evolutionarily Conserved Principles Predict GSE80702
3D Chromatin Organization
Mouse MII oocyte Hi-C Allelic reprogramming of 3D chromatin GSM2203829; GSM2203830
architecture uring early mammalian development
Somatic tissue RNA-seq ENCODE GSE36026
Spermatocyte microarray A Global Expression Switch Marks Pachytene GSE21447
Initiation during Mouse Male Meiosis
Experimental Models: Organisms/Strains
Rhesus monkey Yunnan Key Laboratory of Primate Biomedical ID#100127, ID#100157, ID#100159,
Research, China ID#120675
mouse (PWK/B6) The Jackson Laboratory N/A
mouse B6 Sycp2 null Dr. P. Jeremy Wang Lab N/A
mouse B6 Top6bl null Dr. Qinghua Shi Lab N/A
(Continued on next page)

e1 Molecular Cell 73, 547–561.e1–e6, February 7, 2019


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Software and Algorithms
HiC-Pro Servant et al., 2015 https://github.com/nservant/HiC-Pro
Bowtie2 Langmead and Salzberg, 2012 http://bowtie-bio.sourceforge.net/bowtie2/
index.shtml
TopHat Trapnell et al., 2012 http://ccb.jhu.edu/software/tophat/
index.shtml
Cufflinks Trapnell et al., 2012 http://cole-trapnell-lab.github.io/cufflinks/
HiCPlotter Akdemir and Chin, 2015 https://github.com/kcakdemir/HiCPlotter
WashU Epigenome Browser Zhou et al., 2013 http://epigenomegateway.wustl.edu/
Cluster de Hoon et al., 2004 http://bonsai.hgc.jp/mdehoon/software/
cluster/index.html
Juicer Tools Durand et al., 2016a; Durand et al., 2016b https://github.com/aidenlab/juicer/wiki/
Juicer-Tools-Quick-Start
HICCUPS Durand et al., 2016a; Durand et al., 2016b https://github.com/aidenlab/juicer/wiki/
HiCCUPS
Java TreeView Saldanha, 2004 http://jtreeview.sourceforge.net/
R version 3.4.2 https://cran.r-project.org/
Python version 2.7.13 https://www.python.org/
Trim Galore Babraham Bioinformatics - Trim Galore! https://www.bioinformatics.babraham.ac.
Babraham Bioinformatics uk/projects/trim_galore/
Other
STA-PUT velocity sedimentation cell ProScience INC Cat # 56700-012
separator

ANIMALS

6-month-old and 5-year-old rhesus monkeys were used in this study, and surgeries were performed to obtain testes from anesthe-
tized animals. Spermatogenic cells with the highest purity from four individuals (ID#100127, ID#100157, ID#100159 and ID#120675)
were used for further studies. The animals were individually caged in the animal rooms with a 12-hour light and 12-hour darkness
cycle. The room was maintained with a temperature between 18 to 26 C and with a relative humidity of 40% to 70%. Animals
were fed twice per day with commercial monkey chow (LabDiet, Harlan Laboratories, Inc., USA). Fresh fruits and vegetables were
supplemented once per day. All procedures were approved by the Institutional Animal Care and Use Committee of Kunming Univer-
sity of Science and Technology, and were carried out in accordance with the Guide for the Care and Use of Laboratory Animals. The
animal facility is accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care International.
For the mouse PAC, six 3-month-old F1 mice (generated with male PWK/PhJ mice and female C57BL/6N mice) were used. The F1
mice maintenance and experimental procedures used in current study were carried out according to guidelines of Institutional Animal
Care and Use Committee (IACUC) of Tsinghua University, Beijing, China.

SPERMATOGENIC CELL SEPARATION

For the PAC and RS, we separated them from the testes of 5-year-old rhesus monkeys. A complete spermatogenesis process could
be observed in the seminiferous tubules of testis tissue at this age, through hematoxylin-eosin staining and stage-specific antibody
staining (gH2AX and PNA). Testes of each adult animal were harvested via standard surgical procedure and immediately placed into
DMEM medium. After removing laminae parietalis, the testicular tissues were placed in 10 cm dish with DMEM/F12 (Cat # 11330032,
Life technology) and then cut by surgical scissors into 12 mm3. Small pieces were washed twice in DMEM/F12 to remove blood cell
contamination and further minced through successive cutting. First digestion were incubated in 37 C for 15 min with a DMEM solu-
tion (Cat # 11965-084, Life technology) containing 2mg/ml Collagenase IV (Cat # C5138, Sigma) and 2mg/ml DNase I (Cat # DN25,
Sigma). After spinning down at 600 g, the pellet was then incubated with second enzyme digestion mixture containing 2mg/ml colla-
genase, 2mg/ml DNase I and 0.25% trypsin EDTA (Cat # 25200-114, Life technology) for approximately 15 min. Digestion was
stopped by 10% Fetal Bovine Serum (Cat # 10099-141, Life technology) and the cell suspension was filtered through 100 mm and
40 mm Strainers (Cat # 352340, BD).

Molecular Cell 73, 547–561.e1–e6, February 7, 2019 e2


For the spermatogonia, we separated them from the testes of 6-month-old rhesus monkeys. Only spermatogonia could be
observed in the seminiferous tubules of testes at this age through hematoxylin-eosin staining and stage-specific antibody staining
(PGP9.5). Cells were also collected through two step enzyme digestion. The first step used an enzyme solution in DMEM and the
second step used an enzyme solution in 0.25% trypsin EDTA, both containing 1mg/ml collagenase, 1mg/ml DNase I and
2.5 mg/ml hyaluronidase (Cat # H3884, Sigma).
For mature sperm, we collected them through swimming from the epididymis of the 5-year-old animals in a 37  C water bath.

SPERMATOGENIC CELL PURIFICATION

STA-PUT velocity sedimentation cell separator (Chamber Cat # 56700-012, ProScience INC, Canada) was applied to generating sep-
aration of suspensions of living spermatogenic cells on the basis of differences in cell sedimentation rate through the gravity following
the method previously described (Bryant et al., 2013) and the manual of STA-PUT (2010 edition, ProScience INC, Canada). For adult
monkeys (5-year-old), 1,600ml testis cell suspension (2-5 3 107cell/ml) was used with 3hr sedimentation. For neonate monkeys
(6 months), 600 mL cell suspension was used with 1.5hr sedimentation. According to the visible sedimentation bands, cell suspen-
sion was gradually collected through an automatic collector (Cat # DBS-160F, Jinke, China) after velocity sedimentation. All steps for
separation were performed at 4 C.
For spermatogonia, each replicate was collected from pooled samples from several monkeys, while for other stages each replicate
was from an individual monkey.
gH2AX positive spermatocytes were purified from from Sycp2 null mice (Yang et al., 2006) and Top6bl null mice (Robert et al., 2016)
following the method previously described (Bryant et al., 2013) and the instructions of STA-PUT cell separator.

IMMUNOSTAINING

Small pieces of monkey testis tissue were fixed in Hartman’s solution (H0290, Sigma) or 4% paraformaldehyde solution (PFA, P6148,
Sigma). Paraffinized or OCT-embedded section(5mM) were used for hematoxylin-eosin staining and immunofluorescence staining.
Primary antibodies were listed as: PGP9.5 (ubiquitin C-terminal hydrolase 1, UCHL1, 7863-0504, Abd serotec), 1:1000; gH2AX, 1:500
for mouse (Cat # 16-202A, Merckmillipore) and rhesus monkey (Cat # ab11174, Abcam); PNA (Cat # L7381, Sigma), 1:1000. DAPI
(4’,6-diamidino-2-phenylindole, Cat # D9542, Sigma) were applied for counterstaining of cell nucleus.

a-AMANITIN TREATMENT

We first isolated PAC from 3-month-old C57BL/6J (B6) mice using STA-PUT velocity sedimentation cell separator. Pachytene cells
were cultured according to the protocol previously described (Salle et al., 2009) . Briefly, cells were washed twice with enriched
Krebs-Ringer bicarbonate medium and then cultured with spermatocyte culture medium (Salle et al., 2009) containing 200mg/ml
a-amanitin (Cat # A2263, Sigma) at 32 C for 11 hours. Next, cells were collected and purified through 25% Percoll gradients
(Cat # P4937, Sigma) to further remove dead cells and debris. The majority of viable pachytene cells were applied for Hi-C analysis.
The absence of transcription was verified by BrUTP fluorescent staining following the protocol of a commercial kit (Cat # C10316-1,
Cell-Light EU Apollo567 In Vitro Imaging Kit, RIBOBIO Co, China). The RNA Pol II immunofluorescence staining was performed us-
ing anti-RNA pol II (Cat # 61667, Active motif), 1:1000. Pictures were taken by confocal microscopy (LSM 700, Zeiss, Pleasan-
ton, USA).

SPERMATOCYTE MICRO-SPREADING AND IMMUNOSTAINING

Top6bl null mice spermatocyte chromosome preparations and immunofluorescence were performed as previously described (Peters
et al., 1997; Yang et al., 2011), with the following modifications. Spermatocytes were fixed on slides with 1% PFA containing 0.15%
Triton X-100 in a humidified chamber for overnight, then air-dried and washed the slides twice with 0.4% PhotoFlo. For immunoflu-
orescence, the slides were blocked for 30 min with 13 phosphate-buffered saline (PBS) containing 3% nonfat milk, and then
incubated with primary antibodies against synaptonemal complex protein 3 (SYCP3) (Abcam, ab97672), 1:200, gH2AX (Novus
Biologicals, NB100-384) 1:5000 overnight at room temperature in a humidified chamber. Slides were washed around three times
in 1 3 PBS containing 0.1% Triton X-100. Secondary antibodies (Alexa Fluor 488 Goat anti-Mouse IgG, Molecular Probes
A-21121, 1:100; Alexa Fluor 555 Donkey anti-Rabbit IgG, Molecular Probes A31572, 1:250) were applied for 1 hr at 37 C in a humid-
ified chamber. Both primary and secondary antibodies were diluted in 1 3 PBS containing 3% nonfat milk. After secondary antibody
incubation, three washes were performed in PBST (1 3 PBS containing 0.1%Triton X-100) and the slides were mounted with
VECTASHIELD mounting medium (H-1000, Vector Laboratories). Images were captured using a Nikon C2 Confocal Microscope sys-
tem connected to a CCD camera and analyzed using the Image-Pro Plus software.

e3 Molecular Cell 73, 547–561.e1–e6, February 7, 2019


sisHi-C LIBRARY GENERATION AND SEQUENCING

The sisHi-C library generation was performed as described previously (Du et al., 2017). Briefly, spermatogenetic cells were fixed with
1% formaldehyde at room temperature (RT) for 10 min. Formaldehyde was quenched with glycine for 10min at RT. Cells were washed
with 1XPBS for two times and then lysed in 50ul lysis buffer (10mM Tris-HCl pH7.4, 10mM NaCl, 0.1mM EDTA, 0.5% NP-40 and pro-
teinase inhibitor) on ice for 50 min. After spinning at 3000rpm/min in 4 C for 5 min, the supernatant was discarded with a pipette care-
fully. Chromatin was solubilized in 0.5% SDS and incubated at 62 C for 10 min. SDS was quenched by 10% Triton X-100 at 37 C for
30 min. Then the nuclei were digested with 50U MboI at 37 C overnight with rotation. MboI was then inactivated at 62 C for
20 minutes. To fill in the biotin to the DNA, dATP, dGTP, dTTP, biotin-14-dCTP and Klenow were added to the solution and the re-
action was carried out at 37 C for 1.5 hours with rotation. The fragments were ligated at RT for 6 hours with rotation. This was followed
by reversal of crosslink and DNA purification. DNA was sheared to 300-500bp with Covaris M220. The biotin-labeled DNA was then
pulled down with 10ul Dynabeads MyOne Streptavidin C1 (Life Technology). Sequencing library preparation was performed on
beads, including end-repair, dATP tailing and adaptor-ligation. DNA was eluted twice by adding 20ul water to the tube and incubation
at 66 C for 20 minutes. 9-15 cycles of PCR amplification were performed with Extaq (Takara). Finally, size selection was done with
AMPure XP beads and fragments ranging from 200bp to 1000bp were selected. All the libraries were sequenced on Illumina
HiSeq2500 or HiSeq XTen according to the manufacturer’s instruction. For sperm samples, the lysis buffer (same as that for other
cell types) was added with 0.05% L-a-Lysophosphatidylcholine (Cat # L4129, Sigma).

RNA-SEQ LIBRARY PREPARATION AND SEQUENCING

The RNA-seq libraries were generated using the Smart-seq2 protocol as described previously with minor modification (Picelli et al.,
2014). Cells were lysed in hypotonic lysis buffer (Cat # M334, Amresco), and the polyadenylated mRNAs were captured by the PolyT
primers. After 3–10 min lysis at 72  C, the Smart-seq2 reverse transcription reactions were performed. After pre-amplification and
AMPure XP beads purification, cDNAs were sheared by Covaris and were subject to Illumina TruSeq library preparation. All libraries
were sequenced on Illumina HiSeq 2500 or HiSeq XTen according to the manufacturer’s instruction.

DATA ANALYSIS

Hi-C data mapping


Paired end raw reads of Hi-C libraries were aligned, processed and iteratively corrected using HiC-Pro (version 2.7.1b) as described
(Servant et al., 2015). Briefly, sequencing reads were first independently aligned to the monkey reference genome (rheMac2) or
mouse reference genome (mm9) using the bowtie2 end-to-end algorithm and ‘‘-very-sensitive’’ option. To rescue the chimeric frag-
ments spanning the ligation junction, the ligation site was detected and the 50 fraction of the reads was aligned back to the reference
genome. Unmapped reads, multiple mapped reads and singletons were then discarded. Pairs of aligned reads were then assigned to
MboI restriction fragments. Read pairs from uncut DNA, self-circle ligation and PCR artifacts were filtered out and the valid read pairs
involving two different restriction fragments were used to build the contact matrix. Valid read pairs were then binned at a specific
resolution by dividing the genome into bins of equal size. We chose 100-kb bin size for examination of global interaction patterns
of the whole chromosome, and 40-kb bin size to show local interactions and to perform TAD calling. Then the binned interaction
matrices were normalized using the iterative correction method (Imakaev et al., 2012; Servant et al., 2015) to correct the biases
such as GC content, mappability and effective fragment length in Hi-C data.

RNA-seq data processing


All RNA-seq data were mapped to the monkey reference genome (rheMac2) or mouse reference genome (mm9) by TopHat (version
2.0.11). The gene expression level was calculated by Cufflinks (version 2.0.2) (Trapnell et al., 2012) using the refFlat database from the
UCSC genome browser. Only uniquely mapped reads were kept for further analysis. PRO-seq dataset were obtained from previous
study (Rao et al., 2017).

Reproducibility of sisHi-C data


The correlation between sisHi-C replicates was calculated with 100-kb bins. As the interaction matrix is highly skewed toward prox-
imal interactions, we restricted the analysis to a maximum distance of 50 bins (5Mb) as previously described (Dixon et al., 2012). Then
Pearson correlation coefficient between replicates was calculated for each sample.

Analysis of TADs
40-kb resolution sisHi-C matrices were used for TAD boundaries identification by calculating the insulation score of each bin as pre-
viously described (Crane et al., 2015). Briefly, the insulation score was calculated by sliding a 1Mb X 1Mb square along the diagonal of
the interaction matrix for every chromosome. A 200-kb window was used for calculation of the delta vector. Boundaries with a
‘‘boundary strength’’ less than 0.25 were removed. Insulation scores were plotted around all fibroblast TADs as well as their nearby
regions (+/ 0.5 TAD length).

Molecular Cell 73, 547–561.e1–e6, February 7, 2019 e4


Average interaction heatmap of TADs
All fibroblast TADs were used as representatives of mature TADs. We then plotted the composite interaction frequency by averaging
all TADs along stages of spermatogenesis. The resulted matrices were then normalized by the average levels of the matrix values to
make the sum of matrices for different stages equal.

Identification of conventional chromatin compartments and Refined-A/B


Conventional chromatin compartments A and B were identified with a method described previously (Lieberman-Aiden et al., 2009)
with some modifications. The normalized 100 kb interaction matrices for each stage were used in this analysis. First, the bins that
have no interactions with any other bins were removed before expected interaction matrices were calculated. Observed/Expected
matrices were generated using a sliding window approach (Dixon et al., 2015) with the bin size of 400 kb and the step size of 100 kb.
Principal component analysis was performed on the correlation matrices generated from the observed/expected matrices. The first
principal component of the correlation matrices together with gene density were used to identify A/B compartments. In this analysis,
the correlation matrices were calculated according to the interaction matrices separated according to the location of predicted cen-
tromeres for each chromosome, as the principal component often reflects the partitioning of chromosome arms (Lieberman-Aiden
et al., 2009).
As for Refined-A/B compartment, the calling method were similar to the conventional chromatin compartments, with the matrices
restricted to 10Mb one-by-one instead of whole chromosome arms. The PC1 value generated by those restricted matrices were
referred to as local PC1.

Hi-C interaction heatmap, differential interaction heatmap, conventional correlation heatmap and Refined-A/B
correlation heatmap
HiCPlotter version 0.6.05.compare (Akdemir and Chin, 2015) was used to generate all chromosome-wide as well as the zoom-in
views of interaction frequency heatmap in this study. The ‘‘triangle’’ interaction heatmap to show TADs were generated by WashU
Epigenome Browser (Zhou et al., 2013). The differential interaction heatmaps were calculated as previously described (Du et al.,
2017). The normalized interaction matrices (100-kb and 40-kb bin) with similar sequencing-depth of one stage were subtracted
from the next stage. Both the conventional compartment correlation heatmap and the Refined-A/B compartment heatmap were
generated with Java TreeView according to the corresponding correlation matrices.

Analysis of gene density and H3K4me3 signal in compartments A/B


To examine the relationship between gene density, histone modification and chromatin compartments, the number of genes per
100-kb bins and the H3K4me3 ChIP-seq enrichment (Lesch et al., 2016) (RPKM, 2-kb bin) were computed for the entire genome.
For each compartment region, the average enrichment was computed and shown in boxplots.

Hierarchical clustering analysis


The hierarchical clustering analysis was conducted using the interaction frequency of whole chromosome 1 (100kb bin) using an R
package (ape) based on the Pearson correlation between each pair of datasets. The distance between two datasets was calculated
by (1- correlation).

Analysis of interactions between HT/LT regions


We first identified all regions (500kb) with transcription levels above genome average and at least 3 fold more than nearby regions as
HT regions. We chose 500kb regions which are smaller than conventional compartment A (which has sizes of 1.4-4Mb), but are long
enough to allow high-resolution analysis of Hi-C datasets (25kb x 20 bin). It is also comparable to the average length of Refined-A
regions. The nearby regions with transcription levels below genome average and less than 1/3 of HT regions were identified as LT
regions.
To eliminate the bias of chromatin interaction introduced by genomic distance, we used Observed/Expected (OE) matrix generated
by Juicer (Durand et al., 2016a; Durand et al., 2016b) with 25-kb bin (200 million cis-long reads for each sample) to calculate the
interaction between different HT/HT, LT/LT and HT/LT regions (from 2Mb to 10Mb). We plotted the composite interaction frequency
by averaging pairs of HT/HT, LT/LT and HT/LT regions. The resulted matrices were then normalized by dividing the average levels of
the matrix values to generate the log ratio values.

Loops analysis
The loops in this study were called by HICCUPS (Durand et al., 2016a; Durand et al., 2016b) with the default parameters. We then
calculated the composite interaction frequency by averaging all loops for both monkey fibroblast and PAC cells, respectively. The
resulted matrices were then normalized by the average levels of the matrix values to make the sum of matrices for different stages
equal.

e5 Molecular Cell 73, 547–561.e1–e6, February 7, 2019


The P(s) analysis
The P(s) was calculated with normalized interaction matrices in 100-kb resolution as previously described (Naumova et al., 2013). We
first divided distances into logarithmically spaced bins with increasing factor 1.15: (100kb, 100kb*1.15, 100kb*1.152.). Then for each
bin, we counted the number of interactions at corresponding distances. To obtain the probability P(s), we divided the number of in-
teractions in each bin by the total number of possible region pairs. The P(s) was further normalized so that the sum over the range of
the distances is 1.

Conventional compartment segregation strength


The conventional compartment segregation levels of mutant spermatocytes were calculated based on regions enriched for the CpG
island clusters (CGI forests, representing promoter-rich regions and compartment A) and regions depleted for CpG island (CGI prai-
rie, representing promoter-poor regions and compartment B) (Liu et al., 2018). CpG island rich and poor regions were used as they
are independent of cell types and were shown to well capture most conventional compartments A/B (Jabbari and Bernardi, 2017; Liu
et al., 2018). Segregation strengths were calculated as (AA+BB)/(AB). We chose CGI forest/prairie as they are similar to compart-
ments A/B but are independent of cell types. First, we removed the local interactions that were shorter than 2Mb. Then we assigned
the interactions into two categories: interactions between the same classes (forest versus prairie, F versus P) and the interactions
within same classes (F versus F and P versus P). The ratios of average interaction frequency between those two categories were
calculated for each chromosome (Du et al., 2017).

Molecular Cell 73, 547–561.e1–e6, February 7, 2019 e6

You might also like