**Male Germline Chromatin**

Editors

**Darren Griffin Peter Ellis**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Darren Griffin University of Kent UK

Peter Ellis University of Kent UK

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Genes* (ISSN 2073-4425) (available at: https://www.mdpi.com/journal/genes/special issues/ Male-Germline-Chromatin).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03936-854-9 (Hbk) ISBN 978-3-03936-855-6 (PDF)**

Cover image courtesy of Ben Skinner and Peter Ellis.

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

## **Contents**


#### **Heather E. Fice and Bernard Robaire**

Telomere Dynamics Throughout Spermatogenesis Reprinted from: *Genes* **2019**, *10*, 525, doi:10.3390/genes10070525 ................... **123**

## **About the Editors**

**Darren Griffin** received his Bachelor of Science and Doctor of Science degrees from the University of Manchester and University College London, respectively. After postdoctoral stints at Case Western Reserve University and the University of Cambridge, he landed his first academic post at Brunel University before settling at the University of Kent, where he's been for the last 15+ years. He has worked under the mentorship of Professors Joy Delhanty, Christine Harrison, Terry Hassold, Alan Handyside, and Malcolm Ferguson-Smith. He is President of the International Chromosome and Genome Society, a Fellow of the Royal College of Pathologists, the Royal Society of Biology, and the Royal Society of Arts, Manufacture, and Commerce. He sits on the faculty of CoGen (controversies in genetics) and has previously sat on the board of the Preimplantation Genetic Diagnosis International Society (PGDIS), organizing its annual meeting in 2014. Darren is a world leader in cytogenetics. He performed the first successful cytogenetic preimplantation genetic diagnosis (sexing of IVF embryos) and, more recently, played a significant role in the development of Karyomapping, a universal test for genetic disease in IVF, an approach he now applies to cattle. In his 30+ years of scientific research, he has co-authored over 200 scientific publications, mainly on the cytogenetics of reproduction and evolution, most recently providing insight into the karyotypes of dinosaurs. He is a prolific science communicator, making every effort to make scientific research publicly accessible (both his own and others) and is an enthusiastic proponent for the benefits of interdisciplinary research endeavor. He has supervised over 35 PhD students to completion and his work appears consistently in national and international news. He currently runs a vibrant research lab of about 20 people (including a program of externally supervised students) and maintains commercial interests in the outcomes of research findings, liaising with companies in the agricultural sector in the area of fertility screening. Darren is a member of the Centre for Interdisciplinary Studies of Reproduction (CISoR). He also regularly coordinates the International Chromosome Conferences and the Pig Breeders' Round Table.

**Peter Ellis** is a Lecturer in Molecular Genetics and Reproduction at the University of Kent. Key findings from his works include the identification of novel genes on the mouse Y chromosome that affect sperm head shape and fertility; the discovery of a genomic conflict or arms race between the X and Y chromosomes in mice as they compete to influence offspring sex ratio, which in turn has dramatically affected the structural and functional content of both chromosomes; and the identification of mechanisms regulating meiotic and post-meiotic transcriptional silencing of the sex chromosomes. His laboratory investigates the molecular biology of reproduction, the conflicting roles played by sex-linked genes in regulating this process, and the relationship between DNA damage repair mechanisms and the checkpoints governing meiotic progression.

## *Editorial* **Form from Function, Order from Chaos in Male Germline Chromatin**

#### **Peter J. I. Ellis and Darren K. Gri**ffi**n \***

School of Biosciences and Centre for Interdisciplinary Studies of Reproduction, University of Kent, Giles Lane, Canterbury CT2 7NJ, UK; P.J.I.Ellis@kent.ac.uk

**\*** Correspondence: d.k.griffin@kent.ac.uk

Received: 28 January 2020; Accepted: 9 February 2020; Published: 18 February 2020

**Abstract:** Spermatogenesis requires radical restructuring of germline chromatin at multiple stages, involving co-ordinated waves of DNA methylation and demethylation, histone modification, replacement and removal occurring before, during and after meiosis. This Special Issue has drawn together papers addressing many aspects of chromatin organization and dynamics in the male germ line, in humans and in model organisms. Two major themes emerge from these studies: the first is the functional significance of nuclear organisation in the developing germline; the second is the interplay between sperm chromatin structure and susceptibility to DNA damage and mutation. The consequences of these aspects for fertility, both in humans and other animals, is a major health and social welfare issue and this is reflected in these nine exciting manuscripts.

**Keywords:** spermatogenesis; chromatin; nuclear organisation; DNA oxidation; DNA fragmentation; epigenetic inheritance; histone retention; assisted reproduction; in vitro fertilisation

One of the most fundamental requirements in spermatogenesis is the need to develop male germ cells to undergo radical restructuring of their chromatin. Occurring at multiple stages before, during and after meiosis, it involves coordinated waves of DNA methylation and demethylation. It also involves histone modification, replacement and removal. In this Special Issue, we draw together novel studies and contemporary reviews addressing various aspects of chromatin organization and dynamics in the male germ line, and consider both humans and model organisms. Two major themes emerge from these exciting studies: the first being the functional significance of nuclear organization in the developing germline and the second is the interplay between sperm chromatin structure and DNA damage. The consequence of these aspects for fertility, both in humans and other animals, is a major health and social welfare issue.

Fernanda López-Moncada and colleagues address the question of whether chromosomal reorganization alters gene expression during meiotic prophase [1]. In particular, they show that Robertsonian fusions involving chromosomes bearing nucleolar organizing regions (NOR) perturb their normal organization and nucleolar functionality. In post-meiotic spermatids, Jonathan Riel and colleagues show that Sly deficiency is not the only reason for infertility in mice with deletions on their Y chromosome. Rather, it appears that some other Yq-encoded gene is likely to be required to allow Sly to bind to chromatin and to exert its normal regulatory functions [2].

Four studies examine chromosome organisation in mature sperm. First, Dimitris Ioannou and Helen Tempest show that, while chromosomes in human sperm do indeed to form hairpin loops, as predicted from studies in other species, their centromeres are not organized in the classic "chromocenter" arrangement seen in model species such as mice [3]. Second, Heather Fice and Bernard Robaire confirm that relative sperm telomere length does indeed decrease during ageing in rodents, but, crucially, only in inbred strains [4]. Moreover, the demonstration that relative telomere length changes as sperm pass through the epididymis is a novel one. Third, Ben Skinner and colleagues address the question of

whether chromosome territory organization is conserved between species, demonstrating that mouse chromosomes have retained the same sub-nuclear "address" for over two million years of evolutionary history [5]. Finally, Alexandre Champroux and colleagues turn to the possible deleterious effect of oxidative damage on sperm DNA organization. The surprising finding is that territory organization is largely robust in response to this challenge, with the overall organization of the chromosome territories being maintained even in the face of oxidative DNA damage. However, this organization is then disrupted in response to the treatment, illustrated by the reducing agents, signifying that oxidative damage may perturb chromosome decondensation following fertilization [6].

The theme of DNA damage is covered extensively in our two review articles. While DNA damage is usually regarded as a pathological, abnormal process, Tiphanie Cavé and colleagues review the role of endogenous, naturally-occurring DNA strand breaks created during chromatin remodeling [7]. This is an emerging field with profound implications for our understanding of the processes generating structural variations and polymorphisms within the genome, and the male versus female bias of specific mutational signatures. In a similar vein, but with a more clinical focus, Jordi Ribas-Maynou and Jordi Benet take a look at the differential reproductive effects on male fertility of single and double strand sperm DNA damage, respectively [8]. By their account, single-strand DNA breaks are present as scattered break points throughout the genome, whereas double-strand DNA breaks are mainly localized and attached to the sperm nuclear matrix. Single strand breaks are related to oxidative stress and impede pregnancy rates, whereas double strand breaks may be related to a lack of meiotic DNA repair—or to genome reconfiguration by topoisomerases, as highlighted by Cavé and colleagues—and lead to increased miscarriage rates, low embryo quality and implantation failure during ICSI.

Finally, we are particularly proud of the use of novel methods for studying the interplay between chromatin structure and the susceptibility to DNA damage and mutation. Indeed, this Special Issue boasts three new methodological approaches with Sheryl Homa and colleagues comparing two means of measuring oxidative stress (concluding that both used in tandem are better than one in isolation) [9] and both the Skinner and Champroux papers taking novel approaches to quantify the localization of chromosome territories in asymmetrical nuclei [5,6].

Collectively, these papers serve to highlight the importance of understanding male germline chromatin organisation in order to appreciate how specific regions of the genome may well be exposed to different stressors, remodeled, and activated before or after others immediately following fertilization. This, in turn, has downstream effects on both male germline mutagenesis and for early embryonic development; with profound subsequent implications for understanding natural fertility and improving assisted reproduction techniques.

Taken together, this unique collection of studies will, we hope, serve as a benchmark for a deeper understanding of the fundamental mechanisms perpetuating our germline.

**Acknowledgments:** Ellis is funded by the BBSRC, grant number BB/N000463/1 and the Leverhulme Trust, grant number RPG-2019-194.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Nuclear Integrity but Not Topology of Mouse Sperm Chromosome is Affected by Oxidative DNA Damage**

## **Alexandre Champroux, Christelle Damon-Soubeyrand, Chantal Goubely, Stephanie Bravard, Joelle Henry-Berger, Rachel Guiton, Fabrice Saez, Joel Drevet \* and Ayhan Kocer \***

GReD "Genetics, Reproduction & Development" Laboratory, UMR CNRS 6293, INSERM U1103, Université Clermont Auvergne, 28 Place Henri Dunant, 63000 Clermont-Ferrand, France; alexandre.champroux@uca.fr (A.C.); christelle.soubeyrand-damon@uca.fr (C.D.-S.); Chantal.goubely@uca.fr (C.G.); Stephanie.bravard@uca.fr (S.B.); joelle.henry@uca.fr (J.H.-B.); rachel.guiton@uca.fr (R.G.); fabrice.saez@uca.fr (F.S.)

**\*** Correspondence: joel.drevet@uca.fr (J.D.); ayhan.kocer@uca.fr (A.K.)

Received: 11 September 2018; Accepted: 15 October 2018; Published: 17 October 2018

**Abstract:** Recent studies have revealed a well-defined higher order of chromosome architecture, named chromosome territories, in the human sperm nuclei. The purpose of this work was, first, to investigate the topology of a selected number of chromosomes in murine sperm; second, to evaluate whether sperm DNA damage has any consequence on chromosome architecture. Using fluorescence in situ hybridization, confocal microscopy, and 3D-reconstruction approaches we demonstrate that chromosome positioning in the mouse sperm nucleus is not random. Some chromosomes tend to occupy preferentially discrete positions, while others, such as chromosome 2 in the mouse sperm nucleus are less defined. Using a mouse transgenic model (*Gpx5*−/−) of sperm nuclear oxidation, we show that oxidative DNA damage does not disrupt chromosome organization. However, when looking at specific nuclear 3D-parameters, we observed that they were significantly affected in the transgenic sperm, compared to the wild-type. Mild reductive DNA challenge confirmed the fragility of the organization of the oxidized sperm nucleus, which may have unforeseen consequences during post-fertilization events. These data suggest that in addition to the sperm DNA fragmentation, which is already known to modify sperm nucleus organization, The more frequent and, to date, The less highly-regarded phenomenon of sperm DNA oxidation also affects sperm chromatin packaging.

**Keywords:** mouse sperm chromatin; chromosome organization; nuclear-3D-parameters

#### **1. Introduction**

The mammalian spermatozoon is a highly-differentiated cell produced by the testis during a long and complex process called spermatogenesis. Following successive steps that lead to the multiplication and the production of haploid germ cells through the meiotic program, spermatids undergo a long phase of cyto-differentiation (the so-called spermiogenesis phase) to form highly polarized spermatozoa. Unique characteristics of these cells are featured by the quasi-complete loss of the cytoplasmic content, appearance of the flagella apparatus and drastic size reduction of the nuclear compartment. These major cytological changes give rise to the tiniest mammalian cell type that has the ability to move in order to fulfil its function of delivering to its target, The oocyte, The compacted and, consequently, protected paternal genomic moiety. Up to the spermatid stage the germ cell chromatin presents a somatic organization consisting of short (147 bp) DNA segments wrapped around a histone octamer to form a nucleosome [1]. During spermiogenesis, most (but not all) canonical histone core proteins (H3, H4, H2A, and H2B) are replaced by testis-specific histone variants such as TH2B, H3t, H2AL2 & 5 [2–5]. It is assumed that the inclusion of such variants allows

a more dynamic chromatin structure that permits the upcoming changes. Subsequently, histones, both canonical and testicular variants, are largely replaced by small basic proteins called transition nuclear proteins (Tnps), and find themselves replaced by even smaller and more basic proteins called protamines [6,7]. Protamines and DNA organize themselves into a ring-shaped structure called a toroid, containing up to 100 kb of DNA that ultimately piles up along the chromosomes, greatly increasing the level of the DNA compaction [8–11]. This sequence of events allows a strong nuclear and cell size reduction, when compared to any somatic cell [12]. Together with the fact that these modifications enable optimization of cell mobility, they also contribute to passive protection of the paternal sperm genome in anticipation of its long post-testicular journey to the site of fertilization [13].

Another unique feature of this reshaping of the mammalian sperm, chromatin, is that the supra-organization of the chromosomal chromatin is also tightly ordered and conserved from one sperm cell to another. This has led to the observation that chromosomes are not randomly distributed in the sperm nucleus and that they occupy domains, called chromosome territories (CTs) [14–16]. A limited number of species have been investigated, to date, and for those analyzed (mainly human) not all chromosomes were mapped in the sperm nucleus, with the exception of the porcine sperm [14]. The localization of specific chromosomal regions such as telomeres and centromeres were also investigated in the human sperm nucleus [17,18]. As is the case in somatic cells, sperm cell chromosomes are attached to a nuclear protein scaffold, called the sperm nuclear matrix, which consolidates the structure [19–21]. Here too, The manner in which chromosomes are attached to the sperm nuclear matrix is unique to that cell lineage and is dissimilar to the somatic situation [19,22]. Two non-exclusive theories have been proposed to explain the positioning of chromosomes in the nucleus of a somatic cell. The first is "gene density" with the assumption that gene-poor chromosomes orient themselves toward the nuclear periphery while gene-rich chromosomes are located toward the nuclear interior [23,24]. The second theory, and in our opinion the more pertinent, takes chromosome size into account since, at least in the human sperm, it appears that small chromosomes are located in the center of the nucleus while larger chromosomes are located at the periphery [16,25,26]. Whether the human sperm nuclear organization reflects that of other mammals is a matter of debate.

For many years it was reported that mature spermatozoa do contain residual histones and that the quantity of the so-called persisting histones was species-specific. Indeed, it was estimated that about 1–2% of mouse, hamster, and bull sperm DNA was still associated with histones [27–29] and that this value increased to 15% in human sperm [30–34]. First, attributed to an incomplete, therefore deficient, spermiogenesis program, it was recently reported that persisting histones in the sperm nucleus were not random, but were deliberately excluded from the histone-to-protamine exchange. Although, there is a controversy regarding the extent and quality of nucleosome retention in mammalian spermatozoa it is clear that histones are found in large domains punctuating the protamine-toroidal stacks along the chromosomes and, in addition, nucleosomes persist at each small string of DNA, connecting the adjacent toroids [20]. The consensual explanation for this situation is that these particular paternal regions that maintain a somatic-like organization will be more prone to reactivation early after fertilization at the onset of the developmental program. In support of this hypothesis were the observations that the genes important for the early developmental program were found located in such histone-containing regions [30–32], and that the origins of the paternal DNA replication necessary, prior to the first division of segmentation, were located in the short histone-containing DNA segments, connecting the toroids and is attached to the nuclear matrix [19,35–38]. It is thought that this ordered-organization of the paternal chromosomes in the sperm nucleus is essential after fertilization, during the sequential decondensation phase of the male nucleus into the male pronucleus [16,39].

In recent years, we have shown in a mouse model that these histone-rich regions, particularly those that are attached to the nuclear matrix were mainly localized at the sperm nuclear periphery and at the base of the sperm nucleus towards the so-called annulus domain [35,40]. In agreement with the lower level of condensation and the peripheral easy access of these histone-associated DNA domains we also demonstrated that these regions were particularly susceptible to DNA damage

and in particular to oxidative DNA damage [35]. We also reported that smaller chromosomes were highly susceptible to DNA oxidation [41] in the mouse sperm nucleus. We demonstrated that this was not related to their content of persisting histones, but rather to the more peripheral and basal position of small chromosomes [36]. These observations led to the conclusion that in contrast to human sperm chromosomal organization, which as mentioned above, showed small chromosomes, located more in the central axis of the sperm nucleus, The situation was different in the mouse. This prompted a more precise analysis of the architecture of the mouse sperm nucleus. In the present study, we used three-dimensional fluorescence in situ hybridization (3D-FISH), confocal microscopy, and computational analysis of 3D structures to analyze the topology of at least twelve mouse sperm chromosomes. This has allowed us to propose the largest map of chromosome territories in murine sperm, to date. Our access to *Gpx5*−/<sup>−</sup> transgenic mice, in addition to wild-type controls, allowed us to conduct an analysis of chromatin organization in what now appears to be a frequent type of sperm nuclear damage, i.e., nuclear oxidation [42]. This mouse model was very pertinent to address this question because we reported earlier that *Gpx5*−/<sup>−</sup> males present mild oxidative sperm DNA damage that does not translate to an increase in either sperm DNA fragmentation or nuclear decondensation. This transgenic mouse model was particularly interesting, therefore, as it dissociates the effect of severe sperm DNA damage from the low-grade DNA oxidation situation commonly seen in infertile patients. Indeed, we recently demonstrated that males in two-thirds of couples entering an infertility program, showed mild to severe sperm DNA oxidation. Our aims were then to investigate whether chromosomal 3D parameters including volume and surface area would be affected by DNA oxidation.

#### **2. Results**

#### *2.1. Localization of Chromosome Territories in Murine Spermatozoa*

Previously, we hypothesized that the localization of chromosomes, in the mouse sperm nucleus, could explain their different susceptibility to oxidative damage, as revealed after immunoprecipitation of the oxidized DNA regions, followed by high throughput sequencing approaches [41]. This statement was supported by the fact that we were able to co-localize the smallest murine chromosome (chromosome 19), with a focal point of oxidative DNA damage, in the *Gpx5*−*/*<sup>−</sup> sperm nucleus [41]. To lend support to this statement, we looked at the nuclear distribution of a total of twelve chromosomes (both long and short chromosomes) using the *FISH* assay, in a whole chromosome-painting approach, in both WT and *Gpx5*−/<sup>−</sup> sperm nuclei. Figure 1 shows representative confocal microscopy photographs going through the middle of the sperm head for each chromosome investigated. To facilitate this analysis, we arbitrarily divided the mouse sperm head into four distinct areas, as schematized in Figure 1. For each selected chromosome, a minimum of three hundred and fifty sperm cells were analyzed and preferential chromosome positions were determined. It is clear that the small chromosomes, including chromosomes 17, 18, and 19, localized to the basal part of the sperm nucleus, whereas a long chromosome, such as chromosome 1, localized preferentially to the ventral area (see Figure S1, supplemental data). Chromosome 15 and the X and Y sex chromosomes also clearly localized to the dorsal area (Figure 1). Assignation to a preferential domain was easy for these chromosomes because a clear preference was found for these particular locations (see Table 1). In contrast, assignation to a preferential area was more difficult for some chromosomes. For example, two chromosomes (3 and 12) were statistically equally-assigned to two sperm head areas, namely, basal and ventral for chromosome 3 and basal and apical for chromosome 12 (Table 1). Chromosome 2 was peculiar as it was equally localized among the four distinct areas (Table 1). When the same analysis was carried out using *Gpx5*−/<sup>−</sup> oxidized sperm, it was clear that no difference was recorded (see Table 1).

**Figure 1.** Chromosome mapping in WT mouse sperm nucleus. Schematic representation of a wild-type (WT) mouse sperm nucleus, arbitrarily divided into four regions (apical, dorsal, ventral, and basal). The position of each selected chromosome was detected by fluorescence in situ hybridization (FISH). Green (FITC) staining represents the chromosome position (n = 350 spermatozoa). Nuclei were stained blue with DAPI. Nuclei were captured in Z-stacks by using confocal microscopy and subjected to deconvolution (Huygens software, Hilversum, The Netherlands). Scale bar represents 5 μm (white line). Chr: Chromosome.

**Table 1.** Regional mapping of chromosomes in WT and *Gpx5*−/<sup>−</sup> mouse sperm nuclei. Chromosome positions are assigned, determined in WT and *Gpx5*−/<sup>−</sup> mouse sperm nuclei, using FISH. Spermatozoa (n = 350) were counted for each chromosome studied and per genotype. The orange box denote the main position of chromosome.


Chr: Chromosome. N.D. not-determined.

Taking advantage of the 3D-reconstructed images we examined two topological parameters (volume and surface area), for each chromosome in the WT genetic background. As shown in supplemental Table S1 and supplemental Figure S1, it is clear that there is a linear relationship between the size of a given chromosome and the volume/surface it occupies in the mouse sperm nucleus. Only chromosome 2 behaved in a peculiar manner, since the linear relationship was validated in only 25% of the analyzed sperm—those in which chromosome 2 localized to the basal area (B in supplemental Table S1 and supplemental Figure S1). Strikingly, when chromosome 2 localized to different areas of the sperm nucleus the linear relationships (volume vs. size and surface vs. size) were lost (supplemental Figure S1). This was particularly true when chromosome 2 was located in the ventral (V) and apical (A) areas and to a lesser extent in the dorsal (D) area. Interestingly, contrasting effects were recorded in these two situations, revealing that when chromosome 2 localized to the ventral and apical areas of the sperm nucleus, its footprint (volume/surface) in the sperm nucleus differed from that when localized to the basal area.

#### *2.2. Centromeres, Telomeres, and Histone-Rich Domains Clustered in the Mouse Sperm Nucleus*

Using immunocytochemistry and *FISH*, we further investigated the localization of particular chromosomal subdomains, namely centromeres and telomeres. To do so, we used a pan-centromere specific H3 variant (CENP-A) antibody to detect this ubiquitous centromeric protein (Figure 2A). 3D reconstruction using Imaris software showed that centromeres aligned and clustered along the dorsal and basal ridges of the sperm head (Figure 2B). A similar localization was observed by *FISH* when looking at telomeres (Figure 2C,D) suggesting that in the mouse sperm nucleus, centromeres and telomeres co-localize. No difference in the localization of centromeres and telomeres was recorded when *Gpx5*−*/*<sup>−</sup> sperm nuclei were examined (data not shown). We used three specific histone antibodies (1 canonical and 2 testis-specific variants, respectively, H3, TH2B, and H2A.Z) to corroborate and complete earlier reported partial observations [35] regarding the localization of persisting histones in the mouse sperm nucleus, in immunofluorescence confocal microscopy approaches, associated with 3D Imaris reconstruction. We confirm the basal and dorsal peripheral localization of these persisting histones and their consistently overlapping localization (Figure 3). The 3D Imaris reconstruction, shown in parallel (right panels) in the same Figure, clearly reveals the basal and dorsal ridge localization of these histone-rich domains in what could be designated a "punk-head" distribution. Topoisomerase 2ß, a sperm nuclear matrix protein (Figure 3), as well as the classical cytoskeleton protein ß-tubulin (Figure 3), also fall into these dorsal peripheral and basal ridge domains as was partly shown in the earlier study [30].

**Figure 2.** Representative image of telomere and centromere positions in WT mouse sperm nucleus. The centromere-specific histone H3 variant (CENP-A, red (**A**,**B**)) and telomeric probes ((**C**,**D**), red) were used in immunofluorescence or FISH approaches, respectively. Nuclei were stained blue with DAPI. Nuclei were captured in Z-stack, using confocal microscopy, and subjected to deconvolution (Huygens software, Netherlands). The 3D models were obtained with Imaris software (Bitplane, Switzerland). The set of views per staining represented is a representative nucleus from a pool of 30 spermatozoa. Scale bar in confocal images represents 5 μm (white line).

**Figure 3.** Representative image of chromatin components in WT mouse sperm nucleus. Representative confocal and different views are shown for each component of sperm chromatin in mouse sperm nucleus: Histone H3, histone variant H2A.Z, testis-specific histone variant TH2B, nuclear matrix protein Topoisomerase-II, and ß-tubulin in WT mouse sperm nucleus. Nuclei are captured in Z-stacks using confocal microscopy and subjected to deconvolution (Huygens software, Netherlands). The 3D models were obtained with Imaris software (Bitplane, Switzerland). The set of views per component is a representative nucleus of thirty spermatozoa.

#### *2.3. Oxidative DNA Damage Does Affect 3D-Parameters of the Mouse Sperm Nucleus*

Taking advantage of the confocal images and the power of the Imaris software analysis, we looked in more detail at sperm nuclear 3D-parameters, including volume and surface area, comparing WT and *Gpx5*−*/*<sup>−</sup> spermatozoa. An average value for each parameter (volume and surface area) was obtained from each sample and each condition tested (untreated, NaOH- or DTT-treated) by looking at a pool of thirty spermatozoa. The data are presented in Table 2. Untreated WT spermatozoa showed a mean nuclear volume of 66 μm3 and a mean nuclear surface area of 93.9 μm2. These parameters were significantly different in *Gpx5*−*/*<sup>−</sup> spermatozoa, which had a mean nuclear volume of 54.8 μm3 (*p* < 0.001) and a mean surface area of 80.2 μm2 (*p* < 0.001), revealing a greater state of nuclear condensation. Examination of the detailed shape of the 3D-reconstructed sperm nuclei revealed repeated differences between the WT and *Gpx5*−*/*<sup>−</sup> animals. As shown in Figure 4, with representative photographs of 3D-reconstructed nuclei, *Gpx5*−*/*<sup>−</sup> sperm nuclei present a smoother surface when compared to the more irregular aspect of the WT sperm nuclei. The use of different mild denaturing treatments, namely DTT (2 mM) or NaOH (1.5 N), revealed distinct reactions when WT sperm were compared to *Gpx5*−*/*<sup>−</sup> sperm and confirmed the specific effect of oxidation on the sperm nucleus. As presented in Table 2, when NaOH was used to produce a mild denaturation of the sperm chromatin (by classical breakage effects on the hydrogen bonds linking DNA base pairs), we recorded/*observed* a significant increase in sperm nuclear volume and surface area, in both genetic backgrounds (WT and *Gpx5*−*/*−). However, *Gpx5*−*/*<sup>−</sup> sperm nuclei remained more condensed than WT following treatment with alkali. In contrast, when DTT (a non-ionic detergent that specifically reduces disulfide bonds to free thiols) was used, we observed a marked effect on both sperm nuclear volume and surface area, in the *Gpx5*−*/*<sup>−</sup> mice, as compared with WT controls (Table 2). This is in agreement with the idea that although *Gpx5*−*/*<sup>−</sup> sperm nuclei appear more condensed, they also appear to be significantly less robust when exposed to a mild, reducing environment. These differences in the nuclear reactivity of oxidized or non-oxidized sperm nuclei, when exposed to mild denaturing conditions, can be visualized, as shown in Figure 4. In panel C (Figure 4C), when no denaturing treatment was performed, The *Gpx5*−*/*<sup>−</sup> sperm nuclei presented the smooth aspect, as noted above. When NaOH was used as a mild denaturing treatment, there was no significant change regarding the smooth shape of the sperm nuclei in either genetic background (Figure 4B). However, when mild denaturation was carried out with DTT, it was obvious that the *Gpx5*−*/*<sup>−</sup> sperm nuclei then presented a dense granular aspect (Figure 4C) that was not observed in the WT.

**Figure 4.** *Cont.*

**Figure 4.** NaOH-mediated or DTT-mediated mild denaturation provokes distinct effects on the WT and the *Gpx5*−*/*<sup>−</sup> nuclei.


**Table 2.** Three-dimensional parameters of sperm nuclei according to treatment and genotype.

Volume and surface area of nuclei were calculated from 3D photographs obtained of Z-stack images, generated with the Imaris software (Bitplane, Switzerland). Nuclei were captured in Z-stacks, using confocal microscopy and subjected to deconvolution (Huygens software, Hilversum, The Netherlands). The resulting distribution of the different parameters are shown in the table for each genotype (WT and *Gpx5*−/−). The mean was calculated on thirty spermatozoa per condition. DTT: Dithiothreitol; NaOH: Sodium hydroxide. 'a' represents *p* < 0.001 for WT no treatment condition; 'b' represents *p* < 0.001 for WT NaOH condition; 'c' represents *p* < 0.001 for WT DTT condition; ' d' represents *p* < 0.001 for no treatment/genotype condition.

#### **3. Discussion**

In recent years, it has become apparent that mammalian sperm nucleus organization has implications for fertilization and early embryogenesis [14,15,43–46]. It was shown, mainly in human spermatozoa, that most chromosomes occupy discrete and well-defined territories in a polar/radial distribution that could be partly related to their size [44–46], The shape/volume of the mature sperm cell and the kinetics of the oocyte-driven decondensation program of the paternal nucleus post-fertilization [47,48]. How this highly-ordered organization of the sperm chromatin is achieved, controlled, and maintained in each sperm cell, throughout spermiogenesis and beyond, is still largely unknown. Whether the human sperm chromatin organization applies to murine sperm and how susceptible this organization is to mild nuclear and DNA damage, as represented by the common situation of sperm DNA oxidative damage, are questions we addressed in this study.

Using FISH experiments, we determined the position of a total of twelve chromosomes in the mouse sperm nucleus. Both short and long autosomes and the two sex chromosomes were analyzed. As reported for the human sperm nucleus, and suggested for other species (including mouse, bovine, pig, and rat), using a smaller subset of chromosome probes when compared to the present work [14–17,45,49–56], chromosome positions in the mouse sperm nucleus were not random. This situation seems to be confined to mammals since a tandem head-to-tail organization of sperm chromosomes, in a defined order, was observed in monotremes and marsupials [57,58] while no particular organization was detected in non-mammals, including chicken and planarian spermatozoa [59,60].

Due to this peculiar, asymmetric hook-shape morphology of the mouse sperm head it was difficult to use a polar/radial axis to map the mouse sperm head, as has been performed in other species [15]. We arbitrarily separated the mouse sperm head into four compartments (apical/basal/dorsal/ventral), while still permitting comparative analyses with other species. In the mouse, smaller chromosomes were found to occupy a basal localization, whereas longer chromosomes were preferentially found in the ventral area with the sex chromosomes located in the dorsal area of the sperm nucleus. This appears to be distinct from the human situation since it was shown that small autosomes as well as sex chromosomes occupy a rather central position in human sperm [16]. Some of the CTs appear to be small while others are larger. Our assumption is that it is both related to the respective size of the chromosomes (since we did observe that there is a positive correlation between the size and the volume of the chromosome, as shown in supplementary Table S1). However, it could also be partly related to the number of times by which the chromosomes—which are folded to fit into the tiny nuclear volume—are longer. Although a preferential position could be assigned for most of the chromosomes examined, this did not hold for all chromosomes. Four chromosomes (chr 3, 7, 9, and 12) were equally assigned to two distinct areas, while one chromosome (chr 2) was very plastic and was found evenly distributed among the four arbitrarily-defined nuclear areas. For those chromosomes that were equally distributed between the two distinct nuclear domains, one explanation could arise from the fact that statistically one out of two spermatozoa examined was either a Y-spermatozoon or an X-spermatozoon. The size difference between the sex chromosomes (both localized in the dorsal area) could explain the alternate positions of these autosomes. This hypothesis is strengthened by the observation that overall Y-sperm and X-sperm show a similar nuclear volume (not shown here) suggesting that the necessary adjustment to accommodate the X or Y chromosome size-difference does not rely on nuclear volume variation. Furthermore, when looking at individual chromosome 3D-parameters (i.e., volume and surface area) we observed that chromosomes 3 and 12 (two chromosomes that show equal occupancy of two distinct locations, basal or ventral for chromosome 3 and basal or apical for chromosome 12) have the same footprint, irrespective of their location. This suggests that the nuclear space adjustment necessary to accommodate the X or the Y chromosome does not rely on different folding of individual chromosomes, but rather on different chromosome positions. These hypotheses would require verification using a triad-detection system with probes targeting a chosen autosome, together with probes targeting sex chromosomes. Chromosome 2 is rather intriguing as it distributes equally in any of the four arbitrarily defined nuclear areas. This observation is not unique to murine sperm, since human sperm chromosome 13 showed identical behavior [17]. Although a rather long autosome, it seems that chromosome 2 is considered as an adjustment variable in the mouse sperm nucleus. In addition, we and others have data suggesting that mouse chromosome 2 is a rather accessible chromosome in the mouse sperm nucleus, since it was observed on several occasions that, when purifying murine sperm DNA for high throughput sequencing strategies, one systematically obtained a large excess of chromosome 2 sequences in comparison to other chromosomes [41,61,62]. This suggests a peripheral localization of this chromosome in the mouse sperm nucleus as it does not appear to be less-condensed than other autosomes [41].

Telomeres have recently been assigned a chromosome stabilizing function that is important for reproduction [63] and it is proposed that telomeres are the first chromosomal regions to respond to oocyte decondensing factors that lead to the formation of the male pronucleus [46]. As suggested earlier in mouse sperm nucleus [49,56] and recently confirmed for the human sperm nucleus [17,52,56,64], we showed here that telomeres in murine sperm are also organized in clusters located at the periphery of the sperm nucleus in an edge-like/ridge-like manner, starting from the base of the nucleus and extending along the dorsal side, in close proximity to the peripheral nuclear matrix. With regard to

centromeres, another characteristic domain of chromosomes rich in repeated sequences we found that in the mouse sperm nucleus, they were also located in clusters, at the periphery, with the same edge-like/ridge-like organization. In a previous study, it was shown via *FISH* that the distribution of centromeres in testicular sperm (not fully mature) are clustered at the surface of the heterochromatic chromocenter (schematic representation in Figure 5B). This differed from our study [49] in which fully mature post-testicular (i.e., epididymal) sperm were evaluated. An organization similar to the one we report here was recently described in human sperm nuclei, in which the centromeres were distributed as single clusters [64]. The present localization of centromeres in murine sperm, determined by using the histone H3 variant CENP-A, is in agreement with previous data reporting that histones in mature murine sperm are preferentially located in the basal and dorsal peripheral areas of the nucleus [35]. In view of these results we propose a new model for telomere and centromere organization in murine sperm nuclei (Figure 5C). It would appear that in the mouse, both telomeres and centromeres are closely located at these dorsal peripheral and basal nuclear domains that were shown elsewhere (as well as here) to be domains rich in nuclear matrix attachment components [30] and rich in histone [65,66]. It is interesting to note that the paternal DNA associated with these nuclear regions was shown to be important both for male pronucleus formation and for the first round of DNA replication [19,37] which are early events of embryo development. As it has been well described in a recent review [67], The organization of the sperm nucleus seems to be an important factor for male fertility and embryo development that will require further analysis.

**Figure 5.** Schematic representation of the proposed models of telomere and centromere organization within murine sperm nuclei. Panel (**A**) presents a schematic representation of the murine acrocentric chromosome with two telomere regions (green) at either end of the chromosome and one centromere (red). Panel (**B**) presents a schematic representation of the murine chromosome model in which the centromeres (red) gather in a chromocenter, with the chromosome (light blue) stretching out toward the telomere (green) localized at the peripheral region. In panel (**C**), we present a refined version of the model, based on our observations, which depicts a more segmented organization, with localization of telomeres (green) and centromere (red), throughout the murine nucleus.

Concerning the susceptibility of the sperm chromatin organization to oxidative alterations, gross examination of the nuclear topology of the chromosomes (studied in this work) shows that they are unaffected by the mild oxidative environment present in the *Gpx5*−*/*<sup>−</sup> transgenic mouse strain. This is supported by the fact that we did not record significant differences in the distribution of the chromosomes, in the four arbitrarily defined regions, when comparing WT and transgenic sperm (Table 1). However, a recent study did show that high levels of DNA damage in human sperm (such as significant DNA fragmentation) could disrupt the position of the centromeres [68]. This suggested that chromosome 3D organization may be impacted depending on the level of sperm DNA damage.

When looking at nuclear 3D-parameters, such as nuclear volume and surface area we confirmed, as expected, The susceptibility of the nucleus to oxidative alterations. This is evidenced by the observations that both nuclear volume and nuclear surface area are significantly diminished in the *Gpx5*−*/*<sup>−</sup> spermatozoa, when compared with WT sperm. This is in line with the idea that when the epididymis-secreted GPx5 protein is absent, it leaves more luminal H2O2 that is used by the sperm-nucleus GPx4 (acting here as a disulphide isomerase) to generate disulphide bridges between the sperm nuclear protamines, leading to a greater state of nuclear condensation [69]. The observation, after the 3D-reconstruction, that the *Gpx5*−*/*<sup>−</sup> sperm show a smoother nuclear surface when compared to the WT sperm which has a "goose-bumps" aspect, is interesting as it distinguishes nuclear domains responding differentially to this oxidation-mediated increased condensation. In particular, The use of different, mild, denaturing treatments (alkaline *versus* reductive denaturation) emphasized the point that even though a mildly oxidized sperm nucleus may appear well-condensed (as for *Gpx5*−*/*<sup>−</sup> spermatozoa) it is highly susceptible to mild reductive conditions. This is important in clinical practice as clinicians may be misled when using assays such as the aniline blue or the toluidine blue, to determine the level of sperm nuclear condensation as an indicator of sperm nuclear integrity. Therefore, The type of mild denaturation technique chosen will be critical to correctly determine the level of sperm nuclear integrity. These considerations support our credo that a solid evaluation of sperm nuclear integrity/solidity, prior to assisted reproductive technology (ART), should include several additional tests, addressing the issue of DNA fragmentation, DNA oxidation, and nuclear solidity.

#### **4. Materials and Methods**

#### *4.1. Ethics Statement*

Ethics statement: The present study was approved by the Regional Ethics Committee of Animal Experimentation (CEMEA-Auvergne; Authorization CE99-12) and adhered to the current legislation on animal experimentation in France.

#### *4.2. Animals*

The *Gpx5*−*/*<sup>−</sup> mice were derived, as described originally, from the C57BL/6 genetic line [41,42]. Mice used in this study (eight mice per genotype) were maintained and housed in temperature-controlled rooms with 12-h light/dark cycles. Mice had ad libitum access to food and water. Nine-month-old mice were culled by cervical dislocation and spermatozoa were collected from the caudal segment of the epididymis.

#### *4.3. Immunocytochemistry and Fluorescence in situ Hybridization (FISH) Assays*

All immunofluorescence procedures were performed as previously described [35]. Briefly, spermatozoa were resuspended in a decondensing buffer (2 mM DTT and 0.5% triton X-100 in PBS) and incubated for 45 min, at room temperature. After centrifugation at 500× *g* for 5 min, at room temperature, spermatozoa were resuspended in PBS, numbered, and deposited onto a glass plate. For FISH assays, spermatozoa were recovered as described previously [41]. A fraction aliquot of <sup>10</sup> × 106 spz/mL was centrifuged at 560× *<sup>g</sup>*, for 5 min and re-suspended in 1.25 mL fresh Carnoy's fixative (3:1 ethanol:acetic acid). This spermatozoa-containing solution was spread on the slides (up

to 25,000 spermatozoa/slide) then slides were dried for 1 h, at room temperature (RT), and stored at −<sup>20</sup> ◦C (Superfrost® slides, Thermo Fisher Scientific, Illkirch, France). After 24 h, slides were defrosted at RT and placed in a coplin jar with saline-sodium citrate solution 2X (SSC 2X), for 15 min at 37 ◦C. Slides were dried for 5 min, at RT, and denatured using NaOH 1.5 N (1 min). Slides were then incubated in a coplin jar with SSC 2X for 30 min, at 70 ◦C (±1 ◦C). The coplin jar was left at RT. Slides were successively incubated for 1 min in SSC 0.1X at RT, NaOH 0.07 N at RT, SSC 0.1X at 4 ◦C, and SSC 2X at 4 ◦C. Slides were transferred through a series of ethanol washes for 1 min, each starting with 70%, 95%, and finally 100% ethanol. Slides were left to dry at RT. DNA probes were applied to a sterile coverslip, pre-warmed at 37 ◦C, and sealed using paraffin. Finally, slides were incubated in a dark humidified chamber at 37 ◦C, for 48 h. Mouse chromosome-painting probes (Metasystems, Altlussheim, Germany) and telomere probes (Panagene, Altlussheim, Korea) were prepared according to the manufacturer's instructions. After a 48-h incubation period, The slides were washed with SSC 0.4X for 2 min, at 70 ◦C (±1 ◦C), and 30 s, in SSC 2X, with Tween 0.05%, and for two successive rinses. Vectashield® with DAPI (Vector Laboratories) was added to each slide to counterstain the sperm cell nucleus. Finally, coverslips were mounted, sealed, and slides were stored in the dark at −20 ◦C, until observation.

#### *4.4. Microscopy*

Confocal Z-stacks were captured using a Leica SPE confocal microscope (Leica Microsystems, Wetzlar, Germany) and a 40× oil immersion objective was used for all acquisitions. At least eighty stacks per nucleus were captured and the distance between Z stacks was 0.21 μm. Chromosome territory was assigned after counting not less than three hundred and fifty spermatozoa per chromosome and the percent of spermatozoa presenting a chromosome at one or more given positions was established. A Zeiss microscope Axioplan2 (Carl Zeiss, Oberkochen, Germany) was used to perform these observations.

#### *4.5. Image Analysis Measurements of 3D Parameters*

All the images were deconvoluted using Huygens software (Scientific Volume Imaging, The Netherlands) before analysis. Spermatozoa volume and surface area were measured using Imaris Version 7.6 software (Bitplane AG, Zurich, Switzerland). The mean of each parameter was calculated with at least 30 spermatozoa.

#### *4.6. Statistics*

Mann-Whitney and Spearman correlation analyses were performed using GraphPad Prism® software. The difference was considered significant when *p* < 0.001 (\*\*).

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/9/10/501/s1, Figure S1: Correlation of volume/surface area and size, Table S1: Three-dimensional parameters of sperm chromosomes in WT mouse 3 sperm nucleus.

**Author Contributions:** Authors A.C. and A.K. were involved in the conception and design of the experimental manipulations, assisted with the molecular techniques, and wrote the manuscript. C.D.-S., C.G. and S.B. contributed to the experimental manipulations. J.D. drafted and critically reviewed the manuscript. J.H.-B., R.G. and F.S. also performed a critical review of the manuscript. All authors have read and approved the final manuscript.

**Funding:** This research was funded by the CNRS, INSERM & French Ministry of Higher Education and Research (MESR) through operating grants attributed to JRD.

**Acknowledgments:** The authors would like to thank the CNRS, INSERM, and UCA for their financial support, as well as the former *Région Auvergne* for contribution to this research.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Genetic Instability and Chromatin Remodeling in Spermatids**

#### **Tiphanie Cavé, Rebecka Desmarais, Chloé Lacombe-Burgoyne and Guylain Boissonneault \***

Department of Biochemistry, Faculty of Medicine and Health Sciences, Université de Sherbrooke, Sherbrooke, QC J1E4K8, Canada; tiphanie.cave@usherbrooke.ca (T.C.); rebecka.desmarais@usherbrooke.ca (R.D.); chloe.lacombe-burgoyne@usherbrooke.ca (C.L.-B.)

**\*** Correspondence: guylain.boissonneault@usherbrooke.ca; Tel.: +1-819-821-8000 (ext. 75443)

Received: 27 December 2018; Accepted: 8 January 2019; Published: 14 January 2019

**Abstract:** The near complete replacement of somatic chromatin in spermatids is, perhaps, the most striking nuclear event known to the eukaryotic domain. The process is far from being fully understood, but research has nevertheless unraveled its complexity as an expression of histone variants and post-translational modifications that must be finely orchestrated to promote the DNA topological change and compaction provided by the deposition of protamines. That this major transition may not be genetically inert came from early observations that transient DNA strand breaks were detected in situ at chromatin remodeling steps. The potential for genetic instability was later emphasized by our demonstration that a significant number of DNA double-strand breaks (DSBs) are formed and then repaired in the haploid context of spermatids. The detection of DNA breaks by 3 OH end labeling in the whole population of spermatids suggests that a reversible enzymatic process is involved, which differs from canonical apoptosis. We have set the stage for a better characterization of the genetic impact of this transition by showing that post-meiotic DNA fragmentation is conserved from human to yeast, and by providing tools for the initial mapping of the genome-wide DSB distribution in the mouse model. Hence, the molecular mechanism of post-meiotic DSB formation and repair in spermatids may prove to be a significant component of the well-known male mutation bias. Based on our recent observations and a survey of the literature, we propose that the chromatin remodeling in spermatids offers a proper context for the induction of de novo polymorphism and structural variations that can be transmitted to the next generation.

**Keywords:** spermiogenesis; chromatin remodeling; DNA double-strand breaks; genetic instability; mutations

#### **1. Introduction**

As one can appreciate from this Special Issue, the proper packaging of the male haploid genome involves finely-regulated molecular events, resulting in a near complete replacement of somatic chromatin and the formation of a highly condensed nucleus. Although the final protamine deposition (protamination) yields a genetically and mechanically stable nucleus, it became rather intuitive that the previous chromatin remodeling steps and resulting change in DNA topology [1] entailed potential genetic hazards. Early observations that H4 hyperacetylation occurs in murine spermatids [2,3] provided the first evidence that chromatin most likely undergoes a transient state of increased sensitivity to endonucleases during this process [4]. H4 hyperacetylation also sets the stage for the bromodomain, testis-specific, protein (Brdt)-mediated histone eviction [5,6]. In situ detection of 3 OH ends in mouse and human spermatids confirmed that H4 hyperacetylation is indeed coincidental, or slightly precedes the formation of the DNA strand breaks that were observed in the whole spermatid population [7]. Because of its potentially significant transgenerational impact, establishing the genetic

consequences of this structural transition and the formation of transient DNA strand breaks has been the ultimate objective of our investigation over the past 15 years.

#### **2. DNA Double-Strand Breaks (DSBs) are Intrinsic to the Differentiation Program of Spermatids**

Our initial observation that transient DNA strand breaks were observed in the whole population of mouse and human spermatids ruled out that a canonical apoptotic process was involved, as discussed below. Evidence of a similar surge in DNA strand breakage was also reported in rats [8], drosophila [9], grasshoppers (*Eyprepocnemis plorans*) [10] and in algae (*Chara vulgaris*) [11]. Several methods were used to confirm that the transient DNA strand breakage in spermatids also included a significant proportion of DNA double-strand breaks (DSBs). These included neutral comet assay, pulse-field gel electrophoresis [12], γH2AX labelling [13] and qTUNEL assay, whereby double-strand breaks were specifically labelled in solution following a prior step involving DNA nicks and gap filling [14]. Indirect evidence of a DSB repair response based on γH2AX expression must however be taken with caution, as the latter has also been associated with chromatin alteration [15]. Direct methods were also used by our group to show that transient post-meiotic DSBs also form in the fission yeast, lending strong support to the highly conserved nature of this mechanism [16]. Despite their transient character, the formation of DSBs in the haploid context of spermatids represents a genetic threat, because repair must rely solely on end joining processes, as outlined below [17,18]. Considering the lower DNA repair activity reported in condensing spermatids, potential misprocessing of these DSBs would be expected to further increase genetic instability.

#### **3. Potential Mechanism for DSBs Formation and Repair**

Chromatin structure is a key determinant of the meiotic DSB landscape [19] and hotspot specification in meiosis, where it has been shown to arise from a combination of factors acting upon histones, including the histone methyltransferase PRDM9 [20,21]. In the mature sperm, protamine affords similar but periodic loop-sized protection against the combined activity of an endogenous nuclease interacting with the nuclear matrix-associated topoisomerase IIB (TOP2B) [22]. Similarly, the dynamic character of the chromatin structure transition in spermatids must dictate the genomic distribution of the DSBs during the differentiation steps, and it stands to reason that DNA strand break hotspots regions should be found, although the endonucleases involved have yet to be identified. The potential involvement of TOP2B in the transient formation of DSB has been inferred from the synchronous detection of TOP2B and the tyrosyl-DNA phosphodiesterase 1 (TDP1), which are known to resolve topoisomerase-mediated DNA damage [23]. Such DNA breaks may have been created from the simple hindrance of the TOP2B catalytic cycle during the nuclear condensation process in elongating spermatids. Using RNA interference, the requirement for TOP2 activity in the post-meiotic DSB formation has been recently demonstrated in the ciliate *Terahymena thermophila* [24]. The extent of DSB formation in elongating spermatids is yet unknown. As previously proposed [25], TOP2B is expected to relax all free supercoils generated from nucleosome eviction, reducing the DNA linking number (Lk) in steps of two per catalytic cycle. Although a great number of DSBs should be generated in this case, the DNA ends remain concealed, and it may not elicit a DNA damage response. This is in sharp contrast to the action of an endonuclease that would relax larger domains of unconstrained supercoils from simple strand breakage due to its inability to relegate DNA ends. However, as suggested above in studies concerning mature sperm, a controlled fragmentation process involving the combined action of TOP2 activity and endonucleases in differentiating spermatids should also be considered. For instance, such an interaction between TOP2A and endonuclease G has been shown to be involved in caspase-independent apoptotic DNA fragmentation, since the mitochondrial endonuclease G is translocated to the nuclei during apoptosis [26,27]. Without leading to cell death, it has been proposed that the apoptotic machinery could be borrowed for various differentiation processes, including spermiogenesis [28,29]. Multiple caspases were found to be expressed in the residual bodies of the Drosophila spermatids to remove the unneeded cytoplasmic content during the

process of individualization, but caspases are seemingly kept away from the nucleus [28]. In accordance with an early report from Smith and Haaf [30], our group has found no evidence of complete canonical nuclear apoptosis in the nuclei of mouse spermatids in agreement with the transient character of the DNA strand break formation. Since we observed nuclear translocation of endonuclease G in the nuclei of elongating spermatids (unpublished data), its potential association with topoisomerase in spermatids is currently under investigation, as this could promote a caspase-independent mechanism that would lead to the observed surge in DNA fragmentation. Such a mechanism could be under the control of PARP1 and PARP2, as they were shown to strongly inhibit TOP2B activity of spermatids both in vivo and in vitro [31]. One major hallmark of apoptosis is the resulting mitochondrial outer membrane permeabilization, which releases endonuclease G and apoptosis-inducing factors (AIF) [32–35]. Importantly, although the expression of pro-apoptotic factors has not yet been reported in spermatids, mitochondria are nevertheless known to undergo major structural changes during spermiogenesis. While part of them move to the growing flagellum, other starts to aggregate, and are eventually eliminated by Sertoli cells via phagocytosis or autolysis [36]. It is therefore possible that mitochondrial endonuclease G is released during the autolytic destruction of the outer membrane during the chromatin-remodeling steps. It is worth nothing that even limited, regulated mitochondrial permeabilization can produce DNA damage and genomic instability, without leading to cell death [37], supporting the concept that mitochondrial damage could entail controlled yet reversible DNA fragmentation.

In recent studies, the reversible character of such genome-scale apoptotic-like DNA fragmentation (reversal of apoptosis) has been observed in many instances upon the withdrawal of inducers [38,39]. Striking examples of this are the recovery from global DNA fragmentation observed in the African midge (*P. Vanderplanki*) larva after extreme dehydration [40], similar to the extremotolerant tardigrade species (*R. varieornatus*) [41]. The latter has been shown to express a newly identified, highly basic DNA binding protein (Dsup). The activity of Dsup has been shown to protect transfected cells against radiation and ROS-induced DNA breaks, which is somewhat reminiscent of the protective effect against UV-induced DNA damage that we previously reported for transition proteins [42]. Thus, such a recovery from massive DNA fragmentation, coined "anastasis", indicates that related mechanisms may be operating in spermatids for global DNA repair. However, recovery from apoptosis-like processes can promote mutagenesis and even oncogenic transformation [37], often displaying micronuclei and chromosomal abnormalities [39,43].

Taken together, this compelling evidence suggests that the chromatin remodeling in haploid spermatids, which precludes the use of homologous recombination for templated DSB repair, should create genetic instability [18]. End-joining processes for DSB repair that are likely to operate in a haploid context include single-strand annealing (SSA), microhomology-mediated end joining (MMEJ) or canonical nonhomologous end joining (NHEJ). SSA occurs within repeated sequences and is known to be intrinsically mutagenic [44], whereas, in the absence of canonical NHEJ factors, MMEJ can process the resected DNA ends, using as little as 1-2bp homology when stabilized by PARP [45]. Canonical NHEJ can proceed without sequence homology, and results in insertion, deletion or even chromosomal rearrangement [17]. Whereas meiosis may have evolved mechanisms to prevent these error-prone end-joining processes [18,46], haploid spermatids likely cannot avoid such mutagenic repair mechanisms. However, our initial mapping data indicated that the transient post-meiotic DSBs arise preferentially within repeated elements of the genome, which should minimize the genetic threat associated with the DSB formation [12]. It is important that coding sequences be protected from the global DNA fragmentation process, especially because of the general loss in DNA repair capacity that has been observed as the chromatin remodeling in spermatids proceeds to the final steps [47–49]. In pathological conditions, further alteration in the repair capacity of spermatids could lead to a persistence of DSBs in spermatozoa. Unrepaired DSBs in sperm could be of a lesser concern, given the reported efficient DNA repair activity of the oocytes [50].

#### **4. First Evidence of Genetic Instability in Spermatids**

Trinucleotide repeats (TNRs) are the most unstable DNA sequences, and transgenerational expansion beyond a given threshold has been linked to inherited neuromuscular and neurological disorders in offspring [51,52]. Thus, variations in TNR represent an ideal sentinel to monitor genetic instability as a result of faulty DNA repair or chromatin remodeling. Using a transgenic mouse model, the TNR expansion of a CAG repeat within exon 1 of the human *HD* gene was shown to be limited to post-meiotic events in males, and thus does not involve mitotic replication or homologous recombination between chromosomes [53]. Using this mouse model [54], the purification of spermatids into four distinct populations allowed us to further demonstrate that an increased frequency of longer DNA repeat length occurs just following chromatin remodeling, as was observed in mouse step 15–16 spermatids, which is equivalent to the transition between steps 3 and 4 in human spermiogenesis [55]. Interestingly, based on the increased intensity of the individual repeat length, we estimated that approximately 20% of spermatids displayed a shift to a longer repeat length [56]. In vitro experiments suggested that the increase in free superhelical density that must prevail during histone eviction resulted in expansion at a stabilized hairpin [57]. DNA secondary structures are indeed causative factors of expansion [52], and the free supercoils in spermatids offer an ideal context for hairpin extrusion and stabilization of other alternative non-B DNA conformations at repeated elements, including cruciform and left-handed Z-DNA [58,59]. Because non-B DNA structures are preferential substrates for endonucleolytic incisions [60,61], the striking enrichment of DSBs that we observed at repeated elements of the spermatid's genome may therefore not be surprising. Thus, monitoring TNRs length variation during the chromatin remodeling provided the first experimental evidence that genetic instability is an important feature of differentiating spermatids.

#### **5. DNA Fragmentation in Spermatids and the Male Mutation Bias**

Over the past few years, next generation sequencing (NGS) of parent–offspring trios has confirmed the clear male bias for the transmission of de novo mutations. Male-biased mutations include single-nucleotide variants, small insertions–deletions (indels), and structural variations [62–67]. Not surprisingly, the greater number of replication cycles in spermatogenesis was originally suspected as the leading cause of de novo mutations, because the male-to-female mutation rate ratio correlates with the male-to-female ratio of the number of cell divisions [68].

The relatively recent observation that transient DNA double-strand breaks are part of the differentiation program of spermatids (and the even more recent results on the genome-wide distribution of these breaks) could explain why only a few reports have considered this process in the etiology of the male mutation bias. Several lines of evidence, however, suggest that the formation of DSBs and repair in the haploid context of spermatids are compatible with the recent NGS data. First, the amount of mutations generated during DNA replication is much lower than the polymerase error rate. Hence, the reliability and extent of DNA repair (or DNA repair rate) becomes prominent in the determination of transmittable de novo mutations [69]. Previous reports confirmed the decline in the general repair capacity during spermiogenesis [49], and the limited response to DSBs in elongating spermatids compared to pre-meiotic cells [47,48,70]. Second, our initial screening in mice indicated that a large part of the transient DSBs in spermatids map to repeated elements of the genome, in accordance with the relative abundance of these intergenic regions. Interestingly, however, DSBs arise at a greater frequency in LINEs and microsatellites relative to their normal representation in the genome [12]. Coincidentally, a higher frequency of de novo mutations was found to arise within repetitive DNA sequences [67], and a strong paternal bias has been reported for mutations within microsatellite repeats [71]. Interestingly, we found the density of DSBs to be four times higher in the Y chromosomes than autosomes, which is compatible with the higher mutability reported for the Y chromosome [72]. Third, and as outlined above, DSBs and end-joining repair processes in haploid spermatids are likely to offer a proper context for male-driven rearrangement. De novo indels and structural variations such as retrotransposon insertions and interchromosomal events were shown to arise preferentially in the

paternal germline [66,73], and a similar paternal bias has been reported for copy number variation (CNV) [74,75]. Although replication-based mechanisms could still be responsible for these mutational and structural modifications, they can also arise from the formation of DSBs in the haploid context of spermatids. For instance, in haploid cells, CNV may result from NHEJ, but can also be generated by non-allelic homologous recombination (NAHR) [76], since NAHR is produced by the alignment and subsequent crossover between nonallelic DNA sequence repeats sharing a homology. Both mechanisms require the formation of a DSB [77]. Fourth, it was established that these male-driven variations are generally associated with the transmission of neurodevelopmental disorders [62,63,65,74,78]. For instance, 88% of de novo indels arise on the paternal chromosome and are associated with autism spectrum disorders [65]. When considering only the gene subset, our preliminary gene ontology term analysis showed that DSBs in spermatids arise preferentially within synaptic genes, and with high significance [12]. Hence, the transient DSB formation in spermatids deserves more attention as a potentially significant mechanism by which paternal variation may be transmitted with implications for neurodevelopment.

The transient DSB surge in spermatids may be viewed as a serious threat to the genetic integrity of the differentiating spermatids before it is eventually used for fertilization. However, one must consider that the DSBs are detected in the whole spermatids population and represent the full repertoire of potentially unstable loci seen in a pool of several million cells. The whole genome capture of these DSBs provides a map of their distribution in the whole cell population, if they can be detected by being present in much more than a single cell. A strong DSB hotspot leading to a deleterious mutation or structural variation must therefore be present in a significant subset of spermatids in order to increase the chance of these being selected for fertilization. Pathological conditions that would increase global DNA fragmentation in spermatids could lead to a concomitant increase in the frequency of a given hotspot among cells, thus increasing the chance of that allele being transmitted to the offspring. On the other hand, if a mutational DSBs hotspot is present at a much lower frequency, for instance at 0.01% or lower, then the mutated allele would stand a maximum of 10−<sup>4</sup> chance of being transmitted for each oocyte being fertilized. In such a case, a beneficial or deleterious mutation would be passed on to the next generation over a much longer (or evolutionary) timescale, provided that a hotspot locus was maintained over time in each species. Interestingly, we observed that selected hotspots are shared between two mouse strains (C57BL/6 vs CD1), suggesting that they can be maintained at least over the evolutionary distance between these inbred and outbred strains. A higher frequency (or density) of such weaker hotspots in a chromosome would confer a faster evolutionary global mutability on an evolutionary timescale, as observed for the Y chromosome.

#### **6. DSBs in Lower Eukaryotes**

The apparent conservation of transient DSBs in spermatids between mammals prompted us to investigate whether post-meiotic DSBs may be conserved through the eukaryotic domain, and also be associated with gamete formation in yeast (sporulation) [79]. Being more similar to metazoans, fission yeast (*S. pombe*) represents a better alternative than budding yeasts (*S. cerevisiae*) [80]. In addition, synchronous meiosis can be achieved with the *S. pombe pat1-114* mutant, in which a temperature-sensitive Pat1 (Ran1) protein kinase inhibits meiosis by negatively regulating an RNA-binding protein that controls entry into the meiotic S phase, Mei2. Synchronous meiosis can therefore be induced in a timely and predictable fashion by shifting nitrogen-starved cultures from permissive (25 ◦C) to restrictive (34 ◦C) temperatures [81]. One striking observation is that synchronized *S. pombe* displayed a similar post-meiotic surge in DSBs in the absence of apoptosis, when meiosis is induced in the *pat1-114* mutant, or even in the wild-type *FY435*/*FY436* strain (azygotic meiosis) [16], suggesting that sporulation, much like spermiogenesis, may display a window of genetic instability. Our conclusion is that transient post-meiotic DSBs may be intrinsic to the gamete differentiation program throughout the eukaryotic domain. As outlined above, a major support for this conclusion recently came from the demonstration that DSBs also arise during

post-meiotic steps in the ciliate *Tetrahymena thermophila* [24]. These observations point to the discovery of a highly conserved, physiological mechanism that deserves further investigation regarding its genetic impact and evolutionary consequences. Simple eukaryotic models such as yeast offer the possibility of functional genetics analyses, to identify the endonuclease(s) responsible for the transient DSB formation, and eventually determine their impact on adaptation and evolution over several generations. In-gel nuclease assays in the synchronized pat1-114 mutant have already led to the identification of a candidate mitochondrial endonuclease (Pnu1), an homolog of the *S. cerevisiae* Nuc1p that has been described as part of the caspase-independent apoptotic pathway [82]. Interestingly, the mammalian homolog of Nuc1p is the mitochondrial endonuclease G (Endo G) discussed above, which is also involved in caspase-independent apoptosis. These early observations in yeast lend support to the proposal that the transient post-meiotic DSBs may indeed borrow components of the apoptotic machinery in a controlled, reversible manner.

#### **7. Conclusions and Future Directions**

Given the non-templated DNA repair in haploid spermatids, transient DSB formation may represent an important component of the male mutation bias and the etiology of neurological disorders, adding to the genetic variation provided by meiosis. Repair heterogeneity at these potential hotspots would produce a repertoire of genetic polymorphisms, given the large population of spermatozoa produced over time. In addition to the chromosome reshuffling provided by meiosis, each offspring would also inherit a given set of mutations created by the chromatin remodeling in the spermatid of origin. Because synaptic genes were found to be specifically targeted by DSBs and should therefore harbor more mutational hotspots, pathological conditions (or aging) leading to a global rise in DSB formation would then increase the odds of transmitting de novo variation in neurodevelopmental genes. This hypothesis is thus in agreement with the reported correlation between the father's age at conception and the risk of transmitting neurological disorders. Further investigation should therefore be aimed at deciphering whether mutational DSBs hotspots arise within neurodevelopmental genes and how they are altered under pathological conditions, or following exposure to xenobiotics. Monitoring the distribution and number of DSBs in elongating spermatids should be clearly emphasized, as they represent a much higher threat compared to single-strand breaks. Despite the reported DNA repair capacity of the oocyte, the transgenerational consequences of an increased number of persistent DSBs in sperm deserves some attention for future investigations.

**Author Contributions:** Conceptualization, T.C. and G.B.; investigation, T.C., R.D., C.L.-B., G.G.; writing—original draft preparation G.B.; writing—review and editing, T.C., R.D., C.L.-B., G.B.

**Funding:** This work was supported by a grant from the Canadian Institutes of Health Research (#MOP-136925) to G.B.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Single and Double Strand Sperm DNA Damage: Different Reproductive Effects on Male Fertility**

#### **Jordi Ribas-Maynou \* and Jordi Benet \***

Unitat de Biologia Cel·lular i Genètica Mèdica, Departament de Biologia Cel·lular, Fisiologia i Immunologia, Facultat de Medicina, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain

**\*** Correspondence: jordi\_ri@hotmail.com (J.R.-M.); jordi.benet@uab.cat (J.B.); Tel.: +34-93-581-1773 (J.B.)

Received: 30 December 2018; Accepted: 29 January 2019; Published: 31 January 2019

**Abstract:** Reproductive diseases have become a growing worldwide problem and male factor plays an important role in the reproductive diagnosis, prognosis and design of assisted reproductive treatments. Sperm cell holds the mission of carrying the paternal genetic complement to the oocyte in order to contribute to an euploid zygote with proper DNA integrity. Sperm DNA fragmentation had been used for decades as a male fertility test, however, its usefulness have arisen multiple debates, especially around Intracytoplasmic Sperm Injection (ICSI) treatments. In the recent years, it has been described that different types of sperm DNA breaks (single and double strand DNA breaks) cause different clinical reproductive effects. On one hand, single-strand DNA breaks are present extensively as a multiple break points in all regions of the genome, are related to oxidative stress and cause a lack of clinical pregnancy or an increase of the conception time. On the other hand, double-strand DNA breaks are mainly localized and attached to the sperm nuclear matrix as a very few break points, are possibly related to a lack of DNA repair in meiosis and cause a higher risk of miscarriage, low embryo quality and higher risk of implantation failure in ICSI cycles. The present work also reviews different studies that may contribute in the understanding of sperm chromatin as well as treatments to prevent sperm DNA damage.

**Keywords:** sperm DNA damage; DNA fragmentation; infertility; assisted reproduction; miscarriage; implantation

#### **1. Introduction**

Different fertility societies around the globe and the World Health Organization estimate that infertility is present in between 7% and 15% of couples in reproductive age [1,2]. In a high number of cases female factors and especially female age [3], are the most important causes of infertility, however, different male factors are present in at least 50% of the couples presenting this disorder [4]. Due to the high percentage of incidence in the pathology, recent research suggests that sperm cell and sperm DNA may have a major influence not only in natural conception but also in fertility treatments [5,6].

In front of a fertility disorder or a fertility treatment, microscopic semen analysis measuring sperm concentration, motility and morphology has been the traditional and important first approach to male infertility and, although a high decrease of these parameters had been associated to a lack of achievement of natural pregnancy [7] and nowadays home-based technologies in order to advance the first diagnosis are emerging [8]. However, in most cases these parameters are not indicative of the positive performance of assisted reproduction techniques (ART) [5,9]. In fact, although they are improving, ICSI treatments reached limited implantation rates [10]. Because of that, a deeper study is necessary in most cases to elucidate the alteration in order to design the best treatment in each case.

#### **2. Sperm DNA and Sperm DNA Damage**

Spermatogenesis is a very complex cellular process that implies both meiosis and cell differentiation. The main stage of meiosis is in prophase I where, spermatocytes deliberately produce double-strand DNA breaks (DSB) through Spo11 protein [11,12]. These DSB are necessary for homologous chromosomes to allow DNA recombination. Then, after strand invasion, DSB activate the DNA repair machinery through the protein kinase ataxia-telangiectasia mutated (ATM) in order to repair the free ends and therefore generate the chiasma by homologous recombination and ATM is also responsible of inhibiting the formation of new DSB by Spo11 [12,13]. After meiosis, haploid round spermatids suffer a cell differentiation, loosing most part of their cytoplasm and acquiring midpiece and flagellum in order to possess motility after ejaculation [14]. However, in terms of chromatin, the most important change happening in spermatids is the exchange of histones by protamines, which extraordinarily compact about 85% of the human sperm DNA in toroidal structures tied between them and bond to the nuclear matrix by the matrix attachment regions (MAR regions) (Figure 1). These MAR regions remain compacted by histones and represent a very small part of the genome estimated to be around 15% of the human sperm chromatin [15,16]. This high-grade of DNA compaction with protamines, coupled to a motile architecture of the cell, give the sperm the perfect features to carry male genetic material to oocyte to form the zygote. It is obvious that if this male genetic material contains alterations, these may affect the zygote somehow [17]. In fact, it is undeniable that DNA breaks induce a cellular response in somatic cells leading to an activation of DNA repair machinery, apoptosis or cell transformation, being the basis of cancer and other diseases [18,19]. Different works in embryos analysing the effect of induced DNA breaks in animal sperm cells through radiation observed multiple chromosomal alterations such as chromosome breaks, translocations, fusions and acentric fragments in the zygote [17,20].

**Figure 1.** Schematic structure of the sperm DNA compacted in protamines that form toroid structures (red) linked by MAR regions (matrix attachment regions) compacted in histones (blue) and attached to the nuclear matrix (green). (**A**) represents an intact chromatin. (**B**) represents chromatin with single-strand breaks (red lines). (**C**) represents chromatin with extensive double-strand breaks (red cross). (**D**) represents chromatin with localized double-strand breaks attached to the nuclear matrix (yellow circle).

In the last decade, the previous evidences suggested the incorporation of the sperm DNA fragmentation tests as a promising analysis in male reproduction and multiple studies were performed in the field since then [21]. Regarding natural conception, multiple works show a relation of sperm DNA fragmentation (SDF) to a lack of clinical pregnancy and an increase of time of conception [22–24]. However, after ICSI procedures, opposite results were found by different research groups regarding embryo quality, implantation and pregnancy outcomes, being some studies that show a positive relation of SDF [25–28] and others that show a negative relation of SDF to clinical outcomes [29–33]. This controversy, coupled that only a few studies were conducted in a prospective and double blind manner, led the American Society for Reproductive Medicine to refuse its routine use in 2013 [34]. However, some promising results arisen in the last years might be the explanation why the traditionally measured sperm DNA damage present a lack of predictive power in ICSI.

The debate in sperm DNA fragmentation started regarding which of all DNA analysis techniques, that rely on different mechanisms for DNA breaks detection, was the best for the male infertility diagnosis. Understanding the basis of each technique and the correlations between them is critical to understand their implications in the male fertility diagnosis and to compare between them. Techniques are explained in the following part of the review and are summarized in Table 1.

On one hand, the most used techniques for the analysis of sperm DNA fragmentation have traditionally been the Terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL), Sperm Chromatin Structure Assay (SCSA) and Sperm Chromatin Dispersion (SCD) test. These techniques offer a unique value of sperm with DNA fragmentation, independently of the type (single and double-strand DNA breaks) and the region (toroids compacted in protamines or MAR regions compacted in histones).

TUNEL assay [35] relies on a terminal TdT transferase for the labelling of 3 free ends of DNA, resulting in a higher labelling on fragmented sperm cells. Different modifications have been introduced in the protocol in order to increase its sensitivity in sperm cells, such as the use of a previous DNA decompaction using dithiothreitol (DTT) or the use of flow cytometer [36–38].

SCSA is based on an acid denaturation of the chromatin and staining with acridine orange. When DNA breaks are present, chromatin is more susceptible to denaturation and acridine orange accumulates in the DNA emitting in red fluorescence. When DNA breaks are not present, acridine orange intercalates in the double helix and emits in green fluorescence. Fluorescence is captured using a cytometer in order to determine DNA fragmentation [39].

SCD test uses a sperm lysis solution based on DTT, sodium dodecyl sulphate (SDS) and NaCl to remove the sperm membrane and protamines, that causes the formation of DNA haloes, which allow the differentiation of fragmented and non-fragmented sperm cells [40].


**1.** Techniques for the detection of different types of DNA damage.

**Table**  HDS: High DNA Stainable sperm; TUNEL: Terminal deoxynucleotidyl transferase dUTP nick end labelling; SCSA: Sperm Chromatin Structure Assay; SCD: Sperm Chromatin Dispersion;ICSI: Intracytoplasmic sperm injection. *Genes* **2019**, *10*, 105

On the other hand, Comet assay [41] relies on a DNA decompaction and protein depletion coupled to a single-cell electrophoresis in an agarose micro gel. DNA molecules that contain breaks move towards the cathode and the length of the "comet tail" can be measured to determine the grade of DNA fragmentation at a single cell level. This technique has been applied in multiple different protocols, which usually vary in agarose concentrations and in electrophoresis times [42,43]. As the Comet assay can be performed in alkaline or neutral pH, different types of DNA breaks can be detected (Table 1) (Figure 1): (i) alkaline Comet assay performed in a small electrophoresis time (about four minutes) detect mostly single-strand DNA breaks affecting both toroidal regions and MAR regions in a high number of break points [44,45] and (ii) neutral Comet assay can detect two types of double-strand DNA breaks (Figure 2): (a) extensive DSB, which represent a very small part of total DSB and can be observed as very long comet tails separated from the sperm core; and (b) localized DSB localized and attached to the MAR region, as demonstrated in pulsed-field gel electrophoresis [43–46], being the most common DSB. Although extensive DSB result in longer Comet tails, they cannot be distinguished from localized DSB in a single Comet. However, when a semen sample present high number of sperm cells with extensive DSB (long tails), single-strand DNA damage is also present in a high amount (Ribas-Maynou personal observation). Previous studies had shown that localized DSB represent very few break points in the genome, as long chromatin fibres with a break point in the end can be seen in a detailed neutral Comet image (Figure 2A), which is supported by Kaneko et al., using pulsed field gel electrophoresis [47]. We demonstrated that localized DSB remain attached to the sperm nuclear matrix [45], maybe through a TOP2B or similar protein [45,46], a very important feature taking into account that the nuclear matrix is inherited to the male pronucleus in the zygote [46,48–50], giving a chance to the embryo to repair the DSB.

**Figure 2.** (**A**) Picture and scheme of neutral Comet with localized DSB (double-strand DNA breaks) attached to the nuclear matrix (green). Comet halo consists in non-fragmented chromatin and comet tail is formed by chromatin fibres attached to the nuclear matrix with low number of DNA breaks at the end (arrows). (**B**) Picture and scheme of neutral Comet with extensive DSB. Comet tail is formed by DNA fragments that are not attached to the nuclear matrix. This comet also shows part of localized DNA breaks attached to the MAR region (arrow).

Studies using all the techniques showed that oxidative damage detected by alkaline Comet assay presented a good correlation to TUNEL, SCSA and SCD techniques [23,51,52]. Although these techniques may potentially detect double-strand breaks, a study conducted by our group analysing the same semen samples with five methodologies showed that no correlation was present with the neutral Comet assay [23]. Then, the latter would be the only technique that is able to differentially detect MAR-region double-strand breaks [23,44], whereas TUNEL, SCSA and SCD may detect extensive DSB. A Comet assay variant (two-tailed Comet assay) applying both alkaline and neutral Comet assay in

the same slide by turning it 90º between electrophoresis allows to distinguish single and double-strand DNA breaks on the same sperm cell [53]. However, no studies have been performed comparing these techniques and alkaline or neutral Comet assay separately in order to elucidate if double-strand breaks detected in two-tailed Comet assay correspond to MAR region localized DSB.

#### **3. Oxidative DNA Damage, Alkaline Comet Assay and Pregnancy Achievement**

Using alkaline Comet assay in different cohorts, an study published in 2012 [43] showed that the extensive single-strand DNA breaks were reversely associated to the achievement of natural pregnancy independently of the neutral Comet results (Figure 1 and Table 1). This was confirmed and compared with TUNEL, SCSA and SCD tests in 2013, demonstrating also that alkaline Comet is the most sensitive technique for the prediction of natural pregnancy achievement [23,43]. Which is also in accordance to the numerous studies from other research groups that find similar association in natural pregnancy using TUNEL, SCSA, SCD and Comet assay tests [5,51,54–58].

Single-strand breaks are produced mainly due to reactive oxygen species (ROS) [42,53,59], which may come from exogenous sources such as environmental toxicants, smoking, alcohol, diet, radiation and so forth or from endogenous sources such as an increase of leukocytes, presence of varicocele or even the ROS generated by mitochondria for the movement of sperm cell [60–62]. Free radicals may cause lipid peroxidation, mitochondrial and nuclear DNA base modifications such as 8-OH-guanine and 8-OH-2 -deoxyguanosine (8-OHdG), an oxidized base adduct that destabilize DNA structure and cause a DNA break [63–65]. This affectation does not find a restriction by DNA condensation and therefore may affect both toroids compacted in protamines and MAR regions compacted in histones [44]. Then, if such an extensive damage happens to the sperm DNA due to oxidative stress, the sperm membranes would also be affected and usually sperm motility is lost. Because of that, a strong negative relation between progressive motility and oxidative damage (single-strand DNA damage) analyzed using TUNEL, SCSA, SCD and alkaline Comet [55,61,66].

As mentioned before in this review, controversial results are found in different studies regarding ICSI outcomes: some of them which found predictive value of oxidative damage [25–28] and other with opposite results [29–33]. If single-strand DNA breaks present a correlation to progressive motility and sperm morphology and ICSI procedures use the most motile sperm cells with better morphology, paternal genome should be free of oxidative damage. In this regard, a work by Gosalvez et al. [67] demonstrated that motile sperm organelle morphology examination (MSOME) selected sperm cells were free of DNA damage analysed by SCD test. Moreover, a work using Comet assay suggested that grade I and II sperm cells present lower incidence of oxidative DNA damage than grade III and IV [68]. These results need to be further confirmed in conventional ICSI sperm selection. However, our data suggest that no relation is present between alkaline Comet and embryo quality, embryo kinetics or implantation [69].

#### **4. Double-Strand DNA Damage, Recurrent Miscarriage and Preimplantation Failure in ICSI Cycles**

Analysing the data of the patients and donors with high DSB, a specific profile was observed with low oxidative damage and high neutral comet values in patients with first trimester recurrent miscarriage where all related female factors were discarded and in one subgroup of fertile donors [44]. In a recent study, our group has found that patients with this profile who undergo ICSI treatments produce embryos with a delayed embryo development to blastocyst, which also cause lower implantation rates [69]. Other works also show that double-strand breaks may contribute to a higher implantation failure risk [6,25]. Since implantation failures in ICSI cycles and miscarriages present similar profiles with high DSB, one may think that they might have similar origin. In fact, small number of DNA breaks localized in concrete regions of the genome might induce a cell failure where the affected regions are necessary for the development. In our last study, embryos that achieved implantation

presented faster embryo kinetics than those that did not achieve implantation [69]. In fact, faster embryo kinetics had been associated to embryo euploidy [70–72].

DSB are the most lethal alteration that may happen in a zygote, since paternal and maternal pronucleus remain separated in early mammalian embryos and, therefore, no complementary chain would be available for DNA repair [73–75] and a few number of DSB are sufficient to delay cell cycle [76]. It is important to note that paternal double-strand breaks remain attached to the nuclear matrix and probably to other proteins such as TOP2B [20,46,77] and the nuclear matrix is inherited at male pronucleus until first mitotic division [49,78]. This may be crucial at the zygote, because it may give a chance to correctly repair both free ends of the double-strand break. There is a consensus point that oocyte quality may play a role in this DNA repair, since different studies proved that early embryos are able to repair DNA damage [79–84]. In this sense, in patients with DSB, the most significant delay observed in the embryo kinetics was just after fertilization, indicating that DNA repair machinery may be active in this stage [69]. Recent studies in sperm cells demonstrated that MAR regions are required as a scaffold for DNA replication after fertilization [48] and, in somatic cells, nuclear matrix also is involved in transcription, cell regulation and replication [85,86]. In mammals, inducing DSB in sperm cells and used these sperm cells to fertilize eggs observed chromosomal alterations in paternal genome of the embryo and showing also a delay in the first embryo cleavage [17,20,87]. Moreover, studies inducing double-strand DNA breaks in mice sperm through radiation observed a p53 and p21 related response and less number of foetuses [88,89] or less survival of offspring in a dose dependent manner [90].

#### **5. Prevention of DNA Damage**

The data presented in the studies referenced before supports that oxidative damage may affect the pregnancy achievement capacity due to misbalanced levels of oxidants/antioxidants [61,91].

The use of antioxidants has been widely applied in subfertile males [92]. Several works demonstrated that they are a positive contribution on sperm count, motility, morphology and also proved that they help reducing oxidative DNA fragmentation [93–96]. Although there are very few studies with randomized and placebo controls, Cochrane review suggests that the use of antioxidants causes from 1.8 to 4.6 fold increase in the chances of achieving a natural pregnancy. However, up to a 6.5 fold increase in miscarriages might be observed [97]. In ICSI treatments, it is still not clear if antioxidants could help on improving pregnancy and birth rates [98–100]. High quality studies including different groups of patients are necessary in order to elucidate the need of antioxidants in ICSI procedures.

Treatments for the reduction of double-strand sperm DNA damage should also reduce the miscarriage risk and the implantation failure risk in ICSI cycles, showing also less delay on embryo kinetics. Until our knowledge, no validated treatment reduce the incidence of MAR-region localized DSB. However, a study conducted in humans in 2006 by Schmid and colleagues demonstrated that men with daily caffeine consumption presented increased values of DSB measured with neutral Comet independently of male age in healthy non-smokers [101]. Caffeine is a known inhibitor of DNA repair, as it has been described that inhibits ATM kinase [102,103] and DNA resection in homologous recombination through Rad51 [104,105]. Also, it has been reported to affect cell cycle at both G1/S and G2/M checkpoints and inducing programmed cell death through p53-dependent pathway [106]. Studies in animals reported that caffeine administration to rats caused an impairment of pregnancy [107]. Other studies inducing DNA strand breaks in sperm cells through radiation and cultivating the oocytes and the produced embryos in caffeine demonstrated that chromosome and chromatid aberrations persist in the zygote, indicating oocyte DNA repair is inhibited by caffeine [17]. Since spermatocytes must produce double-strand breaks through Spo11 in prophase I in order to perform DNA recombination and later, they need to repair these DSB. According to previous results, the consumption of caffeine would impair ATM kinase and/or resection of double-strand breaks [104,105] and may induce that a few double-strand breaks would not be repaired, causing that

mature sperm cells present DSB [101]. Further basic studies are needed to explain how a spermatocyte with double-strand breaks can escape the pachytene checkpoint [108,109]. Reducing the incidence of DSB in sperm cell would improve clinical outcomes in terms of miscarriage and implantation in ICSI cycles.

**Author Contributions:** J.R.-M. and J.B. contributed in manuscript writing and revision.

**Funding:** This research was funded by the European Regional Development Fund and Instituto de Salud Carlos III (Economy, Industry and Competitiveness Ministry, Madrid, Spain; Project PI14/Q1 00119) and Generalitat de Catalunya (Project 2017SGR1796).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Automated Nuclear Cartography Reveals Conserved Sperm Chromosome Territory Localization across 2 Million Years of Mouse Evolution**

**Benjamin Matthew Skinner 1,\*, Joanne Bacon 1, Claudia Cattoni Rathje 2, Erica Lee Larson 3,4, Emily Emiko Konishi Kopania 4, Jeffrey Martin Good 4, Nabeel Ahmed Affara <sup>1</sup> and Peter James Ivor Ellis 2,\***


Received: 30 December 2018; Accepted: 28 January 2019; Published: 1 February 2019

**Abstract:** Measurements of nuclear organization in asymmetric nuclei in 2D images have traditionally been manual. This is exemplified by attempts to measure chromosome position in sperm samples, typically by dividing the nucleus into zones, and manually scoring which zone a fluorescence in-situ hybridisation (FISH) signal lies in. This is time consuming, limiting the number of nuclei that can be analyzed, and prone to subjectivity. We have developed a new approach for automated mapping of FISH signals in asymmetric nuclei, integrated into an existing image analysis tool for nuclear morphology. Automatic landmark detection defines equivalent structural regions in each nucleus, then dynamic warping of the FISH images to a common shape allows us to generate a composite of the signal within the entire cell population. Using this approach, we mapped the positions of the sex chromosomes and two autosomes in three mouse lineages (*Mus musculus domesticus*, *Mus musculus musculus* and *Mus spretus*). We found that in all three, chromosomes 11 and 19 tend to interact with each other, but are shielded from interactions with the sex chromosomes. This organization is conserved across 2 million years of mouse evolution.

**Keywords:** nuclear organization; sperm; morphometrics; chromosome painting

#### **1. Introduction**

Studies of the sub-nuclear localisation of chromatin often use fluorescence in-situ hybridisation (FISH) to detect DNA or RNA, or immunostaining to detect proteins. The images are subsequently analysed either manually or using some automated analysis tool. If the nucleus is circular or elliptical, it is commonly divided into concentric shells of equal area and the proportion of signal in each shell is measured (e.g., [1–3]). This has been amenable to automation, allowing analysis of thousands of cells, which, with appropriate statistical treatment, can yield valuable data at a scale that is still beyond the scope of 3D imaging techniques in time and cost.

However, if the nucleus is asymmetric, such as in sperm, a shell analysis is not sufficient. Frequently, nuclei are manually divided into geometric regions, and the number of nuclei with signals in each region are counted. For example, in spatulate sperm, such as pig or human, positions of loci are located into anterior, medial and posterior regions [4–6], or measured by proportional position along each axis [7]. Rodent sperm have a more interesting, falciform, hooked shape: They have

two axes of asymmetry, the anterior-posterior and the dorsal-ventral axis. This means that the location of a FISH signal can—in principle—be unambiguously localised and compared between nuclei. The determination of chromosome position is still manual, with more regions of the nucleus into which a signal may be assigned [8,9], or described without quantitation [10]. This is both time-consuming, and subjective, limiting the numbers of nuclei that can be analysed.

The positions of chromosomes or other loci in gametes (particularly sperm) is of great interest due to both the association of nuclear organisation with fertility in the clinic, in agriculture, and in evolutionary biology. Chromosome position has been linked with infertility in human males; men presenting with fertility problems have less consistent chromosome territories than healthy men [11–13]. Similarly, in farm animals, studies of nuclear organisation have discovered conserved sperm chromosome territories in boars [4], and wider evolutionary studies have shown conservation of some chromosomes, such as the X, from eutherian mammals to marsupial mammals and monotremes [14].

Newer sequencing-based approaches, such as Hi-C are being used to produce 3D maps of chromatin structure across multiple and even single nuclei [15–17]. Validating these results by microscopy is harder due to the number of cells that must be analysed, yet is necessary for our understanding of how chromatin patterns seen across millions of cells relate to chromatin structure within an individual nucleus. Three-dimensional imaging such as confocal microscopy provides high quality position information, but is time-consuming and costly in comparison to 2D fluorescence imaging.

Given this, there is a need to quickly and robustly assay nuclear organisation in 2D fluorescence microscopy images with greater precision than is currently available. Here, we demonstrate the use of automatic landmark detection in nuclei to rapidly localise, aggregate and compare nuclear signals without need for precise detection of the signal boundaries, or extensive manual thresholding and curation. We use this method to investigate the conservation of nuclear organisation between three mouse lineages, *Mus musculus musculus*, *Mus musculus domesticus* and *Mus spretus*. Of these, *M. spretus* has a notably different nuclear shape [18] to the others, being shorter and wider, allowing us to test whether chromosome position is conserved across structurally equivalent regions.

#### **2. Materials and Methods**

#### *2.1. Sample Collection*

We collected sperm from wild-derived inbred mouse strains *Mus musculus musculus* (PWK/PhJ), *M. m. domesticus* (LEWES/EiJ) and *Mus spretus* (STF). All animal procedures were subject to local ethical review by the University of Montana Institute for Animal Care and Use Committee (protocol identification number 002-13JGDBS-011613, approved January 16, 2013). Animals were bred at the University of Montana from mice purchased from Jackson Laboratories (Bar Harbor, ME, USA) or were acquired from Francois Bonhomme (University of Montpellier, France). Animals were housed singly or in small groups, sacrificed via CO2 followed by cervical dislocation, and tissues were collected post mortem for analysis. Sperm were collected and fixed in 3:1 methanol-acetic acid as previously described [18].

#### *2.2. Fluorescence In-Situ Hybridisation (FISH)*

Fixed sperm were dropped on poly-lysine slides, air-dried, and aged at 70 ◦C for one hour. Sperm were swelled in 10 mM DTT in 0.1 M Tris-Hcl for 30 min at room temperature (RT). Slides were rinsed in 2 × saline sodium citrate (SSC) and dehydrated through an ethanol series (70%, 80%, 100%, 2 min at RT). Chromatin was relaxed by incubating slides in 0.1 mg/mL pepsin in 0.01 N HCl at 37 ◦C for 20 min. Nuclei were permeabilized in 0.5% IGEPAL CA-630, 0.5% Triton-X-100 at 4 ◦C for 30 min, and dehydrated through an ethanol series. Slides and chromosome paints for chrX, Y, 11 and 19 (Cytocell, Cambridge, UK, AMP-0XG, AMP-0YR, AMP-11G, AMP-19R) were separately denatured in 70% formamide at 75 ◦C for 5 min, then slides were dehydrated through an ethanol series. Probes were cohybridised in pairs of 4 μL each of: chrX and chrY; chrX and chr19; chr11 and chr19. The probes were added to the slides, coverslips were sealed with rubber cement, and the slides were hybridised for 48 h at 37 ◦C. Coverslips were removed, and slides were washed in 0.7 × SSC, 0.3% Tween-20 at 73 ◦C for 3 min to remove unbound probe, then washed in 2 × SSC for 2 min at RT, rinsed in water and air-dried in the dark. Slides were counterstained with 16 μL VectorShield with DAPI (Vector Labs, Peterborough, UK) under a 22 × 50 mm cover slip and imaged at 100× on an Olympus BX-61 epifluorescence microscope equipped with a Hamamatsu Orca-ER C4742-80 cooled CCD camera and appropriate filters. Images were captured using Smart-Capture 3 (Digital Scientific UK, Cambridge, UK) with fixed exposure times for each fluorochrome.

#### *2.3. Image Analysis*

Analysis was performed using our image analysis software (Nuclear Morphology Analysis, available from http://bitbucket.org/bmskinner/nuclear\_morphology/wiki/Home/, version 1.15.0) for morphometric analysis of mouse sperm shape [18]. Here, we combine nuclear morphometry with FISH signal detection in order to rigorously quantify the distribution of chromosome territories within the asymmetric mouse sperm head. Within our images we detected 1445 PWK nuclei, 906 LEWES nuclei and 712 STF nuclei across all hybridisations (Figure 1B). The number of nuclei with FISH signals detected which were used for chromosome positioning analysis are given in Table S1.

This analysis, which we refer to as nuclear cartography is a form of mesh warping, achieved by overlaying a mesh onto each individual sperm nucleus and quantifying the distribution of the chromosomal signal within each face of the mesh (Figure 1C). This allows accurate, quantifiable 2D analysis of the signal distribution in each cell. Subsequently, since the mesh overlaid onto each sperm head is structurally equivalent, dynamic image warping is used to combine multiple individual nuclear outlines onto the consensus shape of the cell population (Figure 1D). Using this method, signal intensity can be averaged over multiple sperm heads, reducing the effect of background inhomogeneities and revealing the consensus two-dimensional location of the signal in the population as a whole.

For successful warping of the source image, the face of the mesh to which each pixel belongs must be determined. The critical step is the construction of the mesh, such that each face contains a structurally equivalent region of the nucleus. First, we identify key landmarks around the periphery of the nucleus (i.e., the apical hook, tail attachment site, and other areas of maximal curvature), as described previously [18]. Next, semi-landmarks are constructed by spacing a set number of equidistant points between each landmark (Figure 1C-i). These then serve as the peripheral vertices of the mesh. The internal vertices are created by walking through the points pairwise from the tip of the nucleus, and generating a vertex at the centre of the line connecting each pair (Figure 1C-ii). Internal and peripheral vertices are connected into the faces of the mesh (Figure 1C-iii). The same structural mesh is created for the consensus nucleus shape, and for each individual nucleus. An affine transform is applied to image pixels within each face, moving them to their equivalent positions in the consensus mesh. After pixels have been relocated, a gap-filling kernel sets any empty pixel to the average of the surrounding non-zero 8-connected pixels, as long as there are at least 4 non-zero surrounding pixels. This reduces smearing in cases where there is a large size difference between source and consensus mesh faces.

In this way, we warp the original images to fit the consensus nucleus. The warped images can be combined to reveal the locations of consistent nuclear signal. Random noise is averaged out, while consistent signals are reinforced. To avoid bias from higher or lower intensity signals in different nuclei, the FISH images are binarised before warping. Since the individual images are being warped to fit a template shape, it is possible to choose any template with the same underlying graph structure in the mesh. This allows comparison of FISH signal distributions between different hybridisations.

To compare signal distributions between warped signals, we used an open source implementation of a multi-scale structural similarity index measure (MS-SSIM\*) [19,20], which quantifies visual similarity between images [21] on a scale of 0 (no similarity) to 1 (identical images). To further assess adjacency of chromosome territories, we identified the chromosomal signals within the nuclei by thresholding [3], and measured the distances between the centres of mass of co-hybridised chromosomes. Statistical analyses were performed in R 3.5.1 [22], and charts were generated using the cividis colour palette [23].

**Figure 1.** The process of warping fluorescence in-situ hybridisation (FISH) images. (**A**) Examples of un-FISHed nuclei from the three strains, as described in [18]. (**B**) After FISH, nuclei are automatically identified and landmarks are discovered. (**C**) A mesh is created from the consensus nuclear shape; (i) peripheral vertices are evenly spaced between landmarks; (ii) internal vertices divide vertex pairs from the tip; (iii) all vertices are joined. The equivalent mesh is constructed for each nucleus. (**D**) The FISH signal image is transformed to move every pixel to its location in the consensus mesh. The warped images are combined to yield the composite signal image. Mouse strains *Mus musculus musculus* (PWK), *M. m. domesticus* (LEWES) and *Mus spretus* (STF).

#### **3. Results**

#### *3.1. The Sex Chromosomes Have Conserved Position in Mouse Sperm Nuclei*

The process of hybridising FISH probes to sperm nuclei required a considerable swelling step due to the highly compact chromatin. The nuclear area doubles from about 20 μm<sup>2</sup> to about 40 μm2, with the majority of the swelling in the dorsal/ventral axis (Figure S3). This swelling distorts the nuclear shape; our method for automated nucleus and landmark detection [18] was able to identify and orient swelled nuclei successfully, despite the fewer landmarks available.

Confident that we could orient a FISH signal within the nucleus, we applied the new technique to FISH images of mouse sperm from three strains, using chromosome paints for the X and Y chromosomes. These have been previously reported in *M. musculus* strain C57Bl6 to lie under the acrosome [8,9]. Nuclei and signals were detected from the captured images, a consensus nuclear shape was calculated for each strain, and each FISH image was warped onto that consensus shape. A composite image was created by layering each FISH image, effectively providing a heat-map of signal location within the nucleus.

Our results confirm a consistent sub-acrosomal location for both X and Y chromosomes (Figure 2). Following the signal warping onto the population consensus, we observed that both X and Y chromosomes have overlapping territories (Figures 3 and 4).

**Figure 2.** Example images showing the sex chromosome positions within the three strains. Scale bar represents 5 μm.

**Figure 3.** Composite signal distributions for chromosomes X, Y, 11 and 19 in (**A**) PWK, (**B**) LEWES and (**C**) STF. The sex chromosomes occupy a consistent territory apical and dorsal to the centre of mass, generally under the acrosome but rarely extending fully to the periphery of the nucleus. Chromosomes 11 and 19 are more widely distributed, with the predominant location basal and ventral to the centre of mass.

**Figure 4.** Overlay of warped distributions from Figure 3 shows the similarities between chromosome X and Y territories, and 11 and 19 territories in (**A**) *PWK*; (**B**) *LEWES*; and (**C**) *STF*. White shows regions of overlap. Chromosomes X and 19 (and X and 11) are predominantly non-overlapping.

#### *3.2. Chromosomes 11 and 19 Occupy Similar Nuclear Addresses*

With the sex chromosome locations confirmed to be conserved, we decided to examine two further chromosomes, both of which have previously been reported in the literature. Chromosome 19 has been described in C57Bl/6 mice to frequently lie toward the base of the nucleus [8]. Furthermore in Hi-C experiments, chromosomes X and 19 had a low association in *M. musculus* C57BL sperm chromatin; chromosome 19 and chromosome 11 had a moderate association with each other [17]. For this reason, we hypothesised that chr11 and chr19 might share a similar distribution, and that this would be distinct from that of the sex chromosomes.

The composite signal position data are shown in Figure 3. The patterns are indeed different to that of the sex chromosomes. The majority of the signal lies ventral and basal to the centre of the nucleus, yet there are clearly instances of signal throughout the nucleus, from the basal region near the tail attachment point to the apex and partially extending into the hook. Some examples of these positions in individual nuclei are shown in Figure 5.

Although hybridization efficiency was poorer in *M. spretus*, the same patterns are apparent as in the *M. musculus* strains. Interestingly, we observed instances of both chr11 and chr19 below the acrosomal curve, in which the chr19 was generally more elongated than chr11 (see Figure 5B,F). Where chromosome 19 was co-hybridised with chromosome X, we were able to see rare instances of chrX and chr19 lying adjacent, with chrX more internal (Figure S1).

**Figure 5.** Examples of individual chromosome positions for chr11 (**A**,**C**,**E**) and chr19 (**B**,**D**,**F**) in the three strains; the chr11 and chr19 panels do not show the same nuclei. While the majority of the signals for each chromosome were observed ventral and basal of the nuclear centre (column 1), we found territories at the base of the nucleus (column 2), under the acrosome (column 3), and along the ventral surface below the hook (column 4). Scale bar represents 5 μm.

Given the similarity in overall signal distributions, we looked to see if chr11 and chr19 tend to lie adjacent to each other in individual nuclei. Visually, we can see that they are occasionally adjacent, but are not always associated. Measurement of the distance between the chromosome signal centers of mass showed no difference between chr11 and 19 or between chr11 and X, nor did a comparison of individual nucleus warped signal images via a MS-SSIM\*, a technique also used in comparisons of radiological images [24] (*p* > 0.05, Wilcoxon rank sum tests; Figure 6). We conclude that, although chr11 and chr19 have a similar range of possible addresses to occupy within an individual sperm head, they do not necessarily interact, and are no more likely to be adjacent than chromosomes 11 and X. It is however important to appreciate that our data addresses chromosome territories as a whole,

rather than individual loci, and further work will be needed to robustly compare our data with the Hi-C data from [17] (see also Section 4).

**Figure 6.** Chromosomes 11 and 19 do not colocalize within individual nuclei; colocalization of signals shows no difference comparing chr11 and chr19 as when comparing chrX and chr19 by either multi-scale structural similarity index (MS-SSIM\*) (upper) or the distances between the chromosome signal centers (lower).

#### *3.3. Quantification of Signal Positions Reveals Conserved Chromosome Organisation across Species*

In order to quantify the similarity of signal locations both within and between strains, we warped images from all three strains onto the LEWES (domesticus) consensus outline. These consensus warped images were compared using MS-SSIM\*, revealing the similarities in the range of possible nuclear addresses a chromosome could occupy in each strain. The X and Y territories had high structural similarity to each other in all three strains, and had high concordance between strains (Figure 7). Similarly, we saw greater similarity between chr11 and chr19 in all three strains. The pattern was slightly less clear between *M. spretus* and the other strains, presumably due to the lower hybridisation efficiency of the probes. To confirm there was no artefactual bias introduced by the choice of LEWES as the destination shape, we examined the effect of warping signals onto either the PWK or STF consensus outlines, and found that this made little difference in the values obtained (see also Figure S2, Table S2). This demonstrates that our method is robust for comparing differently shaped nuclei as long as we can define structurally equivalent landmarks.

#### **4. Discussion**

We have presented here a new method for quickly and efficiently mapping chromosome position in asymmetric nuclei, such as sperm, based on linking chromosome signals with morphometric information about nuclear structure. Using this analysis, we have been able to measure and quantify differences in chromosome territory position in sperm from three mouse strains. All mouse strains studied here diverged, at most, 3 million years ago [25,26], and the karyotypes of *M. musculus* and *M. spretus* both have 40 chromosomes [27]. *M. musculus* and *M. spretus* are able to produce hybrids in laboratory conditions, of which the female F1 is fertile [28]. We have demonstrated here that orthologous chromosomes adopt similar conformations in the three strains, despite differences in nuclear shape.

**Figure 7.** Similarity of signal distributions in composite warped images measured by MS-SSIM\*, on a scale of 0–1, where 0 indicates no similarity, and 1 indicates identical images. Images were warped in turn onto the consensus shapes of LEWES, PWK and STF. There is high correlation between the MS-SSIM\* scores obtained when images are warped onto different target shapes (see Figure S2). Both within strains and between strains, there is a clear similarity between the distributions of chrX and chrY, and chr11 and chr19, but little similarity between the reciprocal combinations.

#### *4.1. Chromosomes X and Y Have a Conserved Dorsal/Sub-Acrosomal Position*

Both the mouse X and Y chromosomes have been subject to massive amplification of euchromatic sequences. The full sequence of a *M. m. musculus* C57Bl/6 Y chromosome revealed the complex ampliconic structure [29], and demonstrated the presence of similar amplicons on the *M. spretus* Y. These amplicons are thought to arise from genomic conflict in spermatids [30], and copy number measurements of individual ampliconic genes suggests *M. spretus* has generally amplified the same gene families as *M. musculus*, with the exception of X-linked H2al1, which has amplified specifically in the *M. musculus* lineage.

Despite the close evolutionary relationship of *M. musculus* and *M. spretus*, some small rearrangements involving the sex chromosomes have been documented—for example, the Clcn4 gene, X-linked in most mammals including *M. spretus*, is autosomal in *M. musculus* [31], with clear translocation breakpoints surrounding the gene [32].

Given the overall structural similarity of the orthologous chromosomes, it is likely they occupy a similar volume within the nucleus, and are subject to similar conformational constraints. The sex chromosomes have been previously described to adopt a dorsal position in the rodent sperm nucleus [8,9], and have been seen to be sub-acrosomal in human, marsupial and monotreme sperm [14]. It has been suggested that the X chromosome in X-bearing sperm is the first to enter the egg during fertilisation. The position of the Y in marsupials is not reported, but as in mice, it is likely that the Y adopts the same position as X simply because the space is available. In monotremes, the platypus Y chromosomes do show a similar distribution to the X chromosomes [33]. Since the sex chromosomes are different sizes—approximately 90 Mb versus 170 Mb—there must be differences in the chromatin packing to allow them to occupy the same nuclear volume. In future we will be interested to study the impact of chromosome constitution on nuclear morphology.

#### *4.2. Chromosomes 11 and 19 Have a Conserved Ventral/Basal Distribution*

Chromosome 19 has been observed by others to lie in the basal region of the nucleus in approximately two thirds of nuclei based on imaging and manually scoring at least 350 *M. musculus* C57Bl/6 sperm nuclei [8,9]. Our results support these data, and demonstrate conservation of this position across species. The signal in *M. spretus* is less clear, likely due to the cross-species hybridisation, but the pattern is still distinguishable.

Our data from co-hybridisations suggest that although chr11 and chr19 adopt a similar overall location, they do not always lie adjacent within a single nucleus. This indicates that while they have preferred regions of the nucleus, they are mostly unconstrained with regard to each other. Aggregate data from Hi-C experiments in C57Bl/6 sperm [17] have indicated that chr19 is infrequently associated with the X chromosome (and by inference, the Y chromosome), and that chr11 is only moderately associated with both chrX and chr19. It is, however, currently unclear why Hi-C shows chromosome 19 to be more strongly associated with chromosome 11 than the X chromosome, given our data showing that these three chromosome territories are on average equidistant. One potential explanation is that while our measurements focus on the centre of each chromosome territory, interactions occur at the periphery of territories in cells where they abut each other. The mouse sperm head tends to have a DAPI-dense chromocenter core, and that the X/Y and 11/19 regions are deduced to usually lie on opposite sides of this. Potentially, this core forms a barrier to inter-chromosomal interactions (Figure 8). As an analogy, Cersei and Jaime (chromosomes 11 and 19) may both live in the ground floor flat, but they do not take up the exact same physical space, remaining on average a few meters apart. Meanwhile, their upstairs neighbor Daenerys (chromosome X or Y) is roughly equidistant from them, but does not interact with them due to the barrier in between (the centric heterochromatin). However, when averaged across the course of many days, Cersei and Jaime collectively occupy the downstairs flat, while Daenerys occupies the spatially distinct upper floor. A higher resolution investigation of individual loci found to be associated in the Hi-C data will help resolve this distribution.

**Figure 8.** A simple model of how our data may relate individual cells to aggregate measurements. In individual cells, chr11 and chr19 (blue/yellow) frequently lie adjacent, and more rarely further apart. Chromosomes X and Y (purple) lie consistently below the acrosome. In contrast, chromosomes 11 and 19 do not have strictly fixed addresses, but reside interchangeably within the same general area of the nucleus. Thus, chromosomes 11 and 19 colocalise in the aggregate distribution despite not overlapping within any individual nucleus. In this model, the chromocenter core acts as a physical barrier to interchromosomal interactions, explaining why Hi-C detects more 11/19 interactions (indicated by \*) than 11/X or 19/X interactions despite the similar physical distances between the centres of mass of the three territories.

Overall, our measurements tend to support previous Hi-C and FISH findings in laboratory mouse sperm, and provide evidence that the same patterns will be found in *M. spretus*. The concept of spatial synteny—the conserved 3D position of orthologous loci despite karyotypic rearrangements—has been proposed [34], and there is increasing evidence for conserved nuclear organization of genes following chromosomal rearrangements [35]. As we extend our studies, it will be interesting to compare the positions of the full set of chromosomes, to better understand how the shorter and fatter *M. spretus* nucleus maps on the longer, thinner *M. musculus* nucleus. Further comparisons with other mouse strains with greater shape variability will also be of value; for example BALB/c, which have frequent shape abnormalities and aneuploidies [18,36].

Studies of strains with other aneuploidies, chromosomal rearrangements or Robertsonian fusions, which will additionally constrain chromosome territories will be of interest. In humans, no gross morphological differences in sperm nuclei have been seen in men carrying Robertsonian fusions [37]. However, in boars (*Sus scrofa*), while gross nuclear morphology was not perturbed in animals carrying a t(13;17) Robertsonian translocation, differences were apparent in the positions of the affected chromosomes [38]. Extending beyond mice, rats (*Rattus rattus*) have a much thinner hooked sperm nucleus; rat chromosomes have been mapped in developing spermatids from stages 7–13. The nucleus is compressed from a structure which at stage 10 is markedly similar to a mature mouse sperm nucleus [39]. The associated dynamics of nuclear reshaping during spermiogenesis, and chromosome repositioning are an area of active research [10].

#### *4.3. This Method Allows Rapid Screening of Large Numbers of Nuclei*

In this analysis, we examined more than 3000 nuclei, and the method scales to greater numbers with little additional time or user effort after images have been captured. The warping algorithm processed these nuclei in under half an hour on a desktop computer equipped with an Intel i5-2400 processor and 16 Gb memory, with the total user time excluding image capture being a few hours. This is of course experience and hardware dependent, but the key point is that the total analysis time can be measured in hours rather than days. Importantly, our analysis does not rely on extensive manual classification of chromosome position, making it less subjective than current approaches, and amenable to automation. The use of a mesh to warp signals from different nuclei onto a single template shape allows for quantitative measurements of the similarity of signal distributions between images, and in principle will allow us to study small differences in locus position that have been beyond the scope of current scoring systems. Beyond chromosome territory positioning, it is also amenable to the study of single BAC probes, or any small probe generating a punctuate signal, as long as sufficient nuclei are analyzed to generate an aggregate signal; together with Hi-C data this will allow us to study which intra- and inter-chromosomal folding contacts are retained in the sperm head, and address long standing questions of whether sperm chromatin organisation represents an echo of round spermatid chromatin organisation, or prefigures future chromatin folding dynamics in the fertilised zygote.

A further methodological interest would be to identify reliable internal structural features within the nucleus, using DAPI or other stains. Currently we use only peripheral features as landmarks, which puts limits on the accuracy of our mesh when deforming images. More internal structural data would permit more complex morphometric approaches such as Teichmüller mapping, which has been used successfully in analysis (for example) of wing shape in Drosophila species [40].

#### **5. Conclusions**

Here we have demonstrated a new method for locating chromosome paints or other nuclear signals within mouse sperm nuclei, which is in principle also applicable to other asymmetric nuclei, including nuclei with fewer axes of asymmetry, such as spatulate sperm nuclei. We have used this technique to confirm the non-random positioning of the sex chromosomes, and of chromosomes 11 and 19, and demonstrated quantitation of signal positions allowing comparison between different

strains and species. Importantly, we have integrated this method into existing open-source image analysis software designed for other biologists.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4425/10/2/109/s1, Figure S1: Chromosomes X and 19 co-hybridization; Figure S2: Comparison of MS-SSIM \* scores using different warping templates; Figure S3: Examples of swelled and unswelled nuclei; Table S1: Numbers of nuclei with FISH signals analysed in this study., Table S2: Complete MS-SSIM\* comparisons between warped composite images, including the individual similarity components of contrast, luminance and structure.

**Author Contributions:** Conceptualization, B.M.S. and P.E.; Methodology, B.M.S. and P.E.; Software and Validation, B.M.S.; Investigation, J.B., C.C.R.; Data Curation and Formal Analysis, B.M.S.; Visualization, B.M.S.; Supervision and Project Administration, P.E.; Writing—Original Draft, B.M.S. and P.E.; Writing—Review and Editing, B.M.S., C.C.R., J.M.G., E.L.L. and P.E.; Resources, J.M.G., E.L.L., E.E.K.K., N.A. and P.E.; Funding Acquisition, N.A. and P.E.

**Funding:** B.M.S. was supported by the Biotechnology and Biological Sciences Research Council (BBSRC, BB/N000129/1). P.E. and C.C.R. were supported by H.E.F.C.E. (University of Kent) and by the BBSRC (BB/N000463/1). J.M.G. and E.L.L. were supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health (R01-HD073439 and R01-HD094787) and the National Institute of General Medical Sciences (R01-GM098536). E.E.K.K. was supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. (DGE-1313190).

**Acknowledgments:** We thank the animal handling staff at the University of Montana.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
