BIOLOGY OF REPRODUCTION 48, 1193-1201 (1993)

MINIREVIEW
Deoxyribonucleic Acid Loop Domain Tertiary Structure in Mammalian Spermatozoa'
W. STEVEN WARD 2

Division of Urology, Program in Cell and Developmental Biology, Robert Wood Johnson Medical School
New Brunswick, New Jersey 08903-0019

ABSTRACT
In the four decades since the discovery of the basic structure of the DNA double helix, researchers have been investigating
the more dynamic tertiary structures that DNA assumes in the various forms of chromatin. The tertiary structure of DNA is
important because it is directly related to the function of the genome: for the cell to access the information that is present in
the genome accurately and efficiently, the DNA must be in an organized form. This paper reviews the recent work on one
particular enigmatic structural form of eukaryotic DNA, that of the highly condensed spermatozoa. Based on the literature and
on recently completed experiments in the field, a new model for DNA packaging in sperm nuclei is presented. In this model,
each individual DNA loop domain in the sperm chromatin is condensed into a toroid-shaped structure termed the DNA loop
doughnut.

INTRODUCTION organization. Thus, whatever conformation sperm DNA ul-

It was the discovery of the structure of DNA that eluci- timately assumes, it developed from the relatively well-
dated how DNA could function as the biochemical com- understood somatic DNA chromatin structure and must be
ponent of which genes were made. The double-helix struc- related, however distantly. Here, this approach has been
ture made it immediately obvious how complex information used to derive a model for one level of sperm DNA orga-
could be stored, in a linear sequence of codes, and how nization that is not yet understood, the structure of the sperm
the genome could be efficiently replicated and transferred DNA loop domain in its condensed form (Fig. 1).
to the next generation [1]. In the four decades since, much
attention has been devoted to the more complex, tertiary THE HIERARCHY OF SOMATIC CELL DNA
structures the DNA assumes at various levels when it is ORGANIZATION AND ITS RELATIONSHIP TO FUNCTION
packaged into chromatin. These higher-order structural To begin to understand the function of DNA organiza-
motifs have proven to be just as important to DNA function tion in spermatozoa, we must first review the more fully
as the double helix [2,3]. Moreover, it is now becoming defined structural hierarchy of DNA in somatic cells, from
clear that the structural organization of DNA determines the the DNA double helix to mitotic chromosomes.
functional fate, that is, the differentiation, of a cell [2, 4, 5].
While a consensus model for the organization of DNA Double Helix
in somatic cells is now beginning to form [3], it has been The genetic information is coded in a four-letter, linear
only partially defined for the more highly condensed DNA code in one strand of the DNA. The duplicity of the com-
of spermatozoa [6, 7]. The well-documented hierarchy of DNA plementary strands allows for the perfect reproduction of
packaging for the somatic cell is a useful model from which a second copy of the entire genome during cell division
to begin searching for the specifics of sperm DNA organi- [1].
zation. Each level of DNA packaging in somatic cells has a
specific functional jurisdiction. By identifying and compar- Nucleosome
ing the homologous structural motifs in sperm DNA, we Every 200 bp of DNA are coiled twice around an octo-
may discover more about the functions of both sperm nu- mer of histones to form a nucleosome [8]. This serves at
clei and the mammalian genome. least three functions. First, it shortens the meter of DNA
Using somatic cell DNA organization as a paradigm for that must be packaged into the average mammalian cell by
the study of sperm DNA is logical because spermatozoa are a factor of 6 [3]. Second, the octomer may help to control
derived from germ cells that have a somatic cell-type DNA the conformation of the DNA by opening into half-nucleo-
somes when genes are being transcribed [9]. Third, the DNA
Accepted February 10, 1993.
is wound around the histones in a direction that negatively
Received December 11, 1992. supercoils the DNA, that is, in a direction that tends to un-
'Supported by NIH grant HD28501 and by the Edwin A Beer Award (NY Acad wind the double helix [10]. This negative superhelicity al-
Med).
2
Correspondence: Division of Urology, Program in Cell and Developmental Bi-
lows the double helix to be much more easily separated,
ology, MEB-588, Robert Wood Johnson Medical School, New Brunswick, NJ 08903- exposing the template strand, for replication or for tran-
0019. FAX: (908) 418-8013. scription upon removal of the histones [11].

1193
1194 WARD

Mitotic Chromosome
Exactly how the DNA loop domains are compacted into
the mitotic chromosome remains controversial; the two
models that exist have been compared by Pienta et al. [3].
Both contain all the structural motifs discussed above. The
nuclear matrix condenses, coiling the associated DNA loop
domains into a compact form. In the radial loop model, 18
loop domains are wound around a central core to form a
miniband, whereas in the radial coil model, the loop do-
mains are first coiled into a larger, 240-nm fiber that is then
further coiled into stacks. The function of the short-lived
chromosome configuration is to package the newly repli-
cated DNA into units that allow for efficient separation of
the sister chromatids into the two daughter cells, while re-
FIG. 1. DNA is organized into loop domains, anchored to the sperm taining the specific organizational motifs present in the par-
nuclear matrix (right), diagrammed here for a hamster sperm nucleus. It is ent cell. The organized condensation of the loop domains
not known how these DNA loops are packaged in the fully condensed sperm accomplishes both tasks.
nucleus.

SPERM DNA ORGANIZATION AND ITS RELATIONSHIP

TO FUNCTION
The Solenoid
While many recent advances have contributed to defin-
DNA coiled into nucleosomes is further compacted into ing the homologous hierarchy of DNA organization in
a fiber with a diameter of 30 nm. The most accepted model mammalian spermatozoa, the complete hierarchy has not
for how DNA is packaged in this 30-nm filament is the so- yet been elucidated. The levels of sperm DNA organization
lenoid configuration. Six nucleosomes are coiled around so far established are as follows.
each other to form one turn that has a diameter of 30 nm
[12]. This shortens the total length of the genome even fur- Double Helix
ther. The existence of the solenoid remains controversial, The double helix in sperm DNA is almost identical to
but it is the most probable conformation of the 30-nm fil- that of DNA in somatic cells; the genetic code is contained
ament of eukaryotic chromatin. The model for sperm chro- within the four-letter code of DNA. The only documented
matin put forth in this paper is presented with the solenoid differences are the increased length of telomeres in sperm
as the parent structure (i.e., during spermiogenesis), but it DNA [20] and the extent of DNA hypermethylation [21].
is not dependent upon it. This will be discussed in greater
detail below. Protamine-DNA Complex
This level of sperm DNA organization is homologous to
DNA Loop Domains nucleosomes in somatic cells, but it is very different in
structure. The DNA binding proteins of mammalian sperm
The 30-nm solenoid fiber of chromatin is arranged into DNA are protamines, highly basic proteins with long stretches
loop domains of an average of 60 kbp in length, which are of arginine residues [7]. They bind to DNA lengthwise along
attached at their foundations to a structural component of one of the grooves, with each positively charged arginine
the nucleus termed the nuclear matrix [13, 14]. They are residue neutralizing one negatively charged residue of the
attached by specific DNA sequences that are intimately re- phosphodiester backbone of the DNA. This transforms the
lated to function [2, 5]; active genes [2, 5,15] and origins of polyanionic DNA into a neutral polymer, and the DNA-prot-
DNA replication [14,16] are associated with the nuclear ma- amine complexes can then bind together by van der Waals
trix at the bases of the loop domains. The attachment sites forces in a linear fashion, at the molecular level [7]. This
that are related to gene transcription have been proposed model, unlike the homologous structure in somatic cells,
to be transient [4], but this idea has been challenged [17]. the nucleosome, does not confer any supercoiling on the
Razin and colleagues [17,18] have shown that the chicken sperm DNA [6]. This prediction has been confirmed by two
ot-globin gene is attached to the nuclear matrix at different independent laboratories, each of which provided evidence
sites when it is active and inactive. Finally, it has been shown that protamine-bound sperm DNA was not highly super-
that different cell types within the same organism have dif- coiled, if at all [22, 23].
ferent genes associated with the nuclear matrix [2,19], sug- The functional significance of this arrangement may be
gesting that organization of DNA on the nuclear matrix may related to the inertness of sperm DNA. Since the sperm cell
play a significant role in cell differentiation. does not replicate its DNA or transcribe RNA, its DNA has
 SPERM DOUGHNUT LOOP DOMAIN MODEL 1195

no teleological need to be supercoiled (there is one reason PROPOSED MODEL FOR THE SOLENOID EQUIVALENT
for sperm DNA to be very slightly supercoiled, and that is IN SPERM CHROMATIN
packaging, as discussed in the model below). The overrid-
ing evolutionary pressure on protamine is to condense the Problem Being Addressed
sperm genome to a tightly packaged, protected state, and Recent evidence, together with the published data on
the supercoiling can be temporarily sacrificed for the transit. sperm chromatin structure discussed above, makes it pos-
The protamine DNA binding model [7] accomplishes this sible to propose a model for the solenoid equivalent in
task. mammalian spermatozoa. This structural motif is important
because we do not yet understand how the DNA loop do-
mains are packaged into the sperm nucleus (Fig. 1). Unlike
Sperm Solenoid Equivalent
the somatic cell histones, protamines do not seem to de-
Very little published evidence addresses the sperm DNA crease the length of DNA at all, so how are the loops pack-
equivalent of the solenoid in somatic cell DNA. Electron aged into the nucleus? Are they simply laid down linearly
micrographic techniques routinely fail to elucidate any dis- beside each other along the entire length of the sperm nu-
cernable structure in sperm chromatin because of its high cleus, or are they folded or coiled into a novel macro-
electron density [24]. This level of sperm DNA packaging is molecular structure?
the subject of the model described below.
Evidence
Sperm DNA Loop Domains The model must incorporate, and depends upon, all the
Topologically constrained DNA domains were first de- available evidence for sperm DNA structure. This includes
scribed by Risley et al. [23] for amphibian spermatozoa. In the data already discussed: that protamine binds to DNA
their study, the size of the frog loop domains were ap- lengthwise along one of the grooves [7], that DNA loop do-
proximately 25 kb, much smaller than those reported for mains exist (Fig. 1), and that the DNA of sperm nuclei con-
somatic cells. Similarly, hamster sperm DNA has been shown taining protamines is not supercoiled [22, 23]. There are three
to be organized into loop domains attached at their bases additional pieces of evidence that have been included in
to a sperm nuclear matrix [22] (Fig. 1). These loop domains preparing the new model, discussed below.
are about half the size (47 kbp) of the loops found in so- Sperm DNA is packaged in discretefoci. The first piece
matic cells of the same animal (76 kbp), but are otherwise of evidence comes from in situ hybridization [29] and im-
very similar [22]. Moreover, loop domains in both the ham- munohistochemical staining [28] experiments performed on
ster [25] and the chicken [26] have been shown to be at- sperm nuclei for centromeres. In every example, the cen-
tached to the nuclear matrix by specific sequences. This tromeres were located at discrete foci within the sperm nu-
suggests that sperm DNA is specifically organized rather than cleus, rather than being spread out along its length. In situ
randomly packaged into the nucleus. hybridization experiments with probes specific for chro-
These experiments suggest that sperm DNA is organized mosome Y in human sperm have shown similar results
as specifically as that in somatic cells, and that this orga- [30, 31]. These data suggest that the DNA loop domains are
nization may be related to function in the same manner. It folded or coiled into small domains within the sperm nu-
is possible, for example, that the sperm nuclear matrix may cleus, rather than collapsing lengthwise with little or no
provide a cell type-specific organization of the human ge- bending.
nome needed for fertilization and embryonic development, Preliminary experiments from our laboratory using a te-
in the form of DNA loop domain attachments. lomere-specific repeat probe have given similar results. When
condensed sperm nuclei were probed, a pattern of spots
was seen, but decondensed nuclei demonstrated a series
Sperm Equivalent of the Mitotic Chromosome of linear streaks. These data support the idea that the DNA
There may not be a structure equivalent to the mitotic loops are folded or coiled into discrete foci in the sperm
chromosome in sperm nuclei because spermatozoa do not nucleus.
undergo mitosis. However, the fact that they are the recent Sperm chromatin has the appearanceof 60-nm nodules.
products of a meiotic division demands at least consider- Allen et al. [32] have recently developed a technique for
ation of the possibility. It is known, for example, that hu- examining biological structures by atomic force micros-
man sperm DNA can be prematurely condensed into single copy (AFM). Images of amembraneous bull and mouse sperm
chromatids with the same banding patterns as the mitotic nuclei and partially decondensed mouse sperm chromatin
chromosomes in somatic cells [27]. Also, fully condensed provided by this new technique have revealed that the DNA
sperm chromatin retains some centromeric specific pro- is not organized as long, linear bundles of fibers. Instead,
teins also found in somatic cells [28]. These data suggest sperm DNA appears to be coiled into large nodules that
that at least some remnant of the mitotic structure is pres- are somewhat variable in size, and each is large enough to
ent in sperm DNA. contain a replicon-equivalent of DNA. Koehler et al. [33]
1196 WARD

noted a similar arrangement of rat sperm DNA into nod- solenoid configuration, there are 13 supercoils of DNA, two
ules, or beads, with a diameter of 13-25 nm. Arrays of the for each of the six nucleosomes and one for the solenoid
nodules are tightly packed inside the nucleus in a three- itself (Fig. 2A). The first proposal was that as protamines
dimensional arrangement that has not yet been described. replace the histones, the 12 nucleosomal supercoils are re-
Volume of the mammalian sperm nucleus is greater than moved so that the DNA is less curved or more linear, as
previous data suggested. A previous discussion of sperm originally suggested [7]. But the criterion that requires the
DNA organization [6] used volume measurements for sperm fewest changes necessary predicts that the single solenoid
nuclei that were obtained by serial section electron mi- supercoil remains. This simplest model does not fit the data,
croscopy. Allen et al. [34] have since discovered that these however, because the 1200 bp of DNA in one solenoid su-
volumes represent the minimum possible volume for the percoil would form a circle with a diameter of 129 nm,
nucleus, one that reflects the extensive dehydration re- twice the size of the observed round structures. If, on the
quired for embedding the sperm in plastic. New data, ob- other hand, only 11 of the 12 nucleosomal supercoils were
tained by AFM of fully hydrated sperm nuclei, indicate that removed, the DNA would be coiled into two spiraling cir-
the volume of the nucleus is more than twice previous es- cles for each solenoid turn in the chromatin (Fig. 2). The
timates. These studies have shown that sperm chromatin in two spiraling circles of DNA complexed with protamines
vivo is extensively hydrated, and that the larger volumes are would bind to each other by van der Waals forces, as Bal-
consistent with the proposed nodular organization or coiled horn proposed [7].
sperm DNA. Using the previous, smaller measurements, it This first step of the model predicts that sperm DNA is
was calculated that the mouse sperm nucleus contained actually slightly negatively supercoiled, to a degree that is
barely enough volume to contain the naked DNA, with very only 2/13, or 15%, of histone-bound DNA. This is consis-
little room for the bulky three-dimensional structures found tent with the two published reports demonstrating little or
in chromatin [6]. These newer measurements allow the no superhelical density in protamine-bound sperm DNA
possibility that sperm chromatin also contains larger or- [22, 23] because the methods used were not sensitive enough
ganizing structures. to distinguish between a complete lack of superhelical den-
sity versus an 85% decrease. Such a tight regulation of the
THE SPERM DNA LOOP DOUGHNUT MODEL superhelical density of DNA during spermatogenesis re-
quires the presence of topoisomerase. Morse-Gaudio and
Criteriafor the Model Risley [35] have recently demonstrated that topoisomerase
II is present in all cell types during spermatogenesis except
Given all the data discussed above, it is now possible to the spermatozoon itself.
construct a model for sperm DNA structure intermediate Finally, the focal replacement of histones by protamines
between protamine binding and loop domain organization. predicted by this model is consistent with the earlier model
The criteria for the model are twofold. The first is obvious: of Risley et al. [23], who suggested that the supercoiling
it must be consistent with all available data on sperm DNA domains might serve as units of structural transitions dur-
structure. The second criterion is suggested by the fact that ing spermiogenesis.
spermatozoa are derived from stem cells that contain his-
tones, and the organization of sperm DNA must therefore Second Step: Condensation of the DNA Loop Domain
be consistent with the traditional somatic cell hierarchy, in- The next step of the model is the formation of the cir-
cluding nucleosomes and solenoid formation. Therefore, cular structures seen by AFM [32] in sperm chromatin. The
the solenoid equivalent in sperm chromatin ought to con- model predicts that each of these structures represents a
tain as few changes as possible from the parent, solenoid collapsed DNA loop domain. A comparison of the parent
structure of the stem cells. solenoid loop conformation of DNA bound to histones and
the final doughnut loop model of DNA bound to pro-
First Step: Histone Replacement by Protamines tamines is diagrammed in Figure 3. The schematic inter-
The basic tenet of the model is that the 60-nm nodules mediate is drawn only as an instructive diagram and is not
discussed above [32] each represent one DNA loop domain. predicted as a real, functional intermediate that occurs dur-
This focal localization of a single loop domain is consistent ing spermiogenesis. The actual transition must be much more
with the in situ hybridization data already cited. Since we complex, involving transitional proteins that are not con-
know that protamine binds DNA in a linear, side-by-side sidered in this model. The comparison of the histone-so-
fashion, the most efficient way to package the DNA of a lenoid configuration and that of the sperm nuclei is dia-
loop domain into a round structure is in wide, concentric, grammed in Figure 3, for a loop domain of 47 kbp, the
and overlapping circles, as in a wound ball of string. average size reported for hamster spermatozoa [22]. Each
For the purposes of this model, the solenoid configu- of the 39 solenoid turns (each solenoid turn = 13 super-
ration is assumed to be the major conformation of DNA in coils of DNA) in the parent loop domain (Fig. 3, left) even-
the somatic cell 30-nm chromatin fiber. In one turn of the tually becomes two spiraling circles of DNA when replaced
 SPERM DOUGHNUT LOOP DOMAIN MODEL 1197

FIG. 2. Model of protamine replacement in mammalian sperm DNA. A) One coil of DNA in the histone-bound, solenoid configuration has 13 supercoils
of double helix DNA, two for each nucleosome and one for the solenoid itself. The model proposed here predicts that as the protamines replace the
histones during spermatogenesis, all but two of the 13 supercoils are uncoiled, as shown on the right (the DNA double helix has been drawn as a single
line). B) On the left, two coils of DNA in the solenoid conformation (total of 26 superhelical turns) with DNA wrapped around the histone octomers are
diagrammed. On the right, the same amount of DNA is shown bound to protamines (shaded) in an expanded doughnut configuration, drawn to the same
scale (here, the 26 supercoils of the two solenoid turns become four doughnut supercoils). The histones keep the DNA in the relatively bulky solenoid
configuration, while the protamine-complexed DNA concentric circles can bind together into much thinner structures (see Fig. 3C).

by protamines (Fig. 3, center). In the doughnut configu- also condense somewhat to accommodate the condensing
ration, these protamine-bound DNA circles from each arm chromatin (Fig. 3, right). This prediction is consistent with
of the loop are collapsed into a doughnut-shaped nodule the condensation of sperm nuclei during spermatogene-
about 65 nm in diameter (Fig. 3, right). sis.
The model, and the data previously cited, also predict According to the model, this nodule is not a perfect cyl-
concomitant changes in the relationship between the DNA inder, nor evenly and regularly composed. Rather, the model
and the nuclear matrix. First, at some point during sper- predicts that the protamine-complexed DNA circles (Figs.
miogenesis (whether before, after, or during protamine 2B and 3, center) are collapsed in a disorderly fashion, much
replacement is unknown) the number of DNA attachment like the coils of a wound garden hose, forming a three-
sites must increase to decrease the average size of the loop dimensional structure shaped like a doughnut, or torus (Fig.
domain [22, 24]. Second, the sperm nuclear matrix must 3, right). This torus has an outer diameter of 65 nm, an
1198 WARD

FIG. 3. Model of solenoid equivalent in one sperm DNA loop. The model is diagrammed for one loop of DNA that is 47 kbp in length, and the
comparative structures are drawn to scale, viewed from above (top) and from the side (bottom). In the parent cell, the DNA is in the solenoid configuration
(left). As the histones are replaced by the protamines, each turn of the solenoid becomes two concentric circles (center). In the doughnut structure, the
protamine-bound DNA circles are collapsed into a toroid-shaped structure made up of 72 circles of DNA with an average diameter of 65 nm (right). The
schematic intermediate (center) is drawn only as an instructive diagram, and is not predicted as a real, functional intermediate that occurs during sper-
miogenesis. The actual transition must be much more complex, involving transitional proteins that are not considered in this model.

inner diameter of 32.5 nm, and height of 25 nm (Fig. 3,

right). Such a structure has recently been independently
demonstrated experimentally by Hud and Balhorn (per-
FIG. 4. Equivalent levels of DNA packaging in somatic and sperm
sonal communication; Hud NV, Balhorn R, unpublished) chromatin based on the doughnut-loop model. In somatic cells, DNA (A) is
using purified DNA and protamines in vitro. Note that the wound twice around histone octomers into nucleosomes (B), which then
model also predicts that the exact dimensions of the coil into solenoids with six nucleosomes per turn (C). DNA in solenoid form
is attached at intervals of about 60 kb to the nuclear matrix at their bases
doughnut would vary with the size of the DNA loop domain to form DNA loop domains (D). In the somatic cell, these solenoid loop
packaged within it. Sperm DNA loop domains are predicted domains are contained within the nucleus, but when the histones are ex-
to vary in size as do somatic cell loop domains, although tracted, they can be visualized outside the nucleus (E). Active genes are
more closely associated with the somatic nuclear matrix than inactive genes.
this has not been directly addressed. In the sperm nucleus, highly positively charged protamines bind to DNA
lengthwise along the double helix, neutralizing the negative charges of the
DISCUSSION DNA (G). The protamine-bound DNA is coiled with a very slight bend in
the protamine DNA complex into concentric circles (H). These circles of one
With the proposed model for the coiling of the DNA loop loop then collapse into a doughnut (I) in which the neutral DNA protamine
domains in mammalian sperm, it is now possible to con- complexes are tightly packed together by Van der Waal's forces (I, inset).
Each doughnut represents one DNA loop domain attached to the sperm
struct a model for the packaging of the entire haploid ge- nuclear matrix (J). The DNA loop domains of sperm nuclei are smaller than
nome into the sperm nucleus in which sperm DNA loop those of somatic cells.
SPERM DOUGHNUT LOOP DOMAIN MODEL 1199
1200 WARD

domains are packaged as doughnuts bound to the nuclear REFERENCES

matrix (Fig. 4). The doughnut model allows sperm DNA to 1. Watson JD, Crick FHC. Molecular structure of nucleic acids. Nature 1953; 171:737-
be tightly packaged (inset, Fig. 41) compared to the rela- 738.
tively open configuration of somatic cell DNA at the same 2. Getzenberg RH, Pienta KJ, Ward WS, Coffey DS. Nuclear structure and three-
dimensional organization of DNA. J Cell Biochem 1991; 46:1-11.
level of organization (Fig. 4D), and at the same time to be 3. Pienta KJ, Getzenberg RH, Coffey DS. Cell structure and DNA organization. Crit
organized into discrete functional units, the DNA loop do- Rev Euk Gene Express 1991; 1:355-385.
mains (Fig. 4J). 4. Jackson DA. Organization beyond the gene. TIBS 1986; 11:249-253.
5. Mirkovitch J, Gasser SM, Laemmli UK Relation of chromosome structure and
This model is only a first approximation of the data and gene expression. Philos Trans R Soc Lond 1987; 317:563-574.
it will undoubtedly progress through many revisions be- 6. Ward WS, Coffey DS. DNA packaging and organization in mammalian sperma-
fore a consensus, similar to that being formed for histone- tozoa: comparison with somatic cells. Biol Reprod 1990; 44:569-574.
7. Balhorn R. A model for the structure of chromatin in mammalian sperm. J Cell
bound chromatin, is reached. The loop doughnut model Biol 1982; 93:298-305.
for the local packaging of DNA loop domains in sperma- 8. McGhee JD, Felsenfeld F. Nucleosome structure. Annu Rev Biochem 1980; 49:1115-
tozoa still does not explain many unknown aspects of sperm 1156.
9. Chen TA, Smith MM, Le SY, Sternglanz R, Allfrey VG. Nucleosome fractionation
chromatin structure. The next step, for example, is to ask by mercury affinity chromatography. Contrasting distribution of transcriptionally
how the loop doughnuts are arranged within the nucleus. active DNA sequences and acetylated histones in nucleosome fractions of wild-
If sperm chromatin is condensed in a fashion similar to that type yeast cells and cells expressing a histone H3 gene altered to encode a cys-
teine 110 residue. J Biol Chem 1991; 266:6489-6498.
of mitotic chromosomes, it is likely that the loop doughnuts 10. Liu L. DNA topoisomerases-enzymes that catalyze the breaking and rejoining of
are arranged in a circular pattern around a central core [3]. DNA. CRC Crit Rev Biochem 1983; 15:1-24.
11. Liu LF,Wang JC. Supercoiling of the DNA template during transcription. Proc
This possible extension of the loop doughnut model is sim-
Nad Acad Sci 1987; 84:7024-7027.
ilar to a proposal Koehler et al. [33] made for rat sperm 12. Finch JT, Klug A Solenoid model for superstructure in chromatin. Proc Natl Acad
chromatin in 1983. Another aspect of sperm chromatin not Sci USA 1976; 73:1897-1901.
yet fully addressed by this model is the presence of a small 13. Berezney R, Coffey DS. Identification of a nuclear protein matrix. Biochem Bio-
phys Res Commun 1974; 60:1410-1417.
amount of histone proteins in fully condensed sperm nu- 14. Pardoll DM, Vogelstein B, Coffey DS. A fixed site of DNA replication in eucaryotic
clei [36]. It is possible that either the top of the sperm loop cells. Cell 1980; 19:527-536.
remains associated with histones, filling the center of the 15. Cockerill PN, Garrard WT. Chromosomal loop anchorage of the kappa immu-
noglobulin gene occurs next to the enhancer in a region containing topoiso-
toroid loop with nucleosomes. Alternatively, or simulta- merase II sites. Cell 1986; 44:273-282.
neously, the DNA that is associated with the sperm nuclear 16. Vaughn JP, Dijkwel PA, Mullender LHF, Hamlin JL. Replication forks are associ-
matrix may remain in the form of nucleosomes to allow ated with the nuclear matrix. Nucleic Acids Res 1990; 18:1965-1969.
17. Razin SV. DNA interactions with the nuclear matrix and spatial organization of
for greater freedom in bending at this region. replication and transcription. Bioessays 1987; 6:19-23.
Regardless of the unanswered questions, the loop 18. Razin SV, Rzeszowska-Wolny I, Moreau J, Scherrer K. Localization of points of
DNA attachment to the nuclear skeleton inside the domain of chicken a-globin
doughnut model for sperm DNA packaging is consistent with
genes in functionally active and functionally inactive nuclei. Mol Biol 1985; 19:456-
the data available, and it anticipates many possible aspects 466.
of sperm function. It is conceivable, for example, that one 19. Robinson SI, Nelkin BD, Vogelstein B. The ovalbumin gene is associated with
function of the spermatozoon is to transfer the paternal ge- the nuclear matrix of chicken oviduct cells. Cell 1982; 28:99-106.
20. de Lange T, Shiule L, Myers RM, Cox DR, Naylor SL, Killery AM, Varmus HE.
nome to the zygote in a organized but compact form. The Structure and variability of human chromosome ends. Mol Cell Biol 1990; 10:518-
relocation of the proper three-dimensional DNA organiza- 527.
tion may function to ensure that the developing embryo 21. Groudine M, Conkin KF. Chromatin structure and de novo methylation of sperm
DNA: implications for activation of the paternal genome. Science 1985; 228:1061-
can access the paternal genetic information rapidly and ef- 1068.
ficiently. The condensed, rather than open, form of the DNA 22. Ward WS, Partin AW, Coffey DS. DNA loop domains in mammalian spermatozoa.
increases the chances that it will survive the voyage intact. Chromosoma 1989; 98:153-159.
23. Risley MS, Einheber S, Bumcrot DA. Changes in DNA topology during sper-
The doughnut structure fulfills both requirements, keeping matogenesis. Chromosoma 1986; 94:217-227.
DNA sequences in the same focal packets found in somatic 24. Fawcett DW. A comparative view of sperm ultrastructure. Biol Reprod 1970;
2 0 2
(suppl):9 -1 7.
cells, localized DNA loop domains, in a more highly con- 25. Ward WS, Coffey DS. Specific organization of genes in relation to the sperm
densed form than is found in somatic cells. If this loop nuclear matrix. Biochem Biophys Res Commun 1990; 173:20-25.
doughnut model for sperm DNA structure can be more fully 26. Kalandadze AG, Bushara SA, Vassetzky YS Jr, Razin SV. Characterization of DNA
pattern in the site of permanent attachment to the nuclear matrix in the vicinity
verified, this aspect of sperm DNA tertiary structure will lead of replication origin. Biochem Biophys Res Commun 1990; 168:9-15.
us to a more complete understanding of sperm DNA func- 27. Martin RH. A detailed method for obtaining preparations of humans sperm chro-
mosomes. Cytogenet Cell Genet 1989; 35:252-256.
tion. 28. Haaf T, Grunenberg H, Schmid M. Paried arrangement of nonhomologous cen-
tromeres during vertebrate spermiogenesis. Exp Cell Res 1990; 187:157-161.
29. Moens PB, Pearlman RE. Satellite DNA I in chromatin loops of rat pachytene
ACKNOWLEDGMENTS chromosomes and in spermatids. Chromosoma 1989; 98;287-294.
30. Guttenbach M, Schmid M. Determination of Y chromosome aneuploidy in hu-
I would like to thank Dr. Rod Balhorn and Dr. Nick Hud of the Lawrence Liv- man sperm nuclei by nonradioactive in situ hybridization. AmJ Hum Genet 1990;
ermore Laboratory for useful discussions and for sharing their recently submitted 466:553-558.
data. Dr. Hud's experimental results were obtained independently during the writing 31. Wyrobek AJ, Alhborn T, Balhorn R, Stanker L, Pinkel D. Fluorescence in situ
of this manuscript. I would also like to thank Mr. Michael Cooper for drawing all hybridization to Y chromosomes in decondensed human sperm nuclei. Mol Re-
the figures. prod Dev 1990; 27:200-208.
 SPERM DOUGHNUT LOOP DOMAIN MODEL 1201

32. Allen MJ, Pogany GC, Siekhaus W, Balhorn R. Atomic force microscopy studies 35. Morse-Gaudio M, Risley MS. Topoisomerase expression during spermatogenesis.
of mammalian sperm chromatin. J Cell Biol 1990; 111:110 (abstract 613). J Cell Biol 1991; 115:50 (abstract 286).
33. Koehler JK, Wurschmidt U, Larsen MP. Nuclear and chromatin structure in rat 36. Tanphaichitr N, Sobhon P, Taluppeth N, Chalermisarachai P. Basic nuclear pro-
spermatozoa. Gamete Res 1983; 8:357-370. teins in testicular cells and ejaculated spermatozoa in man. Exp Cell Res 1978;
34. Allen MJ, Hud NV, Lee C, Pogany G, Siekhaus WJ, Balhorn R. Analysis of sperm 117:347-356.
nuclear volumes and extent of chromatin compaction by atomic force micros-
copy. J Cell Biol 1991; 115:50 (abstract 289).