| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||
Review |
From the Animal Reproduction and Biotechnology Laboratory, Colorado State University, Fort Collins, Colorado.
| Correspondence to: Rupert P. Amann, 909 Centre Ave, No. 123, Fort Collins, CO 80526 (e-mail: rpalra62{at}comcast.net). |
| Received for publication December 1, 2007; accepted for publication May 18, 2008. |
Understanding the dynamics of spermatogenesis is central to clinical
andrology or to probing environmental effects on human testes. This review
considers what is known about renewal and proliferation of spermatogonia, how
germ cells are organized in cellular associations constituting the cycle of
the seminiferous epithelium, relative frequencies of cellular associations,
durations of the cycle of the seminiferous epithelium and spermatogenesis, and
measurement of daily sperm production. Daily sperm production (DSP) per testis
tends to decline with advancing age. Regardless of age, there is substantial
loss of potential sperm from degeneration of spermatocytes, but not
spermatids. DSP per gram testis parenchyma or DSP per testis cannot be
predicted on the basis of testis size or age of a man. The review shows why
our 1960s data base is neither robust nor precise and suggests how
deficiencies might be rectified. New cellular associations should be defined,
with none representing >15% of the cycle of the seminiferous epithelium.
Then determine when Apale-spermatogonia become committed to
proliferate or how many mitotic divisions occur thereafter. Restudy the
duration of spermatogenesis because the accepted value might be in error by
6 days. Restudying human spermatogenesis will benefit clinicians,
toxicologists, and epidemiologists probing testis function by direct
evaluations or indirectly via evaluations of quantity and quality of sperm
ejaculated. It also will benefit scientists interested in renewal and
proliferation of spermatogonia, or a spermatogonium as a prototype stem
cell.
From the perspective of a clinician or subfertile patient couple, current descriptions of cellular associations, cycle of the seminiferous epithelium, and duration of spermatogenesis might be adequate. However, from the perspective of a scientist interested in renewal and proliferation or differentiation of spermatogonia or of a spermatogonium as a prototype stem cell, current information is imprecise. Perhaps most important, epidemiologists increasingly are probing possible linkages between fetal or adult exposure to a putative or known chemical and aberrations in testis function and seminal characteristics. Such endeavors require robust information on the kinetics and biology of spermatogenesis. Hence, it is appropriate to ask whether current knowledge is adequate.
This review is restricted to spermatogenesis in "normal adults." Prenatal differentiation of primordial germ cells to stem cells and their normal progressive development into gonocytes, undifferentiated A-spermatogonia, and, finally, differentiated A-spermatogonia is a separate subject. However, both normal and aberrant transformation of primordial germ cells can occur. Clark (2007) provides an excellent introduction to the link between germ cells and testicular cancer. Rajpert-De Meyts (2006), Olesen et al (2007), and Veeramachaneni (2008) summarize the linkage among environmental pollutants, changes in spermatogenesis, and testicular cancer.
With this paper, I have 4 goals: 1) provide a brief overview of spermatogenesis, 2) review what is known about the cycle of the seminiferous epithelium, 3) review measurement of daily sperm production and data on quantitative features of spermatogenesis, and 4) discuss what information is needed and why. This required a careful review of primary papers and the consideration of deficiencies in certain papers.
Overview of Spermatogenesis
Spermatogenesis occurs in the seminiferous epithelium, which consists of
Sertoli cells (somatic cells) and several types of germ cells
(Roosen-Runge, 1952; Clermont,
1963,
1966a,b,
1972;
Heller and Clermont, 1964;
Courot et al, 1970;
de Kretser and Kerr, 1994).
Roosen-Runge (1962) reviews
the derivation of terms used to describe spermatogenesis. In humans, the germ
cells are best termed (Roosen-Runge and
Barlow, 1953; Clermont,
1972; Aponte et al,
2005; Ehmcke and Schlatt,
2006; Ehmcke et al,
2006): progenitor Adark-spermatogonia; progenitor
Apale-spermatogonia; committed Apale-spermatogonia;
B-spermatogonia; preleptotene, leptotene, zygotene, pachytene, and diplotene
spermatocytes; secondary spermatocytes; spermatids; and spermatozoa.
(Sometimes preleptotene spermatocytes were termed "young primary
spermatocytes.") The progenitor Adark-spermatogonia
apparently are stem or reserve spermatogonia on the basis of location and
infrequent division, but with an unknown stimulus, they can produce progenitor
Apale-spermatogonia. Progenitor and committed
Apale-spermatogonia look alike, but the latter divide to produce
B-spermatogonia.
Spermatogenesis is the totality of 4 sequential processes (McLachlan et al, 2002; Schlegel and Hardy, 2002; Yoshida et al, 2007):
Regardless of species, there are overarching features of spermatogenesis:
74
days vs 16 days in humans), cohorts of germ cells are layered, with the
youngest near the basement membrane. These "layers" sometimes are
termed generations. Intermingling of cohorts of germ cells is programmed,
because Sertoli cells systematically move maturing spermatids toward the basal
area and, several days later, back toward the tubule lumen. 16 days), appearance of
the cellular association at a given point in a tubule is changing
repetitively, much like frames in a movie with a loop format. Spermatogenesis in humans is different from the process in bulls, mice, rabbits, rats, stallions, or other common mammals. Obvious differences include the 3-dimensional (3D) organization of the seminiferous epithelium and low number of sperm produced daily per gram of testis. The pattern of spermatogonial renewal and proliferation also is different. The difference in sperm production per gram of testis parenchyma is important to clinicians and clinical epidemiologists. Variation among human testes is much greater than that among individual mice, rabbits, or rats used by most experimentalists (Berndtson, 2008). Readers should consult excellent reviews on human spermatogenesis, or comparative aspects of spermatogenesis, such as Ortavant (1959), Clermont (1963, 1966b, 1972), Heller and Clermont (1964), Courot et al (1970), Holstein and Roosen-Runge (1981), Johnson et al (1992, 2000), de Kretser and Kerr (1994), de Rooij and Russell (2000), Schlegel and Hardy (2002), and Ehmcke and Schlatt (2006).
|
|
What Is Known About the Cycle of the Seminiferous Epithelium?
Germ cells develop in consistent groupings termed cellular
associations. In humans, 6 cellular associations are used
(Figure 1), designated with
Roman numerals. Two to 4 cellular associations usually are evident in an
essentially round cross section through a human seminiferous tubule
(Roosen-Runge, 1952;
Clermont, 1963). Cellular
associations can be defined in several ways, but inevitably, this requires
recognition of specific cell types. However, appearance of germ cells is
dependent on methods of fixation, embedment, staining, and visualization.
Therefore, I include procedures with descriptions of what is seen. Newer
methods reveal distinguishing features better than those used 40 years
ago.
Descriptions of germ cells in stained slides are in general agreement (Roosen-Runge and Barlow, 1953; Roosen-Runge, 1962; Clermont, 1963, 1966a,b; Heller and Clermont, 1964; Rowley and Heller, 1971; Holstein and Roosen-Runge, 1981; Hocherau de Reviers et al, 1984). However, the pattern for proliferation of spermatogonia is uncertain, and there are multiple schemes for classification of human spermatids. Because these elements are used to define cellular associations, they are considered first.
Pattern of Spermatogonial Proliferation— Clermont (1966a) described Adark-, Apale-, and B-spermatogonia in sections from biopsies fixed in Zenker-formal and stained with hematoxylin and eosin (H&E) but commented (Clermont 1966b) that some did not look like either A- or B-spermatogonia. He mapped and counted spermatogonial nuclei and "mitotic plates" in different cellular associations (3–5 biopsies, depending on cell type). There was approximately 1 Apale-spermatogonium per Adark-spermatogonium, and Apale-spermatogonia usually were in pairs located near the basement membrane (Clermont 1966a). Calculated ratios of Apale-spermatogonia:B-spermatogonia:preleptotene spermatocytes were 1:2:4 (Clermont 1966b).
The pattern for division of Apale-spermatogonia is uncertain. Apparently, some Apale-spermatogonia proliferate to produce more Apale-spermatogonia, whereas others produce B-spermatogonia. Under the model favored by Clermont (1966b), and most frequently cited, each member of a pair of committed Apale-spermatogonia undergoes only 1 division, to provide 2 B-spermatogonia (Clermont, 1967, 1972; Ehmcke et al, 2006). An alternative model was proposed and favored by Ehmcke and Schlatt (2006); 1 member of a pair of committed Apale-spermatogonia divides to provide a new pair of committed Apale-spermatogonia, whereas the other member of the original pair undergoes a division to provide 2 daughter Apale-spermatogonia (? in Figure 1) which, in turn, divide to produce B-spermatogonia. With either model, 8 preleptotene spermatocytes are produced per original pair of Apale-spermatogonia.
When B-spermatogonia are formed and divide is also uncertain. Clermont (1963, 1966b) concluded that they were formed at the transition between cellular associations VI and I, were obvious in cellular associations I and II, and divided only when forming preleptotene spermatocytes in late association II or early association III. This is depicted in Figure 1 herein. However, Clermont (1966a,b) was inconsistent because in figures 3 and 5 of these papers, B-spermatogonia were depicted only in cellular associations I and II and occasionally early III, whereas figure 10 in both papers clearly shows B-spermatogonia throughout cellular association VI and none in association III. Heller et al (1969) and Rowley and Heller (1971) also concluded that B-spermatogonia were formed sometime during cellular association V and remained through cellular association II.
|
|
Although modeling spermatogonial proliferation was not their goal, Johnson
et al (1987) estimated numbers
of germ cells in 30 testes (testes fixed by perfusion with glutaraldehyde
<15 hours after the sudden death of men in good health; toluidine
blue–stained Epon sections) from measures of volume density and nuclear
diameter. On the basis of Clermont
(1966b), Johnson et al
(1987) assumed that
B-spermatogonia were present throughout cellular associations VI, I, and II
and divided only once to produce 2 preleptotene spermatocytes. They found
11 preleptotene/leptotene spermatocytes per B-spermatogonium, rather than
the expected 2:1 ratio. Johnson et al
(1987) "eliminated the
problem" by rationalization considered unsatisfactory by this
reviewer.
Classification of Spermatids— Pioneering periodic acid–Schiff (PAS; Orths fixative was best) staining of testis tissue, Clermont and Leblond (1955) described how human spermatids develop through a series of 12 steps (Figure 2, column A). The first 7 steps are characterized by subtle changes in acrosome morphology, in that the nucleus remains spherical. During steps 8–12, the nucleus gradually changes shape and shrinks as nucleoprotein condenses; changes in the acrosome continue. However, Clermont (1963) abandoned PAS stain for human testes in favor of Zenker-formal fixation and H&E staining. Zenker-formal preserved details of nuclei in spermatogonia and spermatocytes better than Orths fixative, but gave inconsistent preservation of early acrosomal structures revealed by PAS stain.
Use of H&E-stained sections (Zenker-formal fixation) forced a different scheme to describe changes in spermatids because the acrosome provided little information. On the basis primarily of nuclear characteristics (shape, size, intensity of staining), Clermont (1963) described 4 types of spermatids (Sa, Sb, Sc, and Sd) in H&E sections. Subsequently, this was expanded to 6 types (Sa, Sb1, Sb2, Sc, Sd1, and Sd2) via better distinction of nuclear staining/elongation and sequestration of cytoplasm (Figure 2, column B; Heller and Clermont, 1964; Clemont, 1966b) or even 7 types (distinguishing Sa1 and Sa2 spermatids; Heller et al, 1969). The 6-type classification of Heller and Clermont (1964) is most commonly used.
More detail within spermatids can be resolved when plastic sections are viewed by a light or electron microscope (Figure 2, columns C–E). With light microscopy, bright-field, phase contrast, or differential interference contrast optics each has advantages and limitations (Johnson et al, 1981, 1992; Johnson 1994). With an electron microscope, de Kretser (1969) was able to link changes in fine structure of spermatids with Clermont's (1966b) descriptions for spermatids in H&E-stained sections; 6 steps were described (column C in Figure 2). Holstein (1976) defined and illustrated 8 steps in spermatid development. These illustrations were retained, and finer discrimination of spermatids with spherical nuclei was added, in the superbly illustrated Holstein and Roosen-Runge (1981; some are reproduced in column D in Figure 2).
Examination of Figure 2 reveals that the steps in spermatid development described on the basis of PAS or H&E staining (columns A and B, respectively) are not uniformly distributed over the 24 days (left y axis) required for spermiogenesis. Rather, the distinguishing changes in the acrosomic system are clustered on days 6–10 (column A, steps 4–8). Obvious nuclear changes are clustered on days 9–16 and around day 19. Both de Kretser (1969) and Johnson et al (1992) intentionally fit their observations (columns C and E) to the 6-step classification of Heller and Clermont (1964). In contrast, Holstein (1976) and Holstein and Roosen-Runge (1981) developed their own classification scheme for spermatids and described 8 major steps plus 5 subtle substeps (column D). Looking ahead, use of glutaraldehyde fixation, plastic sections, and toluidine blue stain (column E) could enable discernment of subtle changes in both nuclear and acrosome development with a light microscope, not detectable with H&E staining of paraffin sections (column B).
Cellular Associations— A classification scheme of cellular associations is essential to understanding spermatogenesis because it allows description and quantitation of the seminiferous epithelium and reveals the cycle of the seminiferous epithelium. Any scheme is arbitrary (Roosen-Runge, 1962). Descriptions and illustrations for cellular associations in a human testis presented by Clermont (1963) are widely cited, although enhancement via consideration of 6 rather than 4 types of spermatids (Heller and Clermont, 1964) often is overlooked.
Defining descriptions of cellular associations were based on H&E-stained paraffin sections prepared from Zenker-formal fixed tissue (Clermont 1963; Heller and Clermont, 1964). Spermatids were described from nuclear characteristics as Sa, Sb1, Sb2, Sc, Sd1, and Sd2. A key to discriminating cellular associations is that cellular associations I and II contain 2 generations of spermatids (Sa and Sd1 or Sd2) whereas cellular associations III–V contain 2 generations of primary spermatocytes (preleptotene, leptotene, or zygotene plus pachytene or diplotene; Figure 1). Chromosomal characteristics during meiosis of spermatocytes, location of spermatids, and shape of spermatid nuclei also help. Cellular association VI is defined by presence of metaphase plates of the first or second meiotic divisions, secondary spermatocytes, or both. Examination of Figure 2 using the right y axis provides understanding of the uniformity of spermatids within a cellular association (association I) or variety of subtly different forms (associations III and IV).
The start of cellular association I is evidenced by newly formed spermatids, each with a spherical nucleus. Presence of B-spermatogonia also adds discrimination for associations I and II, although sometimes they are not seen in 1-µm plastic sections (Johnson et al, 1987). Adark- and Apale-spermatogonia usually are present in all cellular associations (Clermont, 1963).
As in all aspects of biology, there are "grey areas," wherein some groups of cells are transitioning to fit the next description. Furthermore, in human testes, there are atypical cellular associations because of missing spermatogonia, spermatocytes, or spermatids, and also from intermingling of associations (Roosen-Runge, 1952; Clermont, 1963; de Kretser, 1969; Johnson et al, 1987). It is likely that the "chaotic nature" of the human seminiferous epithelium has been overemphasized, in that <0.1% of cellular associations are atypical in men with high sperm production (Schulze, 1982; Chaturvedi and Johnson, 1993). However, men with low daily sperm production per gram often are missing a generation or generations of germ cells (Johnson et al, 1992).
With the use of perfusion fixation and staining of plastic sections with toluidine blue, it is likely that 10–12 cellular associations could be discerned routinely in human seminiferous tubules. This is evidenced in Johnson et al (1992:figure 1; note that they allocated uniform intervals to the 6 cellular associations), although they followed Clermont's classification. Also phase contrast or Nomarski differential interference contrast optics would allow resolution of subtle differences in cell morphology not detectable with the bright-field optics available around 1960. Images of each of Clermont's 6 cellular associations, as viewed by Nomarski optics, are in Johnson (1994). Descriptions of the 6 cellular associations by Johnson et al (1992, 2001) are very helpful, but no attempt was made to expand and redefine the classification scheme described by Heller and Clermont (1964).
Relative Duration of Cellular Associations— The relative duration of each of the 6 cellular associations (x axis in Figure 1, right y axis in Figure 2) is calculated from frequency of occurrence of a given cellular association in cross sections through seminiferous tubules. Because several cellular associations are found in a cross section of human seminiferous tubule, Clermont (1963) traced limits of each cellular association within a cross section and cut out the tracings. Total weight of tracings for a given cellular association was expressed relative to total weight for all tracings for that testis.
Relative durations of the 6 cellular associations were 30%, 20%, 6%, 8%, 31%, and 5%, respectively (Clermont, 1963). However, these means were based on only 33–87 tubule cross sections for 1 biopsy from each of 4 individuals. With Clermont's (1963) data, it was calculated that 95% confidence intervals (CIs) for the relative frequencies of cellular associations I–VI encompassed 21%–38%, 13%–26%, 0.5%–12%, 3%–12%, 22%–40%, and 3%–7%, respectively. Data on relative frequency of cellular association reported by Heller et al (1969) or Rowley and Heller (1971) usually are overlooked. Heller et al (1969) examined testis biopsies from 44 healthy men (normal semen and hormonal profiles). They identified and counted germ cells in 30 randomly selected cellular associations per biopsy, and calculated relative occurrence of each cellular association. That data set was enlarged to 100 biopsies in Rowley and Heller (1971), who reported mean frequencies of 23%, 17%, 17%, 13%, 23%, and 7% for cellular associations I–VI, respectively. These values seem different from Clermont's (1963) values of 30%, 20%, 6%, 8%, 31%, and 5%, respectively, but except for associations III and IV, they fall just within the 95% CIs I calculated for Clermont's values. These papers do not mention checking for within-sample repeatability of observations. For these reasons, the relative durations of cellular associations in the cycle of the human seminiferous epithelium should be considered as imprecise values.
Spatial Organization of Cellular Associations— The 3D space occupied by a given cellular association is small. Hence, each cross section viewed microscopically contains several cellular associations (Figure 3; Clermont, 1963; Heller and Clermont, 1964; Kerr and de Kretser, 1981; Johnson, 1994; Johnson et al, 1996). This might be because pairs or clusters of progenitor spermatogonia occupy a smaller area along the basement membrane than in rodents or domesticated animals and might provide committed A-spermatogonia less synchronously.
Historically, organization of cellular associations along the length of a human seminiferous tubule was considered as without pattern (Roosen-Runge, 1952; Clermont, 1963), and clearly different from the sequential juxtaposition of cellular associations in substantial 3D space found in common mammals (eg, rat; de Kretser and Kerr, 1994). However, Schulz (1982) photographed sections of plastic-embedded human testes stained with toluidine blue and made 2D reconstructions (19 tubules in 3 testes; 160–320-µm lengths). These micrographs were used to plot Cartesian coordinates for primary spermatocytes (Schulze and Rehder, 1984), relative to the geometric center of a tubule segment, and apply computer modeling (Schulze et al, 1986). They concluded that at a given point in time, development of spermatocytes was progressively more advanced, in a spiral with decreasing radius, down the length of a tubule and that spermatids displayed a similar pattern in a separate spiral. In their opinion, typically there was an orderly sequence of cellular associations, I–VI (occasionally with a reversal of order), with an oblique orientation. This was depicted (Schulze and Rehder, 1984) as intertwined helical "bands" of sequentially organized cellular associations.
This concept was challenged by Johnson
(1994), who made 3D maps of
cellular associations, rather than primary spermatocytes, on the basis of
successive 22-µm plastic sections (7 tubules from 6 men;
332–597-µm lengths). He did not find a complete sequence of 6
contiguous cellular associations along the length of a tubule, although 2 or 3
consecutive abutting cellular associations were found. Further study
(Johnson et al, 1996; 8
tubules from 6 men, possibly same testes as earlier paper) involved plotting
limits and geometric centers of each cellular association, computerized
reconstructions, and hypothesis testing. Actual sequences, random number
sequences (same total cellular associations as for actual sequences), and
ideal sequences (I–VI, repeatedly) were compared with the use of 2
paradigms. For actual sequences, 2–3 consecutive cellular associations
represented 16% of all cellular associations, but 4–6 consecutively
represented only 8% of all cellular associations; the majority of
abutting cellular associations were nonsequential. Importantly, incidences of
2–3 or 4–6 sequential cellular associations were not significantly
different for actual sequences and random number sequences. Johnson et al
(1996) discussed in detail why
their results might have differed from those of Schulze and Rehder
(1984). Johnson et al
(1996) concluded that the
arrangement of cellular associations in a human seminiferous tubules at any
point in time reflected random occurrences, presumably around puberty. Once
initiated at a site, the commitment process apparently is
"remembered" and replicated every 16 days at that site (or
driven by an unknown signal).
Duration of Spermatogenesis—
A time frame for human spermatogenesis was established by intratesticular
injection of 3H-thymidine to label DNA during synthesis, last in
preleptotene spermatocytes. Progression of those cells or labeled daughter
cells was studied in autoradiograms of specially prepared sections from 11
biopsies taken 4–46 days later (Heller and Clermont,
1963,
1964).
Figure 4 illustrates the
biology underlying this study, and the relationship between 1 cycle of the
seminiferous epithelium and the total process of spermatogenesis (requires
approximately 4.6 cycles). Heller and Clermont
(1963,
1964) concluded that 1 cycle
of the seminiferous epithelium required 16 days and the duration of
spermatogenesis was 74 days. This classic study has not been replicated.
Use of a nonradioactive tracer (eg, Misell
et al, 2006) does not eliminate the need for biopsies.
|
5 days. Hence, during
a given cellular association, multiple events occur within each of the
4–5 different types of germ cells as each evolves, or even divides,
between the beginning and end of a cellular association. Heller and Clermont (1964) reported that neither norethandrolone (depressing spermatogenesis) nor human chorionic gonadotropin affected the time frame for development of germ cells (1 individual per treatment). Animal data summarized by Courot et al (1970) support the contention that altered hormonal status, imposed or natural in seasonal breeders, does not affect the duration of the cycle of the seminiferous epithelium. With rabbits, imposition of planned ejaculation or sexual rest near the time of 3H-thymidine injection did not affect the interval until mature spermatids underwent spermiation (Amann, 1972), but emission/ejaculation hasten transit of sperm through the epididymis. However, there is no study in which individual males were ejaculated frequently or sexually rested for >3 mo and then given 3H-thymidine to allow tracking development of germ cells while the same ejaculation frequency continued. Nevertheless, the conclusion of Ortavant (1959:43) "when spermatogenesis is disturbed, a certain number of [germ] cells degenerate, but those that continue their development until the end [of spermatogenesis] do so with the same speed" remains generally correct.
In humans, the duration of spermatogenesis probably does not change. Testis descent is completed years before initial commitment of Apale-spermatogonia to form B-spermatogonia. This is in contrast to hamsters, mice, and rats. In these animals, spermatogenesis is initiated shortly after birth and before testes are positioned in the scrotum; the interval from initial commitment of spermatogonia through formation of preleptotene spermatocytes is accelerated for the first 10–12 days after birth, and the process then slows to adult timing (van Haaster and de Rooij, 1993). The temporary acceleration of spermatogenesis might be due to elevated temperature in nonscrotal testes.
Measurement of Daily Sperm Production
Overview—
Study of success in spermatogenesis demands measurement of 2 general
features: quantity of sperm produced and quality of each spermatozoon in a
subsample representative of the total population recently produced.
Qualitative success of spermatogenesis is largely independent of number of
sperm produced. Quality of many sperm can be reduced without a detectable
change in number produced (Veeramachaneni et al,
2001,
2007,
2008). Importantly, quality of
ejaculated sperm reflects extra-testicular events as well as
spermatogenesis.
Daily sperm production (DSP) is the number of sperm produced per day by a testis or the 2 testes of an individual (Amann, 1970b). DSP is the quantitative measure for success in spermatogenesis. Number of sperm produced daily per gram of testicular parenchyma (DSP/g), reflects the efficiency of sperm production and is useful in certain comparisons. Prolonged or severe reduction, or both, or failure, in production of committed Apale-spermatogonia is one cause of reduced DSP. The other is an unusually high rate of germ cell death by apoptosis, or other causes, at specific points in spermatogenesis. Either or both problems could detectably reduce DSP per gram testis and overall DSP. With extensive loss of germ cells, both DSP per gram testis and testis weight might diminish. However, even complete elimination of germ cells in 1 tubule (ie, Sertoli cell–only tubule) might not noticeably reduce DSP per gram or DSP per testis. There is no evidence that sexual arousal or ejaculation frequency changes testis weight or DSP per gram on the basis of extensive data for bulls, rabbits, and rams (Amann, 1962, 1970b, 1981; Amann et al, 1974; Almquist, 1982).
Understanding how DSP might be estimated via noninvasive methods in an epidemiologic study is helped by knowing how DSP is measured. DSP can be estimated on the basis of daily sperm output (DSO), provided certain conditions are met (Amann, 1970b, 1981; Amann and Chapman, in preparation). DSP can be measured by morphometric analysis of fixed tissues (Amann and Almquist, 1962; Kennelly and Foote, 1964; Swierstra, 1966) or by enumeration of specific germ cells in homogenates of testicular parenchyma (Ortavant, 1958; Amann and Almquist, 1961).
A "time divisor" is necessary to convert total number of cells of a certain type present at sampling to number produced per day. The time divisor is calculated from the relative durations (%) of each cellular association, the duration (days) of 1 cycle of the seminiferous epithelium, and knowledge of the cellular associations in which the cells counted are found. For several reasons, uncertainty in the time divisor is greater for humans than for bulls, rabbits, rats, etc. A separate section is devoted to appropriate time divisors for humans.
Morphometric Approach— The combination of fixative, embedment, and staining can facilitate identification of germ cells or classification of cellular associations. Glutaraldehyde, osmium, Epon, and toluidine blue (Johnson et al, 1981) or Bouins, methacrylate, PAS, and hematoxylin (Zhengwei et al, 1998) are good, but formalin followed by paraffin should be avoided. Perfusion via the testicular artery is preferable to immersion fixation, because fewer artifacts are introduced and shrinkage of tissue is minimal.
Morphometric approaches to measure DSP are based on 1 or more small samples
of fixed tissue. They are very tedious but allow separate tabulations of
B-spermatogonia and several types of primary spermatocytes, several types of
spermatids, or both, each within its cellular association. DSP can be
calculated from the theoretical yield of sperm from each of several immature
germ cell types and compared with the yield of mature spermatids. Principles
of analysis and sources of error were detailed in Amann
(1970a,b).
Johnson et al (1980a,
1981,
1983) applied this approach to
quantification of human spermatogenesis. Sampling was systematic and results
likely differ little from what might have been obtained had a full dissector
method (Gundersen and Jensen,
1985) been used. Zhengwei et al
(1998) applied a dissector
approach to stereologic analysis of human testis biopsies and estimated
numbers of B-spermatogonia, pachytene spermatocytes, and Sa or
Sd spermatids per Sertoli cell nucleus (they cautioned that
extrapolations to testis totals would be suspect). Johnson et al
(1984c) and Zhengwei et al
(1998) both reported 2.5
Sa-spermatids per Sertoli cell, whereas Rowley and Heller
(1971) reported 1.6.
Shrinkage of tissue during processing can profoundly affect the calculated
number of germ cells per testis and, hence, the value for DSP
(Amann, 1970b). The extent of
shrinkage depends on size of the tissue piece, fixative, and embedment
process. With immersion-fixed tissue embedded in paraffin, final volume ranges
between 0.90 and 0.47 of the fresh value (Amann,
1970a,b;
Zhengwei et al, 1990).
Shrinkage should be measured for each tissue piece. With perfusion-fixed
tissue embedded in plastic, shrinkage might be negligible.
Johnson et al (1981, 1983) concluded that a correction for tissue shrinkage was not necessary for glutaraldehyde, perfusion-fixed human testis tissue embedded in plastic. They found that means for diameter of spherical spermatid nuclei (data for 10 testes), seminiferous tubule diameter (data for 3 testes), and proportion of interstitial tissue were not significantly different in unfixed and plastic-embedded tissues. These measures were made with better than 2% precision.
The issue of tissue shrinkage might be avoided if desired contrasts can be
obtained as a ratio of number of germ cells of a stipulated type (eg,
Sa-spermatid nucleus) per Sertoli cell nucleolus (conspicuous and
only 1 per nucleus). This approach does not provide DSP per gram testis or DSP
per testis but might detect whether a suspected or imposed agent altered
efficiency of sperm production. Indeed, number of spermatids per Sertoli cell
has greater sensitivity or power than measurement of DSP per gram
(Berndtson, 2008). Rowley and
Heller (1971) found that, for
35 normal males, the among-individual coefficient of variation (CV) for mean
number of Sertoli cells per essentially round tubule cross section was 18%,
but the CV for Sa-spermatids per Sertoli cell was 35%.
Variation in number of Sa-spermatids per Sertoli cell is greater
for 50–85-year-old men than for 20–48-year-old men (CVs of 48% and
27%, respectively; calculated from Johnson
et al, 1984c).
Homogenization Approach— Spermatids become extremely resistant to physical destruction (homogenization of unfixed tissue) at a certain point in spermiogenesis (Ortavant, 1958), although condensation of nuclear material is more variable and less complete in human spermatids than in those from other species (Johnson et al, 1980a). Regardless, homogenization-resistant spermatids can be enumerated in an appropriate suspension of unfixed testis with the use of phase contrast microscopy (Amann and Almquist, 1961; Amann and Lambiase, 1969; Amann, 1970b; Amann and Howards, 1980; Johnson et al, 1980a). If glutaraldehyde-fixed tissue is used, nuclei of most testicular cells resist homogenization and can be identified and enumerated with the use of Nomarski optics in the microscope (Johnson et al, 1981). Comparison data suggest that nuclei of few if any germ cells of a given type, or Sertoli and Leydig cells, are destroyed; they can be identified and enumerated (Johnson et al, 1981, 1998).
Use of fixed tissue should be the first choice because more cell types can be enumerated in homogenates of fixed tissue than unfixed tissue. Also, it is likely that immersion of small pieces of tissue (eg, biopsy) in glutaraldehyde would be appropriate to provide samples for homogenization measurement of DSP and also to process for morphometric study or subjective histopathologic examination.
With unfixed tissue, a pair of crucial questions arise: When during spermiogenesis do spermatid nuclei become resistant to the homogenization procedure actually used (buffer, specific homogenizer container and blade configuration, rpm, and volume processed)? Is this a "square-wave" transition from nonresistant immature forms to youngest "resistant form" plus all the more mature cells? With fixed tissue, one might wonder whether 100% of the nuclei of interest survived and were detected in the homogenate. In either case, the lifespan of the germ cells counted must be estimated accurately.
Time Divisor— Total number of spermatids (or other cell type) enumerated in any of these approaches to measuring DSP represents several days' production of sperm. A "time divisor" is needed to convert number of enumerated cells per unit volume, or mass, to number of these cells produced each day (details in Amann, 1970b; Amann and Howards, 1980; Johnson et al, 1980a). It is assumed all cells of interest develop on an identical time frame. This is unlikely. Also, values for the relative and absolute durations of each cellular association and when cells are formed are imprecise. Documenting correctness of any time divisor is difficult, and neither approach described next has been used with humans.
One approach is a labeling study. Rabbit testes were taken at known intervals after injection of 3H-thymidine; homogenization-resistant spermatids were counted; and autoradiograms of testis sections, testis homogenates, or both were evaluated for labeled cells (Orgebin-Crist, 1968; Amann and Lambiase, 1969). Both concluded that spermatids in cellular associations V–VIII were resistant. However, reported time divisors differed by >50% because of laboratory differences in values for the relative frequency of these cellular associations (49% vs 37%) and measured duration of 1 cycle of the seminiferous epithelium (11.3 vs 10.4 days).
The other approach is seeking congruence between calculated DSP and actual number of sperm leaving a testis. For bulls, initially it was assumed that homogenization-resistant spermatids represented 3.27 days' production (Amann and Almquist, 1962). Subsequently, unilateral placement of an indwelling catheter into the rete testis, proximal ductus deferens, or both, with quantitative collection and enumeration of sperm passing through the catheter over a number of days, enabled comparisons with presurgical sperm output in semen collected daily (divided by 2 to represent 1 testis) and DSP measured using the homogenization procedure (Amann et al, 1974). On a within-bull basis (n = 10), numbers of sperm obtained daily by the first 3 methods did not differ significantly. In a separate comparison (n = 13), numbers of sperm obtained via a ductus deferens catheter or ejaculated semen did not differ from DSP, provided the time divisor was revised upward from 3.27 to 5.32 days. Hence, it was concluded that spermatids in cellular associations IV and V, as well as VI–VIII, were resistant to homogenization of unfixed tissue. (Cellular associations are numbered differently in studies by Amann than in Figures 1, 2, 3, 4 herein [see Hochereau et al, 1984:figure 2.2].) Amann et al (1974, 1976) suggested comparable revisions in time divisors for homogenization-resistant spermatids in unfixed tissue from other species. Revised values were: boar, 6.19; bull, 5.32; rhesus monkey, 4.37; rabbit, 5.35; ram, 4.99; Wistar rat, 6.10; and stallion, 6.00 days.
For unfixed human testes, Amann and Howards (1980) considered that Sd1 and Sd2 spermatids (cellular associations I and II; see Figure 2, column B) were resistant to homogenization, and a time divisor of 7.9 days was used. In retrospect, had they studied de Kretser (1969), they might have excluded most or all Sd1 spermatids. Indeed, Johnson et al (1980a) concluded that nuclei of Sd1 spermatids were not resistant to homogenization. They also concluded that values of DSP per gram, measured by counting spermatids in homogenates of fixed tissue, were similar to those obtained with morphometric procedures (putative "gold standard"). Johnson et al (1981) then counted homogenization-resistant spermatids in 10 unfixed testes and separately counted Sa + Sb, Sc, and Sd1 + Sd2 spermatids in homogenates of 10 contralateral fixed testes. To bring congruence in values of DSP per gram, they recommended a time divisor of 2.9 days for homogenization-resistant spermatids in unfixed human testes processed by their procedure.
Collectively, these reports support a conclusion that the
"best" time divisor for calculation of DSP from spermatids
enumerated in homogenates (using Johnson's procedure) of unfixed human testis
tissue would be 2.9 days, on the assumption that 0% of all Sd1
spermatids and 90% of all Sd2 spermatids remain to be counted.
For fixed testes, the best approach is to collectively count Sd1 +
Sd2 spermatids and use a time divisor of 7.9 days
(Johnson et al, 1984a). If
Sa + Sb spermatids are enumerated in homogenates of
fixed human testis, a time divisor of 8.9 days is appropriate
(Johnson et al, 1981).
Values for DSP in Humans—
A man's testicular size, testis volume, or age do not allow meaningful
prediction of his DSP or DSP per gram of testis parenchyma. Testis weight or
volume accounts for 14% of the variation in DSP per testis, with effects
of age removed (Johnson et al,
1984b). Among individuals (n = 132), the CVs for DSP per gram and
DSP per testis were >50%. Hence, estimated testis volume should not be used
to predict DSP and has little utility when allocating individuals to study
groups.
When men in a study of DSP are grouped by age, the mean or modal value for
the younger group usually is significantly greater than that for the older
group (Figure 5A;
Johnson et al, 1984a;
Johnson et al, 1986). This is
not because testes of older men are smaller than those of younger men; on
average, they are not (but see discussion in
Johnson et al, 1984b). Rather,
there is a small, but significant (P < .01), decline in DSP per
gram with advancing age (Figure
5B). Nevertheless, age accounts for only 8% of the variation
of DSP per testis (Johnson et al,
1984b). (Note: men studied by Johnson's group experienced sudden
death; testes were promptly removed and processed, usually by perfusion to
clear blood followed by glutaraldehyde fixative.) Also, Johnson et al
(1984b) concluded that around
33°N latitude of Texas, any summer vs winter difference in mean DSP per
gram was undetectable.
Clinical andrologists recognize the benefit from examination of 2 biopsies from a patient (McLachlan et al, 2007), when biopsy is considered necessary. However, Johnson et al (1980b) opined measurement of DSP per gram (after homogenization of fixed tissue) with a portion of a single biopsy would correctly reflect that individual's DSP pooled across both testes. This conclusion is tenuous. Only 1 testis from each of 12 men was biopsied, and values for DSP per gram in the biopsy were correlated and compared with the mean value for 6 samples (top, middle, lower thirds of both testes) for that individual rather than the value for the corresponding third of the biopsied testis. Furthermore, the important question is whether the 2 values for an individual are comparable, not whether there is a relationship between them or whether a t test finds no significant difference between means pooled across both testes vs the value from 1 biopsy. To compare 2 measures of the same thing (eg, DSP/g), standard regression/correlation or t test approaches are not ideal for reasons discussed by Bland and Altman (1986), Glantz (2002), and Dallal (2007).
Johnson et al
(1980a,b,
1984a) concluded that mean DSP
per gram is not significantly different among the top, middle, or lower thirds
of a testis or between the 2 testes of an individual. Johnson et al
(1980a) reported a CV for DSP
per gram testis of 15%, apparently representing variation among thirds of
a testis averaged across 20 testes. Later, values for the 3 portions of testes
from 10 young men (20–48 years) and 10 older men (50–85 years)
were subjected to 1-way analysis of variance, separately for left and right
testes and each age group. Statistical analysis of their data
(Johnson et al, 1984a:table 1)
showed that DSP per gram in a given third of a testis is not consistently high
or low for all individuals but did not address the question of whether there
is substantial variation, or concordance, among the 3 portions of a given
testis. This could be answered by the intraclass correlation coefficient or
multiple paired comparisons by the Bland-Altman procedure
(Dallal 2007). If raw data
from Johnson et al
(1980a,b,
1984a) are available,
appropriate statistical analysis should remove uncertainty if DSP per gram
usually is concordant among thirds of a testis, adequately represented by a
single biopsy, and usually concordant between testes of an individual.
Publications on DSP in humans are summarized in the
Table. For younger men,
20–40 or 50 years of age, the 95% CIs for weight of the parenchyma
in paired testes (g), DSP per gram (106/day), and DSP per man
(106) tended to be near 28–40, 4.7–6.8, and
150–275. For men over 50 years of age, the respective 95% CIs tended to
be near 25–35, 2.8–5.2, and 90–150. Although original
publications did not state that data for DSP per gram or per testis were not
normally distributed, it is possible that use of a transformed endpoint (eg,
log DSP/g) might reduce the standard error and, hence, width of the
back-transformed 95% CI.
|
In males of a species with a substantial infantile phase, including humans,
bulls, and rabbits but not mice or rats, there is a period of time as
spermatogenesis "gets up to speed"
(Courot et al, 1970). The
change in DSP per gram between puberty (sperm first produced) and sexual
maturity (maximum DSP achieved by an individual) has not been studied
systematically for humans. Available data for humans
(Amann, 1981:table A1) show
that DSP per gram is low in young men as the process starts up. Means were
0.4, 3.5, and 3.9 x 106 DSP/g parenchyma for men
16–18, 19–20, or 21–30 years old, respectively (means for 6,
10, and 40 testes).
It is not known why testes in certain men produce far fewer sperm each day
than those in other individuals. In older men, but not younger men, 40%
of preleptotene/leptotene spermatocytes degenerate before they become
pachytene spermatocytes. However, across men of all ages, a deficiency in
number of pachytene spermatocytes per gram testis parenchyma accounted for
only 50% of the variation in DSP per gram (Johnson et al,
1983,
1987,
2000). In addition, and
regardless of age, 30%–40% of pachytene/diplotene spermatocytes fail to
produce spherical spermatids; apparently the problem is in meiosis-2.
Approximately 70% of the variation in DSP per gram is associated with
degeneration of germ cells during late meiosis. Virtually all spherical
spermatids complete spermiogenesis.
Extensive degeneration of primary or secondary spermatocytes is consistent
with the observation (Johnson et al,
1992) that, in testes with a low DSP per gram, at least 1
generation of germ cells was missing in 50% of cellular associations
examined. Such missing generations were far less common in testes with high
DSP per gram. Also, men with a low DSP per gram tended to have fewer cellular
associations per cross section of seminiferous tubule
(Chaturvedi and Johnson,
1993). Johnson et al
(2000) discounted the concept
that degeneration of germ cells was a mechanism to eliminate cells with
abnormal chromosomes because the magnitude of degeneration was relatively
constant, and they speculated that it might be a mechanism to adjust number of
germ cells to a number sustainable by available Sertoli cells.
Obviously, sperm production is intimately dependent on Sertoli cells. Men
50–90 years old have 30% fewer Sertoli cells per gram testis
parenchyma than men 20–48 years old, and 40% fewer Sertoli cells
per testis (Johnson et al,
1984c,
1986). Nevertheless, numbers
of all germ cells or spherical spermatids per Sertoli cell were similar for
both age groups. Across all 71 men, the correlation between number of Sertoli
cells and DSP was 0.70. These results confirmed earlier studies. However, DSP
per man is independent of number of Leydig cells
(Neaves et al, 1987), probably
because steroidogenic capacity is more dependent on volume of smooth
endoplasmic reticulum than cell number. Indeed, total volume of Leydig cell
smooth endoplasmic reticulum was correlated with DSP per testis (r =
.80; n = 10 men; Johnson et al,
1990).
Planning Studies Where Change in DSP Is an Endpoint— Planning any study should include use of the among-individual CV for the measure(s) of interest, to estimate number of individuals needed per group to detect a difference of a given magnitude among 2 or more groups. From the preceding discussion, it is obvious that CVs for DSP are likely to be large. Berndtson (2008) calculated among-individual CVs for DSP per gram and DSP per testis in publications for humans; his paper includes information for several species. For testis homogenates, Berndtson's values for among-individual CVs in DSP per gram averaged 48% and 89%, for 7 groups of younger men and 4 groups of older men. For DSP per testis, respective CVs were approximately 58% and 104% (5 and 2 groups). For DSP data based on morphometric analysis, only data for younger men were summarized (Berndtson, 2008) and for both DSP per gram and DSP per testis the among-individual CVs averaged 46%. These values far exceed imprecision in values for a given testis because within-individual CVs are <15%, as documented in several of Johnson's publications.
Assumptions needed to use information on CVs to estimate optimal sample size for a study were discussed by Berndtson (1991) and are considered on numerous web sites. For an anticipated magnitude of difference between mean values for a measured attribute, there are tradeoffs between probability or effect of making the wrong conclusion and sample size. Taking the CVs presented above and entering them in table 1 in Berndtson (1991) suggests that an 80% probability of detecting a 30% difference in DSP per testis (or individual) would require at least 63 young men per group, but if one wished to have a 90% probability of detecting a 20% difference, at least 235 young men per group would be needed. For older men, corresponding values are approximately 190 and 540, respectively. If DSP per gram is the endpoint, fewer individuals would be needed. Increasing precision of individual measurements would not decrease these numbers because the crucial feature is interindividual variation. Studies of human DSP per gram or DSP per testis rarely have used >60 individuals per group, so a conclusion of no significant difference might result from a type II error.
What Knowledge Is Needed About the Cycle of the Seminiferous Epithelium?
Cellular Associations—
There are several problems. First, the cycle has been defined in terms of
only 6 cellular associations, of which 3 collectively account for 81% of the
cycle (Figure 1). Early on,
Heller and Clermont (1964)
noted the need for discernment of more cellular associations, but no one has
reported success or failure in this task.
To better understand human spermatogenesis, it is important to develop a
scheme involving 
10 cellular associations, each representing <15% of
the duration of a cycle of the seminiferous epithelium (ie, <2.4 days).
This might be accomplished by use of phase contrast
(Johnson et al, 1981) or
Normaski optics (Johnson,
1994; Johnson et al,
2001), glutaraldehyde fixation and plastic sections stained with
toluidine blue (Johnson et al,
1992), or novel immunochemical stains. Images published by Johnson
and colleagues suggest that identification of distinct but gradually different
steps in spermiogenesis could fill voids in column E of
Figure 2 and allow definition
of perhaps 10 cellular associations. Recall that Johnson's group intentionally
fit identifications of germ cells to fit cellular associations described by
Clermont (1963).
Second, the primary report for the relative frequency of cellular associations (Clermont, 1963), which provides the accepted basis for calculations of both relative and absolute timing of spermatogenesis, was based on a total of 277 tubule cross sections representing 4 individuals. Rather than trying to resolve any difference between values reported by Clermont (1963) and Rowley and Heller (1971), a new classification scheme for cellular associations should be established and their relative frequencies then established on the basis of a robust data set. Properly fixed tissues necessary for this task might be available from Dr Larry Johnson, Department of Veterinary Anatomy and Public Health, Texas A&M University, College Station, Texas.
Repeatability of scoring cellular associations, or estimating their relative frequencies, has never been reported for human testes. This gap should be filled, although for other species, it is evident that scoring can be repeatable. Coefficient of repeatability for triplicate data sets (each based on independent scoring of 800 cross sections through seminiferous tubules per testis) for 8 rabbit testes was .82 (Amann, 1970a). However, in comparing results in Swierstra and Foote (1963), Orgebin-Crist (1968), and Amann and Lambiase (1969), it seems that relative frequencies might differ among observers or laboratories.
Number of tubule cross sections or cellular associations "scored" per testis affects repeatability or precision, as does the technician. Information in Clermont (1963) is insufficient to estimate sampling needs for good precision, but it is likely that 3–17 groups of cellular associations III and VI were encountered in his observations of 33–87 tubules in a given testis. This is too few. Heller et al (1969) and Rowley and Heller (1971) did not address repeatability of observations for percentage of each cellular association in a given biopsy. They arbitrarily evaluated 30 essentially round cross sections for a given biopsy; probably, they found >12 groupings per biopsy only for cellular associations I and V.
Third, concurrent with the above, precision and accuracy of available measurements of the time base underlying the cycle of the seminiferous epithelium might be improved. This is considered in the duration of spermatogenesis section.
Spermatogonia— The pattern and timing of spermatogonial renewal and proliferation to yield primary spermatocytes, and how spermatogonial differentiation is regulated, remain unknown. No model for spermatogonial renewal and proliferation proposed during the last 40 years was based on new data. All used data in Clermont (1966a) for cellular associations, when spermatogonial mitosis occurred, and ratios between cell types. The original, favored pattern proposed by Clermont (1966a,b) was largely unchallenged until recently (Aponte et al, 2005; Ehmcke and Schlatt, 2006; Ehmcke et al, 2006). The pattern of spermatogonial renewal and proliferation depicted in Figure 4 is a plausible model. Because cellular associations when a spermatogonium becomes committed to differentiation or division to form primary spermatocytes are uncertain, the duration of spermatogenesis also is uncertain.
Unanswered questions include: When are different spermatogonia formed? When is a spermatogonium committed to differentiation? Are spermatogonial divisions synchronous. Does a given event always occur in the same cellular association? When and why do many spermatogonia die before producing preleptotene spermatocytes? Study of human spermatogonia would benefit from application of time lapse photography, with confocal or differential interference contrast microscopy to view cultured whole-mount tubules. This would enable linking changes in cell location, cell numbers, and mitotic metaphase plates within the framework of redefined cellular associations. Optical sectioning and imaging should allow 3D study of living human seminiferous tubules. This is not fantasy because the approach has been used with mouse testes (Yoshida et al, 2007). Alternatively, in vitro labeling of DNA in spermatogonia early during short-term culture of intact seminiferous tubules, with appropriate subsequent processing and detection of labeled cells and cellular association in sections, could help establish when A- or B-spermatogonia are getting ready to divide.
|
74 days in humans
(Heller and Clermont, 1964;
Heller et al, 1969). Heller
and Clermont (1964:570) stated
"The whole spermatogenesis which extends over the duration of 4.6 cycles
thus consumes approximately 74 days," and discussed why 4.6 cycles is
correct rather than some other value. Authors citing a value of 64 days as the
duration of spermatogenesis (eg, de
Kretser and Kerr, 1994;
Schlegel and Hardy, 2002)
ignore the fact that spermatogenesis includes proliferation of spermatogonia
(Heller and Clermont, 1963;
Courot et al, 1970). As
evident in Figure 4, an
interval of 9–10 days needs to be added to the time from
3H-thymidine injection to spermiation.
Unfortunately, in a frequently cited review, Clermont
(1972:table 2) gave a value of
64 days as duration of spermatogenesis in humans, although he discussed in
detail the need to include spermatogonial events leading up to preleptotene
spermatocytes within the duration of spermatogenesis and gave an approximation
of this time interval. It is too common for authors to state that
spermatogenesis includes proliferation of committed spermatogonia through
spermiation, yet state that the "entire spermatogenic process"
requires 64 days. Until a better value is available, the duration of human
spermatogenesis should be cited as 74 days with attribution to Heller and
Clermont (1964).
Methodology. Duration of spermatogenesis in humans was measured by injection of 3H-thymidine, followed by testis biopsies to follow progression of labeled germ cells and detection of labeled cells by autoradiography (Heller and Clermont, 1963, 1964). There is no alternative to detecting label in testis biopsies, because detection of labeled sperm in ejaculated semen cannot provide a precise value.
To avoid 3H-thymidine or 5-bromo-2'-deoxyuridine, Misell
et al (2006) had subjects
drink 2H2O 2–3x per day for 14 days and
monitored arrival of 2H-DNA in sperm obtained by masturbation
(after 2 days of abstinence) at 2-week intervals. They correctly noted
that the duration of epididymal transit should be subtracted from the interval
to appearance of labeled sperm in semen but did not do this themselves! In any
case, their approach was fatally flawed. Apparently, the authors failed to
recognize that it was impossible to accurately anticipate when the most
advanced labeled germ cells, or subsequent populations of labeled cells, were
liberated from the seminiferous epithelium. (For the same reason, an early
estimate of epididymal transit time
[Rowley et al, 1970] is
invalid.) This problem was discussed in Amann
(1972) and is discussed in the
next section.
The paradigm used by Misell et al (2006) also precluded accurate timing of when labeled sperm first were available for ejaculation or first were ejaculated because semen was obtained only every 2 weeks. Furthermore, their calculations failed to recognize that spermatogenesis includes proliferation of spermatogonia (Heller and Clermont, 1963; Courot et al, 1970; Clermont, 1972).
With a modern approach, a new scheme for classification of cellular associations, and informed consent of subjects, a new study of the duration of spermatogenesis would provide very important information. The paradigm might include administration of 5-ethynyl-2'-deoxyuridine (EdU; Salic and Mitchison, 2008), followed by 2–3 carefully timed biopsies per testis; immersion fixation of biopsies in glutaraldehyde before or after staining with Alexa594-azide to render the EdU fluorescent, embedment in plastic, staining with toluidine blue (if stained sections preferred over differential interference contrast optics to study cellular associations), and evaluation of labeled germ cells with fluorescence microscopy in the context of new cellular associations. EdU offers superb localization of the label and has been used to study in vivo cell proliferation in mice (Salic and Mitchison, 2008).
Consistency of duration of spermatogenesis. Observed duration of spermatogenesis differs slightly between the 2 testes of an individual rabbit or among individuals (Amann et al, 1965; Amann, 1972), even though mean values across studies (see below) do not differ. On day 30.5 after systemic injection of 3H-thymidine (Figure 6), labeled mature spermatids remained within the testes of all 5 rabbits, with no labeled sperm detected in the efferent ducts on either side. Not until day 33 had labeled sperm been released from all testes. From these and other data, Amann (1972) concluded that the duration of spermatogenesis varied (biological variation and also imprecision in measurement) by 6%–9%, either within or among rabbits, and was not different for rabbits ejaculating daily or at sexual rest. Similar variation between paired human testes, or even within a testis, is anticipated on the basis of variation among men reported by Heller and Clermont (1964).
For laboratory rabbits (several breeds and genetic backgrounds), mean values for the duration of 1 cycle of the seminiferous epithelium of 10.9, 10.7, 10.4, 10.3, and 11.2 days (Swierstra and Foote, 1963, 1965; Amann et al, 1965; Orgebin-Crist, 1965, 1968; Amann and Lambiase, 1969) are generally consistent, even across laboratories. For laboratory rats, available data suggest that the duration of spermatogenesis might differ among strains (Amann, 1970b; Clermont 1972), but the 7%–11% difference might not be real given the small numbers of animals and precision of measurements.
In their classic report for humans, Heller and Clermont (1964) suggested ±1 day of uncertainty in the 16-day duration of 1 cycle of the seminiferous epithelium. Heller et al (1969) gave a value of 74 ± 2 days for the duration of spermatogenesis (4.6 cycles), giving a 95% CI of 70–78 days. It is uncertain how much of this variation represents imprecision of measurements and how much is variation within or among individuals. The duration of spermatogenesis is unlikely to differ substantially among different races of humans and should be more or less uniform for postpubertal men.
Take-Home Message
The cycle of the seminiferous epithelium is described as 6 cellular
associations, of which 2 represent 30% and 31% of the groupings seen. The
effect of this problem will increase as molecular regulation of human
spermatogenesis is studied. It is not known when during spermatogenesis
Apale-spermatogonia become committed to proliferate or what the
number of mitotic divisions involved in producing primary spermatocytes is.
The accepted duration of spermatogenesis might be in error by 6 days.
The andrology community should promptly determine whether my concern is justified. If restudy is deemed appropriate, proceed to do it. This effort should not be considered intellectually inferior, because it will require lateral thinking to draw on diverse techniques and innovation plus dedication to overcome the notion that it is not "modern science." The biomedical community is pursuing the concept of a fetal basis for adult disease involving testis function and the effects of environmental agents on human spermatogenesis. To "get it right," substantial targeted funding (ie, request for proposals) is needed to revisit human spermatogenesis.
Acknowledgments
Drs W.E. Berndtson, L. Johnson, and D.N.R. Veeramachaneni kindly and critically reviewed a draft of this manuscript. Their suggestions are appreciated, but all errors of fact or interpretation are mine. Fifty-one years after I first sectioned and examined bull testes, spermatogenesis remains a challenging topic with unanswered questions. I express appreciation to individuals and funding agencies who have allowed me to wonder and wander, and to colleagues who tried to straighten me out.
References
Almquist JO. Effect of long term ejaculation at high frequency on
output of sperm, sexual behavior, and fertility of Holstein bulls; relation of
reproductive capacity to high nutrient allowance. J Dairy
Sci. 1982; 65:814–823.
Amann RP. Reproductive capacity of dairy bulls. III. The effect of ejaculation frequency, unilateral vasectomy, and age on spermatogenesis. Anat Rec. 1962; 110: 49 –67.[CrossRef]
Amann RP. The male rabbit. IV. Quantitative testicular histology and comparisons between daily sperm production as determined histologically and daily sperm output. Fertil Steril. 1970a; 21: 662 –672.[Medline]
Amann RP. Sperm production rates. In: Johnson AD, Gomes WR, VanDemark NL, eds. The Testis. Vol 1 . New York, NY: Academic Press; 1970b; 433 –482.
Amann RP. The effect of variations in the duration of rabbit spermatogenesis on determinations of sperm epididymal transit time. In: Proceedings of the VII International Congress of Animal Reproduction. 1972;1: 432 –435.
Amann RP. A critical review of methods for evaluation of spermatogenesis from seminal characteristics. J Androl. 1981;2: 37 –58.
Amann RP, Almquist JO. Reproductive capacity of dairy bulls. I.
Technique for direct measurement of gonadal and extra-gonadal sperm reserves.
J Dairy Sci. 1961; 44: 1537
–1543.
Amann RP, Almquist JO. Reproductive capacity of dairy bulls. VIII.
Direct and indirect measurement of testicular sperm production. J
Dairy Sci. 1962;45: 774
–781.
Amann RP, Howards SS. Daily spermatozoal production and epididymal spermatozoal reserves of the human male. J Urol. 1980; 124: 211 –215.[Medline]
Amann RP, Johnson L, Thompson DL Jr, Pickett BW. Daily spermatozoal production, epididymal spermatozoal reserves and transit time of spermatozoa through the epididymis of the rhesus monkey. Biol Reprod. 1976;15: 586 –591.[Abstract]
Amann RP, Kavanaugh JF, Griel LC Jr, Voglmayr JK. Sperm production
of Holstein bulls determined from testicular spermatid reserves, after
cannulation of the rete testis or vas deferens, and by daily ejaculation.
J Dairy Sci. 1974; 57: 93
–99.
Amann RP, Koefoed-Johnsen HH, Levi H. Excretion pattern of labeled
spermatozoa and timing of spermatozoal formation and epididymal transit in
rabbits injected with thymidine-3H. J Reprod
Fertil. 1965;10: 169
–183.
Amann RP, Lambiase JT Jr. The male rabbit. III. Determination of daily sperm production by means of testicula
