Introduction

In the 18th and 19th centuries, the anatomy of descent of the testis was the primary impetus of research, while throughout the 20th century, the hormonal regulation of testicular descent became the main focus. By the 1980s, attempts were being made to integrate the disparate anatomical and regulatory evidence into a unifying schema, culminating in the proposal of the two-stage model, with the recognition that different hormones were regulating the early and later stages of descent [1].

It is now generally accepted that testicular descent occurs in two discrete anatomical and hormonal stages. In the human both phases occur prenatally, with the transabdominal phase between 10 and 15 weeks’ gestation and the inguinoscrotal phase between 25 and 35 weeks of gestation. By contrast, in rodents the transabdominal phase occurs in the third trimester while the inguinoscrotal phase occurs in the first week to 10 days after birth. Apart from these differences in timing, however, the anatomy and hormonal regulation of the two stages of testicular descent are remarkably similar between rodent models and humans [2] (Fig. 1).

Fig. 1
figure 1

Testicular descent in the human foetus vs that in a rodent. In both species, the process occurs in two separate phases: the transabdominal and inguinoscrotal stages. Migration of the gubernaculum is similar, except that in rodents the extracellular matrix in the gubernaculum regresses before migration begins at birth, while in humans this occurs after the gubernaculum reaches the scrotum, and the entire process is prenatal. Also in humans, the last step after descent is closure of the processus vaginalis (to prevent inguinal hernia), while in rodents the processus remains open and a fat pad on the epididymis, which plugs the inguinal canal, prevents herniation (reproduced with permission from J Pediatr Urol [2])

Full size image

Early embryology

The gonads develop on the anteromedial surface of the mesonephros in urogenital ridge, which is attached to the posterior coelomic wall and by thickenings of the attachment cranially (cranial suspensory ligament) and caudally by the genitoinguinal ligament, or gubernaculum. About the time of sexual differentiation in the human (7–8 weeks’ gestation) the mesonephros regresses, leaving the developing ovary or testis on a mesentery, now called the mesovarium or mesorchium. In the free edge of the urogenital ridge, the mesonephric (Wolffian) duct and the paramesonephric (Müllerian) duct develop. The Wolffian duct initially drains the mesonephros, but after regression of the latter structure the duct becomes attached directly to the testis to form the rete testis. Under the action of androgen from the newly formed Leydig cells in the developing testis, the Wolffian duct continues to differentiate into the epididymis and vas deferens. At the caudal end of the Wolffian duct the ureteric bud forms, and cranially to this a second, hormone-dependent bud forms the seminal vesicle. The Müllerian duct in the male regresses under the influence of the hormone secreted by the newly differentiated Sertoli cells, anti-Müllerian hormone (AMH) (also known as Müllerian inhibiting substance (MIS)) [3, 4].

Transabdominal phase

Shortly after sexual differentiation at 7–8 weeks’ gestation, the transabdominal phase of testicular descent occurs between 10 and 15 weeks. In male, the cranial suspensory ligament regresses and the gubernaculum enlarges; while in the female both ligaments persist without obvious changes as the foetus grows. By a combination of gonadal enlargement and the ‘swelling reaction’ in the gubernaculum, the testis remains close to the future inguinal canal, where the gubernaculum is attached to the inguinal abdominal wall (Fig. 2a). By contrast the developing ovary moves relatively further from the inguinal region as the female foetus enlarges.

Fig. 2
figure 2

The two stages of testicular descent. a Before descent the developing testis is held in the urogenital ridge by the cranial suspensory ligament (CSL) cranially and the gubernaculum (G) caudally. The adjacent Wolffian duct (WD) forms the epididymis and vas deferens in a male while the Müllerian duct (MD) forms the uterus and tubes in a female. b At the end of the transabdominal phase (~15 weeks) the testis is held near the future inguinal ring by the swelling reaction in the gubernaculum. The skin just beyond the gubernaculum is over the future external inguinal ring, as the scrotum is remote in the perineum of a mammal. c Inguinoscrotal phase requires the gubernaculum to elongate to the scrotum, under control of androgens and calcitonin gene-related peptide (CGRP) released from the genitofemoral nerve (GFN). After migration is complete, the peritoneum of the processus vaginalis (PV) closes and then completely involutes and disappears (see Fig. 4)

Full size image

The ‘swelling reaction’ in the gubernaculum is caused by cell division in the primitive mesenchymal cells of the distal gubernaculum along with a sudden increase in extracellular matrix molecules, especially glycosaminoglycan and hyaluronic acid [5]. This ‘swelling reaction’ leads to the caudal end of the gubernaculum enlarging to a similar size as the testis where the former is embedded in the abdominal wall [6]. The swollen distal gubernaculum is known as the bulb, and the inguinal abdominal wall muscles differentiate around it to produce the inguinal canal. The proximal attachment of the gubernaculum to the testis and developing epididymis is known as the gubernacular cord, and this also remains short in the male (Fig. 2b), unlike the female where the entire gubernaculum remains long and forms the round ligament.

The peritoneum over the intraabdominal surface of the gubernacular bulb forms an annular diverticulum that grows into the gubernaculum, dividing the gubernaculum into 3 distinct anatomical parts. The infravaginal part (caudal to the peritoneal diverticulum) is the bulb of the gubernaculum, which contains undifferentiated mesenchymal cells. Inside the annular diverticulum, the central column of gubernacular cells differentiates into fibroblasts to form the gubernacular cord, which anchors the intraperitoneal gonad and epididymis to the bulb, which is embedded in the inguinal muscles. Just outside the peritoneum in the gubernaculum the cremaster muscle develops, and with elongation of the diverticulum later the muscle comes to lie around the outside of the diverticulum (which will form the processus vaginalis) in the extra-vaginal part (outside the processus vaginalis).

In 1999, it was discovered that knockout of the insulin-like hormone 3 gene in mice had undescended intraabdominal testes [7, 8], and it has now been confirmed that insulin-like hormone 3 (INSL3) is the primary hormone regulating the swelling reaction. INSL3 secreted from the Leydig cells of the testis stimulates the swelling reaction, and in conjunction with the short gubernacular cord, this provides traction on the testis to keep it close to the inguinal abdominal wall as the foetal abdomen enlarges. INSL3 is secreted in mid gestation which is at the right time to control the swelling reaction, and its receptor (the relaxin family receptor 2, RXFP2) also is located in the gubernaculum at the right time.

INSL3 is a member of the insulin family of related hormones and growth factors, and is synthesised as a 131-amino acid preprotein, which contains a 24-amino acid signalling peptide [9]. In vitro studies of the foetal rat gubernaculum showed that INSL3 stimulated gubernacular growth, with both AMH and testosterone providing some augmented stimulus [10]. Moreover, INSL3 knockout in mice prevented the swelling reaction, so that the gubernacular bulb lacked a central core of undifferentiated mesenchyme at embryonic day 16.5 [7]. When the INSL3 gene was activated in female mice, the ovaries descended to the lower abdomen beside the bladder neck, analogous to transabdominal descent in male mice [11]. INSL3 appears to act through the RXFP2 receptor and then the downstream intracellular signalling involves both NOTCH and Wnt/β-catenin pathways [12–14].

Mutations of INSL3 or its receptor are uncommon in humans with cryptorchidism, which is in keeping with the fact that intraabdominal testes with deficient transabdominal descent are uncommon [15, 16]. However, when considered in proportion to children with impalpable testis only, defects in INSL3 signalling are likely to be common [17].

Although INSL3 is now accepted as the primary hormone controlling the gubernacular swelling reaction and transabdominal descent, there are some pieces of evidence that suggest a role for AMH, particularly in the human. First, in children with mutations of the AMH gene or its receptor, they have undescended testes with persistence of an infantile uterus and tubes, known as the persistent Müllerian duct syndrome (PMDS). The testes are intraabdominal (~70 %) or prolapsed into a hernial sac (~20 %, ‘hernia uteri inguinalis’), or where both testes (along with uterus and tubes) are prolapsed into the same patent processus vaginalis (~10 %, transverse testicular ectopia) [18]. The gubernacular cord in these patients is abnormally long (i.e., >10 cm rather than <½ cm), which prevents the testis being held near the internal inguinal ring as in normal transabdominal descent. By contrast, the swelling reaction is presumed to be normal so that the gubernacular bulb creates an inguinal canal and migration to the scrotum in the inguinoscrotal phase is also normal. The elongated gubernacular cord mirrors a very long round ligament, and allows the testes to flop about in the pelvis and prolapse into the ipsilateral processus vaginalis (to create ‘hernia uteri inguinalis’), or even into the contralateral processus vaginalis (to cause transverse testicular ectopia) [18, 19]. In addition, the extreme mobility of the testis in PMDS may predispose to the reported high frequency of intraabdominal torsion causing vanishing testis in this anomaly [20].

Inguinoscrotal phase

The second phase of testicular descent occurs about 25–35 weeks of gestation in the human foetus and in the first week postnatally in a mouse. The overall process is quite similar in most mammals, allowing for some differences [2]. In human, the gubernaculum which originally ended in the inguinal abdominal wall bulges out to create a future external inguinal ring. The gubernaculum then migrates to the scrotum by remodelling from an inert ligament into an actively migrating and elongating structure, with many analogies to an embryonic limb bud [21–23] (Fig. 2c). What triggers this sudden change in the biology of the gubernaculum is incompletely understood, but there is some evidence that the mammary line may be involved [24].

The possible role of the mammary line in the inguinoscrotal phase is suggested by the anatomy of the marsupial, which separated in evolution from eutherian mammals about 200 million years ago. The mammary line persists over the inguinal ring in both modern marsupials, such as the kangaroo, as well eutherian mammals such as rodents and humans. In the kangaroo and wallaby, the homologous muscle to the cremaster is known as the ilio-marsupialis, and it is supplied by the genitofemoral nerve (GFN) in both sexes (the same as in mammals) [25]. In male, it has a suspensory function similar to the retractile reflex; while in female, the muscle is attached to the breast and is the suspensory muscle of the nipple [26] (Fig. 3).

Fig. 3
figure 3

a Foetal male rodent showing close relationship between gubernaculum and mammary bud (M) before inguinoscrotal phase begins (P penis, S scrotum). b Marsupial male pouch young showing gubernaculum extending into scrotum, which is above the pubis, in the same site as the mammary bud in rodent (a) and female marsupial (c). c Marsupial female pouch young showing muscle of gubernaculum (ilio-marsupialis, homologue of cremaster muscle) attaching to the developing mammary bud

Full size image

Once the close association between the cremaster muscle and GFN and the breast was appreciated in the marsupial, we went back and looked specifically in the foetal rodent, and to our surprise we found that the GFN supplies not only the gubernaculum itself but also the breast bud. Moreover, the breast bud is located just superficial to where the gubernaculum ends in the abdominal wall [24, 27]. The mammary line has inductive properties that control the underlying mesenchyme similar to the apical ectodermal ridge of an embryonic limb bud, as it arises on the side of the embryo precisely between the upper and lower limb buds [28]. Although it is not completely established, we suspect that the mammary line over the end of the gubernaculum provides the initial signalling to enable remodelling and outgrowth of the gubernaculum like an embryonic limb bud at the start of inguinoscrotal descent [29].

During inguinoscrotal descent, the gubernaculum elongates towards the scrotum while the processus vaginalis inside it also elongates, enabling the intraperitoneal testis to leave the abdomen while still remaining inside an extension of the peritoneum. It is likely that extracellular matrix enzymes, which are produced by the gubernaculum or the inguinoscrotal fat pad itself, dissolve the matrix to enable the gubernaculum to elongate in a free tissue plane without obstruction [30]. Once the gubernaculum has reached the scrotum in human several important changes occur. The bulky, gelatinous extracellular matrix of the gubernacular bulb resorbs, leaving a small fibrous remnant. In addition, the fibrous remnant becomes adherent to the inside of the scrotum. In this short time interval, between arrival of the gubernaculum in the scrotum and development of its fibrous connection to the surrounding tissues, is when perinatal torsion of the testis is likely to occur, because of the extreme mobility of the gubernaculum and its contained testis at that time [31].

The final event after the gubernaculum and testis reach the scrotum is closure of the proximal processus vaginalis in the human, thereby preventing inguinal hernia and/or hydrocele. This final stage of testicular descent is seen in humans and primates, but many other mammals have a processus vaginalis that remains patent. In mouse and rat, for example, the inguinal canal remains patent throughout life, and inguinal hernia is prevented by the presence of a large fat pad attached to the head of the epididymis, which effectively plugs the inguinal canal. Patency of the processus vaginalis is demonstrated by the fact that the testis can retract back into the peritoneal cavity even in an adult rat when the retractile reflex is stimulated.

When all the stages of inguinoscrotal descent are seen together, it can be appreciated that there are actually three steps: (1) migration of the gubernaculum and elongation of the processus vaginalis inside it to enable the intraperitoneal testis to reach the scrotum while still inside the processus vaginalis; (2) closure of the proximal processus but not the distal tunica vaginalis, leaving the testis inside a satellite peritoneal cavity within the hemiscrotum; (3) obliteration of the remnant of the processus vaginalis, which enables the spermatic cord to elongate normally after birth (Fig. 4). Failure of migration leads to congenital cryptorchidism, while failure of the processus closure leads to inguinal hernia or hydrocele. Failure of the last step, i.e., complete involution of the PV, is likely to be the cause of acquired cryptorchidism.

Fig. 4
figure 4

Schema showing the sequential processes comprising the inguinoscrotal phase of testicular descent. At the end of the transabdominal phase, the enlarged gubernaculum occupies the future inguinal canal, and must migrate 3–5 cm to the scrotum (a step 1), taking the testis inside the processus vaginalis, which elongates inside the gubernaculum. Failure of this first step causes congenital cryptorchidism. After migration is complete, the processus vaginalis (PV) closes (b step 2), and failure of this causes inguinal hernia or hydrocele. The final (c step 3) process is complete involution of the PV remnant, allowing the spermatic cord to elongate after birth. Failure of this step is the likely cause of acquired cryptorchidism, as the fibrous remnant of the PV prevents the spermatic cord growing normally

Full size image

Androgens control the inguinoscrotal phase, as in both humans and animals with complete androgen insensitivity, the testis remains in the inguinal canal or groin, demonstrating normal transabdominal descent but completely deficient inguinoscrotal descent [1]. It was not appreciated until recently that androgens act in a narrow time window, which in the rat is embryonic days 15–19 [32]. It was assumed for many years that androgens would act directly on the gubernaculum, and hence androgen receptors (AR) should be present in the gubernaculum itself. However, on quantitative analysis of AR localisation using immunohistochemistry we found that AR only appeared in the gubernaculum perinatally, just after the window of androgenic sensitivity controlling postnatal inguinoscrotal descent [33]. Not only were AR absent in the gubernaculum during the foetal ‘programming window’, but they were also absent from the sensory cell bodies of the GFN in the L1-2 dorsal root ganglia until after birth [33]. As the GFN has been proposed to not only supply the gubernaculum and its contained structures (processus vaginalis and cremaster muscle), but also direct gubernacular migration to the scrotum, the mechanism by which androgens control descent via the GFN remains unknown, and it is the subject of current research (see below).

The role of the genitofemoral nerve (GFN)

In 1948, Lewis reported that transection of the GFN in a neonatal rat, just before inguinoscrotal migration occurs, caused cryptorchidism [34]. After repeating Lewis’ study in the 1980’s, Beasley and Hutson proposed that the effect of androgen on gubernacular migration may be via the nerve itself, which may release a neuropeptide [35] to control descent. Calcitonin gene-related peptide (CGRP) was subsequently identified in the sensory nucleus of the GFN after initial proposals that the neuropeptide may be in the motor nucleus proved incorrect [36].

There is now abundant evidence that CGRP synthesised in the GFN sensory branches modulates inguinoscrotal migration of the gubernaculum in rodent models. Both transection of the nerve and ablation of sensory nerves with a specific neurotoxin, capsaicin, not only interfere with gubernacular migration but also sensitise the gubernaculum to CGRP by upregulation of CGRP receptors with the gubernaculum to exogenous CGRP in vitro [37–39]. CGRP causes rhythmic contractibility of the developing cremaster muscle in the gubernaculum [40–42], which is thought to orientate the gubernaculum towards the scrotum by chemotaxis [43]. CGRP stimulates mitosis in the gubernacular bulb in vitro, although androgens are required for gubernaculum to respond [44, 45].

CGRP receptors are present in the rodent gubernaculum and in models of androgen blockade they are upregulated, consistent with androgen controlling CGRP release from the GFN [46], like the flutamide-treated rat and the TFM mouse. In a rat model of congenital cryptorchidism (transcrotal or TS rat), we found that there was no abnormality of androgenic action, but the GFN in this model contained too many sensory nerves and an excess of CGRP [47]. Cryptorchidism in the TS rat appears to be caused by an excess of CGRP in the GFN sensory branches causing downregulation of the gubernacular response to exogenous CGRP, as this can be reversed by transection or capsaicin, the sensory nerve toxin [39, 47].

With clear evidence that the sensory branches of the GFN control gubernacular development via CGRP, and knowing that this is androgen dependent, the failure of AR to be present in either the GFN or the gubernaculum itself in the foetal ‘programming window’ suggested that androgens must masculinise the GFN indirectly [48]. One possibility is that target organs of the GFN other than the GFN itself contains AR and responds to androgenic stimulation by synthesising a neurotrophin that is taken up by the nerve endings to modify their function (Fig. 5). This is already been shown to be the case in an adjacent perineal structure, the bulbocavernosus muscle, a muscle that is important in ejaculation, and that is sexually dimorphic. The bulbocavernosus muscle is thought to masculinise its own nerve supply in response to androgens, by producing neurotrophins to modify the nerve. Both brain-derived neurotrophin factor (BDNF) and ciliary neurotrophic factor (CNTF) have been linked to the bulbocavernosus providing peripheral neurotrophic regulation of the perineal branch of the pudendal nerve [49–54]. On the principle that important signalling systems are likely to be preserved in evolution, we are currently investigating the role of BDNF and CNTF in the GFN [25, 55] (Fig. 5).

Fig. 5
figure 5

Current hypothesis on how testosterone (T) masculinises the genitofemoral nerve (GFN) sensory fibres with cell bodies in the dorsal root ganglion (DRG) at L1, L2. The androgen receptors (AR) during the critical narrow time window are located in the inguinoscrotal fat pad which is supplied by the GFN, and through which the gubernaculum (G) must migrate to reach the scrotum. It is possible that the inguinoscrotal fat pad produces neurotrophins in response to androgen that regulate GFN function, so that it can produce CGRP to modulate the migration

Full size image

Current issues

To understand how androgens act on the GFN to produce CGRP for gubernacular migration in inguinoscrotal descent is an important step needed at present. Once we have finally unravelled the complex mechanism, we will be able to assess what are the likely steps that might be abnormal in cryptorchidism. Another issue still to be resolved is regulation of the remodelling that occurs in the inguinoscrotal fat pad to allow gubernacular migration to the scrotum. We need to know which extracellular matrix enzymes are produced and how they are regulated. At present, we have just begun such a study [30].

Another area still to be explored is how the gubernaculum involutes at the completion of migration, and what controls its fibrous adherence to the inside of the scrotum to prevent perinatal torsion of the testis.

An important current issue is how all the research we have done on the inguinoscrotal phase of descent in animal models relates to the human situation. Certainly there are some who have suggested that the GFN and its neuropeptide, CGRP, may be important in animal models but may not be relevant for testicular descent in humans [56]. It is a truism that extrapolation of biological findings in animal models to the human must be done with care, but the commonality of testicular descent in nearly all modern mammals implies that the important parts of the mechanism are likely to be the same or very similar, once we allow for some minor anatomical differences [2].

The key evidence supporting a role for the GFN in children comes from study of the processus vaginalis, and what controls its perinatal closure to prevent inguinal hernia [57]. As mentioned above, the processus vaginalis (PV) is derived from the specialised peritoneum covering the urogenital ridge [58], and it forms inside the gubernaculum to allow the intraperitoneal testis to exit from the abdomen while remaining inside the peritoneal cavity [59]. We reasoned that the development of the PV and inguinoscrotal descent must be integrated so that testicular descent and PV obliteration could be precisely coordinated in timing. PV closure usually occurs just before birth, when foetal androgens are present, or shortly after birth when the transient postnatal surge of androgen occurs, known as ‘minipuberty’ [60, 61]. As androgens were postulated to act via the GFN in rodents, we looked at inguinal hernia in children and found CGRP-immunoreactive nerve fibres and CGRP receptors in the human processus vaginalis [62]. More importantly, we found in an in vitro system that exogenous CGRP could induce fusion PV excised during inguinal herniotomy by epithelial transformation [63]. The CGRP receptors in the PV are not in the mesothelium itself, but in the underlying mesenchyme, and fusion of the PV can be induced by both CGRP and hepatocyte growth factor (HGF) which may be released by the mesenchyme to trigger the adjacent epithelium to transform into motile fibroblasts [64]. Taken together, all these studies suggest that the GFN is controlling both testicular descent and subsequent PV closure in humans, and that inguinal hernia is likely to respond to medical treatment, such as a local, slow-release injection of CGRP into the inguinal region [65].

The transient postnatal surge in gonadotropins and androgen known as ‘minipuberty’ occurs about 2–6 months of age in humans. It is thought to have a role in male gender identity, by regulating changes in brain function [66]. In addition, it is likely to have an important function in closing the PV (to prevent inguinal hernia), and also to remodel any remaining fibroblasts so that the PV has completely disappeared. This process is likely to be crucial to prevent acquired cryptorchidism, which is often caused by failure of the fibrous remnant of the PV to completely disappear [67]. Spontaneous descent of many acquired undescended testes at puberty suggests that PV involution is under androgenic control, probably indirectly via CGRP release from the GFN. Taken together these observations suggest that the GFN (via CGRP) regulates not only gubernacular migration prenatally, but also PV closure and subsequent obliteration postnatally. A possible cause for congenital cryptorchidism, therefore, is insufficient CGRP release from the GFN prenatally. If minipuberty is deficient, there might also be a deficiency of CGRP release postnatally, predisposing the infant to inguinal hernia, hydrocele and acquired UDT.

Conclusion

There is still much to learn about the regulation of testicular descent, but the evidence suggests that cryptorchidism is likely to be caused by a large number of defects interfering with the very complex mechanism whereby the previously ‘inert’ foetal gubernaculum is triggered to remodel and migrate 3–4 cm from the external inguinal ring to the scrotum [68]. The link between congenital and acquired UDT and inguinal hernia raises the possibility that all three conditions may respond to a local medical treatment, such as a depot injection of CGRP itself or a synthetic analogue.