Research Article

Autologous grafting of cryopreserved prepubertal rhesus testis produces sperm and offspring

See allHide authors and affiliations

Science  22 Mar 2019:
Vol. 363, Issue 6433, pp. 1314-1319
DOI: 10.1126/science.aav2914

Preserving male fertility

Before chemotherapy or radiation treatment, sperm from adult men can be cryopreserved for future use. However, this is not possible for prepubertal boys. Fayomi et al. grafted cryopreserved testicular tissue from castrated pubertal rhesus macaques, placing each animal's own testis sections under the skin of the back or scrotum (see the Perspective by Neuhaus and Schlatt). Grafts grew, produced testosterone, and were able to generate sperm that could fertilize oocytes, in one case resulting in a successful pregnancy. The results hold promise for preserving human fertility, for example, after childhood cancer treatments.

Science, this issue p. 1314; see also p. 1283


Testicular tissue cryopreservation is an experimental method to preserve the fertility of prepubertal patients before they initiate gonadotoxic therapies for cancer or other conditions. Here we provide the proof of principle that cryopreserved prepubertal testicular tissues can be autologously grafted under the back skin or scrotal skin of castrated pubertal rhesus macaques and matured to produce functional sperm. During the 8- to 12-month observation period, grafts grew and produced testosterone. Complete spermatogenesis was confirmed in all grafts at the time of recovery. Graft-derived sperm were competent to fertilize rhesus oocytes, leading to preimplantation embryo development, pregnancy, and the birth of a healthy female baby. Pending the demonstration that similar results are obtained in noncastrated recipients, testicular tissue grafting may be applied in the clinic.

Chemotherapy and radiation treatments for cancer or other conditions can deplete spermatogonial stem cells (SSCs) in the testis, resulting in permanent infertility (14). Before undergoing gonadotoxic treatment, adult men can cryopreserve sperm that can later be used to produce biological children by way of established assisted reproductive technologies. Sperm freezing is not an option for prepubertal boys, who are not yet producing sperm (5, 6). This is an important human health concern because the survival rate of children with cancer is above 80% (7, 8), and 30% of childhood cancer survivors will be infertile as adults (9). The only fertility preservation option available for prepubertal boys is cryopreservation of testicular tissues, which contain SSCs (10, 11).

There are several cell- and tissue-based therapies in the research pipeline that may allow patients to use their cryopreserved testicular tissues to generate sperm and produce biological children (5, 12). Testicular tissue grafting and xenografting are well-tested technologies in which immature testicular tissues, containing SSCs, are grafted ectopically under the skin. Immature testicular tissues from mice, pigs, goats, rabbits, hamsters, dogs, cats, horses, cattle, and monkeys have been grafted under the back skin of immune-deficient nude mice and matured to enable spermatogenesis (13), produce fertilization-competent sperm [as demonstrated in mice, pigs, goats, and monkeys (1416)], and generate live offspring [as demonstrated in mice, pigs, and monkeys (1619)]. Therefore, it is theoretically possible to graft immature testicular tissue from a childhood cancer survivor into an animal host to produce sperm that can be used in the in vitro fertilization clinic to achieve pregnancy. However, the possibility that viruses or other xenobiotics could be transmitted from the animal host to humans needs to be carefully considered (2022).

Three studies have reported autologous grafting of immature testicular tissue in nonhuman primates (2325). Complete spermatogenesis was reported for fresh (24) and cryopreserved (25) testicular tissue grafts in the scrotum but not for grafts under the back skin. Recovery of cryopreserved grafts was low (5%), with complete spermatogenesis observed in only 13 and 17% of seminiferous tubules in two surviving scrotal grafts (25). Sperm function was not tested by fertilization or production of offspring in those studies. Cryopreservation, perhaps for many years, is an essential component of the fertility preservation paradigm for prepubertal cancer survivors.

We modeled the prepubertal cancer survivor in rhesus macaques and report that autologously grafted, frozen, and thawed prepubertal rhesus testicular tissues were matured to produce sperm that were competent to fertilize rhesus oocytes, establish a pregnancy, and produce a healthy graft-derived baby (called “Grady”).

Experimental design

Five prepubertal rhesus macaques were hemicastrated (one testis was removed); testis tissues were cut into small pieces (9 to 20 mm3) and cryopreserved (Fig. 1A) (see supplementary materials and methods). Five to 7 months after hemicastration, the remaining testis was removed and cut into small pieces (9 to 20 mm3). Some of that tissue was designated for fresh tissue grafting and the remaining tissue was cryopreserved. Immediately after removal of the second testis, fresh and cryopreserved tissue fragments (cryopreservation time ranged from 5 hours to 5 months) were autologously grafted under the back skin (three sites fresh and three sites cryopreserved) and under the scrotal skin (one side fresh and one side cryopreserved) (Fig. 1B). Surgical scissors were used to create a skin flap at each site, and four testicular tissue fragments were independently sutured to the subcutaneous aspect of each skin flap (Fig. 1C). Average testicular tissue fragment weight at the time of grafting was 15.34 ± 1.54 mg. Matrigel was injected into four of the six sites on the back (two cryopreserved and two fresh) and into both scrotal sites to stimulate angiogenesis (Fig. 1B).

Fig. 1 Autologous grafting of prepubertal testicular tissue fragments.

(A) Fresh or cryopreserved testicular tissue fragments (9 to 20 mm3) from prepubertal monkeys. (Inset) Higher-magnification image of area demarcated with dashed box. (B) Testicular tissue fragments were grafted under the back skin or scrotal skin, as shown (viewed from the back of animal). Matrigel was added to four of the six graft sites on the back and to both scrotal sites. (C) Four pieces of fresh or frozen-thawed testis tissue were sutured to the subcutaneous aspect of the skin at each graft site.

Pre-graft testicular tissue is immature, lacking spermatogenesis

Histological examination confirmed that testicular tissues of all animals were immature at the time of hemicastration and at the time of castration (Fig. 2, A and B, and fig. S1). Undifferentiated stem or progenitor spermatogonia (types Adark and Apale) and differentiated spermatogonia (type B) were the only germ cells seen in the seminiferous tubules of four animals (13-022, 13-024, 13-026, and 13-030), whereas early meiotic cells (pachytene spermatocytes) were also observed in 0.8% of tubules of 13-008. Adult tissue cross sections are included as a control for comparison (Fig. 2C).

Fig. 2 Histological and immunofluorescent analyses of prepubertal testicular tissue before grafting.

H&E staining indicates that pre-grafted fresh (A) and frozen-thawed (B) testis tissues are immature. In contrast, multiple layers of germ cells with complete spermatogenesis were seen in adult testis tissue controls (C). Arrows indicate undifferentiated type Adark (Ad) and Apale (Ap) stem or progenitor spermatogonia and spermatocytes (Spct). Immunofluorescence staining for (D to F) VASA+ germ cells (red), (G to I) GFRA1+ undifferentiated spermatogonia (white), and (J to L) ACROSIN+ postmeiotic germ cells (green). (M to O) Merged images. DAPI, 4′,6-diamidino-2-phenylindole. Additional staining for UTF1, BOULE, and CREM is shown in fig. S2.

Immunofluorescent staining revealed that VASA+ germ cells were located in the lumen and on the basement membrane of the seminiferous tubules in pre-graft fresh (Fig. 2D) and pre-graft frozen-thawed testis tissue (Fig. 2E). Pre-graft tissues contained GFRA1+ undifferentiated spermatogonia (Fig. 2, G, H, M, and N) but no ACROSIN+ postmeiotic cells (Fig. 2, J, K, M, and N). Seminiferous tubules of adult controls contained multiple layers of VASA+ germ cells (Fig. 2F), including GFRA1+ undifferentiated spermatogonia (Fig. 2I) and ACROSIN+ postmeiotic spermatids (Fig. 2L). Additional markers of undifferentiated spermatogonia (UTF1), spermatocytes (BOULE), and spermatids (CREM) are shown in fig. S2.

Graft growth and endocrine function

A few months after grafting, palpable masses were observed at graft sites on the back (Fig. 3A) and scrotum (Fig. 3B). Graft growth was monitored using calipers, and graft area measurements (length times width) were recorded (Fig. 3, C to F). Graft sizes (back or scrotum) were not affected by cryopreservation (P > 0.05) (Fig. 3, D and E) or addition of Matrigel (comparisons from back only; P > 0.05) (Fig. 3F).

Fig. 3 Testicular tissue grafts increase in size during the 8- to 12-month in vivo incubation period.

By 4 to 5 months after grafting, grafts were easily visualized under the back skin (A) and scrotal skin (B). Calipers were used to monitor graft growth (C to F). All grafts grew during the 8- to 12-month incubation period. Grafts on the back and in the scrotal area grew significantly over time, relative to the first graft size measurement (C). Graft sizes (length times width) on each analysis date were not affected by processing (fresh versus frozen-thawed) in the scrotum (D) or on the back (E) or by addition of Matrigel to the back sites (F). Frozen grafts on the back exhibited a trend toward increasing size throughout the experiment (E), but that increase was not statistically significant. All other grafts exhibited statistically significant increases in size throughout the experiment. Data points are presented as mean ± SEM. *P < 0.05, compared with initial graft size measurement; **P < 0.01, compared with initial graft size measurement within each treatment group.

After grafting, and as animals entered puberty, circulating testosterone (T) levels increased in all monkeys and remained elevated above baseline in four of five recipients, indicating a functional hypothalamic-pituitary-testicular axis (fig. S3, A, C, E, G, and I). Normal pubertal levels of T in rhesus macaques are ~2 ng/ml (26, 27). Circulating follicle stimulating hormone (FSH) levels of all animals were in the normal range (not castrate range) for pubertal and adult rhesus macaques (26, 28), indicating a functional negative-feedback loop from the grafted testicular tissue to the hypothalamus and pituitary (fig. S3, B, D, F, H, and J).

Graft recovery and analysis

Eight to 12 months after grafting, testicular tissues were recovered from all (39 of 39) graft sites. There was no graft in the left scrotum of 13-030 because he ripped open that incision immediately after surgery and destroyed the graft. It was not possible to isolate grafts from individual tissue fragments because the four tissue fragments grafted at each site grew and fused into single large masses weighing an average of 308.61 mg (Fig. 4A and table S1). This represents an approximately fivefold increase in graft weight compared with the ~60 mg of tissue initially grafted at each site (four fragments × 15.34 mg/fragment). Seminiferous tubules that were ~150 to 200 μm in diameter were observed in all grafts (Fig. 4, B and C). Grafts from all animals were fixed for histology and immunohistochemical analyses and, in most cases (32 of 39 grafts), manually dissected and/or digested with collagenase IV to release sperm for fertilization experiments (Fig. 4D). Although the weight of testicular tissue grafts recovered from scrotal skin was greater (P < 0.01) than that of grafts recovered from the back skin (Fig. 4E), there was no difference in graft weight from fresh versus frozen testicular tissue (Fig. 4F), and graft weight was not affected by the addition of Matrigel (comparing back grafts only) (Fig. 4G).

Fig. 4 Recovery of testicular tissue grafts.

Grafts were recovered as one fused tissue, but white fibrous tissue may demarcate margins between individual testis tissue pieces that were originally placed at each graft (A). Seminiferous tubules could be distinguished after gentle teasing apart of the graft tissue with forceps (B). (C) Higher magnification of the boxed area in (B). In (D), the white arrows point to sperm that were released by mechanical dissection of a scrotal graft from animal 13-030. (Inset) Higher magnification of the boxed region. Grafts recovered from under the scrotal skin were larger than those recovered from under the back skin (E). Graft size was not affected by processing (fresh versus frozen) (F) or by addition of Matrigel (comparing back grafts only) (G). Bar graphs are presented as mean ± SEM. P < 0.05 was considered to be significant. N.S., not significant.

Complete spermatogenesis from autologous testicular tissue grafts

Seminiferous tubules from all grafts exhibited complete spermatogenesis with multiple layers of VASA+ germ cells (Fig. 5, A and C) and ACROSIN+ postmeiotic spermatids (Fig. 5, B and C). Additional staining for undifferentiated stem or progenitor spermatogonia (UTF1), spermatocytes (BOULE), and spermatids (CREM) is shown in fig. S4. Hematoxylin and eosin (H&E) staining confirmed complete spermatogenesis in grafts from all experimental animals (Fig. 5D and fig. S5). Most seminiferous tubules (≥70%) demonstrated complete spermatogenesis with elongated spermatids and/or sperm (Fig. 5, E to G, and fig. S5). Complete spermatogenesis with sperm was confirmed in tissues recovered from 100% of graft sites (39 of 39) (table S1). Neither the addition of Matrigel, nor cryopreservation, nor graft location had an impact on the percentage of tubules displaying complete spermatogenesis (spermatids and sperm) (P > 0.05).

Fig. 5 Histological evaluation of spermatogenic development in grafts.

Immunofluorescence staining of recovered graft tissue for (A) VASA+ germ cells (red) and (B) ACROSIN+ postmeiotic cells (green). DAPI counterstain marks all cell nuclei (blue). The merged VASA/ACROSIN/DAPI co-stain is shown in (C). See fig. S4 for additional markers of undifferentiated spermatogonia (UTF1), spermatocytes (BOULE), and spermatids (CREM). (D) H&E staining of post-graft tissues. See H&E staining for grafts from each individual animal in fig. S5. (E to G) Quantification of the most advanced germ cell type in graft seminiferous tubules. Bars represent mean ± SEM. P < 0.05 was considered statistically significant.

Grafts were fibrotic and difficult to dissect exclusively with forceps. After manual dissection, some grafts were digested with collagenase IV to release the remaining sperm. Live sperm were recovered from the majority of grafts (26 of 32). When sperm were recovered and quantified (19 grafts), counts ranged from 60 sperm to 21 million sperm per graft (table S1). Sperm were recovered from fresh and cryopreserved grafts, grafts on the back and in the scrotum, and grafts with or without Matrigel (table S1).

Fertilization and preimplantation embryo development from graft-derived sperm

To model the prepubertal cancer survivor, we selected sperm for functional testing from cryopreserved grafts recovered from under the scrotal skin of monkey 13-008 (table S1). Graft sperm released by mechanical dissection and collagenase IV digestion were suspended separately in human tubal fluid. An aliquot of mechanically dissected sperm was shipped at ambient temperature to the Oregon National Primate Research Center for functional testing by intracytoplasmic sperm injection (ICSI) in May 2017 (table S2). The remaining pre-digest and post-digest sperm were cryopreserved for ICSI experiments performed in October and November 2017. A total of 138 eggs were fertilized by ICSI; 39 of these eggs (28%) cleaved (i.e., progressed to the two-cell stage), and 16 of those 39 (41%) cleavage stage embryos developed into blastocyst stage embryos (Fig. 6, A to F, and table S2).

Fig. 6 Functional evaluation of graft-derived sperm.

Sperm were derived from a cryopreserved graft retrieved from the left scrotum of 13-008 that was recovered 9 months after grafting. Fresh graft-derived sperm were used for the May 2017 ICSI trial. The remaining sperm were cryopreserved and used for the October 2017 and November 2017 trials (table S2). Graft-derived sperm were used to fertilize rhesus oocytes by ICSI (A). The resulting embryos attained two-cell stage by day 1 (B), eight-cell stage by day 2 (C), morula stage by day 6 (D), blastocyst stage by day 10 (E), and hatching blastocyst stage by day 11 in culture (F). Blastocyst embryos were transferred to recipient females and a pregnancy was confirmed by ultrasound on 12 December 2017. Normal fetal development was confirmed by ultrasound on 15 January 2018 (G) and a graft-derived baby (“Grady”) was born by cesarean section on 16 April 2018 (H) (photo from 2-week checkup).

Cleavage (two-cell embryo) rates of 28% after ICSI with graft-derived testicular sperm were lower than in previous experiments using ejaculated sperm in rhesus, but the pace of embryo development (29) and blastocyst development rates were in the expected range (30). The reduced cleavage rate may reflect sperm quality deficits caused by cryopreservation of the testicular tissue before grafting or cryopreservation of graft-derived sperm after graft recovery and before ICSI. However, fresh and frozen testicular sperm produce similar ICSI cleavage rates in the human clinic (31, 32). We observed sperm quality deficits caused by the enzymatic digestion of testicular tissue to retrieve sperm (e.g., detached heads, fragile sperm), and the corresponding cleavage rates in those ICSI trials were 13 and 10%, respectively (table S2, November trial). Thus, mechanical dissection is currently preferred, and future studies are needed to optimize enzymatic digestion protocols. Egg quality deficits have been reported for rhesus cycles performed at the beginning (October) and end (May) of the breeding season (30). In this regard, the ICSI trial that was performed in November with mechanically dissected sperm produced more mature eggs; the cleavage rate was 60%, the blastocyst rate was 67%, and the pregnancy rate was 25% (table S2), which are comparable to rates reported in previous rhesus macaque studies using ejaculated sperm (30).

Pregnancy and live offspring from graft-derived sperm

A total of 11 blastocyst stage embryos (10 fresh and 1 frozen) were transferred into six recipient females (tables S1 and S2). A pregnancy from a fresh blastocyst transfer was confirmed by ultrasound of one recipient female on 15 December 2017 (table S1). A follow-up ultrasound confirmed normal fetal development on 12 January 2018 (Fig. 6G), and a caesarean section was scheduled. Grady was born on 16 April 2018 (Fig. 6H) weighing 471 g and with a fetal/placental weight ratio of 3, which is normal for a near-term female rhesus macaque (3336). Apgar scores were 5 at 1 min, 6 at 5 min, and 7 at 30 min after delivery, which are consistent with previous reports for surgically delivered near-term macaque infants (37). Grady’s 3- and 6-month cage-side behavioral assessments revealed normal social distance behavior with her mother as well as normal social and object play activities (fig. S6).


Autologous testicular tissue grafting is an experimental approach that might be used to produce sperm from human cryopreserved prepubertal testis tissues. This approach was initially described in mice, with the production of sperm from fresh and cryopreserved grafts that were competent to fertilize eggs and produce live offspring (14, 17, 18). Those results have been replicated to some extent in higher primates, but graft survival from autologous cryopreserved primate tissues was low and sperm were not tested functionally by fertilization or with the production of live offspring (2325). These gaps need to be addressed to justify translating autologous testicular tissue grafting to the human clinic (e.g., for childhood cancer survivors).

Our experimental design was similar to previous studies (2325), but a few key differences may be noteworthy. First, we used a lower concentration of dimethyl sulfoxide (DMSO) (5%, 0.7 M) than previous nonhuman primate autograft studies (1.4 M). Differences in DMSO concentration may have implications for tissue toxicity and/or protection from cryopreservation damage. Controlled slow-rate freezing in 0.7 M DMSO is the condition we use in the Fertility Preservation Program at the University of Pittsburgh Medical Center ( and is based on a protocol described by Keros and colleagues (38, 39) for freezing prepubertal human testicular tissues. We did not attempt to optimize freezing conditions in this study but acknowledge that Jahnukainen and colleagues found that survival of rhesus testicular tissue grafts cryopreserved in 1.4 M DMSO was better than that of those cryopreserved in 0.7 M DMSO (40). Complete spermatogenesis was not observed under either freezing condition in that study, perhaps because of the relatively short xenograft incubation time (3 to 5 months). Second, testicular tissue pieces were larger in this study (9 to 20 mm3) than in previous studies (0.5 to 1 mm3). It may seem counterintuitive that larger tissue fragments could vascularize and survive ischemic injury better than smaller tissue fragments, but there may be critical autocrine or paracrine factors in the graft itself that contribute to survival of the tissue. Third, we individually sutured each testicular tissue fragment (four per graft site) to the subcutaneous aspect of the skin rather than depositing a slurry of small pieces in the subcutaneous space. Previous studies indicated that survival of ectopic ovarian tissue xenografts was enhanced by grafting into an angiogenic granulation zone created by injury of the underlying muscle tissue (41). The subcutaneous layer of the skin has a high capillary density (42), and perhaps injury caused by the suture needles and/or skin flap incision was sufficient to promote angiogenic granulation and vascularization of the apposed testicular tissue grafts.

The gold-standard proof of concept for any reproductive technology is the production of functional gametes and live offspring. In this study, we demonstrated that frozen and thawed prepubertal testicular tissue could be matured in vivo by grafting under the back skin or scrotal skin of the same animal to produce functional sperm and a healthy baby. One caveat to our study is that testicular tissues were grafted into castrated animals, which would not usually be the case with prepubertal cancer survivors. We castrated animals in this study because that is the only condition that produced sperm in previous nonhuman primate autologous testicular tissue grafting studies (24, 25). Future studies are needed to confirm that graft development proceeds in a similar fashion in the scrotum or under the skin of individuals with intact testes. A second caveat is that testicular tissues collected before cancer treatment might harbor malignant cells. Therefore, the autologous grafting approach may not be appropriate for childhood leukemia, lymphoma, or testicular cancer survivors. On the other hand, autologous testicular tissue grafting could be appropriate for patients receiving bone marrow transplants for nonmalignant conditions (e.g., β-thalassemia, sickle cell anemia) or patients with solid tumors, including sarcomas and neuroblastomas, that do not metastasize to the testes; >60% of young patients who have frozen their testicular tissues at the Fertility Preservation Program in Pittsburgh and coordinated recruitment sites fall into these two categories (fig. S7).

Testicular tissue grafting and xenografting are extensively tested technologies that have been replicated in numerous mammalian species and may be ready for translation to the human clinic pending demonstration that similar results can be obtained in noncastrated recipients. Complete spermatogenesis from grafted human tissues has not yet been achieved (12). Human tissue studies as well as studies addressing the caveats outlined above are needed to understand the scope, safety, and feasibility of testicular tissue grafting in patients.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

Tables S1 and S2

References (43, 44)

References and Notes

Acknowledgments: We thank the Lab Animal Research Resource (LARR) staff of Magee-Womens Research Institute for animal husbandry and technical support. Histology was performed by the Histology Core of Magee-Womens Research Institute. FSH assays were performed by the Endocrine Technologies Support Core at the Oregon National Primate Research Center. Funding: This work was supported by NICHD grants P01 HD075795 and R01 HD076412 to K.E.O.; a diversity supplement to HD076412 for A.P.F.; NIH P51 OD011092 to the Oregon National Primate Research Center (J.D.H.); and the Magee-Womens Research Institute and Foundation. Author contributions: A.P.F. planned and performed experiments, collected and analyzed data, and drafted the manuscript. K.P. planned and performed experiments and collected data. M.S. performed experiments. H.V.-P. provided fertility preservation patient diagnoses. G.S. and M.L.M. performed rhesus hormone measurements. L.H. and N.R. performed postnatal behavioral and social assessments. C.R., C.H., and J.D.H. planned and performed ICSI experiments and collected and analyzed data. V.R. evaluated the term placenta and placental/fetal weight ratios. I.D. and K.E.O. planned the experiments. K.E.O. performed the experiments, collected data, and wrote the final manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials.

Stay Connected to Science

Navigate This Article