- Email: [email protected]
Inhibition of vascular endothelial growth factor during the postovulatory period prevents pregnancy in the marmoset
Contraception 82 (2010) 572 – 578
Original research article
Inhibition of vascular endothelial growth factor during the postovulatory period prevents pregnancy in the marmoset☆ Hamish M. Frasera,⁎, Keith D. Morrisa , Stanley J. Wiegandb , Helen Wilsona a
Medical Research Council Human Reproductive Sciences Unit, The Queen's Medical Research Institute, EH16 4TJ Edinburgh, UK b Regeneron Pharmaceuticals, Tarrytown, New York, NY 10591, USA Received 11 February 2010; revised 1 April 2010; accepted 16 April 2010
Abstract Background: This study investigated the effects of inhibition of vascular endothelial growth factor (VEGF) during the first postpartum cycle in marmosets housed with a fertile male where a 90% fertility rate is normal. Methods: On resumption of mating, females were treated with either 25 mg/kg aflibercept, a potent VEGF inhibitor, or control Fc protein (n=6 per group) at the time of ovulation. Effects on timing of pregnancy were monitored by measuring plasma progesterone, chorionic gonadotropin (CG) and uterine palpation. Results: In five of six Fc-treated controls, the postpartum rise in progesterone was maintained and followed by a sustained rise in CG by Day 30 posttreatment indicating pregnancy. In all six aflibercept-treated animals, progesterone secretion was suppressed in the treatment cycle and a CG rise did not occur by Day 30. Pregnancy was delayed to the next cycle, significantly extending interbirth interval compared to controls. Posttreatment deliveries and infant development were normal. Conclusion: These results show that stringent pharmacological inhibition of VEGF suppresses luteal progesterone and prevents the successful establishment of pregnancy. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved. Keywords: Corpus luteum; VEGF inhibitor; Progesterone; Pregnancy
1. Introduction Angiogenesis, the formation of new blood vessels via endothelial replication, has been shown to be intense during the first few days of the life span of the corpus luteum (CL) in all species studied, including humans [1]. Angiogenesis is primarily under the regulation of vascular endothelial growth factor (VEGF) [2]. VEGF belongs to a gene family that includes VEGF-A, B, C, D and placenta-derived growth factor (PlGF), of which VEGF-A is the principal form regulating physiological angiogenesis [2]. VEGF action is mediated via two tyrosine kinase receptors, VEGFR1 and VEGFR2. Because of its central role in angiogenesis in health and disease, a large number of compounds have been ☆
Research was supported by the core grant (U.1276.00.002.00003.01) to Medical Research Council Reproductive Sciences Unit to HMF. ⁎ Corresponding author. Tel.: +44 0 131 242 6222; fax: +44 0 131 242 6197. E-mail address: [email protected] (H.M. Fraser).
developed to target VEGF or its receptors [2], which may be used as tools to define its physiological role. We have employed aflibercept (previously known as VEGF Trap), a recombinant chimeric protein comprising portions of the extracellular domains of the human VEGFR 1 and 2 expressed in sequence with the Fc portion of human immunoglobulin [3]. Aflibercept binds all isoforms of VEGF-A, as well as PlGF, with very high affinity, preventing it from binding to and activating VEGF receptors [3]. Studies in which VEGF has been inhibited during the formation of the CL in the marmoset monkey show that angiogenesis is severely suppressed and the resulting CL is largely avascular and nonfunctional [1,4]. Furthermore, inhibition of VEGF suppresses plasma progesterone secretion at all stages of the luteal phase, irrespective of the rate of angiogenesis, indicating an additional action on ovarian function, most likely inhibition of ovarian vascular permeability [4,5]. VEGF is produced by the early CL and is maintained during the establishment of early pregnancy [6–8]. These
0010-7824/$ – see front matter. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.contraception.2010.04.020
H.M. Fraser et al. / Contraception 82 (2010) 572–578
findings suggest that compromise of CL function, secondary to inhibition of angiogenesis and/or vascular permeability, could impair fertility. We have now directly evaluated the effect of transient VEGF inhibition on the establishment of pregnancy in marmosets to test the hypothesis that this would lead to failure in the normal rise of plasma progesterone and result in prevention of pregnancy that was reversible on resumption of normal ovulatory cycles. The marmoset offers major advantages for this study as it has an exceptionally high rate of fertility; breeding female marmosets housed in stable family groups ovulate around 11 days postpartum with a 90% fertility rate [6,9]. In addition, the cellular and molecular regulation of angiogenesis within the ovary and endometrium during the normal reproductive cycle and early pregnancy has been studied extensively in this species [1,6].
2. Materials 2.1. Animals Experiments were carried out under the Animals Act, Scientific Proceedings (1986), and were approved by the local ethical review committee. Common marmosets (Callithrix jacchus) were housed in stable family groups with males of proven fertility. Female marmosets selected for the study had a history of previously giving birth to live young, then reliably becoming pregnant within 2 weeks postpartum on at least the two consecutive occasions immediately prior to recruitment. Interbirth intervals were approximately 155 days. Given a gestation period of 144 days, this indicated that ovulation and a fertile mating occurred approximately 11 days after giving birth. Marmosets meeting these criteria were identified over a 2-year period. To confirm that animals were mating prior to and at the time of expected ovulation, a vaginal lavage was collected from each marmoset on Days 7, 9 and 11 postpartum and examined for motile sperm. The vaginal wash was placed on a slide and examined for the presence of sperm using dark field microscopy. Sperm numbers were scored on a + to +++ basis. 2.2. Treatment To block VEGF, we employed aflibercept (Regeneron Pharmaceuticals, New York) [3]. A single injection of aflibercept (25 mg/kg, s.c.) effectively inhibited VEGF for approximately 10 days in the marmoset, with maximal aflibercept concentrations in the circulation of ∼100 mg/L reached at 31 h, with an elimination half-life of 59 h [5]. Therefore, in the present study, marmosets were treated with aflibercept (25 mg/kg s.c.) or a control protein (recombinant human Fc), n=6 per group, at Day 12 postpartum during the immediate postovulatory period, to block the normal luteal progesterone secretion. Blood samples were collected beginning on Day 9 postpartum using a refined restraint device [10] and
573
continued three times per week until pregnancy was confirmed. Pregnancy was first indicated by a continued elevation of progesterone beyond the 20 days typical of the normal luteal phase, with subsequent presence of circulating chorionic gonadotropin (CG) levels consistently greater than 20 ng/mL. Pregnancy was confirmed by manual palpation at 6 and 10 weeks posttreatment. The day of delivery, total number of offspring and surviving offspring were recorded for each pregnancy. Interbirth intervals were calculated by designating the day of birth prior to treatment as Day 1. Each female in this study was breastfeeding twins from their prior pregnancy at the time of treatment (controls: eight females, four males; treated six females, six males). All mothers were observed to continue to suckle their infants in an apparently normal fashion throughout the treatment period. Infants began taking solid food by 40 days of age and weaning was complete by around 60 days. Body weights of the infants were recorded at approximately 10, 20, 40, 60, 80, and 100 days of age. There is no difference between growth rates between male and female infants [11], so data from both sexes were combined for analysis. Offspring conceived following the treatment cycle also were carefully monitored for any abnormalities and were then weighed routinely at intervals. Body weights were compared to growth curves recorded for infants in the colony during the previous 5 years. At 2–3 years of age, when the females had become adult, blood samples were collected three times a week for at least 2 months and plasma progesterone levels assayed to determine the occurrence of ovulatory cycles. 2.3. Assays Plasma concentrations of progesterone [4] and free aflibercept (aflibercept not already bound to endogenous VEGF) were measured as described previously [5]. Samples for aflibercept ELISA were diluted 1:10,000 where highest concentrations of aflibercept were present and assayed neat in remaining samples. Detection limit of the assay was 1 mg/L. The marmoset does not produce luteinizing hormone (LH) as they do not express the β subunit of LH but instead express the CG β subunit. Marmoset CG was measured using a heterologous RIA [12]. HCG (code no. 75/533) was supplied by the National Institute of Biological Standards and Control (Potter's Bar, Hertfordshire) and was used for radioiodination and as standard. The monoclonal antibody to bovine LH (518 B7) was kindly gifted by Dr. J.F. Rosner. An antibody to the LH alpha chain recognizes the marmoset homologue that is common to CG. The detection limit of the CG assay was 6 ng/mL; all samples were run in a single assay. 2.4. Statistical analyses Fisher's exact test was used to compare the number of first postpartum cycle pregnancies between treated and control groups. With a 90% fertility rate in the first postpartum cycle being expected in controls, a size calculation was carried out assuming complete inhibition
574
H.M. Fraser et al. / Contraception 82 (2010) 572–578
of conception in the aflibercept-treated group. This showed that the minimum number required for statistical significance was five per group, assuming one of five controls failed to become pregnant. Student's t test was employed to examine time to sustained rise (N20 days) in progesterone from day of pretreatment delivery, time to consistent rise in CG from date of ovulation and interbirth intervals. Two-way ANOVA was employed to compare changes in body weight over time, followed by Bonferroni post hoc test. To determine differences in body weight gain per day between weighing intervals, Student's t test was used. Differences were considered to be significant at pb0.05.
3. Results Mating was confirmed by the presence of motile sperm in the vaginal wash in all the marmosets selected for the study. In Fc-treated control marmosets, progesterone increases indicative of ovulation were confirmed to occur at the
expected time in all six animals. Establishment of pregnancy during this first cycle postpartum was observed in five of the six animals, this being associated with a sustained rise in progesterone, followed by a rise in plasma CG to values consistently N20 ng/mL (Fig. 1). Interbirth interval was 155 ±2 (mean±SEM) days in four of the five animals while one Fc-treated control animal miscarried at Day 72 (Table 1). In the remaining Fc-treated control marmoset, progesterone levels were not sustained during the first postpartum cycle, but a second timely ovulation occurred on Day 30 posttreatment and was followed by pregnancy, with an interbirth interval of 173 days. In the aflibercept-treated animals, plasma concentrations of unbound aflibercept were highest at the first sample day then declined steadily, reaching nondetectable levels between Days 10 and 12 postinjection (Fig. 1). All aflibercepttreated marmosets ovulated postpartum as indicated by plasma progesterone N32 nmol/L at the time of treatment. However, none of the treated animals became pregnant during the first postpartum cycle. VEGF inhibition was associated with a failure to sustain a normal luteal phase rise in progesterone, levels being significantly lower than in control animals (b0.05) (Fig. 1). In four marmosets, treatment with aflibercept was followed by a marked and sustained suppression of progesterone. In these animals, the second cycle postpartum cycle was characterized by the occurrence of a preovulatory gonadotrophin surge 20–22 days posttreatment, followed by sustained rises in progesterone and CG similar to those observed in the first postpartum cycle in Fc-treated controls (Fig. 1). The interbirth interval in these animals was 180±2 days, or around 25 days longer than Fc-treated control animals that Table 1 Effects of treatment with control Fc or aflibercept during the first postpartum cycle in marmosets, on occurrence of pregnancy and comparison of interbirth intervals prior to and after treatment
Fig. 1. Effect of treatment with Fc, control (25 mg/kg, s.c.) (open circles) or aflibercept (25 mg/kg, s.c.) (closed circles) at the early luteal phase on plasma progesterone and CG concentrations in the marmoset. Data are centered around the start of treatment (arrow). In controls, the first rise in progesterone is sustained and followed by a rise in CG indicating the cycle was fertile. In treated marmosets, the first ovulatory rise in progesterone is suppressed. The bottom panel shows pharmacokinetics of free aflibercept in plasma. Data are means±SEM (n=5 per group for controls and n=4 treated). The remaining control marmoset ovulated but did not become pregnant until the second cycle postpartum (data not shown).
Animal Pregnancy Interbirth intervals code cycle Pretreatment Treatment
Posttreatment
Control 4W 26Y 25Y 30Y 95Y 15W
Fc First First First First First Second
156 158 152 160 160 151
155 149 158 157 Miscarriage at 72 days 173
155 157 153 NA 157 NA
Aflibercept 853Ra Second 998R Second 994R Second 10W Second 852Ra Second 122Y Second
156 152 153 155 157 150
189 178 180 185 Miscarriage at 50 days 178
155 NA 164 NA NA 154
Letters in the first column are part of the animal identification number. NA, information not available. a These animals showed a partial recovery of progesterone during the treatment cycle.
H.M. Fraser et al. / Contraception 82 (2010) 572–578
became pregnant during the first postpartum cycle, and carried their pregnancies to term (Table 1). In the remaining two marmosets, progesterone levels were also initially suppressed by aflibercept treatment, but began to rise again around Day 12, before falling to follicular phase levels at the end of the treatment cycle, approximately 10–12 days later. These marmosets had the highest progesterone levels at onset of treatment indicating that the CL was relatively more mature. In these animals, a subsequent ovulation occurred around Day 30 posttreatment and was followed by a sustained rise in progesterone and CG (Fig. 2). In one marmoset, pregnancy continued to term, with an interbirth interval of 189 days. In the remaining marmoset, conception also occurred in the second postpartum cycle, as evidenced by sustained plasma progesterone and increased CG, but plasma progesterone fell to follicular phase values at 50 days after ovulation indicating miscarriage, as was also observed in one of the six controls. The subsequent cycle was fertile, resulting in an interbirth interval of 260 days. The number of pregnancies achieved in the first postpartum cycle was significantly less in aflibercept-treated animals (0/6) compared to Fc-treated controls (5/6) (Fisher's
Fig. 2. Plasma progesterone, CG and aflibercept concentrations representing one of two postpartum marmosets treated with aflibercept that exhibited partial recovery of progesterone secretion during the treatment cycle. Data from control marmosets (n=5; open circles) are shown for comparison.
575
exact test; b0.05). Significant delays in time to onset of the sustained rise in plasma CG (b0.005) and progesterone (b0.005) also were observed in the treated marmosets, and interbirth intervals in aflibercept-treated marmosets were significantly longer (pb.0001) than for the Fc-treated controls. Interbirth intervals for subsequent pregnancies were not available for all the marmoset mothers in the study, but for those for whom data were available, the period was within the normal range for both aflibercept-treated animals and Fctreated controls (Table 1). During the study period, infants from the prior pregnancy remained with their mothers and appeared to breastfeed normally. However, body weights were slightly but significantly lower (b0.01) (two-way ANOVA) in breastfeeding infants from the aflibercept-treated group, although the values remained well within the normal range for the colony (Fig. 3A). This attributable to a reduction in the rate of weight gain during the treatment period: body weight gain per day between Days 10–20 postpartum was significantly lower (b0.01) in infants feeding from
Fig. 3. (A) Body weights in infants breastfeeding from mothers receiving Fc (25 mg/kg, s.c.) (open circles, n=12) or aflibercept (25 mg/kg, s.c.) (closed circles, n=11), and the 10% and 90% percentiles (broken line) from 45 normal infants in the colony. Body weight was overall significantly lower in infants feeding from treated mothers. (B) Body weight gain per day over the time intervals normally assigned to monitor development of infants. The 10to 20-day period corresponds to that in which mothers were treated with aflibercept on Day 12. Values show a significantly lower weight gain in infants feeding from treated mothers between Days 10–20 and 20–40. Data are means±SEM. *pb.05. **pb.01.
576
H.M. Fraser et al. / Contraception 82 (2010) 572–578
aflibercept-treated mothers; weight gain between Days 20 and 40 was less affected, although still significantly lower (pb0.05), while weight gained between Days 40 and 60 was comparable in treated and control groups (Fig. 3B). When the females that were nursing during the treatment period became adult, measurement of plasma progesterone concentrations showed that the offspring of both Fc-treated controls and aflibercept-treated mothers exhibited normal ovulatory cycles. By the end of the study period, five Fc-treated control and five aflibercept-treated marmosets had carried their first postpartum pregnancies to term and delivered normally. A total of 14 infants were delivered in each group (Fc-treated controls, one set of twins and four sets of triplets, eight females, six males; aflibercept-treated, one set of twins and four sets of triplets, seven females, seven males). One of the infants in the aflibercept-treated group was a stillborn. As marmosets can only successfully nurse twins, it is standard husbandry practice to euthanize the weakest triplet. Thereafter, infants in both groups developed normally and had body weights within the normal range. At 100 days of age, offspring from Fc-treated control and aflibercept-treated animals were of similar weights (156±8.6 and 148±5.9 g, respectively, means±SEM). One female from an aflibercepttreated mother was euthanized at 200 days of age as a result of weight loss. This was considered by the primate facility manager to be the result of inadequate rearing by the mother and not related to the treatment. When adult, the females born to both treated and control mothers were used in subsequent experiments; plasma progesterone concentrations demonstrating normal ovulatory cycles were observed in all animals from both groups.
4. Discussion This study has demonstrated for the first time in a primate species that stringent pharmacological inhibition of VEGF during the early luteal phase prevents pregnancy. This effect of treatment was rapidly reversible, with the next ovulation occurring 10–20 days following clearance of unbound aflibercept from the circulation and five of the six aflibercept-treated marmosets becoming pregnant at this time. The treatment had no detectable adverse effects on the outcome of pregnancy as healthy infants were delivered and appeared to develop normally after gestations of characteristic length. We have shown previously that the period of intense early luteal angiogenesis is VEGF-dependent, and its inhibition prevents the normal rise in plasma progesterone [1,4]. The progesterone produced by the normal CL functions to transform the proliferative endometrium into a secretory state that renders the uterus receptive to implantation of the embryo. The results of the present study show that stringent VEGF inhibition throughout the first half of the luteal phase precludes the possibility of pregnancy.
In addition to inhibiting luteal angiogenesis and function, follicular development is blocked while unbound aflibercept remains present in the circulation at levels above ∼1 mg/L [1,5,13]. Thus, the posttreatment recovery period must encompass a follicular phase preceding ovulation, ensuring that conception in the recovery cycle is not associated with appreciable exposure to unbound aflibercept. Indeed, in four of the six treated marmosets, the posttreatment ovulation occurred some 20 days after injection of aflibercept. Since the follicular phase in the marmoset lasts 10 days, the treatment was effective in blocking ovarian function for at least 10 days in accordance with plasma levels of aflibercept, which were detectable only up until Days 10–12 postinjection. While VEGF inhibition resulted in a complete functional luteolysis in these four animals, in the remaining two marmosets after the initial 10-day period of suppressed plasma progesterone, levels began to rise again towards values within the range of the normal luteal phase. The most likely interpretation was that this was the result of reactivation of the CL. Despite this recovery, there was still a failure of pregnancy. The partial reactivation of progesterone secretion lasted for approximately 10 days, extending the luteal phase to its normal 20-day duration such that the next ovulation and pregnancy occurred 30 days after treatment. This suggests that the observed partial recovery of progesterone secretion was sufficient to exert negative feedback upon the pituitary gland, delaying follicular development until luteolysis. It was also of note that plasma progesterone levels at onset of treatment were highest in the animals that exhibited a partial recovery of luteal function. This suggests that when luteal angiogenesis and transformation are more advanced, there is a greater capacity for recovery of luteal function once VEGF inhibition ceases. Inhibition of VEGF not only inhibits ongoing luteal and follicular angiogenesis, but it may also act on extant vessels to decrease permeability to macromolecules thereby restricting entry of trophic factors such as CG into the ovarian compartments [14]. Indeed, aflibercept administration in the mid-luteal phase, that is, the postangiogenic period, results in a decline in plasma progesterone secretion [5]. At this same stage of the cycle, VEGF inhibition may also act upon the endometrium to inhibit both angiogenesis and vascular permeability during the proliferative phase in nonhuman primates [15–17]. Thus, a likely dual effect of decreased progesterone and inhibition of endometrial VEGF would result in an endometrium unprepared for implantation that occurs 11 days postovulation in the marmoset [18,19]. It is unlikely with the current treatment schedule that aflibercept inhibited pregnancy by a direct antiimplantation action as it had already cleared from the circulation by this time. In mice, postovulatory injection of immunoneutralizing antibodies to VEGF receptor 2 administered before or after implantation blocks pregnancy. These effects appear to be mediated by direct action on the CL [20] as progesterone replacement maintained pregnancy in antibody-treated ovariectomized animals. Similarly, we have found that
H.M. Fraser et al. / Contraception 82 (2010) 572–578
aflibercept markedly inhibits luteal function when given at the mid-luteal, postangiogenic, phase of the cycle both in marmosets [5] and macaques [21]. However, it has also been reported that treatment of macaques with a neutralizing antibody to VEGF around the time of implantation can reduce pregnancy rate without significantly reducing serum progesterone concentrations, suggesting a primary inhibitory effect on the endometrium [22]. The most likely explanation for this apparent discrepancy is that the anti-VEGF employed failed to completely neutralize endogenous VEGF but was sufficient to affect implantation. Further studies will be required to determine the relative contributions of ovarian and endometrial VEGF to the establishment of pregnancy in primates. Transient inhibition of VEGF in nursing marmosets (Day 12–22 postpartum) had no major deleterious effects on their breastfeeding infants. However, by the end of the treatment period, the infants of aflibercept-treated mothers weighed some 10% less than the infants of mothers receiving the Fc control. Shortly following clearance of aflibercept from the maternal blood, nursing infants again began to gain weight at a normal rate, but mean weights remained slightly below those of controls to Day 100 after birth. The most likely explanation for this finding is that VEGF inhibition affected the quality or quantity of breast milk available to the nursing infants. Deletion of VEGF in the mammary epithelial cells of the mouse compromised postpartum mammary gland development and function, such that the pups of VEGF mutant mothers failed to thrive and attained body weights 50% less than control mice over the period of lactation [23]. Lactating glands of the VEGF-deficient mice exhibited a 50% reduction in vessel density compared to controls and postpartum lobuloalveolar expansion was incomplete. Thus, the reduction in weight gain observed during the treatment period in infants suckling from aflibercept-treated mothers most likely reflects deficiencies in production or composition of breast milk secondary to changes in mammary capillary density and/or permeability. However, possible changes in mammary gland structure and milk quantity and composition were not directly assessed in the present protocol, and a specific analysis of the effects of VEGF inhibition on mammary structure and milk production in nonhuman primates is warranted. In conclusion, using a potent inhibitor of VEGF, we have demonstrated not only that VEGF is essential for normal progesterone secretion by the primate CL but also that VEGF inhibition results in failure of pregnancy that was rapidly reversible following the clearance of free aflibercept from the blood. That previous studies in the marmoset [6] and rhesus macaque [22] failed to consistently prevent pregnancy using neutralizing antibodies to VEGF likely reflects incomplete inhibition of endogenous VEGF in these experiments. In neither of the above studies nor in the present study has VEGF inhibition prior to implantation been associated with significant deleterious effects on offspring of subsequent
577
pregnancies. This appears to be attributable to the fact that resumption of follicular development and ovulatory cycles is not possible as long as VEGF bioactivity remains substantially inhibited, reducing the potential for exposure of the newly conceived embryo to pharmacologically significant levels of VEGF inhibitors following the termination of treatment. Acknowledgments We thank I. Swanston for assays; Regeneron Pharmaceuticals, NY, for the gift of aflibercept; Dr. J. Roser for the LH antibody; the National institute of Biological Standards for the CG; and the staff at the R.V. Short Building for animal care. References [1] Fraser HM, Duncan WC. Regulation and manipulation of angiogenesis in the ovary and endometrium. Reprod Fertil Dev 2009;21:377–92. [2] Ferrara N. Vascular endothelial growth factor: basic science and clinical progress. Endocr Rev 2000;25:581–611. [3] Holash J, Davis S, Papadopoulos N, et al. VEGF-Trap: a VEGF blocker with potent antitumor effects. Proc Nat Acad Sci U S A 2002; 99:11393–8. [4] Wulff C, Wilson H, Rudge JS, Wiegand SJ, Lunn SF, Fraser HM. Luteal angiogenesis: prevention and intervention by treatment with vascular endothelial growth factor trap A40. J Clin Endocrinol Metab 2001;86:3377–86. [5] Fraser HM, Wilson H, Wulff C, Rudge JS, Wiegand SJ. Administration of vascular endothelial growth factor trap in the ‘post angiogenic’ period of the luteal phase causes rapid functional luteolysis and endothelial cell death in the marmoset. Reproduction 2006;132: 589–600. [6] Rowe AJ, Morris KD, Bicknell R, Fraser HM. Angiogenesis in the corpus luteum of early pregnancy in the marmoset and the effects of vascular endothelial growth factor immunoneutralization on establishment of pregnancy. Biol Reprod 2002;67:1180–8. [7] Sugino N, Kashida S, Takiguchi S, Karube A, Kato H. Expression of vascular endothelial growth factor and its receptors in the human corpus luteum during the menstrual cycle and in early pregnancy. J Clin Endocrinol Metab 2000;85:3919–24. [8] Wulff C, Dickson SE, Duncan WC, Fraser HM. Angiogenesis in the human corpus luteum: simulated early pregnancy by HCG treatment is associated with both angiogenesis and vessel stabilisation. Hum Reprod 2001;16:2515–24. [9] Windle CP, Baker HF, Ridley RM, Oerke AK, Martin RD. Unrearable litters and prenatal reduction of litter size in the common marmoset (Callithrix jacchus). J Med Primatol 1999;28:73–83. [10] Greig I, Morris KD, Mathiesen E, Mathiesen R, Buchanan-Smith HM. An improved restraint device for injections and collection of samples from marmosets. Lab Prim News 2006;45:1–5. [11] Abbott DH, Hearn JP. Physical, hormonal and behavioural aspects of sexual development in the marmoset monkey Callithrix jacchus. J Reprod Fertil 1978;53:155–66. [12] Saltzman W, Schultz-Darken NJ, Wegner FH, Wittwer DJ, Abbott DH. Suppression of cortisol in subordinate female marmosets: reproductive and social contributions. Horm Behav 1998;33:58–74. [13] Taylor PD, Wilson H, Hillier SG, Wiegand SJ, Fraser HM. Effects of inhibition of vascular endothelial growth factor at time of selection on follicular angiogenesis, expansion, development, and atresia in the marmoset. Mol Hum Reprod 2007;13:729–36.
578
H.M. Fraser et al. / Contraception 82 (2010) 572–578
[14] Zimmermann RC, Hartman T, Kavic S, et al. Vascular endothelial growth factor receptor 2-mediated angiogenesis is essential for gonadotropin-dependent follicle development. J Clin Invest 2003; 112:659–69. [15] Aberdeen GW, Wiegand SJ, Bonagura TWJ, Pepe GJ, Albrecht ED. Vascular endothelial growth factor mediates the estrogen-induced breakdown of tight junctions between and increase in proliferation of microvessel endothelial cells in the baboon endometrium. Endocrinology 2008;149:6076–83. [16] Fan X, Krieg S, Kuo CJ, et al. VEGF blockade inhibits angiogenesis and reepithelialization of endometrium. FASEB J 2008;22:3571–80. [17] Fraser HM, Wilson H, Silvestri A, Morris KD, Wiegand SJ. The role of vascular endothelial growth factor and estradiol in the regulation of endometrial angiogenesis and cell proliferation in the marmoset. Endocrinology 2008;149:4413–20. [18] Niklaus AL, Murphy CR, Lopata A. Characteristics of the uterine luminal surface epithelium at preovulatory and preimplantation stages in the marmoset monkey. Anat Rec 2000;264:82–92.
[19] Rowe AJ, Wulff C, Fraser HM. Angiogenesis and microvascular development in the marmoset (Callithrix jacchus) endometrium during early pregnancy. Reproduction 2004;128:107–16. [20] Pauli SA, Tang H, Wang J, et al. The vascular endothelial growth factor (VEGF)/VEGF receptor pathway is critical for blood vessel survival in corpora lutea of pregnancy in the rodent. Endocrinology 2005;146:1301–11. [21] Fraser HM, Wilson H, Morris KD, Swanston I, Rudge JS, Wiegand SJ. Vascular endothelial growth factor trap suppresses ovarian function at all stages of the luteal phase in the macaque. J Clin Endocrinol Metab 2005;90:5811–8. [22] Sengupta J, Lalitkumar PG, Najwa AR, et al. Immunoneutralization of vascular endothelial growth factor inhibits pregnancy establishment in the rhesus monkey (Macaca mulatta). Reproduction 2007;133: 1199–211. [23] Rossiter H, Barresi C, Ghannadan M, et al. Inactivation of VEGF in mammary gland epithelium severely compromises mammary gland development and function. FASEB J 2007;21:3994–4004.
Original research article
Inhibition of vascular endothelial growth factor during the postovulatory period prevents pregnancy in the marmoset☆ Hamish M. Frasera,⁎, Keith D. Morrisa , Stanley J. Wiegandb , Helen Wilsona a
Medical Research Council Human Reproductive Sciences Unit, The Queen's Medical Research Institute, EH16 4TJ Edinburgh, UK b Regeneron Pharmaceuticals, Tarrytown, New York, NY 10591, USA Received 11 February 2010; revised 1 April 2010; accepted 16 April 2010
Abstract Background: This study investigated the effects of inhibition of vascular endothelial growth factor (VEGF) during the first postpartum cycle in marmosets housed with a fertile male where a 90% fertility rate is normal. Methods: On resumption of mating, females were treated with either 25 mg/kg aflibercept, a potent VEGF inhibitor, or control Fc protein (n=6 per group) at the time of ovulation. Effects on timing of pregnancy were monitored by measuring plasma progesterone, chorionic gonadotropin (CG) and uterine palpation. Results: In five of six Fc-treated controls, the postpartum rise in progesterone was maintained and followed by a sustained rise in CG by Day 30 posttreatment indicating pregnancy. In all six aflibercept-treated animals, progesterone secretion was suppressed in the treatment cycle and a CG rise did not occur by Day 30. Pregnancy was delayed to the next cycle, significantly extending interbirth interval compared to controls. Posttreatment deliveries and infant development were normal. Conclusion: These results show that stringent pharmacological inhibition of VEGF suppresses luteal progesterone and prevents the successful establishment of pregnancy. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved. Keywords: Corpus luteum; VEGF inhibitor; Progesterone; Pregnancy
1. Introduction Angiogenesis, the formation of new blood vessels via endothelial replication, has been shown to be intense during the first few days of the life span of the corpus luteum (CL) in all species studied, including humans [1]. Angiogenesis is primarily under the regulation of vascular endothelial growth factor (VEGF) [2]. VEGF belongs to a gene family that includes VEGF-A, B, C, D and placenta-derived growth factor (PlGF), of which VEGF-A is the principal form regulating physiological angiogenesis [2]. VEGF action is mediated via two tyrosine kinase receptors, VEGFR1 and VEGFR2. Because of its central role in angiogenesis in health and disease, a large number of compounds have been ☆
Research was supported by the core grant (U.1276.00.002.00003.01) to Medical Research Council Reproductive Sciences Unit to HMF. ⁎ Corresponding author. Tel.: +44 0 131 242 6222; fax: +44 0 131 242 6197. E-mail address: [email protected] (H.M. Fraser).
developed to target VEGF or its receptors [2], which may be used as tools to define its physiological role. We have employed aflibercept (previously known as VEGF Trap), a recombinant chimeric protein comprising portions of the extracellular domains of the human VEGFR 1 and 2 expressed in sequence with the Fc portion of human immunoglobulin [3]. Aflibercept binds all isoforms of VEGF-A, as well as PlGF, with very high affinity, preventing it from binding to and activating VEGF receptors [3]. Studies in which VEGF has been inhibited during the formation of the CL in the marmoset monkey show that angiogenesis is severely suppressed and the resulting CL is largely avascular and nonfunctional [1,4]. Furthermore, inhibition of VEGF suppresses plasma progesterone secretion at all stages of the luteal phase, irrespective of the rate of angiogenesis, indicating an additional action on ovarian function, most likely inhibition of ovarian vascular permeability [4,5]. VEGF is produced by the early CL and is maintained during the establishment of early pregnancy [6–8]. These
0010-7824/$ – see front matter. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.contraception.2010.04.020
H.M. Fraser et al. / Contraception 82 (2010) 572–578
findings suggest that compromise of CL function, secondary to inhibition of angiogenesis and/or vascular permeability, could impair fertility. We have now directly evaluated the effect of transient VEGF inhibition on the establishment of pregnancy in marmosets to test the hypothesis that this would lead to failure in the normal rise of plasma progesterone and result in prevention of pregnancy that was reversible on resumption of normal ovulatory cycles. The marmoset offers major advantages for this study as it has an exceptionally high rate of fertility; breeding female marmosets housed in stable family groups ovulate around 11 days postpartum with a 90% fertility rate [6,9]. In addition, the cellular and molecular regulation of angiogenesis within the ovary and endometrium during the normal reproductive cycle and early pregnancy has been studied extensively in this species [1,6].
2. Materials 2.1. Animals Experiments were carried out under the Animals Act, Scientific Proceedings (1986), and were approved by the local ethical review committee. Common marmosets (Callithrix jacchus) were housed in stable family groups with males of proven fertility. Female marmosets selected for the study had a history of previously giving birth to live young, then reliably becoming pregnant within 2 weeks postpartum on at least the two consecutive occasions immediately prior to recruitment. Interbirth intervals were approximately 155 days. Given a gestation period of 144 days, this indicated that ovulation and a fertile mating occurred approximately 11 days after giving birth. Marmosets meeting these criteria were identified over a 2-year period. To confirm that animals were mating prior to and at the time of expected ovulation, a vaginal lavage was collected from each marmoset on Days 7, 9 and 11 postpartum and examined for motile sperm. The vaginal wash was placed on a slide and examined for the presence of sperm using dark field microscopy. Sperm numbers were scored on a + to +++ basis. 2.2. Treatment To block VEGF, we employed aflibercept (Regeneron Pharmaceuticals, New York) [3]. A single injection of aflibercept (25 mg/kg, s.c.) effectively inhibited VEGF for approximately 10 days in the marmoset, with maximal aflibercept concentrations in the circulation of ∼100 mg/L reached at 31 h, with an elimination half-life of 59 h [5]. Therefore, in the present study, marmosets were treated with aflibercept (25 mg/kg s.c.) or a control protein (recombinant human Fc), n=6 per group, at Day 12 postpartum during the immediate postovulatory period, to block the normal luteal progesterone secretion. Blood samples were collected beginning on Day 9 postpartum using a refined restraint device [10] and
573
continued three times per week until pregnancy was confirmed. Pregnancy was first indicated by a continued elevation of progesterone beyond the 20 days typical of the normal luteal phase, with subsequent presence of circulating chorionic gonadotropin (CG) levels consistently greater than 20 ng/mL. Pregnancy was confirmed by manual palpation at 6 and 10 weeks posttreatment. The day of delivery, total number of offspring and surviving offspring were recorded for each pregnancy. Interbirth intervals were calculated by designating the day of birth prior to treatment as Day 1. Each female in this study was breastfeeding twins from their prior pregnancy at the time of treatment (controls: eight females, four males; treated six females, six males). All mothers were observed to continue to suckle their infants in an apparently normal fashion throughout the treatment period. Infants began taking solid food by 40 days of age and weaning was complete by around 60 days. Body weights of the infants were recorded at approximately 10, 20, 40, 60, 80, and 100 days of age. There is no difference between growth rates between male and female infants [11], so data from both sexes were combined for analysis. Offspring conceived following the treatment cycle also were carefully monitored for any abnormalities and were then weighed routinely at intervals. Body weights were compared to growth curves recorded for infants in the colony during the previous 5 years. At 2–3 years of age, when the females had become adult, blood samples were collected three times a week for at least 2 months and plasma progesterone levels assayed to determine the occurrence of ovulatory cycles. 2.3. Assays Plasma concentrations of progesterone [4] and free aflibercept (aflibercept not already bound to endogenous VEGF) were measured as described previously [5]. Samples for aflibercept ELISA were diluted 1:10,000 where highest concentrations of aflibercept were present and assayed neat in remaining samples. Detection limit of the assay was 1 mg/L. The marmoset does not produce luteinizing hormone (LH) as they do not express the β subunit of LH but instead express the CG β subunit. Marmoset CG was measured using a heterologous RIA [12]. HCG (code no. 75/533) was supplied by the National Institute of Biological Standards and Control (Potter's Bar, Hertfordshire) and was used for radioiodination and as standard. The monoclonal antibody to bovine LH (518 B7) was kindly gifted by Dr. J.F. Rosner. An antibody to the LH alpha chain recognizes the marmoset homologue that is common to CG. The detection limit of the CG assay was 6 ng/mL; all samples were run in a single assay. 2.4. Statistical analyses Fisher's exact test was used to compare the number of first postpartum cycle pregnancies between treated and control groups. With a 90% fertility rate in the first postpartum cycle being expected in controls, a size calculation was carried out assuming complete inhibition
574
H.M. Fraser et al. / Contraception 82 (2010) 572–578
of conception in the aflibercept-treated group. This showed that the minimum number required for statistical significance was five per group, assuming one of five controls failed to become pregnant. Student's t test was employed to examine time to sustained rise (N20 days) in progesterone from day of pretreatment delivery, time to consistent rise in CG from date of ovulation and interbirth intervals. Two-way ANOVA was employed to compare changes in body weight over time, followed by Bonferroni post hoc test. To determine differences in body weight gain per day between weighing intervals, Student's t test was used. Differences were considered to be significant at pb0.05.
3. Results Mating was confirmed by the presence of motile sperm in the vaginal wash in all the marmosets selected for the study. In Fc-treated control marmosets, progesterone increases indicative of ovulation were confirmed to occur at the
expected time in all six animals. Establishment of pregnancy during this first cycle postpartum was observed in five of the six animals, this being associated with a sustained rise in progesterone, followed by a rise in plasma CG to values consistently N20 ng/mL (Fig. 1). Interbirth interval was 155 ±2 (mean±SEM) days in four of the five animals while one Fc-treated control animal miscarried at Day 72 (Table 1). In the remaining Fc-treated control marmoset, progesterone levels were not sustained during the first postpartum cycle, but a second timely ovulation occurred on Day 30 posttreatment and was followed by pregnancy, with an interbirth interval of 173 days. In the aflibercept-treated animals, plasma concentrations of unbound aflibercept were highest at the first sample day then declined steadily, reaching nondetectable levels between Days 10 and 12 postinjection (Fig. 1). All aflibercepttreated marmosets ovulated postpartum as indicated by plasma progesterone N32 nmol/L at the time of treatment. However, none of the treated animals became pregnant during the first postpartum cycle. VEGF inhibition was associated with a failure to sustain a normal luteal phase rise in progesterone, levels being significantly lower than in control animals (b0.05) (Fig. 1). In four marmosets, treatment with aflibercept was followed by a marked and sustained suppression of progesterone. In these animals, the second cycle postpartum cycle was characterized by the occurrence of a preovulatory gonadotrophin surge 20–22 days posttreatment, followed by sustained rises in progesterone and CG similar to those observed in the first postpartum cycle in Fc-treated controls (Fig. 1). The interbirth interval in these animals was 180±2 days, or around 25 days longer than Fc-treated control animals that Table 1 Effects of treatment with control Fc or aflibercept during the first postpartum cycle in marmosets, on occurrence of pregnancy and comparison of interbirth intervals prior to and after treatment
Fig. 1. Effect of treatment with Fc, control (25 mg/kg, s.c.) (open circles) or aflibercept (25 mg/kg, s.c.) (closed circles) at the early luteal phase on plasma progesterone and CG concentrations in the marmoset. Data are centered around the start of treatment (arrow). In controls, the first rise in progesterone is sustained and followed by a rise in CG indicating the cycle was fertile. In treated marmosets, the first ovulatory rise in progesterone is suppressed. The bottom panel shows pharmacokinetics of free aflibercept in plasma. Data are means±SEM (n=5 per group for controls and n=4 treated). The remaining control marmoset ovulated but did not become pregnant until the second cycle postpartum (data not shown).
Animal Pregnancy Interbirth intervals code cycle Pretreatment Treatment
Posttreatment
Control 4W 26Y 25Y 30Y 95Y 15W
Fc First First First First First Second
156 158 152 160 160 151
155 149 158 157 Miscarriage at 72 days 173
155 157 153 NA 157 NA
Aflibercept 853Ra Second 998R Second 994R Second 10W Second 852Ra Second 122Y Second
156 152 153 155 157 150
189 178 180 185 Miscarriage at 50 days 178
155 NA 164 NA NA 154
Letters in the first column are part of the animal identification number. NA, information not available. a These animals showed a partial recovery of progesterone during the treatment cycle.
H.M. Fraser et al. / Contraception 82 (2010) 572–578
became pregnant during the first postpartum cycle, and carried their pregnancies to term (Table 1). In the remaining two marmosets, progesterone levels were also initially suppressed by aflibercept treatment, but began to rise again around Day 12, before falling to follicular phase levels at the end of the treatment cycle, approximately 10–12 days later. These marmosets had the highest progesterone levels at onset of treatment indicating that the CL was relatively more mature. In these animals, a subsequent ovulation occurred around Day 30 posttreatment and was followed by a sustained rise in progesterone and CG (Fig. 2). In one marmoset, pregnancy continued to term, with an interbirth interval of 189 days. In the remaining marmoset, conception also occurred in the second postpartum cycle, as evidenced by sustained plasma progesterone and increased CG, but plasma progesterone fell to follicular phase values at 50 days after ovulation indicating miscarriage, as was also observed in one of the six controls. The subsequent cycle was fertile, resulting in an interbirth interval of 260 days. The number of pregnancies achieved in the first postpartum cycle was significantly less in aflibercept-treated animals (0/6) compared to Fc-treated controls (5/6) (Fisher's
Fig. 2. Plasma progesterone, CG and aflibercept concentrations representing one of two postpartum marmosets treated with aflibercept that exhibited partial recovery of progesterone secretion during the treatment cycle. Data from control marmosets (n=5; open circles) are shown for comparison.
575
exact test; b0.05). Significant delays in time to onset of the sustained rise in plasma CG (b0.005) and progesterone (b0.005) also were observed in the treated marmosets, and interbirth intervals in aflibercept-treated marmosets were significantly longer (pb.0001) than for the Fc-treated controls. Interbirth intervals for subsequent pregnancies were not available for all the marmoset mothers in the study, but for those for whom data were available, the period was within the normal range for both aflibercept-treated animals and Fctreated controls (Table 1). During the study period, infants from the prior pregnancy remained with their mothers and appeared to breastfeed normally. However, body weights were slightly but significantly lower (b0.01) (two-way ANOVA) in breastfeeding infants from the aflibercept-treated group, although the values remained well within the normal range for the colony (Fig. 3A). This attributable to a reduction in the rate of weight gain during the treatment period: body weight gain per day between Days 10–20 postpartum was significantly lower (b0.01) in infants feeding from
Fig. 3. (A) Body weights in infants breastfeeding from mothers receiving Fc (25 mg/kg, s.c.) (open circles, n=12) or aflibercept (25 mg/kg, s.c.) (closed circles, n=11), and the 10% and 90% percentiles (broken line) from 45 normal infants in the colony. Body weight was overall significantly lower in infants feeding from treated mothers. (B) Body weight gain per day over the time intervals normally assigned to monitor development of infants. The 10to 20-day period corresponds to that in which mothers were treated with aflibercept on Day 12. Values show a significantly lower weight gain in infants feeding from treated mothers between Days 10–20 and 20–40. Data are means±SEM. *pb.05. **pb.01.
576
H.M. Fraser et al. / Contraception 82 (2010) 572–578
aflibercept-treated mothers; weight gain between Days 20 and 40 was less affected, although still significantly lower (pb0.05), while weight gained between Days 40 and 60 was comparable in treated and control groups (Fig. 3B). When the females that were nursing during the treatment period became adult, measurement of plasma progesterone concentrations showed that the offspring of both Fc-treated controls and aflibercept-treated mothers exhibited normal ovulatory cycles. By the end of the study period, five Fc-treated control and five aflibercept-treated marmosets had carried their first postpartum pregnancies to term and delivered normally. A total of 14 infants were delivered in each group (Fc-treated controls, one set of twins and four sets of triplets, eight females, six males; aflibercept-treated, one set of twins and four sets of triplets, seven females, seven males). One of the infants in the aflibercept-treated group was a stillborn. As marmosets can only successfully nurse twins, it is standard husbandry practice to euthanize the weakest triplet. Thereafter, infants in both groups developed normally and had body weights within the normal range. At 100 days of age, offspring from Fc-treated control and aflibercept-treated animals were of similar weights (156±8.6 and 148±5.9 g, respectively, means±SEM). One female from an aflibercepttreated mother was euthanized at 200 days of age as a result of weight loss. This was considered by the primate facility manager to be the result of inadequate rearing by the mother and not related to the treatment. When adult, the females born to both treated and control mothers were used in subsequent experiments; plasma progesterone concentrations demonstrating normal ovulatory cycles were observed in all animals from both groups.
4. Discussion This study has demonstrated for the first time in a primate species that stringent pharmacological inhibition of VEGF during the early luteal phase prevents pregnancy. This effect of treatment was rapidly reversible, with the next ovulation occurring 10–20 days following clearance of unbound aflibercept from the circulation and five of the six aflibercept-treated marmosets becoming pregnant at this time. The treatment had no detectable adverse effects on the outcome of pregnancy as healthy infants were delivered and appeared to develop normally after gestations of characteristic length. We have shown previously that the period of intense early luteal angiogenesis is VEGF-dependent, and its inhibition prevents the normal rise in plasma progesterone [1,4]. The progesterone produced by the normal CL functions to transform the proliferative endometrium into a secretory state that renders the uterus receptive to implantation of the embryo. The results of the present study show that stringent VEGF inhibition throughout the first half of the luteal phase precludes the possibility of pregnancy.
In addition to inhibiting luteal angiogenesis and function, follicular development is blocked while unbound aflibercept remains present in the circulation at levels above ∼1 mg/L [1,5,13]. Thus, the posttreatment recovery period must encompass a follicular phase preceding ovulation, ensuring that conception in the recovery cycle is not associated with appreciable exposure to unbound aflibercept. Indeed, in four of the six treated marmosets, the posttreatment ovulation occurred some 20 days after injection of aflibercept. Since the follicular phase in the marmoset lasts 10 days, the treatment was effective in blocking ovarian function for at least 10 days in accordance with plasma levels of aflibercept, which were detectable only up until Days 10–12 postinjection. While VEGF inhibition resulted in a complete functional luteolysis in these four animals, in the remaining two marmosets after the initial 10-day period of suppressed plasma progesterone, levels began to rise again towards values within the range of the normal luteal phase. The most likely interpretation was that this was the result of reactivation of the CL. Despite this recovery, there was still a failure of pregnancy. The partial reactivation of progesterone secretion lasted for approximately 10 days, extending the luteal phase to its normal 20-day duration such that the next ovulation and pregnancy occurred 30 days after treatment. This suggests that the observed partial recovery of progesterone secretion was sufficient to exert negative feedback upon the pituitary gland, delaying follicular development until luteolysis. It was also of note that plasma progesterone levels at onset of treatment were highest in the animals that exhibited a partial recovery of luteal function. This suggests that when luteal angiogenesis and transformation are more advanced, there is a greater capacity for recovery of luteal function once VEGF inhibition ceases. Inhibition of VEGF not only inhibits ongoing luteal and follicular angiogenesis, but it may also act on extant vessels to decrease permeability to macromolecules thereby restricting entry of trophic factors such as CG into the ovarian compartments [14]. Indeed, aflibercept administration in the mid-luteal phase, that is, the postangiogenic period, results in a decline in plasma progesterone secretion [5]. At this same stage of the cycle, VEGF inhibition may also act upon the endometrium to inhibit both angiogenesis and vascular permeability during the proliferative phase in nonhuman primates [15–17]. Thus, a likely dual effect of decreased progesterone and inhibition of endometrial VEGF would result in an endometrium unprepared for implantation that occurs 11 days postovulation in the marmoset [18,19]. It is unlikely with the current treatment schedule that aflibercept inhibited pregnancy by a direct antiimplantation action as it had already cleared from the circulation by this time. In mice, postovulatory injection of immunoneutralizing antibodies to VEGF receptor 2 administered before or after implantation blocks pregnancy. These effects appear to be mediated by direct action on the CL [20] as progesterone replacement maintained pregnancy in antibody-treated ovariectomized animals. Similarly, we have found that
H.M. Fraser et al. / Contraception 82 (2010) 572–578
aflibercept markedly inhibits luteal function when given at the mid-luteal, postangiogenic, phase of the cycle both in marmosets [5] and macaques [21]. However, it has also been reported that treatment of macaques with a neutralizing antibody to VEGF around the time of implantation can reduce pregnancy rate without significantly reducing serum progesterone concentrations, suggesting a primary inhibitory effect on the endometrium [22]. The most likely explanation for this apparent discrepancy is that the anti-VEGF employed failed to completely neutralize endogenous VEGF but was sufficient to affect implantation. Further studies will be required to determine the relative contributions of ovarian and endometrial VEGF to the establishment of pregnancy in primates. Transient inhibition of VEGF in nursing marmosets (Day 12–22 postpartum) had no major deleterious effects on their breastfeeding infants. However, by the end of the treatment period, the infants of aflibercept-treated mothers weighed some 10% less than the infants of mothers receiving the Fc control. Shortly following clearance of aflibercept from the maternal blood, nursing infants again began to gain weight at a normal rate, but mean weights remained slightly below those of controls to Day 100 after birth. The most likely explanation for this finding is that VEGF inhibition affected the quality or quantity of breast milk available to the nursing infants. Deletion of VEGF in the mammary epithelial cells of the mouse compromised postpartum mammary gland development and function, such that the pups of VEGF mutant mothers failed to thrive and attained body weights 50% less than control mice over the period of lactation [23]. Lactating glands of the VEGF-deficient mice exhibited a 50% reduction in vessel density compared to controls and postpartum lobuloalveolar expansion was incomplete. Thus, the reduction in weight gain observed during the treatment period in infants suckling from aflibercept-treated mothers most likely reflects deficiencies in production or composition of breast milk secondary to changes in mammary capillary density and/or permeability. However, possible changes in mammary gland structure and milk quantity and composition were not directly assessed in the present protocol, and a specific analysis of the effects of VEGF inhibition on mammary structure and milk production in nonhuman primates is warranted. In conclusion, using a potent inhibitor of VEGF, we have demonstrated not only that VEGF is essential for normal progesterone secretion by the primate CL but also that VEGF inhibition results in failure of pregnancy that was rapidly reversible following the clearance of free aflibercept from the blood. That previous studies in the marmoset [6] and rhesus macaque [22] failed to consistently prevent pregnancy using neutralizing antibodies to VEGF likely reflects incomplete inhibition of endogenous VEGF in these experiments. In neither of the above studies nor in the present study has VEGF inhibition prior to implantation been associated with significant deleterious effects on offspring of subsequent
577
pregnancies. This appears to be attributable to the fact that resumption of follicular development and ovulatory cycles is not possible as long as VEGF bioactivity remains substantially inhibited, reducing the potential for exposure of the newly conceived embryo to pharmacologically significant levels of VEGF inhibitors following the termination of treatment. Acknowledgments We thank I. Swanston for assays; Regeneron Pharmaceuticals, NY, for the gift of aflibercept; Dr. J. Roser for the LH antibody; the National institute of Biological Standards for the CG; and the staff at the R.V. Short Building for animal care. References [1] Fraser HM, Duncan WC. Regulation and manipulation of angiogenesis in the ovary and endometrium. Reprod Fertil Dev 2009;21:377–92. [2] Ferrara N. Vascular endothelial growth factor: basic science and clinical progress. Endocr Rev 2000;25:581–611. [3] Holash J, Davis S, Papadopoulos N, et al. VEGF-Trap: a VEGF blocker with potent antitumor effects. Proc Nat Acad Sci U S A 2002; 99:11393–8. [4] Wulff C, Wilson H, Rudge JS, Wiegand SJ, Lunn SF, Fraser HM. Luteal angiogenesis: prevention and intervention by treatment with vascular endothelial growth factor trap A40. J Clin Endocrinol Metab 2001;86:3377–86. [5] Fraser HM, Wilson H, Wulff C, Rudge JS, Wiegand SJ. Administration of vascular endothelial growth factor trap in the ‘post angiogenic’ period of the luteal phase causes rapid functional luteolysis and endothelial cell death in the marmoset. Reproduction 2006;132: 589–600. [6] Rowe AJ, Morris KD, Bicknell R, Fraser HM. Angiogenesis in the corpus luteum of early pregnancy in the marmoset and the effects of vascular endothelial growth factor immunoneutralization on establishment of pregnancy. Biol Reprod 2002;67:1180–8. [7] Sugino N, Kashida S, Takiguchi S, Karube A, Kato H. Expression of vascular endothelial growth factor and its receptors in the human corpus luteum during the menstrual cycle and in early pregnancy. J Clin Endocrinol Metab 2000;85:3919–24. [8] Wulff C, Dickson SE, Duncan WC, Fraser HM. Angiogenesis in the human corpus luteum: simulated early pregnancy by HCG treatment is associated with both angiogenesis and vessel stabilisation. Hum Reprod 2001;16:2515–24. [9] Windle CP, Baker HF, Ridley RM, Oerke AK, Martin RD. Unrearable litters and prenatal reduction of litter size in the common marmoset (Callithrix jacchus). J Med Primatol 1999;28:73–83. [10] Greig I, Morris KD, Mathiesen E, Mathiesen R, Buchanan-Smith HM. An improved restraint device for injections and collection of samples from marmosets. Lab Prim News 2006;45:1–5. [11] Abbott DH, Hearn JP. Physical, hormonal and behavioural aspects of sexual development in the marmoset monkey Callithrix jacchus. J Reprod Fertil 1978;53:155–66. [12] Saltzman W, Schultz-Darken NJ, Wegner FH, Wittwer DJ, Abbott DH. Suppression of cortisol in subordinate female marmosets: reproductive and social contributions. Horm Behav 1998;33:58–74. [13] Taylor PD, Wilson H, Hillier SG, Wiegand SJ, Fraser HM. Effects of inhibition of vascular endothelial growth factor at time of selection on follicular angiogenesis, expansion, development, and atresia in the marmoset. Mol Hum Reprod 2007;13:729–36.
578
H.M. Fraser et al. / Contraception 82 (2010) 572–578
[14] Zimmermann RC, Hartman T, Kavic S, et al. Vascular endothelial growth factor receptor 2-mediated angiogenesis is essential for gonadotropin-dependent follicle development. J Clin Invest 2003; 112:659–69. [15] Aberdeen GW, Wiegand SJ, Bonagura TWJ, Pepe GJ, Albrecht ED. Vascular endothelial growth factor mediates the estrogen-induced breakdown of tight junctions between and increase in proliferation of microvessel endothelial cells in the baboon endometrium. Endocrinology 2008;149:6076–83. [16] Fan X, Krieg S, Kuo CJ, et al. VEGF blockade inhibits angiogenesis and reepithelialization of endometrium. FASEB J 2008;22:3571–80. [17] Fraser HM, Wilson H, Silvestri A, Morris KD, Wiegand SJ. The role of vascular endothelial growth factor and estradiol in the regulation of endometrial angiogenesis and cell proliferation in the marmoset. Endocrinology 2008;149:4413–20. [18] Niklaus AL, Murphy CR, Lopata A. Characteristics of the uterine luminal surface epithelium at preovulatory and preimplantation stages in the marmoset monkey. Anat Rec 2000;264:82–92.
[19] Rowe AJ, Wulff C, Fraser HM. Angiogenesis and microvascular development in the marmoset (Callithrix jacchus) endometrium during early pregnancy. Reproduction 2004;128:107–16. [20] Pauli SA, Tang H, Wang J, et al. The vascular endothelial growth factor (VEGF)/VEGF receptor pathway is critical for blood vessel survival in corpora lutea of pregnancy in the rodent. Endocrinology 2005;146:1301–11. [21] Fraser HM, Wilson H, Morris KD, Swanston I, Rudge JS, Wiegand SJ. Vascular endothelial growth factor trap suppresses ovarian function at all stages of the luteal phase in the macaque. J Clin Endocrinol Metab 2005;90:5811–8. [22] Sengupta J, Lalitkumar PG, Najwa AR, et al. Immunoneutralization of vascular endothelial growth factor inhibits pregnancy establishment in the rhesus monkey (Macaca mulatta). Reproduction 2007;133: 1199–211. [23] Rossiter H, Barresi C, Ghannadan M, et al. Inactivation of VEGF in mammary gland epithelium severely compromises mammary gland development and function. FASEB J 2007;21:3994–4004.