Background
Glucocorticoids (GC's) are the effectors of stress and necessarily impact multiple target tissues during normal function. Regulation of GC levels occurs through the hypothalamo-pituitary-adrenal axis (HPA). Recent evidence indicates that the cytokine, leukemia inhibitory factor (LIF) and its functional binding to the high affinity receptor, a heterodimer between low-affinity LIFR and glycoprotein 130 (LIFR:gp130), are players within the HPA cascade of events. The ligands ciliary neurotrophic factor (CNTF), cardiotrophin-1 (CT-1) and a novel neurotrophin set, known alternatively as NNT-1/BSF3 and CLF/CLC, also function through the high affinity LIFR:gp130 heterodimeric receptor but require a third ligand-binding but non-signaling subunit [
1‐
4]. Oncostatin M (OSM) utilizes LIFR:gp130 as an alternative receptor complex for OSM signaling in humans [
5], but not in mice [
6]. Thus, in mice LIFR:gp130 is obligatory for signaling induced by LIF, CNTF, CT-1 and CLF/CLC. Signaling proceeds through the Janus kinase/signal transducers and activators of transcription (JAK/STAT) or the ras-mitogen-activated protein (MAP) kinase pathways [for a review see [
7]].
LIF and LIFR are functionally expressed in rodent tissues that are central to GC production: the hypothalamus [
8] and pituitary [
9‐
12]. LIF works in synergy with corticotropin-releasing hormone (CRH) in the hypothalamus to increase pituitary pro-opiomelanocortin (POMC) and, ultimately, pituitary adrenocorticotrophin hormone (ACTH) [
9,
11,
13‐
16]. LIF also stimulates POMC gene expression directly in the pituitary [
9,
11,
15,
17]. LIF up-regulates pituitary prohormone convertase (PC1). Subsequently, PC1 goes on to facilitate POMC processing to ACTH [
18] within the pituitary. LIF promotes survival of magnocellular vasopressinergic neurons in the hypothalamus [
19] which ultimately also positively influences ACTH level. Thus, it is well documented in rodents that signaling through the LIFR:gp130 heterodimer in the hypothalamus and pituitary leads to adrenal GC production via elevated ACTH. Other LIF effects that may influence the stress response through actions in the rodent nervous system are a LIF induced increase in acetylcholine [
20] and LIF induced decreases in neuropeptide Y and tyrosine hydroxylase [
21]. LIFR:gp130 signaling has strong effects in systems peripheral to but influenced by the HPA including reproductive, skeletal, nervous, neuromuscular, cardiovascular, hematopoietic, immune and metabolic systems, with influences on both development and adult homeostasis [for reviews see [
7,
22‐
27]].
Animal models of LIF over expression and ablation have confirmed importance in stress regulation. Adult mouse LIF over expression mediated by transfer of LIF over expressing hematopoietic cells results in a lethal, multi-systemic phenotype that includes small adrenals with a loss of the innermost cortical layer [
28]. Pituitary specific LIF over expression leads to a Cushing's Syndrome-like condition of GC hyperactivity [
29,
30]. LIF null adult mice have normal or slightly reduced basal ACTH and GC concentrations, and are unable to respond to acute stress by increased ACTH. Basal POMC is reduced, but inducible upon stress [
16]. Thus, an obligatory function of LIF may be in the processing of POMC to ACTH during acute stress.
Null mutation of LIFR results in neonatal death and defects in many systems including absence of glial cells, osteoporosis, glycogen hyper-synthesis, placental defects [
31] and loss of motor neuronal subsets that cripple suckling ability [
32]. In all of the systems affected as a biological consequence of LIFR loss, an explanation of the mechanism of action can be designed around direct action of LIFR in each system. However, when considered as a whole, direct, non-redundant obligatory LIFR function in such diverse systems makes little evolutionary sense. LIFR function within the stress response system could be a unifying element in many of the diverse biological consequences of LIFR loss. Among the defects seen upon developmental LIFR loss, all but the loss of suckling ability and placental changes mirror reported effects of GC excess in the adult. This study explores the effect of alteration of the GC environment during late gestation, with the expectation that suppression of GC effects will improve the LIFR null newborn phenotype and identify LIFR as an obligatory player in concert with the maternal GC surge during late gestation development. To link LIFR function with GC function we have manipulated the late gestation maternal steroid profile by maternal adrenalectomy or RU486 administration to determine if the maternal GC surge impacts fetal well being in the absence of LIFR and functional LIFR:gp130 signaling. RU486 administration during late gestation results in improved development of bone, skeletal muscle, and glial cells in LIFR null fetuses.
Discussion
The results implicate excess fetal GC response in late gestation as the reason for the multi-systemic phenotypes in the absence of LIFR. Maternal adrenalectomy worsens the osteopenia, while maternal hormone suppression through RU486 improves the presumed GC related phenotypes: osteopenia, skeletal muscle integrity and glial development. This was an improvement over the poor prognosis for
Lifr -/- newborns described initially [
31]. In that study,
Lifr-/- pups rarely survived natural delivery, but when they did, they were excluded from the litter and died at variable times beginning one hour following birth. No pups survived through the night following birth. The survival of
Lifr -/- pups following RU486 treatment was similar to the phenotype seen in CNTFR null mice where neuronal development was disrupted which disallowed suckling [
40]. This indicates that at least some of the problems encountered with neuronal development that affect normal suckling were not ameliorated by RU486. Neither adrenalectomy nor RU486 treatment impact the apparent well-being of the
Lifr +/+ or +/- littermates. Excess GC activity due to LIFR loss is an unexpected result since LIF gain-of-function mice develop a Cushing's-like syndrome [
29,
30]. Thus, if the relationship were simple, loss of LIFR function would predict development of GC insufficiency. The observed GC activity excess suggests that LIF and LIFR are integral to GC regulation, but that the relationship is complex and that alterations in GC response are likely to result from altered balance of the GC response cascade induced both by gain and loss of LIFR function.
There is a maternal GC surge beginning on E16.5, where E0.5 is the day of vaginal plug following mating. A primary function of this surge is to allow development to switch from growth to maturation of a number of systems that will allow the pup to survive on separation from the mother. Lung surfactant synthesis and connective tissue maturation allow the lungs to become distensible and capable of coping with high surface tension. In addition, the maternal GC surge effectively leads to fetal glycogen accumulation in the liver for energy demands at birth, increased fetal bioavailable tri-iodothyronine from thyroxine to allow for an increased metabolic rate and thermogenesis required at birth, maturation of the fetal small gut to prepare for digestion, heightened fetal adrenal medullar catecholamine release for control of the above processes, maturation of fetal kidney function and a switch to bone marrow hematopoiesis from the fetal liver [reviewed by [
41]]. The timing of this maternal surge suggests that LIFR function is required at this point in development as a moderator of the consequences of excess GC level.
Mock adrenal surgeries were not performed on the mice. Therefore, we do not know if stress of surgery alone is enough to impact well-being in LIFR null mice. Progesterone was administered in peanut oil following the same regimen as RU486, but resulted in bone loss reminiscent of the results following adrenalectomy (unpublished observation). Thus, peanut oil alone is not responsible for the improvements seen in the RU486 group and the effect of excess progesterone provides further evidence that the LIFR phenotype can worsen through endocrine manipulation.
We previously reported that placental integrity is disrupted at the LIFR null maternal-fetal interface [
31]. In that study, the maternal response to disorganized fetal placental tissue was sufficient to protect the fetus from malnutrition, since there was no detectable disruption of fetal erythropoiesis. However, what this meant for relative transport of RU486 or maternal mediators of fetal hormone balance upon adrenalectomy is unknown. Maternal adrenalectomy on E16.5 was a naïve approach toward blocking the maternal GC surge. It likely had the effect of increasing maternal ACTH levels [
42] that would, in turn, act on the fetal adrenal to synthesize corticosterone. This may have been the source of the modestly elevated corticosterone level seen among the
Lifr null pups in the adrenalectomy group. Although anecdotal, it would appear the RU486 exposed wildtype animal cleared residual late gestation corticosterone following birth more effectively than did the
Lifr -/- littermate. Previous studies measuring circulating insulin and glucose levels in C-section derived E18.5 pups bore no apparent correlation to
Lifr genotype (unpublished observation). At the time it appeared that plasma measurements in pups immediately following C-section were an unreliable reflection of individual condition and were likely to be a reflection of maternal levels. An accurate picture of corticosterone levels will probably require separation from the mother for several hours.
RU486 was first explored as a GR antagonist and subsequently found to be both a GR and progesterone receptor (PR) antagonist, with mild antagonistic effects on the androgen receptor. In certain cases it can function as a mild agonist. RU486 can bind to the hormone-binding domain of both GR and PR. Although still able to localize to the nucleus, it prevents subsequent transcription of GC and progesterone responsive genes. It does not directly antagonize the mineralocorticoid receptor (MR) [reviewed by [
43]].
Because RU486, along with antagonizing the GR antagonizes the PR, some of the RU486 benefit may be due to antagonism of progesterone effects. Progesterone is integral to the hormonal milieu. It is processed to GC's through the actions of 21 hydroxylase and 11 β-hydroxylase [
44] and it is an anti-GC [
45‐
49] and an anti-mineralocorticoid (MC) [
50‐
53]. Therefore, PR antagonism by RU486 could increase GC and MC effects, although this is the opposite of the phenotype seen. Progesterone interacts synergistically with estrogen in bone remodeling [
54], while GC excess is implicated in osteoporosis. LIFR is known to be important in the balance between bone resorption and bone formation acting directly on osteoblasts [
55‐
57]. Bone loss in the untreated control
Lifr -/- pups suggests a heightened response to GC rather than a heightened response to the bone-remodeling effects of progesterone. Whereas, protection by RU486 implies antagonism of GR mediated bone loss, rather than antagonism of PR mediated bone remodeling. In addition, a progesterone receptor knockout mouse model affects only female mice with reported effects limited to organs of reproduction causing anovulation, uterine hyperplasia and inflammation, limited mammary development and impaired sexual behavior [
58]. There was no noticeable difference in phenotype in the
Lifr -/- pups due to gender following any of the treatments (data not shown). Since apparent defects due to LIFR loss arise during the late gestation GC surge, while maternal progesterone levels are declining in anticipation of birth and because phenotypes that have been attributed in the literature to GC excess are lessened through RU486 and the apparent heightened GC response in
Lifr -/- mice affects both sexes equally, it is likely that RU486 attenuation of associated phenotypes is primarily mediated through GR antagonism. However, the interconnectedness of hormone action may also include RU486 mediated PR antagonism as a means of impacting the extent of GR-related effects.
The primary clinical utility of RU486 is PR antagonism to induce early pregnancy abortion. Since the data were collected from pups that were born early by natural delivery (E18.5 versus E19.5–21.5) following RU486 treatment of pregnant females, it can be assumed that the RU486 used was an active PR antagonist, and, by association, an active GR antagonist. The effective dose used in this study was somewhere between 0.1 and 0.2 μg/gram/day for two days. The dose was not administered by weight since litters vary in size and will have a large impact on maternal weight in late gestation. 0.3 μg/gram/day has been reported to induce abortion in 5% of mice during early gestation following 3 days of administration [
59]. The dosage used in this study was able to induce pre-term birth by natural delivery at a time when the pups were able to survive ex utero (E18.5), but was well below the dose used to achieve clinical abortion.
The apparent GC hyper-responsiveness in LIFR null pups is probably not directly due to excess circulating corticosterone in the absence of LIFR. Instead, expression of GR and subsequent GR induced transcription may be more relevant than circulating GC level. Another consideration is that the benefit of RU486 administration may be either central within the HPA or can be due to peripheral GR antagonism. Cardiac dysfunction has not yet been explored in the LIFR null model. However, many of the defects caused by gp130 loss are shared by mice with a null mutation of LIFR, reflecting the heterodimeric relationship of LIFR and gp130 for functional LIFR:gp130 signaling. A recent cardiac muscle specific gp130 knockout model develops normally [
60]. Whereas, non-tissue specific gp130 knockout results in hypotrophic cardiac muscle apparent by E15.5 [
61]. This suggests that the effect of gp130 knockout on cardiac development is mediated outside of cardiac tissue and the primary developmental defect is peripheral to the heart. Given the improvement in skeletal muscle noted in this study following RU486, it would be reasonable to explore a central HPA defect following gp130 loss as the mediator of hypotrophic cardiac development.
Muscle regeneration begins through the muscle precursor satellite cells. Activation is followed by proliferation and fusion with other satellite cells to form new myotubes. Muscle regeneration is a dynamic process necessary in the maintenance of muscle integrity. LIF and CNTF stimulate muscle regeneration in vivo [
62,
63], as do other growth factors; such as insulin-like growth factor [
64,
65]. The power of LIF to regenerate muscle is seen in dystrophin null mdx mice, where exogenous LIF regenerates atrophied diaphragm muscle [
66]. LIF induced signaling appears to be essential in muscle development as seen by muscle atrophy in the absence of LIFR on E18.5. In the absence of LIFR, muscle atrophy may be caused by GC induced alteration of metabolism, which leads to glycogen accumulation, inhibition of protein synthesis and stimulation of protein degradation [
67,
68]. Because low levels of RU486 can fully protect
Lifr -/- fetal muscle from the late gestation GC surge, the balance between hormone function and muscle integrity appears both delicate and direct.
GR expression is specific but widespread within neuronal and glial cell populations. LIF is also responsible for increased GFAP expression [
69], while signaling through LIFR:gp130 is critical in fostering the differentiation of neuronal precursors into astrocytes [
70] mediated through STAT3 [
71‐
74]. Glial cells respond to GC, which, in turn, can affect transcriptional control of GFAP level either positively or negatively [
74]. The present study shows that GC's can play a critical role in hindering astrocyte development in the absence of LIFR as seen through the partial abrogation of this defect following exposure to RU486 in
Lifr -/- mice. RU486 crosses the blood-brain-barrier and is present at only 28% of levels seen in the serum [
75]. Perhaps glial development would be further improved in the presence of higher levels of RU486 were these levels not abortive. The RU486 effect is not likely to be mediated through PR in the
Lifr -/- astrocyte precursors, since astrocytes have low levels of PR's that are only detectable in females [
76].
The vigor displayed at birth by RU486 treated
Lifr -/- pups is striking and suggests sweeping improvement in well-being. Pups of all
Lifr genotypes are able to breathe well at birth and
Lifr +/+ and +/- pups thrive with no apparent lung handicap. This indicates that the improvement due to RU486 is implemented through modest alteration in GC regulation since complete loss of GR function leads to incomplete lung development at birth and impaired survival through atelectasis [
77]. The neural compartment may have been under-protected by RU486 due to partial exclusion by the blood-brain-barrier, while bone and skeletal muscle, tissues exposed to higher levels of RU486, were clearly protected. Hypothalamus and anterior pituitary are also protected by the blood-brain-barrier. Elevation of ACTH in response to RU486 was apparent. Consequently, very low levels of RU486 appear to mediate partial neural normalization. Loss of neuronal subpopulations integral to suckling need to be explored in the RU486 late gestation pups, but the data suggest that the health of this population is not directly influenced by altered GC response.
Authors' contributions
CBW conceived of the study and participated in all aspects including coordination, mouse husbandry, tissue collection, genotyping, histology, immunohistochemistry and data analysis. AMN assisted with mouse husbandry and tissue collection. DL provided pathology expertise and valued discussion. All authors read and approved the final manuscript.