Autotaxin as a novel, tissue-remodeling-related factor in regressing corpora lutea of cycling rats



Autotaxin (ATX) generates lysophosphatidic acid (LPA) from glycerophospholipid via lysophospholipase D (lysoPLD) activity in cooperation with phospholipase A. We studied its expression and possible functional roles in the ovary of nonfertile cycling rats. Immunohistochemistry revealed that ATX was located predominantly in luteal steroidogenic cells of corpora lutea (CL), but not in any follicles. ATX expression was modest in the newest generation of CL and augmented in older generations undergoing structural regression. ATX expression in the whole ovary and lysoPLD activity in circulating blood did not alter during the estrous cycle. Among the LPA receptors examined (LPA1–4), LPA4 was densely present on migratory cells, probably phagocytes, at degenerative foci within regressing CL. Bolus administration of anti-ATX IgG or LPA into ovarian bursa in vivo had little effect on the apoptotic cell death of luteal cells, as evaluated by cleaved caspase 3 expression, but led to altered numbers of neutrophils and macrophages in regressing CL, as evaluated by immunological detection of each cell marker. These treatments, together with bromodeoxy uridine, revealed a stimulatory effect of the ATX/LPA pathway on fibroblast proliferation in regressing CL. The results indicate that ATX is increasingly expressed by structurally regressing CL and has definite local action on phagocyte recruitment and fibroblast proliferation which are responsible for tissue remodeling.


3β-hydroxysteroid dehydrogenase






corpora lutea


equine chorionic gonadotropin


group IVA phospholipase A2


human chorionic gonadotropin


LPA receptor subtypes 1–6


lysophosphatidic acid




lysophospholipase D






phospholipase A


Tris-buffered saline


In the ovaries of nonfertile cycling rats and mice (e.g. Wistar–Imamichi rats), 5–10 mature follicles release ova in response to an ovulatory stimulus and the remaining tissues transform into corpora lutea (CL) every 4 days [1, 2]. CL show structural growth involving the differentiation and proliferation of steroidogenic cells, vasculogenesis and vascularization within a few days. However, when experimental rats and mice do not receive copulation stimulus during the periovulatory period, the formed CL do not gain active progesterone-secreting potency and are destined to undergo structural involution [1, 2]. The former process is assumed to be distinct from ordinary functional regression that is triggered in many species by uterine-derived prostaglandin F, a primary eicosanoid of first-generation lipid mediators [1-4]. The latter process, the so-called ‘structural regression’ of CL, has features somewhat analogous to tissue injury and repair [1-3] and is composed of: (a) the death of CL-constituent cells, primarily luteal steroidogenic and capillary endothelial cells; (b) phagocytotic elimination of dead cells by neutrophils and macrophages; (c) proliferation of residential fibroblasts and filling of the cavity formed after removal of dead cells; and (d) involution of fibrous tissue and scar formation.

To date, intensive studies have focused on the molecular and cellular mechanisms of prostaglandin F-triggered early critical events, which are the inhibition of progesterone secretion and the death of luteal and endothelial cells [1-4]. Therefore, these mechanisms are relatively well understood. However, the mechanisms by which late luteolytic events occur during structural regression have received less attention and are largely unclear [2, 3].

Autotaxin (ATX) is a secreted protein with lysophospholipase D (lysoPLD) activity [5, 6]. This activity metabolizes lysophospholipids, a product of the action of phospholipase A (PLA) on glycerophospholipid, to lysophosphatidic acid (LPA) [7]. LPA belongs to the second-generation lipid mediators whose physiologic and pathologic significance are becoming known [4-6]. LPA acts on a wide range of biological processes, including reproduction, through binding to its specific receptors, LPA1–6 [6, 8, 9]. Our understanding of the generation and roles of ATX and LPA in reproductive tissues in humans and experimental animals is advancing, but remains incomplete [8, 9]. In this article, we show spatially and temporally regulated expression of ATX in cycling rat ovary and its local, but not systemic, roles in regulating phagocytes and fibroblasts within CL tissue. Here, ATX is documented as a novel regulator of tissue demise and remodeling during CL regression that is critical for the structural homeostasis of the organ.


Localization of ATX and other related factors in the ovary

First, we used immunohistochemistry to examine the expression and localization of ATX within the ovary of cycling rats. The immunoreactivity of ATX localized predominantly to CL and interstitial tissue in ovaries harvested on estrous morning (Fig. 1A). It was absent in developing or atretic follicles. Signal intensities in the CL were variable, probably depending on CL generation. In younger CL generations, the signal was slight in seemingly intact sites and almost negative in degenerative/inflammatory sites (Figs 1A and 2B). In older generations of CL undergoing resolution of degenerative/inflammatory reactions and replacement with fibroblasts, the signal was more intense and uniformly distributed. Detailed observation of regressing CL using high-power magnification revealed that large and small luteal cells contained intense signals, whereas fibroblasts and capillary endothelial cells had faint or no signal (Fig. 1B). Preabsorption of the antibody by the blocking peptide reduced specific immunoreactions almost completely, warranting the immunological detection of ATX (Fig. 1C). To investigate the relationship between the extent of ATX expression and the age (generation) of CL, we evaluated the ATX signal semiquantitatively (grade 0–3) using simultaneous classification of the stage (1–4, from the newest to the oldest generations) of CL. There was clearly a progressive increase in ATX signal intensity with CL age (Fig. 1D). However, we found little estrous stage-dependent alteration in general ATX immunoreactivity in CL (data not shown).

Figure 1.

Immunoreactive ATX is expressed predominantly by luteal steroidogenic cells in regressing CL in cycling rats and augments with CL aging. (A) ATX immunostaining of the cycling rat ovary. ATX immunoreactivity is absent in follicles (F) and present in CL at variable intensities depending on the stage of aging, as denoted by the numbers 1–4. (B) In regressing CL, ATX immunoreactivity is intense in large (image_n/febs12565-gra-0001.png) and small (image_n/febs12565-gra-0002.png) luteal cells, faintly positive in fibroblasts (image_n/febs12565-gra-0003.png), and almost negative in capillary endothelial cells (image_n/febs12565-gra-0004.png). (C) Preabsorbed antibody diminished the specific localization of ATX immunoreactivity in the ovary. (D) Staining intensity levels between 0 (negative) and 3 (clear heavy staining) were semiquantitatively evaluated subjectively. Results represent the mean ± SEM (n = 17–64 CL from eight ovarian specimens in the slide).

Figure 2.

Immunohistochemical demonstration of spatial associations between ATX and luteal factors in regressing CL. Serial sections of the cycling rat ovary were immunostained with nonimmune IgG (A) and for ATX (B), GIVA PLA2 (C) and 3β-HSD (D). Positive signals for each antigen were similarly localized to relatively intact regions of CL and interstitial tissues, and there was good association between ATX with other factors. The scale bar in (A) is also applicable to (B)–(D).

To explore the possible relationship between ATX and other luteal factors, the expression of group IVA PLA2 (GIVA PLA2) and 3β-hydroxysteroid dehydrogenase (3β-HSD) was investigated in serial CL sections (stage 2) that contained both degenerated/inflamed foci and seemingly intact sites (Fig. 2). A specimen with normal mouse IgG application served as a pan-negative control (Fig. 2A). ATX signal was evident in intact sites of CL and interstitial tissue, but was rare in degenerated sites (Fig. 2B). The GIVA PLA2 signal was generally weak but appreciable in CL and interstitial tissues (Fig. 2C). This staining behavior was consistent with our original description [10, 11]. A 3β-HSD immunoreactive signal was also present in CL, interstitial tissues and the follicular thecal layer (Fig. 2D). Immunoreactions of steroidogenic acute regulatory protein, another critical steroidogenic factor, were also present in intact luteal cells and interstitial cells (data not shown). Thus, signals of phospholipid-metabolizing enzymes (GIVA PLA2 and ATX) and steroidogenic factors (3β-HSD and steroidogenic acute regulatory protein) showed profound spatial overlap in the luteal cells of aged CL.

Next, to determine the basis of possible local actions of ATK/LPA and to further develop reports on mouse ovarian expression of LPA receptor LPA1–4 mRNAs [12, 13], we specified the cellular localization of LPA1–4 proteins using subtype-specific antibodies. LPA1 and LPA2 signals were faint, but significant in relatively intact regions of regressing CL (Fig. 3A,B). The LPA3 signal was negative within the CL, but was very intense in the smooth muscle cells of arterioles and venules surrounding the CL and in the stroma (Fig. 3C,D). The distribution of LPA4 signals within regressing CL was the most characteristic of the receptor subtypes examined. Its presence was specific to the focus of degeneration/inflammation, but almost absent in seemingly intact sites (Fig. 3E), in contrast to ATX and GIVA PLA2 signals (Fig. 2B,C). Using greater magnification, the signals were found to be in the blood-derived migratory cells, and not in residing steroidogenic cells or fibroblasts (Fig. 3F).

Figure 3.

Immunohistochemical demonstration of LPA1–4 and spatial associations between LPA4 and infiltrating phagocytes. Serial sections (continued from those shown in Fig. 2) of the cycling rat ovary were immunostained for LPA1 (A), LPA2 (B), LPA3 (C, D), LPA4 (E,F), MPO (G) and CD68 (H). Signals for LPA1 and LPA2 are very modest throughout the CL. The signal for LPA3 is intense specifically in large arterioles and venules surrounding the CL and in the ovarian stroma and the signal for LPA4 was localized to the degenerative site of the regressing CL and also vascular muscle cells outside the CL. The spatial association among LPA4, MPO (neutrophil) and CD68 (macrophage) is noted. The 200-μm scale bar in (A) also applies to (B), (C), (E), (G) and (H), whereas the 25-μm scale bar in (D) also applies to (F) and the insets in (G) and (H).

The LPA4 signal was also evident in vascular smooth muscle cells of the extraluteal arterioles and venules, similarly to the LPA3 signal. Because LPA4 signals at the inflamed sites of CL were suspected to be of phagocyte origin, we detected specific markers of neutrophils and monocyte/macrophages, namely myeloperoxidase (MPO) and CD68, respectively. MPO signals, together with polymorphonuclear features (Fig. 3G), and CD68 signals, with larger and irregular shapes, frequently with elongated pseudopods (Fig. 3H), were densely scattered in inflammatory sites. Thus LPA4-containing cells were identified as very likely macrophages and neutrophils.

Expression and activity of ATX in induced ovulation and blood lysoPLD activity during the estrous cycle

To further check whether ATX was involved in follicular dynamics, we used an immature rat model of induced ovulation. The ovaries held only follicles and the stroma, and the development, maturation and ovulatory rupture of follicles could be controlled using exogenous hormone treatment. ATX was slightly expressed only in interstitial tissue and was completely absent in all follicle stages, regardless of the administration of equine chorinonic gonadotropin (eCG; Fig. 4A1). No detectable ATX was induced in either preovulatory [3 h after human chorinonic gonadotropin (hCG); Fig. 4A2] or early ovulating (12 h after hCG; Fig. 4A3) follicles. LysoPLD activity in whole ovarian tissue before gonadotropin treatment was 1.43 ± 0.08 nmol·mg protein−1·h−1 (mean ± SEM, n = 3; Fig. 4B). It decreased during eCG-induced follicular growth (P < 0.05), probably because of the relative augmentation of ATX-negative tissues. Thereafter, the activity was constant throughout periovulatory period (0–24 h after hCG administration), indicating little involvement of ATX in follicular growth and ovulation in the rat.

Figure 4.

ATX expression and tissue lysoPLD activity in induced ovulation, and blood lysoPLD activity during the estrous cycle. In the immature rat model of gonadotropin-induced superovulation there was no ATX expression in preovulatory follicles at 0 h (A1), 3 h (A2) and 12 h (A3) after hCG administration and constant lysoPLD activity in the tissue homogenate (B). Serum lysoPLD activity in adult rats was unaltered during the estrous cycle (D1, diestrus 1; D2, diestrus 2; P, proestrus; E, estrus) (C). Enzyme activity in ovarian tissue (B) and sera (C) expressed as nmol·mg protein−1·h−1 and nmol·μL−1·h−1, respectively. The 200 μm scale bar in (A1) is also applicable for (A2) and (A3). Data are given as mean ± SEM (B, n = 3; C, n = 5).

We next determined blood lysoPLD activity in adult cycling rats and found constant activity (61.6–66.3 pmol·μL−1·h−1) during the estrous cycle (Fig. 4C). This finding, together with the constant immunoreactivity for ATX in CL, suggests little systemic effect of regressing CL-derived ATX on estrous-cycle-associated reproductive dynamics.

Effects of the ATX/LPA pathway on apoptotic cell death of luteal cells

Because ATX expression was induced during follicular luteinization and enhanced as luteal constituent cells began to die, we investigated whether ATX/LPA affected apoptotic activity. We tested the effects of local treatment with anti-ATX IgG, lysophosphatidylcholine (LPC, the primary precursor of LPA) or LPA in the ovarian bursal space on the expression of immunoreactive cleaved caspase 3 in the CL.

In stage 1 CL of intact rats, there were small numbers of activated caspase 3-positive cells (29 cells·mm−2; Fig. S1). The number increased to 138 cells·mm−2 in stage 2 and persisted in later stages. Normal IgG- and vehicle-treated control CL showed a similar temporal change and no difference compared with intact CL. There were no differences between IgG- and anti-ATX-treated groups and very few significant differences between reagent-treated groups and the vehicle-treated group. There was a significant reduction only in stage 4 CL in the LPA-treated group.

Effects of the ATX/LPA pathway on phagocyte recruitment

Because LPA4 signals were very likely present on migratory phagocytes accumulating in degenerative and inflammatory sites, we hypothesized that local endogenous LPA affected the chemotaxis of those cells. We tested the effects of local treatment with anti-ATX IgG or exogenous LPA into ovarian bursa on the numbers of neutrophils and macrophages in CL. Because recent evidence showed that LPC might have a role in the chemoattraction of macrophages [14, 15], the effect of LPC was also tested.

In stage 1 CL in the vehicle-treated group, moderate numbers of MPO-positive neutrophils and CD68-positive macrophages were present (31 and 113 cells·mm−2, respectively; Fig. 5A,B). Both increased significantly in stage 2 CL in the active inflammatory state and were sustained at stage 3 when the inflammatory response began to resolve. The numbers decreased in stage 4 CL undergoing complete resolution of inflammation. Anti-ATX IgG treatment had a modest upregulatory effect on neutrophil infiltration, with statistical significance detected in stages 1 and 4 compared with the normal IgG-treated group (Fig. 5A). Compared with the vehicle-treated group, exogenous LPC and LPA inhibited neutrophil infiltration significantly during stages 1–3. With regard to the dynamics of macrophage recruitment, both LPC and LPA were also inhibitory during stages 1–3 (Fig. 5B). Intriguingly, anti-ATX IgG reduced the number of infiltrated macrophages, especially in stages 3 and 4. Collectively, the ATX/LPC/LPA pathway exerted a negative effect on neutrophil recruitment into CL at stages 1–3. However, both LPC and LPA also inhibited macrophage infiltration during stages 1–3, whereas ATX itself seemed to stimulate it.

Figure 5.

Infiltration of phagocytes into regressing CL is modulated by ATX/LPC/LPA in vivo. Ovaries were treated with reagents for 6 h and thereafter processed for immunostaining of MPO (A) and CD68 (B). The numbers of cells positive for each marker in CL cross-sections were counted in each stage of the regressing CL. Counts were divided by the area (mm2) of the CL sections and expressed as mean ± SEM (n = 6 CL). *P < 0.05 versus vehicle-treated group at each stage. #P < 0.05 versus the normal IgG-treated group at each stage.

Effects of the ATX/LPA pathway on fibroblast proliferation

Finally, we tested the effects of the ATX/LPA pathway on fibroblast proliferation in vivo. In the vehicle-treated group, CL at stage 1 had 32.6% of fibroblasts positive for 5-bromo-2ʹ-deoxyuridine (BrdU; Fig. 6). This proliferative index remained unaltered until stage 3 and increased significantly to 63.2% at stage 4. Another control group with nonspecific IgG treatment showed a similar temporal change. Local application of anti-ATX had no effect at stage 1, but ATX neutralization significantly reduced the BrdU index compared with the IgG-treated control during stages 2–4. The reducing effect of the antibody was greater as the CL stage progressed and was temporally consistent with the expression level of ATX. By contrast, LPA administration significantly increased the BrdU-positive index (53.8 to 63.0% for each of vehicle group) from stage 1 to stage 3, but had no effect at stage 4.

Figure 6.

Fibroblast proliferation in the regressing CL is regulated by ATX/LPA in vivo. (A) Typical image of BrdU-positive fibroblast found in regressing CL. (B) The numbers of BrdU-positive and -negative fibroblasts in each stage of the regressing CL were counted and the ratio of positive cells to the sum of fibroblasts in CL specimens was expressed as the mean ± SEM (n = 6 CL). *P < 0.05 versus vehicle-treated group at each stage. #P < 0.05 versus normal IgG-treated group at each stage.


This study reveals that: (a) the expression of ATX occurs in luteal steroidogenic cells of newly formed CL in the cycling rat ovary and increases progressively until completion of structural involution; (b) the LPA4 signal is densely scattered at inflammatory sites and is overlapping with signals of macrophages and neutrophils, whereas LPA1 and LPA2 signals are faintly present throughout CL; and (c) ATX and LPA do not affect luteal cell apoptosis but modulate the infiltration of phagocytes and stimulate fibroblast proliferation during CL tissue remodeling to scar. This is the first demonstration of spatially and temporally regulated ATX expression in the rodent ovary and its probable contribution to CL demise and the resultant structural homeostasis of the organ.

Early in this research, we observed that a portion of large and small luteal cells contained substantial amounts of steroidogenic and other enzyme proteins, even at the late stage of tissue involution. This indicates that although the majority of CL-constituent cells die and are eliminated via phagocytosis to contribute to structural involution [1, 3], a population of luteal (substantial) cells remains viable and persistently possessed the potential to produce proteins possibly related to tissue demise and remodeling. These proteins include GIVA PLA2 and ATX, which show similar temporal and spatial distribution ([10] and this study) and, very importantly, may create a potential pathway for LPA production [7]. This proposition is supported by biochemical evidence of a metabolic link between these enzymes in injured neurons [16, 17]. What induces and enhances ATX expression in surviving luteal cells is not known, but it is unlikely to be the proestrous prolactin surge that is critical for the structural demise of cycling rodent CL [2, 3]. We found little difference in ATX expression in CL before and after exposure to an endogenous prolactin surge in the afternoon of proestrous, and between the endogenous prolactin-inhibited condition by a dopamine agonist and the exogenous prolactin-supplemented condition (data not shown). It is further notable that the ATX/LPA system in surviving luteal cells does not seem to affect apoptotic cell death in neighboring cells, as suggested by our experimental data.

The specification of LPA4 signals to migratory cells accumulating in degenerative sites led us to test the possible effects of ATX and LPA on phagocytes. Chemical alteration of ATX activity or LPA level in the ovarian local milieu did affect the number of neutrophils and macrophages in regressing CL. Bolus administration of LPC or LPA, at a dose of 1.0 nmol per ovary, inhibited the infiltration of two types of phagocytes. Inhibiting ATX activity increased neutrophil migration in stage 1 CL. LPC had effects similar to LPA, even at one tenth the dose of LPA, suggesting a distinct and direct action for LPC rather than its indirect effect as the precursor of LPA. Collectively, these data mean that the ATX/LPC/LPA pathway is inhibitory for recruitment of the granulocyte in this experimental model in vivo. Although the chemoattractive effects of LPA on human neutrophil [18, 19] and monocytic cell lines in vitro [20] have been reported to be positive, increasing evidence shows negative regulation of cell motility by LPA in cancerous cells that is dependent on a combination of receptor subtypes and cell types [13, 21-23]. The negative chemoattractive effect of the ATX/LPA pathway in regressing CL is further supported by the spatial dissociation of ATX and LPA4 localization and by a negative correlation between increased ATX expression and decreased phagocyte number during the resolution of inflammation (stages 3 and 4). We need to extrapolate the chemoattraction data obtained in vitro into the condition in vivo, because the identity and quantity of the receptor subtypes expressed, and the absolute and relative doses of the two lysophospholipids may result in a more complicated outcome than indicated by the generally clear in vitro data. Neutrophils and monocytes/macrophages show interactive regulation using a variety of signals, including LPC and eicosanoids [24]. This integrated network alters significantly during the progression and resolution of inflammation. We recognize the stimulating effect of ATX on macrophage infiltration, which seems inconsistent with the effects of its substrate and product. Our in vivo data may further involve a novel action of ATX that is specific to monocytes/macrophages and is independent of lysoPLD activity, as postulated for lung epithelial cell migration [25]. Alternatively, a LPC/LPA concentration gradient(s) may be produced by ATX in situ that cannot be accurately mimicked using the current method of reagent administration in different CL stages. The ATX/LPC/LPA pathway undoubtedly affects phagocyte recruitment into regressing CL. Detailed understanding of the definite action mechanism in vivo may require multiple, highly sensitive analytic technologies, including lipidomics.

Unlike phagocyte attraction, the result of ATX/LPA action on fibroblasts is straightforward and reasonable. The mitogenic effect of the ATX/LPA pathway supports the observation of a positive correlation between ATX expression and the proliferative status of the cell in aging CL. Although the LPA receptor subtype responsible is not currently identified, our in vivo findings are strongly supported by detailed studies in vitro showing that the proliferation and migration of fibroblasts are stimulated by the ATX/LPA pathway [26, 27] and GIVA PLA2 [28, 29]. It is of great clinical importance that overstimulation of the ATX/LPA pathway is relevant to the development of liver and lung fibrosis [30-32]. Structural regression of CL must involve a mechanism that terminates ATX/LPA-regulated fibroblast proliferation properly and allows the tissue to shrink and become a scar. Our overall results indicate that the ATX/LPA system is developed in the later stage of cycling CL and has active roles in resolution of the inflammatory response and fibroblastic replacement, rather than the function and death of luteal steroidogenic cells and their subsequent engulfment by inflammatory cells, which occur in the early and middle stages.

The current CL model in cycling rats does not implicate the ATX/LPA system in progesterone-secreting potential. There is an earlier report with bovine CL showing mRNA expression of LPA2 and LPA1, but not LPA3, and a direct inhibitory effect of LPA on luteinizing-hormone-stimulated progesterone release [33], whereas another later study has shown stable and ubiquitous expression of LPA1–4 in luteal cells and direct stimulatory effects of LPA on progesterone release and interferon-τ-regulated gene expression in early pregnancy [34]. CL are highly vascularized tissues whose development and regression is potentially regulated by blood flow [1, 2] and LPA is postulated to have a regulatory role in vascular tone and function [35, 36]. The presence of LPA3 and LPA4 in vascular muscle cells around CL and in interstitial tissues is indicative of an indirect role in CL regulation. In this context, the circulating ATX/LPA system should also be evaluated, because ATX activity and LPA production in sera have clinical implications for normal and pathological pregnancy in humans [37-39]. Our study using a pregnant rat model of functional CL is now in progress and has found enhanced expression of ATX in CL during the second half of pregnancy (unpublished observation). This clearly indicates different expressional regulation and a possible distinct role for ATX in hyperactivated CL compared with regressing CL for the cycle described here.

Finally, in the rat model, we found an absence of ATX in both normally developing and atretic follicles in adults and gonadotropin-controlled immature animals, and we confirmed that there was no fluctuation in tissue lysoPLD activity during hCG-induced ovulation. Follicular fluid from women treated with gonadotropin for in vitro fertilization have been reported to contain significant amounts of LPA and ATX activity [40]. A species-dependent difference may exist in this critical function of the ovary.

In conclusion, ATX is identified as a novel, luteal-regression-related factor that is increasingly produced by a portion of the luteal steroidogenic cell population surviving in structurally regressing CL of cycling rats. This factor modulates the infiltration of phagocytotic neutrophils and macrophages and stimulates fibroblastic cell proliferation, probably via LPA synthesis in large part. The finding provides a novel insight into the mechanism of tissue remodeling associated with structural regression of CL that operates via the ATX/LPA signaling pathway in presumed association with GIVA PLA2.

Materials and methods

Antibodies and chemicals

The specific antibodies for rat ATX, human LPA1 and rat LPA3 were purchased from Cayman Chemical (Ann Arbor, MI, USA). Anti-(human LPA2) (EDG-4) IgG was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-(human LPA4) (GPR23/P2Y9) IgG was from MBL (Woburn, MA, USA). Anti-rat CD68 and mouse nonspecific IgG were from AbD serotec (Kidlington, UK). Anti-human MPO was from Thermo Fisher Scientific (Fremont, CA, USA). A rabbit polyclonal antibody against cleaved caspase 3 (13 amino acid peptide of the C-terminus of mouse and human caspase 3 p18 subunit) was obtained from Trevigen (Gaithersburg, MD, USA). Anti-human GIVA PLA2 was a kind gift from Wyeth Co. (Cambridge, MA, USA). Anti-rat 3β-HSD was generated and characterized in our laboratory [41]. LysoPLD blocking peptide (rat lysoPLD amino acids 573–588), and LPA (1-oleoyl-sn-3-glycerophosphate) were purchased from Cayman. LPC was from Avanti Polar Lipids (Alabaster, AL, USA). eCG and hCG were purchased from Shionogi (Osaka, Japan) and Daiichi-Sankyo (Tokyo, Japan), respectively. The protein assay kit was from BioRad (Hercules, CA, USA). Vectastain Elite ABC staining kit was purchased from Vector Laboratories (Burlingame, CA, USA). The Amersham cell proliferation kit was purchased from GE Healthcare (Little Chalfont, UK). Choline oxidase and horseradish peroxidase were purchased from Funakoshi (Tokyo, Japan) and Toyobo (Osaka, Japan), respectively. TOOS reagent, N-ethyl-N-(2-hydroxy-3-sulfoproryl)-3-methylaniline, was from Dojin (Tokyo, Japan). All other reagents including 3,3ʹ-diaminobenzidine tetrahydrochrolide were of analytical grade.

Animal and tissue preparation

All procedures employed in this study were carried out following the guidelines of the Animal Care and Use Committee of Kitasato University School of Veterinary Medicine (Approval numbers, 11-084 and 12-006). Female Wistar–Imamichi rats were kept in an air-conditioned room with controlled lighting: 14 h of light (0500–1900 h) and 10 h of darkness. Rat chow and tap water were available ad libitum.

The animals used were 3–5 months old, weighed 250–400 g and showed a regular 4-day estrous cycle when examined by vaginal smear cytology. Blood was taken via heart puncture at 0900–1100 h on each day of the estrous cycle. Sera were separated and kept frozen until the lysoPLD assay. Ovaries were sampled from animals estrous morning after cervical dislocation under ether anesthesia.

Immature rats were also used to study the possible involvement of ATX in follicle development and ovulation. Female rats, 25–27 days old and weighing 50–70 g were treated with eCG (intraperitoneally, 0.2 IU·g−1 body weight) to induce follicular growth, followed 48 h later by hCG (10 IU·rat−1) to induce ovulation and luteinization. Rats were killed for ovary sampling during the time course after hCG stimulus, as described previously [42, 43]. Ovaries were processed for ATX immunohistochemistry and to assay tissue lysoPLD activity.


Immunohistochemistry was conducted to localize and evaluate the expression of ATX and other factors in cycling rat ovary that had CL and follicle tissues at different stages. Ovaries were instantly fixed in Bouin's solution overnight at 4 °C, dehydrated and embedded in paraffin [43, 44]. Tissues were serially sectioned at 2- or 6-μm thick, deparaffinized and examined immunochemically.

Sections were immunostained with an antibody against each antigen and a commercial staining kit. Endogenous peroxidase activity was blocked by pretreatment with 0.3% H2O2 in methanol for 30 min. For immunostaining of antigens except cleaved caspase 3, 0.01 m NaCl/Pi pH 7.5 was used as the standard aqueous buffer. For staining of cleaved caspase 3, 0.1 m Tris-buffered saline pH 7.5 was used and, for antigen retrieval, tissue sections were treated in a microwave with 0.01 m citrate buffer for 10 min followed by treatment with 0.1 mm glycine Tris-buffered saline for 30 min. The sections were incubated with the primary antibody at a dilution of 1 : 200 (LPA1–4, GIVA PLA2, MPO and cleaved caspase 3), 1 : 500 (ATX and CD68) or 1 : 10 000 (3β-HSD) overnight at 4 °C. Avidin–biotin–peroxidase complex was revealed by treatment with 3,3ʹ-diaminobenzidine tetrahydrochrolide and H2O2 for the appropriate time (2–6 min), depending on the antigen. The sections were counterstained with hematoxylin and mounted. Neighboring sections for each sample incubated with the nonimmune serum served as negative controls. To validate the specificity of ATX immunodetection, the ATX antibody was first incubated with blocking peptide at 4 °C for 2 h and used as above. Staining of ATX in regressing CL was evaluated by two independent observers using the following subjective scoring: negative (0), weak (1), moderate (2) or intense (3). Individual scores for each slide were averaged and expressed as the relative expression.

Because the ovary of adult rats with repeated estrous cycles contains four or more generations of CL [2, 44], we classified CL into four stages based on their gross and cellular appearance [45]. Stage 1, the newest stage, consisted mainly of intact luteal cells with round and oval nuclei and capillary endothelial cells, rarely of fibroblasts with elongated nuclei and very rarely of focal sites with the onset of an inflammatory-like reaction. Stage 2 consisted of multiple small foci and often a merged large focus with progressing inflammatory reaction; the remaining portion consisted of seemingly intact luteal cells with small vacuoles in the cytoplasm, degenerative endothelial cells and increased numbers of fibroblasts. Stage 3 consisted of irregularly shaped and shrunk luteal cells, marked fibroblasts and sites turning to the resolution of inflammation. In the oldest stage, 4, tissues were much shrunk and composed mostly of fibrous tissues. There were also numerous numbers of immune cells that accumulated into the CL in the stage-dependent fashion, as described in the 'Results' section.

Evaluation of phagocytes infiltration

The infiltration of neutrophils and macrophages was evaluated by immunohistochemical detection of each cell marker. For observation of intact cycling rats, animals were killed on estrus morning (0900–1100 h) and ovaries were harvested. To evaluate the effects of ATX and related lysophospholipids, the experimental treatment was carried out on the morning of the estrus. Animals under ether anesthesia were subjected to bilateral abdominal incisions and the ovarian bursa was exposed. A 100 μL aliquot of vehicle (NaCl/Pi), LPA (10 μm at final concentration), LPC (1 and 10 μm), normal IgG (50 μg·mL−1 in NaCl/Pi) or anti-ATX (50 μg·mL−1 in NaCl/Pi) was injected into one ovarian bursal cavity using a syringe and repeated in another side of the ovary. In this study, the possible effect(s) of secreted ATX was evaluated by the depletion of ATX activity following the challenge of the neutralizing anti-ATX IgG to the given environment. Our experiment determining lysoPLD activity in the sera of pregnant rats demonstrated that an identical final concentration (50 μg·mL−1) of anti-ATX IgG in the assay condition decreased the enzyme activity to 41.9 ± 0.3% that of the IgG-treated control. We ascertained no visible leakage of the injected solution and swelling of the bursa after the successful injection. After the injection, ovaries were positioned back in the abdominal cavity and muscle and skin were sutured separately. Six hours after this treatment, animals were killed by cervical dislocation under light ether anesthesia and ovaries were harvested. The organs were subsequently treated for histological procedures as described above. The numbers of immunoreactions in a CL cross-section, counted under a light microscope, were divided by the area of CL (mm2) and presented.

Fibroblastic cell proliferation assay

The proliferative activity of fibroblasts in vivo was evaluated using a BrdU incorporation assay. BrdU in NaCl/Pi solution (60 μg·100 μL−1·ovarian bursa) with additional reagents including anti-ATX (50 μg·mL−1 in NaCl/Pi), normal IgG (50 μg·mL−1 in NaCl/Pi), LPA (10 μm at final concentration) or vehicle was first prepared. Local treatment was performed in the estrus morning as described above. The animals were killed and organs were harvested 6 h later. Ovarian tissues were fixed in Bouin's fixative and 2-μm thick tissue sections were prepared. BrdU-labeled cells were detected immunochemically using a BrdU in situ detection kit. Both positive and negative immunoreactions for fibroblastic cells were counted with careful distinction from those of other cell types such as steroidogenic cells or capillary endothelial cells. This separate counting was performed in each stage (1–4) of CL (n = 6) obtained from four ovarian samples for each experimental group. The value was expressed as the number of BrdU-positive cells divided by the total number (the sum of positive and negative fibroblasts) and analyzed statistically.

Apoptotic cell death of CL-constituent cells

Apototic cell death in the regressing CL was assayed by histological counting of cells positive for immunoreactive cleaved caspase 3, the final executioner of this form of cell death [2, 3]. CL tissue sections almost identical to those used for phagocyte detection were submitted for immunohistochemistry of active caspase 3 followed by the quantification of apoptotic cells.

LysoPLD activity assay

The lysoPLD activity of sera and ovarian tissue extracts was assayed using a previously described method [46] with some modification. Extracts from whole ovarian tissues of eCG/hCG-treated rats were prepared in ice-cold NaCl/Pi and protein levels were determined using the Bradford assay. Briefly, samples of blood sera or tissue extract (20 μL) were incubated with 2 mm 18 : 0-LPC in the presence of 100 mm Tris/HCl (pH 9.0), 500 mm NaCl, 5 mm MgCl2 and 0.05% Triton X-00 for 5 h at 37 °C. The liberated choline was detected by an enzymatic photometric method using choline oxidase, horseradish peroxidase and TOOS reagent as a hydrogen donor.

Statistical analysis

All numerical values were treated statistically and presented as mean ± SEM (sample numbers are indicated). The means among different groups were analyzed by one-way analysis of variance followed by the Tukey–Kramer multiple comparison test. Values of P < 0.05 was considered to be significant.


The authors thank S. Kitao, H. Satoh and S. Nagashima for help with the experiments, Dr K. Igarashi (TOSOH Corporation) for advice on lysoPLD assay, Dr T. Yonezawa for discussion, and M. Nakata for help in the manuscript preparation. This work was supported partly by Grants-in aid for Scientific Research (No. 23580392) from Japanese Society for Promotion of Science and by the Kieikai Research Foundation (both to SK).