The FOS Transcription Factor Family Differentially Controls Trophoblast Migration and Invasion*

Background: Trophoblast cells invade the uterus during pregnancy to promote blood flow to the conceptus. Results: The transcription factor FOS inhibits trophoblast invasion, whereas FOS-like (FOSL)-1 promotes invasion. Conclusion: The intracellular balance of FOS family transcription factors modulates trophoblast invasive potential. Significance: Understanding the regulation of trophoblast invasion is crucial for determining the etiology of several placenta-associated obstetrical complications. Extravillous trophoblast invasion is a fundamental component of human placentation. Invading trophoblast cells promote blood flow to the conceptus by actively remodeling the uterine vasculature. The extent of trophoblast invasion is tightly regulated; aberrant invasion is linked with several obstetrical complications. However, the transcriptional networks responsible for controlling the extent of trophoblast invasion are not well defined. Previous studies have identified high levels of FOS (FOS, FOSB, FOS-like (FOSL) 1, and FOSL2) proteins in extravillous trophoblast cells. These proteins form part of the activating protein-1 (AP-1) transcription factor complex and are implicated in regulating gene networks controlling cellular invasion in diverse biological systems. Therefore, we hypothesized that FOS family proteins play a role in regulating trophoblast invasion. We assessed expression of FOS family proteins in trophoblast cell lines and human placentae at different gestational ages. FOS, FOSB, and FOSL1 proteins were robustly increased in trophoblast cells subject to wound-based migration assays as well as Matrigel-based invasion assays. FOS knockdown resulted in cessation of proliferation and an induction of migration and invasion concomitant with robust expression of matrix metalloproteinase (MMP) 1, MMP3, and MMP10. Conversely, FOSL1 knockdown abrogated trophoblast migration and invasion and inhibited the production of MMP1, MMP3, and MMP10. In human placenta, FOS was expressed in proximal anchoring villi in conjunction with phospho-ERK. FOSL1 was temporally expressed only in the distal-most extravillous trophoblast cells, which represent a migratory cell population. Therefore, FOS and FOSL1 exert opposing effects on trophoblast invasion.

Extravillous trophoblast invasion during the first half of pregnancy is an essential component of human placentation. Invading trophoblast cells migrate into the uterus, infiltrate the uterine spiral arterioles, displace the endothelium lining these vessels, and destroy the surrounding smooth muscle and elastic laminae (1). This remodeling results in dilated, flaccid conduits capable of supplying the conceptus with sufficient maternal blood during the latter stages of pregnancy. Trophoblast invasion is physiologically normal and is tightly regulated to ensure that the growing conceptus receives a sufficient supply of maternal blood without excessive maternal hemorrhage. Aberrant trophoblast invasion is linked with several obstetrical complications. For example, deficient trophoblast invasion is a key culprit in the etiology of pre-eclampsia and intrauterine growth restriction (2), whereas trophoblast hyperinvasion into maternal tissues is linked with choriocarcinoma, invasive moles, and placenta creta (3).
The transcriptional networks that facilitate trophoblast taxis are not well defined. Because of their inherently invasive nature, trophoblast cells have often been compared with highly invasive carcinoma cells, and in many cases, both exhibit similar gene expression signatures that regulate cellular invasion (4,5). For example, trophoblast cells and carcinomas both express a diverse spectrum of matrix metalloproteinases (MMPs), 4 which facilitate matrix remodeling and promote cellular locomotion in both systems. Therefore, it stands to reason that the transcriptional regulation of many genes associated with cancer metastasis and trophoblast invasion may also be shared. The activating protein-1 (AP-1) complex potentially fits this scenario.
Activation of AP-1 causes rapid modulation of gene expression, including that of several MMP genes, in response to a variety of extracellular stimuli (6,7). The net outcome is the modulation of factors affecting fundamental cellular processes, including proliferation, survival, and migration. AP-1 complexes can consist of JUN (JUN, JUNB, and JUND) and FOS (FOS, FOSB, FOSL1, and FOSL2) family members. Although JUN proteins are capable of homodimerization, FOS family members exclusively heterodimerize with JUN proteins (6). Accumulation of JUN and FOS proteins is promoted by phosphorylation-mediated stabilization of existing proteins as well as transcriptional regulation (8,9). In vitro studies have determined that, in many situations, JUN-FOS heterodimers are more stable and have stronger DNA binding activities than JUN-JUN homodimers (10 -12). Thus, expression of FOS proteins contributes to the robust activity of AP-1-regulated genes.
In the context of cellular invasion, each FOS family member is implicated in the control of genes affecting cell locomotion, particularly in cancer cells (13)(14)(15)(16)(17)(18)(19)(20)(21). However, little is known about the role of FOS family members in the placenta. In the mouse, which exhibits invasive hemochorial placentation akin to humans, FOSL1-deficient mice exhibit placental and extraembryonic abnormalities leading to early embryonic death (22). Similarly, knockdown of FOSL1 in rat placentae and rat-derived trophoblast stem cells causes defects in trophoblast invasion in vivo and in vitro (23). FOS is expressed in mouse placenta, but mice deficient in FOS are viable and fertile despite having reduced placental and fetal weights (15). FOSL2 knock-out mice are also viable but exhibit placental structural and signaling defects that indirectly impact differentiation of embryonic tissues from which they ultimately succumb (24). No defect in placentation has been reported in FOSB knock-out mice. In human placentae, FOS, FOSB, FOSL1, and FOSL2 have been detected spatiotemporally at high levels in extravillous trophoblast cells (25)(26)(27). Moreover, AP-1 transcriptional activity is required for the promotion of MMP production by human trophoblast cells (28,29). However, the biological significance of individual FOS components in human trophoblast cells is not known. To address this, we studied the expression of FOS family members in trophoblast cells in vitro and in placental sections in situ and utilized knockdown approaches to determine the role of individual FOS family members in trophoblast migration and invasion.

EXPERIMENTAL PROCEDURES
Cells-BeWo, Jar, and JEG3 trophoblast cells initially derived from metastatic lesions of choriocarcinoma (30 -32) were obtained from the American Type Culture Collection. The immortalized human first trimester extravillous trophoblast lines HTR-8/SVneo (HTR8) (33), Swan 71 (34), and SGHPL4 (35) were kind gifts from Professors Charles Graham (Queen's University, Kingston, Ontario, Canada), Gil Mor (Yale University, New Haven, CT), and Guy Whitley (St. George's, Univer-sity of London, London, UK), respectively. The TCL1 term trophoblast cell line was kindly provided by Professor Mark Sullivan (Imperial College London, London, UK) (36). BeWo and SGHPL4 cells were maintained in DMEM/F-12 medium (Sigma-Aldrich), whereas JEG3, Jar, HTR8, Swan 71, and TCL1 were maintained in RPMI 1640 medium (Invitrogen). All media were supplemented with 10% fetal bovine serum (FBS), 100 units/ml penicillin, and 100 M streptomycin (all from Sigma-Aldrich). Cells were passaged by light trypsinization prior to reaching confluence and were maintained at 37°C in an atmosphere consisting of 5% CO 2 for no more than 20 successive passages. 293FT human embryonic kidney cells were purchased from the American Type Culture Collection and maintained in DMEM supplemented with 10% FBS, 100 units/ml penicillin, and 100 M streptomycin.
RT-PCR-Relative levels of RNA were assessed by RT-PCR. Total cellular RNA was extracted using TRI Reagent (Sigma-Aldrich) according to the manufacturer's protocol. 500 ng of total RNA was utilized for reverse transcription using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA). cDNAs were diluted 1:10 and subjected to conventional PCR or quantitative RT-PCR analysis using primers detailed in Tables 1 and 2, respectively. Conventional PCR was performed for 30 cycles (denature, 94°C for 30 s; anneal, 55°C for 30 s; extension, 72°C for 30 s), and products were resolved by electrophoresis in ethidium bromide-containing 1% agarose gels. Quantitative RT-PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems). Amplification and fluorescence detection were carried out using an ABI Prism 7500 real time PCR system (Applied Biosystems). Cycling conditions included an initial holding step (95°C for 10 min) followed by 40 cycles of two-step PCR (95°C for 15 s and then 60°C for 1 min) followed by a dissociation step (95°C for 15 s, 60°C for 15 s, and then a sequential increase to 95°C). Relative mRNA expression was calculated by the comparative ⌬⌬Ct method using 18 S rRNA as a reference RNA.
Wound Assay-In vitro scratch wound assays were conducted to assess directional cell migration. They have been described in detail elsewhere (37). Briefly, confluent monolayers of cells were incubated with 5 M mitomycin-C (Sigma-Aldrich) for 2 h to inhibit cell proliferation and washed extensively, and then cells were disrupted by horizontally dragging a pipette tip across the surface of the monolayer. Wells were washed with phosphate-buffered saline (PBS), pH 7.4 to remove detached cells, and fresh medium was added to the wells. Subsequently, an image of the scratch was recorded in phase using a Leica DMI 4000 microscope. To ensure that the image was consistently taken in the same region of the scratch, a small vertical scratch in the monolayer was included and used as a landmark. The time the initial image was taken was considered t ϭ 0 h. Images were captured at multiple times over the course of wound closing. Cells were returned to 37°C, 5% CO 2 between imaging sessions. To measure the area of wound closure in images taken at specified times, cell frontiers bordering the wound were traced using ImageJ software (38). The percentage of wound closure was calculated based on the equation where Wi is the area of wound at t ϭ 0 h and Wz is the area of wound following z h of culture.
To assess FOS family member expression in trophoblast cells following scratch wounding by RT-PCR and Western blotting, multiple scratch wounds were introduced into a monolayer of trophoblast cells by dragging a comb with 20 teeth/inch in a circumferential motion. Following washing with PBS, the medium was changed, and cells were returned to 37°C, 5% CO 2 for 0, 1, 3, or 6 h prior to RNA or protein extraction.
Matrigel-based Invasion Assay-To assess trophoblast invasion through Matrigel, Transwell inserts (6.5-mm diameter, 8-m pore; BD Biosciences) were coated with 400 g/ml phenol red-free Matrigel (BD Biosciences) diluted in serum-free shRNA constructs in pLKO.1 vectors were obtained from Sigma-Aldrich. Control shRNA that does not target any known mammalian gene was obtained from Addgene (pLKO.1-shScramble, plasmid 1864, Addgene, Cambridge, MA). Sequences representing the sense target site for each of the shRNAs used in the analyses are as follows: FOS, GCGGAG-ACAGACCAACTAGAAC; FOSB, GCCGAGTCTCAATA-TCTGTCTC; FOSL1, CCTCAGCTCATCGCAAGAGTAC; Control, CCTAAGGTTAAGTCGCCCTCG. pLX304 lentiviral constructs containing the complete cDNA sequence of FOS were obtained from DNASU plasmid repository (HsCD00434995; DNASU, Tempe, AZ). Empty pLX304 plasmids were obtained from Addgene (plasmid 25890) and used as controls. Third generation lentiviral packaging vectors (pMDLg/pRRE (plasmid 12251) and pRSV-Rev (plasmid 12253)) and a vesicular stomatitis virus glycoprotein envelope vector (plasmid 12259) were acquired from Addgene. The generation of lentiviral particles was reported previously (39). Briefly, 293FT cells were transiently transfected using Lipofectamine 2000 (Invitrogen) with the shRNA-containing transducing vector and the third generation plasmids described above. Culture supernatants containing lentiviral particles were harvested every 24 h for a total of 48 -72 h. Supernatants were centrifuged to remove cell debris, filtersterilized, and stored at Ϫ80°C until used.
In Vitro Lentiviral Transduction-Trophoblast cells were exposed to lentiviral particles for 24 -48 h in the presence of 6 g/ml Polybrene (Sigma-Aldrich) in normal growth medium, selected with puromycin dihydrochloride (Invitrogen; 3.5 g/ml final concentration) for 2-5 days for cells transduced with pLKO.1 plasmids and blasticidin (Invitrogen; 3.5 g/ml final concentration) for 5-7 days for cells transduced with pLX304 plasmids, and then subjected to experimental analyses.
Proliferation Assay-To quantitatively determine the proliferative capacity of manipulated trophoblast cells, shRNA-incorporated cells were plated in triplicate in 96-well plates and incubated at 37°C, 5% CO 2 for 2 h while cells adhered to the plate. This was referred to as t ϭ 0 h. Other triplicate wells were incubated for 24 or 48 h. After the appropriate incubation, cells were fixed in 4% paraformaldehyde, stained using toluidine blue in 5% ethanol, washed extensively, and allowed to dry. Absorbed dye was then released by addition of a solution containing 2% sodium dodecyl sulfate, and the resulting color was gauged by measuring absorbance at 550 nm using a spectrophotometer. The absorbance in these samples was then plotted against a standard curve generated by performing this assay with known cell numbers. Trophoblast cells used in the production of the standard curve were initially counted under a microscope using a hemocytometer and then subjected to a 1:1 dilution series.
Matrigel-based Tube Forming Assay-The ability of trophoblast cells to form tubes when placed on Matrigel has been described previously (40 -42) and is thought to relate in part to their capacity for pseudovasculogenesis. To determine the effect of FOS family members on the ability of trophoblast cells to form tubelike structures on Matrigel, trophoblast cells (1.5 ϫ 10 5 cells/ml) expressing control shRNA or shRNA for FOS, FOSB, or FOSL1 were placed onto gelled Matrigel (ϳ7.5 mg/ml). Images were taken following incubation for 24 and 48 h by light microscopy (Leica DMI 4000). The total numbers of tubes within 400ϫ microscopic fields were recorded. At least four non-overlapping fields were used for analysis.
Immunohistochemistry-Paraffin-embedded blocks of first trimester (gestational age, 6 or 12 weeks) and term (39-week) human placentae were obtained from the Research Centre for Women's and Children's Health Biobank (Mount Sinai Hospital, Toronto, Ontario, Canada). All collections were performed with appropriate consent and were approved by the University of Toronto and the University of Kansas Medical Center research human ethics boards. Placentae that were removed due to elective termination of pregnancies (first trimester tis-sues) or elective caesarian deliveries (term) were deemed to be from healthy pregnancies. Blocks were sectioned at 5-m thickness. Subsequently, sections were deparaffinized and rehydrated in a graded series of ethanol washes, subjected to antigen retrieval at 95°C in Decloaking agent (RV1000M, Biocare Medical, Concord, CA), and probed with antibodies specific for FOS (1:200), FOSB (1:50), FOSL1 (1:100), or phospho-ERK (1:200). The antibodies used for immunohistochemistry were the same as those described above for Western blotting and immunocytochemistry. To identify proliferating cells, some sections were probed for proliferating cell nuclear antigen (diluted 1:200; sc-56, Santa Cruz Biotechnology). Sections were immersed in species-appropriate fluorescence-conjugated secondary antibodies followed by incubation with FITC-conjugated pan-cytokeratin antibody (1:400; Sigma-Aldrich) to identify trophoblast cells and DAPI to localize nuclei. Sections were then mounted with Fluor-G medium and examined using a Leica DMI 4000 microscope.
Statistical Analysis-Statistical comparisons between two means were performed with Student's t test. Comparisons of multiple groups were evaluated with analysis of variance. The source of variation from significant F-ratios was determined with Dunnett's multiple comparison test. Results were deemed statistically significant when p was Ͻ0.05. All experiments were conducted at least in triplicate and were replicated at least two to three times. Graphing and statistical analyses were performed using GraphPad Prism 6.0.

FOS Family Gene Expression in "Quiescent" Trophoblast Cell
Lines-We initially screened seven human trophoblast cell lines for their respective expression of FOS family members, namely three lines derived from choriocarcinoma (BeWo, JEG3, and Jar), three lines from immortalized first trimester extravillous trophoblast cells (HTR8, Swan 71, and SGHPL4), and one line derived by immortalization of term cytotrophoblast cells (TCL1). Each cell line was cultured according to published standard culture conditions. These conditions will be referred to as quiescent for the duration of the report. Under quiescent conditions, FOS and FOSB exhibited low to undetectable mRNA expression. FOSL2 was present at low but detectable levels in all trophoblast lines studied except for the term trophoblast cell line TCL1. FOSL1 expression was not detected in JEG3, Jar, or TCL1 cells; was detected at low levels in BeWo cells; and was constitutively robust in the lines derived from first trimester extravillous trophoblast cells (Fig. 1). For positive controls, BeWo trophoblast cells stimulated with 250 M 8-bromo-cyclic adenosine monophosphate were used.
FOS Family Expression in Trophoblast Cell Lines following Stimulation of Migration-Because of the unique FOS family transcription factor profile in first trimester extravillous trophoblast cell lines, we pursued additional experiments using two of these lines, Swan 71 and HTR8 trophoblast cells. To assess FOS family expression in these cells during migration, confluent monolayers were mechanically disrupted at regular intervals analogous to incorporation of multiple in vitro scratch wounds. In preliminary analyses, we determined that both Swan 71 and HTR8 trophoblast cells begin to move into the wound site ϳ3-4 h following introduction of a scratch wound, and both completely close the wound within 24 h. Immediately following wounding, Swan 71 trophoblast cells exhibited a transient but robust increase in FOS and FOSB transcript levels of ϳ25and ϳ35-fold, respectively, 1 h following wounding compared with unscratched cells ( Fig. 2A; p Ͻ 0.05). FOS and FOSB transcript concentrations were also elevated 3 h following wounding but returned to basal levels by 6 h. Comparatively, FOSL1 and FOSL2 mRNA levels remained unchanged. Similar results were obtained using HTR8 trophoblast cells ( Fig. 2A). Induction of FOS and FOSB proteins paralleled that of mRNA (Fig. 2B). Interestingly, FOSL1, but not FOSL2, protein substantially increased following introduction of the scratch wound. Accumulation of FOSL1 protein was slower than that of FOS or FOSB but was more stable, remaining high even after 6 h. Upon further analysis, it was determined that FOSL1 phosphorylation at Ser 265 (phospho-Ser 265 -FOSL1) also increased following scratch wounding (Fig. 2B). Phosphorylation of FOSL1 at Ser 252 and Ser 265 is known to stabilize FOSL1 by protection from proteasome-mediated degradation (9,44). Therefore, FOSL1 accumulation in human trophoblast cells may also be influenced by post-translational modification.
To determine the localization of FOS, FOSB, and FOSL1 in monolayers subjected to scratch wounding, we performed immunocytochemical analysis for each factor 1 or 3 h following scratch wounding (Fig. 2C). Unwounded cells were used as a control. Cells immediately adjacent to the wound site exhibited remarkably high levels of FOS and FOSB, whereas cells distant from the wound site had low or absent expression of these factors. FOSL1 was increased in both cells adjacent and distant to the wound site, consistent with our earlier observations of constitutively high expression of FOSL1 in quiescent cells (Fig. 2C).
Because JUN family members are heterodimeric binding partners for FOS family members, we also assessed JUN, JUNB,  FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 and JUND expression in Swan 71 and HTR8 trophoblast cells following scratch wounding. JUN, JUNB, and JUND were all modestly but significantly increased by ϳ2-fold 1 and 3 h following scratch wounding in Swan 71 trophoblast cells and returned to basal levels 6 h following wounding. In HTR8 cells, only JUNB was induced by 2-fold, but it returned to basal levels by 6 h following wounding (data not shown).

The Control of Trophoblast Invasion by FOS Family Members
ERK-Mitogen-activated Protein Kinase (MAPK) Regulates FOS Family Transcription Factor Accumulation-MAPKs are commonly linked to the regulation of FOS family members (45,46). Therefore, to ascertain the upstream cascades that induce FOS, FOSB, and FOSL1 in trophoblast cells, we assessed phosphorylation levels of p38, JNK, and ERK at 10, 30, and 60 min following monolayer wounding. Phospho-p38, phospho-JNK, and phospho-ERK were undetectable in quiescent cells. After wounding, all three MAPKs were phosphorylated within 10 min and were maximally phosphorylated after 30 and 60 min (Fig. 3A). We then determined mRNA induction of FOS and FOSB in the presence of the MAPK pharmacological inhibitors U0126 (10 M; inhibits MEK-ERK signaling as shown in Fig.  3A), SB203580 (20 M; preferentially inhibits p38 downstream pathways but does not affect phosphorylation of p38) (47); and SP600125 (20 M; preferentially inhibits JNK phosphorylation as shown in Fig. 3A) 1 h following wounding. Treatment with U0126 profoundly inhibited wound-induced stimulation of FOS and FOSB mRNA by Ͼ75% (Fig. 3, B and C; p Ͻ 0.05). Treatment with SP600125 modestly but significantly decreased FOS and FOSB mRNA induction by 30 and 40%, respectively (Fig. 3, B and C; p Ͻ 0.05), whereas treatment with SB600125 decreased only FOSB mRNA by 30% (Fig. 3C; p Ͻ 0.05). These findings implicate the MEK-ERK cascade as a primary determinant of FOS and FOSB mRNA induction following scratch wounding. To determine the effect of MEK-ERK on FOS, FOSB, and FOSL1 protein accumulation following scratch wounding, we performed Western blotting on cell lysates collected from vehicle (DMSO)-or U0126-treated Swan 71 cells 1 or 3 h following wounding. Unwounded cells were used as controls. As shown in Fig. 3D, inhibition of MEK-ERK profoundly inhibited FOS, FOSB, FOSL1, and phospho-Ser 265 -FOSL1 protein accumulation.
Expression of FOS Transcription Factor Family Members in Trophoblast Cells following Plating on Matrigel-Matrigelbased invasion assays are routinely used to simulate an environment to test the invasive potential of trophoblast cells. However, the intrinsic regulation of trophoblast migration and invasion by dynamic changes in transcription factor expression or activity following plating on Matrigel is poorly understood. Therefore, we assessed expression of FOS family members after plating trophoblast cells on Matrigel. Within 1 h after plating cells on Matrigel, we observed a robust increase in FOS and FOSB mRNA by ϳ60and ϳ240-fold, respectively, compared with quiescent cells (Fig. 4A; p Ͻ 0.05 for each). FOS and FOSB transcripts rapidly declined thereafter and returned to basal levels by 6 h after plating. FOSL1 mRNA increased by ϳ12-fold ( Fig. 4A; p Ͻ 0.05), whereas FOSL2 mRNA did not change significantly after plating on Matrigel. Unlike transcript levels for FOS and FOSB, FOSL1 transcript levels remained increased at 6 h following plating (7-fold). Protein levels of FOS and FOSB were robustly increased by 1 h following Matrigel plating and were still detectable 6 h after plating (Fig. 4B). FOSB was still detectable 24 h after plating, although FOS expression was not detectable. Protein levels of FOSL1 were also increased following plating on Matrigel relative to quiescent trophoblast cells. The increase in FOSL1 protein levels can be partially attributed to the increase in FOSL1 mRNA, although we also observed an increase in phospho-Ser 265 -FOSL1, suggesting that increased protein stability may also play a role (Fig. 4B). FOSL2 protein levels showed modest fluctuations in parallel with our observations for FOSL2 mRNA. Therefore, because FOSL2 protein expression did not change appreciably following scratch wounding or following Matrigel plating, FOSL2 was not assessed in subsequent analyses.

Effect of Individual FOS Transcription Factor Family Members on Trophoblast Proliferation, Migration, and Invasion-To
assess the role of FOS, FOSB, and FOSL1 in various aspects of trophoblast biology, we transduced Swan 71 trophoblast cells with pLKO.1 plasmids encoding shRNAs specific for individual FOS family members. Cells were briefly selected with puromycin and immediately used for experimental purposes to limit perturbations in clonal bias due to putative proliferation defects. The efficacy of shRNA-mediated knockdown for each FOS family member is presented in Fig. 5A. For these experiments, Swan 71 cells expressing control shRNAs or shRNAs targeting FOS, FOSB, or FOSL1 were plated for 1 h on Matrigel. There were no differences in cell viability following individual knockdown of FOS, FOSB, or FOSL1.
We initially assessed proliferation of trophoblast cells following individual knockdown of FOS, FOSB, or FOSL1. Compared with trophoblast cells expressing control shRNA, inhibition of FOSB had no discernable effects on trophoblast proliferation. In contrast, FOS and FOSL1 knockdown resulted in a significant reduction of cell proliferation by 80 and 60%, respectively, compared with control cells (Fig. 5B; p Ͻ 0.05 for both). To determine whether FOS and FOSL1 knockdown affected expression of genes associated with cell cycle progression, we analyzed transcript expression of cyclin-dependent kinase inhibitors (CDKN1A, CDKN1B, CDKN1C, CDKN2A, CDKN2C, and CDKN3) in trophoblast cells expressing shRNAs for FOS, FOSB, or FOSL1. Compared with control cells, knockdown of FOSB had no discernible effect on transcript expression of any of these genes, which is consistent with the fact that the shRNA for FOSB did not affect trophoblast proliferation. In contrast, following knockdown of either FOS or FOSL1, expres-   FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 JOURNAL OF BIOLOGICAL CHEMISTRY 5031 sion of CDKN1A was increased ϳ6-fold compared with control cells (Fig. 5C; p Ͻ 0.05 for both). Levels of the other transcripts were unaffected.

The Control of Trophoblast Invasion by FOS Family Members
Our next analysis involved determining the effect of FOS, FOSB, or FOSL1 on the ability of trophoblast cells to migrate and invade using three independent experiments: scratch wound assay, Matrigel-based invasion assay, and Matrigelbased tube forming assay. Images depicting a wound at the time of scratching (0 h) and 10 h postwounding in cells expressing control shRNA or FOS, FOSB, or FOSL1 shRNAs are shown in Fig. 6A. Following scratch wounding, cells expressing control shRNA migrated into the wound and completely closed it by ϳ20 h. Similar results were obtained using cells expressing FOSB shRNA. Surprisingly, cells expressing FOS shRNA rapidly moved into the wound and completely saturated the wound site by 10 -12 h (Fig. 6B). In contrast, cells expressing FOSL1 shRNA were sluggish in their ability to close the wound site: the gap was still readily apparent even 72 h following scratching. Collectively, these data indicate that cells expressing FOS shRNA had increased migratory potential, whereas cells expressing FOSL1 shRNA had decreased migratory potential.
Similar roles for FOS family members were observed in the invasion assays. For these experiments, cells were plated on Matrigel-coated Transwell inserts, and the ability of cells to traverse the Matrigel and attach to the underside of the Transwell membrane was scored (representative images are shown in Fig. 6C). Compared with cells expressing control shRNA, cells with FOS knockdown exhibited 4.5-fold increased numbers of cells traversing Matrigel (Fig. 6D; p Ͻ 0.05). Cells expressing FOSB shRNA exhibited a modest but significant increase in the number of invaded cells (1.5-fold, p Ͻ 0.05). In contrast, following FOSL1 knockdown, the invasive capacity of trophoblast cells was almost completely inhibited ( Fig. 6D; p Ͻ 0.05).
Finally, we assessed trophoblast morphology on Matrigel using a tube forming assay. Following plating on Matrigel, trophoblast cells clustered tightly together and formed projections that extended into the Matrigel and often connected with adjacent cell clusters analogous to modified tube formations. Images of trophoblast-generated tube formations 24 or 48 h after plating are shown in Fig. 7. Cells expressing control shRNAs as well as cells expressing shRNAs targeting FOS or FOSB immediately (within 6 h) formed tubes on Matrigel. Numbers of tubelike formations were similar among the three groups (Fig. 7). However, cells expressing FOSL1 shRNA were unable to efficiently form tubes but rather formed tight clusters that inefficiently connected to adjacent clusters ( Fig. 7; p Ͻ 0.05).
At 48 h, a striking morphological discrepancy was observed in cells expressing FOS shRNA compared with all other groups. FOS knockdown cells formed more projectiles from cell clusters. Cells in the clusters appeared detached and readily moved into the Matrigel (Fig. 7, white arrows). These morphological observations are consistent with our previous observations that FOS knockdown causes increased migratory and invasive potential. Cells expressing control or FOSB shRNAs formed small projections into the Matrigel (albeit much smaller compared with cells expressing FOS shRNA). In contrast, cells expressing FOSL1 shRNA continued to appear tightly packed with few or no projections extending into the Matrigel, and similar to our observations at 24 h following plating, relatively few tubelike formations were present (Fig. 7).
Because MMP1, MMP3, and MMP10 were dramatically reduced in trophoblast cells expressing FOSL1 shRNA, we per-formed chromatin immunoprecipitation to determine whether FOSL1 is capable of binding to previously characterized AP-1 heptameric sites (TGAGTCA or TGAATCA) located ϳ70 base pairs upstream of the transcription start site of each gene. We found that, relative to nonspecific control antibody, FOSL1 binding was enriched within the proximal promoter regions of MMP1 ( Fig. 9B; 2 FOS Suppresses FOSL1 Expression-Because of the seemingly opposite results observed with trophoblast migration, invasion, and MMP expression in cells expressing FOS versus FOSL1 shRNA, we checked the expression of each FOS family member in Swan 71 cells expressing control, FOS-, FOSB-, or FOSL1-specific shRNAs following 1 h of plating on Matrigel. As expected, cells expressing shRNAs for FOS, FOSB, or FOSL1 dramatically reduced the expression of their intended target (Fig. 10A). FOS knockdown also modestly reduced FOSB expression and vice versa, indicating that these proteins may positively regulate each other. Strikingly, cells expressing FOS shRNA had dramatically enhanced levels of FOSL1 and phospho-Ser 265 -FOSL1 protein (Fig. 10A). Interestingly, we did not observe a change in FOSL1 mRNA expression in FOS knockdown cells, suggesting increased post-translational stabilization of FOSL1 when FOS is inhibited (Fig. 10B). To determine whether FOS is capable of directly reducing FOSL1 phosphor-  FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8 ylation and accumulation, we ectopically expressed FOS in trophoblast cells and noted a marked down-regulation of FOSL1 and phospho-Ser 265 -FOSL1 despite a modest but significant increase in FOSL1 mRNA levels (Fig. 10, C and D; p Ͻ 0.05).

The Control of Trophoblast Invasion by FOS Family Members
Expression of FOS, FOSB, and FOSL1 in Human Placenta-To gain perspective of the spatiotemporal localization of FOS, FOSB, and FOSL1 in human placenta, we performed immuno-histochemistry using antibodies specific for each of these transcription factors on paraffin-embedded sections of human placentae collected at 6 (Fig. 11, A, D, G, J, and M), 12 (Fig. 11, B, E,  H, K, and N), and 39 ( Fig. 11, C, F, I, L, and O) weeks of gestation. Fig. 11, A, B, and C depict images that detail landmark structural components at 6 (Fig. 11A), 12 (Fig. 11B), and 39 ( Fig. 11C) weeks of gestation. At 6 weeks of gestation, trophoblast cells located within the core of the cell columns emanating from villi were intensely immunopositive for FOS (Fig. 11D). This staining closely co-localized with phospho-ERK (Fig. 11J) and the proliferation marker proliferating cell nuclear antigen (not shown). Villous cytotrophoblast cells were also positive for FOS, phospho-ERK, and proliferating cell nuclear antigen but at lower levels than column extravillous trophoblast cells. FOSB was diffusely present within the core of the cell columns (Fig.  11G). Syncytiotrophoblast, representing the outer strata of the trophoblast bilayer on the periphery of the villi, was intensely positive for FOSB at all gestational ages. FOSL1 was lowly expressed at 6 weeks of gestation. Expression was limited to cells on the periphery of the cell columns.
At 12 weeks of gestation, strong expression of FOS and FOSB was present within proximal extravillous trophoblast cells in the core of the cell columns (Fig. 11, E and H, respectively). Similar to our observation at 6 weeks of gestation, FOS expression co-localized with phospho-ERK expression (Fig. 11N). The number of FOS-and FOSB-immunopositive cells seemed to subside near the periphery of the extravillous trophoblast outgrowth where cells seemed detached from the cell column and   were gaining migratory capacity. At this site, FOSL1 was highly expressed (Fig. 11K).
In term placentae (39 weeks), FOS, FOSB, and phospho-ERK were detected in invaded interstitial extravillous trophoblast cells situated within the placental bed (Fig. 11, F, I, and O, respectively). Similar to our earlier observations, FOSB was also detected in syncytiotrophoblast. FOSL1 expression was notably absent in term human placenta (Fig. 11L).

DISCUSSION
In this report, we have demonstrated that FOS family members distinctively contribute to the control of trophoblast invasion. The results insinuate that extravillous trophoblast behavior is highly dependent on the expression of specific FOS family members and that different AP-1 permutations impact extravillous trophoblast phenotypes in discrete ways.
In our initial screen of seven trophoblast cell lines, we identified high levels of FOSL1 in all three first trimester extravillous trophoblast lines. Each of these lines was derived by immortalization of isolated first trimester trophoblast cells (48). They all express cytokeratin, consistent with their classification as epithelial cells, and are highly invasive in vitro but also display features consistent with cells that have transitioned from an epithelial to a mesenchymal phenotype (49). Our in situ analysis identified FOSL1-positive extravillous trophoblast cells at the distal-most edges of the anchoring villi in first trimester placentae, but whether they share mesenchymal fea-tures and are thus related to the first trimester immortalized extravillous cell lines requires further investigation. At this juncture, it is difficult to ascertain whether FOSL1 expression is a consequence or a direct cause of the mesenchymal characteristics associated with human first trimester extravillous cell lines. In support of the latter, FOSL1, but not FOS, was shown to be a key mediator in epithelial-mesenchymal transition in human breast and colon cancer lines (50,51). Moreover, ectopic expression of FOSL1 causes adenocarcinoma cells to adopt morphological changes reminiscent of mesenchymal cells together with modulation of gene expression signatures and enhanced cell motility and invasion (19). Thus, high levels of FOSL1 expressed in first trimester extravillous trophoblast cell lines may be at least partly responsible for their mesenchymal morphology, high levels of in vitro invasiveness, and unique gene expression profiles. In line with this, we found that FOSL1 knockdown substantially affected morphology, motility, and invasiveness of these cells.
Based on comparing relative accrual of transcript and protein, we determined that increased FOS and FOSB expression was likely mediated primarily through transcript induction or stabilization, whereas FOSL1 accumulation was mediated likely by post-translational modifications (e.g. phosphorylation) and protein stabilization. Accumulation of all FOS family transcription factors was dependent on ERK activity, which is consistent with other reports (46). In human placenta, however, only FOS expression correlated with phospho-ERK in extravillous trophoblast cells. FOS and phospho-ERK also co-localized with the proliferation marker proliferating cell nuclear antigen, strongly signifying that an ERK-FOS pathway may be important in stimulating proliferation of extravillous trophoblast cells. Accordingly, FOS knockdown substantially abrogated the capacity of trophoblast cells to proliferate. FOS expression is vital for the proliferative capacity of many cell types in part by regulating the transcriptional activation of cyclin genes and repression of cyclin-dependent kinase inhibitors (52)(53)(54)(55). However, in other cells, FOS seems to be dispensable (56,57) or even down-regulates proliferation (58,59). The effect of FOS on proliferation may depend to some degree on the intracellular abundance of other FOS and JUN proteins (56,60). Interestingly, we found that FOSL1 knockdown also significantly inhibited proliferation albeit to a lesser extent than FOS knockdown. The similar proliferation phenotypes observed with FOS and FOSL1 knockdown may be explained by shared transcriptional targets. Indeed, genetic complementation studies in mice have shown an overlapping function between FOS and FOSL1 in bone formation (61). In this study, we determined that knockdown of either FOS or FOSL1 resulted in a robust increase of the cyclin-dependent kinase inhibitor CDKN1A. CDKN1A encodes p21 (CIP1/WAF1), a potent inhibitor of cell cycle progression, whose expression is negatively associated with trophoblast proliferation (62). Additional studies are needed to determine whether increased expression of CDKN1A is directly responsible for the suppressed proliferation following knockdown of either FOS or FOSL1. In extravillous trophoblast cells, FOS and FOSL1 may each have a role in the regulation of genes governing progression through the cell cycle and consequently are both relevant for the promotion of proliferation.  FEBRUARY 21, 2014 • VOLUME 289 • NUMBER 8

The Control of Trophoblast Invasion by FOS Family Members
One of the most intriguing findings from this study was that FOS actively suppressed the invasive phenotype of trophoblast cells. At first glance, the transient accumulation of FOS following scratch wounding or Matrigel plating seems counterintuitive. However, the capacity to control invasion is a defining feature distinguishing the regulated invasiveness of trophoblast cells from cellular invasiveness during pathological situations like carcinoma progression. The mechanism by which FOS suppresses the invasive phenotype of trophoblast cells is at this point speculative and requires additional studies. However, we noted a substantial accumulation of FOSL1 in trophoblast cells following knockdown of FOS. Moreover, ectopic expression of FOS caused decreased FOSL1 expression. Therefore, it is possible that the repressive actions of FOS on trophoblast invasion may stem from restraining FOSL1 accumulation and will be the subject of future investigations. If this hypothesis is true, this places greater emphasis on FOSL1 as a master regulator of gene expression networks that promote trophoblast invasion. Fittingly, in first trimester human placentae, FOS was highly detected in most extravillous trophoblast cells within the proximal cell columns, consistent with its essential role in trophoblast proliferation. Distally, FOS expression waned whereas FOSL1 expression increased specifically in cells that were detaching from the column and gaining migratory capacity.
Promotion of trophoblast invasion during the first half of pregnancy is essential to establishing a sufficient maternal-fetal blood supply, but equally as important for maternal health is appropriate cessation of trophoblast invasion during the latter half of pregnancy. In term placentae, FOSL1 expression was not detectable in any trophoblast cells, including those that had embedded within the decidua. Conversely, FOS and FOSB were readily detected in these cells. It is possible that cessation of invasion in the latter half of pregnancy is accomplished in part by re-expression of FOS (and FOSB), which actively represses FOSL1 accumulation. Interestingly, high expression of FOSL1 is observed in gestational disorders associated with hyperinvasive trophoblasts, including molar pregnancy and choriocarcinoma (26). In this case, insufficient FOS expression may pathologically fail to disable FOSL1 accumulation in trophoblast cells, causing sustained expression of gene networks that promote invasion.
Putative transcriptional targets of FOS family members include a multitude of genes affecting cell proliferation, survival, and migration. In this report, we studied FOS familymediated control of MMP expression as an illustration of disparate regulation of these genes by individual FOS family members. We focused on MMPs because they are highly linked with cell invasion and tissue remodeling in a variety of biological systems, including extravillous trophoblast cells (63)(64)(65), and several of these proteases contain well characterized AP-1 binding sites within their proximal promoter regions. In this study, we found that MMP1 (collagenase-I), MMP3 (stromelysin-1), and MMP10 (stromelysin-2) were rapidly induced after trophoblast cells were subjected to a wound-based migration assay or plated on Matrigel. These MMPs are situated in series on chromosome 11, and each has a conserved AP-1 site within their respective proximal promoter regions: 5Ј-TGAGTCA-3Ј 71 and 68 bp upstream of MMP1 and MMP3, respectively, and 5Ј-TGAATCA-3Ј 67 bp upstream of MMP10 (66). The structural similarities and chromosomal proximity suggests that these genes arose through duplication. Occupation of this site by AP-1 is essential for transcription of these MMPs (7). In this study, we noted enrichment of FOSL1 at this AP-1 site. We also observed that FOSL1 knockdown in trophoblast cells substantially reduced the expression of MMP1, MMP3, and MMP10, whereas FOS knockdown had the opposite effect. Therefore, FOSL1 binding to this AP-1 site is likely necessary for robust transcription of these genes. However, a high degree of cooperativity with other cis-acting elements, including other putative AP-1 sites located more distally along the promoters of these genes (66), as well as interactions with other transcription factors (67) is also required for cell-and tissue-specific control of MMP expression. Although the regulation of MMPs was used in this study as a paradigm of how individual FOS family members can differentially modulate their expression in trophoblast cells, we predict that other genes regulated by AP-1 transcriptional activity that encode proteins promoting invasion may also display differential regulation by individual FOS family members.
In conclusion, it is clear that individual FOS transcription factor family members have diverse roles in various facets of trophoblast biology. Although FOSB was readily induced along with FOS and FOSL1, we did not observe a substantial phenotypic change in trophoblast behavior following FOSB knockdown. It is possible that other FOS family members are capable of sufficient compensation in its absence. Moreover, although FOSB expression was diffusely present within the extravillous trophoblast population, high levels were also detected in syncytiotrophoblast. Therefore, more information is required to determine exactly what role FOSB plays in placentation. How-ever, we provide compelling evidence that FOS and FOSL1 engage in a physiological "tug of war" to regulate the extent of trophoblast invasion. In vivo, their expression in extravillous trophoblast cells is mutually exclusive and is identified at the right place and at the right time to exert their respective actions on trophoblast proliferation and motility. Our data build on a previous report by our laboratory that identified FOSL1 as a master transcriptional regulator of trophoblast invasion in the rat (23). Therefore, FOSL1 may have a conserved role in the regulation of trophoblast invasion in species exhibiting hemochorial placentation. Much more information is still required to fully understand how FOS and FOSL1 fit into a hierarchical cascade of events controlling trophoblast invasiveness, including whether dysregulation of these factors is implicated in aberrant trophoblast invasion linked with the etiology of placentaassociated obstetrical complications.