Inhibition of eIF5A results in aberrant uterine NK cell function and embryo loss in mice

Abstract

Problem

The role of eukaryotic initiation factor 5A (eIF5A) in feto-maternal immunotolerance is poorly understood.

Methods of study

The effects of N1-guanyl-1,7-diaminoheptane (GC7), an inhibitor of eIF5A, on the proportion and function of natural killer (NK) cell subsets were investigated using flow cytometry, immunofluorescence, CCK8 assay, TUNEL assay, DNA fragmentation analysis, mitochondrial membrane potential assay, and western blotting.

Results

Inhibition of eIF5A by GC7 increased embryo loss and reduced the percentage of NK cells in the uterus and spleen. GC7 treatment caused inhibition of NK cell proliferation in a time- and dose-dependent manner. GC7 also induced apoptosis of NK cells. GC7 treatment increased the protein levels of FasL, bax, p53, and cleaved caspase 3. Moreover GC7 caused loss of mitochondrial membrane potential in NK cells.

Conclusion

Inhibition of eIF5A results in aberrant NK cell function and increased embryo loss.

Introduction

Eukaryotic initiation factor 5A (eIF5A) is a highly conserved protein throughout evolution. It is present in archaea and mammals, but not in eubacteria.¹ Although this protein was originally identified as a translation initiation factor, subsequent studies did not support a role for eIF5A in general translation initiation. To date, its main cellular function remains unclear.²

eIF5A is the only protein that contains a hypusine residue.^2,3 Deoxyhypusine synthase (DHS), the first enzyme involved in the hypusination of eIF5A, catalyzes the transfer of the 4-aminobutyl moiety of spermidine to the ε-amino group of Lys50 of eIF5A. The resulting deoxyhypusine residue is then hydroxylated by a second enzyme, deoxyhypusine hydroxylase, to produce hypusinated eIF5A.⁴ eIF5A has two isoforms in mice and humans: eIF5A1 and eIF5A2. Both isoforms harbor hypusine modification. DHS is the key enzyme for hypusine formation. Among the known inhibitors of DHS, N1-guanyl-1,7-diaminoheptane (GC7) is the most potent and the most specific.⁵

In our previous studies, we identified the differentially expressed proteins in uterine lymphocytes isolated from NOD mice and wild-type BALB/c counterparts using two-dimensional gel electrophoresis and mass spectrometry analysis. Compared to wild-type BALB/c mice, eIF5A expression was significantly lower in uterine lymphocytes in NOD mice that are prone to spontaneous embryo loss. We hypothesized that a lower expression of eIF5A is associated with adverse pregnancy outcome.⁶ In this study, we used the eIF5A inhibitor, GC7, to investigate the function of eIF5A in pregnancy. We found that GC7 injection increased embryo resorption in BALB/c mice. NK cells are the most abundant in decidua and uterine NK (uNK) cells are believed to play an essential role in feto-maternal tolerance.⁷ Precursors of mouse uNK cells do not self-renew in the uterus, but come from secondary lymphoid organs, particularly spleen.^{8, 9} We found that GC7 caused a decreased proportion of NK cells in the uterus and spleen. To further understand the mechanisms, we investigated the effects of eIF5A on NK cells in vitro.

Materials and Methods

Preparation of GC7 compounds

GC7 (259545; Merck, Darmstadt, Germany) was dissolved in 10 mM acetic acid at a concentration of 125 mM as stock solution and stored in aliquots at −20°C.¹⁰

Animal administration and determining the rate of embryo resorption

Female BALB/c mice and male C57BL/6 mice (8–12 weeks old) were purchased from the Shanghai Laboratory Animal Center, Chinese Academy of Sciences (Shanghai, China) and subsequently maintained under specific pathogen-free conditions in the Laboratory Animal Center of Renji Hospital (Shanghai, China). All animal procedures followed the national animal care guidelines and permission was obtained from the ethics committee of Shanghai Jiaotong University. The university’s Institutional Review Board approved all associated data for publication.

Detection of a vaginal plug was chosen to indicate gestational day 0.5 (E0.5). The GC7 stock solution was diluted with saline to obtain the final concentrations before injection. A small mouse catheter (PUFC-C20-10; Instech Laboratories, Plymouth Meeting, PA, USA) is gently inserted into uterine cavity via vaginal and GC7 (6 or 12 mg/kg in a volume of 40 µL) or equal volume of solvent is given at E4.5, E5.5, and E6.5, respectively.

Pregnant mice were sacrificed on E10.5 and the embryo resorption was determined. The resorbed embryos were macroscopically identified by their small size (< 20% of the average size), hemorrhage (a dark brown blood clot at the implantation site) and necrosis.¹¹ The embryo resorption rate was calculated as follows: Resorption rate = (number of resorbed embryos / number of total embryos) × 100.

Purification of lymphocytes and flow cytometry

The pregnant mice treated with or without GC7 were killed at E10.5. Hysterolaparotomy was performed to collect embryo-depleted placentas and associated decidual tissue, including the decidua basalis. The tissues were carefully cut into small pieces (<1mm³) using ocular scissors and filtered through a 50-µm-pore nylon mesh to obtain a single cell suspension. Mononuclear cells were purified with a ficoll-hypaque density medium (catalog No. DH184-1; Dingguo Biotech, Beijing, China). Any contaminating red blood cells that might have persisted in the single cell suspension were eliminated by incubation with red cell lysing buffer (catalog No. 555899; BD Biosciences, Franklin Lakes, NJ, USA).

The splenic tissue was harvested and disrupted by pressing it through a 70 µm cell strainer (352350; BD Biosciences) into a standard petri dish containing mouse lymphocyte separation medium (DKW33-R0100; Dakewe Biotech, Shenzhen, China). Mononuclear cells were isolated by centrifugation of a single cell suspension in gradient medium. Similarly, any red blood cells that contaminated the single cell suspension were eliminated by incubation with red cell lysing buffer.

Anti-CD49b-PE (clone: DX5), anti-CD3-FITC (clone: 145-2C11) and anti-CD45-APC (clone: 30-F11) antibodies were purchased from eBioscience (San Diego, CA, USA). The cells were resuspended in PBS containing 2% fetal bovine serum (FBS). A combination of antibodies were added for extracellular staining for 30 min. Stained cells were assessed by flow cytometry using CXP software (Beckman Coulter, Brea, CA, USA).

Purification of NK cells and cell culture

NK cells were purified under sterile conditions from spleens at gestational day 10.5 using a mouse NK cell isolation kit II (130-096-892; Miltenyi Biotec, Bergisch Gladbach, Germany), according to the manufacturer’s instructions. The purity of the NK cells routinely exceeded 95%, as determined using flow cytometry.^{12, 13}

NK cells were seeded in suspension multiple-well plates and cultured in RPMI 1640 complete medium supplemented with 10% FBS (10099-141; Gibco, Australia), 100 U/ml penicillin, 100 µg/mL streptomycin (SV30010; HyClon, Logan, UT, USA) and 200 U/ml recombinant murine IL-2 (212-12; Peprotech, Rocky Hill, NJ, USA). Cells were seeded at a concentration of 1×10⁶ cells/ml and grown at 37°C in a humidified incubator with 5% CO₂.

GC7 stock solution was diluted with complete medium to obtain the indicated concentrations before use. GC7 is a substrate of serum amine oxidase (SAO); therefore, to ensure SAO to be inhibited, aminoguanidine (396494; Sigma-Aldrich, St Louis, MO, USA), an inhibitor of SAO, was added at a concentration of 1 mM to all cultures (including in the control group).¹⁴ In subsequent experiments, the final GC7 concentrations were 20, 30 and 40 µM, respectively. An equal volume of solvent was used as a negative control.

Immunofluorescence of eIF5A

NK cells were treated for 6 hr with the indicated concentrations of GC7 or without GC7. The cells were then washed and resuspended in cold PBS. The cell suspensions were centrifuged onto the deposition area of the slide using a cytospin apparatus. The slides were fixed in cold acetone for 5 min, and were permeabilized with 0.1% Triton X-100 for 10 min. The cells were incubated with rabbit anti-eIF5A1 (ab32443) or eIF5A2 (ab126735) monoclonal antibody (Abcam, Cambridge, MA, USA; dilution1:50) at 37°C for 1 hr. After washing three times with PBS, the cells were incubated with Alex Flour 488-donkey anti-rabbit IgG (98593; Jackson, West Grove, PA, USA; dilution1:200) for 30 min at 37°C in the dark. The nuclei were stained with 4', 6-diamidino-2-phenylindole (DAPI; GMS10011.2; GenMed Scientifics, Arlington, MA, USA) for 10 min in the dark. Negative controls were established using rabbit Ig of the isotype identical to the primary antibody. The cells were visualized on a LSM 710 confocal laser scanning microscope (Carl Zeiss, Germany).

Cell proliferation assay

Cell growth inhibition rate (IR) was measured using a cell counting kit-8 (CCK8; CK04; Dojindo laboratories, Kumamoto, Japan). NK cells isolated by magnetic affinity cell sorting were seeded onto 96-well plates. After treatment with various concentrations (20, 30 and 40 µM) of GC7 for 6, 12, 18 and 24 hr, 10 µL of CCK8 was added to the cells. The absorbance of the solutions at 450 nm was measured using a microplate reader. The IR was calculated as: IR = [1− (value in the treated samples −value in the blank samples) / (value in the control samples −value in the blank samples)] × 100%. Each assay was performed in triplicate.

Annexin V-PI analysis

We used Alexa Fluor 488-conjugated annexin V and propidium iodide (PI; V13241; Invitrogen, Eugene, OR, USA) staining assessed by flow cytometry to measure the percentage of early apoptotic NK cells. After treatment for 12 hr with the indicated concentrations of GC7 or without GC7, the NK cells were harvested and washed twice with cold PBS. The cells were resuspended in 100 µL of 1× binding buffer supplemented with annexin V and PI, and incubated for 15 min at room temperature in the dark. Flow cytometric analysis was performed as described previously.

TUNEL assay

The terminal deoxynucleotidyl transferase (TdT) mediated dUTP nick end labeling (TUNEL) method was used (in situ cell death detection kit: 11684795910; Roche, Mannheim, Germany) to detect apoptotic cells. NK cells were collected at 12 hr after treatment with or without GC7, as described previously. The cells were resuspended in cold PBS and cytospin slides were prepared. All procedures were conducted following manufacturer’s instructions. Cells were viewed using a confocal microscope.

DNA fragmentation analysis

Fragmented oligonucleosomal DNA was extracted using an apoptotic DNA laddering kit (C1102; Applygen Technologies, Beijing, China). Fragmented DNA with loading buffer was electrophoresed on 1% agarose gel at 1.5 V/cm. DNA in the gels was visualized by Bio-rad molecular imager (GelDoc XR, Bio-Rad, USA) after staining with GelRed (41003; Biotium Inc, Hayward, CA, USA).

Detection of Fas/FasL expression using flow cytometry

The NK cells treated with or without GC7 for 12 hr were harvested and washed twice with PBS, and then supplemented with anti-Fas-FITC (clone: 15A7), anti-FasL-PE (clone: MFL3) antibodies or isotype controls (eBioscience), and incubated for 30 min at room temperature in the dark. The stained cells were analyzed using Beckman flow cytometer. Ten thousand cells from each sample were analyzed.

Measurement of mitochondrial transmembrane potential

Variations in mitochondrial transmembrane potential (Δψm) were examined using mitochondrial membrane potential assay kit with JC-I (C2006; Beyotime Biotechnology, Haimen, China). After the indicated treatments, NK cells were harvested and incubated with JC-1 staining solution at 37°C for 20 min and rinsed twice with PBS. The Δψm was measured using flow cytometry.

Western blotting

Primary antibodies used in western blotting included: rabbit anti cleaved caspase 3 (9664), bax (2772), bcl-2 (2870) antibodies, goat anti p53 (2524) antibody (Cell Signaling Technology, San Jose, CA, USA; dilution 1:1,000), and rabbit anti β-actin (P30002; Abmart, China; dilution 1:2,000). Secondary antibodies include IRDye 800CW-conjugated goat anti-rabbit IgG (926–32211) and donkey anti-goat IgG (926–32214; LI-COR Bioscience, Lincoln, NE, USA; dilution 1:10,000).

NK cells were treated with or without GC7 for 12 hr as described previously. The cells were then suspended in RIPA buffer (9806; Cell Signaling Technology) and sonicated three times in 5 s bursts. The cell lysates were centrifuged at 14,000 ×g for 15 min at 4°C. The supernatants were collected and their protein concentrations were measured using a BCA protein assay kit (P0010; Beyotime Biotechnology, Haimen, China).

Proteins were subjected to 12% sodium-dodecylsulphate polyacrylamide gel electrophoresis (SDS-PAGE) prior to transfer to a PVDF membrane. After blocking with 5% (w/v) fat-free milk for 1 hr, the blotting membrane was incubated with the appropriate primary antibodies listed above for 2 hr, followed by incubation in their corresponding secondary antibodies for 1 hr. Detection of β-actin was used as a loading control. Signals on the blots were visualized using Odyssey Imaging System. The amount of protein in each band on the western blot was analyzed using ImageJ software.

Statistical analysis

The resorption rate of embryos was compared using a χ² test. The cell percentages derived from flow cytometric analysis were further compared using independent sample t-test among groups. Data were present as mean ± standard deviation (SD). A p value of <0.05 was considered significantly different.

Results

Inhibition of eIF5A increased embryo resorption

To assess the effects of eIF5A on fetal resorption, the indicated doses of GC7 (6 or 12 mg/kg/d), or an equal volume of solvent as a negative control was injected at E4.5, E5.5 and E6.5. As shown in Fig. 1, a and b, fetal resorption was significantly higher in the 6 mg/kg/d GC7-treated group mice than in the control group (15.2%; 15 out of 99 vs. 6.5%; 7 out of 107; P<0.05). GC7 caused a further increase in fetal resorption in mice at the 12 mg/kg/d dose compared with control group (17.9%; 20 out of 112; n=10 for each group, P<0.01).

Fig. 1.

Inhibition of eIF5A induced fetal resorption in pregnant mice. Pregnant mice injected with solvent control or GC7 at 6 or 12 mg/kg on E4.5, E5.5, and E6.5. The mice were sacrificed on E10.5. (a) Representative uterine horns of control and GC7-treated mice harvested on E10.5. Resorbed embryos were pointed with arrows. (b) The resorption rate was calculated. * p < 0.05, and ** p < 0.01 compared with the control

Reduction of the uterine and splenic NK cell population by inhibition of eIF5A

To explore the possible impact of eIF5A on NK cells in vivo, we examined NK cell percentage in uterus and spleen by detecting surface markers CD3 and CD49b. We found that the percentage of uterine NK cells from GC7-treated mice was significantly decreased compared with those from solvent-control mice (Fig. 2, b and d). A decrease of similar magnitude was observed in splenic NK cells (Fig. 2, c and e).

Fig. 2.

Inhibition of eIF5A decreases uterine and splenic NK cell populations in pregnant mice. Indicated doses of GC7 (6 or 12 mg/kg/d) or solvent control was injected at E4.5, E5.5 and E6.5. The samples were collected at E10.5. Mononuclear cells were isolated and analyzed by flow cytometry. (a) Cells were gated on CD45+ cells. (b and c) Representative flow-cytometry plots of uterine and splenic samples, respectively. (d and e) Data summary of uterine and splenic NK cells, respectively. * p < 0.05, **p < 0.01 compared with control

eIF5A expression in NK cells

To examine subcellular distribution of eIF5A in the NK cells, we analyzed stained samples using confocal fluorescent microscopy. Both eIF5A1 and eIF5A2 were detected in NK cells. Non-specific staining was assessed using isotype-matched rabbit IgG (Fig. 3a). eIF5A1 was primarily located in the cytoplasm of the untreated NK cells (Fig. 3b); however, it was found to be distributed diffusely throughout the whole cell in some NK cells treated with 20 µM GC7 (Fig. 3c). Crescent-shaped chromatin aggregates that lined the nuclear membrane were observed in some 30 µM treated NK cells, along with the change in location of eIF5A1 (Fig. 3d). Nuclear segregation and fragmentation were observed in some NK cells treated with 40 µM GC7. In addition, eIF5A1 expression exhibited weak pattern (Fig. 3e). Similar trends were observed in the expression of eIF5A2.

Fig. 3.

eIF5A1 expression in NK cells. eIF5A1 expression was assessed using immunofluorescence with a monoclonal rabbit antibody specific for eIF5A1 or an isotype matched control (rabbit IgG). eIF5A1-specific staining displayed green fluorescence as visualized by goat anti-rabbit Alexa Fluor 488. Nuclei were stained with DAPI (blue). (a) Isotype matched control. (b) Negative control. (c) 20 µM, (d) 30 µM and (e) 40 µM GC7 treatment. Bar=20 µm. The last column of panels represents local magnifications of the regions enclosed within boxes on the merged column.

GC7 inhibited the proliferation of NK cells

The effects of eIF5A on NK cell proliferation were evaluated using CCK8 assay. It revealed that NK cell proliferation was significantly inhibited by GC7 at concentrations of 20, 30 and 40 µM in a dose- and time-dependent manner (Fig. 4).

Fig. 4.

Inhibition of eIF5A induced inhibition of NK cell proliferation. The effects of eIF5A on NK cell proliferation were analyzed using CCK8 assay. NK cells were incubated with various concentrations (20, 30 and 40 µM) of GC7 for 6, 12, 18 and 24 hr. The cell growth inhibition rate was calculated.

Inhibition of eIF5A induced apoptosis of NK cells

The effect of eIF5A on NK cell apoptosis was analyzed by various parameters. The percentage of cells showing early apoptosis was quantified using dual staining with annexin V and PI. GC7 significantly increased the percentage of NK cells showing signs of early stage apoptosis (Fig. 5, a and b). The TUNEL assay revealed the presence of late-stage apoptosis by staining free 3'-OH termini using fluorescein labeled nucleotides. These new DNA ends that are generated on DNA fragmentation are typically localized in morphologically identifiable nuclei and apoptotic bodies. In contrast, the normal NK cells that have relatively insignificant number of DNA 3'-OH ends were not stained in the above experiment.¹⁵ Our study indicated that TUNEL-positive NK cells increased significantly in GC7-treated group compared with those in the control group (Fig. 5c). Apoptosis is associated with the fragmentation of chromatin into multiples of the 180 bp nucleosomal unit, known as DNA laddering. In our study, apoptosis was confirmed by typical DNA laddering characteristics (Fig. 5d).

Fig. 5.

Inhibition of eIF5A induced apoptosis of NK cells. (a) Early apoptosis was analyzed using annexin V-PI staining by flow cytometry. (b) Percentage of annexin V+PI− NK cells. * p < 0.05, **p < 0.01 compared with control. (c) TUNEL positive NK cells (green) were visualized using confocal fluorescent microscopy. Cell nuclei were visualized with DAPI staining (blue). Bar=20 µm. (d) DNA fragmentation was assessed using agarose gel electrophoresis in apoptotic NK cells. lane 1, molecular weight marker; lane 2, control; lane 3, 20 µM; lane 4, 30 µM and lane 5, 40 µM GC7 treatment

The level of FasL and Fas protein expression in NK cells

Fas and FasL expression in NK cells were analyzed by flow cytometry. When treated with GC7 at a concentration of 30 and 40 µM, the percentage of FasL⁺ cells in the NK cell population was higher than the baseline level (Fig. 6, a and b). No significant difference was observed in the percentage of Fas⁺ cells between the GC7-treated group and control group (Fig. 6, c and d)

Fig. 6.

Fas and FasL expression in NK cells. NK cells were stained with FITC-conjugated Fas, PE-conjugated FasL antibodies or isotype controls, and analyzed by flow cytometry. (a and c) Representative results of flow cytometry. (b and d) Data summary of flow cytometry. (a and b) FasL. (c and d) Fas. ** p < 0.01 compared with control

Inhibition of eIF5A affected mitochondrial transmembrane potential in NK cells

In normal, undamaged nucleate cells, mitochondria have a high mitochondrial transmembrane potential (ΔΨm). Breakdown of ΔΨm is a characteristic of apoptosis.¹⁶ ΔΨm can be measured using the cell-penetrating lipophilic cationic fluorochrome JC-1. JC-1 stains mitochondria differentially in accordance to their ΔΨm. Active mitochondria with a high ΔΨm accumulate JC-1 aggregates and show red fluorescence, whereas mitochondria with low ΔΨm display the green monomeric form of JC-1.¹⁷ Changes in ΔΨm were evaluated by the JC-1 aggregate/JC-1 monomer fluorescence ratio. GC7 treatment of NK cells resulted in decreased ratio of red over green fluorescence, indicating a loss of ΔΨm (Fig. 7).

Fig. 7.

Inhibition of eIF5A induced loss of mitochondrial transmembrane potential (Δψm) in NK cells. NK cells were stained with JC-1 dye and Δψm was evaluated by the JC-1 aggregate/JC-1 monomer fluorescence ratio. * p < 0.05, and ** p < 0.01 compared with control.

Effect of eIF5A on the expression status of apoptosis regulators

Western blotting was performed using NK cell lysates. Cleaved caspase 3, p53, and bax expression was up-regulated in GC7-treated NK cells and the bax/bcl-2 ratio was also increased (Fig. 8, a and b).

Fig. 8.

Effects of eIF5A inhibitor on the expression of cleaved caspase 3, p53, bax, and bcl-2 in NK cells. (a) Cell lysates extracted from control or the GC7-treated NK cells were used for western blotting. Casp-3, caspase 3. β-actin was used as a loading control. (b) Histogram shows the relative expression level of the proteins as determined by densitometric analysis. * p < 0.05, and ** p < 0.01 compared with control

Discussion

The true physiological functions of eIF5A and its mode of action in mammals remains unclear. Its two isoforms, eIF5A1 and eIF5A2, appear to have different functions. eIF5A1 has been implicated in translation initiation, nuclear export, mRNA decay, cell apoptosis and cell-cycle progression, but most of the findings are controversial or unconfirmed.^{2, 3} eIF5A2 possibly correlates with the aggressiveness of tumors.¹⁸ In this study, we investigated the effect of eIF5A on abortion in mice using GC7, a specific DHS inhibitor, which selectively inhibits eIF5A activation. Previous studies have confirmed that eIF5A hypusination is inhibited by GC7 at the dosage used in this study.^{1, 10, 19}

eIF5A is unique in that it is the only known protein to contain the unusual amino acid, hypusine. Hypusine modification converts an inactive precursor to an active eIF5A, and eIF5A exists mainly as the fully hypusinated form in mammalian tissues.²⁰ Lee and colleagues²¹ presented evidence that hypusinated eIF5A is localized predominantly in the cytoplasm, whereas the non-hypusinated eIF5A precursor was found in nucleus and cytoplasm. Subcellular localization of eIF5A may correlate with its potential function in nuclear export. In our study, similar results were found: eIF5A was mostly localized in cytoplasm of untreated NK cells. However, it was found in both cytoplasm and nucleus after treated with 20 µM GC7 for 6 hr.

uNK cells are thought to be crucial for successful pregnancy. They are key mediators of maternal immune system interactions with fetal cells. uNK cells regulate trophoblast invasion both in vitro and in vivo by production of interleukin-8 and interferon-inducible protein-10 chemokines. uNK cells also control the development of the placenta.²² In addition, they are involved in remodeling of maternal spiral arteries to uteroplacental arteries, which links the placenta to blood vessels to provide a healthy supply line to the fetus.²³ Therefore, this prompted us to explore the effect of GC7 on the NK cells. We compared the CD3⁻CD49⁺ cell percentage in pregnant murine uterus. The uNK cell proportion was significantly lower in the GC7-treated mice than the solvent controls. A similar change in splenic NK cells was observed. Furthermore, we observed a reduction in NK cell proliferation as a result of treatment with GC7 in vitro experiment. GC7 significantly inhibited NK cell proliferation in a dose- and time-dependent manner. These results are in agreement with other studies on the growth inhibitory effects of GC7.²⁴ Recent studies have shown that depletion of eIF5A causes translation elongation, possibly via a mechanism whereby eIF5A interacts physically with the 80S ribosome and the translation elongation factors eEF1A and eEF2.^{2, 3} It has also been proposed that the cell proliferation regulatory properties of eIF5A could be correlated by its reported mRNA transport functions because the mRNAs could encode for proteins involved in the regulation of cell proliferation.²⁵

Apoptosis is a biological phenomenon in which cell death is genetically and biochemically regulated. Apoptosis is also important in the immune system. To determine if apoptosis is also involved in the mechanism of GC7 action, we investigated the role of GC7 in inducing apoptosis in murine NK cells. Apoptosis, a form of programmed cell death, can be seen as a stage-dependent process from its induction to early, intermediate and late-stage apoptotic events. One of the earliest indicators of apoptosis is phosphatidyl serine (PS) translocation from the inner to the outer leaflet of plasma membrane. Annexin V can be used for PS detection. Chromosomal DNA fragmentation is an irreversible step in apoptosis. Two types of apoptosis have been defined: one type results in small DNA fragments associated with a typical ladder in agarose gel electrophoresis which is known as DNA laddering, and in the other type, DNA cleavage may be absent or incomplete in some forms of apoptotic cell death.²⁶ Besides, the TUNEL assay labels the 3'-hydroxyl termini generated during apoptosis. Therefore, TUNEL assay detects the late stage of apoptosis at the single cell level. Apoptosis eventually converge at the level of caspase-3 activation. Inhibition of caspase 3 by synthetic peptide inhibitors often prevents apoptosis induced by various stimuli. Our results showed that inhibition of eIF5A effectively induced the typical features of apoptosis characterized by the increase of annexin V⁺PI⁻ and TUNEL positive NK cells population, up-regulation of activation of caspase 3 and remarkable apoptotic DNA laddering. In addition, apoptotic cells characterized by dense, crescent-shaped chromatin aggregates along the nuclear envelope²⁷ and nucleus disintegrates were observed in GC7-treated NK cells (Fig. 3, d and e). It was reported that over-expression of eIF5A without co-over-expression of DHS and DOHH caused accumulation of unmodified eIF5A precursors and induce apoptosis. In addition, a mutant of eIF5A that cannot be hypusine-modified also increase apoptosis. Therefore, they concluded from it that an accumulation of unhypusinated eIF5A, but not active eIF5A, appears to be proapoptotic.^{28, 29} Consistent with this hypothesis, GC7 can reduce hypusinated eIF5A and increase levels of unhypusinated and induced NK cells apoptosis. However, previous reports about the effect of GC7 on apoptosis are different. It has showed that GC7 pretreatment elicited protective effects against apoptotic death of HUVEC induced by serum starvation.³⁰ GC7 synergistically induces apoptosis in interleukin-3 dependent pro-B cell line (BA/F3) cells, but not in wt-BA/F3 cells.³¹ Some other studies show that GC7 treatment induced apoptosis.²⁹ The different apoptotic responses to GC7 may dependent on differences in the proliferation pattern of cell types, as well as in the mechanisms of GC7-induced apoptosis.

We then explore the pathways underlying GC7-induced apoptosis in NK cells. There are two common pathways that lead to apoptosis, extrinsic and intrinsic pathways. The extrinsic pathway is activated by death receptor ligation, whereas the intrinsic pathway is also called mitochondrial apoptosis. The present study demonstrated the involvement of the extrinsic pathway in the GC7-induced NK cell apoptosis. It is known that Fas/FasL system is a key signal pathway in lymphocyte apoptosis. FasL binding to Fas leads to activation of caspase proteases and initiates apoptosis of the cells that express membrane Fas. This can occur via Fas/FasL interaction on the cell surface in an intracellular or intercellular fashion. FasL plays a crucial role in immune regulation by triggering autocrine suicide or paracrine death in lymphocytes or other target cells.³² We found that NK cells express both Fas and FasL and that there was a significant increase in the percentage of FasL⁺ cells in GC7-treated NK cells.

Furthermore, treatment with GC7 resulted in up-regulation of bax/bcl-2 ratio and disruption of Δψm, which also supports the involvement of the intrinsic pathway in GC7-treated NK cells. The intrinsic pathway emerges from mitochondrial stress, which is mediated by the members of the bcl-2 family. Different members in the bcl-2 family play a role oppositely. Bcl-2, an apoptosis inhibitor located on the mitochondrial outer membrane, prevents cytochrome c release, caspase activation, and cell death. In contrast, bax, another member in the bcl-2 family, is considered the most potent mediator of apoptosis. Bax promotes apoptosis by interacting with bcl-2 and forming a heterodimer. Healthy cells are regulated by the balanced presence of pro- and anti-death bcl-2 family proteins. An increased bax/bcl-2 ratio promotes apoptosis. p53 responds to cell damage in two ways, either arresting the cell cycle allowing ample time for DNA repair or inducing apoptosis. p53 is also involved in the mitochondrial apoptotic pathway. p53 can mediate apoptosis by activating APAF1 and BAX genes expression, the gene products of which stimulate release of Apaf1 protein and cytochrome c from mitochondria.³³ Recent studies have also indicated apoptosis is regulated in a p53-dependent and a p53-independent manners. In our study, we found that GC7 increases the level of p53, suggesting that eIF5A may play a key role in promoting p53-mediated NK cell apoptosis.

In conclusion, this study provide evidence that eIF5A plays a role in preventing NK cell apoptosis and promoting NK cell proliferation. GC7-induced NK apoptosis is mediated by both the extrinsic and intrinsic pathways. Inhibition of eIF5A results in aberrant NK function in feto-maternal microenvironment, which may partially accounts for the GC7-induced abortion in mice. Besides NK cells, several types of immune cells, such as T cells, macrophage, dendritic cell, have been reported as participants in the fetal-maternal interface immune. There may exists a complex dialog among these different cell types in the pregnant uterus.³⁴ Caraglia and colleagues³⁵ reported that interferon α (IFNα) decreased hypusine synthesis of eIF5A, and epidermal growth factor (EGF) treatment can reverse these effects in human epidermoid cancer KB cells, while EGF alone has little effect on hypusine synthesis. A recent report indicated that eIF5A plays a role in tumor necrosis factor α (TNFα) induced apoptosis.³⁶ It is conceivable that there exists a complex signaling network in the response to eIF5A. Further studies are warranted to define the role of eIF5A at the fetal-maternal interface.