eIF2α signaling regulates ischemic osteonecrosis through endoplasmic reticulum stress

Osteonecrosis of the femoral head (ONFH) primarily results from ischemia/hypoxia to the femoral head, and one of the cellular manifestations is the endoplasmic reticulum (ER) stress. To understand possible linkage of ischemic osteonecrosis to the ER stress, a surgery-induced animal model was employed and salubrinal was administered to evaluate the role of ER stress. Salubrinal is a synthetic chemical that inhibits de-phosphorylation of eIF2α, and it can suppress cell death from the ER stress at a proper dose. The results indicated that the ER stress was associated with ONFH and salubrinal significantly improved ONFH-induced symptoms such as osteonecrosis, bone loss, reduction in vessel perfusion, and excessive osteoclastogenesis in the femoral head. Salubrinal also protected osteoblast development by upregulating the levels of ATF4, ALP and RUNX2, and it stimulated angiogenesis of endothelial cells through elevating ATF4 and VEGF. Collectively, the results support the notion that the ER stress is an important pathological outcome in the surgery-induced ONFH model, and salubrinal improves ONFH symptoms by enhancing angiogenesis and bone healing via suppressing the ER stress.


Results
Salubrinal has no significant effects on bone metabolism of normal animals. The body weight of animals in the sham control and salubrinal-treated sham control groups were not significant difference (P > 0.05, Supplemental Fig. 1a). For pDEXA scanning, BMD and BMC of the femurs in these two groups were not significant difference (P > 0.05, Supplemental Fig. 1a,b). The images of H&E staining indicated that the general structure and B.Ar/T.Ar of the femoral heads were not significant difference in these groups (P > 0.05, Supplemental Fig. 1d,e). The result indicated that salubrinal had no significant effects on bone metabolism of normal animals. We thus did not include salubrinal-treated sham control groups in the subsequent experiments.

Salubrinal suppresses bone loss in ONFH.
In the sham control group, the femoral head and neck surface was not detectably altered. However, the femoral head surface in the ON group exhibited "moth eaten" appearance and the femoral neck was thinned. Compared to the ON group, the salubrinal-treated group suppressed osteonecrosis-driven bone loss in the head and neck (Fig. 1a). In the ON group, the values of BV/TV, Tb.N, Tb.Th, and BMD/BMC were decreased, and the value of Tb.Sp was increased (all P < 0.01). Salubrinal significantly increased the values of BV/TV, Tb.N, and Tb.Th, as well as the change of femoral BMD/BMC. It also decreased the value of Tb.Sp (all P < 0.05; Fig. 1b-g). Regarding the results of H&E staining in the femoral head (Fig. 1h,j), the ON group exhibited the lowest B.Ar/T.Ar with the highest percentage of empty lacunae (both P < 0.001). The salubrinal-treated group significantly improved those values (both P < 0.001; Fig. 1i,k).

Salubrinal inhibits osteoclastogenesis in ONFH.
To investigate whether salubrinal affected the bone resorption of osteoclasts in ONFH in vivo, we conducted TRAP staining (Fig. 2a). The osteoclast surface (%) was increased in the ON group (P < 0.01), while this increase was suppressed in the salubrinal-treated group (P < 0.05; Fig. 2b). Regarding TRAP-positive multinuclear cells (more than 3 nuclei) identified as matured osteoclasts in osteoclast formation in vitro assay (Fig. 2c), the ON group significantly increased the osteoclast area ( Fig. 2d) and the osteoclast number ( Fig. 2e) (both P < 0.001), but the salubrinal-treated group significantly suppressed their increases (both P < 0.001). To determine whether salubrinal affected the proliferation and population of osteoclast progenitors, we counted CFU-M/CFU-GM. The numbers of CFU-M/CFU-GM were increased in the ON group (both P < 0.01), while salubrinal significantly reduced them (both P < 0.01; Fig. 2f,g).
To evaluate whether salubrinal affected the function of osteoclast lineage in ONFH, we performed migration, adhesion, and pit formation assays. In the ON group, migration (Fig. 2h) and adhesion (Fig. 2i) of pre-osteoclasts were stimulated, and an increased capacity for bone resorption was observed in the pit formation assay (all P < 0.01; Fig. 2j). Those activities were significantly reduced in the salubrinal-treated group (all P < 0.01; Fig. 2k-m).

Salubrinal promotes osteoblast differentiation in ONFH.
To evaluate the differentiation of osteoblasts and fibroblasts from mesenchymal stem cells, osteoblasts were identified using MacNeal's staining in vivo (Fig. 3a) or ALP staining in vitro (Fig. 3b), and fibroblasts were stained using a HEMA-3 quick staining kit. The number of osteoblasts (Fig. 3a,b) and the number of CFU-F (Fig. 3c) were not significantly different in the sham control and ON groups (all P > 0.05). However, the numbers were significantly increased in the salubrinal-treated group (all P < 0.01; Fig. 3d-f).

Salubrinal promotes osteogenesis by increasing p-eIF2α in vitro.
To evaluate whether salubrinal improved osteogenesis against the ER stress, cell viability and the expression of osteoblast markers were determined. Tunicamycin was used to activate ER stress signaling, and Calyculin A (CA, a specific inhibitor of PP1) was employed as a positive control to salubrinal 25 . We first evaluated the effects of tunicamycin, salubrinal, and CA to MC3T3-E1 cells. The result indicated that the non-toxic dose of salubrinal was 0.2 to 5 μM, while that for CA was 0.05 to 0.2 nM. We thereafter employed those concentrations, and evaluated their effects in the presence of tunicamycin. Tunicamycin inhibited viability of MC3T3-E1cells in a dose-dependent manner at 0.05, 0.1, 0.25, 0.5, 1, and 2.5 μg/ml. Interestingly, administration of 5 μM salubrinal or 0.2 nM CA partially restored cell viability in the presence of 100 ng/ml tunicamycin (both P < 0.001; Fig. 3g). However, salubrinal or CA was unable to rescue MC3T3-E1 cells at higher concentrations of tunicamycin (above 250 ng/ml, all P > 0.05; Supplemental  Fig. 2a).
To investigate the role of eIF2α in the effect of salubrinal on osteogenesis, the cell viability and the expression of osteoblast markers were determined in MC3T3-E1 cells using eIF2α RNA interference. Transfection with eIF2α siRNA blocked the effect of salubrinal on cells viability under ER stress (P < 0.01; Fig. 3h). Furthermore, the protein levels of p-eIF2α, ATF4, ALP and RUNX2 in MC3T3-E1 cells were examined at 48 h after administration of 100 ng/ml tunicamycin with 5 μM salubrinal (Fig. 3i). Tunicamycin significantly inhibited the levels of ALP and RUNX2, while salubrinal increased the levels of p-eIF2α, ATF4, ALP, and RUNX2 (all P < 0.01). Compared to scrambled siRNA, eIF2α siRNA blocked the effect of salubrinal on the expression of p-eIF2α, ATF4, ALP, and RUNX2 under ER stress (all P < 0.05; Fig. 3j-m).

Salubrinal elevates vessel density in ONFH.
To investigate whether salubrinal improves blood perfusion in the ischemic femoral head, we determined the vessels in the femoral head using ink perfusion Percentages of the pit area to total area in each field view from three groups. The asterisks (*, **, and ***) represent P < 0.05, P < 0.01, and P < 0.001, respectively (n = 12).
to 5 μM salubrinal to HUVECs and evaluated salubrinal's effect in the presence of an ER stress inducer, tunicamycin. Tunicamycin inhibited cell viability of HUVECs in a dose-dependent manner (from 250 ng/ml to 5 μg/ ml, all P < 0.05; Fig. 4e). Administration of 5 μM salubrinal restored cell viability in the presence of 500 ng/ml tunicamycin (P < 0.05; Fig. 4f). However, 5 μM salubrinal was unable to rescue the viability of HUVECs that were treated with higher concentrations of tunicamycin (more than 1 μg/ml, all P > 0.05, Supplemental Fig. 3a-c). The results indicated that 500 ng/ml tunicamycin induces significant responses in HUVECs. The dose of 500 ng/ ml tunicamycin was thus used to induce ER stress in HUVECs. Furthermore, compared to scrambled siRNA, eIF2α siRNA transfection blocked the effect of salubrinal for restoring cell viability of HUVECs in the presence of 500 ng/ml tunicamycin (all P < 0.05; Fig. 4g).
To investigate whether salubrinal affected migration and angiogenesis of HUVECs in the presence of tunicamycin, we performed assays for migration, wound healing and tube formation in vitro. In the migration (Fig. 5a) and wound healing ( Fig. 5b) assays, tunicamycin inhibited the migration and wound recovery of HUVECs (both P < 0.05). Salubrinal rescued migration and wound recovery of HUVECs in the presence of tunicamycin (both P < 0.05). However, compared to scrambled siRNA, eIF2α siRNA transfection blocked the effects of salubrinal on migration and wound recovery of HUVECs under ER stress (all P < 0.05; Fig. 5d,e). In the tube formation assay  Salubrinal increases p-eIF2α, ATF4, GRP78 and VEGF, and reduces NFATc1 in ONFH in vivo. The images of immunofluorescence showed that p-eIF2α positive cells (green) were mainly located in the bone marrow cavity (Fig. 6a). Compared to the sham control group, the ON group increased the expression level of p-eIF2α (P < 0.01). The salubrinal-treated group exhibited a further increased level of p-eIF2α (P < 0.01; Fig. 6b). Compared to the sham control group, the ON group increased the levels of p-eIF2α, ATF4, and VEGF, and decreased the levels of GRP78. The salubrinal-treated group increased the level of p-eIF2α, ATF4, GRP78 and VEGF, but it presented a lower level of NFATc1 than that the ON group (all P < 0.05). The level of CHOP was not significantly different in three groups (all P > 0.05) (Fig. 6c-i). Correlation exists among p-eIF2α, angiogenesis, vessel remodeling, and bone healing. To evaluate the correlation among parameters in salubrinal's improving of ONFH symptoms, we selected the p-eIF2α level for representing salubrinal's effect, VEGF level for angiogenesis, vessel number for vessel remodeling, and osteoblast number and osteoclast surface for bone healing. The result indicated that the level of p-eIF2α was positively correlated with the level of VEGF (r = 0.902, P < 0.001) as well as the osteoblast number (r = 0.672, P < 0.01). Furthermore, the vessel number was positively correlated with the osteoblast number (r = 0.613, P < 0.05) and negatively correlated with the osteoclast surface (r = −0.829, P < 0.001).

Discussion
Interruption of the blood supply to the femoral head is a prime cause of ONFH induction 31 . In the rat model of ONFH, ischemic osteonecrosis was surgically induced and the effect of salubrinal was investigated. Compared to the sham control group, the ON group decreased BV/TV, Tb.N, Tb.Th, B.Ar/T.Ar and femoral BMD/BMC, with an increase in empty lacunae and Tb.Sp. As reported in the previous ONFH studies 12,16 , the results herein are consistent with significant bone loss in the femoral head. ONFH also presented the decrease in blood perfusion and the level of VEGF 16,32,33 . It presented the fewer vessel number and smaller vessel area than those in the sham control group. In orthopaedic diseases such as osteoporosis, osteoarthritis, and osteonecrosis, osteoclastogenesis is abnormally activated [34][35][36][37][38] . The result in this study revealed that osteoclastogenesis was also activated in the femoral head of ONFH. The number of TRAP-positive cells was significantly increased in the bone surface of ONFH. Furthermore, osteonecrosis activated the differentiation, migration, adhesion, and activity of osteoclasts, although osteoblast differentiation in the sham control group was not markedly different from that in the ON group.
ONFH primarily results from ischemia/hypoxia to the femoral head. One of the cellular manifestations of ischemia/hypoxia is the ER stress. The ER stress is known to lead to three major signaling pathways, such as the PERK-ATF4 axis, activating transcription factor 6 (ATF6) axis, and inositol-requiring enzyme-1a/X-box binding protein 1 (IRE1a/XBP1) axis 1,26 . A unique feature in disease conditions such as ONFH is that the ER stress may act as a two-edged sword. While a moderate level may assist the reestablishment of cellular homeostasis, the prolonged and unmitigated stress may lead to apoptosis 1,26 . In the present model of ONFH, the ischemic femoral head was unable to acquire an adequate supply of blood, oxygen, or nutrients 20 , and the excessive ER stress was induced. The result in this study showed that osteonecrosis elevated the level of p-eIF2α and ATF4. The elevated level of ATF4, which is induced by the ER stress 1,26 , is reported to promote osteoclastogenesis 39,40 . In the ONFH model, we also observed stimulation in osteoclastogenesis in the ischemic femoral head. Correlation analysis indicated that blood perfusion was negatively associated with osteoclastogenesis. Collectively, ischemia in ONFH induced osteoclastogenesis and interrupted bone healing by inducing excessive ER stress.
Salubrinal elevates the level of p-eIF2α and activates eIF2α mediated signaling 25 , and it stimulates bone metabolism in orthopaedic diseases [27][28][29] . We recently presented several lines of evidence that osteoclastogenesis in osteoporosis was associated with the ER stress, and salubrinal suppressed unloading-induced bone loss 30 . The result in this study showed that salubrinal prevented bone loss in ONFH by suppressing osteoclast activity and promoting osteoblast development. Salubrinal is reported to suppress osteoclastogenesis by reducing the level of NFATc1, and promote osteoblastogenesis by increasing the level of ATF4. The results with NFATc1 and ATF4 in this study are consistent with those in the previous studies 27, 41 .
ATF4 has a dual function in the ER stress pathway. It may serve as a pro-survival factor by promoting transcription of ER chaperones such as GRP78 and reducing the accumulation of misfolded proteins. Alternatively, it may lead to apoptosis by inducing CHOP, which is a pro-apoptotic factor 26 . To evaluate the degree and effect of ER stress in the present ONFH model, we determined the levels of GRP78 and CHOP. Although the level of CHOP was not detectably different, the level of GRP78 in the ONFH model was significantly decreased. The reduction of GRP78 suggests that the ER stress was maintained at the unmitigated level, and the chaperones became unavailable under the prolonged ER stress. Salubrinal attenuated prolonged ER stress through the eIF2α -ATF4 -GRP78 regulatory axis.
Regarding the effects of salubrinal on angiogenesis, we determined the level of VEGF and its action on endothelial cells. The result showed that p-eIF2α increased the expression of ATF4 1, 26 , and ATF4 promoted the secretion of VEGF, a stimulator of angiogenesis 26,[42][43][44][45][46] . Salubrinal protected the proliferation, migration, and angiogenesis of HUVECs against the excessive ER stress, which was induced by tunicamycin. Salubrinal also increased the p-eIF2α level in HUVECs and increased those of p-eIF2α, ATF4, and VEGF in vivo.
To further evaluate the effect of salubrinal on the ER stress, we employed Calyculin A (CA) as a specific inhibitor of PP1 25 . Tunicamycin inhibited the cell viability of MC3T3-E1 cells, while salubrinal and CA suppressed tunicamycin's inhibition. To investigate the in-depth mechanism of salubrinal's action on the ER stress, RNA interference with eIF2α siRNA was conducted. Salubrinal rescued the cell viability of MC3T3-E1 and HUVECs under the ER stress. Transfection of eIF2α siRNA blocked the effect of salubrinal on cells viability under ER stress. Tunicamycin significantly inhibited the levels of ALP and RUNX2, while salubrinal increased those of p-eIF2α, ATF4, ALP, and RUNX2. Treatment with eIF2α siRNA, however, blocked the effects of salubrinal on the levels of p-eIF2α, ATF4, ALP, and RUNX2, as well as migration and angiogenesis of HUVECs in the presence of tunicamycin. These results indicated that salubrinal protects osteoblasts and endothelial cells against the excessive ER stress by modulating the eIF2α signal pathway.
We evaluated toxicity of salubrinal and CA in MC3T3-E1 cells. The result indicated that the non-toxic dose of salubrinal and CA were no more than 5 μM and 0.2 nM, respectively. Collectively, salubrinal not only protected osteogenesis against the ER stress but also had lower toxicity than CA. Furthermore, salubrinal promoted osteoblastogenesis, suppressed osteoclastogenesis, and stimulated angiogenesis through suppressing ER stress signaling by elevating the level of p-eIF2α in ONFH (Fig. 7). In conclusion, ischemia is an inducer of the ER stress in the surgery-induced rat ONFH model, and an excessive ER stress induces osteoclastogenesis and inhibits angiogenesis. Alleviating the ER stress is a potential option to improve symptoms associated with ONFH. The present study suggests that the ER stress is an important pathological outcome in the surgery-induced ONFH model, in which salubrinal improves the ONFH symptoms by alleviating the ER stress. Experimental design. Sixty rats were randomly divided into 4 groups: sham control group (Sham, n = 18), osteonecrosis group (ON, n = 18), salubrinal-treated osteonecrosis group (ON + S, n = 18) and salubrinal-treated sham control group (Sham + S, n = 6). Ischemic osteonecrosis of the bilateral femoral heads were induced according to the procedure previously described 12,47 . Briefly, after anesthesia, a longitudinal incision was made on the skin over the greater trochanter. The gluteus maximus muscle and the gluteus medius muscle were separated from the bone. The joint capsule was transected and the femoral head was dislocated. ONFH was induced by transecting the ligamentum teres and tightly placing a ligature (#3-0 Vicryl, Ethicon) around the femoral neck. After the femoral head was relocated, muscles and skin were sutured with #3-0 and #4-0 stitches, respectively. For sham control and salubrinal-treated sham control groups, rats received the same procedure except for ligating the femoral neck and transecting the joint capsule. In addition, analgesia and antibiotic prophylaxis were applied for the first three postoperative days. After surgery, a dosage of 0.05 mg/kg salubrinal (resolved in propylene glycol) was administered subcutaneously every day for 4 wks to the salubrinal treatment group and salubrinal-treated sham control group 28 . The sham control and osteonecrosis groups were given the vehicle. All animals were euthanised in 4 wks after operation. Bone marrow-derived cells were collected from tibias, and femurs were used for imaging and histology.
Histological assays. Femoral heads were fixed with 10% neutral buffered formalin for 2 days, and decalcified with 10% ethylenediaminetetraacetic acid (EDTA, pH 7.4) for 40 days. Samples were embedded in paraffin and cut into 5-μm thickness coronal slices as previously described 47 . The ratio of trabecular bone area to tissue area (B.Ar/T.Ar), and the percentage of empty lacunae in bone tissue were analyzed using hematoxylin-eosin (H&E) staining 21,49 . Tartrate resistant acid phosphatase (TRAP) staining were used to detect the percentage of osteoclasts 49,50 . MacNeal's staining were used to determine the number of osteoblasts 49,51,52 . For all histological assays, three sections were assessed per femoral head and five 100-400× fields were randomly selected per section. All graphics were measured and analyzed by Cellsense Standard software (Olympus).

Isolation of bone marrow-derived cells and osteoclast development assay.
Bone marrow-derived cells were collected as described previously 47,53 . After euthanasia, we flushed the bone marrow of tibias with DMEM (containing 2% FBS). The cells were separated with Ficoll low-density gradient centrifugation. We seeded cells onto 96-well plates at a density of 1 × 10 5 cells/well, and cultured in MEM-α supplemented with 30 ng/ml M-CSF and 20 ng/ml RANKL for 3 days. On day 4, the culture medium was replaced by MEM-α supplemented with 30 ng/ml M-CSF and 60 ng/ml RANKL, and cells were grown for additional 3 days. Osteoclasts were identified using a TRAP staining kit (Sigma-Aldrich). TRAP-positive multinuclear cells (more than 3 nuclei) were identified as matured osteoclasts 54 . In all cell culture experiments, five random fields per well were selected and photographed for analysis.
Colony-forming unit-macrophage/mononuclear (CFU-M) and colony-forming unit-granulocyte-macrophage (CFU-GM) assays were performed. We seeded bone marrow-derived cells onto 6-well plates at a density of 2.5 × 10 4 cells/well. The culture medium was composed of methylcellulose with 30 ng/ml M-CSF and 20 ng/ml RANKL. Cells were cultured at 37 °C in a 5% CO 2 incubator for 7 days. The colony numbers in CFU-M/CFU-GM were converted to the numbers per tibia 53,55 .
Migration of pre-osteoclasts was performed with a transwell device as described previously 53 . To achieve pre-osteoclasts, we seeded bone marrow-derived cells (2 × 10 6 cells/ml in 6-well plates) in MEM-α supplemented 10% FBS, 30 ng/ml M-CSF and 20 ng/ml RANKL for 4 days. For migration assay, the pre-osteoclasts (1 × 10 5 cells/ well) were resuspended onto the upper chamber of transwells and allowed to migrate to the bottom chamber through an 8-μm polycarbonate filter coated with vitronectin (Takara Bio Inc., Otsu, Shigma, Japan). The medium in the upper chamber was replaced by MEM-α without serum, and MEM-α (consisting of 1% bovine serum albumin and 30 ng/ml M-CSF) were loaded onto the lower chamber. After 6 h incubation, the number of migrated pre-osteoclasts was counted. For adhesion assay, the pre-osteoclasts (1 × 10 5 cells/well) were seeded onto 96-well plates which were coated with vitronectin in MEM-α supplemented with 30 ng/ml M-CSF. The number of adherent pre-osteoclasts was counted after 30-min incubation 56 . For pit formation assay, a UV-sterilized bovine cortical bone slices (120 μm thick) were used in bottom of the 24-well plate 54 . We seeded bone marrow-derived cells onto the plate (1 × 10 5 cells/well), and cultured cells with MEM-α supplemented 10% FBS, 30 ng/ml M-CSF and 20 ng/ml RANKL for two days. On the day 3, the concentration of RANKL was increased to 60 ng/ml, and the medium was changed every 2 days. On day 10, the percentages of the pit area to total area in each field view were calculated. In all above experiments, cells were stained with crystal violet, and five random fields per well were selected and photographed for analysis.

Osteoblast differentiation and colony-forming unit-fibroblast (CFU-F) assays.
In an osteoblast differentiation assay, bone marrow-derived cells were cultured onto 6-well plates (2 × 10 6 cells/well) with osteogenic differentiation medium, and the medium was changed every 2 days. On day 14, cells were stained using an alkaline phosphatase (ALP) staining kit (Sigma) and the percentages of ALP-positive cells were determined 48 .
In a CFU-F assay, bone marrow-derived cells were seeded onto 6-well plates (2 × 10 6 cells/well) with complete MesenCult medium. The medium was changed every 2 days for 14 days. Then, fibroblasts were stained using a HEMA-3 quick staining kit (Fisher Scientific, Waltham, MA, USA) and the colonies with more than 50 cells were counted 57, 58 . Immunofluorescence. Immunofluorescence analysis of the femoral head sections was performed as described previously 50,59 . Bone sections were permeabilized in 0.3% Triton X-100 for 10 min, and blocked in 10% goat serum at room temperature for 30 min. Furthermore, the sections were incubated with special primary antibodies (p-eIF2α 1:100) overnight at 4 °C. Subsequently, the samples were incubated with secondary antibodies conjugated with fluorescence at room temperature for 1.5 h (avoiding light). The expression of p-eIF2α was determined, in which the nuclei were counterstained with DAPI.
Blood perfusion assay. To analyze blood perfusion of the femoral head with ONFH, an ink perfusion assay was performed using Chinese ink as previously described 33,47 . The heart of rat was exposed by surgery after anesthesia. A needle was inserted into the left ventricle for ink infusion, and the right atrium was opened. The circulation were flushed by a heparin-saline solution (25,000 units in 250 ml of 0.9% sodium chloride) until clear liquid flowed from the right atrium. With that, 5% gelatin/ink solution (the ratio between Chinese ink and water was 1:1) was injected into the circulation until the skin of animals became uniformly black. The femoral heads were harvested, and fixed with in 10% neutral buffered formalin for 2 days, and were decalcified in 10% ethylenediaminetetraacetic acid (EDTA, pH 7.4) for 40 days. Samples were embedded in paraffin and cut into 25-μm thickness slices. The quantification of vessel numbers was the vessel numbers in each 200× field view, and vessel area percentages was the ratio of vessel area to total area in each 200× field view.
Transfection. Transient knockdown of endogenous eIF2α was achieved by siRNA which targeting eIF2α. Transfection was performed with Lipofectamine 2000 Reagent (Invitrogen) following the manufacturer's protocol. Briefly, cells were seeded in plates the day before transfection to ensure a suitable cell confluent on the day of transfection. Scrambled siRNA 5′-ACGUGACACGUUCGGAGATT-3′ or eIF2α siRNA 5′-AAGCUACUUCAAUAUCU CTT-3′ were used for transfection in antibiotic free Opti-MEM medium (Invitrogen). Two days after transfection, the transfected cells were then ready for the following experiments.
Cell viability assay. MTT assay was used to evaluate the cell viability as previously described 60 . HUVECs were seeded in 96-well plates at a density of 1 × 10 4 cells/well with EGM2. Four hours later, cells were transfected with 20 nM eIF2α siRNA or scrambled siRNA, and treated with various treatment factors or vehicle, respectively. After 48 h, the absorbance at 570 nm was detected using a μQuant universal microplate spectrophotometer (Bio-tek, Winooski, USA).
Migration and wound-healing assays of HUVECs. Migration of HUVECs was evaluated with trans-well equipment as described previously 50 . HUVECs were transfected with 20 nM eIF2α siRNA or scrambled siRNA. 48 h later, cells (2 × 10 4 cells/well) were suspended in serum-free medium in the upper chambers and incubated them with various treatment factors or vehicle. EGM2 (consisting of 20% FBS) were loaded onto the lower chamber. The number of cells that migrated to the lower surface in 12 h was counted using crystal violet staining. In a wound-healing assay, HUVECs were transfected with 20 nM eIF2α siRNA or scrambled siRNA in 6-well plates for 48 h until grown to 100% confluence. Using a 200-μl pipette tip, a scratch was made on a cell monolayer. The width of scratch was measured on 0 h and 20 h, respectively, and the percentages of scratch recovery were calculated 60 .
Tube formation assay of HUVECs. HUVECs were transfected with 20 nM eIF2α siRNA or scrambled siRNA for 48 h. We coated 24-well plates with Matrigel and incubated them to polymerize at 37 °C for 45 min. HUVECs (2 × 10 5 cells/well) were then seeded on the polymerized Matrigel. Cells were incubated with EGM2-containing treatment agents or vehicle for 12 h to stimulate tube formation 50 . Images of tubes were taken, and the cumulative tube lengths were calculated.
Osteogenesis assay in vitro. Cell viability (MTT assay) and the expression levels of osteoblast markers were determined under the tunicamycin-induced ER stress. MC3T3-E1 cells were seeded in 96-well plates at a density of 1 × 10 4 cells/well in DMEM with 10% FBS. Four hours later, cells were transfected and were exposed to the treatment agents or vehicle. After 48 h, the absorbance at 570 nm was detected in the MTT assay. In addition, 48 h after transfection and administration, the expression levels of p-eIF2α, ATF4, ALP, and RUNX2 inMC3T3-E1 cells were determined by Western blot.