Author contributions: C.P.H.: conception and design, collection of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; V.N., K.G.P., J.G., J.S., Z.Z., and B.D.: collection and/or assembly of data; R.P.: data analysis and interpretation; M.M.: conception and design, manuscript writing, and final approval of manuscript; V.J.D.: conception and design, manuscript writing, data analysis and interpretation, and final approval of manuscript.
Mesenchymal stem cells (MSCs) transplanted into injured myocardium promote repair through paracrine mechanisms. We have previously shown that MSCs over-expressing AKT1 (Akt-MSCs) exhibit enhanced properties for cardiac repair. In this study, we investigated the relevance of Abi3bp toward MSC biology. Abi3bp formed extracellular deposits with expression controlled by Akt1 and ubiquitin-mediated degradation. Abi3bp knockdown/knockout stabilized focal adhesions and promoted stress-fiber formation. Furthermore, MSCs from Abi3bp knockout mice displayed severe deficiencies in osteogenic and adipogenic differentiation. Knockout or stable knockdown of Abi3bp increased MSC and Akt-MSC proliferation, promoting S-phase entry via cyclin-d1, ERK1/2, and Src. Upon Abi3bp binding to integrin-β1 Src associated with paxillin which inhibited proliferation. In vivo, Abi3bp knockout increased MSC number and proliferation in bone marrow, lung, and liver. In summary, we have identified a novel extracellular matrix protein necessary for the switch from proliferation to differentiation in MSCs. STEM Cells2013;31:1669–1682
Stem cell therapy for tissue repair and regeneration has great therapeutic potential [1–4]. Adult stem cells from the bone marrow are one of the most widely used cells for this purpose. Bone marrow contains a heterogeneous population of cells, including mesenchymal stem cells (MSCs), or marrow stromal stem cells. MSCs are nonhematopoietic stem cells that have the potential to develop into differentiated cell types such as osteocytes, chondrocytes, and adipocytes as well as nerve cells . One of the major issues regarding the therapeutic potential of MSCs is their poor survivability in vivo. To this end, our laboratory over-expressed the Akt1 oncogene in MSCs. Injection of Akt1-modified MSCs into the heart after myocardial infarction increased MSC survival and led to major reduction in infarct size, associated with significant improvements in cardiac function . Further research revealed that these therapeutic effects were mediated in part by the release of paracrine factors by these cells. Validation of this paracrine mechanism came from in vivo experiments where concentrated media from Akt1-modified MSCs reduced infarct size and cardiac cell apoptosis in a rat coronary occlusion model . A functional genomic strategy identified several novel proteins that were regulated by Akt1 and potentially could account for the enhanced therapeutic properties of the Akt-MSCs. One of these novel proteins was Abi3bp .
Abi3bp is a relatively novel protein for which the biological function is unclear. A partial fragment of Abi3bp, also known as TARSH and eratin, was identified in a yeast two-hybrid screen with the c-Abl binding protein Abi3 as bait . However, in vivo binding activity between full-length Abi3bp and Abi3 awaits confirmation. Expression of the protein has been observed in the olfactory system; conditioned media containing Abi3bp reduced mitral cell dendritic complexity, which is an important step in the formation of functional circuits . Computational screening followed by in vitro assays identified that a partial fragment of Abi3bp, containing one of the two Fibronectin type-III domains found in the full-length protein, promoted cell attachment and was capable of assembling into an extracellular matrix . In addition, Abi3bp is expressed in normal thyroid and in benign follicular thyroid adenoma but is absent in follicular thyroid carcinoma . Similarly, loss of Abi3bp RNA expression has been reported in both lung cancer cell-lines and lung tumor specimens . Abi3bp has been reported to have both a positive and negative role in senescence [13, 14]. In this study, we show that Abi3bp plays an important role in MSC biology by acting as a “check and balance” in many MSC processes including proliferation, differentiation, adhesion, morphology, and transformation.
Materials and Methods
Full methods are given in supporting information
Abi3bp Knockout Mice
Abi3bp−/+ mice, harboring a neoR replacement of the first exon, were purchased from Taconic. All experiments were performed with littermates and in accordance with institutional guidelines (DLAR and IACUC).
MSC Isolation and Retroviral Transfection
Bone marrow cells from eight 8-week-old wild-type male C57BL/6J mice (Jackson Laboratory, Maine, http://www.jax.org/index.html) or from three 8-week-old male Abi3bp wild-type and knockout mice were collected in MEMα (Invitrogen, NY, http://www.lifetechnologies.com/us/en/home.html) supplemented with 20% (vol/vol) heat-inactivated fetal bovine serum (FBS) and 1× penicillin-streptomycin (Invitrogen, NY, http://www.invitrogen.com) by flushing the marrow cavity with a 27-guage needle. Cells were filtered through a 40 μm cell strainer (Falcon). Mononuclear cells then were isolated from aspirates by Ficoll-Paque (GE Healthcare, Little Chalfont, UK. http://www3.gehealthcare.com/en) gradient centrifugation and cultured. Cells were plated at a density of 20 × 106/9.5 cm2. MSCs were separated from hematopoietic cells based on their preferential attachment to cell culture surfaces. Nonadherent cells were removed 24 hours after plating and the adherent layer washed twice in culture media (MEMα supplemented with 20% [vol/vol] heat-inactivated FBS and 1× penicillin-streptomycin). Media were changed every 2 days and once MSCs were 70% confluent cells were removed by 0.05% (vol/vol) trypsin and replated at 5,000 cells per square centimeter. MSCs were cultured in this fashion until used in an experiment. MSC isolations were performed from Abi3bp wild-type and knockout mice on three separate occasions. In the first isolation, MSCs were used at passage 10, MSCs from the subsequent two isolations were used at passage 5. MSCs at this passage were tested for various MSC, hematopoietic, and endothelial markers as described in Results. MSC senescence was evaluated by β-galactosidase staining (Cell Signaling, MA, http://www.cellsignal.com/index.jsp). Lung and liver MSCs were isolated by fluorescence-activated cell sorting and cultured according to the bone marrow protocol. Differentiation potential of the lung and liver MSCs were evaluated at passage 5.
For smooth muscle differentiation, MSCs were seeded at 5,000 cells per square centimeter in normal growth medium. The next day media were changed for Dulbecco's modified Eagle's medium (DMEM) + 10% heat-inactivated FBS supplemented with 10 ng/ml TGF-β and 50 ng/ml PDGF. Media were changed every 2 days for a total of 6 days, whereupon the cells were fixed and analyzed for smooth muscle actin by immunofluorescence (α-smooth muscle actin, Sigma, MO, http://www.sigmaaldrich.com/united-states.html). For osteogenic differentiation, MSCs were seeded at 10,000 cells per square centimeter in normal growth medium. The next day media were changed for DMEM + 10% heat-inactivated FBS supplemented with 1 μM dexamethasone and 25 μg/ml ascorbic acid-2-phosphate. Media were changed every 2 days for a total of 4 weeks, cells were fixed, and analyzed for osteogenesis using 2% Alizarin-Red (pH 4.2). For chondrogenic differentiation, MSCs were resuspended at a concentration of 1.6 × 107 cells per milliliter. Five microliter aliquots were seeded into 12-well plates and left in a humidified atmosphere for two hours. Differentiation media with or without chondrogenic supplement (R&D Systems, MN, http://www.rndsystems.com) were added. Media were changed every 4 days for a total of 2 weeks, cells were fixed, and analyzed for chondrogenesis using 1% Alcian-Blue in 0.1 M HCl.
siRNA pools and the negative control were purchased from Dharmacon. Transfection was carried out with Dharmafect-I (Dharmacon, PA, http://www.thermoscientificbio.com/dharmacon) according to manufacturer's guidelines. Full details are supplied in supporting information Methods.
Total RNA was extracted using a RNeasy Plus Micro Kit according to the manufacturer's guidelines (Qiagen, Germany, http://www.qiagen.com). Total RNA (500 ng) was converted to cDNA using a high capacity cDNA reverse transcription kit (Applied Biosystems, NY, http://www.invitrogen.com). qPCR was performed by incubating cDNA with a fluorescein amidite (FAM) conjugated gene-specific primer and TaqMan Gene Expression Master Mix (Applied Biosystems).
The staining procedure is described in supporting information Methods.
Cell counting was performed with Promega (WI, http://www.promega.com) CellTiter 96 aqueous nonradioactive cell proliferation assay (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS)) according to the manufacturer's guidelines. Incubations were all 90 minutes at 37°C. MSC lines were all seeded at 5,000 cells per square centimeter and a 48-well format was used.
Fluorescence-Activated Cell Sorting
The antibodies and methods used are described in supporting information Methods.
Coimmunoprecipitation was performed as described previously .
Figures were prepared using CorelDraw. Microscopy images were exported from Axiovision Rel4.8 software. For bright-field images, where necessary color casts were removed using Photoshop using a pure white area of the photo.
Statistical analysis was performed with GraphPad or R. Experiments containing two conditions a t test was performed. Analysis of Variance (ANOVA) was used for experiments with three or more conditions followed by Bonferroni post hoc tests for comparisons between individual groups.
Abi3bp Is an Extracellular Matrix Protein Controlled by Akt and Ubiquitin
Previous work in our laboratory identified that MSCs modified to over-express the oncogene Akt1 show enhanced survival and release of cytoprotective and angiogenic factors after transplantation into the injured myocardium . Microarray analysis found a number of genes of relatively unknown function that were positively induced by Akt1 . One of these genes was Abi3bp.
In order to validate the microarray data we first performed qPCR using MSC, MSC-green fluorescent protein (GFP), and MSC-GFP-Akt1 cells cultured in vitro. Akt1 over-expression in MSCs increased Abi3bp RNA levels by ∼20-fold when compared with MSC and MSC-GFP cells (Fig. 1A). This was accompanied by an increase in Abi3bp protein levels, albeit less dramatic (Fig. 1A). To further test whether Akt1 regulated Abi3bp expression, we used a pool of four siRNAs targeted against Akt1 to deplete the protein from MSC-GFP-Akt1 cells. When Akt1 was depleted Abi3bp RNA expression was reduced by ∼10-fold, validating that Akt1 regulates Abi3bp expression (Fig. 1B). Finally, insulin stimulation following a period of serum withdrawal activated endogenous Akt in MSCs and increased Abi3bp expression when compared with control untreated cells (Fig. 1C).
Akt1 had a substantially greater effect on Abi3bp RNA expression than on the protein, suggesting that a post-translational mechanism regulates Abi3bp levels. To test this hypothesis, we generated a HEK293 cell-line that over-expressed a myc-tagged Abi3bp construct. These cells were treated with either vehicle or the proteasome inhibitor clasto-lactacystin β-lactone for 24 hours after which the cells were exposed to cycloheximide, an inhibitor of protein synthesis. In the presence of cycloheximide alone the half-life of the myc-tagged Abi3bp was ∼1 hour (Fig. 1D). Clasto-lactacystin β-lactone had a dramatic effect, increasing expression of Abi3bp by approximately fivefold. When cells were cultured in both cycloheximide and clasto-lactacystin β-lactone, levels of the myc-tagged Abi3bp did not significantly decrease during the length of the experiment, indicating a half-life in excess of 8 hours (Fig. 1D). Taken together this data indicate that Abi3bp undergoes rapid degradation by a ubiquitin-mediated process.
A previous study has shown that a partial fragment of Abi3bp formed pericellular deposits typical of extracellular matrix (ECM) deposition . We were interested to determine whether the full-length protein would behave in a similar fashion. Abi3bp was secreted by HEK293 cells over-expressing the protein (Fig. 1E). Surfaces exposed to HEK293 cells expressing myc-tagged Abi3bp showed strong ECM staining with a myc-antibody. This was not observed on surfaces exposed to HEK293 cells expressing the control vector (Fig. 1E).
Abi3bp Controls MSC Biology, Differentiation, and Motility
Focal complexes (FC) are small dot-like adhesions that are the first points of contact between the cell and the ECM. FC are transient, and if not stabilized into focal adhesions (FA), are quickly broken down [16–18]. FA turnover is very important for cell spreading, migration, and proliferation. We investigated whether Abi3bp affects FA formation using three cell systems: (a) MSC or (b) MSC-GFP-Akt1 cells expressing either the scrambled control shRNA or the Abi3bp shRNA-83 or (c) MSCs from wild-type and Abi3bp knockout mice. Abi3bp knockdown was achieved by stably expressing one of two shRNA constructs, sh91 or sh83. Stable expression of a scrambled shRNA was used as a control (supporting information Fig. S1A). The wild-type and Abi3bp knockout MSCs used in this study were primary cells as they displayed senescence (supporting information Fig. S1B). Analysis of cell surface markers showed that the MSC isolations were free of hematopoietic and endothelial contamination (no expression of CD45, CD11b, CD34, and CD31—supporting information Fig. S2). Cells were stained with several FC/FA markers, these being phospho-Y118 paxillin and paxillin. Cells were also incubated with phalloidin to stain F-actin stress fibers. Abi3bp knockout MSCs were found to have a significantly lower phospho-paxillin/paxillin ratio for stained foci area, whereas phalloidin intensity was increased when compared with control MSCs (Fig. 1F, top). Similar findings were observed in the MSC-GFP-Akt1 system (Fig. 1F, bottom) and in this cell model phospho-paxillin to vinculin ratios were nearly identical to those for phospho-paxillin:paxillin (supporting information Fig. S1C). Other parameters are shown in supporting information Figure S1D.
MSCs have the potential to differentiate into several cell types. We investigated whether Abi3bp was important in this process. All three MSCs isolations from wild-type or Abi3bp knockout mice were induced with dexamethasone and ascorbic acid for osteogenesis. All the wild-type MSC isolations were capable of undergoing osteogenic differentiation as evidenced by robust alizarin-red staining. Normalization to cell number showed a 20-fold increase in alizarin-red staining in the treated wild-type MSCs (Fig. 2A). However, all three Abi3bp knockout MSC isolations completely failed to show any significant alizarin-red deposition both visually and after normalization to cell number (Fig. 2A). We subsequently measured Runx2 expression, an early marker of commitment to the osteogenic lineage, in our wild-type and Abi3bp knockout MSCs. Wild-type MSCs cultured in differentiation media had a significant twofold increase in Runx2 expression when compared with the vehicle control (Fig. 2A, right hand panel). However, although Runx2 expression in the treated Abi3bp knockout MSCs was higher when compared with vehicle control cells the increase was not significant (Fig. 2A, right hand panel). Furthermore, Runx2 expression in the Abi3bp knockout MSCs was significantly lower than wild-type cells. This data suggests that Abi3bp knockout affects early commitment to osteogenic differentiation. The wild-type MSC isolations also showed robust chondrogenic differentiation; the cells formed pellets with strong deposition of proteoglycans as shown by Alcian Blue staining and normalization to cell number (Fig. 2B, supporting information Fig. S1E). In contrast, Abi3bp knockout MSCs did not form chondrogenic pellets instead forming a cellular net. Normalization to cell number indicated that in contrast to wild-type MSCs proteoglycan deposition in Abi3bp knockout MSCs was highly variable (Fig. 2B). Both wild-type and Abi3bp knockout MSCs were capable of differentiating into smooth muscle; however, the general appearance of the cells was substantively different between the two cell types (Fig. 2C). Finally, we measured adipogenic differentiation in our MSC isolations. All three wild-type MSC isolations generated adipocytes (Fig. 2D). The first isolation had the smallest fold increase, which may be due to their higher passage number. In contrast, Abi3bp knockout MSCs were severely limited in their ability to differentiate into adipocytes (Fig. 2D). Considering that Abi3bp controls FA formation we investigated whether this protein also affected MSC movement. Using a scratch assay MSCs lacking Abi3bp showed higher motility when compared with control cells (Fig. 2E, supporting information Fig. S1F). One might expect MSCs lacking Abi3bp to have a slower motility because of the stabilization of the FAs in these cells. However, these cells were found to have a higher proliferation rate (see below) and this result may be due to an increase in cell number rather than representing more motile cells.
Abi3bp Controls MSC Proliferation
Considering that cell shape is an important determinant for proliferation, we hypothesized that Abi3bp would control this process in MSCs.
Prior to testing this hypothesis, we first characterized MSC, MSC-GFP, and MSC-GFP-Akt1 growth curves using a MTS assay to measure cell number. Both MSC and MSC-GFP cells had virtually identical growth curves (supporting information Fig. S3A and Table S1). Over-expression of the Akt1 oncogene increased MSC growth rate approximately fourfold (supporting information Table S1).
Comparing the growth rates of the wild-type and Abi3bp knockout MSCs indicated that all three Abi3bp knockout MSCs isolations possessed a higher growth rate when compared with their wild-type MSC counterparts (Fig. 3A). Averaging the growth rates of the respective isolations showed that wild-type MSCs have a significantly slower growth rate than Abi3bp knockout MSCs (Fig. 3A). Additional experiments with shRNA showed that Abi3bp knockdown increased the growth rate of MSCs by approximately twofold (supporting information Table S1, Fig. 3B) and by approximately threefold for MSC-GFP-Akt1 cells (supporting information Table S1, Fig. 3B, supporting information Fig. S3B).
Various assays were used to determine whether the effects on cell number were due to changes in cell cycle and proliferation. Flow cytometry was used to measure the percentage of wild-type and Abi3bp knockout MSCs undergoing active proliferation in culture. All three Abi3bp knockout MSC isolations displayed approximately twofold higher incorporation of BrdU, a thymidine analog, when compared with the wild-type MSC controls (Fig. 3C) indicating that Abi3bp knockout increased proliferation. Validation experiments were performed with shRNA knockdown. In MSCs incubated with BrdU Abi3bp knockdown increased the percentage of cells in S-phase (Fig. 3D, supporting information Fig. S3C). Similarly, knockdown of Abi3bp by either shRNA construct in MSC-GFP-Akt1 cells caused a significant increase in the incorporation of BrdU as determined by ELISA (supporting information Fig. S3D). Flow cytometry was used to further investigate the cell cycle. MSC-GFP-Akt1 DNA was stained with 7-AAD and cells ascribed to a phase of the cell cycle using a Watson model. In control MSC-GFP-Akt1 cells, the number of cells in S-phase was ∼2.5-fold lower when compared with cells lacking Abi3bp (Fig. 3E). The number of cells in G0/G1 was significantly lower in the Abi3bp knockdown cells; however, no effect was observed on the G2/M population (Fig. 3E). Effects on the cell cycle were also evaluated by flow cytometry using cells incubated with BrdU. Again Abi3bp knockdown in MSC-GFP-Akt1 cells increased the number of cells in S-phase with a concomitant decrease in G0/G1-phase. No effect was observed on G2/M (Fig. 3F). Finally, immunostaining MSC-GFP-Akt1 cells indicated a greater percentage of cells expressing the marker Ki67, which is expressed only during the active phase of the cell cycle, in Abi3bp knockdown cells when compared with the scrambled control (Fig. 3G). Concentrated media prepared from HEK293 cells expressing the myc-tagged Abi3bp inhibited proliferation of both MSC and MSC-GFP-Akt1 cells when compared with control concentrated media (supporting information Fig. S3E) further validating the antiproliferative effect of Abi3bp.
Abi3bp signaling mechanism
Entry into the cell cycle is controlled by the expression of various cyclins. One of the most important is cyclin-d1. As shown in Figure 4A, all three Abi3bp knockout MSC isolations possessed significantly higher cyclin-d1 expression when compared with their wild-type counterparts (Fig. 4A, supporting information Fig. S4A). The same result was observed in the knockdown system. As shown in Figure 4B, in control MSC-GFP-Akt1 cells, which expressed the scrambled shRNA, cyclin-d1 expression progressively decreased during culture; by day 2, cyclin-d1 levels were ∼30% of basal. Knockdown of Abi3bp prevented any loss of cyclin-d1 expression. To verify that cyclin-d1 expression affects MSC-GFP-Akt1 proliferation stable cell-lines were generated expressing either a control- or a myc-DDK-tagged cyclin-d1 vector (supporting information Fig. S4B). Over-expression of the myc-DDK-tagged cyclin-d1 significantly increased cell number in culture and the number of cells in S-phase (supporting information Fig. S4B).
Cyclin-d1 expression is commonly controlled by Mitogen activated protein kinase (MAPKs) such as Extracellular signal regulated kinase (ERK). Abi3bp knockout increased p-ERK at both 1 and 2 days postseeding in all three MSC isolations when compared with wild-type control cells (Fig. 4A, supporting information Fig. S4A). Similarly, Abi3bp knockdown in MSC-GFP-Akt1 cells increased phospho-ERK2 levels when compared with control cells. Furthermore, Abi3bp knockdown prevented the dephosphorylation observed in the control cells (Fig. 4B). A similar effect was observed with ERK1 but the data did not reach statistical significance. To ascertain whether ERK could affect MSC proliferation, cells of the model MSC-GFP-Akt1 system were made to stably express either a scrambled control shRNA or a shRNA that targeted ERK2, the major ERK isoform in MSCs. Of the two shRNAs, ERK2-shRNA-76 gave the highest level of ERK2 depletion (Fig. 4C, left panel) and had the greatest inhibitory effect on cyclin-d1 expression (Fig. 4C), number of cells in S-phase (Fig. 4C) and growth rate (Fig. 4C, right panels and supporting information Table S1). Similarly, ERK2-shRNA-76 expression in MSC-GFP-Akt1 cells increased the number of cells in G0/G1 (supporting information Fig. S4C). The growth rate of the MSC-GFP-Akt1 cells expressing the ERK2-shRNA-sh76 was virtually identical to unmodified MSCs underlying the importance of ERK for MSC proliferation.
Mitogen activated protein kinase (MEK) is the upstream activator of ERK; the MEK inhibitors U0126 and PD98059 were used to confirm the ERK knockdown experiment and the role of ERK in MSC proliferation. Both U0126 and PD98059 inhibited ERK phosphorylation and cyclin-d1 expression in MSC-GFP-Akt1 cells as determined by immunoblotting (supporting information Fig. S5A). The MEK inhibitors significantly decreased the number of MSC and MSC-GFP-Akt1 cells in S-phase as determined by BrdU incorporation and flow cytometry (supporting information Fig. S5B) further underlining the importance of the ERK pathway in MSC proliferation.
Paxillin phosphorylation, which earlier we showed was regulated by Abi3bp, has been previously correlated both negatively and positively with ERK phosphorylation depending upon cell type [19, 20]. We first verified the staining experiment by immunoblotting. Paxillin phosphorylation at Y188 was found to be negatively regulated by Abi3bp; at both 1 and 2 days postseeding pY118-paxillin was significantly lower in all three Abi3bp knockout MSC isolations when compared with control cells (Fig. 5A, supporting information Fig. S5C). pY118-paxillin increased ∼60-fold in control MSC-GFP-Akt1 cells. However, in Abi3bp knockdown MSC-GFP-Akt1 cells paxillin-Y118 phosphorylation was substantially reduced; at day 2, postseeding paxillin-Y118 phosphorylation was only ∼10-fold above basal (Fig. 5B). We then wished to determine whether there was any link between paxillin and ERK phosphorylation (Fig. 5C). Transfection of MSC-GFP-Akt1 cells with a siRNA pool depleted paxillin protein by more than 90% (Fig. 5D, supporting information Fig. S6A). Interestingly, phospho-ERK1/2 levels were dramatically increased in the paxillin siRNA-treated cells. These effects were not due to transfection or the lipid carrier, neither transfection of a negative siRNA control nor the lipid alone had any effect on paxillin, phospho-ERK1/2, or actin (Fig. 5D). This effect was specific to paxillin, knockdown of another FA protein, vinculin, had no effect on p-ERK (Fig. 5D). Taken together, this indicated that paxillin was sequestering a kinase capable of activating the ERK pathway.
The MEK-ERK pathway is activated by Ras and we determined the activity of this protein by incubating agarose beads coupled to the Ras binding domain of Raf-1, which preferentially binds to the active GTP bound form of Ras, using MSC-GFP-Akt1 protein extracts from control and Abi3bp knockdown cells. Ras activity was found to be significantly increased by Abi3bp knockdown (Fig. 5E). Ras itself is controlled by various pathways, including Grb2-SOS and Src. Grb2-SOS was found not to be important for Ras-MEK-ERK activation in MSCs, peptide-mediated disruption of the Grb2-SOS association had no effect on ERK phosphorylation (Fig. 5E). Inhibition of Src by the inhibitor PP2A inhibited ERK phosphorylation, cyclin-d1 expression, and entry into S-phase (Fig. 5F and 5G, supporting information Fig. S6B) in both MSCs and Akt-MSCs indicating that this tyrosine kinase was responsible for activating the Ras-MEK-ERK pathway.
We then wished to determine how Src was the kinase being sequestered by paxillin. The experiments above showed that Abi3bp knockdown decreased phospho-Y188-paxillin. Considering that paxillin-Y118 phosphorylation is mediated by Src this was an avenue we explored further. Coimmunoprecipitation experiments were used to verify this hypothesis. In control MSC-GFP-Akt1 cells Src coprecipitated with paxillin at both 1 and 2 days postseeding (Fig. 6A). In contrast to the control cells, in MSC-GFP-Akt1 cells lacking Abi3bp the levels of coprecipitated Src were substantially reduced; indicating increased availability of Src which could potentially activate the ERK pathway (Fig. 6A). No coimmunoprecipitation of Src was observed with isotype control antibody (Fig. 6A). Src immunoprecipitates contained equal amounts of Raf-1, a Src target , in control and Abi3bp knockdown MSC-GFP-Akt1 cells (Fig. 6B) indicating that Abi3bp does not control ERK phosphorylation through Src activation of Raf-1.
Abi3bp Binds to Integrin-β1
Given the extracellular nature of Abi3bp and the observation that the protein affected FA formation and related signaling pathways, we surmised that Abi3bp could modulate growth-factor and/or integrin receptors. Abi3bp did not affect growth-factor signaling; equivalent p-ERK1/2 levels were observed in serum starved MSC-GFP-Akt1 cells stimulated with conditioned media prepared from HEK293 cells expressing either a control vector or the myc-tagged Abi3bp vector (Fig. 6C). If the Abi3bp receptor was an integrin, blocking antibodies would increase ERK phosphorylation. We chose to investigate three integrins, α4, α5, and β1 as they are commonly expressed by MSCs. Compared to untreated cells incubation with the isotype control, α4 and α5 blocking antibodies had no effect on ERK phosphorylation. However, incubation with the β1 blocking antibody increased phospho-ERK1/2 levels ∼3.5-fold. Coincubation of the β1 blocking antibody with either α4 or α5 blocking antibody had comparable effects to the β1 antibody alone (Fig. 6D). This result suggested that β1 could be a Abi3bp receptor. To determine whether direct binding could be observed between Abi3bp and integrin-β1, pull-down experiments were performed. Protein extracts from MSC-GFP-Akt1 cells were incubated with conditioned media prepared from fresh serum-free media, serum-free media exposed to HEK293 cells expressing a control vector, or serum-free media exposed to HEK293 cells expressing the myc-tagged Abi3bp vector. Immunoprecipitation with myc resulted in the coprecipitation of integrin-β1 only when myc-tagged Abi3bp was present (Fig. 6E). This experiment was further verified; immunoprecipitation of endogenous Abi3bp from MSC-GFP-Akt1 cells resulted in the coprecipitation of integrin-β1 (Fig. 6E, right panel).
Mice Lacking Abi3bp Show Aberrant MSC Proliferation
Having shown that both Abi3bp knockout and knockdown increased MSC proliferation in vitro, we hypothesized that Abi3bp knockout would affect MSC proliferation in vivo. Abi3bp is expressed in most tissues but is highest in the lung (supporting information Fig. S7A). We analyzed MSC populations by flow cytometry and measured proliferation by BrdU uptake. In the bone marrow we investigated MSCs using two sets of markers; CD45negCD44posCD105pos and CD45negSca-1pos as CD44, CD105, and Sca-1 are known to be expressed on mouse MSCs. First we needed to confirm that CD45negCD44posCD105pos are indeed MSCs. As shown in supporting information Figure S8A, CD45negCD44posCD105pos cells in both wild-type and knockout animals express Sca-1, a mouse MSC marker, but do not possess hematopoietic (CD45, CD34) or endothelial markers (CD31) indicating that the CD45negCD44posCD105pos population are MSCs. CD45negCD44posCD105pos number was increased by Abi3bp knockout (Fig. 7A). Furthermore, Abi3bp knockout enhanced the proliferation of these cells as shown by the approximately threefold increase in BrdU incorporation (Fig. 7B). Similarly, the proliferation rate of CD45negSca-1pos cells was increased in the Abi3bp knockout bone marrow (Fig. 7C). We wanted to see if the cultured MSCs were related to the in vivo MSCs we had characterized. Indeed we found that CD45negCD44pos cells are enriched in the tissue culture plastic attached population following 12 hours of culture, suggesting that the CD45negCD44posCD105pos give rise to the cultured MSCs we investigated above (supporting information Fig. S8B). In the liver we investigated two MSC populations, CD45negCD73posCD90posCD105pos and CD45negSca-1pos. Akin to the bone marrow liver CD45negCD73posCD90posCD105pos MSCs were elevated in the Abi3bp knockout animal (Fig. 7D). Further analysis using BrdU showed that MSCs were ∼3.5-fold more proliferative in the absence of Abi3bp (Fig. 7E). Similarly, Abi3bp knockdown increased the proliferation of liver CD45negSca-1pos MSCs (Fig. 7F). Analogous to the liver, lung CD45neg CD73pos CD90pos CD105pos MSC numbers were elevated and more proliferative in the Abi3bp knockout animal (Fig. 7G, supporting information Fig. S9B). No effect was observed with the kidney CD45negCD73posCD90posCD105pos MSCs (supporting information Fig. S9B). In order to verify that CD45negCD73posCD90posCD105pos cells are MSCs, these cells were isolated from the liver and lung and cultured. These cells displayed typical behavior of MSCs, differentiating into chondrocytes, adipocytes, bone, and smooth muscle (supporting information Fig. S9A).
In this study, we show that Abi3bp is necessary for the co-ordination of proliferation and differentiation of MSCs. (supporting information Fig. S6).
It is clear from our data that Abi3bp is an important determinant of MSC biology. Abi3bp was found to promote the rapid turnover of FAs and reduce cellular stress arising from the ECM. Such changes in cell biology are critically important for differentiation. For example, hepatocyte dedifferentiation occurs via cellular flattening, a process involving the integrin receptor α5β1 and fibronectin . More relevantly, using a microimprinting technique to force MSCs to maintain a rectangular or pentagonal shape affected whether the cells developed down an osteogenic or adipogenic pathway . Similarly, MSC elongation was found to be necessary for myogenic differentiation . The amount of cellular tension, as mediated by RhoA, also affects whether an MSC develops into an adipocyte or osteoblast .
Proliferation and differentiation are often regarded as two sides of the same coin. Differentiation occurs once the cell stops proliferating. Our results support this hypothesis and show that MSCs require Abi3bp to switch from a proliferative to differentiating state. In the absence of Abi3bp, the integrin-β1 is maintained in a nonactive state and the lack of phosphorylated paxillin prevents sequestration of Src and ERK at the plasma membrane, leaving these kinases to activate cyclin-d1 and drive proliferation. Once Abi3bp is present in sufficient amount, integrin-β1 is activated and drives paxillin phosphorylation. The phosphorylated paxillin binds to Src and ERK, and by localizing these kinases at the plasma membrane prevents cyclin-d1 activation in the nucleus. Proliferation stops and the MSC enters into a differentiation pathway, as evidenced by the failure of Abi3bp knockout MSCs to differentiate. This has been observed in other cell types, for example, integrin-β1 activation and paxillin phosphorylation have been shown to be important for Schwann cell differentiation . Proliferation of MSCs was not important for their transformation ability, Akt-MSCs lacking Abi3bp, which had a higher proliferative state than control cells, had a highly significant reduced rate of transformation (supporting information Fig. S7B). This is important information considering the use of MSCs as therapeutic agents.
With respect to the mechanism, it is worth bearing in mind that several experiments rely on the MSC-GFP-Akt1 cell system. In MSC-GFP-Akt1 cells Abi3bp binds to integrin-β1, elevates p-paxillin, which in turn inhibits p-ERK and proliferation through sequestration of Src. These findings are applicable to unmodified MSCs. We found that Abi3bp knockout MSCs were more proliferative than their wild-type counterparts. Similarly, p-ERK and cyclin-d1 were increased by Abi3bp knockout and likewise p-paxillin levels were reduced. Furthermore, by chemical inhibition, we showed that MSC proliferation is dependent upon Src and ERK.
In the Abi3bp knockout mouse, we found elevated MSCs numbers and proliferation in the bone marrow, lung, and liver. However, we did not observe any obvious phenotype. The properties of MSCs in vivo remain elusive . MSCs in the bone marrow exert important effects upon hematopoiesis. In our knockout animal, we do observe a small increase in eosinophil number within the bone marrow suggestive of the possibility that increased MSC numbers may alter the balance of hematopoiesis. Few studies exist regarding the role of MSCs in the lung and liver (reviewed in ). Mouse strains have remarkably dissimilar numbers of MSCs, for example, the bone marrow of FVB mice have 10-fold higher MSC numbers than that of C57BL/6 . Yet there is no correlation with longevity, nor obvious differences in pathology in any organ between these strains suggesting that elevated MSC number may not lead to an obvious phenotype. Considering the lack of phenotype in these tissues we are currently investigating whether injury models will clarify MSC functions in these organs.
In summary, we have identified Abi3bp as a new ECM protein necessary for the switch from proliferation to differentiation in MSCs. By binding to the integrin-β1 Abi3bp blocks MSCs from proliferating and co-ordinates entry into a differentiation mechanism.
We would like to thank the DHVI and John Wong for the use of their flow cytometers. This work was supported by the NIH.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.