|Home | About | Journals | Submit | Contact Us | Français|
The capability to engineer microenvironmental cues to direct a stem cell population toward multiple fates, simultaneously, in spatially defined regions is important for understanding the maintenance and repair of multi-tissue units. We have previously developed an inkjet-based bioprinter to create patterns of solid-phase growth factors (GFs) immobilized to an extracellular matrix (ECM) substrate, and applied this approach to drive muscle-derived stem cells toward osteoblasts ‘on–pattern’ and myocytes ‘off–pattern’ simultaneously. Here this technology is extended to spatially control osteoblast, tenocyte and myocyte differentiation simultaneously. Utilizing immunofluorescence staining to identify tendon-promoting GFs, fibroblast growth factor-2 (FGF-2) was shown to upregulate the tendon marker Scleraxis (Scx) in C3H10T1/2 mesenchymal fibroblasts, C2C12 myoblasts and primary muscle-derived stem cells, while downregulating the myofibroblast marker α-smooth muscle actin (α-SMA). Quantitative PCR studies indicated that FGF-2 may direct stem cells towards a tendon fate via the Ets family members of transcription factors such as pea3 and erm. Neighboring patterns of FGF-2 and bone morphogenetic protein-2 (BMP-2) printed onto a single fibrin-coated coverslip upregulated Scx and the osteoblast marker ALP, respectively, while non-printed regions showed spontaneous myotube differentiation. This work illustrates spatial control of multi-phenotype differentiation and may have potential in the regeneration of multi-tissue units.
The musculoskeletal system comprises multiple tissue types including bone, muscle, tendon, ligament and cartilage as well as their respective tissue interfaces such as bone-to-tendon entheses and muscle-to-tendon junctions. The maintenance and repair of these multi-tissue structures involves the spatial control of stem cell differentiation toward tissue-specific cells, such as osteoblasts, tenocytes and myocytes . This process is regulated by physical and biochemical microenvironmental cues imparted by the interactions of cells with their extracellular matrix (ECM), neighboring cells, and secreted local and systemic signaling molecules, including growth factors (GFs) [1, 2]. Signaling molecules, regulate the pericellular environment where they reside in both the ‘liquid-phase’ (freely diffusing) form and the ‘solid-phase’ (immobilized) form that exists in an equilibrium state between desorption from and adsorption to the ECM and cell surfaces . The unique architecture and biochemical composition of the ECM allows it to sequester (immobilize) and release GFs at picogram to nanogram levels [3–7], and can negatively or positively regulate GF bioactivity and bioavailability . As such, GF sequestration by the ECM immobilizes GFs to specific locations, which in turn imparts the temporal and spatial cues required for directing cell behaviors such as cell adhesion, migration, proliferation, differentiation and apoptosis, which are vital for orchestrating complex processes such as development, maintenance and wound healing [3, 7–19]. Therefore, developing toolsets that can be used to selectively control the physical placement and dosage of multiple exogenous GFs in a physiologically-relevant manner in order to spatially direct a stem cell population toward multiple cell fates simultaneously is a logical consideration for studying stem cell behaviors and may also have direct applications in regenerative medicine.
Prior work reported by our group and by others has shown that ECMs patterned with solid-phase GFs can be engineered to control various aspects of stem cell behavior, including proliferation, migration and differentiation in vitro [8, 13, 14, 16, 20–23] as well as differentiation in vivo . We previously demonstrated that a GF-patterned fibrin ECM created using an inkjet-based bioprinting technology can drive a single stem cell population toward osteoblast and myocyte fates simultaneously, in registration to printed patterns in vitro . In the work presented here, we report on the extension of this approach to spatially drive stem cell differentiation towards a tendon fate simultaneously with osteoblast and myocyte differentiation.
Using solid-phase GFs to direct stem cells to tenocytes in vitro has not been previously reported in literature. Therefore, prior to studying multi-lineage patterning, candidate tendon-promoting GFs had to be identified and validated. Candidate GFs were screened against mouse C3H10T1/2 mesenchymal fibroblasts, C2C12 myoblasts and primary muscle-derived stem cells (MDSCs) using both liquid- and solid-phase immunofluorescence staining for the tendon marker Scleraxis (Scx) [24, 25]. Quantitative PCR studies were subsequently performed to elucidate the mechanism by which stem cells differentiated towards a tendon lineage. Following this, solid-phase presentation of FGF-2 and/or BMP-2 on fibrin-coated glass coverslips using either coarse hand-printing or high resolution, low-dose inkjet bioprinting was used to demonstrate spatial control of stem cell differentiation towards multiple cell fates simultaneously.
Multipotent mouse C3H10T1/2 cells (ATTC, Manassas, VA) were grown in Dulbecco’s Modified Eagle’s Media (DMEM; Invitrogen, Carlsbad, CA), 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and 1% penicillin-streptomycin (PS; Invitrogen, Carlsbad, CA). Mouse C2C12 cells (ATTC, Manassas, VA) were grown in DMEM, 10% bovine serum (Invitrogen, Carlsbad, CA) and 1% PS. Multipotent MDSCs were isolated from primary mouse gastrocnemius muscle biopsies following a modified preplate technique  and were grown in DMEM (high glucose), 10% horse serum (HS; Invitrogen, Carlsbad, CA), 10% FBS, 0.5% Chick Embryo Extract (Accurate Chemical Co, Westbury, NY) as previously described [26, 27]. For myogenic differentiation, cells were grown in low serum containing myogenic differentiation (DMEM, 2% HS, 1% PS) media for 3–5 days. This media is subsequently referred to as myogenic media or myogenic conditions for the remainder of the text. All cells were kept at 37°C, 5% CO2 in a humidified incubator.
Recombinant human BMP-2 (Genetics Institute Inc, Cambridge, MA), FGF-2 (Peprotech, Rockyhill, NJ), FGF-4 (Peprotech, Rockyhill, NJ) and GDF-7 (Prospc Bio, Rehovot, Israel) were reconstituted according to manufacturer’s instructions to 1–2 mg/mL, aliquoted and stored at −80°C. Prior to use, GFs were freshly diluted to the desired concentration in 10 mM sodium phosphate, pH 7.4. For liquid-phase GF experiments, cells were seeded at 2.6–3.1 × 104 cells/cm2 in the presence or absence of GF (1–500 ng/mL) under proliferation (High serum) and myogenic (Low serum) media for 3–4 days. For solid-phase GF experiments, cells were seeded at 3.1–3.6 × 104 cells/cm2 over printed fibrin-coated coverslips under proliferation and myogenic media for 3–4 days.
Homogenous fibrin films were prepared essentially as described by Campbell et al., 2005 . Briefly, 18 × 18 mm epoxy-silanized glass coverslips (Thermo Fisher Scientific, Waltham, MA) were coated with 0.1 mg/mL fibrinogen (Aventis Behring, King of Prussia, PA or American Diagnostica Inc., Stanford, CT)and converted into fibrin by incubating coverslips in 4 U/mL thrombin (Enzyme Research Laboratories, South Bend, IN). Coverslips were then washed with phosphate buffered saline (PBS) and sterile deionized water before air-drying in a laminar flow hood. The thickness of the fibrin films was previously estimated to be approximately 20 nm .
Prior to printing, GFs were freshly diluted to the desired concentration in 10 mM sodium phosphate, pH 7.4. Prior to filling the inkjet with the GF, the printhead was sterilized by rinsing with 70% ethanol followed by sterile deionized water. The bio-ink, consisting of 100–200 μg/ml GF was loaded into the printhead, and printed onto fibrin-coated glass coverslips as previously described [8, 14]. The concentration of inkjetted GFs can be modulated by overprinting, which is achieved by varying the number of times a GF is deposited in the same (x,y) location. In the case of hand-printed GF patterns, 1–2 μL of a 100 μg/mL GF solution was pipetted onto a fibrin-coated glass coverslip instead and a diamond scribe pen was used to mark the droplet perimeter after it had been allowed to air-dry for 1h at 37°C. After printing, fibrin-coated coverslips were incubated in PBS for 5 min followed by serum-free DMEM with 1% PS overnight at 37°C, 5% CO2 to wash off unbound GF prior to cell seeding. The surface concentration of GF present on fibrin-coated coverslips during cell seeding was estimated based on desorption measurements in previous studies [8, 13, 14, 28].
For the experiments investigating scx expression during muscle differentiation, C2C12 cells were grown at 1.55 × 102 cells/cm2 under proliferation conditions and at 2.5 × 103 cells/cm2 under myogenic conditions for 4 days to ensure similar number of cells in both conditions prior to RNA extraction. C3H10T1/2 cells were grown in proliferation medium at 1.5–2.0 × 104 cells/cm2 in the presence or absence of FGF-2 (50 μg/mL) for 36 h and 72 h prior to extraction of total RNA (RNeasy Mini Kit; Qiagen, Valencia, CA). Quantitative polymerase chain reaction analysis for pea3, erm and scx were performed as previously described [29, 30]. Target gene expression was normalized to 18S internal control. Gene expression is displayed as the mean of five independent experiments and represented as mean ± Standard Error Mean (SEM). Statistical analysis was performed as described below.
Cells were washed in PBS, fixed in methanol for 5 min, air-dried and blocked with 10% donkey serum (Jackson Immunoresearch, West Gove, PA) for 20 min at RT. For mouse-on-mouse staining, an additional blocking step was performed by incubating cells with 100 μg/mL donkey anti-mouse FAB (Jackson Immunoresearch, West Gove, PA) for 1 h at RT. Cells were then rinsed with wash buffer (PBS, 0.1%BSA) and incubated with primary antibodies: rabbit anti-scx (10 μg/mL; Abcam, Cambridge, MA), mouse anti-myosin MF20 (1 μg/mL; DSHB, Iowa City, Iowa), mouse anti-α-smooth muscle actin (α-SMA; 1 μg/mL; Abcam, Cambridge, MA) or goat anti-myogenin (2 μg/mL; Santa Cruz Biotechnology Inc, Santa Cruz, CA) overnight at 4°C. Cells were then rinsed three times with wash buffer and incubated with secondary antibodies for 1 h at RT – donkey anti-goat FITC (4 μg/mL; Santa Cruz Biotechnology Inc, Santa Cruz, CA), donkey anti-mouse Dylight 488 nm or donkey anti-rabbit Dylight 549 nm (15 μg/mL each; Jackson Immunoresearch, West Gove, PA). Lastly, cells were rinsed five times with wash buffer and imaged using a Zeiss Axiovert 200M microscope (Carl Zeiss Microimaging, Thornwood, NY) equipped with a Colibri LED light source. Quantification of immunofluorescence staining was performed using Adobe Photoshop 7.0 (Adobe Systems, San Jose, CA). Briefly, the rectangular marquee tool was used to draw a bounding box (Approximately 700 pixels by 700 pixels representing an area 0.9 mm by 0.9 mm in size) and the image histogram tool was used to measure average pixel intensity. Statistical analysis was performed as described below.
Cells were seeded onto GF patterns for 72 h, washed in PBS and fixed for 2 min in 3.7% formaldehyde. Alkaline phosphatase activity (ALP; SIGMAFAST) was detected according to the manufacturer’s instructions (Sigma-Aldrich, St. Louis, MO). Where required, alkaline phosphatase-stained images were converted to CMYK format since this color format is representative of reflected light colors as opposed to emitted light colors (RGB). Since cyan and magenta form the color blue, these channels were added together and inverted. The average pixel intensity was determined using the image histogram tool in Adobe Photoshop 7.0 (Adobe Systems, San Jose, CA).
For both quantitative PCR and immunofluorescence analysis, one-way analysis of variance followed by Fisher’s least significant difference post hoc test using SYSTAT 9 software (Systat Software Inc., Richmond, CA) was performed to determine significance among treatment groups. A p value ≤ 0.05 was considered statistically significant.
Liquid-phase experiments incorporating a combination of 200 ng/mL BMP-2, 500 ng/mL BMP-12/GDF-7 and 50 ng/mL FGF-2 indicated that C3H10T1/2 cells upregulated the tendon marker Scx in the presence of FGF-2 alone but not with GDF-7 or BMP-2 (Supplementary Fig. 1). In addition, FGF-2-treated cells adopted an elongated spindle-like morphology reminiscent of tenocytes (Supplementary Fig. 1). When treated with 50 ng/mL FGF-4, C3H10T1/2 cells showed punctate nuclear staining of Scx transcription factor (Supplementary Fig. 2).
Scx expression in C2C12 cells was examined under proliferation (high serum) as well as myogenic (low serum) conditions to determine if such cells could undergo tendon specification when treated with FGF-2. Under proliferation conditions, increasing amounts of FGF-2 resulted in upregulation of the tendon marker Scx in a dose-dependent manner with punctate nuclear staining of Scx observed in cells treated at 25 ng/mL FGF-2 and 50 ng/mL FGF-2 (Supplementary Fig. 3). Under myogenic conditions, Scx expression was unexpectedly upregulated in nascent myotubes in the absence of FGF-2 (Supplementary Fig. 4). Confocal sectioning studies determined that the thickness of myotubes, which is observed as a bright halo around cells in phase-contrast images (Supplementary Fig. 4), contributed in part to the high levels of Scx (data not shown). Furthermore, quantitative PCR analysis indicated that there was no fold change in scx expression between proliferating C2C12 cells (1.003 ± 0.075 fold change) and differentiating myotubes (1.022 ± 0.209 fold change). In the presence of FGF-2, myotube formation was inhibited and cells showed increased Scx expression when compared to non-myocytes in untreated control (Supplementary Fig. 4).
Tendon specification in MDSCs was examined in cells treated with FGF-2 under proliferation and myogenic conditions. Under proliferation conditions, FGF-2 dose-dependently increased expression of the tendon marker Scx with punctate nuclear staining of Scx occasionally observed at 50 ng/mL FGF-2 (Fig. 1). Similar to C2C12 cells, Scx expression was upregulated in nascent myotubes in the absence of FGF-2. In the presence of FGF-2, myotube formation was inhibited with MDSCs exhibiting lower levels of the myogenic marker myogenin (Supplementary Fig. 5). In addition, Scx expression was upregulated when compared to non-myocytes in untreated control and FGF-2-treated cells did not show increased expression for the myofibroblast marker α-smooth muscle actin (α-SMA; Supplementary Fig. 6).
As previous studies indicated that members of the Ets family of transcription factors such as pea3 and erm were involved in regulation of scx during tendon development in the chick, quantitative PCR analysis of these genes were performed to determine if a similar mechanism was operating in these stem cell populations . 50 ng/mL FGF-2 upregulated pea3 (43.1 ± 27.5 fold change, p = 0.005) and erm (16.5 ± 6.5 fold change, p = 0.178) at 36 h whereas scx levels remained constant. At 72 h, all three genes were upregulated: pea3 (169.6 ± 45.8 fold change, p = 0.008), erm (72.5 ± 16.2 fold change, p = 0.033) and scx (22.8 ± 5.4 fold change, p = 0.006). As such, the prior induction of pea3 and erm suggest that these two genes lie upstream of scx (Fig. 2).
Having demonstrated that Scx expression was upregulated by liquid-phase FGFs, square patterns of FGF-2 (each measuring 1 by 1 mm) were inkjet printed onto fibrin-coated glass coverslips with 2, 6 and 12 overprints to determine if solid-phase GF patterns can spatially direct tendon specification in a dose-dependent manner. Our previous studies have shown that the surface concentration of GF that is deposited can be modulated by overprinting and that such GF patterns can persist for up to 144 hours under standard cell culture conditions [10, 13, 14, 16]. As shown in Fig. 3A, the amount of FGF-2 deposited in 2, 6 and 12 overprints after washing and prior to cell seeding was estimated to be 40.8 pg/mm2, 122.4 pg/mm2 and 244.8 pg/mm2 FGF-2 based on previous studies [8, 13, 14, 28]. Under proliferation conditions (High serum), C3H10T1/2 cells showed upregulation of Scx in response to solid-phase patterning of FGF-2 in a dose-dependent manner (Fig. 3B). Although the lowest dose of solid-phase FGF-2 (40.8 pg/mm2) was not sufficient to induce an increase in Scx expression relative to negative control/non-printed regions (p = 0.872), higher doses of solid-phase FGF-2 resulted in an increase in Scx expression relative to negative control/non-printed regions (p = 0.009 for 122.4 pg/mm2 FGF-2 and p = 0.001 for 244.8 pg/mm2 FGF-2; Fig. 3) in C3H10T1/2 cells. Thus, solid-phase patterning of FGF-2 can spatially control tendon cell fate (Fig. 3).
Similarly, square patterns of FGF-2 (each measuring 1 by 1 mm) were inkjet printed onto fibrin-coated glass coverslips with 5, 10 and 30 overprints (corresponding to an estimated amount of 102 pg/mm2, 203 pg/mm2 and 612 pg/mm2 FGF-2) to determine if multiple stem cell fates could be spatially controlled in a dose-dependent manner within the same construct. Under both proliferation and myogenic conditions, inkjet printed patterns of FGF-2 resulted in a dose-dependent increase in Scx expression (Fig. 4B, C).
Under proliferation conditions, although the lowest dose of solid-phase FGF-2 (102 pg/mm2 FGF-2) was not sufficient to induce an increase in Scx expression relative to negative control/non-printed regions (p = 0.099), higher doses of solid-phase FGF-2 resulted in an increase in Scx expression relative to negative control/non-printed regions (p = 0.01 for 203 pg/mm2 FGF-2 and p = 0.000 for 612 pg/mm2 FGF-2) in C2C12 cells (Fig. 4C). Under myogenic conditions, although lower doses of solid-phase FGF-2 were not sufficient to induce an increase in Scx expression relative to negative control/non-printed regions (p = 0.139 for 102 pg/mm2 FGF-2 and p = 0.053 for 203 pg/mm2 FGF-2), the highest dose of solid-phase FGF-2 resulted in an increase in Scx expression relative to negative control/non-printed regions (p = 0.022 for 612 pg/mm2 FGF-2) in C2C12 cells (Fig. 4C). Within and outside the printed regions, cells fused to form multinucleated myotubes as a result of high cell density leading to spontaneous cell fusion under proliferation conditions or direct myogenic induction, as evidenced by the presence of muscle myosin or MF20 (Fig. 4B). No difference in MF20 staining was observed between printed and non-printed regions (p > 0.05 for all cases). Taken together, these results demonstrate the simultaneous specification of myocyte and tenocyte fates within the same construct in a spatially defined manner in a dose-dependent fashion.
Having demonstrated that multiple stem cell fates could be spatially controlled within the same construct, inkjet bioprinting technology was applied to determine if osteoblast, tenocyte and myocyte fates, representative of a primitive bone-tendon-muscle unit, could be simultaneously specified within the same construct. After 72 h in proliferation media, inkjet printed patterns of BMP-2 and FGF-2 increased ALP and Scx expression, respectively (Fig. 5). On inkjet printed patterns of BMP-2, ALP expression was increased relative to negative control/non-printed regions (p = 0.000) and inkjet printed patterns of FGF-2 (p = 0.000) but no increase in Scx expression was observed relative to negative control/non-printed regions (p = 0.146). On inkjet printed patterns of FGF-2, Scx expression was increased relative to negative control/non-printed regions (p = 0.000) and inkjet printed patterns of BMP-2 (p = 0.003) but no increase in ALP expression was observed relative to negative control/non-printed regions (p = 0.887). Within and outside the printed regions, myotube formation was promoted due to the high density of cells resulting in spontaneous fusion of cells (Fig. 5B). No difference in MF20 staining was observed between printed and non-printed regions (p > 0.05 for all cases). Taken together, these results demonstrate the simultaneous specification of osteoblast, tenocyte and myocyte fates within the same construct in a spatially defined manner.
To rule out the possibility that FGF-2 was directing C2C12 cells towards a myofibroblast fate as opposed to a tenocyte fate, FGF-2 was hand-printed onto a fibrin-coated coverslip and immunostained for the myofibroblast marker α-SMA (Fig. 6). After 72 h in proliferation media, α-SMA, which is transiently expressed during muscle differentiation , was downregulated on hand-printed patterns of FGF-2, indicating that these cells were differentiating towards a tenocyte fate as opposed to a myofibroblast fate.
We previously used our printing approach to demonstrate spatial control of adjacent regions of osteoblast-myocyte differentiation . In particular, when C2C12 cells and MDSCs were cultured under myogenic conditions on BMP-2 patterns printed on fibrin, cells ‘on-pattern’ differentiated toward the osteoblast lineage, whereas cells ‘off-pattern’ differentiated toward the myogenic lineage. The purpose of the research reported here was to extend our prior work to the control of more complex osteoblast-tenocyte-myocyte units, representing primitive but physiologically relevant [33, 34] constructs.
In our original osteoblast-myocyte patterning experiments , osteoblasts were explicitly induced ‘on-pattern’ with solid-phase BMP-2, while myocytes were implicitly induced ‘off- pattern’ using myogenic media. In an effort to explicitly pattern myocytes for this current study, we investigated numerous GFs and signaling molecules implicated in muscle differentiation, including amphoterin/HMGB1, decorin, follistatin, ghrelin, galectin-1, interleukin-4, insulin-like growth factor-1, insulin-like growth factor-2, neuregulin-1 beta 2, sonic hedgehog and wnt3A [7, 35–45]. However, these candidate muscle-promoting cues did not elicit an increased myogenic response relative to control under proliferation conditions in either liquid- or solid-phase experiments (Data not shown). Therefore, we continued to rely on implicit patterning of myocytes off-pattern using either myogenic conditions to induce myogenic differentiation through serum starvation or proliferation conditions to increase cell-cell contact via cell receptors such as N-Cadherins, leading to spontaneous cell fusion and myotube formation [46, 47].
There have been no prior reports on the use of solid-phase GFs to direct stem cells toward tenocytes in vitro. Furthermore, even liquid-phase protocols for differentiating these cell types to a tendon fate have yet to be firmly established. Since previous studies demonstrated that members of the BMP and FGF family of signaling proteins may be involved in tendon formation [31, 48–50], we first screened FGF-2, FGF-4, BMP-2 and BMP-12/GDF-7 in liquid-phase experiments using immunofluorescence staining for the tendon marker Scx  to determine if these GFs could direct multipotent stem cells towards a tendon cell fate. Only FGF-2 and FGF-4 in liquid-phase forms were shown to direct C3H10T1/2 cells, C2C12 cells and MDSCs towards a tendon lineage (Supplementary Figs. 1–5 and Fig. 1). To rule out the possibility that FGF-2-treated cells were differentiating towards myofibroblasts as opposed to tenocytes, FGF-2-treated cells were stained for the myofibroblast marker α-SMA. In these studies, FGF-2-treated cells did not induce upregulation of α-SMA in either C2C12 cells or MDSCs, and in fact, reduced expression of α-SMA slightly (Fig. 6 and Supplementary Fig. 6). Quantitative PCR analysis subsequently determined that the mechanism by which these stem cells differentiate towards a tendon fate may involve members of the Ets family of transcription factors such as pea3 and erm, which may act upstream of the tendon transcription factor scx (Fig. 2), a finding that parallels tendon development in chick .
Unexpectedly, high levels of Scx expression were observed in nascent myotubes (Supplementary Figs. 4 and 5). This was later shown to be contributed, in part, by the thickness of myotubes, as evidenced by the bright halo around cells in phase-contrast images (Supplementary Figs. 4 and 5) and confocal sectioning studies (data not shown). In addition, there was no change in scx gene expression during myogenesis although increased levels of Scx protein in myotubes could be accounted for by post-transcriptional processes. Alternatively, myotubes may show increased Scx levels since Scx is a transcription factor that lies upstream of collagen, a major non-contractile component of muscle. However, this is unlikely since nascent myotubes exhibit little to no nuclear staining of Scx when compared to FGF-2-treated cells (Supplementary Figs. 4 and 5).
FGF-2 was used for all subsequent solid-phase tenocyte patterning experiments using high resolution, low-dose inkjet bioprinting because FGF-4 elicted a lower response in hand-printed and qPCR experiments (Data not shown). In these studies, C3H10T1/2 cells were shown to upregulate the tendon marker Scx in a dose-dependent fashion in response to inkjet printed patterns of FGF-2 (Fig. 3). Similarly, C2C12 cells also dose-dependently upregulated the tendon marker Scx in response to inkjet printed patterns of FGF-2, with spontaneous fusion of myotubes occurring predominantly outside the printed region (Fig. 4). Having demonstrated that solid-phase patterning of tenocytes and myocytes can be engineered under proliferation and myogenic conditions within the same construct, adjacent printed patterns of FGF-2 and BMP-2 were tested and shown to specify osteoblasts, tenocytes and myocytes within the same construct, representing a primitive muscle-tendon-bone unit (Fig. 5). Although FGF-2 displays a clear dose-dependence increase on tendon cell differentiation, some variations in cell responses to printed GF patterns were observed. For example, some C2C12 cells that were ‘off-pattern’ but in close proximity to high doses of FGF-2 or BMP-2 patterns exhibited weak Scx (Fig. 4B, high serum panel) or ALP staining (Fig. 5B), respectively. These variations may stem from desorption of GF from the printed region followed by its readsorption outside the printed region prior to cell seeding or by paracrine signaling from cells ‘on-pattern’ to cells ‘off-pattern’ to differentiate. In addition, the differentiation response of cells within a printed GF pattern was not homogenous throughout as shown by non-uniform Scx staining along with sporadic MF20 staining on printed FGF-2 patterns (Figs. 3–5) and uneven ALP staining on printed BMP-2 patterns (Fig. 5). This may be a function of several factors, including: non-uniform GF distribution within the printed region following inkjet printing and GF drying ; GF desorption followed by readsorption prior to cell seeding [13, 14, 51, 52]; uneven cell density during cell seeding; cell heterogeneity ; or, a combination of all these factors. Furthermore, when multiple GFs are utilized, the dosage of individual GFs was found to be critical for simultaneously specifying multiple cell fates. When similar concentrations of FGF-2 and BMP-2 were inkjet bioprinted in close proximity, no positive ALP staining was observed on printed BMP-2 patterns (Data not shown). This may be attributed to desorption of FGF-2 from the fibrin surface followed by binding to the surface of cells seeded on printed BMP-2 patterns, resulting in the inhibition of BMP-2-induced osteoblast differentiation, which is an effect that is well characterized [14, 54]. This problem was eventually resolved by empirically optimizing the amount of BMP-2 and FGF-2 deposited by inkjet bioprinting, resulting in excess surface concentration of BMP-2 to overcome this inhibitory effect (Fig. 5).
This work has demonstrated that inkjet-based bioprinting technology enables the investigation of spatial control of solid-phase GF-directed differentiation of stem cells toward single or multiple fates in physiologically-relevant engineered microenvironments in vitro. Our prior work also provides support for the application of this technology in vivo . Together, these in vitro and in vivo studies suggest that this technology may have practical implications for both basic scientific research and therapy development and deployment targeting the musculoskeletal system.
This report identified both liquid- and solid-phase FGF-2 as being capable of upregulating the tendon marker Scx in C3H10T1/2 cells, C2C12 cells and MDSCs. Quantitative PCR analysis suggests that members of the Ets family of transcription factors such as pea3 and erm may lie upstream of scx. This report also demonstrates how inkjet bioprinting technology can create persistent GF patterns that direct a single stem cell population towards multiple fates, including tenocytes, myocytes or osteoblasts, within the same construct in a spatially defined manner. This capability not only offers an approach to study multi-lineage differentiation in vitro, but may also be translatable to new therapies to treat disease and trauma of the musculoskeletal system.
We would like to thank James Fitzpatrick for assistance with fluorescence microscopy and Larry Schultz for assistance with GF printing. This work was supported by NIH grants RO1EB004343 and RO1EB007369 as well as funding from the Pennsylvania Infrastructure Technology Alliance (PITA).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.