Utilizing Core–Shell Fibrous Collagen-Alginate Hydrogel Cell Delivery System for Bone Tissue Engineering

Core–shell fiber processing

A dual concentric nozzle (inner 23G and outer 17G) was specifically designed and used to produce a core–shell structured fibrous network of collagen-alginate. The sodium alginate used (A2158; Sigma-Aldrich) had an M/G ratio of 1.67 and a molecular weight of ∼50,000 Da. Solutions of sodium alginate at ratios of 2%–5% (% wt) in water were fed into the outer syringe, while the collagen solution was loaded in the inner syringe. The schematic representation is shown in Figure 1a. The collagen (rat tail type I collagen; First Link; 2.05 mg/mL) solution was prepared by mixing 1 mL of collagen with 100 μL of 10× Dulbecco's modified Eagle medium and a proper amount of 1 N sodium hydroxide to provide a neutral pH to adjust solution conditions for subsequent gelation and providing a nontoxic environment for the loaded cells. Each syringe was attached to an injection pump (KD Scientific) connected through a microtube and the injection rate varied between 20 and 80 mL/h. Core–shell structured collagen-alginate fibers were then injected through the concentric nozzle into a bath containing 50 mM calcium chloride (CaCl₂) for 5 min. The whole injection process was performed under sterile conditions. Following injection, the outer alginate was in contact with the calcium ions, which resulted in crosslinking and gelation to maintain the stability of the fibrous cell carrier.

OPEN IN VIEWER

FIG. 1.

Schematic showing the equipment setup for producing the stem cell-loaded collagen core and alginate shell hydrogel carrier. (a) Each solution of mesenchymal stem cell (MSC) collagen (2) and alginate (3) was advanced at a constant speed from injectors (1), which were directly deposited using the core–shell needle (4, 5) designed for this study under a low concentration (50 mM) of calcium chloride solution bath (6), during which time the alginate was allowed to harden through crosslinking of the Na-alginate with the Ca ion. The MSCs loaded within the collagen inner core became encapsulated within the hardened alginate. The direct deposited cell construction was subsequently incubated at 37°C to allow the gelation of collagen, after which, it was further incubated for cell cultivation for bone tissue engineering. (b) Deposited core–shell hydrogel carrier preserving a continuous fiber structure. (c) Core–shell fibrous hydrogel carrier is easily molded into any determined shape (letter DKU replicated as an example), tuning to the tissue-defect geometry.

Characterizations of core–shell hydrogels

After deposition of the core–shell fibrous structure, the thickness of the core and shell was measured from 10 arbitrarily selected images obtained by optical microscopy. Fibrous hydrogel samples prepared with varying injecting speed (20, 50, and 80 mL/h) or alginate concentration (2%, 3%, and 5%) were used for the measurement. The mechanical properties of the fibrous hydrogels were measured by dynamic mechanical analysis (DMA; MetraVib, DMA25N) using a parallel plate configuration. For this, the deposited fibers were stacked into a cylindrical Teflon mould (12 mm diameter×6 mm height) at a constant volume (0.5 mL of sample). Mechanical testing was carried out using dynamic frequency sweep with frequencies ranging from 0.1 to 10 Hz at 37°C and with strain amplitude of 5%, which was in the linear region of viscoelasticity. Both auto-compression and auto-strain adjustment were applied. The force was ramped from 0.001 to 0.2 N and the maximum allowed strain was set at 10%. The storage modulus (E′) and loss modulus (E″) of the samples were measured. The phase angle delta (tan delta) was computed as E″/E′ (n=3).

The water uptake capacity of the fibrous hydrogel cell carriers was assessed. After immersion of each sample in distilled water for 1 h, the sample was taken out and blotted on a filter paper saturated with distilled water, and the weight was recorded (W_im). After washing and freeze drying the sample, the dried weight was recorded (W_dry). The water uptake capacity was determined as (W_im−W_dry)/W_dry and was expressed as percentage (n=3). Degradation of the samples (0.3 mL for each sample) was tested by immersion of each sample in phosphate-buffered saline (PBS) (n=3). After immersion for various time periods up to 21 days, the sample was obtained, washed, and dried, and the weight after the degradation test was recorded (W_deg). The degradation rate was considered as the percentage of (W_dry−W_deg)/W_dry.

Cell isolation

MSCs derived from rat bone marrow were harvested from the femora and tibiae of adult rats (180–200 g) according to the guidelines approved by the Animal Ethics Committee of Dankook University. The harvested product was centrifuged and the supernatant was collected and suspended within a culture flask containing a normal culture medium; the α-minimal essential medium (α-MEM) supplemented with 10% fetal bovine serum (FBS) and 100 U/mL penicillin and 100 mg/mL streptomycin in a humidified atmosphere of 5% CO₂ in air at 37°C. After incubation for 1 day, the medium was refreshed and cultured until the cells reached near confluence. The cells were subcultured by trypsin and maintained in the normal culture condition. Cells at two to three passages were used for further tests.

Cell encapsulation

For cell encapsulation, sodium alginate powder was sterilized with ethylene oxide and then preserved under sterile conditions. The powder was mixed with distilled water to produce 3% sodium alginate solutions. The collagen solutions in product were in sterile conditions and were used as received. To incorporate MSCs in the collagen hydrogel, the collagen hydrogel was prepared as previously described, incorporating 1×10⁵ cells after the pH reached a neutral value, which was observed by the color change of the collagen solution (from yellow to orange-pink). The cell-loaded collagen solution was put in the inner syringe, whereas the alginate solution was in the outer syringe, and a fibrous structuring was performed in sterile conditions as described above. The amount of collagen present in the core–shell fiber structure was maintained at 0.3 mL for all cases (n=3). As a result, a cell-encapsulated fibrous hydrogel (cell-collagen in the core and alginate in the shell) was obtained. After scaffold construction for 5 min, the cell-encapsulated hydrogel was washed with a culture medium three times to remove the excess CaCl₂, and then cultured in the α-MEM containing 10% FBS and 1% penicillin/streptomycin in a humidified incubator at 37°C under 5% CO₂ for up to 21 days. As the positive control group, 0.3 mL of collagen hydrogel containing 1×10⁵ MSCs was molded in a culture well without the core–shell fibrous structuring (n=3).

Confirmation of cell viability and growth

The number of viable cells in the core–shell fibrous cell carrier or in the collagen gel was measured. At each culture period, hydrogel samples containing cells were washed with PBS twice and dissolved with 1% type I collagenase (Sigma) for 5 min at 37°C. The suspension was then centrifuged to remove the supernatant and the number of viable cells was quantified by Trypan blue screening and using a hemocytometer. The cell proliferative potential was measured using the double-stranded DNA detection kit (Quanti-iT Picogreen; Invitrogen). For this, the hydrogels incorporating cells were washed and dissolved with 1% type I collagenase (Sigma). After collecting the cell pellet, the samples were stored at −20°C. The mixture was then freeze–thawed by two repeating cycles. The cells were quantified according to the manufacturer's specifications by measuring the fluorescence intensity at an excitation of 485 nm and emission of 530 nm. A standard curve was also made according to the manufacturer's instructions and was used for converting fluorescence intensity to DNA content.

Observation of cell morphology

The cell morphology was monitored under light microscopy using an Olympus BX-50 microscope during culture. At each culture period (7, 14, and 21 days), the cells grown on each sample were fixed with 4% paraformaldehyde and then treated with 0.1% Triton X-100. After washing with PBS, Alexa Fluor 546-conjugated phalloidin (Invitrogen A22283) diluted in PBS was added to each sample to stain the F-actin. A ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (Invitrogen) was used to stain the cells and produce blue fluorescence upon binding to the DNA in the nucleus, after which, images were obtained by confocal laser scanning microscopy using a Zeiss LSM 510 microscope.

Determination of cell differentiation by quantitative polymerase chain reaction

MSCs encapsulated within the hydrogel cell carrier were induced to osteogenic differentiation in an osteogenic medium containing 50 μg/mL ascorbic acid, 10 mM β-glycerol phosphate, and 10 nM dexamethasone. Gene expression associated with osteogenesis, including osteocalcin (OCN), bone sialoprotein (BSP), and osteopontin (OPN), was assessed by quantitative polymerase chain reaction (qPCR). After culture for 14 and 21 days, the cell-loaded scaffolds were washed and then dissolved with collagenase. The cell pellet was collected and frozen at −80°C until used. Total RNA was extracted from differentiated cells using the RNeasy Kit (Qiagen) according to the manufacturer's instructions. The concentration of RNA samples was quantified using a Nanodrop 2000 spectrophotometer (Thermo Scientific). RNA samples were transcribed into cDNA by combining the RNA with random hexamer, followed by denaturation at 70°C for 5 min. After denaturation, the samples were mixed with the reverse transcription (RT) premix for complete cDNA synthesis. cDNA was then amplified using RT-PCR premix (Bioneer) at 94°C for 5 min, followed by 35 cycles of 94°C for 1 min denaturation, 57°C for 1 min for annealing, and 72°C during 1 min for extension. Primers used for amplification were OCN: AGGACCCTCTCTCTGCTCAC (forward), AACGGTGGTGCCATAGATGC (reverse); BSP: CTGCTTTAATCTTGCTCTG (forward), CCATCTCCATTTTCTTCC (reverse); OPN: TCCAGCTGACTTGACTCATGG (forward), CCGATGAATCTGATGAGTCCTT (reverse); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH): CCATGTTTGTGATGGGTGTG (forward), GGATGCAGGGATGATGTTCT (reverse). Results were normalized with respect to GAPDH.

In vivo delivery in rat calvarium defect

Tissue compatibility and the bone forming ability of the MSCs delivering hydrogel cell carriers were assessed using a rat calvarium defect model. The protocol of the housing, care, and experimental protocol was approved by the Dankook University Institutional Animal Care and Use Committee, Korea. Nine 11-week-old, 350–400 g healthy male Sprague-Dawley rats were used. During the surgical operation, the animals were anesthetized with intramuscular injections of 80 mg/kg zoletil and 10 mg/kg xylazine in the right quadriceps muscle. All instruments were sterilized before the procedures and all the surgical procedures on animals were carried out under general anesthesia using sterile techniques. Following a midline incision through the skin and the excision of the periosteum using a No. 10 blade with a Bard–Parker scalpel, the cranial bone was exposed. A 5-mm diameter, full-thickness cylindrical defect was created at the midportion of each parietal bone by means of a trephine burr. Two defects were created on each animal, and the defects were randomized into three groups. The defect was filled with 0.1 mL of test sample. Group 1 consisted of the cell carrier only. Group 2 comprised the cell carrier encapsulating nondifferentiated MSCs. Group 3 contained the cell carrier encapsulating differentiated MSCs (n=4). For group 3, the MSCs in the hydrogel construct were cultured under an osteogenic medium for one week. After the implantation, the subcutaneous area was sutured with 4-0 absorbable suture material, and the skin was subsequently sutured with 4-0 non-absorbable monofilament suture material. After surgery, the rats were housed individually in cages under conditions at 20°C–24°C and 30%–70% relative humidity and kept on an alternating 12-h light/12-h dark cycle with standard pellet food and water provided ad libitum. Animals were monitored for signs of infection, inflammation, and adverse effects by visual observation for one week.

Analyses of bone regeneration

At 6 weeks postoperatively, the animals were sacrificed for harvesting of the samples and surrounding tissues. The specimens were harvested from each animal at the time of euthanasia and fixed in 10% of neutral buffered formaldehyde solution for at least 24–48 h. The samples were imaged using a high-resolution microcomputed tomography (μCT) system (Skyscan 1072; Skyscan) to evaluate tissue recovery and bone regeneration. The harvested samples were scanned and serial coronally oriented tomograms were reconstructed from the raw images using NRecon CT Reconstruction software. A cylindrical region of interest (ROI) was precisely positioned over the center of each defect, encompassing all new bone within the defect site. Coronally oriented μCT images were then reformatted in an axial orientation. Reconstructed images over the ROI using a CTAn (Skyscan) and 3D images were created. The total volume of newly formed bone within each ROI was measured by assigning a threshold for the total bone content (including trabecular and cortical bone). Bone surface area and surface density were also recorded. Four samples for each group were measured and the total volume of bone is reported (mm³).

After μCT scanning, specimens were decalcified with the RapidCal™ solution and dehydrated in a graded series of increasing ethanol concentrations (70%–100%), and then trimmed and embedded in paraffin. Tissue sections ∼5 μm in thickness were prepared perpendicular to the cranial bone and the sections were stained with hematoxylin–eosin (HE) and Masson's trichrome (MT) for histopathological examination to check biocompatibility and bone formation. Light microscopy digital images were obtained using Meta-Morph and analyzed with computerized image analysis system (Molecular Devices).

Statistical analysis

Experiments were performed in triplicate, unless otherwise specified. Data are expressed as mean±one standard deviation. Statistical analysis was carried out by one-way analysis of variance with the Fisher's post hoc test and the significance level was considered at p<0.05 or p<0.01.

Generation of core–shell hydrogel cell carriers and properties

Using the device (depicted in Fig. 1a), core–shell structured hydrogel cell carriers could be produced. The core–shell fibrous injection through the co-concentric nozzle and the gathering of fibrous scaffolds were optically visualized (Fig. 1b). A continuous fiber preserving the core–shell structure, where the alginate encapsulated the collagen-MSC inner part was successfully generated. The fibrous hydrogel could be easily molded into any determined shape; a letter DKU using a prefabricated mold is shown as an example (Fig. 1c), indicating the tunability of the core–shell fibrous device to tissue defect 3D geometry.

Controlling the thickness of the core and shell part was possible by changing the alginate concentration and the injection speed, which could significantly affect the nutrient/oxygen penetration and the survival of cells encapsulated. Figure 2 shows the effect of these parameters on the shell and core thickness values. The shell thickness increased as the alginate concentration increased (from 200 μm at 2% alginate to 450 μm at 5% alginate, shown in Fig. 2a). The core thickness decreased at the highest alginate concentration (Fig. 2b). The core–shell morphologies also reflected well the core and shell thickness change with respect to the alginate concentration (Fig. 2c–e). The effect of injection speed was also clearly seen; the shell thickness increased as the injection speed increased (from 100 μm at 20 mL/h to 400 μm at 80 mL/h; Fig. 2f), while the core thickness was not changed (Fig. 2g). The optical images also represent the core–shell thicknesses with varying the injection speed (Fig. 2h–j).

OPEN IN VIEWER

FIG. 2.

Effect on alginate shell thickness of (a–e) alginate concentration and (f–j) injection speed. (a) Shell and (b) core thickness measured with respect to varying alginate concentrations. Optical images showing the core–shell fiber morphologies produced with different alginate concentrations; (c) 2%, (d) 3%, and (e) 5%. (f) Shell and (g) core thickness measured with respect to varying injection speed. Optical images showing the core–shell fiber morphologies produced with different injection speed; (h) 20, (i) 50, and (j) 80 mL/h. Arrows indicate the core thickness. Scale bars=400 μm. Color images available online at www.liebertpub.com/tea

The mechanical properties of the core–shell hydrogel cell carrier were measured by DMA. Samples prepared with varying alginate concentrations were compared. Storage modulus (E′) and loss modulus (E") values recorded at different frequencies are shown in Figure 3. The storage modulus increased as the alginate concentration increased (from 30 kPa at 2% alginate to 50 kPa at 5% alginate; Fig. 3a), suggesting that an increasing alginate concentration improved the ability to withstand plastic deformation. The loss modulus presented very similar values for the different alginate concentrations (Fig. 3b), suggesting that the viscous behavior of the scaffolds was not affected by the alginate concentration. Whereas the storage modulus did not show close dependence on the frequency parameter, the loss modulus showed an almost linear increase as the frequency increased. Moreover, the storage modulus values recorded were higher than the loss modulus values for all cases.

OPEN IN VIEWER

FIG. 3.

Mechanical properties of the core–shell structured fibrous hydrogel cell carriers produced with different alginate concentrations (2%, 3%, and 5%), as measured by a dynamic mechanical analysis; (a) storage modulus and (b) loss modulus were recorded at varying frequency (from 0.1 to 10 Hz). Whereas storage modulus noticeably increased as alginate concentrations increased, not much change was observed in the loss modulus. Color images available online at www.liebertpub.com/tea

Degradation of the core–shell hydrogel cell carrier was measured in PBS over 21 days. The cell carrier obtained with 3% alginate was tested as the representative group. Weight loss measured was ∼4% after 3 days, ∼22% after 7 days, and ∼43% after 14 days (Table 1). Further, the cell carrier was shown to take up water as high as 98%, as taken from the weight change between the hydrogel and freeze-dried samples, demonstrating a typical characteristic of hydrogels.

MSC proliferation, morphology, and osteogenesis

The proliferative potential and viability of MSCs encapsulated in the core–shell hydrogel carrier were investigated. Results were compared with the MSCs cultured within the collagen hydrogel matrix, which has been shown to be an effective 3D matrix for cell growth and population and thus used in this study as a positive control. From DNA analysis, MSCs were shown to actively proliferate for both groups with an on-going increase in the proliferation level over 21 days (Fig. 4a). Whereas the cell proliferation level was comparable during 7 days for both groups, the proliferation in the collagen became significantly higher than that in the core–shell design after 14 and 21 days. Live/dead cell assays revealed comparable levels for both the collagen and core–shell groups, with ∼70% of cell viability at all culture periods (Fig. 4b). Figure 4c shows the shrinkage of the collagen core where MSCs populated and proliferated over 21 days. There was a considerable decrease in the core diameter from 1050 μm (initial) to 900 μm at 7 days, 500 μm at 14 days, and 350 μm at 21 days.

OPEN IN VIEWER

FIG. 4.

(a) Proliferation of MSCs within the core–shell fibrous hydrogel carrier over 21 days, as assessed by the DNA quantification. Results are compared with the MSCs cultured within collagen hydrogel used as a positive control. Cell proliferation comparable at 7 days became different at prolonged periods of 14 and 21 days (core–shell<collagen, *p<0.05, statistically different). (b) Live/dead cells quantified for both groups by trypan blue screening and hemocytometer counting, showing comparable levels for both cases (∼70% survival rate at all time points). (c) Collagen core shrinkage measured for up to 21 days, showing a considerable decrease in the core size from ∼1050 μm (initial) to 900 μm at 7 days, 500 μm at 14 days, and 350 μm at 21 days.

Figure 5 displays representative optical microscopy images of MSCs during culture for up to 21 days when encapsulated within the core–shell fibrous hydrogel carrier as well as those incorporated within collagen gel. MSCs initially displayed a round shape soon after encapsulation (Fig. 5a) and a number of cytoskeletal extensions were evident after 7 days of culture (Fig. 5c), which was similarly observed in the collagen gel control group (Fig. 5b, d). Cells in the core–shell carrier became highly elongated at day 14 (Fig. 5e). Interestingly, the cells showed directional elongation along the fiber axis. The cells in the collagen were also highly elongated, but randomly elongated (Fig. 5f). At 21 days, the cells in the core–shell carrier reached near confluence within the collagen core (Fig. 5g), and a similar behavior was observed in the collagen gel (Fig. 5h). The collagen core part in the core–shell system appeared to shrink substantially at prolonged culture periods.

OPEN IN VIEWER

FIG. 5.

Optical microscopic images of the core–shell hydrogel carrier, encapsulating MSCs at different culture periods (a, c, e, g) and collagen gel control incorporating MSCs at the corresponding culture period (b, d, f, h). MSCs with a round shape initially at the time of seeding (a, b), became spindle shaped at 7 days (c, d), and then highly elongated at 14 days (e, f). At 21 days (g, h), cells proliferated extensively to reach near confluence. Collagen core part in the core–shell carrier appeared to shrink substantially with increasing culture time. Scale bars=350 μm. Color images available online at www.liebertpub.com/tea

The cellular cytoskeletal process and elongation morphology were revealed after immunofluorescence staining of the cells and observation under confocal microscopy, as shown in Figure 6. MSCs with limited cytoskeletal extensions at day 7 (Fig. 6a, b) showed pronounced elongation, forming a highly networked structure among the cells at day 14 (Fig. 6c, d). The cell networks appeared to be formed primarily at the outer surface of the core part, that is, close to the core–shell boundary. Once cell confluence at the surface region was reached, cells grew and proliferated at the inner region of the collagen core part, and at day 21, cells formed more dense networks throughout the collagen core, which was highly oriented along the fiber axis (Fig. 6e, f).

OPEN IN VIEWER

FIG. 6.

Confocal laser scanning microscopy showing the cellular cytoskeletal processes and the elongation morphology within the core–shell fibrous carrier after 7 (a, b), 14 (c, d), and 21 days (e, f). Cell nucleus stained in blue (4′,6-diamidino-2-phenylindole) and the cytoskeleton in red (Alexa Fluor Phalloidin 548). Scale bars=850 μm (a, c, and e) and 350 μm (b, d, and f). Color images available online at www.liebertpub.com/tea

The osteogenic differentiation of the MSCs cultured in the core–shell fibrous carrier was assessed in terms of expression of genes related with bone, including BSP, OPN, and OCN. Figure 7 shows the mRNA expression levels of those genes (expression levels presented by normalizing to collagen gel at 14 days) by qPCR analysis. After culture for 14 days, all genes were expressed at levels significantly higher in the core–shell carrier than in the collagen gel; fold increases were 1.6, 3.0, and 5.5, respectively, for BSP, OPN, and OCN. At 21 days, however, the difference was reduced.

OPEN IN VIEWER

FIG. 7.

Osteogenic differentiation of the MSCs cultured in the core–shell fibrous hydrogel carrier over 14 and 21 days. mRNA levels of bone sialoprotein (BSP) (a), osteopontin (OPN) (b), and osteocalcin (OCN) (c) genes were quantitatively assessed. * Indicates significant differences (p<0.05, by analysis of variance).

In vivo bone forming ability of the MSC delivery system

The tissue compatibility and bone forming ability of the MSC-encapsulated core–shell hydrogel carrier was further assessed using a rat calvarium model. Three different groups were studied: core–shell carrier only, carrier with MSCs as encapsulated, and carrier with MSC encapsulation followed by osteogenic differentiation for 7 days. Throughout the 6-week study period, animals showed a good healing response without adverse tissue reactions and material toxicity, as well as there were no visible signs of internal inflammation on the harvested tissues. Bone regeneration was examined with μCT, as shown in Figure 8. The μCT constructed images revealed new bone formation (orange colored) within the defect region for the three groups (Fig. 8a). Different levels of bone regeneration from the defect margins were noted. Based on the μCT images, the percent bone volume, bone surface area, and bone surface density were quantified. Whereas the percent bone volume of the carrier only was ∼13%, that of MSC-encapsulated carrier showed 25%, and the differentiated MSCs increased the bone volume as high as 45% (Fig. 8b). The bone surface area and bone surface density also showed the effectiveness of the carrier encapsulating MSCs with osteogenic differentiation (Fig. 8c, d).

OPEN IN VIEWER

FIG. 8.

In vivo bone forming ability of the MSC-encapsulated core–shell hydrogel carrier. Three sample groups tested include carrier only (w/o cells), carrier with undifferentiated MSCs (nonosteogenic), and carrier with differentiated MSCs for 7 days in an osteogenic medium (osteogenic). Bone regeneration was analyzed by microcomputed tomography (μCT) after 6 weeks of implantation: (a) μCT constructed images showing the new bone formation (orange color) around the 5-mm-diameter defect region. Quantitative results showing (b) percent of new bone volume, (c) bone surface area, and (d) bone surface density. Significant differences were noticed (*p<0.05 and **p<0.01, by analysis of variance). Color images available online at www.liebertpub.com/tea

New bone formation and defect closure were revealed again by histological views after HE and MT stains, as shown in Figure 9. All three groups showed good biocompatibility with no significant notice of inflammatory cells. HE stains revealed cell and/or tissue in-growth within the defect region. MT stains showed more clearly the newly formed bone within the defect area. The carrier only group showed primarily loose connective tissue formed within the defect with a limited amount of new bone formation at the margin (Fig. 9a). In contrast, when undifferentiated MSCs were encapsulated within the carrier, a level of new bone formation (indicated as NB) was revealed at the center of defect (Fig. 9b). The group using carrier with differentiated MSCs showed a significant level of newly formed bone at the central regions of the defect (Fig. 9c). At higher magnification of the newly formed bone in Figure 9c, the new bone was proven to be mature with a compact structure and lined with many osteoblasts (Fig. 9d).

OPEN IN VIEWER

FIG. 9.

Representative histological images of implanted samples after hematoxylin and eosin (HE) and Masson's trichrome (MT) staining after 6 weeks of implantation; (a) carrier only, (b) with undifferentiated MSCs, (c) with differentiated MSCs, and (d) enlarged image of newly formed bone in (c). (Arrows: defect margins, NB: new bone, HB: host bone). Scale bar=1 mm (a–c) and 200 μm (d). Color images available online at www.liebertpub.com/tea