Is macroporosity absolutely required for preliminary in vitro bone biomaterial study? A comparison between porous materials and flat materials.

Porous materials are highly preferred for bone tissue engineering due to space for blood vessel ingrowth, but this may introduce extra experimental variations because of the difficulty in precise control of porosity. In order to decide whether it is absolutely necessary to use porous materials in in vitro comparative osteogenesis study of materials with different chemistries, we carried out osteoinductivity study using C3H/10T1/2 cells, pluripotent mesenchymal stem cells (MSCs), on seven material types: hydroxyapatite (HA), α-tricalcium phosphate (α-TCP) and b-tricalcium phosphate (β-TCP) in both porous and dense forms and tissue culture plastic. For all materials under test, dense materials give higher alkaline phosphatase gene (Alp) expression compared with porous materials. In addition, the cell density effects on the 10T1/2 cells were assessed through alkaline phosphatase protein (ALP) enzymatic assay. The ALP expression was higher for higher initial cell plating density and this explains the greater osteoinductivity of dense materials compared with porous materials for in vitro study as porous materials would have higher surface area. On the other hand, the same trend of Alp mRNA level (HA > β-TCP > α-TCP) was observed for both porous and dense materials, validating the use of dense flat materials for comparative study of materials with different chemistries for more reliable comparison when well-defined porous materials are not available. The avoidance of porosity variation would probably facilitate more reproducible results. This study does not suggest porosity is not required for experiments related to bone regeneration application, but emphasizes that there is often a tradeoff between higher clinical relevance, and less variation in a less complex set up, which facilitates a statistically significant conclusion. Technically, we also show that the base of normalization for ALP activity may influence the conclusion and there may be ALP activity from serum, necessitating the inclusion of "no cell" control in ALP activity assay with materials. These explain the opposite conclusions drawn by different groups on the effect of porosity.

Introduction

It has been widely accepted that the porosity or pore size of materials influence the cell behavior and also the osteogenesis process [1]. A number of studies have demonstrated this both in vivo (Table A1 in the Appendix) and in vitro (Table A2 in the Appendix). Blood vessels are the major transport means in animals and they supply the nutrients for the growth of bone. Porous materials with interconnected pores provide the space for the ingrowth of blood vessels and hence allow better formation of bone in the implants. This is one of the major reasons for the necessity of porosity for bone formation in animal models (in vivo). Continuing intensive efforts have been made to prepare and characterize highly porous scaffolds for bone substitute applications [2,3,4,5,6,7,8,9,10,11,12,13] and also assess cellular behavior on these materials [4,7,10,12,14].

Table A1

In vivo cell response for cells seeded on 3D scaffolds with different porosities or pore sizes from literatures.

Material a	Implant site	Porosity (%)	Macropore size (μm)	Specific surface area (m2/g)	Assessments b	Conclusion a,b	Reference
BCP	In femurs of rabbits	40, 50%	300, 565	/	Histology	Macroporous biphasic calcium phopshate implants with 565 μm pore diameter and 40% macroporosity represented the optimal association for homogeneous and abundant bone ingrowth.	[62]
BCP	Intramuscularly in goats	macroporosity: 58%; microporosity: 4–24%	<∼400	0.2–9.7	Histology and histomorphometry	The increased specific surface area led to more surface reactivity, which is hypothesized to be essential for osteoinductivity by biomaterials.	[63]
BCP	Intramuscularly in goats	70, 75%	/	0.2, 1.5	Histology and histomorphometry	The BCP scaffold with 75% porosity performed better than the one with 70% porosity in both the amount and rate of bone formation.	[64]
β-TCP	In drill hole defects in cancellous bone of sheep	/	150, 260, 510, 1220	/	Macroscopy, histology and histomorphometry	Samples with an intermediate macropore size (510 μm) were resorbed significantly faster than samples with smaller macropore sizes (150 and 260 μm). However, this fast resorption was associated with a lower bone content and a higher soft tissue content.	[65]
β-TCP	Subcutaneously in mice	25, 65, 75%	/	0.12, 0.18, 0.3	Specific ALP activity, histology	A higher porosity of TCP scaffolds does not necessarily mean a higher ALP activity in vivo. The distribution and size of the pores, as well as the surface structure, might play an important role for osteogenic differentiation in vivo.	[66]
HA	Intramuscularly in goats	60, 70%	400, 800	/	Histology and histomorphometry	The microporous HA scaffolds with 70% porosity and 800 μm pore size yielded more bone than did HA scaffolds with 60% porosity and 400 μm.	[67]
HA	Subcutaneously in mice	65, 80%	/	0.87, 1.63	Histology	The porosity and pore interconnection of osteoconductive scaffolds can influence the overall amount of bone deposition, the pattern of blood vessels invasion and finally the kinetics of the bone neoformation process.	[68]
HA	Subcutaneously in rats	30, 50, 70%	/	/	Histology, ALP acitivity and OCN content	Expanded bone formation was observed earlier for constructs with higher porosity. The ALP activity and OCN production also increased with increasing porosity.	[34]
HA	Subcutaneously in rats	70%	106–600	0.024	Histology, ALP acitivity and OCN content	The ALP activities at 2 wk and the OCN contents at 4 wk after implantation was the highest in the ceramics implants with pore size of 300-400 μm.	[69]
HA	Subcutaneously in rats	Tunnels of 0.7mm or 3mm in diameter	/	/	Histology, Ca and Type II collagen content	The “vasculature-inducing geometry” of the carrier as an extracellular matrix is crucially important for osteogenesis.	[70]
HA, BCP	Intramuscularly in goats	46.5–54.3%	243.9–380.2	0.07–1.60	Histology and histomorphometry	Histomorphometrical results showed that the presence of micropores within macropore walls is necessary to make a material osteoinductive.	[71]
HA, TCP	In cancellous bone of rabbits	60%	50–100 and 200–400	/	Histology	For TCP, bone and tissue ingrowth and implant resorption occurred at a higher rate in the smaller-pored materials compared with the larger-pored ones. For HA, the smaller-pored materials was totally infiltrated by bone or bone marrow after four months but not for the larger-pored ones.	[72]
Ni-Ti alloy	In femurs of rats	46.6, 59.2, 66.1	259, 272, 505	/	Histology and histomorphometry	The porosity of 66.1% showed best bone contact (51%) of the porosities tested.	[73]
PCL	Subcutaneously in mice	/	350, 550, 800	/	Histology, μCT and mechanical testing	Pore sizes between 350 and 800 μm play a limited role in bone regeneration in this tissue engineering model.	[74]
PPF	In cranial defects in rabbits	57–75%	300–500, 600–800	/	Light microscopy, histological scoring analysis and histomorphometric analysis	Scaffold porosity and scaffold pore size were not found to significantly affect the observed tissue response.	[75]
Ti6Al4V	In femurs of rabbits	/	100, 200, 300	/	Histology and histomorphometry	200 μm may be the optimal pore size for laser-textured Ti6Al4V implants.	[76]
Ti6Al4V	In femurs of rats	3, 11, 25%	/	/	Ca concentration	25% porosity samples showed the highest amount of calcium concentration within the pores, suggesting a faster rate of tissue generation and integration compared to samples with lower pore volume.	[77]

BCP: biphasic calcium phosphate, β-TCP: β-tricalcium phosphate, HA: hydroxyapatite, PCL: polycaprolactone, PFF: poly-(propylene fumarate);

ALP: alkaline phosphatase, OCN: osteocalcin, μCT: micro-computed tomography.

Table A2

In vitro cell response for cells seeded on 3D scaffolds with different porosities or pore sizes from literatures.

Material a	Cell	Porosity (%)	Macropore size (μm)	Specific surface area (m2/g)	Assessments b	Conclusion a,b	Reference
β-TCP	Human mesenchymal stem cells	25, 65, 75%	/	0.12, 0.18, 0.3	Protein production, specific ALP activity	In vitro porosity was beneficial for protein production, but did not influence osteogenic differentiation.	[66]
CO3Ap-collagen sponges	MC3T3-E1 cells	48.9, 72.6, 79.2%	50–300	/	Histology	There was no significant difference in the invading osteoblast quantity for composite of apatite and collagen with pores ranging from 50–300 μm and porosities of 49–79%.	[78]
COL-CG	MC3T3-E1 cells	/	85–325	0.00221–0.00845 μm−1	Cell adhesion and infiltration	Scaffolds with a mean pore size of 325 μm were deemed optimal for bone tissue engineering.	[79]
HA	Rat primary bone marrow mesenchymal stem cells	30, 50, 70%	/	/	ALP acitivity and OCN content	The ALP activity and OCN production increased with increasing porosity.	[34]
HA	Primary human osteoblast-like cells	67-76%	/	/	SEM	The introduction of microporosity has no evident effect on cellular morphology at later time points but it seems to play a role in initial cellular anchorage and attachment.	[80]
HA	Rat bone marrow cells	<5, 15, 30%	/	/	Cell attachment, proliferation, total protein content, ALP activity and bone-like nodule formation	The intermediary and final events such as proliferation, protein synthesis, ALP activity, and bone-like nodule formation favored surfaces with a more regular topography, such as that presents in HA with 15% or less of microporosity.	[81]
HA	MC3T3-E1 cells, L132 cells	0.2, 18.3, 25, 80%	1–10, 10–50, 500–600	/	Cell proliferation, viability, SEM, cytochemical staining	The micro-porous HA (internal pore size of 1–10 μm) induced the highest cell growth.	[82]
HA	Autologous human mesenchymal stem cells	75, 88%	200, 500	/	DNA content, ALP activity, histology, SEM and RT-qPCR	The 200-μm pore scaffolds exhibited a faster rate of osteogenic differentiation than did the 500-μm pore scaffolds.	[53]
PCL	Rat marrow stromal cells	84-89%	20–45	/	Cell attachment, spreading and histology	Increasing the thickness of the nanofiber layer (also increasing the pore size) reduced the infiltration of cells into the scaffolds.	[83]
PET	Mesenchymal stem cells	92.9-97.0%	/	/	Cell proliferation, ALP activity, OCN content and SEM	The attachment, proliferation and bone differentiation of MSC was influenced by the fiber diameter and porosity of non-woven fabrics as the scaffold.	[33]
TiO2	Human bone-derived cells	/	0.4, 13, 49	/	Cell proliferation	Smaller pores (0.4 and 13 μm) in TiO2 films enhanced the proliferation of human cells trypsinised from bone in contrast to larger pores (49 μm).	[84]

β-TCP: β-tricalcium phosphate, COL-CG: collagen-glycosaminoglycan, HA: hydroxyapatite, PCL: polycaprolactone, PET: polyethylene terephthalate;

ALP: alkaline phosphatase, OCN: osteocalcin, SEM: scanning electron microscopy, RT-qPCR: reverse transcription quantitative polymerase chain reaction.

Apart from different structures, novel materials of different chemistries are also being developed to have better biocompatibility, osteoconductivity, osteoinductivity, biodegrdation, etc. Ideally, it seems it is the best to compare highly porous materials of different chemistries with exactly the same structural properties such as porosity and pore size. Nevertheless, the precise control of porosity of porous ceramics with high interconnectivity was a challenge [15]. A number of recent studies reported on fabrication technologies to synthesize highly porous scaffold of relatively well defined structures [16,17,18,19,20,21] but most require a certain extent of optimization for each material type. For example, there is significant shrinking of the printed scaffolds during the sintering process. Thus the change in dimensions needs to be precalculated by computer assisted design (CAD) [17]. Besides, these methods require more expertise and equipment compared with dense flat materials which can be prepared relatively simply by pressing the powder together by hydraulic press. Thus it is essential to know whether porosity is absolutely necessary in in vitro studies, especially for comparative preliminary study of materials with new chemistries.

To address this question, we compared the osteoinductivity of dense materials with porous materials using C3H/10T1/2 cells. Osteoinductivity, which is the ability of a material to induce undifferentiated cells into the osteo-lineage [15,22], is an important property for bone substitute materials. Also totipotent, pluripotent or multipotent cells are required to study this property. Adult stem cells, such as mesenchymal stem cell (MSC) model is a good candidate due to its multipotency, relative ease of maintenance and lack of ethical concerns [23,24,25,26]. We used C3H/10T1/2 cells as they possess the properties of MSCs yet it is an established cell line that may provide more reproducible results as compared to commonly used primary stem cells as its purity is difficult to be controlled precisely in different batches of experiments.

On the other hand, Kaplan et al. suggested that higher porosity and pore size result in greater bone ingrowth in vivo and lower porosity stimulates osteogenesis by suppressing cell proliferation and forcing cell aggregation in vitro [1]. In order to evaluate the effect of cell density on the differentiation of cells in vitro, we carried out Bradford protein assay and alkaline phosphatase enzymatic assay for C3H/10T1/2 cells plated with different cell densities.

In short, we compared the Alp gene expression of 10T1/2 cells cultured on porous and dense ceramics (hydroxyapatite, HA, α-TCP and β-TCP) and tissue culture plastic. Furthermore, we assessed the effect of cell plating density on ALP expression. The reasons for the different conclusions drawn by different groups were also investigated.

Experimental Section

Material Synthesis

HA, α-TCP and β-TCP were synthesized by foaming method using 5% hydrogen peroxide (H2O2) solution [27]. The dried calcium phosphate cylinders were sintered using a programmed oven and then cut using a diamond saw using water as the coolant. The discs have a diameter of 10 mm and a thickness of 1 mm. The average porosity estimated from the Archimedes drainage method was about 71–77%. The dense flat ceramics were prepared by pressing powder together in a mold with 15 mm inner diameter using a hydraulic press and sintered according to the temperature profile in the Table A3 in the Appendix. The sintered discs were trimmed into 10 mm discs by cutting the periphery to obtain discs of the same size.

Table A3

Sintering temperature profile for the calcium phosphate ceramics.

HA		α-TCP		β-TCP
Temp. (°C)	Duration (min)	Temp. (°C)	Duration (min)	Temp. (°C)	Duration (min)
0–400	160	0–400	160	0–400	160
400	120	400	120	400	120
400–800	160	400–800	160	400–800	160
800	240	800	240	800	240
800–1100	120	800–1250	180	800–1,100	120
1100	120	1250	360	1,100–900	240
normal cooling		quick cooling		900	360
				normal cooling

Material Characterization and Sterilization

The phases of different material surfaces were identified using an X-ray diffractometer (PW1830, Philips). Measurements were performed using a Cu-Kα X-ray source and the XRD spectrum matched with the database using X'Pert HighScore Plus version 2.0. SEM images were captured using scanning electron microscope (SEM) (JSM-6390, JEOL, Japan). The surface porosities were estimated from the SEM photos using the software OriginPro (details were shown in Section 1 of the Appendix) and the surface crater sizes were measured in Photoshop. The discs of materials were sterilized by autoclaving at 121 °C for 20 min and dried in an 80 °C oven.

Cell Subculturing

C3H/10T1/2 cells (ATCC, CCL-226™) were cultured in 10cm culture dishes (Nunc) containing Dulbecco's Modified Eagle Medium, DMEM (Invitrogen) supplemented with 10% fetal bovine serum (FBS) and antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin) in an incubator at 37 °C under a humidified atmosphere of 95% air and 5% CO2. At 95% confluence, the cells were trypsinized, resuspended in complete medium and counted in a hemocytometer. Then the cells were plated for the assays and medium changed at every three days.

RT-qPCR with Cells from Porous Materials and Flat Materials

9 × 104 cells were seeded on each disc in 48-well plates (Nunc). Wells without materials added (plastic surfaces) were used as controls. RNA from the cells on the material was extracted by TRIzol at day 6 according to the manufacturer's instruction. For cells on α-TCP and β-TCP, modifications were made to the RNA extraction protocol as described previously [28]. Crushing the porous materials in TRIzol was performed for the effective extraction of RNA from cells in the pores of porous material while this is not necessary for dense materials since the cells cannot grow into the dense materials. The subsequent RNA quality check, yield estimation, DNase1 treatment, reverse transcription and qPCR on Alp and β-Actin (the house-keeping gene) were also performed as described [28]. (In this report, “ Alp ” here denotes the gene that codes for alkaline phosphatase in mouse and “ALP” denotes the alkaline phosphatase protein according to the gene nomenclature convention adopted in the biology field. On the other hand, the alkaline phosphatase gene in human is generally denoted by “ALP”).

ALP Enzymatic Assay with Different Cell Plating Densities

Different numbers of cells (1 × 104, 2 × 104 and 4 × 104) were plated in a 96-well cell culture plate (Nunc) at different time so that the assays would be performed with cells after 2 and 7 days of culturing for ALP enzymatic assay and 2, 5, and 7 days of culturing for Bradford protein assay. Cells were washed twice with PBS and lysed with 0.2% Triton X-100 solution. The lysate was mixed with 1 mg/mL p-nitrophenyl phosphate (p-NPP, Sigma) in 0.1 M glycine buffer (pH 10.4) containing 1 mM MgCl2, 1 mM ZnCl2. Absorbance at 405nm was measured after incubation in dark and the amount of p-nitrophenol (p-NP, Sigma) formed was calculated based on a standard calibration curve of p-NP. Bradford assay was also performed using the same lysate and the absorbance value was read at 595 nm. Protein amount was calculated based on a standard calibration curve of bovine serum albumin (BSA). Phase contrast photos of cells were captured the day before the assays.

SEM of Cells on Scaffolds

Cells on porous HA were prepared for scanning electron microscope (SEM) examination using procedures described previously [29]. In brief, C3H/10T1/2 cells were plated on porous HA at 9 × 104 cells per cm2. After 1 week, the porous scaffolds with cells were blotted dry with absorbent papers and washed with phosphate buffered saline (PBS), pH 7.4. They were then fixed with 2.5% glutaraldehyde in PBS, pH 7.4 (for resorbable calcium phosphate, 0.05 M cacodylate buffer, pH 7.4, should be used instead). The scaffolds were rinsed twice with PBS and then with double deionised water (DDI) with subsequent quick freeze in liquid nitrogen and transferred to a freezer dryer for overnight drying. Scanning electron microscope (SEM) (JSM-6390, JEOL, Japan) was utilized to observe the morphology of cells on the porous HA.

DAPI Staining

9 × 104 cells were seeded on each porous HA disc in 48-well plates (Nunc). At day 3, the discs were washed with PBS and then fixed by 4% paraformaldehyde for 5 min at 4 °C. The permeability of cells on the discs was increased by washing 3 times in 0.1% Triton X-100. 4′,6-diamidino-2-phenylindole (DAPI) stain was applied to the discs at 1 μg/mL in 0.1% Triton X-100 for 5 min, followed by washing with Triton X-100. Fluorescence images were taken using Olympus upright fluorescence microscope BX41. 5 to 15 images were taken for each spot due to the limitation of the depth of field of the microscope and cells lying at different depth. The photos at different depth were merged using ImageJ with the “Extended depth of field” plug-in.

Statistical Analysis

P-values were calculated using Student's two-tailed t-test of unequal variance.

Results

Material Characterization

The identity of the materials was confirmed by XRD and morphology was viewed by SEM. Figure 1 shows the XRD spectra of the chosen ceramics. The peaks of the spectra match the most intense peaks in the database and the identity of materials was confirmed. There is no significant difference between the spectra of materials in dense forms and porous forms. The porous materials used are highly porous with considerable degree of interconnections among the pores while the dense materials have flat surfaces as shown in Figure 2. From Table 1, the surface porosities of the three materials are similar but the surface crater sizes are different with HA having the smallest craters while α-TCP the largest. The porous HA, α-TCP and β-TCP materials in this study were made by the same H2O2 foaming method. Yet there is a difference in pore morphology, which resulted from different degrees of shrinkage during the sintering process. To obtain precisely controlled pore sizes, it is necessary to carry out optimization experiments to estimate the shrinkage ratio during the sintering process and also trialed with different synthesis parameters.

Figure 1

XRD spectra of porous and dense: (a) hydroxyapatite, HA; (b) α-TCP and (c) β-TCP discs; and (d) the corresponding matches from the database (the spots on the spectrum indicate the positions of the five most intense peaks from the powder X-ray data of the database).

Figure 2

Surface topography of porous materials and dense materials (HA, α-TCP and β-TCP) revealed by SEM at 30× and 5000×.

Table 1

Surface porosity and crater size of the porous HA, α-TCP and β-TCP (average ± SD).

	Porous HA	Porous α-TCP	Porous β-TCP
Surface porosity a (%)	67.3 ± 1.7	68.1 ± 5.8	66.5 ± 1.5
Crater size b (μm)	203 ± 88	342 ± 134	269 ± 111

n = 4 photos;

n = 60 measurements.

Yield of RNA from Cells on Materials

From Table 2, the amount of RNA extracted from cells on the porous discs is significantly higher than that from dense discs of the same cross sectional area (p < 0.005). This indicates a larger number of cells are present on the porous materials compared with the dense materials. As the same number of cells was plated initially and assuming the dense materials do not induce much higher death rate, there was a larger proportion of cells attached on the porous materials initially and/ or higher proliferation of cells on the porous materials compared with dense materials.

Table 2

Yield of RNA (μg) extracted from cells on porous calcium phosphate (CaP) materials, dense flat CaP materials and flat tissue culture plastic.

	Porous CaP a	Flat CaP a	Flat plastic b
Average	3.47	1.80	2.63
SD	0.89	0.17	0.39

n = 6 with 2 discs from HA,α-TCP and β-TCP each;

n = 2.

Alp Gene Expression

As shown in Figure 3, the Alp mRNA level of 10T1/2 cells was highest when cultured on HA scaffolds. For dense flat materials, the Alp mRNA level was slightly higher for cells on β-TCP than α-TCP while for porous materials, the cells on β-TCP has much higher Alp expression than α-TCP but not statistically significant. Cells on all calcium phosphate materials under test have lower Alp expression compared with cells on flat tissue culture plastics. There were consistent higher Alp mRNA levels for cells on dense flat materials compared with porous materials (p < 0.01 for HA and α-TCP and p < 0.1 for β-TCP).

Figure 3

Relative quantification of gene expression of Alp normalized with β-Actin of 10T1/2 cells seeded on HA, α-TCP, β-TCP and plastic after 6 days of culture (n = 5, technical duplicate in batch 1 and triplicate in batch 2; mean ± SEM; ** p < 0.01, + p < 0.1).

Effect of Cell Density

In order to verify whether the alkaline phosphatase expression of 10T1/2 cells is affected by cell aggregation, experiments with cells plated on plastic at different densities were performed. From Figure 4a, the ALP activity of cells increases from day 2 to day 7, indicating the cells become more differentiated. The ALP activity is higher for cells plated at higher initial plating density after both 2 days and 7 days of culturing. Higher initial plating densities also gave higher Alp mRNA levels (data not shown). From Figure 4b, the protein amount in cells increases with time for low plating density but does not significantly change for high plating density. At day 2, the protein amount is larger for cells plated at higher density. Figure 4c shows the photos of cells captured a day before the assay. The cells increase in number for all initial plating densities from day 1 to day 6. Our results are consistent with the phenomenon that proliferation precedes differentiation in general [30,31]. When the cells are plated at higher cell density, they have less room for proliferation (as shown in Figure 4c and deduced from Figure 4b) and would get into the differentiation program earlier (as shown by the higher ALP activity in Figure 4a).

Figure 4

Effect of cell plating density on ALP expression and total protein amount of 10T1/2 cells seeded on a plastic culture plate (a) ALP enzymatic activity estimated from p-NPP assay after 2 and 7 days of culture, with background subtraction using values obtained in the “no cell” control (n = 3 wells; mean ± SD; + p < 0.1, * p < 0.05, ** p < 0.005); (b) Protein amount after 2, 5 and 7 days of culture estimated from Bradford assay (n = 3 wells; mean ± SD; + p < 0.1, * p < 0.05, ** p < 0.005); (c) Phase contrast photos of 10T1/2 cells seeded on a 96-well plastic culture plate with different plating densities and photos captured the day before the assays (i.e., at day 1, 4 and 6) (scale bar =50 μm).

Distribution and Morphology of Cells in Porous Scaffolds

The cell density on the edges of the pores observed by DAPI staining is higher than the pore surface (Figure 5a) and this may be due to the rougher surface of the edges (Figure 5e and f compared with Figure 5c and d) caused during slicing of blocks into discs. From the SEM photos, the cells in the porous HA span across the pores after one week of culture and only the “outermost” layers of cells may be in direct contact with the material (Figure 5b). The gene expression may be different for cells directly attaching on the material compared to the spanning cells and the percentage of the cells that are directly touching the materials is probably different for different pore sizes and porosities. For most biochemical assays, they are monitoring the overall level of a cellular component of the cell mass instead of an individual cell. Therefore, difference in the actual distribution and arrangement of cell mass may be one of the reasons for the observed difference in Alp mRNA levels of cells seeded on scaffolds with different geometries but with the same chemistry.

Figure 5

(a) 10T1/2 cells viewed under a fluorescent microscope with DAPI staining the nuclei of cells cultured for 3 days; (b) SEM photos of 10T1/2 cells cultured on porous HA for 1 week; SEM photos of (c) pore surface at 1,000×; (d) pore surface at 10,000×; (e) pore edges at 1,000× and (f) pore edges at 10,000×.

Discussion

Effect of Macroporosity of Calcium Phosphate Discs

A number of studies have investigated the effect of porosity on bone growth in vivo and cell proliferation in vitro, but few studies have focused on the osteoinductivity in vitro. Thus this study focused on osteoinductivity of three commonly used calcium phosphate materials: HA, α-TCP and β-TCP. In order to have the largest difference, we compared the extremes: dense materials without macroporosity and highly porous materials prepared by foaming. Among several possible osteomarkers, we chose to assess the Alp mRNA level as Alp expression increased with time but was consistent at different time points in our separate experiments [32]. The results in this study show that the cells on dense flat materials have higher Alp expression compared with porous materials for HA, α-TCP and β-TCP (Figure 3). This is consistent with the conclusion drawn by Kaplan et al. from other materials that lower porosity stimulates osteogenesis by suppressing cell proliferation and forcing cell aggregation in in vitro systems [1]. Tabata et al. demonstrated with non-woven fabrics that osteogenesis in vitro is enhanced by lower porosity [33]. Although porous materials are highly preferred in bone substitute application and cellular behavior on these materials were intensively studied [4,7,10,12,14], our results suggest that high porosity may not be absolutely required for in vitro osteogenesis assay. As variations in porosity of materials may affect the response of cells on them, dense flat materials may be used for preliminary in vitro osteoinductivity study, especially when it is difficult to have precise control of material porosity for a particular synthesis method. In Figure 3, the consistent trend of Alp mRNA level (HA > β-TCP > α-TCP) between porous materials and dense materials also justifies the use of dense materials in preliminary study even though highly porous materials have been used in many studies due to in vivo consideration.

On the other hand, the difference of cell response in 2D and 3D settings has been extensively studied using different models with examples tabulated in Table A4 in the Appendix. It is well accepted that cells in monolayer behave differently compared with 3D cultures or on 3D scaffolds. As shown in Table A4 of the Appendix, 3D culturing can influence the cell attachment, viability, proliferation, gene expression, differentiation, cellular response to chemicals, biomolecule synthesize and maturation. Nevertheless, in most studies, the cells were cultured on substrates with different chemistries in the 2D and 3D settings or the materials on which the cells were cultured were not mentioned explicitly. In this report, we compared the Alp expression of cells on planar surfaces and 3D scaffolds of the same substrate material chemistry.

Figure 7

Schematic showing the relation between porosity, surface area and cell density.

ALP Activity Assay

Though our result agrees with the postulation that lower porosity enhances osteogesis observed for other materials [1], it is opposite to the conclusion of a study using HA with different porosities [34]. In our study, we measured the Alp mRNA level normalized with β-Actin mRNA level while the previous study assessed the ALP activity per disc.

We did not draw a conclusion from ALP activity assay in this report since there was high ALP activity for calcium phosphate samples even for “no cell” control (all conditions were the same except that cells were not added and supposed not to give ALP activity) (Figure 6). Besides this, we investigated whether cell culture medium may contribute to the detected ALP activity with results further discussed in Section 2 of the Appendix. ALP enzymatic assay is widely used in the biomaterial field and this study does not suggest the enzymatic assay cannot be used for biomaterial study. In contrast, we would like to emphasize whether this assay is applicable for a certain type of material depends on the relative background ALP activity because of ALP adsorption from serum, which can be estimated through the use of “no cell” control. This background signal is influenced by various factors such as the material affinity for ALP, the material geometry (e.g., higher background when cell culture medium with serum is easily trapped inside the inner pores and not easily removed during the washing step) and also the pre-treatment before cell plating (e.g., fibronection coating which can enhance cell adhesion and may block ALP adsorption at the same time which would reduce the background). The results are more reliable when the ALP activity in the “no cell” control is much lower than the samples under study.

Figure 6

ALP activity of cells on materials detected using ALP enzymatic assay (9 × 104 cells plated per well of a 48-well plate, assayed after 3 days of culture; n = 2; The experiment was repeated and the high background ALP activity detected for the “no cell” control of porous materials was consistently observed in the repeated experiments).

On the other hand, different numbers of cells can be present on materials of different porosities in the same volume of disc. Thus we hypothesized that the difference in the base of normalization may be one of the reasons for the contrary conclusions. As shown in Table 3, the ALP activity increases with increasing plating density when normalized by protein amount or area covered by the cells but decreases with increasing plating density when normalized by the initial cell number. This shows that the choice of normalization base would affect the conclusions drawn and this is further discussed in Section 3 of the Appendix.

Table 3

Comparison of ALP activity of 10T1/2 cells cultured in a 96-well plate for 7 days and normalized by different attributes.

Number of cells plated per well of a 96-well plate	10,000	20,000	40,000
Cell plating density (thousand cells/sq cm)	31.3	62.5	125
p-NP formed (nmol)	8.92(±0.26)	12.06(±1.10)	16.08(±1.10)
Normalized by initial cell number (nmol/thousand cells)	0.89(±0.03)	0.60(±0.05)	0.40(±0.03)
Normailzed by protein (nmol/mg protein)	10.91(±1.66)	14.68(±1.53)	21.25(±2.26)
Normalized by area covered (nmol/sq cm)	27.87(±0.81)	37.68(±3.43)	50.25(±3.43)

Tradeoff between Higher Clinical Relevance and Lower Variation

For the final clinical application, it seems the best to simulate the in vivo environment in human as closely as possible even for cell culture system. This would imply porous scaffolds should be used in experiments for bone substitute application since porous materials allow blood vessels to grow into the materials which are essential for bone formation in large bone substitutes. Numerous studies have demonstrated successful bone formation in their porous materials to show the novel materials are suitable for bone substitute application. However, in order to show the new material is better than the widely used materials, it is essential to compare with a common reference material. (The use of a reference material or control was discussed in our previous report [32]). Proper scientific control is also required. In other words, it would involve the comparison of cell responses to at least two different materials. To study new material's chemistry or microstructure, the material geometry and structure at macro-scale should be as close as possible for all the materials under study.

Materials with different porosities may have differences in surface area and cell densities (Figure 7). Considering the top surface alone, the surface area of a highly porous material can be 0.92 times larger than flat material, causing considerable difference in cell density. The difference would be even greater if inner pores were also taken into account. From Figure 4, difference in initial cell density can result in significant difference in osteo-marker expression of 10T1/2 mesenchymal stem cells. Therefore, comparable porosity is required for reliable conclusion. Besides, this variation in porosity in replicate samples would cause variation in the gene expression of cells cultured on them, thus may hinder statistically significant conclusions to be drawn. This study validates the use of dense flat materials for preliminary study when precisely defined porous structures are not easily available.

On the other hand, the number of cells to be seeded on the scaffolds may be adjusted in consideration of the actual available surface on the material for cell attachment and growth. For instance, more cells might be seeded on materials with larger surface area to minimize the variations introduced because of the cell density difference. Yet as mentioned in the result section, the actual distribution and arrangement of cell mass may affect the observed difference in Alp mRNA levels. Besides, it may also be technically difficult to adjust the number of cells to be seeded accurately since the surface of different discs may vary even if they are synthesized in the same batch. Hence, it may be possible to reduce the effect of cell density difference by adjusting the initial cell seeding density based on the surface area of materials, but this is not likely to eliminate the problem completely.

Minimizing the variations and improving the reproducibility of experiments are very important in preliminary in vitro study for accurate and reliable conclusions. This lays the foundation regarding the choice of materials for further in vivo experiments which are costly and with limited number of samples that can be handled due to cost and ethical issues. In this study, we have investigated three different materials of single phase (HA, α-TCP, β-TCP). In the actual application, biphasic or even triphasic materials are under intensive research. Besides, new materials with a slight difference in chemical compositions such as metal substitution of calcium in the hydroxyapatite are also developed to optimize the bone formation. Thus a continuum of materials instead of single phase materials may need to be compared and the difference among materials may not be as great as those among materials of single phase. This necessitates the stringent minimization of variations due to geometry, porosity, etc. Hence, although porous materials seem to be better in term of clinical relevance, it may not be always desirable to use porous materials in preliminary in vitro assays when the scaffolds geometry cannot be controlled precisely among the materials under study (Figure 8).

Figure 8

Factors involved in deciding scaffold type.

Conclusions

In contrast to many studies comparing cell behavior of cells on 3D materials with cell monolayer on plastic surface, this study compares cells on flat materials with 3D porous materials of the same chemistry. More cells are present on the porous materials compared with dense flat materials when the same number of cells was initially plated. Flat materials give higher osteo-marker expression of cells than porous materials as proliferation is suppressed and differentiation enhanced, which is supported by the higher ALP expression of cells with higher initial cell plating density. On the other hand, porous materials have a larger surface area which would result in lower initial cell density compared with dense flat materials. Hence, for reliable comparison, it is necessary to use materials of comparable porosities in evaluating materials of new chemistries.

Despite the significant difference in the Alp expression between porous and flat materials, a consistent trend of Alp mRNA level (HA > β-TCP > α-TCP) was observed for both porous and flat materials. This suggests that flat materials could be used for preliminary comparative study of materials with different chemistries to avoid variations in the gene expression due to porosity difference. There is often a tradeoff between higher clinical relevance and less experimental variations in the experiments. Dense flat materials may allow quicker and more reliable early assessment of materials with new chemistries.

This study also suggests that the base of normalization for the ALP enzymatic assay may influence the conclusions drawn. Besides, “no cell” controls are required for reliable conclusions due to the possible ALP adsorption on the materials from the serum. These two technical aspects suggest possible explanations to the contrasting findings from different groups.

Table A4

Cell response in 3D culture compared with 2D from literatures.

Related cell type	Cell	2D material	3D material a	Comparison between 2D and 3D a	Reference
Cancer cell	C4-2B cells	Collagen-coated tissue culture plastic	Electrospun collagen membrane	The cells on electrospun substrates were more resistant to both antineoplastic agents, docetaxel, and camptothecin compared to the cells grown on standard collagen-coated tissue culture polystyrene.	[85]
Chondrocyte	Rat bone marrow cells	Monolayer	3D chitosan or chitosan/gelatin scaffolds	The chitosan scaffolds caused a reduction in alkaline phosphatase production and an increase in the collagen concentration indicating phenotypic changes in the cells after the addition of a chondrogenic medium.	[86]
Chondrocyte	Human articular chondrocytes (HAC)	Monolayer on culture plates	Type II collagen sponges	Three-dimensional expansion of HAC on the scaffolds, as compared with 2D expansion for the same number of doublings, better maintained the chondrocytic phenotype of the expanded cells mRNA) but did not enhance their accumulation of glycosaminoglycan (GAG) following chondrogenic culture. Besides, increasing the HAC seeding density in the scaffolds (from 25 × 103 to 66 × 103 cells/mm3) significantly improved chondrogenesis (up to 3.3-fold higher GAG accumulation and up to 9.3-fold higher type II collagen mRNA).	[87]
Chondrocyte	Autologous chondrocytes and mesenchymal stem cells (MSCs)	Monolayer	3D pellet or fibrin-sealant construct	There was a proliferative effect for MSCs exposed to PRP in monolayer culture and an increase in the expression of chondrogenic markers when cells are exposed to a 3D environment.	[88]
Chondrocyte	Bovine chondrocytes	Monolayer in flasks	3D Minusheet	The Minusheet cultures usually showed a markedly higher mRNA expression than monolayer cultures. Besides, the ratio of type-I to type-II collagen or aggrecan to type-I collagen remained higher in Minusheet 3D cultures than in monolayer cultures.	[89]
Endothelial cell	Primary human artery-derived fibroblasts and human umbilical vein endothelial cells	Monolayer cell sheets	Self-assembled cell-based microtissues	The microtissue has significant enhancement of ECM expression and maturation.	[90]
Fibroblast	Fibroblast cell line GD25b1	Monolayer	3D matrices derived mouse embryo sections or naturally deposited ECM	The relative content of unsaturated fatty acids, which serve as targets of oxidative attack, was observed to be higher in major phospholipids in plasma membranes of 3D cells.	[91]
Fibroblast	Human foreskin fibroblasts	Glass cover slips coated with different proteins or 3D matrix compressed to form localised 2D matrix	3D matrices derived either from detergent-extracted mouse embryo sections or naturally deposited 3D ECM	3D-matrix adhesions differ from focal and fibrillar adhesions characterized on 2D substrates in their content of α5β1 and αvβ3 integrins, paxillin, other cytoskeletal components, and tyrosine phosphorylation of focal adhesion kinase (FAK). Relative to 2D substrates, 3D-matrix interactions also display enhanced cell biological activities and narrowed integrin usage.	[92]
Fibrochondrocyte	Primary fibrochondrocytes	tailored biomimetic surface (C6S surface) on glass cover slip	tailored biomimetic surface on PLGA scaffolds	Human fibrochondrocyte redifferentiation was enhanced by hypoxia in the 3D cultures, independent of hypoxia inducible factor (HIF) transcriptional activity and was shown to potentially involve the transcriptional activation of Sox-9.	[93]
Hepatocyte	Primary rat hepatocytes	Collagen-coated polymeric substrates	3D polymeric scaffold coated with collagen	Albumin and urea assays demonstrated that hepatocytes cultured in the 3D scaffold maintained higher levels of liver specific function over a period of 6 days as compared to the monolayer control. These results may be attributed to the high local concentration of soluble factors within the scaffold, which is important for maintaining the hepatocyte phenotype.	[94]
Hepatocyte	Hepatocyte cell line HepG2	Monolayer	Agar/gelatin sponge	The results showed that the agar–gelatin hybrid sponges induced the formation of 3D HepG2 spheroids with significant liver-specific functions. These spheroids exhibited higher amounts of albumin and urea synthesis than the control monolayer culture. These 3D spheroids were found to be more sensitive to the drug than the control monolayer.	[95]
Myoblast	Rat skeletal myoblasts	Culture plate	PLGA-collagen composite scaffolds	Tensile strain induces higher and faster integrin β1 and ILK expression in 3D cultured rat skeletal myoblasts than in 2D cultures.	[96]
Osteoblast	Rat bone marrow cells	Monolayer	3D homogenized pellet and 3D organotypic explant	The 3D organotypic marrow explant culture resulted in the greatest level of ossification with plate-like bone formations.	[97]
Osteoblast	Osteoblast-like cells MG-63	Monolayer on culture plates	Solid PLGA microspheres	The microspheres give stronger cell attachment, better phenotypic characteristics, higher cell viability and mineralization levels but lower cell count compared with 2D monolayer cell culture over 28 day cell culture studies.	[98]
Tenocyte	Human primary tenocytes	Monolayer	PLGA scaffolds or high-density cultures	Compared to native tendon, decorin and COMP were reduced in 2D and increased in 3D culture almost to ex vivo level.	[99]
Tenocyte	Rat dermal fibroblasts	Monolayer	3D resorbable polyester scaffolds	Direct comparisons between the 2D and 3D studies are difficult, given the differences in experimental variables (i.e. surface chemistry, topography, serum content, seeding density, etc.).	[100]

PRP: platelet-rich plasma, ECM: extracellular matrix, PLGA: poly (lactic-co-glycolic) acid, COMP: a typical non-collagenous protein component of tendon ECM.