Enhanced Ex Vivo Expansion of Human Hematopoietic Progenitors on Native and Spin Coated Acellular Matrices Prepared from Bone Marrow Stromal Cells

The extracellular microenvironment in bone marrow (BM) is known to regulate the growth and differentiation of hematopoietic stem and progenitor cells (HSPC). We have developed cell-free matrices from a BM stromal cell line (HS-5), which can be used as substrates either in native form or as tissue engineered coatings, for the enhanced ex vivo expansion of umbilical cord blood (UCB) derived HSPC. The physicochemical properties (surface roughness, thickness, and uniformity) of native and spin coated acellular matrices (ACM) were studied using scanning and atomic force microscopy (SEM and AFM). Lineage-specific expansion of HSPC, grown on these substrates, was evaluated by immunophenotypic (flow cytometry) and functional (colony forming) assays. Our results show that the most efficient expansion of lineage-specific HSPC occurred on spin coated ACM. Our method provides an improved protocol for ex vivo HSPC expansion and it offers a system to study the in vivo roles of specific molecules in the hematopoietic niche that influence HSPC expansion.


Introduction
Human stromal cells, residing in hematopoietic niches in the BM, provide cellular and molecular signals that are essential for the regulation of hematopoiesis [1]. Many tissue engineered biological and nonbiological scaffolds have been used for ex vivo expansion of HSPC. Bioengineered 2D and 3D scaffolds, which mimic the in vivo functional properties of hematopoietic niches, are gaining importance in the research and clinical trials [2]. These scaffolds mainly comprise ACM that are prepared by decellularizing either the whole tissue [3], organ, or cultured stromal cells [4]. The ACM comprises specific extracellular matrix (ECM) components, such as fibrous and nonfibrous collagen, fibronectin or laminin, glycosaminoglycans (GAGs), proteoglycans, growth factors, and cytokines [5]. These components individually or collectively provide biological cues that regulate proliferation, differentiation, and survival of stem and progenitors cells of different types in vivo [6]. Some cell types that have been expanded on ACM-based substrates include mesenchymal stem cells (MSCs) [7], neuronal cells [8], osteocytes [9], and embryonic stem cell-derived hepatic cells [10].
A recent study has shown that substrates containing a cocktail of fibronectin, heparin sulphate, tropocollagen I, hyaluronic acid, and cofibrils of collagen I bound to heparin or hyaluronic acid can be used for the adhesion, proliferation, and differentiation of human HSPC [11]. A few earlier studies have used these matrix components individually for the same purpose [12,13]. However, utilization of the entire ACM as substrates or scaffolds for ex vivo lineage-specific HSPC expansion has been limited.
We have previously shown that ACM, prepared from a mouse BM stromal cell line , can mimic the endosteal and/or vascular niche-like properties of human BM and support ex vivo expansion of UCB HSPC [4]. Our results showed that MS-5 cell line derived ACM could support a 2 Stem Cells International significant level (up to 80-fold) of common myeloid progenitor (CMP) cell expansion [4]. In order to develop a more clinically relevant nonxenogeneic model, we report here the potential of ACM generated by human BM stromal cell line (HS-5) on supporting HSPC expansion. HS-5 cells have a fibroblast-like morphology and secrete significant levels of various cytokines, such as granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage-CSF (GM-CSF), macrophage-CSF (M-CSF), Kit Ligand (KL), macrophageinhibitory protein-1 alpha, interleukin-6, interleukin-8, and interleukin-11 (IL-6, IL-8, and IL-11) into the culture media [14]. Intact HS-5 cells and their conditioned media have been shown to support the proliferation of HSPC in serumdeprived conditions [14]. For clinical use, it would be preferable to be able to manufacture a standardised ACM product, and we have therefore compared the effects of native HS-5 ACM to a sonicated and solubilised ACM preparation that can be spin-coated onto cell culture surfaces.

Materials and Methods
2.1. Preparation of CD34 + Cells. UCB samples were collected from full-term delivery with informed and signed consents following the local ethical guidelines of the institution. Each cord blood sample was collected, by the trained clinicians, into the blood collection bags containing 49 mL of citrate phosphate dextrose anticoagulant (CPDA). Standard protocols were followed for isolation of the target cell (CD34 + ) population from UCB for multilineage ex vivo expansion using the previously reported method [15]. Briefly, the UCB was processed by density gradient centrifugation using Ficoll Paque (GE Healthcare Bio-Sciences, Sweden) to obtain the total mononuclear cells (MNCs). CD34 + cells were isolated by using CD34 MicroBead Kit, in AutoMACS (Miltenyi Biotec Inc., Germany) by the magnetic assisted cell sorting (MACS) method. Observed cell enrichment was more than 90%. The percentage purity of the MACS-sorted cells was determined by a preexpansion flow-cytometry assay for CD34 + CD45 lo , CD34 + CD38 − , and CD34 + CD133 + cells surface markers that represent different levels of HSPC differentiation. The isolated UCB CD34 + cells were further cultured on ACM substrates. Phenotypic analysis of expanded cells was performed by flow cytometry and functional analysis by colony forming unit assays. GlutaMAX. Cultured cells were maintained in 5% CO 2 , 20% O 2 , and 90% humidity in standard cell culture incubators with regular media change. The cells were grown until they were 80-90% confluent before splitting to the next passage.

HS-5 Cells for HSPC Expansion.
HS-5 cells were maintained in complete growth medium as discussed above. After the cells reached 80% confluency, they were shifted to a media containing 1% FBS to limit the mitotic activity of the cells. These mitotically inactive cells were used as cellular control to expand UCB derived HSPC. All experiments were performed with cells maintained within 4-6 passages. Blank tissue culture plates (TCP) without any matrix or cells were used as negative control throughout all the expansion and matrix characterization experiments.

ACM Preparation
2.3.1. Native ACM. Native ACM was prepared by decellularization of HS-5 cells. Cells were maintained in complete growth medium for cellular control as described in Section 2.2.2. Before decellularization, cells were further grown for additional 4 days after confluency for better ECM deposition. All procedures for decellularization were done under sterile conditions as mentioned elsewhere [15] (summarised in Figure 1). Briefly, equal numbers of HS-5 cells grown on 24-well plates (Becton Dickinson Biosciences Labwares, USA) were washed with phosphate buffer saline (PBS) and then incubated in MilliQ water until the shape of cells became round. The cells were treated with 0.02 M NH 4 OH and the residual cellular matrices were retained on the surface. ACM was dried and washed with PBS to remove traces of NH 4 OH. The ACM obtained on 24-well plates were used as native ACM substrates.

Spin Coated ACM.
For spin coating, ACM suspensions obtained from 60 mm dishes were pooled and briefly sonicated and total protein content was estimated. The fine matrix suspension was immediately processed for spin coating the 24-well tissue culture plates (TCP) for expansion purpose. Based on the total estimated matrix protein in each well of 24-well plate (see Supplementary  Figure 1). The process was repeated three times to get maximum layering. The coated plates and coverslips were dried overnight in the laminar flow cabinet, washed in PBS with Pen/Strep, and used for ex vivo HSPC expansion and ACM characterization process. Each expansion experiment was done in freshly prepared and spin coated ACM substrate. The amount of protein coated and the parameters of spin coating were kept standard for each experiment. Step 1 Step 2 Step 3 Step 4 (c) Spin coating process: step 1-extraction and collection of acellular matrices, step 2-sample kept on tissue culture surfaces to be coated, which is held in vacuum on a stub, step 3-spinning at fixed rpm, and step 4-coating and drying under sterile conditions. microscope (Axioplan 2, Carl Zeiss, Germany). Image acquisition and processing was done using AxioVision software (Version Release 4.4, Carl Zeiss, Germany).

Scanning Electron Microscopy.
Morphological characterization of the native and spin coated ACM was performed by SEM. Both substrates were rinsed in PBS, fixed in 2.5% glutaraldehyde, washed and dehydrated in increasing concentrations of ethyl alcohol (50%, 60%, 70%, 80%, 90%, and 100%), and finally kept in acetone. The fixed and dehydrated samples were then subjected to critical point drying (CPD) and sputter coating with a thin layer of gold. The samples were examined at a range of 5-10 kV in the microscope (3400N, Hitachi, Japan).

Atomic Force Microscopy.
Freshly prepared native and spin coated ACM was dried in a dehumidified desiccator for 16 h and subjected to AFM measurements. AFM analysis was carried out in 200-300 kHz tapping modes using a multiprobe imaging microscope Multiview 4000 TM (Nanonics Imaging Ltd., Israel) equipped with a hybrid scanner. Images were collected at scanning frequencies of 5 kHz using glass tip cantilevers at 512 × 512 pixels resolution. Scan regions of 5 × 5 or 2 × 2 m 2 were used for data collection and data of multiple scan regions were gathered and averaged (as indicated in Section 3). Root-mean-square values were used to calculate average surface roughness. For calculation of the coating thickness, a scratch on the native or spin coated surface was made to expose the underlying plastic or glass surface and differences in the data values of the scratched and exposed surfaces were used to calculate the thickness of the matrix. For final image reconstruction and calculation of height variations, thickness, and surface roughness, software provided by the manufacturers (WSxM v4.3) was used.

Cytochemical Characterization.
Various cytochemical stains were used to stain the native and spin coated ACM for detection of major ECM components. The presence of collagenous matrix in the native and spin coated ACM was demonstrated by staining with Masson's Trichrome kit (Sigma, USA) as instructed by the manufacturer. Similarly, fibrillar collagen was detected by staining the native and spin coated ACM with Picro Sirius Red. Proteoglycans and ECM-bound GAGs were detected by staining with Safranin-O and Alcian Blue, respectively. Details of each of these staining protocols have been described in the supplementary information section.
2.6. Immunofluorescence Characterization. Immunofluorescence staining was performed to examine the expression of major ECM proteins in native and spin coated ACM. Samples were fixed in 4% formaldehyde for 20 min at RT followed by blocking in 3% BSA for 1 hr. The fixed native and spin coated ACM samples were incubated with the appropriate dilution of Phalloidin actin Alexa Flour-488 (Molecular Probes, Invitrogen, USA) in 1% BSA in PBS for 45 min in dark at RT. All the antibodies used in immunofluorescence studies were purchased from Sigma. Primary antibodies against fibronectin (1 : 500), laminin (1 : 200), and collagen I (1 : 500) were added to the samples. After an hour of incubation, samples were washed and incubated with appropriate anti-mouse isotype IgG-Alexa Flour-594 (1 : 500) and anti-rabbit IgG-Alexa Flour-594 (1 : 500) secondary antibodies (Molecular Probes, Invitrogen). After an hour of incubation, samples were washed with 1x PBS and mounted in Vectashield mounting medium containing 4 ,6-diamidino-2-phenylindole (DAPI). Images were captured using acoustooptical beam splitter confocal microscope (Leica Microsystems, Germany).

HSPC Expansion on HS-5 Derived ACM.
24-well TCPcoated with native or spin coated HS-5 ACM were used for expansion experiments. Serum-deprived intact HS-5 cells were used as a cellular control and normal TCP wells without any matrix or cells were used as negative controls. Magnetically sorted HSPC with >90% CD34 + cell enrichment was suspended in Stemline II serum free media (Sigma) containing 50 g/mL penicillin and streptomycin at a density of 1 × 10 4 viable cells/mL. The medium was supplemented with growth factors as indicated-thrombopoietin (TPO, 25 ng/mL), stem cell factor (SCF, 25 ng/mL), interleukin-6 (IL-6, 50 ng/mL), and Flt3 ligand (Flt3, 40 ng/mL) (Millipore, USA). 1 mL cell suspension was placed in each well of the 24well plate in triplicate for each substrate, that is, native ACM, spin coated ACM, TCP, and HS-5 cellular control. All cultures were placed in an incubator with an atmosphere of 5% O 2 , 5% CO 2 , and 95% humidity at 37 ∘ C for 8 days, undisturbed and without a media change. After 8 days, the cells were harvested and were processed for immunophenotypic analysis by flow cytometry and functional CFU-c assays.

Flow Cytometry.
Flow cytometry for multilineage cell surface markers was performed on HSPC before and after expansion in order to calculate the fold change in the numbers of lineage-specific progenitors. CD34 + CD45 lo phenotype was used for all progenitors committed for hematopoiesis: CD34 + CD133 + and CD34 + CD38 − for cells with an early commitment to granulocyte, erythrocyte, monocyte, and megakaryocyte lineages. 7-Amino-actinomycin D (7-AAD) was added to measure the cell viability. The methodology for HSPC pre-and postexpansion flow cytometry analysis was followed as described elsewhere [15]. Briefly, 2 × 10 4 HSPC cells (pre-and postexpanded) were added to Fluorescence-Assisted Cell Sorting (FACS) tubes containing 100 L of PBS/2% FACS solution, and appropriate antibodies and viability dye were added. Stained cells were incubated at RT in the dark for 20 min, followed by washing with 1x PBS containing 2% FCS to remove excess antibody. Finally the cell pellets were resuspended in 300 L fixative solutions (PBS 2%, FCS 1%, formaldehyde 2 g/mL, and actinomycin D) and stored at 4 ∘ C in the dark. Data acquisition and analysis were performed within three days in FACS Calibur flow cytometer using CellQuest software. Antibodies used in FACS were purchased from Becton Dickinson Biosciences, USA. For determining the mean and standard error values, data from three independent experiments with triplicate samples in every experiment was used. For determining the fold increase in total cell number and lineage-specific cells the calculations were done as follows:   5.0 (GraphPad S) and a value of ≤ 0.05 was considered as significant.

Phase Contrast Microscopy.
After HS-5 cells had been decellularized by treatment with NH 4 OH, the cell debris was removed leaving behind the ACM. Phase contrast micrographs of HS-5 cells before (Figure 2(a)) and after (Figure 2(b)) NH 4 OH treatment clearly demonstrate the absence of cell nuclei in the decellularized native ACM (Figure 2(b)). The spindle-shaped cells absorb NH 4 OHcontaining water. They become round and burst due to excessive accumulation of water leaving behind some cellular components within the ECM. Complete decellularization can be confirmed by the absence of intact nuclei. Hence, decellularization efficiency was further evaluated by staining with DAPI for the absence of intact DNA. The DAPI staining of intact HS-5 cells and decellularized ACM is shown in Supplementary Figure S1.

Scanning Electron
Microscopy. SEM analysis of the native and spin coated ACM showed the microdetailing and architecture of the matrix surfaces. Native ACM (Figure 3(a)) exhibit the presence of thick bulky structures, which are the deposited matrix component after decellularization. The porous surface structure is visible as irregular shape measuring few m in diameter. On the other hand, spin coated ACM (Figure 3(b)) exhibited more uniform layer with the presence of nanosized globular structures on the entire surface.

Atomic Force
Microscopy. AFM analysis of native and spin coated ACM was performed to understand the topographical differences (thickness, roughness, and uniformity) on the surface of the matrices. To analyze the 2D and 3D  (Figure 4(e)), and that for spin coated ACM were at +12 nm and −4 nm, respectively (Figure 4(f)). The high range of difference in the values of surface height suggests that native ACM surface was more irregular than spin coated ACM. The spin coated ACM showed repetition of similar surface height profile in a line scan.
The thickness of matrix deposited on the native and spin coated ACM was measured by AFM. The average thickness was calculated to be equal to the difference between the scratched and the exposed glass surface and the average matrix deposit level. In native ACM, different levels of thickness were observed to be consistent with the mixed symmetries present on the surface. The average thickness was calculated by scanning random segments in different frames and multiple scans were used for the final calculation in both the substrates. The representative AFM image of the thickness profile of native ACM is shown in Figure 5(a) and that of spin coated ACM is shown in Figure 5(b). The thickness of the native ACM was calculated to be 583 ± 130 nm and that of spin coated ACM was 160 ± 11 nm (Supplementary Table 2

Cytochemical Characterization.
Cytochemical staining of substrates showed that both native and spin coated ACM depict similar retention of matrix constituents such as collagen (Figures 6(a) and 6(b)), fibrillar collagen (Figures 6(c) and 6(d)), proteoglycans (Figures 6(e) and 6(f)), and GAGs (Figures 6(g) and 6(h)). In contrast to the native ACM, where the stains are retained by deposited matrix leaving spaces in between, the spin coated ACM exhibit more uniform stain retention. The differences in the physicochemical properties of the substrates did not seem to have affected the qualitative distribution of matrix proteins on the two substrates. Cytochemical staining images for Masson's trichrome, Picro Sirius Red, Alcian Blue, and Safranin-o on blank TCP are shown in Figure S3.

Immunofluorescence Characterization.
Antibody staining of native and spin coated ACM confirmed the presence of all the three major ECM molecules, that is, fibronectin (Figures 7(a) and 7(b)), collagen type I (Figures 7(c) and 7(d)), and laminin (Figures 7(e) and 7(f)) in both the substrates. No nuclear staining (as shown by DAPI staining) or presence of cytoplasmic actin (as shown by Alexa Fluor-488 phalloidin actin staining) could be seen. Native ACM exhibited structural features that were similar to the native stromal structure, whereas the spin coated ACM exhibited uniformly distributed nonfibrous particulate matrix on the surface (Figures 7(e) and 7(f)). Control images of immunofluorescence studies are shown in Figure S4.

HSPC Expansion on ACM
3.4.1. HSPC Cell Morphology. After 8 days of HSPC expansion, images were taken on a light microscope (Axiovert Live cell 200M, Carl Zeiss, Germany). The maximum increase in cell number after 8 days of expansion in culture was observed on spin coated ACM followed by native ACM (Figure 8). HSPC cell morphology was best retained in native and spin coated ACM as compared to the control substrates.

Immunophenotypic Analysis of HSPC Expansion.
The flow cytometry data was obtained from three different UCB derived HSPC and the average cell viability and percentage of each lineage-bearing cell were calculated. The corresponding dot plots for CD34 + CD45 lo , CD34 + CD38 − , and CD34 + CD133 + cells, before and after expansion on the substrates, have been shown in Figure 9. Fold expansion was calculated according to the formula discussed earlier in Section 2.
Our results indicate that spin coated ACM could better support the expansion of CD34 + CD133 + bearing progenitors 44-fold on spin coated ACM versus 35-fold on native ACM in comparison to the controls (Figure 10(a)). Spin coated substrates were also good for expanding committed HSPC (CD34 + CD45 lo ), as compared to TCP and cellular control but not for undifferentiated CD34 + CD38 − progenitors ( Figure 10(a)).

Colony Forming Cells Analysis of HSPC Expansion.
CFU-c data were obtained from three independent experiments of HSPC expansion on ACM substrates and controls. Comparison of results of expansion between different substrates is represented in Figure 10(b). Although expansion of all cell types (CFU-GM, BFU-E, CFU-GEMM, and total viable cells) was observed on both ACM, the expansion of BFU-E (170-fold) was efficient in spin coated ACM as compared to the TCP and cellular controls. CFU-GM (120fold) and CFU-GEMM (70-fold) were more efficient and significant on spin coated ACM as compared to the native ACM, TCP, and cellular controls (Figure 10(b)).

Discussion
The discovery of HSPC and its application for restoration of normal hematopoiesis in malignant and nonmalignant, haematological disorders [2] has been a landmark in the field of cell transplantation. Very often, the success of this treatment is limited due to the low numbers of repopulating stem cells in the graft. This deficiency can be improved by the expansion of HSPC ex vivo so that the requirement of a large number of uncommitted primitive cells to repopulate the BM and restoration of normal hematopoiesis is met. The fact that the hematopoietic niche regulates HSC selfrenewal and proliferation in vivo has inspired researchers to recreate niche like in vitro conditions for achieving lineagespecific HSPC growth. The hematopoietic niche in BM is a complex biological milieu; it comprises several cellular and molecular components, soluble growth factors, and nonsoluble ECM [6]. Both the cellular and extracellular counterparts of the hematopoietic niches have been investigated extensively to understand their stem cell regulatory activities and properties [16]. It has been demonstrated that niche components can regulate adhesion, proliferation, differentiation, and survival of HSPC in the BM [6]. ECM molecules anchor growth factors and they provide mechanical strength to the niche for facilitating interactions between other niche components [17].
Earlier studies have focussed on individual ECM components for HSPC expansion, which do not represent the in vivo complexity of the hematopoietic niche [18][19][20]. For example, a cocktail of different ECM molecules did not give significant ex vivo expansion of HSPC probably because they did not provide all the required extracellular signals necessary for HSPC expansion in vivo [11]. Our study has succeeded on this account because we have used the entire ACM, which potentially contain all ECM molecules along with associated growth factors and other regulatory molecules that are required for HSPC expansion. In this report, we have demonstrated that ACM derived from human BM stromal cell line, both in native or spin coated forms, could efficiently support ex vivo HSPC expansion. We have also shown that these ACM are enriched in fibronectin, laminin, collagen, and ECM-associated growth factors as observed in the ECM of in vivo hematopoietic niches [21,22]. Previous studies have demonstrated that HSPC can enhance their functions by directly binding to ECM molecules; for example, fibronectin enhances CFU-GM and BFU-E progenitor cell growth ex vivo by binding to 4 and 5 integrins on these cells [23]; similarly immobilized fibronectin used as 3D scaffolds [24] or coatings and fibrillar collagen-I associated with glycophosphoprotein and osteopontin support the expansion of UCB derived HSPC [13]. We have demonstrated the enhanced expansion of HSPC on HS-5 derived ACM. Our results suggest that HS-5 derived ACM mimic the in vivo properties of the vascular niche in the BM. To our knowledge, this is the first report on the use of ACM derived from human stromal cells that significantly supports the expansion of UCB derived HSPC. An additional advantage of our results is that HS-5 derived ACM can be used in the clinical studies where xenogeneic substrates are not permitted due to immunological considerations. We also demonstrate that, along with biochemical properties, the topographical and physical features also play a critical role in HSPC growth and survival. We could observe an improvement in the lineage-specific HSPC expansion of committed progenitor cells population on spin coated HS-5 ACM as compared to native ACM. One possible explanation is that the topographical differences in surface roughness, thickness, and the uniform nanoglobular surface architecture of the substrates could be responsible for this functional improvement. With the spin coating, the matrix proteins are more uniformly distributed which probably helps in improving the total surface area of interaction of the cells with the matrix components. The thickness of the matrix is also known to affect the fate of the cells [25]. It has been shown that thinner matrices having higher elastic modulus and stiffness can better mimic the rigid bone microenvironment [25]. The nanoglobular surface features of tissue-engineered substrates are also known to support better osteogenic functions. Surface microtopography and roughness have also been reported   to play an important role in the differentiation of MSCs [26] and embryonic stem cells [27] and in the adhesion, spreading, proliferation, and differentiation of osteogenic cells [28]. Based on our results, we propose that topographical features of spin coated ACM that are thinner (3.5-fold) and more rough (2-fold) than native ACM can support a better expansion of committed progenitor of hematopoietic cell populations. These topographical features such as uniformity, stiffness, and roughness of the spin coated ACM could have contributed to the improved expansion in spin coated ACM. Our data confirm that an optimum substrate topography and roughness are necessary for achieving enhanced availability of the nutritive medium and growth factors to the cells, and, therefore, such surfaces are better than others for supporting all cell functions.
Recent reviews also highlight the use of various technologies in the development of efficient biomaterials, using one or more ECM components for tissue engineering applications [29][30][31]. Such advancements in biomaterial designing can be applied to extract maximum potential of human tissue-specific ACM without compromising the functional properties of individual ECM components to get enhanced expansion of lineage-specific HSPC cells ex vivo.

Conclusion
Our report describes a simple and novel method to prepare native and spin coated ACM that mimic functional properties of the hematopoietic niche in human BM. ACM prepared by our method can provide better understanding of physicochemical and topographical properties of hematopoietic niches and reveal novel regulatory roles of soluble and ECM bound signalling molecules. Since the matrix can be prepared in advance and stored, this can be used as a simple coating material with a known concentration, thereby minimizing the variation that is seen when using the feeders cells as culture system. The native and spin coated ACM can be stored in sterile conditions for a longer period as compared to conventional biological substrates. Our methodology also provides a general protocol to prepare ACM from other cell sources (e.g., embryonic fibroblasts and foreskin fibroblasts) and replace their use as feeder layers to minimise variations in various tissue engineering experimental setups.