Neurogenesis of Rhesus adipose stromal cells

Stem cells are clonogenic cells that have the capacity for self-renewal and multilineage differentiation. During embryogenesis, totipotent embryonic stem cells that are derived from the blastocyst give rise to ectoderm, mesoderm and endoderm lineage cell populations (Weissman, 2000). It has become evident that stem cells persist in adult tissues, although they represent a rare population localized in small niches (Woodbury et al., 2002). Adult stem cells are not totipotent, but they are capable of self-renewal and differentiation into multiple specialized cell types (Kopen et al., 1999). In tissues from postnatal animals, stem cells have been successfully isolated from liver, intestine, bone marrow and brain (Wei et al., 2000). Neural precursors that differentiate into neurons, astrocytes and oligodendrocytes may hold significant therapeutic potential for the replacement of damaged or diseased neural tissue resulting from congenital neuropathological conditions, brain injuries and neurodegenerative disorders. Although regional neurogenesis continues throughout the lifespan of rodents and humans, the number and availability of neural stem cells (NSCs) is limited in the postnatal central nervous system (CNS) (Barami et al., 2001; Shetty and Turner, 1996). The transplantation of NSCs has been shown to provide functional improvement in vivo (Barami et al., 2001; Borlongan et al., 1997; Masada et al., 1997; Shetty and Turner, 1996).

Recent evidence indicates that bone marrow stem cell transplantation effectively prevented the progression of neurological disease signs in some functional studies, if it is performed at an early stage in the disease (Jin et al., 2002). Subpopulations of bone marrow cells may serve as an alternative source of stem cells for the treatment of CNS disease, whereby mesenchymal stem cells differentiate into various lineages of brain cells. It has been shown that cells isolated from both bone marrow and umbilical cord blood (CB) can give rise to neural cells in vitro (Black and Woodbury, 2001; Kohyama et al., 2001; Reyes and Verfaillie, 2001; Deng et al., 2001; Sanchez-Ramos et al., 2000; Sanchez-Ramos et al., 2001; Woodbury et al., 2000; Colter et al., 2000; Colter et al., 2001) and in vivo (Azizi et al., 1998; Kopen et al., 1999; Mezey et al., 2000; Brazelton et al., 2000). Many studies have shown that bone marrow-derived cells can give rise to neural cells as well as many tissue-specific cell phenotypes, including hematopoietic, skeletal muscle, hepatic, heart and vascular endothelial cells (Terskikh et al., 2001; Gussoni et al., 1999; Petersen et al., 1999). The results of these studies have shown that host tissue-specific microenvironment conditions may be essential for the multilineage transdifferentiation of bone marrow-derived stem cells (BMSCs). BMSCs also have been used as vehicles for gene delivery to various tissues including the brain (Ding et al., 1999; Jin et al., 2002; Park et al., 2001; Suzuki et al., 2000; Kang et al., 2003). These findings suggest that bone marrow cells are a potential source of brain progenitor cells and have clinical importance in applications for tissue engineering and also as vehicles for gene therapy.

Adipose tissue has been identified as an alternative source of pluripotent mesenchymal stromal cells (Patrick, 2000; Zuk et al., 2001). Cells isolated from adipose tissue are self-renewing and can be induced to differentiate along several mesenchymal tissue lineages, including adipocytes, osteoblasts, myocytes and chondrocytes (Zuk et al., 2001; Halvorsen et al., 2001; Erickson et al., 2002). Adipose tissue, like bone marrow, is derived from the embryonic mesoderm and contains a heterogeneous stromal cell population (Zuk et al., 2002). Recently, MSCs isolated from the adipose tissue of rats were differentiated into neuron-like cells expressing neuronal markers (Kang et al., 2003; Safford et al., 2002). Therefore, adipose tissue may serve as an alternative source of pluripotent stromal cells capable of neural differentiation and, as such, may have application for the treatment of neurologic disorders.

Because little is known about the biologic, differentiation or engraftment properties of mesenchymal stem cells in higher order animals, we have begun to isolate and characterize the biological properties of these cells from the adipose tissue of rhesus monkeys. Macaques share greater biological similarities with humans than most other species with which stem cell research is being conducted, and therefore provide an unmatched opportunity to research diseases that afflict humans. The goals of these studies were as follows: (1) to isolate and characterize the growth and mesodermal differentiation capabilities of adipose tissue stem cells compared with pBMSC and (2) to investigate the differentiation potential of these cells along neural lineages in vitro. Our results show that adipose tissue is a viable source of mesenchymal stem cells in non-human primates that are capable of multilineage differentiation along mesodermal and neural lineages in vitro.

All animal procedures conformed to the requirements of the Animal Welfare Act and protocols were approved before implementation by the Institutional Animal Care and Use Committee (IACUC) of Tulane University. The animals were housed under conditions approved by the Association for the Assessment and Accreditation of Laboratory Animal Care International. Healthy Rhesus monkeys (Macaca mulatta, 15-20 kg, 10 to 15 years old) of both sexes were used for these studies. Activities related to animal care were performed as per standard Tulane National Primate Research Center operating procedures. All animals were negative for simian retrovirus (SRV) and simian T cell leukemia virus (STLV).

Non-human primate adipose tissue was obtained under local anesthesia. The raw adipose tissue was processed according to established methodologies to obtain a stromal vascular fraction (Zuk et al., 2002). To isolate stromal cells, samples were washed extensively with equal volumes of phosphate-buffered saline (PBS), and digested at 37°C for 30 minutes with 0.075% collagenase (Sigma, St Louis, MO). Enzyme activity was neutralized with α-Modified Eagle's Medium (MEM) (Invitrogen, Gaithersburg, MD), containing 10% FBS and centrifuged at 1200 g for 10 minutes to obtain a high-density cell pellet. The pellet was resuspended in red blood cell (RBC) lysis buffer (Biowhittaker, Walkersville, MD) and incubated at room temperature for 10 minutes to lyse contaminating RBCs. The stromal cell pellet was collected by centrifugation, as described above, and incubated overnight at 37°C/5% CO₂ in α-MEM medium containing 10% FBS.

For neural lineage potential comparison studies, rhesus BMSCs were obtained from 2-3 ml aspirates from the femur. The aspirate was diluted 1:2 in PBS and marrow cell fraction was obtained by centrifugation over 50% Percoll (Pharmacia LKB, Piscataway, NJ) at 1100 g for 30 minutes at 20°C. The nucleated cells were collected from the interface, diluted with two volumes of PBS, and collected by centrifugation at 900 g. The cells were resuspended, counted and plated at a concentration of 150-200 cells/cm² onto Nunclon culture dish (Nunc, Naperville, IL). The cells were cultured in α-MEM supplemented with 10% FBS (Atlanta Biological, Lawrenceville, GA) and 1% penicillin/streptomycin antibiotic solution. Medium was replaced first at 24 hours and then every third day thereafter.

To verify the multipotential differentiation of mesenchymal characteristics of pATSCs, cells were subjected to differentiation in conditions known to induce adipogenic, osteogenic and chondrogenic lineages in human cells. Before culture in the induction medium, cultures were grown to at least 80% confluence.

For adipogenic differentiation, pATSCs were induced by passaging cells at a 1:10 dilution in control medium and supplemented 10 ng/ml insulin and 10^-9 M dexamethasone. Adipogenic differentiation was visualized by the presence of highly refractive intracellular lipid droplets in phase contrast microscopy or staining by Oil-Red O. To induce osteogenic differentiation, the cultures were fed daily with control medium to which was added 10 mM β-glycerophosphate, 50 ng/ml ascorbic acid and 10^-9 M dexamethasone for 3 weeks. Mineralization of the extracellular matrix was visualized by staining of the cultures with Alizarin Red S (2% w/v Alizarin Red S adjusted to pH 4 using ammonium hydroxide) for 5 minutes at room temperature followed by a wash with water. Chondroblast differentiation was induced by differentiation medium supplemented with 6.25 μg/ml insulin, 10 ng/ml transofrming growth factor β1 (TGF β1) and 50 ng ascorbate-2-phosphate in control medium for 3-4 weeks. After differentiation, the cultures were washed and fixed in 4% paraformaldehyde and stained for glycosaminoglycans using 0.1% Safranin O.

Undifferentiated pATSCs and pBMSCs cultured at high densities spontaneously formed spherical clumps of cells that were isolated in 0.25% trypsin (Invitrogen). We also collected free floating neurospheres that were released from the cell culture surface into the culture media. The spheres of cells were transferred to a Petri dish and cultured in Neurobasal medium (NB) (Invitrogen) supplemented with B27 (Invitrogen), 20 ng/ml bFGF, and 20 ng/ml EGF (Sigma) for 4-7 days. The culture density of the spheroid bodies was maintained at 10-20 cells/cm² to prevent self aggregation.

For neural lineage differentiation, neurospheres derived from pATSCs were layered on PDL-laminin double-coated chamber slide (Lab Tek, Nalge/Nunc). Spheres were cultured and maintained for 10 days in NB media containing only the B27 supplement. During differentiation, 70% of the media was replaced every 4 days. The cells were examined at 10 days after differentiation by immunocytochemistry, western blot and reverse transcription polymerase chain reaction (RT-PCR). All data to be shown are representative of at least three different experiments.

For phenotypic characterization by flow cytometry, undifferentiated pATSCs, pATSC-derived neurospheres and adherent cells were harvested by trypsinization, washed twice with PBS and suspended at a concentration of 1×10⁶ cell/ml and incubated with antibodies to the following antigens: CD3, CD4, CD8, CD11b, CD13, CD90, CD164, CD133, CD59 and HLA-1 for 20 minutes. For FACS analysis, we used primary antibody directly conjugated with APC, or FITC. Monoclonal antibodies to CD34, CD3, CD4 and CD8 were used to identify cells as hematopoietic. The stained cells were thoroughly washed with two volumes of PBS and fixed in neutralized 2% paraformaldehyde solution. For an isotype control, nonspecific mouse or rabbit IgG (DAKO, Chemicon or Santa Cruz) was substituted for the primary antibody. The labeled cells were analyzed on a FACScan argon laser cytometer (Becton Dickinson, San Jose, CA).

Before and after neural differentiation of pATSCs, total cellular RNA was isolated with Trizol (Invitrogen) reverse transcribed into first strand cDNA using oligo-dT primer and amplified by 35 cycles (94°C, 1 minute; 55°C, 1 minute; 72°C, 1 minute) of PCR using 20 pM of specific primers. PCR amplification was performed using the primer sets. All primer sequences were determined using established human GeneBank sequences for genes indicative of neural lineages or control genes. Duplicate PCR reactions were amplified using primers for GAPDH as a control for assessing PCR efficiency and for subsequent analysis by 1.5% agarose gel electrophoresis.

Primer sequences for all the aforementioned genes were as following (Table 1).

To assess the efficiency of neural differentiation and compare the levels of expression of brain-derived neurotrophic factor (BDNF) and microtubule associated protein-2 (MAP2ab) expression in differentiated pATSCs and pBMSCs, quantification was performed using real-time RT-PCR. Total cellular RNA was isolated using conventional protocol. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers and probe (5′FAM and 3′TAMRA) were purchased from Applied Biosystems (Foster, CA). Quantitative real-time RT-PCR was performed using this kit according to the manufacturer and an ABI7700 Prism Sequence Detection System. Primer and probe sequences were designed using Primer Express software (PE-Applied Biosystems, Warrington, UK) using gene sequences obtained from the GeneBank database. All probes are designed with a 5′fluorogenic probe 6-FAM and a 3′quencher TAMRA. The expression of human GAPDH was used to standardize gene expression levels.

For analysis of neural differentiation of pATSC neurospheres, differentiated cells were fixed with 4% paraformaldehyde, and incubated with 10% goat serum to prevent nonspecific antibody binding. The cells were incubated overnight at 4°C with antibodies. For detection of differentiated neuronal or glial cell proteins, we used several species-specific monoclonal antibodies directed against glial acidic fibrillary protein (GFAP) (1:2000, Dako, Carpinteria, CA), MAP2ab (1:250, Sigma, St Louis, MO), nestin (1:250, Sigma), Neu N (1:500, Sigma), NF160 (1:500, Sigma) and myelin basic protein (MBP) (1:250, Chemicon, Temecula, CA). After extensive washing in PBS, the cells were incubated for 30 minutes with FITC or Alexa Fluor 568 conjugated secondary antibodies (1:250, Molecular Probe, Eugene, OR). Controls in which primary antibodies were omitted or replaced with irrelevant IgG resulted in no detectable staining. Specimens were examined using a Leica TCS SP2 laser scanning microscope equipped with three lasers (Leica Microsystems, Exton, PA). Immunocytochemical studies were repeated at least three times.

Protein extracts were prepared from undifferentiated or differentiated pATSCs by the treatment of lysis buffer containing 20 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mM phenylmethylsulfonylfluoride, 10 μg/ml aprotinin and 1 mM sodium orthovanadate. Total protein (30-40 μg/ml) was resolved on 12.5% acrylamide gel and electroblotted onto Polyvinyldiethylfluoride (PVDF) membrane (Amersham). The blot was probed with either mouse anti-nestin (1:500) or mouse anti-MAP2ab antibodies (1:500). Immunoreactive bands were detected using horseradish peroxidaseconjugated anti-mouse IgG antibodies (Amersham) and visualized by enhanced chemiluminescence (Amersham).

Samples for gene array analysis were prepared from total RNA and microarray analysis was performed following the manufacturer's recommendations. Fragmented cRNA (15 μg) was hybridized 16 hours at 45°C to the HG-U95A array for the comparison study (Affymetrix, Santa Clara, CA). After hybridization, the gene chips were automatically washed and stained with streptavidinephycoerythrin by using a fluidics station. Finally, the probe arrays were scanned at 3 μm resolution using the Genechip System confocal scanner made for Affymetrix by Agilent. Affymetrix Microarray Suite 4 was used to scan and analyze the relative abundance of each gene as derived from the average difference of intensities. Analysis parameters used by the software were set to values corresponding to moderate stringency. The threshold values to determine the present (P) or absent (A) call were set as follows; α1=0.05, α2=0.065, τ=0.015. Fluorescence intensity was measured for each chip and normalized to average fluorescence intensity for the entire chip. Output from the microarray analysis was merged with the Unigene or Genebank descriptor and stored as an Excel data spreadsheet. The definition of increase or decrease, or no change of expression for individual genes was based on ranking the Difference Call from two comparisons (2×1), namely, no change (NC) of expression for individual genes was merged with the Unigene or GeneBank descriptor and stored as an Excel data spreadsheet. The definition of increase (I), or no change (NC) of expression for individual genes was based on ranking the Difference Call from the two comparisons (2×1) namely, No change=0, Marginal Increase/Decrease=1/-1, Increase/Decrease=2/-2. The final rank referred to summing up the two values corresponding to the Difference Calls and the value varied from -6 to 6. The cut-off value for the final determination of Increase/Decrease was set as 3/-3. Evaluation of the reproducibility of paired experiments was based on calculation of the coefficient of variation (CV) (SD/mean) for fold change (FC). The CV of FC must be less than or equal to 1.0. Finally, genes with an FC over 1.5 were considered significant. These cut-off values represented a conservative estimate of the numbers of genes whose expression levels differed between samples. Gene categorization was based on a literature review.

Within two to three passages after the initial plating of the primary culture, pATSCs appeared as a monolayer of broad, flat cells (20-30 mm in diameter) (Fig. 1A). When the cells approached densities over 80%, the cell morphology changed to a more spindle-shaped, fibroblastic morphology (Fig. 1B). pBMSCs isolated from femurs and tibias showed heterogeneous groups of low contrast flat cells together with smaller, more spindle-shaped cells (Fig. 1C). Furthermore, both the pBMSC and pATSC populations had many small, round cells that were attached to or growing on these cells; these might have been cells from the hematopoietic lineage or the recently described rapidly self-renewing stem cells (data not shown) (Colter et al., 2000; Colter et al., 2001). Single-cell pATSC cultures generated CFU (colony forming unit) clones after 2 weeks of culture, indicating the potential for self-renewal (Fig. 1D). For evaluation of pATSCs-derived single-cell clones pluripotency, single clone-derived progenies were plated at 50 cells/cm² and induced differentiation to adipogenic, osteogenic and neurogenic. After differentiation cells were stained with Oil Red O and Alzarin Red S to detect fat and calcium deposits. For neural lineage detection, cells were immunostained with GFAP, Tuj and MBP antibodies. The results showed that a high percentage of clonal populations were adipogenic (82%), osteogenic (64%) and neurogenic (79%) potential positive (Fig. 3). Of 20 clones analyzed, 17 clone-derived showed multilineage potential, as shown by their differentiation along several mesodermal and neural lineages.

pATSCs did not spontaneously differentiate during in vitro culture expansion. The differentiation potential of rhesus pATSCs appears very similar to human pATSCs (see Table 2) (Kang et al., 2003). Using lineage-specific differentiation culture media, expanded pATSCs were capable of generating adipocytes as indicated by the accumulation of neutral lipid vacuoles (Fig. 1E). Osteogenic lineage capacity was detected by an increase in calcium deposition, as identified by Alzarin Red S (Fig. 1F). Chondrogenic induction of pATSCs, under the micromass conditions, resulted in cell condensation after induction and was followed by spheroid or nodule formation by 3-4 weeks. Nodules at this time point stained positively using Safranin O staining solution. pATSCs chondrogenesis was absolutely dependent on high cell density and induction conditions (Fig. 1G). Bone nodule formation was dependent on the presence of TGFβ1 and could not be induced in monolayer cultures. To analyse clonally derived populations of pATSCs, we performed low density cell culture in 10-cm dishes (50-100 cells). More than 95% of the CFU derived from single cell differentiated to mesodermal lineages (adipogenic, osteogenic and chondrogenic) after 20 days in lineage specific induction media. The pATSCs retained their multilineage potential for as long as 7 weeks of culture (data not shown). Cumulative population doublings were measured with respect to passage number in multiple animal samples. Our data were consistent with the observed lag time upon initial culture; pATSCs underwent an average of three population doublings before the first passage. An average of 1.5-2 population doublings was observed on subsequent passages, and a linear relationship between cumulative population doubling and passage number was observed (Fig. 1H). After their initial plating at 200-300 cells/cm², the pATSCs adherent cells reached confluence within 1-2 weeks. No detectable loss of self-renewal capacity could be observed through passage 13 (2-3 months in culture), which is markedly different from rhesus pBMSC cells (B.A.B., unpublished observations).

pATSCs and pBMSCs were induced towards the neurogenic lineage through neurosphere formation and final differentiation on PDL-laminin-coated substrate in NB media that was supplemented with B27, basic fibroblast growth factor (bFGF) and epidermal growth factor (EGF). Neural differentiation was analyzed through the detection of the expression of neuronal markers (MAP2ab, Neu N and NF160), in addition to GFAP as a marker of astrocytes. During neurogenic induction in NB media, both cell populations undergo a marked morphological change from elongated fibroblast morphology to compact, spheroid bodies, which expand to larger spheroid bodies as the total cell number expands (Fig. 2A,E). After detachment of the spheroid bodies from substrate, neural induction was performed neural induction for 4 days through suspension culture in Petri dishes and then the intact NS or dissociated NS were layered on the PDL-laminin-coated chamber slide and cultured for an additional 10 days. As soon as the cells were layered on laminin coated surface, the spheroid cell mass began to adhere and spread across the growth surface, forming long chains of cellular processes (Fig. 2B,C,F,G) and, finally, the cell processes began to exhibit secondary branching with multiple extensions (Fig. 2D,H).

To fully characterize the pATSC-derived neurospheres further, we performed both immunocytochemistry and western blot analysis for specific antigens indicative of neural cell lineages. The data from these analyses indicate that pATSC-derived neurospheres express high levels of nestin, MAP2ab, GFAP and CD133 (Fig. 4A-D). We next assessed the levels and the pattern of induction of MAP2ab expression in pATSC-NS from 0 to 6 days of neurosphere formation. The results indicated that MAP2ab expression was significantly induced through the fourth day of NS formation and then decreased (Fig. 5). The decreased expression of MAP2ab protein may be derived from apoptosis or death of inner cells within the neurospheres. The pATSC-NS cells differentiated on the laminin-coated surfaces for 10 days. The increase in neural lineage related protein expression on neural induction was confirmed using RT-PCR analysis (Fig. 6), and double immunostaining for MAP2ab (Fig. 7A,E,F), Neu N (Fig. 7B,D), NF160 (Fig. 7A,B), astrocyte marker, GFAP (Fig. 7C), and nestin (Fig. 7G) as well as western blot analysis (Fig. 8A). We failed to detect the expression of MBP in any of these assays, suggesting that pATSCs did not differentiate into oligodendrocytes in vitro. Control populations of undifferentiated pATSCs did not express detectable levels of the assessed neuronal, oligodendrocyte or astrocyte protein markers, confirming the specificity of our neural differentiation methods and immunocytochemical staining protocol. However, RT-PCR analysis confirmed the expression of nestin in undifferentiated pATSCs. The expression of markers characteristic of more mature neuronal subtypes, choline acetyltransferase (ChAT) or GAD₆₅, was not observed in pATSCs by RT-PCR (Fig. 6).

For comparison of neural differentiation potential between pATSCs and pBMSCs, western blot analysis was performed using MAP2ab and nestin antibodies; real-time RT-PCR was carried out with a MAP2ab-specific primer and probe set. We detected very low levels of nestin protein expression in NS derived from pBMSCs, however, in pATSC-derived NS it was highly expressed (Fig. 8A). Quantitative RT-PCR analysis indicated that MAP2ab expression level in the NS differentiated from pATSCs were twofold higher than those of pBMSC-NS (Fig. 8B). After neural induction and differentiation, analysis indicated that both nestin and MAP2ab were expressed at higher levels at both the RNA and protein levels in differentiated pATSCs than those in differentiated pBMSCs (Fig. 8).

To explore the phenotypic characteristics of the isolated pATSCs, and pATSC-NS, we performed flow cytometry using primary antibodies against surface epitopes. The flow results indicated that undifferentiated pATSCs were negative for the hematopoietic markers CD3, CD4, CD8, CD34 or CD45. They do express the cell-surface epitopes CD13, CD90, CD59 and HLA-1, and very low levels of CD11b. Comparison of the flow cytometric results for the undifferentiated rhesus pATSCs and pATSC-NS indicates that the expression of two CD marker antigens differ; CD90 (Thy-1) and CD164 (Sialomucin). Specifically, pATSC-derived neurospheres expressed high levels of CD90 and CD164, which were induced during neurosphere formation (Table 3).

Using our neural differentiation protocol, a variable number of cells underwent differentiation, although the response exceeded 70% of the starting population. In an attempt to optimize differentiation, we modified the neural differentiation protocol. The addition of BDNF (10 ng/ml) to the NB (supplemented medium B27, bFGF and EGF) increased the proportion of cells displaying neuronal characteristics and the response was more consistent (data not shown). After the incorporation of BDNF to the neural differentiation medium, neurotrophic factor receptor Trk B was induced in pATSCs-NS and differentiated pATSCs as indicated by RT-PCR analysis (Fig. 6, lane Q). Flow cytometric analysis of cells differentiated in the presence of BDNF showed that this cell population stained positive for the neuronal marker MAP2ab (53% positive cells), nestin (49%) and the astrocytic marker GFAP (55%) (Fig. 9).

We compared the in vitro neurogenic differentiation ability of pATSCs and pBMSCs. Neural lineage differentiation required that pATSCs and pBMSCs were cultured 1-2×10⁴ cell/cm² in NB medium supplemented with B27. We also compared mesodermal lineage differentiation capability using lineage specific induction medium. The differentiation efficiency of these two cell types was more or less similar when induced along adipogenic and osteogenic lineages. However, pBMSCs did not undergo efficient chondrogenic differentiation under the conditions used in this study, suggesting distinctions in the differentiation capacities between pATSCs and pBMSCs (Table 1). Neuroprogenitors (Neurospheres) can be expanded with bFGF, EGF and BDNF, and more extensive differentiation induced by removal of cytokines and growth on PDL-laminin-coated surfaces. Populations of differentiated pATSCs and pBMSCs have morphological and phenotypical characteristics of astrocytes (GFAP), neurons (MAP2ab and NF60) and neuronal precursor cell marker (nestin). Cells positive for MBP (oligodendrocytic) were not generated in stem cells isolated from either adipose tissue or bone marrow. Quantitative RTPCR analysis confirmed increased expression of BDNF in undifferentiated pATSCs compared with that in undifferentiated pBMSCs (data not shown).

In order to analyze the gene expression pattern, we performed oligonucleotide microarray analysis. The gene expression profile in pATSC control cells was compared with pATSC-NS. Total RNA was harvested from both cultures and gene expression profiles were compared using Affymetrix HG-U95a microarray (22,000 genes and ESTs). Affymetrix Microarray Suite 4.1 was used to scan and analyze the relative abundance of each gene. The signal output from each gene from the pATSC control profile was plotted against the pATSC-NS profile (Fig. 10), and the correlation coefficient (r) was calculated for each comparison. The analysis of the gene expression levels showed that less than 1% of the total genes were expressed at greater than 2.2-fold different levels in pATSC and pATSC-NS, as indicated by the r value (=0.8). In the online supplemental data, Table 4 shows a partial list of genes that are highly relevant for neural lineage development that were either induced (total number of genes=25) or inhibited (total number of genes=12) in differentiated pATSCNS when compared with undifferentiated pATSC control cells. Table S1 gives a partial list assembled into gene function of non-neuronal genes that are either upregulated (total number of genes=260) or downregulated (total number of genes=569) expressed in pATSC-NS compared with naïve pATSC.

The data from this study show that cells with multipotent adult progenitor characteristics can be isolated from different organs in non-human primates. The cells have similar morphology, phenotype and in vitro differentiation ability, and they have highly similar gene expression profiles (Wieczorek et al., 2003). Primary cultures of bone marrow and adipose tissue are a heterogeneous population containing small numbers of hematopoietic cells, pericytes, endothelial cells and smooth muscle cells (Zuk et al., 2001). However, Gronthos et al. reported that the frequency of these other cells appears to diminish quickly through serial passages in culture (Gronthos et al., 2001). Also, these cells have the ability to differentiate into multiple lineages, including osteogenic, adipogenic, chondrogenic and neural differentiation. We used pATSCs of 3-4 passages for the experiments and observed that more than 95% of cells expressed CD59 and HLA-1, but not CD34, CD3, CD4 or CD8. Also, the low levels of nestin and GFAP expression in undifferentiated ATSCs in the rhesus model are consistent with the previous reports in stromal cells of bone marrow origin (Deng et al., 2001; Kang et al., 2003) and suggest that ATSCs may retain a native potential for neural differentiation. pATSCs following the induction of neural differentiation revealed biologic and morphologic characteristics of neural lineages. The high percentage of cells undergoing neural differentiation in our study indicates that neural differentiation is occurring within a broad population of pluripotent cells. While immunocytochemical evidence from us and others (Zuk et al., 2001; Zuk et al., 2002) suggests that neurogenesis can occur in culture from rhesus adipose stromal cells, it remains to be investigated whether these neurons are electrically active and functional with all essential characteristics of mature CNS neurons (Song et al., 2002). In accordance with previous studies, stem and progenitor cells from adult adipose stromal tissue retain the capacity to differentiate toward mesenchymal and nonmesenchymal lineages, similar to bone marrow stromal cells (Sanchez-Ramos et al., 2000; Woodbury et al., 2000; Zuk et al., 2001; Erickson et al., 2002; Safford et al., 2002). Recent work on BMSCs undergoing early neurogenic differentiation has confirmed the expression of nestin, an intermediate filament protein thought to be expressed at high levels in neural stem cells (Sanchez-Ramos et al., 2000). Consistent with this, nestin expression was detected in noninduced pATSCs as well as in those induced under several neurogenic media conditions, suggesting the differentiation potential of pATSCs into a neural stem cell phenotype. Recently, nestin expression has also been observed in myogenic cells, endothelial cells and hepatic cells, indicating that it cannot be used as a marker for putative neurogenic differentiation of non-neural stem cells. After differentiation pATSCs showed a neuron-like morphology and the increased expression of three neuron specific proteins, NSE, MAP2ab and NF160 and mature astrocyte marker GFAP. MAP2ab and Neu N expression is thought to coincide with terminal differentiation of pATSCs. Also, coexpression of Neu N, MAP2ab and early neural precursor marker nestin may indicate potential neurogenic capacity in pATSCs. Compared with pATSCs, neural-lineage-induced pBMSCs have lower levels of MAP2ab expression indicating that neural lineage potential of the pATSCs may be greater than that of pBMSCs.

Analysis of pATSCs and pBMSCs properties has identified many common biological characteristics between the two populations. Importantly, we have also observed several distinct properties in the two populations that suggest they are very similar but not identical (Table 1). Also, our cell culture experience with pATSCs indicates that screening lots of serum, a requirement for efficient pBMSC culture, is not necessary for efficient expansion and differentiation. Our data indicate that pBMSCs did not undergo chondrogenic differentiation under the conditions used in this study, suggesting distinct differentiation capacities between the two stem cell lineages. Immunocytochemical analysis also identified differences in the surface epitope profiles of pATSC and pBMSC populations, and our data and others indicate that these distinctions between ATSC and BMSC populations may also extend down to the gene level (Zuk et al., 2002). BDNF and MAP2ab real-time RT-PCR data showed that the neural differentiation capabilities of pATSCs may be significantly higher than those of pBMSCs.

The differentiation potential of pATSCs may result from the commitment of lineage-specific precursors rather than the presence of a multipotent stem cell population. To verify proliferation and differentiation potential of pATSCs, we isolated clones derived from single pATSCs cells in 96-well plates. After 14 days culture, the plates were stained with 0.5% crystal violet in methanol for 5 minutes and we counted the number of colonies that were more than 2 mm in diameter. Around 80-90% of single clone cultures generated crystal violet-positive CFU clones after high potency of self renewal. For evaluation of multipotency of pATSCs-derived single cell clones, we passaged cloned population to population doubling (PD) 100 to PD 150. Single clone-derived ATSCs were plated at 50 cell/cm² and cultured and induced to adipogenic, osteogenic and neurogenic lineage differentiation. A high percentage of clonal populations were demonstrated to retain adipogenic (82%), osteogenic (64%) and neurogenic (79%) differentation potential (Fig. 3). Of 20 clones analyzed, 17 clones showed pluripotential potency along several mesodermal and neural lineages. This result indicates that a high ratio of subpopulation of pATSCs have the characteristics of multipotential and pluripotential in vitro, because individual progenitor cells are capable of self-renewal and can generate daughter cells capable of differentiating into the major mesodermal and ectodermal lineage. It also showed that more than 75% of subpopulation of pATSCs had neurogenic potential.

To investigate patterns of gene expression during the process of neurosphere formation and culture, cDNA microarray analysis was performed. Labeled cDNA targets were hybridized with the microarray, and 829 clones (downregulated gene in pATSCNS=569, unregulated gene in pATSC-NS=260) that differed by more than twofold intensity in at least one pairwise comparison were selected. That 22,000-gene microarray is representative of all the unique human gene sequences that were available at the time the array was produced. Many of the genes with the highest intensity values (Z score >8) in pATSC control and pATSC-NS were ESTs or unnamed genes. A discussion of all the named genes that were increased or decreased in the pATSC-NS related to pATSC control is impractical within the context of this paper. These include oncostatin M receptor, HIF-1 responsive RTP801, PDGF C, cellular retinoic acid binding protein 2, TGF β receptor 1, and syntaxin 1A. Furthermore, several genes including insulin-like growth factor beta polypeptide, dickkopf homolog3, achaetescute complex like 1, and erythropoietin that are expressed in neural stem cells were highly downregulated in pATSC-NS compare with pATSC control. Table S1 provides categories of genes that showed differences in expression levels (increased or decreased in pATSC-NS compared with pATSC) between two samples. Several genes that have been reported to be important for neural lineage were highly expressed in pATSC-NS (Table 3) (Wright et al., 2003). A broad variety of cellular functions are represented, including signaling, structural elements, cell cycle control and apoptosis, DNA function, transcription and translation, transport activity, cell adhesion, growth and trophic factors, general metabolic proteins and enzymes, as well as catalytic activity related genes. Because entire genome sequences or even large clustered EST sequences from many of the non-human primates used as experimental models in biomedical research are not currently available, interspecies microarray hybridization studies represent one possible way to identify genes within a transcriptome and profile the expression levels. The microarray technology platform used here in uniform and data can be normalized from different experiments (real-time RTPCR, immunocytochemistry). Furthermore, in previous studies (Marketa et al., 2003), the proportion of common genes shared between humans and macaque species might be higher than between two different monkey species.

Adipose stromal cells are easily obtained from patients and are, therefore, one of the most clinically practical sources of stem cells in adults. Moreover, these cells have few practical, ethical or immunological barriers to their clinical application and are promising materials for future cell and gene therapies. Devine et al. (Devine et al., 2001) reported that the intravenous injection of mesenchymal stem cells into baboons (papio anubis) was not associated with significant toxicity and the cells were capable of homing to and establishing residence in the bone marrow for an extended period of time. Recently, transplantation of human ATSCs improved functional deficits in ischemic brain injury induced by MCAo (Kang et al., 2003). Intracerebral grafting of BDNF-transduced hATSCs significantly improved motor recovery of functional deficits in MCAo rats. The data from this study indicate that transplanted hATSCs survive, migrate and improve functional recovery after recovery of stroke, and that genetically engineered hATSCs can express biologically active gene products and, therefore, can function as effective vehicles for therapeutic gene transfer to the damaged brain.

In summary, we have identified and characterized non-human primate adipose tissue stromal cells that contain progenitor cells that have a potential for in vitro expansion and neural differentiation.

We are extremely grateful to Xavier Alvarez for his help with the confocal microscopy, and Louis Martin and the staff of the flow cytometry core laboratory for their help with flow cytometric analyses. We would also like to thank James Munoz for helpful discussions. The work was supported by grant number RR00164 from the National Center for Research Resources, National Institutes of Health, a grant from the State of Louisiana Millennium Health Excellence Fund and the Louisiana Gene Therapy Research Consortium.