In vitro neurogenesis by progenitor cells isolated from the ... - Nature

© 2000 Nature America Inc. • http://medicine.nature.com

ARTICLES

In vitro neurogenesis by progenitor cells isolated from the adult human hippocampus NEETA SINGH ROY1, SU WANG1, LI JIANG4, JIAN KANG4, ABDELLATIF BENRAISS1, CATHERINE HARRISON-RESTELLI1, RICHARD A. R. FRASER2, WILLIAM T. COULDWELL3, AYANO KAWAGUCHI5,6, HIDEYUKI OKANO5,6, MAIKEN NEDERGAARD3,4 & STEVEN A. GOLDMAN1


Departments of 1Neurology and Neuroscience and 2Neurosurgery, Cornell University Medical College, 1300 York Ave. Room E607, New York, New York 10021, USA Departments of 3Neurosurgery and 4Cell Biology, New York Medical College, Valhalla, New York 10585, USA 5 Division of Neuroanatomy, Osaka University Graduate School of Medicine, and 6 Core Research for Evolutional Science and Technology, Japan Science and Technology Corporation, 2-2 Yamadaoka Suita, Osaka, 565-0867Japan Correspondence should be addressed to S.A.G.; email: [email protected]

Neurogenesis persists in the adult mammalian hippocampus. To identify and isolate neuronal progenitor cells of the adult human hippocampus, we transfected ventricular zone-free dissociates of surgically-excised dentate gyrus with DNA encoding humanized green fluorescent protein (hGFP), placed under the control of either the nestin enhancer (E/nestin) or the Tα1 tubulin promoter (P/Tα1), two regulatory regions that direct transcription in neural progenitor cells. The resultant P/Tα1:hGFP+ and E/nestin:enhanced (E)GFP+ cells expressed βIII-tubulin or microtubule-associated protein-2; many incorporated bromodeoxyuridine, indicating their genesis in vitro. Using fluorescence-activated cell sorting, the E/nestin:EGFP+ and P/Tα1:hGFP+ cells were isolated to near purity, and matured antigenically and physiologically as neurons. Thus, the adult human hippocampus contains mitotically competent neuronal progenitors that can be selectively extracted. The isolation of these cells may provide a cellular substrate for re-populating the damaged or degenerated adult hippocampus.

Dividing hippocampal progenitor cells have been reported in adult animals ranging from chickadees to humans1–7. In rodents, hippocampal neurogenesis can be modulated by stress8, enrichment9, exercise10 and learning11. Among primates, both adult macaques6,12 and humans7 show histological evidence of neurogenesis in the dentate gyrus. Hippocampal cells have been found in suspension cultures derived from both adult rats13 and humans14; these can expand in response to fibroblast growth factor (FGF)-2, include multipotential founders13, and are capable of heterotopic integration into other regions of granular neurogenesis, such as the olfactory subependyma15. Yet despite the widespread incidence of hippocampal neurogenesis in adult animals, human hippocampal progenitor cells have never been isolated. As a result, no assessment of the abundance, factor-responsiveness or regenerative capacity of these cells has been possible. To identify and extract neuronal progenitors from the adult human hippocampus, we transfected ventricular zonefree dissociates of surgically resected adult human hippocampi with plasmid DNA bearing the gene for humanized green fluorescent protein (hGFP), placed under the control of regulatory sequences for the genes encoding either the early neuronal protein Tα1 tubulin16–18, or the neuroepithelial protein nestin19,20. We isolated the fluorescent progenitors obtained to purity using fluorescence-activated cell sorting (FACS). Cells defined by either the nestin enhancer or Tα1 tubulin promoter divided in vitro, and both types gave rise to antigenically typical, functionally appropriate neurons. NATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000

Neurons arise in cultures of the adult human dentate gyrus To characterize mitotic cell types that can be collected from the adult human hippocampus, we obtained papain dissociates of surgically resected hippocampus from eight male patients, 5–63 years old. Four patients had temporal lobe resections for medication-refractory epilepsy21; two were subjected to decompressive resection during or after extra-axial meningioma removal, one sample was obtained during aneurysm repair, and one, during decompression for traumatic edema. For each, the dentate gyrus was dissected free from the temporal ventricular zone, and the two were removed separately. Dentate tissue was dissociated using papain/DNase (ref. 13); the resultant cultures were grown in Dulbecco’s modified Eagle medium/F12/N2, with 20 ng/ml FGF-2 and 2% platelet-depleted fetal bovine sera (PD-FBS). A mitotic marker, bromodeoxyuridine (BrdU; 10 µg/ml) was added at 6 hours of in vitro culture. Cultures were fixed after 1, 2 or 4 weeks in vitro, and their resident phenotypes were assessed by immunolabeling for one of two neuronselective proteins, βIII-tubulin/TuJ1 (ref. 22,23) or microtubule-associated protein (MAP)-2 (ref. 24). Among randomly chosen, low-power-magnification fields of dissociated hippocampus (10 per patient), 14.5 ± 6.8% (mean ± s.e.m.; n = 3 patients) of the scored cells expressed βIII-tubulin/TuJ1-immunoreactivity (βIII-tubulin+/TuJ1+ cells) at 1 week in culture (Fig. 1). At that time, MAP-2, which appears later in neuronal development than TuJ1, was expressed by 2.3 ± 1.3% of the hippocampal cells (Map-2+ cells). By 30 days in vitro, the 271


ARTICLES Fig. 1 Adult human hippocampus contains mitotic neuronal progenitor cells. a and b, Monolayer dissociate of adult human dentate gyrus, removed from a 33-year-old man after temporal lobectomy. Phase (a) and fluorescence (b) images of a cluster of neurons at 7 d in vitro, labeled with antibody against neuronal MAP-2 (b, red). c, Hippocampal culture from a 35-year-old; the culture was exposed to BrdU in vitro, then fixed and immunostained for BrdU as well as βIII-tubulin. d–f, TuJ1+ (red) and BrdU+ (green) neurons, generated by mitotic neurogenesis from hippocampal progenitors. Scale bar represents 50 µm.


percentage of MAP-2+ cells in each plate increased to 6.13 ± 1.4%, and that of TuJ1+ cells increased to 14.9 ± 7.8%. The relatively high initial incidence of cells expressing βIIItubulin, which appears early in neuronal ontogeny, together with the later maturation of MAP-2+ cells, indicated that these neurons arose from precursors, rather than from resident neurons that survived dissociation. The high proportion of TuJ1+ cells that incorporated BrdU (BrdU+ cells), indicating their mitogenesis in vitro, confirmed this.

c

Mitotic neurogenesis persists in adult hippocampal cultures A substantial proportion of the antigenically confirmed neurons incorporated BrdU from the culture media, indicating their genesis in vitro (Fig. 1). BrdU+/βIII-tubulin+ neurons constituted 25.4 ± 5.9% of all neurons defined by βIII-tubulin at 7 days in vitro. At this same time, BrdU+/MAP-2+ cells were rare, consistent with the relatively late appearance of MAP-2 after neuronal mitogenesis. But by 14 days in vitro, 3.0 ± 1.0% of all cells in these cultures expressed MAP-2, and of these, 22.5 ± 7.3% were BrdU+. The persistence of neurogenesis in these dissociates was also manifested by the sustained neuronal incorporation of BrdU in cultures to which the marker was first added after 5 days in culture. Among plates first exposed to BrdU on day 5 in vitro and

a

b

c

d

e

f

Fig. 2 The Tα1 tubulin promoter is transcriptionally active in adult hippocampal neuronal precursor cells. Hippocampal cultures derived from a 63-year-old man were transfected at 5 days in vitro CDIV with pPTα1:hGFP, to identify neuronal precursor cells and their young neuronal daughters. P/Tα1:hGFP+ cells were photographed at 7 (a–d) and 14 (e and f) days in vitro after transfection (DIV). Left, phase contrast images; right, fluorescence images. Scale bar represents 30 µm. 272

b

a

d

e

f

fixed on day 7, 8.2 ± 1.9% of the cells expressed TuJ1; this was almost identical to the 8.9 ± 2.8% incidence of TuJ1+ neurons in the plates exposed to BrdU from the outset (n = 10 fields/sample). Among those TuJ1+ neurons counted in the late BrdU-addition plates, 13.2 ± 7.6% co-labeled with BrdU when fixed at 7 days in vitro. Thus, new hippocampal neurons continued to be generated for at least the first week in culture. Moreover, neurogenesis was sustained beyond the first week, as reflected in an increasing number of neurons as a function of time in vitro: Among a sample of hippocampal cultures fixed at serial time points after plating, the average numbers of βIII-tubulin/TuJ1+ and total cells/field increased from 2.0 ± 0.2 and 23.6 ± 1.0, respectively, at 7 days in vitro (n = 141 low-power fields), to 10.3 ± 0.7 and 136.4 ± 7.1 at 30 days in vitro (n = 142). Yet despite this more-than500% increase in the number of neurons/field over the month in vitro, the commensurate increase in the total cell number ensured that the net percentage of TuJ1-defined neurons in these cultures was constant: 6.7 ± 0.6% at 7 days in vitro, and 8.9 ± 1.1% at 30 days in vitro. These data indicated that neurons could be generated from dividing progenitors in cultures of the adult human hippocampus. P/Tα1 tubulin identifies neuronal progenitor cells To identify and separate neuronal progenitors from the adult hippocampus, we used a strategy of promoter-based FACS described for isolating progenitor cells from the fetal brain. With this approach, mixed cell dissociates are transfected with plasmid DNA bearing fluorescent transgenes placed under the control of cell-specific promoters; the cell types of interest fluoresce and can be isolated by FACS. To establish the feasibility of this strategy for selecting progenitor cells from the central nervous system, a separation vector was constructed by placing the gene encoding hGFP (ref. 25) under the control of the Tα1 tubulin promoter (P/Tα1), an early neuronal regulatory sequence16,17. When the resultant plasmid P/Tα1:hGFP was transfected into dissociated forebrain cultures, P/Tα1:hGFP was found to be strongly expressed by precursors and very young neurons, but not by glia18. This allowed the use of FACS to enrich the transfected progenitors, based upon their P/Tα1 tubulin-driven GFP fluorescence. Using this strategy, neuronal precursor cells could be identified and selected from both the fetal and adult rat brain18,26. Here, we extended this strategy to enrich neuronal progenitors from the adult human brain, by sorting cells obNATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000


ARTICLES


Fig. 3 Adult hippocampal progenitors sorted by either E/nestin:EGFP or P/Tα1:GFP give rise to neurons. Cells were derived from the dissociated dentate gyrus of a 20-year-old man. a, Cells were transfected with E/nestin:EGFP (green), fixed 4 d later, and co-stained for neuronal βIII-tubulin (red). b, Cells in the culture were sorted, and the E/nestin:EGFP+ cells were allowed to differentiate in 5% FBS, then they were fixed and stained for βIII-tubulin 3 d after FACS. c and d, Hippocampal cells were transfected with P/Tα1:hGFP (green) and stained for βIII-tubulin (red): c, unsorted, at 4 d in vitro; d, a week after FACS (total of 14 d in vitro). The P/Tα1:GFP+ cells express MAP2 as well as βIII-tubulin, and incorporated BrdU during the first 7 d in vitro. Scale bars represent 30 µm.

tained from the adult hippocampus after transfection with pP/Tα1:hGFP. Before attempting to isolate adult hippocampal progenitor cells, we needed to first establish that P/Tα1:hGFP could indeed identify them. We used four adult hippocampi for this; these included resected tissue from male patients 5, 20, 33 and 50 years old. All the samples were dissociated and plated, and randomly selected plates were transfected on day 1 in vitro with P/Tα1:hGFP plasmid DNA. Within 4 days after transfection, a discrete population of fiber-bearing, initially bipolar cells expressed GFP (Fig. 2). These cells expressed neuronal βIII-tubulin, and matured to express MAP-2 over the first 10 days in vitro. The P/Tα1:hGFP+ cells failed to express either glial fibrillary acidic protein (GFAP) or oligodendrocytic O4 (not shown), supporting the idea of neuronal specificity of the P/Tα1:hGFP selection cassette. Furthermore, the P/Tα1:hGFP+ cells included mitotically generated neurons; both BrdU+ and BrdU– Tα1:hGFP+ cells were confirmed as being neuronal by their co-expression of βIII-tubulin (Fig. 3).

a

b

c

d

e

f

g

h

i

Fig. 4 The nestin enhancer drives expression in nestin protein-positive adult hippocampal cells that divide and give rise to βIII-tubulin+ neurons. Left and middle columns, phase contrast (a,d,g) and fluorescence (b,e,h) images of E/nestin:EGFP+ hippocampal cells (middle column, green). Right column, expression of human nestin protein (c and f, red) and, in a matched culture, immunostaining for βIII-tubulin (i, red); all three cells had incorporated BrdU (blue) during their first week in culture, and were fixed at 7 d in vitro. Arrows, E/nestin:EGFP+ cells that incorporated BrdU, reflecting their division in vitro. Scale bar represents 50 µm. NATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000

a

b

c

d

Nestin enhancer also identifies mitotic hippocampus progenitors To identify potentially less-committed neural progenitor cells of the adult human hippocampus, we used an expression vector encoding GFP placed under the control of the nestin enhancer, composed of the second intron of the nestin gene20. The 637base-pair region between bases 1,162 and 1,798 of the rat nestin gene is sufficient to target gene expression to neuroepithelial progenitor cells27, and is conserved between rats and humans. It was placed upstream of the minimum promoter of heat shock protein-68 (ref. 28), a sequence that shows no basal activity unless an enhancer is placed in its vicinity; the resultant E/nestin:P/hsp68 construct was ligated to enhanced (E)GFP–polyA to yield the selection plasmid E/nestin:EGFP. The neural progenitor cell-specific expression of this transgene has been confirmed in transgenic mice (A.K. et al., manuscript submitted). This E/nestin:EGFP construct recognized a relatively primitive population of initially flat and bipolar hippocampal cells, which constituted 0.53 ± 0.20% of the cells assayed (n = 30 fields, 10 each from three patients). These cells were typically nestin+/TuJ1±/GFAP–/O4– at 1 day in vitro. In the week after transfection, 62.5 ± 2.9% of the E/nestin:EGFP cells developed TuJ1 immunoreactivity. Among these, 21.0 ± 15.2% incorporated BrdU during that week, indicating both their progression from a nestin to βIII-tubulin+ neuronal phenotype, and their mitogenesis in vitro (Fig. 4). Almost one-third of the E/nestin:EGFP+ cells remained morphologically undistinguished. Of these, a small number developed into GFAP+ astrocytes, but in the conditions we used, which included 2% PD-FBS in DMEM/F12/N2, almost all of the remainder continued to express nestin protein. Thus, E/nestin:EGFP identified adult hippocampal cells that were mitotically competent and able to generate new neurons. They seemed to constitute a neuronal progenitor population analogous to, if not identical to, the hippocampal cell ‘pool’ defined by P/Tα1:hGFP. FACS selection of Tα1:hGFP+ progenitor cells Using FACS, we next isolated P/Tα1:hGFP+ hippocampal cells and enriched them to purity, from 1-ml aliquots of dentate dissociates prepared from four patients. Manual counting of these cells on a hemocytometer yielded an average of 197,500 ± 72,169 dentate cells/sorted sample, whereas flow cytometry yielded 402,634 ± 205,833 cells/sample (all values, mean ± s.e.m.). The 273



ARTICLES

a

b

c

d

e

f

g

Fig. 5 Neural progenitors identified by P/Tα1:hGFP and E/nestin:EGFP can be isolated by FACS. a–d, P/Tα1:hGFP-based sorting of hippocampal dentate cells derived from two male patients, 5 (a and b) and 20 (c and d) years old. The cells were transfected with either P/Tα1:lacZ (a non-fluorescent control; a and c), or pP/Tα1:hGFP (b and d). An average of 1.36% and 1.69% of the cells derived from the 5- and 20-year-old patients, respectively, achieved an arbitrary threshold of P/Tα1:hGFP fluorescence intensity, which was calibrated to that achieved by 0.03% of control cells. e–h , Dentate gyrus cultures from the 5-year-old were transfected with either P/CMV:lacZ (a non-fluorescent control; e) or E/nestin:EGFP (f), and the cells were sorted. In this typical example, 1.69% of the cells sorted by E/nestin:EGFP achieved threshold fluorescence, which was calibrated to that achieved by 0.01% of control cells. g and h, Low-power magnification phase (g) and fluorescent (h) micrographs of the E/nestin:EGFP+ cells from f, 2 h after FACS; most sorted cells visibly expressed GFP. For all FACS, GFP fluorescence intensity (FL1) was plotted against cell size (forward scatter, FCS).

expressed nestin protein, and did not otherwise develop a differentiated phenotype in the week after FACS.

h

larger number obtained through flow cytometry likely included damaged cells and free nuclei, even after gating to avoid debris. The hemocytometer counts, in contrast, were limited to viable cells, as assessed morphologically and confirmed by trypan blue exclusion. Of the samples transfected with P/Tα1:hGFP, an average of 3,324 ± 1,113 cells were assayed as GFP+ by FACS; given the highstringency cut-off for fluorescence assignment that we used, essentially all of these could be confirmed visually as expressing GFP. Thus, the P/Tα1:hGFP+ isolates effectively included 1.11 ± 0.35% of the cell population assayed by hemocytometer (Fig. 5). During the weeks after FACS, these cells sorted by P/Tα1:hGFP matured into morphologically and antigenically characteristic neurons (Fig. 3). One week after FACS, we switched both the sorted plates and their unsorted controls from base media containing 2% PD-FBS and 10 ng/ml FGF-2, in which they were raised initially to allow progenitor division, to 5% FBS and 20 ng/ml brain-derived neurotrophic factor, to encourage neuronal differentiation and survival21. Fully 73.2 ± 6.6% of the sorted cells expressed βIII-tubulin immunoreactivity by a week after FACS, whereas only 14.5 ± 6.8% and 26.9 ± 1.9% of the cells in the unsorted dissociates did so at 7 and 14 days in vitro, respectively. Each of these comparisons yielded a significant increase in βIII-tubulin/TuJ1+ cells as a result of FACS based on P/Tα1:hGFP (P < 0.01, one-way ANOVA with Boneferroni adjustment) (Fig. 6). Among the sorted βIII-tubulin+ cells, 19.1 ± 0.2% incorporated BrdU during their initial week in low-serum culture supplemented with FGF-2, indicating that adult dentate cells sorted by P/Tα1:hGFP were still mitotically competent. Furthermore, of the TuJ1– cells in the sorted ‘pool’, essentially all 274

FACS selection of hippocampal progenitor cells We next used FACS to isolate E/nestin:EGFP+ cells from both juvenile and adult hippocampal dissociates, again using samples dissected free from their overlying ventricular epithelium (n = 3 male patients 5, 33 and 50 years old). As a result, the mostly nestin-positive ependymal/subependymal ‘pool’ of the ventricular zone29 would not have been present in these samples. The target population of the E/nestin:EGFP plasmid was instead intended to be the nestin-expressing cells of the adult human dentate gyrus, which were expected to include neural precursors just as in rodents13. Using 1-ml dissociates of adult dentate gyrus, an average of 405,768 ± 209,852 cells were counted as single events by the sorter in purification mode. Of 273,333 ± 156,950 viable dentate cells/sorted sample confirmed by hemocytometer, 11,331 ± 10,737, or 1.96 ± 1.26%, were gated in purification mode as E/nestin:GFP+ cells (Fig. 5). These cells sorted by E/nestin:EGFP+ initially expressed nestin protein ubiquitously. Within the week after FACS, however, 61.0 ± 7.6% of the ‘pool’ sorted by E/nestin developed βIII-tubulin immunoreactivity, whereas only 24.8 ± 10.9% of the cells in unsorted controls expressed βIII-tubulin at that point (P < 0.01, Student’s t-test) (Fig. 6). In the 5% FBS/brain-derived neurotrophic factor environment to which the cells were switched after FACS, most of the cells sorted by E/nestin:EGFP+ that were TuJ1– continued to express nestin; only rarely were glia found in the sorted ‘pool’, even at a week after FACS. Among the TuJ1+ cells sorted by E/nestin:EGFP, 21.7 ± 3.3% incorporated BrdU during their first week in culture. Thus, like their counterparts separated by Tα1:hGFP, the cells sorted by E/nestin:EGFP matured as neurons in the weeks after FACS. Response of progenitor-derived neurons to depolarizing stimuli To establish the ability of newly generated adult hippocampal cells to respond in a neuronal manner to depolarizing stimuli, we loaded selected cultures (n = 4, derived from 2 brains) with the calcium indicator dye fluo-3, and exposed them to 60 mM potassium during confocal microscopy. Astrocytic responses to depolarization were minimal in these culture conditions, as has been noted30. In contrast, neuron-like cells verified as such showed rapid, reversible increases in cytosolic calcium in response to potassium, consistent with the activity of neuronal voltage-gated calcium channels21,30 (Fig. 7). We then confirmed the neuronal phenotype of these cells antigenically, by immunostaining for NATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000


ARTICLES

Sodium currents in progenitor-derived neurons To assess the development of neuronal fast sodium currents by P/Tα1:hGFP-defined progenitor-derived neurons, we transfected adult hippocampal cultures (n = 3, from 2 patients) with P/Tα1:hGFP, and then assessed the GFP+ cells by whole-cell patch-clamp recording. We identified P/Tα1:hGFP+ cells by their persistent fluorescence, then patch-clamped and recorded in a voltage-clamped configuration during voltage steps (Fig. 7). There were voltage-activated sodium currents, which form the ionic basis of the neuronal action potential, in each of four P/Tα1:hGFP+ cells recorded. In these, the maximum sodium current was 617 ± 136 pA (n = 4). In contrast, none of 13 non-fluorescent cells assayed in these unsorted cultures, which included a variety of glial morphologies, showed substantial current-induced sodium current. Although voltage-gated sodium channels have been found in glia31,32, hippocampal astrocytes do not express these channels in sufficient numbers to mediate the fast sodium currents of neuronal depolarization33,34. Discussion Our results indicate that the adult human hippocampus contains mitotically competent progenitor cells that can give rise to new neurons. Our findings extend a recent report of hippocampal neurogenesis in tissue sections of adult human brain7, both by identifying viable hippocampal progenitor cells and by establishing a means for their specific extraction in a manner that allows their physiologically appropriate maturation. By transfecting hippocampal dissociates with plasmid DNA encoding GFP expressed under the control of two early neural regulatory sequences, the nestin enhancer and the Tα1 tubulin promoter, we were able to identify and sort neuronal progenitors from the native adult human hippocampus, in substantial numbers and purity. These cells remained mitotically competent after FACS, and matured as functionally competent neurons. To our knowledge, this is the first reported targeted selection and

% Tuj1+ cells


βIII-tubulin. Further immunolabeling showed that a distinct subpopulation of these βIII-tubulin+ cells had incorporated BrdU, indicating their in vitro mitogenesis.

Unsorted E/nestin:EGFP

Unsorted

P/Ta1hGFP

Selection plasmid

Fig. 6 FACS based on P/Tα1:hGFP and E/nestin:EGFP each enrich neuronal progenitor cells from the adult hippocampus. Using stringent FACS criteria intended for cell-type purification, the percentage of TuJ1+ cells increased from 26.9 ± 1.9% in unsorted cultures (filled bars) to 73.2 ± 6.6% by 1 week after FACS based on P/Tα1:hGFP (n = 3 runs; shaded bar, right), and from 24.8 ± 10.9 to 61.0 ± 7.6% after FACS based on E/nestin:EGFP (shaded bar, left). Each sort yielded significant enrichment of TuJ1+ cells (P < 0.01, after Boneferroni adjustment). Essentially all of the P/Tα1:hGFP+/TuJ1– cells, and most of their E/nestin:EGFP+/TuJ1– counterparts, expressed nestin protein. NATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000

extraction of mitotically competent neuronal progenitor cells from any nonventricular region of the adult human brain or spinal cord. The sortings based on P/Tα1:hGFP and E/nestin:EGFP produced similar results, and yielded an unexpected abundance of both phenotypes, even in high-stringency sorting conditions. The average transfection efficiency of 9.40 ± 0.9% indicates that our cytometric yields of P/Tα1:GFP and E/nestin:EGFP-defined progenitors, 1.11 ± 0.35% and 1.96 ± 1.26% respectively, may reflect much larger ‘pools’ of potentially neurogenic cells in the adult human dentate gyrus. These estimates are substantially higher than the incidence of human hippocampal neurogenesis indicated by histological sections7, indicating that the dentate gyrus has a relatively large ‘pool’ of mitotically quiescent progenitor cells. However, our numbers may include dentate cells capable of resuming an early neuronal transcriptional program when removed to a permissive culture environment, without cell cycle re-entry; these may exceed the actual number of dentate progenitor cells that can divide to give rise to new neurons in vivo. Although nestin and Tα1 tubulin are expressed at different stages in neuronal ontogeny (nestin, by uncommitted as well as phenotype-restricted progenitors19,35; and Tα1 tubulin, by their neuronally restricted daughters17), their promoters seemed to recognize mostly overlapping cell types in the adult hippocampus. Yet this is not unexpected in the relatively homogeneous cell environment of the dentate gyrus: The astrocytic component of the dentate is minor, and oligodendrocytes are scarce in the granular layer. As a result, the nestin gene may be transcribed by about the same ‘pool’ of granule cell neuronal progenitors in the adult hippocampus as the Tα1 tubulin gene. The typical fate of E/nestin-defined hippocampal cells was neuronal in the conditions we used. Nonetheless, both neuronally committed and multipotent progenitors may be recognized by E/nestin:EGFP (ref. 35). We made no attempt at further lineage analysis of the E/nestin:EGFP+ cells, or to assess their clonal derivatives, as our main purpose was to isolate neuronal progenitors from the adult hippocampus. It is possible that some of these may have been persistent neural stem cells13,36; this will require further study of the lineage potential and self-renewability of these cells. The therapeutic potential of adult hippocampal neurogenesis may be profound. The existence of a mitotic progenitor ‘pool’ in the human dentate indicates that neuronal recruitment to this ‘pool’ might be modulated pharmacologically or experientially in adulthood, as in rodent models6,9,10. Just as learning is associated with increased hippocampal neurogenesis in rodents11, the targeted induction or supplementation of this progenitor ‘pool’ might directly influence learning and memory in humans. Our ability to target and extract adult hippocampal progenitor cells in substantial numbers should permit hitherto unapproachable studies of their mitotic capacity, growth control, maturation competence and engraftability. These studies in turn should provide us new insights into the function of these cells in the normal adult human, their potential for exogenous activation and induced neurogenesis, and their utility as agents for hippocampal reorganization and repair. Methods Adult human hippocampal dissociation and culture. Adult human brain tissue was obtained during temporal lobectomy, as described21,30. The hippocampus was dissected free from the ventricular surface, and the dentate gyrus was then dissected from the rest of the tissue. This sample was cut 275



ARTICLES

a

b

c

d

e

f

Fig. 7 Hippocampal cells sorted by P/Tα1:hGFP develop into physiologically mature neurons. a–d, Progenitors sorted by P/Tα1:hGFP develop neuronal calcium (Ca2+) responses to depolarization. Cells sorted by P/Tα1:hGFP, loaded with the calcium indicator dye fluo-3, 10 d after FACS show uniform maturation into fiber-bearing cells of neuronal morphology. Same field in a after exposure to glutamate (b), after return to baseline after media wash (c) and after exposure to a depolarizing stimulus of 100 mM KCl (d). The neurons showed rapid, reversible increases of more than 300% in cytosolic calcium in response to potassium, consistent with the

activity of neuronal voltage-gated calcium channels. Scale bar represents 50 µm. e and f, Whole-cell patch-clamp shows voltage-gated sodium currents in P/Tα1:hGFP+ dentate neurons. A representative cell 14 d after FACS based on P/Tα1:hGFP, identified visually as a progenitor-derived neuron on the basis of its residual GFP expression (e), was patch-clamped in a voltage-clamped configuration, and its responses to voltage steps (upper lines, in mV) were recorded (f). The fast negative deflections after depolarized voltage steps (f) are typical of the voltage-gated sodium currents of mature neurons.

into pieces of about 2 mm on each edge, rinsed twice with PIPES solution (120 mM NaCl, 5 mM KCl, 25 mM glucose and 20 mM PIPES), and dissociated in papain/DNase, as described37,38. The dispersed cells were collected and rinsed with DMEM/F12/N2 containing 20% PD-FBS (Cocalico, Reamstown, Pennsylvania) to stop the enzymatic dissociation, and resuspended at a concentration of 1 × 107 cells/ml in DMEM/F12/N2 with 2% PD-FBS and 10 ng/ml FGF-2 (Sigma). The cells were plated at a concentration of 0.1 ml/dish into 35-mm Falcon Primaria plates coated with 2 µg/cm2 laminin, at 37 °C in 5% CO2. Then, 4 h later, an additional 0.7 ml DMEM/F12/N2, 2% PD-FBS and 10 ng/ml FGF-2 were added to each plate.

cells after 5–7 days in vitro by liposomal transfection, using lipofectin in Opti-MEM (Life Technologies) as described18. After a 6-hour transfection period, the reactions were terminated by adding 10% PD-FBS in DMEM/F12/N2. Then, 2 h later, the cells were returned to serum-free DMEM/F12/N2 with 10 ng/ml basic fibroblast growth factor. Images of GFP were obtained using an Olympus IX70 microscope. GFP expression was maximum 6–7 d after transfection, and the cells were sorted at that time. Our net transfection efficiency, with P/CMV:lacZ as positive control, averaged 9.40 ± 0.9% (n = 40 fields in four plates, derived from two patients).

BrdU labeling and immunocytochemistry. Cultures were exposed continuously to 10 µg/ml BrdU beginning 1 d after plating. Cells were fixed with 4% paraformaldehyde after 7, 14 or 28 d in vitro. They were immunostained first for BrdU, and then for either nestin, βIII-tubulin or MAP-2. βIIItubulin was detected using monoclonal antibody TuJ1 (from A. Frankfurter), and MAP-2 by a rabbit antibody against MAP-2 (from S. Halpain), each as described23,39. Nestin was detected using rabbit antibody against human nestin (1:1,000 dilution; from U. Lendahl), and detected using biotinylated antibody against rabbit IgG (1:200 dilution) and Texas Red–avidin (Vector Laboratories, Burlingame, California).

Flow cytometry and sorting. Flow cytometry and sorting of hGFP+ cells was done on a FACS Vantage (Becton Dickinson, San Jose, California). Cells were washed with calcium- and magnesium-free Hank’s balanced salt solution, then dissociated for 5 min at 37 °C in 0.05% trypsin–EDTA. The reaction was terminated by the addition of DMEM/F12/N2 with 10% FBS. The cells (2 × 106/ml) were then analyzed by light forward- and right-angle(side-) scatter, and assessed for GFP fluorescence through a bandpass filter of 510 ± 20 nm, as the cells traversed the beam of a Coherent INNOVA Enterprise II Ion Laser (488 nm; 100 mW). Sorting was done using a purification-mode algorithm. Cells transfected with E/nestin:lacZ were used as a control to set the background fluorescence; a false-positive rate of 0.1–0.3% was accepted to ensure adequate yield. Cells transfected with either E/nestin:EGFP or P/Tα1:hGFP were sorted at a rate of 1,000–3,000 cells/s. GFP+ cells were then plated onto laminin-coated, 24-well plates, in DMEM/F12/N2 with 5% FBS and BrdU. At 2 and 7 days after FACS, the sorted cultures were fixed and stained for BrdU, with either TuJ1/ βIII tubulin or MAP-2; selected plates were also stained for O4 or GFAP as described37.

Construction of plasmids P/Tα1:hGFP and E/nestin:EGFP. Expression vectors were constructed that encoded GFP, under the control of either the Tα1 tubulin promoter, or the nestin enhancer. First, hGFP (ref. 40) was placed under the control of the Tα1 tubulin promoter to yield pP/Tα1:hGFP as described18 (P/tα1 plasmid from F. Miller). Next, EGFP was placed under the control of the nestin enhancer to generate E/nestin:EGFP. For this, the enhancer element of the second intron of the rat nestin gene, spanning bases 1,162–1,798 (ref. 27), was placed upstream of the minimum promoter of heat shock protein-68 (hsp68; ref. 28); the resultant E/nestin:P/hsp68 construct was ligated to EGFP-polyA (Clontech, Palo Alto, California), to yield E/nestin:EGFP. The construction and neuroepithelial expression of this transgene will be reported (A.K. et al., manuscript submitted). Transfection. The plasmid constructs were introduced into the cultured 276

Data analysis. Experimental endpoints included the proportion of cells expressing each antigenic marker (all nominally GFP+ after sorting) as a function of time after FACS. At each time of sampling, the proportions of antigenically defined neurons in the sorted cultures were compared with those of unsorted control cultures that were similarly dispersed, but replated without sorting, after adjusting their cell density to that of the sorted ‘pool’ after FACS. For each combination of treatment (sorted or unsorted) NATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000


ARTICLES


and immunolabeling, the number of stained and unstained cells were counted in 10 randomly chosen fields, in each of three triplicate cultures. Calcium imaging. For identification of neurons physiologically, selected fields were challenged with a depolarizing stimulus of 60 mM potassium, during which their cytosolic calcium levels were measured. Calcium imaging used confocal microscopy of cultures loaded with fluo-3 acetoxymethylester (fluo-3; Molecular Probes, Eugene, Oregon), as described21,30. A BioRad MRC600 confocal scanning microscope, equipped with an argon laser and coupled to a Nikon Diaphot 300 microscope, was used to obtain images of the fluo-3 signal. These neurons show a mean calcium increase of more than 400% to 60 mM potassium in vitro; this is in contrast to an astrocytic calcium response of less than 20%, and undetectable oligodendroglial responses30. Here, we assigned neuronal identity to cells showing depolarization-induced increases in calcium of 300% or more. Electrophysiology. Images of P/Tα1:GFP+ fluorescent neurons were first obtained using Imaging Workbench (Axon Instruments, Burlington, California). Excitation was provided by 480-nm monochromated light (DeltaSCAN-1, PTI). Emission was low-pass-filtered (515 nm). An intensified CCD video camera (IC-100, PTI, South Brunswick, New Jersey) (ref. 34) coupled to an Olympus BX50 microscope was used to obtain images. Whole-cell voltage-clamped recordings (J.K. et al., 1998) were then made of the labeled neurons, using differential inference contrast optics at 23–24 °C. Patch electrodes with a resistance of 4–7 MΩ were pulled from KG-33 glass capillaries (inside diameter, 1.0 mm; outside diameter, 1.5 mm; Garner Glass, Claremont, California), using a P-97 electrode puller (Sutter Instrument, Novato, California). Any seal resistance less than 5 GΩ was rejected. The pipette solution contained 120 mM K-gluconate, 10 mM KCl, 1 mM MgCl2, 10 mM HEPES, 0.1 mM EGTA, 0.025 mM CaCl2, 1 mM ATP, 0.2 mM GTP and 4 mM glucose, pH 7.2. The extracellular solution contained 130 mM NaCl, 3 mM KCl, 1.25 mM NaH2PO4, 2 mM MgCl2, 2 mM CaCl2 and 10 mM glucose. A holding potential of –60 mV and voltage ‘steps’ of 10 mV with 100-ms durations were applied to the recorded cells through the patch electrodes. Recordings were made using Axopatch 200B and PCLAMP 7.0. The software was used to subtract capacitative and/or linear leak currents. Signals were sampled every 50 µs. Data were further processed with Origin 4.1 (Microcal, Northhampton, Massachusetts) and CorelDraw 7.0 (Corel, Ontario Canada). Acknowledgments We thank H.M. Keyoung for comments on the manuscript. This work was supported by the National Multiple Sclerosis Society, the Aitken Charitable Trust, the G. Harold and Leila Y. Mathers Charitable Foundation, the Human Frontiers Scientific Program, and National Institute of Neurological Disorders and Stroke grants R01NS29813 and R01NS33106.

RECEIVED 12 AUGUST; ACCEPTED 13 DECEMBER 1999 1. Altman, J. & Das, G.D. Autoradiograpbhic and histological evidence of postnatal hippocampal neurogenesis in rats. J. Comp. Neurol. 124, 319–335 (1965). 2. Kaplan, M.S. & Hinds, J.W. Neurogenesis in the adult rat: electron microscopic analysis of light radioautographs. Science 197, 1092–1094 (1977). 3. Bayer, S., Yackel, J. & Puri, P. Neurons in the rat dentate gyrus granular layer substantially increase during juvenile and adult life. Science 216, 890–892 (1982). 4. Barnea, A. & Nottebohm, F. Seasonal recruitment of hippocampal neurons in adult free-ranging black-capped chickadees. Proc. Natl. Acad. Sci. USA 91, 11217–11221 (1994). 5. Gould, E., McEwen, B., Tanapat, P., Galea, L. & Fuchs, E. Neurogenesis in the dentate gyrus of the adult tree shrew is regulated by psychosocial stress and NMDA receptor activation. J. Neurosci. 17, 2492–2498 (1997). 6. Gould, E., Tanapat, P., McEwen, B., Flugge, G. & Fuchs, E. Proliferation of granule cell precursors in the dentate gyrus of adult monkeys is diminished by stress. Proc. Natl. Acad. Sci. USA 95, 3168–3171 (1998). 7. Eriksson, P. et al. Neurogenesis in the adult human hippocampus. Nature Med. 4, 1313–1317 (1998). 8. Gould, E., Cameron, H., Daniels, D., Wooley, C. & McEwen, B. Adrenal hormones suppress cell division in the adult rat dentate gyrus. J. Neurosci. 12, 3642–3650 (1992). 9. Kempermann, G., Kuhn, H. & Gage, F. More hippocampal neurons in adult mice

NATURE MEDICINE • VOLUME 6 • NUMBER 3 • MARCH 2000

living in an enriched environment. Nature 386, 493–495 (1997). 10. van Praag, H., Kempermann, G. & Gage, F. Running increases cell proliferation and neurogenesis in the adult mouse dentate gyrus. Nature Neurosci. 2, 266–270 (1999). 11. Gould, E., Beylin, A., Tanapat, P., Reeves, A. & Shors, T. Learning enhances adult neurogenesis in the adult hippocampal formation. Nature Neurosci. 2, 260–265 (1999). 12. Kornack, D. & Rakic, P. Continuation of neurogenesis in the hippocampus of the adult macaque monkey. Proc. Natl. Acad. Sci. USA 96, 5768–5773 (1999). 13. Palmer, T., Takahashi, J. & Gage, F. The adult rat hippocampus contains primordial neural stem cells. Molec. Cell Neurosci. 8, 389–404 (1997). 14. Kukekov, V. et al. Multipotent stem/progenitor cells with similar properties arise from two neurogenic regions of adult human brain. Exp. Neurol. 156, 333–344 (1999). 15. Suhonen, J., Peterson, D., Ray, J. & Gage, F. Differentiation of adult hippocampus-derived progenitors into olfactory neurons in vivo. Nature 383, 624–627 (1996). 16. Miller, F., Naus, C., Durand, M., Bloom, F. & Milner, R. Isotypes of α-tubulin are differentially regulated during neuronal maturation. J. Cell Biol. 105, 3065–3073 (1987). 17. Gloster, A. et al. The Tα1 α-tubulin promoter specifies gene expression as a function of neuronal growth and regeneration in transgenic mice. J. Neurosci. 14, 7319–7330 (1994). 18. Wang, S. et al. Identification and enrichment of forebrain neuronal precursor cells by fluorescence-activated sorting of ventricular zone cells transfected with GFP regulated by the Tα1 tubulin promoter. Nature Biotechnol. 16, 196–201 (1998). 19. Frederiksen, K. & McKay, R.D. Proliferation and differentiation of rat neuroepithelial precursor cells in vivo. J. Neurosci. 8, 1144–1151 (1988). 20. Zimmerman, L. et al. Independent regulatory elements in the nestin gene direct transgene expression to neural stem cells and muscle precursors. Neuron 12, 11–24 (1994). 21. Pincus, D.W. et al. FGF2/BDNF-associated maturation of new neurons generated from adult human subependymal cells. Ann. Neurol. 43, 576–585 (1998). 22. Lee, M., Rebhun, L. & Frankfurter, A. Posttranslational modification of class III βtubulin. Proc. Natl. Acad. Sci. USA 87, 7195–7199 (1990). 23. Menezes, J.R. & Luskin, M.B. Expression of neuron-specific tubulin defines a novel population in the proliferative layers of the developing telencephalon. J. Neurosci. 14, 5399–5416 (1994). 24. Bernhardt, R. & Matus, A. Light and electron microscopic studies of the distribution of microtubule-associated protein 2 in rat brain: a difference between dendritic and axonal cytoskeletons. J. Comp. Neurol. 226, 203–221 (1984). 25. Chalfie, M., Tu, Y., Euskirchen, G., Ward, W. & Prasher, D. Green fluorescent protein as a marker for gene expression. Science 263, 802–805 (1994). 26. Wang, S., Roy, N., Benraiss, A., Harrison-Restelli, C. & Goldman, S. Promoterbased isolation and purification of mitotic neuronal progenitor cells from the adult mammalian ventricular zone. Dev. Neurosci. 22, 167–176 (2000). 27. Lothian, C. & Lendahl, U. An evolutionarily conserved region in the second intron of the human nestin gene directs gene expression to CNS progenitor cells and to early neural crest cells. Eur. J. Neurosci. 9, 452–462 (1997). 28. Rossant, J., Zirngibl, R., Cado, D., Shago, M. & Giguere, V. Expression of a retinoic acid response element-hsp/lacZ transgene defines specific domains of transcriptional activity during mouse embryogenesis. Genes Dev. 5, 1333–1344 (1991). 29. Roy, N. et al. Promoter-targeted selection and isolation of neural progenitor cells from the adult human ventricular zone. J. Neurosci. Research 59, 321–331 (2000). 30. Kirschenbaum, B. et al. In vitro neuronal production and differentiation by precursor cells derived from the adult human forebrain. Cerebral Cortex 4, 576–589 (1994). 31. Sontheimer, H., Mintuen, J.E., Black, J.A., Ransom, B. & Waxman, S. Two types of Na(+)-currents in cultured rat optic nerve astrocytes: changes with time in culture and with age of culture derivation. J. Neurosci. Res. 30, 275–287 (1991). 32. Sontheimer, H., Black, J.A., Ransom, B.R. & Waxman, S. Ion channels in spinal cord astrocytes in vitro. I. Transient expression of high levels of Na+ and K+ channels. J. Neurophysiol. 68, 985–1000 (1992). 33. Tse, F.W., Fraser, D.D., Duffy, S. & MacVicar, B.A. Voltage-activated K+ currents in acutely isolated hippocampal astrocytes. J. Neurosci. 12, 1781–1788 (1992). 34. Kang, J., Jiang, L., Goldman, S. & Nedergaard, M. Astrocyte-mediated potentiation of inhibitory synaptic transmission. Nature Neurosci. 1, 683–692 (1998). 35. Lendahl, U., Zimmerman, L.B. & McKay, R.D. CNS stem cells express a new class of intermediate filament protein. Cell 60, 585–595 (1990). 36. McKay, R. Stem cells in the central nervous system. Science 276, 66–71 (1997). 37. Roy, N. et al. Identification, isolation and enrichment of oligodendrocyte progenitor cells from the adult human subcortical white matter. J. Neurosci. 19, 9986–9995 (1999). 38. Doetsch, F., Caille, I., Lim, D., Garcia-Verdugo, J. & Alvarez-Buylla, A. Subventricular zone astrocytes are neural stem cells in the adult mammalian brain. Cell 97, 703–716 (1999). 39. Barami, K., Iversen, K., Furneaux, H. & Goldman, S. A. Hu protein as an early marker of neuronal phenotypic differentiation by subependymal zone cells of the adult songbird forebrain. J. Neurobiol. 28, 82–101 (1995). 40. Levy, J., Muldoon, R., Zolotukhin, S. & Link, C. Retroviral transfer and expression of a humanized, red-shifted green flourescent protein gene into human tumor cells. Nature Biotechnol. 14, 610–614 (1996).

277