Concurrent isolation and characterization of oligodendrocytes, microglia and astrocytes from adult human spinal cord

073h-574X/93 $h.(XJ+o.tXJ PL!rgamon Press Ltd 0 lYY3 ISDN

CONCURRENT ISOLATION AND CHARACTERIZATION OF OLIGODENDROCYTES, MICROGLIA AND ASTROCYTES FROM ADULT HUMAN SPINAL CORD Scorr *The Miami Project

R. WHI~TEMORE,*~‘$~HENRY R. SANON* and PATRICK M., and Departments

of tNeurologica1 Surgery and $Physiology and Biophysics. School of Medicine. Miami, FL 33136, I7.S.A.

WOOD*? llniversity

of Miami

(Received IS June 1993: accepted 7 July 199.1)

Abstract-A cellular preparation of highly enriched oligodendrocytes was obtained from adult human spinal cord by Percoll gradient centrifugation followed by either differential adhesion or fluorescence-activated cell sortingafter immunostaining with an antibody against galactocerebroside (01). The adherent and 01-negative cell fractions were >96% microglia. The non-adherent and 01-positive fractions were >96% positive for the oligodendrocyte markers 04 and 01, @-2% positive for glial fibrillary acidic protein, and were devoid of neuronal or microglial markers. If the oligodendrocyte fraction was co-cultured with purified dissociated rat dorsal root ganglion neurons, the oligodendrocytes adhered to the axons and their numbers increased over a 4 week period. However, myelin sheaths were not produced around axons in these cultures. In contrast, if the oligodendrocyte cell fraction was grown alone in culture for >3 weeks, the number of oligodendrocytes decreased and a layer of astrocytes developed underneath the oligodendrocytes. The oligodendrocytes could be eliminated from these cultures by subsequent passaging, thus producing cultures of pure astrocytes. The astrocytes accumulated both K+ and glutamate with kinetic properties similar to those reported for rodent astrocytes. We suggest that these astrocytes arose in part from an 04/01-positive precursor which did not initially express glial fibrillary acidic protein. These results define a relatively simple method by which highly enriched populations of oligodendrocytes, astrocytes and microglia can be obtained from adult human spinal cord. Key words: human

CNS. spinal cord, glia.

To study the function of CNS glia, methods have been developed to isolate highly enriched populations of neonatal rat astrocytes21 rat microglia,9 and both neonatal29 and adult30,36 rat oligodendrocytes. Recent evidence would suggest, however, that the biology of primary adult human cells in culture can be different from other non-human vertebrates. For example, adult human, monkey and rodent Schwann cell proliferation and myelination are clearly regulated by distinct factors.22 Moreover, human and rodent fibroblasts differ significantly in their ability to be immortalized, suggesting intrinsic differences in their cell cycle contro1.28%31,42We observed differences between adult rat and human oligodendrocytes both in their growth requirements as well as their ability to be stably infected with viral oncogenes (unpublished results). To better understand the biology of adult human glial cells, we have adapted reported purification procedures to allow concurrent enrichment of highly purified populations of oligodendrocytes, microglia and astrocytes from adult human spinal cord. Initial studies showed that the human oligodendrocytes would associate with neurites from embryonic rat DRG neurons and that the astrocytes accumulated extracellular K+ and had a Na+-dependent, high affinity glutamate uptake mechanism. The ability to isolate adult human glial cells should facilitate studies of their specific physiological properties. EXPERIMENTAL

PROCEDURES

Materials

Primary antisera raised against the indicated antigens were obtained from the following sources: galactocerebroside (GalC) - Barbara Ranscht, La Jolla Cancer Institute; rabbit polyclonal glial EiAuthor to whom correspondence shouldbe addressed at: The Miami Project, University of Miami School of Medicine. 1600 NW 10th Avenue (R-48). Miami, FL 33136, U.S.A. Abbreviations: bFGF, basic fibroblast growth factor; D/F, Dulbecco’s modified Earle’s medium/Ham’s F-12 medium (50% v/v); DiI-ac-LDL, DiI conjugated acetylated low density lipoprotein; DRG, dorsal root ganglion; FACS, fluorescence-activated cell sorting; FBS, fetal bovine serum; GFAP, glial fibrillary acidic protein: HS. horse serum; IGF-I, insulin-like growth factor I; L-15, Leibowitz’s L-15 medium; PDGF, platelet-derived growth factor. 7.55

756

\. K. Whittemorc

P/ trl.

fibrillary acidic protein (GFAP) .- Incstar (Stillwater, MN); mouse monoclonal GFAP 17’ED I3 1 Eldon Geisert Jr. University of Alabama: 01 and 04 - Melitta Schachner. University 01 Heidelberg; 5-bromo-2’-deoxyuridine - Becton Dickinson (San Jose, CA). Dil-conjugated acetylated low density lipoprotein (Dir-ac-LDL) was purchased from Biomedical Technologies, Inc. (Stoughton, MA). fluorescein-conjugated secondary antibodies from Cappel-Organon Teknika (Malvern, PA), fetal bovine serum (FBS) from Hyclone Laboratories, Inc. (Logan, UT). Basic fibroblast growth factor (bFGF) was generously provided by Judith Abraham, Scios, Inc. (Mountain View, CA) and platelet-derived growth factor (PDGF) was purchased from Amgen (Thousand Oaks, CA). Powdered media, enzymes and all other chemicals were obtained from Sigma (St Louis. MO). --

Cell culture

Adult human spinal cords were aseptically obtained from organ donors with the cooperation of The Organ Procurement Team of the University of Miami School of Medicine and the Medical Examiners Offices of Dade, Broward, Palm Beach, Collier, Monroe and St Lucie Counties. Spinal cords were removed from donors ranging from 14-52 years old and immediately placed in RPM1 medium at 0°C. Post mortem times between clamping of the aorta and placing the spinal cord into ice in RPM1 medium varied between 20 and 45 min. The time between removal of the spinal cord and initiation of the glial fraction isolation ranged between 30 min and 16 hr, but the spinal cords remained on ice in RPM1 during this interval. A glial fraction was prepared by Percoll gradient centrifugati0n.l” Briefly, spinal cords were dissected free of meninges, sliced into small fragments and incubated for 90 min in Earl’s balanced salt solution containing 0.25% trypsin and 50 kg/ml DNAse I. Following incubation, the digestion was arrested with horse serum, centrifuged and the tissue was dissociated in Leibowitz’s L-15 medium (L-15) containing 10% horse serum by trituration with a reduced bore pipette (0.5 mm). The dissociated cells were mixed 60:40 (v:v) with 80% Percoll containing 0.25 M sucrose and 20 mM sodium phosphate buffer (pH 7.4). After centrifugation at 31,ooO g for 45 min, the band containing the glial cells (above the red blood cell band and below the band of myelin debris) was removed, diluted with L-15, centrifuged and resuspended in growth media. A purified oligodendrocyte fraction was obtained either by fluorescence-activated cell-sorting (FACS) after labelling with 01 antibody on a Becton Dickinson FACScan equipped with FACStar PLUS research software6j4’ or by differential adhesion to cell culture plastic overnight. K2’ Gently swirling the plates 3-4 times was sufficient to dislodge any non-adherent cells not firmly attached. The plates were then gently rewashed with fresh medium and the two non-adherent fractions combined. The Ol-positive/nonadherent and 01-negative/adherent cell populations were nearly pure oligodendrocytes and microglia, respectively (see below). Oligodendrocyte fractions were grown at a density of 3x104 cells/cm2 on poly-L-lysine coated dishes at 37°C in humidified 5% CO2-95% air in 50% Dulbecco’s modified Earle’s medium(SO% Ham’s F-12 (D/F) containing 2.5% heat-inactivated FBS, 20 &ml insulin, and 5 @ml bFGF. Alternatively, oligodendrocytes weregrown in B16 medium5 containing 2.5% heat-inactivated FBS and 5 ng/ml bFGF. B16 medium allowed better survival of differentiated ohgodendrocytes. All media contained 100 U/ml penicillin and 100 pg/ml streptomycin. Oligodendrocytes cultured in the absence of bFGF failed to survive. In contrast, oligodendrocytes could be co-cultured with neurons for at least 5 weeks in the absence of bFGF (see below). When grown in culture for characterization, the non-adherent fraction and the 01-negative cell population from the FACS enrichment were grown in the D/F-based medium. Purified DRG neuron cultures were prepared as described,18 except that the neurons were cultured on 12 mm glass coverslips pre-coated with pnly-L-lysine and 1amini.n. The purified human oligodendrocytes (50,000 cells/cover&p) were added when the neuronal cultures were 3 weeks old. To generate populations of primary adult spinal cord astrocytes, enriched oligodendraeyte preparations were grown in D/F-based medium for longer than 3 weeks. The number of oligodendrocytes gradually decreased, and a confluent layer of astrocytes developed under the remaining oligodendrocytes. The astrocytes were paasaged when confluent and the remaining oligodendrocytes did not survive, resulting in a pure population of astroeytes (see below).

Purification of adult human spinal cord glial cells

757

Immunohistochemistry

GFAP, 01 and 04 immunoreactivities were determined as described previously.33 Microglia were detected by incubating live cultures overnight with 1 p,g/ml DiI-ac-LDL,l’ which can be detected with rhodamine filters. Some of these cultures were subsequently immunohistochemitally stained for 04,Ol or GFAP expression using fluorescein-conjugated secondary antibodies. Other cultures were double-stained for GFAP and 01 expression using a rabbit polyclonal GFAP antibody with anti-rabbit fluorescein-conjugated secondary antibodies and the monoclonalO1 antibody with rhodamine-conjugated anti-mouse secondary antibodies. Non-specific immunofluorescence was always assessed in parallel using the appropriate species-specific immunoglobulins. Following the reactions, cells were photographed with a Zeiss IM inverted microscope. Co-expression of oligodendrocyte

and astrocyte antigens in the non-adherent fraction

To assess the purity and analyse potential co-expression of glial antigens in the non-adherent cell fraction, cells were double-stained in suspension either with 04 and GFAP antibodies or 01 and 04 antibodies. In the former case, cells were rinsed with L-15 containing 10% heat-inactivated horse serum (L-15/10% HS) (4°C) and then incubated for 30 min at 4°C in undiluted medium conditioned by the 01 hybridoma cells containing 0.1% NaN3. After three rinses at 4°C with L- 1500% HS, cells were incubated with rhodamine-conjugated anti-mouse antibodies in L-l 5110% HS, rinsed as above and fixed with 4% paraformaldehyde in PBS for 5 min. Following two rinses with PBS, one with PBS containing 0.2% Triton-X 100, and one with L-15/10% HS, cells were incubated at 4°C for 30 min with GFAP antibodies diluted 1:lO in L-15/10% HS. After three rinses with L-15/10% HS, the cells were incubated with fluorescein-conjugated anti-rabbit secondary antibodies. To distinguish between 01 and 04 immunoreactivities, cells were initially incubated with 01 antibodies and fluorescein-conjugated anti-mouse secondary antibodies as described for 04, except that the cells were not fixed following the second antibody step. Instead. the cultures were subsequently incubated with 04 antibodies and rhodamine-conjugated anti-mouse antibodies and fixed as described above. Cells were mounted on glass microscope slide in Citifluor containing 0.1 ~.LMHoechst dye, coverslipped and visualized under UV, fluorescein and rhodamine optics. Only those cells in which an intact nucleus (Hoechst dye) could be unambiguously identified were subsequently scored for 04, 01 and/or GFAP expression. Glutamate uptake assays

Na+-dependent glutamate uptake assays were modified from Hertz et al.” Cells grown in 24 well plates were washed twice with either ice cold Na-Ringers (150 mM NaCl, 3 mM KCl, 1 mM CaCl2,0.6 mM MgCl2,1.7 mM KH2PO4,8 mM Na2HP04,6 mM glucose) or Ch-Ringers in which 150 mM choline chloride was substituted for NaCl. Cells were then incubated for 30 min at 37°C in the respective Ringer’s solutions containing concentrations of L-glutamate ranging from lop6 to 5~10~~ M. Cells were transferred to an ice-water bath, and the respective media replaced with 1 ml ice-cold medium containing 1 &i/ml 3H-glutamate. Following an incubation at 37°C for 8 min, the cells were washed twice with the appropriate ice cold Ringers without labelled glutamate, solubilized in 1N NaOH, neutralized with 1N HCl, and counted by liquid scintillation for “H-glutamate uptake. Cells at each glutamate concentration were analysed in triplicate and high affinity, Na+-dependent uptake was defined at the difference between Na- and Ch-Ringers. Three independent preparations of cells were analysed and kinetic constants (Km and I/m,& determined by double-reciprocal plots and linear regression analysis. K+ uptake assays

Potassium uptake assays, measured with the potassium analog s6Rb, were modified from Tas et a1.34 Culture media from the astrocytes, grown in 24 well cell culture clusters, was aspirated and

the cells rinsed twice and equilibrated in a physiological salt solution (PSS), comprised of 145 mM NaCl, 3 mM KCl, 1 mM CaCl2,0.6 mM MgCl2,l mM Na2HP04,6 mM glucose and 15 mM HEPES (pH 7.4) for 30 min at 37°C. This solution was aspirated and replaced with 1.0 ml of PSS containing 4.5 PCi of 86RbCl. After incubation for 2,5,10 or 15 min, the cultures were placed in an ice-water

bath and quickly rinsed six times with ice-cold II.29 M sucrose, 10 mM Tris-HCI (pH 7.4). Cells were Iysed with 1 ml of 1 N NaOH, and the lysate cent~fuged at 13,000g for 5 min at 22°C’. Aliquots of the supernatants were assessed for 86Rb by gamma radiation counting. Three independent preparations were analysed in triplicate, and rates of uptake were determined from the linear portion of the respective uptake curves.

RESULTS

The suspensions of adult human glial cells obtained by Percoll gradient centrifugation consisted of roughly 50% oligodendrocytes and 50% microglia, as determined by immunostaining (Fig. 1A, and below), These cell types couid be further enriched to near homogeneity by FACS or by differential adhesion. Two independent preparations ( 1 - sacral and lower lumbar spinal cord from a 39-year-old male who suffered a traumatic head injury, and 2 - sacrai and lower lumbar spinaf cord from a 35-year-old female who suffered a cerebral vascular accident) were enriched by FACS using the 01 antibody. The per cent of 01-positive cells obtained in these preparations was 97.5and 96.3%)and represented 482 and 47.5% of the total initial cell pop~at~on, respectively (see also Fig. 1). The composition of the sorted 01-positive and 01-negative populations was analysed by assessing 01 and GFAP immunoreactivity and phagocytosis of DiI-ac-LDL, 5 days after plating. The sorted 01-positive population was nearly 100% positive for 01 and did not bind DiI-ac-LDL (Fig. lB), and a few scattered cells expressed GFAF’ immunoreactivity in some preparations (data not shown). The 01-negative sorted population was predominantly DiI-ac-LDL-positive microglia (Fig. IC), although occasional Ol-positive OIigodendrocyteswere observed (data not shown). Due to the unpr~ictable procurement of human spinal cords from organ donors, it was not practical to routinely use the FACS equipment at the University of Miami Flow Cytometery

sp‘““L

L”lU

‘Gus,

vy

KC,‘“11

~,iauicm

~~,~,‘,,~~~~~“~,

Iabelled with fluorescein conjugated 01 antibody and enriched by fluorescke-activated‘& sorl (FACS). The unsorted (A), 01-positive (B) and 01-negative (C) cellular fractions were grown for 5 days at which time the cultures were double-stained for the presence of 01-~unorea~ivjty using fluorescein (green)-conjugated secondary antibodies, and Di~-~~nju~atedacetylated Iow densi~li~~ote~ (BIac-LDL).whichfluoresce-ssimilarlyto rhodamine (red) and is snecificatlveodocvtosed bvmicroelia.‘* The photomkographs are aUdouble kposures of flukeskein and*rhoda&e fluo&xence: Nate &e heterogeneity of the unsorted cells which contain a mixture of both o~~~endro~tes and micro&t. The Ol-positive fraction is highly enriched in digodendrocytes white the U-negative fraction represents highly a purified microgtialpopulation. Bar=50 km.

Purification of adult human spinal cord glial cells Table

1. Immunohistochemical characterization oligodendrocyte cultures Number

Marker Hoechst GFAP 04 01104 OUGFAP Unknown

Exp. #l

of enriched

of positively-stained

Exp. #2

Exp. #3

109 3 103 103 0 3

14s 4 139 139 0 2

106 0 105 n.t. 0 1

759 human

cells Mean(%) 1ot) 1XZO.9 Y6.5?1.4 95.2 (94.5, 95.9) 0 1.4kO.3

Adult human spinal cord cell suspensions were fractionated on Percoll gradients and the glial cell fraction was plated on uncoated tissue culture plastic overnight. Non-adherent cells were removed, centrifuged and immunohistochemically double-stained in suspension for the presence of 01 and 04 or 04 and GFAP. In addition, the nuclear Hoechst dye was used to determine the total number of cells. Only those cells in which nuclei could be unequivocally identified were further analysed. Cells in which nuclei were detected but did not stain with 01,04 or GFAP were classified as “unknown”. The spinal cords used derived from a 27-year-old male (Exp.#l), a 22-year-old male (Exp. #2) and a 17-year-old male (Exp. #3), all of whom suffered traumatic head injuries. Abbreviations: Exp. #experiments #l-3; Mean %-the mean percentage of cells expressing the indicated antigen(s)+SEM, except for the 01104 data where the mean % (range) is presented; GFAP-glial fibrillary acidic protein, n.t.-not tested.

Facility. For this reason, a differential adhesion enrichment protocol was evaluated. Of the non-adherent cells, 96.5+- 1.4% (n=3) of the cells stained for the oligodendrocyte-specific antigens 01 and/or 04 (Table 1). In these experiments where double staining was performed, all 04-positive cells co-expressed 01. Of the remaining cells, 1.8r+O.9% of the cells expressed the astrocyte intermediate filament GFAP, and in 1.4-+0.3% of the cells, none of these antigens could be unequivocally detected. The small percentage of GFAP-positive cells observed in some of the 01-positive fractions is consistent with previous data. l6 These presumptive astrocytes were not detected in all preparations and never exceeded 2.8% of the non-adherent cells. If the adherent fraction was grown in culture for 5 days and double stained for DiI-ac-LDL and GFAP or DiI-ac-LDL and 04, >96% of the cells were microglia. Fewer than 1% (0.23?0.22%; l/524 total cells counted from two independent preparation) were GFAP positive, while 4.0% 1.2% (19/519 total cells counted from two independent preparations) were 04-positive. With increased time between removal of the spinal cord and the beginning of the isolation, cells recovered from the Percoll gradient tended to give lower yields with a higher percentage of microglia. No viable oligodendrocytes were obtained from adult human spinal cords which remained on ice for >ll hr (data not shown). Change in the dbtribution of oligodendrocytes

and astrocytes with increased time in vitro

At 5 days in culture, nearly 100% oligodendrocytes were detected in either the 01-sorted (Fig. 1B) or the non-adherent (data not shown) fraction. With increasing time in culture (up to 3 weeks), the percentage of 01-positive cells decreased (Figs 2A and C) and the monolayer of flat cells had grown under the remaining oligodendrocytes (Figs 2B and D). Apparent greater oligodendrocyte survival was seen in B16- as compared to D/F-based medium, although the number of oligodendrocytes declined over time under both culture conditions. The majority of the cells in the underlying monolayer stained positively for GFAP (Figs 2B and D), confirming the identity of the monolayer cells as astrocytes. Cells with long processes which expressed both 01 (Fig. 2C) and GFAP (Fig. 2D) immunoreactivities were detected, although these were not common. The purified astrocytes could be passaged at least six times (the latest passage examined), although their growth rate slowed. Interestingly, the passaged astrocytes down-regulated GFAP expression at confluency, but re-initiated it during cell proliferation after passaging (data not shown). Purified adult rat oligodendrocytes have previously been shown to survive, proliferate and form myelin sheaths when co-cultured with purified embryonic rat DRG neurons.30,41 In a single

differential adhesion, grow for 3 &e&s in BWbased medium. A, B and C, D are the same fields in which Ql- (A, C) and GFAP (B, D) fhtorescence are detected with rhodamine- and fluorescein-conjugated secondary antibodies, respectively. There is a decreased number of ohgodendrocytes compared with younger cultures (compare Figs 2A and C with Fig. 13). Moreover, a monolayer of astrocytes which show a highly heterogenous morphology bps grown beneath the ohgodendrocytes. These astrocytes were not present in 5 day cultures. In C and D, the arrows indicate a cell which expresses both Ol- and GFAP-immunoreactivity, the open arrows identify an 01 -positive, GFAP-negative cell and the arrowhead marks an astrocyte which is 01-negative. Bar=50 pm.

Purification

of adult human spinal cord glial cells

761

Fig. 3. Adult human oligodendrocytes survive and proliferate in co-culture with rat dorsal root ganglion neurons. Phase contrast (A, C. E) and immunofluorescent (B. D. F) images of adult human oligodendrocytes after immunostaining with 01 antibody. A, B: A culture stained I day after plating the oligodendrocytes. The small round somata of the oligodendrocytes are seen adhering to the axons of the much larger neurons. 01 immunostaining shows that more than 99% of the small adherent cells are oligodendrocytes. C-F: A culture stained 3 weeks following the addition of the oligodendrocytes of the neuronal cultures. Note the increase in survival and apparent size of the oligodendrocytes in comparison to A and B. The field shown in C and D was taken from the central area of the culture where the oligodendrocytes formed clusters in which the growth of processes are not easily seen. The field shown in E and F was taken from the periphery of the culture where the oligodendrocyte density is lower and only dorsal root ganglion neurites and not cell bodies are found. There, the growth of processes from an oligodendrocyte (arrow) is readily observed. Bar= 100 pm.

experiment, the purified human oligodendrocytes were co-cultured with embryonic rat DRG neurons to test the degree to which the human cells would display the same functional capacity as the rat oligodendrocytes. This experiment was carried out without bFGF in the medium. The day after initial plating, we observed that human oligodendrocytes adhered well to the rat axons (Figs 3A and B).The human oligodendrocytes appeared healthy over the 5 week course of the experiment and appeared to increase in number during the first three weeks of co-culture (compare Figs 3A and B with 3C and D). By 3 weeks, the oligodendrocytes were able to extend processes along the axonal surface (Figs 3C-F). This was especially apparent at the edges of the culture where the cell density was much lower (Figs 3E and F). However, after 5 weeks in co-culture, myelin sheaths were not formed by the human oligodendrocytes, as assessed by Sudan black staining. In contrast. both proliferation and extensive myelination were observed in control adult rat oligodendrocyte/DRG co-cultures run in parallel (data not shown). Uptuke of glutamate and Kf by adult human spinal cord astrocytes

The Km and Vmax for Na+-dependent glutamate uptake averaged 12.0724.23 ~.LLM and 11.1354.88 nmol/min/mg protein, respectively (n=4). K+ uptake was 12.1 -Cl.1 nmol/min/mg protein (n=3). DISCUSSION Enrichment of adult human spinal cord glial cells

The purities of the oligodendrocyte preparations obtained from adult human spinal cord by Percoll gradient centrifugation followed by FACS enrichment with 01 antibodies were similar to

762

S. K. Whittemore

rf nl.

those reported previously for adult rat spinal cord. 30,41The 01-negative fraction was not characterized in those studies, but present results indicate that in human spinal cord preparations it consists of >96% microglia. Differential adhesion has been used previously following Percoll gradient centrifugation to enrich for adult rat and human oligodendrocytes from corpus callosum16,20+“h.37 and the purity in the present study was also similar to that reported by those authors. Microglia from neonatal rodent forebrain cultures have also been isolated by differential adhesion,9 but this approach required weeks in culture before isolation. The procedure described here affords a rapid way in which to simultaneously isolate highly enriched populations of both oligodendrocytes and microglia from adult human CNS. The differential adhesion protocol proved as effective as FACS in purifying both adult human oligodendrocytes and microglia. The fact that neither neurons nor astrocytes were initially detected in significant numbers in either fraction indicates that these cells died during the dissociation and subsequent gradient centrifugation. Adult CNS neurons are very sensitive to axotomy,l potentially accounting for their loss during the enrichment procedure. It is conceivable that the mature astrocytes did not survive because loss of their extensive cytoplasmic processes resulted in damage too severe to repair. Despite high levels of insulin in the culture medium, bFGF was essential for the survival of adult human spinal cord oligodendrocytes when grown alone in culture. However, as assessed by 5-bromo-2’-deoxyuridioe incorporation, labelling indexes above 20% were not obtained, even in the presence of bFGF (data not shown). Thus, the temporal decrease of oligodendrocytes reflects a slow loss under these culture conditions. Unlike neonatal optic nerve oligodendrocyte precursors,24 a combination of bFGF and PDGF did not enhance the labelling index above that seen with bFGF alone. Other investigators have shown that bFGF is a mitogen, but not a survival factor. for adult rat oligodendrocytes,37*41 and these differences may be species-dependent or result from the lower cell plating density used in the present study. A recent report has suggested that bFGF is not a mitogen for adult human white matter oligodendrocytes,3 but it is difficult to directly compare present data with those studies as they utilized a mixed population of 04+/01+, /04+/O 1 cells and the plating densities were not given. While IGF-I stimulates rat oligodendrocyte proliferation,23 it was ineffective at enhancing cell division in 04-positive adult human spinal cord (data not shown) or cerebral cortical3 cells. These apparent species differences in growth requirements underscore the importance of characterizing human cell populations when modelling clinical conditions in vitro. In contrast to the slow loss of oligodendrocytes observed over time in cultures of adult human oligodendrocytes alone, we observed an increase in oligodendrocyte numbers in co-cultures with embryonic rat sensory neurons. These results suggest that the neurons are able to provide both survival and mitogenic factors to the oligodendrocytes, even in the absence of exogenous bFGF. These data further imply a similarity between adult rat and human oligodendrocytes in their response to neuron-derived survival and mitogenic factors. The failure of the human oligodendrocytes to form compact myelin sheaths could reflect an inability of the human cells to recognize the appropriate signals on the rat axons. However, in this regard, it has been observed that transplanted human oligodendrocytes can produce myelin around mouse axonsll and that normal dog oligodendrocytes can recognize and myelinate the axons of the myelin-deficient rat.* Alternatively, myelin formation by adult human oligodendrocytes may proceed at a slower rate than that by rodent or canine oligodendrocytes. The human oligodendrocyte-rat DRG co-culture could thus provide an approach for the study of molecular mechanisms involved in human oligodendrocyte proliferation and survival. The astrocytes which developed from the non-adherent fraction could have arisen entirely from the small percentage of GFAP-positive cells which were contaminants in the cell preparation, from 04/01-positive cells which had the capacity to differentiate along the astrocytic lineage, or from both cell populations. The first possibility is unlikely as 2.1 Xl O5cells/30 mm well were routinely seeded, which suggests that on average 4000 of these were GFAP-positive (see Table I). The doubling time of the primary astrocytes isolated by passaging the 3 week oligodendrocyte cultures was between 4 and 5 days, and by 3 weeks the monolayer was nearly confluent. Assuming five cell divisions, the number of astrocytes arising from GFAP-positive cells in the culture at confluency could maximally have been 1.3~10~ (4000X2”). As a confluent culture contained 5.2X105 astrocytes, and astrocytes developed from all 01-sorted and non-adherent fraction cultures, even those initially devoid of GFAP-positive cells, we suggest that some of these astrocytes developed from

Purifi~tion

of adult

human

spinal

cord gtial cells

763

04/01-positive precursors (i.e. oligodendrocytes). This hypothesis is supported by the identification of cells that expressed both 01 and GFAP immunoreactivities (Figs 2C and D). Also consistent with this suggestion, the culture medium used contained 2.5% serum, which has been shown to stimulate astrocyte proliferation from 04-positive precursors in the neonatalz9*3s and adult39 rodent CNS. Other investigators have postulated the development of adult rodent, bovine2s*26 and human27 astrocytes from precursors which were initially GFAP-negative. In the adult rat optic nerve, the oligodendrocyte-type 2 astrocyte precursor (O-2A) cell has been characterized as 04-positive, 01-negative and vimentin-negative.39 All of the 04-positive cells isolated from adult human spinal cord were also 01-positive, and are therefore presumably oligodendrocytes and distinct from the precursor phenotype identified in the adult rat optic nerve. The conversion of oligodendrocytes into astrocytes would require that the oligodendrocytes de-differentiate and re-differentiate along the astrocyte pathway. De-differentiation of adult rat oligodendrocytes has been reported. 4o The conversion of de-differentiated oligodendrocytes into astrocytes remains to be conclusively demonstrated. CNS glutamate concentrations in the extracellular space have been estimated to range between 2 and 20 FM, and can transiently reach much higher levels following injury.4,7 Astrocytes play a key role in the regulation of extracellular glutamate levels. 7,14,32While the Km values for glutamate uptake are in accord with those previously reported for neonatal rodent astrocytes, the values for Vmaxwere an order of magnitude below those reported by other investigators.8*12*1” The reasons for the discrepant Vmax values between neonatal rodent and adult human astrocytes may reflect either species or developmental differences. Astrocytes have also been implicated as having as important role in regulating extracellular K+ levels in the CNS, and can accumulate potassium by a number of distinct mechanisms. 38 The I(+ uptake levels determined here for adult human spinal cord astrocytes were similar to those reported previously for fetal human34 and neonatal rati7*t9 cerebral cortical astrocytes. Thus, at least by these two physiologi~l parameters, the astrocytes isolated from adult human spinal cord are functionally comparable to previously characterized astrocytes from other species and/or developmental stages. The present study, as well as recent evidence,22,28,31*42suggest that the biology of human cells is significantly different from that of non-human vertebrates, especially with regards to the control of cell prol~eration. These data underscore the need to concu~ently examine primary human cells when extrapolating results from cell culture to mechanisms functioning in the intact human CNS. The methods described here afford a relatively rapid procedure to obtain highly enriched populations of oligodendrocytes, astrocytes and microglia from the adult human CNS. These purified cellular populations should prove very useful in studies to delineate factors involved in the physiology and neuropatholo~ of adult human CNS cells in viva. Acknowledgements-We are grateful to Les Olson and J. D. Waters of the University of Miami Organ Procurement Team and the Medical Examiners Offices of Dade, Broward, Collier. Monroe, Palm Beach and St Lucie Counties. without whose extensive cooperation this research could not have been undertaken. We thank Dr Joseph T. Neary for his help with the K+ and glutamate uptake assays, Robin Mendelsohn for excellent technical assistance and Linda White for insightful suggestions throughout the course of this project. This work was supported by The American Paralysis Association, The Miami Project to Cure Paralysis, The Florida High Technology Council, NS26887 (SRW) and The National Multiple Sclerosis Foundation RG-2210-A-2 (P.M.W).

REFERENCES 1. Aguayo A., Rasminshy M., Bray G. M., Carbonetto S., McKerracher L., Villegas-Perez M. P., Vidal-Sanz M. and Carter D. A. (1991) Degenerative and regenerative responses of injured neurons in the central nervous system of adult mammals. Phil. Trans. R. Sot. Lond. 331,337-343. 2. Archer D., Hoffman R., Miletic V. and Duncan 1. (1991) Myelination in the myelin deficient rat by transplanted canine cells. Sot. Nerdrosci. A b. 17, 1149. 3. Armstrong R. C., Dorn H. H., Kufta C. V., Friedman E. and Dubois-Dalcq M. E. (1992) Pre-oligodendrocytes from adult human CNS. J. Neurosci. X&1538-1547. 4. Benveniste H. (1989) Brain mi~odialysis. I. Neuroche~. 52,1741-1750. 5. Brewer G. J. and Cotman C. W. (1989) Survival and growth of hippoc~pal neurons in defined medium at low density: advantages of a sandwich culture technique or low oxygen. Brain Res. 494,65-74. 6. Duncan I. D., Paino C., Archer D. R. and Wood P. M. (1992) Functional capacities of transplanted cell-sorted adult ohgodendrocytes. Devl Neurosci. 14114122. 7. Erecinska M. and Silver I. A. (1990) Metabolism and role of glutamate in mammalian brain. frog. Nerrrobiol. 35, 245-296.

5. K. Whitkmorc

764

andSeifert W. (1991) Characterizatio;lofglutamotc uptakesystemsin astrocytepritnar~ cultures IIXNII r;ti I?MII. Glia 4293-304. Frei K., Bodmer S., Schwerdel C. and Fontana A. (lY86) Astrocyte-derived interleukin 3 as a growth factor for rnicroghai cells and peritoneal macrophages. J. Neuroinwnuno/. 137,352 l-3527. Guilian D., Johnson B., Krebs J. F.. George J. K. and Tapscott M. (1991) Microglial mitogens are producal m the developing and injured mammalian brain. J. Cell Biol. ll2,323-333. Gumpel M., Lachapelle F., Gunsmuller A.. Banloy M.. Baron van Evercooren A. and Baumann N. (1987) Tr;~nsplantation of human ohgodendrocytes into shiverer brain. Ann N. Y. Acad. Sci. 495,71*5. Hansson E. (1986) Primarv cultures from defined brain areas. III. Effects ofseeding time on [iH]-t.-glutamate transport and glutamine synthetasc activity. Bruin Re.t 389. 203-209. Hansson E. and Ronnback L. (lY91) Receptor regulation of the glutamate. GABA, and taurine high-affinity uptake into astrocytes in primary culture. Bruin Re.\. W&21 S-21 I. Hertz L. (1979) Functional interactions between neurons and astrocytes. 1. Turnover and metabolism of putative amino acid transmitters. Prog. Nruroh~ol. 13, 277-323. Hertz L.. Schousboe A., Boechler N.. Mukerji S. and Federoff S. (1978) Kinetic characteristics of the glutamate uptake into normal astrocytes in cultures. Nenrochm~. Rex 3, l- 14. Kim S. U. (1985) Antigen expression by glial cells grown in culture. J. Neuroimmunol. 8,255-2X2. Kimelberg H.. Bowman C.. Biddlecome S. and Bourke R. S. (1979) Cation transport and membrane potential properties of primary astroglial cultures from neonatal rat brains. Brain Rex 177,533-550. Kleitman M., Wood P. M. and Bunge R. P. (1991) Tissue culture methods for the study of myelination. In: r ‘ftlfurirrl: Nerve Cells (eds G. Banker & K. Goslin), pp. 337-376. MIT Press, Cambridge. Latzkovits L.. Torday C.. Labourdette Cl., Pettman B. and Sensenbrenner M. (1988) Sodium and potassium uptake 111 primary cultures of proliferating rat astroglial cells induced by short-term exposure to an astroglial growth factor. Nelrrochem. Res. 13,837-84X. Lisak R. P., Hirayama M., Kuchmy D., Rosenzweig A., Kim S. U., Pleasure D. and Silberberg D. H. (1983) Cultured human and rat oligodendrocytes and rat Schwann cells do not have immune response gene associated antigen (la) on their surface. Brain Rex 289,285-292. McCarthy K. D. and DeVellis J. (19x0) Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissue. J. Ceil Biol. 85,89&902. Morrissey T. K.. Kleitman N.. Goddard M.. Aebischer P. and Bunge R. P. (1991) Schwann cell differentiation in vitro: differences between cells from adult rat and human nerve. Sot. Neurosci. Ab. 17,376. Mozell R. L. and McMorris F. A. (1991) Insulin-like growth factor I stimulates oligodendrocyte development and myehnation in rat brain aggregate cultures. J. Neurosci. Res. 30,382-390. Nohle M. (1991) Points of controversy in the O-2A lineage: clocks and type-2 astrocytes. Glia 4,157-l 64. Norton W. T. and Farooq M. (1989) Astrocytes cultured from mature brain derive from glial precursor cells. .I. Neurosci. 9,769-77.5. Norton W. T., Farooq M., Chiu F.-C. and Bottenstein J. E. (1988) Pure astrocyte cultures derived from cells isolated from mature brain. Glia 1,403-414. Perzel’ova A. and Mares V. (1993) Appearance of GFAP-positive cells in adult human brain cultures spontaneously decelerated in growth. Glia 7,237-244. Radna R. L., Caton Y., Jha K., Kaplan P., Li G., Traganos F. and Ozer H. L. (1989) G rowth of immortal simian virus SV40 tsA-transformed human fibroblasts is temperature dependent. Molec. Cell. Biol. 9,3093-3096. Raff M. C.. Miller R. H. and Noble M. (1983) A glial progenitor cell that develops in vitro into an astrocyte or an oligodendrocyte depending on the culture medium.-N&e %3,390-396. Rosen C.. Bunae R. P..Ard M. and Wood P. M. (1989) ,_. Twe 1 astrocvtes inhibit myelination of adult rat oligodendrocytes in vitro. J. Neu;osci. 9,337 1-3379. Sager R.. Tanaka K., Lau C. C.. Ebina Y. and Anisowicz A. (1983) Resistance of human cells to tumorigenesis induced by cloned transforming genes. Proc. nafn. Acud. Sci. U.S.A. 80,7601-7605. Schousboe A. (1981) Transport and metabolism of glutamate and GABA in neurons and glial cells. Inr. Rev. Nrurobiol. 25 l-45. Sweetnam P. M., Sanon H. R., White L. A.. Brass B. J., Jaye M. and Whittemore S. R. (1991) Differnetial effects of acidic and basic fibroblast erowth factors on spinal cord cholinergic, GABAergic and glutamatergic neurons. J. Neurochem. 57,237-249. Tas P. W. L., Massa P. T. and Koschel K. (1986) Preliminary characterization of an Na +, K’.. Cl- co-transport activity in cultured brain astrocytes. Neurosci. Leff. 70,369~373. Trotter J. and Schachner M. (1989) Cells positive for the 04 surface antigen isolated by cell sorting arc able to differentiate into astrocytes or ohgodendrocytes. Devl Brain Res. 46,115-122. Vick R. S., Chen S.-J. and DeVries Cl. H. (1990) Isolation, culture and characterization of adult rat oligodendrocytes. .I. Neurosci. Rex 25, 524534. Vick R. S. and De Vries G. H. (1992) Mitotic potential of adult rat oligodendrocytes in culture. J. Neuroscr. Rex 33, 68-74. Walz W. and Hertz L. (1983) Functional interactions between neurons and astrocytes. II. Potassium homeostasis at the cellular level. Prog. Neurobiol. 30,133-1X3. Wolswijk G. and Noble M. (1989) Identification of adult-specific glial progenitor cell. Development 105,387-400. Wood P. M. and Bunge R. P. (1991) Origin of remyelinating cells in the adult central nervous system: the role of the mature oligodendrocyte. Glia 4,22_-232. Wood P. M. and Mora J. (1993) Source of remyelinating astrocytes. In Advances in Neurology, Vol. 59 (ed. F. J. Seil). pp. 11>123. Raven Press, New York. Wright W. E., Pereira-Smith 0. M. and Shay J. W. (1989) Reversible cellular senescence: implications for immortalization of normal human diploid fibroblasts. Molec. Cell. Biol. 9,3088-3092.

K. FlottB.

Y. 10. 11. 12. 13. 14. IS. 16. 17. 18. 19.

20.

21. 22. 23. 24. 2s. 26. 21. 28. 29. 30. 31. 32. 33.

34. 3.5 36 37. 38. 39 40. 41. 42.

er rrl.