Analysis of Stem Cell Lineage Progression in the Neonatal Subventricular Zone Identifies EGFR+/NG2 Cells as Transit-Amplifying Precursors


  • Author contributions: T.C., K.O., and C.P.B.: collection of data and analysis, data interpretation, manuscript writing; T.F.: collection and analysis of the data; C.M., G.H.-W., K.W., and V.E.: collection of the data; F.C.: conception and design, data analysis and interpretation, manuscript writing. T.C. and K.O. contributed equally to this article.

  • First published online in STEM CELLS Express March 26, 2009


In the adult subventricular zone (SVZ), astroglial stem cells generate transit-amplifying precursors (TAPs). Both stem cells and TAPs form clones in response to epidermal growth factor (EGF). However, in vivo, in the absence of sustained EGF receptor (EGFR) activation, TAPs divide a few times before differentiating into neuroblasts. The lack of suitable markers has hampered the analysis of stem cell lineage progression and associated functional changes in the neonatal germinal epithelium. Here we purified neuroblasts and clone-forming precursors from the neonatal SVZ using expression levels of EGFR and polysialylated neural cell adhesion molecule (PSANCAM). As in the adult SVZ, most neonatal clone-forming precursors did not express the neuroglia proteoglycan 2 (NG2) but displayed characteristics of TAPs, and only a subset exhibited antigenic characteristics of astroglial stem cells. Both precursors and neuroblasts were PSANCAM+; however, neuroblasts also expressed doublecortin and functional voltage-dependent Ca2+ channels. Neuroblasts and precursors had distinct outwardly rectifying K+ current densities and passive membrane properties, particularly in precursors contacting each other, because of the contribution of gap junction coupling. Confirming the hypothesis that most are TAPs, cell tracing in brain slices revealed that within 2 days the majority of EGFR+ cells had exited the cell cycle and differentiated into a progenitor displaying intermediate antigenic and functional properties between TAPs and neuroblasts. Thus, distinct functional and antigenic properties mark stem cell lineage progression in the neonatal SVZ. STEM CELLS 2009;27:1443–1454


The subventricular zone (SVZ) of the rodent postnatal brain is the main germinal region that generates olfactory bulb inhibitory neurons throughout adulthood. Three main precursors have been identified in the adult SVZ: slowly dividing stem cells, rapidly proliferating transit-amplifying precursors (TAPs), and neuroblasts [1]. Unlike stem cells, TAPs do not self-renew extensively in vivo. However, epidermal growth factor (EGF) promotes greatly their proliferation in vivo and it elicits the generation of clones from both stem cells and TAPs in vitro [2]. Stem cells express glial fibrillary acidic protein (GFAP) and the Lewis X/stage-specific embryonic antigen-1 (LeX/SSEA-1) [1, 3], whereas TAPs and neuroblasts express the homeobox transcription factor distalles 2 (Dlx2) [2]. In contrast to the adult SVZ, stem cells in the neonatal SVZ cannot be identified solely on the basis of antigen expression and their lineage progression has not been analyzed. For example, polysialylated neural cell adhesion molecule (PSANCAM) is expressed mainly by neuroblasts in the adult SVZ, but it is found in most cells including glial precursors in the neonatal SVZ [4]. Only two cell types have been distinguished in the neonatal germinal epithelium. The predominant one consists of small, migratory, highly proliferative, Dlx2 immunopositive cells, which include neuronal and glial progenitors [4, 5]. The second type is represented by larger, less mobile, and proliferative cells which are direct descendant of embryonic radial glia precursors [5] and after birth will give rise to astroglial stem cells and other cell types [5, 6]. Similar to TAPs in the adult SVZ, some Dlx2 immunopositive cells in the neonatal SVZ are EGF receptor (EGFR) immunopositive and divide rapidly [7]. However, this cell population also expresses markers of oligodendrocyte precursors such as neuroglia proteoglycan 2 (NG2) and oligodendrocyte lineage transcription factor 2 (Olig2); functional evidence that they represent TAPs remains lacking [8, 9].

In addition to antigenic characteristics, distinct functional properties appear during lineage progression. For example, gap junctions are associated with progenitor proliferation [10, 11] and may promote stem cell self-renewal [12] and cell cycle synchronization [13]. Gap junctions decline during neuronal differentiation [14] and are absent in postmitotic and migrating neurons [13]. Voltage-dependent Ca2+ entry emerges during neuronal differentiation [15, 16], whereas Ca2+ signaling in radial glia cells mainly involves connexin hemichannels, purinergic receptors, and Ca2+ release from intracellular stores [17]. In contrast, both neuroblasts [18, 19] and precursors [11] in the SVZ express voltage-dependent K+ channels, which in many cell types control cell cycle progression and proliferation [20].

Here we use multiple approaches to identify putative stem cells in the neonatal SVZ. We showed that the majority of clonogenic cells expressing high levels of EGFR have antigenic and functional characteristics of TAPs and differ in vitro from neuroblasts in their electrophysiological properties and ability to form gap junctions. Furthermore, by cell tracing in situ, we provided evidence that within 2 days most cells expressing high levels of EGFR lose stem cell potential and undergo differentiation into a previously uncharacterized intermediate progenitor before giving rise to neuroblasts.


Tissue Dissection and Cell Culture

Neonatal CD1 mice (between P4 and P9) were decapitated, in accordance with the local ethical guidelines for the care and use of laboratory animals (Charles River, Germany, The SVZ of the lateral ventricle was dissected and dissociated as previously described [16, 21]. The dissociated cells were processed for sorting either immediately or after being plated overnight at a density of 105 cells per milliliter in NS-A (EuroClone S.p.A., Siziano, Italy, medium supplemented with 10 ng/ml fibroblast growth factor-2 (FGF-2; R&D Systems Inc., Minneapolis, as previously described [21].

Fluorescence-Activated Cell Sorting and Clonal Analysis

For EGFR staining, cells were stained with 20 ng/ml EGF-Alexa (EGF-A; Molecular Probes Inc., Eugene, OR, as previously described [21].

For live immunostaining, cells were first incubated with primary PSANCAM, LeX/SSEA1, or NG2 antibodies and secondary antibodies conjugated to fluorescent fluorophores (Molecular Probes Inc.) and then processed for EGF-A staining as previously described [21].

Sorting was performed on either a Vantage flow cytometer or a FACSAria (Becton, Dickinson and Company, Franklin Lakes, NJ, with single cell precision purity mask. Fluorescence-activated cell sorting (FACS) gates were set using unstained cells (autofluorescence), cells preincubated with 20 ng/ml EGF (Sigma-Aldrich, St. Louis, MO,, and cells immunostained only. Dead cells were detected by propidium iodide (1 μg/ml; Sigma-Aldrich) exclusion.

For immunocytochemistry, sorted cells were plated on chamber slides (Nunc, Rochester, NY,, coated with growth factor-reduced matrigel (Becton, Dickinson and Company) in Neurobasal A medium (Invitrogen, Carlsbad, CA, containing 2% B27 (Gibco, Grand Island, NY, and 1% fetal calf serum, and fixed after adhesion.

For clonal analysis, cells were plated by FACS-automated cell deposition in 96-well plates (Nunc) in medium containing 20 ng/ml EGF and/or 10 ng/ml FGF-2 as previously described [21]. Neurosphere formation was scored after 10 days.

For electrophysiology and calcium imaging, sorted cells were plated onto 12-mm coverslips (VWR International GmbH, Darmstadt, Germany, coated with growth factor-reduced matrigel in medium containing EGF and FGF-2 at densities of 30–60 per mm2 and 200–400 per mm2 for precursors and neuroblasts, respectively. Cultures were analyzed 24–48 hours after plating when they had recovered from the stress of sorting without having undergone differentiation.


The following antibodies were used at the indicated dilution: mouse monoclonal to LeX/SSEA1, 1:200 (Developmental Studies Hybridoma Bank, Iowa City, IA,∼dshbwww); mouse monoclonal to PSA-NCAM, 1:200 (Upstate, Charlottesville, VA,; goat polyclonal to Doublecortin, 1:500 (Santa Cruz Biotechnology Inc., Santa Cruz, CA,; rabbit polyclonal to green fluorescent protein (GFP), 1:200 (Molecular Probes Inc.); mouse monoclonal to GFP, 1:200 (Chemicon, Temecula, CA,; sheep polyclonal to EGFR, 1:200 (Upstate); mouse monoclonal to EGFR, 1:100 and 1:300 (Sigma-Aldrich); rabbit polyclonal to phospho histone H3 (pHH3), 1:500 (Upstate); rabbit polyclonal to Ki67, 1:200 (Abcam, Cambridge, U.K.,; rabbit polyclonal to Dlx2, 1:500 (Chemicon); rabbit polyclonal to GFAP, 1:600 (DAKO, Glostrup, Denmark,; mouse monoclonal to RC2, 1:100 (Developmental Studies Hybridoma Bank); rabbit polyclonal to Olig2, 1:200 (Chemicon); rat monoclonal to NG2 [22], 1:10 (a gift from Dr. J. Trotter, University of Mainz, Mainz, Germany).

Organotypic Slice Preparation and Infection

Brains embedded in low melting point agarose gel (4% in phosphate-buffered saline [PBS]) were cut into 300-μm-thick forebrain coronal sections containing the SVZ using a vibratome (CU65 Cooling Unit and HM650V Vibratome; Microm, Walldorf, Germany, at 0°C in slicing medium (in mM): 150 sucrose, 40 NaCl, 4 KCl, 7 MgCl2, 1.25 NaH2PO4, 0.5 CaCl2, 10 glucose, 26 NaHCO3, gassed with 95% O2 and 5% CO2. Slices were cultured as previously described [21]. Slices were infected by pipetting 1 μl viral dilution onto the lateral ventricle and analyzed after 1–3 days as indicated in the text.


Cells were fixed and immunostained as previously described [23]. Primary antibodies were detected using Alexa 488, Alexa 555, or Cy3 conjugated secondary antibodies (Molecular Probes Inc.). Images were acquired with a conventional fluorescence microscope (DMIRBE microscope, 40× objective; Leica, Germany, 4′,6-Diamidine-2′-phenylindole-dihydrochloride (1:1,000; Roche Diagnostics GmbH, Roche Applied Science, Mannheim, Germany, was used for nuclear counterstaining. For each antigen the number of immunopositive cells was quantified from at least two independent experiments and for each experiment an average of 100 cells across multiple fields were analyzed with a conventional fluorescence microscope.


P7 mice were anesthetized with an intraperitoneal injection (2 ml/kg body weight) of sodium-pentobarbital (Narcoren; Merial, Germany) and perfused with a solution of 4% paraformaldehyde (PFA). Brains were removed, left overnight in 4% PFA at 4°C, and incubated in 30% sucrose for 24 hours. After PBS washing and embedding in 4% low melting agarose, brains were cut into 40-μm-thick coronal vibratome sections and processed for immunohistochemistry as previously described [21]. Images were acquired using a laser scanning confocal microscope (TCS SP2 scanning head and inverted DMIRBE microscope, 40× oil immersion HCX PL APO objective, Leica Confocal Scan software; Leica). The proportion of Ki67+/pHH3 cells within the EGFR+ and EGFR populations was calculated as the difference between the total number of Ki67+ cells and the number of pHH3+ cells in each population. The counts of Olig2+/EGFR+ cells obtained from slices were confirmed by counting the number of double immunopositive cells in dissociated SVZ cells that after dissection had been plated and immunostained as described earlier.

Infected organotypic slices were fixed with 4% PFA/3% sucrose in PBS for 3 hours at room temperature (RT) and transferred into 30% sucrose in PBS before immunohistochemistry. Images were acquired using a laser scanning confocal or a conventional fluorescence microscope as described earlier. For each type of analysis and antigen, we used four slices representing at least three independent experiments. An average of four GFP+ cells per slice was analyzed.

Ca2+ Imaging

Cells were loaded at RT for 20 min with 2–4 μM Fluo3-AM (Molecular Probes Inc.) in Ringer's solution as previously described [16]. Ca2+ imaging was performed at RT in Ringer's solution in an open-topped perfusion chamber (Life Imaging Services, Basel, Switzerland, mounted in an inverted fluorescent microscope (IX70; Leica) equipped with a charge coupled device camera (ImagoQE; TILL Photonics GmbH, Graefelfing, Germany, and a software interface (TILLvisION 4.0; TILL Photonics GmbH). Excitation light was generated by a monochromator (Polychrom IV; TILL Photonics GmbH) coupled to a xenon short arc light source (USHIO, Tokyo, Japan, Fluo-3 fluorescence intensity (F) obtained from regions of interest was background subtracted and normalized to baseline fluorescence (F0). Fluctuations in Ca2+ levels larger than 5% were defined as Ca2+ transients. Where indicated, 40 mM K+ was applied by replacing half Ringer with Ringer containing 80 mM K+ in substitution for Na+. All data came from at least three independent experiments from different cell preparations.

Patch Clamp

Whole-cell patch clamp recordings and fluorescent imaging were performed at RT as previously described [16]. Solutions for intracellular outward current recording (in mM): 10 NaCl, 110 KCl, 1 EGTA, 0.1 CaCl2, 10 HEPES, 4 MgATP, 0.4 Na2GTP (pH 7.2, 265 mOsm/l); and extracellular outward current recording (in mM): 113 NaCl, 5 KCl, 1 MgCl2, 2 CaCl2, 10 HEPES, 25 glucose (pH 7.4, 280 mOsm/l). To block K+ currents, KCl and NaCl were replaced with tetraethylammonium chloride (TEACl, 118 mM; Sigma-Aldrich). Solutions for intracellular Ca2+ current recording (in mM): 90 CsCl, 20 TEACl, 5 EGTA, 2 MgCl2, 10 HEPES, 8 glucose, 2 MgATP, 0.4 Na2GTP, 10 phosphocreatine (pH 7.3, 300 mOsm/l); and extracellular Ca2+ current recording (in mM): 135 NaCl, 10 BaCl2, 1 MgCl2, 10 HEPES (pH 7.3, 293 mOsm/l). Alexa-fluor 594 hydrazide (10 μM; Molecular Probes Inc.) was added to the internal solutions. Organotypic slices were continuously superfused with the following (in mM): 125 NaCl, 3.5 KCl, 1.3 MgCl2, 2.4 CaCl2, 1.2 NaH2PO4, 26 NaHCO3, 10 glucose, saturated with 95% O2 and 5% CO2 (pH 7.35, 320 mOsmo/l). Membrane capacitance (CM), zero-current resting potential (VR), and membrane resistance (RM) were determined in the first 3 min of recording by the Multiclamp 700 software interface using a 5 mV hyperpolarizing step of 10 ms at holding potential (Vh) = −70 mV. Capacitative and leakage currents were reduced by subtracting the average current in response to P/4 hyperpolarizing pulses. All data have been corrected for junction potentials that were calculated with Clampex (for K+ currents: 5.2 mV in sorted cells, 4.2 mV in slices; for Ca2+ currents: 4.8 mV in both systems). Drugs were applied either with a pressure ejection system (PY800 Pneumatic Picopump; World Precision Instruments, Sarasota, FL,, or by rapid bath solution exchange using a multibarrel perfusion pencil (AutoMate Scientific, Berkeley, CA, Meclofenamic acid (MFA) was obtained from Sigma-Aldrich. Data are presented as mean values ± SE. Statistical significance was calculated using Student's t test for either paired or independent samples.


Both Stem Cells and Neuroblasts Are PSANCAM Immunopositive

EGF in vitro and in vivo promotes the proliferation of TAPs and to a lesser extent of stem cells [2]. Consistent with this, double immunostaining with specific antibodies revealed that, in contrast to EGFR cells, most EGFR+ cells in the neonatal (P7) SVZ are cycling (Ki67+; Fig. 1A). Furthermore, a larger fraction of cycling EGFR+ precursors undergoes mitosis, as indicated by the number of pHH3+ cells, than correspondent Ki67+/EGFR. This suggests that either the cell cycle of EGFR cells is slower or they do not undergo mitosis in the SVZ (Fig. 1A). To better characterize SVZ EGFR+ cells we next used a fluorescently tagged EGF to purify them from the remaining SVZ cells. Using this approach we have previously shown that cells expressing high levels of EGFR (EGFRhigh) in the periventricular germinal epithelium of both the embryonic and adult telencephalon are highly enriched in multipotent and long-term self-renewing precursors [21]. Immunostaining with EGFR antibodies revealed that 84.3% ± 1.0% of the EGFRhigh cells displayed a strong immunoreactivity, whereas a weak immunofluorescent signal could only be detected in 15.7% ± 4.2% of the remaining cells expressing low levels of EGFR (EGFRlow; Fig. 1B), showing the differences in EGFR expression between the two populations. To investigate the relationship between EGFRhigh cells and other SVZ cell types, we next sorted neonatal and adult (P48) SVZ cells according to EGFR expression and immunoreactivity to PSANCAM (PS; supporting information Fig. S1A); in the adult SVZ it is mostly associated with neuroblasts [24]. After overnight treatment with exogenous FGF-2, EGFRhigh neonatal cells represented only around 5.8% of the viable cells (supporting information Table S1) as previously reported for adult and embryonic tissue [21]. At both ages, whether analyzed after tissue dissociation or after 24 hours of culturing in medium containing FGF-2, most SVZ cells, including virtually all EGFRhigh cells, were PS+ (supporting information Fig. S1A; Table S1). As expected [21], clone-forming cells were observed in both the EGFRlow and EGFRhigh populations. However, their incidence was almost 40-fold higher in the latter, either when isolated directly after dissection (Fig. 4C) or after overnight FGF-2 treatment (supporting information Table S1). Surprisingly, in both populations, most clonogenic cells were PS+ (supporting information Table S1) and, unlike their embryonic counterparts [21], around 90% of EGFRhigh cells required exogenous EGF to undergo clone formation (Fig. 1D). Confirming the differences between the two populations and the specificity of our sorting procedure, doublecortin was expressed by most PS+/EGFRlow, but not PS+/EGFRhigh cells (Fig. 1C), whereas quantitative real-time polymerase chain reaction analysis of RNA revealed stronger expression of Mash1 in the latter than in the former population (not shown). In line with the FACS analysis, immunostaining of brain slices revealed that the majority of PS+ cells in the neonatal SVZ are EGFR and that EGFR+ cells also strongly express PSANCAM (Fig. 1E).

Figure 1.

Neuroblasts and EGFRhigh cells both express PSANCAM. (A): Proportion of the EGFR+ and EGFR cells which are cycling (Ki67+/phospho histone H3 [pHH3]), mitotic (pHH3+), and nonproliferating (Ki67), quantified on neonatal subventricular zone (SVZ) slices after double immunostaining with EGFR and Ki67 or pHH3 antibodies. (B, C): Photomicrographs of neonatal EGFRhigh and EGFRlow cells sorted after dissections using fluorescence-activated cell sorter (B) and PS+/EGFRlow and PS+/EGFRhigh cells sorted at day in vitro (DIV)1 (C) after immunostaining as indicated. Arrows in B indicate a weakly immunopositive cell. Scale bar = 50 μm (B); 20 μm (C). (D): Quantitative analysis of clone-forming PS+/EGFRhigh and PS+/EGFRlow cells, sorted at DIV1 and plated in the presence of EGF and FGF-2 as indicated. (E): Confocal photomicrographs of a neonatal SVZ coronal section and higher magnification images of the boxed areas (E1, E2) upon double immunostaining as indicated. Scale bar: large image, 100 μm; insets, 25 μm. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; DCX, doublecortin; EGFR, epidermal growth factor receptor; FGF, fibroblast growth factor; LV, lateral ventricle; PS or PSANCAM, polysialylated neural cell adhesion molecule.

As both astroglial stem cells and TAPs are clonogenic [2], and a subset of adult EGFRhigh cells expresses GFAP and LeX/SSEA1 (LeX) [21], some neonatal EGFRhigh cells may also represent astroglial stem cells. Indeed, when neonatal SVZ cells were sorted according to EGFR expression and LeX immunoreactivity, around 15% of EGFRhigh cells, representing 0.8% of the viable cells, were found LeX+ (supporting information Fig. S1B). Both LeX+ and LeX EGFRhigh cells were similarly capable of forming clones (supporting information Table S1). However, they significantly differed in terms of GFAP and Dlx2 expression that were predominant in the LeX+/EGFRhigh and LeX/EGFRhigh populations, respectively (Fig. 2A, 2B). Instead, a similar analysis with antibodies to the NG2, which is expressed in oligodendrocyte precursors and in a subset of SVZ precursors [7], showed no overlap between the two cell populations (supporting information Fig. S1C). Consistently, immunofluorescence on brain slices showed that virtually all EGFR+ cells were NG2 and most (94.2% ± 3.3%) did not express the transcription factor Olig2, which is known to be coexpressed with NG2 (Fig. 2C, 2D) [7].

Figure 2.

Antigenic characterization of sorted cells. (A): Neonatal LeX+/EGFRhigh cells and LeX/EGFRhigh cells sorted at DIV1 and immunostained with GFAP or Dlx-2 antibodies. Scale bar: 25 μm. (B): Quantitative analysis of the immunostaining shown in (A). (C, D): Confocal photomicrographs of neonatal subventricular zone coronal sections upon double immunostaining with EGFR and NG2 (C) or Olig2 (D) antibodies. Higher magnification images of the boxed areas are shown in (C1) and (C2) and in the insets of (D). White arrowheads, black arrowheads, and arrows indicate single EGFR+, NG2+, or Olig2+ cells, respectively. Scale bar: large images, 50 μm; insets, 25 μm. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; Dlx-2, distalles 2; EGFR, epidermal growth factor receptor; GFAP, glial fibrillary acidic protein; LeX, Lewis X; LV, lateral ventricle; NG2, neuroglia proteoglycan 2.

Taken together, the analysis of clone formation and the pattern of antigen expression indicated that clonogenic EGFRhigh cells in the postnatal SVZ can be divided into putative TAPs (PS+/LeX/Dlx2+), representing more than 80% of EGFRhigh cells, and stem cells (PS+/LeX+/GFAP+). As the majority of PS+/EGFRlow cells are doublecortin+ neuroblasts, hereafter we will refer to PS+/EGFRhigh and PS+/EGFRlow cells as precursors and neuroblasts, respectively.

Functional Properties of Precursors and Neuroblasts

During embryonic development, spontaneous Ca2+ events occur in both mitotic and postmitotic neural precursors [16, 17, 25, 26]. Therefore, we next used Ca2+ imaging to investigate whether sorted precursors and neuroblasts were functionally distinct. Cells were imaged 1–2 days after plating (DAP). In these conditions most precursors proliferated without undergoing differentiation, whereas only few neuroblasts divided (data not shown). Sorted neuroblasts displayed a small bipolar cell body and formed clusters and chains because of the presence of laminin [27] (Fig. 3A4). In contrast, precursors had a larger elongated soma and did not cluster, although both isolated cells and cells in contact were present at DAP2 (Fig. 3A2). Both populations showed spontaneous Ca2+ activity at irregular intervals (Fig. 3A) albeit with clearly different kinetics (supporting information Fig. S2), suggesting that different mechanisms regulate Ca2+ homeostasis in the two groups. Application of a high [K+] depolarizing solution produced a fast rise in the [Ca2+]i in the majority (79%, n = 43) of the neuroblasts (Fig. 3A3), with a mean increase in fluorescence of 75.3% ± 13.7% (n = 34). Both spontaneously active and silent neuroblasts responded to depolarization. In contrast, only few precursors (9%, n = 86) responded to high [K+] (Fig. 3A1), with a mean fluorescence increase of 57.7% ± 26.7% (n = 8). These data suggested differences in densities of K+ channels and/or voltage-dependent Ca2+ channels (VDCCs) between the two populations. Therefore, we next investigated this issue using whole cell patch clamp. Current injection in current clamp did not evoke action potentials in either cell type (supporting information Fig. S3A), consistent with the absence of inward Na+ current upon a depolarizing step protocol in voltage clamp (Fig. 3B; supporting information Table S2). Instead, the same protocol in both populations activated slow voltage-dependent outward currents, typical of K+ currents, during voltage clamp steps to 30 mV or more depolarized potentials. The density of this current at its peak was significantly larger in precursors than in neuroblasts (Fig. 3B, 3C; supporting information Table S2). Analysis of the kinetic properties showed that outward currents in the two cell types had comparable half-activation potentials, voltage dependence, and identical inactivation kinetics (supporting information Fig. S4A, S4B). These outward currents were blocked to a similar extent in both precursors and neuroblasts by tetraethylammonium (TEA, 118 mM), a general blocker of K+ currents (supporting information Fig. S4C–S4F). Voltage clamp protocols, with and without a hyperpolarizing prepulse, were used to isolate an A-type current, which was present in all neuroblasts tested (n = 12) and in 80% of the progenitor cells (n = 10). Both cell types showed a comparable density of A-type currents (supporting information Fig. S4G), which represent about half of the total depolarization-activated outward current in both groups (supporting information Fig. S4H). The residual outward current represents a delayed rectifier current, which was completely blocked by 4-aminopyridine (4-AP) at 3 mM in both cell types (supporting information Fig. S4L, S4M). The A-type component was also blocked by 4-AP at only 3 mM but not at 0.5 mM in neuroblasts (supporting information Fig. S4N).

Figure 3.

Precursors and neuroblasts are functionally distinct. (A): Fluorescence changes (F/F0) of Fluo-3-loaded cells in control and upon 40 mM K+ (HiK) application within three representative PS+/EGFRhigh (A1) and PS+/EGFRlow (A3) cells. Differential interference contrast photographs of PS+/EGFRhigh (A2) and PS+/EGFRlow (A4) cells upon whole-cell patch clamp recordings at day after plating 2. Scale bar = 20 μm. (B): Representative whole-cell currents. (C): Current density to voltage relationship (IV) (**, p < .01; n = 17 per group). Mean Ca2+IV obtained with a 155-ms ramp protocol in (D) PS+/EGFRlow (n = 6) and (E) PS+/EGFRhigh cells (n = 4). Abbreviations: EGFR, epidermal growth factor receptor; PS, polysialylated neural cell adhesion molecule.

We next analyzed the expression of VDCCs in sorted precursors and neuroblasts. Depolarizing ramp (Fig. 3D) and step (supporting information Fig. S3B) protocols in 10 mM extracellular Ba2+ elicited a small voltage-dependent inward current in all neuroblasts tested (n = 6), but not in isolated precursors (n = 5), where only a small outward current at positive potentials was observed (Fig. 3E), which likely represents a residual delayed rectifier K+ current not blocked by the 20 mM intracellular TEA (see above). Taken together, these data showed functional differences between precursor cells and neuroblasts; indeed, although both express delayed rectifying and A-type K+ channels, precursors have a higher density of TEA-sensitive K+ currents and exhibit A-type currents which are more sensitive to 500 μM 4-AP typical of A-type potassium channel isoform 1.4 (Kv1.4) but not Kv3.3 or 4.3 (see IUPHAR channel compendium for channel nomenclature). Furthermore, functional VDCCs are present exclusively in neuroblasts.

The passive membrane properties of precursors and neuroblasts also differed. As neural precursors can form functional gap junctions, we recorded on the same coverslip from isolated precursors and those contacting neighboring cells (supporting information Fig. S5A) [28–30]. Whether isolated or in contact, precursors had a significantly larger CM, smaller RM, and a more hyperpolarized VR than neuroblasts (supporting information Table S2). However, in precursors these parameters depended on their contact with neighboring cells. Compared with isolated cells, precursors contacting each other displayed a twofold larger CM (p < .001), decreased RM (p < .001), and a more hyperpolarized VR (p < .01) (supporting information Table S2; Fig. S5B, S5C). These differences were not observed in the presence of the gap junction blocker MFA (100 μM), which reversibly reduced CM and increased RM without modifying the VR in all the precursors in contact tested (Table 1; supporting information Fig. S5D, S5E). The density and voltage dependency of K+ currents were not significantly affected by cell contact or MFA in either cell group (Table 1; supporting information Fig. S5F, S5G), as expected for the involvement of gap junctions coupling between similar cell types. These effects of MFA on passive membrane properties were specifically observed in precursors in contact but not in isolated precursors (Table 1) or neuroblasts (data not shown), suggesting that they are most likely due to a specific block of gap junctions rather than to a nonspecific effect of this drug [31].

Table 1. Effect of MFA on the electrophysiological properties of precursor cells
inline image

Thus, the differences in passive membrane properties between isolated precursors and precursors in contact are likely due to the presence of gap junctions.

Our sorted precursors consisted mostly of TAPs with some astroglial stem cells (see earlier). To directly assess the electrophysiological properties of stem cells, we also recorded from isolated LeX+/EGFRhigh precursors. We found that outward K+ currents (supporting information Fig. S5H, S5I) and passive membrane properties (supporting information Table S3) of selected LeX+/EGFRhigh were significantly different from those measured in neuroblasts but not in PS+/EGFRhigh precursors. Thus in vitro neuroblasts and precursors displayed distinct electrophysiological properties, whereas antigenically distinct precursors do not.

Lineage Progression of EGFR+ Cells In Situ

We next investigated whether EGFR+ cells behave as TAPs by analyzing their lineage progression in situ. To this end, we infected acute neonatal telencephalic slices with a lentivirus expressing enhanced GFP (eGFP) under the control of the EGFR promoter (pEGFR-eGFP; supporting information). Two days after infection, some eGFP+ cells were visible in a restricted region along the border of the lateral ventricle but not in the striatum (Fig. 4A). Most eGFP+ cells had a small, round, regular shape, whereas the remaining eGFP+ cells, similar to EGFRhigh cells in vitro, were larger with an elongated shape. All eGFP+ cells were EGFR+ and the vast majority was EGFRhigh cells (Fig. 4B–4D).

Figure 4.

EGFR+ cells behave in situ as transit-amplifying precursors. (A): Photomicrographs of organotypic slices immunostained with GFP antibodies 2 days (DIV2) after lentiviral transduction. Note the eGFP+ cells are all localized in the SVZ and not in the Str. Scale bar: 100 μm. (B): Fluorescence-activated cell sorting analysis of infected slices dissociated at DIV2. (C): Percentage of plated cells undergoing clone formation within the indicated sorted populations. (D–F): Confocal photomicrographs of organotypic slices upon double immunostaining with the indicated antibodies at DIV2 (D, E) and DIV3 (F). Scale bar = 25 μm. Abbreviations: D, dorsal; DAPI, 4′,6-diamidino-2-phenylindole; DCX, doublecortin; DIV, day in vitro; EGF, epidermal growth factor; eGFP, enhanced green fluorescent protein; EGFR, EGF receptor; I, internal (septum); L, lateral; NG2, neuroglia proteoglycan 2; PI, propidium iodide; Str, striatum; SVZ, subventricular zone; V, ventral.

Next, we sorted eGFP+/EGFR+, eGFP, EGFRhigh, and EGFRlow cells from slice cultures dissociated at day in vitro (DIV)1 and DIV2 as well as EGFRhigh and EGFRlow cells from the dissociated neonatal SVZ and performed clonal analysis on each group. We found that the clonogenic ability of noninfected EGFRhigh cells progressively decreased over the culturing period from around 40% directly after dissociation to around 20% at DIV2 (Fig. 4C). At DIV2, the decrease in clonogenic ability was even more pronounced in eGFP+/EGFR+ cells, although they still expressed EGFR (Fig. 4D). Instead, culturing did not affect the clonogenic activity of EGFRlow cells. Around 2% of EGFRlow cells and eGFP cells (representing cells expressing low levels of EGFR and/or noninfected cells) isolated from DIV2 slices were clonogenic as their counterparts isolated from the dissociated SVZ (data not shown). This suggested that the loss of clonogenic activity in eGFP+/EGFR+ and EGFRhigh cells isolated from DIV2 slices is not a consequence of culturing, but it likely reflects the normal process of differentiation and cell cycle exit of TAPs, which are overrepresented in these groups. The difference in clone ability between these groups at DIV2 is likely due to the fact that newly born cells are still entering the EGFRhigh but not the eGFP+/EGFR+ group over this time period. In support of this hypothesis, we found that the proportion of EGFR+ cells coexpressing Ki67, which identifies cycling precursors, drops from around 60% in acute slices to only 28% in DIV2 slices (supporting information Fig. S6A, S6B, S6E). However, at this age, around 50% expressed the radial glial marker RC2 [32] and most EGFR-expressing cells were still Dlx2+ (supporting information Fig. S6C–S6E). Interestingly, strongly RC2+ cells displayed a size larger than that of less immunoreactive cells, suggesting that RC2 expression is slowly downregulated during differentiation. Consistent with our previous analysis, eGFP+/EGFR+ cells on DIV2 slices were immunonegative for Olig2 or NG2 (Fig. 4E) and the majority (85% ± 7.6%) were doublecortin (Fig. 4D). Furthermore, after additional 24 hours, most of the transduced cells (88.1% ± 7.9%) turned doublecortin+ and started to downregulate the expression of EGFR, which at DIV3 was expressed by 70.0% ± 7.9% of the eGFP+ cells (Fig. 4F). Taken together, these results showed that most EGFR+ cells only transiently display neural stem cell properties and give rise to neuroblasts by generating an intermediate progenitor, which expresses Dlx2 but is Olig2/NG2/doublecortin.

EGFR+ Cells Analyzed In Situ Consist of Two Functionally Distinct Cell Types

We found that after 2 days the majority of eGFP+/EGFR+ cells in neonatal slices have lost their clonogenic ability and represent noncycling Dlx2+/Olig2 cells that will eventually become doublecortin+ neuroblasts. To better define the identity of eGFP+/EGFR+ cells at DIV2 we next used electrophysiology to investigate their functional properties. As aforementioned, eGFP+/EGFR+ cells could be divided into small and large. The first ones (Fig. 5A3, 5A4), representing more than 60% of eGFP+ cells, displayed a small CM (4.3 ± 0.5 pF) and a high RM (1,843 ± 228 MΩ), in accordance with their small size (Fig. 5B, empty circles), similar to sorted neuroblasts (supporting information Table S2; Fig. S5). Their VR was −33.33 ± 0.95 mV (Fig. 5C, empty circles). Small eGFP+/EGFR+ cells in situ resembled sorted neuroblasts also in terms of densities of outwardly rectifying currents (137.69 ± 46.63 pA/pF) and IV relationship (compare Fig. 5D, 5E and Fig. 3B, 3C). In contrast, large cells (Fig. 5A1, 5A2) had a CM (23.96 ± 1.71 pF) and a RM (865.80 ± 341.13 MΩ) (Fig. 5B, filled circles) more similar to that of sorted precursors. Finally, their VR (−53.8 ± 7.8 mV; Fig. 5C, filled circles) was much more hyperpolarized than that of small eGFP+ cells.

Figure 5.

Electrophysiological properties of eGFP+ cells. (A): DIC (A1, A3) and fluorescence (A2, A4) photomicrographs of large (A1, A2) and small (A3, A4) eGFP+ cells (indicated by arrowhead or patch electrode) upon patch-clamp recordings in DIV2 slices. (B, C): Plots of RM versus CM ([B], n = 18) and of VR versus CM ([C], n = 13). Empty and filled circles represent small and large eGFP+ cells, respectively. Representative currents (D) and average IV ([E], n = 8) evoked by voltage steps, as in Figure 3B, in small eGFP+ cells. (F): Ca2+IV from 155-ms ramps in eGFP+ cells (n = 4). (G): Differential interference contrast (G1), GFP (G2), and Alexa-594 (G3) images of a large cell upon recording (CM = 25 MΩ); note the dye coupling. Scale bar = 20 μm. Abbreviations: CM, membrane capacitance; eGFP, enhanced green fluorescent protein; I, current; RM, input resistance; V, voltage; VR, zero-current resting potential.

Consistent with their high RM, no dye coupling was observed when Alexa fluor 594 hydrazide was introduced in small eGFP+/EGFR+ cells (n = 12) with the patching pipette. In contrast, upon patching large eGFP+/EGFR+ cells, we observed dye diffusion during the recording (2 of 13) (Fig. 5G). Notably, coupled cells had the smallest RM values (Fig. 5B). Finally, using the solutions and protocols described earlier, we were able to elicit large inward Ba2+ currents in striatal neurons [33] (not shown), but not in either small or large eGFP+/EGFR+ cells (n = 5; Fig. 5F), showing that functional VDCCs are not present in EGFR-expressing cells in situ.

Thus, our patch-clamp analysis in brain slices, 2 days after infection with the pEGFR-eGFP lentivirus, identified two different populations of eGFP+/EGFR+ cells: a majority of small cells showing passive membrane properties and outward currents similar to neuroblasts purified by FACS, and the rest were large cells whose morphology, larger CM, and occasional gap-junction coupling resembled sorted precursors. The fact that small eGFP+/EGFR+ cells were mostly doublecortin (Fig. 4D) and did not display functional VDCCs suggested that they were not yet fully differentiated neuroblasts.


In this study we report the first direct antigenic and functional characterization of TAPs in the neonatal murine SVZ. Figure 6 summarizes the electrophysiological characteristics associated with lineage progression of stem cells to neuroblasts.

Figure 6.

Lineage progression from stem cells to neuroblasts. Shading represents changes in the levels of expression of markers. Clonogenic ability, gap junction coupling, and large potassium currents are transiently expressed by TAPs. Voltage-dependent Ca2+ channels, reduction of K+ current density along with changes in cell size, and passive membrane properties are associated with differentiation. Abbreviations: CM, membrane capacitance; Dcx, doublecortin; EGF, epidermal growth factor; EGFR, EGF receptor; GFAP, glial fibrillary acidic protein; LeX, Lewis X; NG2, neuroglia proteoglycan 2; PSANCAM, polysialylated neural cell adhesion molecule; RM, input resistance; TAPs, transit-amplifying precursors.

Both Stem Cells and Neuroblasts Express PSANCAM

Our finding that most clonogenic precursors in the postnatal SVZ express PSANCAM is consistent with previous reports showing that PSANCAM during development and at early postnatal stages is expressed by the majority of perinatal SVZ cells including neuroblasts [1], glial progenitors [34, 35], and a subset of NG2+ cells defined as TAPs [7]. Despite being downregulated after birth, PSANCAM remains elevated in neurogenic areas [34] and has been used to isolate from various brain regions highly proliferative precursors displaying multipotency in vitro [36, 37] and in vivo [38]. In particular, Marmur et al. found that in the postnatal cortex all multipotent precursors are PSANCAM+, whereas in the SVZ the multipotent cells equally consist of PSANCAM and PSANCAM+ precursors. We instead found the vast majority of clonogenic SVZ precursors to be PSANCAM+, with few PSANCAM clone-forming cells. It is possible that the sequential immunopanning technique used by Marmur et al. did not allow the purification of cells expressing lower PSANCAM levels. Indeed, we found a wide range of PSANCAM expression in SVZ cells especially when analyzed immediately after dissection, indicating that enzymatic dissociation may also affect PSANCAM expression.

EGFR-Expressing Cells in the Neonatal SVZ Represent Stem Cells and TAPs

We showed here that EGFRhigh cells are not homogeneous but are composed of two different populations: the majority expressing Dlx2 but not markers associated to the glial or neuronal lineage, and a smaller subset displaying antigenic characteristics of astroglial stem cells [1–3]. This is in agreement with a recent characterization of GFAP-expressing cells in the postnatal SVZ [9]. As Dlx2 expression and clone-forming ability together define TAPs in the adult SVZ, we concluded that most clonogenic EGFRhigh cells in the neonatal SVZ represent TAPs. Importantly, this conclusion was based on antigenic and functional characterization of EGFRhigh cells and in particular on the analysis of their lineage progression in situ.

The analysis of NG2+ cells in the neonatal SVZ has previously revealed that as TAPs they express Mash1 and proliferate rapidly [7]. Furthermore, although in vivo most NG2+ cells are oligodendrocyte precursors [39], they also contribute GABAergic interneurons to the hippocampus [7]. The relationship between NG2+ cells and the EGFRhigh cells is unclear. Several lines of evidence suggested that, although they both express Mash1 and proliferate rapidly, NG2+ and EGFRhigh cells represent two distinct cell populations. First, using specific antibodies, we and others [9] were unable to detect the colocalization of EGFR with NG2. Furthermore, only few EGFR+ cells expressed Olig2. Second, it has been shown that in vivo EGFR+ cells proliferate more than NG2+ cells [9]. Third, we showed that putative TAPs within the EGFRhigh population do not express markers associated with astroglial stem cells such as LeX and GFAP. Instead, it was reported that the only NG2+ cells capable of clone formation, a defining property of TAPs [2], were LeX+ [7]. The interpretation of this previous observation is complicated by the fact that it refers to cells isolated from the whole brain and not specifically from the SVZ. Furthermore, the incidence of clonogenic cells, and therefore putative TAPs, within the NG2+ population is unknown. However, if clonogenic NG2+ cells in the SVZ will be confirmed to be LeX+, it will be a further indication that they are distinct from the EGFRhigh cells described here as TAPs.

Isolated Stem Cells and Neuroblasts Are Functionally Different

We showed here that, besides antigenic and clonogenic characteristics, neuroblasts and precursors show distinct electrophysiological properties. Compared with neuroblasts, isolated precursors display a significantly larger CM and lower RM and are more hyperpolarized. Our measurements of CM and RM in isolated precursors are comparable to those of GFAP+ cells recorded in situ in the presence of a gap junction blocker [11] and of intermediate progenitors measured in cortical embryonic slices (531 MΩ) [40], which is consistent with the predominance of TAPs in these populations. The VR of our isolated precursors (−37 mV) is also comparable to that previously reported in precursor cells in vitro [15, 41] and closely matches the onset of K+ channel activation in these cells. A more hyperpolarized VR, measured in SVZ GFAP+ precursors in culture [42] and in situ [11], was due to the presence of an inward rectifier K+ conductance. We observed only depolarization-activated outward K+ currents in neuroblasts and precursors. The higher density of outwardly rectifying K+ current in precursors than in neuroblasts may be responsible for their slightly more negative VR. This property may facilitate precursor proliferation, as blockers of delayed rectifier K+ channels have been shown to suppress precursor proliferation [42]. In addition, EGF, as other tyrosine kinase receptors [43], regulates the phosphorylation state of delayed rectifier K+ channels [44, 45], increasing their currents [46]. Considering the differences in EGFR expression between stem cells and neuroblasts, a differential modulatory effect of EGF on delayed rectifying K+ current is expected and may also explain the clonogenic effect which we observed for EGF, specifically on EGFRhigh cells via phosphorylation and activation of TEA-sensitive delayed rectifier channels.

The presence of electrical coupling and the absence of VDCCs in isolated precursors further define them as electrically more immature than neuroblasts. Indeed, astroglial precursors are electrically coupled [11, 13], whereas L-type VDCCs correlate with and promote the neuronal differentiation in vitro [15, 16].

We found that, although antigenically different, PS+/EGFRhigh cells, most of which represent TAPs, and LeX+/EGFRhigh cells in vitro share a similar electrophysiological profile. Probably, this is due to the fact that sorted cells were maintained in the presence of high concentrations of EGF, which in culture may induce LeX+/EGFRhigh cells to generate TAPs and support their extensive proliferation while delaying lineage progression. Instead, in situ, in the absence of sustained EGFR signaling, small and large tagged EGFR+ cells displayed completely different passive membrane properties. They also differed with respect to gap-junction coupling that was confined to large cells. Interestingly, large EGFR+ cells were significantly more hyperpolarized than cultured precursors, with a VR intermediate between that of neuroblasts and that of GFAP+ cells recorded in acute SVZ slices (−88 mV) [11]. This suggested that EGFR+ large cells are radial glia-like precursors, which is consistent with the intense RC2 immunoreactivity observed in these cells. However, within 2 days more than 70% of the EGFR+ cells, though still expressing Dlx2, had exited the cell cycle and become smaller. Because small cells displayed highly homogenous electrophysiological properties and represented the majority of EGFR+ cells at this stage, they likely represent the nonclonogenic cells isolated from DIV2 slice cultures. Their antigenic profile, the passive membrane properties, and K+ current density indicate that such cells, despite being mostly doublecortin and not expressing VDCCs, belong to the neuronal lineage. This is confirmed by the fact that 1 day later most eGFP+ cells have differentiated into doublecortin+ neuroblasts. Interestingly, it has been shown that during the regeneration of the adult SVZ, neuroblasts are generated within a similar time frame [1].


Taken together, our data show that as in the corresponding adult germinal area, EGFRhigh cells are mostly TAPs and to a lesser extent astroglia stem cells. Importantly, we found that although the neonatal SVZ is a site of active gliogenesis [47] in this region, as in the adult SVZ, stem cells give rise to TAPs that mainly generate neuroblasts. Finally, we show that progression through this lineage is paralleled by distinct functional changes. These data identify distinct antigenic and functional properties of postnatal stem cells and their progeny which are likely to shape their physiology and lineage progression.


We thank Dr. Jaqueline Trotter for discussion and helpful suggestions and for kindly providing the neuroglia proteoglycan 2 antibody. T.C., K.O., and F.C. acknowledge the support of Landesstiftung Baden-Württemberg.


The authors indicate no potential conflicts of interest.