We tested the differentiation potential of two unique sets of huNSPCs isolated at equivalent stages of gestation 
to determine whether they might differ in propensity to form differentiated neurons and glia and could be used to test the hypothesis that stem cell electrophysiological properties reflect biases in fate potential. In order to quantitatively determine progenitor cell fate potential, we induced differentiation of SC27 and SC23 huNSPCs and characterized differentiated cell phenotypes as an indicator of the fate potential of the progenitors. We measured generation of neurons and astrocytes, but not oligodendrocytes since they are not efficiently generated by either set of cells in our usual differentiation conditions 
. In comparison to SC23 huNSPCs, SC27 cells generated numerous cells with compact cell bodies and extensive processes that resembled neurons by phase contrast microscopy (, top panels) and expressed the neuronal marker MAP2 (, middle panels). In contrast, SC27 cells formed fewer cells that stained with the astrocytic marker GFAP than SC23 cells (, bottom panels).
HuNSPCs differ in neurogenic and gliogenic potential.
The identities of cells differentiated from huNSPCs were confirmed by determining the presence of neuron and astrocyte markers by immunostaining and assessing cell morphological criteria. We counted the percentages of cells that extended long processes and expressed the neuronal marker MAP2 or, for a more stringent standard of neuronal phenotype, co-expressed two neuronal markers (MAP2 and doublecortin)(Materials and Methods and Fig. S1a
). In both cases quantitation revealed that SC27 huNSPCs generate greater numbers of neurons than SC23 cells (). Undifferentiated cortical stem/progenitor cells can express the typical astrocytic marker GFAP (e.g. 
), making it difficult to distinguish these cells from astrocytes. We labelled cells with GFAP or a combination of GFAP and sox2, a stem/progenitor cell marker, to determine the numbers of GFAP-positive cells (), true astrocytes (GFAP-positive and sox2-negative)( and Fig. S1b
), putative progenitor cells (GFAP-positive and sox2-positive)(Fig. S1b
), and stem/progenitor cells (GFAP-negative and sox2-positive) (Fig. S1b
). SC27 huNSPCs had a lower propensity to generate astrocytes () and GFAP-positive/sox2-positive cells (Fig. S1b
) than SC23 cells, but had higher levels of cells that were only sox2-positive than SC23 cells (Fig. S1b
). Quantitation of differentiated cell phenotypes revealed statistically significant differences in the neuron and astrocyte differentiation potential of SC27 and SC23 huNSPCs (). SC27 and SC23 cells therefore could be used to test the hypothesis that electrophysiological properties indicate fate potential of undifferentiated huNSPCs.
Although SC27 and SC23 huNSPCs vary in differentiation potential, the undifferentiated cells are remarkably similar. Both sets of cells were isolated at the same stage of gestation, have comparable morphology (, top panels), and express similar levels of the stem cell markers nestin and sox2 (
and , bottom panels). The undifferentiated NSPCs were in suspension when assessed by DEP to determine electrophysiological properties and there was no significant difference in the diameters of SC27 and SC23 cells in suspension (SC27 14.22+/−0.23 µm, SC23 14.26+/−0.25 µm, p
0.90). Further analysis using confocal microscopy determined that there was no difference in cell perimeters of SC27 and SC23 cells (SC27 55.0+/−2.6 µm, SC23 52.1+/−1.8 µm, p
0.37). The similarities in diameter and perimeter suggest that SC27 and SC23 cells do not significantly differ in morphology when in suspension as used for DEP measurements. Both sets of huNSPCs are viable over several hours in the low ionic strength iso-osmotic buffer used for DEP (Fig. S1c
), as previously demonstrated for mouse NSPCs and encompassing the time necessary for DEP analysis of electrophysiological properties 
Unique sets of undifferentiated huNSPCs that are similar in morphology and marker analysis differ in dielectric properties.
Electrophysiological measurements of undifferentiated NSPCs were obtained using the DEP-Well system as described in detail previously 
(see also Materials and Methods
). Briefly, the responses of NSPCs to DEP forces were assessed over a full spectrum of frequencies ranging from 2 kHz to 20 MHz and DEP spectra acquired for each set of cells. DEP spectra of SC27 and SC23 huNSPCs demonstrate unique responses of the cells to specific DEP frequencies (), which is not due to size differences between the cells since their diameters are almost identical (SC27 vs. SC23, p
0.90; inset, ). Our previous work indicated that neurogenic mouse NSPCs experience positive DEP and are attracted to electrodes at higher frequencies than gliogenic mouse NSPCs, resulting in DEP curves for neurogenic cells that are right-shifted compared to those for their gliogenic counterparts 
. The huNSPCs analyzed here demonstrated the same pattern; SC27 cells that are more neurogenic produced DEP spectra that are right-shifted compared to the spectra of SC23 cells.
Measured DEP spectra of huNSPCs were fit to the single shell dielectric model 
to determine specific membrane capacitance (Cspec) and conductance (Gspec) values for the cells 
. These calculations take cell size into account, so the values of capacitance and conductance are cell size independent. Membrane capacitance is a measure of the ability of the membrane to store charge and generate a dipole in a frequency-dependent manner in DEP and is governed by both morphology and composition. The membrane capacitance values of SC27 and SC23 huNSPCs were significantly different from each other, with values for SC27 cells ~23% lower than those of SC23 cells (SC27 7.6+/−0.3 mF/m2
vs. SC23 9.9+/−0.2 mF/m2
, p<0.01). These cells are similar in many regards but differ in fate potential, suggesting that the membrane capacitance of huNSPCs inversely correlates with their neurogenic capacity (). Neurogenic fate potential in is shown by MAP2 staining of differentiated cells since this single phenotypic marker reflects the difference in fate potential between SC27 and SC23 huNSPCs shown by multiple neuronal markers () and multiple marker analysis of the cells analyzed in gave the same results (data not shown).
NSPC membrane capacitance correlates with neurogenic potential but membrane conductance does not.
To further test the association of membrane capacitance with NSPC neurogenic capacity, we analyzed mouse NSPCs that differ in ability to generate neurons. Mouse NSPCs isolated from the cerebral cortex at earlier stages of embryonic development (embryonic day 12, E12) formed more neurons upon differentiation than cells isolated at a later developmental stage (embryonic day 16, E16)(Fig. S2
). E12 and E16 mouse NSPCs were of similar size (E12 11.6+/−0.6 µm; E16 12.0+/−0.3 µm, p
20 or more cells) and show no differences in the expression of nestin, sox2, or GFAP 
. Neurogenic mouse NSPCs exhibited lower membrane capacitance than their gliogenic counterparts (E12 8.2+/−0.5 mF/m2
vs. E16 10.7+/−0.6 mF/m2
), showing that membrane capacitance also reflects fate bias of mouse NSPCs (). The inverse correlation between membrane capacitance and neurogenic potential of both human and mouse NSPCs shows that membrane capacitance is a label-free electrophysiological parameter that indicates NSPC fate bias.
The neurogenic capacity of huNSPCs decreases with continued passaging, so we used cells at different passage numbers to test whether huNSPC membrane capacitance dynamically predicts changes in cell fate potential. With increasing passage number, SC27 and SC23 huNSPCs significantly decrease neuron generation and increase membrane capacitance (, Cspec SC23 p7 vs. p22 p<0.01, p7 vs. p28 p<0.01, n
3 or more separate experiments with different sets of cells; %MAP2-positive SC23 p7 vs. p28 p<0.05; SC27 p7 vs. p19 p<0.01, p7 vs. p28 p<0.01, p19 vs. p28 p<0.01, n
6000 or more cells from 3 or more separate experiments). These data from two different sets of cells confirm that the biophysical parameter membrane capacitance continues to inversely correlate with and successfully predict huNSPC neurogenic potential as it dynamically shifts over continued cell passaging.
Although membrane capacitance correlates with neurogenic potential of NSPCs, membrane conductance does not. Membrane conductivity describes the potential of the membrane to transmit charge. When analyzed across several human and mouse NSPCs that differ in fate potential, the differences in specific membrane conductivity per unit area (Gspec) did not consistently correlate with the ability of the cells to generate neurons (). Furthermore, analysis of Gspec and neurogenic potential over passage of huNSPCs failed to show a consistent correlation between these measures (Fig. S3
). These findings suggest that the electrophysiological property membrane capacitance, rather than membrane conductance, reflects the fate potential of both human and mouse NSPCs.
The DEP spectra of neurogenic human NSPCs () and DEP trapping curves of neurogenic mouse NSPCs 
occur at higher frequencies than those of their respective gliogenic counterparts, making it possible that NSPCs with different fate biases have unique crossover frequencies (frequency at which there is no net induced DEP force). The crossover frequency is of particular interest since DEP can separate cells that differ in this parameter. We used the changing fate potential of huNSPCs over passaging to test whether the crossover frequency can reflect huNSPC fate bias. The crossover frequency of huNSPCs significantly changed over increasing passage number and directly correlated with neuron generation from the cells (, crossover frequency huNSPC p7 vs. p28 p<0.01, p22 vs. p28 p<0.05, n
3 or more separate experiments with different sets of cells; the cells in are the same as shown in so the %MAP2-positive cells are included again here to demonstrate the correlation of crossover frequency with neurogenic ability). The size of the cells did not significantly change over passage number (SC23 diameter at p7 was 14.7+/−0.4 µm and at p28 was 15.7+/−0.4 µm, p
0.094; similarly, SC27 diameter at p7 was 14.6+/−0.3 µm and at p28 was 14.7+/−0.4 µm, p
0.823;) and computer simulations predicted that a difference in diameter of more than 10 µm would be necessary to shift the crossover frequency as seen for the SC23 cells at different passages (J. Lu, unpublished data). We tested whether the positive association of crossover frequency with neurogenic potential also applies to mouse NSPCs and found that neurogenic E12 mouse NSPCs had a higher crossover frequency than E16 gliogenic mouse NSPCs (). These data show that the direct correlation of crossover frequency with neuronal fate bias holds true for both human and mouse NSPCs.
NSPC crossover frequency reflects neurogenic potential.