In vivo imaging of adult neurogenesis
Supplemental Text:
Lentivirus transduction labels thousands of new cells. Such large numbers are critical in this setting because of the increased chances of getting labeled cells within the small imaging window. Since lentiviruses transduce numerous cell types in the SVZ niche (both mitotic and postmitotic), it seemed as a limited method for labeling precise birthdates. Therefore, it was critical to evaluate the "noise" level of birth dates in the GFP-expressing population in the OB. I thus carried out three different control experiments to assess this noise.
Control 1
In order to determine the time-frame of birth dates that the virally labeled neurons represent, I compared GFP expression to BrdU labeling. BrdU labeling served as the "gold standard" for labeling precise birth dates as the half-life time of BrdU is estimated to be several hours. I analyzed GFP and BrdU in coronal slices from the OB at different times (5, 10, 45 and 90 DPI) after virus injection (Suppl. Fig. 2a). Specifically, I compared the relative GFP expression and BrdU labeling in the intrabulbar anterior commissure (aci) vs. the rest of the OB layers (i.e. GCL, EPL and GL). The aci represents newborn cells since it is where newborn neurons enter the OB. The GCL, EPL and GL are the target sites representing the more stable neurons. As expected, at 5DPI BrdU appeared almost exclusively in the aci (Suppl. Fig 2b,f),and with consecutive days the cell numbers in the aci gradually decreased. At 10 DPI I found an equal number of cells in and out of the aci and by 90 DPI the aci had virtually no BrdU-labeled cells (Supp. Fig2c-f). GFP expression profile was surprisingly similar to that of the BrdU. Much like BrdU, the relative portion of GFP expressing neurons in the aci as compared to the target layers is high at 5DPI, evenly distributed at 10DPI and decreases in 45 and 90 DPI (Suppl. Fig. 2b-e, g). Although infected progenitor cells clearly continue to arrive into the OB even at 90 DPI, their numbers seem to dwarf the numbers of cells that are produced in the first month. The exact number is difficult to quantify since there is still high variability in the number of labeled cells between injected animals. I next carried out an additional control to analyze the level of GFP labeled newborn neurons arriving at the OBby time-lapse imaging.
Control 2
To determine the relative numbers of new neurons arriving to the GL, I carried a time-lapse imaging experiment of PGN migration in the GL. I compared PGN migration of newborn neurons at different times after virus injection. Time lapse imaging of newborn neurons at 10 DPI (24 hours intervals, total duration: 2-4 days) revealed a great deal of neuronal migration of newborn PGNs in the GL. Within 24-hour time windows, only 25% of the newborn neurons were stable and 75% (121/162 neurons, n=16 mice) were still migrating (Suppl. Fig. 3a,e). Several weekslater (starting at 36 DPI), the majority of cells (71%, 55/77 cells, n=5 mice) were stable (Suppl. Fig. 3b,e). By 90 DPI, almost all neurons (96% 87/91 neurons, n=4 mice) were docked in place and did not migrate between imaging sessions (Suppl. Fig. 3c,e).
Since these observations are based on cells appearing and disappearing from a limited imaging window, they might still reflect only changes in GFP expression or alternatively cell death. To rule out this possibility, I imaged neurons at shorter intervals where cells could be captured during migration. In all animals imaged (n=6 mice at intervals of 4.5 and 6 hours) cells could be clearly seen migrating both within and into the imaged volume (Suppl. Fig. 3d). This experiment shows that less neurons arrive into the GL with increasing days after virus injection.
Control 3
To partially test whether 45 and 90 DPI neurons are morphologically different from the existing neurons, random PGN's in a slice preparation were labeled intracellularly and their morphology compared to those of GFP labeled cells (Suppl. Fig. 4). A total of 10 PGNs were filled and compared to our database of 45 DPI and 90 DPI GFP expressing neurons. Larger variability was evident in the random filled neurons because a few neurons had morphology that was never seen in the newborn population (see e.g. the neuron with multiple dendritic trees in Suppl Fig.4b). However,most cells were generally similar in appearance. In addition, the newborn population was tested by others using classical electrophysiology and no differences were found (see references 15,16 in main text).
Taken together, these results support the argument that most labeled cells at 45DPI and at 90DPI represent stable neurons.
1