Fig 1.
Experimental paradigm and construct design.
(A) Timeline of experimental manipulations used in these studies. (B) Map outlining the genetic maps of the various retroviral vectors used in these studies. Each viral vector has the coding sequence of one or two genes in its multiple cloning site, which sits 5’ (upstream) of an IRES-eGFP sequence. The vectors are identified in name by their respective inserts: empty vector, Sox2, Pax6, and SEP (Sox2-E2A-Pax6). (C) Western blot performed on lysates from HEK293T cells transfected with the vectors, confirming protein production of each gene product at the expected molecular weights when stained with antibodies against Pax6, Sox2 and GFP. Lane 1: pLIA-Pax6, lane 2: pLIA-SEP, lane 3: pLIA-Sox2, lane 4: pLIA-GFP, lane 5: untransfected HEK cells, lane 6: molecular weight standards.
Table 1.
Antibodies used in the study.
Table 2.
Methods for Antibody Staining used in this study.
Fig 2.
Sox2, Pax6, and Ascl1 demonstrate the hierarchy of progenitor cell progression during MeBr-induced regeneration in mouse olfactory epithelium (OE).
(Col. A) Sox2, (Col. B) Pax6, (Col. C) Ascl1 (Mash1), and (Col. D) TuJ1 (neuron-specific tubulin) immunostaining in control mouse OE and at 1, 2, 4 and 7 days post MeBr-lesion (dpl). Note the reappearance and patterning of transcription factor expression. CK14 labeling marks the horizontal basal cells in the uninjured epithelium and the cells that derive from them at the early stages in the reconstitution of the epithelium, including ones that transition into Sox2/Pax6 (+)/CK14 (-) GBCs. Regeneration of olfactory sensory neurons is marked by the expression of neuron-specific tubulin (marked by Tuj1 staining). The vast majority of spared cells express both Pax6 and Sox2 up through 4 dpl. By this time Ascl1-expressing GBCs have made a limited reappearance and a few Tuj1 (+), neurons have emerged. Dashed lines mark the basal lamina. Scale bar in (A, uninjured) equals 10 μm and applies to all panels.
Fig 3.
Identification of cellular targets for replication-incompetent retroviruses, during lesion-induced regeneration of the OE.
(A-F) At 1 day post-MeBr lesion GBCs and duct/gland cells are predominant proliferating populations and putative retroviral targets as shown by immunostaining of tissue sections of OE from 1 day post-MeBr lesion with antibodies to Sox2 (which mark GBCs) and to Sox9 (which marks gland/duct cells) along with anti-CK14 and Ki-67, the latter marking proliferating cells. Single solid arrows mark Ki-67 (+), dividing cells that are Sox2 (+)/CK14 (-) in A-C and Sox9 (-)/CK14 (-) in D-F, and are found at a remove superficial to the basal lamina and correspond to GBCs or HBCs caught in transition to GBCs. All of the CK14 (+) cells are also Sox2 (+). In A-C, the solid arrow with one asterisk indicates a dividing HBC, while in D-F the solid arrow with two asterisks marks an HBC that is not dividing but now expresses Sox9, in addition to the Sox2 expression documented above. The presence of Sox2 (+)/Sox9 (+) basal cells may represent HBCs that are differentiating into duct/gland cells, which is observed after MeBr lesion (Leung et al., 2007). In A-F the hollow arrows indicate a proliferating putative duct/gland cell (in D-C) and a marker-confirmed duct/gland cell (in D-F). (G-I) Staining of sections of the normal mouse olfactory epithelium with Sox2, Sox9, and CK14 to mark specific cell types along with markers of proliferation, either Ki-67 labeling or EdU incorporation, for comparison with the composition of the epithelium after injury. On the basis of these staining patterns, Sox2 (+) cells located at a distance superficial to the basal lamina with weak or no CK14 (+) labeling are classified as GBCs, while those basal cells that are attached to the basal lamina and strongly positive for CK14 are identified as HBCs. Duct/gland cells or cells that are Sox9 (+) form a separate category. (J) The criteria enumerated above were used to count the percent of dividing cells that are GBCs, which are the vast majority under the lesion conditions used here. Dashed lines indicate the basal lamina. Scale bar in K corresponds to 10 μm and applies to all photomicrographs.
Fig 4.
Sox2 transduction produces the largest clones while Pax6 produces the smallest.
(A-D) GFP expression in representative whole mounts, demonstrating typical clone size following transduction with EV, Sox2, Pax6, and SEP retroviral vectors, respectively. Scale bar in (D) corresponds to 50 μm and applies to all photos. (E) The range in cell number per clone is depicted as a box-and-whiskers scatter plot, and the median clone size is indicated by the horizontal black line within the box. The number of clones analyzed were as follows: EV– 49, Sox2–111, Pax6–150, and SEP– 165. Non-parametric ANOVA was statistically significant (Kruskal-Wallis, p < 0.05) as were post-hoc pairwise comparisons using Conover-Iman tests of multiple comparisons using rank sums; * designates p < 0.05.
Fig 5.
Sox2 transduction increases the number of cells proliferating early on in the regeneration process, compared to control.
(A-D) Proliferation of transduced cells was assessed, across viral constructs at 8 days post-infection by labeling with GFP and EdU (to identify cells in S-phase at the time of euthanasia). Hollow arrows indicate examples of GFP (+)/EdU (+) cells. Note the large number of double-labeled cells with Sox2 transduction (B). Dashed lines mark the basal lamina, and the scale bar in (D) corresponds to 20 μm and applies to all of the images. (E) The mean number of EdU (+) cells is plotted as a percentage of the total number of GFP (+) cells (percentages were derived from count data justifying arcsin-transformation to allow for multiple testing by Kruskal-Wallis ANOVA, with Dunn’s method for pairwise comparisons, * p < 0.05).
Fig 6.
Pax6- and Sox2-encoding retroviral vectors differentially alter the cellular composition across clones.
(A-H) Tissues harvested 3 weeks after EV, Pax6, SEP and Sox2 RVV-transduction were stained with GFP, along with PGP9.5 (a neuronal marker) to illustrate the range in outcomes for each condition. Cell identity among clones was then determined relative to PGP9.5 staining. GFP-labeled cells were identified as sustentacular cells (Sus) when their cell bodies were found superficial to the band of PGP9.5 (+) OSNs, while GFP-labeled cells were identified as basal cells if they were deep to the band of neurons. GFP (+) cells were identified as duct/gland cells (D/G) when they were flattened, oriented as a chain of cells along the apical-basal axis, and/or extended deep to the basal lamina into the lamina propria (cf. Fig 3G and 3H). GFP (+) cells were identified as OSNs, if they co-labeled with PGP9.5. (E-H) CD54 was used to immunostain HBCs along with PGP9.5. (E-H) There is substantial variation in clonal composition. (E) The arrow indicates a mixed clone composed of OSNs and a large group of Sus cells atop the neurons. (F) The arrow indicates a Sus-only clone where the GFP-labeled cells sit atop the neurons. (G) The solid arrow with asterisk points to gland cells located deep to the HBCs and basal lamina. (H) Another example of a mixed clone composed of Sus cells (arrow) and neurons (hollow arrow with asterisk). Dashed lines mark the basal lamina (A-H) and the scale bar in (H) corresponds to 20 μm and applies to all of the panels. (I) For the various cell types, the data from each and every clone was used to generate a box-and-whiskers scatter plot; the median for each distribution is marked by the horizontal black line within the box. In the order shown in the graph they are: for basal cells– 1, 1, 1, 0; for OSNs– 8, 2, 0, 3; for Sus cells– 3, 2, 0, 1; for D/G (duct/gland cells)– 0, 0, 0, 0, respectively. Note the break in the ordinate of the graph to accommodate those clones that contain a markedly greater number of neurons, which occur exclusively with Sox2 transduction. Because all of the datasets shown here failed the Shapiro-Wilk test for normality, a Kruskal-Wallis One-Way ANOVA on Ranks was used, followed by Dunn’s Method for pairwise multiple comparisons. (J) For the group of clones as a whole, the percentage of clones that contain neurons for each form of transduction are indicated in the bar graph, and multiple pairwise comparisons (indicated by the horizontal lines) are significantly different (Fisher Exact test with Holm-Sidak multiple comparison correction, * p < 0.05).
Fig 7.
Sox2 transduction enhances Ascl1 expression, whereas Pax6 transduction leads to a reduction in Ascl1 expression and a reciprocal increase in the prevalence of downstream neural precursor markers.
(A) A proneural bHLH cascade operates during olfactory neurogenesis and defines a hierarchy of GBCs as they progress from transit amplifying (TA) progenitor cells into differentiated neurons, by way of two immediate neural precursor (INP) stages. (B-M) At 8-days post-infection (9-days post-lesion), tissue was immunolabeled for Ascl1, Neurog1, and NeuroD1 along with GFP in clones derived from EV (B, F, J), Sox2 (C, G, K), Pax6 (D, H, L) and SEP (E, I, M) RVV-transduction. Hollow arrows mark GFP (+) cells within a given clone that co-express Ascl1, Neurog1, or NeuroD1. In contrast, solid arrows mark non-transduced, i.e., GFP-lacking, cells that express these factors. Dashed lines mark the basal lamina and scale bar in K corresponds to 20 μm in panels B-M. (N-P) Quantitative analysis of the effect of the various transduction conditions on the percentage of cells in a clone that express Ascl1 (N), Neurog1 (O) and NeuroD1 (P). For the graph, the percentage of cells marked by joint expression of the individual transcription factor and GFP within each clone are graphed in a box-and-whiskers scatter plot; the median number of each distribution is marked by the horizontal black line within the box. For Ascl1+ GBCs, the median percentages for EV vs. Sox2 vs. Pax6 vs. SEP are 0, 36, 0, 20, respectively. For Neurog1+ GBCs, the median percentages were 0, 37, 17, 25, respectively. Finally, for NeuroD1+ GBCs, the median values were 0, 50, 33, 19, respectively. Multiple pair-wise comparisons (horizontal line) were found to be statistically significant using Dunn’s method (* p < 0.05) following the demonstration of overall significance using Kruskal-Wallis One-Way ANOVA tests. (Q) Percentage of clones that contain cells positive for labeling with the bHLH factors. To compare percentages statistically, data were first arcsin transformed prior to normality testing, standard parametric ANOVA, and post-hoc Holm-Sidak pair-wise testing (**p < 0.01). Asterisks signify comparisons to EV, except for the comparison in number of Ascl1-conatining clones between Sox2 and Pax6 vectors, which is designated by the orthogonal lines.
Fig 8.
Conditional deletion of Sox2 by way of RVV transduction produces smaller clones that contain fewer neurons than control.
(A) Cartoon illustration of retroviral construct used for transduction with Cre recombinase. (B, C) Representative clones from whole mount tissue harvested 3 weeks after infection. (B) X-gal staining demonstrates the typical clone size from control R26fl(stop)-LacZ indicator mice transduced with Cre. (C) GFP fluorescence expressed from the RRV, demonstrates the typical clone size from Sox2flox/flox transduced mice. (D) Representative histological section stained to detect β-galactosidase activity within a clone following Cre-transduction in a R26fl(stop)-LacZ control mouse; note the large numbers of neurons, which are situated broadly along the middle of the epithelium. (E-G) Representative sections from Cre/GFP-expressing clones in Sox2flox/flox mice that have been immunostained to characterize clonal composition. Dashed lines mark the basal lamina. (H) The range in the number of cells/clone is depicted in the form of a box-and-whiskers scatter plot, and the horizontal black line within the box indicates the median clone size, where the median values were 19 and 3, respectively. The distributions are significantly different (Mann-Whitney-U test, *** p < 0.001). (I) The proportion of infected clones containing neurons or not is depicted in the form of a stacked bar graph, gray and black respectively. The proportions are significantly different between wildtype and knockout (Z test, *** p < 0.001) (J) Compositional analysis plotting the number of the different types of cells in each clone, by a box-and-whiskers scatter plot; median clone size is indicated with a black line. Median values for control vs. knockout for basal, OSN, Sus, and D/G are 0 vs. 1, 2 vs. 0, 3 vs. 1, and 0 vs. 0, respectively. Note that both the number of neurons and the number of Sus cells are significantly reduced by Cre-mediated recombination in floxed Sox2 mice as shown by Dunn’s method for pairwise comparisons (indicated by the horizontal lines, * p < 0.05). Scale bar in (C) corresponds to 50 μm and also applies to B, scale bars in D-G correspond to 10 μm.
Fig 9.
Pax6 expression remains unaltered within Sus cells and HBCs upon conditional deletion of Sox2.
Clones from tissue harvested 3-weeks after Cre RVV-transduction of bigenic R26fl(stop)-LacZ; Sox2flox/flox mice were immunolabeled for the neuronal marker PGP9.5 (A) or the HBC marker CD54 (B), along with Pax6 and GFP. In clones containing GFP (+)/ PGP9.5 (-) Sus cells (A), situated above the neural strata (arrows), Pax6 expression remains indistinguishable from surrounding untransduced Sus cells. In clones containing GFP (+)/CD54 (+) HBCs (B), Pax6 levels remain indistinguishable from untransduced HBCs (arrows). Arrowheads mark the basal lamina and scale bar corresponds to 20 μm and applies to all panels.
Fig 10.
CreERT2-mediated excision of Sox2 in HBCs reduces clone size and neuron numbers.
Tamoxifen was administered to unlesioned (A) and lesioned (B) control K5CreERT2; Sox2flox/+; R26fl(stop)-LacZ mice, as well as to unlesioned (C) and lesioned (D) experimental K5CreERT2; Sox2flox/flox; R26fl(stop)-LacZ mice; tissue was analyzed three weeks into epithelial recovery. Sections are X-gal-stained to detect β-galactosidase activity within clones, stained for cytokeratin 14 to mark HBCs, and immunolabeled for Sox2 to confirm gene recombination. (A) In the mice that are heterozygous for the floxed Sox2 allele, the excision of one copy of Sox2 had no discernable effect in the absence of lesion. (A’) The inset shows Sox2 labeling in β-galactosidase-marked CK14 (+) basal cells, as expected (double thin arrows). (B, B’) In the heterozygous animals, excision of one copy of Sox2 had no effect on the composition of the clones that arise following MeBr-lesion from the activated HBCs. X-gal-labeled Sus cells maintain their prominent expression of Sox2 following their regeneration from the HBCs that have undergone genetic recombination, as expected (double thin arrows). (C) In the unlesioned, homozygous floxed Sox2 mice, the absence of Sox2 has no apparent effect. (C’) The inset shows the absence of Sox2 labeling in β-galactosidase-marked CK14 (+) basal cells, as expected (single arrows). (D, D’) However, when the Tamoxifen-treated, homozygous floxed-Sox2 mice were exposed to MeBr, the X-gal-stained clones resulting from lesion-induced activation of the HBCs were smaller and contained fewer neurons. X-gal-labeled Sus cells lack their usual prominent expression of Sox2 following their regeneration from the HBCs that have undergone genetic recombination, as expected (single arrows). (E) Comparing lesioned tissue between floxed Sox2 homozygotes and heterozygotes, the range in the number of cells/clone is depicted in the form of a box-and-whiskers scatter plot, and the median clone size is indicated by the horizontal black line within the box. The median value of clone size between heterozygote vs. homozygote is 10 vs. 4, respectively. The difference between homozygotes and heterozygotes is statistically significant (Mann-Whitney U test, *** p < 0.001). (F) Compositional analysis plotting the relative number of different cell types in each clone, reveals that the percentage of neurons (OSNs) is significantly decreased in the floxed Sox2 homozygotes (white bars) compared to heterozygotes (black bars) (Mann-Whitney U-test, *** p < 0.001), while the percentage of HBCs is increased (Mann-Whitney U-test, *** p < 0.001). Scale bar in A is 25 μm and applies to all panels except A’ and C’. The dashed lines mark the basal lamina.