The CMZ expresses mRNA for sema3fa and its known receptors

We reported previously that sema3fa mRNA is present within the zebrafish CMZ [12], a stem cell niche in the peripheral retina that allows for ongoing growth of the zebrafish eye through adulthood [15]. Here, we used whole mount (WM) and fluorescent slide in situ hybridization (ISH) to further characterize sema3fa mRNA in the CMZ, and to identify potential receptors through which the protein acts. sema3fa mRNA is present in CMZ progenitors as this domain emerges at about 48 hours post-fertilization (hpf) [16], and remained through the end of embryogenesis (72 hpf) (Fig 1A) and on to larval stages (Fig 1B and 1C). sema3fa mRNA was present throughout much of the CMZ, though appeared to be absent from the most distal CMZ where the stem cells reside (Fig 1A–1C, black arrows). Indeed, double fluorescent ISH (FISH) for cyclin d1 (ccnd1), a marker of proliferative progenitors in the CMZ [17], alongside sema3fa revealed considerable overlap, with little or no expression in the distal retinal stem cell domain (S1A and S1B Fig; white arrows). Of note, while we reported previously that sema3fb is expressed along with sema3fa within the temporal zebrafish eye primordium at 18 hpf, expression of sema3fb is lost subsequently in the embryonic neural retina and mRNA is not present in the post-embryonic CMZ [18].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Expression of mRNAs for Sema3fa and its receptors in the larval and juvenile CMZ.

Plastic sections through the eyes of WM ISH of 3 (A, D, G, J, M, P) and 7 (B, C, E, F, H, I, K, L, N, O) dpf larvae for sema3fa (A-C) and potential receptors, nrp2a (D-F), nrp2b (G-I, P), plxna1a (J-L), and plxna1b (M-O). The CMZ (red dashed lines) is shown enlarged at 7 dpf (B, C, F, I, L, O). Peripheral-most CMZ lacks expression (arrows B, C, F, I). Q) Double FISH for nrp2b (red) and ccnd1 (green) mRNA in the 72 hpf CMZ. R) Schematic representing general organization of ccnd1 (blue) and nrp2b (yellow) expression in the CMZ. The two expression domains exhibit some overlap (purple). S-X) Double FISH on sections of 1-month retinas with a DNP riboprobe for sema3fa (S) and DIG probe for her4.1 (T; merge in U), and with a DIG probe for sema3fa (V) and DNP probe for nrp2b (W; merge in X). Expression of sema3fa in CMZ (dotted yellow line) indicated by yellow arrows. Orientation bar in A applies to A, D, E, G, H, J, K, M, N. D, dorsal; GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; L, lens; ONL, outer nuclear layer; outer plexiform layer, OPL; RPE, retinal pigment epithelium; V, ventral. Scale bar in A is 50 µm for A, D, E, G, H, J, K, M, N and 10 µm for all other panels.

https://doi.org/10.1371/journal.pgen.1011748.g001

Sema3s signal by binding to a Nrp-PlxnA heteroreceptor complex [14,19]. To determine if progenitors within the CMZ have the necessary receptors to respond to Sema3fa, we asked by WM ISH whether mRNAs for Nrp and Plxn were expressed by CMZ progenitors at 72 hpf and 7 days post fertilization (dpf). We focused on receptors that show high binding affinity to Sema3fa and are thought to mediate its actions [13,14]; Nrp2 and Plxna. We examined mRNA expression of nrp2 (nrp2a, and nrp2b) and plxna (plxna1a, plxna1b, plxna2, plxna3 and plxna4), and then performed plastic tissue sectioning to investigate expression within the CMZ.

At 48 hpf, nrp2a and nrp2b mRNAs are expressed by cells of the retinal ganglion cell (RGC) layer, and in the case of nrp2b by the CMZ [20]. The mRNAs of the receptors were generally restricted within the CMZ. In transverse sections, nrp2a (Fig 1D–1F) and nrp2b (Fig 1G–1I and 1P) were expressed in the central CMZ at both 72 hpf (Fig 1D, 1G and 1P) and 7 dpf (Fig 1E, 1F, 1H and 1I), and were reduced or absent from the distal CMZ (Fig 1F and 1I; arrows). Of note, paralogs nrp1a and nrp1b were also expressed by CMZ progenitors at both 72 hpf and 7 dpf (S1C–S1F Fig).

Zebrafish have two paralogs of plxnal, plxna1a and plxna1b. In the early optic vesicle (18 hpf), plxna1a is expressed by retinal pigment epithelial (RPE) and dorsal eye progenitors and plxna1b by some neural progenitors [21]. In transverse sections, mRNAs for the two paralogs were present throughout the CMZ at both 72 hpf and 7 dpf (Fig 1J–1O). Of note, we found mRNA for plxna2 (S1G Fig) and plxna3 [14,22], but not plxna4 (S1I Fig), within the CMZ at 72 hpf, while at 7 dpf mRNAs for all three receptors were present in the CMZ (S1H and S1J Fig). Thus, CMZ progenitors express the necessary receptors to respond to the Sema3fa the CMZ cells produce. To determine whether cycling progenitors might respond to Sema3fa, we performed double FISH for ccnd1 and nrp2b, the latter as a representative receptor that transduces Sema3fa signals both in vitro and in vivo [11,14,23]. nrp2b FISH signal partially overlapped with that of ccnd1; nrp2b mRNA was expressed by cells on the central side of the ccnd1 mRNA domain, and by more centrally located CMZ cells that presumably were exiting or had exited the cell cycle (Fig 1Q and 1R). These data argue that mRNAs for the receptors are generally in cells closer to the differentiated central retina than those that express sema3fa mRNA (Fig 1R). Importantly, in one-month old juvenile fish, sema3fa continued to be expressed in the proliferative CMZ marked by her4.1 (Fig 1S–1U), alongside progenitors that express the mRNA for the Nrp2b receptor that would allow them to respond to Sema3fa (Fig 1V–1X).

Eye size is reduced in sema3fa^ca304 fish

To assess a role for Sema3fa within the CMZ we took advantage of a CRISPR mutant allele we generated and previously characterized, sema3fa^ca304/ca304, which has a 2 base pair deletion in exon 2 that introduces a frame shift and is predicted to produce a truncated protein; Western analysis indicates that the mutant is null for Sema3fa protein [12]. Recently, we found that loss of Sema3fa results in a mild embryonic eye phenotype, with a small but significant disruption in the production of amacrine cells specific to the dorso-temporal retina [18]. Of note, the retinas of the mutants are properly laminated by 72 hpf and the RGC layer, and outer inner nuclear layer (INL) that contains the soma of bipolar and horizontal cells, appear unimpacted by the loss of Sema3fa. As such, we could reasonably assess a role for Sema3fa in the larval CMZ without concern that developmental disruptions were responsible for CMZ phenotypes we observed in the sema3fa^ca304 fish. Additionally, we took advantage of the fact that while sema3fa mRNA is restricted to the temporal eye embryonically [18], it is present in the CMZ of all eye quadrants by 72 hpf. As such, any CMZ-related phenotype that was present in the nasal CMZ would be unlikely to have arisen due to an embryonic defect.

Because the CMZ stem cell niche in the peripheral retina allows for ongoing growth of the zebrafish eye through adulthood, we asked whether eye growth was impacted in the absence of Sema3fa. Importantly, we reported previously that the size of the embryonic eye at 72 hpf was not impacted in the sema3fa^ca304 allele [12]. To determine if post-embryonic growth beyond 72 hpf was impacted by Sema3fa loss, we measured the size of the left eye viewed dorsally at 10 dpf (Fig 2A, 2B and 2E) and laterally at 1 month (Fig 2C, 2D and 2F). When normalized to the nose to swim bladder length, there was no significant difference in eye area between the two genotypes at 10 dpf (Fig 2A, 2B and 2E), but the eye area normalized to head size was significantly smaller in juvenile mutant fish as compared to their WT counterparts (Fig 2C, 2D and 2F). These data suggest an eye growth defect that emerges as the mutant fish age. In support, the circumference of the 20 dpf eye, as indicated by the circumferential length of the IPL measured in frozen retinal sections, was significantly shorter in sema3fa mutant fish than WT (N = 4 independent experiments; WT 443.6 ± 48.1 µm (S.D.), n = 26 eyes; sema3fa-/- 369.9 ± 78.4 µm, n = 25 eyes; p = 0.0002 Mann Whitney U test). In summary, the smaller eye that develops over time in the sema3fa mutants points to a role for Sema3fa within the CMZ.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Larval eye size and CMZ proliferating progenitors impacted in the sema3fa^ca304 mutant.

A-D) Dorsal view at 10 dpf (A, B) and lateral view at 1 month (C, D) of eyes (outlined in white dashed line) in WT (A, C) and sema3fa^ca304 (B, D) fish. E-F) Eye area normalized to nose to swim bladder length at 10-11 dpf (E) and to area of the head (measured from the nasal-most aspect of the head to the anterior boundary of the gills) at 1 month (F) for WT and sema3fa^ca304 fish. Pooled from N = 2 independent replicates. G-K) Cumulative EdU labelling of WT and sema3fa^ca304 CMZ by bathing 72 hpf larvae in an EdU bath for 1-12 hours. One hour EdU labelling (G, G′, H, H′) of the CMZ in transverse retinal sections, also processed for PCNA immunolabelling (G, H, G′′, H′′) and for Hoescht to label nuclei (blue). The yellow arrows indicate uniform, intense PCNA immunolabel across the CMZ in WT (G′′) but lowered PCNA immunoreactivity in the mutant central CMZ (H′′). Graphs showing the percentage of Hoescht-labelled nuclei in the CMZ that are PCNA+ (I), the area of intense PCNA expression normalized to CMZ area (J), and the percentage of CMZ Hoescht-labelled cells that are EdU+ for each time point (K). Data pooled from N = 2 independent replicates. L-O) Retinal cryosections of 7 dpf CMZ showing EdU-labelled progenitors after a 4-hour EdU pulse (L, M) and immunolabelling for pHH3 (N, O). P-T) Graphs representing quantitation from retinal sections of the area of PCNA immunolabelling for the dorsal and ventral retina (P; N = 2 independent replicates), numbers of EdU+ cells (Q; N = 3 independent replicates) and the area of the CMZ they occupy (R; N = 3 independent replicates) from horizontal retinal sections, numbers of pHH3 + cells (S; N = 3) and the mean distance between them (T; N = 3) from horizontal retinal sections. U-W) Quantitation (W; N = 3 independent replicates) of the numbers of EdU+ progenitors (8 hour pulse) in Sema3fa-Myc-positive and Sema3fa-Myc-negative dorsal CMZ in transverse retinal sections of heat-shocked 5 dpf larvae injected at the one-cell stage with a construct with sema3fa-myc under the control of the hsp70 promoter. The ventral (U) and dorsal (V, V′) CMZ of a single retinal section from a heat-shocked larvae, showing little (U) or Myc (V′; green) immunolabelling for Sema3fa-myc+ along with Hoechst (blue) and EdU (U, V, V′; red) label. For all graphs, values represent individual embryos, error is standard deviation, and Mann Whitney-U test was used to test statistical significance. IPL, inner plexiform layer; OPL, outer plexiform layer. Scale bar in C represents 500 µm for A, B and 1000 µm for C, D; scale bar in G is 10 µm for G-H, L-O and U-V.

https://doi.org/10.1371/journal.pgen.1011748.g002

Disruption of the organization of proliferating cells with Sema3fa loss

Alterations in proliferation could explain the reduced eye size in sema3fa mutants. To address this issue, we performed immunolabelling for Proliferating Cell Nuclear Antigen (PCNA) at 72 hpf; nuclei that stain uniformly and intensely for PCNA are in late G1 and early S-phase [24]. While most cells of the WT and sema3fa mutant CMZ expressed PCNA (Fig 2G, 2G′′, 2H, 2H′′ and 2I), the area of intense PCNA immunoreactivity normalized to the area of the CMZ was reduced in sema3fa^ca304 fish as compared to WT (Fig 2H′′ and 2J). Note that the CMZ was defined anatomically as the most peripheral region of the retina closest to the lens that extends until the edges of the inner and outer plexiform layers [25]. The elongated shape of progenitor cells further helped to demarcate the borders of the CMZ. To understand, if this reduction reflected changes in cycling progenitor numbers we performed EdU cumulative labelling [26,27], where 72 hpf larvae were incubated for progressively longer times between 1–12 hours in an EdU bath prior to using Click-iT to identify cells that were in or passed through S-phase over the course of the incubation period. These data were represented as a percentage of Hoescht-labelled cells in the CMZ that were Edu+ (Fig 2K). The number of CMZ cells that were in S-phase over an hour period, as normalized to the number of Hoescht-labelled cells in the CMZ, was reduced significantly in mutants vs. WT (Fig 2G′, 2H′ and 2K). This was also true for longer EdU labelling periods (Fig 2K). Indeed, the maximum labelling index (growth fraction, GF) as the proportion of cells that comprise the proliferating population of the CMZ, revealed by the plateau of the cumulative curve (Fig 2K), appeared reduced in the sema3fa mutants. This plateau was achieved in both genotypes by 4 hours (Fig 2K). The cumulative EdU labelling graphs allowed us to estimate the time when the maximum number of EdU+ cells were labelled (Tm), cell cycle length (Tc) and the length of S-phase (Ts) (slope = GF/Tc + Y; Tm = Tc-Ts). The first time point when the curve approached a slope of zero was taken as the Tm. To determine approximate cell cycle values, we fitted the first part of each curve, prior to reaching GF (1–4 hours), with a linear regression to identify the slope (WT, 0.719 = 4/Tc + .447, Tc = 15 hours, Ts = 11 hours; sema3fa-/-0.686 + 4/Tc + .322, Tc = 11 hours, Ts = 7 hours; N = 2 independent replicates, n = 6–8 larvae/point). These data suggest that the cell cycle and time spent in S-phase is potentially shorter in the mutants. This data, combined with the observation that the maximum labelling index is smaller in the mutants, may indicate that the more rapidly dividing mutant progenitors leave the cell cycle to give rise to post-mitotic neurons sooner than their WT counterparts.

With further development, the CMZ stabilizes and the numbers of mitotic progenitors in the retinal periphery decreases as the CMZ shrinks in size [15,28]. We examined if CMZ progenitor behaviour remained altered in the mutant fish. Similar to what we found at 72 hpf, at 7 dpf the area of the intense PCNA label was significantly reduced in sema3fa mutants as compared to WT (Fig 2P). Interestingly, when we performed a 4 hour EdU pulse at 7 dpf, which the cumulative labelling data (Fig 2K) suggests should catch most or not all dividing cells in the CMZ, we found slightly more S-phase labelled progenitors in sema3fa mutants than WT (Fig 2L, 2M and 2Q). The number of mitotically dividing cells in the late G2 and M phases of the cell cycle, as identified by phospho histone H3 (pHH3) immunolabelling of retinal sections (Fig 2N, 2O and 2S), however, was similar between the two genotypes. We noticed that the small numbers of CMZ EdU+ cells present at 7 dpf were distributed more broadly within both the nasal and temporal mutant CMZ than in WT (Fig 2L and 2M). Indeed, the average area of the CMZ occupied by EdU+ cells in sema3fa^ca304 fish was increased significantly when compared to WT (Fig 2R). The same was true of pHH3 + cells, which were clustered in the WT CMZ but spread out within the mutant CMZ (Fig 2N and 2O), as quantitated by measuring the average distance between pHH3 + cells (Fig 2T).

To test whether Sema3fa directly controls progenitor proliferation we used a heat shock promoter to express a myc-tagged form of Sema3fa within the retina at 5 dpf to avoid disrupting embryonic retinogenesis. We injected a hsp70:sema3fa:p3E-MTpA Tol2 construct along with transposase mRNA into one-cell embryos, with expression of eGFP in the heart provided by myl7:egfp as the positive transgenesis marker. At 5 dpf, the embryos with green hearts were put at 38°C for 60 minutes to induce Sema3fa-myc protein expression. An 8-hour Edu pulse was performed 48 hours later, which should have caught all CMZ progenitors passing through S-phase. Note that this heat shock approach induced robust expression of Sema3a-myc protein in zebrafish embryos [29]. Regardless of whether there were no or significant numbers of Myc+ Sema3fa-expressing cells within the CMZ, even for the ventral (Fig 2U) and dorsal (Fig 2V′) CMZ of a single retinal section, similar numbers of EdU positive cells were present (Fig 2U, 2V and 2W). These data argue that Sema3fa is not sufficient to induce progenitor proliferation.

CMZ area and cell numbers are unchanged by Sema3fa loss

To understand why proliferating progenitors are disrupted with the loss of Sema3fa we next asked whether the CMZ was grossly normal in the mutants. We first asked if the CMZ emerged normally during development by performing ISH for the retinal progenitor marker vsx2 at 52 hpf [30] (Fig 3A and 3B). By this time point in most of both WT and mutants vsx2 mRNA was present in the CMZ and mainly absent from the differentiating, central neural retina. These data argue that CMZ development is not delayed in sema3fa mutants. By 72 hpf, however, while vsx2 mRNA was localized to the more distal WT CMZ, it was present more broadly within the sema3fa^ca304 CMZ (Fig 3C and 3D). To quantitate this difference, we normalized the area of the vsx2 expression domain within the CMZ to the area of the CMZ as measured in horizontal retinal sections. Importantly, the anatomical areas of both the nasal and temporal CMZ did not appear impacted substantially by the loss of Sema3fa (Fig 3E). The vsx2 domain, however, made up a significantly larger portion of the sema3fa^ca304 CMZ than seen in WT (Fig 3F), suggesting possible alterations in the progenitors. Importantly, the vsx2 domain expansion was evident in both the nasal and the temporal eye and so is unlikely to arise from defects in embryonic retinal development, as sema3fa mRNA is not present in the nasal embryonic retina [18]. Note that when we considered only the most highly expressing vsx2 progenitors of the distal CMZ, there was only an increase in the vsx2 area in nasal (Fig 3G), but not temporal, dorsal or ventral (Fig 3G and 3H) CMZ of mutants as compared to WT.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Gross CMZ development and anatomy unimpacted by Sema3fa loss.

A-D) Retinal sections of WM ISH for vsx2 that becomes restricted to the proliferative retinal periphery in WT (A) and sema3fa^ca304 (B) fish at 52 hpf. At 72 hpf (C-D), vsx2 is expressed at high levels by distal progenitors of the CMZ in both genotypes, but is also expressed at low levels broadly through the mutant nasal CMZ (arrows). Note normal expression of vsx2 by INL cells in both genotypes. n’s represent numbers of fish with represented expression pattern. E-H) Quantitation of the vsx2 expression domain within the CMZ. In sections through the central retina, the vsx2 ISH expression domain and CMZ were outlined, using the edge of the inner and outer plexiform layers to mark the central limit of the CMZ; average CMZ area (E), total vsx2 area normalized to CMZ area (F), and high vsx2 expression area normalized to CMZ area (G, H). I-J) Hoechst labelling of nuclei in horizontal cryostat retinal sections of 7 dpf WT (I) and sema3fa^ca304 (J) nasal (I, J) and temporal (I′, J′) CMZ. K-O) Quantitative analysis of WT and sema3fa^ca304 CMZ at 7 dpf from horizontal sections; (K) CMZ area; (L) Number of Hoechst-labelled nuclei; (O) Average distance between nearest neighbours, represented in M and N. INL, inner nuclear layer; IPL, inner plexiform layer; nr, neural retina; L, lens; RPE, retinal pigment epithelium. For all graphs, data is from N = 3 independent replicates, error bars are standard deviation, values represent individual embryos, and statistics represent Mann Whitney U-test. Scale bar in A is 150 µm for A, B and 40 µm for C, D, I-J, M, N.

https://doi.org/10.1371/journal.pgen.1011748.g003

To determine whether the size of the CMZ and numbers of progenitors remained similar between genotypes with ongoing retinal growth, we assessed in a masked manner Hoechst-labelled nuclei in horizontal cryostat sections of WT and mutant 7 dpf eyes (Fig 3I and 3J). In the absence of Sema3fa, the average areas of the CMZ in both the nasal and temporal retina were not significantly different between WT and mutants (Fig 3K). In agreement, the total number of Hoescht-labelled nuclei within the CMZ was similar in the two genotypes (Fig 3L). Further, there was no significant difference in the average distance between nuclei in the WT and mutant CMZ (Fig 3M–3O). These data suggest that the numbers and packing of cells within the CMZ were not impacted by the loss of Sema3fa.

Disruption in the expression of zone markers in the sema3fa mutant CMZ

The dispersion of proliferative markers (Fig 2) and the expansion of vsx2 mRNA (Fig 3C and 3D) in the sema3fa^ca304 CMZ led us to ask whether the spatial organization of the CMZ into transcriptional zones was impacted by the loss of Sema3fa. Previous reports suggest that the CMZ can be divided into four spatially discrete zones based on the unique expression of genes [31] (Fig 4C). We used the previous descriptions of gene expression within the CMZ of fish and frogs to characterize the zonal organization of the CMZ in the absence of Sema3fa [25,32,33]. These analyses were performed by WM ISH on 3 and 7 dpf zebrafish larvae, followed by plastic tissue sectioning to visualize and quantitate expression domains within the CMZ. The advantage of the ISH approach was that it allowed us to understand the spatial features of gene expression within the mutant CMZ. To compare the size of expression domains in retinal sections, we normalized the area of the gene expression domain to that of the CMZ in that retinal section.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Disruption of retinal progenitor cell domain with loss of Sema3fa.

A-E) Zone 1 markers bmp4 (D, E) and col15a1b, at low (A, B) and high magnification (A′, B′), from plastic horizontal (A, B) and transverse (D, E) sections of WM ISH preparations. Schematic of the zonal structure of the CMZ (C). F-G) Average col15a1b (F; N = 2 independent experimental replicates) and dorsal bmp4 (G; N = 3 independent experimental replicates) expression domain areas normalized to CMZ area. H-M) Immunolabelling of distal CMZ cells for Crb2a/zs4 (H, I) and aPKC (J, K), and plastic sections of WM ccnd1 ISH (L, M). Arrowheads point to label (H, I, J, K) in the distal CMZ that contains stem cells. n’s represent numbers of embryos analysed that show WT-like expression (H-L), and dispersed ccnd1 expression (M; arrow). N) Fold change of col15a1b mRNA of sema3fa^ca304 relative to WT 72 hpf eyes. Each dot is an independent biological replicate. Paired two-tailed t-test performed on delta Ct values. O) Average normalized ccnd1 ISH expression domain area for 72 hpf nasal/temporal CMZ in horizontal sections (O; N = 3 independent replicates). P) Fold change of ccnd1 (Q) mRNAs of sema3fa^ca304 relative to WT 72 hpf eyes. Each dot is an independent biological replicate. Paired two-tailed t-test performed on delta Ct values. Q) Average normalized ccnd1 ISH expression domain area in 72 hpf dorsal CMZ in transverse sections (N = 3 independent replicates). For graphs in F, G, O, Q points are individual embryos, error is standard deviation, and statistics are Mann Whitney U-test. IPL, inner plexiform layer; L, lens; N, nasal; OPL, outer plexiform layer; RPE, retinal pigment epithelium; T, temporal. Scale bar in A is 100 µm for A, B; 30 µm for A′, B′, D, E, H-M.

https://doi.org/10.1371/journal.pgen.1011748.g004

Zone 1 is the distal region of the CMZ where retinal stem cells reside [25,32]. A marker for the stem cells of Zone 1 stem cells, col15a1b [34], was present in the peripheral-most region of the CMZ in both genotypes (Fig 4A, 4A′, 4B and 4B′). In the nasal CMZ, however, the col15a1b ISH domain was broadened in the mutants as compared to WT (Fig 4A′, 4B′ and 4F). In agreement, RT-qPCR analysis of cDNA collected from surgically isolated 72 hpf eyes indicated that col15a1b mRNA levels were upregulated significantly in the sema3fa^ca304 retinas as compared to WT (Fig 4N). Both genotypes, however, expressed similarly a second marker of Zone 1, bone morphogenetic protein 4 (bmp4), which is present in the distal portion of the embryonic mouse CMZ [35] (Fig 4D, 4E and 4G). In agreement, immunolabelling for polarity markers in the distal CMZ, the apical adherens junctions complex proteins Crumbs2a (Crb2a/zs4) [36,37] (Fig 4H and 4I) and atypical Protein Kinase C (aPKC) [38] (Fig 4J and 4K) appeared unaltered in Zone 1 of the CMZ. These data suggest that retinal stem cells are present in absence of Sema3fa, though some minor alteration of the nasal CMZ stem cell zone may be present.

Zone 2 was assessed by the expression of ccnd1 [17]. At 3 dpf, the ccnd1 domain in WT was compact and ovoid (Fig 4L), however, in sema3fa^ca304 fish the domain’s shape was irregular, with expressing cells located towards the central edge of the CMZ next to the neural retina (Fig 4M; white arrow). Further, cells appeared to show variable levels of ccnd1 expression, with more centrally-located progenitors showing reduced ccnd1 mRNA (Fig 4L and 4M; compare yellow arrows in M). We quantitated this expansion by measuring the areas of the CMZ and ccnd1 mRNA domains and found that the proportion of both the nasal and temporal CMZ that expressed ccnd1 was expanded significantly in mutants relative to WT fish (Fig 4O). While the area of the ccnd1 domain was expanded in the mutants, suggesting disorganization of the domain, the levels of ccnd1 mRNA were slightly decreased in dissected 72 hpf sema3fa^ca304 vs. WT eyes, as assessed by RT-qPCR (Fig 4P), potentially reflecting progenitors expressing variable ccnd1 mRNA levels. A disorganized ccnd1 + Zone 2 was also observed in the mutant CMZ in the dorso-ventral axis; a greater portion of the mutant CMZ expressed ccnd1 than seen in WT (Fig 4Q). Thus, Zone 2 was impacted in all quadrants of the mutant retina. Additionally, disruption of the ccnd1 + progenitors remained as the eye matured, with a significantly larger ccnd1 ISH domain normalized to the CMZ area present in the dorsal CMZ of 7 dpf mutants than WT (WT 34.4% ± 7.2%, n = 9 larvae; sema3fa^ca304 51.6% ± 7%, n = 9 larvae; N = 2 independent replicates, p = 0.0002, Mann Whitney U test).

We next assessed Zone 3, which contains cells that are leaving or left the cell cycle. Cells in both genotypes expressed mRNA for cyclin dependent kinase 1c (cdkn1c). Cdkn1c inhibits G1 Cyclin/Cdk complexes and is a negative regulator of cell proliferation, impairing movement of cells into S-phase of the cell cycle [39]. The mutant CMZ showed an irregularly shaped cdkn1c CMZ expression domain as compared to the compact domain of WT (Fig 5A and 5B), as reflected in significantly increased areas of cdkn1c expression in both the nasal and temporal CMZ of mutants (Fig 5C). Yet, similar to ccnd1, cdkn1c expression was slightly downregulated in the sema3fa^ca304 72 hpf eyes as compared to WT as assessed by RT-qPCR (Fig 5D). These data suggest that the cell cycle genes may be expressed at slightly lower levels in individual cells of the mutant CMZ, a phenotype not readily captured with the sections of WM ISH. Together with the ccnd1 data our results suggest that the organization of the progenitors within the sema3fa^ca304 CMZ is disrupted.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Disrupted organization of committed CMZ progenitors in the absence of Sema3fa.

WM ISH for Zone 3 markers shown in 7 µm plastic sections (A, B, E, F, I′, J′, K, L), z-stack of confocal optical sections (M, N), or lateral WM (I, J). A-H) cdkn1c (A, B) and atoh7 (E, F) ISH at 72 hpf, graphs of cdkn1c (C) (pooled data from N = 2 independent replicates) and atoh7 (G) (pooled data from N = 4 independent replicates) mRNA domain areas normalized to CMZ area; Points for graphs are independent biological replicates, errors are standard deviation, and statistics are Mann Whitney U-test. Fold change of cdkn1c (D) and atoh7 (H) mRNA levels in 72 hpf eyes relative to WT as measured by RT-qPCR (paired Student’s t-test of delta CT values, with each data point an average of 3 technical replicates of RT-qPCR performed on mRNA isolated from n = 30 eyes). I-J) Zone 3 hes6 marker in the 72 hpf eye in WM (I, J) and in plastic section (I′, J′). Expansion of hes6 in situ label to the most peripheral CMZ adjacent to the lens in sema3fa^ca304 eye (yellow arrows). n’s refer to the numbers of eyes with significant (J) or little or no (I) hes6 ISH label in the CMZ immediately adjacent to the lens. K, L) Plastic section of Zone 3 neurod4 mRNA in the 72 hpf CMZ showing expansion (yellow arrows) of the neurod4 label towards peripheral CMZ. M, N) Confocal images of 72 hpf CMZ processed for WM double FISH for ccnd1 (red) and neurod4 (green). There is more co-expression in mutant than WT (compare arrows in M and N). n’s refer to the numbers of larvae showing WT levels of co-expression (M) and enhanced co-localization (N). O, P) Double FISH for ccnd1 (O, P) and atoh7 (O′, P′), and merge (O′′, P′′; green-ccnd1, red-atoh7) for WT (O) and mutant (P) 72 hpf CMZ. Q-R) Quantitation of Hoescht-labelled nuclei in ccnd1+ (Q) and atoh7+ (R) expression domains in transverse retinal sections processed for double FISH for ccnd1 and atoh7. Points for graphs are individual larva, errors are standard deviation, and statistics are Mann Whitney U-test (N = 2 independent experiments). Scale bar in A is 20 µm for all panels except I, J (100 µm).

https://doi.org/10.1371/journal.pgen.1011748.g005

To further assess the organization of CMZ progenitors, we investigated the expression of transcription factors that turn on in cells of Zone 3 as they become post-mitotic and begin to differentiate. atoh7 is a transcription factor of the basic helix-loop-helix family that is essential for the genesis of RGCs [40]. As described previously, at 72 hpf atoh7 mRNA was restricted to the central region (Zone 3) of the WT CMZ [41]. Like cdkn1c, the atoh7 expression domain in the sema3fa^ca304 CMZ was irregular in shape (Fig 5E and 5F), and the proportion of the CMZ that expressed atoh7 was increased significantly in the mutant vs. WT in both the nasal and temporal retina (Fig 5G). Additionally, a significant increase in atoh7 mRNA level was observed by RT-qPCR (Fig 5H).

To corroborate the disorganization seen in atoh7 expression in mutant CMZ at 72 hpf, we analyzed the expression of two additional CMZ Zone 3 markers. First, we assessed hes6 mRNA expressed in differentiating cells of the central CMZ [42]. In lateral views, hes6 mRNA was present in the peripheral portion of the CMZ adjacent to the lens in both the nasal and temporal retina of sema3fa^ca304 but not WT fish (Fig 5I and 5J, yellow arrowheads). This expanded expression domain in mutants was evident in horizontal sections, where hes6 was expressed by cells close to the lens (compare yellow lines, Fig 5I′ and 5J′). To quantitate the hes6 phenotype we measured on lateral WM images the area of the hes6 ISH that sat within the inner CMZ, ringed by the inner plexiform layer. This value was then normalized to the eye area for each embryo. A significant increase in the normalized hes6 ISH area was seen in the sema3fa mutant vs. WT eyes (pooled data from N = 2 independent replicates; WT hes6:CMZ, 20.5% ± 6.0% (standard deviation), n = 25 larva; sema3fa^ca304 29.3% ± 7.0%, n = 25 larva; p < 0.0001, Mann Whitney U test). A similar expansion towards the peripheral CMZ was evident with neurod4, as we had noted previously [18], a transcription factor of the basic helix-loop-helix family that mediates neuronal differentiation as well as amacrine cell fate specification in the developing retina (Fig 5K and 5L; compare yellow lines) [43].

While Zone 2 and Zone 3 marker domains were expanded in the CMZ of sema3fa^ca304 fish, we found the CMZ area was unchanged from WT (Fig 3E). Possibly in mutants there is some intermingling of the progenitors of the two zones. Alternatively, some progenitors may express markers of both zones. To address these possibilities, we first performed double FISH on WM WT and sema3fa^ca304 5 dpf fish for ccnd1 (red), a Zone 2 marker, and neurod4 (green), a Zone 3 marker. If progenitors intermingled, we expected mixed, but separate red and green FISH labels. In contrast, progenitors that expressed more than one zone marker would be yellow. Analysis of confocal images indicated that for the WT CMZ the Zone 2 and 3 markers were largely separate, though there was overlap of neurod4 ISH signal with the central edge of the ccnd1 domain (Fig 5M, arrow). The region of overlap was considerably expanded in the sema3fa^ca304 CMZ (Fig 5N, arrow). We also performed double FISH for ccnd1 and atoh7 mRNA on transverse 72 hpf retinal sections. While these two genes are thought to lie in Zones 2 and 3, respectively, we found in the WT CMZ that atoh7 mRNA sat within a small, centrally-biased region of the ccnd1-expressing Zone 2 (Fig 5O). Thus, certain previously characterized markers of Zone 3 progenitors may show some degree of overlap with progenitors of Zone 2. In the sema3fa^ca304 CMZ, both ccnd1 and atoh7 mRNA domains were expanded (Fig 5P), as described earlier. The FISH approach allowed us to use Hoescht label nuclei to directly determine whether the increase in area was due to more cells expressing either of the two markers. We counted the number of nuclei in ccnd1- and atoh7- expressing domains (Fig 5Q and 5R). We found that atoh7 appeared to be expressed by almost twice the number of cells (Fig 5R), but there no significant change in the numbers of ccnd1-expressing cells in the mutant as compared to WT (Fig 5Q). In both WT and mutants the atoh7 mRNA domain sat within the ccnd1 + domain, though the atoh7 mRNA signal occupied a larger extent of the ccnd1 + Zone 2 in the mutant fish (see double-headed arrows in Fig 5O′ and 5P′). These data suggest in mutants there is disrupted gene expression, where more ccnd1 + progenitors turn on Zone 3 markers such as atoh7, and as a result, progenitors that express atoh7 or neurod4 sit within regions of the CMZ they do not normally occupy. Thus, the loss of Sema3fa may alter the numbers of committed progenitors that express key genes that regulate the production of distinct neuron types.

Mutant CMZ produce smaller radial cohorts of progeny

The disrupted spatial localization of mutant progenitors could impact neurogenesis by altering the extrinsic signals and cell-cell interactions encountered by a progenitor. To address this possibility, we investigated neurogenesis by EdU pulse chase labelling. We exposed 7 dpf WT and mutant fish to EdU for 4 hours, and let the fish grow for two weeks. Retinas were sectioned and processed for EdU, and the numbers and laminar identity of cells within individual radial cohorts of the central retina assessed (Fig 6A–6D). Of note, in larval retinas the only proliferating cells other than those of the CMZ are a small number of neurogenic Müller glia that provide new rod cells for the central retina [44]; lone rod progenitors within the central outer retina were not included in our CMZ analysis. Thus, most EdU-labelled cells in the retina adjacent to the CMZ at 20 dpf would have derived from CMZ progenitors. We collected 12 µm cryostat sections through the entire eye and counted, in a fashion masked to genotype, the number and layer distribution of EdU+ cells in each individual radially arranged cohort of cells in every second retinal section. We found that at 20 dpf, the average cohort size in the central retina was significantly smaller in mutants than in WT (Fig 6E), suggesting that any one mutant CMZ progenitor on average produces fewer retinal neurons than its WT counterpart. One possibility to explain the smaller cohort size is that newborn neurons die preferentially in sema3fa mutants. Yet, we found no significant difference in the mean number of TUNEL positive cells in 7 dpf eyes of WT vs. sema3fa mutants (WT 2.0 ± 1.5 (S.D) TUNEL+ cells, n = 10 embryos; sema3fa-/- 1.8 ± 1.1 TUNEL+ cells, n = 10; p = 0.78, Mann Whitney U test).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Altered size and cell composition of CMZ-derived radial cohorts with loss of Sema3fa.

EdU labelling of retinas of 20 dpf zebrafish larvae incubated for 4 hrs with EdU at 5 dpf. A-D) Examples of radial cohorts of EdU-labelled cells in the central retina of separate WT (A, B) and sema3fa^ca304 (C, D) fish. Panels are separate examples from different fish. E) Number of EdU+ cells (counted in every second section) in each radial cohort averaged per retina (WT, n = 17 larvae; mutant, n = 14 larvae; N = 2 independent replicates). F-H) The averaged distribution of EdU+ cells across the three major layers of the neural retina within individual radial cohorts of each larvae; outer nuclear layer (ONL) (F), inner nuclear layer (INL) (G), and retinal ganglion cell layer (RGCL) (H). A larger percentage of cells of the radial cohorts reside within the RGC layer in the mutant as compared to the WT. I) Layer distribution of cells for each radial cohort of a representative WT and sema3fa^ca304 retina. J-M) Percentages of cells in each retinal lamina were calculated for individual radial cohorts, and these percentages averaged for each retina (WT, n = 17 larvae; mutant, n = 14 larvae). The standard deviations for these averages are graphed in J-M. For all graphs, data points represent individual embryos, error bars are standard deviation. Statistical analyses use Mann Whitney U-test for E-H and a Levene’s test for the equality of variance for J-M. INL, inner nuclear layer; ONL, outer nuclear layer; RGCL, retinal ganglion cell layer. Scale bar is 50 µm for A-D.

https://doi.org/10.1371/journal.pgen.1011748.g006

Retinal clones from both embryonic and CMZ progenitors can vary in size [15,31,45]. Yet, the distribution of progeny from multiple cohorts across the three retinal nuclear layers is comparable to the relative numbers of cells amongst the three layers; ~ 25% in the outer nuclear layer (ONL) that contains the photoreceptors, ~ 50% in the inner nuclear layer (INL) where the somas of interneurons and Müller glia reside, and ~25% in the RGC layer [45]. Note that we showed previously that sema3fa mutant retina exhibit normal lamination [12]. In investigating the layer distribution of cells (Fig 6F–6H), the cells of the RGC layer made up a greater percentage of the radial cohorts in the mutant retinas than was seen for WT (Fig 6H). Further, in comparing the cell composition of individual radially-oriented cell cohorts (as in Fig 6A and 6B) for a single representative WT and mutant eye, we observed more variability in the mutant vs. WT cohorts (Fig 6I and 6J). These data suggest that the distribution of labelled cells across laminae for an individual mutant cohort often did not prescribe to the ~ 25:50:25 ratio that was typically seen in the WT fish. To test this possibility, we compared statistically the variance (standard deviation) of the average percentage distribution of cells in each of the three retinal layers for each eye (Fig 6K–6M) and found that the contribution of RGCs to individual cohorts was significantly more variable in mutants than WT (Fig 6M).

In the absence of Sema3fa newborn retinal cells appear delayed in their exit from the CMZ

Interestingly, the EdU-labelled radial cohorts were often closer to the CMZ in the mutant retinas than WT (Fig 7A and 7B). To quantitate this phenotype, we measured the distance between the peripheral-most edge of the CMZ and the lateral edge of each radially oriented EdU-labelled cohort in the 20 dpf retinas (Fig 7D; d₁). We found that in sema3fa^ca304 fish the distance of the radial cohorts from the CMZ was reduced significantly (Fig 7C). To determine if newly born cells were slowed in their departure from the CMZ in the absence of Sema3fa, we measured the average distance of Edu-labelled progeny from the CMZ in central retinal sections after an 8-hour EdU pulse at 6 dpf and assessed 3 days later (Fig 7D; d₂). The leading edge of the EdU labelled cells was significantly further away from the peripheral edge of both the dorsal (Fig 7E) and ventral (Fig 7F) CMZ in WT vs. mutants. These data suggest that in the absence of Sema3fa newly born cells are delayed in their movement out of the CMZ. If true, one might expect the cells that express a potential receptor for Sema3fa, Plxna3 [14], would be found more distally within the sema3fa^ca304 CMZ than observed in their WT counterparts. This is what we observed in plastic sections of 72 hpf retinas, where there was an expansion of plxna3 label into more distal CMZ (Fig 7G and 7H; arrows). To quantify this phenotype, we measured the area of the plxna3 ISH domain and normalized it to the area of the CMZ (Fig 7K). We found a significant increase in the area of the CMZ take up by plxna3 ISH label in mutant as compared to WT CMZ. To confirm these data, we performed double FISH for the Zone 2 marker ccnd1 and plxna3 on retinal sections from WT and sema3fa^ca304 fish at 72 hpf. In this way we could compare plxna3 ISH label to that of a more peripherally expressed CMZ marker. In WT sections, there was a clear separation of the ccnd1 and plxna3 expression domains, with the plxna3 domain located more towards the central retina (Fig 7I). In contrast, in the mutant CMZ plxna3 cells sat immediately adjacent and within the ccnd1 domain (Fig 7J), arguing that Sema3fa-responding cells fail to move robustly out of the CMZ and that potentially progenitors begin to express a receptor normally expressed by centrally-located CMZ cells that have left the cell cycle.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Sema3fa may promote smooth movement of progenitors through CMZ by downregulating cell adhesion.

A-B) Radial EdU+ cohorts (green) that have moved away (dotted arrows) from the CMZ in WT (A) and sema3fa^ca304 retina (B). A 4 hour EdU pulse was performed at 7 dpf and retinal sections from larvae fixed at 20 dpf processed for EdU labelled cells. C) Average distance (d₁; D) of EdU+ radial cohorts from the CMZ in sema3fa^ca304 is reduced significantly as compared to WT. N = 3 independent replicates. D) Schematic showing measurement of the distance (d₁) moved from the central CMZ edge by radial cohorts of EdU+ cells (green) 2 weeks post-EdU label, and the distance (d₂) moved from the peripheral CMZ edge by the central-most cohort of EdU+ cells (green) 3 days post-EdU label. E-F) Graphs showing the average distance (d₂; D) of each radial cluster of EdU-labelled CMZ-derived cells from the peripheral-most CMZ, as measured in 9 dpf dorsal (E) and ventral (F) retina. N = 2-3 independent replicates. G-H) Retinal sections of the CMZ from WM ISH for plxna3 mRNA in WT (G) and sema3fa^ca304 (H) 72 hpf fish. n’s represent the numbers of larva showing either WT (G) or expanded (H; arrows) plxna3 expression. I-J) WT (I) and sema3fa^ca304 (J) CMZ processed by double FISH for plxna3 (green) and ccnd1 (red) mRNA. plxna3-expressing CMZ cells are found ectopically (arrows) in the peripheral CMZ and co-express ccnd1 (yellow). n’s represent the number of individual embryos that exhibit separated (I) or overlapped (J) ccnd1/plxna3 expression domains. K) Quantitation of plxna3 expression domain normalized to CMZ area of 72 hpf horizontal retinal sections (N = 3 independent replicates). L-O) Immunolabel for ß-catenin (L, M) and Cdh2 (N, O) in the 72 hpf CMZ. Arrows indicate reduced immunolabel of the central CMZ in WT but not mutant. n’s indicate the numbers of fish showing either preferential, bright label of the more distal CMZ (L, N) or immunolabel distributed across both the distal and central CMZ (M, O). P) Quantitation of the percentage of the CMZ exhibiting robust Cdh2 immunostaining fluorescent label. Data from N = 4 independent replicates. Q-S) Cdh2 immunolabelling of the CMZ in larvae injected at the 1-cell stage with a hsp70:sema3fa-myc construct and heat-shocked at 38°C for 60 minutes at 52 hpf and fixed 24 hours later. Quantitation (S) of the domain of high Cdh2 expression as a percentage of CMZ area in CMZ exhibiting Myc+ cells (R, R’; red) and no Myc+ cells (Q). Data from N = 3 independent replicates. For all graphs, dots represent individual fish, error is standard deviation, and Mann Whitney U-test was performed for statistical analysis. T) Schematic of the CMZ of WT and sema3fa^ca304 fish showing normal retinal stem cell (RSC) localization to the CMZ periphery, but disrupted organization of committed and proliferating (retinal progenitor cells; RPC) in the mutants. Fewer neurons (purple) are produced and move less distance from the CMZ in mutants. We propose a model whereby Sema3fa negatively regulates cell adhesion molecule expression by progenitors. As such, sema3fa^ca304 CMZ progenitors are more adhesive and do not move smoothly through the CMZ (red cells, arrows). Boxes show impact on the zonal organization of the CMZ. Scale bar in A is 100 µm for A, B and 25 µm for G-J, L-O, Q-R.

https://doi.org/10.1371/journal.pgen.1011748.g007

A possible explanation for the disorganization of the CMZ in sema3fa mutants is that normally Sema3fa negatively regulates the expression by progenitors of cell adhesion molecules. Indeed, SEMA3F inhibits E-cadherin mediated cell adhesion of cultured breast cancer cells [46]. In this scenario, by keeping the levels of adhesion molecules on progenitors low, Sema3fa would help the progenitors move smoothly through the CMZ in a distal to central direction. To explore whether adhesive signaling was potentially upregulated in mutant CMZ progenitors we performed immunolabelling for the adhesive signaling proteins ß-catenin and N-Cadherin (Cdh2) (Fig 7L–7O). ß-catenin controls the adhesion of retinal precursor cells in the embryonic mouse retina [47] and is required to maintain proliferation of embryonic zebrafish retinal progenitors [48], while zebrafish CMZ progenitors express Cdh2 [25]. We directly assessed protein expression within the CMZ by immunolabelling, rather than performing Western blot analyses of surgically isolated eyes, as both proteins are also expressed in the neural retina. Interestingly, we found both proteins appeared upregulated within the mutant CMZ as compared to WT, with an expansion of the immunolabelling to more central regions of the CMZ (Fig 7L–7O; compare arrows). Indeed, a greater percentage of the CMZ exhibited robust fluorescent label for Cdh2 in the absence of Sema3fa as compared to WT (Fig 7P).

To test this model further, we turned to the heat shock approach described earlier, which results in mosaic expression of Sema3fa-myc in the CMZ, which we verified by Myc immunolabelling. We predicted that overexpression of Sema3fa-myc would cause cells adjacent to Sema3fa-myc+ cells to downregulate expression of Cdh2. Indeed, while Cdh2 appeared evenly distributed in the more distal-central CMZ of injected fish that showed no Myc-immunolabelling within the CMZ upon heat shock (Fig 7Q), in CMZ with Myc+ cells, Cdh2 immunolabelling was uneven and showed gaps in expression (Fig 7R and 7R′; arrows). Indeed, the area of strong Cdh2 expression was decreased in CMZ that showed Myc+ cells as compared to the non-Myc expressing CMZ (Fig 7S). These data support the idea that Sema3fa negatively regulates adhesion between CMZ progenitors.

Ethics statement

All animal protocols were performed ethically as approved by the University of Calgary Animal Care Committee (ACC Certification AC23-0127).

Zebrafish husbandry

Zebrafish (Danio rerio) were housed according to standard conditions on a 14-hour light/10-hour dark cycle at 28°C. We incrossed wild type (WT) Tupfel Long Fin (TL) fish, sema3fa^ca304/ca304 fish, and for the latter their corresponding WT line produced from the same clutch used to generate the specific sema3fa^ca304/ca304 fish line [12]. Their embryos were raised in E3 medium supplemented with 0.25 mg/L methylene blue and staged by convention [64,65]. For certain experiments, pigment synthesis was inhibited by adding 0.003% (w/v) 1-phenyl-2-thiourea (PTU) to the E3 solution at 12–24 hpf.

Statistics

The underlying numerical data for all graphs and summary statistics in the text have been deposited in a Dryad repository: https://doi.org/10.5061/dryad.pc866t21r [72]. All statistical analyses were completed on Prism 10 software (GraphPad). An unpaired, non-parametric Mann Whitney U test was used for statistical analyses of data, except where indicated. A Levene’s test for the equality variance was used to compare the standard deviations of average cohort distributions between retinal laminae for individual eyes. Data sets for quantitation were analyzed, where possible, in a fashion where the researcher was masked to the genotype/experimental condition. N refers to the number of repeated experiments for a data set. The number of biological replicates or individual larvae used for a data set is the “n”.