Morphological characterization of human DGCs in neurologically healthy control subjects
Previous descriptions of the general morphological features of human DGCs are available in the literature [11, 20]. However, the extent to which the morphology of these cells varies depending on their positioning within the GCL remains poorly studied. To thoroughly characterize the morphology of DGCs located in distinct sub-regions of the GCL, morphometric determinations were performed on outer and inner DGCs separately. Compared to outer DGCs, inner DGCs exhibited shorter dendritic trees (U1, 127 = 1259, p < 0.001) (Fig. 1a-d) and reduced dendritic branching (Repeated measures ANOVA, Greenhouse–Geisser Interaction F1, 125 = 13.582, p < 0.001) (Fig. 1e). The soma area of inner DGCs was smaller than that of outer DGCs (U1, 127 = 1533, p = 0.020) (Fig. 1a-c, and f). Furthermore, inner DGCs presented fewer ending-tips (U1, 127 = 1435, p = 0.004) (Fig. 1g). The dendritic complexity (DCI index) revealed that inner DGCs presented less complex dendritic trees (U1,127 = 1477, p = 0.0499) (Fig. 1h), and a smaller dendritic span (U1,127 = 1500, p = 0.013) (Fig. 1i) than outer ones. A reduced percentage of inner DGCs with more than one primary apical dendrite was also observed (X21,127 = 14,997; p < 0.001) (Fig. 1j). Accordingly, inner DGCs showed fewer proximal branches than outer DGCs (1st order: U1,127 = 1284, p < 0.001; 2nd order: U1,127 = 1357, p = < 0.001) (Fig. 1k and l), despite having a similar number of distal branches. Pearson´s correlation test (Additional file 3: Figure S3 and Additional file 5: Extended data) revealed no effect of subject´s age on the morphology of DGCs in neurologically healthy control subjects.
Taken together, these results indicate that inner DGCs exhibit a less complex morphology than their outer counterparts—a finding that is compatible with a more immature neuronal phenotype [12, 15, 31].
Morphological alterations in the DGCs of patients with AD
To study the morphology of DGCs throughout AD progression, we studied patients distributed along the six neuropathological stages of the disease (Braak-Tau stages (I-VI)). For graphical representation and statistical analyses, subjects were grouped into the following categories: Control, Braak-Tau I/II, Braak-Tau III/IV, and Braak-Tau V/VI [32]. We analyzed total DGCs (Fig. 2) and outer/inner (Additional file 4: Figure S4) DGCs separately.
The mixed-effects model analysis revealed an effect of Braak-Tau stage (F = 8.823, p = 0.002) and cell positioning (F = 34.24, p < 0.001) on total dendritic length. Moreover, a statistically significant interation between Braak-Tau stage*cell positioning (F = 2.742, p = 0.042) pointed to differential vulnerability of this parameter in inner/outer DGCs to progression of the disease (Additional file 5: Extended data). In this regard, compared to control subjects, the total dendritic length of DGCs decreased progressively as Braak-Tau stages advanced (K3,639 = 43.86, p < 0.001), although it remained unchanged in patients at Braak-Tau I/II stages (Fig. 2a’-d’, and e). Although similar alterations were observed in outer DGCs (Additional file 5: Extended data), inner DGCs showed a decreased total dendritic length only at Braak-Tau V/VI stages (Additional file 4: Figure S4a and Additional file 5: Extended data). Sholl´s analysis revealed progressively decreased dendritic branching in AD patients at Braak-Tau III/IV stages and onwards (Repeated measures ANOVA, Greenhouse–Geisser Interaction F3,635 = 19.685, p < 0.001) (Fig. 2f). A similar decrease was observed in outer DGCs (Additional file 5: Extended data), although alterations in dendritic branching of inner DGCs were observed only in patients at Braak-Tau V/VI stages (Additional file 4: Figure S4g-h and Additional file 5: Extended data). The mixed-effects model analysis revealed an effect of Braak-Tau stage (F = 3.375, p = 0.048) on DGC soma area (Additional file 5: Extended data). In this respect, this parameter was reduced in AD patients at Braak-Tau I/II and V/VI stages (K3,639 = 37.43, p < 0.001) (Fig. 2a’-d’ and 2 g). Similar alterations were observed in outer DGCs (Additional file 5: Extended data), but no changes in this parameter were detected in inner DGCs.
The mixed-effects model analysis revealed an effect of Braak-Tau stage (F = 5.995, p = 0.012) and cell positioning (F = 38.931, p < 0.001) on the number of ending-tips (Additional file 5: Extended data). Post-hoc analyses revealed that this parameter decreased progressively as the disease advanced (K3,639 = 26.16, p < 0.001), reaching statistical significance in patients at Braak-Tau III/IV stages and onwards (Fig. 2h). In this case, similar alterations were observed for outer and inner (Additional file 4: Figure S4c and Additional file 5: Extended data) DGCs when these cells were analyzed separately.
The mixed-effects model analysis revealed an effect of Braak-Tau stage (F = 11.789, p = 0.001) and cell positioning (F = 4.139, p = 0.042) on the DCI. DGCs showed a progressive decline in this parameter (K3,639 = 47.07, p < 0.001), which reached statistical significance in patients at Braak-Tau III/IV and V/VI stages (Fig. 2i). Similar reductions in the DCI were observed in outer and inner DGCs (Additional file 4: Figure S4d and Additional file 5: Extended data). Similarly, the mixed-effects model analysis revealed an effect of Braak-Tau stage (F = 7.217, p < 0.001) and cell positioning (F = 59.121, p < 0.001) on the maximum dendritic span, which was reduced in patients at Braak-Tau V/VI stages (K3,639 = 19.96, p < 0.001) (Fig. 2j).
The percentage of DGCs with more than one primary apical dendrite was higher in AD patients regardless of Braak-Tau stage (Pearson X2-test, X2(3,639) = 12,844; p ≤ 0.005) (Fig. 2k and Additional file 5: Extended data), thereby indicating that this parameter is markedly altered early in disease progression. The number of proximal branches increased in patients at Braak-Tau I/II stages (1st order; K3,639 = 9.804, p = 0.02), whereas that of distal branches was reduced in patients at Braak-Tau III/IV and V/VI stages (Fig. 2l-m) (3rd order, K3,639 = 22.82, p < 0.001, 4th order, K3,639 = 37.84, p < 0.001; and 5th order, K3,639 = 13.69, p = 0.003). Similar alterations were observed in outer and inner DGCs (Additional file 4: Figure S4i-k and Additional file 5: Extended data).
To rule out the effect of any potential age-driven or inter-individual variation on the morphological changes detected, we performed Pearson´s correlation tests (Additional file 3: Figure S3) and a mixed-effects model analysis (Additional file 5: Extended data). None of the morphological parameters examined in the inner/outer/total DGCs showed statistically significant Pearson´s correlations with age (Additional file 3: Figure S3). Moreover, the mixed-effects model analyses revealed no major effect of age or inter-individual variations on any of the parameters studied (except for the maximum dendritic span, which showed limited subject and age-dependent variations). Detailed results of these analyses are included in the Additional file 5: Extended data.
Taken together, these results indicate that most of the morphological alterations exhibited by DGCs in AD patients start to be observed at Braak-Tau III/IV stages. However, two parameters, namely the reduction in the area of the soma and the presence of several primary apical dendrites, are altered in the initial stages of the disease (Braak-Tau I/II stages). Strikingly, the Braak-Tau stages in which inner and outer DGCs show specific morphological alterations differ, thereby suggesting the putative differential vulnerability of some of their morphological features to specific pathological mechanisms triggered in the AD brain.