Impact of Increased APP Gene Dose in Down Syndrome and the Dp16 Mouse Model

Gene dose-dependent increases in fl-hAPP in AD-DS and DS.

To evaluate hAPP expression and processing in brains of DS and AD-DS, and to compare the findings to those in other populations, we examined cases of non-demented (ND) populations without DS (male: n=12, age 74–94; female: n=8, age 58–102), DS cases with AD (AD-DS; male: n=4, age 46–70, female: n=7, age 45–62), DS without AD (DS; female: n=4, age 42–62) and AD without DS (AD; male: n=10, age 64–89; female: n=11, age 80–94). The ages for DS and AD-DS cases correspond to those at which people with DS are at markedly increased risk for AD, as supported by neuropathological staging (Table 1). Approximately 50% of people with DS are diagnosed with AD-DS at an age that is two or three decades younger than for sporadic AD [51], explaining the use of NDs in this age range for both the AD and DS cases. The demographics of cases are given in Table 1. APOE allele status was recorded in all AD cases, and in all but one ND; the E3 allele was the most frequent, with only one E4 homozygous AD female. Of a total of 16 DS cases, APOE status was recorded in 9. As in the general population, the E3 allele was most frequent; none were E4 homozygotes.

Protein lysates were prepared from the frozen frontal cortex. fl-hAPP and hCTFs levels were analyzed by SDS-PAGE and western blots. Anti-Tubulin, anti-Actin, and anti-GAPDH antibodies were tested as loading controls. Anti-actin was selected for quantification of human samples since this antibody showed the most consistent pattern of signals with respect to the amount of total protein loaded per lane (Figure 1a, d). We first analyzed the samples in a sex-matched manner. There was no consistent difference in the banding pattern for fl-hAPP when comparing AD-DS, DS, and AD with ND (Figure 1a, d). However, consistent with increased APP dose, there was an increase in the amount of fl-hAPP in DS independent of the diagnosis of AD. In AD-DS, significant differences were observed in both males and females (Figure 1b, e, Table S1). In males, fl-hAPP averaged ~2-fold in comparison to ND (p = 0.0119) (Figure 1b); in both AD-DS and DS females the ratios were increased to ~1.5-fold relative to ND (p = 0.0115) (Figure 1e; Table S1). In AD cases, the levels of fl-hAPP in males and females varied when normalized to the actin control, but neither group differed significantly from ND (male p > 0.9999, female p=0.4467) (Figure 1a, b, d, e, Table S1). These data are evidence that fl-hAPP was increased to levels proportional to gene dose in both male and female adults with AD-DS and DS.

hCTF levels in AD-DS and DS.

hCTFs were detected at apparent molecular weights between ~11 and 17 kDa in all the human postmortem brain samples (Figure 1a, d). There was variability in the levels and relative intensity of longer (i.e., pC99 and C99; also known as β-CTFs) versus shorter (pC89, C89, pC83, C83; α-CTFs) forms, across samples. hCTFs were increased in both males and females with DS, with or without AD, in comparison to ND (Figure 1a, c, d, f). In AD-DS males, compared to ND males, there was a significant (p = 0.0095) ~ 2.5-fold increase in hCTFs relative to actin. There was also a significant ~ 2.5-fold increase in AD-DS males compared to AD males (p = 0.0203) (Figure 1a, c). However, this increase was not significant compared to fl-hAPP (p > 0.9999) (Table S1; Figure S1a). Increases in hCTFs were also detected in both AD-DS and DS females. Compared to female ND, hCTFs were approximately 5-fold increased in AD-DS and DS relative to actin (AD-DS, p = 0.0149, DS, p = 0.0307), and female AD-DS cases had significantly increased hCTFs compared to female AD cases (~2.8-fold, p = 0.0312) (Figure 1 d, f). Moreover, the ~ 3- to 4-fold increases with respect to fl-hAPP (Figure S1b) were significant [ND: CTF/APP = 1.070 ± 0.1394 (average ± SEM); AD-DS = 2.938 ± 0.3911 [p=0.0391]; DS: 3.789 ± 0.7698 [p=0.0493] (values were normalized to the ND average hCTF/fl-hAPP) (Table S1). In those with AD but not DS, the levels for hCTFs varied in males and females (Figure 1a, d), but on average showed no significant difference relative to ND when normalized to actin (Figure 1c, f) or in proportion to fl-hAPP (Figure S1a, b, Table S1). These data demonstrate that while the increase in hCTFs in the brains of people with AD-DS was present in both males and females, in females the increases exceeded those predicted based on APP dose, as reflected in the levels of fl-hAPP.

Increased Aβ peptide levels in AD-DS, DS, and AD: comparison to ND.

We measured concentrations of Aβ38, 40, and 42 in brain lysates using the V-PLEX Aβ Peptide Panel 1 (4G8: Meso Scale Discovery). Table S2 reports on the absolute levels of each Aβ peptide in ND, AD-DS, DS, and AD samples in the assay. Though individual samples varied, increases in average absolute levels relative to the ND values were detected for each of the Aβ peptides examined in each set of DS samples (Figure 1g–i). With respect to Aβ42 the increases in AD-DS males and females with AD-DS and DS averaged ~ 45 to ~100-fold the ND values (absolute values: ND = 18.73 ± 12.08 pg/ml (average ± SEM); male AD-DS = 856.0 ± 143.6 [p vs ND = 0.0002]; female ND= 6.339± 3.168; female AD-DS= 658.1 ± 258.1 [p vs ND = 0.0017]; female DS= 603.1 ± 87.30 [p vs ND = 0.0028])(Table S2). These values were increased in proportion to fl-hAPP levels by ~ 20-fold in AD-DS males and ~40-fold in AD-DS and DS females (Table S3); all increases with respect to fl-hAPP were statistically significant (statistics: vs sex-matched ND; AD-DS males p = 0.0059; AD-DS females p = 0.0109; DS females p= 0.0094) (Figure S1c, Table S3). When comparing ND males versus ND females, or AD-DS males versus AD-DS females, Aβ42 did not show statistically significant differences in absolute levels (for ND, p = 0.2463; for AD-DS, p = 0.3152) (Figure 1g); the same comparisons between sexes failed to show significant differences when expressed with respect to fl-hAPP (for ND, p = 0.2303; for AD-DS, p = 0.2544) (Figure S1c; Table S3). Statistically significant increases of Aβ40 were also observed in male and female AD-DS (Figure 1h, Table S2); relative to sex-matched ND: ~42-fold in male AD-DS (p = 0.0214), ~45-fold in female AD-DS (p = 0.0103) (Figure 1h, Table S2). Aβ40 levels were also significantly increased in proportion to fl-hAPP; relative increases to the sex-matched ND: ~15-fold in male AD-DS (p = 0.0214), ~50-fold in female AD-DS (p = 0.0103) (Figure S1d, Table S3). Comparing males and females with AD-DS, no statistically significant difference was observed in the absolute values of Aβ40 (p > 0.9999) or in proportion to fl-hAPP levels (p = 0.7360). Lesser but significant increases of ~10-fold were detected in DS females (p = 0.0433 versus female ND) (Figure 1h, Table S2); for these subjects, the increase in Aβ40 with respect to fl-hAPP was not significant (p = 0.2534, Figure S1d, Table S3). For Aβ38 relative to the sex-matched ND, the concentration was increased in both male AD-DS (p = 0.0045) and female AD-DS (p = 0.0012) (Figure 1i), averaging ~30-fold in males and ~50-fold in females (Table S2). Lesser increases were present in DS females relative to ND females; they averaged 19-fold (Figures 1i, Table S2) but they were not statistically significant (p = 0.1068 relative to female ND). While the increase in Aβ38 relative to fl-hAPP was statistically significant for AD-DS males (p = 0.0154), this was not the case for AD-DS or DS females (p = 0.0601, p > 0.9999, respectively) (Figure S1e, Table S3).

We also measured Aβ peptide levels in those with sporadic AD. Aβ42 levels were increased in both males (~15-fold enrichment) and females (~30-fold) (Figure 1g. Table S2); while in males the increase was significant with respect to the sex-matched ND (p = 0.0061), in females the very strong trend failed to reach statistical significance (p = 0.0608). However, in neither males nor females were the increases significantly greater than predicted by fl-hAPP in comparison to the sex-matched ND (male, p = 0.161; female, p = 0.1598) (Figure S1c, Table S3). Levels of Aβ42 in AD did not show significant differences between males and females with respect to concentration (p = 0.7013) or with respect to fl-hAPP (p = 0.1964). Aβ40 levels were increased in AD males and females by ~4-fold, a result that was significant in males (p = 0.0045) but not in females (p = 0.3908) (Figure 1h, Table S2). Relative to fl-hAPP neither the increases in males nor females were significant (Figure S1d; Table S3). For Aβ 38 levels, while average increases of ~ 4- to 9-fold were detected in AD males and females, respectively (Figure 1i, Table S2), the increases were not significant as compared to sex-matched ND (males p = 0.7777, females p = 0.0596, respectively) (Figure 1i) or with respect to fl-hAPP (Figure S1e, Table S3).

Comparing AD-DS versus AD, both males and females with AD-DS had much higher levels of all three Aβ species than in AD males and females (Table S2). Indeed, a several-fold increase was documented for each Aβ species (~3-fold for Aβ42, ~11-fold for Aβ 40 and ~6-fold for Aβ38; Table S2); the increases in Aβ42 and Aβ40 in AD-DS significantly exceeded those predicted by fl-hAPP (Table S3). These data are evidence that the AD-DS brain differs from the AD brain in much larger increases in Aβ species.

The ratio of Aβ42 to Aβ40 provides indirect estimates for the activity of γ-secretase, as well as the regulation of Aβ metabolism and clearance [37, 64, 65]. In an earlier study, the kinetics of synthesis and clearance used the SILK method, which radiolabels newly synthesized proteins and follows their appearance over time in CSF [66]. While similar kinetics were observed for Aβ42, Aβ40, and Aβ38 in patients without amyloid plaques, in those with plaques there was an increase in the fractional turnover rate for Aβ42 relative to Aβ40 and Aβ38 [66]. Positive correlations for an increase in the ratio of Aβ42/Aβ40 turnover rates with plaque load and plaque growth rate combined to suggest that Aβ42 is more readily incorporated into amyloid plaques than Aβ40 or Aβ38 [66–68], a posit also consistent with the decrease in the CSF Aβ42/40 ratio in AD [69].

We explored the hypothesis that as for AD, due to sporadic, late-onset disease as well as familial AD (FAD), in AD-DS there would also be an increase in the Aβ42/Aβ40 ratio [70, 71]. As expected, in males and females with sporadic AD the Aβ42/Aβ40 ratio was significantly increased compared to the sex-matched ND (males p = 0.0014, females p = 0.0024) (Figure S1f). In females with DS, a significant increase in the Aβ42/Aβ40 ratio was also observed (p = 0.0291). However, concomitant large increases in both Aβ42 and Aβ40 in AD-DS males and females reduced the ratio (Figure 1g, h, Figure S1f, g). The Aβ42/Aβ38 ratio was also significantly increased in males and females with sporadic AD compared to the sex-matched ND (males p = 0.0008, females p = 0.0209) (Figure S1g). Females with DS also showed a significant increase in Aβ42/Aβ38 (p = 0.0353); however, relatively larger increases in Aβ38 in males and females with AD-DS blunted this ratio (Figure 1i, Figure S1f, g). Thus, while in AD increases in the Aβ42/40 and Aβ42/38 ratios were confirmed, the concomitant and greater increases in Aβ40 and Aβ38 precluded this demonstration in AD-DS.

APP and Its Cleavage Products in Partial Trisomy and Mosaic DS.

To further evaluate the impact of APP gene dose in AD pathogenesis in DS, we examined fl-hAPP, hCTFs, and Aβ peptides in the brain lysate of an individual with partial trisomy 21 (PT-DS) (Table 1). In extremely rare cases [72–74], only a portion of HSA21 is present in excess [75], resulting in PT-DS. This person was the subject of a previous report of PT-DS in which the genome was shown to harbor the normal 2 copies of hAPP. While demonstrating many of the clinical features of DS, molecular analysis revealed an increase in copy number for only 18.8Mb, extending from position 28.12 Mb to the telomere of HSA21; excluded from the duplication was a 1.95Mb region containing APP (26.17 Mb). This and another report highlighted the important role of hAPP gene dose in AD pathogenesis in DS [27, 28] because neither case manifested declining cognition in old age or the pathology of AD at postmortem examination. This male PT-DS reported by Doran et al. [28] was a 65-year-old man followed over 7 years during which time normal plasma levels of Aβ40, and undetectable plasma levels of Aβ42, were documented. On PiB-PET scanning he registered SUVr values lower than for AD and typical DS subjects and that was similar or lower than healthy controls. Postmortem exam showed the absence of neuritic plaques except for trace amounts in CA1 and NFTs judged to be Braak stage III [28]. The brain lysate from this patient demonstrated, as expected, the typical banding pattern for fl-hAPP with no increase in the level of fl-hAPP, in the levels of Aβ42, Aβ40 or Aβ38, or the ratios of Aβ42/40 or Aβ42/38 (Figure 1a, b, g–i; Figure S1c–g). Surprisingly, there was a marked increase in the levels of hCTFs (Figure 1c) with values comparable to those in males with AD-DS; indeed, the ~2:1 ratio of CTFs to fl-hAPP was greater than seen in AD-DS males (Figure S1a).

Another rare type of DS results from mosaicism for HSA21 (~2% of total cases) in which cases the extra chromosome is present in some but not all cells, the percent of which may vary across tissues [14]. In a single female case of mosaic DS examined herein (aged 48, Plague Stage B, Braak Stage III) (Table 1) we expected to find that the extent to which neurons harbored the extra chromosome would be reflected in increases in fl-hAPP and its products. However, whereas fl-hAPP was not increased, there was an increase in hCTFs and the ratio of hCTFs to fl-hAPP; the values recorded approximated those in DS and AD-DS females (Figure 1d–f, Figure S1b). Consistent with the presence of amyloid plaques on postmortem exam, we detected an increase in the absolute levels of Aβ42 and Aβ40, but not Aβ38 (Figure 1g–i). The increase in Aβ42 and the ratio of Aβ42/40 (Figure S1f) were equivalent to those in DS and AD-DS females, while the Aβ42/38 ratio exceeded the same ratio in DS and AD-DS females (Figure S1g).

Summarizing the data for DS, the levels of fl-hAPP and its products are impacted in the context of full trisomy for HSA21 as well as in the cases of PT-DS and mosaic DS examined herein. These findings point to APP dose as one factor for regulating the levels of fl-hAPP and suggest that increased APP dose influences other processes, possibly including processing and clearance of fl-hAPP. Given the data for the mosaic DS case, the further possibility is raised that aneuploidy in a minority of cells may induce changes in hAPP products.

Evidence for Sex Differences in hAPP processing in DS: studies in AD-DS, PT-DS, and DS-mosaic cases.

Our initial analysis (Figure 1a–f and Figure S1a, b) indicated that male and female cases with AD-DS differ in the levels of CTFs. To further address possible sex effects, we reexamined the levels of fl-hAPP and hCTFs in male and female ND and AD-DS cases on the same blot (Figure 2a–g). When normalized to the actin control and expressed relative to the male ND, the female ND cases showed a slight increase in fl-hAPP. There was a significant increase in fl-APP in the female AD-DS cases compared to both ND males (p-value = 0.0029) and ND females (p-value = 0.0223; Figure 2b). Male AD-DS cases also showed an increase, albeit smaller than shown in Figure 1b (1.6-fold vs 2.1-fold the ND male control) and insignificant using a two-way ANOVA followed by Tukey post hoc analysis. Note, however, that comparing male ND to male AD-DS there was a significant difference when evaluating this data using a one-way ANOVA followed by Dunn’s multiple comparison post hoc test (p-value = 0.0375). The increases in both male and female AD-DS thus approximated APP dose. There was no significant difference for the sex and genotype interaction (two-way ANOVA, p=0.5148) (Figure 2a, b). On the other hand, evaluation of hCTFs in female AD-DS compared to male AD-DS cases revealed a significant difference (post-hoc Tukey’s multiple comparisons test, p = 0.0217), as well as for the sex and genotype interaction (two-way ANOVA, p= 0.0230) (Figure 2a, c). Thus, while the increase in AD-DS males were ~2.5-fold the ND males, for AD-DS females it was ~ 4-fold the ND females. After expressing the level of hCTFs with respect to fl-hAPP, the difference in male and female AD-DS was again apparent. While for male AD-DS there was no significant difference relative to ND subjects, the increase in female AD-DS was significant relative to both ND females and ND males (two-way ANOVA followed by post-hoc Tukey’s multiple comparisons test, female AD-DS vs female ND: p = 0.0024; female AD-DS vs male ND p= 0.0158) (Figure 2d).

The enzymatic processing of fl-hAPP leads to the production of hCTFs of different molecular weights [76]. To better understand fl-hAPP processing in DS we evaluated the distribution of hCTF isoforms, quantifying shorter bands, corresponding to α-hCTFs (i.e., pC89, C89 plus pC83, C83) and longer ones, corresponding to β-hCTFs (i.e., pC99 and C99). The increase in hCTFs in female AD-DS was due to increases in both shorter (α-hCTFs; Figure 2e; p-value < 0.0001) and longer forms of hCTFs (β-hCTFs; Figure 2f; p-value < 0.0001). In males the, the increases relative to ND cases was only significant (p-value = 0.0444) for the shorter forms of hCTFs (Figure 2e, f). The ratio between shorter and longer hCTFs showed no significant differences comparing AD-DS to ND or comparing males to females (Figure 2g). To further characterize the distribution of hCTF fragments in samples, and explore the findings for hCTFs in the PT-DS and DS-mosaic cases, we analyzed a representative subset of AD-DS cases, and the PT-DS and DS-mosaic cases, using 16% tricine gels (Figure S2a–d). The PT-DS and DS-mosaic cases demonstrated a relatively larger fraction of α-hCTFs as compared to β-hCTFs (Figure S2a–d). Finally, we also compared AD-DS to AD by examining CTFs in AD cases (Figure S2e–h). No significant differences in the relative levels of α-hCTF species, β-hCTF species, or in the ratio between α-/β-hCTF species were found in comparing AD cases to sex-matched NDs. In summary, sex-based changes in the levels of the hCTFs were seen in AD-DS, with increased levels of hCTFs in females with full trisomy as well as in the mosaic DS. While in AD-DS cases both α-hCTFs and β-hCTFs were increased, the PT-DS and DS-mosaic cases demonstrated greater increases in α-hCTFs.

Examining the levels of APP processing enzymes in AD, AD-DS, PT-DS, and DS-mosaic cases.

To further explore hAPP processing, and in particular, to ask if the increase in hCTFs in females with DS was correlated with changes in the levels of processing enzymes, we examined the relative levels of ADAM10, BACE1, and 2 and components of the γ-secretase complex, including presenilin (PSEN) 1 and 2, nicastrin, and presenilin enhancer (PEN) 2 (Figure S2i–k) [76]. Brain lysates were subjected to SDS-PAGE and immunoblotting with antibodies to each of the enzymes listed (Figure S2i–k). As compared to ND females, significant increases were present in AD-DS females in the average levels of BACE2 (3-fold the ND; p = 0.0439), Nicastrin (mature) (1.4-fold the ND; p = 0.0024), and PEN2 (1.5-fold the ND; p = 0.0255) (Figure S2i: one-way ANOVA followed by a multiple comparison test). Mature BACE1 levels showed a trend to an increase in AD-DS females compared to female ND cases; statistical significance was observed by t-test between AD-DS and ND females but not with ANOVA (p = 0.0567). There were no significant differences for protein levels in male AD-DS versus their ND controls for any of the proteins examined (Figure S2i, j). For proteins, whose levels were significantly different in female AD-DS versus ND cases we calculated the ratio of the protein versus fl-hAPP; no significant differences relative to fl-hAPP were seen for BACE2, the mature forms of BACE1 and nicastrin or PEN2. Nor did AD-DS males show a difference (Figure S2k). We noted, however, that in the DS Mosaic case increases with respect to actin and relative to ND were detected for BACE2, nicastrin, and PEN 2; moreover, mature BACE1 levels were also increased.

In summary, female AD-DS cases demonstrated differences relative to male AD-DS in the levels of proteins that contribute to the processing of APP via BACE and the γ-secretase complex. They are further evidence for sex-linked differences in AD-DS females. Nevertheless, whether and how the changes are linked to increased hCTFs in DS females is unclear. Remarkably, an extra copy of the BACE2 gene located on HSA21 was correlated with a significant increase in the levels of the protein in the brains of females but not males with AD-DS. Indeed, neither AD-DS males nor the PT-DS male had significantly increased levels of BACE2, though their genomes harbor an extra copy of BACE2.

The phosphorylation status of Tau at Thr212 (pT212) correlates with tau pathology.

Hyperphosphorylation of Tau and the presence of these species in neurofibrillary tangles is a hallmark in AD pathogenesis, a relationship defined in part by the Braak stage [56, 77, 78]. Table 1 reports the Braak stage determined in our cases. To extend our studies comparing AD with AD-DS we investigated the phosphorylation of Tau, specifically focusing on pT212. Thr212 is targeted for phosphorylation by DYRK1A, whose gene is also encoded on HSA21. In addition to tau, many substrates are targeted by DYRK1A, including APP and PSEN, raising interest in the possibility that DYRK1A may contribute to AD pathogenesis, especially in the context of DS [13]. Thr212 is hyperphosphorylated in the AD brain; increases in phosphorylation are equal to or greater than other p-tau species in bands migrating at both greater than 65 kDa (high molecular weight [HMW] tau) and less than 65 kDa (low molecular weight [LMW] tau) [79]. Immunoblot analysis of levels of pT212 tau employed the same methods used for APP. pT212 tau was detected as a set of prominent bands at apparent molecular weights ranging from ~50 to 68 kDa, values consistent with earlier reports [79, 80]. Little if any pT212 tau was detected in either male or female ND cases (Figure 2 h–j, l). In AD-DS cases pT212 Tau was readily detected in both males and females, as compared to the sex-matched ND cases (males p = 0.0001, females p = 0.0252). Total Tau levels were detected using Tau 5, a pan-tau antibody that labels all six isoforms, with major bands ranging in apparent molecular weight from ~50 to 65 kDa. Interestingly, total Tau levels were also significantly increased in AD-DS cases compared to the sex-matched ND cases in both males and females (one-way ANOVA, followed by Dunn’s multiple comparison test, males: p = 0.0049, females: = 0.0049) (Figure 2h, i, k, m). In contrast, we found little if any pT212 tau signal in the DS females without AD; total Tau levels were slightly but insignificantly increased. In the AD cases, the levels of pT212 tau were less than in AD-DS; there was a significant increase of pT212 in males (p = 0.0041) but not in females (p = 0.2595) (Figure 2j, l). Different from AD-DS, total Tau was not significantly increased in either male or female AD cases (males p > 0.9999, females p = 0.3947) (Figure 2k, m). Consistent with his Braak stage, the PT-DS case showed no detectable signal for pT212 tau. Thus, even though this person’s genome carried three copies of DYRK1A there was no evidence of this activity with respect to pT212. In both PT-DS and DS Mosaic cases, no significant increase in total Tau was detected (Figure 2j, l).

To determine whether the increased levels of pT212 and total Tau in AD-DS cases (Figure 2h–m) showed a sex difference between males and females, representative samples of male and female ND cases (n=6 each), all the AD-DS cases (male n= 4, female n=7) and the PT-DS and DS-mosaic cases were run on the same gel. The levels of pT212 levels (Figure S2l, m) in AD-DS show no significant difference between males and females (p = 0.6229). Nor did we observe sex-based differences in total Tau expression in ND or AD-DS cases (ND p = 0.3946, AD-DS p = 0.4804) (Figure S2n, o). The levels of pT212 and total Tau in the PT-DS and DS-mosaic cases were equivalent to the ND cases. In summary, increases in pT212 and total tau point to additional significant differences between AD-DS and AD. That the changes in AD-DS versus DS were correlated with increased Braak stage raises the possibility that increasing levels of pT212 and total tau mark increasing tau pathology during the progression of AD in this population.

Examining gene expression at the level of mRNA in the young and aged brain.

Measurement of APP products in postmortem human brain provides an important ‘snapshot’ of APP expression but does not inform as to how age impacts these values, or which can be implicated in neurodegenerative phenotypes. Examining APP gene dose effects in model systems complements human studies by documenting age-related changes, enabling comparisons between the levels of APP products in human and model brains, and deciphering the contributions of individual APP products to phenotypes. The Dp16 mouse model of DS has proven useful in examining relevant developmental and cognitive phenotypes [57] but a detailed evaluation of possible neurodegenerative changes has not been reported. The Dp16 model builds on the extensive synteny between the long arm of HSA21 and mouse chromosomes 16. It was created with chromosomal engineering and through partial duplication of mouse chromosome 16 carries an extra copy of all mouse chromosome 16 genes with a human homolog; in total, the Dp16 genome contains 105 genes also present on HSA21 [81]. mAPP is present in three copies in these mice, as are several other genes with a possible role in the biology of AD, including Dyrk1A.

We first explored transcriptional alterations in Dp16 mice using the nCounter mouse AD panel (NanoString Technologies). This panel reports RNA levels of 770 target genes selected for relevance to AD based on postmortem studies of human AD samples [61, 82–84]. We assessed total mRNA extracted from the hippocampus of Dp16 and 2N mice at 4 and 19 months (n=3). Nine housekeeping genes (Cnot10, Fam104a, Supt7l, Csnk2a2, Aars, Lars, Tada2b, Ccdc127, Asb7) were used as internal controls to normalize data. Four comparisons of gene expressions were conducted: 2N versus Dp16 mice at 4 months; 2N versus Dp16 at 19 months, 2N at 4 months versus 9 months, Dp16 at 4 months versus 19 months. Results are shown with a cut-off at p-values < 0.05 and for changes that were increased or decreased ≥ 25% (Table 2, Table S4). In comparing Dp16 to 2N, nine genes satisfied cut-off criteria at both 4 months and 16 months; at 4 months, 4 genes were decreased and 5 were increased. At 19 months only increases were detected. Three of the upregulated genes at both ages (App, Syjn1, and Psmg1) are encoded on the trisomic region; the increase in their RNA levels was in proportion to gene dose at both ages. Genes on other chromosomes were also significantly impacted in Dp16; interestingly, the RNA changes for these genes at 4 months showed no overlap with those present at 19 months (Table 2). Comparing RNA levels in 2N mice between 19 months and 4 months showed changes in transcripts for 12 genes that met cut-off criteria (Table S4), there were 4 decreases and 8 increases. We also detected 12 genes whose transcripts differed in Dp16 mice, comparing 19 months versus 4 months; in each case, the level of the RNA was increased and no genes on the duplicated segment met the cut-off (Table S4). To validate the findings, we selected four genes that showed significant changes across comparisons (Olfm3, Per1, Ctss and Psmg1 [Dp16 versus 2N at 4 months]; Apha1a, Smad4, Fxyd5 and Psmg1 [Dp16 versus 2N at 19 months]) and performed RT-qPCR using cDNA synthesized with the total mRNA extracted from the hippocampus. In the 4-month comparison, as predicted, expression of Olfm3 and Per1 showed significant decreases and Ctss and Psmg1 showed significant increases (n=3, p-value = 0.05, one-tailed Mann-Whitney test; Figure S3a–d). In the 19-month comparison, Apha1a, Fxyd5 and Psmg1 showed the predicted significant increases (n=3, p-value = 0.05, one-tailed Mann-Whitney test; Figure S3e, g, h) but Smad4 did not (Figure S3f). Thus, except for Smad4, these observations validated the findings from the nCounter mouse AD panel. Taken together, these data are evidence for transcriptional differences between Dp16 and 2N mice as well as for changes during aging in both genotypes.

Next, we separately analyzed the expression of mAPP and seven genes encoded in the trisomic region that support neuronal function and/or are related to the pathology of AD: Dyrk1a, Kcnj6, Synj1, Rcan1, Bace2, Itsn1, Sod1 [5, 11, 14, 76, 85, 86]. mRNAs extracted from cortex of 2N and Dp16 mice at ages 4 and 19 months were analyzed using RT-qPCR (n=4 and 5, respectively, at each age). We found significantly increased expressions of all eight genes duplicated in Dp16 mice compared to 2N mice at both 4 and 19 months (Figure 3a–h), with changes in each case approximating the increase in gene dose. At 19 months, 7 of the 8 genes showed statistically significant increases (Figure 3a–h); the increases were more variable with respect to gene dose (range: ~1.3 to 2.0-fold the 2N value). Bace2 mRNA levels, while increased, failed to reach significance. To further characterize gene expression, and to examine a region not impacted in AD-DS, we also analyzed mRNA from the cerebellum of 12-month-old 2N and Dp16 mice. In the cerebellum, mRNAs for App, Dyrk1a, Kcnj6, Synj1, and Rcan1 were increased significantly in Dp16 mice in comparison to 2N; the increases were equal to gene dose (mApp, Dyrk1a) or slightly greater than gene dose (Kcnj6, Rcan1, Synj1; ~2-fold 2N) (Figure S3i–m). While mRNA levels for Bace2, Itsn1, and Sod1 were increased, the changes were not statistically significant (Figure S3n–p).

These data are evidence for age-related and regional differences in transcriptional regulation in Dp16 mice involving genes both within and distinct from the duplicated segment. Changes were registered in the expression of genes that encode products whose functions are linked to neurons and the biology of AD. Notably, increased levels of the mRNA for mApp approximated the increase in mApp dose across all brain regions and ages tested. The picture points to a widely disrupted transcriptional network whose contributions may impact many distinct molecular pathways during aging.

Increased levels of mApp and its products in young and aged Dp16 mice.

To examine age-dependent effects downstream from mApp dose in the Dp16 mouse, we measured the levels of fl-mAPP and its products in the cortex at 2, 4, 16, and 19 months of age and in the hippocampus at 4 and 19 months. For analysis of mouse lysates, we used Tubulin or GAPDH as internal controls due to reproducibility and the correlation with the normalized total protein lysates. In cortex and hippocampus, the banding patterns for fl-mAPP and mCTFs did not appreciably differ between 2N and Dp16 mice; they corresponded to those in human samples expect that in mouse samples the relative amounts of shorter (α-mCTFs) versus longer fragments (β-mCTFs) was more uniform (Figure 3i), consistently demonstrating a predominance in α-mCTFs. fl-mAPP and its mCTFs were significantly increased in male Dp16 mice, as compared to 2N control, in cortex and hippocampus at 4 and 19 months of age; only the increase of fl-mAPP at 4 months in hippocampus failed to reach significance (Figure 3i–k). Only in the hippocampus of the Dp16 mouse at 19 months was the ratio of mCTFs/ fl-mAPP significantly increased compared to 2N mice (p = 0.0066, p = 0.0415; at 19 months and 4 months, respectively) (Figure S4a). Based on increased mApp gene dose and given increased mApp mRNA levels at age 4 months, we anticipated finding increased levels of fl-mAPP and its products at an early age. This expectation was confirmed for both fl-mAPP and mCTFs in the cortex at age 2 months wherein increases were present in Dp16 females compared to 2N females (Figure S5a–c). The increase in fl-mAPP (~1.3-fold the 2N level) was proportional to gene dose; for mCTFs, the levels were ~2-fold than those in 2N mice.

Mouse Aβ38, 40, and 42 were also measured in the cortex of 2N and Dp16 males at 4 and 19 months using the V-PLEX Aβ Peptide Pane 1 (4G8: Meso Scale Discovery, Rockville, MD). Aβ38 levels were under the detection limit in both Dp16 and 2N mice (data not shown). Aβ40 was increased in Dp16 versus 2N samples at both ages (Figure 3l). The increases relative to 2N mice were 1.8-fold at 4 months (p=0.0286) and 1.8-fold at 19 months (p = 0.0079). Interestingly, there was a significant increase of Aβ40 in both 2N and Dp16 at 19 months compared to 4 months (2N: 1.7-fold, p = 0.0159, Dp16: 1.6 -fold, p = 0.0159). Comparing Dp16 versus 2N samples, there was no significant change in the ratio of Aβ40/fl-mAPP at either 4 months or 19 months; however, the ratio in Dp16 mice at 19 months was significant with respect to both 2N and Dp16 mice at 4months (Figure S4b). A similar pattern was evident for Aβ42 (Figure 3m) but with Aβ42 present at much lower levels than for Aβ40. The increases relative to 2N mice were 1.4-fold at 4 months (p=0.0286) and 1.6-fold at 19 months (p = 0.0079). There was a significant increase of Aβ42 in both 2N and Dp16 at 19 months compared to 4 months (2N: 1.2-fold, p = 0.0238, Dp16: 1.6 -fold, p = 0.0159). Comparing Dp16 versus 2N samples, there were no significant changes for Aβ42/fl-mAPP at 4 months or 19 months. However, a significant increase was observed in Dp16 mice at 19 months vs Dp16 at 4months (Figure S4c). We calculated the ratio of Aβ42/40 in Dp16 and 2N mouse samples. There was no significant change between 2N and Dp16 at either 4 or 19 months of age, however, there was a small but significant decrease comparing Dp16 at 19 months versus 2N at 4 months (Figure S4d). In summary, both 2N and Dp16 mice showed age-dependent increases in Aβ40 and Aβ42 (Figure 3l, m). The increase in the ratios of Aβ40/fl-mAPP and Aβ42/fl-mAPP in old Dp16 mice (Figure S4a, b) is evidence suggesting accumulation of these species during aging albeit not beyond that predicted by App gene expression, as registered by fl-mAPP.

Next, we hypothesized that the sex effect seen for DS in humans would also be present in mice; fl-mAPP and mCTFs in male and female Dp16 and 2N mice were examined in the cortex at 8 and 14 months. At each age, there were strong trends for increases for both fl-mAPP and mCTFs in Dp16 males and females relative to 2N mice (Figure S5d–f). The increases in fl-mAPP were somewhat greater in females, but there were no statistically significant differences between male and female Dp16 mice for either fl-mAPP or mCTFs (at p = 0.05 threshold). Examining further the possible correspondence of sex differences in AD-DS, we also evaluated the levels of BACE1 and 2 in 8-month-old Dp16 and 2N mouse cortex. There was a modest increase in BACE2 in Dp16 versus 2N mice, but neither male nor female Dp16 mice differed significantly with respect to each other or 2N mice (Figure S5g, i). However, as for females with AD-DS, female Dp16 mice showed a significant difference in mature BACE1 levels compared to male Dp16 mice by t-test, but not by ANOVA (p= 0.0009, p = 0.1748, respectively) (Figure S5h, j). There was no significant difference in the levels of immature forms of BACE1 (Figure S5h, k). Finally, as for DS samples, we examined the levels of the pT212 tau epitope in the cortex and hippocampus of male 2N and Dp16 at 4 and 19 months. Between 4 and 19 months, Dp16 samples showed on average an increase in pT212; relative to the 2N level at 4 months, the increase in cortex in Dp16 at 19 months was significant (Figure S5l, m). In hippocampus, the value in the 19 month Dp16 mouse was significantly increased relative to that in the 4 month Dp16 mouse. Total Tau levels failed to show significant differences for age or genotype (Figure S5l, n).

In summary, studies of mApp gene expression in the Dp16 mouse brain demonstrated both similarities and differences with respect to those in the AD-DS and DS brain. While neither sex-dependent difference in CTF levels nor the increase of total Tau were recapitulated, consistent gene-dose linked changes in the levels of mApp mRNA, fl-mAPP, mCTFs, and age-related increases in Aβ and Tau phosphorylation draw important parallels with AD-DS.

Effects of APP gene dose on neurodegeneration in Dp16 mice.

Loss of specific populations of vulnerable neurons is characteristic in AD and AD-DS, (reviewed [87]). Among vulnerable neuronal populations are those in layer II of the entorhinal cortex (EC) [88, 89], catecholaminergic neurons the locus coeruleus (LC) [90, 91], and cholinergic neurons in the basal forebrain magnocellular complex (BFCN) [92, 93]. To ask if the same populations are impacted in the Dp16 mouse, and to test for the necessity for increased mApp gene dose for loss of neurons, we normalized the mApp gene dose in Dp16 mice by producing Dp16 mice (mApp ++-) on the same strain background. We then examined Dp16, Dp16 (mApp ++-), and 2N mice using unbiased stereology in the entorhinal cortex, locus coeruleus, and medial septum of the BFCN. By an average age of 16 months there were significant decreases in the number of neurons in Dp16 mice compared to 2N mice in BFCNs (p-value = 0.0474; Figure 4a panels, b) and LCNs (p-value = 0.0095; Figure 4d); however, there were no significant differences between 2N and Dp16 (mApp ++-) mice in either region. The analysis also revealed a significant difference in EC (p=0.014). Although post hoc analysis did not indicate which group(s) were different compared to 2N mice, the average number in Dp16 mice was lower than in either 2N or Dp16 (mApp++-) mice. To examine p-Tau epitopes in Dp16 and 2N mice, we immunostained using a well-established antibody for IHC (PHF-1), which recognizes the phosphorylated sites at Ser396 and Ser404. At an average age of 16 months, accumulation of pTau-ir was readily seen in the Dp16 mouse cortex but not in Dp16 (mApp ++-) mice; (Figure 4e panels) [94]. In Dp16 mice, quantitation of p-Tau-ir demonstrated an increase in pixel intensity (scale ranged from 0 = black to 255 = white) with respect to 2N mice, where the lower the number, the darker the pixel value, indicating greater immunoreactivity. ANOVA analysis revealed a significant difference between the groups (p-value = 0.0167; Figure 4f); however, post hoc analysis did not indicate which group(s) were significantly different compared to 2N mice. Nevertheless, the values for Dp16 (mApp++-) approximated those of the 2N mouse. These findings are evidence that increased mApp dose is necessary in Dp16 mice for the degeneration of neurons vulnerable in AD-DS and for increased levels of an immunostaining marker of tau pathology.

To confirm the age-related degenerative nature of the reductions in neuron number we examined the same populations in the younger Dp16 mice. There were no significant differences at 4 months of age between the values in Dp16 normalized to 2N mice: Dp16 (0.9532 ± 0.0791) vs 2N (1.0 ± 0.3580) mice in ECNs (p-value = 0.8603); nor between the normalized numbers of LCNs: Dp16 (0.8658 ± 0.06901) vs 2N mice (1.0 ± 0.07367; p-value = 0.2544). Likewise, there was no significant difference (p-value = 0.9776) between Dp16 (1.005 ± 0.1172) and 2N (0.9998 ± 0.1012) mice in the number of BFCNs at 10 months of age. Similarly, PHF-tau immunoreactivity, using pixel intensity as a measure of densitometry, was not significantly different between Dp16, (176.5 ± 3.525) and 2N mice (186.4 ± 0.7686) in the entorhinal cortex at 4 months of age. These findings recapitulate age-related neurodegenerative phenotypes seen in AD-DS [89, 91, 95].

To document the expected reductions in APP products in the Dp16 (mApp ++-) mouse, we measured the levels of mApp mRNA, fl-mAPP, mCTFs, and Aβ40 in all three genotypes (2N; Dp16 and Dp16 (mApp ++-)). We examined these measures in the cortex at age 4 months – i.e., before the onset of degeneration. Relative to 2N mice, in Dp16 mice the level of mApp mRNA was increased about 2-fold and increases in the other APP products approximated those reported above. No significant differences were registered between 2N and Dp16 (mApp++-) mice in any measure (Figure S5o–s). These data are evidence for gene-dose linked mApp expression in the Dp16 model and confirm that increased levels of this gene are necessary for neurodegeneration. However, they do not identify a specific APP product as responsible.

Investigation of APP gene dose on microglial and astrocytes cells in Dp16 mice.

There is increasing recognition of a prominent role played by inflammation in the biology of AD the involvement of astrocytes and microglial cells [96, 97]. A recent report documented changes in microglial cells in young Dp16 mice, with changes detected as early as postnatal day 22 [98]. It reports in microglia cells a scavenging phenotype accompanied by a loss of dendritic spines in surrounding neurons. We assessed the status of activated microglial phenotypes, including ramifications of processes and size of microglial soma in the cortex of aged mice (16 months). We found that total process length, when normalized to the number of cells, was significantly decreased in Dp16 (p-value = 0.0252), but not in Dp16 (mApp++-) (p-value = 0.2095) mice, as compared to 2N mice (Figure 4g). However, there was no significant difference in the size of microglia in Dp16 or Dp16 (mApp ++-) mice compared to 2N mice (Figure 4h). These data are evidence for increased microglial activation in Dp16 versus 2N mice. Since process length, but not the microglia size, was a function of mApp dose, microglial activation may be under the influence mApp as well as other genes in Dp16 mice.

Astrocytes are involved in several processes in both the developing and mature brain [99], with astrogliosis present in AD-DS. To assess the possibility that astrocyte activation would characterize the young adult Dp16 brain, we again examined mice at age 4 months. Immunohistochemistry probing the entorhinal cortex show an elevated level of GFAP immunoreactivity in both parenchyma tissue and near the vasculature of the Dp16 mouse brain (Figure 5a), but not in the medial septum (Figure 5b). Stereological analysis revealed a significant increase in the number of GFAP⁺ astrocytes in Dp16 vs Dp16 (mApp++-) mice in the entorhinal cortex: Dp16 = 1.39 x 10⁵ cells/mm3; Dp16 (mApp++-) = 0.63 x 10⁵ cells/mm3); [(Dp16 versus 2N: p-value < 0.01; Dp16 versus Dp16 (mApp++-): p-value <0.001)] (Figure 5a). However, stereological analysis revealed no significant increase medial septum in the number of GFAP⁺ astrocytes in Dp16 vs Dp16 (mApp++-) mice: Dp16= 1.11 x 10⁵ cells/mm3; Dp16 (mApp++-) = 1.03 x 10⁵ cells/mm3); [(Dp16 versus 2N: p-value =0.90; Dp16 versus Dp16 (mApp++-): p-value =0.92)] These data are evidence of an effect of the App gene dose on astrocyte activation in the entorhinal cortex, but not medial septum of Dp16 mice. Whether or not astrocytosis affects medial septum at later ages and what other regions are affected will require additional studies [100].

Investigation of APP gene dose on endosomes in Dp16 mice.

Dysregulation of early endosomes, with increased recruitment of Rab5 and enlargement, is an early feature of DS and is consistently present in AD-DS and AD [23, 24]. To explore a role for APP gene dose in this phenotype we asked if changes could be detected in young adult mice – i.e., at a time when increased mApp expression was documented in Dp16 mice and demonstrated to be normalized in Dp16 (mApp++-) mice. We examined the entorhinal cortex (Figure 5c–f) and medial septum (Figure 5g–j) of mice at age 4 months, comparing 2N mice to Dp16 and Dp16 (mApp++-) mice. Tissue sections were stained with an antibody to Rab5 and random fields of the entorhinal cortex and medial septum were examined by fluorescent confocal microscopy and computer-aided morphometric analysis. Early endosomes with immunoreactivity for Rab5 in Dp16 were enlarged with an average size of 0.07 μm², compared to 0.06 μm² for both the Dp16 (mApp++-) and 2N mice in the entorhinal cortex. Early endosomes with immunoreactivity for Rab5 in Dp16 were also enlarged in the medial septum (Figure 5h) with an average size of 0.08 μm², compared to 0.06 μm² for both the Dp16 (mApp++-) and 2N mice. The enlargement of early endosomes in Dp16 mice in both the entorhinal cortex (Figure 5d) and medial septum (Figure 5h), though modest, was statistically significant, p < 0.05 with respect to both 2N and Dp16 (mApp++-) mice. More revealing was the frequency distribution of individual Rab5 immunoreactive early endosomes in the entorhinal cortex (Figure 5d) and medial septum (Figure 5h). This showed that the observed increase in average puncta size was caused by an increase in the size of endosomes whose sizes were greater than 0.30 μm². Though constituting a minority of total endosomes, in these size classes the Dp16 mice had approximately twice as many puncta as for either 2N or Dp16 (mApp++-) mice, which was statistically significant in the medial septum (Figure 5h; p-value < 0.05). In the entorhinal cortex the largest class of puncta (>0.50) was increased more than three-fold (Figure 5b). Whether or not endosomal enlargement is greater at later ages and registered in other brain regions is yet to be determined.

In summary, studies in the Dp16 mouse demonstrated the presence of degenerative hallmarks of AD-DS. In each case, the phenotype was normalized by reducing the mApp gene dose. The findings are evidence that increased mApp gene dose is necessary for several neurodegenerative phenotypes in this model of AD-DS.