The Alzheimer’s gene SORL1 is a regulator of endosomal traffic and recycling in human neurons

Background Loss of the Sortilin-related receptor 1 (SORL1) gene seems to act as a causal event for Alzheimer’s disease (AD). Recent studies have established that loss of SORL1, as well as mutations in autosomal dominant AD genes APP and PSEN1/2, pathogenically converge by swelling early endosomes, AD’s cytopathological hallmark. Acting together with the retromer trafficking complex, SORL1 has been shown to regulate the recycling of the amyloid precursor protein (APP) out of the endosome, contributing to endosomal swelling and to APP misprocessing. We hypothesized that SORL1 plays a broader role in neuronal endosomal recycling and used human induced pluripotent stem cell derived neurons (hiPSC-Ns) to test this hypothesis. We examined endosomal recycling of three transmembrane proteins linked to AD pathophysiology: APP, the BDNF receptor Tropomyosin-related kinase B (TRKB), and the glutamate receptor subunit AMPA1 (GLUA1). Methods We used isogenic hiPSCs engineered to have SORL1 depleted or to have enhanced SORL1 expression. We differentiated neurons from these cell lines and mapped the trafficking of APP, TRKB and GLUA1 within the endosomal network using confocal microscopy. We also performed cell surface recycling and lysosomal degradation assays to assess the functionality of the endosomal network in both SORL1 depleted and overexpressing neurons. Finally, we analyzed alterations in gene expression in SORL1 depleted neurons using RNA-sequencing. Results We find that as with APP, endosomal trafficking of GLUA1 and TRKB is impaired by loss of SORL1. We show that trafficking of all three cargo to late endosomes and lysosomes is affected by manipulating SORL1 expression. We also show that depletion of SORL1 significantly impacts the endosomal recycling pathway for APP and GLUA1 at the level of the recycling endosome and trafficking to the cell surface. This has a functional effect on neuronal activity as shown by multi-electrode array (MEA). Conversely, increased SORL1 expression enhances endosomal recycling for APP and GLUA1. Our unbiased transcriptomic data further support SORL1’s role in endosomal recycling. We observe altered expression networks that regulate cell surface trafficking and neurotrophic signaling in SORL1 depleted neurons. Conclusion Collectively, and together with other recent observations, these findings suggest that SORL1 is a broad regulator of retromer-dependent endosomal recycling in neurons, a conclusion that has both pathogenic and therapeutic implications for Alzheimer’s disease.


Methods
We used isogenic hiPSCs engineered to have SORL1 depleted or to have enhanced 27 SORL1 expression. We differentiated neurons from these cell lines and mapped the trafficking of 28 APP, TRKB and GLUA1 within the endosomal network using confocal microscopy. We also 29 performed cell surface recycling and lysosomal degradation assays to assess the functionality of 30 the endosomal network in both SORL1 depleted and overexpressing neurons. Finally, we 31 analyzed alterations in gene expression in SORL1 depleted neurons using RNA-sequencing.

33
Results We find that as with APP, endosomal trafficking of GLUA1 and TRKB is impaired by loss 34 of SORL1. We show that trafficking of all three cargo to late endosomes and lysosomes is affected 35 by manipulating SORL1 expression. We also show that depletion of SORL1 significantly impacts

272
For all imaging experiments, data was collected and analyzed in a blinded manner. Data was 273 assessed for significance using parametric two-tailed unpaired Student's t-tests or two-way 274 ANOVA tests. Data is represented as mean + standard deviation to show the spread of the data.

275
Significance was defined as a value of p < 0.05. All statistical analysis was completed using

431
We tested whether loss of SORL1 expression affected the size of Rab11+ recycling endosomes.

432
Interestingly, we observed a significant increase in the size of Rab11+ recycling endosomes,    another key pathology of AD: neurodegeneration, a slowly progressive process that begins with 533 synaptic dysfunction characterized by glutamate receptor loss, which then progresses to synaptic 534 loss before ultimately, over years, leading to widespread neuronal cell death (Selkoe, 2002).

536
The early endosome is considered the central station in the sorting and trafficking of cargo 537 throughout the many stations of the endo-lysosomal system. While the early endosome is the 538 station that is affected first and foremost in AD, it is not surprising that a primary dysfunction in 539 this central station will secondarily influence trafficking throughout the system including the

592
Interestingly, when we further interrogated our RNA-seq data for pathways enriched in synaptic function using the SynGo database (Koopmans 2019), we observed an upregulation of 594 differentially expressed genes in synaptic pathways (Supplemental Fig. 5). This data suggests 595 that SORL1KO neurons may attempt to compensate for altered trafficking of synaptic receptors 596 by upregulating gene expression.

597
Our unbiased transcriptomic screen further supported that neurotrophic signaling and cell 598 surface recycling pathways are impacted by SORL1 deficiency (Figure 7). The SORLA 599 cytoplasmic tail has been shown to translocate to the nucleus and activate transcription in a 600 reporter gene assay (Bohm et al., 2006). Despite this, distinct genes regulated by SORLA are 601 not known. Rather than looking for a direct effect on gene regulation, our goal for the analysis 602 was to determine the global effect of SORL1 loss or overexpression on neuronal networks.

603
Indeed, our data does not show that the specific cargo proteins described here are differentially

610
Trafficking of glutamate receptors is an event that is critical for preventing synaptic dysfunction 611 and synaptic loss, thus our results link SORL1 to AD's early-stage neurodegenerative process.

612
Since retromer-dependent glutamate receptor recycling has been shown to occur independent of 613 APP (Temkin et al., 2017), our previous and current results suggest that SORL1 mutations can, 614 at least in principle, drive two key AD pathologies, amyloid pathology and synaptic pathology, 615 through parallel mechanisms (Small and Petsko, 2020).
We summarize our findings in Figure. 8. However, our study has certain limitations. For example,

617
our results encompass only one human genome. Future studies will benefit from looking at 618 SORL1 deficiency or overexpression in multiple human genetic backgrounds. Furthermore, in this 619 work we are describing purely neuronal phenotypes although SORL1 is expressed in other CNS 620 cells. Future work looking at cell-autonomous and non-cell autonomous mechanisms of SORL1 621 depletion or overexpression in human glial or brain organoid models will be informative.

623
In this work, we report that SORL1 depletion affects endosomal trafficking by retaining cargo in       Representative immunofluorescent images of WT and SORL1KO neurons showing increased colocalization of (a) TRKB (green) and (b) GLUA1 (green) with EEA1 (red). All neurons were immunolabeled with MAP2 (far-red) and counterstained with DAPI (blue). Scale bar: 10µm. In all cases, quantification of colocalization was represented as Mander's correlation co-efficient (MCC). 1 WT and 2 SORL1KO isogenic clones were used for these experiments and 10 images per clone per genotype were analyzed. Data represented as mean ± SD. Data was analyzed using parametric two-tailed unpaired t test. Significance was defined as a value of *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. neurons. Scale bar: 10µm. In all cases, quantification of colocalization was represented as Mander's correlation co-efficient (MCC). 1 WT and 2 SORL1KO isogenic clones were used for these experiments and 10 images per clone per genotype were analyzed. Data represented as mean ± SD. Data was analyzed using parametric twotailed unpaired t test. Significance was defined as a value of *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. 10 images per genotype were analyzed. Data represented as mean ± SD. Data was analyzed using parametric two-tailed unpaired t test. Significance was defined as a value of *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. for these experiments. 10 images per clone per genotype were analyzed. Data represented as mean ± SD. Data was analyzed using parametric two-tailed unpaired t test and two-was ANOVA. Significance was defined as a value of *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. (b) (red). Scale bar: 5µm. Intensity of APP and GLUA1 measured using ImageJ software. Data is presented as a ratio of surface intensity to total intensity. 2 WT and 2 SORL1KO clones were used in these experiments. 16 images per clone per genotype were analyzed. Data represented as mean ± SD. Normally distributed data was analyzed using parametric two-tailed unpaired t test. Significance was defined as a value of *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. (c) Multielectrode array (MEA) analysis of WT and SORL1KO neurons at early (d27) and late (d66) time points of differentiation. 1 WT and 1 SORL1KO clone was used for these experiments.
Scale bar: 10µm. In all cases, quantification of colocalization was represented as Mander's correlation coefficient (MCC). 1 WT and 2 SORL1KO isogenic clones were used for these experiments and 10 images per clone per genotype were analyzed. Data represented as mean ± SD. Significance was determined using parametric two-tailed unpaired t test and was defined as a value of *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. represented as mean ± SD. Normally distributed data was analyzed using parametric two-tailed unpaired t test.

Supplemental figure 3.
Loss of SORL1 does not change protein expression of the compartments or cargo analyzed in this study as analyzed by Western blot. Representative blots in (a), quantification in (b). There are significantly more down regulated genes than upregulated genes (p<2.2e-16). Grey circles represent genes that are not significantly differentially expressed.

Supplemental figure 5 .
Loss of SORL1 expression alters synaptic pathways. Analysis of bulk RNA-sequencing data indicates alterations in synaptic pathway functioning. We conducted gene ontology analysis of DEGs in WT and