Optical induction of autophagy via Transcription factor EB (TFEB) reduces pathological tau in neurons

Pathological accumulation of microtubule associated protein tau in neurons is a major neuropathological hallmark of Alzheimer’s disease (AD) and related tauopathies. Several attempts have been made to promote clearance of pathological tau (p-Tau) from neurons. Transcription factor EB (TFEB) has shown to clear p-Tau from neurons via autophagy. However, sustained TFEB activation and autophagy can create burden on cellular bioenergetics and can be deleterious. Here, we modified previously described two-plasmid systems of Light Activated Protein (LAP) from bacterial transcription factor—EL222 and Light Responsive Element (LRE) to encode TFEB. Upon blue-light (465 nm) illumination, the conformation changes in LAP induced LRE-driven expression of TFEB, its nuclear entry, TFEB-mediated expression of autophagy-lysosomal genes and clearance of p-Tau from neuronal cells and AD patient-derived human iPSC-neurons. Turning the blue-light off reversed the expression of TFEB-target genes and attenuated p-Tau clearance. Together, these results suggest that optically regulated TFEB expression unlocks the potential of opto-therapeutics to treat AD and other dementias.


Introduction
Among the various microtubule-associated proteins (MAP), tau (encoded by MAPT) predominately localizes to axons where it binds to microtubules. Tau is known to promote nucleation, stabilization, and prevent disassembly of microtubules 1 . However, tau is susceptible to many post-translational modifications 2 , with phosphorylation being one of the well-studied modifications 3,4,5 . Upon hyperphosphorylation, tau linearly decreases its affinity to microtubules causing depolymerization 6 , which has been the prevailing hypothesis, that tau's loss-of-function causes neurodegeneration 7,8 . These dissociated forms of tau can self-assemble into paired-helical filaments (PHFs) gaining further potential to aggregate as Neurofibrillary tangles (NFTs) -a classical neuropathological hallmark of Alzheimer's disease (AD) and related tauopathies 9 . Alternatively, hyperphosphorylated and pathological tau (p-Tau) has been shown to acquire gain-of-toxic function in triggering synaptotoxicity relevant to AD 10 . While AD is the most common form of tauopathy and sixth leading cause of death in the United States 11 , NFT pathology is also the primary etiology in many, but rare tauopathies such as Progressive Supranuclear Palsy (PSP), Pick's disease (PiD), Corticobasal Degeneration (CBD), Fronto-temporal Dementia and Parkinsonism linked to Chromosome-17 tau-type (FTDP-17T) and others 12 . Because of the exponential rise in tauopathy related deaths, there is an urgent need to find intervention(s) against tauopathies.
A plausible strategy to prevent p-Tau from becoming pathological is to promote its degradation via autophagy in "at risk" neuronal populations. Impairment of autophagic processes has been implicated in several neurodegenerative disorders. However, there are still several reasons for considering autophagy's potential in clearing p-Tau: (i) Multiple reports have suggested a causal link between failure in autophagic processing and AD-related pathologies 13,14,15,16,17,18,19 , (ii) Autophagy disposes of potentially toxic intracellular protein aggregates too large for proteasomal removal 20,21 , and performs trophic functions 22,23 (sometimes referred to as "programmed cell survival"), (iii) autophagy removes depolarizes mitochondria 24 and shows cytoprotective interactions with stressed endoplasmic reticulum 25,26,27,28,29,30 that are of direct consequence for tauopathies 31,32,33,34,35,36 , (iv) previous work from our group provides compelling evidence that autophagy prevents spurious inflammasome activation 37,38 , which when left uncontrolled, could drive tau pathology and cognitive impairment 39 , and (v) chronic promotion of autophagic processing can enhance clearance of p-Tau and rescue neurotoxicity in a mouse model of tauopathy 40 . We have demonstrated that induction of autophagy via chemical or genetic means lead to clearance of inflammation-induced p-Tau 41 in neuronal cells. Transcription Factor EB (TFEB) regulates transcription of an entire CLEAR (Coordinated Lysosomal Expression and Regulation) network, which consists of a consensus site predominately found in the promoter regions of autophagylysosomal genes 42,43 . Thus, when TFEB localization is nuclear, it leads to robust increase in lysosome biogenesis, and results in accelerated degradation of autophagic substrates 30,44,45,46 . Phosphorylation of Ser211 in TFEB by mammalian target of rapamycin complex 1 or mechanistic target of rapamycin complex 1 (mTORC1) is one of the key regulators of nuclear localization, as the pS211 phosphorylation prevents TFEB from entering into the nucleus 47 .
However, the limitation of pro-autophagy studies is their focus on the continual activation of autophagy. While autophagy is generally thought to promote survival as discussed above, under certain conditions sustained autophagic-flux can lead to cell death 48,48 . Furthermore, prolonged activation of autophagy proteins (e.g. LC3 and BECN1) and vacuoles in response to ischemic stroke/reperfusion in vivo, or oxygen-glucose deprivation (OGD) in vitro leads to significant cell death 49 . Interestingly many autophagic processes do not significantly affect cell health until days after the injury, indicating that prolonged activation is critical for cell death to occur 50,51 . Another example, constitutive activation of the δ2 glutamate receptor was demonstrated to cause Purkinje cell death in Lurcher mice via activation of autophagy 52 . Thus, for elderly tauopathy patients with comorbid conditions such as ischemia and vascular dementia, sustained activation of autophagy could exacerbate cell death rather than reduce it. Therefore, it is important to develop tunable systems to turn-on/turn-off autophagy in neurons with optimum spatio-temporal control.
In the case of autophagy, which requires the coordinated expression and function of a host of proteins, optical induction of transcription by a master regulator offers such a system 53 . Here we optimized an optical induction system based on light-responsive bacterial transcription factor 52 to drive TFEB expression in a number of different mammalian expression systems that display pathological tau. For the first time, our group has shown that optically controlled TFEB efficiently expresses in human AD neurons, up-regulates TFEB target genes, and efficiently targets and reduces pTau.

Cytomegalovirus (CMV) promoter and the nuclear localization signal (NLS) sequence derived from cMyc shows robust gene expression with light.
We chose a light-inducible gene expression system that utilizes an engineered version of EL222, a bacterial transcription factor that contains a Light-Oxygen-Voltage (LOV) protein 72,54 (reiterated in Fig. 1A). This system was previously optimized with five copies of the bacterial EL222-binding Clone 1-20 base pairs (C120) 5 sequence 55,56 . This consensus site acts like a promoter region for the EL222 binding and drives the expression of any genes inserted downstream of C120 (FIG 1A-B).
In order to recapitulate Motta-Mena et al., EL222 system, we first tested their original two-plasmid system pVP-EL222, (that we call 'light activated protein' or 'LAP'); and their pC120-Fluc (Firefly Luciferase reporter that we call 'lightresponse element' or 'LRE'), where the luciferase gene was inserted downstream of LRE. As reported previously 72 , we also observed robust luciferase activation upon blue light illumination in HEK293T cells ( Fig. 1C-D). However, using the original pSV40-SV40 NLS-LAP in our neuro2a (N2a) cell line, we show more than a two-fold decrease in luciferase expression compared to HEK293Ts. Therefore we created two different EL222 light-responsive systems by replacing the SV40 promoter with a CMV promoter and including an additional cMyc nuclear localization signal (NLS) sequence (Fig. 1B). Strikingly, we observe statistically significant, over two-to four-fold luciferase expression upon blue-light stimulation compared to dark controls in both HEK293T and N2a's ( Fig. 1C-F). We also observed a different degree of luciferase expression with different promoter and NLS combinations with the pCMV-LAP-2x NLS showing the robust induction of luciferase expression in N2a cells. This difference was likely due to a weaker SV40 promoter in the LAP plasmid used in the original system 72 compared to more robust CMV promoter, as previously reported 57

Constitutively active TFEB (not light inducible) clears various types of pathological tau with equal efficiency in cellular models of tauopathy.
Now that the optogenetic system parameters have been set, there are a few other factors that need to be considered within our neuronal tauopathy model. First, as mentioned above, there are many forms of tauopathy. We wanted to determine whether TFEB has the ability to target different forms of pathological tau (pTau) and lead to its clearance via autophagy in neuronal cells. Tau gene (MAPT) in humans encodes six different isoforms that are contrasted by exons 2, 3 and 10 60 . Exon 10 encodes second microtubule binding repeat, thereby resulting in tau with either three (3R -without second repeat)-or four (4R -with second repeat) microtubule binding repeats of 31-32 amino acids in the carboxy terminal half 85 . Tau isoforms are also generated in either having or not having one (1N), two (2N), or zero (0N) amino terminal inserts of 29 amino acids each in the N-terminal half of the protein 85 . In normal adult brain, the relative amounts of 3R tau and 4R tau are approximately equal, however in neurodegenerative tauopathies, the ratio of 3R:4R is often altered 61 . Besides altered isoform rations, post-translational modifications, such as phosphorylation in tau, can also affect tau's function and contribute to disease pathogenesis in tauopathies 86 . We and others have previously shown that by TFEB degrades pTau via beclin-1 dependent autophagy pathway 31,32 . However, the effectiveness of TFEB on different isoforms of tau with different disease-modifications has not been tested.
To test this, we co-transfected TFEB with different forms of tau; (1) pCMV-0N3R -non-mutant tau, but over-expression can lead to Pick's Disease (PiD) 62  Next, we determine whether or not different tags (FLAG versus GFP) to the TFEB have any differential effects in clearing pTau via autophagy. Coexpression of T231D/S235D tau with either pCMV-TFEB3xFLAG or pCMV-TFEB-GFP showed that GFP tagged TFEB was slightly efficient in clearing mutant tau than 3xFLAG tagged TFEB ( Fig. 2C-D). Given that TFEB has to be nuclear for it to be functionally efficient and to drive the expression of genes in the CLEAR network, we next assessed the effects of S211 phosphorylation in TFEB in clearing mutant tau. TFEB-GFP with the S211A mutation, which prevents phosphorylation by mTORC1 70 thus promotes TFEB's nuclear entry, had the greatest effect in reducing T231D/S235D mutant pTau (FIG2 C-D).
Together, these results suggest that genetically facilitating the nuclear entry of TFEB does provide added advantage in enhancing the autophagic clearance of T231D/S235D tau.

Optogenetically expressed TFEB activates CLEAR network genes in neuronal cells.
To determine the optimal parameters for the optogenetically-driven TFEB (Opto-TFEB) expression in N2a cells, we co-transfected N2a cells with the two different LAP's and pLRE-TFEB(S211A)GFP plasmids, stimulated the cells with blue light for 12 h, fixed in 4% PFA, immunostained for VP16 and GFP and performed confocal microscopy. Double immunofluorescence analysis (for VP16 and GFP), with the substitution of a CMV promoter and additional cMyc NLS, we observed a significant increase of TFEB expression (as revealed by anti-GFP staining) with blue-light stimulation compared to dark control ( Fig. 3A-B). As expected, the VP16 staining was detectable in pCMV-LAP (Fig. 3A).
After having established that TFEB expression can be induced optogenetically (Opto-TFEB) in neuronal cells, we set to determine whether lightinduced TFEB expression is functionally active and display expected transcriptional function. Based on the TFEB's preferential binding to the CLEAR consensus site ( 5' GTCACGTGAC 3' ), we next utilized a previously published 71 , reporter plasmid pCLEAR-FLuc to assess the functional activation of Opto-TFEB.
The pCLEAR-FLuc plasmid consists of four replicates of the CLEAR consensus sequence upstream of the luciferase gene, serving as a DNA-binding activity readout for TFEB. We transiently co-transfected pCLEAR-FLuc with pLAPs, and pLRE-TFEB(S211A)GFP in N2a cells and stimulated the cells with blue light overnight (12 h). Shortly after luciferin treatment, whole cell culture plates were imaged using IVIS Lumina Series III. We observed a significant increase in TFEB DNA-binding activity (enhanced CLEAR-luciferase signal) only in cells that optogenetically expressed TFEB with light, and not in the dark control ( Fig.3C-D).
Together, our results suggest that Opto-TFEB is functionally active in driving expected transcriptional activity of genes in the CLEAR network.

Opto-TFEB reduces pathological tau in neuronal cells
Because we have shown Opto-TFEB can drive transcriptional regulation of the CLEAR network (responsible for autophagy flux and lysosomal biogenesis), then it should in theory increase the autophagic flux and reduce the levels of misfolded p-Tau upon light stimulation. We took a multiple approach in establishing that Opto-TFEB is capable of clearing specifically phosphorylated species of Tau. First, we used confocal microscopy to measure the levels of pTau in N2a cells that expressed Opto-TFEB. We overexpressed human mutated 0N3R-T231D/S2345D tau along with pCMV-LAP2xNLS and pLRE-TFEB(S211A)GFP. Upon cell-by-cell analysis, we observed that for every Opto-TFEB + cell there is little to no Tau12 staining (Fig. 4A). Using Zeiss software to measure total geometrical intensity levels we saw significant increase in Opto-TFEB levels compared to dark. To perform an unbiased quantification, we performed quantitative morphometry for Tau12 levels using high-content and automated Cellomics® microscopy. We observed a significant decrease in overall Tau12 intensity levels in Opto-TFEB + cells compared to dark control (FIG4 C-D). Next, to validate the Cellomics® data, we performed Western blot analysis for total human tau (Tau12) in N2a cells. Strikingly, we observed statistically significant increase in TFEB-GFP expression (Fig. 4D-E) and reduction in the levels of total tau (Tau12) (Fig. 4D-F)  To assess the efficacy of Opto-TFEB in sAD2.1 cells, it was necessary to clone Opto-TFEB plasmids (pCMV-LAP2xNLS and pLRE-TFEB(S211A)GFP) into a lentiviral system and subsequently co-transduce sAD2.1 iPSNs. Using confocal microscopy and analyzing with Zeiss software for overall geometric intensities, we note there is a significant decrease in pTau (AT8) levels in our light control compared to dark (FIG5 A-B). We also wanted to make sure that the reduction is due strictly because of Opto-TFEB induction. To do so, we measured the gene expression levels of three known TFEB targets; PTEN 38 , CTSF, and MCOLN1 40 . (FIG5 C) Compared to Dark, we show significant increase in TFEB expression as well as the three targets we chose. Lastly, we took a look at overall levels of pTau (AT8 and AT180) at the total protein level (FIG5 D-E). In conclusion, for the first time we show that light controlled / optogenetic TFEB has the ability to reduce pTau in a highly respectable tauopathy model.

DISCUSSION
TFEB is a master transcriptional regulator of autophagy and lysosome biogenesis. However, studies have revealed TFEB constitutes and interacts with a variety of biological functions, including the inflammatory process, stress responsive pathways 74 , oxidative stress 75 , and metabolic regulation 76 .
It has been shown in many studies how autophagy contributes to human AD diseased iPSNs and the endogenous phenotypes/disease characteristics. For example, Reddy et al. generated iPSC-derived human forebrain cortical neurons from AD patients with M146L and A246E mutations, as well as a presenilin-1 (PS-1) knockdown in control neurons 77 . Using the same CLEAR-luciferase reporter assay, they found a reduction in CLEAR activity in their forebrain cortical iPSNs as well as decrease in LC3II levels in PS-1 knockdown neurons, advocating decreased autophagy flux. Another group, again used iPSC-derived forebrain cortical neurons studied the amino acid metabolite homocysteine (Hcy), and found that exposure to Hcy caused reduced autophagic activity by a means of increase in mTORC1 activity therefore reduction in TFEB activity by phosphorylation of TFEB by mTORC1 78 .
Therefore, in order to setup a system that employs TFEB as a therapeutic target, it cannot be constitutively active or termed in an active/nuclear state.
Here we describe a new blue light inducible TFEB gene expression system that works well in mouse neuronal cell lines and human AD iPSCs derived into mature neurons. Not only have we shown successful light controlled gene expression, but we also effectively enhanced the autophagy flux, specifically targeting and reducing pathological tau in our AD iPSNs. The benefits of using diseased iPSNs in this study, is that when derived into a neuronal population, the AD phenotype is displayed 98,79,80,81 . This avoids the hurdle of low tri-plasmid transfection efficiency. Now that we have successfully shown the benefits of neuronal optoTFEB induction, it may be interesting to see what remunerations could be achieved when cell-type specific optoTFEB is accomplished. This would answer the question if it poses more beneficial characteristics if optoTFEB only was induced in glia cells, like astrocytes or microglia, using a GFAP or CX3CR1 promoter to drive the LAP transgene. It also may be necessary to induce OptoTFEB in various time-points, to assess cell toxicity and or the ability of OptoTFEB to be beneficiary in a dose-like dependent manner. Altogether, our data strongly suggest that OptoTFEB efficiently expresses in AD iPSNs, up-regulates TFEB target genes, and efficiently targets and clears pTau.

Vector construction
All constructs were cloned using NEB HIFI Assembly Kit (NEB # E5520S) with restriction enzymes and PCR amplification. Briefly, the original episomal plasmids gifted by Motta-Mena, (pVP-EL222 and pGL4-C120-mCherry) were For Firefly luciferase activities, 4XCLEAR-luciferase reporter plasmid #66800, purchased from addgene. D-Luciferin, Potassium Salt (ThermoFisher # L2916) was reconstituted in water and was added (1:100) to each well, 3 -4 mins after addition of substrate, 24-well plate samples were analyzed through the IVIS Lumina Series II with system software.

Western blotting (WB) and immunocytochemistry (ICC)
WB -Cells were lysed by RIPA buffer (Thermo #89900), incubated on ice for 30 mins then centrifuged at 20,000 × g for 15 min. Cell lysate supernatants were then sonicated for 20secs at 30%, then subjected to SDS-PAGE usage, transferred to PVDF membranes and detected using the ECL method (Pierce).

LAMP1.
ICC-Cells were plated on coverslips coated with laminin, once cells were ready for fixation, they were fixed in 4% PFA, blocked with 0.2%triton and 10% donkey serum, incubated in primary overnight in 4 o C (5% DS), secondaries were incubated for 1hr at RT. Incubated in DAPI for 10 mins, and mounted to slides using fluoromount(CAT#). Immunofluorescence confocal microscopy was carried out using Zeiss LSM 510 Meta microscope. Histo and profile analysis was performed using ZEISS ZEN imaging Software.

Gene expression analysis
RNA from cells was extracted using the TriZOL reagent as described by the manufacturer (Thermo Fisher Scientific). Total RNA (20 ng/µL) was converted to cDNA using the High Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific) and amplified using specific TaqMan assays (catalog # 4331182; Thermo Fisher Scientific). GAPDH (catalog # 4352339E, Thermo Fisher Scientific) was used as a housekeeping gene for normalization. qRT-PCR assays were run on the StepOnePlus® Real-Time PCR System (Thermo Fisher Scientific) and the statistical analyses were performed using Prism.

Cellomics®-based high-content imaging analysis
Cells were plated in 96 well plates transiently transfected with pCMV-T231D/S235D (phosopho-mimicking tau), pCMV-LAP2xNLS, and pLRE-TFEB(S211A)GFP. Twenty-four hours later, cells were incubated with conditioned medium from BV2's, as previously described, then subsequently induced with light (470nm) for 12 hours. Cells were fixed in 4% PFA, blocked with 0.2%triton and 10% donkey serum, incubated in primary for one hour at RT (5% DS), secondary was incubated for 1hr at RT. Incubated in DAPI for 10 mins and analyzed through cellomics machine.

Statistics
Unless otherwise indicated, comparisons between the two groups were done via unpaired t test; comparisons between multiple treatment groups were done via one-way or two-way analysis of variance (ANOVA) with indicated multiple comparisons post-hoc tests. All statistical analyses were performed using GraphPad Prism®.