Alzheimer’s disease-relevant tau modifications selectively impact neurodegeneration and mitophagy in a novel C. elegans single-copy transgenic model

Background A defining pathological hallmark of the progressive neurodegenerative disorder Alzheimer’s disease (AD) is the accumulation of misfolded tau with abnormal post-translational modifications (PTMs). These include phosphorylation at Threonine 231 (T231) and acetylation at Lysine 274 (K274) and at Lysine 281 (K281). Although tau is recognized to play a central role in pathogenesis of AD, the precise mechanisms by which these abnormal PTMs contribute to the neural toxicity of tau is unclear. Methods Human 0N4R tau (wild type) was expressed in touch receptor neurons of the genetic model organism C. elegans through single-copy gene insertion. Defined mutations were then introduced into the single-copy tau transgene through CRISPR-Cas9 genome editing. These mutations included T231E and T231A, to mimic phosphorylation and phospho-ablation of a commonly observed pathological epitope, respectively, and K274/281Q, to mimic disease-associated lysine acetylation. Stereotypical touch response assays were used to assess behavioral defects in the transgenic strains as a function of age, and genetically-encoded fluorescent biosensors were used to measure the morphological dynamics and turnover of touch neuron mitochondria. Results Unlike existing tau overexpression models, C. elegans single-copy expression of tau did not elicit overt pathological phenotypes at baseline. However, strains expressing disease associated PTM-mimetics (T231E and K274/281Q) exhibited reduced touch sensation and morphological abnormalities that increased with age. In addition, the PTM-mimetic mutants lacked the ability to engage mitophagy in response to mitochondrial stress. Conclusions Limiting the expression of tau results in a genetic model where pathological modifications and age result in evolving phenotypes, which may more closely resemble the normal progression of AD. The finding that disease-associated PTMs suppress compensatory responses to mitochondrial stress provides a new perspective into the pathogenic mechanisms underlying AD.

are necessary for neuronal health (28). Although it is likely that tau pathology affects 84 mitochondrial biology, the underlying mechanisms are not well understood, nor it is known 85 how tau modified at disease relevant sites differ from its wild type form in causing 86 mitochondrial abnormalities leading to neurodegeneration. 87 To understand the role of tau in the context of AD per se, tau transgenic models have 88 been developed in C. elegans (29-31), D. melanogaster (32,33)   were used to study the impact of tau and mutant tau expression on neuronal morphology 105 and mitochondrial phenotypes, with the advantage of being able to assess age-106 dependence in a relatively short time frame (38, 39). 107 Our results clearly demonstrate that wild-type tau has little effect at baseline, but that AD-108 relevant tau PTMs selectively impact sensory neuron function and morphology and 109 mitochondrial handling. Moreover, age exacerbates defects in one of the tau mutant 110 strains, but not the others. This leads us to conclude that using our single copy tau model  The wild-type background strain was Bristol-N2. Other  Nematodes were maintained at 20 0 C on Nematode Growth Media (NGM) plates made 175 with Bacto Agar (BD Biosciences). The plates were seeded with live E. coli OP50-1 176 bacterial strain (cultured overnight at 37 o C at 220 rpm) and allowed to grow overnight, as 177 previously described (45). For experimental assays, after synchronization by standard 178 procedure with sodium hypochlorite, 4 th larval stage (L4) hermaphrodites (characterized 179 by the appearance of a "Christmas tree vulva") were selected and moved to test plates.

180
The day after moving was considered adult day 1, and animals were assayed on day 3 181 and day 10. Animals were transferred daily to avoid mixed population until they stop laying 182 eggs.    Touch sensitivity assay 203 The behavioral response to being touched by an eyelash was adapted from an assay    Mitochondrial stress assays 220 For paraquat (PQT) mild stress assays, synchronized 2-day old adult and 9-day old adult 221 hermaphrodites were exposed to 8 mM PQT (51, 52) in NGM plate for overnight at 20 o C.

222
Animals were picked from the respective (treated and control) plates the next day and 223 imaged, as described below.

224
Neurodegeneration assay 225 For imaging, animals were mounted by placing them in 3% agarose pads on glass slides 226 and immobilized with 1 mM tetramisole hydrochloride (Sigma). Imaging was performed 227 using Confocal Laser Scanning Confocal microscope (Olympus 1X61) and FV10-ASW

266
All statistical analyses were conducted using Prism 8.0 (GraphPad Software), with alpha-267 error level of p < 0.05 considered to be significant. Data were averaged and represented 268 as mean ± standard error (mean ± SEM) unless otherwise noted. In general, group 269 differences were analyzed with either one-way or two-way ANOVA depending upon the   Fig. 1E-H), which mediate the behavioral response to light touch.

287
The 0N4R fusion to Dendra2 will be referred to hereafter as TauT4. Dendra2 was also 288 expressed alone (Fig. 1A-D), and these negative control strains responded to light touch 289 similarly to the wild-type N2 strain at both day 3 and day 10 of adulthood (Fig. S1).
290 Surprisingly, TauT4 worms exhibited normal touch responsiveness as both young day 3 291 post-reproductive adults (Fig. 2B) and older day 10 adults (Fig. 2C). In order to address 292 the effect of tau PTMs, CRISPR-Cas9 gene editing (43, 44) was used to introduce 293 phosphomimetic T231E, phosphoablation T231A, and acetylmimetic K274/281Q 294 mutations into the TauT4 ORF ( Fig. 2A). For simplicity, these mutants will be referred to 295 as T231E, T231A and K274/281Q. Our results clearly demonstrate that T231E exhibited 296 subtle but significant defects in touch responsiveness at both day 3 and day 10, while 297 K274/281Q was different from the Dendra2 control only at day 10 ( Fig. 2B,C). However, 298 between day 3 and day 10, the touch sensitive phenotype of K274/281Q worsened 299 significantly (p = 0.01). This may indicate either a ceiling effect of T231E or a sensitized 300 K274/281Q progression with age. In contrast, T231A was indistinguishable from TauT4.

301
The differences between the disease-associated mutants and TauT4 represents a novel 302 observation and a first-in-kind platform for studying the effect of pathologic tau 303 modifications in the absence of baseline defects. Finally, since survival plots of the 304 various strains used in this work were statistically indistinguishable, we were able to rule 305 out any phenotypic age-dependence being due to a change in lifespan (Fig. S2E, F). 306 We also evaluated several other stereotypical behavioral measures that have been 307 shown to be influenced by age but do not involve touch cell neurons, including thrashing 308 in liquid (Fig. S2A, B) and basal locomotion on solid media (Fig. S2C, D). Taken together, 309 these data suggest that the effect of pathological, AD-relevant tau expression in touch 310 sensory neurons is restricted to the behavioral response to light touch. We found that T231E strongly and significantly increased the 322 incidence of overextension from ~4% to ~40% by day 3 of adulthood (Fig. 3G). However, 323 the TauT4 and K274/281Q mutants were not significantly different from Dendra2 controls 324 in day 3 adults (Fig. 3G).

325
In addition to overextension defect, other neuritic abnormalities develop with age, such 326 as branching, guidance defects, beading, and breakage ( Fig. 3C-F). While none of the 327 strains were significantly different in terms of these defects at day 3 (data not shown), 328 both T231E and K274/281Q exhibited an increased incidence of overextension, guidance, and gap defects at day 10, but were not different with respect to branching or 330 beading compared to the Dendra2, TauT4 or T231A mutant strains (Fig. 3H-K). It was 331 intriguing that age exacerbated the overlap defect in K274/281Q (p = 0.05, between day 332 3 and day 10), which mirrored its effect on touch sensitivity, but that T231E had reached 333 its maximum penetrance by day 3 of adulthood. These results suggested to us that this 334 model is appropriate to detect subtle differences in pathology and "disease" progression 335 as a function of specific tau PTMs. can be used as a dual excitation ratiometric mitophagy reporter, as we expand upon 346 below ( Fig. 5 and 6). However, here we used single wavelength excitation-emission 347 imaging of mito-mKeima in the appropriate channel to visualize mitochondrial structure, 348 such as shown in Fig. 4. Under these image acquisition conditions, mitochondria are 349 visible, but mitochondria that have been engulfed by acidic vesicles are not (for 350 convenience, heretofore we will refer to these structures as "mitolysosomes"). Based 351 upon these images, mitochondria were categorized into four levels, from normal tubular-352 reticular morphology through increasing degrees of fragmentation (Fig. 4A-D and   353 Methods). Neuronal mitochondria from day 3 adult animals had generally tubular-reticular 354 morphology, and their distribution was independent of tau genotype (data not shown).

355
However, by day 10 of adulthood, all of the strains contained some fragmented 356 mitochondria, consistent with age-associated remodeling, but it was clear that T231E and 357 K274/281Q were significantly more fragmented than Dendra2, TauT4 or T231A (Fig. 4E).

358
Pathologic tau modifications suppress stress-induced mitophagy 359 Next, we employed mito-mKeima in dual excitation mode in order to assess organelle proteases. These characteristics allows a mitophagy index to be calculated using dual 371 excitation ratio imaging that reflects the relative amount of mitochondria that have 372 undergone engulfment and fusion with acidic vesicles. In addition, because these 373 mitolysosomes are spectrally and morphologically distinct (Fig. 5), we can also assess 374 their absolute abundance and size.

375
In PLM neurons, pathologic tau modifications T231E and K274/281Q had little effect on 376 baseline mitophagy, but decreased the number of mitolysosomes in young adults (Fig.   377 5). We also note an apparent increase in the mitophagy index and reduction in the number 378 of mitolysosomes with age that reached significance in Dendra2, TauT4, and T231A, but 379 not in T231E and K274/281Q (Fig. 5G, H, I, J).

395
A characteristic hallmark of the AD brain is the presence of tau with PTMs defined as 396 pathological, that likely contribute to the onset and progression of the disease.

397
Phosphorylation of tau at specific epitopes is widely appreciated to contribute to AD (6, which could be reflective of a deficit in mitophagy. Therefore, we measured the relative 442 amount of mitochondria that were engulfed and fused with acidic compartments, as well 443 as the absolute abundance of mitochondria in acidic compartments ("mitolysosomes"). 444 Interestingly, we found that in control animals, the mitophagy index, a measure of relative 445 mitolysosome to mitochondria abundance, increased with age, and the number of 446 mitolysosomes decreased (Fig. 5). The factors critical for the effective turnover of cumulative vesicle fusion, and so we need to temper our conclusion to reflect this caveat.

455
Nevertheless, we were able to stimulate mitophagy using PQT at day 10 to a similar 456 extent as day 3 (Fig. 6), confirming that, at a minimum, the ability to generate a robust 457 response to oxidative stress is maintained in older wild type animals.

458
Our results demonstrate a striking abolition of PQT-induced mitophagy in the AD-relevant 459 T231E and K274/281Q mutants (Fig. 6). This observation is consistent with defective 460 mitophagy being a prominent feature in age-related disorders (80), including AD (81), and 461 contributing to premature aging such as observed in Werner's syndrome patients and 462 invertebrate Werner's disease models (82). It is also particularly intriguing that the T231E 463 and K274/281Q do not appear to exhibit the same age-dependence as Dendra2,TauT4,464 or T231A. This could be interpreted to mean that these mutants exhibit characteristics 465 that appear in older adults. Their inability to response appropriately to oxidative stress -466 at both a young and old age -suggests that the mitochondria in fact do have baseline 467 defects, albeit at a level that is not discernable in the absence of stress. The recent finding 468 that mitophagy enhancement can suppress AD-related phenotypes in tau transgenic 469 animals lends support to this idea (83).

470
It will be of interest to determine whether the tau mutants described here are perceived 471 as stressors, and hence cause activation of a retrograde response, such as has been In conclusion, to our knowledge this is the first study to clearly demonstrate that single 484 copy expression of tau with AD-associated PTMs impairs neuronal function and structure 485 in an age-dependent manner. In addition, the effect of tau modifications on stress-induced 486 mitophagy could lead to cumulative metabolic defects and energetic crises with age. One 487 advantage of our single-copy model is that it allows us to quantitatively measure subtle 488 deficits and discriminate between the effects of distinct PTMs. For example, we 489 demonstrate that T231E presents with a neuronal functional (and morphological) deficit