*2.3. Acetylation of Tau Inhibits Heparin-Induced Aggregation*

Heparin has been widely used to induce and accelerate Tau aggregation [11]. Utilizing a truncated Tau construct, Ambadipudi et al. demonstrated that heparin promotes Tau fibrillation via LLPS [37]. Similarly, we observed that heparin induces LLPS of full-length wt Tau and facilitates subsequent protein aggregation (Figure 3A–C,F). In contrast, Ac-Tau failed to undergo heparin-induced LLPS in the same experimental conditions (Figure 3D). Additionally, Ac-Tau (relative to wt Tau) exhibited a dramatic decrease in the fibrillation rate as reported by Th T fluorescence (Figure 3F). Residues in the VYINK280K281 and VQIVK311 regions of the Tau microtubule binding repeats (R1–R4), which we identified as Tau acetylation sites (Figure 1), are also known interaction sites for heparin [39]. Thus, the observed effects of acetylation on Tau heparin-induced aggregation can be attributed to the loss of binding to heparin.

**Figure 3.** Acetylation disfavors heparin-induced Tau aggregation. (**A**,**B**) Heparin accelerates non-acetylated wt Tau LLPS. The presence of heparin (5 μM heparin: 20 μM Tau ratio) results in more droplets and larger fused droplets at the bottom of the dish. Droplet formation was not observed for Ac-Tau in the absence (Figure 2B,D) or presence of heparin (24 h, (**D**)). After 24 h incubation, an abundance of irregularly-shaped oligomers/aggregates was observed for wt Tau but not for Ac-Tau (**C**,**D**). No aggregation/LLPS was observed for the heparin control (**E**). Ac-Tau displayed minimal aggregation compared to wt Tau as reported by Th T fluorescence assay (**F**). Scale bars represent 10 μm. See Section 3 for details.

Although heparin accelerates wt Tau LLPS, it is unknown whether heparin is equally distributed in the Tau-rich and Tau-poor phases (which we think to be unlikely). LLPS, nevertheless, allows Tau to co-localize and thereby concentrate, with the Tau-rich condensed phase facilitating Tau aggregation nucleation and/or seeding.

#### *2.4. Acetylation of Tau Prevents Access to LLPS-Mediated Microtubule Assembly*

A recent report by Hernandez-Vega et al. suggests that Tau phase separated droplets (induced using the crowding agents PEG, Ficoll or dextran) can initiate microtubule assembly [35]. To assess LLPS-mediated microtubule assembly by wt Tau and Ac-Tau independent of crowding agents, we performed our phase separation experiments in low-salt conditions. After mixing rhodamine-labeled and unlabeled tubulin heterodimers with wt Tau, we observed an initial increase in solution turbidity. The ensuing dynamic microtubule assembly was visible by fluorescence microscopy within 1 h of incubation (Figure 4A). In contrast, Ac-Tau neither displayed turbidity nor detectable microtubule assembly up to 18 h of incubation (Figure 4B). Whereas previous studies have shown that acetylation reduces Tau's ability to bind to microtubules [20], our data clearly demonstrates that the failure of Ac-Tau to undergo LLPS affects its potential for microtubule assembly.

**Figure 4.** Tau acetylation prevents access to LLPS-mediated microtubule assembly. (**A**,**B**) Fluorescence microscopy images showing microtubule assembly from mixtures of rhodamine-labeled and unlabeled tubulin heterodimers with Tau or Ac-Tau, respectively. Scale bars represent 10 μm. See Section 3 for details.

Ourin vitro data indicate that Ac-Tau is less prone to aggregation as compared to wt Tau. Cryo-EM structures of AD patient-derived filaments indicate that Tau residues 306–378 form the amyloid core [40]. The stable core is composed of several β-strands that pack intra- and inter-molecularly, with β1 (306VYINK311) in close proximity to β8 [40]. Our results show that in Ac-Tau, K311 (β1), and K375 (β8) are both acetylated; we speculate that this influences interactions within the amyloid core, and contributes to inhibition of Tau aggregation. Further experiments on the acetylation of the amyloid core residues will be needed to directly assess the effect of Tau acetylation on the amyloid structure.

Recent reports suggest that Tau aggregation is accelerated through LLPS [36,37]. Our data clearly show that acetylation decreases or abolishes Tau LLPS. Our findings are consistent with an LLPS-mediated model of aggregation (but do not prove whether such a mechanism is operative in vivo). Since acetylation reduces the propensity of Tau to undergo LLPS, we conclude that acetylation in vivo is unlikely to enhance or lead directly to condensation-mediated aggregation, in contrast to the demonstrated effect of hyperphosphorylation [36]. It is, however, possible that combinations of phosphorylations and acetylations can favor LLPS and/or aggregation; future experiments with Tau bearing homogeneous PTMs will be needed to address this conclusively.

Tau participates in microtubule formation and stabilization, and Tau LLPS has been shown as a mechanism by which a Tau-rich condensed phase can recruit tubulin dimers and facilitate their assembly [20]. Acetylation at key Tau sites that interfere with tubulin binding would affect this function, as would acetylation that disfavors partitioning of Tau into a Tau-rich phase. Thus, we speculate that the primary contribution of Tau acetylation to cellular dysfunction is not through a gain-of-function mechanism, such as toxic aggregation, but through a loss of physiologic function mechanism (i.e., reduced binding to tubulins/microtubules, and decreased LLPS-mediated initiation of microtubule assembly; Figure 5).

**Figure 5.** Model for Tau's loss of physiologic function and gain of pathologic dysfunction linked to its ability to undergo LLPS as modulated by acetylation.

Less direct effects on physiologic Tau function may also be important. Many of the same lysines (K254, K311, and K353) implicated as sites of ubiquitination [41] are also sites of acetylation and, thus, might be involved in evading the ubiquitin-lysosome proteasomal degradation machinery. Acetylation has been shown to inhibit Tau degradation by inhibiting its ubiquitination [21], and results in the accumulation of Tau, including hyperphosphorylated Tau. The presence of lysine deacetylase (SIRT1) has been shown to inhibit neuronal loss in an AD mouse model and deletion of SIRT1 results to pathologic levels of Tau in vivo [42]. Cross-talk between the different PTMs has also been reported. For example, hypoacetylation of Tau at key KIGS motifs in the *R1*-*4* regions increases vulnerability to hyperphosphorylation, which leads to filament aggregation [23]. Hyperphosphorylation of Tau has been reported to enhance Tau LLPS. However, other reports also show that hyperphosphorylation reduces microtubule assembly [14]. Thus, LLPS-mediated mechanisms by hyperphosphorylated Tau could be detrimental for both function and dysfunction pathways (Figure 5). We plan to carry out further experiments on hyperphosphorylated Tau to assess how this PTM of Tau can modulate microtubule assembly and protein aggregation, both in LLPS and non-LLPS conditions. Nevertheless, we speculate that the hyperacetylation of Tau is detrimental to Tau function, but not instrumental to LLPS-mediated Tau dysfunction (Figure 5). It would also be interesting to know the cross-talks between hyperphosphorylation and hyperacetylation in LLPS-mediated microtubule assembly and promotion of pathologic fibrils. Can hyperphosphorylated Tau also recruit hyperacetylated Tau into droplets? If so, this might explain the presence of hyperacetylated Tau in pathological inclusions of hyperphosphorylated Tau.

In conclusion, our data affirm the importance of electrostatics in Tau LLPS. Furthermore, we show that hyperacetylation disfavors Tau LLPS and, as a consequence, LLPS-facilitated aggregation. Finally, by preventing access to LLPS-mediated microtubule assembly and stabilization, hyperacetylation contributes to Tau dysfunction primarily through a loss-of-function mechanism.
