Tyrosine Inhibits TyrRS-mediated DNA Repair and Induces Neuronal Oxidative DNA Damage


 Human aging and neurodegenerative diseases accumulate oxidative DNA damage-associated mutations in neurons. Circadian-regulated tyrosine (Tyr) is increased during aging and in Alzheimer’s Disease (AD). Tyr exacerbates the cognitive decline in the elderly and AD patients. Tyrosyl-tRNA synthetase (TyrRS) that activates Tyr for protein synthesis and participates in DNA repair is depleted in the affected brain regions of AD patients through an unknown mechanism. Here, we found that increased Tyr levels decrease the nuclear and neurite levels of TyrRS in neurons and cause oxidative DNA damage. Although Tyr inhibits protein synthesis at the elongation step, dopamine (DA)- a neurotransmitter derived from Tyr increases TyrRS levels. We previously showed that Tyr inhibits TyrRS-mediated activation of poly-ADP-ribose polymerase 1 (PARP1), a modulator of DNA repair. We now found that trans-resveratrol (trans-RSV) that binds to TyrRS mimicking ‘Tyr conformation’ decreases TyrRS, inhibits DNA repair and induces neurotoxicity. Conversely, cis-RSV binds to TyrRS mimicking a ‘Tyr-free conformation,’ increases TyrRS, facilitates DNA repair, and protects neurons against multiple neurotoxic agents in a TyrRS-dependent manner. Our results suggest that increased Tyr levels may have causal effects in human aging and neurocognitive disorders and offer a plausible explanation to divergent results obtained in clinical trials using RSV.


Introduction
TyrRS is a member of the evolutionarily conserved family of aminoacyl-tRNA synthetases (aaRSs) that activates the aromatic amino acid (AAA) Tyr for protein synthesis 1 . TyrRS has a moonlighting function in DNA repair as well 2,3 and is essential for cell survival 4 . In neurons, brainderived neurotrophic factor (BDNF) stimulates the de novo synthesis of TyrRS 5 . Protein synthesis is required for long-term memory formation 6,7 , and consistently, the brain protein level of TyrRS is depleted in the affected brain regions of AD patients 8 through an unknown mechanism. Tyr is circadian-regulated with the highest serum levels in the morning and lowest at midnight (sleep time) 9,10 . Sleep stimulates human brain protein synthesis 11 , memory formation 12,13 , and neuronal DNA repair 14 . Beyond feeding and protein synthesis 15 , Tyr level is modulated by the circadian activities of tyrosine hydroxylase (TH) 16 , tyrosine aminotransferases (TAT) 17,18 , and gut microbiota 19 . However, serum levels of Tyr increase during human aging [20][21][22] , and increased Tyr level is a biomarker for the development of type 2 diabetes (T2D) in humans 23 , a major risk factor for dementia 24 and AD 25 . Genetic mutations that increase the levels of Tyr (tyrosinemia) [26][27][28][29] or its precursor L-phenylalanine (Phe, phenylketonuria [PKU]) 30 cause multiple health problems, including cognitive deficits in children. Moreover, treatment with Tyr exacerbates the cognitive decline in the elderly [31][32][33] and AD patients 34 , and dysregulated Tyr metabolism shortens lifespan in tyrosinemia patients 35 through unknown mechanisms.
We previously showed that Tyr inhibits TyrRS-dependent activation of PARP1 36 , a major modulator of DNA repair 37,38 . Here, we found that increased Tyr levels decrease the nuclear and neurite levels of TyrRS and cause neuronal oxidative DNA damage, potentially through the inhibition of protein synthesis at the elongation step. However, the neurotransmitter derived from Tyr (dopamine) stimulates the de novo synthesis of neuronal TyrRS. Furthermore, consistent with our previous findings that cis-RSV evokes a 'Tyr-free conformation' 39 , and trans-RSV mimics a 'Tyr conformation' 36 in TyrRS, we now show that cis-and trans-RSV have opposite effects on TyrRS levels and neuronal DNA repair. cis-RSV protects the neurons against Tyr and other neurotoxic agents-induced depletion of TyrRS and DNA damage. Therefore, our results provide a potential molecular basis for the age-associated accumulation of neuronal oxidative DNA damage [40][41][42][43] and Tyr-induced cognitive impairments in elderly and AD patients [31][32][33][34] .

TyrRS is decreased in the hippocampal tissue samples of human AD patients.
AD decreases brain protein synthesis at the elongation step in humans [44][45][46][47] through an unknown mechanism. Recently published human brain proteome showed decreased TyrRS and phenylalanyl-tRNA synthetase beta (PheRS) levels in AD-affected brain regions 8 ( Supplementary Fig. 1a and b). We validated the depletion of TyrRS and PheRS in the hippocampal region of AD patients. However, the level of PheRS was not affected ( Fig. 1a and Supplementary Table 1). Our re-analysis of a second brain proteome 48 showed that the protein levels of TyrRS and PheRS correlate with cognitive performance in humans ( Supplementary   Fig. 1c) 48 . Conversely, their decrease correlates with AD status and Braak stages ( Supplementary   Fig. 1d) 48 . Intriguingly, a meta-analysis of human brain transcriptomic data from AD patients did not show any changes in the mRNA levels of TyrRS or PheRS 49,50 . While PheRS translation declines in an age-dependent manner (Supplementary Fig. 1e) 51 , TyrRS did not correlate with any known biomarkers of AD or other neurodegenerative diseases (Supplementary Fig. 1f) 48 , indicating that TyrRS protein level might be modulated by hitherto unknown factors.

Tyrosine is increased during aging and in neurocognitive disorders in humans.
Aging is the highest risk factor for neurodegenerative diseases 52 and intriguingly, the incidence of AD and other dementias is higher in women than in men 53 . Tyr level increases during human aging [20][21][22] and consistently, Tyr levels are also increased in human AD brain tissues [54][55][56] . Although there is no sex difference in Tyr levels in children 57 or AD patients 58 , interestingly, young women have lower serum Tyr levels than young men 21,57,59 . However, menopause increases Tyr levels 60 , resulting in a significant increase in Tyr levels in older women from ~60-70 yr 20 Table 2). Beyond neurodegenerative diseases 52 , aging in humans is associated with increased incidents of other diseases as well 61 . Consistently, our literature analysis showed that Tyr and/or Phe are increased during delirium, heart failure (HF), Parkinson's Disease (PD), autism spectrum disorders (ASD), cardiovascular diseases (CVD), and other metabolic disorders in humans (Supplementary Fig. 2b and Supplementary Table 3). Interestingly, increased levels of Tyr and/or Phe inhibit aminoacyl-tRNA synthesis in the brain ex vivo 62,63 and inhibit protein synthesis 64 and induce DNA damage 65,66 and promote oxidative stress in brain 67 in vivo in rat. Because human AD brain has decreased levels of TyrRS (Fig. 1a), and AD decreases brain protein synthesis at the elongation step [44][45][46][47] , and aged neurons 40 and AD brain 42 accumulate increased levels of oxidative DNA damage, and Tyr inhibits brain protein synthesis 64 , we hypothesized that increased levels of Tyr negatively regulate the protein levels of TyrRS and might cause neuronal oxidative DNA damage.
Moreover, the levels of TyrRS were restored in 16-24 hr ( Fig. 1e and Supplementary Fig. 3e), suggesting that the effect of Tyr on TyrRS protein levels is reversible. The neurotransmitter dopamine (DA), which is decreased during aging 68 , and in the affected brain regions of AD patients 54 , is generated from Tyr (L-Tyr L-DOPA DA). DA dephosphorylates eukaryotic elongation factor 2 (eEF2) 69 and activates neuronal protein synthesis 70 . Because BDNF activates eEF2 71 and stimulates the de novo synthesis of neuronal TyrRS 5 (Supplementary Fig. 3f and g), we tested if DA would also increase neuronal TyrRS levels. As expected, DA increased the protein levels of TyrRS ( Fig. 1f and Supplementary Fig. 3h), and the effects of DA were abrogated by rapamycin (rapa) (Fig. 1g). Unlike DA 69 , treatment with Tyr inhibited eEF2 (Fig. 1h). This data suggests that while Tyr decreases TyrRS protein levels, BDNF and DA increase it (Fig. 1i).

cis-RSV and trans-RSV have opposite effects on TyrRS levels in neurons.
We previously showed that trans-RSV competes with Tyr to bind TyrRS 36  increased TyrRS protein levels ( Fig. 2a and Supplementary Fig. 4a). cis-RSV rescued the effects of trans-RSV and Tyr-mediated decrease in TyrRS levels in a dose-dependent manner ( Fig. 2b and c). High concentration trans-RSV decreased PheRS, not PheRS levels, while low concentration cis-RSV increased their levels (Fig. 2d). Further, cis-RSV-mediated increase in TyrRS level was abrogated by adding cycloheximide (CHX) (Fig. 2e). Ser51 phosphorylation of eukaryotic initiation factor 2 alpha (p-eIF2) and Thr56 phosphorylation of eEF2 (p-eEF2) inhibit protein synthesis at the initiation and elongation steps, respectively. While cis-RSV triggered a transient increase in the levels of p-eIF2 (Supplementary Fig. 4b and c), trans-RSV sustained the inhibition of eIF2 (Supplementary Fig. 4b and c). Consistent with a previous report 81 , trans-RSV inhibited eEF2 (Fig. 2f), but in contrast, cis-RSV activated eEF2 through dephosphorylation ( Fig. 2f). Similarly, treatment with an eEF2K activator (nelfinavir) 82 depleted neuronal TyrRS which was rescued by cis-RSV ( Fig. 2g and Supplementary Fig. 4d). We previously showed that genetic reduction of eEF2K attenuates age-related memory deficits in mice 83 . Consistently, we found that TyrRS and PheRS were increased in the brain tissue samples of eEF2K +/mice ( Fig.   2h). To determine the mechanism of cis-RSV-mediated activation of eEF2, we tested if cis-and trans-RSV modulate the interaction of TyrRS with eEF2. We found that cis-RSV facilitated the interaction of eEF2 with TyrRS and protein phosphatase 2 (PP2A), whereas trans-RSV and Tyr decreased their interactions (Fig. 2i). These data suggest that cis-RSV and Tyr modulate the de novo synthesis of TyrRS at the elongation step, similar to BDNF 71 and DA 69 . Furthermore, similar suggesting that neuronal TyrRS is a potential target of multiple neurotoxic agents.

Pharmacological activation of protein synthesis increases TyrRS levels in neurons.
Phosphorylation of eIF2 modulates motor and cognitive function 84-88 , and neuronal survival 89 .
Pharmacological activation of mRNA translation using integrated stress response inhibitor (ISRIB) protects against age-related memory deficits 90 . Consistently, treatment with ISRIB (5-50nM) increased the protein levels of TyrRS and PheRS (Supplementary Fig. 5a) and rescued trans-RSV-mediated depletion of TyrRS (Supplementary Fig. 5b). Interestingly, ISRIB stimulated the dephosphorylation of eEF2 (Supplementary Fig. 5a), potentially mediated through increased levels of TyrRS. However, higher doses of ISRIB (250-500nM) decreased TyrRS levels ( Supplementary Fig. 5c), consistent with some reported adverse effects of ISRIB 91,92 . Because circadian-regulated Tyr 9,10 decreased neurite TyrRS (Fig. 1b), and sleep regulates synaptic protein synthesis 93 , we wondered if the synaptic protein level of TyrRS is circadian-regulated. Our reanalysis of the mouse circadian proteomic 93 and metabolomic 94 data showed that synaptic protein level of only TyrRS (among the aaRSs) is circadian-regulated and is inversely correlated with Tyr levels (Supplementary Fig. 5d). Further, re-analysis of the human metabolome 95,96 showed that sleep deprivation increases the serum levels of Tyr (Supplementary Fig. S5e). Collectively, these data suggest that Tyr is a potential endogenous modulator of synaptic and nuclear TyrRS.

Tyrosine induces oxidative DNA damage in neurons and cis-RSV protects against it.
Human aging and neurodegenerative diseases accumulate oxidative DNA damage-associated mutations in neurons through unknown mechanisms [40][41][42][43] . 3a and b). While cis-RSV rescued Tyr-induced accumulation of -H2AX, and 8-oxo-dG, trans-RSV itself caused the accumulation of -H2AX and 8-oxo-dG ( Fig. 3a and b). Tyr also decreased the levels of 8-oxoguanine-DNA glycosylase (OGG1) (Fig. 3c), analogous to the reported decrease in the levels of OGG1 in sleep-deprived humans 98 and in the AD brain tissues 99 .
cis-RSV rescued Tyr-mediated depletion of OGG1 (Supplementary Fig. 6a). 8-oxo-dG is highly mutagenic, driving a G . C A . T transversion 100 , and mutagenic frequency is reported to increase in aged neurons 40 and AD brain tissues 43 Fig. 6b and c). cis-RSV protected against D-Tyr-mediated neurotoxicity but not trans-RSV (Fig. 3f). Emerging works show a critical role of histone poly-ADP-ribosylation factor (HPF1) in DNA repair 105 . Although cis-RSV did not affect the HPF1 levels, treatment with Tyr and trans-RSV decreased the levels of HPF1 ( Fig. 3g and Supplementary Fig. 6d and e), indicating multiple mechanisms for Tyr-mediated accumulation of oxidative DNA damage in neurons. Consistently, we found that HPF1 is decreased in the hippocampal tissues from AD patients (Fig. 3h).

cis-RSV is neuroprotective and trans-RSV is neurotoxic.
Because cis-and trans-RSV have opposite effects on neuronal oxidative DNA damage, we hypothesized that they would have differential effects on neuronal survival under stress conditions.
We analyzed cis-and trans-RSV effects on the survival of rat primary cortical neurons exposed to different stress agents to test this hypothesis. As expected, the effect of trans-RSV on NMDAmediated neurotoxicity showed a concentration-dependent dual response, where low concentrations of trans-RSV (≤10M) evoked protective effects, but the higher concentrations (≥25M) exacerbated the toxicity (Fig. 4a). In contrast, cis-RSV protected against NMDAmediated toxicity in a concentration-dependent manner (Fig. 4b). Hence, the concentrationdependent dual response of trans-RSV on neuroprotection is consistent with trans to cis conversion at low concentrations 36,74 , increasing TyrRS levels (Fig. 2a), and the retention of 'trans/Tyr conformation' at high concentrations 36,39 , causing TyrRS depletion (Fig. 2a). We also found that cis-RSV (50M) suppressed the neurotoxicity induced by a DNA-damaging agent (etoposide, ETO) ( Supplementary Fig. 7a), oxidative stress (H2O2) (Supplementary Fig. 7b), and mitochondrial inhibition (MPP + ) (Supplementary Fig. 7c) but trans-RSV (50M) did not protect against these neurotoxic agents (Supplementary Fig. 7a-c). To test if the observed effects of cis-and trans-RSV on neurotoxicity are mediated via TyrRS, we carried out siRNA knockdown of TyrRS (siRNA TyrRS ) in rat cortical neurons. TyrRS knockdown (Supplementary Fig. 3d) blunted the neuroprotective effects of cis-RSV but did not diminish the toxicity of trans-RSV (50M) upon NMDA treatment (Fig. 4c). These results suggest that the neuroprotective effect of cis-RSV (and low-dose trans-RSV) is TyrRS dependent, but the neurotoxic effect of trans-RSV is TyrRS independent. Moreover, we found that trans-RSV (50M) by itself was neurotoxic in the rat primary cortical neuron cultures (Fig. 4c), whereas cis-RSV protected against the neurotoxicity induced by trans-RSV in a dose-dependent manner (Fig. 4d). Furthermore, a high concentration of trans-RSV and D-Tyr increased the levels of cleaved caspase-3 (a marker of apoptosis) in rat primary cortical neurons (Supplementary Fig. 7d and e). In contrast, cis-RSV decreased the levels of caspase-3 cleavage (Supplementary Fig. 7d). As expected, treatment with either ISRIB, eEF2K inhibitor or DA protected against neurotoxic effects of trans-RSV (Supplementary Fig.   7f-h), indicating a critical role of protein synthesis in neuronal survival.

cis-RSV and trans-RSV have opposite effects on the auto-PARylation of PARP1.
We previously showed that Tyr inhibits the auto-poly-ADP-ribose(PAR)ylation of PARP1 36 while cis-RSV induced a 'Tyr-free conformation' in TyrRS 39 to stimulate the auto-PARylation 36 .

cis-RSV stimulates the deADP-ribosylation of neuronal chromatin.
DNA damage activates PARP1/HPF1-dependent serine ADP-ribosylation of histones 108 , and removal of these 'ADP-ribose chromatin scar' is essential for neuronal survival 109,110 . Importantly, the brain samples of AD patients have increased levels of nuclear ADP-ribosylation 111 , suggesting a potential role of 'ADP-ribose chromatin scar' in AD-related neurodegeneration. Although auto-PARylation dissociates PARP1 from the chromatin 112 , HPF1 inhibits the auto-PARylation 113 , and increases trans-PARylation of chromatin 113,114 . Because trans-RSV inhibited the auto-PARylation ( Fig. 4f), we hypothesized that trans-RSV would increase PARP1-dependent trans-PARylation of the chromatin. As expected, trans-RSV increased the association of PARP1 with chromatin and increased the levels of PARylated proteins in the chromatin fraction (Fig. 5a). Further, low concentrations of trans-RSV (≤10M) and cis-RSV prevented the interaction of PARP1 with histone H3 while the higher concentrations (≥25M) of trans-RSV increased it ( Supplementary   Fig. 8a). Although cis-RSV-mediated auto-PARylation of PARP1 resulted in its removal from the chromatin, unexpectedly, we found that cis-RSV activated the deADP-ribosylation of the chromatin fraction along with higher levels of TyrRS ( Fig. 5a and Supplementary Fig. 8b). ADPribosyl-acceptor hydrolase 3 (ARH3) removes 'ADP-ribose chromatin scar' 109,110 , and as expected, cis-and trans-RSV had differential effects on the recruitment of ARH3 to the chromatin ( Fig. 5a and Supplementary Fig. 8c). Despite having increased levels of nuclear ADPribosylation 111 , the levels of ARH3 remained unchanged in the hippocampal region of human AD patients ( Supplementary Fig. 8d), indicating that ARH3 may not be functional in human AD brain tissues in the absence of TyrRS. Consistently, we found that TyrRS interacted with ARH3 ( Fig. 5b), suggesting a potential novel role of TyrRS in the removal of 'ADP-ribose chromatin scar' (Fig. 5a) that enhances neuronal DNA repair and survival.

'Trapped' PARP1 inhibits DNA repair and mediates the neurotoxic effects of trans-RSV.
Suicidal crosslinking of PARP1 to the damaged DNA 115 causes cytotoxicity 116 , and therefore, cell survival depends on removing 'trapped' PARP1 on the broken DNA by either ablation 38,117 or auto-PARylation 112,118,119 . Because trans-RSV caused DNA damage ( Fig. 3a and b) and inhibited the auto-PARylation of PARP1 (Fig. 4f), and ablation of PARP1 rescues 'trapped' PARP1-
Although there is no sex differences in Tyr levels in children 57 , whether increased levels of Tyr in children with ASD 156,157 or mutations of amino acid transporter (LAT1) that increase the levels of Tyr/Phe in the brain 153,154 contribute to the increased incidence of mutations 158 and dysregulated protein synthesis in ASD 154 will be of future research interest. Therefore, Tyrmediated induction of 8-oxo-dG and γH2AX (Fig. 3a-c), and Tyr-mediated depletion of TyrRS shown here may have causal effects in human aging, motor, cognitive, and metabolic disorders ( Supplementary Fig. 9).
Most recent clinical trials showed that low-dose trans-RSV (50mg/dose, 6M peak plasma concentration) that gets converted to cis-RSV 36,74 and induces a 'Tyr-free conformation' in TyrRS 39 protected against human heart failure 159 where high Tyr levels negatively correlate with survival 160,161 . In contrast, high-dose trans-RSV (500mg/dose, 68.5M peak plasma concentration 76 ) that maintains 'trans/Tyr conformation' in TyrRS and induces oxidative DNA damage ( Fig. 3a and b)              hr at room temperature, respectively. Immobilon ECL Ultra Western HRP Substrate (WBULS0500, Millipore) and a luminescent image analyzer (ChemiDoc Imaging System, Bio-Rad) were used to detect proteins. Quantification of western blots was done using ImageJ (Version 1.53c).

Use of Publicly Available Proteomics Data for the Analysis of TyrRS Levels in Human Brain
The proteomic data for TyrRS in human brain samples were obtained from the public databases as mentioned below. The graphical representation for biweight midcorrelation (BICOR) score of TyrRS protein level in the brain was created by retrieving and analyzing data from a large-scale proteomic database associated with a previously published work in Nat Med 26, 769-780 (2020) 48 . The published proteomic analysis 48  the motor cortex (MCx) and cerebellum (CER) were identified using mass spectrometry from donors (n = 9 AD cases, n = 9 asymptomatic controls). Statistical significance was determined using a global false discovery rate (FDR) threshold of 5%, i.e., the largest set of proteins with an average local FDR ≤ 5% were deemed significant.