Excessive copper impairs intrahepatocyte trafficking and secretion of selenoprotein P

Selenium homeostasis depends on hepatic biosynthesis of selenoprotein P (SELENOP) and SELENOP-mediated transport from the liver to e.g. the brain. In addition, the liver maintains copper homeostasis. Selenium and copper metabolism are inversely regulated, as increasing copper and decreasing selenium levels are observed in blood during aging and inflammation. Here we show that copper treatment increased intracellular selenium and SELENOP in hepatocytes and decreased extracellular SELENOP levels. Hepatic accumulation of copper is a characteristic of Wilson’s disease. Accordingly, SELENOP levels were low in serum of Wilson’s disease patients and Wilson’s rats. Mechanistically, drugs targeting protein transport in the Golgi complex mimicked some of the effects observed, indicating a disrupting effect of excessive copper on intracellular SELENOP transport resulting in its accumulation in the late Golgi. Our data suggest that hepatic copper levels determine SELENOP release from the liver and may affect selenium transport to peripheral organs such as the brain.

Selenium homeostasis depends on hepatic biosynthesis of selenoprotein P (SELENOP) and SELENOP-mediated transport from the liver to e.g. the brain. In addition, the liver maintains copper homeostasis. Selenium and copper metabolism are inversely regulated, as increasing copper and decreasing selenium levels are observed in blood during aging and inflammation. Here we show that copper treatment increased intracellular selenium and SELENOP in hepatocytes and decreased extracellular SELENOP levels. Hepatic accumulation of copper is a characteristic of Wilson's disease. Accordingly, SELENOP levels were low in serum of Wilson's disease patients and Wilson's rats. Mechanistically, drugs targeting protein transport in the Golgi complex mimicked some of the effects observed, indicating a disrupting effect of excessive copper on intracellular SELENOP transport resulting in its accumulation in the late Golgi. Our data suggest that hepatic copper levels determine SELENOP release from the liver and may affect selenium transport to peripheral organs such as the brain.
Selenium is an essential nutrient and trace element that is of vital importance for human health. Twenty-five human genes encode for selenoproteins, all of which contain selenocysteine (Sec) residues at their active sites which are catalyzing redox reactions 1 . Among these, selenoprotein P (SELENOP) is the most important one for distributing selenium from the liver to organs such as the brain and testes to maintain their selenium supply and local selenoprotein expression [2][3][4] . Approximately 60% of the total selenium concentration in serum is bound to SELENOP. Upon loss of SELENOP, mice develop severe seizures and ataxia due to selenium deficiency in the brain 5,6 . Receptors of the LRP (low-density lipoprotein receptor-related protein) family including LRP8 (APOER2) are mediating SELENOP endocytosis 7 . Unlike other selenoproteins that carry single Sec residues, SELENOP contains up to ten, one at its N-terminal domain with redox activity and nine at its C-terminal domain which is cleaved off to release selenium within target cells 8 . While uptake of SELENOP by target tissues has been extensively studied, less attention has been paid to the release of SELENOP from hepatocytes. Ingested selenium is first transported to the liver and either used for local selenoprotein expression or for synthesis of SELENOP for further distribution within the body. SELENOP transcription is mediated by many factors including inflammatory mediators 9,10 . SELENOP is a glycoprotein which is further processed by both N-and O-glycosylation 11 . Newly synthesized glycoproteins destined for export such as SELENOP are translocated into the endoplasmic reticulum (ER) for early processing and quality control before further glycosylation and excretion via the Golgi apparatus takes place 12 .
Besides selenium, the liver is also the main organ for maintaining homeostasis of additional trace elements such as copper. The ATP-driven transporter ATP7B (ATPase copper transporting β) is responsible for balancing hepatic and systemic copper levels. Under basal conditions, copper is shuttled by antioxidant 1 copper chaperone (ATOX1) to the trans-Golgi and bound to ATP7B to metallate newly synthesized ceruloplasmin (CP), the major copper containing protein in serum. Holo-CP, which binds up to six copper atoms, operates as a carrier that distributes copper to peripheral organs 13 . Excess copper results in ATP7B translocation from the Golgi to endo-lysosomal compartments to store copper or to release it into the bile 14,15 . We have previously shown that copper inhibits hepatic selenoenzyme activity of the glutathione peroxidase (GPX) and thioredoxin reductase (TXNRD) families both in vitro and in vivo 16 . Thus, an adequate selenium supply could be limited on the functional level of selenoenzyme activity by copper. Circulating selenium and copper concentrations are negatively correlated, e.g. during aging copper levels are increasing while selenium levels are decreasing 17 . During inflammation, the same picture emerges as CP is an acute phase protein increasing in the circulation while SELENOP is a negative acute phase reactant declining in the circulation 10,18 . SELENOP is a candidate target protein for interference of copper with selenium homeostasis as it represents the main route of selenium excretion from the liver. Based on this, we hypothesized a direct effect of copper on SELENOP excretion resulting in the inverse regulation of the trace elements in hepatocytes and blood. Accordingly, we aimed to study whether copper excess affects intracellular selenium levels, and SELENOP distribution and excretion of cultured hepatocytes and whether the findings are compatible with data from genetic models of ATP7B mutations in rats and Wilson's disease patients.
Here we show that copper treatment increases intracellular selenium and decreases extracellular SELENOP levels using the liver-derived cancer cell line HepG2. The same effect is observed in cultured primary murine hepatocytes and supported by an untargeted secretome approach in which SELENOP is particularly downregulated by copper. Accumulation of copper in liver is a characteristic of Wilson's disease and LPP rats, which serve as a disease model. Accordingly, SELENOP levels are relatively low in serum of patients with Wilson's disease and LPP rats, especially in those with low CP concentrations. The CP promoter contains the SNP rs11708215, which is associated with increased serum copper and CP concentrations. An analysis of a large cohort study (EPIC-Potsdam) shows a congruent positive correlation of SNP rs11708215 (coded per G allele) with serum SELENOP levels. Overall, the results indicate that genetic variations modulating hepatic copper concentrations cause changes in the amount of circulating SELENOP which most probably affects the selenium distribution to peripheral organs such as the central nervous system and endocrine glands.

Copper blocks SELENOP excretion from HepG2 cells
To study the cross-talk of the trace elements selenium and copper, HepG2 cells were cultured for up to 72 h with CuSO 4 or Cu(His) 3 in combination with selenite or selenomethionine (SeMet). None of the chosen copper concentrations in combination with or without 50 nM selenite showed cytotoxic effects ( Supplementary Fig. 1a) 16 . The copper treatment with CuSO 4 (Fig. 1a) or Cu(His) 3 ( Supplementary Fig. 1b) resulted in a more than 20-fold increase of cellular copper levels which was independent of the selenium supply of the cells. Both selenite and SeMet treatment upregulated the cellular selenium concentration by a factor of 4 ( Fig. 1b, S1c). In previous studies, we identified an overall reduction of enzymatic activity of selenoproteins by copper 16 . Unexpectedly and in contrast to the reduced selenoprotein activity, selenium levels were increased in copper-treated cells by a factor of 1.5 (Fig. 1b, S1c). This effect was independent of the copper and selenium source. For further experiments, mainly selenite and CuSO 4 were used as selenium and copper sources, respectively.
HepG2 cells excrete SELENOP into the medium. A 24 h incubation with increasing copper concentrations up to 500 µM, resulted in a concentration-dependent decrease of SELENOP excretion independent of the selenium status (Fig. 1c). Also upon copper treatment for 72 h, extracellular SELENOP concentrations were reduced (Fig. 1d, upper Western Blot; Supplementary Fig. 1d) as indicated by the copper-induced fold changes of 0.3-0.7 (Fig. 1e, white bars). In parallel, SELENOP accumulated within copper-treated cells (Fig. 1d, lower Western Blot) which was quantified as a fold change of 1.3 and 1.7 for selenite-and SeMet-treated cells, respectively (Fig. 1e). This effect was much more pronounced for fully glycosylated SELENOP with a molecular weight of~65 kDa represented by the upper band in Western Blot analyses (Fig. 1d, blue box, upper band) which was quantified as fold change of 2 and 2.7 for selenite-and SeMet-treated cells, respectively (Fig. 1e, S1e). In contrast to the intracellular SELENOP accumulation, qPCR analyses revealed a mild downregulation of SELENOP mRNA while LRP8 mRNA was slightly upregulated by copper (Fig. 1f). However, protein concentrations of the SELENOP transporter LRP8 in the membrane/organelle fraction were not modulated by copper treatment ( Supplementary Fig. 1f).
To exclude that the observed effects are restricted to HepG2 cells, a 24 h ex vivo copper treatment of primary murine hepatocytes was performed resulting in intracellular copper accumulation and reduced extracellular SELENOP concentrations compared to cells without copper treatment (Fig. 1g, h). When using colon-derived HT-29 cells which do not excrete measurable amounts of SELENOP, copper treatment and concomitantly higher intracellular copper levels did not increase but even decrease intracellular selenium concentrations ( Supplementary Fig. 1g, h) indicating that the copper effect on selenium accumulation depends on SELENOP and its accumulation. To further test this hypothesis, an siRNA-mediated knockdown of SELE-NOP was generated in HepG2 cells. Both extracellular and intracellular SELENOP levels were substantially reduced in SELENOP knockdown cells (Fig. 1i, j). In addition, neither a copper-induced effect on extra-or intracellular SELENOP (Fig. 1i, j) nor on cellular selenium concentrations ( Fig. 1k) was detectable upon knocking down SELENOP. Downregulation of SELENOP did not change the cytotoxicity of copper levels up to 1 mM in comparison to cells expressing SELENOP (Fig. 1l). Overall, the data indicate that copper increased the intracellular selenium content by blocking SELENOP excretion which appears to be specific for liver cells.
To characterize the time-course of the copper-related inhibition of SELENOP excretion, the incubation time with selenium and copper was reduced to 6 h. Already after 6 h of treatment, measurable amounts of SELENOP were excreted into the medium which were reduced by copper (Fig. 2a, b). In parallel, selenium started to accumulate inside the cells (Fig. 2c). To reverse the copper-induced effects on SELENOP excretion, we treated the cells 24 h prior to harvest with the two copper-specific chelators BCS and TTM 16 . This treatment partially rescued the inhibition of SELENOP excretion by copper (Fig. 2d). However, chelator treatment only without additional copper was not able to increase SELENOP excretion. To further test for the specificity of the inhibition of SELENOP excretion by copper, additional cations, namely zinc and iron, were used. In contrast to copper, zinc treatment reduced both intracellular and extracellular SELENOP concentrations indicating that zinc might inhibit SELENOP expression (Fig. 2e, S2a). This effect was confirmed on mRNA level, where downregulation of SELENOP and upregulation of LRP8 was even more pronounced after zinc than after copper treatment ( Supplementary  Fig. 2d). Fold changes were still relatively small (0.6 for SELENOP and 1.4 for LRP8) in comparison to the strong upregulation of metallothionein MT2A mRNA expression in response to zinc but also to copper with fold changes of 90 and 20, respectively ( Supplementary  Fig. 2e). Iron increased intracellular SELENOP protein concentrations, but did not affect mRNA levels, intracellular selenium or extracellular SELENOP levels indicating that iron might affect cellular turnover of SELENOP (Fig. 2e, Supplementary Fig. 2b). As all three cations are able to modulate the cellular redox state towards oxidation, we further treated cells with H 2 O 2 to mimic this effect. As observed for zinc, a decrease of both extra-and intracellular SELENOP was detected in H 2 O 2 treated cells (Fig. 2e, Supplementary Fig. 2c). Thus, the identified inhibitory effect on SELENOP excretion was specifically observed for copper only and does not appear to be driven by a general modulation of the cellular redox status.

Hepatic SELENOP excretion is suppressed by copper in vivo
Based on the cell culture data, we aimed to test for in vivo relevance of our results. Accordingly, we were wondering whether a hepatic copper accumulation which was studied in cell culture would affect SELENOP excretion. Hepatic copper accumulation takes place in Wilson's disease caused by mutations of ATP7B. The LPP rat is a Wilson animal model with genetic ablation of ATP7B and growing rats develop different stages of Wilson's disease (classified as affected, disease onset, and diseased) over time, as described earlier 19,20 . Already at the early, affected stage, copper strongly accumulated in the liver of homozygous rats which was maintained in all disease stages (Fig. 3a). The two early stages were characterized by lower plasma copper concentrations (Fig. 3b) and very low CP oxidase activity in plasma (Fig. 3c) in comparison to heterozygous control rats. In the diseased stage, circulating copper concentrations increased again back to levels of control rats due to damage of liver tissue (Fig. 3b) while CP concentrations stayed low (Fig. 3c). Selenium and SELENOP levels were downregulated during the course of the disease (Fig. 3d, e). Plasma levels of CP and SELENOP were positively correlated (Pearson's r = 0.522; p-value = 0.011). The treatment of diseased LPP rats with the copper chelator methanobactin reduced circulating copper concentrations ( Fig. 3f) as expected 20 and accordingly resulted in higher circulating selenium and SELENOP concentrations (Fig. 3g, h). Based on these in vivo results combined with the in vitro studies, we can clearly conclude that copper interferes with selenium homeostasis and hepatic SELENOP excretion.

Relationship of copper and selenium serum biomarkers in humans
We next analyzed the selenium status in serum samples of Wilson's disease patients at the time point of diagnosis before the onset of therapeutic interventions. Overall, the clinical parameters of the patients were very heterogeneous, e.g., a very broad range of serum copper concentrations was observed. Based on their serum CP, the patients were divided into two groups with normal (>90 µg/L) or low (<90 µg/L) serum CP concentrations (Fig. 4a) which was confirmed by measuring the functional holo-CP reflected by CP oxidase activity (Fig. 4b). Low CP is one of the parameters used for the diagnosis of Wilson's disease 21 which most probably indicates a higher hepatic copper accumulation as shown for the rat Wilson model (Fig. 3). In addition, a third group ('diseased') was established for those patients with high clinical serum parameters such as C-reactive protein (CRP) and hepatic transaminases indicating already a more severe disease state resulting in copper leakage from destroyed hepatocytes (for patient's characteristics see table S1). To further characterize the patients' copper status, total and free copper were analyzed. In line with the CP concentrations, total copper levels were reduced in the 'low CP' group in comparison to the 'normal CP' group ( Fig. 4c). Free copper levels confirmed the division of patients into the 'diseased' group with higher concentrations in comparison to the two other groups (Fig. 4d). The SELENOP concentration was significantly decreased in the serum of patients with low serum copper and CP concentrations in comparison to the 'normal CP' group but was heterogeneous within the 'diseased' group ( Fig. 4e). The overall serum selenium concentration showed the same picture as for SELENOP but was also reduced in the 'diseased' group ( Fig. 4f). CP is the main transport protein for copper from the liver to peripheral organs, meaning that low circulating CP levels would indicate that copper is retained in the liver. The SNP rs11708215 is located in the promoter region of the CP gene and has been described to enhance hepatic CP expression and CP secretion 22 which would as a consequence result in more efficient copper export from the liver and a concomitant decrease in hepatic copper levels. The analysis of EPIC-Potsdam samples revealed that the SNP rs11708215 (coded per G allele) was positively correlated with serum levels of copper 23 but also with SELENOP ( Fig. 4g). This indicates that more SELENOP can be excreted from the liver when hepatic copper levels are low.

Copper modulates the secretome of HepG2 cells and strongly interferes with SELENOP release
To explore the mechanisms behind copper-induced effects on blocking SELENOP excretion we used the HepG2 cell model. The overall effects of copper treatment on the secretome of HepG2 cells were analyzed by quantitative LC-MS/MS. The analysis led to the identification of 1,179 proteins in the medium of HepG2 cells. Gene ontology (GO) enrichment analyses revealed that downregulated proteins were involved in glycosaminoglycan, heparin, and sulfur compound binding and possess oxidoreductase activity ( Supplementary Fig. 3a). Upregulated proteins also had oxidoreductase and hydrolytic activity, were involved in carbohydrate binding and belonged to the peroxiredoxins. The GO enrichment for cellular compartments confirmed the reliability of the results mainly representing proteins located to exosomes or vesicles ( Supplementary Fig. 3b) including SELENOP. Copper treatment significantly upregulated 195 and downregulated 75 proteins which together accounted for 23% of the detected proteins in the secretome. SELENOP was one of the most significantly and consistently downregulated proteins in response to copper treatment (Fig. 5a). Overall, the secretome analysis showed that there was no general inhibition of secretory processes by copper. Only 6.4% of the whole secretome were downregulated in the extracellular space in response to copper. One of the upregulated proteins was CP, the main extracellular copper binding protein 13 . The release of α1-antitrypsin (AAT, SerpinA1) a highly abundant liver-derived serum protein 24 was unaffected by copper ( Supplementary Fig. 3c, Supplementary Data 1) in agreement with the majority of secreted proteins in this analysis. Next to SELENOP, proteins of the apolipoprotein family (APOB, E, and H) were consistently downregulated as well. For APOE, we could confirm an early extracellular decrease after 6 h of copper treatment (Fig. 5b) and a concomitant intracellular APOE accumulation (Fig. 5c) showing the same pattern for APOE as observed for SELENOP. APOE has been described to bind to SELENOP and to block its excretion 25 . Thus, the copper-induced intracellular APOE accumulation may were measured using TXRF with 1 mg/L yttrium as standard element for 300 s and normalized to protein content or with 1 mg/L gallium as standard element for 1000 s, respectively. c The CP oxidase (CPO) activity in plasma of rats was measured photometrically. e, h Plasma SELENOP was analyzed by Western Blot and normalized to Ponceau (P) staining (n = 5-8). Data are depicted as mean ± SD. Biological replicates are indicated by individual dots. Statistical analyses were based on one-way ANOVA with Bonferroni's post-test *p < 0.05; **p < 0.01; ***p < 0.001 vs. ATP7B +/control rats (a-e) or by two-tailed t test compared to untreated rats (f-h). Source data are provided as a source Data file.
contribute to the reduced SELENOP excretion. Based on this, we generated an APOE knockdown in HepG2 cells reducing intracellular APOE levels to 60% of controls (Fig. 5d). SELENOP excretion was substantially increased upon APOE knockdown under basal conditions but was still reduced by copper treatment (Fig. 5e).

SELENOP accumulates in the late Golgi in response to copper treatment
As secretory proteins are processed in the Golgi, we used the inhibitors brefeldin A and monensin to interfere with intracellular protein shuttling. Brefeldin A blocks the transport of secretory proteins from the ER to the Golgi complex, and thus completely abolishes Golgi-resident glycosylation processes 26 . Accordingly, no distinct glycosylation of intracellular SELENOP was detectable in brefeldin A-treated cells (Fig. 6a). There was no copper-dependent effect on intracellular SELENOP or selenium levels after brefeldin A treatment (Supplementary Fig. 4a, b). Next, monensin was used to inhibit protein transport from the medial to the trans Golgi complex 27 . Monensin-treated cells showed strong intracellular SELENOP accumulation which, however, was restricted to the two lower bands of SELENOP (Fig. 6b). The fully glycosylated form of SELENOP with a size of 65 kDa which was most sensitive to copper treatment (Fig. 1d, e) was not detectable after , and selenium (f) concentrations of Wilson's patients measured using photometric activity, TXRF with 1 mg/L gallium as standard element for 1000 s or a fluorometric method (normal CP n = 8; low CP n = 11; diseased n = 4). Data are depicted as mean ± SD. Biological replicates are indicated by individual dots. Statistical analyses were based on one-way ANOVA with Bonferroni's post-test. *p < 0.05; **p < 0.01; ***p < 0.001 vs. 'normal CP', # p < 0.05; ## p < 0.01 vs. 'low CP ' . g Results from EPIC-Potsdam samples for SNP rs11708215 (coded per G allele) and its correlation with serum SELENOP from linear regression using an additive genetic model adjusted for age at recruitment and sex. Serum SELENOP levels were natural log-transformed and standardized. Source data are provided as a source Data file. monensin treatment indicating that this glycosylation step takes place in the trans or late Golgi. In line with this, copper treatment did not increase intracellular SELENOP but even decreased SELENOP and selenium levels in monensin-treated cells ( Supplementary Fig. 4c, d).
To further specify the compartment where the copper-sensitive glycosylation of SELENOP takes place, deglycosylation experiments using PNGase F and EndoH were performed 28 . PNGase F digestion cleaves off all N-glycans and accordingly results in the deglycosylation of all three SELENOP forms (Fig. 6c). A comparable lack of glycosylation was observed when treating cells with tunicamycin which inhibits N-linked glycosylation 29 and accordingly also substantially reduced SELENOP glycosylation ( Supplementary Fig. 4e). In tunicamycin-treated cells, copper reduced intracellular SELENOP levels ( Supplementary Fig. 4e) as observed in monensin-treated cells. In contrast, EndoH more specifically cleaves off early Golgi N-linked sugar chains, while late Golgi N-linked sugars are EndoH resistant. The copper-sensitive glycosylation of SELENOP turned out to be EndoH resistant (Fig. 6d) and accordingly appears to be established in the late Golgi (Fig. 6e). Cellular fractionation experiments revealed that the Cu-induced SELENOP accumulation was undetectable in the cytosol but was enriched in the membrane/organelle fraction which also contained GOLGIN97, an established marker for the trans Golgi network (TGN) (Fig. 6f). This accumulation was detectable already after 1 h of copper treatment and increased over time (Supplementary Fig. 4f). In parallel with SELENOP, glycosylated APOE increased in the membrane/organelle fraction (Fig. 6f, Supplementary Fig. 4g). Immune fluorescence experiments also identified a copper-induced increase of SELENOP in the perinuclear region showing a co-localization with GOLGIN97 (Fig. 6g, h). These results indicate that copper interferes with the late Golgi cargo process resulting in accumulation of SELENOP.

Discussion
We provide herein data from cell culture, rodent models, and Wilson's disease patients indicating strong evidence for a connection between copper and selenium homeostasis. We have consistently shown that hepatic copper accumulation reduces SELENOP excretion from hepatocytes which represents about 60% of circulating SELENOP 2 . Copper treatment resulted in lower SELENOP concentrations in cell culture media as well as in the circulation of whole organisms (Fig. 7). The untargeted secretome approach performed in the culture medium of HepG2 cells identified SELENOP as one of the most strongly downregulated proteins in response to copper (Fig. 5a). The comparison of the secretome of HepG2 cells with that of primary human hepatocytes indicated a high overlap 24 . Among the secreted proteins several previously described hepatokines including SELENOP have been detected 24 . Also in murine hepatocytes, SELENOP release was reduced by copper (Fig. 1h). SELENOP has an essential role as selenium distributing protein to organs such as brain and testes [2][3][4] . Lowering circulating SELENOP concentrations has major consequences for the selenium supply of these peripheral organs which is most important in populations with a suboptimal selenium supply, which also prevails in the European population 30 . As Wilson's disease is caused by hepatic copper accumulation, we addressed the question whether the selenium status of these patients is impaired. We observed lower amounts of circulating SELENOP especially in patients with low serum CP concentrations which is an indicator for a more pronounced hepatic copper accumulation (Fig. 4e). So far, SELENOP concentrations have not been analyzed in serum samples of Wilson's patients. Only one study characterized the selenium status of Wilson's patients and found lower circulating selenium concentrations in male patients fitting to our SELENOP and selenium results being reduced in patients with lower circulating CP concentrations (Fig. 4e, f). This is also in line with the data obtained from the LPP rat model of Wilson's disease. In rats of the disease onset and disease groups, we observed a decline of both circulating selenium and SELENOP concentrations (Fig. 3d, e). Interestingly, a tendency for lower selenium concentrations was also observed in Wilson's patients with dominating neuropsychiatric symptoms 31 , which could hint for reduced selenium transport to the brain via SELENOP. It has been well described that SELENOP knockout mice suffer from severe seizures and ataxia due to selenium deficiency in the brain 5,6 . Based on this, it can be speculated that reduced selenium levels in the brain contribute to the development of neurological symptoms in Wilson's patients. In addition to Wilson's disease, cholestatic diseases such as primary biliary cholangitis, and primary sclerosing cholangitis (PSC) are accompanied by hepatic copper accumulation 32 . For example, in PSC hepatic selenium concentrations increase while serum selenium levels decrease 33 , fitting to what we observed in Wilson's patients. Under conditions of Wilson's disease, ATP7B-deficient hepatocytes lack the most effective defence mechanism against copper which is efficient copper export and therefore they have to rely on alternative mechanisms to survive copper overload that includes upregulation of copper binding proteins such as metallothionein or glutathione, allowing the cells to buffer increasing copper levels 34 . Also in our cell culture model, copper treatment resulted in transcriptional upregulation of metallothionein which was, however, less pronounced as compared to zinc-treated cells (Supplementary Fig. 2e). A further increase of copper exceeding the buffer capacity results in oxidative stress. SELENOP has been described to have antioxidant properties 8 and, thus, SELENOP accumulation could be an attempt to combat Cuinduced oxidative stress which would explain the necessity to stop its excretion under conditions of copper overload. We neither observed a difference in copper-induced cytotoxicity nor in total cellular copper concentrations upon SELENOP knockdown (Fig. 1l) indicating that SELENOP is not essential for limiting copper-induced oxidative stress in HepG2 cells. In line with this, SELENOP knockout mice do not show any difference in overall hepatic copper levels in comparison to wild type mice 35 . In LPP rats, oxidative stress is only a characteristic of the liver of the 'diseased' group and, thus, does not correlate with the reduced SELENOP excretion which was already observed in rats of the 'disease onset' group (Fig. 3e). Also the chosen cell culture conditions with a copper supply of 100 µM for up to 72 h do not result in oxidative stress but only in a very mild Nrf2 activation 16 . This Nrf2 activation appears to result in the observed transcriptional changes with a copper-but also a zinc-induced downregulation of SELENOP and an upregulation of LRP8 mRNA (Fig. 1f, Supplementary Fig. 2d) which both have been described to be mediated by Nrf2 36,37 . Accordingly, the accumulation of both intracellular selenium and SELENOP was not observed in cells treated with other redox-active metals such as iron or zinc or by treating cells with different concentrations of H 2 O 2 (Fig. 2e) pointing towards a copper-specific mechanism which goes beyond a mere oxidative stress-mediated effect. Copper specificity was further supported by the reversal of the copper-induced SELENOP accumulation by co-treating with the copper chelators BCS or TTM (Fig. 2d) in HepG2 cells and by methanobactin in LPP rats (Fig. 3f-h). Thus, the copper-mediated SELENOP accumulation does not appear to be mainly driven by redox-dependent processes.
Prior to excretion, SELENOP is processed in the ER/Golginetwork 12 . The SELENOP molecule contains heparin binding regions, two histidine-rich domains and several N-and O-glycosylation sites which are post-translationally modulated 11,38 . The post-translational modifications are supposed to protect the Sec residues within SELE-NOP by reducing their reactivity. However, they are not modulating the amount of SELENOP excreted from HepG2 cells 12 . In contrast, heparin binding to SELENOP increased its excretion while APOE binding to the heparin binding regions reduced SELENOP excretion from hepatocytes. Accordingly, circulating APOE bound to SELENOP is involved in enhancing the binding of SELENOP to LRP8 on the surface of target cells 25 . While membrane localization of LRP8 was unaffected by copper ( Supplementary Fig. 1f), we observed lower extracellular and higher intracellular APOE concentrations in copper-treated cells (Fig. 5a-c) which indicates an intracellular APOE accumulation. Copper has been well described to modulate lipid metabolism particularly cholesterol biosynthesis 34 . Copper can bind to APOE 39  on the APOE genotype, expression of metallothionein and, thus, intracellular copper binding capacity is modulated 40 . In addition, disruption of copper delivery to mitochondria increases APOE expression and secretion 41 . An APOE knockdown in HepG2 cells enhanced SELE-NOP release into the medium (Fig. 5e) which has been also shown before in the serum of APOE knockout mice 25 . Thus, APOE accumulation and binding to SELENOP provides a mechanistic link how copper could interfere with SELENOP excretion. The ER/Golgi-network is also crucial for maintaining copper homeostasis as trafficking between membrane compartments is key to balance copper uptake and excretion 42 . ATOX1 ferries copper to ATP7A and ATP7B in the TGN for the metallation of newly synthesized copper-containing enzymes. Interestingly, this is the compartment in which SELENOP accumulates in response to copper treatment (Fig. 6f-h). But under conditions of copper exposure ATP7A/B translocate to post-Golgi membrane vesicles to either enhance copper efflux or to transiently store excess copper 43 . These processes are specific for hepatocytes because only in these cells a biliary copper secretion is possible. We did not observe a copper-induced selenium accumulation in intestinal cells (HT-29, Supplementary  Fig. 1h) but rather a downregulation which is also the case in neurons (differentiated LUHMES cells) and partially in astrocytes (CCF-STTG1 cells) 44 . This indicates that the liver-specific shuttling of copper appears to be important for mediating SELENOP accumulation. Hepatic copper accumulation in Wilson's patients has been suggested to impair protein synthesis and general protein secretion indicated e.g., by the reduced release of functional coagulation factors 45 . Herein, no clear picture emerged regarding secretion of coagulation factors by HepG2 cells as some were less secreted but most of them were unaffected by copper (Supplementary Data 1). Based on our secretome data, we can exclude a general reduction of released proteins because the major part of the secreted proteins was unaffected by copper (including e.g., α1-antitrypsin; Supplementary Fig. 3c) and only 6.4% of the whole secretome were downregulated by copper. But obviously, there is an enrichment of glycoproteins in the group of proteins less secreted upon copper treatment (53% of the downregulated proteins are glycoproteins while only 40% of the upregulated proteins belong to this group; categorized based on GlyCosmos, https://glycosmos. org/glycoproteins/), which includes SELENOP and APOE 11,46 . Based on the results from the PNGase and EndoH digestion of lysates (Fig. 6c, d), it can be concluded that the copper-sensitive full glycosylation of SELENOP takes place in the late Golgi (Fig. 6e). It has been suggested that SELENOP glycosylation might function to prevent proteolysis of SELENOP, thus, stabilizing intracellular levels 47 . We observed no copper-induced intracellular SELENOP accumulation in cells treated with the glycosylation inhibitor tunicamycin (Supplementary Fig. 4e), indicating that this glycosylation is important for the observed copperinduced effects. Several proteins of the conserved oligomeric Golgi (COG) complex have been shown to interact with ATP7A/B indicating that Golgi homeostasis is modulated in response to the copper status. The association of ATP7 paralogs with the COG complex remained after challenging cells with copper 48,49 . But so far, no mechanism has been identified that explains how copper modulates Golgi shuttling and cargo sorting within cells. Just recently, it was proposed that copper-induced effects on mitochondria could be the main driver for the observed differences in the cellular secretome 50 . Accordingly, it appears most likely that hepatic SELENOP accumulation is driven by copper-induced modulation of intra-Golgi transfer and/or glycosylation processes resulting in SELENOP accumulation in the Golgi.
From the Wilson's disease samples (rat and human), we can conclude that a low serum copper concentration is positively correlated with low circulating SELENOP levels (humans: Pearson's r = 0.425; pvalue = 0.07) which is supposed to be caused by hepatic copper accumulation. Another proxy for circulating copper is CP which is mainly secreted from liver and represents 80-95% of circulating copper in healthy individuals. The reduction of circulating holo-CP concentrations is one of the early symptoms of Wilson's disease caused by the impaired transfer of copper via ATP7B to CP and concomitant hepatic copper accumulation 51 . In the LPP rat model, serum levels of CP oxidase activity and SELENOP were positively correlated (Pearson's r = 0.522; p-value = 0.011). Several genetic polymorphisms have been described in the CP gene. One of those, SNP rs11708215 (minor allele frequency for GG: 0.19) located in the CP gene promoter was strongly associated with increased levels of circulating CP (p = 9 × 10 −10 ) as compared to the AA genotype. The results were replicated in an independent cohort in which the relationships remained significant also after adjusting for potential covariates including cholesterol levels in blood 22 . In line with this, the analysis of EPIC-Potsdam samples revealed that serum levels of SELENOP were correlated with SNP rs11708215 (Fig. 4g) which has been shown to positively correlate with serum copper 23 and CP concentrations 22 . The effects of the genetic polymorphisms of CP on hepatic copper needs to be studied further but points towards a genetic component modulating the relationship between hepatic copper and SELENOP release (Fig. 7). Additional conditions resulting in the upregulation of CP expression and downregulation of hepatic copper concentrations are non-alcoholic fatty liver disease (NAFLD) and overnutrition 52 . In contrast to the described steady-state conditions, the secretome analyses herein identified a negative correlation of extracellular CP and SELENOP because of increased CP and decreased SELENOP release (Fig. 5a). Interestingly, additional negative acute phase proteins including transthyretin (FC of 0.72), thyroxine-binding protein (FC of 0.76), and alpha-fetoprotein (FC of 0.65) reacted in the same manner as SELENOP. Also, HepG2 cells treated with IL-6 start to enhance CP release while dampening extracellular SELENOP levels 10 . Thus, dynamics of adaptation processes regulating both copper and selenium homeostasis appear to be connected and modulated depending on the health status but also the genetic background of individuals.

Methods
All research in this study complies with all relevant ethical regulations. Animal procedures were approved in case of mice by the Ministry of Environment, Health and Consumer Protection of the federal state of Brandenburg, Germany and conducted following national guidelines and institutional guidelines of the German Institute of Human Nutrition Potsdam-Rehbruecke, Germany. Animal experiments with LPP rats were approved by the government authorities of the Regierung von Oberbayern, Munich, Germany, and all animals were treated according to the guidelines for the care and use of laboratory animals of the Helmholtz Center Munich. Studies on Wilson's patients were approved by the Medical Ethics Committee of the University Hospital Heidelberg, Germany (Reference no. S-565/2011; DRKS00031527) and carried out in accordance with the declaration of Helsinki.

Cell culture
Cell culture experiments with hepatocarcinoma-derived HepG2 from a male donor (ACC 180 German Collection of Microorganisms and Cell Cultures (DSMZ)) and adenocarcinoma-derived HT-29 cells from a female donor (ACC 299 DSMZ) were performed as previously described 16 with minor modifications. In addition to previously described experiments Cu(His) 3 (CuSO 4 (Merck, Darmstadt, Germany; 451657) and L-histidine (Merck; H6034) 1:3; pH 7.4) was used. The two different inhibitors of Golgi-ER shuttling, brefeldin A (0.1 µg/ml; Biomol, Hamburg, Germany; Cay11861) or monensin (0.5 µM; Sigma-Aldrich/Merck; M5273) and the inhibitor for protein glycosylation tunicamycin (1 µg/ml; Abcam, Cambridge, UK; ab120296) were added 24 h prior to harvest of the cells. ZnSO 4 (Merck; 1088830500) and FeCl 2 (Merck; 1038610250) were used in combination with selenite (Na 2 SeO 3 ; Thermo Fisher Scientific; 15670510) for up to 72 h. Hydrogen peroxide (Merck; 1072090250) was added to selenium-supplied cells for 6 h in phenol red-free RPMI (Thermo Fisher Scientific, Waltham, USA; 11835030) to induce oxidative stress. For secretome analyses, HepG2 cells were cultured for 48 h in Opti-MEM (Thermo Fisher Scientific; 11058-021) supplemented with 50 nM selenite. After indicated incubation times, media were collected and cells were harvested. Medium samples and cell pellets were frozen in liquid nitrogen and stored at −20°C until further procedure.

siRNA-mediated knockdown in HepG2 cells
HepG2 cells were transfected with either SEPP1 silencer select predesigned siRNA (Ambion, Thermo Fisher Scientific, s12698), APOE silencer select pre-designed siRNA (Ambion, s536402) or silencer select negative control #1 siRNA (Ambion) as described by the manufacturer. Briefly, the transfection reagent lipofectamine 2000 (Thermo Fisher Scientific; 11668019) and the siRNAs were diluted in Opti-MEM (Thermo Fisher Scientific; 11058-021). The mixture of prediluted siRNA and lipofectamine was incubated at room temperature for 20 min and was added to the cells seeded in RPMI medium (Thermo Fisher Scientific; 21875091) without antibiotics at a final concentration of 1.5 µg or 5 pmol for lipofectamine and siRNA, respectively. Transfection was carried out for 18 h, before treatment with copper and/or selenium took place as described for additional 72 h. For cytotoxicity assays, transfected cells were pre-treated with selenite for 24 h and co-treated with selenite and increasing copper concentrations for additional 24 h.

Cytotoxicity assay
For the MTT assay, cells were seeded in 96-well plates. After up to 72 h of incubation with the trace elements, the medium was discarded and medium without FCS containing 0.5 mg/ml thiazolyl blue tetrazolium bromide (MTT; Merck; M2128) was added to the wells. After 1 h, the medium was discarded and followed by a 10 min shaking step with DMSO (Carl Roth, Karlsruhe, Germany; 7029) to dissolve the obtained formazan crystals. Absorption was measured at 550 nm with 690 nm as reference wavelength, using a microplate reader (Synergy H1, Biotek, Bad Friedrichshall, Germany).

Primary murine hepatocytes
Six male C57BL/6JCtrl mice were housed in a 12:12 h light/dark cycle at 22°C and constant humidity at 55%, fed a nutrient-rich standard chow (V1534-300, Ssniff, Soest, Germany) ad libitum and were sacrificed in deep isoflurane (Abbott, Wiesbaden, Germany) anesthesia by cervical dislocation at the age of 20 weeks. Primary, mouse hepatocytes were isolated as described earlier 53 . After isolation and seeding, cells were treated with 50 nM Na 2 SeO 3 in combination with or without 10 µM CuSO 4 for 24 h before harvesting.

LPP rat model
Rat liver homogenate and plasma samples were generated and provided as previously described 54,55 . Briefly, rats were housed in a 12:12 h light/dark cycle at 22°C and constant humidity at 55% and maintained on an ad libitum Altromin 1314 diet (Altromin Spezialfutter) and tap water. Some rats of the diseased group received the copper chelator methanobactin for 8 days 20 . Rats were anesthetized with isoflurane (Abbott) and blood was withdrawn to sacrifice the animals. Tissue samples were mixed with 0.1% Triton-X 100 (Carl Roth; 3051.4) and 2 µg/mL protease inhibitor (Merck; 539134) and centrifuged at 14,000 × g (10 min, 4°C). The supernatant was used for further analysis.

Wilson's patients
Serum samples were obtained from 23 patients at the time point of diagnosis of Wilson's disease recruited between 2011 and 2019 at the University Hospital Heidelberg, Germany as part of the clinical trial 'Biochemical and genetic markers of liver diseases'. Informed consent was obtained by participants. Patient characteristics are shown in table S1. Sex of participants was determined based on self-report.
Patients were classified into two groups (normal or low) based on their circulating CP concentration (90 µg/L as cut-off) 21 . Patients were further subdivided into 'severe disease' based on their inflammatory parameters such as serum CRP and hepatic transaminases as marker for liver pathology.

Deglycosylation using PNGase F and EndoH
HepG2 cell lysates were used for deglycosylation experiments using PNGase F (New England Biolabs, Frankfurt, German; P0704S) and EndoH (New England Biolabs; P0702S). 20 µg of protein was diluted with 10 × glycoprotein denaturing buffer and denatured at 100°C for 10 min followed by cooling and a 10 s centrifugation step. Afterwards, 10 × GlycoBuffer 2 and 10% NP-40 (the latter only in case of PNGase F) were added to the mixture prior to the addition of 500 U enzyme and incubation for 1 h at 37°C. A preparation without PNGase F or EndoH was used as negative control for each sample.

Mass spectrometry analysis
Secretome samples were dissolved in 5% SDS in water and digested using the filter-aided sample preparation method 57 . In short, proteins were reduced with 100 mM dithiothreitol (Thermo Fisher Scientific; 165680250) at 60°C for 30 min, transferred to 30 kDa Microcon Centrifugal Filter Units (Merck; MRCF0R030), washed several times with 8 M urea and once with digestion buffer (25 mM triethylammonium bicarbonate (TEAB) (Merck; T7408), 0.5% sodium deoxycholate (SDC) (Merck; D6750)) prior to alkylation with 10 mM methyl methanethiosulfonate (Thermo Fisher Scientific; 23011) in digestion buffer for 30 min. Samples were digested with trypsin (Pierce MS grade Trypsin, Thermo Fisher Scientific;90057; estimated ratio 1:20) at 37°C overnight and two additional rounds of trypsin digestions were performed and incubated for two hours each. Peptides were collected by centrifugation. Digested peptides were purified using High Protein and Peptide Recovery Detergent Removal Spin Column (Thermo Fisher Scientific; 88306) according to the manufacturer instructions. SDC was removed by acidification with 10% trifluoroacetic acid (Thermo Fisher Scientific; 434161000). Samples were desalted (Pierce peptide desalting spin columns, Thermo Fisher Scientific; 89852) and dried prior to reconstitution in 3% acetonitrile (Thermo Fisher Scientific; 047138.M1) and 0.2% formic acid (FA) (Thermo Fisher Scientific; 270480250).
Samples were analyzed three times on an Orbitrap Fusion™ Tribrid™ mass spectrometer interfaced with Easy-nLC1200 liquid chromatography system (LC-MS/MS; Thermo Fisher Scientific). Peptides were trapped on an Acclaim Pepmap 100 C18 trap column (100 μm × 2 cm, particle size 5 μm; Thermo Fisher Scientific; 164213) and separated on an in-house packed analytical column (75 μm × 35 cm, particle size 3 μm; Reprosil-Pur C18; Dr. Maisch, Ammerbuch, Germany; r13.b9.) using a gradient from 5% to 80% acetonitrile in 0.2% FA over 90 min. Precursor ion mass spectra were acquired at 120,000 resolution, m/z 400-1400 and maximum injection time of 50 ms. MS2 analysis was performed in a data-dependent mode using CID spectra of the 20 most intense precursor ions in each MS scan, recorded in the ion trap at collision energy setting of 35. Precursors were isolated in the quadrupole with a 1.2 m/z isolation window, charge states 2 to 7 were selected for fragmentation and the dynamic exclusion was set to 20 s and 10 ppm.
Proteomic datasets were processed by MaxQuant ver. 1.6.10.43 58 . Data were searched against reference proteome of Homo Sapiens (101,702 entries, June 28 2022, UP000005640) downloaded from Uniprot. MaxQuant-implemented database was used for the identification of contaminants. Protein identification was done using these MaxQuant parameters as follows: mass tolerance for the first search 20 ppm, for the second search from recalibrated spectra 4.5 ppm; maximum of 2 missed cleavages; maximal charge per peptide z = 7; minimal length of peptide 7, maximal mass of peptide 4600 Da; carbamidomethylation (C) as fixed and acetylation (protein N-term) and oxidation (M) as variable modifications with the maximum number of variable modifications per peptide set to 5. Trypsin with no cleavage restriction was set as a protease. Mass tolerance for fragments in MS/MS was 0.5 Da, taking the 8 most abundant peaks per 100 Da for search (with enabled possibility of co-fragmented peptide identification). FDR filtering on peptide spectrum match was 0.01 and only proteins with at least one identified unique peptide were considered further. Proteins were quantified using MaxLFQ function 59 with at least one peptide ratio required for pair-wise comparisons of protein abundance between samples. Proteins identified as contaminants were removed before any further interpretation of data. Statistical analyses were performed in Perseus (v1.3.0.4). t test-based statistics were applied on normalized and logarithmized protein ratios to extract the significantly regulated proteins. Functional annotation of proteins by Gene Ontology (GO) was done using DAVID Gene Ontology enrichment analysis 60 .

ELISA for SELENOP and CP in human serum samples
Human CP concentrations were measured using sandwich ELISA as described 18 . Briefly, serum samples of Morbus Wilson patients were pre-diluted 1:200 and 50 µl were used for CP quantification. A commercially available human CP standard (Ceruloplasmin 187-51-10, Lee Biosolutions, Maryland Heights, MO, USA) served as calibrator. The detection of an HRP-labeled antibody was performed applying 100 µl of 3,3′,5,5′-tetramethylbenzidine (TMB, Surmodics IVD, USA). The spectrophotometric read out was recorded within 10 min at 450 nm using a NanoQuant Infinite 200 Pro microplate reader (Tecan Group AG, Männerdorf, Schweiz). Serum SELENOP concentrations were measured using a validated, commercial sandwich ELISA (selenOtest TM , selenOmed GmbH, Berlin, Germany). Sample preparation and assay procedure was performed according to manufacturer's instructions.

CP oxidase activity
CP oxidase activity was measured in plasma samples of LPP rats and in serum samples of Wilson's patients which were either diluted 1:2 (ATP7B +/-, Wilson's patients) or 1:5 (ATP7B -/rats) in 0.9% NaCl (Carl Roth; 3957). 5 µl diluted plasma samples were mixed with 75 µl acetate buffer (0.1 M sodium acetate, pH 5; Roth; 3856) and were incubated for 5 min at 30°C. Afterwards, 20 µl of a 7.88 mM o-diansidine dihydrochloride solution (preheated to 30°C) (VWR; A17175) were added to each well followed by another shaking step at 30°C. For each sample, a background value was generated by adding 150 µl 9 M H 2 SO 4 (VWR; 1.00731.2500) after 10 min which was subtracted from values generated by stopping the reaction after 60 min. The absorbance of the formed purple-red chromophore was measured at 550 nm with 690 nm as reference wavelength using a microplate reader (Synergy H1, Biotek) using GenS software version 3.11.

Measurement of trace element concentrations
Trace element concentrations were measured in rat plasma and tissue samples, human serum, and cell lysates using a total reflection X-ray fluorescence (TXRF) spectrometer (Bruker Nano GmbH, Berlin, Germany) and TEsprit software version 1.0. As internal standard for cells and rat tissue, 1 mg/L yttrium (Merck Millipore, Burlington, MA, USA; 119809) and for serum or plasma 1 mg/L gallium (Alfa Aesar/ Thermo Fisher Scientific, Kandel, Germany; 88066) was used. 10 µL of each sample were placed on siliconized (except plasma) quartz glass carriers and dried at 40°C. Samples were measured in duplicates for up to 1000 s.

Measurement of free copper
The concentration of free copper was quantified by a fluorometric method using a copper-binding fluorescent peptide (FP4) based on Young et al. 62 . and coupled to carboxyfluorescein as fluorophore. Serum samples were pre-diluted (1:20) in assay buffer (50 mM HEPES, pH 7.5 63 ,) and stored at −80°C. An aliquot of 20 μL was added to 80 μL assay buffer containing FP4 (final concentration 10 nM) at room temperature. Free serum copper concentrations were calculated based on the fractional saturation of the sensor 64 , determining the maximal and minimal fluorescence signal of FP4 after addition of 2 mM EDTA or 2.2 mM CuSO 4 , respectively, and using a dissociation constant for the FP4-copper-complex of 0.37 pM.

Association between candidate SNPs and SELENOP in EPIC-Potsdam
The EPIC-Potsdam cohort study consists of 27,548 participants recruited between 1994 and 1998 from the general population in Potsdam and surroundings, Germany. We used a random sample of 2500 individuals from 26,444 participants who provided blood samples at baseline. Of these, participants with prevalent diabetes, myocardial infarction or stroke were excluded. Further exclusion criteria were missing genetic data and missing data on trace element measurements, leaving 2204 individuals for analyses. DNA extraction from buffy coat, genotyping, and quality control within EPIC-Potsdam has been described in detail elsewhere 23 . Two candidate SNPs from the CP gene region were selected from previous GWAS analysis on serum copper levels in EPIC-Potsdam (rs11708215) and GWAS meta-analysis (rs34951015) 23 . Serum SELENOP levels were natural log-transformed to normalize the right-skewed distributions and standardized. We assumed an additive genetic model, adjusted for age at recruitment and sex.

Immunofluorescence
Cells were grown on glass coverslip with the indicated trace elements for 24 h, fixed in 4% methanol-free paraformaldehyde (Thermo Fisher Scientific; 043368.9 L) for 10 min at room temperature and washed three times with PBS, permeabilized with Triton X-100 0.1% for 15 min at room temperature, rewashed with PBS and blocked with BSA 2% diluted in PBS Tween 20 0.1% (PBST) for 30 min at room temperature. Cells were incubated overnight at 4°C with primary antibodies diluted in BSA 1% in PBST. The next day, cells were washed three times with PBS, incubated with fluorophore-conjugated secondary antibodies in 1% BSA-PBST for 1 h at room temperature, washed with PBS and then stained with DAPI (Thermo Fisher Scientific; D3571) for 5 min at room temperature, and mounted with ProLong™ glass anti-fade medium (Thermo Fisher Scientific; P3698). The following antibodies were used: mouse anti-SELENOP (1:200 56 ;), rabbit anti-GOLGIN97 (1:200; Abcam; Ab8340), and secondary antibodies anti-mouse AlexaFluor 488® (1:500; Thermo Fisher Scientific; A11029) and anti-Rabbit AlexaFluor 647® (1:500; Thermo Fisher Scientific; A32795).

Statistics
Data are given as mean ± SD. The numbers of independent biological replicates are indicated in the figures as dots. Statistical analyses were performed using GraphPad Prism version 8 (San Diego, CA, USA). Mean differences between two groups were analyzed using Student's two-tailed t test. Differences between three or more groups were analyzed using one-way, two-way or three-way ANOVA followed by post hoc analysis through Bonferroni's multiple comparison test depending on the type of analyzed samples. Linear correlations were performed using the Pearson test. Data distribution was considered normal if it passed the Shapiro-Wilk normality test. A p-value below 0.05 was considered statistically significant.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral. proteomexchange.org) via the PRIDE partner repository 61 with the dataset identifier PXD036300. As reference proteome of Homo Sapiens UP000005640 was used. The remaining data generated in this study are provided in the Supplementary Information. Source data are provided with this paper.