Proteomics reveals a global phenotypic shift of NK cells in HCV patients treated with direct‐acting antivirals

Chronic hepatitis C virus (HCV) infections compromise natural killer (NK)‐cell immunity. Direct‐acting antivirals (DAA) effectively eliminate HCV, but the long‐term effects on NK cells in cured patients are debated. We conducted a proteomic study on CD56+ NK cells of chronic HCV‐infected patients before and 1 year after DAA therapy. Donor‐variation was observed in NK‐cell proteomes of HCV‐infected patients, with 46 dysregulated proteins restored after DAA therapy. However, 30% of the CD56+ NK‐cell proteome remained altered 1 year post‐therapy, indicating a phenotypic shift with low donor‐variation. NK cells from virus‐negative cured patients exhibited global regulation of RNA‐processing and pathways related to “stimuli response”, “chemokine signaling”, and “cytotoxicity regulation”. Proteomics identified downregulation of vesicle transport components (CD107a, COPI/II complexes) and altered receptor expression profiles, indicating an inhibited NK‐cell phenotype. Yet, activated NK cells from HCV patients before and after therapy effectively upregulated IFN‐γ and recruited CD107a. Conversely, reduced surface expression levels of Tim‐3 and 2B4 were observed before and after therapy. In conclusion, this study reveals long‐term effects on the CD56+ NK‐cell compartment in convalescent HCV patients 1 year after therapy, with limited abundance of vesicle transport complexes and surface receptors, associated with a responsive NK‐cell phenotype.

Modern IFN-free therapies rely on direct-acting antivirals (DAA) that eliminate HCV within 8-12 weeks [28].Several publications suggested successful DAA treatments normalized NKcell functions, including the expression of activating receptors, cytotoxicity, and cytokine production [29][30][31][32].However, another study examining patients up to week 12 after therapy showed only a transient reactivation of NK cells, followed by their renewed dampening down to pretreatment levels [33].Consistent with this, we also found that NK-cell immune responses in HCV patients before and after treatment did not differ significantly from those of healthy controls [24], and HCV clearance did not completely restore their altered cytokine and chemokine milieu [34,35].Furthermore, we have already shown that chronic HCV infection leads to long-lasting imprints not only in NK cells [24] but also in mucosa-associated invariant T cells [36].More recently, we reported that the imprint of unconventional Tcell responses persists in acute HCV infection despite successful early antiviral treatment [37].In conclusion, features of immune exhaustion persist after treatment, suggesting that chronic HCV infection might result in irreversible damage to the immune system [38].However, a systematic description of the NK-cell phenotype in cured patients is missing.
In this study, we evaluated the molecular phenotype of the whole CD56 + NK-cell compartment from chronic HCV patients before, during, and 1 year after DAA therapy.HCV patients received the same antiviral therapy with the well-characterized DAA combination of sofosbuvir and ledipasvir [39][40][41], and clinical proteomics characterized their NK-cell compartment compared to that of healthy individuals.This approach identified a set of dysregulated NK-cell proteins normalized upon DAA therapy, and surprisingly, proteomics showed that 30% of NK-cell proteomes in HCV patients were found altered with low donor-variation even 1 year after therapy.These long-lasting changes do not seem to affect the overall responsiveness of NK cells to K652 target cells, but impact surface receptor profiles of CD56 + NK cells following chronic HCV infection.

NK-cell proteome response in DAA-treated HCV patients
This study aims to complement proteome information on the NKcell compartment in HCV patients that undergo DAA therapy.For this, a group of six patients (three women and three men) with confirmed chronic HCV infections that received an IFN-free DAA therapy using sofosbuvir and ledipasvir were defined.In order to minimize global effects of secondary diseases on the immune system, only patients without liver cancer or fibrosis were studied.Before therapy (baseline, BL), patients showed normal lymphocyte numbers, but increased levels of alanine aminotransferase, aspartate aminotransferase, gamma-glutamyltransferase, and FibroScan (kPa) of liver stiffness.At the end of DAA therapy (week 8), the HCV RNA level in the patients' blood dropped to an undetectable level and stayed undetectable until 48 weeks after treatment (fu48), demonstrating the sustained virological response.As expected, DAA therapy in these patients widely normalized the liver function as indicated by alanine aminotransferase, aspartate aminotransferase, gamma-glutamyltransferase, and FibroScan levels (Fig. 1A and Supporting Information Table 1).Thus, DAA therapy improved the health status of all patients, and none of them have had HCV recurrence or developed HCC along with this study.
To further characterize the NK-cell phenotype by proteomics, the CD56 + NK cells were sorted by flow cytometry (Supporting Information Fig. 1A) from these HCV patients before (BL) and after treatment (fu48).Additionally, we sorted CD56 + NK cells from six healthy individuals who had no history of chronic viral infections for comparative analyses.Following NK cell sorting from peripheral blood mononuclear cell (PBMC) fractions, NKcell pellets were stored at −80°C until the start of proteome analyses.As we could not detect a shift in CD56 dim/bright ratios compared to healthy donors (Supporting Information Fig. 1B), the major CD56 dim and CD56 bright NK-cell subsets were analyzed as a whole to robustly characterize individualized NK-cell proteomes.The collected NK-cell samples were then examined by label-free quantitative proteomics.For that, all 18 individual NK-cell samples (6 BL, 6 fu48, and 6 healthy) were subjected 3 times to 1D-LC-MS-based peptide sequencing using long-LC-gradients for an optimal resolution of proteomes.Thus, NK-cell proteome information was obtained for each patient in triplicates allowing a quantitative evaluation of their individual CD56 + NK-cell compartments (Fig. 1B and Supporting Information Table 2).
First we performed correlation analyses using all 54 proteome data sets ( NK cells from the same patients (BL, fu48) and healthy controls were isolated from blood samples by flow cytometry and then analyzed by proteomics (label-free LC-MS/MS), measuring triplicates for each patient/donor sample.MaxQuant (MQ) and R were used to calculate patient-and donor-specific abundances of NK-cell proteins.Student's t-test was used to compare the expression of each protein across the groups, and the corresponding p-values were indicated in the example.We only included in this study information on proteins that have been robustly characterized by mass spectrometry in at least 15 of 18 patients/donors.By this, 4042 NK-cell proteins could be identified and quantified in direct-acting antivirals (DAA)-treated patients in comparison to healthy individuals.outliers and were found highest between technical replicates, proving the robustness of this analytical workflow (Supporting Information Fig. 2).To test variances between the donors, interindividual and intergroup analyses were performed based on Pearson correlation coefficients (Fig. 2A).Donor-variation was significantly higher in the group of chronic HCV-infected patients before therapy than in healthy individuals.Mean and median correlations in the healthy control group as well as in the patient group after therapy were approximately 0.98, whereas the mean and median correlations in the patients group before treatment (BL) were approximately 0.95.Application of the t-test to the correlations with p-values <0.0001 indicated significant differences in mean correlation between the before treatment and the healthy group and between the before treatment and the after treatment group.The p-value for correlations within the healthy group compared to the after treatment group was 0.14.This indicated an interindividual heterogeneity of CD56 + NK-cell immunity in the phase of chronic HCV infections, which, however, was compensated 1 year after DAA therapy in the cured patients, reaching the homogeneity of healthy individuals.
With respect to intergroup variations, we found a low correlation between healthy donors and chronic HCV-infected patients, confirming the altered NK-cell phenotype in HCV infection.Notably, NK-cell proteomes of cured patients remained still notably different in comparison to the healthy group even 1 year after therapy.This was unexpected as the donor-variations between HCV patients normalized upon viral clearance resulting in similar low donor-variations as in the group of healthy individuals.A more detailed look at the individual responses in the group of HCV-infected patients indicated two classes of patients: Donor-variations were high for females (BL_1-BL_3), whereas lower for males (BL_4-BL_6), indicating a more homogenous NK-cell response toward DAA therapy in male patients.Additionally, we performed principal component analyses (PCA) to confirm that HCV patients before treatment are largely heterogeneous, and patients after treatment are more homogeneous Correlations are generally high, that is, greater than 0.93, and red indicates lower correlations, green higher correlations.The red dashed boxes refer to correlations of donors from the same group and can be used as an indicator for the interindividual variance within each group (healthy, before and after treatment).The correlations in the group before treatment are lowest suggesting a higher interindividual variance.The blue dashed boxes highlight the correlations in comparison to healthy donors indicating the variance of NK-cell proteomes of HCV patients before as well as after therapy.(B) Unsupervised hierarchy clustering of NK-cell proteomes using Perseus.Total proteome data (all 4042 proteins) were normalized with Z-score and then used for unsupervised hierarchy clustering in Perseus.The symbols indicate sex (round for females, triangles for males) and HCMV status (green for negative, red for positive, gray for data not available).The ages of healthy donors and patients when donating blood are indicated in brackets.The red-blue color scale represents the normalized abundance of each protein, with red indicating high abundance and blue indicating low abundance.and clustered together.In this PCA analysis, we also integrated available information on the CMV status, which, however, did not segregate patients into their respective groups.In contrast, the HCV infection status separated HCV patients clearly from the group of healthy donors, with the cured patients being particularly close (Supporting Information Fig. 3).Ultimately, we used total proteome signatures for unsupervised hierarchical cluster analyses.Using this approach, CD56 + NK-cell proteomes segre-gated healthy individuals from HCV patients and confirmed that proteomes from cured HCV patients clustered closely together (Fig. 2B).This is different from pretherapy (BL) patients, whose profiles showed notable heterogeneities.NK-cell proteomes of those patients were less consistent and indicated subclusters, although no correlation with age, sex, or CMV status was evident with respect to these subclusters.To clarify which NK-cell functions mostly contribute to the observed heterogeneity, we determined those proteins whose expression showed the highest donor-variation in the BL group (Supporting Information Table 2, column I).Proteins with the highest variations (SD > 1.5) allowed protein network building by the STRING database indicating their functional association (Supporting Information Fig. 4).Interestingly, the network enriched the endosomal and lysosomal machinery and HLA classes I and II components (HLA-A, HLA-B, CD74, B2M) that were associated with HCV viral load and liver diseases before [42].
Ultimately, we aimed to elucidate the underlying mechanism of donor-variation in female patients by profiling plasma samples from further HCV patients that underwent the same successful DAA therapy (six females, seven males).The protein abundance levels of 92 plasma proteins were determined in a targeted approach (Olink technology) and provided information on HLA-regulatory cytokines, including IFN-γ, IL-2, IL-6, and TGFα/β (Supporting Information Fig. 5).However, none of these measured cytokines indicated higher donor-variations in plasma protein levels in women as compared to men.Instead, this approach indicated a higher gender-independent variation of IFN-γ and MCP-4, which was found widely reduced at the end of therapy.Among the 92 plasma components, only Sirtuin-2 (SIRT2) indicated a higher donor-variation in female patients than in male patients (F-test p-value 0.034, adjusted p-value 0.688, data not shown).Interestingly, SIRT2 plays a role in sex hormone synthesis pathways and also co-determines the activity of NK cells [43].Thus, SIRT2 levels may indeed contribute to the observed gender-specific NK-cell variations but requires further validation in a larger cohort.
Taken together, NK-cell proteomes indicate highest donorvariations in chronically HCV-infected patients before treatment, which is reduced to the level of healthy individuals in cured patients 1 year after DAA therapy.NK-cell proteomes from cured patients remained variant from those of healthy individuals, indicating long-term effects on the NK-cell phenotype imprinted by chronic HCV infections.

Identification of CD56 + NK-cell proteins altered in HCV-infected and cured patients
We then determined NK-cell proteins that differentiate patients before and after DAA therapy from healthy individuals.We performed Student's t-tests for triplicate MS data belonging to each protein and created volcano plots to visualize altered NK-cell protein levels of HCV patient groups in comparison to healthy individuals.These volcano plots integrate information on the relative regulation of 4042 NK-cell proteins as well as the significance of regulation indicated by high p-values (from Supporting Information Table 2).We then determined the number of CD56 + NKcell proteins with altered abundances in chronic HCV infections.Stringent threshold criteria (minimal fold change s0 = 0.2, false discovery rate [FDR] = 0.01) classified 115 proteins significantly downregulated before treatment and 192 proteins significantly upregulated in NK cells from HCV-infected patients in compari-son with healthy individuals (Fig. 3A).Thus, proteomics identified 307 regulated proteins in the CD56 + NK-cell compartment by chronic HCV infection.
Next, we applied the same selection criteria to inspect of how DAA therapy impacts the number of regulated CD56 + NK-cell proteins.Regulations of NK-cell proteins from cured patients were found in a similar dynamic range as observed before therapy, but notably more proteins had to be classified as significantly altered based on low donor-variations and corresponding high p-values.In total, about 30% of the examined CD56 + NK-cell compartment was affected 1 year after DAA therapy with 458 significantly downregulated and 776 significantly upregulated proteins (Fig. 3B).We then asked the question whether changes in the CD56 + NK-cell compartment are similar or different in HCV patients before and after DAA therapy.In particular, we were interested in the extent to which DAA therapy can compensate for the changes observed in chronically HCV-infected patients.The longitudinal effect of DAA therapy is summarized in a Sankey diagram (Fig. 3C) and an animation (Supporting Information Video 1).The Sankey diagram links all 4042 protein regulations as defined in the volcano plots for the group of patients before and after therapy.In this way, it can be determined, for example, to what extent altered protein expressions are normalized by DAA therapy and end up in the group of unregulated proteins as compared to healthy individuals.Of note, only 46 out of the 307 NK-cell proteins with altered abundances in chronically HCV infected patients were found rescued 1 year after therapy.Of those, 30 were identified upregulated before therapy including IFN-dependent responses as exemplified by antiviral MX2 levels.On the other hand, 16 proteins were found repressed only in infected patients suggesting their role in immune evasion (Supporting Information Table 3).Several of these proteins have RNA binding functions (DDX19A, ZRANB2, LAS1L, SRP9, TXNL4A, and RPS29), whereas ARHGAP31, CORO7, TUBB8, and GTPBP3 are GTPase-related proteins.Interestingly, DDX19A senses viral RNA [44], and its downregulation is instrumental for HCV to reduce inflammasome-dependent antiviral responses.
In contrast, most of the 307 proteins that were up-or downregulated before treatment remained up-(n = 162) or downregulated (n = 99) in the cured patients, respectively.Moreover, about half of these proteins showed no normalization or even an increase in their altered expression compared with healthy subjects (Supporting Information Video 1).Additionally, normally abundant NK-cell proteins of HCV-infected patients became either downregulated (n = 359) or upregulated (n = 614) after DAAtreatment.Patients at fu48 (after therapy) show less heterogeneous NK-cell proteomes as compared to the proteomes at BL (before therapy), and generally more proteins were classified as significantly regulated based on their p-values.Thus, the dynamics of protein regulation in this patient group suggests a phenotypic shift rather than a slow normalization of previously deregulated NK-cell proteins.The observed phenotypic shift includes important antitumor and antiviral regulators.For instance, NK cells of cured HCV patients express less CD47 but more Cyclin L2 (CCNL2).CD47 is a systemic regulator of NK-cell homeostasis and important for NK-cell responses in viral infection [45], and CCNL2 is required for tumor control [46].Furthermore, POTEE, which is most significantly upregulated in cured patients, suggests a shift toward mTORC2-dependent NK-cell maturation [47][48][49].
Taken together, DAA therapy rescued only 46 of 307 CD56 + NK-cell proteins that were regulated under the condition of chronic HCV infection.Instead, the total numbers of regulated NK-cell proteins were surprisingly found increased 1 year after DAA therapy, affecting 30% of the CD56 + NK-cell compartment.

Long-term effects on NK-cell processes and pathways in DAA-cured HCV patients
Proteome data and the number of regulated proteins suggested a phenotypic shift of the NK-cell compartment, and therefore, we asked the question, which processes and pathways mostly con-tribute to this effect in DAA-cured HCV patients.We speculated that long-lasting effects on the NK-cell phenotype are accompanied by the regulation of proteins with impact on gene expression.Indeed, proteomics gave insight into 368 RNA processing proteins (Supporting Information Table 4 and Fig. 4), out of which 100 proteins were significantly dysregulated in the NK cells of chronically HCV-infected patients compared with healthy controls.After DAA treatment, 84 (84%) of the dysregulated RNA processing proteins in NK cells from chronic HCV patients were sustained dysregulated in the NK cells from cured patients.Only 16 proteins (16%) were rescued in NK cells of cured patients to the level of healthy controls.
To better understand the altered NK-cell phenotype of DAAcured HCV patients, we then validated whether distinct processes or pathways are in general up-and/or downregulated.For this, all 4042 quantified proteins were classified based on their fold changes and p-values into 5 "quantiles" whereby the middle Out of the 4042 proteins, 368 proteins (9%) were annotated with "RNA processing" (GOBP), out of which 100 proteins were significantly dysregulated (s0 = 0.2, false discovery rate [FDR] = 0.01) in the NK cells of chronically HCV-infected patients (baseline [BL]) compared with healthy controls.After direct-acting antivirals (DAA) treatment, 84 (84%) of the dysregulated RNA processing proteins in NK cells from chronic HCV patients were sustained dysregulated in the NK cells from cured patients.Only 16 proteins (16%) were rescued in NK cells of cured patients to the level of healthy controls.
quantile Q3 represents no regulation, and the two lower (Q1, Q2) and upper quantiles (Q4, Q5) represent moderate and most robust down-and upregulations in comparison to healthy individuals, respectively (Supporting Information Fig. 6).Proteins of each quantile were then used for enrichment analyses based on gene ontology (GO) and KEGG to recognize functional consequences following DAA therapy.
This approach revealed a limited number of biological processes (GOBP) and pathways (KEGG) deregulated in CD56 + NK cells after DAA therapy (Fig. 5).Interestingly, most of the top 10 significant processes were detected in Q1 indicating those to be somewhat impaired in HCV patients 1 year after DAA therapy.In particular, four of those have relevance for NK immunity, namely, "Response to stimulus", "Regulation of NK cell mediated cytotoxicity", "Intracellular protein transport", and "Chemokine signaling".In contrast, the majority of top enriched processes and pathways from patients before treatment were upregulated (showed up in Q4 or Q5), for example, tRNA aminoacylation, amino acid (aa) activation, response to ATP (Supporting Information Fig. 7).Furthermore, some NK-cell immunity-related processes were already moderately downregulated in HCV-infected patient (enriched in Q2) and become more regulated after DAA therapy (enriched in Q1).

Effector molecule and transport phenotype of DAA-treated HCV patients
To evaluate the specific impact on functions as part of processes and pathways, we generated plots for each of the 4042 CD56 + NK-cell proteins presenting regulatory and statistical information on their protein levels in comparison to the group of healthy individuals (Supporting Information Table 2 and Supporting Information File 1).As "Regulation of NK cell mediated cytotoxicity" is central to antiviral and antitumoral immunity, we first evaluated the content of lytic granules.We found all granzymes (Gzm A, B, H, K, and M) at normal levels in comparison with healthy donors (Fig. 6A).Further, perforin levels indicated a moderate but not significant increase, whereas Granulysin, which induces target cell apoptosis through increasing intracellular calcium and decreasing intracellular potassium [50], was more abundant in CD56 + NK cells of cured patients.In contrast, CD107a (LAMP1), which is required for the release of perforin and granzymes into target cells [51], was found significantly downregulated suggesting a degranulation phenotype in patients' samples both before and after treatment.Indeed, complexes required for the release of lytic granules [52] and enriched in "molecular transport" were also found downregulated.In particular, components from the cytoplasmic coat protein (COP) complex I (COPI) that regulates vesicle budding from the Golgi apparatus directed to the ER, and COPII that transports vesicles from the ER to the Golgi apparatus [53,54].All detected components of COPI (COPA, COPB1, COPB2, COPE), COPII (SEC23A, SEC24C), and clathrin (CTLB and CLTC), which regulate the vesicle transport between Golgi, endosomes, and the surface membrane [55], were found consistently downregulated after therapy (Fig. 6B).Protein levels of all COP components were found lowest in the NK cells of patients 1 year after therapy suggesting a sustained phenotypic shift.Indeed, the probability that the same regulation pattern over the three donor groups occurs in all eight detected components of COPI and COPII by chance is notably low ((1/6) 7 ≈ 0.0000036).
To shed first light on the functional relevance of a lower amount of vesicle transport proteins, we evaluated NK cells in an additional group of ten patients (six women, four men).Consistent with patients analyzed by proteomics, frequencies of NK cells were not affected in this group in comparison to healthy individuals, both before and after therapy (Supporting Information Fig. 8A).Further, we found no differences in their activation status (CD69) as well as normal granzyme B levels, in line with data from proteomics (Fig. 6A, Supporting Information Fig. 8B/C).Then, NK cells in PBMC fractions were confronted with K562 cells to characterize early (1 h) CD107a mobilization by flow cytometry.Interestingly, NK cells from HCV patients before and after therapy could efficiently recruit CD107a to the cell surface (Supporting Information Fig. 8B/C).Likewise, NK cells from HCV patients were also found well able to induce IFN-γ responses.As transport components may affect the cellular localization of receptors, we next selected Tim-3 (HAVCR2) and CD244 (2B4) for surface exposure measurements.Tim-3 was found downregulated in our proteome study before therapy (see Fig. 7), and CD244 (2B4), although not detected by proteomics, was reported to be downregulated in response to IFN-α and ribavirin therapy [56].Interestingly, flow cytometry revealed reduced surface expression levels (MFI) of both receptors in all HCV patients in comparison to healthy individuals.This reduction was observed before therapy and, interestingly, remained reduced after therapy (Supporting Information Fig. 8B/C).In conclusion, the reduced abundances of vesicle transport complexes COPI/II components do not limit the early CD107a-dependent degranulation but may affect surface receptor transport.

Signaling response phenotype of DAA-cured HCV patients
Next, we evaluated regulated components that enriched "Response to stimulus" and "Chemokine signaling".Of note, the functional diverse group of chemokine signaling was well covered by this proteome study (Supporting Information Table 6).In fact, very few chemokine signal components showed elevated levels, and these were limited to the period before or after therapy.For instance, the small GTPase Rac1, known to limit cytotoxicity and granule secretion [57], was only upregulated before therapy, whereas phospholipase C beta 2 was notably increased after therapy.Phospholipase C beta 2 inhibits the virus-induced expression of pro-inflammatory cytokines [58].Concomitantly, most significant downregulations were observed of β-arrestin 2 (ARRB2) and the GTPases RAP1A and RAP1B.β-arrestin 2 is an activator of cGAS-STING signaling and target of viral immune evasion [59].RAP1 GTPases control immunological synapse formation and are required for receptor-mediated chemokine and cytokine responses of NK cells [60].As results indicated the regulation of signaling pathways, we examined the group of surface receptors based on the in silico human surfaceome definition [61].In total, this proteome study identified 112 surface proteins, of which 34 were significantly (p < 0.05) regulated in NK cells before therapy and 55 were significantly regulated 1 year after therapy (Supporting Information Table 7).This group includes 19 functionally described activating and inhibitory NK-cell receptors as well as surface proteins with elusive immune functions in NK-cell immunity (Table 1).
Before therapy, the abundances of activating and inhibitory receptors indicate a lower NK-cell responsiveness as described before (Fig. 7A): Three activating receptors were found downregulated (CD59, CD45, CD317) in the NK cells of chronic HCV patients, whereas only two activating receptors were   After therapy, the inventory of surface receptors shifted even further toward a lower NK-cell responsiveness (Fig. 7B): Only the expression of SLMF7 and HAVCR2 normalized, whereas all other receptors mentioned before remained regulated with the exception of DNAM-1.Of note, we could show by flow cytometry that the surface expression of HAVCR2 (Tim-3) was still downregulated after therapy (Supporting Information Fig. 8B/C), indicating that normalized total protein levels of this receptor do not coincide with normalized signaling capabilities.Further, the activating receptor DNAM-1 changed the direction of regulation and was now found downregulated after therapy.Additionally, leukocyte surface antigen CD47 and the activating receptors fractalkine (CXCL1), HLA class II histocompatibility antigen gamma chain (CD74), and HLA class II histocompatibility antigen gamma chain (SELPLG/CD162) were found downregulated only after therapy.
Taken together, surface receptor profiles favor inhibitory phenotypes of CD56 + NK cells before therapy (Fig. 7C).Most of the activating receptors were downregulated, and all identified inhibitory receptors were found upregulated in the NK cells of chronically HCV-infected patients.DAA therapy normalized the abundance of a few surface receptors, but overall the receptor profiles of cured patients 1 year after therapy are not restored to the levels of healthy individuals (Fig. 7D).

Discussion
DAA have revolutionized anti-HCV treatment and allowed to cure more than 95% of patients.This study characterized patients treated with the combination of sofosbuvir and ledipasvir that are well characterized.Ledipasvir targets the NS5A replication complex, and sofosbuvir inhibits the NS5B polymerase activity of HCV [87][88][89].Clearance of HCV protects patients from virus-driven pathologies and basically can improve NK-cell immunity after DAA treatment [29,[90][91][92].However, other studies indicated no significant difference in NK-cell responses before and after DAA therapy [93], and the impact on the NK-cell repertoire diversity was irreversible, suggesting that imprints on NK-cell compartments persist for years despite clearance of HCV [24,38].Furthermore, other immune cells, including mucosa-associated invariant T cells are irreversibly impacted by acute [37] and chronic HCV infection [34].These observations indicate the importance of studying effects on the cellular and soluble immune system and immune cell plasticity following both acute and chronic viral infections.
Generally, patient-specific data on long-term effects upon viral infections are rare and not accomplished at the systems level.Here, we now characterized a group of six HCV genotype 1 positive, noncirrhotic patients by means of a clinical proteome pipeline before and almost 1 year after DAA therapy.Our gating strategy excluded CD56 − NK cells that can expand in HCV patients and share similarities with CD56 dim NK cells [26].Unlike some previous reports [17,94,95], patients included in this study did not show a shift in the frequency of CD56 dim and CD56 bright NK cells with about 90% and 10% both in healthy and HCV patients, respectively (Supporting Information Figs.1B and 8A).
Correlation analyses showed that the CD56 + NK-cell proteomes of HCV patients differ significantly from those of healthy individuals, not only before but also especially after therapy.Before therapy, NK-cell proteomes of patients indicated a notable higher interindividual variation.This heterogenicity normalized within the patient cohort 1 year after DAA therapy to the level of healthy individuals, which is in full accordance with the DAAdependent reduction of DEV reported before [24].Of note, PCA analyses relate this heterogeneity to the three female patients, whereas the three male patients are more similar to each other, suggesting sex-based differences in NK-cell responses to chronic HCV infection.In this line, cluster analyses faultlessly discriminate healthy from HCV patients but revealed restrictions for clearly segregating untreated HCV (BL) patients and indicated potential subclusters, which share variant similarities to the more consistent cluster of cured (fu48) patients.These subclusters could not be correlated with age, gender, or HCMV status in the BL group (Fig. 2) and thus may indicate variant NK-cell immune states in chronic HCV infection.Interestingly, NK-cell proteins with the highest donor-variations in the BL group can be functionally associated in a protein-protein-interaction network and enrich lysosomal and endosomal as well as HLA classes I and II components.HLA genotypes have been associated with viral load and more recently with susceptibilities in HCV risk groups and now may support the further classification of chronic HCV patients [96][97][98].In this context, it is reasonable to speculate on the role of sex hormones and sex-specific cytokine levels to explain donor-variations in the NK-cell proteomes of female patients.Profiling plasma samples of female and male DAA-treated HCV patients did not indicate the importance of HLA-regulating cytokines but discovered notable donor-variation of SIRT2 in female patients before therapy.Interestingly, SIRT2 plasma levels can predict the outcome of assisted reproduction in men [99] and play a role in progesterone, estradiol, and testosterone synthesis [100].As SIRT2 is also impacting the activity of liver NK cells [43], it would now be interesting to characterize in larger cohorts whether SIRT2 plasma levels indeed define or even co-regulate NK-cell proteins in HCVinfected patients.Nevertheless, cluster analyses indicated also many consistent and group-specific protein regulations.For example, the functional uncharacterized NK-cell protein C3orf86 was found consistently highest expressed in all HCV patient samples, whereas POTEE (see below) was only consistently high expressed in cured patients.C3orf86 was recently introduced as a biomarker in cancer [101] and in NK cells additionally may indicate ongoing or past chronic infections.
In particular, this study identified CD56 + NK-cell proteins with altered abundances in chronic HCV infected patients as a repository to better understand NK-cell immunity along DAA therapy (Supporting Information Table 2).Surprisingly, only 46 protein abundances were found to be rescued in cured patients, indicative for limited plasticity of NK-cell immunity.Indeed, DAA-rescued proteins included prototypic IFN responses as exemplified by MX2; an IFN-stimulated protein involved in antiviral function during HCV infection [102,103].In the same line, it is tempting to speculate about the role of 16 downregulated and DAA-rescued proteins as part of immune evasion mechanisms (Supporting Information Table 3).Indeed, the here-identified RNA binding proteins could interfere with the translation and replication of HCV or promote the translation of antiviral proteins.Additionally, regulated GTPase-related proteins might facilitate HCV replication [104] or participate in the polarization of cytolytic effectors [105] and cell migration [106].Downregulation of these proteins could be either a response of NK cells to inhibit HCV replication or an immune evasion strategy of HCV to impair NK-cell immunity by interfering with migration and degranulation.Furthermore, NPC2 promotes postlysosomal export of cholesterol [107], and cholesterol transport between endosome and lysosome is essential for HCV replication [108].Accordingly, the downregulation of NPC2 in chronic HCV patients could be used by NK cells to block the HCV life cycle.In conclusion, DAA therapy in combination with cellular OMICs provides a promising concept to observe immune evasion "in reverse" and may complement missing information on how HCV controls immunity.
CD56 + NK-cell proteomes of all patients (BL and fu48) were found clearly different in comparison to healthy individuals.Essentially, many moderate or not significant regulations before therapy were found more pronounced and significant after therapy, indicating a global phenotypic shift of the CD56 + NK-cell compartment.Enrichment analyses revealed that only a limited number of dysregulated pathways at BL were rescued after DAA therapy, and these include tRNA aminoacylation, ama activation, and activation of signaling protein activity involved in unfolded protein response (Supporting Information Fig. 7).Conversely, pathways and processes with importance for NK-cell immunity were found downregulated after DAA therapy (fu48) suggesting a limited responsiveness and cytotoxic capacity of NK cells in DAAcured patients (Fig. 5): (i) Chemokine signaling is restricted.The most significantly dysregulated KEGG pathway in patients after DAA-treatment was "chemokine signaling pathway".Of all pathway components, PLCB2 was the top upregulated, and ARRB2, RAP1A, and RAP1B were the top downregulated proteins (Supporting Information Fig. 9A).PLCB2 inhibits the virusinduced expression of pro-inflammatory chemokines [58], whereas ARRB2 promotes cGAS-STING signaling and IFNβ production [59].Both RAP1A and RAP1B are required for mobilization and clustering of LFA1, and RAP1B is essential for NKG2D, Ly49D, and NCR1-mediated cytokine and chemokine production [60,109].Taken together, the chemokine-dependent NK-cell immunity favors a more restricted responsiveness in chronic HCV patients, even 1 year after viral clearance.
Although we could not directly detect and characterize the expression of all receptors (e.g., NCRs, NKG2A, and NKG2D), we could validate CD16 expression levels between chronic HCV infected patients and healthy donors as described before [24].On the other hand, this study complemented information for multiple other receptors.The sustained downregulation of activating receptors (CD59, CD317, CD45, and CD11a) and sustained upregulation of inhibitory receptors (CD300A, GPR56, and LAIR1) suggest that receptor-related functions within NK cells from cured HCV-infected patients are limited.In addition, two activating receptors, CD74 and CX3CL1, were shifted from unregulated to downregulation along DAA therapy.CD74 is a receptor for macrophage migration inhibitory factor (MIF) and stimulates NF-κB-mediated inflammatory responses [116].CX3CL1 has been shown to promote the adhesion and migration process of leukocytes [67].Downregulation of these two receptors suggests that the abilities for adhesion and migration, as well as response to MIF stimulation, are impaired in the NK cells from cured patients.Furthermore, the expression of the activating receptor BTN3A1, which stimulates NKp30-mediated cytokine production [71], was sustained upregulated, implying that NK-cell responses may be trained for higher cytokine production in HCV patients.Other regulated surface receptors can either activate or inhibit NK-cell immunity based on bound ligands: In chronic HCV infection, SIGLEC7 was reported to promote NK-cell immunity [76].Therefore, the sustained downregulation of SIGLEC7 can serve as proof-of-concept and supports an impaired NKcell phenotype in DAA-cured patients.Next, CD47 has been reported to regulate NK-cell effector protein expression and activation against tumor cells [82] as well as inhibit the expression of IFN-γ and effector proteins against lymphocytic choriomeningitis virus infection [83].Therefore, the shift of CD47 to downregulation after treatment suggests that the NK cells of cured HCV patients might show impaired immunity against cancer cells but trained immunity against virus infection.This is in line with previous reports showing that the risk for tumor development in cured HCV patients is still present or even slightly higher [117][118][119][120][121].
Finally, HAVCR2 (Tim-3) has been reported to enhance IFNγ expression and to inhibit cytotoxicity in HCV infection [78,79].DAA therapy rescued the amount of cellular HAVCR2 to the level of healthy donors; however, flow cytometry revealed sustained low surface expression levels suggesting limited surface receptor transport (see below).HAVCR2 (Tim-3) negatively regulates cytotoxicity and plays a role in NK-cell-mediated immune-tolerance [122].Therefore, reduced surface levels of HAVCR2 (Tim-3) may contribute to the observed NK-cell responsiveness after DAA therapy, as described here (CD107a and IFN-γ) and in previous reports [78,79], which potentially will come along with reduced immune regulatory capacities.Likewise, the identified sustained lower surface expression of CD244 (2B4) may be part of this NK-cell phenotype.Downregulation of 2B4 has been observed in the first weeks of IFN therapy of HCV patients, coinciding with improved cytotoxic NKcell responses and viral clearance [56].Further, CD244 inhibitory signaling is related to exhausted NK-and T-cell phenotypes in chronic infection and cancer [123].Interestingly, our proteomic data corroborate that 2B4 plays an inhibitory rather than activating role after chronic HCV infections.2B4 activation signaling is realized via the SAP (SLAM-associated protein adapter) inhibitory signaling pathway that includes the phosphatases SHP-1 (PTPN6) and SHP-2 (PTPN11) [123].Indeed, proteomics revealed SAP and SHP-2 to be counter-regulated in NK cells after therapy, that is, NK cells are SAP low /SHP-2 high (Supporting Information Fig. 9B), suggesting inhibitory 2B4 signaling.In conclusion, altered surface receptor inventories and surface exposures indicate an aberrant NK-cell responsiveness and preparedness.(iii) Altered COPI/II vesicle transport does not affect NK-cell degranulation.Degranulation of perforin and granzymes toward target cells is an essential step for NK-cell cytotoxicity [124,125].Although we have not found a downregulation of granzymes and perforin in this proteomic study (Fig. 6), we did observe a sustained downregulation of the degranulation marker CD107a in NK cells before and after treatment.The latter, however, is not consistent with a previous report showing CD107a recovery to the level of healthy controls [27].The difference might be due to methodological differences where we measured the total amount of CD107a in resting NK cells, whereas the previous study focused on the surface expression of CD107a on stimulated NK cells [27].As NK cells from patients after DAA treatment still have the ability to degranulate and kill K562 cells [24,29,93], the lower expression level of CD107a does obviously not disrupt the delivery of lytic granules.In fact, we found NK cells of HCV patients well able to recruit CD107a to the surface within 1 h after activation, indicating that the reduced amount of CD107a is at least not limiting initial degranulation.At later time points, degranulation with CD107-negative vesicles that carry a 15 kDa instead a 9 kDa Granulysin variant might play an additional role [126].Proteomic data could not discriminate these variants in this study but found the total Granulysin levels increased after therapy indicating good capacities to induce target cell apoptosis in cured patients [50].One of the most striking proteome findings is the consistent and persistent downregulation of components belonging to the vesicle transport complexes COPI, COPII, and clathrin after therapy.COP complexes regulate vesicle transport at ER and Golgi [127,128], whereby COPI vesicles either contain cargo proteins for retrieval to the endoplasmic reticulum (ER) or proteins that cycle between the ER and the Golgi [127,128].The efficient mobilization of CD107a at the surface of K562-activated NK cells indicated no role of COP-vesicles for the degranulation response of cured patients.Instead, COPI/II complexes have known roles in receptor recycling and thus can play a role in the surface exposure of receptors.Interestingly, we found a sustained lower surface expression of HAVCR2 (Tim-3), although the total amount of cellular Tim-3 analyzed by proteomics was found widely normalized 1 year after therapy (see Fig. 7 and Supporting Information Fig. 8).Therefore, it would now be of interest to elucidate the supposed roles of COP-complexes on immune receptor turnover and localization which to date is widely uncharacterized.
Additionally, we found several mechanisms that potentially regulate NK-cell immunity in the chronic HCV patients: Notably, SUMO1 was upregulated, whereas SUMO2 and SUMO3 were downregulated in NK cells of patients both before and after DAA treatment (Supporting Information Fig. 9C), indicating that NK cells shift to mono-SUMOylation and thereby can counteract STAT1-dependent IFN signaling [129].Next, this study found higher expression of POTEE and RICTOR as part of the mTOR signaling pathway (Supporting Information Fig. 9D).NK-cell education relies on the modulation of mTOR signaling activities that in turn control NK-cell responsiveness through NK cell activating surface receptors [48].Further, sustained upregulation of ubiquitin E3 ligase TRAF6 (Supporting Information Fig. 9E) that targets the assembly of respiratory chain [130-132], as well as dysregulated components of respiratory chain complexes [133] (Supporting Information Fig. 9F), might be the underlying mechanism resulting in the metabolic switch as described for viral infections.
But how can HCV cause a prolonged change of the CD56 + NK-cell proteome that persists after virus clearance?One reason could be the long-lasting changes in the cytokine milieu and other immune cells that we and others have reported [38].Furthermore, the HCV core envelope protein may regulate epigenetics in host cells [134].This might also involve NK cells and their development.Recent studies found that HCMV induces long-lasting alteration of NK-cell epigenetic modifications on surface receptors and cytokines loci [135,136].Of note, HCV is also present in the bone marrow of patients [137,138] where it can get in contact with CD34 + hematopoietic stem cells [139].Therefore, it is reasonable to assume that HCV influences epigenetics in NK-cell progenitors and thus has long-term effects on NK-cell development.Indeed, proteome data support this hypothesis including the regulation of the POTEE/T-bet axis that can indicate an incomplete maturation [140,141].Further, we observed the downregulation of antiviral proteins, including PRMT5 and HMGN2 (Supporting Information Fig. 9G), whereby PRMT5 restricts hepatitis B virus replication through epigenetic repression of DNA transcription [142,143].Therefore, the downregulation of PRMT5 favors epigenetic modifications that can inhibit NK-cell antiviral immunity.Finally, this study revealed RNA processing proteins as the largest functional group with long-lasting alterations in DAA-cured HCV patients.Out of 100 dysregulated RNA processing proteins, the vast majority (84) remained dysregulated after DAA therapy (Supporting Information Table 4 and Fig. 4).Host RNA processing proteins are known targets of viruses to facilitate their replication and for the intervention of antiviral responses [144,145].In NK cells, global regulation of RNA-processing proteins is more likely to be the result of aberrant epigenetics and probably represents a major mechanism for altering 30% of NK-cell proteomes in DAAcured patients.Yet, in this study, NK cells of cured patients were found well armed with effector molecules but revealed limited inventories for vesicle degranulation and receptors required for NK-cell activation.Therefore, the identified regulations shall also be considered as part of a trained immunity or memory NK-cell phenotype.For instance, CD47 (see above) was found downregulated only after DAA therapy, indicating reduced antitumor immunity in favor of trained antiviral immunity in cured patients.Memory NK cells were already reported for MCMV and HCMV [146][147][148], and more recently also HCV-specific memory NK cells were confirmed in patients obtaining the same DAA therapy as used in this study [149].It would now be important to investigate (i) which of the changes described in this study remain further stable in virus-free patients at even later time points and (ii) which of those contribute to HCV-specific or general NK-cell memory phenotypes.
In conclusion, this proteome study complements information on deregulated human NK-cell functions in chronic HCV infections and, importantly, provides evidence of a global phenotypic shift of the CD56 + NK-cell compartment.The here-identified regulations might be rescued at even later time points than 1 year, however, more likely constitute a resource for the further characterization on how chronic HCV infections impact the plasticity and memory phenotype of human NK cells.

Data limitations and perspectives
This proteomic study of HCV patients, before treatment and 1 year after antiviral therapy, is descriptive by nature and provides socalled short-fat OMIC data of the total CD56 + NK-cell compartment from only a limited number of six patients.Overall, 30% of the NK-cell proteome was found consistently altered in all patients after therapy, raising the key question to which extent this constitutes a new functional NK-cell phenotype.As suggested by the reviewers, selected processes were investigated in more detail by orthogonal approaches using samples from two additional patient cohorts.We now can conclude that the reduced amount of transport complex proteins does not affect early NK-cell degranulation, whereas on the other hand, the amount of the co-stimulatory receptors Tim-3 and 2B4 was validated to be notably reduced at the surface of NK cells 1 year after therapy.However, for the vast majority of proteins that were found to be regulated, it remains open at this stage whether and what functional consequences arise for NK-cell-mediated immunity in various disease settings.In this context, it is particularly exciting whether and which adaptations extent our understanding of NK-cell memory and potentially influence the responsiveness of NK cells "lifelong".

Patients' information
Six patients with chronic HCV infection without liver cirrhosis as well as six healthy controls were recruited at the Department of Gastroenterology, Hepatology and Endocrinology at Hannover Medical School, Germany.The patients and healthy donors involved in the proteome study belong to the age group 40-60 years for which no age-dependent effects on NK immunity have been described so far.However, the healthy donors included in this age group are on average younger, which is a possible limitation of this study.Informed consent was obtained from all participants.The study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethics Committee of Hannover Medical School (Study numbers: 2148-2014 and 2604-2014).Patients with evidence of HCC or cirrhosis were excluded from this study as well as pregnant participants or patients with psychiatric disorders.All patients received sofosbuvir/ledipasvir based on diagnosed chronic HCV infection genotype 1.The clinical characteristics of the chronic HCV patients and the healthy controls are summarized in Supporting Information Table 1.The HCV patients were studied longitudinally before, during, and after an 8-week DAA treatment regimen with a combination of sofosbuvir and ledipasvir.All patients cleared the HCV infection after 8 weeks.PBMCs from chronic HCV patients were collected, isolated, and cryopreserved at treatment start (BL), along therapy and 1 year after the end of therapy (fu48).PBMCs from healthy controls were isolated and cryopreserved in the same manner for comparative analysis.
Two additional groups of patients that received sofosbuvir/ledipasvir were used for the validation of NK-cell responsiveness and the analyses of plasma cytokine levels.Both groups contained similar numbers of women and men.For the NK-cellresponsiveness study, a total of ten HCV patients at the age of 46-67 years (six women/four men) and five healthy individuals (two women/three men) at the age of 29-56 years were included.For the cytokine study, plasma isolated from 13 HCV patients (6 women/7 men) at the age of 37-64 years was used.

PBMC and NK-cell isolation
Cryopreserved PBMCs from HCV patients and healthy donors were thawed, washed twice, and cells were stained for flow cytometric analysis and cell sorting.All stainings were performed for 40 min at 4°C in the dark using the following antibodies: LIVE/DEAD Fixable Green Dead cell Stain Kit, Thermo Fisher or DAPI; CD14-FITC, clone M5E2, BD Biosciences; CD19-FITC, clone HIB19, BD Biosciences; CD56-BV421, clone NCAM16.2,BD Biosciences; and CD3-BV786, clone UCHT1, BD Biosciences.PBMCs were washed twice after the staining and stored on ice until acquisition and sorting on a FACSAria Fusion cell sorter (BD Biosciences).PBMCs were sorted for live CD3 − /CD14 − /CD19 − /CD56 + NK cells.Flow cytometry data were analyzed with Flow Jo software V.9.8 (Treestar Inc., Ashland, OR, USA).After cell sorting, NK cells were once centrifuged for 5 min at 4°C with 1000 × g and resuspended in cold PBS.Then NK cells were centrifuged for 10 min at 4°C with 750g, the supernatant was removed, and NK cells were snap frozen at −80°C for subsequent proteomic analysis.

NK-cell proteome analyses (LC-MS/MS)
NK cells were purified from each patient and donor (ranging from 0.5 to 2.2 million, see Supporting Information Table 1).NK cells were then lysed in 100 μL of 8 M urea in 1 M triethylammonium bicarbonate complemented with benzonase for 30 min at 25°C.Cysteines were irreversibly modified after lysis, by adding and incubating 5 mM tris(2-carboxyethyl) phosphine (TCEP) and then 10 mM methyl methanethiosulfonate for each 30 min at 25°C.To digest proteins into peptides, 1 μg combined LysC/trypsin reagent (Promega) was added to each sample and incubated for 5 h at 25°C, following by dilution of urea by 700 μL MilliQ water and incubation at 37°C overnight.
Generated peptide samples were acidified with 10% formic acid at pH 3, and peptides were purified by using Oasis reverse phase columns (Waters Corporation) and afterward resuspended in 0.2% trifluoroacetic acid/3% acetonitrile.Purified peptides of each NK-cell patient sample were analyzed separately by a Dionex UltiMate 3000 RSLCnano LC system (Thermo Scientific) connected to an LTQ Orbitrap Fusion Tribrid Fourier transform mass spectrometer (Thermo Scientific).For each technical replicate, 800 ng peptides, which is the equivalent of about 100.000 primary NK cells, were used for peptide sequencing by LC-MS.Peptides were loaded onto a C18 precolumn (3 μm RP18 beads, Acclaim, 75 μm × 20 mm), washed for 5 min with 3.5% MS buffer B (0.1% formic acid, 80% acetonitrile) at a flow rate of 6 μL/min, and separated on a C18 analytical column (3 μm, Acclaim PepMap RSLC, 75 μm × 25 mm, Dionex) at a flow rate of 0.3 μL/min via a linear 180 min gradient from 3.5% MS buffer B (0.1% formic acid, 80% acetonitrile) to 31.3%MS buffer B, followed by a 40 min gradient from 31.3% to 62.5% MS buffer B and a 10 min gradient from 62.5% to 90% buffer B. The LC system was operated with the Chromeleon software (version 6.8, Dionex) embedded in the Xcalibur software suite (version 3.0.63,Thermo Scientific, Dreieich, Germany).Robust electrospray ionization of peptides was done by using a stainless steel emitter (Thermo Scientific).Eluting peptides were analyzed by a Thermo Orbitrap Fusion Lumos Tribrid (Thermo Scientific) mass spectrometer in a data-dependent acquisition mode.A full survey scan MS (from m/z 350 to 1500) was acquired in the Orbitrap with a resolution of 120,000.The AGC target for MS1 was set as standard with maximum injection times of 50 ms.Peptide fragmentation was carried out using the higher energy collisional dissociation mode in the ion trap.Ions with charge states 2, 3, or 4 were isolated in 3-s cycles, fragmented using higher energy collisional dissociation fragmentation with 34% normalized collision energy and detected at the ion trap with a mass isolation window of 1.6 (m/z).

Label-free protein quantification
The MS/MS raw data files were processed by MaxQuant [150] (version 1.6.2.3) with integrated search engine Andromeda searching against a UniProtKB/Swiss-Prot protein database of Homo sapiens.The parameters used are as follows: methyl methanethiosulfonate (H(2)CS) as fixed modifications, maximum number of modifications per peptide as 3, first search peptide tolerance as 10 ppm and main search peptide tolerance as 5 ppm, maximum mis-cleavage as 1, and the label-free quantification was turned on.Label-free quantification minimum ratio count was set to 1, and the rest of the parameters are used as default settings.Significantly regulated proteins presented in this study were manually validated in the raw spectra data.
The statistical assessment of MaxQuant data was achieved using software Perseus [151] (version 1.6.2.3), and figures were produced using the package ggplot2 [152] in R [153].Then the GO and KEGG annotation file were added.Then all the abundance values were log2 transformed, filtered to have at least 15 out of 18 valid values in at least one group.Pearson correlation coefficients between all donors were computed based on the NK-cell proteomes.Mean and median correlations within donor groups (healthy, before and after treatment) were determined, and correlations in different groups were compared based on the t-test.After filtering, the median of three technical replicates for each sample was taken, and the remaining missing values were replaced by imputed Gaussian distribution values (width: 0.3: downshift: 1.8).Based on the imputed data, Pearson correlations of all the NK-cell proteomes were calculated, and PCA was performed to visualize the projection of data sets.All the 4042 proteins were normalized with Z-score and then used for unsupervised hierarchical clustering in Perseus software, with distance set as "Euclidean," linkage as "Average," and constraint for column tree as "preserve order." Finally, to visualize the differential expression of proteins in chronic HCV patients before and after treatment, the Hawaii plot (multiple volcano plots) was created in Perseus software using the fold changes and Student's t-test p-values.p-Values were corrected for multi-comparisons by permutation-based FDR using the default settings of Perseus, which were then used as a threshold to define significantly regulated proteins.Then a stringent threshold (level A: minimal fold change s0 = 0.2, FDR = 0.01) and a less stringent threshold (level B: s0 = 0.1, FDR = 0.05) were applied to determine the significantly regulated protein.

Pathway enrichment of selected proteins
The global regulations of pathways and functions were analyzed based on the method from Pan et al. [154] and R script from Voigt et al. [26].For each comparison, all proteins were divided into five regulation groups according to both levels A and B regulation, and direction of fold change.Two-sided Fisher's exact test was then performed for assessing the p-value of all the GO and KEGG categories.NK-cell-related functions or pathways that were significantly (p < 0.05) enriched in at least one of the groups were filtered and used to construct a heatmap.

Protein quantification of plasma proteins
The patients' plasma samples were thawed, and 50 μL of each sample were pipetted for analysis via proximity extension analysis (PEA, Olink AB).A predefined panel simultaneously measuring 92 inflammation-related proteins in plasma was used (Target 96 Inflammation Panel).Briefly, in PEA, a pair of oligonucleotidelabeled antibodies, Proseek probes, bind to the target protein in the plasma sample.When the 2 Proseek probes are in close proximity, a new PCR target sequence is formed by a proximitydependent DNA polymerization event.This complex is subsequently detected and quantified using standard real-time qPCR.The generated normalized protein expression (NPX) unit is on a log2 scale where a larger number represents a higher protein level in the sample.The Ct values from a DNA extension control are subtracted from the measured Ct value of the analyte followed by an interplate control correction.Finally, a correction factor is subtracted to yield an NPX value that is log2 transformed.The limit of detection for each assay is calculated based on the negative controls plus three SDs.Detailed information regarding limit of detection, reproducibility, and validation is available on the manufacturer's homepage (Olink).The F-test for equality of variances was performed in R to compare the variances of these NPXs in female and male patients.The p-values were corrected by Benjamini-Hochberg for multiple testing.

Stimulation of PBMCs, staining, and spectral FACS analysis
PBMCs were thawed and were transferred into RPMI MEDIUM 1640 + GlutaMAX (Gibco) supplemented with 10% FCS (Serana) and 2 mM L-glutamine (Gibco).For each donor and per condition, 1 million cells were seeded in a 96-round bottom well plate (Thermo Scientific).Cells were placed in an incubator for 16 h at 37°C and 95% humidity.Next, K562 cells were added in a 10:1 ratio and incubated for 6 h at 37°C and 95% humidity.CD107a-BV421 (BioLegend) was present during the entire time of the assay.Brefeldin A (1:1000; BioLegend) and Monensin (1:1500; BioLegend) were added for the last 5 h of stimulation.After incubation, staining for flow cytometry was performed in the dark and as follows: Extracellular antibody mix (BD: CD16-BUV563 clone 3G8, BioLegend: CD56-AF647 clone 5.1H11, CD69-BV785 clone FN50, TIM-3-PE clone F38-2E2, CD3-PE/Fire640 clone SK7, CD4-PE/Fire640 clone SK3, CD19-PE/Fire640 clone HIB19, CD14-PE/Fire640 clone G3D3, CD244-BV605 clone C1.7) and Zombie Aqua (BioLegend) for live/death staining were added, and samples were incubated for 35 min at 4°C followed by a washing step with RPMI.Next, samples were fixed with Fix/Perm (from eBioscience Foxp3/Transcription Factor Staining Buffer Set, Invitrogen) for 35 min at room temperature followed by two washing steps with wash buffer from the same buffer set.Next, intracellular antibody mix (BD: IFN-γ-BV480 clone B27, Granzyme B-AF700 clone QA16A02) was added and incubated for 30 min at room temperature followed by two washing steps with RPMI.Samples were resuspended in RPMI prior to acquisition with the spectral FACS Cytek Aurora (Cytek).For FACS panel setup, single dye-coupled antibodies were measured followed by unmixing.The Cytek Aurora was equipped with the red, blue, and UV Laser and was calibrated with SpectroFlow QC beads.NK cells were gated as alive and CD14 − /CD19 − /CD3 − /CD4 − /CD56 + .Data evaluation and interpretation were performed using FlowJo (version 10. ).We thank all the participants for donating their blood samples in this study.We thank J. Wissing for excellent technical assistance.Open access funding enabled and organized by Projekt DEAL.

Conflict of Interest:
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

BFigure 1 .
Figure 1.Molecular phenotyping of HCV-infected patients.(A) Blood samples from six ledipasvir/sofosbuvir-treated HCV patients before (baseline, BL), along (1, 4, 8, 12, 20 weeks) and 48 weeks after therapy (fu48) were collected to characterize viral load (HCV RNA) and liver function (alanine aminotransferase [ALT] and FibroScan).Wilcoxon test (ALT) and unpaired t-test (FibroScan) were used to calculate significances (*p ≤ 0.05).(B)NK cells from the same patients (BL, fu48) and healthy controls were isolated from blood samples by flow cytometry and then analyzed by proteomics (label-free LC-MS/MS), measuring triplicates for each patient/donor sample.MaxQuant (MQ) and R were used to calculate patient-and donor-specific abundances of NK-cell proteins.Student's t-test was used to compare the expression of each protein across the groups, and the corresponding p-values were indicated in the example.We only included in this study information on proteins that have been robustly characterized by mass spectrometry in at least 15 of 18 patients/donors.By this, 4042 NK-cell proteins could be identified and quantified in direct-acting antivirals (DAA)-treated patients in comparison to healthy individuals.

Figure 2 .
Figure 2. Correlation and cluster analyses of CD56 + NK-cell proteomes from HCV patients.(A) Heatmap showing the Pearson correlations of NK-cell proteomes between donors.Correlations are generally high, that is, greater than 0.93, and red indicates lower correlations, green higher correlations.The red dashed boxes refer to correlations of donors from the same group and can be used as an indicator for the interindividual variance within each group (healthy, before and after treatment).The correlations in the group before treatment are lowest suggesting a higher interindividual variance.The blue dashed boxes highlight the correlations in comparison to healthy donors indicating the variance of NK-cell proteomes of HCV patients before as well as after therapy.(B) Unsupervised hierarchy clustering of NK-cell proteomes using Perseus.Total proteome data (all 4042 proteins) were normalized with Z-score and then used for unsupervised hierarchy clustering in Perseus.The symbols indicate sex (round for females, triangles for males) and HCMV status (green for negative, red for positive, gray for data not available).The ages of healthy donors and patients when donating blood are indicated in brackets.The red-blue color scale represents the normalized abundance of each protein, with red indicating high abundance and blue indicating low abundance.

Figure 3 .
Figure 3. Differentially expressed NK-cell proteins in chronic HCV patients before (baseline [BL]) and after (fu48) direct-acting antivirals (DAA)therapy in comparison to healthy donors.Volcano plots of NK-cell protein expressions in HCV patients before (A) and after (B) treatment in comparison to healthy individuals.Colored dots indicate significantly regulated proteins (s0 = 0.2, false discovery rate [FDR] = 0.01).Selected top-regulated proteins are annotated.(C) Sankey diagram visualizing the conversion of 4042 NK-cell protein regulations as defined before (A) and after (B) treatment.

Figure 4 .
Figure 4. Sustained dysregulation of RNA processing proteins in cured HCV patients.Out of the 4042 proteins, 368 proteins (9%) were annotated with "RNA processing" (GOBP), out of which 100 proteins were significantly dysregulated (s0 = 0.2, false discovery rate [FDR] = 0.01) in the NK cells of chronically HCV-infected patients (baseline [BL]) compared with healthy controls.After direct-acting antivirals (DAA) treatment, 84 (84%) of the dysregulated RNA processing proteins in NK cells from chronic HCV patients were sustained dysregulated in the NK cells from cured patients.Only 16 proteins (16%) were rescued in NK cells of cured patients to the level of healthy controls.

Figure 6 .
Figure 6.Regulation of lytic vesicle and vesicle transport proteins in HCV therapy.Box and jitter plots showing protein levels of lytic vesicle-(A) and vesicle transport-associated proteins (B) from healthy individuals (H) and HCV patients before treatment (baseline [BL]) and after treatment (fu48).Each data point indicates the median value of three technical replicates.The color of the dot indicates different donors, and the shape of the dot indicates sex of these donors.Box plots show the median and quantiles for each group.The p-values of Student's t-test are indicated (n = 6) and together with p-values adjusted for multiple testing based on Benjamini-Hochberg correction allow the comparison between all three groups at the single protein level (Supporting Information Table5).

Figure 7 .
Figure 7. Abundances of activating and inhibitory NK-cell receptors in HCV patients.NK-cell proteome analyses characterized the abundances of activating and inhibitory NK-cell receptors in HCV patients before (A and C) and after ledipasvir/sofosbuvir treatment (B and D).The surface receptors are labeled based on their potential role in NK cells: activating receptors, integrins, inhibitory receptors, and unclear (receptors regulating both activation and inhibitory functions).Increased (↑) and decreased protein abundances (↓) as well as nonsignificantly altered protein levels (≈) as observed in the NK-cell proteomes are indicated in black.Regulatory information by flow cytometry from a control patient group are indicated in red, accordingly.Figures were created with BioRender (free version) and Inkscape (version 1.2).
8.2).Acknowledgments: This work was supported by the China Scholarship Council (CSC) Grant #201506220146 (to W.B.), the Office of the China Postdoctoral Council (OCPC) grant (to W.B.), the International Research Training Group 1273 (to J.H.), the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)-under Project number 158989968-SFB 900 (to M.C. and H.W.) and under Germany's Excellence Strategy-EXC 2155-project number 390874280 (to M.C), the German Federal Ministry of Education and Research (BMBF) within the framework of the CompLS research and funding concept (grant 031L0294C) (to M.C.), and the German Centre for Infection Research (DZIF) (to M.C.

Author contributions:
Lothar Jänsch and Heiner Wedemeyer conceived the study.Lothar Jänsch, Julia Hengst, Anke Kraft, Marco van Ham, Markus Cornberg, and Heiner Wedemeyer designed experiments.Wenjie Bi, Jasmin Mischke, Sophie Engelskircher, Moana Witte, and Julia Hengst performed experiments and analyzed data.Wenjie Bi, Marco van Ham, and Lothar Jänsch wrote the original manuscript.Wenjie Bi, Anke Kraft, Frank Klawonn, Marco van Ham, Markus Cornberg, Heiner Wedemeyer, Julia Hengst, and Lothar Jänsch reviewed and edited the manuscript.Ethics statement and patient consent: The study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethics Committee of Hannover Medical School (Study numbers: 2148-2014 and 2604-2014).Informed consent was obtained from all participants.

Table 1 .
Dysregulated surface receptors in NK cells of chronically HCV-infected patients before and after therapy.

Table 1 .
(Continued) Surface proteins that have significantly altered expression in chronic HCV infected patients before (BL) and 1 year after (fu48, in bold) ledipasvir/sofosbuvir therapy as compared with healthy donors (H).Abbreviation: BL, baseline. Note: