Serum HBV pregenomic RNA exhibited opposite associations with NKdim and NKbright cell immunity in treatment-naïve chronic hepatitis B patients

Abstract Hepatitis B virus (HBV) pregenomic RNA (pgRNA) is a new biomarker that reflects HBV replication, but its relationship with natural killer (NK) cell immunity in chronic hepatitis B (CHB) is unknown. We assessed serum HBV pgRNA levels in 323 CHB patients by reverse transcription-polymerase chain reaction, assessed cytokine production and activation and inhibitory markers of NK cells by flow cytometry, and measured serum cytokines by enzyme-linked immunosorbent assays (ELISAs). Among the different CHB phases, the serum HBV pgRNA level was highest in the immune-tolerant (IT) and immune-active (IA) phases. Regarding NK and NKdim cells, HBV pgRNA was negatively associated with frequencies, but positively associated with NKp44 and NKp46 expression (activation markers). Regarding NKbright cells, serum HBV pgRNA was positively associated with frequency and programmed cell death protein 1 (PD1) expression (inhibitory marker), but negatively associated with NKp44 and NKp46. Serum HBV pgRNA was not associated with NKp30 (activation marker) on NK cells or subsets. Lastly, serum HBV pgRNA was positively correlated with the levels of serum IL-7 and IL-12P40 (NK cell-promoting cytokines) and negatively correlated with serum prostaglandin E2 (PGE2) level (which negatively regulates NK cells). In conclusion, we found varied relationships between serum HBV pgRNA and NK cells and subsets, indicating that HBV pgRNA may play a complicated role in NK cell-related immunity, providing new information on HBV and host immunity.


Introduction
Hepatitis B virus (HBV) is one of the most common viruses that can infect the liver, especially in Asia. Chronic HBV infection can cause prolonged liver injury, which may progress to liver cirrhosis, hepatic failure, and even hepatocellular carcinoma [1]. HBV is a non-cytopathic virus and its mechanism of liver injury is thought to involve inducing immune inflammation [2].
Natural killer (NK) cells belong to the innate immune system and play crucial antitumor and antivirus roles. They directly target cells, causing apoptosis or osmotic cell lysis via cytotoxic mechanisms [3]. Based on the expression of surface marker CD56, NK cells can be subtyped into CD56 bright (NK bright ) cells with high expression of CD56 and CD56 dim (NK dim ) cells with low expression of CD56. The majority of NK cells in peripheral blood are NK dim cells with typical cytotoxicity and low cytokine expression. In contrast, NK bright cells, which are usually non-cytotoxic and express high levels of cytokines, are significantly increased in inflammatory sites [4].
NK cells have an important and early role in HBV infection. In chimpanzee models of acute HBV infection, HBV DNA is significantly reduced by non-T cells, especially NK cells, long before liver infiltration

Quantification of serum HBV pgRNA
The serum HBV pgRNA level was assessed using an HBV pgRNA Kit (SUPBIO, Guangzhou, China). Briefly, serum HBV RNA was isolated, treated with DNase I, and reverse transcribed according to the manufacturer's instructions. qPCR was then performed without using reverse transcriptase. The detection limit was 25 copies/ml.

Statistical analysis
Serum HBV pgRNA was expressed as log 10 copies/ml, and presented as median (lower and upper quartiles). To compare the patient groups, the Wilcoxon signed-rank test was used for continuous variables, Mann-Whitney U test was used for comparison between two groups, and Kruskal-Wallis (K-W) test were for comparison among multiple groups (if the P-value of K-W test was significant, then Bonferroni method was used for multiple comparison). The χ 2 test was used for categorical variables. Correlations were assessed using Spearman correlation analysis. Additionally, the associations of serum HBV pgRNA with various factors were determined by linear regression analyses. All the analyses were performed using SPSS version 23 (IBM Corp., U.S.A.) and P<0.05 was considered significant.

Associations of serum HBV pgRNA level with clinical-virological characteristics
Based on international guidelines, the patients were divided into IT (n=29), IA (n=190), IC (n=48), and GZ (n=56) groups. Demographic and clinical-virological characteristics of the patients and HCs are presented in Table 2. The median serum HBV pgRNA level in the IT, IA, IC, and GZ groups was 5.54 (0, 7.81), 5.71 (0, 9.00), 1.39 (0, 8.55), and 1.76 (0, 7.17) log 10 copies/ml, respectively. It was significantly higher in the IT and IA groups than the IC and GZ groups using K-W test and Bonferroni method analyses (P<0.001) ( Figure 1A).
Consistent with the distribution of serum HBV pgRNA across the patient groups and its positive correlations with clinical-virological characteristics (Figure 1), the HBsAg level and HBV DNA load were also higher in the IT and IA groups than the IC and GZ groups, and highest in the IT group using the K-W test and Bonferroni method analyses. Liver inflammation (ALT and AST) and fibrosis (FibroScan value) were also highest in the IA group among the four groups. TBIL and ALB were higher in the IT group than the IC and GZ groups. While GLB was not different among the groups (Supplementary Figure S1).

Associations with NK cell and subset frequencies
The immune system plays important roles in virus clearance and liver injury during chronic HBV infection. In our previous study, we found that serum HBV pgRNA was associated with T helper (Th) cell immunity. Therefore, here we explored the relationship between HBV pgRNA and NK cell immunity.
Firstly, the associations of serum HBV pgRNA with NK cell and subset frequencies were analyzed. Serum HBV pgRNA was negatively correlated with NK and NK dim cell frequencies, but positively correlated with NK bright cell frequency (r=−0.188, −0.192 and 0.315, respectively, all P<0.001) in the Spearman correlation analysis (Figure 2A-C). Both univariate and multivariate linear regression analyses showed that serum HBV pgRNA was negatively associated with NK cell frequency (B = −0.073 and −0.658, P=0.001 and 0.022, respectively) but positively associated  with NK bright cell frequency (B=2.489 and 2.885, both P<0.001). It was inconsistently associated with NK dim cell frequency based on univariate (B=−0.076, P<0.001) and multivariate (B=0.602, P=0.039) linear regression analyses (Supplementary Table S2). In summary, serum HBV pgRNA tended to be negatively correlated with NK and NK dim cell frequencies, but positively correlated with NK bright cell frequency.
To gain full understanding of the relationship between serum HBV pgRNA and NK cell immunity, we analyzed the data further using the K-W test and Bonferroni method, assessing the differences in NK cell frequencies among patient groups. NK and NK dim cell frequencies were significantly lower in the IA group than the GZ group, while NK bright cell frequency was higher in the IA group than the IC and GZ groups, which was consistent with the varying correlations between serum HBV pgRNA and NK cell and subset frequencies (Supplementary Figure S2A-C).

Associations with antiviral cytokines produced by NK cells and subsets
IFN-γ and TNF-α are important antiviral cytokines produced by NK cells in HBV infection, so we explored the associations between serum HBV pgRNA and these cytokines. The results showed that serum HBV pgRNA was positively correlated with IFN-γ produced by NK, NK bright , and NK dim cells (r=0.185, 0.214, and 0.185, and P=0.0016, 0.0009, and 0.0042, respectively), but there were no correlations with TNF-α produced by NK, NK bright , or NK dim cells (P>0.05) in the Spearman correlation analysis ( Figure 2D Table S2).
Additionally, by the K-W test and Bonferroni method, IFN-γ levels produced by NK cells and subsets were higher in the IA group than the IT and GZ groups and that of NK bright cells was also higher in the IA group than the IC group. The TNF-α level produced by NK cells and subsets did not differ among the groups, which is consistent with the lack of correlation between serum HBV pgRNA and TNF-α level (Supplementary Figure S2D-I).  Table S3). In summary, varying associations were found between serum HBV pgRNA and the activation markers on different NK cells.

Associations with activation markers on NK cells and subsets
In addition, the expression of the activation marker NKp44 on NK and NK dim cells was higher in the IA group than the GZ group, and NKp30 on NK bright cells was lower in IC group than the IA and GZ groups using the K-W test and Bonferroni method analyses. There were no differences in NKp46 on NK cells or subsets among the four groups (Supplementary Figure S3).

Associations with inhibitory markers on NK cells and subsets
NKG2A, PD1, Tim3, and LAIR1 are inhibitory markers of NK cells. The associations of serum HBV pgRNA with these markers on NK cells and subsets were analyzed. Serum HBV pgRNA was positively correlated with PD1 on NK, NK bright , and NK dim cells (r=0.332, 0.341 and 0.339, and P<0.0001, respectively), but negatively correlated with Tim3 on NK, NK bright , and NK dim cells (r=−0.320, −0.303 and −0.325, and P<0.0001, respectively) and LAIR1 on NK and NK dim cells (r=−0.209 and −0.220, and P=0.002 and 0.0012, respectively) in the Spearman correlation analysis. However, it was not correlated with NKG2A on NK, NK bright , and NK dim cells (P>0.05) ( Figure 4A-L). Serum HBV pgRNA was not only positively associated with PD1 on NK bright cells in both the univariate and multivariate linear regression analyses (B=0.030 and 0.001; both P<0.01), but also positively associated with NKG2A and Tim3 on NK bright cells in multivariate linear regression analysis (B=0.116 and 0.048; P=0.001 and 0.020, respectively). It was also negatively associated with LAIR1 on NK and NK dim cells in univariate linear regression analysis (B=−0.019 and    Table S4). Thus, serum HBV pgRNA tended to be positively associated with inhibitory markers on NK bright cells, but negatively associated with inhibitory markers on NK dim cells.
Furthermore, regarding inhibitory markers, PD1 expression on NK cells and subsets was higher in the IA group compared with the IT, IC and GZ groups, while Tim3 expression was lowest in the IA group. No between-group differences in LAIR-1 or NKG2A on NK cell or subsets were found, except that NKG2A expression on NK bright cells was higher in the IT group than the IA and IC groups using the K-W test and Bonferroni method analyses (Supplementary Figure S4).

Associations with serum cytokines and chemokines
IL-7, IL-12, and PGE2 are critical molecules for the development and function of NK cells. Therefore, we explored the associations of serum HBV pgRNA with the serum levels of these cytokines (Supplementary Figure S5A-C). Serum HBV pgRNA was positively correlated with IL-7 and IL-12P40 (r=0.230 and 0.296, and P=0.0238 and 0.0032, respectively), which promote NK cell development and cytotoxicity. However, it was negatively correlated with PGE2 (r=−0.247 and P=0.0015), which inhibits NK cell function and activation. Additionally, the serum IL-7 level was highest in the IT group, the IL-12 level was higher in the IT and IA groups than the IC group and the PGE2 level was lower in the IA group than the IT and IC groups using the K-W test and Bonferroni method analyses (Supplementary Figure S5DF).

Discussion
The present study explored the relationship between serum HBV pgRNA and NK cell immunity in CHB patients. We found that serum HBV pgRNA had varied relationships with different NK cell subsets. Although serum HBV pgRNA was negatively associated with NK and NK dim cell frequency, it was positively associated with activation markers (NKp44 and NKp46) and negatively associated with inhibitory markers on these cells. In contrast, it was positively associated with NK bright cell frequency and cytokine production of NK bright cells but negatively associated with activation markers (NKp44 and NKp46) and positively associated with inhibitory markers on these cells. Moreover, it was positively associated with serum IL-7 and IL-12P40 (NK cell-promoting cytokines), but negatively associated with serum PGE2 (which negatively regulates NK cells).
NK cells in the peripheral blood are divided into two major subsets according to CD56 expression, referred to as CD56 dim (NK dim ) cells and CD56 high (NK bright ) cells. Most NK cells are NK dim , while <15% are NK bright [17]. NK cells defend against virus infection and tumors, and the two subsets of NK cells play different roles in immunity. Upon stimulation by T cells, dendritic cells (DCs) and monocytes, NK bright cells promptly produce high levels of chemokines and cytokines to regulate the immune response and act as a bridge between innate and adaptive immunity [18][19][20]. However, NK bright cells do not readily express the intracellular cytotoxicity molecules granzymes A and B and perforin [21]. Therefore, they are thought to have poor cytotoxicity and more immune-modulatory activity. In contrast, NK dim cells contain more cytotoxic granules and are more efficient regarding cytotoxicity and antibody-dependent cellular cytotoxicity (ADCC) [22][23][24]. In this study, serum HBV pgRNA had different relationships with NK, NK dim and NK bright cell parameters, which might be related to the different characteristics of these NK cell subsets.
Laroni et al. found that NK bright cells could be induced to suppress CD4 + T-cell proliferation via the use of natural cytotoxicity receptors (NCRs) [25]. Morandi et al. found that NK bright cells decreased T-cell proliferation by releasing adenosine [26]. NK bright cells were also found to be cytotoxic toward T cells in daclizumab-treated patients [27]. T cells, especially virus-specific CD8 + T cells, play a crucial role in the antiviral response to HBV and pathogenesis of HBV infection. NK cells have been shown to have a negative effect on T cell immunity during chronic HBV infection [28][29][30]. Boni et al. observed an increase in NK bright cells and a low level of HBV-specific T cells in patients with chronic HBV infection, and the authors proposed that NK bright cells might promote HBV chronicity by killing T cells in a TNF-related apoptosis-inducing ligand (TRAIL)-dependent process. After antiviral therapy reduced HBV DNA and cleared HBsAg, HBV-specific T cells were no longer suppressed by NK cells [31]. Peppa et al. showed that activated NK cells highly expressed TRAIL and thereby hindered the function of HBV-specific T cells and promoted their apoptosis in HBV-infected inflammatory livers, preventing antiviral immunity in CHB [32]. Our previous study showed that Th1 immunity and cytotoxic lymphocytes (CTLs) were negatively correlated with the serum HBV pgRNA level [15]. In the current study, HBV pgRNA had a positive relationship with NK bright cell frequency and cytokine production. Therefore, we speculate that HBV pgRNA might suppress Th1 cell antiviral immunity by enhancing NK bright cell functions, which would contribute to HBV persistence in CHB patients. However, it is hard to explain why serum HBV pgRNA was positively associated with activation markers on NK and NK dim cells. This needs further research. [33]. NK cells contribute to HBV clearance [34] and their early activation can enhance HBV clearance in woodchuck models and individuals with acute HBV infection [35,36]. During acute HBV infection, antiviral cytokines mainly come from NK cells that accumulate in the liver [37]. However, during chronic HBV infection, NK cell immunity is gradually reduced, followed by enhancement of the adaptive immune response [9]. Zhang et al. found a lower level of circulating NK cells in CHB patients than HCs, especially during the IA phase of CHB, which might be due to the NK cells being more prone to apoptosis under the proinflammatory conditions during this phase, and HBV may significantly inhibit NK cell proliferation [38]. Tjwa et al. demonstrated that the NKp30 + NK cell frequency dramatically decreased in CHB patients, which was associated with the HBV level [8]. We observed that serum HBV pgRNA level, a new HBV marker, was negatively associated with NK and NK dim cell frequency, suggesting that serum HBV pgRNA might also had negative association with NK cell proliferation.

NK cells recognize cells infected by viruses and clear the virus infection by producing cytokines
Activation and inhibition of NK cells are strictly regulated to ensure full reactivity, appropriate immune monitoring, and self-tolerance, while avoiding excessive activity that can lead to inflammation or autoimmune diseases. A shift in the balance between activation and inhibition regulates NK cell responses. NK cells up-regulate inhibitory signals when interacting with peripheral tissues. When they come across target cells, the increase in activating receptor stimulation and/or the lack of inhibitory signals shifts the balance towards NK cell activation [39]. HBV infection can change the activation state and receptor expression pattern of NK cells. The up-or down-regulation of certain activation and inhibitory markers was reportedly correlated with the serum HBV DNA level [8]. Li et al. demonstrated that low HBV load and liver inflammation up-regulated NKp46 on NK cells in CHB, which facilitated inhibition of HBV replication and liver inflammation [40]. Alter et al. revealed that NKp30 and NKp46 down-regulation on NK cells was strongly related to HCV clearance [41], suggesting an indirect link between these natural cytotoxic receptors on NK cells and virus clearance. Similar to these previous results, we also found associations of serum HBV pgRNA with activation and inhibitory markers on NK cells. It was very interesting to observe that though the serum HBV pgRNA level was negatively associated with NK and NK dim cell frequency, it was positively correlated with activation markers on these cells. Regarding NK bright cells, the serum HBV pgRNA level was positively associated with NK bright cell frequency and cytokine production, but negatively associated with activation markers and positively associated with inhibitory markers on the cells, indicating a complex association with NK bright cells. All of these results demonstrated that the serum HBV pgRNA level had varied relationships with NK and subset cells, indicating that although HBV pgRNA might suppress NK and NK dim cell proliferation, it may promote the activation of these cells, while it may promote both the proliferation of NK bright cells and their function (in some respects, though it is negatively and positively associated with activation and inhibitory markers, respectively).
In mouse models, IL-7 is confirmed as an important cytokine for NK cell development in the thymus (with decreased NK cell generation during IL-7 production deficiency), but it does not affect NK cell development in the bone marrow [42]. Thymic NK cells regulate T and B cells in an IL7-dependent manner [43]. IL-12 induces NK cells to produce cytokines and enhances NK cell cytotoxicity, and it slightly increases the proliferation of resting peripheral blood NK cells [44]. In contrast, PGE2 inhibits NK cell function and activation and can promote antigenic immune escape via PGE2 receptors 2 and 4 (EP2 and EP4) on NK cells [45,46]. Bonavita et al. reported that PGE2 inhibited the early activation of NK cells and subsequently blocked T-and B-cell recruitment [47]. We also found that serum HBV pgRNA was associated with these molecules, which may partly explain how HBV pgRNA affects NK cell immunity.
In summary, we showed that the serum HBV pgRNA level had opposite associations with NK dim and NK bright cell frequency and function, suggesting that HBV pgRNA may play a complicated role in regulating NK cell-related immunity. Although the NK and NK dim cell frequencies were suppressed, serum HBV pgRNA was still positively associated with NK bright cells, which might limit T-cell responses, helping HBV to escape host immunity. However, the study is an observational study, and further experiments are still required to understand the molecular mechanism behind these findings. Understanding the molecular mechanism by which HBV evades host immune responses might help to develop new immunotherapeutic strategies to cure chronic HBV infection.

Data Availability
The data presented in the present study are available on request from the corresponding author.