Natural Killer Cells and Host Defense Against Human Rhinoviruses Is Partially Dependent on Type I IFN Signaling

Rhinovirus (RV), the causative agent of the common cold, causes only mild upper respiratory tract infections in healthy individuals, but can cause longer lasting and more severe pulmonary infections in people with chronic lung diseases and in the setting of immune suppression or immune deficiency. RV-infected lung structural cells release type I interferon (IFN-I), initiating the immune response, leading to protection against viruses in conjunction with migratory immune cells. However, IFN-I release is deficient in some people with asthma. Innate immune cells, such as natural killer (NK) cells, are proposed to play major roles in the control of viral infections, and may contribute to exacerbations of chronic lung diseases, such as asthma. In this study, we characterized the NK cell response to RV infection using an in vitro model of infection in healthy individuals, and determined the extent to which IFN-I signaling mediates this response. The results indicate that RV stimulation in vitro induces NK cell activation in healthy donors, leading to degranulation and the release of cytotoxic mediators and cytokines. IFN-I signaling was partly responsible for NK cell activation and functional responses to RV. Overall, our findings suggest the involvement of NK cells in the control of RV infection in healthy individuals. Further understanding of NK cell regulation may deepen our understanding of the mechanisms that contribute to susceptibility to RV infections in asthma and other chronic lung diseases.


INTRODUCTION
Respiratory viruses, particularly rhinoviruses (RV), typically cause only mild, self-limited infections in healthy individuals, but the consequences of infection can be much more serious in people with chronic lung diseases and in the setting of immune suppression or immune deficiency (Glezen et al., 2000;Versluys et al., 2010;Costa et al., 2011;Jacobs et al., 2013). For example, RV are implicated in both the induction of asthma and exacerbations of established asthma, representing the most common cause of asthma-related death (Longini et al., 1984;Beasley et al., 1988;Gern et al., 1996;Papadopoulos et al., 2002a;Papadopoulos et al., 2002b;Hansbro et al., 2008). Asthmatics are no more likely than healthy individuals to develop a cold, but are substantially more likely for the infection to spread to the lower respiratory tract, causing more severe symptoms (Corne et al., 2002). This raises important questions regarding the mechanisms by which healthy people are able to mount an effective host response to RV. Understanding these processes that occur in healthy individuals might help to explain why RV infections can have such severe outcomes in people with asthma.
Bronchial epithelial cells (BECs) are the primary point of contact between infecting respiratory viruses and the host (Bals and Hiemstra, 2004). During infections, BECs release type I (IFN-I) and III IFNs, which are integral for the induction of innate immune response, conferring protection against viruses in conjunction with migratory immune cells (Hansbro et al., 2008;Message et al., 2008). Some investigators report that RV-induced IFN-I release is deficient in asthma Corne et al., 2002;Wark et al., 2005;Hansbro et al., 2008), though other reports have been unable to confirm these findings (Sykes et al., 2014;Xi et al., 2015). Cell-mediated immunity to RV is also important in the control and clearance of infection and may be abnormal in asthmatics (Message and Johnston, 2001). Innate immune cells, such as natural killer (NK) cells have been proposed to play a major role in both asthma (Message et al., 2008;Culley, 2009), and the control of viral infections in general (Biron et al., 1999;Biron and Brossay, 2001;Barnig et al., 2013;Lam and Lanier, 2017); however, little is known about the role NK cells play in RV infections.
NK cells are large, granular lymphocytes involved in the innate immune response to infected or malignant cells (Trinchieri, 1989;Alter et al., 2004;Vivier et al., 2008;Bryceson et al., 2010;Montaldo et al., 2013). Without prior exposure or antigen sensitisation, NK cells are able to detect and kill infected or aberrant cells, and produce chemokines and cytokines that prime and shape adaptive immunity (Kay, 1985;Colucci et al., 2003;French and Yokoyama, 2003;Bryceson et al., 2010). Interestingly, circulating NK cells appear to reflect the maturation status of lung-resident NK cells, with a phenotype capable of both cytokine production and cytotoxicity (Ivanova et al., 2014).
Human NK cells are identified as CD56 + lymphocytes that lack markers specific to other lymphocytes, such as CD3 on T cells, CD14 on monocytes, and CD19 on B cells (Herberman and Ortaldo, 1981;Robertson and Ritz, 1990;Bancroft, 1993;D'Arena et al., 1998;Papamichail et al., 2004). Two distinct subsets of human NK cells have been identified and classified based on the cell-surface density of CD56; these subsets appear to have distinct phenotypic and functional properties (Lanier et al., 1986;Cooper et al., 2001a;Sedlmayr et al., 2004;Moretta, 2010). The CD56 dim subset of NK cells appear more naturally cytotoxic, but have a reduced ability to produce cytokines upon stimulation (Cooper et al., 2001a;Di Santo, 2008); while the CD56 bright subset of NK cells are distinguished by potent production of both Th1 and Th2 cytokines in response to stimulation, but have reduced cytotoxic capacity (Lanier et al., 1986;Cooper et al., 2001b).
NK cell recognition of target cells relies on an assortment of both activating and inhibiting receptors that detect microbial products, cytokines, stress signals, and MHC class I molecules (Lanier, 2000;Smyth et al., 2000;Moretta et al., 2001;Radaev and Sun, 2003). NK cell activation does not rely on an individual receptor, but on a combination of activating and inhibiting signals, with activation occurring when different combinations of receptors are ligated to overcome a threshold of activation (Bryceson et al., 2005;Lanier, 2008). NK cell surface receptor repertoire has been reported to vary between subsets, and thus subset activation may vary based on the stimulus (Cooper et al., 2001a;Sedlmayr et al., 2004).
NK cells can lyse target cells through the directed exocytosis of cytolytic granules containing perforin and granzyme at the immunological synapse between the NK cell and target cell (Robertson and Ritz, 1990;Cooper et al., 2001a;Artis and Spits, 2015). In particular, granzyme B (GzymB) has been shown to be required for the rapid induction of NK cell-mediated apoptosis of target cells (Shresta et al., 1995). The release of these cytolytic granules can be indicated through the presence of cell surface CD107a (Alter et al., 2004;Bryceson et al., 2010).
With the strong association between respiratory viral infections and asthma exacerbations, it has been speculated that inappropriate immune activation, possibly due to defective IFN-I response, may be involved in the susceptibility of asthmatics to more persistent and severe infections (Wark et al., 2005;Message et al., 2008;Culley, 2009). Therefore, this study aimed to determine whether RV can activate NK cells, and the extent to which NK cell activation and function in vitro are IFN-I dependent.

Participants
All volunteers completed a detailed questionnaire regarding respiratory symptoms, prior medical diagnoses, and medication use. Healthy participants had no symptoms or prior self-reported physician diagnoses of respiratory disease (including asthma) and had not experienced respiratory infection symptoms within the preceding month. All participants underwent skin prick testing (SPT) against a panel

Rhinovirus Generation and Titration
RV16 stocks were generated by passage in Ohio HeLa cells, as previously described (Sanders et al., 1998). RV16 was purified over Optiprep gradient (Sigma-Aldrich) as described by Xi et al. (2015). Viral titer was determined using TCID50, and RV16 at an MOI of 1 was used in culture (Pritchard et al., 2012).

ELISA
Cell-free culture supernatants were collected and stored in deep-well plates at −30°C, until required. IFNg and GzymB concentrations in culture supernatant were measured using ELISA. IFNg assays used commercially available paired Abs and recombinant cytokines (BD Pharmingen; limit of detection = 3.91 pg/ml). GzymB assays used a commercially available ELISA kit (R&D Systems; limit of detection = 39.1 pg/ml), according to the manufacturer's instructions.

Statistical Analysis
Data were analyzed using Graphpad Prism (version 8.1) using Wilcoxon matched-pairs signed rank tests to compare differences between paired samples. Data are represented as median (interquartile range -IQR). A p value of <0.05 was considered statistically significant.

RV16, But Not IFN-I Signaling, Causes Minor Changes in NK Cell Populations
To determine whether RV could activate NK cells, and the extent to which NK cell activation and functional changes are IFN-I dependent, PBMCs from healthy volunteers were cultured for 1 h in the presence or absence of B18R to block IFN-I signaling, prior to culture for 24 h in the presence or absence of RV16 stimulation. Following this, flow cytometry was used to identify and evaluate the frequency of CD56 + NK cells, along with CD56 dim and CD56 bright NK cell subsets ( Figure S1).
Our results demonstrate that neither RV16 stimulation, nor blocking of IFN-I signaling, altered the frequency of lymphocytes ( Figure 1A), though RV16 induced a small increase in CD56 + NK cell frequency that was less apparent when IFN-I signaling was blocked ( Figure 1B). Stimulation with RV16 induced minor changes in the distribution of CD56 dim and CD56 bright NK cell subsets, with a decrease in the frequency of CD56 dim NK cells and a corresponding increase in the frequency of CD56 bright NK cells ( Figure 1C). These changes in the distribution of the CD56 dim and CD56 bright NK cell subsets were not substantially altered when IFN-I signaling was blocked.
RV16 Induces Intense NK Cell Activation, Which Is Partly Dependent on IFN-I Signaling NK cell activation was assessed based on cell surface CD69 expression. Both an increase in the frequency of CD69 + cells and the expression intensity of CD69 can be used to assess NK cell activation (Draghi et al., 2007;Du et al., 2010;Souza-Fonseca-Guimaraes et al., 2012;Barnig et al., 2013). RV16 stimulation of PBMC for 24 h led to substantial and significant increases in the proportion of NK cells expressing CD69, though this occurred to a lesser extent in the absence of IFN-I signaling (Figure 2A, left). Blocking IFN-I signaling had a larger impact on the percentage of CD69 + cells in the CD56 bright subset ( Figure  2A, right) than in the CD56 dim subset (Figure 2A, middle). RV16 also increased the median fluorescent intensity (MFI) of CD69 surface expression on NK cells ( Figure 2B), especially the CD56 dim subset ( Figure 2B, middle).

RV16 Induces NK Cell Cytolytic Granule
Release Which Is Partly Dependent on IFN-I Signaling NK cell degranulation was assessed based on CD107a surface expression. CD107a lines the cytolytic granules that are secreted during cytolysis, and appearance at the cell surface is upregulated following stimulation, correlating with target cell lysis (Alter et al., 2004;Bryceson et al., 2005;Bryceson et al., 2010). The presence of CD107a at the cell surface is an indicator of NK cell release of cytotoxic granules (Haworth et al., 2011;Barnig et al., 2013). RV16 stimulation of PBMC cultures resulted in significant increases in the proportion of CD56 + NK cells expressing cell surface CD107a ( Figure 3A, left). Blocking of IFN-I signaling in vitro led to a lower frequency of degranulating CD56 + NK cells, both in the presence and absence of RV16 stimulation. These trends were also observed in both the CD56 dim ( Figure 3A, middle) and CD56 bright NK cell subsets ( Figure 3A, right). Stimulation with RV16 had no significant effect on the MFI of CD107a surface expression on CD56 + NK cells ( Figure 3B, left). There were no significant changes in the level of CD107a surface expression in the CD56 dim NK cell population ( Figure 3B, middle). However, in the CD56 bright NK cell population ( Figure 3B, right), blocking of IFN-I signaling resulted in a small increase in the MFI of CD107a surface expression, in both the presence and absence of RV16 stimulation ( Figure 3C).

RV16 Induces Small Changes in Both the Percentage of GzymB-Producing NK Cells and Their Intracellular GzymB
The cytolytic granules released through directed exocytosis contain proteins, such as GzymB, which is important in NK cell-mediated apoptosis of target cells (Shresta et al., 1995). RV16 stimulation of PBMC cultures resulted in small, but statistically significant, increases in the proportion of CD56 + NK cells producing GzymB ( Figure 4A, left). This was most apparent in the CD56 bright NK cell population ( Figure 4A, right). Blocking of IFN-I signaling in vitro with B18R did not significantly alter the frequency of GzymB-producing NK cells ( Figure 4A). RV16 stimulation also increased the intracellular GzymB MFI of CD56 + NK cells ( Figure 4B, left), with changes observed in both the CD56 dim ( Figure 4B, middle) and CD56 bright NK cell subsets ( Figure 4B, right). Blocking of IFN-I signaling caused only minor changes in these responses.

RV16 Induces IFNg-Producing NK Cells in an IFN-I Dependent Manner
NK cells are known to produce IFNg in response to other viruses, including influenza viruses (Du et al., 2010). IFNg activates multiple pathways associated with direct antiviral functions and immunoregulation, and promotes downstream protective immune responses (Biron and Brossay, 2001). Herein, we have found that RV16 stimulation resulted in a significant increase in the frequency of IFNg-producing CD56 + NK cells ( Figure 5A, left), and in the intracellular IFNg MFI ( Figure 5B, left). These trends were observed in both CD56 dim ( Figure 5A, middle; 5B, middle) and CD56 bright NK cell subsets ( Figure 5A, right; 5B, right). The increase in frequency of IFNg-producing cells due to RV16 stimulation was most prominent in the CD56 bright NK cell subset ( Figure 5A, right). When IFN-I was blocked, there was a significant decrease in the frequency of RV16-stimulated IFNgproducing CD56 + NK cells, which was reflected in both the CD56 dim and CD56 bright NK cell subsets ( Figure 5A). This indicated that the RV16-stimulated increase in the frequency of IFNg-producing NK cells was only partially dependent on IFN-I signaling.

IFN-I Signaling Is Involved in the RV16-Stimulated Release of GzymB and IFNg
ELISA techniques were used to quantify RV16-stimulated GzymB and IFNg release. Herein, we found that RV16 stimulation of PBMCs resulted in significant GzymB and IFNg release into the culture supernatants ( Figure 6). When IFN-I signaling was blocked, there was a decrease in the concentration of released GzymB, in both the presence and absence of RV16 stimulation ( Figure 6A). In the presence of RV16 stimulation, blocking of IFN-I signaling also decreased IFNg concentrations in culture supernatant ( Figure 6B).

NK Cells Make a Large Contribution to the Production of GzymB and IFNg
As there are multiple cell types in PBMC (other than NK cells) that can respond to viral stimuli, the amount of GzymB and IFNg in culture supernatant cannot be wholly attributed to NK cells (Hornung et al., 2002). Thus, we next determined the relative levels of GzymB and IFNg produced by other cells types (T and NKT cells) versus NK cells. In order to do this, surface staining was used to identify T cells and NKT cells ( Figure S1), and the iMFI was calculated for GzymB and IFNg production for each of these cell types, as described by Darrah et al. (Darrah et al., 2007). We then scaled this relative to the size of each population by multiplying the iMFI by the frequency of each cell type of the total lymphocyte population. We found that RV16 stimulation significantly upregulated the relative iMFI of GzymB + cells in NK cells, T cells, and NKT cells, both in the presence and absence of IFN-I signaling ( Figure 7A). In the NK cell and NKT cell populations, the relative iMFI of GzymB + cells increased when IFN-I signaling was blocked in the absence of RV16 stimulation ( Figure 7A, left and right). In all the conditions tested, the relative iMFI of GzymB + cells was highest in NK cells ( Figure 7A, left). It is worth noting that, unlike in the ELISA results, there was no significant decrease in RV16-stimulated GzymB when IFN-I signaling was blocked.
RV16 stimulation significantly upregulated the relative iMFI of IFNg + cells in all three of the populations tested, in both the presence and absence of IFN-I signaling ( Figure 7B). In RV16stimulated cultures, blocking of IFN-I signaling resulted in a significant decrease in the relative iMFI of IFNg + NK cells and T cells, but not NKT cells ( Figure 7B). The trends observed in NK cells and T cells reflect the levels of IFNg detected in the culture supernatant ( Figure 7B, left and middle). Stimulation with RV16 caused the largest change in the relative iMFI of IFNg + cells in the NK cell population ( Figure 7B, left).

DISCUSSION
This study aimed to investigate the possible role for NK cells in the immune response to RV infection and determine the extent to which RV16-induced activation and function of NK cells in vitro is dependent on IFN-I signaling. The key findings to emerge were that in cultured PBMCs from healthy people, RV16 stimulation affected NK cell activation and function, in a manner that was partially regulated by IFN-I signaling. IFN-I signaling partly contributed to RV-stimulated NK cell activation. Blocking of IFN-I signaling in PBMC cultures prior to RV16 stimulation reduced, but did not eliminate, NK cell activation. The magnitude of the effect varied between donors, but was generally modest. IFN-I has previously been shown to play both a direct and indirect role in the activation of NK cells (Biron et al., 1999;Cooper et al., 2001b;Hansbro et al., 2008). RV16 induces PBMCs to release IFN-I into the supernatant within 24 h (Khaitov et al., 2009). DC-mediated activation of NK cells involves both IFN-I dependent and independent mechanisms (Benlahrech et al., 2009). This is consistent with the NK cell activation observed in this study.
Some respiratory viruses, such as influenza, can directly interact with NK cells to elicit immune responses (Ennis et al., 1981;Sireń et al., 2004;Hwang et al., 2013). The use of PBMC cultures in this study did not allow us to determine the contribution of direct interactions between RV16 and NK cells, nor whether RV16 activates NK cells indirectly via other cells, such as APCs. Future experiments could determine this by studying purified NK cells.
RV16-stimulated NK cell degranulation was reduced when IFN-I signaling was blocked in vitro, as shown in Figure 3. This correlates with previous research into murine models of viral infection, with IFN-I signaling shown to contribute to the degranulation of NK cells during MCMV infection (Orange   and Biron, 1996b;Nguyen et al., 2002). Interestingly, blocking IFN-I signaling reduced the proportion of CD107a + NK cells both in virus-stimulated and unstimulated cultures. The latter observation may be attributed to autologous DCs inducing NK cell degranulation, as described by others (Hornung et al., 2002;Walwyn-Brown et al., 2018). Alternatively, it is possible there is a certain amount of constitutive IFN-I signaling that produces low-level NK cell degranulation. It has been suggested that CD107a protects degranulating cells from their own cytolytic granules, providing a basis for constitutive CD107a expression (Cohnen et al., 2013). There were no notable differences between the degranulation of CD56 dim and CD56 bright NK cell subsets. GzymB is a cytotoxic mediator released from NK cells to cause lysis of target cells (Fehniger et al., 2007). While RV16 stimulation increased the frequency of GzymB-producing NK cells, blocking of IFN-I signaling did not have a significant impact. Despite the frequency of GzymB-producing NK cells not changing significantly, the amount of intracellular GzymB (which was significantly increased in response to viral stimulation) was lower in the absence of IFN-I signaling.
Previous studies have observed similar trends in murine vaccinia virus infection, where the addition of IFN-I directly stimulated NK cell production of GzymB (Martinez et al., 2008). While constitutive expression of GzymB was noticeably higher in CD56 dim NK cells, similar trends were still observed in CD56 bright NK cells. RV16 stimulation induced GzymB release into culture supernatant, in a manner partially dependent on IFN-I signaling. Importantly, NK cells were observed to produce more GzymB than T cells or NKT cells.
Stimulation by viruses and cytokines induces NK cell production of IFNg (Cooper et al., 2001b;Papadopoulos et al., 2002b;Maroof et al., 2008). RV16 stimulation significantly increased both the frequency of IFNg-producing NK cells and their level of intracellular IFNg. Blocking IFN-I signaling resulted in a reduction in the frequency of IFNg-producing NK cells in RV16-stimulated cultures. This indicates that IFN-I signaling plays a role in RV16-stimulated IFNg production by NK cells, and is consistent with previous studies on the role of IFN-I on NK cell IFNg in other viral infections (Biron et al., 1999;Gary-Gouy et al., 2002;Martinez et al., 2008;Kronstad et al., 2018). Notably, almost three times as many CD56 bright NK cells were producing IFNg in response to RV16 stimulation, than CD56 dim NK cells. RV16 stimulation also significantly increased the amount of released IFNg, and this was also partly dependent on IFN-I signaling. Importantly, NK cells were found to be more responsive to both RV16 stimulation and IFN-signaling, than T cells or NKT cells. Subsequent studies could elucidate the variations in NK cell surface receptor repertoire, specifically the density of IFNAR, that may contribute to the differential effects of RV on the CD56 dim and CD56 bright NK cell subsets (Cooper et al., 2001a;Sedlmayr et al., 2004).
The limitations of this study must be acknowledged. Firstly, the use of PBMCs as opposed to lung immune cells for in vitro experiments. However, NK cells in the lung seem to primarily consist of circulating rather than tissue resident cells (Marquardt et al., 2017). In addition to this, RV stimulation of PBMCs has been shown to be a suitable in vitro model in which to observe immune responses (Message and Johnston, 2001;Papadopoulos et al., 2002b;Hornung et al., 2002;Xi et al., 2015;Xi et al., 2017). It is also worth noting that the sample size of this study was small (n=12); despite this, there was enough statistical power identify significant differences between groups. The age of participants in this study (21.6 ± 2.8 years) was also restricted. Future studies should be conducted in larger cohorts, with a broader age range, to confirm these findings and account for interindividual variation in response to viral stimulation. This study only assessed NK cell response at a single time point with a single MOI of one serotype of human RV. There are over 150 serotypes of human RV (Glanville and Johnston, 2015). These serotypes are categorized into major and minor subtypes, based on their method of cell entry; however, even serotypes that share a common method of cell entry can follow different endocytic pathways and release of viral genome at different locations within the infected cells (Blaas and Fuchs, 2016), leading to diversity in the elicited immune response (Wark et al., 2009). Despite this, recognition of pathogen associated molecular patterns that are highly conserved across RV serotypes, such as ssRNA, by pattern recognition receptors triggers the activation of the innate immune response (Triantafilou et al., 2011). Activation of PBMCs in response to RV stimulation has also been shown to be dose-dependent (Gern et al., 1996). Future studies using multiple MOIs and different serotypes of RV could address the question of dose-dependent and serotype-dependent differences in NK cell responses. It is also important to acknowledge that immune cells, including NK cells, do not interact with RV in circulation, but at the airway epithelium. Thus, further studies should be conducted in co-cultures of respiratory epithelial cells and NK cells, in order to observe the NK cell response to virally infected respiratory epithelial cells. Future studies should also assess the levels of RV-specific neutralizing antibodies in the serum of each participant. A deficiency in RV-specific neutralizing IgG antibodies, specifically those targeting the VP1 viral capsid protein, has previously been associated with increased risk of exacerbations in patients with COPD (Yerkovich et al., 2012). Future studies should examine whether variations in RV-induced NK cell activation are associated with susceptibility to colds. A study of this nature would require a large cohort of participants to have sufficient study power.
In conclusion, we demonstrated that RV16 stimulates NK cells in vitro, and that this response is partially, but not completely, regulated by IFN-I signaling. These results established that RV16 stimulation of PBMCs leads to NK cell activation, degranulation, cytotoxic mediator production, cytokine production, and the release of cytotoxic mediators and cytokines into the culture supernatant. These aspects of NK cell activation and function were all partially dependent on IFN-I signaling. While deficient IFN-I signaling may play some role in the susceptibility of asthmatics to more persistent and severe infections, our findings also indicate that further studies need to examine other cytokines and APC function, and how this impacts on NK cell function. This study provides an important foundation for future studies into NK cell activation and function in asthma.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENTS
The studies involving human participants were reviewed and approved by The University of Queensland Human Research Ethics Committee and Metro South Human Research Ethics Committee. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
SH and JU contributed to the conception and design of the study. SH and YX conducted the experiments and statistical analysis. SH wrote the manuscript. YX and JU contributed to the revision of the manuscript and approved the submitted version. All authors contributed to the article and approved the submitted version.

FUNDING
The National Health and Medical Research Council (NHMRC) of Australia supported this work via Project grant APP1128010.