Possible role of highly activated mucosal NK cells against viral respiratory infections in children undergoing haematopoietic stem cell transplantation

Infection is the leading cause of non-relapse-related mortality after allogeneic haematopoietic stem cell transplantation (HSCT). Altered functions of immune cells in nasal secretions may influence post HSCT susceptibility to viral respiratory infections. In this prospective study, we determined T and NK cell numbers together with NK activation status in nasopharyngeal aspirates (NPA) in HSCT recipients and healthy controls using multiparametric flow cytometry. We also determined by polymerase chain reaction (PCR) the presence of 16 respiratory viruses. Samples were collected pre-HSCT, at day 0, +10, +20 and +30 after HSCT. Peripheral blood (PB) was also analyzed to determine T and NK cell numbers. A total of 27 pediatric HSCT recipients were enrolled and 16 of them had at least one viral detection (60%). Rhinovirus was the most frequent pathogen (84% of positive NPAs). NPAs of patients contained fewer T and NK cells compared to healthy controls (p = 0.0132 and p = 0.120, respectively). Viral PCR + patients showed higher NK cell number in their NPAs. The activating receptors repertoire expressed by NK cells was also higher in NPA samples, especially NKp44 and NKp46. Our study supports NK cells relevance for the immune defense against respiratory viruses in HSCT recipients.

through the production of cytotoxic granules containing molecules such as perforins and granzymes that lyse infected cells. Other important features of NK cells are the production of cytokines including interferons, their enhancement of local immune responses by directly acting on target cells, and their role in the recruitment and activation of other immune cells, including T cells and macrophages 17 .
NK cells recognize virally infected cells through the integration of signaling from activating and inhibitory germ line encoded receptors on the NK cell surface 17 . Membrane bound Killer-cell Activation Receptors (KARs) work together with inhibitory Killer Immunoglobulin-like Receptors (KIRs), in order to regulate the NK cells functions on infected or transformed cells. The interaction between the KIR family on NK cells and HLA Class I molecules results in inhibitory signals. KARs, like the natural cytotoxicity receptors (NCRs), comprising NKp30, NKp46, and NKp44 are key receptors in the recognition and elimination of virally infected and tumor cells 18,19 .
Because immune cells in the nasal mucosa can control and regulate viral infections via killing infected respiratory epithelial cells, low numbers of NK and or T cells in the nasopharyngeal aspirates (NPAs) of HSCT pediatric recipients could be associated with an increased susceptibility to viral respiratory infections. Although blood immune reconstitution after an HSCT has been widely studied [20][21][22][23][24][25][26] , very little is known about the immune reconstitution in HSCT recipients' mucosa. In the present study, our goals were: 1) to determine NK and T cell numbers in the nasal passages of HSCT recipients; 2) to determine the activation phenotype of NPAs' NK cells using multiparametric flow cytometry; 3) to determine the presence and functional profile of these cells within the context of a viral infection; and 4) to assess the effects of viral infections on nasal immune reconstitution.

Materials and Methods
Study participants. Prospective study including patients under 18 years of age who received allogenic HSCT at La Paz University Hospital (Madrid, Spain), from January 2017 to June 2018. Exclusion criteria included chronic lung disease prior to HSCT. HSCT was performed according to current centre protocols, using bone marrow or peripheral blood stem cells as haematopoietic stem cell source. Acute and chronic graft versus host disease (GVHD) were defined and graded as previously reported 27,28 .
Age-and sex-matched controls were recruited among healthy patients undergoing elective surgery. Cases and controls were also matched according to procedure date (HSCT and elective surgery, respectively) ± 1 week.
All research was performed in accordance to local regulations. Informed consent in accordance with the Declaration of Helsinki was obtained from the parents of all children. The study was approved by the Clinical Research Ethics Committee at La Paz University Hospital. Study samples. Study samples were peripheral blood (PB) and NPAs that were collected prior to the conditioning regimen and after HSCT at days 15 and 30 for PB mononuclear cells and before the conditioning regimen and at days 0, 10, 20 and 30 after HSCT for NPAs. The following tests were performed in each sample type: • PB: total, T and NK cell counts.
• NPAs: total T and NK cell counts, natural cytotoxicity receptors, viral detection.
For NPA collection, 1 mL of 0.9% saline solution was instilled into each nostril and the nasopharynx was suctioned using a sterile catheter with a mucus trap. NPAs were not performed if the patient had active bleeding or his platelet count was below 20.000/mm3. NPA samples containing blood were considered not valid and discarded. Each specimen was sent to the Respiratory Virus and Influenza Unit at the National Microbiology Center (ISCIII, Madrid, Spain) for viral analysis and to the Translational Research in Paediatric Oncology, Hematopoietic Transplantation & Cell Therapy Unit at La Paz Hospital Research Institute (Madrid, Spain) for flow cytometry assays. NPAs were processed within 24 hours after collection. In the control group, NPAs were collected on the day of surgery. Viral detection by Rt-pcR. Upon reception of the samples, three aliquots were prepared and stored at −80 °C. Both the reception and the NPA sample processing areas are separated from those defined as working areas. Three independent RT-PCR assays were performed to detect sixteen respiratory viruses as previously published by our group [29][30][31] . Influenza A, B and C viruses were detected by using previously described primer sets only to amplify influenza viruses in a multiplex PCR assay 29 . A second multiplex PCR was used to detect parainfluenza viruses 1 to 4, human coronaviruses 229E and OC43, enteroviruses and HRV 30 . Presence of RSV A and B types, HMPV, HBoV and adenoviruses was established using a third multiplex RT-nested PCR−BRQ method 31 . Cell characterization by flow cytometry. For flow cytometry assays, NPA samples were filtered with a 70 µm cell strainer (Falcon), centrifuged at 1400 rpm (5 min) and suspended in 100 µl of PBS. Total sample volume was acquired using a Navios flow cytometer (Beckman Coulter).

Results
HSct recipients and control group characteristics. During the study period, 27 HSCT recipients were recruited. Their main characteristics are summarized in Table 1. The median age of patients was 7.7 years (IQR 9.2).
All HSCT were allogenic. Twenty-five patients underwent HSCT for the first time (93%) and two received their second HSCT (7%).
Eighteen healthy children were included in the control group, 10 male (56%) and 8 female (44%), with a median age of 8.7 years (IQR 9). There were no significant differences regarding age and sex between HSCT recipients and healthy controls.

Viral infections.
A total of 83 samples were collected from the 27 HSCT recipients, and 77 were valid for viral studies (median number of valid samples per patient: 3; IQR 2). Twenty-five samples (32%) were positive, and 16 of 27 HSCT recipients (60%) had at least one viral detection. Among HSCT recipients with viral infection, the median number of positive samples per patient was 1 (IQR 1). HRV was isolated in 21 samples (84% of positive NPA) from 12 patients, followed by adenovirus and parainfluenza type 1 (two positive samples from two different patients each, 8%). There were no viral coinfections among HSCT recipients. Detailed information regarding positive samples is given in Table 2.
Infections caused by HRV were symptomatic in 2 of 12 patients (17%): one had low-grade fever and the other persistent rhinorrhea. Both patients with adenovirus infections had fever, mucositis and elevated levels of C-reactive protein (above 100 mg/L). Infections by parainfluenza type 1 virus were also symptomatic (one patient with fever and another with laryngitis and pneumonia). None of the patients required admission to the intensive care unit (ICU) nor died as a result of a viral infection. There were no differences regarding age between HSCT recipients with and without viral infections (median [IQR] 7.5 [8.8]  www.nature.com/scientificreports www.nature.com/scientificreports/ A total of 17 samples from healthy controls were analyzed, and viruses were identified in 4 (24%): two single infections (HRV and adenovirus) and two coinfections (HRV and HBoV, HRV and adenovirus) ( Table 2). Controls with viral infections were younger, but this difference did not reach statistical significance (median [IQR] 4.1 [6.6] vs. 8.9 [8.5] years, p = 0.07). All infections were asymptomatic. No significant differences were found regarding viral isolation rate between patients and healthy controls (32% and 24% of samples, respectively, p = 0.57).
NPA cellular composition of HSCT recipients. The NPAs of patients prior to HSCT conditioning contained fewer T and NK cells as compared with healthy controls p = 0.0132 and p = 0.120, respectively (Fig. 1A). Additionally, PCR + patients prior to HSCT conditioning showed statistically significant higher numbers of NK cells in NPAs than PCR− patients (p = 0.006). Those differences were not observed in healthy controls or in the T cell populations isolated from NPAs.
We further analyzed T and NK cell kinetics after the HSCT. We observed an increase in T cells in the NPA samples at infusion day. This increase is reverted at day 10 post HSCT, when NPA cell counts reach levels similar to those observed prior to the HSCT. This effect was observed both in PCR+ and in PCR− patients (Fig. 1B and Suppl. Table 1). However, we did not observe that peak with NK cells. There were no differences regarding NPA T and NK cell counts over time in patients with and without respiratory infection.   www.nature.com/scientificreports www.nature.com/scientificreports/

Natural cytotoxicity receptors (NCRs) expression in NPA's NK cells. The NK cells found in the NPA
samples had a highly activated status as determined by flow cytometry analysis (Fig. 3A). The percentage of NK cells expressing different activating receptors was higher in PCR + NPA samples than in PCR− samples for all the receptors analyzed, and especially significant for NKp44 and NKp46 receptors (Fig. 3B). The highly activated status was observed both in healthy donors and HSCT patients' NPAs.
HSct recipient outcomes. The outcomes of HSCT recipients are summarized in Table 3. There were no significant differences according to the presence of viral infection. The number of T or NK cells in patients' NPAs pre-, during-or after-HSCT was not correlated with a higher risk of GVHD, graft failure, ICU admission or death (Suppl . Tables 3 and 4).

Discussion
The nasal mucosa is the first place within the respiratory system to be exposed to inhaled viral pathogens. Therefore, nasal immune cells are likely to play an important role in early innate immune responses to those environmental factors, both in healthy individuals and in immunocompromised patients like HSCT recipients. NK cells are thought to play a particularly important role in this context, since they are able to influence both innate and adaptive immune responses 32,33 and are the first lymphocyte subset to appear in PB after HSCT [20][21][22][23][24][25][26] . While macrophages and dendritic cells have been identified in the nasal submucosa 34 , and neutrophils have been identified in the nasal cavity 35 , the immune cells present in NPAs have not been fully characterized. To phenotype NPA cells, previous researchers have relied on cell differential analysis of cytocentrifuge slides stained with haematoxylin and eosin. However, flow cytometry may be a superior method to fully characterize NPA immune cells since it allows for a precise quantification of cell populations as well as for phenotype studies, making it possible to determine the activation status of such specific cell populations.
Our study showed that HSCT patients, as expected -due to conditioning regimens and underlying diseasehad fewer T and NK cells in their NPAs as compared to healthy controls. Interestingly, those patients exposed to viral pathogens showed statistically significant higher NK cell counts in their NPAs, which is in accordance with their role in the "front line" defense against viral infections. The phenotypes and functions of NK cells differ depending on the source organ or tissue [36][37][38] . Our current research further explores the particular phenotype of nasal NK cells, which show a higher activation status than those found in PB, comparing the results obtained in this study with previous publications from our group 24 . This was especially true for those individuals with positive viral detection in their nasal secretions. Okada et al. recently reported that NK cells found in the nasal passages of mice belong to the conventional NK cell linage and characteristically demonstrate an immature and activated phenotype 39 . Our results constitute the first evidence of such observations in humans.
In our study, we observed an unexpected T cells increase in NPAs on HSCT day 0 samples, despite conditioning regimens. We attribute it to the fact that all NPA samples were collected in the morning, and almost all patients receiving non-manipulated grafts (60% of our cohort) had finished their infusions prior to NPA collection. Pre-transplant conditioning, such as high-dose chemotherapy is toxic to many cells and render the endothelium vulnerable 40 which could lead to increased T cells in NPA of many patients.
Natural killer cell activity is regulated through the balance of signals from inhibitory and activating receptors. The natural cytotoxicity receptors (NCRs) NKp46 (NCR1), NKp44 (NCR2), and NKp30 (NCR3), as well as NKG2D, are the main activating receptors involved in mediating NK cell function in health and disease 18,19 . The www.nature.com/scientificreports www.nature.com/scientificreports/ most overexpressed activating receptors we found in NK cells of NPAs were NKp44 (NCR2) and NKp46 (NCR1). This is consistent with Shemer-Avni et al. 's findings 41 . In nasal lavage samples of patients infected with respiratory viruses, they found NKp46 as the most abundantly expressed NCR. However, in their RTqPCR analysis they detected no NKp44 expression and very little of NKp30. Altogether, our findings point to NKp44 as a key player in nasal mucosa elimination of virally infected cells by NK cells.
Conditioning regimens used for allogeneic HSCT, high dose chemotherapy and/or radiotherapy, induce serious damage in mucosal, humoral and cellular immune function. Moreover, those regimens can harm   www.nature.com/scientificreports www.nature.com/scientificreports/ the host's thymopoiesis resulting in further delayed immune recovery 25,26 . Collectively, these factors predispose the host to a variety of infections. In fact, despite the routine use of prophylactic antimicrobials in the peri-transplant period 42,43 , infections occur is about 80-85% of HSCT recipients and are one of the leading causes of non-relapse-related mortality after allogeneic HSCT 44,45 .
More than half of the HSCT recipients included in our study presented a viral respiratory infection at some point. To date, most studies of these infections in pediatric HSCT recipients have focused on symptomatic individuals, reporting an incidence of 5-11% [46][47][48] . Asymptomatic respiratory infections in children undergoing HSCT have been barely studied. A prospective study of 33 patients detected respiratory viruses in 24% of them before HSCT, mainly HRV and adenovirus 49 . In our series, infections caused by HRV were the most common, and were not associated with increased morbidity. These findings are in concordance with those reported by Srinivasan et al. in children with cancer or post-HSCT and upper or lower respiratory infection 13 and asymptomatic children before HSCT 49 . Other authors have reported implementing routine extensive pre-HSCT lung screening, including bronchoalveolar lavage for the detection of viruses, bacteria and fungi. In that study, almost one third of asymptomatic patients had viral respiratory infections, mainly caused by rhinovirus, and none of these worsened during HSCT 50 . On the other hand, infections caused by adenovirus and parainfluenza virus were less frequent but more often symptomatic in our patients. Adenovirus has been described as a cause of lower respiratory tract infection with high mortality after HSCT in studies involving symptomatic children 46,51 . We did not find a correlation between viral infection and HSCT outcomes, which could be caused by small sample size and low incidence of symptomatic infection.
Important strengths of our study include its prospective design and the determination of number, activation phenotype and functional profile of NK cells together with respiratory virus detection, assessing HSCT recipients over time and comparing them with healthy controls. Potential limitations of our study relate to sample size and losses to follow-up. NPA cell counts have great variability so larger cohorts would be desirable to draw stronger conclusions. Many patients did not tolerate well NPAs and withdrew during the course of the study. The small number of patients further precluded correlation of viral detection and transplantation outcomes, and limits our ability to examine the implications of asymptomatic viral respiratory infections in HSCT recipients. NPAs were collected up to 30 days after HSCT, and some infections occurring later may have been missed. However, most respiratory viral infections in HSCT recipients occur soon after transplantation, and adverse clinical outcomes are more likely to occur in the same period 46,51 .
In conclusion, we provide data suggesting NK cells as major effectors in respiratory viral control in the nasal mucosa. We also characterized this NK cell population finding a highly activated phenotype compared with that of their PB counterparts. Further study of this unique mucosal immune cell population will be beneficial in assessing the effects of viral infections on upper respiratory immune responses in HSCT recipients as well as in healthy individuals.