Introduction

Influenza is a leading cause of respiratory infections worldwide (Beigel 2008; Taubenberger and Kash 2010). A significant proportion of the world’s population is infected by the virus every year. In addition, the emergence of a novel influenza virus leads to the occurrence of pandemic events as was the case for the pandemic influenza A (H1N1) 2009 virus (CDC 2009; Peiris et al. 2009; Taubenberger and Kash 2010). On April 2009, the circulation of a novel influenza virus was discovered in Mexico and documented shortly thereafter throughout the rest of the world (CDC 2009; Gomez-Gomez et al. 2010; Perez-Padilla et al. 2009). The majority of persons infected with this novel virus had a self-limiting illness, and many asymptomatic infections were demonstrated through serologic surveys (Chen et al. 2010; Lee et al. 2010). An increase in the severity of influenza infections was observed amongst diabetic or obese individuals, pregnant women as well as in those having other underlying disorders (Campbell et al. 2010; Gomez-Gomez et al. 2010; Siston et al. 2010; Zarychanski et al. 2010). However, many patients with severe illness did not have any identifiable risk factors (Campbell et al. 2010; Fabbiani et al. 2011; Gomez-Gomez et al. 2010). As such, the reason why some patients develop severe symptoms when infected with this virus while others experience only mild symptoms and self-limiting infection remains unclear.

The clinical outcome of influenza virus infections depends on a complex set of host–pathogen interactions that take place early on during the initial antiviral immune response. In inter-pandemic seasonal influenza infections, the vaccine-primed adaptive immune system is largely responsible for establishing a prompt antiviral response, which in most cases will eradicate the virus with minimal impact on the person’s health. On the other hand, the clinical outcome of influenza virus infections occurring during pandemics is likely to be determined by innate antiviral responses as the adaptive immune system is not likely to have encountered the novel influenza virus. Natural killer (NK) cells are part of the innate immune system and are considered the first line of defence against viral incursions as they do not require prior antigenic sensitization (Hamerman et al. 2005). Cytotoxicity and cytokine release by NK cells are controlled by activating and inhibitory signals which arise from a diverse array of cell surface receptors. Killer-cell immunoglobulin-like receptors (KIR) are by far the most diverse and complex cell surface receptors modulating NK cells. KIR proteins are encoded by at least 17 different genes located on 19q13.4 (KIR2DL1, KIR2DL2, KIR2DL3, KIR2DL4, KIR2DL5A, KIR2DL5B, KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5, KIR2DP1, KIR3DL1, KIR3DL2, KIR3DL3, KIR3DS1 and KIR3DP1); however, some KIR genes share a locus and are thus mutually exclusive as is the case for KIR2DS3/2DS5 and KIR3DL1/3DS1 (Gomez-Lozano et al. 2002; Pyo et al. 2010; Trowsdale et al. 2001). These genes encode for proteins having either short cytoplasmic tails which provide activating signals (KIR2DS1–5 as well as KIR3DS1) or long cytoplasmic tails (KIR2DL1–5 and KIR3DL1–3) which inhibit NK cell alloreactivity (Campbell and Colonna 2001; Renard et al. 1997). KIR proteins interact with classical human leukocyte antigen (HLA) class I ligands such as HLA-A (KIR3DL2), HLA-B (KIR3DL1 and KIR3DS1) and HLA-C (KIR2DL1, KIR2DL2, KIR2DL3 and KIR2DS2) as well as with non-classical HLA ligands such as HLA-G (KIR2DL4) (Cantoni et al. 1999; Dohring et al. 1996; Gumperz et al. 1996; Moretta et al. 1993). The ligands of the remaining KIR proteins have not been determined to date.

KIR genes exhibit allelic polymorphism and are arranged into haplotypes which exhibit variation in gene content (Pyo et al. 2010; Shilling et al. 2002; Uhrberg et al. 1997). Inferred functional capabilities distinguished KIR haplotypes having a single activating KIR gene (KIR2DS4) and a fixed set of inhibitory genes (group A haplotypes) from those exhibiting greater variability in the number of genes (group B haplotypes) (Uhrberg et al. 1997). A recent genomic study has demonstrated that most gene content variations arise from different combinations of a fixed and relatively limited set of centromeric and telomeric haplotypic motifs (Pyo et al. 2010). This same study also proved that certain KIR genes (KIR2DL5, KIR2DS3/2DS5) could be found in either the centromeric motif (KIR2DL5C and KIR2DS3/2DS5C) or a telomeric motif (KIR2DL5T and KIR2DS3/2DS5T).

This diverse yet predetermined immune receptor repertoire enables NK cells to respond rapidly to many viral infections. An important role for these cells in the early events following respiratory virus infections has been previously suggested (Capelozzi et al. 2010; Culley 2009; Du et al. 2010; Jost et al. 2011; Mao et al. 2010). Animal models have shown that NK cells recruited to the lungs after experimental influenza virus infections contribute importantly at reducing viral replication (Stein-Streilein et al. 1983; Trammell and Toth 2008). However, NK cell activation has also been shown to contribute to inflammatory lung injury (Capelozzi et al. 2010; Harker et al. 2010; Okamoto et al. 2002).

Studies of KIR gene involvement in HIV infection have shown that vigorous NK cell alloreactivity as predicted from activating gene content might be advantageous during initial antiviral responses but detrimental in chronic states of activation as in chronic infections (Alter et al. 2007; Carrington et al. 2008; Gaudieri et al. 2005; Jennes et al. 2006; Martin et al. 2007; Paximadis et al. 2011). Thus, it could be conceived that in respiratory virus infections as in other viral incursions, NK cell responses should be kept in balance in order to avoid lung damage yet allow for proper viral clearance. Dysregulation of innate immune responses or predetermined receptor repertoires capable of eliciting potent NK cell reactions might contribute to greater inflammatory damage and to the development of severe lower respiratory tract disease. In this study, we assess differences in KIR genotype between healthy subjects and those suffering from mild and severe forms of influenza in the search for associations between KIR genes and disease.

Material and methods

Patients

Patients admitted to the Hospital Central ‘Dr. Ignacio Morones Prieto’ between April 2009 and April 2010 with respiratory symptoms and lower respiratory tract involvement due to A (H1N1) 2009 pandemic influenza were eligible for inclusion in this study (severe influenza group; n = 45). In addition, 26 patients with upper respiratory tract symptoms and confirmed A (H1N1) 2009 pandemic influenza infection not requiring hospitalization were also included (mild influenza group). Previously published KIR genotypic data produced on 300 healthy blood donors sampled by the same hospital and sharing the same demographic and ethnic background as our study population were used as a reference population (Alvarado-Hernandez et al. 2011). Demographic and clinical information for influenza patients was obtained from medical records and interviews. The study was reviewed and approved by the hospital’s Research and Ethics Committee. Informed consent was obtained in writing from participating patients or their next of kin prior to enrolment.

Influenza virus detection

Viral screening was carried out as part of the pandemic influenza public health emergency measures (Gomez-Gomez et al. 2010). Respiratory secretions obtained from pharyngeal or nasopharyngeal swabs were used for RNA extraction using the High Pure Viral RNA Kit (Roche Diagnostics, Indianapolis, IN, USA). Pandemic influenza virus was detected using either the CDC’s real-time protocol or an RT-PCR assay previously developed by our research group (Gomez-Gomez et al. 2010). Specificity assays showed no cross-amplification of seasonal A (H1N1), A (H3N2) and influenza virus B strains.

KIR genotyping and haplotype inference

DNA was obtained from a 5-ml blood sample from patients in a heparinised tube according to established protocols (Garcia-Sepulveda et al. 2010). The presence of KIR genes was determined using a previously published sequence-specific priming PCR (SSP-PCR) amplification assay capable of detecting the 17 KIR genes in 16 reactions (Alvarado-Hernandez et al. 2011). This SSP-PCR approach did not enable us to distinguish between KIR2DL5A and KIR2DL5B, nor between KIR2DL5C and KIR2DL5T or KIR2DS3/S5C and KIR2DS3/S5T. KIR haplotype motifs (cA01, cB01, cB02, cB03, tA01 and tB01) and extended haplotypes could be assigned using a deterministic approach based solely on genotyping results as described in the original publication and allowing for the inherent ambiguities that are described with greater detail in the original publication (Pyo et al. 2010). Approximately 98 % of our healthy controls and 100 % of our patient genotypes could be assigned a haplotype motif or extended haplotype using this approach. KIR genotype nomenclature followed previously established conventions (Gonzalez-Galarza et al. 2011).

Statistical analysis

Demographic and clinical characteristics were analysed using descriptive statistics. Medians were used for continuous variables; categorical variables were described through percentages and compared using the chi-squared test or Fisher’s exact test. A P value <0.05 was considered statistically significant. No corrections for multiple comparisons were made. All statistical analyses were performed with SPSS for Windows release 14.0.0 (SPSS, Inc. 2005; Chicago, IL, USA) and EpiInfo (version 6, Centers for Disease Control and Prevention, Atlanta, GA, USA).

Results

Clinical and demographic characteristics

We analysed 45 blood samples from patients hospitalized with influenza-associated lower respiratory tract infection (severe illness). Of these, 28 (62.2 %) patients were male and 17 (37.8 %) were female, their median age being 36 years (range, 3–67 years). Twenty-six patients comprised the ‘mild illness’ group of which 10 (38.5 %) were male and 16 (61.5 %) were female, their median age being 28 years (range, 12–62 years). Symptoms observed in all influenza patients included cough (93 %), fever (85.9 %), myalgias (74.6 %), dyspnea (74.6 %), arthralgias (73.2 %), headache (71.8 %) and rhinorrhea (62 %). Co-morbidities were more common in patients with severe influenza (68.9 %) compared to those presenting with mild illness (15.4 %, P < 0.001). The most common underlying disorders were obesity (17 patients, 23.9 %), type II diabetes (9 cases; 12.7 %) and hypertension (9 cases, 12.7 %). Thirty (66.6 %) patients in the severe influenza group required admission to the intensive care unit, and 18 (40 %) died.

KIR gene carrier frequencies in patients with A (H1N1) 2009 pandemic influenza infection

The percentage of patients having each KIR gene (carrier frequency) was calculated by direct counting and compared to the control population (Table 1). The four framework genes (KIR3DL3, KIR3DP1, KIR2DL4 and KIR3DL2) were present in 100 % of the samples. Carrier frequencies for most genes were similar between patients with mild or severe infection and the Mexican mestizo reference population. However, subjects exhibiting influenza infections (both mild and severe groups combined) had higher frequencies of KIR2DL5 (P = 0.034), KIR2DS5 (P = 0.001) and KIR3DS1 (P = 0.025). When analysed in separate groups, these KIR gene associations only remained statistically significant when comparisons were restricted to subjects having severe forms of the disease and the reference population (KIR2DL5, P = 0.02; KIR2DS5, P = 0.017 and KIR3DS1, P = 0.03).

Table 1 KIR gene carrier frequencies in patients with pandemic influenza A (H1N1) 2009 infection and healthy donors

The frequency of individuals harbouring both KIR3DL1 and KIR3DS1 genes was higher amongst influenza-infected patients than in the reference population (56.3 versus 37.7 %, respectively; P = 0.004). This finding remained statistically significant when subjects having severe forms of the disease were compared to controls (57.8 versus 37.7 %, respectively; P = 0.01). In contrast, individuals harbouring KIR3DL1 in the absence of KIR3DS1 were more common in the reference population than amongst influenza-infected patients (56.7 versus 42.3 %, respectively; P = 0.028), a finding that remained statistically significant when the reference population was compared to the severe influenza infection group (56.7 versus 40 %, respectively; P = 0.036).

Analysis of the number of activating genes present in the study participants showed a bimodal distribution (Fig. 1a). Overall, most individuals included in the study had either one (33.9 %) or four (22.4 %) activating genes. A higher proportion of individuals having ≥3 activating KIR genes was observed amongst the severe influenza group (68.9 %) in comparison to healthy controls (47.7 %; P = 0.008). Comparisons of inhibitory KIR gene carrier frequencies (disregarding framework genes) exhibited a similar behaviour where the proportion of individuals having five inhibitory genes was higher amongst subjects with severe influenza in comparison to healthy controls and to patients having mild forms of the disease (Fig. 1b). The frequency of individuals having five inhibitory KIR genes amongst the mild influenza group was twice as high as in healthy controls (34.6 versus 17.3 %, respectively; P = 0.03). However, the proportion of individuals having five inhibitory KIR genes was thrice as high amongst patients having severe influenza than in healthy controls (53.3 versus 17.3 %, respectively; P = 0.001). Other activating and inhibitory KIR gene number stratifications were also evaluated but found not to be statistically significant.

Fig. 1
figure 1

Activating and inhibitory KIR genes in patients with pandemic influenza A (H1N1) 2009 infection and healthy donors. a Proportion of influenza patients and healthy donors according to the number of activating KIR genes. b Proportion of influenza patients and healthy donors according to the number of inhibitory KIR genes (disregarding framework genes)

KIR genotypes and haplotypes in patients with pandemic influenza infection

In total, 23 different KIR genotypes were observed in the influenza patients that were studied, all of which had been reported previously (Table 2). Twelve genotypes were seen in more than one instance and encompass more than 84.5 % of our study population (n = 60 subjects). The frequencies of these KIR genotypes were, in general terms, similar between the different study groups. However, the frequency of genotype 6 (which includes all activator and inhibitory KIR genes) was higher in the severe influenza group than in the mild illness group and the control population (8.8, 3.8 and 2.0 %, respectively; this difference was statistically significant between patients with severe influenza and the control group, P = 0.03). On the contrary, genotype 4 (characterised by the presence of KIR2DL2/S2 and KIR2DS4) was less frequent amongst patients with severe influenza infection compared to those with mild influenza and the control population (2.2, 11.5 and 12 %, respectively); however, this difference was not statistically significant. The frequency of classical KIR haplotype groups (A and B) amongst influenza patients had a similar distribution to that of the previously described healthy control population. KIR haplotype motifs and extended haplotypes were inferred from our genotyping data based on previously published KIR gene content analysis of human chromosomes and shown in Table 3 (Pyo et al. 2010). The presence of the combination of KIR genes contained in the cB01 and cB03 haplotype motif was higher amongst influenza-infected patients (39.4 and 55 %, respectively) in comparison to healthy controls (21 and 40.3 %, respectively), P = 0.001 and P = 0.026, respectively. However, when the influenza cohort was stratified by severity, this comparison only remained statistically significant for the severe influenza group (P = 0.001 and P = 0.002, respectively). As a direct consequence, extended KIR haplotypes bearing the aforementioned centromeric motifs cB01|tA01 and cB01|tB01 were also found at a higher frequency in influenza-infected patients (34 and 30 %, respectively) in comparison to healthy controls (17.3 and 16.7 %; P = 0.002 and P = 0.013, respectively). Again, on stratification, this frequency remained statistically significant only for the severe influenza group (37.8 and 31.1 %, P = 0.001 and P = 0.02, respectively). Interestingly, the frequency of the cB03|tA01 extended haplotype proved to be higher amongst severe influenza patients in comparison to healthy controls and patients with mild influenza (53.3, 34 and 30.8 %, respectively; P = 0.012).

Table 2 Frequency of KIR genotypes seen in patients with pandemic influenza A (H1N1) 2009 and healthy donors
Table 3 Frequency of KIR haplotype motifs and extended haplotype carriers in patients with pandemic influenza A (H1N1) 2009 infection and healthy donors

Discussion

Starting on 2009, there has been a worldwide dissemination of a novel strain of influenza A (H1N1) virus (CDC 2009; Fowlkes et al. 2011; Perez-Padilla et al. 2009). Mexico was the first country to suffer the effects of the epidemic caused by this virus, and as in other pandemics, more severe infections were observed, particularly amongst young adults (CDC 2009; Gomez-Gomez et al. 2010; Perez-Padilla et al. 2009). Clinical severity of influenza infections is determined by both viral and host factors. Host cell tropism, virulence and drug-resistance mutations are well-known pathogen features which relate to clinical severity. Host factors that determine the clinical severity of viral incursions are mainly (but not exclusively) dependent on the immune response. For influenza virus infections, the precise host mechanisms responsible for the observed variability in clinical severity have not been defined as of yet. Nevertheless, accumulating evidence highlights the importance of immune responses, especially those of the innate immune system. As NK cells play an important role during the initial response to viral pathogens, the relevance of these cells with regard to the clinical severity of viral infections has been the focus of much recent attention (Amadei et al. 2010; Brenner et al. 1989; Jost et al. 2011; Mao et al. 2010). It has previously been suggested that differences in the innate immune system’s ability to respond to certain pathogens could explain the propensity of some individuals to develop more severe symptoms following viral infections (Holmskov et al. 2003). In vitro studies have demonstrated that NK cells are capable of eliciting potent antiviral responses within 24 h after exposure to influenza A virus (He et al. 2004). In addition, IFN-γ levels in nasopharyngeal lavage fluid have been shown to correlate with influenza virus titres and clinical symptoms (Kaiser et al. 2001). KIR genes represent the most complex, polymorphic and diverse group of cell surface receptors capable of modulating NK cell responses (Khakoo and Carrington 2006). The presence of specific KIR genes and KIR gene combinations has been associated with susceptibility to (and progression of) different viral infections such as HIV, hepatitis C, Ebola, herpes simplex and CMV (Ahlenstiel et al. 2008; Estefania et al. 2007; Martin et al. 2002; Paladino et al. 2007; Stern et al. 2008; Wauquier et al. 2010). However, very little is known regarding the influence of specific genes on the susceptibility to influenza infections in humans. The possible involvement of KIR genes in influenza was assessed using an in vitro model of infection with influenza A virus (Ahlenstiel et al. 2008). In this study, differences in human NK cell activity related to distinct KIR/HLA genotypes were observed. It was shown that NK cells of HLA-C1 homozygous subjects lacking KIR2DL2, KIR2DL3 and KIR2DS2 secreted larger amounts of IFN-γ more rapidly than their HLA-C2:KIR2DL1/S1 counterparts.

NK cell receptor repertoire analyses have demonstrated that patients with seasonal and pandemic influenza infections possess low numbers of absolute peripheral blood NK cells (Denney et al. 2010; Jost et al. 2011). However, these studies fail to provide information on the severity of the observed infections. In addition, a recent study has suggested that the absence of ligands for KIR3DL1/S1 and KIR2DL1, as well as the presence of KIR2DL2/L3 ligands, may be associated to severe infections caused by pandemic influenza A (H1N1) 2009 virus (La et al. 2011).

In order to assess the role of KIR genes in acute influenza infections, we have compared the KIR gene, haplotype and genotype content of patients having both mild and severe influenza infections against those of healthy controls having the same ethnicity and geographic background. The frequency of several individual KIR genes proved to be statistically associated with increased disease severity as was shown for KIR2DL5, KIR2DS5 and KIR3DS1. Although these genes are all present in the telomeric tB01 haplotype motif, the frequency of this association of genes did not prove to differ between the study groups. Instead, it was the centromeric cB01 and cB03 haplotype motifs which exhibited significantly higher frequency amongst influenza-infected patients and particularly amongst patients having severe disease. Other than the framework genes KIR3DL3 and KIR3DP1, the cB01 and cB03 haplotype motifs have KIR2DL5, KIR2DS5, KIR2DP1 and KIR2DL1 in common. However, KIR2DP1 and KIR2DL1 did not prove to be different amongst the three study groups. In all, these findings suggest that haplotype motifs bearing this KIR2DL5C/KIR2DS3S5C combination of genes are associated with greater severity of influenza illness as well as their corresponding extended haplotypes (cB01|tB01, cB01|tA01, cB03|cB01). Ligands for KIR2DL5 and KIR2DS5 have yet to be defined; this hinders the interpretation of the possible KIR/ligand interactions responsible for this effect. However, as some alleles of KIR2DL5 are known for not being expressed, it could be conceived that KIR2DS5 is responsible for the detrimental effect and that KIR2DL5’s statistical significance might arise from linkage disequilibria to KIR2DS5. This effect is not noted on the opposite end of the KIR2DS5 gene (KIR2DP1) as this pseudogene is virtually present in all Mexican mestizos studied (95 %).

The importance of KIR3DS1 in severe forms of the disease is highlighted not only by the statistical significance that the gene has on its own but also by the way it contributes to the significance of the cB01|tB01 extended haplotype. In this extended haplotype, the effect of the centromeric KIR2DL5 and KIR2DS5 genes adds to the significance of the telomeric KIR3DS1 gene. In addition, the relevance that this activating KIR gene has is further stressed by the contrasting effect that its inhibitory counterpart imposes. While the proportion of individuals bearing KIR3DS1 is higher amongst patients with severe forms of the disease with regard to the healthy controls, the proportion of individuals having KIR3DL1 but no KIR3DS1 is contrastingly lower than in controls. It is possible that the statistical significance of the 3DL1+/3DS1+ combination seen in patients with severe influenza might arise from the fact that most individuals having KIR3DS1 were 3DL1+ as the individual contribution of KIR3DL1 did not achieve statistical significance. In patients with HIV infection, KIR3DS1 and KIR3DL1 are important factors associated to control of viremia and progression of infection (Alter et al. 2009; Long et al. 2008; Martin et al. 2002). On the other hand, as suggested by La et al. (2011), the presence or absence of 3DL1/3DS1 ligands may be a relevant factor that needs to be considered to understand the role of these genes in association to severe influenza infections. Future studies are planned to address this KIR gene association in a larger cohort of influenza-infected patients.

As the aforementioned KIR haplotypes have a high content of activating and inhibitory genes, we also analysed whether the number of activating or inhibitory KIR genes had any relevance. We observed that the proportion of patients with severe influenza infection who had ≥3 activating KIR genes was higher than that seen in our healthy controls. This finding is similar to that described for other activating KIR genes and their association with death in Ebola virus infections (Wauquier et al. 2010). However, others have demonstrated an association between activating KIR genes, especially KIR2DS5 in the context of chronic hepatitis C virus infections (Khakoo et al. 2004). The high number of activating KIR genes found in haplotypes that include these motifs may result in a higher activation capacity of NK cells in these individuals resulting in an increased inflammatory response while this higher activation of the immune response might also be considered to lead to NK cell/CD8 lymphocyte exhaustion/depletion as described elsewhere (Hofmeyer et al. 2011; Wodarz et al. 1998).

Contrastingly, we also noted that the proportion of individuals having ≥5 inhibitory genes was higher than that seen in our healthy controls. Together, these findings suggest that KIR genotypes capable of encoding for a diverse KIR protein repertoire might in fact enable NK cells to enhance potent immune responses and increase clinical severity of influenza infections. In this study, we found an association between both activating and inhibitor genes and severe influenza infection. This could indicate that the total number of activating genes by itself may not result in a worse outcome of infection through activation or inhibition signals in NK cells. Rather, distinct combinations of KIR genes might be more influential than quantitative properties of KIR genes to balance activation and inhibition signals on NK cells in the immune network during infection.

Our findings provide further evidence of the clinical relevance of NK cells and KIR genes in the host response to influenza virus infections. Limitations to our study include the relatively low number of subjects which precluded the inclusion of other variables, such as antiviral use or presence of underlying disorders, in a multivariate statistical analysis. It is possible that other aspects of KIR diversity which were not analysed, such as allelic polymorphism and KIR protein expression levels, as well as the presence or absence of KIR ligands, may also influence the clinical manifestations of influenza. Also, a larger sample size is required to be able to make adjustments for multiple comparisons. Future studies are being planned to address these shortcomings through both genotypic and phenotypic characterization, including HLA typing. Nevertheless, our preliminary data should be useful to design hypothesis-driven studies that aim to validate these results. A better understanding of the genetic factors that contribute to susceptibility to influenza virus pathogenesis may aid in the development of immune intervention strategies directed towards controlling the severity of disease.