2.1. The “NorthernFlu” Consortium
To better understand the etiology of influenza epidemics in Norway, Sweden, Finland, Estonia, Latvia and Lithuania a “NorthernFlu” consortium was established in 2017. It consists of epidemiologists, clinicians and researchers from Norway, Sweden, Finland, Estonia, Latvia and Lithuania. The consortium aims to determine the impact of meteorological, societal, and genetic factors on IV stability and transmissibility and to improve prediction, prevention and treatment of severe IV infections.
2.3. Statistical Analysis
The FluNet database provided weekly statistics of specimens positive for influenza viruses, whereas meteorological factor data were measured daily. Therefore, we adjusted each meteorological factor (median) per week.
In order to investigate the potential relationship between meteorological factors and influenza activity (number of IV positive specimens) in all analyzed countries, correlation analysis and machine learning based modeling were applied. The Pearson correlation coefficient (r) between each meteorological factor and influenza activity was calculated using the “stats” R package with a threshold of significance, p = 0.05.
For the machine learning modeling, Random Forest (RF) was optimized to find the most predictive meteorological factors of influenza activity. More specifically, for each country, the time-series measurements for all six meteorological factors were used as input features (explanatory variables) and the number of IV-positive specimens as an output variable for the model. The optimal Random Forest hyper-parameters, which are estimated from the data and used to train the most accurate model, were identified using Bayesian optimization (mlrMBO R-package, version: 1.1.1), with five times repeated 10-fold cross validation. The optimized RF model (with the lowest root-mean-square error (RMSE) from cross-validation) was used to estimate the contribution of each meteorological factor as a predictive performance of IV activity. This was done by random permutation of the measurements of one of the meteorological factors, while keeping measurements for all other factors constant, followed by estimation of the impact of this procedure on model accuracy (i.e., percentage mean decrease in accuracy). RF R-package version 4.6-14 was applied for the model training, with the following hyper-parameters used for optimization: Number of trees, number of randomly sampled features at each tree split (mtry), and minimum size of terminal nodes.
To find correlation between colder winters and increases in the number of influenza-positive specimens for six countries, we applied the Cox–Stuart statistical test. In particular, the median winter temperatures (2010–2018) were calculated by taking a median of daily temperatures between 1st of December and the 28th of February. Then, the median winter temperatures were ordered by the number of total IV-positive specimens for each country each year. A Cox–Stuart statistical test was applied to the ordered median winter temperatures to find any significant increasing or decreasing trends (threshold of significance was set to p = 0.05).
2.4. Experimental Validation
Madin–Darby canine kidney (MDCK) cells were grown in Dulbecco’s Modified Eagle’s medium (DMEM; Gibco, Paisley, Scotland) supplemented with 100 U/mL penicillin and 100 μg/mL streptomycin mixture (Lonza, Cologne, Germany), 2 mM l-glutamine, 10% fetal bovine serum (FBS; Lonza, Cologne, Germany). The virus growth medium (VGM) contained 0.2% BSA, 2 mM l-glutamine, and 1 μg/mL l-1-tosylamido-2-phenylethyl chloromethyl ketone-trypsin (TPCK)-trypsin (Sigma-Aldrich, St. Louis, USA) in DMEM.
Human telomerase reverse transcriptase-immortalized retinal pigment (RPE) cells were grown in DMEM-F12 supplemented with 100 U/mL penicillin and 100 μg/mL streptomycin mixture, 2 mM l
-glutamine, 10% FBS, and 0.25% sodium bicarbonate (Sigma-Aldrich, St. Louis, USA) as described previously [15
]. The virus growth medium (VGM) contained 0.2% BSA, 2 mM l
-glutamine, 0.35% NaHCO3
, and 1 μg/mL TPCK-trypsin (Sigma-Aldrich) in DMEM-F12 (Gibco, Paisley, Scotland).
Human primary macrophages were derived from leukocyte-rich buffy coats from healthy blood donors. In particular, monocytes were isolated as described previously [16
]. Monocytes were seeded in 96- or 6-well plates and cultured in serum free macrophage media (Gibco, Paisley, Scotland) supplemented with 10 ng/mL granulocyte macrophage colony stimulating factor (GM-CSF; Sigma-Aldrich, St. Louis, USA) and 100 U/mL penicillin and 100 μg/mL streptomycin mixture at 37 °C and 5% CO2
for 7 days, polarizing the monocytes into macrophages of the acute pro-inflammatory M1-phenotype. Before stimulation, the media was replaced with fresh GM-CSF free macrophage media.
GFP-expressing influenza A/PR8-NS116-GFP strain (PR8-GFP) was purified by centrifugation in sucrose gradient as previously described [17
]. The purified virus was incubated at different temperatures for 48 h. Alternatively, the virus was exposed to UVC (λ = 254 nm) or to UVB (λ = 302 nm) using Hoefer UVC 500 Ultraviolet Crosslinker (20 J/cm2
) or VM25/30/GX trans-illuminator as UV sources, respectively. Resulting viruses were titered on MDCK cells using plaque assay as described previously [15
]. The virus titers were determined by calculating the plaque forming units (PFU) for each sample and were expressed as PFU/mL.
RPE cells were infected with the viruses at a multiplicity of infection (moi) of 1. After 24 h GFP expression was measured in infected cells using a fluorescent microscope (Zeiss Observer Z1, Zaventem, Belgium). Viability of cells were measured using Cell Titer Glow assay (Promega, Madison, USA). The luminescence was read with a PHERAstar FS plate reader (BMG Labtech, Ortenberg, Germany).
The A/Helsinki/P18/2009(pdm09) virus was isolated from nasopharyngeal aspirates of Finnish patients in 2009 [19
]. The viral genome was sequenced (GeneBank accession numbers: JQ173161–JQ173168). The virus was exposed to different temperatures, UVB or UVC radiation. Resulting viruses were titered on MDCK cells using plaque assay as described previously [15
]. The virus titers were determined by calculating the plaque forming units (PFU) for each sample and expressed as PFU/mL.
Macrophages were infected with mock, temperature/UV-treated or untreated viruses for 36 h. The viability of cells were measured using Cell Titer Glow assay using a PHERAstar FS plate reader.