Analysis of the Impact of African Dust Storms on the Presence of Enteric Viruses in the Atmosphere in Tenerife , Spain

Airborne viruses and their relation to dust storms, as a possible route for dispersion, have not been widely investigated. There are, however, studies that have described the airborne dispersal of pathogenic viruses and their potential impact on public and agronomical health. Atmospheric samples were collected in an urban area of Tenerife during 2009, 2010, 2012 and 2013 and screened for the presence of enteric viruses using PCR and sequencing. The potential relationship of viral data with African dust storms and other climatic variables (viz., the seasonality, origin of the air mass and PM levels) was analyzed. Enteroviruses and Rotaviruses were detected in 15.4% (20/130) and 36.9% (48/130) of the samples, respectively. No significant statistical relationships were observed with African dust storms or the origin of the air masses, although higher percentages of positives were obtained for dust storm days. Enterovirus detection was significantly linked to warmer seasons, and PM2.5 levels showed an inverse correlation with a rotaviral presence. This is the first multi-year report to describe the presence of Enterovirus and Rotavirus genetic sequences in air samples collected in an outdoor urban environment. The data illustrates the need for source region sampling to determine links and the influence of the weather and climatic and regional wind patterns on long-range atmospheric dispersion of viruses in future research efforts.


INTRODUCTION
Dust storms are a climatic phenomenon, originating in arid and desert regions of the planet, and are the main source of atmospheric dust on Earth (Middleton and Goudie, 2001).On a global scale, the primary source region is the Sahara-Sahel, which is believed to transmit more than 50% of all aerosolized dust to Earth's atmosphere at an estimated range of 400 to 2,200 Tg year -1 (Huneeus et al., 2011).Atmospheric dust carries diverse materials, including minerals, anthropogenic aerosols, and microorganisms that may be aerosolized from soils and congregated from various ecosystems during dust cloud movement.Bacterial concentrations in soil samples have been estimated in ≈ 10 6 microorganisms per gram (Kellogg and Griffin, 2006).Fungal concentrations may be at a similar magnitude, and virus concentrations are typically one to two orders of magnitude lower than bacterial concentrations (Griffin, 2007;Gonzalez-Martin et al., 2013).Most microorganisms are probably only transported relatively short distances due to their attachment to large soil particulate matter (PM) and many others may lose their viability during transport due to physical sources of stress (ultraviolet exposure, dehydration, etc.) (Griffin et al., 2011).However, some are able to resist adverse conditions experienced during transport and reach new niches many kilometers away (Hara et al., 2015).Knowledge about Saharan air mass transportation routes and the possible dispersion of microorganisms has driven studies about the potential relationship between dust storm events and the spread of pathogens (Gonzalez-Martin et al., 2014).
Among microorganisms, viruses may be more easily aerosolized due to their smaller size and potential attachment to finer sized particulate matter, but they are also very sensitive to adverse conditions during transport.For example, UV radiation has been identified as an effective way to inactivate viral particles (Tseng and Li, 2005), but it does not eliminate the possibility to detect their genomes, as previously demonstrated by Lee and colleagues.They tested an aerosolized vaccine with inactivated Influenza virus and the viral RNA was easily identified by regular molecular biology methods (Lee et al., 2011).So, there is, however, airborne dispersal of viruses.A large number of viruses have been detected in atmospheric samples, such as the Newcastle virus (Hietala et al., 2005) that infects poultry; or the porcine viruses PRRSV (Porcine Reproductive and Respiratory Syndrome Virus) and SIV (Porcine Influenza Virus) (Hermann et al., 2006;Weesendorp et al., 2008).These viruses have been studied to determine their dissemination routes inside of farms to limit substantial economic losses.In the eighties, transmission and dispersion of FMDV (Foot-and-Mouth Disease Virus) throughout Europe became a huge problem, and Gloster demonstrated that under certain circumstances, long-range (~60 km) atmospheric transmission of the virus was possible and it was recently demonstrated using statistical methodology (Gloster et al., 2005;Sanson et al., 2011).
Research in Asia have indicated a similar trend of airborne dispersion of FMDV in Korea and Japan and the potential relationship with dust storms emanating from Asian deserts (Ozawa et al., 2001;Joo et al., 2002;Sakamoto and Yoshida, 2002).Studies that have shown that some viruses are still viable after extended long-range transport have been recently reviewed (Gonzalez-Martin et al., 2014).
Regarding human health, pathogenic viruses such as Enteroviruses or Adenoviruses, whose mode of transmission is the fecal-oral route, have been isolated from aerosols sampled within hospital facilities (Tseng, 2010).Although to-date there is no confirmation of their capacity to be transported long distances, there are studies on Enterovirus outbreaks that point to the atmosphere as route of transmission (Ho et al., 1999;Lin et al., 2002).Other viruses responsible for epidemics have been investigated to identify its routes of dispersion.The Severe Acute Respiratory Syndrome (SARS) was found in air samples taken inside a clinic where infected individuals were hospitalized (Booth et al., 2005).Influenza viruses have been identified in aerosolized particles in a hospital emergency department, whose size was within the "breathable" fraction (Blachere et al., 2009) and other hospital environments (Tseng, 2010).Respiratory Syncytial Virus (RSV) has been isolated from air filters in a children's daycare facility (Prussin et al., 2016).Moreover, recent studies have addressed possible long-range transmission of these types of virus.Chen detected Ambient Influenza and Avian Influenza viruses in air samples collected from farms and a year later they confirmed their presence in Asian dust storms impacting Taiwan (Chen et al., 2009(Chen et al., , 2010)).
The Canary Islands are a Spanish archipelago formed by seven main islands, located between 100 and 500 km off the west coast of North Africa and they are frequently affected by African dust storms (Dorta et al., 2002).To date, studies conducted on the microbiological aspects associated with dust storms in the Canary Islands are practically nonexistent, although there are some reports about the potential consequences on human health such as possible correlations with elevated rates of allergies and asthma (Garcia-Carrasco et al., 2001;Sanchez-Lerma et al., 2009).In addition, previous studies have focused on the idea of the airborne route as an alternative dispersion of the human enteric viruses (Dennehy, 2000;Tseng, 2010;Thornley et al., 2011;Wan et al., 2012).These data initiated our interest in the research presented here whose focus was the presence of enteric viruses in the atmosphere over Tenerife during several years of study and their potential sources.The two main objectives were to detect and identify viral genomes of enteric viruses (Adenovirus, Enterovirus, Rotavirus and Norovirus) present in the air during different atmospheric conditions (with and without the influence of African dust) and to analyze possible correlation between the results observed and the climatic variables (frequency of dust storms, origin of air masses, PM levels).

Air Sampling
Air samples were collected on the roof of the University Institute of Tropical Diseases and Public Health of the Canary Islands (≈ 550 m.a.s.l.; 28°28ʹ43.71ʹʹN;-16°19ʹ 17.27ʹʹW).This facility is in an urban area within the center of San Cristobal of La Laguna city, located close to Tenerife North Airport.Two different air samplers were used, a Mattson-Garvin sampler (Barramundi Corporation, Homosassa, FL, USA) and an Omni 3000 (Innovaprep, Drexel, MO, USA).The switch to the Omni 3000 was due to a malfunction of the Mattson-Garvin unit.The Mattson-Garvin sampler used a 140 mm Petri dish containing 45 L of glycine buffer, pH 6.5.Sampling was performed continuously for two hours at flow rate of 28,2 L min -1 resulting in the screening of 3,396 L of air.Afterwards, the remaining volume of glycine buffer, ranging from 30-40 mL, was transferred aseptically to a 50 mL sterile tube.The Omni 3000 (Innovaprep, Drexel, MO, USA) collects air samples at a flow rate of 300 L min -1 , using cartridges containing 10 mL of distilled water.Sampling was performed for 20 min, which resulted in the screening of 6,000 L of air.All samples were kept under refrigeration until processed and the time between sampling and processing never exceeded 72 hr.

Samples Concentration
Concentration of air samples was performed through centrifugation at 5000 X g for 15 min in a Heraeus® Multifuge® 1 L-R (Thermo Scientific, Waltham, MA, USA) using Macrosep® Advance Centrifugal 30 K tubes obtaining a final volume of ≈ 3 00 µL (PALL, Port Washington, NY, USA).Samples collected with the Mattson-Garvin were split into two sub-samples (≈ 20-25 mL each) for the centrifugation step and the recovered volume was later unified (≈ 500 µL).

Nucleic Acid Extraction
The QIAamp® Ultrasens® Virus kit (Qiagen, Valencia, CA, USA) was used as it allows the joint extraction of RNA and DNA.The protocol was performed following the manufacturer instructions.Quantification of nucleic acids was conducted using a NanoDrop ND-1000 (Thermo Scientific) spectrophotometer, and ranged from 8 to 228 ng µL -1 for DNA and from 6 to 130 ng µL -1 for RNA.

PCR and Sequencing
Samples were screened for Human Adenoviruses, Enteroviruses, Rotaviruses and Noroviruses.The Adenovirus primer target was the hypervariable region 1-6 (HVR 1-6 ), based on Lu and Erdman (2006).Enterovirus detection was based on the amplification of the untranslated region UTR-5' (Casas et al., 1997).Rotavirus identification was performed using a VP6 gene target (Gray and Iturriza-Gómara, 2011).Prior to amplification, the rotavirus sample aliquot was mixed with DMSO to a final concentration of 7%, heated at 97°C for 5 min and chilled afterwards in an ice-ethanol bath to achieve the complete denaturalization of double stranded RNA (Gouvea et al., 1990;Iturriza-Gómara et al., 1999).Detection of Noroviruses was performed by targeting the RNA polymerase gene, following Moe and Wang publications (Moe et al., 1994;Wang et al., 1994).Primer sequences are detailed in Table S1 of Supplementary Material.The QIAquick® PCR Purification kit (Qiagen, Valencia, CA, USA) was used to purify amplicons of expected band sizes and were later sequenced in a Genetic Analyzer Applied 3500 (Life Technologies, Carlsbad, CA, USA) using a Big Dye Terminator v3.1 sequencing (Life Techonologies, Carlsbad, CA) kit.Sequences were analyzed with MEGA software and screened within the GenBank database using BLAST software from NCBI.

Climatic Variables
African dust intrusions were monitored over four different years (2009, 2010, 2012 and 2013) using the forecasts provided by AEMET (Spanish Meteorological Agency) based on climatic models: NAAPS, BSC-DREAM8b v2.0, Skiron, and ECMWF (CALIMA, 2013).The origin of all air masses sampled was verified using the same forecast models as well as the retro-trajectories provided by the HYSPLIT model (Fig. 1) Air mass trajectories were classified into three categories: African, Marine and European.African corresponded to those aerosols that originated and/or whose trajectories crossed part of the North African continent.Marine air masses had a predominant trajectory over the Atlantic Ocean; and European's were considered those European and Marine aerosols, whose trajectories crossed the European continent and the Atlantic Ocean.PM values were obtained from the Network for the Control and Surveillance of Air Quality in the Canary Islands (Canary Islands Government, 2013).

Statistics
Comparison between viral results and climatic variables was performed using IBM-SPSS Statistics (version 20.0) software using a level of significance of p < 0.05.Values of quantitative variables did not have a normal distribution; therefore, they were expressed as median and rank (minimum and maximum values).Values of qualitative variables were expressed as frequency and percentage, and comparisons were made using Pearson's Chi-squared test.Fisher's exact test was used in those cases were the expected frequency was lower than five.

RESULTS AND DISCUSSION
The study of long-range atmospheric transmission of microorganisms is an emerging field of research.Recent interest regarding their capacity to survive long-range dispersion, the possibility of airborne metabolic activity, their role in atmospheric nucleation and the significant role that dust storms serve as a vector has fostered a number of scientific endeavors (Amato et al., 2005;Griffin, 2007;Bowers et al., 2011;Joly et al., 2013;Schuerger et al., 2013;Gonzalez-Martin et al., 2014).However, airborne viral composition in open-field studies is still a lightly explored field.Here, we investigated the presence of enteric viruses in the outdoor atmospheric environment and the possible influence of African dust storms in the Canary Islands.Sixtyfive samples were collected with Mattson-Garvin sampler from 2009 till 2010 and another sixty-five were collected using the Omni 3000 during 2012 and 2013, generating a total of one hundred and thirty air samples.Detailed information about positive samples is shown in Table 1.

Viral Results
Two out of the four enteric viral groups analyzed were detected in the air samples collected: Enteroviruses and Rotaviruses.Regarding Enteroviruses, 15.4% (20/130) of the samples were classified within the Picornavirales order.Eleven positive samples were collected with the Mattson-Garvin sampler (55%, 11/20) and the other nine with the Omni 3000 sampler (45%, 9/20).Nineteen out of the twenty positive samples were identified within the Enterovirus genus, and among them, ten samples corresponded to Enterovirus A, five to Enterovirus B and two to Enterovirus C.Although one sample could not be classified as Enterovirus due to their low identity value to the GenBank results, in all cases, the closest matches were among the Enterovirus genus.
Previous research has studied aerial dispersion of Enteroviruses, but mostly in indoor environments (Couch et al., 1970;Meschievitz et al., 1984;Dick et al., 1987;Jennings et al., 1988;Myatt et al., 2003).Several studies were conducted in the 1970s in which air samples from wastewater-irrigated fields were screened using culture-based and fluorescent antibody methods and different Enteroviruses species were identified, to include Echovirus 7 and Coxsackievirus type B (Teltsch and Katzenelson, 1978;Teltsch et al., 1980).In 2010, Tseng detected Enteroviruses in a pediatric area of a hospital in Taiwan at similar rates as observed in this study, 15% (n = 33) (Tseng, 2010).
Forty-eight samples out of the 130 (36.9%) were positive for Rotaviruses and twenty-nine of them (29/48) were identified as Rotavirus A. Thirty-three (50.8%, 33/48) of the positives were samples collected with the Mattson-Garvin sampler and fifteen (23.1%, 15/48) with the Omni 3000.Seven samples (7/130) were simultaneously positive  The remaining sequences that could not be deposited are available by contacting the author.For additional information see Table S2 in the Supplementary Material.
for both targets (Enterovirus and Rotavirus), five of them collected with the Mattson-Garvin sampler and two with the Omni 3000 sampler.Sequences over 200 bp were submitted to GenBank (Ac.No. KU821663-KU821677).
Regarding Rotaviruses, reports are scarcer in relation to their airborne dispersal.A recent study in which air samples were collected in hospital rooms with Rotavirus-infected children, 75% of samples (n = 61) were positive for the pathogen's genome.This data indicates that the airborne route could be an important way of dispersion and would explain the explosive outbreaks that this virus is known to cause (Dennehy et al., 1998;Dennehy, 2000).Nevertheless, until this date, no study has been published regarding the presence of this viral group in outdoor atmospheric environments.However, despite identifying this group of viral genomes in the air samples described in this report, the issue of their viability, and therefore their potential pathogenicity, remains unsolved and will be addressed in future efforts.If viability were demonstrated for the viral groups detected in this study, then co-presence would indicate an increase in health risk if the viruses where proven to be pathogenic.Considering this issue, Ijaz and colleagues performed several indoor atmosphere studies focused on the influence of climatic conditions on Rotavirus survival (Ijaz et al., 1994(Ijaz et al., , 1985)).They found an 80% survival of airborne Rotaviruses when temperatures were ~20°C and relative humidity (HR) was 50%.Mean values of temperature and HR in Tenerife are 21.2°C and 63%, respectively (AEMET, 2014), which is suitable for Rotavirus atmospheric survival.Furthermore, a prediction model has been recently developed to assess hospitalizations due to Rotavirus gastroenteritis (Hervás et al., 2014).In that study, association of different climatic variables with Rotaviruses was evaluated and a positive correlation to atmospheric pressure was noted (higher pressures were related to higher viral prevalence).The authors theorized that elevated atmospheric pressure decreases the mobility of airborne particles, favoring their settlement.Additionally, since high pressure decreases evaporation, resulting humidity levels would favor survival of viruses (Hervás et al., 2014).Accordingly, the frequent influence of the Azores High over the Canary Islands archipelago (Dorta Antequera, 1996) could enhance atmospheric survival for Rotaviruses.
Positive samples for Enteroviruses and Rotaviruses were detected from both samplers used, but differences were noticed.Statistical analyses showed a positive correlation between the Mattson-Garvin sampler and Rotavirus detection, which may be due to the glycine buffer used.Glycine buffer could act as a better preservative for Rotaviruses (or genomes) than distilled water utilized by the Omni 3000.
Regarding Adenoviruses and Noroviruses, both have been detected in previous indoor studies and the airborne route has been considered in multiple gastrointestinal outbreaks (Marks et al., 2000(Marks et al., , 2003;;Said et al., 2008;Kuo et al., 2009;Aziz, 2010;Tseng, 2010;Wan et al., 2012;Moon et al., 2013;Xu et al., 2013;Nenonen et al., 2014).Only one publication has reported the detection of Norovirus genomes in an outdoor environment (Brooks et al., 2005).The study analyzed the influence and risks from aerosols generated during the application of biosolids and 3 out of 350 air samples collected were positive for Noroviruses; however, no positive sample was detected further than 5 m from the application sites.

Climatic Variables and Viral Data Comparison
While differences in the frequency of viral detection between variables were noted, statistical analysis indicated that the observations were heterogeneous in nature.Considering weather conditions, 46.2% (60/130) of the samples were collected on days under the influence of a dust storm, while 53.8% (70/130) were collected on days under different climatic conditions.No significant differences were noted for dust vs non-dust days regarding the viral results observed, although, 55% (11/20) of Enterovirus and 54.2% (26/48) of Rotavirus were detected during dust days.Previous publications have reported higher concentrations of microorganisms during dust storms days, using culture-based and real-time PCR methods (Griffin et al., 2001(Griffin et al., , 2003(Griffin et al., , 2006;;Chen et al., 2010) while others, besides perceiving the same increase, found similar composition patterns for different climatic conditions (Smith et al., 2013).However, simultaneous detection of Enteroviruses and Rotaviruses in the same sample was more frequent during dust days (6/7).
Regarding the origin of the air masses, 46.2% (60/130) of the samples were considered to be African (equal percentage of dust day samples), 45.4% (59/130) were catalogued as Marine and 8.4% (11/130) as European.Similar percentages of positives were observed for both, Enteroviruses and Rotaviruses, in relation to origin.Thus, 55% (11/20) and 54.2% (26/48) of the positives were obtained from African air masses, 40% (8/20) and 42% (20/48) from marine air masses and 5% (1/20) and 4% (2/48) from European air masses.No significant differences were noted that could infer a relationship between the origin of the air mass and the viral data.Considering the specific origin of African air masses and results obtained, with Morocco being the most frequent source of African dust, these data correlate with other previously published analyses, where it was considered the primary source of African dust in the Canary Islands, based on fingerprinting of deposited dust and analyses of back-trajectory data (Coude-Gaussen et al., 1987;Bergametti et al., 1989;Grousset et al., 1992;Kandler et al., 2007;Muhs, 2012).
Warmer seasons produced a higher relative percentage of positive samples (63.4%, 26/41) than cooler ones (39.2%, 35/89).This trend is opposite to observations obtained in Korea, using metagenomics, by Whon and colleagues, who found fluctuations in the airborne viral concentrations, with a peak during autumn-winter (Whon et al., 2012).Two seasonal groups were analyzed (autumn/winter vs spring/summer) to perform statistical analysis and significant differences were obtained for Enteroviruses, with a higher probability of being detected during spring/summer season, using Pearson's chi-squared test (p = 0.003; RR 0.239; IC95% 0.089-0.642).These results seem to correlate with prior analyses of seasonality, where Enterovirus outbreaks have been linked to summer and autumn (Stalkup and Chilukuri, 2002) and Rotaviruses outbreaks (in temperate climates) linked to autumn and winter, with a peak at the end of the winter (in tropical environments they can occur throughout the year) (Blacklow, 2013).Although the airborne transmission route is not currently considered as substantial, the data obtained in the present work seems to coincide with those seasonal trends.
Mean PM 10 value for dust days was four times higher than for non-dust days (44.2 and 11.6 µg m -3 , respectively), and the limit recommended by the European Directive (European Commission, 2008), 50 µg m -3 , was exceeded 12.3% (16/130) of the sampled days.In the case of PM 2.5 , the mean value for the samples collected during dust days, 18.4 µg m -3 , was more than twice the value obtained for non-dust days, 7 µg m -3 .Regarding the limit recommended by the World Health Organization (WHO, 2006) 25 µg m -3 , it was surpassed 10% (13/130) of the days.When viral results were contrasted with the PM 10 levels, no statistical difference was observed, while when analyzed regarding PM 2.5 values, an inverse correlation was detected for Rotaviruses (p = 0.047; RR 2.186; IC95% 1.005-4.578).Levels of particulate matter and their influence on morbidity and mortality have been previously reviewed (de Longueville et al., 2013) and conflicting results have been noted.While some authors have found a clear correlation between parameters (Villarrubia et al., 2010;Chan and Ng, 2011;López-Cadelis et al., 2014), others have reported negative correlation (Kashima et al., 2012).Considering the methodology used to detect airborne microorganisms, there is the potential for high concentrations of particulate matter acting as an inhibitor for PCR-based methods.In this case, experiments to detect viral, bacterial and fungal genomes from air samples proved to be successful with concentrations of particulate matter up to 50 µg per sample (Fabian et al., 2010;Hospodsky et al., 2010).In the present work, a significant correlation was noted between PM 2.5 and Rotavirus results, showing that lower PM levels increased the probability of detection.This trend would agree with other studies that have observed that high concentrations of particulate matter may inhibit molecular biology tools utilized to detect airborne microorganisms (McDevitt et al., 2007).In our sample set, no positive sample was obtained when the highest PM levels were recorded.

CONCLUSIONS
Although viability issues are yet to be examined, the results presented in this work report the presence of Enterovirus and Rotavirus genomic sequences in air samples from an outdoor urban environment.The hypothetical influence of African dust storms on the airborne viral composition was not confirmed.Increasing the sampling period and including samples from source regions could provide information about whether the detected microorganisms originated in Africa or in other locations.The detection of Enteroviruses was seasonally influenced, but this apparent trend should be examined in future efforts in consideration of outbreaks occurring in potential source regions.Earlier described associations between desert dust storms and an increase in particulate matter (PM) levels in the Canary Islands were confirmed, and trends were noted in the viral presence, although they were not statistically significant.In addition, the results warrant investigating potential modifications of existing nucleic extraction methodologies to increase extraction efficiencies and to limit PCR inhibitor carryover.

Fig. 1 .
Fig. 1.Examples of air mass trajectories (HYSPLIT).On the left, an air mass collected on March 30 2009 originated over the North Atlantic Ocean.On the right, an air mass collected on June 26 2012, originated over North Africa.

NEGa
Sample code: A = Mattson -Garvin samples and B = Omni3000.b Seasons: SPR = spring, SUM = summer, AUT = autumn, WIN = winter.c DUST: YES = dust day, NO = no dust day.d Origin: AFR = African, MAR = Marine, EUR = European.e PM levels: N = null value.f Viruses: NEG = Negative result.Positive results are indicated with the species identification.Sequences over 200 bp were submitted to GenBank and accession numbers are attached to the virus ID.

Table 1 .
List of positive samples and associated climatic variables.