Ice nucleators have shorter persistence in the atmosphere than 1 other airborne bacteria 2

Ice nucleation is a means by which the deposition of an airborne microorganism can be accelerated 10 under favourable meteorological conditions. Analysis of 56 snow samples collected at the high altitude 11 observatory Jungfraujoch (3580 m a.s.l.) revealed an order of magnitude larger dynamic range of ice nucleating 12 particles active at -8 °C (INPs-8) compared to the total number of bacterial cells (60 % was on average living). 13 This indicates a shorter atmospheric residence time for INPs-8. Furthermore, concentrations of INPs-8 decreased 14 much faster, with an increasing fraction of water precipitated from the air mass prior to sampling, than the 15 number of total bacterial cells. Nevertheless, at high wind speeds (> 50 km h) the ratio of INPs-8 to total 16 bacterial cells largely remained in a range between 10 to 10, independent of prior precipitation, perhaps 17 because of recent injections of particles in regions upwind. Based on our field observations, we conclude that ice 18 nucleators travel shorter legs of distance with the atmospheric water cycle than the majority of bacterial cells. 19 Pseudomonas syringae, a prominent ice nucleating bacterium, was successfully isolated from 3 of 13 samples 20 analysed. Colony forming units of this species constituted a minor fraction (10) of the numbers of INPs-8 in 21 these samples. Overall, our findings expand the geographic range of habitats where this bacterium has been 22 found and corroborates theories on its robustness in the atmosphere and its propensity to spread and to colonise 23


Introduction
The nucleation of ice in clouds is a process of primary relevance both for the radiative budget of clouds and for the development of precipitation (Cantrell and Heymsfeld, 2005;Mülmenstädt et al., 2015;Murray et al., 2012).
Most ice nucleating particles (INPs) active at moderate supercooling in the atmosphere are of biological origin (Murray et al., 2012).Pseudomonas syringae was the first organism found to produce an ice nucleation active molecule (Maki et al., 1974) and to occur as an active biological INP in clouds (Sands et al., 1982).As it is also a plant pathogen it received and continues to receive attention from biologists in the perspective of improving the protection of crops from the spread of microbial diseases (Lamichhane et al., 2014;Lamichhane et al., 2015).
The combination of both roles, of an ice nucleator and of a plant epiphyte and pathogen, sparked the bioprecipitation hypothesis (Morris et al., 2014;Sands et al., 1982).Part of the hypothesis is the idea that ice nucleation activity contributes preferentially to the deposition of the organisms with this property helping them to return to plant surfaces where they can proliferate (Morris et al., 2013).In fact, favouring the growth of ice crystals could be a powerful and effective means for airborne microorganisms to reduce their residence time Biogeosciences Discuss., doi:10.5194/bg-2016-496, 2016 Manuscript under review for journal Biogeosciences Published: 24 November 2016 c Author(s) 2016.CC-BY 3.0 License.
aloft.Previous research has illustrated that bacterial strains capable of nucleating ice can deposit rapidly under laboratory simulated cloud conditions, corresponding to cold temperature regime and supersaturation (Amato et al., 2015).Modelling studies suggest that the atmospheric residence time of singularly airborne bacterial cells is highly reduced if such bacteria act as condensation and ice nuclei in clouds (Burrows et al., 2009).Furthermore, snowfall has been shown to enrich for the presence of ice nucleation active strains of P. syringae compared to a range of other environmental contexts including within clouds (Morris et al., 2008;Morris et al., 2013).
Therefore, more direct evidence from field observations on the selective deposition of ice nucleation active microorganisms with precipitation is wanting.
Here we explore, through analysis of snow samples collected at a high altitude station, whether the capability to induce the formation and growth of ice crystals makes a discernible difference to the atmospheric residence time of INPs compared to the majority of bacterial cells, which are not INPs.We have sought to isolate the prominent ice nucleation active bacterium P. syringae at high altitude and compare its abundance with numbers of INPs -8 .
We have also carried out phylogenetic analysis in an attempt to identify potential sources of this bacterium in the atmosphere.

Sampling, counting of bacteria and INPs
Fifty six snow samples were collected at the High Altitude Research Station Jungfraujoch (7°59'06'' E, 46°32'51'' N, 3580 m a.s.l., Switzerland) during 11 short sampling campaigns from March to September 2013 and May to October 2014.Samples were collected with a 0.1 m 2 Teflon-coated tin for periods of 1.5 to 8 hours depending on precipitation intensity.The tin was rinsed several times with 70 % ethanol and sterile Milli-Q water between samples to avoid cross-contamination.Snow samples were melted at room temperature (about 16 °C) and analysed immediately on site for concentration of INPs in a drop freeze assay with the LINDA device as described in Stopelli et al. (2014).The sample was adjusted to physiological conditions (9 ‰ NaCl) and two filters were prepared for later analysis of bacterial number concentration.For each filter, 20 mL of sample were passed through the active area (6 mm diameter) of a 0.22 µm black polycarbonate membrane (Whatman).The filter was rinsed with 3 mL of sterile phosphate-buffered saline (PBS) and the staining agents were added: 10 µL SYBR green (100x) for total cell count, and 10 µL SYBR green and 10 µL propidium iodide (in Milli-Q water, 1‰ wt/wt) to facilitate counting cells that are live (i.e with intact membranes) and dead or dying (with permeable membranes).After incubation in the dark for 10 min, the stains were filtered away and the filter column rinsed with 3 mL of sterile water.Once dry, the filters were put on glass slides, 15 µL of antifading agent (5 mL PBS, 5 mL glycerol, 10 mg p-phenylendiamine) was added for preservation in dark at -20 °C until analysis.Blanks were periodically prepared from the Milli-Q water used to rinse the tin.Blank counts were generally less than 10 % of sample counts.Filters were analysed with a fluorescence microscope (Leica DM2500) with a 100x ocular lens and an objective with 10x10 10 µm grids.Bacterial cells were recognised by shape and size.Each time, at least 10 fields and 300 cells were counted.In filters stained solely with SYBR green all cells were visible under blue light, while in filters treated with SYBR green and propidium iodide only living cells were counted.Discuss., doi:10.5194/bg-2016Discuss., doi:10.5194/bg- -496, 2016 Manuscript under review for journal Biogeosciences Published: 24 November 2016 c Author(s) 2016.CC-BY 3.0 License.

Selective isolation and phylogenetic characterization of Pseudomonas syringae
Enough precipitation volume was available in 13 samples in a total of 15 samples collected in 2014 to study the presence of culturable P. syringae.The method employed for the selective isolation of P. syringae and its metabolic characterisation is described in Monteil et al. (2014) and in Morris et al. (2008).In addition to the above described procedure, the snow collecting tin was sterilised by dry heat (150 °C) for 20 min.Samples were concentrated 500 times by filtration of about 1 litre of melted snow across sterile 0.22 µm pore size polycarbonate filters and successive resuspension of the particles by stirring the surface of the filter into 2 mL of filtered precipitation water.The concentrated samples were dilution-plated on KBC, King's medium B supplemented with boric acid, cephalexin and cycloheximide, to isolate and calculate the abundance of P. syringae.Plates were incubated at 20-25 °C.Putative strains of P. syringae were purified on KB media (without the antibiotics and boric acid) where their production of fluorescent pigment could be observed under UV light (312 nm) and were then tested for the absence of arginine dihydrolase and for the absence of cytochrome C oxidase.Those that were negative for both the oxidase and arginine tests were suspended in sterile phosphate buffer and sent to the Plant Pathology Unit, INRA, Montfavet, France for the molecular identification using PCR primers specific for P. syringae: Psy-F: 5'-ATGATCGGAGCGGACAAG-3'; Psy-R: 5'-GCTCTTGAGGCAAGCACT-3' (Guilbaud et al., 2016).The strains confirmed to be P. syringae were also tested for their ice nucleating activity.After 3 days of growth on KB, suspensions of pure colonies corresponding to 10 8 cells mL -1 in 0.9 % NaCl were incubated 1 h in melting ice and subsequently tested for ice nucleation activity between 0 and -8 °C, with the freezing apparatus LINDA at 200 µL per tube (about 2•10 7 cells in each tube).The capacity of strains of P. syringae to induce a hypersensitive response (HR), indicative of the presence of a functional type III secretion system used in pathogenicity by the strains, was determined on tobacco plants (Nicotiania tobacum) by infiltrating bacterial suspensions of 48 h cultures at approximately 10 8 cells mL -1 into the leaves of the plant.After 24-48 h of incubation HR was revealed by the appearance of necrotic lesions at the site of infiltration.
This analysis of partial cts gene sequences was performed as described previously using P. syringae reference strains (Berge et al., 2014), atmospheric strains (Amato et al., 2007;Joly et al., 2013;Vaitilingom et al., 2012) and the 24 strains from this study that were positive with the PCR specific for P. syringae.Alignment of sequences was made with DAMBE (version 5.6.8) and a Neighbour Joining tree was built with MEGA (version 5.05; Tamura et al., 2011).

Environmental parameters and statistics
The fraction of water vapour lost from an air mass prior to sampling was calculated from the relative abundance Univariate and non-parametric statistics were carried out with PAST software version 2.17 (Hammer et al., 2001).

Concentration variability of bacterial cells and INPs -8
Concentrations of total bacterial cells in the 56 samples collected over the course of 18 months (excluding the winter period between 2013 and 2014) were mostly between 3.0•10 3 (10 th percentile) and 3.9•10 4 mL -1 (90 th percentile), a range which is coherent with previous observations carried out on precipitation and cloud water samples collected at several places around the world (Bauer et al., 2002;Joly et al., 2014;Šantl-Temkiv et al., 2013;Sattler et al., 2001;Vaïtilingom et al., 2012).Concentrations of INPs active at -8 °C or warmer (INPs -8 ) (Fig. 1, data taken from Stopelli et al., 2016) presented a similar seasonal trend but with an order of magnitude larger dynamic range (DR, i.e. the ratio between the largest and the smallest values) (10 th percentile: 0.2 mL -1 ; 90 th percentile: 68.1 mL -1 ).
The DR of atmospheric particles or molecules is largely determined by their atmospheric residence time: DR is larger for species with shorter residence times (e.g. 222Rn) than for those with longer residence times (e.g.CO 2 ; see for example Figure 2 in Griffiths et al., 2014).In fact, species with longer residence time have a higher chance to mix and integrate more sources and over wider areas.Furthermore, a longer residence time leads to a higher background concentration of a species in the atmosphere.Changing inputs due to changing source strength, or changing losses due to changing sink strength, only make smaller differences to this larger pool.
Hence, the relatively small DR of bacterial cells compared to that of INPs -8 suggests that bacterial cells have a longer residence time in the atmosphere than INPs -8 .This is likely to be due to the capacity of some bacteria and other particles carrying a biological ice nucleation active fragment (O' Sullivan et al., 2015) to incite ice formation and growth at -8 °C.
Surprisingly, the fraction of living cells among total bacterial cells was on average 0.6, with small standard deviation (0.1), despite the harsh environmental conditions such as low temperature (sometimes down to -25 °C) and intense solar radiation.The fraction of living cells was neither related to the fraction of water lost prior to sampling nor to wind speed, and had no relation with the number of INPs -8 .This finding does not exclude that bacteria can constitute an important fraction of INPs -8 .Rather, it suggests that i) the ice nucleating capability can be retained beyond cellular death under atmospheric conditions (Amato et al., 2015;Möhler et al., 2007) and/or ii) it can be linked not only to entire cells but also to cellular fragments or molecules released from cells (O' Sullivan et al., 2016;Pummer et al., 2015).
In an earlier study we had already found evidence for INPs being more efficiently deposited from precipitating clouds than the majority of particles larger than 0.5 µm (Stopelli et al., 2015).This evidence was based on the comparison of INPs -8 in precipitation with that of particles in air.In this work, by comparing number concentrations of both INPs -8 and bacteria in precipitation, we corroborate with field data the hypothesis that the ability to foster the formation and growth of ice crystals increases the chance of an airborne particle to be deposited.With an increasing fraction of water precipitated from air masses prior to arrival at the observatory, Biogeosciences Discuss., doi:10.5194/bg-2016-496,2016 Manuscript under review for journal Biogeosciences Published: 24 November 2016 c Author(s) 2016.CC-BY 3.0 License.
the number concentration of INPs -8 decreased much faster than that of bacteria (Fig. 2).Therefore, a more rapid loss from the atmosphere with precipitation is one factor contributing to the greater dynamic range of INPs -8 .
But, can we also find evidence for shorter time scales in the replenishment of their atmospheric concentrations?

The influence of wind speed on INPs -8 and bacteria
Wind speed is an important factor associated with enhanced number concentrations of INPs -8 in air masses at Jungfraujoch (Stopelli et al., 2016).At high wind speeds (> 50 km h -1 ) numbers of INPs -8 were 10 -2 to 10 -3 times the number of bacterial cells in precipitation, independent of prior precipitation (Fig. 3, red symbols).At lower wind speeds, the number of INPs -8 decreased much more rapidly than the number of bacterial cells (blue symbols).We interpret this observation in the following way.High wind speeds at Jungfraujoch could be associated also with high wind speeds in source regions upwind (e.g.: Swiss Plateau, Po Valley) which promote the aerosolisation of INPs -8 and bacterial cells (Fig. 2; Lindemann and Upper, 1985).The atmospheric residence time of total bacterial cells is longer than that of INPs -8 .Therefore, their background number concentration is relatively large and stable and it is not changed much by the additional cells aerosolised at high wind speeds in the region.The shorter atmospheric lifetime of INPs -8 means their background number concentration tends to be smaller, relative to the emission strength of these particles.Consequently, numbers of INPs -8 increase more substantially above background values at high wind speed, than is the case for bacterial cells.Hence, the ratio of INPs -8 to bacterial cells remains large, even when a substantial fraction of water had already precipitated from the air mass prior to sampling.At lower wind speeds, concentrations of INPs -8 are not restored from regional sources and we see more clearly the effect of a preferential washout of INPs -8 relative to bacterial cells (blue symbols in Fig. 3).The median ratio of INPs -8 to bacterial cells we observed at Jungfraujoch was 6.6•10 -4 , very close to what Joly et al. (2014) had found over the course of a year in cloud water at Puy de Dôme (5.5⋅10 -4 ).

Abundance and diversity of P. syringae
Pseudomonas syringae was successfully detected in three of the 13 samples analysed for the presence of this bacterium.These samples had over 1000 bacterial cells mL -1 and more than 10 INPs -8 mL -1 and were from clouds that had precipitated less than 70% of their water vapor prior to sampling (Fig. 2).Two-thirds of the strains (16/24), after culture in the laboratory, produced at least 1 cell in a suspension of 2•10 7 cells that was ice nucleation active at temperatures warmer than -8 °C.The freezing onset temperature of all ice nucleation active strains was warmer than -5 °C and was -2.1 °C for the most active strain (Table 1).The precipitation samples from which these strains were isolated were characterised by a relatively warmer onset of freezing (median onset temperature of -5.0 °C for samples with P. syringae vs. -6.8°C for those without P. syringae, p = 0.02, Mann-Whitney test).Although P. syringae was found in the samples with the largest numbers of INP -8 and with the warmest onset freezing temperature, only 2, 4, and 45 colony forming units (i.e.culturable cells) of P. syringae were present per litre of sample.This corresponds to 2 orders of magnitude less than what has been found in snowfall at lower altitudes (Monteil et al., 2014).We suspect that this can be due both to the preferential removal of P. syringae with precipitation as soon as it develops at lower altitudes and to the larger distance of Jungfraujoch from the sources where P. syringae entered into air masses.This possibly leads to longer exposure of P. syringae to UV radiation and desiccation, reducing its culturability when collected at higher altitude.The This is the first time this bacterium has been isolated at such altitudes (3580 m a.s.l.), and therefore this result expands the established limits for P. syringae's dissemination and survival under atmospheric conditions.
Sequencing of the cts gene for phylogenetic analyses was conducted for all 24 strains of P. syringae from culture plates from the precipitation samples (Table 1) to obtain insight into the possible origin of these strains.For all dates where P. syringae was isolated, the strains in each precipitation event were genetically diverse and represented a broad range of known phylogenetic groups (Table 1, Supplemental Fig. 1).This high diversity suggests that the process of entry of P. syringae into air masses moving up to Jungfraujoch involves either multiple events from a wide range of sources along the trajectory or a few entry events from common sources that harbor a high diversity of P. syringae that can readily be wafted into the air.Leaf litter, for instance, is one substrate that could have been a source of a high diversity of P. syringae, since it contains a high density and high genetic diversity of P. syringae, many of which are ice nucleation active, no matter the geographic origin and trajectory of the air masses (Berge et al., 2014;Monteil et al., 2012).
All but 3 of the 24 strains had functional type III secretion systems (hypersensitivity test).This suggests that they had potential to cause plant disease on some crops and illustrates the potential extent of spread of diseases caused by this pathogen.

Conclusions
Based on observations of INPs -8 and bacterial cells in 56 precipitation samples collected at Jungfraujoch we have shown that ice nucleation activity at -8 °C and warmer temperatures contributes to shorter atmospheric residence times due to a greater probability for wet deposition.For bacterial cells that are disseminated via the atmosphere, this property is advantageous because, by enhancing deposition, the bacteria reduce the time of of 18 O and 16 O in precipitation, expressed as δ 18 O.More details on this calculation are provided in Stopelli et al, 2015.Wind speed was measured at Jungfraujoch by MeteoSwiss and data was stored as 10-minute averages.Biogeosciences Discuss., doi:10.5194/bg-2016-496,2016 Manuscript under review for journal Biogeosciences Published: 24 November 2016 c Author(s) 2016.CC-BY 3.0 License.