California State University, Northridge

Home , Thermoactinomycetaceae

CALIFORNIA STATE UNIVERSITY, NORTHRIDGE

Identifying Changes in the Active, Dead, and Dormant Microbial Community Structure

Across a Chronosequence of Ancient Alaskan Permafrost

A thesis submitted in partial fulfillment of the requirements

For the degree of Master of Science in Biology

Alexander Burkert

December 2017

The thesis of Alexander Burkert is approved:

______

Dr. Kerry Cooper Date

______

Dr. Gilberto Flores Date

______

Dr. Rachel Mackelprang, Chair Date

California State University, Northridge

ii Acknowledgements

There are many people who have contributed to the success of this work and who deserve acknowledgement and thanks.

To my advisor, Dr. Rachel Mackelprang of the College of Math and Science at CSU Northridge, who took a chance on a student with very little experience in microbiology. I have been afforded more opportunities in her lab than I had ever expected from this program. From collecting my own permafrost samples in Fairbanks, Alaska to presenting my research in Nuuk, Greenland and even learning a little bit of beekeeping along the way. Her dedication to science is an inspiration.

To the members of my committee: if Dr. Gilberto Flores, of the College of Math and Science at CSU Northridge, had a dollar for every question of mine that he answered, he would be a very wealthy man. I appreciate his ability to get to the heart of the problem and give me guidance without directly telling me what to do. I also thank Dr. Kerry Cooper, of the College of Math and Science at CSU Northridge, for his jovial persona and his valuable input in my thesis project.

To our collaborators, thank you to Tom Douglas from the Alaska Projects Office of the Cold Regions Research and Engineering Laboratory for his assistance collecting permafrost core samples and to Mark Waldrop of the United States Geological Survey for his help performing soil chemistry analysis on our samples.

To the community of brilliant scientist in my lab (The Mackelprang Gang) and in my cohort at CSU Northridge who have provided both camaraderie and a place to commiserate with others who understand the late nights and failed experiments that led to this work.

Lastly to my family and friends who have been there to keep me grounded and have encouraged me at every step of the way, I thank you.

Alex Burkert

iii Table of Contents

Signature Page ...... ii Acknowledgements ...... iii List of Figures ...... v List of Tables ...... ix Abstract ...... x 1. Introduction ...... 1 2. Materials & Methods ...... 9 Permafrost Sample Collection ...... 9 Permafrost Subsampling ...... 12 Soil Chemistry ...... 12 Cell Separation for Enumeration via Microscopy ...... 12 Live/Dead Staining ...... 13 DAPI Staining ...... 13 Cell Enumeration...... 13 Separation of Biomass from Soil Matrix...... 14 Depletion of DNA from Dead Cells via Propidium Monoazide Treatment ...... 15 Endospore Enrichment via Lysozyme Enzyme Treatment ...... 15 DNA Extraction...... 17 PCR Amplification and Sequencing ...... 17 Statistical Analysis ...... 18 3. Results ...... 20 Soil Chemistry ...... 20 Cell Enumeration...... 20 16S rRNA gene-based community analysis ...... 24 Depletion of DNA from Dead Cells via Propidium Monoazide Treatment ...... 32 Endospore Enrichment via Lysozyme Enzyme Treatment ...... 33 4. Discussion/Conclusion ...... 39 Literature Cited ...... 46 Appendix A: Supplementary Material ...... 55

iv List of Figures

Figure 1. Experimental strategy— To obtain direct counts for live, dead, and total cells, we first separated cells from soil using a Nycodenz density centrifugation and stained using either a Live/Dead differential stain or DAPI stain before counting cells via fluorescence microscopy. In addition, we separated biomass from soil using a gravity separation technique and treated with either a propidium monoazide treatment (to deplete DNA from dead organisms) or a lysozyme enzyme treatment (to enrich for endospores). We conducted 16S rRNA gene amplicon sequencing for each treatment plus an untreated control for all age categories to observe changes in alpha diversity, beta diversity, and taxa abundance in the active dead, and dormant populations...... 8

Figure 2. Map of Alaska with the marked location of the United States Cold Regions Research and Engineering Laboratory (CRREL) Permafrost Tunnel Research Facility. The tunnel is located 16 km north of Fairbanks, Alaska. (Figure from Cyzewski et al 2010)...... 10

Figure 3. Photograph taken from the entrance to the CRREL Permafrost tunnel...... 10

Figure 4. Schematic drawing of the CRREL Permafrost tunnel. The age of exposed permafrost increases with distance from the tunnel portal. (Figure from Mackelprang et al 2017)...... 11

Figure 5. Diagram showing the core sampling strategy used in the CRREL Permafrost Tunnel. The cores were dug into the wall at three depths. Microbiological studies were conducted with the second and third depth cores which were interior to the wall and less likely to suffer from contamination or thaw. The outermost core was used for soil chemistry analysis. (Figure from Mackelprang et al 2017)...... 11

Figure 6. Direct cell counts as determined by fluorescent microscopy using SYTO 9 (B), propidium iodide (C), and DAPI (D). Total counts were highest in the intermediate aged category for the live counts (B, Kruskal Wallis, X2(2) = 46.25, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01), dead counts (C, Kruskal Wallis, X2(2) = 53.16, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01), and total counts gdw-1 (D, Kruskal Wallis, X2(2) = 53.58, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01). The total counts gdw-1 were lowest for the oldest samples (D, Dunn’s post hoc test, p < 0.05). (** = p < 0.01, * = p < 0.05). Values show averages of five replicates and error bars show standard error of the mean...... 22

Figure 7. Direct cell counts determined by fluorescent microscopy showed that the proportion of live cells increased with increasing age. The intermediate aged samples had a significantly higher proportion of live cells than the youngest samples (Kruskal Wallis, X2(2) = 9.84, p < 0.01, Dunn’s post hoc test, ** = p < 0.01). Values show averages of five replicates and error bars show standard error of the mean...... 23

Figure 8. Alpha diversity estimates based on 16S rRNA gene sequences for samples depleted of DNA from dead cells across the three age groups were measured using QIIME. There were significant differences between age groups for observed species (A,

v ANOVA, F (2,6.99) = 246.8, p < 0.01), Shannon (B, ANOVA, F (2,7.72) = 239, p < 0.01), Chao1 (C, ANOVA, F (2,6.65) = 247.98, p < 0.01), and PD Whole Tree (D, ANOVA, F (2,7.46) = 363.4, p < 0.01). The youngest samples were most diverse (Tukey HSD post hoc test, ** = p < 0.01), followed by the oldest samples (Tukey HSD post hoc test, ** = p < 0.01), and the intermediate aged samples were least diverse (Tukey HSD post hoc test, ** = p < 0.01). Values show averages of five replicates and error bars show standard error of the mean...... 25

Figure 9. Alpha diversity estimates based on 16S rRNA gene sequences for the endospore-enriched samples and non-enriched controls across the three age groups were measured using QIIME. The endospore-enriched samples were less diverse than the non- enriched controls from the intermediate aged samples for observed species (A, Student’s t-test, t (7.99) = -4.83, ** = p < 0.01), Shannon, (B, Student’s t-test, t (7.82) = -3.13, * = p < 0.05), Chao1 (C, Student’s t-test, t (7.97) = -4.47, ** = p < 0.01), and PD Whole Tree (D, Student’s t-test, t (7.35) = -4.67, ** = p < 0.01). For PD Whole Tree, the endospore- enriched samples were also less diverse than the non-enriched controls from and the youngest aged samples (D, Student’s t-test, t (5.78) = -3.37, * = p < 0.05) and the oldest aged samples (D, Student’s t-test, t (7.83) = -2.65, * = p < 0.05). Values show averages of five replicates and error bars show standard error of the mean...... 26

Figure 10. Community composition of samples differ based on age and treatment. Principal Coordinates Analysis (PCoA) based on weighted Unifrac values of 16S rRNA gene sequences. Samples differed significantly by age (A, PerMANOVA, R2 = 0.83, F = 135.64, p < 0.001) and treatment for the 19K aged category (B, PerMANOVA, R2 = 0.51, F = 5.65, p < 0.001), the 27K aged category (C, PerMANOVA, R2 = 0.42, F = 3.87, p < 0.01), and the 33K aged category (D, PerMANOVA, R2 = 0.37, F = 3.14, p < 0.01). .... 28

Figure 11. Community composition of samples differ based on ice content. Principal Coordinates Analysis (PCoA) based on weighted Unifrac values of 16S rRNA gene sequences. Samples differed significantly by ice content (PerMANOVA, R2 = 0.21, F = 15.59, p < 0.001)...... 29

Figure 12. Non-metric multidimensional (NMDS) scaling based on weighted Unifrac values of 16S rRNA gene sequences. Age (PerMANOVA, R2 = 0.89, p < 0.001), ice content (PerMANOVA, R2 = 0.33, p < 0.001), and DOC (PerMANOVA, R2 = 0.86, p < 0.001, DOC data from Mackelprang et al 2017) significantly correlate to NMDS coordinates. pH did not significantly correlate to NMDS coordinates (p > 0.05)...... 30

Figure 13. Average relative abundance of 16S rRNA gene amplicons at the phylum level ranked by abundance. The phyla with significant differences based on age and treatment, included the seven most abundant phyla: Proteobacteria, Actinobacteria, Firmicutes, Bacteroidetes, Chloroflexi, Acidobacteria, Planctomycetes as well as Chlamydiae (Linear Mixed Effect Model, p < 0.01)...... 31

Figure 14. Firmicutes relative abundance was significantly greater in the endospore- enriched samples compared to the non-enriched controls for the youngest (Mann- Whitney-Wilcoxon test, U = 0, ** = p < 0.01) and intermediate aged category (Mann-

vi Whitney-Wilcoxon test, U = 2, * = p < 0.05). These differences were driven by the classes Bacilli and Clostridia. Values show averages of five replicates and error bars show standard error of the mean...... 34

Figure 15. Bacilli were significantly more highly represented in the endospore-enriched samples compared to the non-enriched controls for the youngest (Mann-Whitney- Wilcoxon test, U = 0, ** = p < 0.01), intermediate (Mann-Whitney-Wilcoxon test, U = 2, * = p < 0.05) and oldest samples (Mann-Whitney-Wilcoxon test, U = 1, * = p < 0.05) while the Clostridia were underrepresented (Mann-Whitney-Wilcoxon test, U = 2, * = p < 0.05). Change in relative abundance of Bacilli and Clostridia in endospore-enriched samples compared to the non-enriched controls across the three age categories. A negative value implies underrepresentation in the endospore-enriched samples compared to the non-enriched control while a positive value implies overrepresentation. Values show averages of five replicates and error bars show standard error of the mean...... 35

Figure 16. Actinobacteria relative abundance was significantly greater in the endospore- enriched samples compared to the non-enriched controls from the youngest aged category (Mann-Whitney-Wilcoxon test, U = 0, ** = p < 0.01). This was driven by the family Micrococcaceae within the Actinomycetales as well as the families Gaiellaceae, and Solirubrobacteraceae. Values show averages of five replicates and error bars show standard error of the mean...... 36

Figure 17. Chlamydiae relative abundance was significantly greater in the endospore- enriched samples compared to the non-enriched controls from the youngest aged category (Mann-Whitney-Wilcoxon test, U = 0, ** = p < 0.01). This trend was driven by an increase in the relative abundance of the family Chlamydiaceae. Values show averages of five replicates and error bars show standard error of the mean...... 37

Figure 18. Direct cell counts determined by fluorescent microscopy did not show significant differences in the ratio of live to dead cells across the three age categories (Kruskal Wallis, X2(2) = 1.38, p > 0.05). Values show averages of five replicates and error bars show standard error of the mean...... 55

Figure 19. Alpha diversity estimates based on 16S rRNA gene sequences for non - depleted controls across the three age groups were measured using QIIME. There were significant differences between age groups for observed species (A, ANOVA, F (2,7.11) = 196.67, p < 0.01), Shannon (B, ANOVA, F (2,7.83) =192.37, p < 0.01), Chao1 (C, ANOVA, F (2,7.77) = 188.64, p < 0.01), and PD Whole Tree (D, ANOVA, F (2,7.38) = 386.71, p < 0.01). The youngest samples were most diverse (Tukey HSD post hoc test, ** = p < 0.01), followed by the oldest samples (Tukey HSD post hoc test, ** = p < 0.01), and the intermediate aged samples were least diverse (Tukey HSD post hoc test, ** = p < 0.01). Values show averages of five replicates and error bars show standard error of the mean...... 56

Figure 20. Community composition of samples differ based on age and treatment. Principal Coordinates Analysis (PCoA) based on unweighted Unifrac values of 16S rRNA gene sequences. Samples differed significantly by age (A, PerMANOVA, R2 =

vii 0.38, F = 17.26, p < 0.001) and treatment for the 19K aged category (B, PerMANOVA, R2 = 0.259, F = 1.87, p < 0.001), the 27K aged category (C, PerMANOVA, R2 = 0.24, F = 1.66, p < 0.001), and the 33K aged category (D, PerMANOVA, R2 = 0.22, F = 1.49, p < 0.001)...... 57

viii List of Tables

Table 1. Percent ice content and pH across the three time periods. There were no significant differences between age categories for ice content (ANOVA, F (2, 5.94) = 2.06, p > 0.05) or pH (ANOVA, F (2, 12) = 0.6, p > 0.05. Values are averages of five biological replicates...... 20

Table 2. The average DNA concentration (ng/uL) decreased by ~50% (Students t-test, t (26.79) = 3.11, ** = p < 0.01) in the propidium monoazide treated group (DNA depleted samples) compared to the untreated group (non-depleted controls). Significance shows pairwise comparison between DNA concentration in DNA depleted samples and non- depleted controls...... 32

Table 3. Average percent difference in relative abundance between the endospore- enriched samples and non-enriched controls across the three age categories (Mann- Whitney-Wilcoxon test, ** = p < 0.01, * = p < 0.05). A negative value implies underrepresentation in the endospore-enriched sample compared to the non-enriched control while a positive value implies overrepresentation. Values show averages of five replicates and error bars show standard error of the mean...... 38

Table 4. Average percent difference in relative abundance between the endospore- enriched samples and non-enriched controls across the three age categories (Mann- Whitney-Wilcoxon test, ** = p < 0.01, * = p < 0.05). A negative value implies underrepresentation in the endospore-enriched sample compared to the non-enriched control while a positive value implies overrepresentation. Values show averages of five replicates and error bars show standard error of the mean...... 58

ix Abstract

Identifying Changes in the Active, Dead, and Dormant Microbial Community Structure

Across a Chronosequence of Ancient Alaskan Permafrost

Alexander Burkert

Master of Science in Biology

Permafrost— perennially frozen soil— hosts a diversity of microorganisms. Permafrost microbial communities survive and reproduce for millennia despite extreme conditions such as water stress, subzero temperatures, high salinity, and low nutrient availability.

Most studies targeting permafrost microbial communities use DNA-based methods such as metagenomic and 16S rRNA gene sequencing. However, constant subzero temperatures may preserve DNA from dead organisms for extended periods of time making it difficult to distinguish between active, dead, and dormant cells. This is particularly concerning in increasingly ancient permafrost because dormancy may be a survival strategy and DNA from dead cells may accumulate over time. To circumvent this hurdle, we applied live/dead differential staining coupled with microscopy, endospore enrichment, and selective depletion of exogenous DNA and DNA from dead cells to permafrost microbial communities across a Pleistocene permafrost chronosequence (19 kyr, 27 kyr, and 33 kyr). Cell counts and analysis of 16S rRNA gene

x amplicons from live, dead, and dormant cells revealed how communities differ between these pools and how they change over geologic time. We found clear evidence that cells capable of forming endospores are not necessarily dormant and that the propensity to form spores differed between taxa. Specifically, Bacilli are more likely to form endospores in response to long-term stressors associated with life in permafrost than members of Clostridia, which are more likely to persist as vegetative cells over geologic timescales. We also found that exogenous DNA preserved within permafrost does not bias DNA sequencing results, since its removal did not significantly alter microbial community composition. Lastly, the results of our cell enumeration confirmed a previous study that permafrost age and percent ice content acted as drivers of microbial cell abundance and diversity. Total cell counts and alpha diversity decreased between our youngest and oldest samples, while the proportion of live cells increased in older samples compared to younger samples suggesting that the permafrost environment selects for organisms adapted to survive. Taken together, these data contribute to our understanding of how microbial life adapts and survives in the extreme permafrost environment across geologic timescales.

xi 1. Introduction

A combination of low water potential, limited exchange of nutrients and waste, constant subzero temperatures, and background radiation make permafrost an extreme environment1. Despite these challenges, bacteria, archaea, and fungi survive and reproduce within the frozen subsurface conditions of permafrost soils for millenia2 3 4 5 6 7

8. Most organisms isolated from permafrost are psychrotolerant or mesophilic, rather than psychrophilic; although they live in frozen conditions, it is not their ideal growth temperature. This supports the idea that permafrost microbial communities are a

“community of survivors”9 that have been forced to adapt to subzero conditions rather than organisms that innately prefer extreme cold.

Terrestrial permafrost is an analog for frozen extraterrestrial environments because of the conditions they share including a lack of oxygen, constant subzero temperatures, limited access to nutrients, and limited water availability10 11. Permafrost soils range in age from our current epoch, the Holocene, which began 11,700 years before present to the Pleistocene which began 2.5 million years before present12. The evidence of viable organisms in ancient terrestrial permafrost is of interest to those searching for potential life in even older cold environments on distant worlds13 14 15 16. If we can identify predictable changes among microbial communities in response to age within terrestrial permafrost, we may be able to uncover strategies necessary for survival through geologic time on other cryogenic cosmic bodies.

In addition to permafrost’s importance to the field of exobiology, it also has implications that are relevant to biogeochemical cycles on Earth. Permafrost contains approximately twice as much carbon as the atmosphere currently contains, an estimated

1 1672 Pg of stored organic carbon, in the form of frozen plant material and debris17 18 19.

Climate change is warming our planet and thawing these soils, allowing microorganisms to degrade this massive carbon pool20. Upon thaw, microbial metabolic processes release this stored carbon into the atmosphere in the form of the greenhouse gases CO2 and CH4.

The question remains how distinct microbial communities in varying permafrost soils worldwide will respond to thaw. Through meta-analysis, Schädel et al (2014) found that the release of stored carbon in the form of CO2 could be rather quick. By investigating thaw related carbon degradation in aerobic laboratory incubations from permafrost soils across the Arctic they found that up to 90% of original carbon could be degraded in the duration of the only 50 years at a temperature of 5°C21. Additionally, a pan-Arctic meta- analysis by Treat et al (2015) investigated CH4 and CO2 production in anoxic soils in response to vegetation cover, soil properties, and environmental conditions. While total carbon degradation was reduced by 75-85% compared to oxic conditions, the maximum

22 CH4 production was 4 – 5 times higher in saturated anoxic soils compared to dry soils .

Permafrost thaw can often lead to soil saturation, thereby altering microbial community structure and increasing production of the much more potent greenhouse gas CH4.

Understanding the factors that structure microbial communities in permafrost soils will allow us to build more effective models to predict the impact of permafrost thaw in the global carbon budget.

Currently, much of our understanding of permafrost communities stems from culture independent techniques such as 16S rRNA gene amplicon sequencing and metagenomics23. Microbial communities within permafrost vary widely across the globe2.

For instance, the number of viable microorganisms in permafrost soils, measured through

2 culture dependent plate counting and via fluorescence microscopy, varies globally from

105-108 cells per gram of dry weight24. Arctic permafrost hosts a diversity of taxa, which include Firmicutes, Deltaproteobacteria, Alphaproteobacteria, Actinobacteria,

Chloroflexi, Acidobacteria, and Bacteroidetes2 25 26 27 28 29 though the relative abundance of these groups varies substantially by location. These organisms are metabolically diverse and include heterotrophs, anaerobes, methanogens, sulfur-oxidizers and reducers, iron reducers, nitrogen fixers, nitrifiers, and denitrifiers2 5 23 29 30 31.

Permafrost soils are not homogenous, therefore taxonomic differences observed across Arctic permafrost are expected. Several factors have been identified as potential drivers of permafrost microbial community composition. In permafrost, ice content32 directly affects microbial community composition. Additionally, permafrost age acts as a strong selective pressure controlling microbial community composition5 33 34. Pleistocene permafrost samples have a lower number and diversity of culturable bacteria compared to samples from the Holocene35 36. In addition, Kao-Kniffin et al compared 16S rRNA genes from permafrost microbial communities across a chronosequence of drained lake basins ranging in age from 50 to 5,000 years in Barrow, Alaska 37. They found that permafrost underlying the older lake basins had a distinct microbial community, dominated by gram-positive bacteria, compared to permafrost underlying younger lake basins. Soil pH also has a direct influence over microbial community composition in temperate soils 38 39 and in active layer Arctic soils40 41, but we do not know the extent to which pH drives community composition in permafrost soils. While the differences observed in permafrost globally could be due to ice content, age, or pH, without a

3 comprehensive survey it is unclear if other factors contribute to microbial community composition in permafrost.

Adaptations to extreme cold include alterations to cellular membranes, altered protein structure, genomic redundancy, production of compatible solutes, and entering a viable but non-culturable state. Cold temperatures limit diffusion across membranes therefore common cold response mechanisms include increasing the ratio of polyunsaturated/saturated fatty acids, upregulation of genes involved in LPS biosynthesis and peptidoglycan synthesis42 43, and upregulation of genes involved in membrane transport44. Additionally, subzero temperatures limit protein flexibility and activity.

Several psychrophilic organisms isolated from permafrost have altered amino acid residues to limit hydrogen bonding and salt bridge formation allowing proteins to remain flexible at colder temperatures6 45. Genomic redundancy, in the form of increased abundance and diversity of chaperones (which serve to correctly fold proteins) and tRNAs (an increased number of which can compensate for slow diffusion in cold temperatures), can contribute to cold temperature adaptation. As temperatures decrease, ice crystals can nucleate within the cytoplasm rupturing the cell. Many cold adapted organisms produce antifreeze proteins46, trehalose47 48, or compatible osmotic solutes such as, glycine, betaine, and mannitol49 to limit ice nucleation within the cytoplasm.

Also, by entering a viable but non-culturable state (VBNC), in which a bacterium maintains a low level of metabolic activity but does not divide, an organism can survive harsh conditions50 51. Lastly, by investigating metagenomes from a permafrost chronosequence, Mackelprang et al (2017) found that genes associated with stress response, horizontal gene transfer, chemotaxis, sensing the environment, scavenging

4 from dead organisms, and dormancy all increased in abundance as sample age increased suggesting these functions are related to survival in permafrost34.

One challenge associated with studying microorganisms using DNA-based techniques is the presence of DNA from dead microorganisms. DNA from dead organisms can be preserved over geologic timescales due to the frozen environment52.

DNA-based approaches do not distinguish between these preserved dead and living cells.

Theoretically, this problem can be overcome using approaches such as metatranscriptomics and metaproteomics. However, due to the short lifespan of RNA transcripts, and the fact that older permafrost samples have low biomass with slow metabolic rates, these strategies have only been successful in permafrost samples younger than approximately 2,000 years23 29 53. There have been few studies which have worked around these challenges and successfully characterized the active population of permafrost microbial communities using stable isotope probing54, extracellular enzyme assays55 and incubation experiments56.

Another challenge using DNA-based techniques is that they are not able to distinguish between vegetative cells and dormant forms such as endospores. Endospores are extremely hardy, non-reproductive structures formed by a few bacterial taxa (mostly from the phylum Firmicutes) in response to adverse environmental conditions (typically carbon, nitrogen, or phosphate starvation)57 58. Extreme environments with multiple biological stressors are known to select for endospore-forming taxa because in these environments the survival benefits of endospore formation outweigh its energetic costs59.

Endospores have unmatched survivability and longevity60 61 due to their protective spore coat, low water content, an inner membrane which is impermeable to potentially toxic

5 chemicals, and the presence of small acid-soluble proteins in the core which bind and condense DNA protecting it from UV light and DNA damaging chemicals57.

Mackelprang et al (2017) investigated a chronosequence of Late Pleistocene permafrost located in Alaska and observed an increase in relative abundance of Firmicutes, particularly of the classes Bacilli and Clostridia (both of which contain lineages of known endospore-formers) and an increase in the abundance of functional genes involved in endospore formation as sample age increased. Endospore-forming Firmicutes are commonly observed in high abundance in Canadian High Arctic permafrost samples26, suggesting that the ability to form endospores could be responsible for increasing long- term survivorship of these groups in older permafrost. However, Firmicutes in permafrost are not necessarily dormant. Hultman et al (2015) used RNA to DNA ratios to show that

Firmicutes were actively transcribing their 16S rRNA genes in Holocene permafrost29 demonstrating that it is not possible to predict whether these potential endospore forming organisms are dormant based solely on 16S rRNA gene amplicon sequencing.

While endospore formation may contribute to long-term survivorship of

Firmicutes in permafrost, it is unclear whether this strategy is optimal across geologic timescales. Despite resistance to extreme conditions and long-term survivorship, DNA within an endospore can still accumulate damage62. Typically, this would be repaired upon germination by DNA repair machinery57. However, damage accumulated over geologic timescales may be beyond the ability of repair enzymes to remedy. Johnson et al showed that endospore-forming taxa most likely accumulate DNA damage over geologic timescales (>100kyr) which could prevent their long-term ability to germinate.

They removed damaged DNA from permafrost samples of increasing age before

6 amplification of a 4kb 23S rRNA gene amplicon and found that the representation of endospore-forming organisms decreased in older samples. They also found non- endospore-forming Actinobacteria were capable of long-term viability, metabolic activity, and DNA repair in ancient permafrost samples as old as 500kyr63. Willerslev et al confirmed that Actinobacteria were more highly represented compared to endospore- forming organisms in samples of increasing age (>100kya)33. Therefore, it remains unclear whether endospore formation is an optimal survival strategy in permafrost for timescales dating back beyond the Late Pleistocene.

To better understand the role of dormancy in long-term survival, the extent to which cells remain viable in increasingly ancient permafrost, and the effect that DNA from dead cells has on microbial community structure inferences in permafrost, we isolated the active, dead, and dormant cell populations from a Late Pleistocene permafrost chronosequence. Specifically, we asked the following questions: (1) Does the number and proportion of live cells within permafrost change with increasing age? (2)

Are there significant differences in the live and dead communities over geologic time?

(3) To what extent are the endospore-forming taxa actually dormant? To address these questions, we coupled 16S rRNA gene amplicon sequencing with a trio of methods (two of which have never been applied to permafrost) to quantify and identify active, dead, and dormant organisms across a chronosequence of ancient permafrost—live/dead staining, depletion of DNA from dead cells, and endospore enrichment (Figure 1).

Figure 1. Experimental strategy— To obtain direct counts for live, dead, and total cells, we first separated cells from soil using a Nycodenz density centrifugation and stained using either a

Live/Dead differential stain or DAPI stain before counting cells via fluorescence microscopy. In addition, we separated biomass from soil using a gravity separation technique and treated with either a propidium monoazide treatment (to deplete DNA from dead organisms) or a lysozyme enzyme treatment (to enrich for endospores). We conducted 16S rRNA gene amplicon sequencing for each treatment plus an untreated control for all age categories to observe changes in alpha diversity, beta diversity, and taxa abundance in the active dead, and dormant populations.

8 2. Materials & Methods Permafrost Sample Collection

We collected samples from the United States Cold Regions Research and

Engineering Laboratory (CRREL) permafrost tunnel research facility located 16 km north of Fairbanks, Alaska (64.951oN, - 147.621oW) (Figure 2). The tunnel extends 110m horizontally at a depth of ~15m into a hillside exposing a chronosequence of late

Pleistocene permafrost (Figure 3 & Figure 4).64 65 The temperature of the tunnel is maintained by refrigeration at -4oC. In April 2016, we sampled from three locations inside the tunnel representing three age categories: 19kyr, 27kyr, and 33kyr as determined previously by radiocarbon dating34. After removing the sublimated surface layer (~5cm) from the walls of the tunnel, we collected five replicate cores per age category using an approximately 7.5 x 5 cm key hole saw attached to a power drill as described previously (Figure 5)34. Cores were shipped back to California State

University, Northridge (CSUN) on dry ice and stored at -20°C.

Figure 3. Photograph taken from the entrance to the CRREL Permafrost tunnel.

Figure 4. Schematic drawing of the CRREL Permafrost tunnel. The age of exposed permafrost increases with distance from the tunnel portal. (Figure from Mackelprang et al 2017).

Figure 5. Diagram showing the core sampling strategy used in the CRREL Permafrost Tunnel.

The cores were dug into the wall at three depths. Microbiological studies were conducted with the second and third depth cores which were interior to the wall and less likely to suffer from contamination or thaw. The outermost core was used for soil chemistry analysis. (Figure from

Mackelprang et al 2017).

11 Permafrost Subsampling

For subsampling, we placed cores on autoclaved foil at room temperature for 10 minutes to allow the outer layer to soften. Surface contamination was removed by slicing the outer layer with an autoclaved knife exposing the uncontaminated frozen interior. We sub-sectioned the remaining uncontaminated material using a fresh knife into sterile 50 mL falcon tubes and whirlpack bags in preparation for downstream treatment.

Soil Chemistry

Permafrost soil pH was determined using litmus paper with a range from 0.0 to

14.0. We suspended 2.5 g of permafrost sample in 4 mL of sterile distilled water. We dipped pH strips into the soil suspension and observed the color change compared to a pH chart. For dry weight measurements, pre-weighed aliquots of permafrost soils were put into a drying oven set at 80ºC. We weighed the samples each day until the weight remained constant for two consecutive days. The percent dry weight was calculated by dividing the final value by the initial weight. Percent ice content was recorded as the percent lost to evaporation.

Cell Separation for Enumeration via Microscopy

For cell enumeration, cells were separated from the permafrost soil matrix using

Nycodenz density cushion centrifugation, as described previously66 67 68 69 70. To remove cells from soil debris, we disrupted 1.5 g of soil in a mild detergent consisting of 2 mL of

0.05% Tween 80 and 50 mM tetrasodium pyrophosphate buffer (TTSP)67 71 and sonicated for 1 minute at 20V using a QSonica ultrasonicator with a 1/8 in probe.

Sonicated samples were centrifuged at 750 x g for 7 min at 4°C to remove large particles and debris. We extracted 600 uL of the supernatant and layered this over 600 uL of 1.3

12 g/L Nycodenz solution in a 2 mL tube. The tubes were centrifuged at 14,000 x g for 30 min at 4°C. We transferred 600 uL of the upper and middle phase containing bacterial cells into a sterile 2 mL tube and centrifuged at 10,000 x g for 15 min at 4°C. The supernatant was discarded and the pellet was resuspended in 1 mL of 0.85% NaCl solution.

Live/Dead Staining

The Live/Dead BacLight Bacterial Viability Kit (Invitrogen Detection

Technologies, Carlsbad, CA) was used to differentially stain live and dead cells. We added 3 uL of a 1:1 mixture of 3.34 mM SYTO 9 and 20 mM propidium iodide solution to 1 mL cell suspensions as per the manufacturer’s protocols. Stained suspensions were incubated at room temperature for 15 minutes in the dark to allow the dyes to permeate cells and bind to DNA.

DAPI Staining

DAPI staining was performed to obtain total cell counts. After removal of soil debris (as described above in the section on Cell Separation for Enumeration via

Microscopy), we added 3 uL of 14.3 mM DAPI stock solution to each 1 mL cell suspension. Stained suspensions were incubated at room temperature for 15 minutes in the dark to allow the dye to permeate cell membranes and bind to DNA.

Cell Enumeration

We diluted and vacuum filtered the stained suspensions onto a 25mm diameter

0.2 um pore size black polycarbonate membrane, which we placed on a slide with sterile forceps. Samples were observed at 100X magnification on a single focal plane using a

Zeiss Axio Imager M2 fluorescence microscope coupled to an Apotome 2.0 System with

13 appropriate filters for each stain. We counted fifteen fields of view for live/dead and

DAPI stained cells for each sample72 73. The average number of cells per field of view was multiplied by the area of the filter, the dilution factor, and corrected for dry weight to calculate the average number of cells per gram of dry weight.

Separation of Biomass from Soil Matrix

While Nycodenz density centrifugation is effective at removing soil debris, making it ideal for microscopic visualization, it is biased against endospores and heavily attached cells74. To retain these cells for DNA-based analyses, we opted for a less biased separation protocol based on Wunderlin et al 201675. We disrupted 5 g of sample in 25 mL of 1% sodium hexametaphosphate buffer (SHMP) and sonicated for 1 minute at 20V using a QSonica ultrasonicator with a 1/4 in probe. The samples were left for 15 minutes to allow large particles and debris to settle before transferring the supernatant to a clean

50 mL tube. We added 15 mL of 1% SHMP to the pellet and sonicated this mixture again for 1 minute at 20V using a QSonica ultrasonicator with a 1/4 in probe. The mixture incubated for another 15 minutes to allow debris to settle and then we combined the new supernatant with the supernatant from the previous step. To further remove large particles and debris, we centrifuged the combined supernatant at 20 x g for 2 minutes.

Following this centrifugation, we divided the supernatant equally into two 50 mL tubes as an experimental and a control group. These tubes were centrifuged at 10,000 x g for 15 minutes to pellet biomass. We discarded the supernatant and the biomass pellet was either stored at -20°C to await DNA extraction (in the case of the control pellets) or immediately used for downstream treatments.

14 Depletion of DNA from Dead Cells via Propidium Monoazide Treatment

To remove exogenous DNA and DNA from membrane compromised cells, we treated samples with propidium monoazide (PMAxx, Biotium Inc., Hayward, CA), which is a DNA-intercalating dye similar to the nucleic acid dye propidium iodide76. It is selectively permeable, passing through the impaired membranes of dead cells, but it is unable to penetrate the membranes of living cells. In the presence of intense bright light, the azide group enables propidium monoazide to covalently cross link double stranded

DNA, preventing its amplification via PCR77.

We resuspended the extracted biomass pellets in 500 uL of 0.85% NaCl solution and placed them in clear 1.5 mL microcentrifuge tubes. We added 2.5 uL of 20 mM propidium monoazide solution to each microcentrifuge tube resulting in a final concentration of 100 uM. We increased the concentration from the commonly used 50 uM due to the presence of leftover soil debris following cell extraction as recommended for environmental samples by Heise et al78 79. Tubes were incubated in the dark at room temperature for 10 minutes. After incubation, we placed the tubes on a sheet of foil in an ice bucket to prevent warming. A 500 W halogen work lamp was placed 20 cm above the samples for 15 minutes. Every five minutes, we mixed the samples gently to ensure even light distribution. Following light exposure, we centrifuged samples at 10,000 x g for 15 minutes and discarded the supernatant. We stored the pellets at -20°C until use in downstream DNA extractions.

Endospore Enrichment via Lysozyme Enzyme Treatment

To separate endospores from vegetative cells, we used a lysozyme enzyme treatment involving three steps: physical, enzymatic, and chemical cell lysis. The first

15 physical treatment uses heat to lyse vegetative cells. Second, lysozyme dissolves the cell membrane followed by a solution of sodium hydroxide (NaOH) and sodium dodecyl sulfate (SDS) to further disrupt cellular membranes. Finally, a DNAse treatment is used to degrade the DNA from ruptured cells.

We resuspended the extracted biomass pellets with 900 uL of 1X Tris – EDTA buffer (10 mM Tris and 1 mM EDTA; pH 8) and placed them into 2 mL tubes. Tubes were placed in a heat block at 60°C for 10 min with shaking at 80 rpm. After incubation, we let the tubes cool for 15 min to 37°C before adding 100 uL of lysozyme solution (20 mg/mL in 1X TE Buffer) and incubating in a heat block at 37°C for 60 min with shaking at 80 rpm. After lysis 250 uL of 3N sodium hydroxide and 100 uL of 10 % sodium dodecyl sulfate was added to the sample, reaching a final volume of 1.35 mL, which we incubated at room temperature for 60 min with shaking at 80 rpm. After the final incubation, we centrifuged the solution at 10,000 x g for 15 minutes to pellet cell debris and discarded the supernatant. We resuspended the pellet with 450 uL of sterile water,

50 uL of 1X DNAse reaction buffer and 1 uL DNAse enzyme and left it for 15 min to remove DNA from lysed vegetative cells. Following the DNase treatment, we centrifuged the tubes for 10,000 x g to pellet endospores and discarded the supernatant. The pellet was then resuspended in 1 mL 0.85% NaCl solution to wash any residual DNAse enzyme. We centrifuged the suspension one last time at 10,000 x g for 15 min, discarded the supernatant, and stored the pellet at -20°C in a sterile Eppendorf tube until it was used for downstream DNA extraction.

16 DNA Extraction

We performed DNA extractions using a modified bead-beating protocol capable of lysing endospores, cysts, and cells with thickened cell walls, all of which are known to exist in permafrost13 29 81 82. We resuspended pellets in 775 uL of lysis buffer (0.75M sucrose, 20 mM EDTA, 40 mM NaCl, 50 mM Tris) and transferred them to a MP Bio

Lysis Matrix E Tube. We added 100 uL of 20 mg/mL lysozyme and incubated the samples at 37°C for 30 minutes. Following incubation, 100 uL of 10% SDS was added and the samples were homogenized in an MP Biomedicals FastPrep 24 homogenizer for

20 seconds at 4.0 m/s. We placed the samples in a heat block at 99 °C for 2 minutes and allowed them to cool at room temperature for 5 minutes. We added 25 uL of 20 mg/mL

Proteinase K and incubated samples at 55 °C overnight. The next day we centrifuged the tubes at 10,000 x g for 15 minutes and transferred the supernatant to a clean 2 mL

Eppendorf tube. We used a FastDNA Spin Kit for Soil (MP Biomedicals, Santa Ana, CA) for DNA extractions from the lysed cells following the manufacturer’s protocols but omitting the initial lysis step. The DNA was cleaned using a PowerClean DNA Clean-

Up (Mo Bio, Carlsbad, CA) and quantified using a Qubit 2.0 fluorometer (Thermofisher

Scientific, Canoga Park, CA).

PCR Amplification and Sequencing

We amplified the variable region four (V4) of bacterial and archaeal 16S rRNA genes as described by Caporaso et al but with the addition of 2 uL of 20 mg/mL bovine serum albumin in each PCR reaction. We used the golay barcoded Illumina (San Diego,

CA, USA) primer set 515F/806R with an added degeneracy to enhance amplification of archaeal sequences on the 515F primer and thermal cycling steps recommended by the

17 Earth Microbiome Project protocol version 4.1383. We pooled the amplified PCR products from triplicate reactions for each sample at approximately equal concentrations, as measured using a PicoGreen dsDNA Quantification Assay Kit (Thermofisher

Scientific, Canoga Park, CA). We quantified the pooled 16S rRNA gene amplicons by qPCR using an Illumina Library Quantification Kit (Kapa Biosystems, Wilmington, MA) on a CFX96 Real Time PCR Detection System (Bio-Rad, Hercules, CA). 16S rRNA gene amplicons were sequenced with a 2 x 150 bp v2 Reagent Kit using an Illumina

MiSeq instrument located on campus at CSUN.

We demultiplexed and quality filtered raw fastq data using the open source software package Quantitative Insights Into Microbial Ecology (QIIME)84. Forward and reverse sequences were merged and all sequences that passed quality filtering were de novo clustered into operational taxonomic units (OTUs) at 97% sequence identity using

USEARCH85. We assigned taxonomy using the RDP classifier with a confidence score of 0.584. For phylogenetic metrics of diversity, a phylogenetic tree was constructed using

FastTree87 as implemented in QIIME. We rarefied samples to an equal depth (N = 5000 sequences/sample) for all subsequent analyses. One OTU was abundant (>3% relative abundance) only in the blank samples and samples which had the lowest DNA yields.

This OTU, from the genus Burkholderia, was removed from all other samples because it was likely a result of laboratory contamination88.

Statistical Analysis

We tested for significant differences in soil chemistry characteristics between age categories using an ANOVA and Tukey Honest Significant Difference post hoc test which adjusts p-values for multiple comparisons using the ‘aov’ function in R.

18 Significant differences in the ratio of live to dead cells, the proportion of live cells, and the direct cell counts for each stain between age categories was tested with a Kruskal

Wallis and a Dunn’s post hoc test for nonparametric data using the ‘PMCMR’ package in

R89. P-values were corrected using the False Discovery Rate (FDR).

We calculated alpha diversity using OTU richness, the Shannon Diversity Index,

Chao1 Index, and phylogenetic diversity90. Differences in alpha diversity between ages were tested using ANOVA and a Tukey Honest Significant Difference post hoc test using the ‘aov’ function in R. Differences in alpha diversity between treatments for each age category was tested with a t-test using the ‘t.test’ function in R. We calculated pairwise differences in community composition (beta diversity) using phylogenetic metrics

(UniFrac) via PerMANOVA in R using the ‘adonis’ function in the ‘vegan’ package8888.

Differences between samples were visualized using principal coordinate analysis (PCoA) using the ‘phyloseq’ package in R93. We also tested for significant correlations between soil chemistry characteristics and non-metric multidimensional scaling coordinates via

PerMANOVA using the ‘adonis’ function in the ‘vegan’ package in R.

Differences in the relative abundance of specific taxa between treatments and age categories were tested using a linear mixed effect model on rank transformed taxa abundances and nested treatment as a factor within age using the ‘nlme’ package in R94.

P-values were corrected using the FDR. The specific taxa indicated were tested for various pairwise comparisons between treatments and untreated controls for each age category using a Mann-Whitney-Wilcoxon test using the ‘wilcoxon.test’ function in R.

Differences in DNA concentrations between propidium monoazide treated samples and untreated controls were tested with a t-test using the ‘t.test’ function in R.

19 3. Results

Soil Chemistry

We measured ice content in permafrost samples and found that it ranged from 35

– 51%, though differences were not significant among age categories. pH decreased across the chronosequence from 7.4 to 7 though there were no significant differences between age categories (Table 1).

Table 1. Percent ice content and pH across the three time periods. There were no significant differences between age categories for ice content (ANOVA, F (2, 5.94) = 2.06, p > 0.05) or pH (ANOVA, F (2, 12) = 0.6, p > 0.05). Values are averages of five biological replicates.

Samples Ice Content (%) pH

19K 34.72 ± 4.32 7.4 ± 0.2

27K 51.32 ± 6.87 7.2 ± 0.4

33K 41.84 ± 1.30 7.0 ± 0.0

Cell Enumeration

We performed cell counts across the chronosequence using live/dead staining coupled with fluorescent microscopy. Average counts for live cells (stained with SYTO

9) ranged from 3.6 x 106 to 9.2 x 106 cells [gram dry weight (gdw)-1], for dead cells

(stained with propidium iodide) counts ranged from 1.7 x 107 to 4.5 x 107 cells gdw-1, and total counts (stained with DAPI) ranged from 2.3 x 107 to 4.7 x 107 cells gdw-1. Cell counts were highest in the intermediate aged samples for each category [(Live, Kruskal

20 Wallis, X2 (2) = 46.25, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p

< 0.01, Figure 6B) (Dead, Kruskal Wallis, X2 (2) = 53.16, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01, Figure 6C) and (Total, Kruskal Wallis, X2

(2) = 53.58, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01,

Figure 6D)]. We found the fewest cells in the oldest age category—counts were significantly lower compared to both the intermediate (Dunn’s post hoc test, p < 0.01,

Figure 6D) and youngest samples (Dunn’s post hoc test, p < 0.05, Figure 6D). The ratio of live cells to dead cells did not change significantly across the chronosequence (Kruskal

Wallis, X2 (2) = 1.38, p > 0.05, Supplementary Figure 18). However, the proportion of live cells increased significantly from 14% to 26% between the youngest and intermediate aged samples (Kruskal Wallis, X2(2) = 9.84, p < 0.01, Figure 7). In the oldest samples, 25% of the cells were live though this was not significantly different than the values observed for the youngest or intermediate aged samples.

Figure 6. Direct cell counts as determined by fluorescent microscopy (A) using SYTO 9 (B), propidium iodide (C), and DAPI (D). Total counts were highest in the intermediate aged category for the live counts (B, Kruskal Wallis, X2(2) = 46.25, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01), dead counts (C, Kruskal Wallis, X2(2) = 53.16, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01), and total counts gdw-1 (D, Kruskal

Wallis, X2(2) = 53.58, p < 0.001, Dunn’s post hoc test, 19K – 27K p < 0.01 27K – 33K p < 0.01).

The total counts gdw-1 were lowest for the oldest samples (D, Dunn’s post hoc test, p < 0.05). (**

= p < 0.01, * = p < 0.05). Values show averages of five replicates and error bars show standard error of the mean.

Dunn’s post hoc test, ** = p < 0.01). Values show averages of five replicates and error bars show standard error of the mean.

23 16S rRNA gene-based community analysis

Sequencing of the 16S rRNA gene from 19 kyr, 27 kyr, and 33 kyr old permafrost

(n = 5 replicate samples x 3 age categories x 4 treatments = 60 samples total) from the

CRREL permafrost tunnel yielded 4,833,270 sequences representing 3,962 unique OTUs.

We used four alpha diversity metrics to assess diversity: observed species, Shannon

Index, Chao1 Index, and PD Whole Tree. Overall, alpha diversity estimates were highest in the youngest category and lowest in the intermediate aged category (ANOVA, p < 0.01

Supplementary Figure 19A-D). For each permafrost sample, we performed endospore- enrichment and depletion of DNA from dead cells. DNA depletion did not change alpha- diversity estimates—there were no significant differences in samples depleted of DNA from dead cells compared to non-depleted controls (Student’s t-test, p > 0.05). Similar to non-depleted controls, alpha diversity in DNA depleted samples was highest in the youngest category, followed by the oldest category, and lowest in the intermediate category (ANOVA, p < 0.01, Figure 8A-D).

Unlike the results for samples depleted of DNA from dead cells, endospore enrichment reduced alpha-diversity estimates compared to non-enriched controls. In the intermediate aged category, this was significant for all diversity metrics (Student’s t-test,

Figure 9A-D). For the PD Whole Tree metric, this trend was also significant for the youngest age category (Student’s t-test, t (5.78) = -3.37, p < 0.05, Figure 9D) and the oldest age category (Student’s t-test, t (7.83) = -2.65, p < 0.05, Figure 9D).

Figure 8. Alpha diversity estimates based on 16S rRNA gene sequences for samples depleted of

DNA from dead cells across the three age groups were measured using QIIME. There were significant differences between age groups for observed species (A, ANOVA, F (2,6.99) = 246.8, p < 0.01), Shannon (B, ANOVA, F (2,7.72) = 239, p < 0.01), Chao1 (C, ANOVA, F (2,6.65) =

247.98, p < 0.01), and PD Whole Tree (D, ANOVA, F (2,7.46) = 363.4, p < 0.01). The youngest samples were most diverse (Tukey HSD post hoc test, ** = p < 0.01), followed by the oldest samples (Tukey HSD post hoc test, ** = p < 0.01), and the intermediate aged samples were least diverse (Tukey HSD post hoc test, ** = p < 0.01). Values show averages of five replicates and error bars show standard error of the mean.

25 Figure 9. Alpha diversity estimates based on 16S rRNA gene sequences for the endospore- enriched samples and non-enriched controls across the three age groups were measured using

QIIME. The endospore-enriched samples were less diverse than the non-enriched controls from the intermediate aged samples for observed species (A, Student’s t-test, t (7.99) = -4.83, ** = p <

0.01), Shannon, (B, Student’s t-test, t (7.82) = -3.13, * = p < 0.05), Chao1 (C, Student’s t-test, t

(7.97) = -4.47, ** = p < 0.01), and PD Whole Tree (D, Student’s t-test, t (7.35) = -4.67, ** = p <

0.01). For PD Whole Tree, the endospore-enriched samples were also less diverse than the non- enriched controls from and the youngest aged samples (D, Student’s t-test, t (5.78) = -3.37, * = p

< 0.05) and the oldest aged samples (D, Student’s t-test, t (7.83) = -2.65, * = p < 0.05). Values show averages of five replicates and error bars show standard error of the mean.

26 PerMANOVA of weighted Unifrac values showed significant differences based on age (R2 = 0.83, F = 135.64, p < 0.001, Figure 10A) and ice content (R2 = 0.21, F =

15.59, p < 0.001, Figure 11). pH did not differ significantly among samples

(PerMANOVA, p > 0.05). Age (R2 = 0.89, p < 0.001), ice content (R2 = 0.33, p < 0.001), and DOC (R2 = 0.86, p < 0.001, DOC data taken from Mackelprang et al 2017) significantly correlated with NMDS coordinates (Figure 12). While age was still the primary driver of community composition, communities of the endospore enriched samples were distinct (PerMANOVA, p < 0.01, Figure 10 B-D). In addition,

PerMANOVA of unweighted Unifrac values also showed significant differences based on age (R2 = 0.38, F = 17.26, p < 0.001, Supplementary Figure 19A) and treatment

(Supplementary Figure 20B-D).

Figure 10. Community composition of samples differ based on age and treatment. Principal

Coordinates Analysis (PCoA) based on weighted Unifrac values of 16S rRNA gene sequences.

Samples differed significantly by age (A, PerMANOVA, R2 = 0.83, F = 135.64, p < 0.001) and treatment for the 19K aged category (B, PerMANOVA, R2 = 0.51, F = 5.65, p < 0.001), the 27K aged category (C, PerMANOVA, R2 = 0.42, F = 3.87, p < 0.01), and the 33K aged category (D,

PerMANOVA, R2 = 0.37, F = 3.14, p < 0.01).

Figure 11. Community composition of samples differ based on ice content. Principal

Coordinates Analysis (PCoA) based on weighted Unifrac values of 16S rRNA gene sequences.

Samples differed significantly by ice content (PerMANOVA, R2 = 0.21, F = 15.59, p < 0.001).

Figure 12. Non-metric multidimensional (NMDS) scaling based on weighted Unifrac values of

16S rRNA gene sequences. Age (PerMANOVA, R2 = 0.89, p < 0.001), ice content (PerMANOVA,

R2 = 0.33, p < 0.001), and DOC (PerMANOVA, R2 = 0.86, p < 0.001, DOC data from

Mackelprang et al 2017) significantly correlate to NMDS coordinates. pH did not significantly correlate to NMDS coordinates (p > 0.05).

30 Phyla that differed significantly based on age and treatment (endospore enrichment and depletion of DNA from dead cells) using a linear mixed effect model included Proteobacteria, Actinobacteria, Firmicutes, Bacteroidetes, Chloroflexi,

Acidobacteria, Planctomycetes, and Chlamydiae (FDR corrected p < 0.01, Figure 13).

Chloroflexi, Acidobacteria, Planctomycetes as well as Chlamydiae (Linear Mixed Effect Model, p

< 0.01).

31 Depletion of DNA from Dead Cells via Propidium Monoazide Treatment

We used a propidium monoazide treatment to remove DNA from membrane compromised cells and extracellular DNA. Depletion of DNA from dead cells did not change the relative abundance of taxa for any of the age categories (Mann-Whitney-

Wilcoxon test, p > 0.05). However, depletion decreased DNA yield by ~50% (Student’s t-test, t (26.79) = 3.11, p < 0.01, Table 2). Although there were no significant differences in taxa abundance, we observed several trends: Acidobacteria, Alphaproteobacteria,

Chloroflexi, and Chlamydiae were consistently more abundant in samples depleted of

DNA from dead cells compared to non-depleted controls. Actinobacteria were consistently less abundant in depleted samples compared to the non-depleted controls.

Table 2. The average DNA concentration (ng/uL) decreased by ~50% (Students t-test, t (26.79)

= 3.11, ** = p < 0.01) in the propidium monoazide treated group (DNA depleted samples) compared to the untreated group (non-depleted controls). Significance shows pairwise comparison between DNA concentration in DNA depleted samples and non-depleted controls.

Samples Non-Depleted DNA Depleted t -value p-value Controls (ng/uL) Samples (ng/uL) TOTAL 1.84 ± 0.24 ** 0.90 ± 0.19 ** t (26.79) = 3.11 p < 0.01

19K 2.44 ± 0.11 1.53 ± 0.45 t (4.18) = 1.96 p > 0.05

27K 1.90 ± 0.60 0.70 ± 0.09 t (4.18) = 1.96 p > 0.05

33K 1.19 ± 0.10 ** 0.47 ± 0.10 ** t (7.99) = 4.96 p < 0.01

32 Endospore Enrichment via Lysozyme Enzyme Treatment

We used a lysozyme enzyme treatment to lyse vegetative cells and enrich for endospores. This treatment increased the relative abundance of three phyla—Firmicutes,

Actinobacteria, and Chlamydiae. As might be expected, this trend was the strongest among Firmicutes (specifically from endospore-forming classes Bacilli and Clostridia).

Overall, endospore enrichment increased the relative abundance of Firmicutes in all age categories. This trend was significant for the youngest (Mann-Whitney-Wilcoxon test, U

= 0, p < 0.01, Figure 14) and intermediate (Mann-Whitney-Wilcoxon test, U = 2, p <

0.05, Figure 14) age categories. However, Bacilli and Clostridia responded differently to enrichment. Endospore enrichment increased the relative abundance of Bacilli across all age categories and grew more pronounced in the older samples (youngest: 5.8%, intermediate: 9.3%, oldest: 18.6%). These data were significant for each age category

(Mann-Whitney-Wilcoxon test, U = 2, p < 0.05, Figure 15). In contrast, endospore enrichment decreased the relative abundance of Clostridia. The youngest aged category exhibited a slight increase of ~1.7% in endospore enriched samples compared to controls, while the oldest samples had an almost 10% decrease in the endospore enriched samples compared to controls (Mann-Whitney-Wilcoxon test, U = 2, p < 0.05, Figure 15). These trends were driven by the families Planococcaceae, Thermoactinomycetaceae,

Bacillaceae, and Paenibacillaceae for the Bacilli and the family Clostridiaceae for the

Clostridia (Table 3).

Endospore enrichment increased the relative abundance of Actinobacteria in the youngest age category from 22.4% to 32.8% (Mann-Whitney-Wilcoxon test, U = 0, p <

33 0.01, Figure 16). This increase was driven by the families Micrococcaceae within the

Actinomycetales, as well as the Gaiellaceae and Solirubrobacteraceae (Table 3).

Chlamydiae relative abundance increased in the youngest age category from 0.7% to 2.0% as a result of endospore enrichment (Mann-Whitney-Wilcoxon test, U = 0, p <

0.01, Figure 17). This trend was driven by an increase in the relative abundance of the family Chlamydiaceae. All other taxa including Proteobacteria, Alphaproteobacteria,

Deltaproteobacteria, Bacteroidetes, Acidobacteria, Chloroflexi, and Planctomycetes decreased in relative abundance due to the endospore enrichment (Supplementary Table

4).

Figure 14. Firmicutes relative abundance was significantly greater in the endospore-enriched samples compared to the non-enriched controls for the youngest (Mann-Whitney-Wilcoxon test, U

= 0, ** = p < 0.01) and intermediate aged category (Mann-Whitney-Wilcoxon test, U = 2, * = p

< 0.05). These differences were driven by the classes Bacilli and Clostridia. Values show averages of five replicates and error bars show standard error of the mean.

Figure 15. Bacilli were significantly more highly represented in the endospore-enriched samples compared to the non-enriched controls for the youngest (Mann-Whitney-Wilcoxon test, U = 0, **

= p < 0.01), intermediate (Mann-Whitney-Wilcoxon test, U = 2, * = p < 0.05) and oldest samples

(Mann-Whitney-Wilcoxon test, U = 1, * = p < 0.05) while the Clostridia were underrepresented

(Mann-Whitney-Wilcoxon test, U = 2, * = p < 0.05). Change in relative abundance of Bacilli and Clostridia in endospore-enriched samples compared to the non-enriched controls across the three age categories. A negative value implies underrepresentation in the endospore-enriched samples compared to the non-enriched control while a positive value implies overrepresentation.

Values show averages of five replicates and error bars show standard error of the mean.

Figure 16. Actinobacteria relative abundance was significantly greater in the endospore- enriched samples compared to the non-enriched controls from the youngest aged category

(Mann-Whitney-Wilcoxon test, U = 0, ** = p < 0.01). This was driven by the family

Micrococcaceae, within the Actinomycetales, as well as the families Gaiellaceae and

Solirubrobacteraceae. Values show averages of five replicates and error bars show standard error of the mean.

Figure 17. Chlamydiae relative abundance was significantly greater in the endospore-enriched samples compared to the non-enriched controls from the youngest aged category (Mann-

Whitney-Wilcoxon test, U = 0, ** = p < 0.01). This trend was driven by an increase in the relative abundance of the family Chlamydiaceae. Values show averages of five replicates and error bars show standard error of the mean.

37 Table 3. Average percent difference in relative abundance between the endospore-enriched samples and non-enriched controls across the three age categories (Mann-Whitney-Wilcoxon test, ** = p < 0.01, * = p < 0.05). A negative value implies underrepresentation in the endospore- enriched samples compared to the non-enriched controls while a positive value implies overrepresentation. Values show averages of five replicates.

Taxa Family 19K (%) U-value 27K (%) U-value 33K (%) U-value Actinobacteria Micrococcaceae 2.4 ± 0.5 ** 0 0.0 ± 0.0 9.5 0.0 ± 0.0 11 Solirubrobacteraceae 1.5 ± 0.4 3 0.6 ± 0.3 4 0.1 ± 0.1 5 Gaiellaceae 1.0 ± 0.3 * 1 -0.2 ± 0.1 8 0.1 ± 0.1 8.5 Bacilli Planococcaceae 2.4 ± 0.9 4 -8.5 ± 1.0 * 1 4.0 ± 2.6 8 Thermoactinomycetaceae 0.4 ± 0.1* 0 20 ± 2.5 * 2 7.8 ± 2.2 6 Bacillaceae 0.7 ± 0.1 * 0 -2.5 ± 1.2 8 1.0 ± 0.4 9 Paenibacillaceae 2.5 ± 0.4 * 0 0.4 ± 1.6 12 5.6 ± 2.7 4 Clostridia Clostridiaceae 3.1 ± 0.4 ** 0 -2.5 ± 1.5 5 -9.2 ± 1.2 ** 0

38 4. Discussion/Conclusion

In this study, we present strong evidence that potential endospore forming organisms observed in DNA-based studies are not always dormant. We found that the tendency to undergo sporulation among endospore-forming organisms was taxa dependent. When faced with the stressors of our Late Pleistocene chronosequence,

Bacilli were more likely to form endospores compared to members of Clostridia, which were more likely to persist as vegetative cells over geologic timescales. Additionally, we found that exogenous DNA from dead organisms preserved within permafrost does not bias DNA sequencing results, since its removal did not significantly alter microbial community composition. Lastly, our results confirm that age and ice content act as primary drivers of microbial abundance and community composition in permafrost soils.

We found that Bacilli were more likely to exist as endospores in increasingly ancient permafrost, while members of Clostridia were more likely to remain vegetative.

Previous studies have found that high levels of endospore forming microorganisms, including Firmicutes, are common in Canadian High Arctic permafrost26 and that these organisms increase in relative abundance over time34 80 suggesting dormancy is a possible survival strategy. In 2014, Tuorto et al used stable isotope probing and found that

Firmicutes were not observed in the permafrost soil active community implying that they are likely dormant4. However, new studies suggest that endospore-forming taxa are not necessarily dormant. Hultman et al (2015) used RNA to DNA ratios to show that

Firmicutes in permafrost are more active than what would typically be inferred based

DNA-sequencing alone29. Our data corroborate this trend and suggest that, among

Firmicutes, the tendency to form endospores is taxa specific. While endospore

39 enrichment increased the relative abundance of Bacilli, it decreased the relative abundance of Clostridia. This trend grew more pronounced as sample age increased. For

Bacilli, this trend was driven by the families Planococcaceae, Thermoactinomycetaceae,

Bacillaceae, and Paenibacillaceae which are predominantly aerobic organisms, most of which are also endospore formers31. Three of the four genera identified from the family

Thermoactinomycetaceae are thermophilic. This family increased by 20% in the endospore-enriched samples from the intermediate aged category. For Clostridia, this trend was driven by the family Clostridiaceae which decreased by over 9% in the endospore-enriched samples compared to non-enriched controls from the oldest aged category. Clostridiaceae are Gram positive, motile, obligately anaerobic, endospore- forming rods with metabolic strategies ranging from fermentation to chemoorganoheterotrophy to chemolithoautotrophy31. The limited oxygen and constant subzero temperatures of permafrost could be stressors which contribute to endospore formation for groups of Bacilli, while Clostridia—which are anaerobes—do not respond in this way. It also begs the question what traits do members of the family Clostridiaceae possess which make them suited to survive as vegetative cells in permafrost over geologic timescales? It could be that being able to move towards a food source, being able to deal with limited or no oxygen, and having a varied way to access energy are all advantageous adaptations which allow them to remain vegetative in the extreme conditions of Late Pleistocene permafrost. Therefore, it seems likely that endospore formation could be a viable survival strategy for Bacilli against the conditions of ancient permafrost, at least for the timescales observed in this study. However, over increased timescales endospores can accumulate DNA damage that is beyond the ability of repair

40 enzymes to fix 33 57 63. If we were to increase our investigation to permafrost samples beyond the Late Pleistocene, it is likely that there would be a certain sample age at which

DNA damage limits survival and the relative abundance of dormant Bacilli begins to decline. Alternatively, as sample age increases, the relative abundance of vegetative

Clostridia might continue to increase suggesting they are suited to survival in permafrost conditions for geologic timescales dating beyond the Late Pleistocene.

Though our endospore treatment was designed to enrich for true endospores, it also enriched for two other phyla—Actinobacteria and Chlamydiae—but only in the youngest age category. Actinobacteria were previously found to resist this treatment perhaps due to their ability to form dormant forms and spore-like structures95. In our samples, members of families Micrococcaceae, Solirubrobacteraceae, and Gaiellaceae were overrepresented in our endospore-enriched group compared to non-enriched controls. While none of these taxa are known to form spores, they are all known to be able to survive radiation, starvation, and extreme dessication31 63, which may also confer the ability to resist lysozyme treatment. Enrichment of non-endospores may also be due to the smaller cell size and thicker membranes that are common among permafrost microorganisms10 82.

Unexpectedly, Chlamydiae were also enriched for in the endospore-enriched samples compared to non-enriched controls from the youngest aged category. This was driven exclusively by the family Chlamydiaceae, which are not known to have resting states31. All genera within this class are obligate intracellular symbionts of

Acanthamoeba. Acanthamoeba are a genus of single-celled eukaryotes commonly found in freshwater and soil. They exist in a free-living form and as a stress-resistant dormant

41 cyst,96 which could account for the increase in the relative abundance of Chlamydiae in the endospore-enriched samples. There are several studies which have found intact and viable Acanthamoeba cysts in permafrost soils ranging in age from the Holocene to the

Pleistocene97 98. Though we did not sequence eukaryotic marker genes, it is possible that there are Acanthamoeba in our samples.

To determine whether the DNA from dead organisms significantly biases DNA sequencing results, we depleted DNA from membrane-compromised cells and exogenous

DNA from our samples. This treatment did not alter microbial community composition suggesting that exogenous DNA and DNA from membrane compromised cells does not bias DNA sequencing results. The is supported by the observation that propidium monoazide treatment reduced DNA concentrations by ~50%, consistent with the previous estimate that up to 40% of environmental DNA is made up of ‘relic DNA’ from dead cells99. Lennon et al (2017) used a DNase treatment to investigate the effects of ‘relic

DNA’ from dead organisms on estimates of microbial diversity in several environmental samples100. They used their data to validate a simulation modelling how and when DNA from dead organisms will bias sequencing results and found that if rates of death and rates of degradation are proportional among taxa then the relic DNA pool should not significantly bias DNA-based studies. Although Carini et al (2016) found significant bias in 16S rRNA gene sequences from environmental samples as a result of propidium monoazide treatment99, the limited number of changes in taxa abundance observed in our permafrost samples is consistent with other several other studies that used this approach on complex environments77 101 suggesting that rates of turnover are proportional between taxa and the relic DNA pool does not bias estimates of microbial diversity in permafrost.

42 One potential limitation using the propidium monoazide approach on environmental samples is that soil particles can prevent light penetration and limit efficacy. This is unlikely to be a concern for our samples given intermittent mixing during treatment as well as a biomass extraction step and an increased PMA concentration suggested for particle-rich environmental samples78. Another potential complication using this approach is that successive rounds of PCR amplification can reduce the degree of differences in abundance between starting templates over increased cycle numbers102 which would make slight differences difficult to observe using relative abundance data alone.

Similar to prior observations using 16S rRNA gene amplicon sequencing33 34 and direct cell counts via fluorescence microscopy5 32, our cell count data and comparisons of alpha and beta diversity demonstrate that age and ice content drive microbial community composition and abundance in permafrost. Among our three age categories, the intermediate had the highest number of live, dead, and total cells. However, the oldest samples had the lowest number of total cells. Direct counts ranged from 2.3 x 107 to 4.7 x

107 cells gdw-1 and were consistent with previous studies from Arctic permafrost using similar techniques2 5 26 103 104. We also observed that the proportion of live cells increased from ~15% to ~25% in our intermediate aged category compared to the youngest age category. Alpha diversity decreased in an age dependent manner between the youngest and oldest samples, but was lowest in our intermediate aged category. The decrease in alpha diversity observed in our intermediate aged samples likely reflect the high ice content in those cores, which has been shown previously to decrease diversity in permafrost32. The age dependent decrease in alpha diversity and cell counts between our

43 youngest and oldest samples indicate that the harsh conditions of the permafrost environment are selective against certain organisms over geologic timescales. Therefore, the observed increase in the proportion of live cells with increased sample age could indicate a community of organisms which are adapted to the permafrost environment.

Our findings are consistent with a previous publication studying metagenomes from the

Fox Tunnel chronosequence, though direct bacterial cell counts increased by an order of magnitude. This is likely due to increased cell recovery as a result of our biomass extraction protocol, which used sonication (rather than vortexing) to remove cells from soil particles34.

The Live/Dead staining approach uses membrane permeability as a proxy for viability105 and has been used extensively in environmental samples106 107 108 including permafrost34 56. One potential drawback using this approach, shown by Kirchhoff et al

(2017), is that live cells can incorrectly stain dead under growth conditions which increase membrane potential109. The study demonstrated that growing cells in anoxic and dark conditions for two hours resulted in an increased membrane potential causing live cells to stain dead, though this effect was negated by returning the cells to oxic conditions. We suggest that this is unlikely to impact our samples. Permafrost is a stable environment in which membrane potentials are well maintained and we stained under aerobic conditions in the light. However, if our treatment affected membrane potentials, this would result in an underestimate of the number of live cells that are able to persist for geologic timescales in the extreme conditions of permafrost.

This study builds on previous work aimed at understanding how microbial communities adapt to the extreme conditions of permafrost over geologic timescales. We

44 used two strategies never applied to permafrost (endospore-enrichment and depletion of

DNA from dead organisms) to show that Firmicutes represented in DNA based studies are not always dormant. We found that preserved exogenous DNA from dead organisms does not bias DNA sequencing results, since its removal using propidium monoazide did not significantly alter microbial community composition. We confirmed that both microbial cell counts and diversity decreased between our youngest and oldest samples and were primarily controlled by sample age and ice content. In addition, we observed an increase in the proportion of live cells in our older samples suggesting that the permafrost environment is selective for certain taxa which are adapted to survive. The methods used in this study could be applied to samples from many environments to identify active, dead, and dormant community members. The survival strategies which were identified in this study may be common across different types of permafrost, though expanding investigations to older permafrost samples and permafrost with different biogeochemical properties will be essential to building a model for how microorganisms in permafrost survive for geologic timescales. In addition, these data may yield clues to how life (if it exists) survives on other cryogenic extraterrestrial bodies in our solar system and beyond.

45 Literature Cited

1. D’Amico, S. et al. Psychrophilic microorganisms: challenges for life. EMBO Rep. 7, 385–389 (2006).

2. Jansson, J. K. & Taş, N. The microbial ecology of permafrost. Nat. Rev. Microbiol. 12, 414–425 (2014).

3. Gilichinsky, D. Permafrost. in Encyclopedia of Environmental Microbiology (John Wiley & Sons, Inc., 2003).

4. Tuorto, S. J. et al. Bacterial genome replication at subzero temperatures in permafrost. ISME J. 8, 139–149 (2014).

5. Steven, B., Léveillé, R., Pollard, W. H. & Whyte, L. G. Microbial ecology and biodiversity in permafrost. Extremophiles 10, 259–267 (2006).

6. Mykytczuk, N. C. S. et al. Bacterial growth at −15 °C; molecular insights from the permafrost bacterium Planococcus halocryophilus Or1. ISME J. 7, 1211 (2013).

7. Bakermans, C., Tsapin, A. I., Souza-Egipsy, V., Gilichinsky, D. A. & Nealson, K. H. Reproduction and metabolism at − 10°C of bacteria isolated from Siberian permafrost. Environ. Microbiol. 5, 321–326 (2003).

8. Rivkina, E. M., Friedmann, E. I., McKay, C. P. & Gilichinsky, D. A. Metabolic Activity of Permafrost Bacteria below the Freezing Point. Appl. Environ. Microbiol. 66, 3230–3233 (2000).

9. Friedmann, E. I. Permafrost as microbial habitat. Viable Microorg. Permafr. 21–26 (1994).

10. Soina, V. S., Vorobiova, E. A., Zvyagintsev, D. G. & Gilichinsky, D. A. Preservation of cell structures in permafrost: A model for exobiology. Adv. Space Res. 15, 237– 242 (1995).

11. Amato, P., Doyle, S. M., Battista, J. R. & Christner, B. C. Implications of subzero metabolic activity on long-term microbial survival in terrestrial and extraterrestrial permafrost. Astrobiology 10, 789–798 (2010).

12. Faul, H. GEOLOGIC TIME SCALE. GSA Bull. 71, 637–644 (1960).

13. Gilichinsky, D. a. et al. Microbial Populations in Antarctic Permafrost: Biodiversity, State, Age, and Implication for Astrobiology. Astrobiology 7, 275–311 (2007).

14. Anderson, D. M., Ugolini, F. C. & Gatto, L. W. Antarctic Analog of Martian Permafrost Terrain. Antarct. J. U. S. 7, 114 (1972).

46 15. Rivkina, E. et al. Microbial life in permafrost. Adv. Space Res. 33, 1215–1221 (2004).

16. Gilichinsky, D., Rivkina, E., Shcherbakova, V., Laurinavichuis, K. & Tiedje, J. Supercooled Water Brines Within Permafrost—An Unknown Ecological Niche for Microorganisms: A Model for Astrobiology. Astrobiology 3, 331–341 (2003).

17. Schuur, E. a. G. et al. Climate change and the permafrost carbon feedback. Nature 520, 171–179 (2015).

18. Tarnocai, C. et al. Soil organic carbon pools in the northern circumpolar permafrost region. Glob. Biogeochem. Cycles 23, GB2023 (2009).

19. Tarnocai, C. et al. Soil organic carbon pools in the northern circumpolar permafrost region. Glob. Biogeochem. Cycles 23, GB2023 (2009).

20. Mackelprang, R. et al. Metagenomic analysis of a permafrost microbial community reveals a rapid response to thaw. Nature 480, 368–371 (2011).

21. Schädel, C. et al. Circumpolar assessment of permafrost C quality and its vulnerability over time using long-term incubation data. Glob. Change Biol. 20, 641– 652 (2014).

22. Treat, C. C. et al. A pan-Arctic synthesis of CH4 and CO2 production from anoxic soil incubations. Glob. Change Biol. 21, 2787–2803 (2015).

23. Mackelprang, R., Saleska, S. R., Jacobsen, C. S., Jansson, J. K. & Taş, N. Permafrost Meta-Omics and Climate Change. Annu. Rev. Earth Planet. Sci. 44, null (2016).

24. Psychrophiles: From Biodiversity to Biotechnology | Rosa Margesin | Springer.

25. Steven, B., Pollard, W. H., Greer, C. W. & Whyte, L. G. Microbial diversity and activity through a permafrost/ground ice core profile from the Canadian high Arctic. Environ. Microbiol. 10, 3388–3403 (2008).

26. Steven, B. et al. Characterization of the microbial diversity in a permafrost sample from the Canadian high Arctic using culture-dependent and culture-independent methods. FEMS Microbiol. Ecol. 59, 513–523 (2007).

27. Taş, N. et al. Impact of fire on active layer and permafrost microbial communities and metagenomes in an upland Alaskan boreal forest. ISME J. 8, 1904–1919 (2014).

28. Schostag, M. et al. Distinct summer and winter bacterial communities in the active layer of Svalbard permafrost revealed by DNA- and RNA-based analyses. Front. Microbiol. 6, (2015).

29. Hultman, J. et al. Multi-omics of permafrost, active layer and thermokarst bog soil microbiomes. Nature 521, 208–212 (2015).

47 30. Nikrad, M. P., Kerkhof, L. J. & Häggblom, M. M. The subzero microbiome: microbial activity in frozen and thawing soils. FEMS Microbiol. Ecol. 92, (2016).

31. BERGEY’S MANUAL OF DETERMINATIVE BACTERIOLOGY (7th ed.). Am. J. Public Health Nations Health (1964). doi:10.2105/AJPH.54.3.544-c

32. Gilichinsky, D. A. Permafrost Model of Extraterrestrial Habitat. in Astrobiology 125– 142 (Springer, Berlin, Heidelberg, 2002).

33. Willerslev, E. et al. Long-term persistence of bacterial DNA. Curr. Biol. 14, R9–R10 (2004).

34. Mackelprang, R. et al. Microbial survival strategies in ancient permafrost: insights from metagenomics. ISME J. (2017). doi:10.1038/ismej.2017.93

35. Gilichinsky, D. A. et al. Long-term preservation of microbial ecosystems in permafrost. Adv. Space Res. Off. J. Comm. Space Res. COSPAR 12, 255–263 (1992).

36. Gilichinskiy, D. A., Khlebnikova, G. M., Zvyagintsev, D. G., Fedorov-Davydov, D. G. & Kudryavtseva, N. N. Microbiology of sedimentary materials in the permafrost zone. Int. Geol. Rev. 31, 847–858 (1989).

37. Kao-Kniffin, J. et al. Archaeal and bacterial communities across a chronosequence of drained lake basins in arctic alaska. Sci. Rep. 5, (2015).

38. Lauber, C. L., Hamady, M., Knight, R. & Fierer, N. Pyrosequencing-Based Assessment of Soil pH as a Predictor of Soil Bacterial Community Structure at the Continental Scale. Appl. Environ. Microbiol. 75, 5111–5120 (2009).

39. Chu, H. et al. Soil bacterial diversity in the Arctic is not fundamentally different from that found in other biomes. Environ. Microbiol. 12, 2998–3006 (2010).

40. Wagner, R., Zona, D., Oechel, W. & Lipson, D. Microbial community structure and soil pH correspond to methane production in Arctic Alaska soils. Environ. Microbiol. 19, 3398–3410 (2017).

41. Zhang, T., Wang, N.-F., Liu, H.-Y., Zhang, Y.-Q. & Yu, L.-Y. Soil pH is a Key Determinant of Soil Fungal Community Composition in the Ny-Ålesund Region, Svalbard (High Arctic). Front. Microbiol. 7, (2016).

42. Durack, J., Ross, T. & Bowman, J. P. Characterisation of the Transcriptomes of Genetically Diverse Listeria monocytogenes Exposed to Hyperosmotic and Low Temperature Conditions Reveal Global Stress-Adaptation Mechanisms. PLOS ONE 8, e73603 (2013).

43. Gao, H., Yang, Z. K., Wu, L., Thompson, D. K. & Zhou, J. Global Transcriptome Analysis of the Cold Shock Response of Shewanella oneidensis MR-1 and

48 Mutational Analysis of Its Classical Cold Shock Proteins. J. Bacteriol. 188, 4560– 4569 (2006).

44. Bakermans, C. et al. Proteomic analysis of Psychrobacter cryohalolentis K5 during growth at subzero temperatures. Extremophiles 11, 343–354 (2007).

45. Ayala-del-Río, H. L. et al. The Genome Sequence of Psychrobacter arcticus 273-4, a Psychroactive Siberian Permafrost Bacterium, Reveals Mechanisms for Adaptation to Low-Temperature Growth. Appl. Environ. Microbiol. 76, 2304–2312 (2010).

46. Celik, Y. et al. Microfluidic experiments reveal that antifreeze proteins bound to ice crystals suffice to prevent their growth. Proc. Natl. Acad. Sci. 110, 1309–1314 (2013).

47. Kandror, O., DeLeon, A. & Goldberg, A. L. Trehalose synthesis is induced upon exposure of Escherichia coli to cold and is essential for viability at low temperatures. Proc. Natl. Acad. Sci. 99, 9727–9732 (2002).

48. Phadtare, S. & Inouye, M. Genome-Wide Transcriptional Analysis of the Cold Shock Response in Wild-Type and Cold-Sensitive, Quadruple-csp-Deletion Strains of Escherichia coli. J. Bacteriol. 186, 7007–7014 (2004).

49. Casanueva, A., Tuffin, M., Cary, C. & Cowan, D. A. Molecular adaptations to psychrophily: the impact of ‘omic’ technologies. Trends Microbiol. 18, 374–381 (2010).

50. Chattopadhyay, M. K. Cold-adaptation of Antarctic microorganisms – possible involvement of viable but nonculturable state. Polar Biol. 23, 223–224 (2000).

51. Vishnivetskaya, T. A. et al. The resistance of viable permafrost algae to simulated environmental stresses: implications for astrobiology. Int. J. Astrobiol. 2, 171–177 (2003).

52. Margesin, R. Permafrost Soils. (Springer Science & Business Media, 2008).

53. Coolen, M. J. L. & Orsi, W. D. The transcriptional response of microbial communities in thawing Alaskan permafrost soils. Front. Microbiol. 6, (2015).

54. Martineau, C., Whyte, L. G. & Greer, C. W. Stable Isotope Probing Analysis of the Diversity and Activity of Methanotrophic Bacteria in Soils from the Canadian High Arctic. Appl. Environ. Microbiol. 76, 5773–5784 (2010).

55. Gittel, A. et al. Site- and horizon-specific patterns of microbial community structure and enzyme activities in permafrost-affected soils of Greenland. Front. Microbiol. 5, (2014).

49 56. Hansen, A. A. et al. Viability, diversity and composition of the bacterial community in a high Arctic permafrost soil from Spitsbergen, Northern Norway. Environ. Microbiol. 9, 2870–2884 (2007).

57. Nicholson, W. L., Munakata, N., Horneck, G., Melosh, H. J. & Setlow, P. Resistance of Bacillus Endospores to Extreme Terrestrial and Extraterrestrial Environments. Microbiol. Mol. Biol. Rev. 64, 548–572 (2000).

58. Foster, S. J. & Johnstone, K. Pulling the trigger: the mechanism of bacterial spore germination. Mol. Microbiol. 4, 137–141 (1990).

59. Filippidou, S. et al. A Combination of Extreme Environmental Conditions Favor the Prevalence of Endospore-Forming Firmicutes. Front. Microbiol. 7, (2016).

60. Cano, R. J. & Borucki, M. K. Revival and identification of bacterial spores in 25- to 40-million-year-old Dominican amber. Science 268, 1060–1064 (1995).

61. Kennedy, M. J., Reader, S. L. & Swierczynski, L. M. Preservation records of microorganisms: evidence of the tenacity of life. Microbiol. Read. Engl. 140 ( Pt 10), 2513–2529 (1994).

62. Setlow, P. Mechanisms for the prevention of damage to DNA in spores of Bacillus species. Annu. Rev. Microbiol. 49, 29–54 (1995).

63. Johnson, S. S. et al. Ancient bacteria show evidence of DNA repair. Proc. Natl. Acad. Sci. 104, 14401–14405 (2007).

64. Hamilton, T. D., Craig, J. L. & Sellmann, P. V. The Fox permafrost tunnel: A late Quaternary geologic record in central Alaska. GSA Bull. 100, 948–969 (1988).

65. Bjella, K., Tantillo, T., Weale, J. & Lever, J. Evaluation of the CRREL Permafrost Tunnel. (2008).

66. Amalfitano, S. & Fazi, S. Recovery and quantification of bacterial cells associated with streambed sediments. J. Microbiol. Methods 75, 237–243 (2008).

67. Portillo, M. C., Leff, J. W., Lauber, C. L. & Fierer, N. Cell Size Distributions of Soil Bacterial and Archaeal Taxa. Appl. Environ. Microbiol. 79, 7610–7617 (2013).

68. Eichorst, S. A. et al. Advancements in the application of NanoSIMS and Raman microspectroscopy to investigate the activity of microbial cells in soils. FEMS Microbiol. Ecol. 91, (2015).

69. Lindahl, V. & Bakken, L. R. Evaluation of methods for extraction of bacteria from soil. FEMS Microbiol. Ecol. 16, 135–142 (1995).

50 70. Poté, J., Bravo, A. G., Mavingui, P., Ariztegui, D. & Wildi, W. Evaluation of quantitative recovery of bacterial cells and DNA from different lake sediments by Nycodenz density gradient centrifugation. Ecol. Indic. 10, 234–240 (2010).

71. Kallmeyer, J., Smith, D. C., Spivack, A. J. & D’Hondt, S. New cell extraction procedure applied to deep subsurface sediments. Limnol. Oceanogr. Methods 6, 236– 245 (2008).

72. Kepner, R. L. & Pratt, J. R. Use of fluorochromes for direct enumeration of total bacteria in environmental samples: past and present. Microbiol. Rev. 58, 603–615 (1994).

73. Epstein, S. S. & Rossel, J. Enumeration of sandy sediment bacteria: search for optimal protocol. Oceanogr. Lit. Rev. 9, 759 (1995).

74. Holmsgaard, P. N. et al. Bias in bacterial diversity as a result of Nycodenz extraction from bulk soil. Soil Biol. Biochem. (2011). doi:10.1016/j.soilbio.2011.06.019

75. Wunderlin, T., Junier, T., Paul, C., Jeanneret, N. & Junier, P. Physical Isolation of Endospores from Environmental Samples by Targeted Lysis of Vegetative Cells. J. Vis. Exp. (2016). doi:10.3791/53411

76. Nocker, A., Cheung, C.-Y. & Camper, A. K. Comparison of propidium monoazide with ethidium monoazide for differentiation of live vs. dead bacteria by selective removal of DNA from dead cells. J. Microbiol. Methods 67, 310–320 (2006).

77. Nocker, A., Sossa-Fernandez, P., Burr, M. D. & Camper, A. K. Use of Propidium Monoazide for Live/Dead Distinction in Microbial Ecology. Appl. Environ. Microbiol. 73, 5111–5117 (2007).

78. Heise, J., Nega, M., Alawi, M. & Wagner, D. Propidium monoazide treatment to distinguish between live and dead methanogens in pure cultures and environmental samples. J. Microbiol. Methods 121, 11–23 (2016).

79. Bae, S. & Wuertz, S. Discrimination of Viable and Dead Fecal Bacteroidales Bacteria by Quantitative PCR with Propidium Monoazide. Appl. Environ. Microbiol. 75, 2940–2944 (2009).

80. Steven, B., Pollard, W. H., Greer, C. W. & Whyte, L. G. Microbial diversity and activity through a permafrost/ground ice core profile from the Canadian high Arctic. Environ. Microbiol. 10, 3388–3403 (2008).

81. Niederberger, T. D., Steven, B., Charvet, S., Barbier, B. & Whyte, L. G. Virgibacillus arcticus sp. nov., a moderately halophilic, endospore-forming bacterium from permafrost in the Canadian high Arctic. Int. J. Syst. Evol. Microbiol. 59, 2219–2225 (2009).

51 82. Soina, V. S., Mulyukin, A. L., Demkina, E. V., Vorobyova, E. A. & El-Registan, G. I. The Structure of Resting Bacterial Populations in Soil and Subsoil Permafrost. Astrobiology 4, 345–358 (2004).

83. Caporaso, J. G. et al. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J. 6, 1621–1624 (2012).

84. Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336 (2010).

85. Edgar, R. C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461 (2010).

86. Wang, Q., Garrity, G. M., Tiedje, J. M. & Cole, J. R. Naïve Bayesian Classifier for Rapid Assignment of rRNA Sequences into the New Bacterial Taxonomy. Appl. Environ. Microbiol. 73, 5261–5267 (2007).

87. Price, M. N., Dehal, P. S. & Arkin, A. P. FastTree: Computing Large Minimum Evolution Trees with Profiles instead of a Distance Matrix. Mol. Biol. Evol. 26, 1641–1650 (2009).

88. Salter, S. J. et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol. 12, 87 (2014).

89. Thorsten Pohlert. The Pairwise Multiple Comparison of Mean Ranks Package (PMCMR). R Package (2014).

90. Faith, D. P. Conservation evaluation and phylogenetic diversity. Biol. Conserv. 61, 1–10 (1992).

91. Lozupone, C. & Knight, R. UniFrac: a New Phylogenetic Method for Comparing Microbial Communities. Appl. Environ. Microbiol. 71, 8228–8235 (2005).

92. Dixon, P. & Palmer, M. W. VEGAN, a package of R functions for community ecology. J. Veg. Sci. 14, 927–930 (2003).

93. McMurdie, P. J. & Holmes, S. phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLOS ONE 8, e61217 (2013).

94. José Pinheiro and Douglas Bates. Package ‘nlme’ : Linear and Nonlinear Mixed Effects Models. R Package (2017).

95. Wunderlin, T., Junier, T., Roussel-Delif, L., Jeanneret, N. & Junier, P. Endospore- enriched sequencing approach reveals unprecedented diversity of Firmicutes in sediments. Environ. Microbiol. Rep. 6, 631–639 (2014).

96. Marciano-Cabral, F. & Cabral, G. Acanthamoeba spp. as Agents of Disease in Humans. Clin. Microbiol. Rev. 16, 273–307 (2003).

52 97. Podlipaeva, I., Shmakov, L. A., Gilichinskiĭ, D. A. & Gudkov, A. V. [Heat shock protein of HSP70 family revealed in some contemporary freshwater Amoebae and in Acanthamoeba sp. from cysts isolated from permafrost samples]. Tsitologiia 48, 691– 694 (2006).

98. Shmakova, L. A. & Rivkina, E. M. Viable eukaryotes of the phylum Amoebozoa from the Arctic permafrost. Paleontol. J. 49, 572–577 (2015).

99. Carini, P. et al. Relic DNA is abundant in soil and obscures estimates of soil microbial diversity. bioRxiv 043372 (2016). doi:10.1101/043372

100. Lennon, J. T., Placella, S. A. & Muscarella, M. E. Relic DNA contributes minimally to estimates of microbial diversity. bioRxiv 131284 (2017). doi:10.1101/131284

101. Nocker, A., Richter-Heitmann, T., Montijn, R., Schuren, F. & Kort, R. Discrimination between live and dead cellsin bacterial communities from environmental water samples analyzed by 454 pyrosequencing. Int. Microbiol. Off. J. Span. Soc. Microbiol. 13, 59–65 (2010).

102. Suzuki, M. T. & Giovannoni, S. J. Bias caused by template annealing in the amplification of mixtures of 16S rRNA genes by PCR. Appl. Environ. Microbiol. 62, 625–630 (1996).

103. Rivkina, E., Gilichinsky, D., Wagener, S., Tiedje, J. & McGrath, J. Biogeochemical activity of anaerobic microorganisms from buried permafrost sediments. Geomicrobiol. J. 15, 187–193 (1998).

104. Singh, P. et al. Bacterial communities in ancient permafrost profiles of Svalbard, Arctic. J. Basic Microbiol. n/a–n/a doi:10.1002/jobm.201700061

105. Bernardini J. N., La Duc M. T., Diamond R., Verceles J. Fluorescence-Activated Cell Sorting of Live Versus Dead Bacterial Cells and Spores. NASA Tech Briefs 22 (2012).

106. Leuko, S., Legat, A., Fendrihan, S. & Stan-Lotter, H. Evaluation of the LIVE/DEAD BacLight Kit for Detection of Extremophilic Archaea and Visualization of Microorganisms in Environmental Hypersaline Samples. Appl. Environ. Microbiol. 70, 6884–6886 (2004).

107. Bianciotto, V., Minerdi, D., Perotto, S. & Bonfante, P. Cellular interactions between arbuscular mycorrhizal fungi and rhizosphere bacteria. Protoplasma 193, 123–131 (1996).

108. Biggerstaff, J. P. et al. New methodology for viability testing in environmental samples. Mol. Cell. Probes 20, 141–146 (2006).

53 109. Kirchhoff, C. & Cypionka, H. Propidium ion enters viable cells with high membrane potential during live-dead staining. J. Microbiol. Methods (2017). doi:10.1016/j.mimet.2017.09.011

54 Appendix A: Supplementary Material

Figure 18. Direct cell counts determined by fluorescent microscopy did not show significant differences in the ratio of live to dead cells across the three age categories (Kruskal Wallis, X2(2)

= 1.38, p > 0.05). Values show averages of five replicates and error bars show standard error of the mean.

Shannon (B, ANOVA, F (2,7.83) =192.37, p < 0.01), Chao1 (C, ANOVA, F (2,7.77) = 188.64, p

< 0.01), and PD Whole Tree (D, ANOVA, F (2,7.38) = 386.71, p < 0.01). The youngest samples were most diverse (Tukey HSD post hoc test, ** = p < 0.01), followed by the oldest samples

(Tukey HSD post hoc test, ** = p < 0.01), and the intermediate aged samples were least diverse

(Tukey HSD post hoc test, ** = p < 0.01). Values show averages of five replicates and error bars show standard error of the mean.

Figure 20. Community composition of samples differ based on age and treatment. Principal

Coordinates Analysis (PCoA) based on unweighted Unifrac values of 16S rRNA gene sequences.

Samples differed significantly by age (A, PerMANOVA, R2 = 0.38, F = 17.26, p < 0.001) and treatment for the 19K aged category (B, PerMANOVA, R2 = 0.259, F = 1.87, p < 0.001), the 27K aged category (C, PerMANOVA, R2 = 0.24, F = 1.66, p < 0.001), and the 33K aged category (D,

PerMANOVA, R2 = 0.22, F = 1.49, p < 0.001).

57 Table 4. Average percent difference in relative abundance between the endospore-enriched samples and non-enriched controls across the three age categories (Mann-Whitney-Wilcoxon test, ** = p < 0.01, * = p < 0.05). A negative value implies underrepresentation in the endospore- enriched sample compared to the non-enriched control while a positive value implies overrepresentation. Values show averages of five replicates.

Taxa 19K (%) U-value 27K (%) U-value 33K (%) U-value Proteobacteria -4.3 ± 2.2 6 -4.0 ± 0.3 ** 0 -6.2 ± 2.1 5 Alphaproteobacteria 3.6 ± 0.7 6 -3.8 ± 0.3 ** 0 -0.6 ± 0.3 9 Deltaproteobacteria -8.8 ± 0.3 ** 0 -0.2 ± 0.1 * 1.5 -6.0 ± 0.7 ** 0 Bacteroidetes -3.9 ± 0.7 ** 0 0.0 ± 0.0 4.5 -2.9 ± 0.9 ** 0 Acidobacteria -3.2 ± 0.3 * 0 -0.1 ± 0.0 * 1.5 -0.1 ± 0.1 4.5 Chloroflexi -1.4 ± 0.3 ** 0 -0.6 ± 0.1 ** 0 -0.1 ± 0.1 7 Planctomycetes -1.3 ± 0.1 ** 0 -0.8 ± 0.1 * 0 -0.3 ± 0.1 4