Antonie van Leeuwenhoek73: 9–14, 1998. 9c 1998Kluwer Academic Publishers. Printed in the Netherlands.

Broad-scale analysis of soil microbial community DNA from Uplandgrasslands

Christopher D. Clegg�, Karl Ritz & Bryan S. GriffithsSoil Plant Dynamics Unit, Cellular and Environmental Physiology, Scottish Crop Research Institute, Invergowrie,Dundee DD2 5DA, Scotland, U.K.; (� author for correspondence)

Received 24 January 1997; accepted 4 September 1997

Key words:genetic diversity, microorganisms, soil community DNA, upland grassland, community structure

Abstract

We have applied a broad-scale approach to the analysis of DNA extracted from soils which support characteristicgrasslands at an upland site in the UK. To test for the degree of coherence between microbial and vascular com-munities, grasslands were characterised as ‘improved’, ‘semi-improved’, or ‘unimproved’, according to the degreeof management they had received and consequent botanical composition. Microbial DNA was extracted directlyfrom the grassland soils and analysed by three techniques: (i) thermal denaturation, which profiles the guanine andcytosine (G+C) base distribution within the community; (ii) cross hybridisation of the DNA which measures thedegree of similarity between the samples; (iii) measurement of reassociation kinetics of denatured DNA, whichprovides a measure of the complexity of the DNA. Thermal denaturation revealed significant differences in the%G+C composition of the communities. DNA from the improved soil had the highest median %G+C value, whilstthat from the unimproved soil had the lowest. The relative distribution of G+C bases also differed significantlybetween the samples from the three grasslands. Cross hybridisation of DNA from the different soils also indicatedsignificant differences in the degree of similarity between the DNA from the grasslands, with unimproved showing59% similarity to improved. Indices from the cross hybridisation assay suggested that, in terms of complexity, thesamples ranked unimproved> semi-improved> improved. Reassociation kinetics supported this conclusion, butthe rates of reassociation were such that less than 40% reassociation occurred over a 31-day period, thus preventingcalculation of Cot1=2 .

Introduction

Micro-organisms are key components in soils, con-tributing important roles in the decomposition oforganic matter, nutrient cycling and soil structure. Inrecent years there has been a move from treating micro-bial populations in soils as a single biomass entity to theapplication of molecular techniques to describe specif-ic components of such populations. Analysis of thecompositional structure of soil microbial communitiesis hampered by its inherent complexity and the knowl-edge that a minority of soil bacteria, as few as 2–4%(Olsen & Bakken, 1987), and an unknown proportionof soil fungi maybe isolated by standard culturing tech-niques.

Molecular techniques have been used to study thecomplexities of microbial populations in soil. Suchstudies have tended towards the development of probesfor phylogenetically related species (e.g. Stahl et al,1988; Mobarry et al, 1996) or rRNA sequence analy-sis in population studies (e.g. Giovannoni et al, 1990).Although much information is obtained from specificorganisms and their sequences, it does not provide aholistic view of a microbial community. This can beachieved by broad scale approaches utilising the DNAextracted from microbial populations within the soil.Such broad-scale analyses include the determinationofthe relative abundance of the nucleic acids guanine andcytosine (G+C) measured using thermal denaturationprocedures (Torsvik et al, 1990a; Leser et al, 1995)

10

or density gradient centrifugation (Holben & Harris,1995); cross hybridisation of different DNA samples(Lee & Fuhrman, 1990; Ritz & Griffiths 1994; Grif-fiths et al, 1996); and reassociation kinetics (Torsviket al, 1990a, 1990b). Such approaches allow us todetermine degrees of microbial community similarityand complexity, and overall shifts in the communitystructure.

In this study we report on the application of ther-mal denaturation, cross hybridisation and reassociationkinetics of microbial DNA extracted from soils takenfrom three characteristic vegetation sequences at anupland grassland site within the U.K. The vegetationsequences at the site differ as a result of land man-agement practices such as grazing and fertiliser input.The hypothesis was that the soil microbial communi-ties under the different swards would be different dueto the selection pressures exerted on the communitiesby the different practices and botanical composition ofthe corresponding vegetation.

Methods and Materials

Soils

Soil samples were taken from a site at Borrowdalein Cumbria, U.K. (national grid reference - NY (35)236 122). At the site there were 3 grassland types,unimproved [National Vegetation Classification: U4a(sandy silt loam, pH 4 )], semi-improved [U4b (sandysilt loam, pH 4.8)] and improved [MG6 (sandy siltloam, pH 5.3)], as determined by vegetation sequenceclassification (determined by Dr. J. Rodwell and Dr.S. Edwards, Lancaster University, U.K.). Soil sampleswere collected by removing cores and sieving througha 2 mm sieve before storage at -20�C prior to use.

Extraction of microbial DNA from soils

Microbial DNA was obtained from soil by directextraction as described previously (Clegg et al, 1997),quantified by the diphenylamine reaction (Lichtenstein& Draper, 1985), and stored at -20�C until required.

Thermal denaturation of DNA

Approximately1.5�g of soil extracted DNA was dilut-ed in 650�l 0.1 x standard saline citrate (SSC: Sam-brook et al, 1989) and placed in a 1 cm light path quartzcuvette. Using a PTP-C Peltier Temperature Program-

mer attached to a Perkin Elmer Lambda 2S UV/VISSpectrometer, DNA samples were heated from 45 to 90�C at a rate of 1�C min�1 and changes in absorbancewere logged at 260 nm. Individual absorbance pro-files were normalised on a scale of 0–100% basedon absorbance values at 45–90�C respectively. Threereplicate melts were carried out per sample. Melt pro-files of the DNA were then parametised by fitting ageneralised logistic equation to the data between 60–90 �C (Ritz et al, 1997). The resulting curves are afunction of the parameters ‘b’ (the slope of the meltcurve or spread of the %G+C profile), ‘m’ (the pointof inflexion or median %G+C content), and ‘t’ (thedegree of symmetry of the profile about the median;Ritz et al, 1997) which were compared by analysis ofvariance. Melting profiles were converted to %G+Cdistribution according to Mandel et al, (1970).

Cross hybridisation of soil DNA

Aliquots of 500 ng of DNA were denatured at 90–95�Cfor 5–10 min followed by rapid cooling on ice and dotblotted onto Hybond-N nylon membranes (AmershamInternational, U.K.). The wells in the dot blot apparatus(BioRad) and DNA spots were washed with 50�l 5 xSSC. Membranes were allowed to air dry before DNAwas bound by crosslinking under UV and baking for30 min at 80�C. To generate the probes for the hybridi-sation reaction, 200 ng of soil extracted DNA was firstdigested using 20 unitsRsa1 (GibcoBRL) in a reactionmix of 20�l according to manufacturers instructions.DNA probes were prepared from the digested DNAusing a random primer labelling kit (Amersham Inter-national, U.K.). The probe generation reaction wasallowed to proceed overnight at 30�C. Probe yieldwas determined following manufacturers instructions.

Hybridisations were carried out using the hybridis-ation buffer recommendedby the manufacturer (Amer-sham International, U.K.). Membranes were allowedto pre-hybridise for 1 h at 60�C. The hybridisationbuffer was removed and replaced with fresh solutionat 0.3 ml cm�2 membrane. Denatured labelled probewas added to the hybridisation buffer at 10 ng ml�1

and allowed to hybridise at 60�C for 20–24 h. Mem-branes were washed and signal generation determinedaccording to manufacturers instructions. Intensity ofthe hybridisation signals on the membranes was deter-mined by measuring light output from each dot usinga luminometer (Dynotech, California).

Similarity indices, S, of the DNA samples werecalculated as previously described (Lee and Fuhrman,

11

1990; Ritz & Griffiths, 1994). Mean and 95% confi-dence intervals were calculated using a bootstrap pro-cedure (Griffiths et al., 1996). If the target and probe areequally diverse then the reciprocal cross of the targetprobe hybridisation give the same observed similarity(S) value. If one sample is more diverse than the other,then S is not equal and the couplets are asymmetric;the probe with the highest S value is the more diverseof the two samples and the true degree of similarity isgiven by the lowest value of S (Lee & Fuhrman, 1990;Ritz & Griffiths, 1994).

Reassociation

Reassociation rates of soil DNA andEscherichia coliDNA were determined essentially according to themethod of Torsvik et al (1995). DNA was shearedin 0.1 x SSC by sonication with a microprobe to pro-vide fragments of about 700–1000 bp. About 30�gsheared DNA in 40�l 0.1 x SSC was mixed with 30�l 20 x SSC and 30�l dimethyl sulphoxide, DMSO(Sigma) such that the final concentrations were 6 xSSC and 30% DMSO. The final mix was placed in a1 mm path length cuvette and degassed by placing in avacuum until bubbles of air ceased to appear. Solutionswere then overlayeredwith mineral oil and the cuvettescapped to avoid loss by evaporation. The DNA wasmelted by heating the solutions to 90�C for 5 min in aPTP-C Peltier Temperature Programmer attached to aPerkin Elmer Lambda 2S UV/VIS Spectrometer. Thetemperature was then reduced to 50�C (about 25�Cbelow the melting temperature) and held for 30 minand absorbance values logged every 30 s. Thereafter,absorbance values were logged every 5 min. A con-trol blank made up without DNA was also included tocheck for baseline drift of the spectrophotometer. Thereaction was allowed to proceed for 31 days. Reasso-ciation rates were determined from the point after thecollapse (Torsvik et al, 1995). The percentage reasso-ciation was plotted against log Cot (moles l�1 x sec).

Statistics

Results were considered significant at P� 0.05 usingthe analyses described under the appropriate section.

Table 1. DNA yields and absorbance ratios extracted from theBorrowdale soils

Soil yield (�g g�1 A260:230 A260:280

dry soil)

Improved 10.5 1.50 1.64

Semi-improved 23.0 1.26 1.56

Unimproved 24.0 1.32 1.62

Table 2. Parameters for the melting curves of the DNAextracted from the three Borrowdale grasslands (n=3)

Parameters

Grassland b m t

Improved 0.50 78.4 2.88

Semi-improved 0.57 76.5 3.75

Unimproved 0.42 74.69 2.27

LSD 0.06 1.84 1.85

E. coli 0.79 74.0 3.42

LSD: least significant difference

Results

DNA extraction from soil

The yields and purities of the DNA extracted from thethree different soil types at Borrowdale are given inTable 1.

Thermal denaturation of soil extracted DNA

Melting profiles of DNA revealed differences in themicrobial DNA samples from soils. Analysis of theparameters for the curves revealed significant differ-ences in ‘b’ and ‘m’ (Table 2). The greatest ‘b’ val-ue, i.e. that showing greater dominance of a par-ticular proportion of G+C, was that from the semi-improved soil, and the least from the unimprovedsoil. The %G+C distribution of the DNA from thethree soils is represented in Figure 1. There was aconsistent and significant shift in the median %G+Ccomposition (i.e. ‘m’) of the DNA extracted fromthe unimproved, semi-improved and improved vegeta-tion sequences respectively.Hyperchromicity values of28.1, 19.2 and 20.4 were determined for DNA extract-ed from the improved, semi-improved and unimprovedgrasslands respectively when melted in 0.1% SSC.E.coli DNA gave a hyperchromicity value of 36.

12

Figure 1. % G+C distribution profiles of microbial community DNAderived from improved (—-), semi-improved (- - -) and unimproved(� � �) grasslands at Borrowdale, andE. coli (- x -).

Table 3. Similarity indices (95% confidence intervals) for thereciprocal cross hybridisation of DNA for the three soils at Bor-rowdale (n=3)

target

imp s-imp unimp

probe: imp – 92 (76–110) 59 (53–68)

s-imp 206 (193–219) – 77 (72–82)

unimp 221 (136–408) 109 (74–199) –

key; imp: improved; s-imp: semi-improved; unimp: unimproved

Cross hybridisation of DNA

Similarity indices for the cross hybridisation resultsare given in Table 3. All the crosses of DNA fromthe three grassland types were asymmetric. The low-er of the two similarity indices in reciprocal cross-es is indicative of the similarity, e.g. DNA from theimproved grassland was 92% and 59% similar to thatfrom the semi-improved and unimproved grasslandsrespectively.

Reassociation of soil DNA

The signals measured in the spectrophotometer for soilDNA samples were somewhat variable over the timeintervals measured during the latter stages of the reas-sociation. Thus, mean values were taken over uniformlog intervals of 0.5, and are plotted in Figure 2. The rateof reassociation of the soil DNA was considerably lessthan that of the single strain DNA fromE. coli (Fig-ure 2). ForE.coli, 100% reassociation occurred within24 hr, giving a Cot1=2 of 0.86. The kinetics of reas-sociation of DNA from improved and semi-improvedsamples were very similar, and consistently slower forthe unimproved sample (Figure 2). After 31 days, the

Figure 2. Reassociation kinetics of microbial community DNA fromimproved (#), semi-improved (M) and unimproved (E ) grasslandsat Borrowdale, andE. coli (- x -).

extent of reassociation for the samples was about 41%,38% and 24% for the improved, semi-improved andunimproved DNA respectively.

Discussion

The results from the three assays suggested that therewere distinct differences in the microbial communitystructure in the soils taken from the different vege-tation sequences. The DNA was obtained by directextraction from the soils, as opposed to fractionationof the different microbial components, as this is con-sidered to give the best representation of the microbialcommunity (Harris, 1994). Thus, the results obtainedin this study are representative of that part of the totalmicrobial population which was lysed and the DNAsubsequently extracted and purified; this is likely to berepresentative of a major fraction of the total bacterialcommunity, but is perhaps less representative of thefungal community (Harris 1994). Some humic impu-rities were however co-extracted with DNA from thesoil samples, as indicated by the A260:230 ratios andhyperchromicity values.

Results from the melting profiles and %G+C dis-tribution indicated some differences in the DNA com-position of the soils. Melting profiles which are sim-ilar indicate that the microbial DNA from the soils iseither the same in terms of microorganisms present and

13

their abundance, or that the communities are composedof different organisms, but with coincidentally simi-lar %G+C compositions. Alternatively, melting curveswhich differ are indicative either that the same microor-ganisms are present, with differences in their relativeabundance, or the communities are composed of differ-ent microorganisms. Thus %G+C profiling techniquesneed to be checked with cross-hybridisation data todistinguish between these cases.

Percentage similarity indices for cross hybridisa-tion of DNA from the different vegetation sequenceswere asymmetric. Asymmetric indices provide infor-mation about the relative complexity of the DNA sam-ples (Lee & Fuhrman, 1990; Ritz & Griffiths, 1994).Their conclusion was that when reciprocal hybridis-ations were asymetric, this was due to differences inthe complexity of the two DNA samples, and that thelower of the indices for the probe was an indicatorof the ‘true’ degree of similarity. In the context ofthis study we determined that the DNA extracted fromthe improved grassland was 92% similar to that ofthe semi-improved grassland DNA, and 59% similarto that of the unimproved grassland (Table 3). Addi-tionally, when the probe is more complex than thetarget, this gives the greater similarity value (Lee &Fuhrman, 1990; Ritz & Griffiths, 1994). The resultstherefore indicate that in terms of the relative com-plexity of the DNA extracted from the grasslands, thesamples ranked thus: unimproved> semi-improved>improved (Table 3).

A problem in studying microbial communities isthat similarity indices for cross hybridisations are com-plicated by both the number of species, and their rela-tive abundance. An example of this is the cross betweenthe improvedand semi-improvedDNA. The % similar-ity, from the reciprocal hybridisation crosses here indi-cate that the true degree of similarity of the improvedand semi-improved DNA is 92%, although the sim-ilarity indices clearly indicate that the crosses wereasymmetric. This may be interpreted that the num-bers of ‘species types’ are essentially the same for thetwo samples, however, their relative abundance dif-fers. The assay is a broad scale approach to studyingchanges in the relative structure of the microbial pop-ulations and does not offer a high degree of resolution.Indeed, an absolute determination of the true similari-ty between communities may only be possible throughrRNA sequence analysis.

Torsvik et al, (1994) demonstrated increasing ratesof reassociation of DNA in samples of increasing com-plexity. This correlates with our original hypothesis

and findings in the cross hybridisations. An intrinsicelement of the cross hybridisation assay involves an18-hour hybridisation, which for the control hybridi-sation is effectively a form of reassociation assay. Inthis study we obtained a Cot1=2 for E. coli of 0.86 moll�1 s�1, which compares favourably with the valueof 0.72 reported by Torsvik et al (1994). The reas-sociation data suggests that in terms of complexity,improved and semi-improved samples were very sim-ilar, and unimproved was more complex. The rates wemeasured were slow in comparison to those of Torsviket al (1994) who determined Cot values of prokaryoticDNA from the bacterial fraction. In this study we deter-mined Cot values on DNA extracted directly from soilwhich invariably contained some DNA of eukaryoticorigin.

The greater yields of microbial DNA obtained fromthe unimproved and semi-improved grassland soils arenot necessarily a reflection of a greater diversity ofmicroorganisms within the soils. It is possible that asoil may harbour fewer numbers of microorganismsthan another soil, but the number of different typesof microorganisms is greater. Further, in this study theunimproved and semi-improved grassland soils yieldedsimilar quantities of microbial DNA, more than doublethat obtained from the improved grassland. However,the cross hybridisation indices (Table 3) suggest thatthere is a greater similarity between the improved andsemi-improved DNA (92%), with different extractionyields, than that of the semi-improved and unimprovedDNA (77%) which had similar extraction yields.

Combining the results of the three assays provides aclearer perspective on differences between the micro-bial populations within the different soils. Soil fromthe unimproved grassland is clearly different to thatfrom improved, being more complex, having a low-er average %G+C content and a wider distribution of%G+C spread and showing only 59% similarity. Dif-ferences between the semi-improved and the improvedare less distinct, indicative perhaps of the transition thesemi-improved represents between the two extremes.

The results from this study demonstrate that a broadscale approach to the analysis of soil microbial popula-tions can distinguish between different microbial com-munities prevailing in the grassland soils. The next stepis to increase the resolution of the analyses to elucidatewhich components of the community are responsiblefor these differences.

14

Acknowledgements

We would like thank Alexandra Holmes for technicalassistance and Jim McNicol for statistical assistance.The work was funded by the Scottish Office Agri-culture, Environment and Fisheries Department undertheir flexible-funded MICRONET programme.

References

Clegg CD, Ritz K & Griffiths BS (1997) Direct extraction of micro-bial community DNA from humified upland soils. Lett. Appl.Microbiol. 25: 30–33

Giovannoni SJ, Britschgi TB, Moyer CL & Field KG (1990) Geneticdiversity in Sargasso sea bacterioplankton. Nature 345: 60–63

Griffiths BS, Ritz K & Glover LA (1996) Broad-Scale approaches tothe determination of soil microbial community structure: Appli-cation of the community DNA hybridisation technique. Microb.Ecol. 31: 269–280

Harris D (1994) Analyses of DNA extracted from microbial com-munities. In: Ritz K Dighton J & Giller KE (Ed) Beyond theBiomass: Compositional and Functional Analysis of Soil Micro-bial Communities (pp. 111–118) John Wiley, Chichester, U.K.

Holben WE & Harris D (1995) DNA-based monitoring of totalbacterial community structure in environmental samples. Mol.Ecol. 4:627–631

Lee S & Fuhrman JA (1990) DNA hybridisation to compare speciescompositions of natural bacterioplankton assemblages. Appl.Environ. Microbiol. 56: 739–746

Leser TD, Boye M & Hendriksen NB (1995) Survival and activityof Pseudomonas sp. strain B13(FR1) in a marine microcosmdetermined by quantitative PCR and an rRNA-targeting probeand its effect on the indigenous bacterioplankton. Appl. Environ.Microbiol. 61: 1201–1207

Lichtenstein C & Draper J (1985) Genetic engineering of plants.In: Glover DM (Ed) DNA Cloning, Volume 2 (pp. 67–120). IRLPress, Oxford

Mandel M, Igambi L, Bergendahl J, Dobson ML & Scheltgen E(1970) Correlation of melting temperature and caesium chloridebuoyant density of bacterial deoxyribonucleic acid. J. Bacteriol.101: 333–338

Mobarry BK, Wagner M, Urbain V, Rittmann BE & Stahl DA (1996)Phylogenetic probes for analysing abundance and spatial organi-sation of nitrifying bacteria. Appl. Environ. Microbiol. 62: 2156–2162

Olsen A & Bakken LR (1987) Viability of soil bacteria: optimisationof plate counting technique and comparison between total countsand plate counts within different size groups. Microb. Ecol. 13:59–74

Ritz K & Griffiths BS (1994) The potential application of a communi-ty hybridisation technique for assessing changes in the populationstructure of soil microbial communities. Soil Biol. Biochem. 26:963–971

Ritz K, Griffiths BS, Torsvik VL & Hendriksen NB (1997) Analysisof soil and bacterioplankton community DNA by melting profilesand reassociation kinetics. FEMS Microbiol. Lett. 149: 151–156

Sambrook J, Fritsch EF, and Maniatis T (1989) Molecular Cloning:A laboratory Manual, second edition. Cold Spring Harbor Labo-ratory Press, Cold Spring Harbor, New York

Stahl DA, Flesher B, Mansfield HR & Montgomery L (1988) Use ofphylogenetically based hybridisation probes for studies of rumi-nal microbial ecology. Appl. Environ. Microbiol. 54: 1079–1084

Torsvik V, Daae FL & Goksøyr J (1995) Extraction, purificationand analysis of DNA from soil bacteria. In: Trevors JT & vanElsas JD (Eds) Nucleic acids in the environment: Methods andapplications (pp 29–48). Springer Verlag, Berlin

Torsvik V, Goksøyr J, & Daae FL (1990a) High diversity in DNA ofsoil bacteria. Appl. Environ. Microbiol. 56: 782–787

Torsvik V, Goksøyr J, Daae FL, Sørheim R, Michalsen J, & SalteK (1994) Use of DNA analysis to determine the diversity ofmicrobial communities. In: Ritz K Dighton J & Giller KE (Ed)Beyond the Biomass: Compositional and Functional Analysis ofSoil Microbial Communities (pp. 39–48) John Wiley, Chichester,U.K.

Torsvik V, Salte K, Sørheim R, & Goksøyr, J. (1990b) Comparisonof phenotypic and DNA heterogeneity in a population of soilbacteria. Appl. Environ. Microbiol. 56: 776–781