P. F. LONG, H. G. WILDMAN and G. E. AMPHLETT^l

Dept. Natural Products Discovery and ^1Statistics, Glaxo Group Research Ltd., Greenford, Middlesex, UB6 OHE, UK

Code Number: AC94001
File Sizes:
     Text:   30K
     No associated graphics files

Application of molecular methods in microbial ecology has allowed the detailed population study of many microbial communities (Holben et al., 1988; Sayler and Layton, 1990). Amplification and comparison of target sequences, particularly within ribosomal RNA genes, has shown there to be a greater diversity of microorganisms in environmental samples than previously found by more traditional, culture dependent techniques (Pace et al., 1986; Giovannoni et al., 1990; Ward et al., 1990). Therefore, environmental samples such as soils could still offer a bountiful supply of potentially novel microorganisms ripe for exploitation in natural product screening programmes. However, a problem exists in our inability to culture many of these microorganisms whose presence can only be illustrated after direct extraction of genetic material from environmental samples (Ward et al, 1990).

It has been suggested that electrostatic interactions caused by polyvalent cations can hold bacteria in close association with soil particles (Edwards and Bremner, 1965; Stotzky, 1986). This has led to the introduction of methods that use detergents and ion exchange resins to increase the number of bacteria isolated from soils (McDonald, 1986; Herron and Wellington, 1990; Hopkins et al., 1991; Jacobsen and Rasmussen, 1992). Increasing the numbers of bacteria recovered may also lead to a change in diversity of the cultured population (Long and Wildman, 1993). This has had varying degrees of success depending on the soil types studied (Ogram et al., 1987; Steffan et al., 1988; Tsai and Olson, 1991). A previous analysis has suggested that the physico-chemical characteristics of soil has a significant effect on the successful outcome of this technique (Long and Wildman, 1992).

It is common to use an analysis of variance assuming a fixed effects statistical model to analyse the effects of pretreatments on the isolation of actinomycetes from soil. Comparison of the mean values from different soil types, both with and without pretreatment, then allows the effect of the pretreatment to be assessed. However, although different soil types may be used, the conclusions will only be valid for the soil samples selected. If the assumption that the soils are randomly selected from the population is acceptable, then using a model which allows for a random soil effect would yield conclusions which are more generally valid.

Here we use analysis of variance to assess the data generated from the total viable counts of actinomycetes isolated from different soils, using different isolation media, and pretreatments including detergents and a dry heat method. We consider a number of models which make different assumptions about the fixed and random nature of the various factors .

MATERIALS and METHODS

Soil Material. Thirteen soil samples were collected from various locations, which are shown in Table 1. All soils were stored at +4øC prior to use.

----------------------------------------------------------------
---
Soil     Description of Location
number
-------------------------------------------------------------------

7642     Cliff face
7643     Molehill on vineyard
7644     Rhizoplane of Chenin Blanc vine
7645     Rhizoplane of Cabinet Sauvignon vine
7647     Mountain summit
7648     Mountain summit
7649     Forest clearing
7650     Forest on mountain slopes
7651     Forest clearing
7652     Hill top
7653     Garden
7654     Hill top
7655     Scree
----------------------------------------------------------------
Table 1. Soil material
----------------------------------------------------------------


Soil treatments. All soils were air dried at room
temperature to constant weight before the following treatments were
tested:-
     1. 1g soil suspended in 100ml sterile distilled water by
        shaking at 250rpm for 30mins at 25 C.
     2. 1g soil suspended in 100 sterile Defined Germination Media
        (DGM, Labeda and Shearer, 1990) by shaking at 250rpm for 30
        mins at 25 C.
     3. 1g soil suspended in l00ml of 6% (w/v) yeast extract broth
        (Oxoid) containing 0.05% (w/v) sodium dodecyl sulphate
        (SDS, Sigma) by shaking at 250rpm for 30 mins at 37 C
        (Nonomura and Hayakawa, 1988).
     4. 1g soil suspended in 100ml of 0.1M tetrasodium
        pyrophosphate (Na4P2O7, Sigma) by shaking at 250rpm for 30
        mins at 25 C (Long and Wildman, 1992).

     The four dispersal techniques were also carried out on samples
that had been preheated at 120 C for 1 hour (Nonomura and Hayakawa,
1988).


     Isolation Media. Three media recommended for the
isolation of actinomycetes were used. These were: ammonium
chloride-glycerol agar (ACGA) (Okazaki et al., 1987),
oatmeal agar (Kuster, 1959) plus soil extract (OSEA) and
arginine-vitamins agar (AVA, Nonomura and Ohara, 1969). Nystatin
(Sigma) and cycloheximide (Sigma) were added to the media at 50
ugml^-1 each to inhibit the growth of fungi. 250ml of each medium
was used to fill 20cm x 20cm bioassay plates. The plates were
divided into quarters. Each treated soil suspension was diluted in
sterile distilled water to 10^-4. 0.1ml aliquots were spread
inoculated onto each quarter of the bioassay plates. The plates
were then left to dry for 2hrs in a laminar flow cabinet before
incubation at 28 C for 8 weeks.


Enumeration of Actinomycetes and Statistics. Spore forming
colonies and isolates that exhibited a tough stable, or fragmenting
colony type were classed as actinomycetes.


     The experiment had a full factorial design. Counts from each
quarter of an isolation plate were scored. All 342 plates were
counted giving 1248 data values. Analysis of variance was performed
on the data using the SAS statistical package.


RESULTS

Data Handling. Many of the isolation plates did not yield
any viable actinomycete counts after the 8 week incubation period.
Since there were so many zero values in the data, the assumptions
underlying analysis of variance were not met. By taking the mean of
the four replicates for each treatment combination, there was a
reduction in the proportion of zero values in the data set. This
data set could then be analysed using analysis of variance and
furthermore, no important information was lost by taking the mean
of the data in this way.


     Fixed Effects Model. Analysis of variance showed there
were significant differences between the levels of all the factors
(p<0.0001). For example, the use of dry heat gave a
significantly lower total viable count of actinomycetes recovered
from the soils than when the soil samples were left unheated. All
the two factor interactions were also significant (p<0.004). For
example, although most of the soil samples (except #7655) gave a
reduction in total viable counts of actinomycetes on heating, the
magnitude of the difference between the heated and unheated samples
significantly depended on the soil sample being tested. However,
results in Table 2 also indicated that the total mean viable counts
of actinomycetes recovered from each soil was not consistent with
the dispersal fluids used.

-------------------------------------------------------------------
                       DISPERSAL AGENTS
SOIL                         
          Distilled   Defined       0.1M      0.05%   
Overall mean
          water     germination    Na4P2O7    SDS
                      medium
7642      23.1         59.1         21.6       16.3       30.1
7643      97.0         93.6         35.0       24.8       62.6
7644      51.3         48.5         17.4       10.3       31.9
7645      24.6         16.2         10.3       18.8       17.5
7647       5.7          0.8          1.2        1.0       2.1
7648       0.5          1.4          5.5        0.5       2.0
7649       6.6         11.9          3.0        2.2       5.9
7650      10.1         16.2         25.0        5 7      14.3
7651      22.7          4.8         35.8        7.9      17.8
7652      24.1         29.6         14.8       56.6      31.3
7653      54.4         43.2         34.2       21.5      38.3
7654      23.4         19.8         36.3        6.8      21.6
7655       1.5          8.7          2.6        2.8       3.9
------------------------------------------------------------------
Overall   26.5         27.2         18.7       13.5
mean
-------------------------------------------------------------------

Table 2. Effect of different dispersal agents on the total
viable counts of actinomycetes isolated from thirteen soils. 10^4 x
total viable counts of actinomycetes (mean number of colony forming
units/g air dried soil, 4 counts per pretreatment).
-------------------------------------------------------------------


For example, soil #7643 gave the highest mean value recovered using
sterile distilled water as the dispersal fluid, while with soil
#7642 it was with defined germination medium, this was also seen
with the Na4P2O7 (soil #7654) and SDS (soil #7652).
     The use of dry heat had an effect on the total viable counts
of actinomycetes recovered from the soils (Table 3).
-------------------------------------------------------------------
Soil       Unheated     Heated     Overall
                       120C/1hr    mean
-------------------------------------------------------------------
7642        35.0         25.1       30.1
7643        82.1         43.2       62.6
7644        38.8         25.0       31.9
7645        24.4         10.5       17.5
7647         3.0          1.3        2.1
7648         2.4          1.5        2.0
7649         8.0          3.9        5.9
7650        20.1          8.4       14.3
7651        21.1         14.5       17.8
7652        33.9         28.6       31.3
7653        43.1         33.5       38.3
7654        23.8         19.4       21.6
7655         1.1          6.9        3.9
-------------------------------------------------------------------
Overall     25.9     17.0
mean
-------------------------------------------------------------------

Table 3. Effect of a dry heat pretreatment on the total viable
counts of actinomycetes recovered from thirteen soils. See Table 2.



Heating reduced the overall mean viable count from 25.9 x 10^4
CFU/g air dried wt to 17.0 x 10^4 CFU/g air dried wt. All the soil
samples showed a reduction with dry heat, with the exception of
soil #7655, where the mean viable count increased (Table 3).
However, the different dispersal agents did have an effect on the
mean recovery values (Table 4). Defined germination media gave the
highest mean recovery (27.3 x 10^4 CFU/g air dried wt) whereas SDS
gave the lowest (8.5 x 10^4 CFU/g air dried wt) on heated samples.
-------------------------------------------------------------------

Dispersal         Un-        Heated          Overall
agent             heated     120C/1h         mean
-------------------------------------------------------------------
Dist. water        34.5        18.6          26.5
DGM                27.1        27.3          27.2
O.1M Na4P2O7       23.7        13.6          18.7
0.05% SDS          18.4         8.5          13.5
-------------------------------------------------------------------
Overall mean       25.9        17.0
-------------------------------------------------------------------
Table 4. Effect of different dispersal agents on the
recovery of actinomycetes from thirteen soils after a dry heat
pretreatment. See Table 2 (DGM: Defined germination medium)



The use of different isolation media effected the total viable
counts of actinomycetes (Table 5). From the overall mean values,
soil #7643 had the largest recovered population of actinomycetes,
whereas soils #7647 and #7648 had the lowest. There was a
difference between the overall mean viable counts between each
media. The complex medium OSEA gave the highest recovery of
actinomycetes (26.4x10^4 CFU/g air dried wt) whereas the defined
medium AVA gave the lowest (15.8x10^4 CFU/g air dried wt).
-------------------------------------------------------------------
Soil                      MEDIA                 Overall
mean         --------------------------------------------------
--------------- 
          AVA        ACGA         OSEA
          _____________________________   
7642      8.0        42.7         39.4          30.1
7643     39.2        81.5         67.2          62.6
7644     24.9        26.0         44.7          31.9
7745     12 6        20 0         19.8          17 5
7647      2.8         1.7          2.0          2.1
7648      2.8         1.2          2.0          2.0
7649      5.0         5.2          7.6          5.9
7650     24.2        10.4          8.1          14.3
7651     10.8        18.8         23.8          17.8
7652     20.0        31.6         42.2          31.3
7653     41.9        27.8         45.3          38.3
7654     12.4        21.3         30.9          21.6
7655      0.5         1.4          9.8          3.9
-------------------------------------------------------------------
Overall     l5.8     22.3          26.4
mean
-------------------------------------------------------------------

Table 5. Effect of three different isolation media on the
total viable counts of actinomycetes isolated from thirteen soils.
See Table 2 (AVA: arginine vitamins agar, ACGA: ammonium chloride
glycerol agar, OSEA: oatmeal soil extract agar).


-------------------------------------------------------------------

Media     Un-      Heated     Overall
        heated     120C/1h     mean
-------------------------------------------------------------------
AVA      22.4        3.2       15.8
ACGA     27.2       17.4       22.3
OSEA     28.2       24.5       26.4
-------------------------------------------------------------------
Overall  25.9       17.0
mean
-------------------------------------------------------------------

Table 6. Effect of a dry heat pretreatment on the total viable
count of actinomycetes recovered on three different isolation
media. See Tables 2 and 5
-------------------------------------------------------------------


The use of dry heat effected the recovery of actinomycetes on each
medium (Table 6). The total viable counts were less with heated
samples than with unheated. Again, although the overall mean heated
counts were lower than the unheated, the largest recovery of
actinomycetes was obtained using OSEA (24.5x10^4 CFU/g air dried
wt). The use of different dispersal agents effected the recovery of
actinomycetes on each medium. However, this effect was not
consistent (Table 7).
-------------------------------------------------------------------
Dispersal     AVA     ACGA     OSEA      Overall
agent                                    mean
-------------------------------------------------------------------

Distilled     22.0     23.5     34.1     26.5
DGM           14.5     36.8     30.4     27.2
O.1M          15.2     16.9     23.9     18.7
Na4P2O7
0.05%         11.4     11.9     17 0     13 5
SDS
-------------------------------------------------------------------
Overall       15.8     22.3     26.4
mean
-------------------------------------------------------------------

Table 7. Effect of different isolation media on the total
viable count of actinomycetes recovered using four dispersal
agents. See Table 2.



On AVA and OSEA the greatest counts were obtained using distilled
water as the dispersal agents. However, on ammonium chloride
glycerol agar the highest counts were obtained using defined
germination medium. SDS consistently gave the lowest counts.


Random Effects Model. From the fixed effects model all the
factors chosen in this study (soil, media, heat treatment,
dispersal agents) were found to be significant, but the conclusions
are only valid for the soils selected. If we can assume that the
soils and media are a random selection from all the soils and media
in the population, then we can assume these effects are random.
Hence the conclusions we draw should apply whatever the sample of
soils and media we select. Both the pretreatments and heat method
are assumed to be fixed. Analysis of variance was performed to
analyse the data set under these conditions. There was a
significant soil-dispersal agent interaction (p=0.004: Table 2),
i.e., the recovery of viable actinomycetes using the
dispersal agents depended on the soil being tested. There was also
a significant interaction between the heat treatment and the
dispersal agent (p=0.024; Table 4). The effect of heat on the total
viable counts of actinomycetes depended on the dispersal agent
used. Hence, distilled water gave a reduction on heating, whereas
there was a slight increase on heating with DGM. The recovery of
actinomycetes was significantly dependent on the soil sampled
(p=0.002). However, using the random effects model, analysis of
variance suggested no overall differences in total viable counts of
actinomycetes between dispersal agents used (p=0.270), the
isolation media (p=0.192), or whether the soil was heated or not
(p=0.10).


     Although we may assume that the medium is a random effect,
only three media have been selected therefore, the power of the
statistical tests to make significant conclusions is low. Assuming
that the only random would be the soils tested, then were still
significant soil-dispersal agent (p=0.004; Table 2), dispersal
agent-heat interactions (p=0.038; Table 4) and differences amongst
the soils (p=0.002). There are likely to be overall significant
differences between the isolation media (p>0.026), between
heated and unheated samples (p>0.011), but not between the
dispersal fluids (p>0.075). Unfortunately, with this model it is
not possible to obtain exact p values for some of the effects and
so we can only make general conclusions.


DISCUSSION


It has been estimated that only 12% of bacteria in the environment
have been cultured (Bull et al, 1992). There is potentially a large
pool of untapped organisms that could be available for use in
natural product screening programmes. The challenge for microbial
ecology is to isolate these organisms whose presence so far has
only been highlighted by the use of molecular techniques.
Typically, microorganisms are isolated after selective pressure has
been applied to the source material such as a pretreatment, or by
the use of selective isolation media. Our strategy in this report
has been to take the first steps towards reassessing methods
commonly used for the isolation of actinomycetes in an attempt to
see just how good the treatments actually are.


     In this study, 13 soils collected from a wide variety of
different locations were subjected to both physical and chemical
pretreatments as well as selective isolation media. Soil dilution
plating was carried out by dividing large bioassay plates into
quarters. This was to encourage the growth of as many of the soil
actinomycetes as possible. Inoculation of the same volume of soil
into standard size Petri dishes may have resulted in the
inhibition, or masking of some organisms by others (Williams and
Vickers, 1988). Gross changes in the actinomycete population of
each soil was assessed by analysis of variance on the total viable
counts of actinomycetes recovered and compared using both fixed and
random effects statistical models.


     Fixed effects models are the most widely used, certainly in
actinomycete ecology, to assess how effective pretreatments or
selective media are. A drawback of this model is that conclusions
based on the experimental data can only be applied to the actual
soils tested, and recommendations to use pretreatments or media for
other soils (which all too often happens) are not fully justified.
Hence, many methods for the isolation of organisms are not as good
as they should be because treatments have been recommended solely
on the basis of the results of an experiment that went well.
Indeed, the reverse might also be true. Pretreatments/media may
have been rejected because the results of a particular fixed
effects experiment were poor, which may be attributable to the soil
studied. This may be a possible reason why so few microorganisms in
the environment have been cultured. Evidence for this hypothesis
can be found from this study. For example, the fixed effects model
predicts a significant difference between the dispersal fluids
used. In the literature all the dispersal fluids used in this study
have been quoted as increasing the numbers of actinomycetes
isolated in comparison to water. From the random effects model,
however, where the soils were considered the only random effects,
there was no significant difference between the dispersal fluids.
This highlights the unpredictable nature of the use of detergents,
or spore activation agents in actinomycete isolation work.
Interestingly, there was a soil-dispersal agent interaction in both
fixed and random effects models. It may well be that we have to
understand the physico-chemical nature of soil before we can
effectively use such dispersal agents.


     Theoretically, it seems that the best model to use would be
the one where just soil samples are regarded as random, but this
model fails to yield exact p values. However, by considering all
models we can conclude that there are clear differences among the
soil samples, significant soil-dispersal agent and dispersal
agent-heat interactions, and some evidence of overall differences
between media and between heated and unheated samples.


     We recommend that when assessing a new pretreatment or
selective medium for the isolation of actinomycetes that a full
factorial experiment be designed and the results analysed using an
analysis of variance where the source material be considered as a
random effect in the model.


     As many soils as possible should be looked at. From this study
the random effects model did not give exact p-values. This may not
be critical since the first stage in assessing a new method would
be an estimate of gross changes in the total viable counts in a
population. We have not been able to find in the actinomycete
literature, a selective isolation method quoted where a change in
the diversity of the population has not been accompanied by a large
swing in the total viable counts. Once a trend from the full
factorial experiment has been established, then a smaller
experiment can be designed to look at changes in the diversity of
the population as a result of the treatment. It is important that
this experiment be small, yet include all the factors investigated
in the larger experiment. This is necessary because accurate
identification of bacteria, particularly actinomycetes, is a very
labour intensive process and so is not practical on large numbers
of organisms.


     It is hoped that following such a strategy when assessing
pretreatments or selective media will increase the numbers of novel
organisms isolated in the future, because a more accurate
prediction of treatment effects should result.


ACKNOWLEDGEMENTS. The authors would like to thank Miss J. S.
Whybrew for typing the manuscript. The soils used in this study
were imported into the UK under licences issued by the Ministry of
Agriculture, Fisheries and Food.


REFERENCES


Bull, A. T., M. Goodfellow & Slater, S. H. (1992).
Biodiversity as a source of innovation in biotechnology. Ann.
Rev. Microbiol., 46: 219-252


Edwards, A. P. & J. M. Bremner (1965). Dispersion of mineral
colloids in soils using cation exchange resins. Nature, 205:
208-209


Giovannoni, S. J., T. B. Britschgi, C. L. Moyer & K. G.
Field (1990). Genetic diversity in Sargasso Sea bacterioplankton.
Nature, 344: 60-63


Herron, P. R. & E. M. H. Wellington (1990). New
methods for extraction of streptomycete spores from soil and
application to the study of lysogeny in sterile amended and
nonsterile soil. Appl. Env. Microbiol., 56: 1406-1412


Holben, W. E., J. K. Jansson, B. K. Chelm & J. M. Tiedje
(1988). DNA probe method for the detection of specific
microorganisms in the soil bacterial community. Appl. Env.
Microbiol., 54: 703-711


Hopkins, D. W., S. J. Macnaughton & A. G. O'Donnell (1991).
A dispersion and differential centrifugation technique for
representatively sampling microorganisms from soil. Soil Biol.
Biochem., 23: 217-225


Jacobsen, C. S. & O. F. Rasmussen (1992). Development and
application of a new method to extract bacterial DNA from soil
based on separation of bacteria from soil with cation-exchange
resin. Appl. Env. Microbiol., 58: 2458-2462


Kuster, E. (1959). Outline of a comparative study of
criteria used in characterisation of actinomycetes. Int. Bull.
Bact. Nomen. Tax., 9: 98-108


Labeda, D. P. & M. C. Shearer (1990). Isolation of
actinomycetes for biotechnological applications In: D. P. Labeda
(ed.) Isolation of Biotechnology Organisms from Nature.
McGraw-Hill, pp. 1-19


Long, P. F. & H. G. Wildman (1992). Strategy for the
use of pretreatments in the isolation of non-streptomycete
actinomycetes from soil. Actinomycetes, 4: 59-64


McDonald, R. M. (1986). Sampling soil microfloras:
Dispersion of soil by ion exchange and extraction of specific
microorganisms by elutriation. Soil Biol. Biochem., 18:
399406


Nonomura, H. & M. Hayakawa (1988). New method for the
selective isolation of soil actinomycetes In: Y. Okami, T. Beppu &
H. Ogawara (eds.) Biology of Actinomycetes '88. Japan Sci.
Press, pp. 288-293


Nonomura, H. & Y. Ohara (1969). The distribution of
actinomycetes in soil. VI. A selective plate culture isolation
method for Microbispora and Streptosporangium
strains. J. Ferm. Technol., 47: 463 -469


Ogram, A., G. S. Sayler & T. Barkay (1987). The extraction
and purification of microbial DNA from sediments. J. Microbiol.
Meth., 7: 57-66


Okazaki, T., R. Enokita, H. Miyaoka, T. Takatsu & A. Tomkata
(1987). Chloropolysporins A, B and C, novel glycopeptide
antibiotics from Faenia interjecta sp.nov. J.
Antibiot., 40: 917-923


Pace, N. R., D. A. Stahl, D. J. Lane & G. J. Olsen (1986).
The analysis of natural microbial populations by ribosomal RNA
sequences. Adv. Microb. Ecol., 9: 1-55


Sayler, G. S. & A. C. Layton (1990). Environmental
application of nucleic acid hybridisation. Ann. Rev.
Microbiol., 44: 625-648


Steffan, R. J., J. Goksoyr, A. K. Bej & R. M. Atlas
(1988). Recovery of DNA from soils and sediments. Appl. Env.
Biol., 54: 2908-2915


Stotzky, G. (1986). Influence of soil mineral colloids on
metabolic processes, growth, adhesion and ecology of microbes and
viruses. In: P. M. Haung & M. Schnitzer (eds.) Interactions of
Soil Minerals with Natural Organisms and Microbes. Soil Sciences
Society of America, pp. 305-428


Tsai, Y-L. & B. H. Olson (1991). Rapid method for direct
extraction of DNA from soil and sediments. Appl. Env.
Microbiol., 57: 1070-1074


Ward, D. M., R. Weller & N. M. Bateson (1990). 16S RNA
sequences reveal numerous uncultured microorganisms in a natural
community. Nature 345: 63-65


Williams, S. T. & J. C. Vickers (1988). Detection of
actinomycetes in natural habitats - problems and perspectives. In:
Y. Okami, & H. Ogawara (eds.). Biology of Actinomycetes '88.
Japan Sci. Press, Tokyo, pp. 265-270.



Copyright 1994 C. E. T. A.