Notes on the characterization of prokaryote strains for taxonomic purposes - CORE
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
View metadata, citation and similar papers at core.ac.uk brought to you by CORE
provided by Digital.CSIC
International Journal of Systematic and Evolutionary Microbiology (2010), 60, 249–266 DOI 10.1099/ijs.0.016949-0
Taxonomic Notes on the characterization of prokaryote strains
Note for taxonomic purposes
B. J. Tindall,1 R. Rosselló-Móra,2 H.-J. Busse,3 W. Ludwig4 and P. Kämpfer5
Correspondence 1
DSMZ-Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Inhoffenstraße 7B,
Brian J. Tindall D-38124 Braunschweig, Germany
bti@dsmz.de 2
Grup de Microbiologia Marina, Departament d’Ecologia I Recursos Marins, IMEDEA (CSIC-UIB),
C/Miquel Marqués 21, E-07190, Esporles, Spain
3
Institut für Bakteriologie, Mykologie und Hygiene, University of Veterinary Medicine Vienna,
Veterinärplatz 1, A-1210 Vienna, Austria
4
Lehrstuhl für Mikrobiologie, Technische Universität München, Am Hochanger 4, D-85354
Freising-Weihenstephan, Germany
5
Institut für Angewandte Mikrobiologie, Justus-Liebig-Universität Giessen, Heinrich-Buff-Ring 26-
32 (IFZ), D-35392 Giessen, Germany
Taxonomy relies on three key elements: characterization, classification and nomenclature. All three
elements are dynamic fields, but each step depends on the one which precedes it. Thus, the
nomenclature of a group of organisms depends on the way they are classified, and the
classification (among other elements) depends on the information gathered as a result of
characterization. While nomenclature is governed by the Bacteriological Code, the classification
and characterization of prokaryotes is an area that is not formally regulated and one in which
numerous changes have taken place in the last 50 years. The purpose of the present article is to
outline the key elements in the way that prokaryotes are characterized, with a view to providing an
overview of some of the pitfalls commonly encountered in taxonomic papers.
INTRODUCTION placement of a genus in a family should be mentioned, and
this can be extended to the other hierarchical levels as these
The characterization of a strain is a key element in
become defined. Although this approach may appear
prokaryote systematics. Although various new methodo-
novel, with much emphasis currently being placed on the
logies have been developed over the past 100 years, both
species, the advent of 16S rRNA gene sequencing forces us
the newer methodologies and those considered to be
to choose between primers that are specific for members of
‘traditional’ remain key elements in determining whether a
the Archaea or for members of the Bacteria, so the first step
strain belongs to a known taxon or constitutes a novel one.
in that direction is already routine in many laboratories.
In the case of a known taxon, a selected set of tests may be
However, such a classification system is only possible if
used to determine whether a strain has been identified as a
strains are comprehensively and properly characterized. A
member of an existing taxon. However, in the case of a further key element is the way in which datasets are
strain or set of strains shown to be novel taxa, they compared and it is here too that some degree of guidance
should be characterized as comprehensively as possible. and a discussion of the potential problems needs to be
The goal of this characterization is to place them within provided. In the case of species, various recommendations
the hierarchical framework laid down by the have been made with respect to the ways in which species
Bacteriological Code (1990 revision) (Lapage et al., may be delineated and it is important to consider these
1992), as well as to provide a description of the taxa. aspects when considering how new strains are to be placed
Strains should be allocated to species (and/or subspecies), in novel species. However, far too little attention has been
but the nature of the ‘species name’ (a binomial or paid to the way in which taxa above the rank of species
combination) dictates that it must also be assigned to a should be characterized and modern prokaryote taxonomy
genus. The genus may be either an existing or a novel would benefit from paying greater attention to the ranks
genus. The Bacteriological Code also recommends that the above those of species.
Abbreviations: AFLP, amplified fragment-length polymorphism; DDH, The purpose of the present article is to deal with current
DNA–DNA hybridization; LPS, lipopolysaccharides; MLSA, multilocus methodologies and to outline how these methods should
sequence analysis; MLST, multilocus sequence typing. be best used and implemented. It does not set out to guide
016949 G 2010 IUMS Printed in Great Britain 249B. J. Tindall and others
the reader as to how the results should be interpreted, published. Data may be supplied by colleagues who do not
although there are some aspects that are not widely wish to be co-authors, but have given their consent for the
appreciated, where an indication of the value of the dataset results to be published, or they may be supplied as part of a
may be helpful. The article is divided into genetic scientific service. In both cases, the source of the data
(including protein sequence-based methods) and pheno- should be given clearly and unambiguously in order to
typic methods. The latter include aspects such as cell shape, make it obvious that the data were not collected by the
size, physiological and biochemical tests, as well as methods authors. Methods must be given.
of chemical analysis of the cell.
The importance of types
Information on the strains being studied
It should be remembered that in prokaryote taxonomy
In a publication, the information presented on the the inclusion of types is of central importance. Although
strains being studied should be as complete as possible. not laid down by the Code, since the Code deals with
It should include the location from which the strains matters of nomenclature and not matters of taxonomy, it is
were isolated (geographical references may also include taxonomic common sense to include all type strains that
GPS or latitude/longitude data where possible), the are relevant to a study. Where members of different genera
strain designations (including any culture collection are included and not all species belonging to those genera
accession numbers) and any information that relates to can be taken into consideration, the inclusion of the type
the environment from which the strains were isolated species of the genera under study is of utmost importance.
(e.g. pH, salinity, temperature, chemical composition of Naturally, the type strain of that type species must be used.
the environment, depth of the water column or soil It must be emphasized that the type species of the genus is
profile). However, care should be taken in using such data the most important reference organism to which a novel
to formulate conclusions regarding the ecological signifi- species has to be compared if it is considered to be a
cance of the novel strain in the environment without member of the same genus. Other species of the same
further studies (role of the organism in the environment, genus may be misclassified and may be subject to
cell counts, etc.). Note that it is unacceptable to cite a reclassification in the future. Similarly, a species placed in
culture collection number if that strain has not been a new genus must be compared with species of closely
obtained directly from a collection (e.g. DSM 1234 vs related taxa, which must include the type species of the
derived from strain DSM 1234 and supplied by x). This genera under study. In the case of comparisons across
information may also be required by culture collections or families, the type genera must be included, and by
databases such as GenBank, DDJB or EBI/EMBL and it is extension, the type strains of the type species of the type
the responsibility of the author(s) to make sure that all the genera. Many papers published in the IJSEM seriously
information is consistent in order to avoid the accumula- violate this elementary taxonomic principle. It should be
tion of errors. Where reference is made to a strain as a borne in mind that the types of certain taxa may be
type strain, this should be clearly indicated in the organisms that are pathogenic to humans, animals or
publication, database entry (GenBank/DDBJ /strain5 plants. Not all researchers have access to the facilities or
‘,strain_id.’ /note5‘,type strain of. ,Genus. ,spe- permits for handling such organisms. This should be taken
cies.’ or EMBL /strain5‘type strain: ,strain_id.’), or into consideration before research is begun and also by the
the culture collection accession form. The current policy of editors of manuscripts that would normally require
GenBank, DDBJ and EBI/EMBL is not to accept new comparative studies with such organisms.
taxonomic names until they have appeared in print. In order
for the database staff to update the names, it is important
that the strain can be accurately identified by using the
GENETIC-BASED METHODS
associated information submitted to the database. This
information will also enable the entry to be linked to the Modern prokaryote taxonomy has been strongly influenced
relevant publication. by developments in genetic methods. The elucidation of
the structure of DNA and the deciphering of the genetic
code were major steps forward in biology. The long-term
Source of the data in the publication goal of researchers in the 1950s was to be able to gain rapid
The data presented in a manuscript may be derived from a access to the genome sequence data, a goal that was realized
number of different sources. Original data should be based by the mid 1990s. Even prior to the elucidation of whole
on results collected using the methods described in the text. genome sequences, the widely differing DNA G+C values
So that others may repeat the experimental work described within the prokaryotes were recognized as having a direct
in a paper, the methods used should be given clearly or a link to codon usage (De Ley, 1968), a topic that is
reference should made to another publication in which the becoming more transparent as more genomes are
methods are fully described. When data are extracted from sequenced. The development of nucleic acid hybridization
the literature, clear and unambiguous references should be methods (DNA–DNA and DNA–RNA) has allowed the
made to the publication(s) in which the data were first indirect comparison of gene sequences. The introduction
250 International Journal of Systematic and Evolutionary Microbiology 60Characterization of strains for taxonomic purposes
of the analysis of the 16S rRNA gene by cataloguing (Fox et CLUSTAL_X, CLUSTAL W, CLUSTAL X2, CLUSTAL W2,
al., 1977), reverse transcriptase-sequencing (Sanger et al., MEGA, T-COFFEE, MUSCLE), followed by manual editing.
1977; Lane et al., 1988) and finally PCR-based gene Evaluate the alignment of new and reference sequences
sequencing (Saiki et al., 1988) has provided a useful with respect to primary and secondary structure
working hypothesis on which other elements may be consensus. The alignment MUST be made available to
compared when investigating the taxonomy and evolution the editors and reviewers at the time of manuscript
of prokaryotes. It is realistic to assume that the recognition submission – this should be extended to the deposition
of novel taxa often centres on the use of 16S rRNA gene- of the alignments as supplementary data as outlined
based techniques. Despite the widespread use of 16S rRNA below. Examples of programs that can display secondary
gene sequencing, there are a number of points that need to structure for sequence editing are ARB and PHYDIT
be considered when evaluating the data. 16S rRNA gene (http://plaza.snu.ac.kr/~jchun/phydit/), which has been
sequences are one of the most widely used datasets. discontinued in favour of jPHYDIT (http://plaza.
Although there is justification for using 23S rRNA gene snu.ac.kr/~jchun/jphydit/index.php; Jeon et al., 2005).
sequences, this dataset is currently much smaller and the RnaViz (De Rijk et al., 2003; http://rnaviz.sourceforge.
16S rRNA gene sequence presently remains the gene net/) was also developed specifically to display RNA
sequence of choice. As more whole genome sequences secondary structure. The Gutell Lab. at the University of
become available, a greater selection of genes with different Austin Texas maintains the Comparative RNA website
degrees of resolution will become available. and project (http://www.rna.ccbb.utexas.edu/) and this
resource is an excellent source of reference secondary
Recommendations for sequence analysis of the RNA structures.
16S rRNA gene Pairwise nucleotide sequence similarity values should
be calculated with caution. The following should be
Primary data quality considered:
Use (almost) complete sequences.
Check the quality of new and reference sequences: (i) the way in which the similarity was calculated
ambiguities, primary and secondary structure con- should be given in detail. The following
sensus violations, overlay of potential cistron programs are recommended for similarity
heterogeneities (direct PCR fragment sequencing). calculations: ARB, PHYDIT and jPHYDIT.
The quality of the sequences should also be checked Programs such as CLUSTAL or PHYLIP give the
against a set of properly aligned sequences (see dissimilarity values. EzTaxon (www.eztaxon.
below). The quality should be checked before org) provides a web-based tool.
sequences are deposited in databases, used in (ii) pairwise similarity values obtained from local
publications or sent to culture collections along alignment programs, such as BLAST and FASTA,
with type strain deposits. should not be used. These programs are
Remember that the sequence databases are full of primarily useful for database searches.
(iii) corrected evolutionary distance (e.g. Jukes and
incorrectly labelled and poor quality sequences.
Cantor model) should not be used for pairwise
There is NO justification for using a sequence of
similarity calculations.
poor quality/dubious origin simply because it is in
(iv) full-length sequences should always be used.
the database. When characterizing new taxa, a
taxonomist should use the best quality data Assignment to defined taxa
available, including resequencing if appropriate. Overall sequence similarity values (ranges) might be
sufficient if comprehensive high quality reference
Multiple alignment
datasets are available. There is extensive documented
The use of expert-maintained seed alignments evidence that two strains sharing less than 97 % 16S
comprising only high quality sequence data is highly rRNA gene sequence similarity are not members of
recommended, e.g. ARB (www.arb-home.de), RDP the same species (Amann et al., 1992; Collins et al.,
(http://rdp.cme.msu.edu/), SILVA (www.arb-sil- 1991; Fox et al., 1992; Martı́nez-Murcia & Collins,
va.de) and LTP (www.arb-silva.de/projects/living- 1990; Martı́nez-Murcia et al., 1992). There have been
tree/). The European rRNA database at the suggestions that this value should be set higher;
Department of Plant Systems Biology, University of however, it should be remembered that the values
Gent, is no longer updated (as of February 2007) but should be based on almost full-length, high quality
a link is provided to the SILVA website. A limitation sequences. 16S rRNA gene sequences alone do not
is that seed alignments may not be universally describe a species, but may provide the first
compatible with some of the programs used by indication that a novel species has been isolated
authors of articles in the IJSEM. Alternatively, high (less than 97 % gene sequence similarity). Where
quality sequences that were not previously aligned 16S rRNA gene sequence similarity values are more
can be obtained from public databases and aligned than 97 % (over full pairwise comparisons), other
using robust multiple alignment programs (e.g. methods such as DNA–DNA hybridization or
http://ijs.sgmjournals.org 251B. J. Tindall and others
analysis of gene sequences with a greater resolution sequences should be removed or the organism in
must be used. These methods must also be question should be resequenced.
correlated with the characterization based on Establish sequence conservation profiles for the
phenotypic tests. While the resolving power of the group of interest and higher ranks (Ludwig &
16S rRNA gene with respect to the delineation of Klenk, 2001; Peplies et al., 2008).
novel species has been extensively debated, less Use these profiles as filters to recognize branch
attention has been paid to other taxonomic ranks, attraction effects probably resulting from plesio-
such as the genus. 16S rRNA gene sequence morphic sites. Alignment columns should be succes-
similarities between strains belonging to the same sively removed according to positional variability and
genus may vary dramatically from ~97 % in members the resulting tree topologies compared. Applying no
of the family Enterobacteriaceae and some members of and 50 % filters has proved to be a reasonable
the Actinomycetes, etc., to much lower levels, for compromise. The filters MUST be clearly and
example, in the genera Deinococcus and Hymenobacter. unambiguously identified otherwise the work is
This variation may be in part due to either an NOT reproducible. When taxonomic interpretation
overinflation of the number of genera (high similarity is based on filters/masks, these MUST be identified in
values) or a failure to recognize the presence of the alignments that accompany the paper.
additional taxa (low similarity values). Apply alternative treeing methods (distance matrix,
At values above ~95 % 16S rRNA gene sequence maximum-parsimony, maximum-likelihood meth-
similarity (over full pairwise comparisons), taxa ods). The latter two are to be preferred; distance
should be tested by other methods to establish matrix methods should be used for raw screening
whether separate genera are present. In BOTH only (Ludwig & Klenk, 2001; Peplies et al., 2008).
cases, the establishment of novel species or new The use of different methods of evaluation using the
genera (irrespective of the degree of sequence same dataset does not identify effects such as gene
similarity) should be clearly and unambiguously transfer or convergent or parallel evolution.
documented. In the case of species, authors should Credibility should be given to other datasets
document differences between the novel species and (alternative genes and non-sequence based meth-
existing species within the genus. If the genus is too ods) that have been shown to reflect evolution.
heterogeneous (e.g. members of the genera Bacillus or As many high quality sequence data as possible should
Clostridium), then the authors should provide a be included and balanced datasets (with respect to the
reasonable scientific justification for using only a phylogenetic spectrum) should be preferred to reduce
limited set of species from the genus. However, the bias resulting from branch attraction effects (Ludwig
type species of the genus in question must be & Klenk, 2001; Peplies et al., 2008). New maximum-
included. In the case of genera, reasonable attempts likelihood software tools are available (PHYML, RAXML),
should be made to differentiate the new genus from which allow the inclusion of a reasonable number of
other genera that are ‘closely related’. Many genus sequences.
descriptions published in recent years do not Remember that all methods are based on underlying
distinguish the new genus from existing genera, assumptions. If the dataset violates those assumptions
violating an elementary principle in taxonomy. A (which is often difficult to test), then the tree
95 % gene sequence similarity ‘lower cut-off window’ produced may well be the most parsimonious or
for genera seems reasonable, but in practice it may not most likely, but it is not necessarily the ‘correct’
be easy to implement. A key issue is often the answer in terms of the true evolution.
interpretation of chemotaxonomic data and few Never use sequences from single distantly related
scientists now have extensive experience of this organisms as (an) outgroup(s) (branch attraction!).
procedure. A recent article (Yarza et al., 2008) has The group of interest is the ingroup, the remaining
also indicated that a ‘lower cut-off window’ of 95 % taxa included in the tree serve as outgroups for that
16S rRNA gene sequence similarity may be reasonable, ingroup.
but it must be emphasized that this was based on the Always include sequences of the appropriate types –
evaluation of taxa that are delineated by a broad type strains for species, the type strain of the type
spectrum of methods and not on 16S rRNA gene species for genera, etc. (this applies to ALL datasets!).
sequences alone.
Presentation of trees
Graphical representation – phylogenetic trees If properly described (via the methods section and in the
Use (almost) complete high quality sequences. legends to the figures), it is acceptable to visualize only
Use high quality alignments. the part(s) of trees based on comprehensive datasets (as
Do not mix full and partial sequences. recommended) that is (are) of importance in the
Never truncate an alignment to the region covered context of the taxonomic assignment or description.
by partial sequences (due to a reduction in the If properly described (via the methods section and in
information content) – instead the truncated the legends to the figures), it is acceptable to show
252 International Journal of Systematic and Evolutionary Microbiology 60Characterization of strains for taxonomic purposes
consensus trees visualizing local support (resolved an extensive collection of literature on DNA–RNA
topology) or discrepancies (multifurcation, shading, hybridization, this method was eventually replaced by
circles) that have been constructed by applying 16S rRNA gene sequencing. Those consulting the older
different treeing methods and parameters. literature should remember that there is generally a good
Alignments must be made available as supplementary correlation between the results obtained in DNA–RNA
material. experiments and 16S rRNA gene sequencing.
Trees could be included as supplementary material.
Recommendations for DNA–DNA hybridization (reasso-
Details of the settings used for some methods, such a ciation) experiments. DNA–DNA hybridization (DDH) is
maximum-parsimony or maximum-likelihood ana- to be performed in cases where the new taxon contains
lysis, are rarely given – one can only assume that the more than a single strain, in order to show that all
‘default settings’ have been used. The settings used in members of the taxon have a high degree of hybridiza-
all tree-construction methods should be clearly tion among each other. DDH is necessary when strains
indicated. share more than 97 % 16S rRNA gene sequence
Only bootstrap proportions of 70 or higher should be similarity. If the new taxon shows this high degree of
included in the dendrogram. similarity to more than one species, DDH should be
performed with all relevant type strains to ensure that there
Other methods of representing the results of
is sufficient dissimilarity to support the classification of the
sequence analyses
strain(s) as a new taxon. DDH can be performed using a
The graphical representation (trees) of sequence data
number of techniques. Most of them have been validated
is currently a two-dimensional representation.
and show comparable results (Grimont et al., 1980;
Heat maps have also been used to some effect in the
Rosselló-Móra, 2006).
analysis of large datasets and are a further devel-
opment of the shaded similarity matrices that have The method (with all modifications) must be clearly
been commonly used in numerical taxonomy in the cited. If a novel method is used, the authors must
past (Lilburn & Garrity, 2004; Sneath & Sokal, 1973). provide evidence that the new method produces
Three or multi-dimensional representations of the data comparable results to the established methods.
may be preferable to two-dimensional representations. DDH data must be provided for the type strain of
the proposed novel species with all other strains of
Other genes the proposed novel species, and for the new type
As indicated above, there is growing interest in the use strain with the type strain(s) of the closest related
of other genes with a greater degree of resolution species with validly published names. At least one
(protein-encoding genes) to resolve issues that are not reciprocal value for DDH of the closest related
solved by 16S rRNA gene sequencing. species to the new type strain must be provided.
Typically multiple genes are used and the sequences Note that when applying spectrophotometric meth-
are either compared as individual datasets or ods carried out in solution, no reciprocal values can
combined in concatenated sequences. In such cases, be obtained as none of the DNAs are labelled (De
the criteria laid down for the compilation of reliable Ley et al., 1970). Standard deviations of at least
datasets and expertly maintained alignments are three analyses must be given.
important. DDH is optional for new taxa encompassing a single
strain that shares less than 97 % 16S rRNA gene
Sequences of other conserved protein-encoding genes, sequence similarity to the closest neighbour.
typically housekeeping genes, can provide a higher resolu- In prokaryote systematics, DDH methods have largely
tion than 16S rRNA gene sequences and can complement concentrated on the use of relative binding ratio
DNA–DNA relatedness or 16S rRNA gene sequence data for (RBR) methods. The results should be given as
taxonomic analysis at the species level. It can be expected percentage binding or percentage hybridization. It
that multilocus sequence analysis (MLSA) and multilocus has been recommended that only DNA from strains
sequence typing (MLST) schemes will become available sharing a difference in the melting temperature (DTm)
through publicly accessible databases. However, it is of 5 uC or less should be included. Methods using the
essential that the species delineations based on MLSA difference in melting point of the heterologous hybrid
schemes are corroborated with DNA–DNA hybridization compared with the homologous hybrid (DTm) have
data, thereby also validating the MLSA scheme itself. rarely been used in studies with prokaryotes, but have
been widely used in zoology.
Nucleic acid hybridization methods
Hybridizations are commonly performed at
The technique of DNA–DNA or DNA–RNA hybridization Tm230 uC and/or Tm215 uC (stringent conditions).
was introduced into prokaryote systematics from the 1960s Tm is calculated on the basis of the known DNA G+C
onwards (Brenner et al., 1967; De Ley et al., 1966; Johnson content of the strains to be hybridized. The temper-
& Ordal, 1968; McCarthy & Bolton, 1963; Pace & ature at which the hybridization is carried out must be
Campbell, 1971; Palleroni et al., 1973). Although there is clearly stated.
http://ijs.sgmjournals.org 253B. J. Tindall and others
A DDH value equal to or higher than 70 % has been able to substitute for DDH. ANI has been demon-
recommended as a suitable threshold for the defini- strated to correlate with DDH, where the range of ~
tion of members of a species (Brenner, 1973; Johnson, 95–96 % similarity may reflect the current boundary
1973; 1984; Wayne et al., 1987), but this value should of 70 % DDH similarity (Goris et al., 2007). ANI may
not be used as a strict species boundary (Ursing et al., substitute for DDH analyses in the near future.
1995). A single species must embrace all the strains Gene content
that cannot be clearly discriminated by a stable The rate at which this area can develop is dependent on
phenotypic property. A single species may embrace the number of sequenced genomes that are available.
groups of strains with DNA–DNA hybridization Such studies rely on the distribution of particular genes
values of less than 50 % that are indistinguishable by and also on good annotation of the sequenced genomes
means of other properties tested at the time. (Snel et al., 1999; Huson & Steel, 2004).
A single species may contain several genomic groups Where representatives of only a small number of taxa
or genomovars (Ursing et al., 1995) that may be have been studied, the coverage may be insufficient to
reclassified as novel species once clear and stable
delineate taxa adequately. In the case of taxa where a
discriminative phenotypic properties are found.
larger number of strains have been sequenced, more
DNA G+C content detailed analyses are already possible.
In the apparent absence of ways of defining or
DNA G+C content is still a useful parameter and its differentiating higher taxa, it is often not clear
relationship to codon usage is clearly illustrated in genome whether a particular taxon may be over differentiated
analysis. It is also an important prerequisite for determin- (e.g. members of the family Enterobacteriaceae) or
ing the conditions used in DNA–DNA hybridizations. under differentiated (e.g. members of the genus
While DNA G+C content may be fairly constant in a Deinococcus). Pre-conceived concepts and nomencla-
group, there are notable exceptions, particularly for tures can easily influence the interpretation of data.
obligate intracellular parasites. Deviations from the values
obtained for other members of the group may also be an Multiple (gene) aligned sequence datasets
indication of problems with the strain being studied. The The comparison of large datasets of aligned sequences
methods of choice are now HPLC-based (Ko et al., 1977; is usually based on the concatenation of the dataset. In
Tamaoka & Komagata 1984; Mesbah et al., 1989). such cases, there appears to be some preference for the
Although thermal stability of the native DNA and caesium selection of genes that strengthen a particular
chloride density-gradient centrifugation are alternative viewpoint. When comparing genomes across all
methods, these are now largely of historical interest (see members of the Bacteria or between the Bacteria and
Johnson & Whitman, 2007; De Ley, 1970). the Archaea, the major problem appears to be the
limited dataset since there are only a small number of
Use of whole genome sequences genes that appear to be universally distributed (Dagan
The drop in the price of sequencing whole genomes, & Martin, 2006).
together with the technical advances that have been made Different algorithms or the selection of a smaller
suggest that routine sequencing of prokaryote genomes will subset of genes may give different results (Wolf et al.,
be realistic in the near future. Further advances need to be 2001).
made in the annotation of the genome. The pilot phase As in the case of 16S rRNA gene sequencing, particular
study of the Genomic Encyclopedia of Bacteria and attention must be given to accurately reporting the
Archaea project (GEBA; http://www.jgi.doe.gov/pro- quality of the alignment, to the annotation of the
grams/GEBA/index.html) is currently examining the feas- genome and to any form of secondary restrictions (use
ibility of sequencing all available type strains. The current of masking/filters) used in the dataset.
plan is for the information to be published as short genome
There is increasing interest in the use of such multilocus
reports in the open access journal Standards in Genomics
sequencing methods to define species. However, it is clear
Sciences (http://standardsingenomics.org). A key issue that
that estimating the influence of recombination within a gene
remains is the reliable annotation of all genes in a genome
pool is not an easy task. In some cases strains appear to have
since identifying gene homologies (preferably orthologues)
clonal origin, whereas others appear to be almost freely
is of central importance in taxonomy. In principle, there
combining. The development of the concept of the ecotype
are three basic approaches:
indirectly puts emphasis on the process of ecological
Genome indexes selection on the organism, where it is the expressed
There are several indexes that are obtained by phenotype (at the level of the organism) rather than the
comparing pairwise genomes that could be used in gene that is exposed to direct selection (Cohan, 2002).
taxonomy. Noteworthy are the Average Nucleotide
Nucleic acid fingerprinting
Identity (ANI; Konstantinidis et al., 2006) and
Maximal Unique Matches (MUM; Deloger et al., In general, these methods provide information at the
2009) indexes as they have been hypothesized to be subspecies and/or strain level. Examples for these
254 International Journal of Systematic and Evolutionary Microbiology 60Characterization of strains for taxonomic purposes
techniques are: amplified fragment-length polymorphism relevant traits should be listed in the protologue of each
PCR (AFLP), macrorestriction analysis after pulsed field gel taxon being described. At the rank of species and
electrophoresis (PFGE), random amplified polymorphic subspecies, more than one representative strain should
DNA (RAPD) analysis, rep-PCR (repetitive element be included in order to determine which characteristics
primed PCR, directed to naturally occurring, highly are stable and which are variable. The inclusion of
conserved, repetitive DNA sequences, present in multiple suitable positive and negative controls should be empha-
copies in the genomes) including REP-PCR (repetitive sized, particularly where test conditions differ from those
extragenic palindromic-PCR), ERIC-PCR (enterobacterial given in standard reference works. The issue of the use of
repetitive intergenic consensus sequences-PCR), BOX-PCR more than a single strain is raised at regular intervals
(derived from the boxA element), (GTG)5-PCR and (Christensen et al., 2001; Felis & Dellaglio, 2007). The
ribotyping. A major disadvantage of some of these number of strains to be used is not (and cannot be) laid
fingerprint-based methods is that the results are often very down by the Bacteriological Code (since it deals with
difficult to compare when they have been obtained in nomenclature and not taxonomy or the characterization of
different laboratories due to the lack of standardization. strains/taxa), but this does not diminish the importance of
Exceptions include AFLP and ribotyping since these using more than a single strain. Morphological, physio-
approaches have been adequately standardized. DNA logical and biochemical traits should be carefully evaluated
fingerprinting methods are of limited value for species to determine those that are common (or even unique) to
descriptions, but when used properly they can be the taxon in question. Properties that are variable may be
valuable for identification at the species and subspecies included in the protologue since they may indicate
levels. These typing techniques can however be very subgroupings within the taxon in question.
useful to demonstrate whether or not isolates of a novel
taxon are members of a clone.
Morphological criteria
Cell shape and size. The diversity in cell form and its
PHENOTYPIC CHARACTERIZATION underlying structural basis are now becoming
The phenotype of an organism is typically taken to be apparent, largely due to a combination of improved
restricted to parameters such as cell shape, colony methods of visualizing prokaryote cells, structural
morphology, biochemical tests, pH and temperature studies at the subcellular and molecular levels, and
optima, etc. However, such a definition is too limited or genome sequencing. Although light microscopy may
simplistic and does not reflect the true scope of be adequate for describing coarse features of the cell,
phenotypic characteristics that can now be easily exam- electron microscopic images have a higher resolu-
ined. Typically the chemical composition of the cell (fatty tion. Depending on the organisms concerned, scanning
acid, polar lipid and respiratory lipoquinone composition, electron micrographs may be used to show the
amino acid composition of the peptidoglycan of the cell morphology of whole cells, while transmission electron
wall of Gram-positive bacteria, presence and size of micrographs may be used to determine the infrastruc-
mycolic acids, polyamine pattern, etc.) is included under a ture of the cell envelope or the presence of cytoplasmic
separate heading, chemotaxonomy, but it is in essence a inclusions, internal membrane structures, etc.
part of the phenotypic characterization of an organism. A Cell shape and form should be adequately described
recently published collection of phenotypic methods has and supported by appropriate photographs (including
incorporated references to standard chemotaxonomic electron micrographs). Photographs must be of good
methods under the category of phenotypic characteriza- quality and features discussed in the text must be
tion (Tindall et al., 2008). This publication includes, and is evident in the photograph(s). In some cases, cells are
an extension of, the work by Smibert & Krieg (1994). This simply rods with a uniform size and evenly rounded
view is adopted here and expands on principles outlined by ends or cocci with a regular spherical shape. In other
two ad hoc committees of the ICSB (Wayne et al., 1987; cases, the ends of rods may have characteristic forms
Murray et al., 1990). Other references to methods useful for or cocci that have just divided may have relatively flat
the phenotypic characterization of prokaryotes include adjoining surfaces. Spirilla and spirochaetes have
Bascomb (1987), Blazevic & Ederer (1975), Holdeman et different forms, with different amplitudes and wave-
al. (1977) and Barrow & Feltham (2004). This list is not length. In extreme cases, cells may form spirals and
complete and reference to the appropriate published their diameter may be characteristic. Vibrios may also
minimal standards (a list of which is given later) should be of different lengths and have different amplitudes.
be made where they are available. Light microscope photographs of cells are best
prepared by covering microscope slides with a thin
layer of agar. Such slides may be used either while still
Morphology, physiology and biochemistry
wet or air-dried and stored for later analysis. Wet
Examinations of morphological, physiological and bio- mounts should be examined for areas where cells are
chemical properties are the oldest tools for the well contrasted and form a single layer between the agar
characterization and classification of prokaryotes. All surface and the coverslip (Pfennig & Wagener, 1986).
http://ijs.sgmjournals.org 255B. J. Tindall and others
Special attention should be given to characteristic distinguishes between cell-wall structures that allow
features of the cell, such as apolar growth, the a dye complex to be washed out of the cell and those
presence of stalks, or the fact that they undergo a life that retain it. The structural basis of this reaction
cycle (Caulobacter spp., Pirellula spp., etc.). The has been described using electron microscopy
formation of prosthecae, budding or branching are also studies (Beveridge & Davies, 1983; Davies et al.,
important characteristics, as is the formation of cell 1983). It should be noted that certain organisms
aggregates (e.g. diplococci, sarcinas, chains of cells). with a defective Gram-positive cell-wall structure
Spore formation, endospores or exospores. In the may also stain Gram-negative (Wiegel, 1981) and
case of endospore-forming organisms, the location that the reaction to the Gram stain may alter as the
of the spores and their size in relation to the size of cells age. However, such properties are not ran-
the vegetative cell should be noted. In the case of domly distributed, but are found in restricted
exospores, features such as the formation of sporangia groups of taxa.
and the shape of the spores should be noted. In certain groups, such as members of the class
The location of flagella should be noted with care: Halobacteria, Gram staining may only work after the
polar, subpolar or laterally inserted. Their occurrence fixation of the cells. The stability of the cells at low salt
in one location (tufts of flagella) or distributed at concentrations or in the absence of salt may be a more
different locations should be noted. Special care important feature than the Gram stain.
should be taken to check whether strains are motile Acid-fast staining has been used in the past for
by standard tests or under the light microscope before organisms that typically contain mycolic acids. A
the strain is examined in an electron microscope since detailed examination of the nature of the mycolic
flagella may be lost during fixation. acids should be undertaken in such cases.
Motility should be described accurately since both Lipophilic cellular inclusions (e.g. polyhydroxybutyric
the form and speed may be of significance. Some acid) may be stained with Sudan Black.
organisms may be rapidly motile, others only slowly. Extracellular polymers may be visualized by suspend-
A distinction should be made between motility caused ing cells in a solution of India Ink and observing wet
by flagella and gliding or other forms of motility. mounts under the microscope. Bright haloes around
Intracellular structures visible under the light micro- the cells indicate the presence of extracellular material,
scope such as gas vacuoles, sulphur granules or such as polysaccharides.
polyhydroxybutyrate granules should be investigated
using appropriate methods. Recommendations for phenotypic analyses
When grown on solid surfaces, the shape and size of
the colony, together with other features of the colony The methods (with all modifications) must be clearly
given. If novel methods are used, the authors must
form, should be accurately noted. Standard textbooks
provide evidence that the new method produces
cover the diversity of colony shape and size, including
comparable results to established methods.
features such as the nature of the outer edge, whether
Physiological and biochemical tests should be
the colony is opaque or translucent and whether the
carried out in test media and under conditions that
colony has any visible additional structures/features.
are identical or at least comparable. When different
Some strains form rapidly spreading colonies, while
methods are used, authors must provide evidence
others are fairly compact.
that they give comparable results.
In cases where colonies are not formed, it is still
Scientists are encouraged to use tests that may not be
important to record whether cell material is pigmen-
covered by API (bioMérieux) tests or Biolog (Biolog
ted. Cellular pigments may be soluble either in water
Inc.) test plates. Such tests may be documented in the
or organic solvents. Typical pigments include carote-
original literature or in minimal standards. For certain
noids (which turn blue in the presence of concen-
groups of organisms, the degradation of polymers is
trated sulphuric acid), flexirubins (which change an important characteristic and should be tested by
colour reversibly under acid and alkaline conditions), appropriate methods. The use of multi-point inocu-
(bacterio)chlorophyll [which is soluble in organic lators and multi-well plates can be helpful for
solvents, but becomes water soluble following saponi- handling large numbers of strains.
fication; treatment with acid results in the formation It is important to differentiate between methods that
of (bacterio)phaeophytin], melanin and pyocyanin require the demonstration of enzyme activity
(fluorescence at 360 nm, reversible colour change at (Bascomb, 1987; Blazevic & Ederer, 1975), growth
acid/alkaline pH); this list is not exhaustive. on the substrate or metabolic activity without
associated growth (Bochner, 1989).
As a general recommendation, authors should
Staining behaviour of the cell
include strains of the most closely related taxa for
The Gram stain (Gram, 1884) is one of the oldest comparison in their phenotypic analysis rather than
forms of staining the cells of prokaryotes and using the data from previously published work. The
256 International Journal of Systematic and Evolutionary Microbiology 60Characterization of strains for taxonomic purposes
comparisons must include the type strain of the and members of the Erysipelothrix/Holdemania group. All
type species of the appropriate genera. Exceptions other murein-containing bacteria so far analysed exhibit
may be made where laboratories have collected the the A-type peptidoglycan. The amino acid composition of
data used in previous publications themselves using the peptide side chain, including the characteristic diamino
identical methods. acid is usually common to all species of a genus. However,
Accurate citation of the source of the methods is a higher degree of variability has been detected in the mode
important, particularly where standard reference of cross-linkage between the peptide side chains, which
works may give more than one method. Authors often differ among species of certain genera (e.g. members
must clearly attribute the tests to the relevant of the genus Microbacterium), but may also differ between
reference work in which the test is described in detail. strains of a single species as reported for Micrococcus luteus.
In cases where more than one method for performing Analysis of the peptidoglycan structure is a requirement for
a particular test is given in a reference work, the all members of novel Gram-positive genera when they are
authors must indicate which of the methods was used, described and at least the amino acid composition should
e.g. lipase activity was determined using method 2 be provided for every novel Gram-positive species
listed in Tindall et al. (2007). described. In the majority of cases, the amino acid
All references to methods used should give the composition of the peptide side chain of the type species
source in which the method is described in detail, of a novel genus may be shared by future species assigned
allowing others to verify the method and the results to the genus and hence it should be listed in the genus
obtained. In many instances, reference is made to a description as a characteristic trait. The complete
work that itself also refers to another paper where composition of the peptidoglycan of a novel species of a
the method may, or may not, be described in detail. recognized genus should be in agreement with the
This can make it difficult to locate the original characteristics of the genus and may provide differences
publication and description of the method. in the interpeptide bridge that are useful for differentiation
of the novel species from other species. A list of peptido-
glycan variations can be found at http://www.dsmz.de/
Chemical characterization microorganisms/main.php?content_id=35. This system is
Chemical characterization of the cell (traditionally referred slightly different to that developed by Schleifer & Kandler
to as chemotaxonomy) deals with various structural (1972).
elements of the cell including the outer cell layers Pseudomurein, a characteristic peptidoglycan, has been
(peptidoglycan, techoic acids, mycolic acids, etc.), the cell detected in some Gram-positive staining members of the
membrane(s) (fatty acids, polar lipids, respiratory lipoquin- Archaea, in which N-acetyl muramic acid is replaced by N-
ones, pigments, etc.) or constituents of the cytoplasm acetyl talosuronic acid (König et al., 1982; 1983).
(polyamines).
Respiratory lipoquinones. Respiratory lipoquinones are
As a general recommendation, authors have to study the
widely distributed in both anaerobic and aerobic organisms
chemotaxonomic features of the most closely related taxa
within the Bacteria and Archaea. These may be divided into
for comparison, especially where novel genera are being
two basic structural classes, naphthoquinones and
proposed. The discriminatory power of the different
benzoquinones. A third class includes the benzothiophene
methods may vary (see specific comments below).
derivatives, such as Sulfolobus quinone and Caldariella
quinone, but data available to date indicate that this class is
Peptidoglycan. The discriminatory power of the structure
restricted to members of the order Sulfolobales (Tindall, 2005).
of peptidoglycan is apparently restricted to Gram-positive
bacteria, whereas no variation has been reported among
members of the phylum Proteobacteria and phylum Benzoquinones
Bacteroidetes (Schleifer & Kandler, 1972). Analyses of the Benzoquinones include ubiquinones, rhodoquinones
peptidoglycan structure can be performed at different and plastoquinones.
levels. The simplest analysis is the determination of the Ubiquinones appear to be restricted to members of
characteristic diamino acid in the cross-linking peptide. the classes Alphaproteobacteria, Gammaproteo-
Analysis of the peptidoglycan type (A type: cross-linkage of bacteria and Betaproteobacteria (Tindall, 2005).
the two peptide side chains via amino acid three of one Reports to the contrary should be treated with
peptide subunit to amino acid four of the other peptide caution. Within members of the class Alphapro-
subunit; B type: cross-linkage of the two peptide side teobacteria, the predominant ubiquinone is Q-10,
chains via amino acid two of the one peptide subunit to although some taxa are known with Q-9 or Q-11.
amino acid four of the other peptide subunit), mode of Members of the class Betaproteobacteria typically have
cross-linkage (direct or interpeptide bridge and amino Q-8, while there is a greater degree of variation within
acids in the bridge) and complete amino acid composition the members of the class Gammaproteobacteria.
provide more detailed information. B type peptidoglycan is Quinones Q-7 to Q-14 may be found in different
characteristic of all genera of the family Microbacteriaceae taxa within the members of the class
http://ijs.sgmjournals.org 257B. J. Tindall and others
Gammaproteobacteria, although there appears to be determined by 16S rRNA gene sequence analysis
no evidence of organisms producing only Q-10 as the (Tindall, 2005).
major ubiquinone in members of this class. Members It would appear that there are no differences in the
of the genus Legionella typically synthesize more than qualitative variation between strains within a species.
one ubiquinone, with chain lengths between Q-10 and Demethyl-, monomethyl-, dimethylmenaquinones
Q-14 (Collins and Gilbart, 1983; Karr et al., 1982). and menathioquinones also have characteristic distri-
Some members of the classes Alphaproteobacteria and bution patterns.
Gammaproteobacteria may also synthesize menaquin-
ones and rhodoquinones. Their relative concentra-
tions are usually dependent on the substrates utilized Methodological aspects
and whether a strain is using oxygen as a terminal Respiratory quinones should be extracted from cell
electron acceptor. Based on the report of Hiraishi et material taking care to avoid extremes of pH,
al. (1984), there are betaproteobacterial taxa such as exposure to strong light or highly oxidizing conditions.
Rhodocyclus tenuis and Rubrivivax gelatinosus that, in Pre-screening using TLC will identify the various
addition to the synthesis of ubiquinones, also respiratory lipoquinone classes: ubiquinones, rhodo-
synthesize menaquinone 8 (MK-8). quinones, menaquinones, Sulfolobus-/Caldariella-quin-
There is evidence for side chain modifications of the ones, menathioquinones and demethyl-, monomethyl-
ubiquinones in some methanotrophs (Collins, 1994). and dimethylmenaquinones.
Separation of simple quinone mixtures may be
undertaken using reverse phase TLC, but a properly
Naphthoguinones calibrated reverse phase HPLC column provides
The vast majority of evolutionary lineages within the greater reproducibility and allows more accurate
Bacteria and Archaea that are known to produce quantification (Collins, 1994; Tindall, 2005).
respiratory lipoquinones synthesize naphthoquinone If quinones are detected that cannot be identified,
(menaquinone) derivatives. These include menaquin- information such as UV-visible spectroscopy results
ones, demethylmenaquinones, monomethylmenaquin- and accurate reports of the behaviour of the quinones
ones, dimethylmenaquinones and menathioquinones in TLC and HPLC systems would be helpful. Full
(Collins, 1985; Collins, 1994; Tindall, 2005). structural identification can usually be achieved by a
The side chain lengths recorded to date of known combination of MS and NMR (Collins, 1994).
respiratory lipoquinones range from 5 to 15 isopre-
noid units. Hydrophobic side chains of lipids. Hydrophobic side
The isoprenoid side chains are often fully unsaturated, chains of intact lipids are typically obtained by alkaline or
but certain groups have characteristic, stereospecific acid hydrolysis of the intact lipids. A variety of side chains
patterns of hydrogenation (saturation) in the side has been recorded: members of the Bacteria contain a
chain. Typically certain members of the class diverse range of hydrophobic side chains in their lipids,
Deltaproteobacteria and members of the class whereas members of the Archaea have only isoprenoid-
Halobacteria synthesize menaquinones in which one based side chains.
point of saturation occurs at the aliphatic end of the
side chain. The isoprenoid side chains in members of Isoprenoid-based ether-linked lipids
the class Halobacteria are octaprenyl; in members of Isoprenoid ether-linked side chains are currently
the class Deltaproteobacteria they are penta- to only known to occur in members of the Archaea.
heptaprenyl (Collins, 1994). The stereochemistry of the glycerol-isoprenoid side
In the high G+C Gram-positives, the isoprenoid side chain is opposite to that found in the glycerol linked
chains show different patterns of hydrogenation. In to the hydrophobic side chains in members of the
some cases, the first position of saturation is at the Bacteria.
second isoprenoid double bond (from the naphtho- In members of the Archaea, the intact glycerol
quinone ring), followed by position three. In other isoprenoid ethers are normally analysed and are
cases, the second position of saturation is at the typically divided up into diethers, hydroxylated
aliphatic end of the chain. There is currently a single diethers, macrocyclic diethers, tetraethers, and polyol
report of a tetra-hydrogenated menaquinone in which derivatives of the tetraether. The presence of unsat-
saturation is only present at the aliphatic end (Collins, uration or penta- and hexa-cyclic rings is particularly
1994). important in the further modification of the iso-
Terminal ring structures have also been reported prenoid side chain.
(Collins, 1994). Despite reports to the contrary, there is no reliable
Fully saturated side chains have been reported in evidence that members of the Archaea have significant
certain taxa within the Archaea. amounts of ester-linked fatty acids in their lipids.
The chain length, degree of saturation (if any) and Care should be taken in choosing the methods of
position of saturation closely follows groupings as hydrolysis since some compounds (in particular
258 International Journal of Systematic and Evolutionary Microbiology 60Characterization of strains for taxonomic purposes
hydroxylated diethers) produce artefacts when hydro- a variety of polar lipids found in specific groups of
lysed under strongly acid conditions (Ekiel & Sprott, bacteria. These compounds appear to be formed by
1992; Sprott et al., 1990). On the other hand, some the condensation of a fatty acid with a molecule with a
ether-linked lipids are notoriously stable and specially small molecular mass, such as an amino acid. Known
developed methods are required to cleave all head examples include sphingolipids (Naka et al., 2003),
groups (Kumar et al., 1983; Nishihara et al., 2002). capnines (Godchaux & Leadbetter, 1984) and alkyl-
The methods used must be fully documented. amines (Anderson, 1983). Long chain diols (Pond et
Ether-linked lipids can be rapidly detected using TLC al., 1986) have also been reported to be a significant
methods (Tornabene & Langworthy, 1979). They may component in the lipids of Thermomicrobium roseum.
also be detected by HPLC (Demizu et al., 1992; Where such compounds are known to occur,
Ohtsubo et al., 1993; Koga et al., 1998), high appropriate methods of analysis should be undertaken
temperature GC (Nichols et al., 1993) or MS (de to ascertain whether they are actually present. The
Souza et al., 2009; Macalady et al., 2004). There methods used must be documented.
appear to be no universal methods applicable to all
compounds. Non-isoprenoid based-ether-linked lipids
In addition to fatty acids, there is a diverse range of
Fatty acids and their derivatives members of the Bacteria that synthesize ether-
Fatty acids, ester-linked to the glycerol are typical linked lipids. Typically they are either straight chain
constituents of almost all members of the Bacteria. or simple branched (not isoprenoid) side chains or
The Sherlock MIS system (MIDI Inc.) provides a mono-unsaturated derivatives, with the point of
comprehensive database, but this is certainly not unsaturation adjacent to the ether bond (i.e.
complete and there are some discrepancies that plasmalogens or vinyl ethers). Plasmalogens/vinyl
need to be clarified or compounds that are currently ethers are hydrolysed under acid conditions to give
not included in the database. dimethyl acetals that elute with the fatty acid methyl
When determining the fatty acid patterns of strains, esters prepared by acid catalysis during GC. Where
the cultivation conditions of the strains should be such compounds are known or suspected to occur in a
identical prior to fatty acid extraction. There may be group, efforts should be made to record their
exceptions in cases where it is impossible to grow presence/absence. There is evidence that dimethyl
the organisms under the same conditions. These acetals have been reported to occur in some organisms
must be carefully documented. where hydroxylated fatty acids elute with the same
When reporting fatty acids, all components that retention time and may have been mistakenly
constitute 1 % or more of the fatty acids must be identified as dimethyl acetals instead of the hydroxy-
reported. In cases where major peaks are not lated fatty acid (Moore et al., 1994; Helander &
identified, they will not be included in the peak- Haikara, 1995). For example, 1,1-dimethoxy-12-
naming table. In such instances, their presence must methyl-pentadecane (C15 : 0 anteiso-DMA) exhibits
be reported and the equivalent chain-length (ECL) an ECL of 20.0008 in GC analyses compared with
given. This will allow any future work on the C14 : 0 2-OH (Kämpfer et al., 2000).
elucidation of the structure to link to the ECL given Non-plasmalogen ethers in members of the Bacteria
in publications. include both mono- and diethers (Langworthy et al.,
When comparing patterns of fatty acids, it should be 1983, Rütters et al., 2001). They do not co-
remembered that they generally do not fluctuate chromatograph with mono- or diethers from mem-
significantly within a taxonomic group. Thus care bers of the Archaea, but methods applicable to the
should be taken of reports of branched chain fatty analysis of such compounds are similar to those used
acids in a group that otherwise synthesizes straight for members of the Archaea.
chain and unsaturated fatty acids. Similarly the
presence or absence of hydroxylated fatty acids is Polar lipids There is a vast diversity of polar lipids now
generally characteristic and their unexpected absence known to be present in prokaryotes and in many cases their
or presence should be treated with caution. structures have yet to be fully elucidated. There is currently
When comparing fatty acid patterns, it is important to no collective work that adequately covers all aspects of
make appropriate comparisons. The Sherlock MIS prokaryote lipids, although the work of Ratledge &
(MIDI) system extracts fatty acids from intact cells Wilkinson (1988) is a good starting point. Their
(Miller, 1982) and comparisons with results that have biosynthesis is also not fully understood. It should be
been obtained from the lipid fraction(s) extracted from emphasized that the polar lipid diversity is associated with
cells or on fatty acid methyl ester mixtures that have the cell membrane(s) and is not limited to just
previously been separated into different classes by TLC phospholipids. Given the large diversity, it is important
are not sensible and can falsify any interpretations. to document the lipids present by providing an image of
In additional to simple fatty acids, there is growing the thin layer plate stained with a reagent that will allow all
evidence for the importance of modified fatty acids in lipids to be visualized.
http://ijs.sgmjournals.org 259You can also read
