EXPERIMENTAL EVOLUTION, LOSS-OF-FUNCTION MUTATIONS, AND "THE FIRST RULE OF ADAPTIVE EVOLUTION"

Page created by Dwayne Torres
 
CONTINUE READING
Volume 85, No. 4              THE QUARTERLY REVIEW OF BIOLOGY                                   December 2010

 EXPERIMENTAL EVOLUTION, LOSS-OF-FUNCTION MUTATIONS,
      AND “THE FIRST RULE OF ADAPTIVE EVOLUTION”

                                             Michael J. Behe
      Department of Biological Sciences, Lehigh University, Bethlehem, Pennsylvania 18015 USA
                                        e-mail: mjb1@lehigh.edu

                                         keywords
            experimental evolution, adaptation, mutation, loss of function, malaria,
                                        Yersinia pestis

                                                  abstract
           Adaptive evolution can cause a species to gain, lose, or modify a function; therefore, it is of
        basic interest to determine whether any of these modes dominates the evolutionary process under
        particular circumstances. Because mutation occurs at the molecular level, it is necessary to
        examine the molecular changes produced by the underlying mutation in order to assess whether
        a given adaptation is best considered as a gain, loss, or modification of function. Although that
        was once impossible, the advance of molecular biology in the past half century has made it
        feasible. In this paper, I review molecular changes underlying some adaptations, with a partic-
        ular emphasis on evolutionary experiments with microbes conducted over the past four decades.
        I show that by far the most common adaptive changes seen in those examples are due to the loss
        or modification of a pre-existing molecular function, and I discuss the possible reasons for the
        prominence of such mutations.

     Adaptation by Gain, Loss, or                         Yet he realized that the changes that were
       Modification of Function                           selected to adapt an organism to its envi-
                                                          ronment did not have to be ones that con-
I N THE ORIGIN OF SPECIES, Darwin
  (1859) emphasized the relentlessness of
natural selection:
                                                          ferred upon it a new ability, such as sight or
                                                          flight. For instance, while observing some
                                                          barnacles, Darwin discovered unexpected
 [N]atural selection is daily and hourly scru-            cases of the gross simplification of an or-
 tinising, throughout the world, every varia-             ganism (Stott 2003):
 tion, even the slightest; rejecting that
 which is bad, preserving and adding up all                  The male is as transparent as glass . . . In
 that is good; silently and insensibly work-                 the lower part we have an eye, & great testis
 ing, whenever and wherever opportunity                      & vesicula seminalis: in the capitulum we
 offers, at the improvement of each organic                  have nothing but a tremendously long pe-
 being in relation to its organic and inor-                  nis coiled up & which can be exserted.
 ganic conditions of life. (p. 84)                           There is no mouth no stomach no cirri, no

                     The Quarterly Review of Biology, December 2010, Vol. 85, No. 4
                Copyright © 2010 by The University of Chicago Press. All rights reserved.
                                   0033-5770/2010/8504-0002$15.00

                                                      419
420                       THE QUARTERLY REVIEW OF BIOLOGY                            Volume 85

  proper thorax! The whole animal is re-         protein that is the target of the mammalian
  duced to an envelope . . . containing the      immune response might sustain a missense
  testes, vesicula, and penis. (p. 213)          mutation, causing an amino acid substitu-
Adaptive evolution can just as easily lead to    tion at a recognized epitope of the protein,
the loss of a functional feature in a species    thereby decreasing or eliminating the im-
lineage as to the gain of one. Over the          mune response while preserving protein
course of evolutionary history, snakes have      function, which could be considered a sim-
lost legs, cavefish have lost vision, and the    ple modification of a pre-existing mo-
parasitic bacterium Mycoplasma genitalium        lecular element; 3) the bacterium might
has lost its ability to live independently in    acquire a gene or gene cluster by horizon-
the wild, all in an effort to become better      tal transfer that leads to expression of a
adapted to their environments. Whatever          new surface feature that hides the epitope
variation that sufficiently aids a particular    from the mammalian immune system,
species at a particular moment in a partic-      which would be a gain of a functional mo-
ular environment— be it the gain or loss of      lecular feature. The focus of this review is
a feature, or the simple modification of         on the gain, loss, or modification of func-
one—may be selected. Since species can           tional molecular features underlying adap-
evolve to gain, lose, or modify functional       tation, no matter whether the phenotypic
features, it is of basic interest to determine   manifestation strikes us as a loss or gain of
whether any of these tends to dominate           a property.
adaptations whose underlying molecular              There are, of course, myriad functional
bases are ascertainable. Here, I survey the      molecular features of a cell, including those
results of evolutionary laboratory experi-       of all the major biomolecular categories: nu-
ments on microbes that have been con-            cleic acids, proteins, polysaccharides, lipids,
ducted over the past four decades. Such          small metabolites, and so forth. However, the
experiments exercise the greatest control        functional molecular features of only two cat-
over environmental variables, and they           egories—nucleic acids and proteins—are di-
yield our most extensively characterized re-     rectly coded by the genome, and thus only
sults at the molecular level.                    these features can be directly affected by mu-
                                                 tation; all other biological features are only
                                                 affected indirectly by mutations, through
      distinctions among adaptive                their effects on nucleic acids and proteins. In
                mutations                        this review, I focus on adaptive evolution by
   Adaptation is often viewed from two dis-      gain, loss, or modification of what I term
tinct aspects: the phenotypic aspect and         Functional Coded elemenTs (FCTs). An FCT is a
the molecular aspect. In order to avoid          discrete but not necessarily contiguous re-
confusion concerning how adaptive evolu-         gion of a gene that, by means of its nucleo-
tion proceeds, these two aspects must be         tide sequence, influences the production,
kept separate. A phenotypic adaptation           processing, or biological activity of a particu-
might have any number of underlying mo-          lar nucleic acid or protein, or its specific
lecular bases, resulting from either the         binding to another molecule. Examples of
gain, loss, or modification of a molecular       FCTs are: promoters; enhancers; insulators;
feature. As an example, consider a bacte-        Shine-Dalgarno sequences; tRNA genes;
rium that evolves virulence in a mamma-          miRNA genes; protein coding sequences;
lian host. If the adaptive phenotype is          organellar targeting- or localization-
deemed to be the gain of pathogenicity,          signals; intron/extron splice sites; codons
then that could arise in several ways. For       specifying the binding site of a protein for
instance: 1) a gene for a bacterial protein      another molecule (such as its substrate,
that is the target of the mammalian im-          another protein, or a small allosteric regu-
mune response might be deleted, which            lator); codons specifying a processing site
would be the loss of a functional molecular      of a protein (such as a cleavage, myristoyl-
feature of the bacterium; 2) a gene for a        ation, or phosphorylation site); polyade-
December 2010                    EXPERIMENTAL EVOLUTION                                         421

nylation signals; and transcription and            a gene or regulatory region, since this pro-
translation termination signals. Examples          cess does not by itself produce a new func-
of general nucleic acid features that are          tional element, although it may, like other
not FCTs include base composition, strand          events classified here as “modification-of-
asymmetry, distance between coded ele-             function,” be a step to future gain-of-FCT
ments, and the like. They may enter into           events. It also includes mutations that may
consideration indirectly only if they cause        act by more amorphous means, such as rear-
an FCT to be lost or gained—for example,           ranging gene order, or changing the dis-
if altering the base composition also elim-        tance or orientation between two interacting
inates the activity of an enhancer.                elements (e.g., the deletion of a stretch of
   With these considerations in mind, I            DNA that brings two interacting transcrip-
classify adaptive mutations as belonging to        tion factors closer, or that affects their orien-
one of three modes:                                tation with respect to each other). This
   1) A “loss-of-FCT” adaptive mutation is a       category is dubbed “modification-of-function”
mutation that leads to the effective loss of       rather than “modification-of-FCT ” in order
the function of a specific, pre-existing,          to emphasize that the mutation may not nec-
coded element, while adapting an organ-            essarily exert its influence through alteration
ism to its environment. The loss of the            of a discrete coded element, as in the above
ability of a frame-shifted gene to produce a       amorphous examples.
functional product, of an altered promoter            In order to fall under any of the above
to bind a transcription factor, or of a mu-        three categories, a mutation must be adap-
tated protein to bind its former ligand, are       tive. Therefore, if, for instance, a mutation
examples of loss-of-FCT mutations.                 causes a transcription factor binding site to
   2) A “gain-of-FCT” adaptive mutation is         be formed or lost, but the mutation is ei-
a mutation that produces a specific, new,          ther neutral or deleterious, it is excluded
functional coded element while adapting            from consideration. Although assignment
an organism to its environment. The con-           of an adaptive mutation to one of the
struction by mutation of a new promoter,           modes above may occasionally be ambigu-
intron/exon splice site, or protein process-       ous, the assignment is straightforward in
ing site are gain-of-FCT mutations. Also in-       the great majority of cases in which molec-
cluded in this category is the divergence by       ular changes are elucidated, as we shall see
mutation of the activity of a previously du-       below.
plicated coded element.
   3) A “modification-of-function” adap-           illustrations of modes of adaptation
tive mutation is a mutation whose defining            To better our understanding of the
property is negative—while adapting an or-         kinds of mutations that fall into these three
ganism to an environment, it does not lead         modes, it is useful to look at illustrations
to the loss or gain of a specific FCT. This        from outside the realm of the experimen-
definition is intended to be broad enough          tal evolution of microbes. Consider, for in-
to act as a “catch-all” for anything that falls    stance, human mutations that have been
outside the first two modes. It includes           selected in the past 10,000 years under the
point mutations as well as other mutations         pressure of the malarial parasite Plasmo-
that have a quantitative effect on a pre-          dium falciparum. The reason for using such
existing FCT, increasing or decreasing its         examples as these is that, because of its
strength, for instance, or shifting its activity   threat to human health, the interaction of
somewhat (such as allowing a protein to            malaria with humans has been very well-
bind a structurally-related ligand at the          studied at the molecular level.
same site as its normal substrate), but               If a person of typical northern European
without effectively eliminating it. The            descent visited a malarious region of the
category “modification-of-function” also           world and contracted the disease, she
includes the simple duplication of fea-            would likely view the resistance to malaria
tures, without divergence of activity, such as     among many of the native peoples as an
422                               THE QUARTERLY REVIEW OF BIOLOGY                                        Volume 85

                                                      TABLE 1
                               Human mutations selected for resistance to malaria
                                                                               Mode of
Clinical phenotype                     Underlying mutation                    adaptation            Reference
Hb S                       ␤6 glu val                                             G        Cavalli-Sforza et al. (1994)
Hb C                       ␤6 glu lys                                             M        Modiano et al. (2001)
Hb E                       ␤26 glu lys                                            M        Weatherall et al. (2002)
Hereditary persistence     Deletion/point mutations in control regions of         L        Forget (1998)
  of fetal hemoglobin        hemoglobin gamma chain gene
Thalassemia                Deletion/point mutations in either ␣ or ␤              L        Flint et al. (1986)
                             hemoglobin genes
G6PD deficiency            Point mutations, deletions of G6PD gene                L        Ruwende et al. (1995)
Loss of Duffy antigen in   Point mutation switching off production of Duffy       L        Pogo and Chaudhuri (2000)
  red blood cells            antigen specifically in red blood cells
Band 3 protein             Deletion                                               L        Kennedy (2002)
  deficiency
Abbreviations:
  G - adaptive gain of functional coded element
  L - adaptive loss of functional coded element
  M - adaptive modification of function

unquestionably positive trait: the ability to                 Kd ⬃ 1 mM (Behe and Englander 1979)—
withstand a debilitating disease. However,                    protein binding site has appeared.
because of the detailed knowledge of the                         Another population indigenous to some
molecular bases of malaria resistance that                    malarial regions has a different point mu-
has been accumulated over past decades,                       tation in their hemoglobin. In this in-
that is not the way I would categorize it                     stance, the sixth codon of the ␤ chain has
here. Many persons indigenous to malari-                      mutated from glutamic acid to lysine. Al-
ous regions are resistant, but they have ac-                  though the altered hemoglobin (HbC)
quired their resistance through different                     does not aggregate as sickle hemoglobin
molecular mechanisms (Table 1). Al-                           does, it confers resistance to malaria for
though the mutations are all adaptive, dis-                   reasons that are unclear. Because appar-
tinctions may be made among them. The                         ently no new, discrete, coded molecular
most well-known population of resistant                       feature has been developed, this is catego-
persons is heterozygous for sickle hemo-                      rized as a “modification-of-function” adap-
globin. In one of the two genes that they                     tive mutation; no FCT has been lost or
inherited from their parents for the ␤                        gained.
chain of hemoglobin, the sixth codon des-                        A third population of natives in some
ignates valine rather than the glutamic                       malarial regions is comparatively healthy
acid that the other ␤ chain gene codes.                       due to a different kind of mutation. They
When the red blood cell is invaded by the                     have a condition called thalassemia, which,
malarial parasite, the mutant, hydrophobic                    like sickle hemoglobin and HbC, also con-
valine residue allows hemoglobin mole-                        fers a measure of resistance to malaria. In a
cules to aggregate with each other into                       thalassemic person, however, one of the ␣
long microtubular structures. This poly-                      or ␤ chain hemoglobin genes that is inher-
merization is associated with a measure of                    ited from a parent is nonfunctional, for
resistance to malaria (Friedman 1978), al-                    any one of a variety of reasons. In some
though the exact mechanism for resistance                     thalassemic individuals, a whole gene has
in vivo is still a matter of debate (Verra et                 been deleted, whereas, in others, a frame-
al. 2007). This is an example of an adaptive                  shift mutation has caused the gene to
gain-of-FCT mutation because a codon                          produce a nonfunctional protein. In still
helping to specify a new, albeit weak—with                    others, the genetic control region near
December 2010                   EXPERIMENTAL EVOLUTION                                     423

one of the globin genes has suffered a           new metabolic functions in laboratory or-
change that makes it nonfunctional, and so       ganisms” (Hall 1983) and Microorganisms as
it produces no protein. Any of these would       Model Systems for Studying Evolution (Mort-
be an example of a loss-of-FCT mutation.         lock 1984a)—were both about twenty years
   Other populations are also malaria-           old at the time, thus indicating a lack of
resistant due to loss-of-FCT mutations,          recent work in that subfield.
but in genes other than the globin genes.
Take, for instance, the gene that codes             metabolism of unusual compounds
for glucose-6-phosphate dehydrogenase               I will begin by examining the early
(G6PD). Any one of hundreds of separate          work reported in Microorganisms as Model
known mutations in the coding and con-           Systems for Studying Evolution, edited by
trol regions of the gene has eliminated the      Mortlock (1984a) (Table 2). Four of the
functional enzyme, or has significantly de-      first five chapters concern several species
creased its catalytic activity. Other loss-of-   of bacteria in which cultures are evolved
FCT mutations include the deletion of a          to metabolize an unusual chemical food
gene for Band 3 protein, and the loss of a       source that the unevolved species cannot
promoter site for expression of Duffy anti-      grow upon. In general, in each of the
gen in red blood cells.                          investigations, the first mutation selected
   In summary, human genetic lines in ma-        is either one that inactivates a repressor
larious regions of the world have adapted        gene, thus turning the gene it regulates
to an environment that includes the malar-       into a constitutively-expressed gene, or,
ial parasite, thereby giving rise to a number    less often, one that increases the copy
of molecular changes (Table 1). They have        number of a particular metabolic gene.
done so by different modes, which can be         (The identities of some mutations in this
classified as adaptive mutations involving       early work were inferred by the authors,
loss- or gain-of-FCT, or modification-of-        not demonstrated.) The amplified or de-
function. As we shall see below, microbes        regulated gene is one that typically codes
in laboratory evolution experiments also         for an enzyme that metabolizes a related
adapt in these ways.                             compound and that has a limited ability
                                                 to metabolize the novel substrate. The
Evolution Experiments with Bacteria              increase in the concentration of enzyme
   Rather than performing an exhaus-             brought about by deregulation or gene
tive review of laboratory evolution exper-       amplification allows it to metabolize
iments, I will use several recent, influential   enough of the novel material to permit
reviews as jumping-off points for my survey      the mutated bacterium to grow, albeit
of the relevant literature, starting with        sometimes slowly. Similar results were re-
“Evolution experiments with microorgan-          ported by Clarke (1984) in Chapter 7,
isms: the dynamics and genetic bases of          which focuses on mutations in the meta-
adaptation” (Elena and Lenski 2003). In          bolic pathway of Pseudomonas that metab-
their review, Elena and Lenski (2003) were       olizes amides. During the course of many
particularly interested in laboratory studies    experiments, she obtained a variety of
that were “open-ended and long-term,”            point mutations in the regulatory protein
such as they themselves had conducted.           for the amidase gene and in the pro-
They eschewed reviewing experiments that         moter region for the amidase gene, as
“targeted specific, often new, metabolic         well as in the gene itself.
functions, even though this work is of great        Constitutive mutations are loss-of-FCT mu-
interest” (Elena and Lenski 2003:458). In-       tations because the organism has lost the
stead, for that topic, they referred readers     function of a coded feature that controls syn-
to two previous reviews. Interestingly, even     thesis of the enzyme. Once the bacterium
though their own review was quite up-to-         has mutated, allowing it to utilize a novel
date, the two reviews on specific metabolic      substrate sufficiently enough to grow, fur-
functions that they cited—“Evolution of          ther selection can often improve the growth
424                               THE QUARTERLY REVIEW OF BIOLOGY                                      Volume 85

                                                     TABLE 2
      Work reported in Microorganisms as Model Systems for Studying Evolution (Mortlock 1984)
                                                                                           Mode of
Chapter          Laboratory phenotype                    Underlying mutation              adaptation    Reference
   1         Xyltitol catabolism by Klebsiella     Inactivation of a ribitol                  L          Mortlock
                                                     dehydrogenase repressor gene                        (1984c)
             l-arabitol catabolism by Klebsiella   Inactivation of a ribitol                  L
                                                     dehydrogenase repressor gene
             Increased catabolism of xylitol       Apparent point mutations of ribitol        M
                                                     dehydrogenase
             Increased catabolism of xylitol       Apparent alteration of ribitol             M
                                                     dehydrogenase gene promoter
             Increased catabolism of xylitol       Apparent increase in copy number           M
                                                     of ribitol dehydrogenase gene
             Xylitol transport across              Deregulation of d-arabitol operon          L
                membrane
   2         Increased xylitol catabolism by       Apparent increase in copy number           M           Hartley
                Klebsiella                           of ribitol dehydrogenase gene                        (1984)
             Increased xylitol catabolism by       Point mutation in ribitol                  M
                Klebsiella                           dehydrogenase protein
             Increased xylitol catabolism by       Transfer of Klebsiella rdh operon to       M
                E. coli                              E. coli plus gene amplification
             Increased xylitol catabolism by       Transfer of Klebsiella rdh operon to       M
                E. coli                              E. coli plus point mutation
   4         d-lyxose isomerase activity by        Uncertain; perhaps due to                  L          Mortlock
                Klebsiella                           constitutive d-mannose                              (1984b)
                                                     isomerase
             d-lyxose isomerase activity in E.     Uncertain; perhaps due to                  L
                coli                                 constitutive d-mannose
                                                     isomerase
             d-arabinose isomerase activity by     Constitutive l-fucose isomerase            L
                Klebsiella
             d-arabinose isomerase activity in     Point mutation-induced l-fucose            M
                E. coli                               pathway
   5         Growth on l-1,2-propanediol of        Possible alteration of                     M         Lin and Wu
                E. coli                               oxidoreductase promoter                             (1984)
             Growth on d-arabinose of E. coli      Either constitutive expression of         L
                                                      fucose genes, or dual control of       G
                                                      regulator protein by l-fucose
                                                      and d-arabinose
             Growth on d-arabitol of E. coli       Constitutive production of                 L
                                                      oxidoreductase in already-
                                                      mutated strain 3
             Growth on xylitol of E. coli          Constitutive production of d-xylose        L
                                                      pathway
             Growth on ethylene glycol of E.       Constitutive use of l-1,2-                 L
               coli                                   propanediol pathway
   6         Growth of E. coli on lactose          Point mutation in cryptic ␤-               M         Hall (1984)
               (after deletion of lacZ)               galactosidase ebgo; also rendered       L
                                                      constitutive by frameshifts and
                                                      IS-disruption of ebg repressor
             Growth on lactulose                   Second point mutation in ebgo              M
             Increased growth on lactulose         Point mutation in ebg repressor            M
                                                      allowing lactulose as inducer
             Growth on lactobionate                Third point mutation in ebgo               M
                                                                                                           continued
December 2010                                EXPERIMENTAL EVOLUTION                                             425

                                                     TABLE 2
                                                      Continued
                                                                                             Mode of
Chapter           Laboratory phenotype                   Underlying mutation                adaptation   Reference
    7         Growth in presence of                Possible point mutation in                   M          Clarke
                butyramide of P. aeruginosa          regulator protein gene amiR                           (1984)
              Growth in presence of                Possible point mutation in                   M
                succinate/lactamide                  regulator protein gene amiR plus
                                                     catabolite repression mutant
              Increased production of              Mutation in promoter                         M
                amidase
              Growth on butyramide,                Point mutations in amidase                   M
                valeramide, acetanilide
    8         Resistance of yeast to allyl         Point mutations in alcohol                   M        Wills (1984)
                alcohol                              dehydrogenase
    9         Rescue of leuD deletion              Availability of cryptic leuD-like gene       M          Kemper
                                                                                                           (1984)
Abbreviations:
  L - adaptive loss of functional coded element
  M - adaptive modification of function

rate through point mutations in the gene for                      recruitment of cryptic proteins
the overproduced enzyme. These additional                        The evolution of the ability of the bacte-
mutations are modification-of-function mu-                    ria described above to metabolize novel
tations, because no FCT is gained or lost.                    compounds depends on mutations in
Increased-copy-number mutants are also                        known proteins of known metabolic path-
modification-of-function mutants, because                     ways. Several chapters in Microorganisms as
there is no evidence that any of the extra                    Model Systems for Studying Evolution (Mort-
copies have diversified, which would then                     lock 1984a), however, describe examples
place it in the class of gain-of-FCT muta-                    where previously unsuspected, cryptic pro-
tions.                                                        teins were recruited to take over the func-
   An interesting variation on this pattern                   tion of a known protein whose gene had
is reported in Chapter 5 by Lin and Wu                        been deliberately deleted.
(1984). Mutants of E. coli and S. typhi-                         In Chapter 9, Kemper (1984) describes
murium (S. enterica var Typhimurium) gain                     the ability of the product of a gene of Sal-
the ability to metabolize the unusual sub-                    monella typhimurium, which he called newD,
strate d-arabinose by altering the specificity                to replace a protein coded by the leuD gene
of a regulatory protein. Normally, the en-                    of the leucine biosynthesis pathway. NewD
                                                              ordinarily binds tightly to a larger protein
zyme fucose isomerase is induced in these
                                                              subunit that Kemper dubbed SupQ, just as
bacteria when some fucose enters the cell
                                                              LeuD binds to a larger subunit, LeuC. In
and binds to a positive regulatory protein,                   order to allow NewD to restore leucine
which then turns on the gene for the isom-                    biosynthetic function when leuD is deliber-
erase. The regulatory protein of some mu-                     ately deleted, the gene for SupQ also has to
tants, however, responds to both fucose                       be deliberately knocked out. Kemper ar-
and d-arabinose. It turns out that the un-                    gues that supQ/newD was not simply a gene
usual substrate d-arabinose can be metab-                     duplication of leuC/leuD (the genes of the
olized by enzymes of the fucose pathway,                      leucine pathway) (Stover et al. 1988). How-
and, because the protein has apparently                       ever, at that time, sequence data was more
gained an additional binding site for the                     difficult to come by than it is today. It is
novel substrate, the mutation is classed as                   now known that the S. typhimurium genome
gain-of-FCT.                                                  contains a homolog of LeuD with 38%
426                      THE QUARTERLY REVIEW OF BIOLOGY                         Volume 85

amino acid sequence identity. It seems         logs, unlike the previous cases of LacZ/
likely that the protein that Kemper discov-    Ebg and LeuD/NewD.
ered that could substitute for LeuD was           The work of Patrick et al. (2007) is not in
this homolog. The S. typhimurium genome        itself an experimental evolution study, as
also contains a homolog of LeuC with 52%       all of the manipulations were performed
amino acid sequence identity; it seems         by the investigators and no spontaneous
likely that SupQ was this homolog. No such     mutations arose. Nonetheless, it does dem-
LeuC or LeuD homologs exist in E. coli.        onstrate much potential redundancy in the
Because newD is a homolog of LeuD that         E. coli genome that may contribute to its
already binds a homolog of LeuC, this is a     evolvability, as the authors suggest. It may
modification-of-function event.                be instructive to consider how newer work
   In Chapter 6, Hall (1984) discusses his     would be classified if the events that
work over the previous decade with an en-      Patrick et al. (2007) investigated were to
zyme he dubbed Ebg, which stands for           occur in the wild. If, to adapt an organism
evolved ␤-galactosidase. He had shown that     in nature, a gene were initially lost by mu-
E. coli that had had its lacZ gene (which      tation, that of course would be a loss-of-
codes for a ␤-galactosidase) intentionally     FCT event. If subsequent rescue in nature
deleted could be mutated so that the al-       required the overexpression of a poten-
tered Ebg protein metabolized the sugar        tially compensating gene, as in Patrick et
lactose in its stead. In subsequent years,     al.’s (2007) experimental cases, then that
however, Hall (1995) showed unexpect-          could occur in at least several ways: 1) a
edly that Ebg was an unrecognized homo-        repressor element of the compensating
log of LacZ, with 38% sequence identity.       gene could be deleted, rendering the gene
The Ebg active site was very similar to that   constitutive, which would be a loss-of-FCT
of LacZ, and acquired its ability to metab-    mutation; 2) an existing promoter of the
olize lactose only when the Ebg active site    compensating gene could have its se-
mutated to one identical to that of an ac-     quence altered, binding a transcription
tive LacZ. Because the unevolved enzyme        factor more strongly and thus leading to
could already metabolize lactose at a low      overexpression, which would be classed as
rate, the adaptive point mutations that        a modification-of-function mutation, as an
increased its activity are categorized as      existing FCT was altered, not gained or
modification-of-function ones. Mutations       lost; 3) a new promoter could be con-
in the ebg repressor that allowed the ebg      structed by mutation to increase the tran-
␤-galactosidase protein to be constitutively   scription rate of the gene, which would be
expressed are loss-of-FCT.                     a gain-of-FCT mutation; 4) the compensat-
   In further experiments, both Kemper         ing gene could increase in copy number,
(1984) and Hall (2003) deleted the homol-      which by itself would be classed as a mod-
ogous recruited protein genes ebg and          ification-of-function mutation; or 5) a copy
newD, but were unable to recruit any of the    of the compensating gene could diverge in
thousands of other cellular proteins to re-    sequence to more efficiently assume the
place them. Recently, Patrick et al. (2007)    activity of the deleted gene, which would
employed functional genomics tools to sur-     be classified as a gain-of-FCT mutation. Fi-
vey 107 single-gene knockout strains of E.     nally, the effect of the deleted gene may be
coli that could not grow on M9-glucose me-     compensated in the wild not by increasing
dium, transfecting each one with a plasmid     the expression of its potential comple-
library that contained every E. coli open-     ment, but instead by deleting or reducing
reading frame. They discovered that about      the expression of one or more other genes,
20% of the knockout strains could be res-      which would be an adaptive loss-of-FCT
cued by overexpression of at least one         mutation (this possibility was not investi-
other E. coli gene. Patrick et al. (2007)      gated in the excellent work of Patrick et al.
found that the missing gene product and        [2007]). As these examples make clear,
its suppressor were generally not homo-        any of multiple molecular evolutionary
December 2010                    EXPERIMENTAL EVOLUTION                                     427

pathways could underlie the rescued                eral others. In subsequent work, Cooper et
phenotype. To understand whether an                al. (2001) discovered that twelve of twelve
adaptation represents a gain, loss, or             cell lines showed adaptive IS-mediated de-
modification of a function, the molecular          letions of their rbs operon, which is in-
events underlying the adaptation must              volved in making the sugar ribose. Thus,
first be understood.                               the adaptive mutations that were initially
                                                   tracked down all involved loss-of-FCT.
  the work of lenski and colleagues                   Several years later, when the cultures
   In the 1990s, investigators began to per-       had surpassed their 20,000th generation,
form the “open-ended and long-term” evo-           Lenski’s group at Michigan State brought
lution experiments that are the main focus         more advanced techniques to bear on the
of a review by Elena and Lenski (2003).            problem of identifying the molecular
Certainly, the most extensive and longest          changes underlying the adaptation of the
running investigation has been under-              E. coli cultures. Using DNA expression pro-
taken by Lenski himself, who has been con-         files, they were able to reliably track down
tinuously growing cultures of E. coli in his       changes in the expression of 1300 genes of
laboratory since the late 1980s and moni-          the bacterium, and determined that 59
toring the facets of their evolution (Lenski       genes had changed their expression levels
2004). Diluting a portion of the previous          from the ancestor, 47 of which were ex-
day’s culture a hundred-fold, each day can         pressed at lower levels (Cooper et al 2003).
potentially see 6 to 7 new generations of          The authors stated that “The expression
bacteria (26.6 ⬇ 100). At intervals, Lenski        levels of many of these 59 genes are known
freezes a portion of a bacterial generation        to be regulated by specific effectors includ-
and stores it, so that, later, the descendant      ing guanosine tetraphosphate (ppGpp)
generations can be straightforwardly ana-          and cAMP-cAMP receptor protein (CRP)”
lyzed and compared head-to-head with               (Cooper et al 2003:1074). They also noted
their ancestors. Over the years, the num-          that the cellular concentration of ppGpp is
ber of generations has approached 50,000.          controlled by several genes including spoT.
With a cumulative population size of about         After sequencing, they discovered a non-
1014 cells, Lenski’s investigation is large        synonymous point mutation in the spoT
enough and long enough to give solid, re-          gene. When the researchers examined ten
liable answers to many questions about             other populations that had evolved under
evolution.                                         the same conditions for 20,000 genera-
   The results of a number of Lenski’s pa-         tions, they found that seven others also had
pers over the term of the experiment are           fixed nonsynonymous point mutations in
very briefly summarized in Table 3, includ-        spoT, but with different substitutions than
ing his early work (not discussed here) in         the first one that had been identified, thus
which the underlying molecular bases of            suggesting that the mutations were de-
adaptive phenotypes had not yet been               creasing the protein’s activity.
identified. Schneider et al. (2000) exam-             The group then decided to concentrate
ined multiple insertion sequence (IS) mu-          on candidate genes suggested by the phys-
tations that became fixed in the bacterial         iological adaptations that the cells had
populations within ten thousand genera-            made over 20,000 generations. One such
tions. The mutations were of the kind              adaptation was a change in supercoiling
caused by IS elements, including insertions        density; therefore, genes affecting DNA to-
and recombinations, that led to genetic            pology were investigated (Crozat et al.
inversions and deletions. By examining the         2005). Two of these genes, topA and fis, had
DNA sequence of the E. coli in the neigh-          sustained point mutations. In the case of
borhood surrounding the IS elements, the           topA, the mutation coded an amino acid
investigators saw that several genes involved in   substitution, whereas, with fis, a transver-
central metabolism were knocked out, as well       sion had occurred at the fourth nucleotide
as some cell wall synthesis genes and sev-         before the starting ATG codon. The topA
428                                 THE QUARTERLY REVIEW OF BIOLOGY                                          Volume 85

                                                         TABLE 3
                      Selected results of Richard Lenski’s long-term E. coli evolution project
                                                                                          Mode of
Laboratory phenotype                                   Underlying mutation               adaptation        Reference
Mean fitness of evolved strains improves by       unknown                                    —        Lenski et al. (1994)
   37% on minimal glucose medium over
   2000 generations; most improvement
   comes early.
Over 10,000 generations, fitness and cell         unknown                                    —        Lenski and Travisano
   volume increase rapidly, then slow down;                                                             (1994)
   fitness increases in discrete steps; authors
   liken this to “macroevolution.”
Over 10,000 generations, mutator strains          Probable defect in methyl-directed        —*        Sniegowski et al.
   evolve in 3 of 12 populations; probably          mismatch repair pathway                             (1997)
   hitchhike with beneficial mutations.
After 6000 generations, two phenotypes, L         unknown                                    —        Rozen and Lenski
   and S, coexist in balanced polymorphism.                                                             (2000)
   Fitness is frequency dependent.
Study of nine IS mutants that are fixed after     IS mediated insertions, deletions,        L,L       Schneider et al.
   10,000 generations. IS had inserted into          rearrangements                         L,L         (2000)
   several genes (pykF, nadR, pbpA-rodA, hokB/
   sokB) or caused rearrangements.
12 of 12 lines show deletion of rbs operon        Repeated deletion of rbs operon            L        Cooper et al. (2001)
   mediated by IS150. Selective value of
   deletion is 1%-2%.
DNA expression arrays showed 59 gene              Eight separate point mutations in          M        Cooper et al. (2003)
   changes in parallel in two strains grown         spoT, which controls cellular
   for 20,000 generations. Many genes               concentration of ppGpp
   controlled by CRP and ppGpp.
Sequence random genes from 10,000 and             No mutations in nonmutator                —*        Lenski et al. (2003b)
   20,000 generations                                strains. Some point mutations
                                                     in mutator strains, but no
                                                     adaptive effect
Mutations affecting DNA topology: topA            topA H33Y; fis A-C transversion            M        Crozat et al. (2005)
   mutant decreases protein activity; fis            four nucleotides before start           M
   mutant decreases amount of protein made           ATG
Four genes (pykF, nadR, pbpA-rodA,                Same four genes repeatedly suffer         M,M       Woods et al. (2006)
   hokB/sokB) first identified by IS insertions      selectable mutations at different      M,M
   are examined in other populations.                sites
Comparison of protein profiles of evolved E.      malT locus suffers multiple                L        Pelosi et al. (2006)
   coli vs. ancestor                                 deletions, substitutions
Citrate utilization under aerobic conditions      unknown                                    —        Blount et al. (2008)
*Identified or inferred mutations are not categorized because they are neutral, not adaptive.
Abbreviations:
  G - adaptive gain of functional coded element
  L - adaptive loss of functional coded element
  M - adaptive modification of function

mutation decreased the activity of the en-                       served to be disrupted by IS elements,
zyme, while the fis mutation decreased the                       thereby leading to greater fitness in the
amount of fis gene product produced.                             growth medium, were then sequenced in
Moving the mutations into the ancestor                           twelve culture lines that had evolved for
improved its fitness in minimal glucose me-                      20,000 generations (Woods et al. 2006). All
dia.                                                             lines were observed to have point muta-
  The genes that had earlier been ob-                            tions in or near one or more of the genes;
December 2010                  EXPERIMENTAL EVOLUTION                                      429

however, the genes were mutated at a vari-      constitutive expression on a multicopy
ety of points. In other words, there was        plasmid of a citrate transporter gene, citT,
strong parallelism at the level of which        which normally transports citrate in the
gene it is adaptive to mutate, but little or    absence of oxygen, was responsible for elic-
no parallelism for a particular mutation in     iting the phenotype (Pos et al. 1998). If the
a particular gene. The fact that multiple       phenotype of the Lenski Cit⫹ strain is
point mutations in each gene could serve        caused by the loss of the activity of a nor-
an adaptive role—and that disruption by IS      mal genetic regulatory element, such as a
insertion was beneficial—suggests that the      repressor binding site or other FCT, it will,
point mutations were decreasing or elimi-       of course, be a loss-of-FCT mutation, de-
nating the protein’s function.                  spite its highly adaptive effects in the pres-
   In an investigation of global protein pro-   ence of citrate. If the phenotype is due to
files of the evolved E. coli, Lenski’s group    one or more mutations that result in, for
discovered that the MalT protein of the         example, the addition of a novel genetic
maltose operon had suffered mutations in        regulatory element, gene-duplication with
8 out of 12 strains (Pelosi et al. 2006).       sequence divergence, or the gain of a new
Several mutations were small deletions          binding site, then it will be a noteworthy
while others were point mutations, thus         gain-of-FCT mutation.
suggesting that decreasing the activity of         The results of future work aside, so far,
the MalT protein was adaptive in minimal        during the course of the longest, most
glucose media. This was tested by moving        open-ended, and most extensive laboratory
the mutation into the ancestral strain,         investigation of bacterial evolution, a num-
which subsequently gained in fitness.           ber of adaptive mutations have been iden-
   Recently, Lenski’s group reported the        tified that endow the bacterial strain with
isolation of a mutant E. coli that had          greater fitness compared to that of the an-
evolved a Cit⫹ phenotype. That is, the          cestral strain in the particular growth me-
strain could grow under aerobic condi-          dium. The goal of Lenski’s research was not
tions in a culture of citrate (Blount et al.    to analyze adaptive mutations in terms of
2008). Wild E. coli cannot grow under such      gain or loss of function, as is the focus here,
conditions, as it lacks a citrate permease to   but rather to address other longstanding evo-
import the metabolite under oxic condi-         lutionary questions. Nonetheless, all of the
tions. (It should be noted that, once inside    mutations identified to date can readily be
the cell, however, E. coli has the enzymatic    classified as either modification-of-function
capacity to metabolize citrate.) The pheno-     or loss-of-FCT.
type, whose underlying molecular changes
have not yet been reported, conferred an                    an ambiguous case
enormous growth advantage because the              A fascinating case of concurrent gain-
culture media contained excess citrate but      and loss-of FCT can be seen in the work of
only limited glucose, which the ancestral       Zinser et al. (2003). The authors examined
bacteria metabolized. Blount et al. (2008)      a mutant of E. coli that thrived when the
marshaled evidence to show that multiple        culture was starved, but that was outcom-
mutations were needed in the population         peted under conditions of growth. They
before the final mutation conferred the         demonstrated that the result depended on
ability to import citrate; the activating mu-   two mutational events: 1) an initial inser-
tation did not appear until after the           tion of an IS element between the tran-
30,000th generation.                            scriptional promoter of the cstA gene and a
   As Blount et al. (2008) discussed, several   nearby CRP box that regulated it; and
other laboratories had, in the past, also       2) inversion of the sequence flanked by the
identified mutant E. coli strains with such a   new IS element and another IS element
phenotype. In one such case, the underly-       ⬃60 kb distant and upstream of the ybeJ-
ing mutation was not identified (Hall           gltJKL-ybeK operon. The inversion brought
1982); however, in another case, high-level     the ybeJ gene of the mutant, which encodes a
430                       THE QUARTERLY REVIEW OF BIOLOGY                          Volume 85

portion of an amino acid transporter, under      recruit as bacteria do, except that which
the influence of the CRP box that previously     comes from their cellular hosts. 2) Because
regulated the cstA gene and that encodes a       of their small genome size, most of their
starvation-inducible oligopeptide permease.      genome is required for their life cycle; rel-
In the new arrangement, the ybeJ gene was        atively few viral genes are dispensable.
actively transcribed, while the cstA gene        3) For RNA viruses, the mutation rate is
no longer was. Interestingly, the ybeJ gene      greatly increased, as much as a million-fold
product helped the bacterium grow under          over that of cells. This allows various mu-
conditions of starvation as expected, but        tation “solutions” to a selective challenge
the authors demonstrated that expression         to present themselves much more rapidly
of the cstA gene itself would also be bene-      than for cells.
ficial under the same conditions. Nonethe-          Table 4 summarizes evolution experi-
less, the inverted arrangement was selected      ments with viruses discussed in the reviews
because it possessed the greatest net fitness.   above or in related papers. Most of the
   This example points to unavoidable am-        experiments discussed below disturbed vi-
biguity in the classification of some adap-      rus growth in various ways and followed
tive evolutionary events. One can view the       their evolutionary reaction to the distur-
adaptation investigated by Zinser et al.         bance. The papers are categorized by the
(2003) in several ways: 1) as a modification-    general selective pressure that was brought
of-function event (i.e., the insertion of a      to bear on the viruses.
duplicate IS element) followed, as a result
of the inversion, by the loss of a functional     recovery from multiple bottleneck-
coded element controlling the cstA gene               induced deleterious mutations
(the CRP box) and the gain of a functional          Two laboratories (Burch and Chao 1999;
coded element controlling the ybeJ gene          Bull et al. 2003) subjected two different
(the same CRP box); or 2) since no FCT           viruses—an RNA virus (␸6) and a DNA
was gained or lost in the inversion but          virus (T7)—to repeated population bottle-
rather was simply rearranged, the inversion      necks (i.e., severe reductions in population
sums to a second modification-of-function        size that cause the loss of genetic variation)
mutation, albeit one in which there is now       of just one or a few viruses. Because of the
a potential for a recombination switch un-       high mutation rate of RNA viruses, each
der conditions of starvation vs. growth.         virus has a high probability of having a
                                                 mutation, most likely deleterious, and,
 Evolution Experiments with Viruses              through many bottlenecks, the researchers
   Because of their ability to rapidly re-       were able to accumulate many mutations
produce to enormous population sizes,            in the virus. Because of the lower rate of
viruses, like bacteria, are well-suited to       mutation in DNA viruses, the workers us-
experimental evolutionary studies, and           ing phage T7 added a mutagen to the
such studies have been well-reviewed in          growth medium in order to increase its
recent years (Elena and Lenski 2003; Elena       mutation rate.
and Sanjuan 2007; Bull and Molineux                 Burch and Chao (1999) showed that ␸6
2008; Elena et al. 2008). Several general        suffered a ten-fold loss of fitness after 20
features distinguish viruses from bacteria       bottleneck generations of repeatedly pick-
and alter evolutionary expectations. 1) Viral    ing individual plaques grown on a plate of
genomes are usually much smaller than            bacterial host, and then using the plaque
bacterial genomes, often by a thousand-          to seed a new plate. When subsequently
fold, and one consequence of this is that        allowed to grow at larger population
mutations are much easier to identify by         sizes, the phage was able to recover fit-
sequencing for viruses than for bacteria.        ness by compensatory mutations; how-
Another consequence, however, is that vi-        ever, because the molecular natures of
ruses do not have deep reserves of homol-        the mutations were not determined, they
ogous genes or other genetic material to         cannot be classified here.
December 2010                    EXPERIMENTAL EVOLUTION                                          431

   Bull et al. (2003) were able to accumu-         ments, thus suggesting that the mutations
late hundreds of mutations in bacterio-            compensate for growth in the particular
phage T7, and to reduce its fitness to about       environment. All mutations were nucleo-
0.0001% of the wild type. During the recov-        tide substitutions; no deletions or inser-
ery phase of the experiment, two separate          tions were seen. Twelve substitutions were
populations of phage gained a factor of            synonymous, fifteen were nonsynonymous,
about 100-fold and 10,000-fold in fitness dur-     and three were in intergenic regions.
ing 115 and 82 passages, respectively. Both           Bull et al. (1997) have studied the adapta-
populations were continuing to recover             tion of replicate lineages of the single-stranded
fitness when the experiment was halted.            DNA virus ␸X174 to high temperature
Among the mutations accumulated during             (43.5°C) on several hosts. In one study of 119
the bottleneck phase of the experiment were        total substitutions at 68 sites in the virus, over
several small and large deletions in nones-        half were identical with substitutions in other
sential genes, one small insertion, and about      lineages, and no deletions or insertions were
400 point mutations, almost equally divided        reported. Adaptation improved fitness in pri-
between missense mutations and silent mu-          mary lines by about 4000-fold with S. typhi-
tations. There were also several nonsense          murium as host, and by about 100,000-fold with
mutations in nonessential genes.                   E. coli as host. In another study, Wichman et al.
   In the recovery phase, there were also sev-     (1999) grew ␸X174 on S. typhimurium at ele-
eral large and small deletions, as well as a few   vated temperature (43.5°C). Half of the muta-
dozen point mutations, most of which were          tions in one line also appeared in a second line
missense mutations that were likely subject        grown under identical conditions. Most
to positive selection. Recovery-phase muta-        changes were amino acid substitutions, but
tions involving large or frame-shifting dele-      one common mutation was an intergenic de-
tions in genes can, with reasonable confi-         letion of 27 nucleotides. Since no discrete mo-
dence, be classified as loss-of-FCT. Selectable    lecular features were attributed to the deleted
point mutations are more difficult to classify     intergenic region, that deletion can be consid-
definitively. None were reported to have           ered a modification-of-function—no func-
formed identifiable, functional coded ele-         tional coded element was lost. No discrete new
ments, such as new promoter sites or protein       molecular features were identified within mu-
processing sites. However, the possibility re-     tated coding regions, but further biochemical
mains that some new functional coded ele-          analysis would be required to determine if a
ment that is not readily identifiable from its     functional coded feature—such as a new pro-
nucleic acid sequence may have arisen in           tein binding site—were gained. In the absence
one or more cases. To rigorously test such         of such evidence, selectable point mutations
a possibility, careful biochemical analysis        are tentatively classified as “modification-of-
would be required. In the absence of ev-           function” in Table 4.
idence for gain-of-FCT, selectable point
mutations are tentatively classified as                     adapting to a new host
“modification-of-function” in Table 4.                Another category of selective pressure is
                                                   that of viruses forced to adapt to a novel
     adapting to a new environment                 host organism in the laboratory. The adap-
   A second category of selective pressure is      tation of a virus to a new host in the wild
that of viruses placed in new environments.        can, of course, be a major— even cata-
Cuevas et al. (2002) described the compe-          strophic— epidemiological event, as the ex-
tition of two neutrally-marked strains of          amples of HIV and, potentially, H1N1 show.
the single-stranded RNA vesicular stomati-         Nonetheless, as emphasized throughout this
tis virus (VSV) when grown on the same             review, to understand how adaptation pro-
medium and the same cells, but at varying          ceeds, the phenotypic and molecular as-
population sizes. There was remarkable             pects must be kept separate. The ability
parallelism in the exact nucleotide changes        of a virus to grow on an alternate host
for the two strains over multiple experi-          is a phenotype that may have a variety of
432                                  THE QUARTERLY REVIEW OF BIOLOGY                                            Volume 85

                                                        TABLE 4
                                     Summary of selected viral evolution experiments
                                                                                          Mode of
Investigator action                               Underlying mutation                    adaptation         Reference
Viruses subject to drift
  Bacteriophage ␸6 subject to drift in    Recovery by growth in larger populations           —        Burch and Chao (1999)
     bottleneck populations                 showed many small steps; no sequence
                                            data
  Bacteriophage T7 repeatedly passed      Several dozen point mutations; several            L,M       Bull et al. (2003)
     through single-virus population        large and small deletions in recovery
     bottlenecks                            phase
Viruses adapting to new environments
  Two neutrally-marked strains of         Extensive parallel convergence at the             M         Cuevas et al. (2002)
     VSV competed at different              nucleotide and amino acid level; some
     population ratios                      co-variations noted
  Bacteriophage ␸X174 adapted to          Multiple convergent substitution                  M         Bull et al. (1997)
     high temperature growth                mutations across several lineages
  Two populations of bacteriophage        Multiple convergent mutations, mostly             M         Wichman et al. (1999)
     ␸X174 adapted to high                  substitutions, in two populations
     temperature and novel host             accumulated in different orders
Viruses adapting to new hosts
  ␸X174 adapted to growth on              Amino acid substitutions at   5 sites in          M         Crill et al. (2000)
     Eschericia and Salmonella hosts        major capsid attachment     protein affect
                                            host preference
  ␸X174 adapted to growth on E. coli      Amino acid substitutions at   10 sites in         M         Pepin et al. (2008)
    strains differing in                    major capsid attachment     protein affect
    lipopolysaccharide attachment           host preference
    site
  ␸6 adapted to growth on 15              Amino acid substitutions at 9 sites in            M         Duffy et al. (2006)
    Pseudomonas syringae strains             spike attachment protein affect host
                                             range
  Broad-specificity ␸6 adapted to         Single substitution in spike attachment           M         Duffy et al. (2007)
     growth on single Pseudomonas            protein gene narrows host range
     syringae strain
Viruses manipulated to be defective
  Deletion of 19 intercistronic           One revertant deleted 6 nucleotides;              G,G       Olsthoorn and van
     nucleotides from RNA virus MS2          another duplicated an adjoining 14-                        Duin (1996)
     containing Shine-Dalgarno               nucleotide sequence; missing
     sequence and two hairpins               functional coded elements
                                             substantially restored
  4 nucleotide deletion in lysis gene     Reading frame restored by deletions,              G,G       Licis and van Duin
     of MS2                                  insertions                                                  (2006)
  Randomized operator sequence of         Modification-of-function mutations                M         Licis et al. (2000)
     MS2                                     appeared, but operator hairpin not
                                             restored
  Deletion of T7 viral ligase gene        Loss-of-FCT and modification-of-function          L,M       Rokyta et al. (2002)
                                             mutations in several genes; new ligase
                                             activity not obtained from host
  Bacteriophage T7 forced to              9 of 16 promoters altered sequence;               M         Bull et al. (2007)
    replicate using T3 RNA                   several missense modification-of-
    polymerase                               function changes
  T7 RNA polymerase gene moved            Loss of one E. coli polymerase                    L,M       Springman et al. (2005)
    toward end of viral genome               termination site; facilitated
                                             rearrangement of RNAP gene closer
                                             to the beginning of the viral genome
                                                                                                                      continued
December 2010                           EXPERIMENTAL EVOLUTION                                                433

                                                      TABLE 4
                                                       Continued
                                                                                Mode of
Investigator action                               Underlying mutation          adaptation        Reference
Viruses forced to be interdependent
  Separate viruses, f1 and IKe,        In media containing both antibiotics,       L,M      Sachs and Bull (2005)
     engineered to carry distinct         phages co-packaged into f1 protein       G
     antibiotic resistance markers        coats; two-thirds of IKe genome
                                          deleted, second antibiotic gene
                                          captured by f1
Abbreviations:
  G - adaptive gain of functional coded element
  L - adaptive loss of functional coded element
  M - adaptive modification of function

possible underlying molecular bases. As                      were nucleotide substitutions; seven of those
reviewed below and as observed in exper-                     were synonymous, and, of the remaining 14,
imental evolutionary studies, viruses have                   10 occurred in the major capsid protein
presumably adapted to new hosts by mod-                      gene F. Since the substitutions apparently
ification-of-function mutations rather than                  did not form a new FCT or cause the loss of
by gain- or loss-of-FCT ones.                                one, but simply modified the strength of at-
   Many experimental studies have been                       tachment, they are modification-of-function
performed investigating the adaptation of                    mutations.
viruses to new hosts, and several recent                        Duffy et al. (2006) examined the evo-
papers illustrate the topic of gain- or loss-                lution of the RNA bacteriophage ␸6 on
of-FCT that this review is concerned with.                   15 Pseudomonas syringae strains. Nine mu-
Commonly, although not exclusively (see                      tations were isolated that affected host
Yin and Lomax 1983; Subbarao et al. 1993;                    range, and all were amino acid substitu-
Agudelo-Romero et al. 2008), mutations in                    tions in the host-attachment spike-protein
coat proteins underlie the phenotypic vari-                  P3. One virus strain carrying a substitution
ation. Crill et al. (2000) studied the adap-                 of P3 that broadened the ␸6 host range was
tation of ␸X174 to Eschericia and Salmonella                 used for further studies (Duffy et al. 2007).
hosts. Over 11 days of selection on a host,                  When the mutant strain was grown exclu-
up to 28 substitutions, as well as a 2-base                  sively on an alternate permissive host, it
insertion and 27-base deletion, were ob-                     acquired a further single nucleotide substitu-
served in all genes except one. The authors                  tion in the gene for the spike protein. The
determined that nonsynonymous nucleo-                        substitution conferred increased growth on
tide substitutions at 5 sites in the major                   the alternate host and re-narrowed the host
capsid protein gene F affected host prefer-                  range.
ence. The altered amino acid residues lie                       The work reviewed above presents an op-
on the virion surface and correlate strongly                 portunity to further clarify the classification
with attachment efficiency but, intrigu-                     categories introduced in this paper. Host ad-
ingly, are at sites on the protein other than                aptations could potentially be classified in
the putative carbohydrate binding site                       several ways. Some investigators might cate-
identified by X-ray crystallographic studies                 gorize the adaptations reported by Duffy et
(McKenna et al. 1994). Pepin et al. (2008)                   al. (2007), for instance, as gains and losses of
revisited host adaptation of ␸X174, but in a                 the function of binding very specific host
more specific context. They examined the                     ligands. As I discussed earlier in this paper,
adaptation of the virus to three strains of E.               however, I classify shifts in ligand preference
coli that differed from each other only by a                 as “modification-of-function” events if the
single sugar group in the lipopolysaccha-                    binding occurs at the same site on the pro-
ride site of attachment. Of 21 mutations, all                tein, as in the examples discussed here. From
434                        THE QUARTERLY REVIEW OF BIOLOGY                             Volume 85

this view, the functional coded element is the     loss-of-FCT mutation. The 14-nucleotide
protein binding site, which can be modified        duplication led to the formation of new
either by amino acid substitution directly at      functional coded elements (it did not
the site or elsewhere in the protein. Consid-      simply repeat pre-existing elements), so
ered in this way, a functional site is not         it is not just a modification-of-function
gained or lost as host preference is altered,      mutation. The subsequent point muta-
but is instead modified.                           tions that improve activity are modifica-
                                                   tion-of-function.
  recovery from introduced defects:                   Another paper from van Duin’s labora-
       the rna bacteriophage ms2                   tory investigated the evolution of phage
   Another category of selective pressure is       MS2 in which a 4-nucleotide sequence was
the intentional introduction by the investiga-     deleted from a 38-nucleotide intercistronic
tor of defined defects in a virus. Olsthoorn       region between the coat protein gene and
and van Duin (1996) deleted a 19-nucleotide        the replicase gene (Licis and van Duin
intercistronic segment of RNA phage MS2            2006). The gene for the lysis protein spans
between the stop codon for the maturation          the intercistronic region and overlaps into
protein gene and the start codon for the coat      the coat protein and replicase genes, so
protein gene; this segment contains the            that the deletion introduces a frameshift
Shine-Dalgarno sequence and usually forms          into it. Furthermore, the deletion destabi-
two hairpins. The synthesis of the coat pro-       lizes the operator loop, which is needed to
tein is especially sensitive to the presence of    bind a dimer of coat protein, initiating cap-
a hairpin containing its initiator codon, and      sule formation, as well as regulating ex-
the investigators reported that the titer of the   pression of the replicase gene. The mutant
altered phage dropped ten orders of magni-         phage decreased in fitness by a factor of
tude (Olsthoorn and van Duin 1996). Two            106.
separate kinds of revertants were isolated—           Most initial revertants at relatively low ti-
one that had deleted 6 nucleotides that            ters first repaired the frameshift by inserting
coded the final two residues of the matura-        a nucleotide in this region. One deleted two
tion protein, and another that contained a         more nucleotides, which also restored the
duplication of an adjoining 14-nucleotide          correct reading frame. Several further rever-
sequence. Remarkably, the deletion rever-          tants were obtained by increasing the initial
tant substantially restored the coat pro-          population size of mutant viruses and passag-
tein initiator hairpin and the Shine-              ing the virus. In particular, insertion rever-
Dalgarno sequence and, subsequently,               tants were obtained, one of which (revIN4)
approximated them more closely by ac-              had inserted four nucleotides exactly at the
cumulating several point mutations. The            site where they initially had been deleted.
duplication revertant partially restored both      These are all gain-of-FCT mutations. After
native hairpins and the Shine-Dalgarno             further passages, most revertants acquired
sequence, and subsequently accumulated             point mutations that repaired damaged fea-
point mutations to more closely approxi-           tures, including the operator loop. Only one
mate the starting phage. The titers of the         revertant achieved a fitness equal to the wild
reconstructed phages were very similar to the      type phage after multiple passages, and this
starting phage.                                    was a derivative of revIN4 that went on to
   Both the 6-nucleotide deletion and the          revert completely to the wild type.
14-nucleotide duplication are gain-of-                A third paper from van Duin’s lab con-
FCT mutations since they both produced             cerning MS2 examined the evolutionary
new coded molecular features in the vi-            consequences of partially randomizing the
rus that did not exist in the immediate            sequence of the operator hairpin (Licis et
precursor; that is, the virus that sustained       al. 2000). The fitness of the initial mutants
the deliberately-deleted 19 nucleotides.           decreased anywhere from 103 to 107. Most
The 6-nucleotide deletion did not inacti-          revertants failed to reconstruct the opera-
vate the maturation protein, so it is not a        tor. The two best revertants, with 2% and
You can also read