Directed Mutation in Bacteria?
See a general article on mutants.

Updated and corrected: 17/8/13.


This article discusses unusual results obtained from a mutation experiment and this leads onto a discussion of
adaptive mutation and quantum biology. The article makes no novel claims or assertions, but leaves the reader
to draw their own conclusions. The experiment was inspired by Hill (2000) who used mammalian cells and here
we use the anaerobic bacterium
Bacteroides fragilis. Populations of cells exposed to the stress of antibiotic
resistance develop increased resistance, due largely to genetic mutation. Repeatedly exposing a fresh sample
of the same population causes increased resistance to develop over time, even though non-exposed control
populations do not develop resistance, even when old. Interestingly, the immediate siblings of cells exposed to
antibiotic also developed significantly increased levels of resistance even though no classical signal could have
conceivably passed from the exposed siblings to the non-exposed siblings. The possibilities of error, unforeseen
classical mechanisms and quantum entanglement of DNA are discussed.


Mutation is generally considered to be a random phenomenon. DNA contains lots of information that must be
copied from every mother cell to every daughter cell over the aeons. Like any information storage and
replication system, DNA replication does not have 100%
fidelity - errors occur during copying and storage. For
living things these inevitable errors are crucial in the long run as they enable
mutation, adaptation to change
and evolution. For this reason bacteria will actually regulate the degree of DNA-replication fidelity - they are
programmed to reduce fidelity when stressed, increasing mutation rate in order to increase the odds of
producing cells more resistant to whatever is causing the stress. Some regions of the DNA are adjusted more
than others, such that certain genes will have an enhanced mutation rate but not others - an effect called
adaptive mutation. Adaptive mutation can create the impression that bacteria mutate intelligently. For example,
lac operon in Escherichia coli is a region of DNA that produces proteins (including enzymes) needed if E. coli
is to grow on the sugar lactose in the absence of other sugars. If this region is disabled by a mutation and then
E. coli is grown with no other energy and carbon source than lactose than the lac operon will mutate back to
working order at a rate which is much higher than the normal mutation rate and higher than that in some other
parts of the genome. However, the
lac genes are not precisely targeted, rather certain regions of the genome
hypermutable (they have increased mutation rate) whereas other regions continue to mutate at the
normal frequency. It maybe that more vital parts of the genome are protected from hypermutation, whilst more
expendable parts are allowed to hypermutate, which would make sense since many mutations are
disadvantageous and knock out the genes in which they occur, but certainly the bacteria do not develop lactase
enzymes on demand!

Here we shall describe experiments that suggest a different mechanism altogether. We subjected a proportion of
cells in a population of the bacterium Bacteroides fragilis to antibiotic stress and found that a sub-lethal stress
induced antibiotic resistance to a subsequent lethal doses of antibiotic, as expected, but that resistance was
also induced in that part of the population that had NOT been exposed to antibiotic! The implication appears to
be that when bacteria are stressed they can somehow send a signal to their genetic relatives (siblings) to induce
specific resistance to the stressor they encounter in some of their siblings, even when they are physically
separated from their siblings in a separate sealed container. This is directed mutation and has been reported
before in mammalian cells (see review by Hill, 2000).

Materials and Methods

The basic method is illustrated in fig.1.

Establishing culture flasks

Bacteroides fragilis
(American type culture collection (ATCC) strain 25285 also designated National Type
Culture Collection (NTCC) 9343) were grown on agar plates overnight. Two single colonies were selected, each
into 3 ml broth and grown overnight. From each of the two 3 ml inoculates, 1 ml was placed into 100 ml broth.
Thus, two separate culture flasks were established, one as an independent control and one test flask.
Bacteroides fragilis is an anaerobe and was grown under anaerobic conditions using the Anoxomat Mk II system
(Spiral Biotech), in which air is replaced by a mixture of 85% nitrogen, 10% carbon dioxide and 5% hydrogen.
Each flask was maintained in a separate sealed jar containing the anaerobic environment. To further eliminate
the exchange of gaseous molecules between the two flasks, they were grown in separate incubators (although
the incubators were next to one another in the same room).

Applying mutator stress

  • At weekly intervals, 1 ml was withdrawn from the control flask (flask A) and 100 ul inoculated into 10 ml
    broth and shaken for 12h in anaerobic conditions in an anaerobic jar (sub-culture A).

  • At the same time, also at weekly intervals, 1 ml was withdrawn from the test flask (flask B) and 100 ul
    inoculated into 10 ml broth with 0.1 ug/ml of tetracycline (this is a sub-lethal dose since the minimum
    inhibitory concentration (MIC) of 0.5 to 1 ug/ml) and shaken for 12h in anaerobic conditions (sub-culture
    B+). In addition 100 ul were inoculated into 10 ml broth without antibiotic and shaken for 12h in anaerobic
    conditions (sub-culture B-). These two subcultures were grown in separate anaerobic jars. A further 100
    ul was also withdrawn from flask A, to ensure that both flasks received identical treatment and discarded (i.
    e. 200 ul was withdrawn from both flasks, but no stressor was applied to the inoculum from flask A).

  • Next, 100 ul samples from each of the three 10 ml sub-cultures were plated onto 3 agar plates with 0.5
    ug/ml tetracycline (a potentially lethal dose for wild type bacteria) and 3 agar plates without antibiotic (a
    total of 18 plates) and resultant colonies were counted after three days.

  • The pre-stressed (B+) and non-stressed (B-) sibling populations from flask B were kept strictly separate at
    all times. This was imperative! During the three days of growth on agar, the 6 test plates (B+, three +tet,
    three -tet) were in one anaerobic jar and the 6 control plates (B-, three +tet, three -tet) were kept in a
    separate anaerobic jar in separate incubators. (Thus, one incubator housed the control flask (B-) and
    control plates (B-), whilst another housed the test flask (B+) and test plates (B+)). These two incubators
    were situated next to one another, about one metre apart. [The 6 flask A plates were also kept separate,
    though as it transpired this was of no importance, so long as the B+ are kept strictly separate from the B-].

This procedure was repeated weekly (every 5-7 days) for 8 weeks (or until the peak mutation rate has been
obtained – see results). Note that 20 ml of fresh medium was added to both flasks, at week 4 only. We
distinguish between those test plates containing cells previously stressed by tetracycline as the stressed
population (B+) and the parent culture from which samples were also drawn and tested without prior exposure to
tetracycline comprise the unstressed sibling population (B-). The totally separate culture in flask C is a control
population to which no stressors were applied.

Estimating cell viability

When 0.1 ml samples were plated onto tetracycline agar plates, 0.1 ml samples were also plated onto
tetracycline free agar plates to estimate cell viability. In each case confluent growth was obtained in the absence
of tetracycline, even at week 8, although in the last couple of weeks the layer of growth was visibly thinner,
indicating a decline in viable numbers (though this was not accurately quantified).


The whole experiment was replicated five times and analysed by ANOVA. Our null hypothesis is that there
should be no difference in the number of resistant colonies obtained for both the stressed and the sibling
populations. On the basis of random mutation we expect the alternative hypothesis that there will be significantly
more resistant colonies in the stressed population than the sibling population. We expect there to be no
significant difference between the sibling and control populations.

The control groups showed essentially zero tetracycline resistant colonies, which is to be expected, since we were
plating out 0.1 ml of bacteria in stationary phase (about 10^8 colony forming units per plate) and normal
spontaneous mutation rates vary from about 10^-4 to 10^-10 per gene per generation, which are extremely low
and so would require many more plates to enable likely detection.

The increase in mutation rate can be explained by two possible mechanisms. Firstly, it may be due to the aging of
the cell population, since cells were maintained in continuous culture without the introduction of fresh medium
except at the 4 week time point, so the cells may be becoming exhausted or stressed as they are maintained in
stationary phase. Stress has been reported to increase mutation rate [refs]. The decline in the number of resistant
colonies at week 7 probably indicates that the cells are dying and losing viability. The second possible explanation
will be given below. [?]

The sibling population should show the same null result as the control population if mutation is entirely random.
However, the sibling population showed significantly more resistant colonies than the control group by week 4 and
the numbers peaked at week 7, corresponding to the peak in the stressed population, before similarly declining.
The numbers of resistant colonies obtained with the stressed population were significantly greater than that
obtained in the sibling population. Thus, the sibling population exhibited a definite response to the tetracycline,
albeit a weaker response than the stressed population, even though the sibling population never came into
contact with tetracycline. This suggests that the stressed population is somehow sending a signal to the sibling
population, increasing their mutation rate, either globally and non-specifically or specifically to enhance resistance
to tetracycline - our experiment does not discriminate between these two possibilities.

Since the stressed and sibling populations were in separate anaerobic jars, which are tightly sealed, it is doubtful
that a chemical gaseous signal could pass from one jar to the other. Furthermore, such a signal would reach the
control population with equal probability. To further eliminate chemical cross-talk we placed the stressed
population jar in a separate incubator from the sibling population jar. With each population in dishes in tightly
sealed anaerobic jars in physically separate incubators (that were nevertheless only about one metre apart) the
likelihood that a chemical signal could be sent from one population to the other seems vanishingly small. Also a
light signal seems unlikely since the two jars were in separate closed incubators and, again, such a signal could
potentially also be detected by the control population.

This result was highly consistent, occurring to a similar degree in all five replicates and with a similar time course.
The only exception was that in the first replicate the control group showed a similar, albeit very weak, response,
with a mean of 1.33 mutant colonies detected in week three, dwindling to zero by week 7 with 12 resistant colonies
detected in total. All other control groups gave absolutely zero resistant mutant colonies. This suggests that a
weak but demonstrable signal may have in fact reached the control group in this instance, though it may be that
for some reason we happened to detect spontaneous mutants in this instance, though this is extremely unlikely
unless the mutation rate was intrinsically higher in this control group.

The inescapable conclusion seems to be that a signal was passed from the stressed populations specifically to
their sibling population, which contained closer kin than the control population, since the number of generations
separating the stressed and sibling populations was much fewer than that between the sibling and control
populations. Thus, the signal appears kin specific. The signal is also mysterious as it can cross solid barriers over
a distance of at least one metre for perhaps up to twelve hours or more after the populations have been
separated - twelve hours corresponding to the period of sub-lethal stress.

The question remains: what is the physical nature of this signal? Hill (2000) found a very similar phenomenon, with
a similar time course, in mammalian cells subjected to either of thioguanine or ethionine or high temperature
stress. Not only did the incidence of mutations increase over time, despite transferring the cells to fresh culture
(however changes in mammalian cell-lines in continuous culture are known to occur) but Hill also observed that
the rate of mutation increased in non-stressed sibling populations. Hill suggested that the 'mysterious signal' may
be the phenomenon of quantum entanglement.

Quantum entanglement

Quantum entanglement is a well-established phenomenon in the sub-atomic world. For instance, two photons or
an electron and a positron that are produced as a pair from a given 'point' in spacetime will behave as a single
quantum entity described by a single set of quantum mechanical equations. This is not surprising, when one
considers that the wave functions of two particles that are physically close to one-another overlap. When this
overlap is significant then the two wave functions effectively merge into a single wave function. This happens , for
example, to the two electrons in a helium atom - the two electrons become indistinguishable in principle and it is
more accurate to picture them as a single two-photon state. This phenomenon gives rise to the measurable
exchange energy.

Recall that the wave-function describes the probabilistic behaviour of an ensemble of such particles in identical
states (we could almost equivalently imagine that it determines the probabilistic behaviour of the single particle
pair, but this places us on controversial ground regarding the nature of probability, so it is easier to consider a
population of such particle states). Suppose we produce a pair of photons of opposite polarity, one horizontally
polarised and the other vertically polarised, then it is possible in principle for each photon to exist in both the
vertical and horizontal states simultaneously, due to quantum superposition, until a suitable measurement forces
one of the particles to assume a definite state. The state assumed by a given photon upon performing such a
measurement is random, unless we assume that the photon has an intrinsic polarity independent of measurement,
which however invokes a hidden variable and current experimental research makes the existence of such hidden
variables appear highly unlikely. However, once one photon randomly becomes say, a horizontally polarised
photon, then its partner instantaneously becomes vertically polarised, since the two must be in opposite states.
The measurement changed the wave function, instantly changing the state of each particle from a superposition
into a single eigenstate. An interesting thing is that when two particles are produced as such an overlapping pair,
then their wavefunctions remain combined or entangled no matter how far apart they are - they can be travelling in
opposite directions at the speed of light but will remain in a connected or coherent quantum state until one of the
particles interacts and is forced into a definite state, then the state of the other particle is determined
instantaneously ('ghostly action at a distance') and the entanglement is broken.

Quantum entanglement in cells?

Such quantum entanglement has also been observed in atoms and even large molecules. However, for larger
particles, such as these, the entanglement becomes more fragile and ephemeral. In particular thermal noise easily
disrupts it, as one or other of the particles gains or loses thermal kinetic energy. In a cell this might happen when a
protein interacts with vibrating water molecules. Entanglement on atoms and molecules typically involves cooling
the particles down to very low temperatures. In any case, entanglement diminishes with distance travelled by the
particles and time elapsed, as it becomes increasingly likely that disruption will have occurred.

However, cells have an unusual property, often overlooked, that may favour the persistence of quantum
coherence: the electromagnetic fields generated by large biomolecules, such as DNA, lipids and proteins tends to
preserve order by causing polar water molecules to align in ordered arrays, reducing thermal noise. It at least
becomes a possibility then that quantum coherence may occur in biological molecules in cells and persist for long
periods of time. How long is difficult to estimate, clearly diffusion is important to cell function so clearly thermal
fluctuations occur to a significant degree in cells, but these are expected to be much less near large structures
such as DNA macromolecules.

There is a conceivable possibility that when DNA replicates, that the two daughter molecules produced are in a
state of quantum entanglement. Presumably, processes such as transcription will disrupt this coherence, but there
is a potential for a change in one base pair to cause an instantaneous change in the base pair of the sibling
molecule. In a population of such cells, we would expect such changes to occur sometimes but not always, as not
all will remain coherent for the necessary length of time. We would expect the number of sibling cells or DNA
molecules remaining entangled to decline over time and with spatial separation between them.

There is currently no proof that such entanglement occurs or that it is of biological importance, but it offers one
potential explanation as to why tetracycline resistance appears in the sibling cells that have never encountered
the antibiotic, even if this occurs at a lower frequency than in the populations of sibling cells that are stressed.
Indeed we would expect the frequency in non-stressed siblings to be lower. The control culture, being more
distantly related in time would be expected to exhibit no such entanglement.

As it stands, this study is simply a pilot that has never been followed up and we are not claiming that the results
observed are indeed due to quantum entanglement, but simply that no explanation has yet been found. If further
experiments did indeed verify this work and the work of Hill and show unequivocally that sibling cells can
communicate DNA base changes to their immediate siblings then we have a new evolutionary mechanism. Indeed,
such a mechanism would be highly advantageous, as if part of a population encounters a given stressor, then
there is a strong likelihood that the remainder will at some future time, in which case having a few cells already
pre-adapted would offer an immense selective advantage.

Is there any evidence for quantum entanglement in sibling cells?

An interesting study (ref) has shown that in cultures of 3T3 fibroblasts when one cell divides into two daughter
cells, the two daughter cells move off at random, since such cells migrate and turn randomly in the absence of a
directional stimulus. However, the random pattern of one daughter tended to be the mirror image of the other for a
considerable distance and time until both cells eventually differed in their paths! It is as if the two cells were in an
entangled state, mirroring each other like our entangled photon pair. However, we can not rule out that this was
due to some symmetrical pattern in the cytoskeleton of the dividing cell, causing the cytoskeletion of each sibling
to initially resemble the mirror image of the other. The distribution of the cytoskeleton and tensions placed upon it
are important in distinguishing direction of movement in crawling cells. However, this was an ingenious experiment
which also involved tracking cell migration in darkness (since mammalian cells respond to illumination by changing
their behaviour) and it warrants further attention. The importance of quantum coherence in cells remains
controversial and is essentially unknown. It should be born in mind though that biological molecules are not too
large to demonstrate 'strange' quantum phenomena normally associated with sub-atomic particles and it should
also be born in mind that many molecular processes are ultimately quantum phenomena, such as the absorption
of a photon by a pigment molecule (which can not be accurately described in purely classical terms) or the
formation of a chemical bond causing a conformational change in a protein. Quantum biology is a reality, but it
remains to be seen whether or not life has found a way to exploit some of the stranger quantum behaviours.

Superposition of states in DNA

The bases which encode the message of DNA (cytosine (C), guanosine (G), adenosine (A) and thymine (T))
exhibit a phenomenon called tautomerism. In tautomerism, a molecule can exist in more than one chemical form -
the various forms being called tautomers. These forms differ only in the position of a single proton. The proton
may be bound to an imine group, -NH, (imino form) to form an amino group, -NH2, in the amino form or to a
carbonyl or keto group, -C=O, forming an alcohol group, C-OH, in the enol form (in which the C atom attached to
the alcohol group also forms a double bond with another atom as in an alkene).

Possible tautomers of the four bases of DNA are shown below. Can you decide which are amino, imino, keto and
enol forms?
Figure 2: Resistant colonies growing on agar with a normally lethal dose of tetracycline, plotted
against age of the non-stressed parent cultures. Green line, diamonds: control population from
flask C; blue, squares: tetracycline non-stressed siblings from flask B; red, triangles:
tetracycline pre-stressed siblings. Mutations are those that have acquired tetracycline
Results and Discussion

The populations remained healthy throughout the duration of the experiment and spontaneous mutation rates
were negligible

The data are shown in the graph below. As expected, the control population (flask C) never grew on the
tetracycline plates (with the exception of one or two colonies on one occasion). This is shown by the green
curve (diamonds) in fig. 2 below. On plates lacking tetracycline, they grew to confluence. Thus, this untreated
culture remained healthy and viable throughout the duration of each experiment and the acquisition of
tetracycline resistance due to spontaneous mutation was negligible.

Samples from the pre-stressed (B+) and non-stressed (B-) sibling populations also always grew to confluence
on agar plates containing no tetracycline, as expected. This confirms that they remained viable throughout the
duration of the experiment. However, as mentioned below, viability of the cultures seemed to diminish at week 8
onwards as the cultures became old.

The pre-stressed population (B+) became more tetracycline resistant over time

The red line (triangles) shows the mean number of resistant colonies from the pre-stressed population (B+) (of
2 to 3 plates) that grew on agar containing the lethal dose of tetracycline, whilst the green line (diamonds)
shows the number that grew in the control (flask/population C). It is clear that the pre-stressed population had a
higher rate of resistance, as expected since these would be selected for during culture, but also that the
numbers of resistant individuals increased over time, peaking at week seven and then declining, presumably as
the culture medium began to age causing the cells to gradually lose viability. This shows that there would be no
point continuing the trials for a longer period of time without some kind of sustainable culture system with
medium replenishment (such as a chemostat).

The sibling population (B-) to the pres-stressed cells also became more resistant over time despite never
encountering antibiotic and never encountering stressed cells from population B+

The blue line (squares) shows the number that grew in the sibling population (B-) from which samples were
drawn and subjected to tetracycline stress. Note that only those bacteria that grew in the red line case were
exposed to tetracycline, these were the stressed populations (samples), neither the control group nor the
siblings were exposed to antibiotic. Recall that the pre-stressed population (B+) was kept completely separate at
all times from the non-stressed sibling population (B-). The number of mutants selected for increased over time
in both the stressed populations and its non-stressed sibling population. The control group (C) only showed a
response, developing very small numbers of resistant colonies in only one of five replicates and no resistant
colonies were detected in any of the other four control groups.

sibling population showed an increase in tetracycline-resistant cells without being exposed to tetracycline
and the pattern over time mirrors that of the stressed population, except that the actual mutation rates are only
about 30% as high. The same pattern was obtained in all five replicates of the experiment, without fail.
Anomalous mutation experiment - summary of method
Figure 1. The method used was one of serial assay. Two parent cultures, A and B, were
established, representing distant siblings (distant in time). At weekly intervals samples were
withdrawn from each parent culture and split into two sibling populations: { A1, A2 } represent
one sibling group descended from parent population A, { B1, B2 } sibling populations sampled
from parental population B. A1 and A2 were both grown in standard medium as was B1, but B2
was incubated in a sub-lethal concentration of tetracycline (yellow). After incubation, samples of
all four daughter populations (A1, A2, B1 and B2) were plated onto normal blood agar (top row of
plates in red) and blood agar with a lethal concentration of tetracycline (lower row of orange
plates). This sampling of populations A and B was repeated weekly for 9 weeks. No daughters of
population A developed resistance to tetracycline, with only a single exception, representing a
low rate of spontaneous mutation. Siblings of population B stressed by prior exposure or
selection with tetracycline (B2) produced a significant fraction of tetracycline-resistant cells.
remarkably, siblings of this group which never encountered antibiotic (B20 also developed a
significant, though lower, fraction of resistance. the entire experiment was repeated three times
with identical results. Trials 2 and 3 took additional measures to separate populations A and B
into separate incubators, with no observable effect on the results obtained.

Hill, M, 2000. Review: Adaptive state of mammalian cells and its nonseparability suggestive of a quantum system.
Scripta Medica, 73: 211-222.

Davies, P.C.W. 2004. Does quantum mechanics play a non-trivial role in life? BioSystems 78: 69–79.

Article last updated: 14/3/2015
DNA base tautomerism
Above: DNA base tautomers. Adenine usually exists in the amino form (a) but is occasionally found in the imino form
(A*); thym,ine is usually found in the keto form (T) but occasionally in the enol form (T*); cytosine is usually found in
the amino form (C) but occasionally as the imino form (C*) and guanine is usually found as the keto form (G) but
occasionally as the enol form (G*). The -NH imino group at the bottom of each structure has the H replaced by a
sugar residue in DNA. Quantum mechanically, the proton can change position by quantum tunneling its way
through the potential energy barrier which would otherwise deter the switch.

These less common altered forms of the bases have profound effects on DNA. Normally A pairs with T and C with
G. However A* pairs with C, C* with A, T* with G and G* with T. This means that if a base flips into its less common
tautomer during DNA synthesis or during transcription to mRNA, an error may occur and the code may be altered.
Proof-reading may remove this error, but if it doesn't then a mutation or faulty protein may result. For example, an
A, which would normally pair up with T, may flip to A* when it is being read and the complementary strand will then
have a C put in place of the correct T. If this complementary strand is later read, then the error will propagate.

The key is this: quantum mechanics tells us that a base need not be in one or the other state, but a mixture of the
two! That is an A, for example, can also be an A* - the two states, which are quantum waves, become added by
superposition. Such superpositions are bound to exist since the only difference is a movement of the proton
within the molecule, which is not much different to moving an electron within an atom. Such quantum superpositions
are inherently unstable and when they interact strongly with the environment, such as when a water molecule
collides with them, or perhaps when DNA or RNA polymerase reads the base, it will collapse into one or the other, A
or A*, with A being more likely. The jostling about of water molecules due to thermal motion is expected by many to
make such states short-lived. However, it is becoming ever more apparent that quantum superpositions in cells may
be much more stable than previously thought, since the electromagnetic fields of large macromolecules, such as
DNA, tends to cause a local crystalline order in the water near the macromolecule, such that little jostling actually
occurs. How stable these states are exactly remains to be determined.

There is another additional relevant quantum phenomenon related to superposition -
quantum entanglement.
When a DNA strand is being read by DNA polymerase, a hypothetical possibility is that the template base and the
complementary base added to the complementary strand could both be in a state of superposition after pairing.
This means that even after DNA synthesis is complete we have to consider the possibility that the identity of the
base (e.g. A or A* pairing with either T or C respectively) remains uncertain or non-determined. It is not obvious
whether or not an entire base-pair can be in a state of superposition, such that we can not be certain whether we
have A-T or A*-C, or whether the uncertainty resides only on the adenine (A or A*). Requiring a superposition
between C and T clearly requires extensive chemical modification and it is not intuitive to expect C to T transitions.
In either case a mutation may be passed on (if a strand containing A* is read by DNA polymerase then this A/A*
complimentary strand may have a C inserted instead of a T). What is more relevant is whether or not that
uncertainty can persist in two daughter DNA molecules that have been recently synthesised from the same parental
DNA molecule. It has been suggested that the wavefunctions of daughter DNA molecules may overlap during
synthesis such that the two daughter strands are created in a state of quantum entanglement. However, this is
possibly not the case due to the semi-conservative nature of DNA synthesis. Let us assume that a normal parental
DNA molecule undergoing synthesis has the A read whilst in the A* state. An A* on one parental strand will result in
a G being inserted in the complimentary strand, giving an A*-G pair. However, we might expect the other parental
strand to have the normal T and then one daughter cell would inherit the usual A-T pair. It should be remembered,
however, thatr entanglement is not an intuitive phenomenon and ultimately the question as to whether or not DNA
molecules can be entangled must be answered empirically.

The potential mechanism is not clear, but let us draw an analogy with photons. When photons are created in pairs
they are entangled from birth, such that the overlap in their wave functions means that they are governed by the
same combined wave function (their two wave functions superpose in some way) as a result of the quantum
indistinguishability of two particles in close proximity (it is impossible to say which photon is which). Similarly, the two
electrons of the helium atom have significantly overlapping wave functions due to their close proximity and we can
never say which electron is electron as we have instead a kind of two-electron entity which leads to observable
properties in helium atom spectra. With photons, this entanglement can persist as the two photons fly apart in
opposite directions and can persist indefinitely over a limitless distance until one of the photons interacts strongly
with another system. We know that such photon pairs are formed with opposite polarity, e.g one is horizontally
polarised (state H), the other vertically polarised (state V). However, whilst entangled both photons are in a
superpositional state and so are a mixture of both H and V polarisations. Upon interacting strongly, for example
passing through a polarisation measuring device, one of the photons will then lose its superposition and its
polarisation will become either H or V. Simultaneously, its partner will assume the opposite polarity, no matter how
far apart they are; this has been called 'ghostly action at a distance' and all current empirical evidence suggests the
reality of entanglement. (Entanglement generally becomes weaker over time and distance however). Similar
entanglement has been found in large molecules (including small diamonds), so sceptics take note!

So, suppose that sibling DNA molecules can become entangled (we await proof of this or otherwise). What are the
implications on evolution, e.g. the evolution of antibiotic resistance? Suppose possessing A* (equivalent to G)
instead of A conferred a beneficial mutation that gave a bacterium increased resistance against an antibiotic.
Clearly, those cells in which the superposition collapses into A*, though in the minority, will have a distinctive
survival advantage. If siblings were entangled then this would also determine the base in the sibling cell and its level
of resistance. This could account for observations in which siblings of stressed cells adapt to that stress at higher
than expected rates even though they are not themselves stressed. This would be of clear evolutionary advantage
in a population by preparing the population for likely stressors.

What could cause such a wave function collapse? Is there a potential mechanism by which exposure to antibiotic will
cause the sequence of the DNA to be determined, whether including an A or A*? Supposing our A/A* was on a
quiescent gene which was switched off until antibiotic stress was encountered. Once the gene is transcribed its
state would be determined by the strong molecular interaction with the RNA polymerase molecule and it would
presumably be read as either an A or an A*. Clearly by the time we have a functional protein there can be no doubt
which amino acid sequence it possesses. Once as an A* the base will remain as an A* if measured again soon
after, since A* would be a stationary eigenstate. Frequent repeated 'measurements' would maintain the base as A*.
However, if left to its own devices then it could eventually return to a superposition A/A* state.