ped

May 3, 1953

Dear Luca:

I have been able to make a hasty review of your draft. As you say, the
writing is very rough, but I have no qualmds about your ability to write
excellent English prose (viz. your WHO paper). While I would welcome an op-
portunity to go over the final paper, do not hesisate to submit it at your
own judgment if necessary. I have a little difficulty in viewing the paper
as a whole, and my comments now are on particulars.

p. l: the term genetic adaptation is satisfactory and simple; I would think
‘pre-adaptation" was more expressive if coupled with your explanations. In
fact, one may contrast genetic pre~adaptation with genetic and physialogical
post—adaptations respectively.

p. 2: I think too much weight is given (as an immediate response, no doubt] tp
Ydudkin's position. I think it can be expressed as an attempt to obliterate the
distinctions by postulating so unstable a genotype that genetie local changes
could not be perceived. I would give it the statis of an ad hoe compromise, rather
than equal weight with I and III. Your comment on II and III as oversimplifica-
tions would be a good introduction to the necessdity for detailed analysis:

"III can also be phrased in terms which would include the pre-adaptation theory.

If the population is studied as a whole, and no attempt made to resolve its elements,
the enzyme balance of the mass will doubtless be found to show adaptive adjustments
of the kind subsumed under this hypothesis. For an underatadding of the genetic

basis of such changes, howver, we t peumedx mist examine the adaptive responses
of individual cells, and we shall that the apparently directed, postadaptations
of whole populations are best interpreted in terms of pre-adaptive changes of

small numbers of its constituent cells.

p.2 bottom or intra~- vs. inter-clonal variation

p.3 The suggested figure is an excellent idea. However, I would distingu&ah
1A and 1B vs. 2A and 2B (instead of 1,2,3, 4) to make later reference easier.
It should be pan— pointed out that, with reference to survival in presence of
drug, 1A and B are not meaningfully different, but that other criteria may ulti-
mately be found by which they can be distingusihed (elg. responses to much higher
levels of drug). Eagle's work would be put on p.3, if anywhere, as a possible
case a >proaching §. I would ignore Sevag, as gibberish. Instead of “distinction
between models 3 and 4 may seem possible", elementary.

p.3 Clonal appearance of mutants: pedigree— along lines of Zelle's study of
smooth-rough variation in Salmonella typhimurium (J. Inf. Bis. 71:131-152 1942)

Why not "variance analysis", and a "crude variance analysis" in place of
"fluctuation test"

I would ABtsay "fully vitdated by"--*subject to*# is a sufficient criticism.

Is not the denial of a Po&sson distribution seimply a formal satement of
Hinshelwood's objection? In consbructing the Poisson series, one does not
have to postulate an equal a priori probablity, but simply a low probability
and one that is independent of neighbours (environmental effects) and sibs
(heritable). I.E., when I sample the populations, I have no coneern for the
a priori probabilities, but need only consider the result. ‘ith large humbers
of cells, and a low expectation per trial, can one not ignore the a priori
probabilities and simply consider the total successes/ total trials? I am
not expressing this very well, and may perhaps be wrong in my thinking (of
which I must do some more on this question). There has been no discussion of
this in the preseht cantext |[perhaps becuase of the consideratiom given], but
AM May 4

there is something very mich like it in the discussion of mechanisms of
exponential disinfection: See Irwin, 1942 J HY¥g. 42:328; Chick 1930 in

Brit. M.R.C.: System of Bacteriology, and discussion by Topley&Wilson,

Princ. Bacteriology & Imnunity,1946ed. p.140. I do not believe there has

been a serious attempt to combine the stochastic and distributional approaches,
as a realistic theory perhaps should try.

After a night's relaxation, I am fairly sure that my objection is correct.
The "fluctuation test" is a comparison of intra~ and inter-culture variances.
In the absence of some correlating factor (heredity or environment) these variances
should be the same. The problem would then reduce to the question of a non-
Poissonian distribution of samles from the same culture. It is beyond the immeddate
limits of my ingenhity to devise a distribution of cells from which properly
drawn samples will show other than sampling error. If a non-Poisson distribution
were found, we would immediately suspect either the sampling process, or the assay
method as the source of the heterogeneity. One would, I presume, get a negative
binomial distribution af the samplee-were sample sizes were not uniforms&f, but
distributed in a logarithmic series. However, one can get a negative binomial out
of almost any data, since one has an extra parameter. I do not propose that this
analysis #s incorrect, but only that it is a restatement of the others.

I am a weak biometrician, and perhaps for this reason I am always suspicious
ofvthe generality of any biometric test. (This applies to kinetics as well). One
can often disqualify the npll hypothesis, but it is often difficult to predict
whether a major or a trivial modification of the null hypothesis will be needed.
The present discussion (i.e. of Luria&Delbruck)is an excellent example.

To return to p.4: I do not think Newcombe's method should be discussed at all
as a fluctuation test, since it is not the variance that is important, but the
more direct demonstration of statistical clones. Of course the spread plates
will show a high variance, and the unspread a low one, but 46 it is not the
difference in variance, but the difference in mean, which is emphasized.()This is
unimportant). I do not see that the objection can be taken very seriously (effect
of spreading): the inoculum is so large that there is a confluent growth whether
or not the plate is respread, although I suppose there is still some local inhome
geneity. Would it helped the argument to have added the phage first, and respread
after, perhaps, 20 minutes when the seen sensitive cells can be shown to be already
doomed? My only reservation to this proof is that it still relies on a statis-
tical demonstration of the clones, and does not isolate them as such.

p5:Unless we are referring to different work, Law(Nature 4/12/52),as I recall,
studied resistance in vivo, and showed that a cluster of resistant cells occurred
in one line of an unselected tumor. My memory may not be right.

p6. I would prefer not to describe the other medhods summarized here as "indirect
selection", although this would be perfectly legitimate. Sib selection would have
beenf” perhaps a better mme for the replica-plating result. For¥ the present,
could you write "Drug-rests&ant variants have also been securedppy selection for
correlated characters .... yt theabsense of the drug:
Should one add here experiments which might be done baséd on the linear accumlation
of mutants in the chemostat (Novick and Szilard?). Unfortunately, there is very
little with drug resistance along this line, Bryson and Szybalski having disregarded
this opportunity/ to use their turbidostat.

pe7 I am greatly disappointed that Roger Stand#@r should have quoted this work

of Doudoroff's in the present context. If one cons&ders that the toxic stimulus is
not salt concentration, but rapid change in salt concentration, there is no
meaningful physiological adaptation at all. This has been borne out more recently

by Anderson's experiments on osmotic shock with phage (Ann Inst Pasteur Jan. 53).

It is doubtful whether a free phage particle could be thought of as using an adaptive
mechanism. I would either leave out this reference, orgexplain the detailed interpretation.

To 7, after Dauermodifikationen, I would add: Some caution must be exercised in postu-
lating similar mechanisms for bacteria, etc.# It is important to keep in mind that

the distindtive feature of Daaermod., the gradual loss of resistance on later cultivation
---- is here a property of individual lines of descent, and that experiments of a similar
sort but based on tnresolved populations may reflect a gradual change of the population,

but a sudden, discrete mtational change in a few individuals subsequently selected¥ But
Rerhaps you make this point sufficiently later on,

p7 bottom I would prefer you nqt quote my very casual experiment with chloraminobenzoate
(Strandskov) since Demerec! published statament is available. If you like, you can quote
my review as dealing with the general question.

p8 I would put Eagle in after Baskett on p.7 with the proviso that no test was made of
the heritability of the adaptation. In fact, Doudoroff's case can be used as a good example
where two mechanisms are operating (though the physiological can be vaewed as either

simple or spurious), as may well be the case also with Eagle, and a good deal of Hinshel-
wood's material.

p8: girect genetic effects.... I am rather fond of this section, and pleased you are

able to work it in. Would it be better to put Dauermodifikationen after thig, instead
of as above?

p9zline 32 of mating between compatible éells* instead of fusion.... I would omit
sentence on inhewitance of F+, but if you like the JG¥ paper can be inserted as an
addnl. reference for the whole story. lLwoff will not like expression "lyogenic phage"-——
I don't care. temperate may do as well. In fact, "lytic" phage will also transduce, if
the recipient is protected by previous infection with the temperate.

p 10. middle on S?...: Excelleht!
ll: azide: but Az™ x S™ has been used.

13: I had forgotten about colicin as an antibiotic. I think it would be worth while

to say two or three lines, expecially about the qualitative specificyt of the different
resistance types in recombination tests. Almost anything is a modified phage these
days; 1 shouldn't be surprised if streptomycin is claimed to be such sometime !¢

13: Tatum long ago worked oyt the inhibition of strain K-12 by Valine, and its antago~
nism df isoleucine (a better statement than complete medium ), had isolated valine-resis-—
tant mutants and done just a few crossing experiments with them. Unfortunately, I do not
have the details, and doubt if they have been published. One locus was found, I think, to
the right of TL but I haven't heard any? more about it in 5 years. Probably, it would be
better not to quote any of this without a reference, but for my own curiosity I will
question Tatum about it.

13: In Salmonella SY transduction ca 100-1000X control.

I am very much disturbed by the repeated citathih of the pneumococcus transformation
as an example of directed mutation, by way of a justification of drug-induced adaptations
(cf. the SGM symposium; Hobby in Bact. Revs. March 1953; originally perhaps largely due to
Dobzhansky: see Amer. Natur. 87:123, v. recently, as well as his book). For this reason,
as much emphasis as possible should be given to the pn.tr. as a me species of genetic
transfer. I think this is already implicit in your discussion; I do not know whether more
should beput in in the beginning.

15: ~Hughes- OK. G4rd critique of teehnique should perhaps be téned down, but not excised.
Could some of his variations have occurred in subsequeht clones to the initial isoilion?

16: I am not sure Jf I understand the last} 6 lines. Have you noticed 5° reversion to S%
in W-1177? Was it not before there were only 3% ST that you noticed a discrepancy in
counts with and without am? iDon&t put this in paper, but repliea—phating is of course
ideally suited to finding this kind of mtant] Have you studied such reversions genetical
ly? One could add that 3% ST grades continously} into S¢; it would be worth seeing
whether a small amount of streptomycin would accelerate the growth rate. If so, there

would be no special point to studying reversion to §° in the difficult case, when more
typical sd would give a more direct answer. sanee S°-»S° has been measured in several

cases (Demerec; Berta&h; Catlin}/ and, I think, Newcombe).

17 P2: This is a very important paragraph. I fully agree with your conclusions. The
emphasis might be that we should by no means ignore the more difficult cases, but that
it has not been shown that any of these are in fundamental contradiction with the genetic
theory. I think it is questionable whether extra~nuclear effects play an appreciable
rule in adaptations of the protozoa. Une tends to emphasize the unique.

18 blencorrhage = gonorrhea?. I do not think we have to quote specifie cases, but
this depends on whether drug-resistance will be considered at the sympoisum from any
other viewpoints.

18-19 In the April 1953 Jour. Bact. Morton and Klédn Bhve an extensive discussion of
synergism (which they claim is always genetic {and 1 think doubtful]) and anat-
antagonism (always, of course, physiological). Unless you know of an earlier paper
of theirs to quote on the sans theme, I think it would be best to regard this as
coming too late.

This has all the makip@s of an interesting work, Luea, and I am only sorry that so

many circumstances impose this rush. Perhaps we should, after completing this, underkake
to do a more thorough job together, But let us postpone discussion of it untile we can
do so in person.

Back to p. 5: in Dittrich's expt: membrane filters/ cellophane sheets.

In second paragraph, the wording is especially difficult. I suggest:
"Another approach, based on clonal appearanee occurrence of resistant mutants, depends
on secondary differences sometimes observed among divfey¢ At independent mutations for
resistance to the sama drug./ If resistants appear in clones, such differences...."

a ae mene on te ene

You will have noticed a minor error at the heading of one of our tables (7 should be

+ or . , 4 forget which. I% can be corrected easily in reprints; perhaps you should
Notify effitops to include it in Corrigenda. In view of the symmetry of the table, it
should confuse no one who gces as far as to dtudy it.

|Aeat)s alas, hastily,

a

NES
/, Joshua Lederberg