April 14, 1952
My dear Hayes:

Your letter of the 8th has just now arrived. I judge that you will but
recently have received mine of the 5th. This crossing of ideas as well as
correspondence is coming to be a regular thing.

I think my letter may have anticipated some of the points raised in yours
(or vice versa). However, my note, upon a re~reading appears rather obscure,
so that I will try to clarify it. I shall not be surprised to find that this
messgge will again cross yours, but this kind of exchange of ideas is abso-
lutely indispensable.

First of all, I don't think a detailed comparison of experimental protocols
is essential. There is no doubt we have all run into the same thing, both with
respect to the differential fertility of F+ x F~ as compared with F+ x F+, and
the effet of F-polarity on the segregation ratios. I noticed this about 3
months ago, but was rather puzzled that BM-F— x TL-F+ was much more fertile
than BM-~F+ x TL-F+ or F- which are about the same. With some further compariscns
using other BY and TL stocks, I've come to the conclusion that there is a gradi-
ent of F+ potency. Of the stocks you have, the F- would be a zero grade,

58-161 would be +1, and TL-F+ +2. The fertility of a combination would depend
on the degree of separation of the parents (along the lines of the Hartmann
theory of relative saxuality in the algae). In addition, S” cells generally
can function (almost) exclusively as relative F- . This would account for

the very low yields of BM-F+ S- x TL-F+3S" on streptomgcin agar, as compared
with the cross x TL-B-ST (\/~1177), There is also an effect of segregation
ratio of the S’ factor, but this is not enough to account for the differehce.

As to the segregation ratios, the effect of F-polanity was under our noses
almost three years but (as usual?) we nissed it until the compatibility story
came to light. Filial TL- cuktures extracted frombheterozygotes were observed
to give sxkts inverted segregations in crosses with 58161. I thought then that
this was evidence for chromosomal aberrations of a simple sort (that the
filial stock had acquired the chromosome order of the other parent). Of course,
there was no more to it than the F+ character of the filial. All of the results
(yours and curs) are compatible with the assumption that the gametic contribu-

tion of the F+ parent is defective in some way. We have some stocks which are
of higher Fa+ grade than TLF+, and in these crosses it is again the TL parent's
contribution that is favored among the protetrophs.

I hope that you will underatand this counter—Machiévellian gambit (shall we
cali this "Sciencemanship-- according tc Potter"?). But seriousky, I am sure
that after we understand each other's terms, we shall fand ourselves in mech
closer agreement that might appear at first.

There is no doubt that there is some unorthodoxy fin the sexual mechanism of
E. coli, and you are certainly missing no opportunity to expose it. My experi-
ence with persistent diploids has given me a certain orientation as to the
probabke nature of this peculiarity. You may or may not agree with xk the
idea, but I think it more important we don't get into a merely semantic
embroil.

We are agreed that the zygote is defective, especially with respect
to the genetic contribution of the F+ parent. We can resolve our discussion
in these terms: is the defect large or small, compared to the whole genotype;
does it originate prior to gamete formation (as I believe you hold), or subse-
quent to the union of complete gametes? The first question is more readily
answered, as the independent evidence from the diploids is quite clearcut.

Faced solely with the prototroph results, and the effect of F-polarity, I
probably would have entertained the same hypothesis as you suggest, that recom-
bination is restricted to small blocks of genes. Of all the markers that have
been tested, however, only two have proved to be deficient in the diploids:
the linked pair- Mal and S, All others have regularly appeared in heterozygous
condition from appropriate crosses. When the Het factor is present, some 10%

of the prototrophs are usually diploid, and in their general behavior they re-
semble the prototroph sequence, so that I have been confident that the diploids
are valid representatives of the zygotes from which the prototrophs normally issue.
It requires the consideration, simply, that K-12 has just one chromosomes to see
how a defect limited to a relatively small region can pebburb the apparent segre-
gation behavior of all the other markegs: the defective chromosome is simply
inviable in the haploid condition. I have fairly direct evidence that F-polarity
influences the choice of chromosomes to be eliminated, ile., determines whether
the diploids of a given cross will_be BM-TL+Lac+Vj5 Mal+S°/BM+TL-Lac-V,".. ee

or BM-TL+Lac+V,°.. . ./BM+TL-Lac-V, “Mal-s". This is a mysterious phenomenon, but

its unorthodoxy is at a somewhat different level from the question of sexuality.

In a very few instances, I have obtained diploids that webe heterozygous flor all

of the markers.These in turn have been used in diploid x haploid crosses (via a few
obscure tricks) in which the elimination does not seem to occur, so that the
derived diploids are, in turn, "complete".

The timing of the defect is somewhat more difficult to be precise about. I
had worried about the possibility that the defect was primary (1.e., your hypo-
thesis). The evidence on which I had decided that it probably was not is not com
pelling, but it still makes sense to me. This has to do with the occurrence of
crossing-over in the fopmation cf the diploids. Some of them are, for example,
Lac-/Lac— instead of the expected Lac-/Lac+ [there is rather detailed proof that
they are diploid homozygous Lac~/Lac~ rather than hemizygous Lac-/.. like the
Mal-1. These can be accounted for by the occurrence of crossing-over within the
2ygotea prior to establishment. of the persistent diploids. A similar sort of thing
seems to happen in the defective segment, so that somes of the diploids, still hemi-
zygous for this segment are Mal+S™ or Mal-S° instead of the expected arrangements.
A gamete defective for the whole sapment could not engender such crossovers. On the
other hand, if they debive from "less defective" gametes, why don't these give
types hemizygous for S, but not for Mal, or vice versa? Such types simply have not
occurred. This is the main reason I have pesferred to think of a "segmental elimina-
tion", rather than a primary defect, but I may be wrong.

All of this complexity is a bit hard to take, and I have often wondered whether
I might not be on the wrong track altogether. But I can interpret the few complete
diploidgs' bshavior in no otherway than a straightforward sexual process. That this
might gc wrong in details is obvious, but it stands as the fundamental pattern.
This discussion hss reached a point where wm even s morphological demonstration
of the sexual process would hardly alter the issue (unless of course it turns out
to be in no wag a typical mating reaction). In the present state of our knowledge,
descriptive accounts are going to s&&ind up better than premature interpretations,
Do we know enough yet of the biology of transformations to be very secure in evolu-
tionary comparisons with sex? I am afraid that much more has been written into
my descriptions of K-12 recombination as a sexual process than I had intended, but
I knew of no better context to described what one could accomlish by way of a
genetic analysis. That is, to my mind, all of our discussion has concerned the
description of the sex process that K-12 happens to have. Whether you want to call
something a gene-donor or a gamete is immaterial, so long as you have a clear
concept and description of its behavior. I've said all this before, in my 1947 paper.
If some of the gametes are partially defective, so be it, (but I don't think so}.

Sincerely,

Joshua Lederberg