Rearranging the centromere of the human Y chromosome with phiC31 integrase.

Abstract

We have investigated the ability of the integrase from the Streptomyces phiC31 'phage to either delete or invert 1 Mb of DNA around the centromere of the human Y chromosome in chicken DT40 hybrid somatic cells. Reciprocal and conservative site-specific recombination was observed in 54% of cells expressing the integrase. The sites failed to recombine in the remaining cells because the sites had been damaged. The sequences of the damaged sites indicated that the damage arose as a result of repair of recombination intermediates by host cell pathways. The liability of recombination intermediates to damage is consistent with what is known about the mechanism of serine recombinase reactions. The structures of the products of the chromosome rearrangements were consistent with the published sequence of the Y chromosome indicating that the assembly of the highly repeated region between the sites is accurate to a resolution of about 50 kb. Mini-chromosomes lacking a centromere were not recovered which also suggested that neo-centromere formation occurs infrequently in vertebrate somatic cells. No ectopic recombination was observed between a phiC31 integrase attB site and the chicken genome.

Full text

Page 1

Rearranging the centromere of the human
Y chromosome with fC31 integrase
Sunir Malla, Felix Dafhnis-Calas, John F. Y. Brookfield, Margaret C. M. Smith
and William R. A. Brown*
Institute of Genetics, Queen’s Medical Centre, Nottingham NG7 2UH, UK
Received September 4, 2005; Revised and Accepted October 6, 2005
ABSTRACT
We have investigated the ability of the integrase from
the Streptomyces fC31 ‘phage to either delete or
invert 1 Mb of DNA around the centromere of the
human Y chromosome in chicken DT40 hybrid
somatic cells. Reciprocal and conservative site-
specific recombination was observed in 54% of
cells expressing the integrase. The sites failed to
recombine in the remaining cells because the sites
had been damaged. The sequences of the damaged
sites indicated that the damage arose as a result of
repair of recombination intermediates by host cell
pathways. The liability of recombination intermedi-
ates to damage is consistent with what is known
about the mechanism of serine recombinase reac-
tions. The structures of the products of the chromo-
some rearrangements were consistent with the
published sequence of the Y chromosome indicating
that the assembly of the highly repeated region
between the sites is accurate to a resolution of
about50kb.Mini-chromosomeslackingacentromere
were not recovered which also suggested that neo-
centromere formation occurs infrequently in verte-
brate somatic cells. No ectopic recombination was
observed between a fC31 integrase attB site and
the chicken genome.
INTRODUCTION
Manipulation of large scale chromosome structure, often
referred to as chromosome engineering, has become an impor-
tant tool in experimental genetics. In mice, deletions and
inversions are being used in screens for recessive mutations
(1) and translocations or inversions (2) are being exploited as
models for the mutations underlying various human genetic
diseases, particularly those involved in somatic malignancies.
In Drosophila somatically induced mitotic recombination is
used in mosaic analysis of the effects of mutations that are
lethal or pleiotropic (3). Construction of such rearrangements
requires the use of site-specific recombinases. In all cases, of
which we are aware, chromosome rearrangements in metazoa
have employed either the Cre recombinase isolated from the
bacteriophageP1orthe FLPrecombinaseisolated fromthe 2m
plasmid of Saccharomyces cerevisiae.
The Cre and the FLP recombinases are members of the
integrase family of site-specific recombinases. These proteins
use tyrosine as their active site nucleophile and the recomb-
ination reaction proceeds through a Holliday junction inter-
mediate (4). Both proteins catalyse reversible reactions
between identical sites of 34 bp in length, termed loxP sites
in the case of Cre. The reversibility of the reactions however
limits the utility of these enzymes. In particular it is difficult to
use either Cre or FLP to promote inversions or translocations
irreversibly except by supplying the recombinase transiently
and selecting for the desired product. It is technically straight-
forward to do this in tissue culture but impractical in whole
organisms. Although some have suggested that this type of
problem may be overcome using mutant target sites we are
unaware of any reports of the successful use of such sites in the
engineeringoflargescalechromosomerearrangementsinvivo.
A second, more practical problem is posed by the toxicity and
mutagenicity of Cre (5). Although widely ignored it is now
clear that Cre has growth inhibitory effects and the ability to
cause widespread DNA damage when expressed at high levels
in mouse cells (5). The damage is thought to arise as a result of
interactions between Cre and sequences similar to loxP;
termed loxP pseudo-sites, in the mouse genome. These unde-
sirable properties mean that Cre expression may have side
effects in some live animal studies and in general that Cre
will only be able to be used at levels which are toxicitylimited.
A unidirectional, non-toxic alternative to Cre would therefore
be generally useful.
*To whom correspondence should be addressed. Tel: +441158493244; Fax: +441159709906; Email: william.brown@nottingham.ac.uk
Present address:
Margaret C. M. Smith, Institute of Medical Sciences, University of Aberdeen, Aberdeen AB25 2ZD, UK
? The Author 2005. Published by Oxford University Press. All rights reserved.
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access
version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press
areattributedastheoriginalplaceofpublicationwiththecorrectcitationdetailsgiven;ifanarticleissubsequentlyreproducedordisseminatednotinitsentiretybut
only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org
Nucleic Acids Research, 2005, Vol. 33, No. 196101–6113
doi:10.1093/nar/gki922

Page 2

A second family of site-specific recombinases uses serine
as the active site nucleophile (6). The best characterized
members of the serine recombinase family are the gd resolvase
and Hin invertase. These proteins bring two recombination
sites together to form a synapse in which all four DNA strands
are cleaved and in which phosphor di-ester links are formed
between active site serine residues and each strand of the
participating DNA molecules. Strands are exchanged as a
result of a ligation reaction that occurs after relative rotation
of the two halves of the protein–DNA complex (7). The serine
recombinase family has been divided into at least three
different classes of protein that differ in their domain organ-
ization and requirements for accessory proteins (6). The large
serine recombinase class includes members that promote
site-specific recombination without requiring any accessory
proteins (8). The integrases encoded by the Streptomyces bac-
teriophage fC31 (9) and Mycobacterium ‘phage BxbI (10)
integrate and excise the respective viral DNA molecules
into and from the host chromosome and are the best under-
stood examples of this class of protein. The integration reac-
tion promoted by each of these proteins occurs between two
sites, termed attachment (att) sites; attB and attP, about 40 bp
in length to yield product sites; attL and attR, and is promoted
bythe integrase acting alone. The excision reactions which use
attL and attR to reform attP and attB are thought to require the
integrases and accessory proteins. Thus these integrases acting
alone promote unidirectional site-specific recombination
reactions and are, thus, potentially ideal for promoting pre-
cisely those reactions for which the current set of reagents; Cre
and FLP, are unsuited. In addition there are many members of
the serine integrase family (6) and they probably have sim-
ilar general properties. Several, including fC31 integrase,
have been shown to be able to promote site-specific integra-
tion in eukaryotic cells (11–14). However it has been reported
that vertebrate genomes contain large numbers of pseudo
attachment sites for fC31 integrase (15). Pseudo-sites
would have the potential to compromise the specificity of a
desired manipulation. As in the case of Cre, pseudo-sites
could also act as targets for the ectopic activity of the integ-
rase and thus could give rise to a mutagenic background.
Such concerns may have discouraged others from using the
fC31 integrase and other serine recombinases for genome
engineering.
The centromeres of the human sex chromosomes have been
extensively mapped and sequenced and thus are structurally
the bestcharacterizedvertebrate
Here experiments are described in which the integrase of
the Streptomyces fC31 ‘phage has been used to manipulate
the long range structure of the centromere of the human
Y chromosome. Chicken DT40 cells were used to engineer
a mini-chromosomederivativeofthe human Ychromosome to
contain attB and attP sites flanking the centromeric alphoid
DNA. The sites, separated by ?1 Mb, were introduced in
either respective orientation and this enabled investigation
of the ability of fC31 integrase to promote either deletion
or inversion of the centromeric interval. It does so acc-
urately and efficiently in about half of the reactions but in
the rest of the sites are damaged during the rearrangements
and either fail to complete the reaction or rearrange
unpredictably. Pseudo-sites were not detected in the chicken
genome.
centromeres(16,17).
MATERIALS AND METHODS
Plasmid construction
Plasmids were constructed by standard methods. In each case
the att sites were engineered into plasmids using synthetic
DNA provided by Invitrogen. The plasmids were checked
by restriction site mapping and in all cases by sequencing
across the att sites. The hygromycin resistance gene used
waspresentin thecounter
thymidine kinase fusion (HyTk) (18). The puromycin resis-
tance gene (19) was a gift of Jean-Marie Buerstedde. The
CCAG promoter (20) was a gift of Ian Chambers (Edinburgh
University). The nuclear localization signal used to tag the
fC31 integrase (8) was derived from the large T antigen of
SV40 virus. It included the residues MPKKKRKV and was
placed at the caroboxy terminus.
selectablehygromycin–
Cell culture
The starting mini-chromosome D2.5 was originally described
elsewhere (21). It was moved from Chinese hamster ovary
cells, in which it was isolated, into DT40 cells by microcell
fusion (22). DT40 cells and DT40 somatic cell hybrids were
maintained and electroporated as also described previously
(22) except that the medium used was RPMI 1640 including
446 mg/l L-alanyl-L-glutamine with 10% foetal bovine serum
(FBS),1%chickenserum,10-5M2-mercaptoethanol,10U/ml
penicillin and 10 mg/ml streptomycin. This medium gave
cleaner selection after addition of antibiotics than our earlier
DMEM based medium. Following electroporation with the
targeting or fC31 integrase expression constructs the cells
were plated out in 96-well dishes. Selection was applied
18 h later. Colonies were isolated after 12–14 days. The pro-
portion of cells expressing eGFP was determined using a
fluorescence activated cell sorter.
PCR,fluorescent insitu andfilter hybridization analysis
Conventional agarose, pulsed field gels and filter hybridization
were as described previously (22). The sequences of the pri-
mers used in the PCR to check the breakpoints in the fC31
mediated rearrangements are given in Supplementary Table 1.
Fluorescence in situ hybridization (FISH) was by standard
methods as described (22) except that the alphoid DNA
was directly labelled with Alexafluor-594 dUTP and the
second sequence was labelled with biotin and detected
using Streptavidin conjugated to Alexafluor-488. The Alex-
afluor conjugates were from Molecular Probes. These
improvements led to a technique that was faster, more sensi-
tive and gave less background than before. Restriction
enzymes used in the eGFP analysis of the attP and attL
sites were Asp718 and BglII and in the puromycin analysis
of the attB and attR sites were NcoI, Asp718 and BglII.
Targeting sequences were constructed by PCR from cloned
DNA and the respective primers are given in Supplementary
Table 1.
Estimation of reaction rates
We can estimate the combined rate of the two possible reac-
tions by considering the reaction as follows.
Let us consider an initial cell P and two daughter cells I and
II. There are three things that can happen to the cell as it goes
6102 Nucleic Acids Research, 2005, Vol. 33, No. 19

Page 3

from P to I or from P to II. It can either reach final state A,
or reach final state B, or can remain uncommitted. Let us
call the probabilities of these three outcomes PA, PBand
1 ? PA? PB. We are interested in the state of the final
descendants of the population of cells started from P. In par-
ticular, we are interested in the mean and variance of the
proportion of cells that are of type A. The mean is obviously
PA/(PA+ PB). Let us suppose that the expected variance
between replications is V. What is the contribution of cell I
to this variance? If the cell is of type A, then all its descendants
will be of type A. If I is A, then half the population will
inevitably be of type A, whereas, on average, a proportion
of PA/(PA+ PB) of I’s descendants would be expected to
be type A. Thus, I being A contributes {0.5 [1 ? PA/
(PA+ PB)]}2to the overall variance, or [0.5 PB/(PA+ PB)]2.
But the probability of cell I is of type A is PA. Thus, the
contribution to the overall variance created by cell I changing
to A is 0.25 PAPB
to the variance of cell I changing to B is 0.25 PBPA
(PA+ PB)2. If, however, I has remained undifferentiated its
expected contribution to the total variance is V/4. Thus, in
total, the contribution of cell I to the total variance is
2/(PA+ PB)2. By symmetry, the contribution
2/
0:25½ðPAþ PBÞðPAþ PBÞ=ðPAþ PBÞ2
þ ð1?PA?PBÞV?
¼ 0:25½ðPAPB=ðPAþ PBÞ þ ð1?PA?PBÞV?
The contribution from both cell I and cell II is thus 0.5 [(PAPB/
(PA+ PB) + (1 ? PA? PB)V].
But, by definition, this is also V, so
0:5½ðPAPB=ðPAþ PBÞ þ ð1?PA?PBÞV? ¼ V
Call (PA+ PB), x, so that mean A is PA/x and mean B is PB/x.
Now PAPB/x + (1 ? x)V ¼ 2V and so V(1 + x) ¼ PAPB/x.
However, if we sample binomially N events where the
probability of A is PA/x and probability of B is PB/x, the
expected variance is PAPB/Nx2. Thus, in our distribution by
comparing the mean with the variance we can estimate N, the
effective sample size given x. And N can tell us what x is in the
following way.
Since V ¼ PAPB/[x(1 + x)] ¼ PAPB/Nx2, it follows that
N ¼ (1 + x)/x.
For example, if the change to the final state was always in
the first division), then x ¼ 1 and N is 2, the variance would be
determined by binomial sampling of the two cells I and II. Any
lower rate of resolution (x) will increase N and thus decrease
the variance.
RESULTS
Strategy
We used sequence targeting and telomere directed chromo-
some breakage in DT40 hybrid somatic cells to create two
mini-chromosome derivatives of the human Y chromosome in
each of which the centromere was flanked by attB and attP
sites (Figure 1A and B). In one: CA4 (Figure 1C), the sites
were placed in an opposite orientation with respect to one
another and in the other: XP4 (Figure 1D), the sites were
placed in a parallel orientation. The attB sites were first tar-
geted into the centromeric array of alphoid DNA by telomere
directed chromosome breakage of a mini-chromosome called
D2.5 (21) to generate a mini-chromosome called DD16
(Figure 1B). The attP sites were then targeted into the
DYZ5 array using replacement constructs (Figure 1B).
Restriction analysis established that the attB site was at the
end of a 160 kb alphoid DNA fragment while the attP site in
CA4 was 420 kb from the centromeric end of the DYZ5 array
and the attP site in XP4 was only 40 kb from the centromeric
end of the array. The distance between the alphoid and DYZ5
arrays is 700 kb and thus the distance between the attB
and attP sites in both CA4 and XP4 is ?1 Mb. Details of
the routine manipulations and mapping are described in
Supplementary Data.
In order to help identify any chromosomes that had been
rearranged as a result of site-specific recombination we placed
a promoterless eGFP gene adjacent to the attP site such that
site-specific recombination would place the eGFP gene
(Figure 1B) under the control of the b-actin promoter origi-
nally upstream of the puromycin gene in the attB containing
construct. We used a hygromycin resistance gene fused to a
herpes simplex virus thymidine kinase (HyTk) gene driven by
a CCAG promoter (20) as the marker to select for stable
transfectants containing the attP site. We ultimately intend
to use the engineered mini-chromosomes for experimental
substitutionofcentromericsequencessoweflankedthecoding
region of this marker with a loxP site and an attP site for the
fBT1 integrase (23). As we will describe elsewhere the pres-
ence of the loxP and fBT1 att sites in the targeting construct
should allow us to use Cre and the fBT1 integrase to introduce
additional sequences into the engineered chromosomes inde-
pendently of any rearrangement that the fC31 integrase may
induce.
Expression of the fC31 integrase in cells containing a single
chromosome with the sites oriented in an opposite orientation
would be expected to invert the centromeric interval between
the two sites (Figure 1C). The action of the fC31 integrase in
G2 cells with two sister chromatids containing sites in the
opposite orientation has the potential however to generate
more complex outcomes. A single site-specific recombination
event between two sister chromatids would generate a larger
metacentric chromatid (referred to in the text as an inter-
chromatid maxi) containing attP and attR sites and a mini-
chromatid containing attB and attL sites and only the inter-
vening DNA. Both would contain a single centromere and
would be expected to be stable (Figure 1C). Of the three
types of rearranged chromosome generated by the integrase
acting on the sites in opposite orientations only the inter-
chromatid mini retains a un-rearranged puromycin resistance
gene and so cells with this mini-chromosome would, uniquely,
be expected to retain resistance to this antibiotic.
Intra-chromatid site-specific recombination between attP
and attB sites oriented in the same orientation would be
expected to generate a circular centric fragment and an acen-
tric linear fragment (Figure 1D). However a single exchange
event between sister chromatids containing sites in the same
orientation, would generate the acentric fragment produced
by the intra-chromatid event but would also generate a linear
dicentric fragment containing attB, attP and attR sites. Such a
chromosome could be rearranged by a second site-specific
recombination event to yield a circular dicentric chromatid
(Figure 1D). The fate of these two types of dicentric
Nucleic Acids Research, 2005, Vol. 33, No. 19 6103

Page 4

Figure 1. Engineering a human mini-chromosome with attB and attP sites that flank the centromere and the products of site-specific recombination between these
sites. (A) Outline of scheme for flanking the centromere of a human mini-chromosome with attB and attP sites. The mini-chromosome D2.5 was derived from the
humanYchromosomebytelomere directedchromosomebreakageinChinesehamsterovarycells (21)andthentransferredintochickenDT40cellsbysomaticcell
fusion.IntheexperimentsdescribedinthisarticleanattBsitewasintroducedintothecentromericalphoidDNAarraybytelomeredirectedchromosomebreakageto
generatethemini-chromosomeDD16andthenanattPsitewastargetedtothecentromereproximalDYZ5arrayineachofthetwopossibleorientationstogeneratethe
mini-chromosomesusedinthisworktermedCA4andXP4.DetailsaredescribedinthetextandinFigure1oftheSupplementaryData.(B)Detailsoftheconstructs
used to target the DYZ5 array. A fC31 integrase attP site was introduced into the DYZ5 array in each of the two possible orientations using constructs containing a
hygromycin–thymidine kinase fusion resistance gene, flanked by a loxP site for the Cre recombinase and an attP site (indicated as attP00) for the fBT1 integrase.
Immediatelyadjacent to the fC31 attP site on each construct was a promoterlesseGFP gene that could act as an indicatorfor the site-specific recombination event.
The DYZ5unitrepeatis 20kbin lengthand thearrayis composedof?25such repeats.IntheCA4 chromosome theconstructtargeted420 kbfromthe centromeric
end of the array while in the XP4 chromosome the construct targeted 40 kb from the centromeric end. (C) Possible products of site-specific recombination between
attachment sites oriented in opposite orientations. The upper section of this illustration indicates the product of an intra-chromatid recombination event. The lower
sectionindicatesthepossibleproductsthatareuniquetoasingleinter-chromatidrecombinationevent.(D)Possibleproductsofsite-specificrecombinationbetween
attachment sites oriented in the same orientation. The upper section of this illustration indicates the products of an intra-chromatid recombination event. The lower
section indicates the possible products that are produced by inter-chromatid recombination events.
6104 Nucleic Acids Research, 2005, Vol. 33, No. 19

Page 5

rearranged chromatids in the next division cannot be predicted
with certainty but studies in budding yeast [reviewed in (24)]
suggest that kinetochore polarity is established during S-phase
of the cell cycle and thus it would seem likely that these
fragments would be effectively dicentric for at least one
cycle and would be structurally unstable in the next mitosis.
Site-specific recombination between sites flanking the
centromere in an opposite orientation
Clone CA4, containing the mini-chromosome targeted with
the fC31 att sites in an opposite relative orientation, was
transfected with a construct that allowed expression of a
derivative of the fC31 integrase tagged at the carboxy termi-
nus with a nuclear localization signal (Figure 2A). Stably
transfected clones were analysed for site-specific recombina-
tion first of all by using a fluorescent activated cell sorter to
measure the proportion of cells that expressed eGFP. This
varied (Figure 2B) between no detectable expression and
almost complete expression. Western blotting was used to
measure the amount of integrase expressed by the individual
clones (Figure 2C). This indicated that there was no relation-
ship between the levels of integrase expression and of induced
fluorescence. Six clones that expressed different levels of
eGFP were analysed using direct techniques. Filter hybridiza-
tion after gel electrophoresis using probes for either the eGFP
gene or the puromycin resistance genes that respectively
flank the attP or attB sites in the un-rearranged CA4 mini-
chromosome (Figure 2D) indicated an explanation for the
eGFP fluorescence data. The simplest of the six clones was
number 15 in which 100% of the cells were expressing eGFP.
This clone contained restriction fragments consistent with
intra-chromatid inversion with no residual un-rearranged
mini-chromosome. PCR (Figure 2E) confirmed the presence
of attL and attR and absence of attB and attP sites indicated by
the blot. However, clones 6 and 10 expressed no eGFP and
contained attP and attR sites but no attB and attL sites at the
level of sensitivity afforded by filter hybridization. This com-
bination of sites suggested that these clones arose as a result of
an inter-chromatid rearrangement but that they had lost the
cells containing the inter-chromatid mini-chromosome prod-
uct and instead contained only the reciprocal inter-chromatid
maxi (Figure 1C). For this explanation to be valid the cells
containing the inter-chromatid mini-chromosome that would
have also been generated by inter-chromatid recombination
must have been lost from the population. One explanation for
this loss was that the cells containing this inter-chromatid
mini-chromosome grew more slowly than the cells containing
the inter-chromatid maxi as a result of the small size of the
inter-chromatid mini-chromosome activating a cell division
checkpoint that delayed the cell cycle of this lineage within
what would have been a mosaic cell line. Clones 2 and 8
contain a mixture of sites consistent with their containing a
mixture of the intra-chromatid inversion and inter-chromatid
maxi. However, Clone 2 contain no eGFP bright cells and
this may reflect the instability of expression of the subtelom-
eric b-actin promoter. Clone 4 uniquely contained attB, attP,
attL and attR sites and could have been explained by the
presence of either incomplete rearrangement, a mixture of
inter-chromatid mini- and maxi-chromosomes or by a com-
bination of each of these types of chromosomes. However the
alphoid DNA in this chromosome was found by long range
mapping to be rearranged (data not shown) and so the
chromosome was not studied further.
Cytogenetic or long range mapping studies was carried out
in order to confirm the results of the blotting and PCR analysis.
Thestructure ofthehypothetical inter-chromatidmaxiinclone
6 was confirmed by FISH (Figure 3A) using probes for the
alphoid DNA (red) and for a subtelomeric sequence, cY29
(green) (25) on the short arm of the human Y chromosome.
Conventional FISH had insufficient resolution to provide
unambiguous evidence for the inversion in clone 15 and so
pulsed field gels and restriction enzyme mapping with an
enzyme, PmeI, that does not cut in either the alphoid DNA,
the DYZ5 array or in the DYZ5 targeting construct enzyme
was used to confirm the presence of the DYZ5-alphoid DNA
and DYZ5-telomere junction fragments predicted by the maps
of the un-arranged chromosome (Figure 3E and F).
The second type of experiment designed to investigate the
site-specific recombination between sites with an opposite
relative orientation was carried out under conditions where
a unique product would be predicted: we investigated whether
we could recover the inter-chromatid mini-chromosome that
was missing in the products of the above experiment. Accord-
ingly clone CA4 was transfected with the fC31 integrase
expression construct and stably transfected clones were iso-
lated in the presence of puromycin. We checked these by
PFGE and confirmed the presence of a mini-chromosome
of about 1 Mb in size. Detailed analysis of three such clones
(mini-4, mini-56 and mini-81) by gel electrophoresis and filter
hybridization and by PCR is shown in Figure 2F, G and H. At
the level of sensitivity afforded by filter hybridization none of
these threeclonescontained the attR siteindicativeof either an
intra-chromatid inversion or an inter-chromatid maxi and all
contained an attL site characteristic of the inter-chromatid
mini. Clones mini-56 and mini-81 also contained an attP
site as detected by gel electrophoresis and filter hybridization
which indicated that the rearrangement was incomplete. Quan-
titation of the signal strength in the eGFP cognate fragments
indicated that the rearrangement was 70 and 74% complete
respectively. The mini-chromosome detected by FISH in the
inter-chromatid mini-4 cells (Figure 3B)usinganalphoid(red)
and DYZ5 probe (green) was, as anticipated, smaller than the
starting mini-chromosome (Figure 3C).
Incomplete site-specific recombination is associated
with mutation of recombination sites
The DNA used in the analysis illustrated in Figure 2G and
2H was extracted from the cloned cells approximately one
month after transfection with the integrase expression con-
struct. This data indicated that in the clones mini-56 and
mini-81 the rearrangement reaction had not occurred in all
of the cells in the population. In order to test the idea that
the rate of rearrangement was slow and had not yet continued
to completion for kinetic reasons the cells were cultured for
another month in the presence of zeocin. Analysis at the end of
the incubation (data not shown) was not significantly different
from that shown in Figure 2G and H. These results indicated
that the reaction had proceeded to completion but that a frac-
tion of the cells in the population were failing to rearrange.
We were using an internal ribosome entry site construct
Nucleic Acids Research, 2005, Vol. 33, No. 196105

Page 6

Figure 2. fC31 integrase mediated rearrangement of the attB and attP targeted human mini-chromosome CA4. (A) CCAG 30-NLS fC31 integrase IRES Zeo; the
plasmid used to direct expression of the NLS tagged fC31 integrase. (B) eGFP expression in clones derived following stable integration of CCAG 30-NLS fC31
integrase IRES Zeo into DT40 cells containing the CA4 mini-chromosome. The bars indicate the proportion of eGFP bright cells at three and five weeks following
transfection as measured by FACS. (C) 30-NLS fC31 integrase expression in the clones shown in B determined by western blotting. (D) Assay of site-specific
recombination around the puromycin resistance and eGFP genes by restriction enzyme digestion, gel electrophoresis and filter hybridization following stable
expressionof30-NLSfC31integraseinaselectionoftheclonesthatwereanalysedinBandC.(E)AssaybyPCRofattB,P,LandRsitesintheclonesanalysedinD.
(F) PFGEofthreeinter-chromatidmini-chromosome containingclonesisolatedfromCA4 cells after stableexpressionof30-NLSfC31integrasein the presenceof
puromycin. 49B(A)A9 is a mini-chromosome described in an earlier publication (22) and included as a molecular weight marker. (G) Assay of site-specific
recombinationaroundthepuromycinresistanceandeGFPgenes,intheclonesthatwereanalysedinF,byrestrictionenzymedigestion,gelelectrophoresisandfilter
hybridization following stable expression of 30-NLS fC31 integrase in the presence of puromycin. (H) Assay by PCR of attB, attP, attL and attR sites in the clones
analysed in F and G.
6106Nucleic Acids Research, 2005, Vol. 33, No. 19

Page 7

(Figure 2A) conferring resistance to zeocin to drive expression
of the integrase gene and had maintained the cells under
continuous selection for integrase expression. Therefore, it
seemed unlikely that variegated integrase expression was a
cause of the incomplete reaction and so we wondered whether
all of the attB and attP sites were intact. Of these sites the attP
sites would be present exclusively on any un-rearranged
starting mini-chromosome while the attB sites would be
present on both the un-rearranged and inter-chromatid mini-
chromosomes. Accordingly ten copies of both the attP and
attB sites from each of clones mini-56 and mini-81 were
sequenced. Each collection of attB and attP sites contained
copies that had been mutated by deletions and substitu-
tions (Table 1). The level of substitutions was such as to be
consistent with PCR error. Taken together these observations
suggested that the clones mini-56 and mini-81 consisted of
amixtureofrecombined
chromosomes and that the incomplete reaction was caused
by the presence of either a mutated attB or attP site on the
un-rearranged mini-chromosome. In order to test this idea the
cell line containing the inter-chromatid mini-chromosome
mini-81 was sub-cloned and six sub-clones that contained
attB and attL sites (Figure 4; clones 7–12) were identified.
The structure of the mini-chromosomes was checked in these
sub-clones using sequence tagged sites (STSs) (Figure 5) (26)
from the short arm of the Y chromosome and this showed that,
as predicted, the Y chromosome distal short arm as detected
by sY20 and sY37was absent. These clones contained the STS
and un-recombinedmini-
E
alphoid
550
441
351
276
213
kb
F
Yp
TEL
DYZ5 array
attR attL
loxP attP''
CCAG
HyTk
EGFP
Puro
1 kb
PmeI
PmeI
PmeI
420Kb DYZ5 fragment
Yp
300 Kb alphoid fragment
TEL
DYZ5 array
attB
loxP attP''
550Kb DYZ5 fragment
CCAG HyTkEGFP
attP
Puro
PmeI PmeIPmeI
CA4
clone15
DYZ5
CA4
clone15
CA4
clone15
BB
CCAA
DD
Figure 3. Establishing the identity of the fC31 integrase mediated chromosome rearrangements by FISH and by restriction site mapping. (A) FISH analysis of the
inter-chromatid maxi-chromosome in the cell line CA4 clone 6 using probes against the alphoid DNA (red) and the telomere of the short arm of the human Y
chromosome (25) (cY29) (green). (B) FISH analysis of the inter-chromatid mini-chromosome in the puromycin resistant cell line mini-4 containing an inter-
chromatidmini-chromosomeusingprobesagainstthealphoidDNA(red)andtheDYZ5sequence(green).(C)FISHanalysisofanun-rearrangedchromosome(inthe
celllineXP4)usingprobesagainstthealphoidDNA(red)andtheDYZ5sequence(green).ThechromosomesintheCA4celllinearesimilar.Theimageofthemini-
chromosome is slightly confusing because the two alphoid signals (red), deriving from the sister chromatids, appear to be more centrally located than the DYZ5
sequenceandtobeamoreobviousdoublet.Thewellresolveddoubletisacharacteristicofthetelomericpositionofthesequencesbutthefactthattheyappearinasub-
telomericpositionmayreflectthewaythechromosomeisfolded.Thisdistributionofthetwosequenceswasobservedconsistently.(D)FISHanalysisofthecircular
deletion mini-chromosome in the cell line XP4 DEL 14 using probes against the alphoid DNA (red) and the DYZ5 sequence (green) of chromosomes.
(E) Arrangement of PmeI sites around the alphoid and DYZ5 sequences in the cell lines, CA4 and CA4 clone 15 following fC31 integrase mediated intra-
chromatidinversionoftheintervalbetweentheattBandattPsites.(F)FilterhybridizationandrestrictionenzymeanalysisofthearrangementofthePmeIsitesaround
the alphoid and DYZ5 sequences in the cell lines DD16, CA4 and CA4 clone 15 that was isolated after introduction of a fC31 integrase expression construct in the
absence of puromycin.
Nucleic Acids Research, 2005, Vol. 33, No. 19 6107

Page 8

detecting the DYZ5 sequence sY61 and therefore contained a
mini-chromosome with a structure consistent with an inter-
chromatd mini-chromosome. We also identified six clones
that contained the STS’s from the distal short arm of the
chromosome (Figure 4; clones 1–6) and which were predicted
to contain mini-chromosomes that had not rearranged. These
were typed for attB, attP and attL and in all of them the attL
site was absent and an attB sitewas present.However intwo of
Table 1. Sequences of attachment sites recovered from cell lines described in this study
Cell lineSite sequencedSequence detected Frequency or source of sequence
DD16fC31
CA4
attB
attB
attP
attB
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
CCGCGGTGCGGGTGCCAGGGCGT
CCGCGGTGCGGGTGC
CCGCGGTGCGGGTGCCAGGGCGT
Not detected
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGAGTTCTCTCAGTTGGGGGCGT
Not detected
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
CCGCGGTGCGGGTGCCAGGGCGTG
CCGCGGTGCGGGTGCCAGGGCGTGCCC
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACTGGGGTAACCT
TAGTAGTGtCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGAGTTCTCTCAGTTGGGGGCGT
Not detected
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
CCGCG
CCGCGGTGCGGGTGCCAGGGCGTGCCC
CCGCGGTGCGGGTGCCAGGGCGTGC
CCGCGGTGCGGGTGCCAGGGCGTGCC
CCGCGGTGCGGGTGCCAGGGC
CCGCGGTGCGGGTGCCAGGGCGTGC
CCGCGGTGCGGGTGCCAGGGCGTGCC
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACaGGGGTAACCT
TAGTAGTGCCCCgACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGAGTTCTCTCAGTTGGGGGCGT
Not detected
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
CCGCGGTGCGGGT
CCGCGGTGCGGGTGCCAGGGCGTG
CCGCGGTGCGGGTGCCAGGGCGT
CCGCGGTGCGGGTGCCAGGGCGTGCC
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACT GGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
Not detected
TAGTAGTGCCCCAACTGGGGTAACCTTTGGGCTCCCCGGGCGCGTACTCCA
CCGCGGTGCGGGTGCCAGGGC
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGcTCTCTCAGTTGGGGGCGTA
Not detected
TAGTAGTGCCCCAACTGGGGTAACCTTTGGGCTCCCCGGGCGCGTACTCCA
CCGCGGTGCGGGTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCGTACTCC
CCGCGGTGCGGGTGCCAGGGCGG
CCGCGGTGCGGGTGCCAG
CCGCGGTGCGGGTGCCAGGGCG
CCGCGGTGCGGGTGCCAGGGCGTGCCCTT GGCTCCCCGGGCGCGTACTCC
CCGCGGTGCGGGTGCCAGGGCGTGCC
CCGCGGTGCGGGTGCCA
CCGCGGTGCGGGTGCCAGGGCGTG
CCGCGGTGCGGGTGCCAGGGC
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTTCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACTGGGGTAACC TTGAGTTCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACTGGGGTAACCTTTGAGTcCTCTCAGTTGGGGGCGTA
TAGTAGTGCCCCAACTG
Not detected
TAGTAGTGCCCCAACTGGGGTAACCTTTGGGCTCCCCGGGCGCGTACTCCA
9/9
20/20
10/10
7/10
1/10
1/10
1/10
Mini-4
GGCTCCCCGGGCGCGTACTCC
CCCCGGGCGCGTACTCCC
GCTCCCCGGGCGCGTACTCC
attP
attL
attR
attB
PCR product
Mini-56 1/10
1/10
8/10
8/10
1/10
1/10
PCR product
GGCTCCCCGGGCGCGTACTCC
CGGGCGCGTACTCC
attP
TTCTCTCAGTTGGGGGCGTA
attL
attR
attB
Mini-81 2/10
1/10
2/10
1/10
1/10
1/10
1/10
1/10
8/10
1/10
1/10
PCR product
GCTCCCCGGGCGCGTACTCC
CCGGGCGCGTACTCC
TCCCCGGGCGCGTACTCC
CTCCCCGGGCGCGTACTCC
CCGGGCGCGTACTCC
CTCCCCGGGCGCGTACTCC
CCCCGGGCGCGTACTCC
attP
TTCTCTCAGTTGGGGGCGTA
attL
attR
attB
XP4nls 101/9
1/9
5/9
1/9
1/9
8/9
1/9
GGGCTCCCCGGGCGCGTACTCC
CCCCGGGCGCGTACTCC
GCTCCCCGGGCGCGTACTCC
CCCCGGGCGCGTACTCC
attP
attL
attR
attB
attP
PCR Product
9/9
8/9
1/9
XP4nls 11
TCCCCGGGCGCGTACTCC
attL
attR
attB
PCR product
1/9
1/9
1/9
1/9
1/9
1/9
1/9
1/9
1/9
7/10
1/10
1/10
1/10
XP4nls 14
TTGGGCTCCCCGGGCGCGTACTCC
GGGCTCCCCGGGCGCGTACTCC
CCCCGGGCGCGTACTCC
CTCCCCGGGCGCGTACTCC
GGGCTCCCCGGGCGCGTACTCC
GCTCCCCGGGCGCGTACTCC
TCCCCGGGCGCGTACTCC
attP
TTGAGTTCTCTCAGTTGGGGGCGTA
attL
attR
PCR product
EachrowinthetablerepresentsasequenceofeitheranattachmentsitesequencedasaPCRproductorasclonedDNAandthisisindicatedintherighthandcolumn.
This column also indicates the frequency with which the respective sequence was recovered when cloned. Deletions are indicated by gaps in the sequence and
substitutions by lower case letters. Thus a continuous sequence of upper case letters indicates that the sequence was recovered intact.
6108Nucleic Acids Research, 2005, Vol. 33, No. 19

Page 9

the clones the attP site was present and in four was unde-
tectable. We suggest that the four clones in which the attP site
was undetectable included a deletion extending from the
site which had destroyed one or both of the PCR primer bind-
ing sites. The attB sites and, wherepossible, the attP sites were
sequenced in the six sub-clones containing the un-rearranged
mini-chromosome and in every sub-clone the attB site had
been damaged by one of the deletions seen in the PCR
products from uncloned cells that had been subsequently
isolated in bacterial clones. The attL sites in each of the six
sub-clones containing the inter-chromatid mini-chromosome
were also sequenced and in every case the sequence was
consistent with accurate and conservative site-specific
recombination.
Damage to recombination sites is associated with
failed site-specific recombination reactions
These observation that the recombination sites were found to
be mutated in chromosomes that had failed to recombine
raised four questions, firstly whether the sites were mutated
before introduction of the integrase? We sequenced attB and
attP sites from clone CA4 before the introduction of the inte-
grase. These results (Table 1) demonstrated that both sites
were intact in the starting cells. The second question was
whether the sites were mutated when exposed to integrase
alone or did the mutations only arise when both the attB
and attP sites were exposed to integrase in the same cell?
Weaddressedthisquestionbyintroducinganintegraseexpres-
sion constructinto aDD16 clone thatcontainsonlythe attB site
and sequencing ten copies of the attB site after one month in
culture. The sequence was not detectably mutated (Table 1).
The third question was whether mutations were seen only in
the substrate sites or were also present in the product sites.
Accordingly we sequenced PCR products containing attL sites
isolated from the inter-chromatid mini-chromosome contain-
ing clones mini-56 and mini-81 and in each case the sequence
was consistent with accurate site-specific recombination bet-
ween undamaged sites. Similarly we sequenced attL and attR
sites from the intra-chromatid inversion clone 15 described
above and showed that both were as predicted by accurate site-
specific recombination.
These results thus indicate that the substrates were being
damaged when exposed to the integrase only when both were
presentinthesamecellandthatdamagewasspecifictothesub-
strate sites. We interpret these observations mechanistically to
suggest that the intermediate of the fC31 integrase reaction is
a substrate for the host DNA damage response. Our observa-
tions suggest that repair of the double-strand break introduced
by the integrase destroys the ability of the sites to participatein
subsequent site-specific recombination reactions. The final
question was whether the presence of mutated recombination
sites was specific to the clones that were failing to rearrange
completely? Therefore we sequenced ten copies of the attB
site in the mini-4 clone and of these, three sequences included
a deletion (Table 1). We also sequenced the attL site in this
clone and, consistent with the results for clones mini-56 and
mini-81, found this to be intact. These results suggest that the
sequence of reactions leading to mini-4 was as follows. One
sister undertook an abortive attempt at recombination and this
left it with damaged attB sites and an intact attP site. The other
sister, containing an intact attB site necessary for site-specific
recombination then participated in a productive inter-
chromatid site-specific recombination reaction with the intact
attP site present on its sister which led to the mini-4 with
sequences that we detect.
No detectable ectopic interaction between a native
attB site on a mini-chromosome and pseudo-attP
sites present in the chicken genome
The inter-chromatid mini-chromosomes, containing attB sites,
were present in cells expressing the fC31 integrase and so we
could use such chromosomes to test the possibility that there
might be pseudo-attPsites in the chicken genome available for
recombination with the attB site on the mini-chromosome.
Such ectopic recombination would be predicted to lead to
integration of the mini-chromosome into the host genome
or to mini-chromosome loss. Therefore we cultured cells
containing inter-chromatid mini-4 for one month and mea-
sured both copy number by FISH (Figure 5A) and structural
stability by PFGE (Figure 5B). This analysis provided no
evidence for any rearrangement or structural instability of
the mini-chromosome and was thus consistent with an absence
of ectopic site-specific recombination. However in order
to interpret these observations we needed to show that the
attB site present on the mini-chromosome and the integrase
present in the cell were both active. Therefore we assembled
attB
attL
attP
12
3
4
5
6
78
9
10 1112
Figure 4. Sub-cloning of the cell line inter-chromatid mini-81 establishes that it is a mixture of cells containing rearranged and un-rearranged mini-chromosomes.
ThisfigureshowsPCRanalysisof12sub-clonesofthecelllinecontainingtheinter-chromatidmini-chromosomemini-81analysedin Figure3.STSmarkerssY20,
37, 61 STS’s from the short arm of the Y chromosome and the attachment sites, attB, attL and attP were analysed. The starting clone CA4 and the cell line with the
intra-chromatid inversion CA4 clone 15 were analysed as controls.
Nucleic Acids Research, 2005, Vol. 33, No. 19 6109

Page 10

a plasmid attP-neo (Figure 5C) that contained a promoterless
G418 resistance gene and introduced it into the clone 4
containing the inter-chromatid mini-chromosome and selec-
ted for G418 resistant cells. We were able to recover many
thousands of G418 resistant clones. We checked 12 of these
for site-specific recombination using specific amplification
across the recombinant attR and attL sites and found
that each was a site-specific recombinant (Figure 5D). The
behaviour of the chromosome thus provides no evidence
for the idea that there are pseudo-attP sites in the chicken
genome.
Site-specific recombination between sites flanking the
centromere generates a circular mini-chromosome,
is often incomplete and is also associated with
mutation of attB or attP sites
The results of the previous experiments demonstrated that the
fC31 integrase functioned with useful efficiency in promoting
site-specific recombination across distances as far as 1 Mb in
vertebrate cells. We also wanted to demonstrate that it could
be used to promote centromere deletion and thereby provide
the basis of an assay for centromeric DNA. Therefore we took
cells containing the XP4 engineered mini-chromosome and
transfected them with the fC31 integrase expression construct,
selected for stably transfected clones in the absence of pur-
omycin selection and analysed them by filter hybridization
analysis and PCR (Figure 6A and B). The results were
more complicated than one might have initially predicted
but in large part this could be explained by ascribing the
complications to the behaviour of the dicentric chromosomes
produced by an inter-chromatid site-specific recombination
(Figure 1D). Clones XP4 DEL 10, 12 and 14 were the most
straightforward. The puromycin probe detected an attR frag-
ment and a small amount of un-rearranged attB fragment. The
eGFP probe detected residual un-rearranged attP. This pattern
of fragments suggests that these clones contained the circular
centric fragment, a variable fraction of cells with the
un-recombined chromosome but lacked the linear acentric
fragment predicted to contain the attL site. We analysed clones
DEL 12 and 14 by FISH and the results bore this out
(Figure 3D). In each of these clones we could see a small
fragment that hybridized with alphoid DNA and weakly
with DYZ5. We could also see variable amounts of both
the starting chromosome and cells lacking any hybridizing
material (Figure 6C). Clone 8 lacked hybridizing material
completely. The presence of cells lacking hybridizing mini-
chromosome in these clones could be explained in two ways.
One was that the original site-specific recombination event
was exclusively intra-chromatid, that this went almost to
completion but that the circular centric mini-chromosome
was mitotically unstable. A second explanation was that the
circular mini-chromosome was mitotically stable and that
the cells lacking any hybridizing material arose as a result
of a different mechanism. We investigated which of these
two explanations were correct by culturing clones DEL 12
and 14 for one month in the absence of selection; the results
shown for clone DEL 14 (Figure 6C) shows that there is no
detectable loss of the centric fragment. There is also no evi-
dence for any further recombination indicating that as with the
inversions analysed in Figure 3 the reaction went to comple-
tion by the time of the first analysis. Sequencing (Table 1)
across the attB and attP sites in the clones XP4 DEL 10 and 14
detected damaged sites consistent with the notion that, as with
the inversions, damage to the substrate sites was preventing
Figure5.Amini-chromosomewithafunctionalattBsiteisstableinDT40cells
in the presence of fC31 integrase. (A) Stability of mini-chromosomes in clone
CA4 inter-chromatid mini-chromosome clone 4 analysed using FISH with
probes for the alphoid DNA and the DYZ5 sequence after the indicated time
in culture in the presence and absence of selection with puromycin. Zeocin,
selectingforintegraseexpression,waspresentinbothofthecultures.Thefilled
bars indicate the proportion of cells containing a mini-chromosome and the
empty bars indicate the proportion of cells lacking a mini-chromosome.
(B) Mini-chromosomesin clone 4remainedstructurally intactafter fourweeks
in culture as judged by pulsed field electrophoresis. (C) Diagrammatic repre-
sentationofthesite-specificrecombinationreactionbetweentheattBsiteinthe
mini-chromosomes in clone number 4 and a plasmid containing attP and a
promoterless gene conferring resistance to the antibiotic G418; fCattPneo.
(D) attB sites on mini-chromosomes in clone number 4 are able to undergo
site-specific recombination after one month in culture as judged by PCR. Cells
from clone 4 containing the inter-chromatid mini-chromosome were electro-
porated with the plasmid fCattPneo and G418 resistant clones isolated. DNA
extracted from 12 such clones was analysed by PCR across the products of
the site-specific recombination reaction between fCattPneo and the resident
attB site.
6110 Nucleic Acids Research, 2005, Vol. 33, No. 19

Page 11

complete reaction. In order to account for the cells that
lack any detectable hybridizing material we need to con-
sider the consequences of an inter-chromatid recombination
(Figure 1D). The inter-chromatid recombination event gives
rise to at least two types of predictably unstable and dicentric
fragments. If the broken fragments had been degraded or
had failed to be incorporated into the reforming nucleus at
telophase then we could account for the presence of the cells
lacking any hybridizing material. Broken fragment are recom-
binogenic and would also be expected to integrate into the
chicken genome and such a combination of processes explains
theexistenceofaclonesuchasXP4DEL15whichcontainsan
attB site and no eGFP cognate DNA. These considerations
leave us with clones DEL 11 and 13. Clone DEL 11 has not
undergone any detectable site-specific recombination. We
sequenced both the attB and attP sites in this clone (Table 1)
and discovered that all of the recovered attB sites had been
mutated thus providing an explanation for the failure of detec-
table site-specific recombination. Clone DEL 13 contains what
appeared to be an aberrantly rearranged attR fragment. One
explanation for this reaction was that it arose as a result of
faulty site-specific recombination reaction in which damage to
one or both of the sites had occurred before inter-strand liga-
tion.InordertoinvestigatethispossibilityweusedinversePCR
to recover the fragment. Sequence analysis however showed
that this was not an inter-strand ligation product but that the
attB site was intact. This fragment may therefore have arisen
either as a result of repair of the attB component of an inte-
grase reaction intermediate into the host genome or as a result
of damage of an inter-chromatid dicentric derived fragment.
DISCUSSION
The results of our investigations may be stated qualitatively as
demonstrating that the fC31 integrase promotes efficient long
range chromosome rearrangements but that the reaction fails
to go to completion because the sites are sometimes mutated in
the course of the attempt at recombination. The mutations are
mainly deletions but sometimes we detect base substitutions
(Table 1). However the substitutions are detected rarely in the
cloned products of the PCR, are not present in directly
sequenced attachment sites and are therefore unlikely to rep-
resent mutations introduced by the repair of the attachment
sites. In addition to mutations occurring within the attachment
sites we have also detected larger scale rearrangements as
judged by the presence of aberrantly sized restriction enzyme
fragments in the products of the rearrangements. We have
detected no evidence of ectopic recombination between an
engineered attB site and ectopic pseudo-sites resident in the
host genome suggesting that the protein is not reacting with
pseudo-sites. Exclusive damage to the participating sites
detracts from the potential utility of the fC31 integrase
becauseitreduces theefficiencyofthe process. Amoreserious
concern is the extent to which the DNA damage response may
cause intermediates in the reaction to recombine non-
homologously with the host cell genome. We detected one
such potential rearrangement in the studies of the inversion
reaction (Clone CA4 clone 4). However, such mutagenic
events are a small minority of those that we observe.
In order to consider the results quantitatively we turn to
the data from the analysis of fifteen clones isolated from
the clone XP4 deletion experiments. Of these; three contained
rearranged fragments and one lacked any detectable mini-
chromosome DNA. These four clones could all be composed
predominantly of derivatives of the inter-chromatid events
discussed with reference to Figure 1D. Alternatively they
could be derived by chromosome rearrangements originating
from repair of the fC31 integrase reaction intermediates.
XP4
DEL8DEL10DEL11
DEL12
DEL13 DEL14
DEL15
23.1
9.4
6.6
4.4
2.3
2.0
kb
attL
attB
attP
attR
Probe: eGFP Probe: Puro
0
10
20
30
40
50
60
0 2 weeks4 weeks
0 copies
1 copy
Unrearranged XP4
DEL14
% cells
with
indicated
chromosome
constitiution
A
B
attB
attP
attL
attR
XP4
100bp ladder
DEL8
DEL10
DEL11
DEL12
DEL13
DEL14
CA4nls15
100bp ladder
DEL15
200bp
300bp
200bp
300bp
XP4
DEL8DEL10 DEL11
DEL12
DEL13DEL14
DEL15
C
Figure 6. Deleting the centromere of the human mini-chromosome XP4
with the fC31 integrase. (A) Assay of site-specific recombination around
the puromycin resistance and eGFP genes by restriction enzyme digestion,
gel electrophoresis and filter hybridization following stable expression of
30-NLS fC31 integrase. (B) Assay by PCR of attB, attP, attL and attR sites
in the clones analysed in A. (C) Stability of mini-chromosomes in clone XP4
derived deletion clone DEL 14 analysed using FISH with probes for the
alphoid DNA and the DYZ5 sequence after the indicated time in culture.
The circular mini-chromosome could be readily discriminated from the
un-rearrangedchromosomeonthebasisofsizeandmoreeasilybytheintensity
of staining with the DYZ5 probe (as indicated in Figure 3).
Nucleic Acids Research, 2005, Vol. 33, No. 196111

Page 12

This first alternative seems likely to hold for at least some
clones butcannot beprovenand so these clones cannot beused
in any quantitative discussion. In the remaining clones we
estimated the extent of rearrangement by comparing the inten-
sities of the attB and attR containing fragments recognized by
hybridization with the puromycin probe (Figure 6A). This
indicated that the average extent of rearrangement was 54%
although this varied widely between clones. The variation in
the extent of rearrangement arises in two ways. These can be
easily understood by considering the site-specific recombina-
tion reaction. This may be productive or unproductive but it is
irreversible.Ifthe integrase isintroduced into a G1cell and the
reaction proceeds before the subsequent S-phase then all of the
cells in the clone derived from this cell will be of one type or
other. However if the integrase reacts slowly with the target
sites and the reaction takes many cell divisions to go to com-
pletion then the clone will be a mixtureof bothoutcomes. Drift
in the composition of the population arising from differences
in growth rate between sub-clones however may also give rise
to differences between the clones in the relative proportions of
the productive and unproductive events. The long term culture
experiments indicate that differences in growth rate are small
and for the purposes of a crude estimate may be excluded. If
we make this simplification then we can use the variance of the
proportion of productively recombined attB site in the deletion
reactiontoestimatethe rateofreaction percelldivisionas0.86
percelldivision(seeMaterials andMethodsforadiscussionof
the calculation). This is an upper estimate but despite the small
sample size and systematic imprecision of the calculation
would seem to justify more elaborate experiments aimed at
measuring the rate of site-specific inter-chromosomal translo-
cations in this as well as other systems.
Attachment site mutation arises only when the two sites are
present in the same cell as the integrase. This observation
suggests that the attachment sites are mutated as a result of
the host cell DNA repair machinery recognizing the fC31
integrase nucleoprotein complex as it undergoes recombina-
tion. A likely stage in recombination that could be targeted is
the cleaved DNA intermediate undergoing strand exchange.
Such repair is consistent with the structure of the intermediates
and the kinetics of serine recombinases promoted reactions
(7,27,28). Furthermore both the BxbI (10) and fC31 (9) inte-
grases will relax a supercoiled plasmid containing an attach-
ment site in the presence of a short linear double-stranded
DNA molecule containing the reciprocal site. These results
indicate that the DNA bound integrase subunits whilst cova-
lently linked to the cleaved DNA can iterate strand exchange
before effecting ligation. However it is important to point out
that we are studying reactions between substrates separated by
?1 Mb. Further work will be required to establish how the
level of attachment site mutation varies with the distance
between the two sites and the experimental cell line.
A third variable is the site-specific recombinase itself. The
fC31 integrase is a member of a large class of serine recom-
binases (6) and it would seem reasonable to imagine that small
differences between the kinetics of the reactions promoted by
different members of this class of protein and between the
structures of the different recombination synapses would
give rise to differences in the susceptibility of the different
intermediates to the reactions that we suggest are promoted by
the cellular repair pathways.
Our results thus raise the obvious question: what utility does
the fC31 integrase have for engineering chromosome rear-
rangements in vertebrate cells? This question cannot be
addressed in isolation but only by consideration of the alter-
natives. As discussed in the introduction Cre is a powerful
mutagen and at levels that are not detectably mutagenic pro-
motes translocations in vertebrates only slowly (29). We are
not aware of any study that has subject Cre to the scrutiny that
our work has applied to the fC31 integrase. If the fC31 inte-
grase were to be more efficient and less damaging mutagen
than Cre then fC31 integrase would be a general alternative to
Cre. Cre and Flp are also reversible and so are unsuited to
experiments where the rearrangement needs to be irreversible.
Our results prove that fC31 integrase can be used for such
experiments.
Our work was initiated as part of an ongoing project aimed
at investigating vertebrate centromeres and the results also
have implications for the current attempts to close the
human genome sequence and for the study of centromeres.
Firstly the sizes of the inter-chromatid mini-chromosomes are
consistent with the current assembly of the sequence of the
proximal short arm of the human Y chromosome. Such exten-
sively repeated regions of human chromosomes are difficult to
assemble into reliable contigs and our observations provide
confidence in the accuracy of such assemblies. The resolution
of the pulsed field gels is only about 50 kb and we cannot
exclude errors in assembly or gaps smaller than this. Second
the loss of the acentric fragments generated in the centromere
deletion reactions formally demonstrates that DT40 cells do
not form neo-centromeres on linear chromosome fragments
with two telomeres at a high frequency.
In summary we have shown that that the fC31 integrase
promotes efficient, irreversible, site-specific long range chro-
mosome rearrangement in vertebrate cells. No evidence of
ectopic recombination between a fC31 integrase attachment
site and genomic pseudo-sites was found. The fC31 integrase
is therefore unique in its proven ability to promote irreversible
chromosome engineering reactions. However further work is
required to establish the full range of chromosome engineering
reactions for which it is the most suitable reagent.
SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online.
ACKNOWLEDGEMENTS
This work was supported by the BBSRC under the Exgen
program, the EU (FP5 Contract QLK3-CT-2000-00785 and
FP6 Proposal GENINTEG) and a Nottingham University stu-
dentship to SKM. We thank Alistair Chambers for comments
on the manuscript and Jean-Marie Buerstedde for support.
Funding to pay the Open Access publication charges for this
article was provided by JISC.
Conflict of interest statement. None declared.
REFERENCES
1. Yu,Y. and Bradley,A. (2001) Engineering chromosomal rearrangements
in mice. Nature Rev. Genet, 2, 780–790.
6112 Nucleic Acids Research, 2005, Vol. 33, No. 19

Page 13

2. Forster,A., Pannell,R., Drynan,L.F., Codrington,R., Daser,A.,
Metzler,M.,Lobato,M.N.andRabbitts,T.H.(2005)Theinvertorknock-in
conditional chromosomal translocation mimic. Nature Methods.,
2, 27–30.
3. Lee,T. and Luo,L. (2001) Mosaic analysis with a repressible cell marker
(MARCM) for Drosophila neural development. Trends Neurosci.,
24, 251–254.
4. Guo,F., Gopaul,D.N. and van Duyne,G.D. (1997) Structure of Cre
recombinase complexed with DNA in a site-specific recombination
synapse. Nature., 389, 40–46.
5. Loonstra,A., Vooijs,M., Beverloo,H.B., Allak,B.A., van Drunen,E.,
Kanaar,R., Berns,A. and Jonkers,J. (2001) Growth inhibition and DNA
damage induced by Cre recombinase in mammalian cells. Proc. Natl
Acad. Sci. USA., 98, 9209–9214.
6. Smith,M.C.M. and Thorpe,H.M. (2002) Diversity in the serine
recombinases. Mol. Microbiol., 44, 299–307.
7. Li,W.,Kamtekar,S.,Xiong,Y.,Sarkis,G.J.,Grindley,N.D.andSteitz,T.A.
(2005) Structure of a synaptic gammadelta resolvase tetramer covalently
linked to two cleaved DNAs. Science., 309, 1210–1215.
8. Thorpe,H.M. and Smith,M.C.M. (1998) In vitro site-specific integration
of bacteriophage DNA catalyzed by a recombinase of the resolvase/
invertase family. Proc. Natl Acad. Sci. USA., 95, 5505–5510.
9. Smith,M.C., Till,R. and Smith,M.C.M. (2004) Switching the
polarity of a bacteriophage integration system. Mol. Microbiol.,
51, 1719–1728.
10. Ghosh,P., Kim,A.I. and Hatfull,G.F. (2003) The orientation of
mycobacteriophage Bxb1 integration is solely dependent on the central
dinucleotide of attP and attB. Mol. Cell., 12, 1101–1111.
11. Groth,A.C., Olivares,E.C., Thyagarajan,B. and Calos,M.P. (2000)
A phage integrase directs efficient site-specific integration in
human cells. Proc. Natl Acad. Sci. USA., 97, 5995–6000.
12. Thyagarajan,B., Olivares,E.C., Hollis,R.P., Ginsburg,D.S. and
Calos,M.P. (2001) Site-specific genomic integration in mammalian
cells mediated by phage phiC31 integrase. Mol. Cell Biol., 21,
3926–3934.
13. Olivares,E.C., Hollis,R.P. and Calos,M.P. (2001) Phage R4 integrase
mediates site-specific integration in human cells. Gene., 278,
167–176.
14. Stoll,S.M. and Calos,M.P. (2002) Extrachromosomal plasmid vectors
for gene therapy. Curr. Opin. Mol. Ther., 4, 299–305.
15. Thyagarajan,B., Guimaraes,M.J., Groth,A.C. and Calos,M.P. (2000)
Mammalian genomes contain active recombinase recognition sites.
Gene., 244, 47–54.
16. Schueler,M.G., Higgins,A.W., Rudd,M.K., Gustashaw,K. and
Willard,H.F. (2001) Genomic and genetic definition of a functional
human centromere. Science., 294, 109–115.
17. Skaletsky,H.,Kuroda-Kawaguchi,T.,Minx,P.J.,Cordum,H.S.,Hillier,L.,
Brown,L.G., Repping,S., Pyntikova,T., Ali,J., Bieri,T. et al. (2003)
The male-specific region of the human Y chromosome is a mosaic of
discrete sequence classes. Nature., 423, 825–837.
18. Lupton,S.D., Brunton,L.L., Kalberg,V.A. and Overell,R.W. (1991)
Dominant positive and negative selection using a hygromycin
phosphotransferase-thymidine kinase fusion gene. Mol. Cell Biol.,
11, 3374–3378.
19. delaLuna,S.,Soria,I.,Pulido,D.,Ortin,J.andJimenez,A.(1988)Efficient
transformation of mammalian cells with constructs containing a
puromycin-resistance marker. Gene., 62, 121–126.
20. Niwa,H., Yamammura,K. and Miyazaki,J. (1991) Efficient selection for
high-expression transfectants with a novel eukaryotic vector. Gene.,
108, 193–201.
21. Shen,M.H., Yang,J., Loupart,M.L., Smith,A. and Brown,W.R.A. (1997)
Human mini-chromosomes in mouse embryonal stem cells. Hum. Mol.
Genet., 6, 1375–1382.
22. Yang,J.W., Pendon,C., Yang,J., Haywood,N., Chand,A. and
Brown,W.R.A. (2000) Human mini-chromosomes with minimal
centromeres. Hum. Mol. Genet., 9, 1891–1902.
23. Gregory,M.A., Till,R. and Smith,M.C.M. (2003) Integration site for
StreptomycesphagephiBT1anddevelopmentofsite-specificintegrating
vectors. J. Bacteriol., 185, 5320–5323.
24. Nasmyth,K.(2002)Segregatingsistergenomes:themolecularbiologyof
chromosome separation. Science., 297, 559–565.
25. Cooke,H.J., Brown,W.R.A. and Rappold,G.A. (1985) Hypervariable
telomeric sequences from the human sex chromosomes are
pseudoautosomal. Nature., 317, 687–692.
26. Vollrath,D., Foote,S., Hilton,A., Brown,L.G., Beer-Romero,P.,
Bogan,J.S. and Page,D.C. (1992) The human Y chromosome—a
43-interval map based on naturally-occurring deletions. Science.,
258, 52–59.
27. Dhar,G., Sanders,E.R. and Johnson,R.C. (2004) Architecture of the hin
synaptic complex during recombination: the recombinase subunits
translocate with the DNA strands. Cell., 119, 33–45.
28. Leschziner,A.E. and Grindley,N.D. (2003) The architecture of the
gammadelta resolvasecrossoversitesynapticcomplex revealedby using
constrained DNA substrates. Mol. Cell., 12, 775–781.
29. Zong,H., Espinosa,J.S., Su,H.H., Muzumdar,M.D. and Luo,L. (2005)
Mosaic analysis with double markers in mice. Cell., 121, 479–492.
Nucleic Acids Research, 2005, Vol. 33, No. 19 6113