Cloning and sequencing of a highly productive, endotheliotropic virus strain derived from human cytomegalovirus TB40/E.

Cloning and sequencing of a highly productive,
endotheliotropic virus strain derived from human
cytomegalovirus TB40/E
Christian Sinzger,13 Gabriele Hahn,23 Margarete Digel,1 Ruth Katona,1
Kerstin Laib Sampaio,1 Martin Messerle,3 Hartmut Hengel,4
Ulrich Koszinowski,5 Wolfram Brune6 and Barbara Adler5
Correspondence
Christian Sinzger
christian.sinzger@
med.uni-tuebingen.de
1Institut fu¨r Medizinische Virologie, Eberhard-Karls-Universita¨t, Tu¨bingen, Germany
2Laboratoriumsmedizin, Klinikum Ingolstadt, Germany
3Abteilung fu¨r Virologie, Medizinische Hochschule Hannover, Germany
4Institut fu¨r Virologie, Heinrich-Heine-Universita¨t, Du¨sseldorf, Germany
5Max von Pettenkofer-Institut fu¨r Virologie, Ludwig-Maximilians-Universita¨t Mu¨nchen, Germany
6Fachgebiet Virale Infektionen, Robert Koch-Institut, Berlin, Germany
Received 4 July 2007
Accepted 17 October 2007
Human cytomegalovirus (HCMV) strain TB40/E, replicates efficiently, exhibits a broad cell tropism
and is widely used for infection of endothelial cells and monocyte-derived cells yet has not been
available in a phenotypically homogeneous form compatible with genetic analysis. To
overcome this problem, we cloned the TB40/E strain into a bacterial artificial chromosome (BAC)
vector. Both highly endotheliotropic and poorly endotheliotropic virus clones, representing three
distinct restriction fragment patterns, were reconstituted after transfection of BAC clones
derived from previously plaque-purified strain TB40/E. For one of the highly endotheliotropic
clones, TB40-BAC4, we provide the genome sequence. Two BACs with identical restriction
fragment patterns but different cell tropism were further analysed in the UL128-UL131A gene
region. Sequence analysis revealed one coding-relevant adenine insertion at position 332 of
UL128 in the BAC of the poorly endotheliotropic virus, which caused a frameshift in the
C-terminal part of the coding sequence. Removal of this insertion by markerless mutagenesis
restored the highly endotheliotropic phenotype, indicating that the loss of endothelial cell tropism
was caused by this insertion. In conclusion, HCMV strain TB40/E, which combines the high
endothelial cell tropism of a clinical isolate with the high titre growth of a cell culture adapted
strain, is now available as a BAC clone suitable for genetic engineering. The results also suggest
BAC cloning as a suitable method for selection of genetically defined virus clones.
INTRODUCTION
One of the hallmarks of human cytomegalovirus (HCMV)
is its very broad cell tropism in vivo, which allows for
systemic spread of this herpesvirus into virtually any organ
during acute infection. Therefore, there has been increasing
interest in studying HCMV infection of endothelial cells
(EC) and monocyte-derived cells, which are assumed to
contribute greatly to the haematogenous spread of the
virus (Gerna et al., 2002; Hertel et al., 2003; Jarvis &
Nelson, 2002; Sinzger & Jahn, 1996). This has been
facilitated by the introduction of EC-propagated HCMV
strains that have preserved their natural broad cell tropism
in cell culture. One of these strains, TB40/E, has been used
by several laboratories as a highly endotheliotropic and
macrophage-tropic strain (Allal et al., 2004; Bentz et al.,
2006; Hertel et al., 2003; Homman-Loudiyi et al., 2003;
King et al., 2006; Moutaftsi et al., 2004; Reeves et al., 2005;
Reinhardt et al., 2005). However, genetic analyses have
revealed that this strain is genetically heterogeneous despite
several rounds of plaque purification, which argued against
its use for genotype–phenotype analyses (Dolan et al.,
2004). Generation of enhanced green fluorescent protein-
tagged mutants of TB40/E by conventional homologous
recombination resulted in random selection of variants
with either high or low EC tropism (Laib Sampaio et al.,
2005), indicating that it is not only genotypically but also
phenotypically heterogeneous.
3These authors contributed equally to this work.
The GenBank/EMBL/DDBJ accession number for the sequence
reported in this paper is EF999921.
Journal of General Virology (2008), 89, 359–368 DOI 10.1099/vir.0.83286-0
0008-3286 G 2008 SGM Printed in Great Britain 359

Genetic investigation of cytomegaloviruses (CMV) has
improved because of cloning of CMV genomes into
bacterial artificial chromosomes (BAC) (Borst et al.,
1999; Messerle et al., 1997). This technique (i) greatly
facilitates genetic manipulation of viral genes in the context
of the viral genome, (ii) enables amplification of the
genome in the absence of selective pressure, and (iii) yields
clonal viral genomes without the need of plaque purifica-
tions (Brune et al., 1999; Messerle et al., 2000). Here, we
report generation of BACs derived from TB40/E in order to
yield genetically pure and highly endotheliotropic clones of
this virus and to test the hypothesis that the phenotype of
variants contained within TB40/E is determined by the
UL128-UL131A gene region previously linked to loss of EC
tropism during extended fibroblast adaptation (Adler et al.,
2006; Hahn et al., 2004; Wang & Shenk, 2005b).
METHODS
Cells. Human foreskin fibroblasts (HFF) were cultured in minimum
essential medium (MEM; Gibco) containing 5% fetal calf serum
(FCS), 2.4 mmol glutamine l21 and 100 mg gentamicin ml21
(designated fibroblast medium), and were used for experiments at
passage 10–25. Human umbilical vein endothelial cells (HUVEC)
were cultured in RPMI 1640 containing gentamicin (100 mg ml21),
heparin (5 IU ml21), endothelial cell growth supplement (50 mg
ml21; Becton Dickinson) and human serum (10%; seronegative for
HCMV) (designated EC medium), and were used for experiments at
passage 4–7. All cell preparations tested negative by 4,6-diamidino-2-
phenylindole (DAPI) staining for mycoplasmas.
Viruses. HCMV strain TB40/E was derived in our laboratory from
throat wash of a bone marrow transplant recipient by propagation for
5 passages in fibroblasts and 22 passages in EC (Sinzger et al., 1999).
For preparation of virus stocks, HFF were infected at an m.o.i. of 0.1
p.f.u. per cell. Supernatants of infected cultures were harvested 6 days
after infection and stored at 280 uC after removal of cell debris by
centrifugation for 10 min at 2800 g. The infectious titre in HCMV
preparations was determined by TCID50 assays in fibroblasts on 96-
well plates (Mahy & Kangro, 1996).
Generation of BACs. The EC-propagated HCMV strain TB40/E was
cloned as a BAC in Escherichia coli as described previously (Hahn
et al., 2002). Briefly, 107 HFF were transfected with 35 mg plasmid
pEB1097 containing a tk-gpt-bac-cassette flanked with HCMV
homologous sequences of US1-US2 (AD169 nt 192 648–193 360;
GenBank accession no. X17403) on the right side and US6-US7
(AD169 nt 195 705–197 398) on the left side of the cassette. After
24 h, the monolayer was infected with TB40/E at an m.o.i. of 5. Three
rounds of selection with 100 mM xanthine and 25 mM mycophenoloic
acid followed. Circular episomal DNA was extracted using the
method of Hirt (1967) and electroporated into E. coli DH10B using a
Bio-Rad Gene Pulser II (2.5 kV, 25 mF, 200 V). Bacteria were then
plated onto agar plates containing 12.5 mg chloramphenicol ml21.
After 24 h, colonies were picked and grown in liquid culture for BAC
preparation. DNA of 18 clones was used to reconstitute infectious
virus by transfection of fibroblasts (MRC-5) as reported previously
(Borst et al., 1999). The BAC cloned TB40/E strains were referred to
as TB40E-BAC1–18, respectively. Guided by a restriction fragment
length analysis (RFLA) of the BAC genomes, nine clones (nos 1, 2, 3,
4, 6, 9, 10, 12 and 18) were chosen for phenotypic characterization.
For the generation of mutants and sequencing, genomic DNA from
reconstituted and phenotyped viruses HCMV-TB40E-BAC1 and
HCMV-TB40E-BAC4 was retransformed into E. coli and then used
for downstream applications. This procedure has the advantage that
genomic changes due to reconstitution in fibroblasts are represented
in the sequence of the retransformed clones. To allow distinction
from the initial clones, the reconstituted BAC clones were assigned
TB40-BAC1 and TB40-BAC4, and the respective viruses were assigned
HCMV-TB40-BAC1 and HCMV-TB40-BAC4. Omission of the suffix
E is also meant to avoid mistaking BAC1 as an endotheliotropic clone.
Regarding virus titre production and EC tropism, HCMV-TB40E-
BAC4 and HCMV-TB40-BAC4 were indistinguishable (data not
shown).
Markerless BAC mutagenesis. The markerless replacement of
TB40-BAC1-UL128 sequences (nt 288–368) by the respective TB40-
BAC4 sequences was done according to the ‘en passant’ method
previously published by Tischer et al. (2006). A recombination
fragment was amplified from plasmid pEPKan-S using forward and
reverse primers that contained 60 and 59 nt, respectively, of homology
to the TB40-BAC4-UL128 sequence and 25 or 22 nt, respectively, of
homology to pEPKan-S. This recombination fragment was introduced
into TB40-BAC1 by a first Red recombination, resulting in a selectable
BAC with an ISceI restriction site and a kanamycin cassette flanked by a
duplication of the TB40-BAC4-UL128 target sequence. After successful
kanamycin selection, all non-HCMV sequences were removed from
this BAC by an intrabacterial ISceI digest and a subsequent Red
recombination, resulting in scarless ‘repair’ of the UL128 gene in the
background of TB40-BAC1. Using this two-step (‘en passant’)
mutagenesis protocol, two independently generated BACs were
generated and termed TB40-BAC1-UL128repair-1 and TB40-BAC1-
UL128repair-2. Using the same approach, an adenine residue was
inserted at position 332 of UL128 in TB40-BAC4, resulting in a BAC
termed TB40-BAC4-UL128insA332.
RFLA. Viral DNA was digested with either EcoRI, BamHI, XbaI or
HindIII. DNA samples were then separated by electrophoresis on 1%
agarose gels, stained with ethidium bromide and visualized by
transillumination with UV light.
Focus expansion (FE) assays. The capability of reconstituted
viruses to grow in EC cultures was quantified by FE assays as
described previously (Sinzger et al., 1997). Briefly, frozen infected cells
were thawed, washed and co-cultured in 96-well plates together with
either uninfected HFF or uninfected HUVEC. To quantify HCMV
replication in EC cultures (FEHUVEC), 26104 uninfected HUVEC per
well were co-cultured with serial dilutions of infected fibroblasts (104–
100) for 5 days at 37 uC with 5% CO2 in EC medium. To determine
HCMV replication in fibroblast cultures (FEHFF), 26104 uninfected
HFF per well were co-cultured in the same way, using fibroblast
medium. After 5 days of co-cultivation, cells were fixed with cold
methanol and HCMV immediate-early (IE) antigen was detected by
indirect immunoperoxidase staining with monoclonal antibody E13
(Biosoft), peroxidase-conjugated goat anti-mouse-IgG Fab92 and the
chromogen diaminobenzidine (Sigma). All tests were done in
quadruplicate. Stained dishes were analysed with an Axiovert 135
microscope (Zeiss). Infectious foci were defined as clusters of three or
more antigen-positive cells. The number of infected cells in the largest
focus in each of the four parallel tests was counted. The highest and
the lowest counts were always eliminated and the mean values of the
remaining two counts were defined as the FE value of the respective
strain, thus ensuring highly reliable results (Sinzger et al., 1997).
Determination of infection efficiency with cell-free virus. HFF
and HUVEC grown on gelatin-coated 96-well plates (mclear; Greiner)
were washed with fresh MEM medium with 5% FCS for 30 min and
then incubated with the respective virus preparation at an m.o.i. of
0.7. After infection (2 h), virus preparations were replaced by fresh
cell culture medium and cells were incubated overnight. Cells were
C. Sinzger and others
360 Journal of General Virology 89

then fixed with 80% acetone for 5 min at room temperature. For
immunofluorescence detection of infected cells, the fixed cultures
were subsequently incubated with monoclonal antibody (mAb) E13,
directed against HCMV-IE antigen (Biosoft) and Cy3-conjugated
goat anti-mouse-IgG Fab92 (Jackson ImmunoResearch). Finally,
nuclei were counterstained with DAPI. Stainings were read under a
Zeiss Axiovert 200 microscope and documented using Axiovision
software.
Nuclear localization assay. HFF and HUVEC grown on gelatin-
coated 96-well plates (mclear; Greiner) were washed with fresh MEM
medium with 5% FCS for 30 min and then incubated with the
respective virus preparation at an m.o.i. of 5 p.f.u. per cell. After
infection (1 h), virus preparations were replaced by fresh cell culture
medium and cells were incubated for an additional 5 h. Cells were
then fixed with 80% acetone for 5 min at room temperature. For
immunofluorescence detection of virus particles, the fixed cultures
were subsequently incubated with mAb XP1 directed against the
capsid-associated HCMV-tegument protein pp150 (Behringwerke)
and Alexa Fluor 488-conjugated goat anti-mouse-IgG Fab92
(Molecular Probes). Remaining binding sites for mouse antibodies
were blocked with mouse serum. For immunofluorescence detection
of microtubules, cells were then incubated with Cy3-labelled mouse
anti-b-tubulin (Sigma). Finally, nuclei were counterstained with
DAPI. Stainings were visualized under a Zeiss Axiovert 200
microscope and documented using Axiovision software.
DNA sequence analyses. Sequence comparison of the UL128-
UL131A gene region from TB40-BAC1 and TB40-BAC4 was done by
SequiServe. Complete sequencing of TB40-BAC4 was done by
Macrogen custom sequencing service using shotgun sequencing with
six times coverage and primer walking to fill the remaining gaps.
Sequence data were analysed and aligned using CLUSTAL W (Chenna
et al., 2003) and BioEdit v.5.0.9 (Hall, 2001). The nucleotide sequence
was annotated using Lasergene SeqBuilder software (DNASTAR) and
deposited at GenBank (accession no. EF999921).
RESULTS
Generation, reconstitution and phenotypic
characterization of BAC clones from HCMV strain
TB40/E
BAC clones were generated from the highly endothelio-
tropic HCMV strain TB40/E in order to serve two
purposes: (i) the availability of TB40/E-derived BACs
would greatly facilitate genetic manipulations of this virus;
(ii) reconstitution of virus from TB40/E-derived BAC
clones would overcome the previously reported problem of
genetic heterogeneity found within TB40/E despite
repeated rounds of plaque purifications.
For the generation of TB40/E-derived BACs, the US2-US6
genome region of the viral genome was replaced with the
selectable F-ori-containing plasmid pEB1997 by homolog-
ous recombination in TB40/E-infected HFFs, and success-
fully recombined viral genomes were enriched by selection
with xanthine/mycophenolic acid. Circular viral DNA was
extracted from infected HFFs and transformed into E. coli
DH10B. After transfection of DNA from 18 bacterial clones
into HFFs, infectious virus could be reconstituted from all
TB40-BACs, and based on an initial RFLA nine TB40-BACs
were chosen for further phenotypic and genotypic analyses.
RFLAs of DNA from these viruses confirmed the previously
reported finding that TB40/E was not genetically homo-
geneous despite repeated plaque purifications. Three
distinct restriction fragment profiles were found after
digestion with enzymes BamHI, EcoRI, HindIII and XbaI
(Table 1). Profile 1 was represented by three HCMV-TB40-
BAC clones (1, 2 and 4); profile 2 was represented by five
HCMV-TB40-BAC clones (3, 6, 10, 12 and 18); and profile
3 was only represented by HCMV-TB40-BAC clone 9.
Differences are not due to different orientations of US and
UL segments in the TB40-BAC clones, as RFLAs have been
performed with the reconstituted viruses that are known to
contain all four isoforms at equal stoichiometry (Kilpatrick
& Huang, 1977, McVoy & Ramnarain, 2000). Therefore,
the three RFLA profiles most likely represent true genetic
variants present in TB40/E.
The EC tropism of all successfully reconstituted infectious
clones was determined by FE assays. Only two virus clones
(reconstituted from TB40-BAC4 and TB40-BAC12) could
spread efficiently in HUVEC monolayers as determined by
the number of infected cells per focus after 7 days of
cultivation. (Fig. 1a). Unexpectedly, the genetic pattern as
determined by RFLA did not correlate with the phenotype
Table 1. Genotype and phenotype of viruses reconstituted from TB40/E-derived BACs
RFLA profile: for each enzyme, the different restriction fragment patterns are assigned with capital letters in the order of appearance. The three
different combinations of restriction profiles were assigned with numbers in the order of appearance. EC tropism: the number of infected cells/focus
after 7 days of focal expansion from one productively infected cell in HUVEC monolayers is given as FEHUVEC.
BAC1 BAC2 BAC3 BAC4 BAC6 BAC9 BAC10 BAC12 BAC18
RFLA pattern 1 1 2 1 2 3 2 2 2
BamHI A A B A B C B B B
EcoRI A A B A B C B B B
HindIII A A B A B B B B B
XbaI A A A A A A A A A
EC tropism Low Low Low High Low Low Low High Low
FEHUVEC ,3 ,3 ,3 100 ,3 ,3 ,3 44 ,3
HCMV strain TB40/E-derived BAC clones
http://vir.sgmjournals.org 361

(Table 1). For further phenotypic characterization, recon-
stituted viruses HCMV-TB40-BAC1 and HCMV-TB40-
BAC4 were chosen, because they represented a maximal
difference in EC tropism on the background of an identical
RLFA pattern (Fig. 1b), thus making them particularly
suitable for genotypic comparisons.
To test whether HCMV-TB40-BAC4 and HCMV-TB40-
BAC1 were suitable to represent the phenotypic differences
described previously for the parental strains TB40/E and its
poorly endotheliotropic counterpart TB40/F, respectively
(Sinzger et al., 2000), we analysed the infection capacity of
cell-free virus in HUVEC cultures and the efficiency of
nuclear translocation. As expected, the phenotypic differ-
ence in FE assays was also reflected in the ‘cell-free’
infection mode. When HUVEC were infected for 24 h at an
m.o.i. of 0.7 p.f.u. per cell as normalized on HFF, only
HCMV-TB40-BAC4 infected HUVEC at high efficiency
(35%), whereas ,0.1% of infected cells were found to
have HCMV-TB40-BAC1 (Fig. 2a). This correlated well
with a difference in the efficiency of nuclear translocation
of virions after EC entry. When virus particles were
visualized at 6 h after infection by indirect immunofluor-
escence against the capsid-associated HCMV-tegument
protein pp150, only particles of HCMV-TB40-BAC4
Fig. 1. Phenotype and genotype of viruses reconstituted from
TB40/E-derived BAC clones. (a) Detection of viral IE antigens in
HUVEC monolayers 7 days after co-culture of productively
infected HFF with an excess of uninfected HUVEC indicator cells
(ratio 1 : 3000). (b) Restriction fragment patterns of DNA from
reconstituted TB40/E-BACs numbered 1, 3, 4 and 12.
Arrowheads point to fragments differing between these clones.
(a)
(b)
Fig. 2. Phenotype of BAC clones 1 and 4 derived from HCMV
TB40/E. (a) Infection efficiency of HCMV-TB40-BAC1 and
HCMV-TB40-BAC4 in fibroblasts (HFF) and endothelial cells
(HUVEC) was compared by detection of viral IE antigens
(pUL122/123) 24 h after infection at an infection multiplicity of
0.7. Viral antigen was visualized by indirect immunofluorescence
(Cy3, red nuclear signals). Counterstaining was done with DAPI
(blue nuclear signals). (b) Nuclear localization efficiency of
incoming virus was compared in HUVEC at 6 h after infection at
an infection multiplicity of 5. Viral structural antigen pUL32 was
visualized by indirect green immunofluorescence (punctate sig-
nals). Cellular tubulin was visualized by indirect red immunofluor-
escence (tubular signals). Counterstaining was done with DAPI
(blue nuclear signals).
C. Sinzger and others
362 Journal of General Virology 89

translocated efficiently (.50%) towards the nucleus of
infected HUVEC, whereas particles of HCMV-TB40-BAC1
remained in the periphery of the cell (nuclear translocation
,1%, Fig. 2b). HCMV-TB40E-BAC12 behaved similarly
to HCMV-TB40-BAC4 (data not shown), further corrob-
orating the correlation between nuclear translocation and
successful infection. Single-step growth curves showed that,
apart from a slight delay in fibroblasts, HCMV-TB40-
BAC4 greatly resembles the parental virus TB40/E with
regard to release of high titres from both infected
fibroblasts and EC as compared with the more cell-
associated HCMV strain FIXBAC (Fig. 3).
Sequence comparison of TB40-BAC4 and
TB40-BAC1 within the UL128-UL131A gene region
and genetic transfer of TB40-BAC4-UL128 into
TB40-BAC1
Interstrain differences in EC tropism between highly
passaged HCMV strains and low passage HCMV strains
have been linked to the genomic region UL128-UL131A.
To analyse whether interstrain differences within this
genomic region also determine cell tropism differences
between HCMV-TB40-BAC1 and HCMV-TB40-BAC4, we
compared the respective DNA sequences of these BACs.
Fragments of the respective gene region were amplified
from TB40-BAC1 and TB40-BAC4 using suitable primers,
and the nucleic acid sequence was then determined and
compared. Not a single base difference was found between
the two BACs in UL130 and UL131A, whereas two single-
nucleotide variations were detected in UL128. When
compared to TB40-BAC4, TB40-BAC1 showed one
adenine-to-cytosine exchange at nucleotide position 282
and an adenine insertion at nucleotide position 332 of the
UL128 gene sequence (Fig. 4a). The base exchange A282C
is located in the first intron and was therefore unlikely to
alter the amino acid sequence of the respective protein. In
contrast, the 332A insertion is located within the second
exon and causes a frame shift resulting in a truncated
pUL128 (Fig. 4b).
Next, we tested the hypothesis that the additional adenine
at position 332 of UL128 is responsible for the loss of EC
tropism of TB40-BAC1-derived viruses. We used the
markerless ‘en passant’ mutagenesis previously published
by Tischer et al. (2006) to remove the inserted adenine
specifically from the TB40-BAC1 genome without leaving
any further alteration (Fig. 5a). The obtained TB40-BAC1-
UL128repair was reconstituted in fibroblasts, grown to
high titres in HFF to avoid any selection for EC tropism,
and then compared to HCMV-TB40-BAC1 and HCMV-
TB40-BAC4 for infectivity in HUVEC. HUVEC and HFF
were infected with all virus preparations at an m.o.i. of 1
p.f.u. per cell (normalized in HFF), and the ratio of
infected cells was determined by immunofluorescent
detection of IE antigen at 24 h after infection. The relative
infection efficiency in HUVEC/HFF was 0.79 for HCMV-
TB40-BAC4, 0.01 for HCMV-TB40-BAC1 and 0.84 for
HCMV-TB40-BAC1-UL128repair (Fig. 5b and c). To
further corroborate this finding, the reverse experiment
was performed, inserting an adenine at position 332 of
UL128 of TB40-BAC4. The reconstituted virus of this
mutant had a relative infection efficiency in HUVEC/HFF
of 0.01 (Fig. 5b and c), thus resembling HCMV-TB40-
BAC1 which also carries the additional adenine.
Obviously, an insertion of adenine at position 332 of
UL128 was the only coding difference between TB40-BAC4
and TB40-BAC1 in the previously described tropism-
relevant gene region, and removal of this additional
adenine was sufficient to restore EC tropism in the
genomic background of TB40-BAC1.
Fig. 3. Single-step growth curves of HCMV-TB40/E, HCMV-
TB40-BAC4 and HCMV-FIXBAC. Fibroblasts and EC were
infected at an infection multiplicity of 1. Supernatant was collected
daily from 3 to 7 days post-infection. The titre of infectious virus
released by infected cell cultures was determined on fibroblast
monolayers by limiting dilution analyses.
HCMV strain TB40/E-derived BAC clones
http://vir.sgmjournals.org 363

Sequence alignment of TB40-BAC4 with various
HCMV genomes
After successful generation of TB40/E-derived BACs and
proof of their suitability for genetic modification by
markerless removal or insertion of a single nucleotide in
TB40-BAC1 or TB40-BAC4, respectively, we sought to
obtain the genomic sequence and reading frame annotation
of the highly endotheliotropic TB40-BAC4 as a basis for its
future use by the scientific community. Sequencing was
performed by a commercial service using a BAC DNA
preparation that had been tested for its integrity by RFLA,
virus reconstitution after transfection and phenotypical
testing of the reconstituted virus in HUVEC.
From the obtained BAC sequence, non-HCMV sequences of
the pEB1997 vector backbone were removed, and the
sequence was positioned to start with the US segment
adjacent to the BAC cassette (US7–US34), followed by the
repeat regions and the UL segment. It is noteworthy that two
neighbouring open reading frames (ORFs), IRS1 and US1,
are missing in the TB40-BAC4 sequence. This is surprising
because US1 and a part of US2 were used as the left flanking
homology arm for the insertion of the BAC vector into the
TB40/E genome. Indeed, a short part of US2 is present in
TB40-BAC4, but US1 and IRS1 are absent. The reason for
this is unknown, but it seems likely that it was caused by an
illegitimate recombination event during the insertion of the
BAC vector by homologous recombination in fibroblasts.
Interestingly, a similar unanticipated deletion was found in
the HCMV FIXBAC (Murphy et al., 2003b), which was
constructed using the same BAC vector and flanking
homologous arms (Hahn et al., 2002). With regard to
previously reported TB40/E-variants, TB40-BAC4 resembles
the Bart strain in that UL141 has a frameshift insertion at
codon 63 (Tomasec et al., 2005); however, unlike strain Bart,
UL144 andUL145 are intact and thus TB40-BAC4 is identical
to the TB40/E sequence published by Dolan et al. (2004) in
all three genes. The entire TB40-BAC4 sequence was
annotated in analogy to other previously annotated HCMV
strains and has been deposited under the designation TB40-
BAC4 at the GenBank database (accession no. EF999921).
In order to get an idea about the relation of TB40-BAC4 to
the other published HCMV genomes, sequences from eight
HCMV strains were aligned using CLUSTAL W and similarity
plots were performed comparing each genome with the
consensus sequence obtained from the CLUSTAL W alignment.
To enable an alignment, all genomes except TB40-BAC4
were transformed to prototype orientation in advance.
Similarity plots showed that TB40-BAC4 is perfectly
collinear with the HCMV ‘consensus sequence’ and has no
gross deletions, except the US2-US6 region, which has been
replaced by the BAC cassette and the adjacent IRS1-US1
region. A more detailed phylogenetic analysis of several
glycoproteins showed that most of them can be individually
classified into well defined genotypes except UL73 which is
too polymorphic. TB40-BAC4 is closely related to FIXBAC
in UL73, UL74, UL75, UL115 and UL120, closely related to
Merlin in UL55 and UL119, closely related to AD169 in
UL119 and UL120, closely related to TR-BAC in UL75 and
UL119, closely related to PH-BAC in UL100 and closely
related to Toledo in UL74 (Fig. 6).
Few genomic regions of TB40-BAC4 deviate from all other
strains: RL6 is almost completely missing in TB40-BAC4,
whereas it is present in all other strains. UL10 is highly
divergent with an identity at the protein level of only 86.8%
between TB40-BAC4 and AD169. pUL77 of TB40-BAC4
lacks 12 aa, resulting in an amino acid identity of 97.8%.
pUL84 is 97.1% identical at the amino acid level including a
4 aa deletion in TB40-BAC4. pUL28 of TB40-BAC4 shows
moderate divergence in the N-terminal half, resulting in
98% identity at the amino acid level.
Fig. 4. Comparison of the ORF UL128 of TB40-BAC1 and
TB40-BAC4 (a) on the level of the experimentally determined DNA
sequence and (b) on the level of the predicted amino acid
sequence. Grey shading highlights nucleotides or amino acids that
differ in TB40-BAC1 as compared with TB40-BAC4.
C. Sinzger and others
364 Journal of General Virology 89

Taken together, the information from the sequence
comparison demonstrated that, apart from the replacement
of the US1-US7 gene region by the BAC cassette, plus defects
in RL6 and UL141, TB40-BAC4 appears to possess a
relatively intact HCMV genome, which is collinear to all
other strains and shares well defined glycoprotein genotypes.
DISCUSSION
The highly endotheliotropic strain TB40/E has been widely
used for HCMV infection of EC, macrophages and
dendritic cells. However, molecular analyses have been
hindered by the fact that TB40/E was not available as a
molecular clone. Furthermore, TB40/E was reported to be
genetically heterogeneous (Dolan et al., 2004). By genera-
tion of TB40/E-derived BACs we have now solved both
problems. TB40-BAC4 together with the genome sequence
is now available for reconstitution of this highly endothe-
liotropic virus in a genetically pure and manipulable form.
The phenotypic and genotypic analysis of virus strains
reconstituted from nine TB40/E-derived BACs confirmed
the previous notion about heterogeneity of the parental
virus. Heterogeneity was observed in restriction fragment
patterns and cell tropism, though they did not correlate. A
closer analysis of the UL128-UL131A genomic region
revealed an additional nucleotide in the UL128 reading
frame of TB40-BAC1, which leads to a frameshift, a
truncated protein product and a loss of EC tropism. Thus,
it is apparent that BAC cloning is suitable to obtain
genetically defined virus clones even in cases where plaque
purification failed to serve this purpose.
Identification of the subtle mutations underlying the cell
tropism difference between RFLA-identical clones HCMV-
TB40-BAC1 and HCMV-TB40-BAC4 further confirmed
previous reports, which demonstrated that loss of EC
tropism during extended cell culture propagation is due to
changes in the UL128-UL131A gene region (Akter et al.,
2003; Adler et al., 2006; Hahn et al., 2004; Patrone et al.,
2005; Wang & Shenk, 2005b). Targeted comparison of both
BAC clones in the UL128-UL131A gene region revealed a
single coding-relevant mutation only in UL128, and
transfer of the UL128-BAC4 genotype into the background
of TB40-BAC1 fully restored a high EC tropism in
Fig. 5. Genetic marker transfer from the TB40-BAC4-UL128 sequence into the TB40-BAC1 background. (a) Removal of the
additional UL128-332A from the TB40-BAC1 genome by markerless mutagenesis yielded the recombinant BAC TB40-BAC1-
UL128repair. Likewise, insertion of an adenine at position 332 of UL128 in the TB40-BAC4 genome yielded the mutant BAC
TB40-BAC4-UL128insA332. (b) Cell tropism of viruses reconstituted from TB40-BAC1-UL128repair and TB40-BAC4-
UL128insA332 as compared with wild-type viruses reconstituted from TB40-BAC1 and TB40-BAC4. Fibroblasts (HFF) and
endothelial cells (HUVEC) were infected at an infection multiplicity of 0.7. Viral IE antigen was visualized by indirect
immunofluorescence at 24 h after infection (red nuclear fluorescence). Counterstaining was done with DAPI (blue nuclear
fluorescence). (c) Quantitative evaluation of three experiments as shown in panel (b). Given are mean values of the ratios of
infection efficiencies in HUVEC and HFF (+/–SD).
HCMV strain TB40/E-derived BAC clones
http://vir.sgmjournals.org 365

HCMV-TB40-BAC1-UL128repair, which was indistin-
guishable from the phenotype of HCMV-TB40-BAC4.
Vice versa, insertion of an adenine residue at the respective
site in TB40-BAC4 destroyed its endotheliotropic pheno-
type. This is the first example illustrating restoration of the
endotheliotropic phenotype in a poorly endotheliotropic
HCMV UL128 mutant by orthotopic markerless repair of
the respective gene, thus supporting previous findings with
rescue of HCMV strain Merlin by transcomplementation
(Hahn et al., 2004). As a recent sequence comparison of 34
clinical isolates did not detect any major alterations of the
UL128 coding sequence such as frameshifts or premature
stop codons among these isolates (Baldanti et al., 2006),
the adenine insertion in TB40-BAC1 was most probably
acquired during the initial propagation of TB40/E in
fibroblast cultures. Obviously, although other gene regions
were also reported to contribute to EC tropism as
evidenced by loss of EC tropism after deletion of the
respective ORF (Dunn et al., 2003; Pretsch et al., 2005),
changes due to cell culture adaptation apparently target
preferentially the UL128-UL131A gene region. An explana-
tion for this is provided by recent reports on the inhibition
of virus release of strain AD169 after repair of a defective
UL131A or insertion of an intact UL131A ORF (Adler et al.,
2006; Wang & Shenk, 2005a). Loss of the protein complex
formed by gH/gL and the pUL128-131A gene products
obviously provides a growth advantage in fibroblasts at the
cost of a restricted cell tropism.
Maintenance of broad cell tropism, including EC tropism,
is often at the cost of low titre virus production (Adler et al.,
2006). HCMV-TB40-BAC4 is different as it combines both
a high EC tropism and high titre virus release (Fig. 3), thus
making it particularly suitable for applications where large
amounts of virions are required, like electron microscopy or
proteomic analyses of cell-free virus particles. The molecular
determinants that allow for efficient release of infectious
progeny while maintaining EC-tropism are widely
unknown, albeit some hints may be derived from the whole
genome comparison of TB40-BAC4 with other HCMV
strains. One region where TB40-BAC4 differs from all other
strains is the RL11 family. Particularly, TB40-BAC4 shows a
loss of RL6 and a high degree of variation in UL10. Although
some researchers have suggested RL6 may not be translated
(Murphy et al., 2003a), this issue deserves further invest-
igation. Alternatively, the minor deviation of TB40-BAC4 in
UL28, UL77 and UL84 might also account for the unique
features of this strain. The availability of TB40/E-derived
BACs and the genome sequence of TB40-BAC4 should
facilitate investigations of the contribution of these genes by
the construction of the respective mutants.
In conclusion, we have generated a BAC derived from the
HCMV strain TB40/E that combines the high titre growth
of a cell culture-adapted strain with the broad cell tropism
of a clinical HCMV isolate. Analysis of its genomic
sequence showed a close relationship to the endothelio-
tropic strain FIXBAC, but also revealed that TB40-BAC4
shares the sequence of certain highly polymorphic genes
with each of the other strains. The usability of TB40-
derived BACs for genetic manipulations was proved by a
marker transfer from TB40-BAC4 to TB40-BAC1 and vice
versa, thus demonstrating the importance of the C-
terminal part of pUL128 for EC tropism.
Fig. 6. Genomic comparison of TB40-BAC4 with other HCMV strains. Genotype trees obtained from phylogenetic analyses of
the coding sequence of selected glycoproteins.
C. Sinzger and others
366 Journal of General Virology 89

ACKNOWLEDGEMENTS
This work was supported by grants from the DFG through SI 779/3-2,
AD 131/2-3, HE 2526/6-2, SFB 421 project B14 (W. B.), and SFB 587
project A13 (M.M.).
REFERENCES
Adler, B., Scrivano, L., Ruzcics, Z., Rupp, B., Sinzger, C. &
Koszinowski, U. (2006). Role of human cytomegalovirus
UL131A in cell type-specific virus entry and release. J Gen Virol 87,
2451–2460.
Akter, P., Cunningham, C., McSharry, B. P., Dolan, A., Addison, C.,
Dargan, D. J., Hassan-Walker, A. F., Emery, V. C., Griffiths, P. D. &
other authors (2003). Two novel spliced genes in human cyto-
megalovirus. J Gen Virol 84, 1117–1122.
Allal, C., Buisson-Brenac, C., Marion, V., Claudel-Renard, C., Faraut, T.,
Dal Monte, P., Streblow, D., Record, M. & Davignon, J. L. (2004).
Human cytomegalovirus carries a cell-derived phospholipase A2
required for infectivity. J Virol 78, 7717–7726.
Baldanti, F., Paolucci, S., Campanini, G., Sarasini, A., Percivalle, E.,
Revello, M. G. & Gerna, G. (2006). Human cytomegalovirus UL131A,
UL130 and UL128 genes are highly conserved among field isolates.
Arch Virol 151, 1225–1233.
Bentz, G. L., Jarquin-Pardo, M., Chan, G., Smith, M. S., Sinzger, C. &
Yurochko, A. D. (2006). Human cytomegalovirus (HCMV) infection
of endothelial cells promotes naive monocyte extravasation and
transfer of productive virus to enhance hematogenous dissemination
of HCMV. J Virol 80, 11539–11555.
Borst, E. M., Hahn, G., Koszinowski, U. H. & Messerle, M. (1999).
Cloning of the human cytomegalovirus (HCMV) genome as
an infectious bacterial artificial chromosome in Escherichia coli: a
new approach for construction of HCMV mutants. J Virol 73,
8320–8329.
Brune, W., Menard, C., Hobom, U., Odenbreit, S., Messerle, M. &
Koszinowski, U. H. (1999). Rapid identification of essential and
nonessential herpesvirus genes by direct transposon mutagenesis. Nat
Biotechnol 17, 360–364.
Chenna, R., Sugawara, H., Koike, T., Lopez, R., Gibson, T. J., Higgins,
D. G. & Thompson, J. D. (2003). Multiple sequence alignment with the
Clustal series of programs. Nucleic Acids Res 31, 3497–3500.
Dolan, A., Cunningham, C., Hector, R. D., Hassan-Walker, A. F., Lee, L.,
Addison, C., Dargan, D. J., McGeoch, D. J., Gatherer, D. & other
authors (2004). Genetic content of wild-type human cytomegalovirus.
J Gen Virol 85, 1301–1312.
Dunn, W., Chou, C., Li, H., Hai, R., Patterson, D., Stolc, V., Zhu, H. &
Liu, F. (2003). Functional profiling of a human cytomegalovirus
genome. Proc Natl Acad Sci U S A 100, 14223–14228.
Gerna, G., Percivalle, E., Baldanti, F. & Revello, M. G. (2002). Lack of
transmission to polymorphonuclear leukocytes and human umbilical
vein endothelial cells as a marker of attenuation of human
cytomegalovirus. J Med Virol 66, 335–339.
Hahn, G., Khan, H., Baldanti, F., Koszinowski, U. H., Revello, M. G. &
Gerna, G. (2002). The human cytomegalovirus ribonucleotide
reductase homolog UL45 is dispensable for growth in endothelial cells,
as determined by a BAC-cloned clinical isolate of human cytomegalo-
virus with preserved wild-type characteristics. J Virol 76, 9551–9555.
Hahn, G., Revello, M. G., Patrone, M., Percivalle, E., Campanini, G.,
Sarasini, A., Wagner, M., Gallina, A., Milanesi, G. & other authors
(2004). Human cytomegalovirus UL131–128 genes are indispensable
for virus growth in endothelial cells and virus transfer to leukocytes.
J Virol 78, 10023–10033.
Hall, T. (2001). BioEdit, 5.0.6 edn. Raleigh: Department of
Microbiology, North Carolina State University.
Hertel, L., Lacaille, V. G., Strobl, H., Mellins, E. D. & Mocarski, E. S.
(2003). Susceptibility of immature and mature Langerhans cell-type
dendritic cells to infection and immunomodulation by human
cytomegalovirus. J Virol 77, 7563–7574.
Hirt, B. (1967). Selective extraction of polyoma DNA from infected
mouse cell cultures. J Mol Biol 26, 365–369.
Homman-Loudiyi, M., Hultenby, K., Britt, W. & Soderberg-Naucler, C.
(2003). Envelopment of human cytomegalovirus occurs by budding
into Golgi-derived vacuole compartments positive for gB, Rab 3, trans-
golgi network 46, and mannosidase II. J Virol 77, 3191–3203.
Jarvis, M. A. & Nelson, J. A. (2002). Mechanisms of human
cytomegalovirus persistence and latency. Front Biosci 7, d1575–d1582.
Kilpatrick, B. A. & Huang, E. S. (1977). Human cytomegalovirus
genome: partial denaturation map and organization of genome
sequences. J Virol 24, 261–276.
King, C. A., Baillie, J. & Sinclair, J. H. (2006). Human cytomegalovirus
modulation of CCR5 expression on myeloid cells affects susceptibility
to human immunodeficiency virus type 1 infection. J Gen Virol 87,
2171–2180.
Laib Sampaio, K. L., Cavignac, Y., Stierhof, Y. D. & Sinzger, C. (2005).
Human cytomegalovirus labeled with green fluorescent protein for live
analysis of intracellular particle movements. J Virol 79, 2754–2767.
Mahy, B. W. J. & Kangro, H. O. (1996). Virology Methods Manual,
pp. 35–37. San Diego: Academic Press.
McVoy, M. A. & Ramnarain, D. (2000). Machinery to support genome
segment inversion exists in a herpesvirus which does not naturally
contain invertible elements. J Virol 74, 4882–4887.
Messerle, M., Crnkovic, I., Hammerschmidt, W., Ziegler, H. &
Koszinowski, U. H. (1997). Cloning and mutagenesis of a herpesvirus
genome as an infectious bacterial artificial chromosome. Proc Natl
Acad Sci U S A 94, 14759–14763.
Messerle, M., Hahn, G., Brune, W. & Koszinowski, U. H. (2000).
Cytomegalovirus bacterial artificial chromosomes: a new herpesvirus
vector approach. Adv Virus Res 55, 463–478.
Moutaftsi, M., Brennan, P., Spector, S. A. & Tabi, Z. (2004). Impaired
lymphoid chemokine-mediated migration due to a block on the
chemokine receptor switch in human cytomegalovirus-infected
dendritic cells. J Virol 78, 3046–3054.
Murphy, E., Rigoutsos, I., Shibuya, T. & Shenk, T. E. (2003a).
Reevaluation of human cytomegalovirus coding potential. Proc Natl
Acad Sci U S A 100, 13585–13590.
Murphy, E., Yu, D., Grimwood, J., Schmutz, J., Dickson, M., Jarvis,
M. A., Hahn, G., Nelson, J. A., Myers, R. M. & Shenk, T. E. (2003b).
Coding potential of laboratory and clinical strains of human
cytomegalovirus. Proc Natl Acad Sci U S A 100, 14976–14981.
Patrone, M., Secchi, M., Fiorina, L., Ierardi, M., Milanesi, G. & Gallina, A.
(2005). Human cytomegalovirus UL130 protein promotes endothelial
cell infection through a producer cell modification of the virion. J Virol
79, 8361–8373.
Pretsch, R., Kleining, V., Schierling, K., Hahn, G., Sinzger, C.,
Mertens, T. & Winkler, M. (2005). Human cytomegalovirus pUL24,
part of a novel protein complex, is necessary for endothelial cell
tropism. In 30th International Herpesvirus Workshop. Turku, Finland.
Reeves, M. B., Lehner, P. J., Sissons, J. G. & Sinclair, J. H. (2005). An
in vitro model for the regulation of human cytomegalovirus latency
and reactivation in dendritic cells by chromatin remodelling. J Gen
Virol 86, 2949–2954.
Reinhardt, B., Schaarschmidt, P., Bossert, A., Luske, A.,
Finkenzeller, G., Mertens, T. & Michel, D. (2005). Upregulation of
HCMV strain TB40/E-derived BAC clones
http://vir.sgmjournals.org 367

functionally active vascular endothelial growth factor by human
cytomegalovirus. J Gen Virol 86, 23–30.
Sinzger, C. & Jahn, G. (1996). Human cytomegalovirus cell tropism
and pathogenesis. Intervirology 39, 302–319.
Sinzger, C., Knapp, J., Plachter, B., Schmidt, K. & Jahn, G. (1997).
Quantification of replication of clinical cytomegalovirus isolates in
cultured endothelial cells and fibroblasts by a focus expansion assay.
J Virol Methods 63, 103–112.
Sinzger, C., Schmidt, K., Knapp, J., Kahl, M., Beck, R., Waldman, J.,
Hebart, H., Einsele, H. & Jahn, G. (1999). Modification of
human cytomegalovirus tropism through propagation in vitro is
associated with changes in the viral genome. J Gen Virol 80,
2867–2877.
Sinzger, C., Kahl, M., Laib, K., Klingel, K., Rieger, P., Plachter, B. &
Jahn, G. (2000). Tropism of human cytomegalovirus for endothelial
cells is determined by a post-entry step dependent on efficient
translocation to the nucleus. J Gen Virol 81, 3021–3035.
Tischer, B. K., von Einem, J., Kaufer, B. & Osterrieder, N. (2006). Two-
step red-mediated recombination for versatile high-efficiency marker-
less DNA manipulation in Escherichia coli. Biotechniques 40, 191–197.
Tomasec, P., Wang, E. C., Davison, A. J., Vojtesek, B., Armstrong, M.,
Griffin, C., McSharry, B. P., Morris, R. J., Llewellyn-Lacey, S. & other
authors (2005). Downregulation of natural killer cell-activating ligand
CD155 by human cytomegalovirus UL141. Nat Immunol 6, 181–188.
Wang, D. & Shenk, T. (2005a). Human cytomegalovirus UL131
open reading frame is required for epithelial cell tropism. J Virol 79,
10330–10338.
Wang, D. & Shenk, T. (2005b). Human cytomegalovirus virion
protein complex required for epithelial and endothelial cell tropism.
Proc Natl Acad Sci U S A 102, 18153–18158.
C. Sinzger and others
368 Journal of General Virology 89