Phylogenetic analysis of the main neutralization and hemagglutination determinants of all human adenovirus prototypes as a basis for molecular classification and taxonomy.

JOURNAL OF VIROLOGY, Dec. 2005, p. 15265–15276 Vol. 79, No. 24
0022-538X/05/$08.00�0 doi:10.1128/JVI.79.24.15265–15276.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.
Phylogenetic Analysis of the Main Neutralization and Hemagglutination
Determinants of All Human Adenovirus Prototypes as a Basis
for Molecular Classification and Taxonomy
Ijad Madisch, Gabi Harste, Heidi Pommer, and Albert Heim*
Institut fu¨r Virologie, Medizinische Hochschule Hannover, Hannover, Germany
Received 8 July 2005/Accepted 29 September 2005
Human adenoviruses (HAdV) are responsible for a wide spectrum of diseases. The neutralization � deter-
minant (loops 1 and 2) and the hemagglutination � determinant are relevant for the taxonomy of HAdV.
Precise type identification of HAdV prototypes is crucial for detection of infection chains and epidemiology. �
and � determinant sequences of all 51 HAdV were generated to propose molecular classification criteria.
Phylogenetic analysis of � determinant sequences demonstrated sufficient genetic divergence for molecular
classification, with the exception of HAdV-15 and HAdV-29, which also cannot be differentiated by classical
cross-neutralization. Precise sequence divergence criteria for typing (<2.5% from loop 2 prototype sequence
and <2.4% from loop 1 sequence) were deduced from phylogenetic analysis. These criteria may also facilitate
identification of new HAdV prototypes. Fiber knob (� determinant) phylogeny indicated a two-step model of
species evolution and multiple intraspecies recombination events in the origin of HAdV prototypes. HAdV-29
was identified as a recombination variant of HAdV-15 (� determinant) and a speculative, not-yet-isolated
HAdV prototype (� determinant). Subanalysis of molecular evolution in hypervariable regions 1 to 6 of the �
determinant indicated different selective pressures in subclusters of species HAdV-D. Additionally, � deter-
minant phylogenetic analysis demonstrated that HAdV-8 did not cluster with -19 and -37 in spite of their
having the same tissue tropism. The phylogeny of HAdV-E4 suggested origination by interspecies recombina-
tion between HAdV-B (hexon) and HAdV-C (fiber), as in simian adenovirus 25, indicating additional zoonotic
transfer. In conclusion, molecular classification by systematic sequence analysis of immunogenic determinants
yields new insights into HAdV phylogeny and evolution.
Adenoviridae viruses are nonenveloped, double-stranded DNA
viruses with an icosahedral capsid (59). Human adenoviruses
(HAdV) are classified into six species (HAdV-A to HAdV-F)
which were defined historically as subgenera on the basis of hem-
agglutination properties (10, 17, 55). Subsequently, oncogenic
properties and DNA homology were also used to define subgen-
era (species) (59). The six HAdV species consist of 51 HAdV
types, defined mainly by neutralization criteria (10, 64). Several
simian adenoviruses (SAdV), which cannot productively infect
humans, are also members of the species HAdV-A (SAdV-2, -4,
-6, -9, -10, -11, and -14), HAdV-B (SAdV-21), HAdV-C (SAdV-
13), HAdV-E (SAdV-22 to -25), and HAdV-F (SAdV-19).
Precise HAdV typing is essential for epidemiological studies
and quality control of diagnostics. Moreover, the greater knowl-
edge about differences in virulence and organ tropism among the
adenovirus serotypes has increased the medical value of HAdV
classification (59). For example, the more dangerous lower respi-
ratory tract illness is associated primarily with HAdV-B3, -B7,
-B21, and -E4 infection (11, 13, 16, 21), whereas the less severe
upper respiratory tract illness is frequently caused by HAdV-C1,
-C2, -C5, -B3, and -B7 (15, 21). Thus, typing of HAdV isolated
from a respiratory tract sample or feces sample may predict the
clinical course. Infections of the eye with HAdV-D8, -D19, and
-D37 may cause severe epidemic keratoconjunctivitis, whereas
mild follicular conjunctivitis and pharyngoconjunctival fever are
most frequently caused by HAdV-B3, -B7, and -E4 (6, 37). Dis-
seminated disease in highly immunosuppressed patients, e.g., al-
logeneic stem cell transplant recipients, is associated with HAdV-
A31, -B3, -B7, -C1, -C2, and -C5, most frequently with the last
three types (20, 32, 54, 57, 63). Acute hemorrhagic cystitis and
acute renal failure in infants, young children, and immunosup-
pressed patients are associated primarily with the serotypes
HAdV-B11, -B21, -B34, and -B35 (3, 12, 20, 23, 38, 57). Due to
this background and the wide spectrum of diseases which are
caused by HAdV of differing organotropism and virulence, a
deeper understanding of the phylogenetic relationships of HAdV
types is crucial for the development of molecular classification
criteria.
All adenoviruses contain three diagnostically useful anti-
gens, which are part of the three capsid proteins: hexon
(polypeptide II), penton (polypeptide III), and fiber (polypep-
tide IV). The main type-specific epitope, the ε determinant,
consisting of loop 1 (L1) and loop 2 (L2) on the hexon protein,
reacts with type-specific antisera in neutralization tests (NT)
(24), which are the classical reference method for typing. Cases
of failing neutralization with available antisera are an unre-
solved issue of NT because these require extensive cross-neu-
tralization studies for defining a new HAdV type. The knob
region of the fiber protein, which includes the type-specific �
determinant, has hemagglutinating properties which are used
for hemagglutination inhibition (HI) tests. HI is preferred by
many laboratories, because it is more convenient and rapid.
However, HI cannot differentiate all 51 HAdV because of
* Corresponding author. Mailing address: Institut fu¨r Virologie,
Medizinische Hochschule Hannover, Carl-Neuberg-Str. 1, D-30625
Hannover, Germany. Phone: 49-511-5324311. Fax: 49-511-5325732.
E-mail: Heim.Albert@mh-hannover.de.
15265

several cross-reactions (59). As HAdV have a high capability
for intraspecies recombination and occasional interspecies re-
combination, HI results of clinical isolates can be contradictory
to NT results (“intermediate strains”). Intermediate strains are
recombinant viruses between two serotypes presenting hexon
neutralization epitopes of one serotype and fiber knob hem-
agglutination epitopes of another serotype. Probably, recom-
bination of HAdV may give rise to new types and even species.
Retargeting of HAdV types with favorable gene expression
and persistence properties (as for example HAdV-C5) with
fiber knobs of other HAdV types holds promise for improving
HAdV gene therapy and vaccination vector approaches (9, 35).
A complete phylogenetic analysis of the fiber knobs of all 51
HAdV should facilitate this approach.
Nowadays, PCRs are frequently used for rapid diagnosis of
adenovirus infections (5, 19, 36, 43). Therefore, attempts were
made to use PCR also for genotyping of HAdV. Techniques
included type-specific PCRs and multiplex PCR protocols us-
ing species-specific primers (28, 42, 43, 45, 69) as well as com-
binations of a generic PCR with restriction digestion of the
amplicons (4, 30, 50, 61). These approaches permit rapid mo-
lecular typing of a limited number of HAdV types or at least
species identification. However, new insights into molecular
evolution and epidemiology of HAdV cannot be attained, and
new HAdV types cannot be identified.
Based on extensive sequence analysis of the neutralizing
epitope ε (L1 and L2 of the hexon capsid protein) and the
hemagglutination epitope �, we developed criteria for molec-
ular classification of HAdV and for the identification of new
prototype isolates. Sequencing of the short L2 proved to be
sufficient for molecular typing whereas additional L1 and �
determinant sequence data should be generated for the iden-
tification of new prototype isolates. L2 sequencing can be rec-
ommended as a second step for precise type identification after
species identification by diagnostic PCRs.
MATERIALS AND METHODS
HAdV prototype strains and cells. HAdV prototype strains HAdV-B50
(ATCC VR-1501), -C1 (VR-1), -D8 (VR-1368), -D15 (VR-1092), -D20 (VR-
255), -D22 (VR-1100), -D23 (VR-258), -D25 (VR-223), -D27 (VR-1105), -D28
(VR-226), -D29 (VR-1107PI/RB), -D30 (VR-273), -D32 (VR-625), -D33 (VR-
626), -D36 (VR-913/275), -D37, -D38 (VR-988), -D39 (VR-932), -D42 (VR-
1304), -D43 (VR-1305), -D44 (VR-1306), -D49 (VR-1407), and -D51 (VR-1502)
were obtained from the American Type Culture Collection (ATCC). HAdV-
B14, -D10, -D13, and -D30 prototype strains were obtained from our collection
at the German National Reference Laboratory for Adenoviruses, Hannover
Medical School, Hannover, Germany, and had been typed previously with cross-
neutralization. In a preliminary quality control study using a multiplex PCR
protocol (69), the virus delivered by ATCC as HAdV-B50 (VR-1501) turned out
to be a member of species D, whereas HAdV-D51 (VR-1502) was a species B
virus. In the following correspondence with ATCC, it was stated that ATCC
believes that stocks of VR-1501 (HAdV-B50) and VR-1502 (HAdV-D51) have
been switched. Subsequently, VR-1501 and VR-1502 were deleted from the
ATCC catalogue. For this study, HAdV-B50 (VR-1501) was relabeled as HAdV-
D51 and vice versa. All viruses were propagated on A549 cells (ATCC, CCL-185)
on 75-cm2 cell culture flasks. When the cytopathogenic effect was above 50%,
cells were frozen at �70°C, and DNA was extracted with the QIAGEN blood kit
(QIAGEN, Hilden, Germany).
Generic HAdV PCR. HAdV DNA was amplified using either a conventional
PCR protocol which was published as a first step of a nested PCR (4) or a
real-time PCR protocol (26).
PCR amplification of hypervariable hexon loops. The hexon L1 region was
amplified as described previously (43). For the PCR amplification of hexon L2
two primer pairs, one for the species HAdV-B and the other for the species
HAdV-C and HAdV-D, were developed (Table 1) with the help of a multiple
alignment with GenBank sequence data flanking the L2 region (HAdV-B3, -B7,
-B14, -D9, -D10, -D17, -D19, -D23, -D24, -D26, -D45, -D46, -D47, and -D48).
Amplification of L2 comprises two primer pairs in individual PCRs, one for the
species HAdV-C and -D (CDL and CDR), which produces a 322-bp product,
and a primer set for the species B (BL and BR) with a 580-bp amplicon
(Table 1). Both primer pairs were designed for an identical annealing tempera-
ture of 54°C as calculated with MeltCalc software to permit amplification with a
single thermal profile (53). Additionally, L1 and L2 primer pairs were designed
for HAdV-A18 to obtain the immunogenically relevant sequences for the phy-
logenetic analysis (Table 1). These pairs have the same PCR conditions as the
corresponding L1 and L2 protocols (Table 1). PCR was performed in a total
volume of 100 �l consisting of HotStar Mix (QIAGEN), 1 �M of each primer,
and 3 �l of the purified HAdV DNA. The PCR program starts with activation of
the “hot start” DNA polymerase for 15 min at 95°C, followed by 40 cycles
consisting of denaturation at 94°C for 20 s, primer annealing at 54°C for 20 s, and
elongation at 72°C for 40 s, followed by a final extension step of 72°C for 5 min.
The sensitivity of the PCR protocols was determined by testing serial dilutions of
cell culture-derived HAdV-B3, -C2, and -D8 previously quantified by a generic
HAdV real-time PCR protocol (25).
PCR amplification of the fiber knob region. For amplification of HAdV-D, a
new sense primer, AdFiDL, and an established antisense primer, AdD2, in the
TABLE 1. Amplification primer pairs for loop 1 (generic) and loop 2 of the hexon capsid protein (species B, C, and D)
and � determinant of the fiber protein (A18, B, and D)
Species Primer Gene Gene region Sequence (5�–3�) Ampliconsize (bp) Positions
Virus for
positions
Generic H1.1 Hexon ε det.a (loop 1) TTTGACATCCGCGGCGT �800 301–317 HAdV-C2
H2.1 GTGTTTCTGTCTTGCAA 1133–1117
B BL Hexon ε det. (loop 2) TTGACTTGCAGGACAGAAA �590 1146–1164 HAdV-B3
BR CTTGTATGTGGAAAGGCAC 1772–1754
C and D CDL Hexon ε det. (loop 2) GTTGACTTGCAAGACAGAAA �322 1188–1207 HAdV-D42
CDR AAACTCYTCCAYAGGTTGGC 1549–1529
A18 A18L1L Hexon ε det. (loop 1) TGGTCCCAGTTTTAAGCC �656 327–344 HAdV-A12
A18L1R TGGTATGACAGCTCTGT 1016–1000
A18 A18L2L Hexon ε det. (loop 2) GGCCTGATGTATTACAACAG �575 907–926 HAdV-A12
A18L2R TTCGGTGGTGGTTAAAAGG 1519–1501
B FiBL Fiber � det. (knob) TACCCCTATGAAGATGAAAGCA �1,064 43–64 HAdV-B3
FiBR GGAGGCAAAATAACTACTCG 1147–1128
D FiDL Fiber � det. (knob) ATGTCTCACTCAAGGTGGGA �1,000 179–198 HAdV-D8
ADD2 GCTGGTGTAAAAATCAATAAAGA 1219–1198
F FiFL Fiber � det. (knob) TATGGACTTAGGAGACGGT �773 2125–2143 HAdV-F40
FiFR CGTTCATTATTTCGAACGC 2934–2916
a det., determinant.
15266 MADISCH ET AL. J. VIROL.

E4 region were combined (69), resulting in a 1,000-bp amplicon (Table 1). The
sense primer was designed with the help of a multiple alignment of GenBank
sequence data of the 5� end flanking the � determinant (sequences of HAdV-B3,
-B7, -B11, -B14, -B16, -B21, -B34, -B35, -D8, -D9, -D15, -D17, -D19, -D28, -D30,
and -D37). For species B, a new primer set (AdFiBR and AdFiBL) was devel-
oped (Table 1), which amplifies the same region as do the species D primers and
requires the same thermal cycling profile. PCR was performed in a total volume
of 100 �l as described above with a final concentration of 2 �M MgCl2. The
thermodynamic profile was equivalent to the previously described hexon PCR
program except for the annealing temperature of 56°C and the elongation step
for 90 s. New primer pairs were designed for species HAdV-B (FiBL/FiBR) and
species HAdV-F (FiFL/FiFR) (Table 1). The PCRs were performed as described
above. The sensitivity of the PCR protocols was determined by testing serial
dilutions of cell culture-derived HAdV-B3 and -D8 previously quantified by a
generic HAdV real-time PCR protocol (26). Furthermore, the fiber knob region
of HAdV-A18 was sequenced after amplification with primers located in the
fiber (5�-GGCTATGGCTGATGGTG-3�) and 3� inverted terminal repeat re-
gion (5�-GTTTGGGCGGATGAGG-3�).
Gel electrophoresis. PCR products were separated in a 2% agarose gel (1% in
the case of fiber amplicons) for 60 min at 120 V. DNA extraction from the
agarose gels was performed with the QIAGEN gel extraction kit according to the
manufacturer’s recommendations.
Sequencing. Cycle sequencing was performed with rhodamine-labeled dideoxy-
nucleotide chain terminator (DNA sequencing kit; ABI, Warrington, England) and
analyzed on an ABI Prism 310 automatic sequencer (Applied Biosystems). PCR
primers were used for the sequencing reactions.
Phylogenetic analysis. L1, L2, and fiber knob region sequences were compared
by sequential pairwise alignment with the Clustal algorithm implemented in the
BioEdit software package (version 6.0.5) (62) and adjusted manually to conform
to the optimized alignment of deduced amino acid sequences. Phylogenetic
relationships were inferred from the aligned nucleic acid sequences by the neigh-
bor-joining method implemented in the programs DNAdist and Neighbor
(PHYLIP, version 3.6; University of Washington, Seattle) (22) using the Kimura
two-parameter substitution model (31) and a transition/transversion ratio of 10.
Support for specific tree topologies was estimated by bootstrap analysis with
1,000 pseudoreplicate data sets. Branch lengths in consensus trees were calcu-
lated by the maximum-likelihood quartet-puzzling method, using the nucleotide
substitution model of Tamura and Nei (60) as implemented in Tree Puzzle 5.0
(58). The pairwise nucleotide identity matrix was calculated with BioEdit (ver-
sion 6.0.5). The histogram of the pairwise nucleotide comparisons was plotted
with the program PNCH (34). Similarity plots were calculated based on different
matrices (PAM250 and EBLOSUM60) with a window of 10 amino acids with the
software Plotcon included in the EMBOSS package (47). Bootscans were per-
formed with the software SimPlot (version 3.5.1) (33) with a window of 200 bp
(20-bp step) based on a Kimura two-parameter substitution model (31) with a
transition/transversion ratio of 2.0.
For making multiple alignments of L1, L2, and fiber knob regions the follow-
ing GenBank sequences were used: for hexon, HAdV-C1 (X67709), -C2
(AJ293903), -B3 (X76549), -E4 (X84646), -C5 (AF542130), -C6 (X67710), -B7
(Z48571), -D9 (AF161562), -D10 (AB023548), -B11 (AY163756), -A12
(X73487), -B14 (AB018425), -B16 (X74662), -D17 (AF108105), -D19
(AF161565), -B21 (AJ012091), -B34 (AB052911), -D23 (AB023552), -D26
(AB023554), -A31 (X74661), -B35 (AB052912), -D37 (AB023555), -F40
(X51782), -F41 (X51783), -D45 (AB023556), -D46 (AB023557), -D47
(AB023558), -D48 (U20821), and SAdV-E25 (BK000413); for fiber knob
(� determinant), HAdV-C1 (AB108423), -C2 (J01917), -B3 (X01998), -C5
(M18369), -C6 (AB108424), -B7 (AF104384), -D8 (AB162768), -D9
(X74659), -B11 (AY163756), -A12 (X73487), -B14 (AB065116), -D15
(X72934), -B16 (U06106), -D17 (Y14241), -D19 (U69131), -D28 (Y14242),
-B21 (U06107), -A31 (X76548), -B34 (AB073168), -B35 (U32664), D37
(U69132), -E4 (X76547), -F40 (M28822), -F41 (M60327), and SAdV-E25
(BK000413).
Nucleotide sequence accession numbers. The sequences of L1 of the ε deter-
minant of HAdV-B14 (AB018425), -B50 (AJ864518), -D13 (AJ749847), -D15
(AJ821891), -D20 (AJ749848), -D25 (AJ749849), -D27 (AJ749850), -D28
(AJ749851), -D29 (AJ749852), -D30 (AJ749853), -D32 (AJ749854), -D33
(AJ749855), -D36 (AJ749856), -D38 (AJ749857), -D39 (AJ749858), -D42
(AJ821893), -D43 (AJ821894), -D44 (AJ749859), -D49 (AJ821895), and -D51
(AJ821896) have been deposited in GenBank, as well as the L2 sequences
of HAdV-A18 (AJ821897), -B14 (AJ749860), -B50 (AJ749861), -D13
(AJ745880), -D15 (AJ745881), -D20 (AJ745882), -D22 (AJ745883), -D25
(AJ745884), -D27 (AJ745885), -D28 (AJ745886), -D29 (AJ745887), -D30
(AJ745888), -D32 (AJ745889), -D33 (AJ745890), -D36 (AJ745891), -D37
(AJ745892), -D38 (AJ745893), -D39 (AJ745894), -D42 (AJ745895), -D43
(AJ745896), -D44 (AJ745897), -D49 (AJ745898), and -D51 (AJ745899). Fi-
ber knob (� determinant) sequences of HAdV-A18 (AJ841699), -B50
(AJ811465), -D10 (AJ811442), -D13 (AJ811443), -D20 (AJ811444), -D22
(AJ811445), -D23 (AJ811446), -D24 (AJ811447), -D25 (AJ811448), -D26
(AJ811449), -D27 (AJ811450), -D29 (AJ811451), -D30 (AJ831473), -D32
(AJ811452), -D33 (AJ811453), -D36 (AJ811454), -D38 (AJ811455), -D39
(AJ811456), -D42 (AJ811457), -D43 (AJ811458), -D44 (AJ811459), -D45
(AJ811460), -D46 (AJ811461), -D47 (AJ811462), -D48 (AJ811463), -D49
(AJ811464), and -D51 (AJ811460) have also been submitted to GenBank.
RESULTS
Phylogenetic analysis of � determinant L1 and L2. The type-
specific neutralization epitope (ε determinant) consists of L1,
which includes six hypervariable regions (HVR1 to HVR-6),
and of L2 with the seventh HVR (14). Sequences of the hy-
pervariable L1 region were generated with the generic primer
pair H1.1-H2.1, which has an amplicon size of �800 bp (Table
1). All 45 HAdV prototype viruses of species B, C, D, and A18
were amplified successfully with the new L2 primer sets
(Table 1), and all amplicons had the expected length of 322 bp
in the case of species C and D, 590 bp in the case of species B,
and 656 bp in the case of HAdV-A18. L1 and L2 amplicons of
prototype strains were directly sequenced and aligned with
previously available GenBank sequence data for human ade-
novirus prototype strains (for nucleotide sequence accession
numbers see Materials and Methods). Phylogenetic trees for
L1 based on nucleotide acid positions 19228 to 19828 and for
L2 based on positions 19989 to 20272 (both referring to the
HAdV-C2 hexon sequence) were constructed with several phy-
logeny reconstruction algorithms (neighbor joining, maximum
likelihood, and maximum parsimony) in order to determine
the phylogenetic relationships among HAdV on the species
level. The phylogenetic trees constructed by different methods
were congruent in overall structure (Fig. 1 depicts neighbor-
joining trees), as were trees constructed with deduced amino
acid sequences (compare Fig. 1 and 2).
As expected, phylogenetic trees of both L1 and L2 con-
firmed that human adenoviruses clustered together in major
groups corresponding to their six species, with the exception of
the only virus of species E, HAdV-E4 (Fig. 1 and 2). HAdV-E4
clustered with the species HAdV-B and had the closest rela-
tionship to HAdV-B16 with a high bootstrap value both in
nucleic acid and in amino acid phylogenetic trees (100%, Fig. 1
and 2), which supports previously published cross-neutraliza-
tion data (65). Groups of HAdV-A and HAdV-F were clus-
tering together in one monophyletic group, which correlates
well with their common tissue tropism (Fig. 1 and 2) (8). This
group clusters together with species HAdV-C.
Identical � determinants of HAdV-D15 and HAdV-D29.
Genetic divergence of HADV-D15 and HAdV-D29 was as low
as 0.0% (nucleic acid sequence) in L1 (Fig. 1A) and 0.4% in
L2 (Fig. 1B), resulting in a single amino acid substitution
(E422Q), whereas genetic divergence of other HAdV-D sero-
types was in the range of 2.5% to 23%. HAdV-D15 and
HAdV-D29 also exhibit extensive cross-neutralization (1, 10,
29, 66). Moreover, sequence variability between different
strains of HAdV-D15 was in the same range as the difference
between HAdV-D15 and HAdV-D29. For example, three dif-
ferent bases resulting in two amino acid substitutions in L2
were found between the HAdV-D15 strain Morrison, available
VOL. 79, 2005 PHYLOGENY OF IMMUNOGENIC ADENOVIRUS DETERMINANTS 15267

in the GenBank database (X76707), and the HAdV-D15 strain
305 (955; CH. 38; V-215-003-014, ATCC VR-1092) sequenced
additionally in this study (AJ745881). Therefore, L1 sequenc-
ing, L2 sequencing, and classical neutralization typing concor-
dantly indicate that HAdV-D15 and HAdV-D29 are not dis-
tinct serotypes. For comparison, the second highest genetic
identity in L2 (97.5%) was observed between HAdV-D39 and
-D43, which can be clearly discerned by cross-neutralization.
There were eight bases different in the L2 region, resulting in
a single amino acid difference between these two serotypes
(D435N) (Fig. 1B and 2B). Furthermore, sufficient antigenic
divergence between HAdV-D39 and -D43 was plausible be-
cause of L1 sequence divergence (7.4% nucleic acid divergence
resulting in 16 amino acid exchanges) (Fig. 1A and 2A). These
molecular evolutionary data indicated that divergence of
HAdV-43 and HAdV-39 into distinct types was the product of
a relatively recent evolutionary event.
Evolution of neutralization determinants of species HAdV-D.
Species HAdV-D consists of 31 serotypes (counting HAdV-
D15 and HAdV-D29 as a single neutralization serotype), and
three main subclusters were observed in the L1 amino acid
sequence phylogenetic tree (Fig. 2A). Therefore, a subanalysis
of the genetic variability related to the HVRs in L1 of the three
main subclusters of HAdV-D was performed in order to de-
termine mechanisms of molecular evolution leading to the
selection of a multitude of neutralization serotypes. In all three
subclusters HVR1 to HVR6 were found (Fig. 3) with the main
genetic variability found in HVR4 and HVR5. Similarity plots
indicated that HAdV-D prototype viruses of subcluster 1 di-
verged mainly by additional molecular evolution of HVR3
(Fig. 3A). In contrast to subcluster 1, viruses of subcluster 2
diverged mainly by additional molecular evolution of HVR1
(Fig. 3B), and viruses of subcluster 3 diverged in both HVR1
and HVR2 (Fig. 3C).
Molecular classification criteria derived from � determi-
nant phylogeny. The phylogenetic trees of the L1 and L2
regions (Fig. 1 and 2) demonstrated that typing of HAdV is
feasible by sequencing either L1 or L2 amplicons because of
sufficient genetic distances. The amplicon of the L2 PCR is
shorter and easier to sequence than that of the L1 PCR. As the
nucleic acid sequence phylogenetic tree of L2 is identical in
overall structure to the amino acid sequence tree of L2, we
suggest L2 sequencing and comparison of nucleic acid se-
quence data as routine molecular typing methods. After
HAdV-D15 and -D29 were counted as a single neutralization
serotype, nucleic acid sequence identity of serotypes of the
same species was in the range of 61.5% to 97.5% in the L2
coding region, with the lowest intraspecies nucleic acid identity
found between types HAdV-B14 and -B16 and the highest
between HAdV-D39 and -D43 (Fig. 1). For comparison, nu-
FIG. 1. Phylogenetic analysis of the nucleic acid sequences of ε determinant L1 (600 bp referring to HAdV-C1) (A) and L2 (283 bp referring
to HAdV-C1) (B) hexon sequences (neighbor-joining tree, Kimura two-parameter matrix). The bootstrap values were generated with 1,000
pseudoreplicates. For nucleotide sequence accession numbers see Materials and Methods.
15268 MADISCH ET AL. J. VIROL.

cleic acid identity between serotypes of different species was
lower, between 51.9% and 67.0%, with the lowest interspecies
identity found between HAdV-D37 and -F40 and, as an excep-
tion, the highest (93.0%) found between HAdV-B16 and -E4
(Fig. 1). Distribution of nucleic acid sequence pairwise identity
scores of the L2 region between serotypes clearly resulted in
two peaks, representing heterologous serotypes of the same
species and heterologous serotypes of different species (Fig. 4).
A nucleic acid sequence divergence of �2.5% in the L2 region
compared to the next homologous prototype sequence in the
data bank is proposed as a criterion for molecular identifica-
tion of a serotype, because the lowest genetic divergence be-
tween two serotypes in the L2 region was 2.5% (HAdV-D39
and -D43) (Fig. 1B). Type identification of a clinical isolate is
achieved in the case of a �2.5% genetic divergence of the L2
sequence compared to the most closely related prototype se-
quence, if the next closely related sequence of another proto-
type of the same species (in the case of HAdV-E4, HAdV-
B16) has a sequence divergence of at least �2.5%. If these
criteria are not fulfilled, a deduced amino acid sequence of the
clinical isolate has to be aligned with prototype data bank
sequences, for example, by using the FASTA Internet server. If
the L2 amino acid divergence compared to the next data bank
sequence is �1.2%, typing is accomplished (Fig. 2B). However,
if the L2 amino acid divergence compared to the next data
bank sequence is �1.2%, typing is not feasible and isolation of
a new HAdV type may be suspected.
Phylogenetic analysis of the fiber knob � determinant. Am-
plification of the fiber knob � determinant of all HAdV species B,
D, and HAdV-A18 was achieved with newly developed primer
sets (Table 1). New sequence data were generated by direct se-
quencing and aligned with sequence data previously published in
GenBank (nucleic acid positions 31959 to 32766 of HAdV-C2).
Phylogenetic trees of all 51 HAdV built by several methods
(neighbor joining, maximum likelihood, and maximum parsi-
mony) were identical in overall structure, as were nucleic acid
sequence trees (data not shown) and deduced amino acid se-
quence trees (Fig. 5A depicts a neighbor-joining tree). Trees
confirmed several phylogenetic relationships suggested by hem-
agglutination properties of the HAdV species. Adenovirus fiber
knob sequences clustered together in major groups corresponding
to the six HAdV species, with the exception of species HAdV-D,
which formed three clusters. However, division of species HAdV-D
into three clusters was not confirmed by bootstrapping. Sur-
FIG. 2. Phylogenetic analysis of the deduced amino acid sequences of L1 (A) and L2 (B) hexon sequences (neighbor-joining tree, Kimura
two-parameter matrix). The bootstrap values were generated with 1,000 pseudoreplicates. Three HAdV-D subclusters which were used for
calculation of similarity plots (Fig. 3) are depicted. For nucleotide sequence accession numbers see Materials and Methods.
VOL. 79, 2005 PHYLOGENY OF IMMUNOGENIC ADENOVIRUS DETERMINANTS 15269

prisingly, pairwise nucleotide comparison of fiber knob se-
quences resulted in three peaks (Fig. 5B), with peak 1 repre-
senting intraspecies relationships, whereas interspecies
relationships were divided into two groups (peaks 2 and 3 in
Fig. 5B), suggesting a two-step model of HAdV species evolu-
tion. Peak 2 is formed by the interspecies relations between
species A and F and between species D and C, including
HAdV-E4, whereas peak 3 consisted of all other interspecies
relationships. Several phylogenetic relationships of species
HAdV-D corresponded well to HI cross-reactivity (HAdV-D8
and -D9; HAdV-D10, -D19, and -D37; HAdV-D13, -D38, and
-D39; HAdV-D15, -D22, and -D42; HAdV-D20 and -D47;
HAdV-D24, -D32, -D33, and -D46) (Fig. 5) (59). As already
suggested by cross-neutralization experiments and HI cross-
reactions (59), phylogenetic relations of HAdV-D fiber knob
sequences did not completely coincide with hexon L1 and L2
sequences (Fig. 2 and 5A). Overall, intraspecies genetic diver-
gence in the fiber knob region was lower than for hexon L1 and
L2, with several prototype sequences almost identical (as low
as 0.0% difference; Fig. 5A and B).
Phylogenetic criteria for species classification. A � determi-
nant species criterion should correlate excellently with the
classical species criteria, because the fiber knob region (� de-
terminant) is responsible for the hemagglutination properties
of the HAdV. The phylogenetic tree of the fiber knob region
demonstrated that a criterion for classification of HAdV spe-
cies can be easily deduced with as a single exception species
HAdV-E clustering with HAdV-C (Fig. 5). Based on an iden-
tity matrix of the complete data set of the � determinant
sequences, we calculated the maximum intraspecies divergence
of each species (47%, nucleic acid divergence in species B),
and the minimum interspecies divergence between all species
was computed (52%, A to F). Therefore, a nucleic acid se-
quence divergence of�50% from the next data bank prototype
sequence can be suggested as a criterion for species definition
(Fig. 5B). Based on this species criterion, HAdV-E4 is a mem-
ber of the HAdV-C species with an average identity of 56% to
the other members of the HAdV-C (54.1% to 56.9%). Species
identification criteria can also be derived from hexon sequence
data sets. For example, many diagnostic laboratories use ge-
neric PCR protocols which amplify a conserved region at the 5�
FIG. 3. Similarity plots of three clusters from species HAdV-D
with a window of 10 amino acids. (A) Cluster 1; (B) cluster 2; (C) clus-
ter 3. Members of the three clusters are shown in Fig. 2A. Hypervari-
able regions (HVR1 to HVR6) are marked in the similarity plots.
FIG. 4. Distribution of pairwise nucleotide identity scores of L2.
Frequency (y axis) represents the number of all nucleotide compari-
sons between different prototypes with an identical genetic diversity.
Peak 1 corresponds to comparisons of heterologous strains (different
serotype) of the same major phylogenetic cluster (same species). In
contrast to this, peak 2 corresponds to comparison of heterologous
strains (different serotype) of different phylogenetic clusters (different
species). The asterisk represents the 99% identity between HAdV-D15
and HAdV-D29.
15270 MADISCH ET AL. J. VIROL.

end of the hexon gene closely adjacent to the L1 coding region
(4, 26). Therefore, we evaluated whether amplicon sequences
of these PCRs are appropriate for molecular classification.
Figure 6 shows that genetic divergence permits reliable species
identification. A generic PCR protocol proposed by Allard et al.
(4) allows species identification with a nucleic acid sequence di-
vergence of �9% from the next data bank prototype sequence
(Fig. 6A). In the case of a generic real-time PCR protocol (26)
with its smaller amplicon size, sequence divergence of �4.8% is
required (Fig. 6B). Both protocols even allow a final typing deci-
sion, but it is reliable for only six members of the species A, E, and
F (criterion, �1.5% nucleic acid sequence divergence from the
next homologous prototype sequence) (Fig. 6).
Molecular identification of intermediate strains. Amplifica-
tion of the immunogenic � determinant in combination with
the L2 PCR could be a useful alternative for molecular iden-
tification of HAdV intermediate strains because the main neu-
tralization epitope and the hemagglutination epitope can be
sequenced. Four clinical isolates previously typed as interme-
diate strains with the classical methods (NT and HI) were
analyzed with the newly developed combination of the L2
sequencing and � determinant sequencing. In accordance with
previous results of classical typing methods, isolates were iden-
tified as intermediate strains HAdV-15H9 (two isolates),
HAdV-30H44, and HAdV-37H13. The L2 nucleic acid se-
quences of four analyzed intermediate strains were 100% iden-
tical to the prototype strains. Their � determinant sequences
were 98% to 100% identical to the prototype strains, and the
deduced amino acid sequences were 97% to 100% identical. In
the case of HAdV-30H44, the fiber � determinant of HAdV-
D44 was 100% identical to HAdV-D48. This finding correlates
well with cross-reactivity in HI between these two viruses (52).
HAdV-D30H44 was typed previously with HAdV-D44-specific
HI antiserum when HAdV-D48 was not yet isolated, and an
HAdV-D48-specific antiserum was not available to us. There-
fore, a final differentiation between HAdV-D44 and -D48 was
not feasible with both molecular and classical methods.
Molecular evolution of the main antigenic determinants of
species HAdV-B and HAdV-E. Species B has been previously
divided into two subspecies (“subgenera”) because of obvious
striking differences in restriction fragment patterns and, in part,
tissue tropism. Detailed analysis of the molecular phylogeny of
FIG. 5. (A) Phylogenetic analysis of the deduced amino acid sequence from � determinant (269-amino-acid fiber knob sequence referring to
HAdV-C2), with a neighbor-joining method tree based on a Kimura two-parameter matrix. The bootstrap values were generated with 1,000
pseudoreplicates. For nucleotide sequence accession numbers see Materials and Methods. (B) Distribution of pairwise nucleotide identity scores
of fiber knob (� determinant). Frequency (y axis) represents the number of all nucleotide comparisons between different prototypes with an
identical genetic diversity. Peak 1 corresponds to comparisons of heterologous strains (different serotype) of the same major phylogenetic cluster
(same species). In contrast to this peak, peaks 2 and 3 correspond to comparison of heterologous strains (different serotype) of different
phylogenetic clusters (different species). Peak 2 is formed only by interspecies identities between HAdV-C (including HAdV-E) and -D, as well
as between HAdV-A and -F.
VOL. 79, 2005 PHYLOGENY OF IMMUNOGENIC ADENOVIRUS DETERMINANTS 15271

the main immunogenic determinants showed a more complex
pattern. In hexon L1 and L2 trees subspecies B1 viruses
HAdV-B3 and -B7 were as closely related as anticipated by
the subspecies concept, whereas other subspecies B1 viruses
(HAdV-B21 and -B50) clustered with subspecies B2 viruses,
HAdV-B21 with HAdV-B11 and -35 and HAdV-B50 with
HAdV-B14 and -34, supported by high bootstrap values
(Fig. 2). Subanalysis of genetic variability related to HVR1 to
HVR6 in L1 indicated that HAdV-B3 and -B7 diverged mainly
by selection of immunogenic epitopes in HVR5 (Fig. 7A). By
contrast, HAdV-B14, -B34, and -B50 diverged mainly by mo-
lecular evolution in HVR1 and HVR4 (Fig. 7B), and HAdV-
B11, -B21, and -B35 diverged mainly in HVR1 and HVR2
(Fig. 7C). Phylogenetic relations of HAdV-B fiber knob se-
quences did not coincide with hexon L1 and L2 sequences,
indicating intrasubspecies and intersubspecies recombination
events in the molecular phylogeny of species B HAdV (com-
pare Fig. 2 and 5A). The close phylogenetic relationship in the
species B assumed by HI cross-reactions between HAdV-B7,
-B11, and -B14 and between HAdV-B34 and -B35 was con-
firmed (Fig. 5A) (59).
In the ε determinant (L1 and L2) HAdV-B16 formed a sub-
cluster with HAdV-E4 and with the more distantly related SAdV-
E25 but was not closely related to the other species HAdV-B
prototype viruses. However, in the fiber knob region HAdV-B16
clustered with the other species B viruses whereas both HAdV-E4
and SAdV-E25 clustered with species HAdV-C.
DISCUSSION
Previously, HAdV subgenera, now species, were defined by
their hemagglutination properties; later oncongenicity, fiber
length, and genotyping (restriction fragment analysis) data
were also included as subgenus criteria. Phylogenetic analysis
of fiber knob (� determinant) sequences of all 51 HAdV con-
firmed the species concept (Fig. 5A). From our sequence data,
DNA homology criteria for species identification were de-
duced (�50%, Fig. 5). However, phylogenetic relationships
between species C and D and between species A and F were
closer (peak 2 in Fig. 5B) than between other species, resulting
in a unique three-peak pattern of the pairwise nucleotide com-
parison plot (Fig. 5B).
This finding indicated a more recent divergence in the mo-
lecular evolution of the enterotropic species HAdV-A and
HAdV-F, as well as in the evolution of HAdV-C and -D, and
suggested two main steps of molecular evolution of HAdV
species. The species concept of HAdV is also supported by the
molecular phylogeny of the ε determinant L1 and L2 (Fig. 1
and 2), but the two interspecies relationship peaks were not
observed in the pairwise nucleotide comparison plot of L2
(Fig. 4). However, the recent divergence of species HAdV-A
and -F is also supported by hexon L1 and L2 phylogenetic trees
(Fig. 1 and 2), whereas in the case of species HAdV-C and -D
a close relationship could not be confirmed by hexon sequence
data. Probably, the higher selection pressure resulted in a more
FIG. 6. Phylogenetic analysis of generic hexon PCR sequences (nucleic acid sequences): amplicons of conventional PCR (4) (A) and real-time
PCR (26) (B) with neighbor-joining method based on a Kimura two-parameter matrix. Both phylogenetic trees show that sequence divergence is
sufficient for species identification as well as type identification of HAdV-A, HAdV-E, and HAdV-F viruses. For nucleotide sequence accession
numbers see Materials and Methods.
15272 MADISCH ET AL. J. VIROL.

rapid evolution of the neutralization determinant ε than of
the � determinant. A subanalysis of the genetic variability of
HVR1 to HVR6 indicated complex patterns of molecular evo-
lution of the ε determinant (Fig. 3), leading to the emergence
of multiple neutralization types of species HAdV-D. This may
have covered up older events of species divergence in the ε
determinant data set. Another hypothesis for the evolution of
species HAdV-C and -D viruses would be a recombination
event leading to divergence of these species by introducing a
more diverse hexon sequence and subsequent evolution in
both ε and � determinants, resulting in the multiple types of
these species.
Historically, HAdV-E4 was classified as a species of its own
mainly because of its difference in GC content from that of
species HAdV-B (10), although HAdV-E4 is closely related to
species HAdV-B as demonstrated by phylogenetic studies of
the hexon and several other genome regions except the fiber
gene (8, 25, 44). Recently, the complete genome of HAdV-E4
was sequenced and the close relationship with simian adeno-
viruses was demonstrated (most closely to SAdV-E25) (46). A
zoonotic transmission event was suggested as the possible or-
igin of HAdV-E4 (46). Therefore, we included SAdV-E25 in
our phylogenetic analysis of HAdV. Surprisingly, in the neu-
tralization determinant ε, HAdV-E4 was more closely related
to HAdV-B16 (supported by a bootstrap value of 100%) and
far more distantly related to SAdV-E25, which also clustered
with species HAdV-B (Fig. 1 and 2). By contrast, phylogenetic
analysis of the fiber knob region confirmed the close relation-
ship of HAdV-E4 with SAdV-E25 (supported by a bootstrap
value of 100%) and both viruses clustered with (but were more
distantly related to) species HAdV-C (bootstrap value 100%)
(Fig. 5A). Thus, our phylogenetic analysis supported both an
interspecies recombination mechanism in the phylogeny of
both HAdV-E4 and SAdV-E25, combining a species HAdV-
C-like fiber gene with a species HAdV-B genetic backbone,
and a zoonotic transmission event as suggested by Purkayastha
et al. (46). However, in the ε determinant (L1 and L2),
HAdV-E4 is very closely related to HAdV-B16 as indicated by
phylogenetic trees (Fig. 1 and 2) and also by cross-neutraliza-
tion experiments (27). Therefore, we tried to analyze the phy-
logenetic relationships of HAdV-B16, HAdV-E4 (AY594253),
SAdV-E25 (BK000413), and HAdV-B7 (AY495969). Unfortu-
nately, a complete genomic sequence of HAdV-B16 was not
available in GenBank. Phylogenetic analysis of available HAdV-
B16 sequences of E1A (AY490821), 13S protein (AY490821),
putative pVIII protein and E3 (AB073632), putative 11-kDa pro-
tein (AY509991), VA RNA I and II (U10674), and RNA preter-
minal protein (U52564) sequences in addition to hexon and fiber
knob sequences demonstrated that the close relationship of
HAdV-B16 and HAdV-E4 is restricted to the hexon region (data
not shown). Therefore, a bootscan analysis of the complete
HAdV-B16 hexon sequence, compared to HAdV-E4, HAdV-
B11, HAdV-B7, HAdV-B3, and SAdV-E25, was performed.
Bootscan results indicated an intrahexon recombination event in
the origin of HAdV-B16. For nucleotide positions 1 to 1400 of the
hexon gene (including the ε determinant), HAdV-B16 had the
highest similarity to HAdV-E4, whereas for positions 1400 to
2823 HAdV-B16 was more closely related to HAdV-B11.
In general, intraspecies recombination is observed much more
frequently than interspecies recombination (55, 67). Therefore,
FIG. 7. Similarity plots of three clusters from species HAdV-B
with a window of 10 amino acids. (A) Cluster formed by HAdV-B3
and -B7; (B) cluster formed by HAdV-B14, -B34, and -B35;
(C) cluster formed by HAdV-B11, -B21, and -B35. Hypervariable
regions (HVR1 to HVR6) are marked in the similarity plots.
VOL. 79, 2005 PHYLOGENY OF IMMUNOGENIC ADENOVIRUS DETERMINANTS 15273

most intermediate strains belong either to species B or to species
D, both of which include more serotypes (41 of 51) than all other
species do. Several cases of strong HI cross-reactions have been
described, for example, between HAdV-D10, -D19, and -D37 and
between HAdV-D24, -D32, -D33, and -D46 (59), which may
suggest that some of these prototype viruses are recombination
variants sharing an almost identical fiber knob region. For exam-
ple, fiber knob sequences of HAdV-D44 and -D48 were 100%
identical, as well as HAdV-D19 and -D37 in the � determinant,
which are closely related to HAdV-D10 (Fig. 5). Therefore, a
previous recombination event can indeed be hypothesized in the
origin of some HAdV prototypes, e.g., HAdV-D44 and -D48 or
HAdV-D19 and -D37. This result is in complete accordance with
the organotropism and virulence of HAdV-D19 and -D37, which
both cause epidemic keratoconjunctivitis. On the other hand,
HAdV-D8 neither clustered with HAdV-D19 and -D37 nor
cross-reacted in HI, although it shares similar organotropism and
virulence. Recently, sialic acid and CD46 were identified as cel-
lular receptors interacting with the fiber knob of HAdV-D37 (7,
68), whereas the receptor of HAdV-D8 was not yet definitely
identified. The phylogenetic relationship of the HAdV-D8 fiber
knob to HAdV-D9 suggests CAR as the putative receptor (Fig. 6)
(48). This indicates that factors other than cellular receptors are
significant for epidemic keratoconjunctivitis.
From a type identification perspective, fiber knob (� deter-
minant) sequencing is clearly secondary to hexon (ε determi-
nant) sequencing (compare Fig. 2 and 5A), as is hemaggluti-
nation inhibition testing compared to neutralization testing
(27, 59, 65). In a diagnostic setting highly sensitive detection of
HAdV DNA is currently achieved in many laboratories with
extensively validated, generic PCR amplifying the conserved 5�
end of the hexon gene (4, 26). Sequencing of these amplicons
can be proposed as the first step of a two-step molecular typing
scheme. Phylogenetic analysis of these amplicon sequences
indicated that species identification is unequivocally feasible
(Fig. 6). For unequivocal type identification sequencing of the
hypervariable neutralization determinant ε is required as a
second step.
This is the first work including sequences and phylogenetic
analysis of both loops of the neutralization determinant ε of all
51 HAdV. Phylogenetic analysis demonstrated that either L1
or L2 sequencing is sufficient for precise typing (Fig. 1). A
�2.5% nucleic acid sequence divergence was deduced as an
unequivocal criterion for molecular typing from L2 sequence
data. Sequencing of the bigger L1 (compare Fig. 1A and 1B) or
use of deduced amino acid data for L2 did not improve type
identification (compare Fig. 1B and 2B). Therefore, our results
clearly support L2 sequencing as the most simple but sufficient
approach for molecular typing of all 51 HAdV. Recently, a
sequence-based approach for typing HAdV of species D and E
was proposed (56). However, the relevant neutralization de-
terminant ε was not covered by this approach, nor was an
application to HAdV species other than D and E achieved.
Another group suggested L2 sequencing for typing all HAdV
without defining sequence divergence criteria for type identifica-
tion or defining the length of the sequence region (51). Further-
more, molecular identification of new prototype HAdV was not
possible because of the lack of type identification criteria.
If typing of a wild-type isolate is not feasible with the newly
defined L2 molecular typing criterion, isolation of a new HAdV
prototype can be suspected. In this case it is important to gain
additional L1 sequence data because of the bigger contribution
of L1 to the ε determinant. A �2.4% L1 nucleotide divergence
from the next prototype strain (Fig. 1A) in combination with
�4.2% amino acid divergence (Fig. 2A) strongly supports
identification of a new HAdV neutralization prototype. As a
next step, we suggest that the fiber knob region (�) should be
sequenced, in order to predict the immunogenic properties
(neutralization and hemagglutination inhibition) of the pro-
posed new prototype sufficiently. However, it should be kept in
mind that major cross-reactions in hemagglutination inhibition
tests and even identical sequences of the � determinant have
not been contradictory to identification of a new prototype
(Fig. 5A) (10, 59). Other techniques should be used for con-
firming identification of a new HAdV prototype, e.g., genotyp-
ing by restriction fragment analysis (2) and classical cross-
neutralization techniques, as taxonomic criteria for type
identification have not been modified (10).
Virus typing and subsequent definition of new types on the
basis of sequence data of immunogenic epitopes have been
applied successfully to human enteroviruses and resulted in the
identification of several new types (39–41). Probably, a molec-
ular typing concept as suggested by us for HAdV will also
facilitate the identification of new HAdV prototype viruses.
Regarding the suggestions for molecular discovery of new
HAdV types, we checked whether the most recently described
HAdV types, 50 and 51, would have been indicated as new
types (18). The sufficient nucleic acid sequence divergence of
the new serotypes HAdV-B50 (16% divergence from the next
heterologous serotype HAdV-B34) and -D51 (20% divergence
from HAdV-D15 and HAdV-D29) in the L2 region (Fig. 1B)
would have clearly indicated the isolation of two new types.
This is confirmed by L1 sequences (Fig. 1A) and � determinant
sequences (Fig. 5A). Thus, our sequencing results and criteria
also confirm that these isolates are new HAdV types.
In contrast to HAdV-B50 and -D51, molecular typing is not
feasible with the two serotypes HAdV-D15 and -D29. This
does not seem to be a drawback of the proposed molecular
typing concept because the neutralization method also cannot
reliably differentiate these two HAdV (1, 29, 66). Therefore, it
was already suggested that the new definition of the serotype
HAdV-D29 as a distinct serotype was premature in 1962 (49,
66). Using the proposed molecular typing criteria, HAdV-D29
(AJ811451) is not a separate serotype but a hemagglutination
variant (intermediate strain) of HAdV-D15. By contrast, an-
other study comparing the L2 sequences of HAdV-D15 and
HAdV-D29 indicated sufficient sequence divergence for mo-
lecular typing (51), contradictory to our results and cross-
neutralization results (1, 29, 66). This may be explained by the
inclusion of hexon sequences adjacent to the immunogenic L2,
which do not contribute to the neutralization epitope. More-
over, our L2 typing result is clearly supported by the phyloge-
netic analysis of L1, which codes for a bigger part of the
neutralization epitope (0.0% sequence divergence). As HAdV-
D29 differs from HAdV-D15 by its hemagglutination properties
as usually intermediate strains do, we tried to identify the phylo-
genetic origin of the hemagglutination epitope by sequencing of
the � determinant. Phylogenetic analysis indicated a close rela-
tionship of HAdV-D29 (AJ811451) to a previously published data
bank sequence of HAdV-D30, strain BP-7 (AF447393). There
15274 MADISCH ET AL. J. VIROL.

were only two bases different between the sequences of HAdV-
D29 and -D30, resulting in a single amino acid substitution
(S104F). This implied that HAdV-D29 and -D30 should have
HI cross-reactions. However, HI cross-reactions have never
been reported between these two viruses (49, 59, 65). Sequenc-
ing of two different strains of HAdV-D30, the BP-7 strain (ATCC
VR-273) (AJ831473) and a prototype strain of the German ade-
novirus reference center, demonstrated that these two had 100%
identical sequences which were quite divergent from the HAdV-
D30 sequence (AF447393) in the data bank (39% nucleic acid
sequence divergence) and from the HAdV-D29 sequence in the
data bank (AJ811451) (61.6%). Phylogenetic analysis demon-
strated a different clustering of the newly generated HAdV-D30
sequences, which had the closest relationship to HAdV-D49
(Fig. 5A). This relationship is supported by one-way HI cross-
reactions between HAdV-D30 and -D49 (52). In summary, our
results imply that HAdV-D29 is an intermediate strain originating
from a recombination between HAdV-D15 (ε determinant) and a
so-far-not-isolated, unknown HAdV prototype (� determinant)
phylogenetically related to HAdV-D25. The latter result may
justify the acknowledgment of HAdV-D29 as a distinct serotype.
In conclusion, molecular typing by sequencing of immuno-
genic HAdV determinants is an adequate substitute for clas-
sical neutralization and hemagglutination inhibition typing.
This approach will also help to identify new types more easily.
The complete data set of the main immunogenic determinants
ε and � may facilitate development of gene therapy vectors and
give new insights into the complex molecular evolution of
HAdV including multiple recombination events.
REFERENCES
1. Adrian, T., B. Bastian, W. Benoist, J. C. Hierholzer, and R. Wigand. 1985.
Characterization of adenovirus 15/H9 intermediate strains. Intervirology 23:
15–22.
2. Adrian, T., G. Wadell, J. C. Hierholzer, and R. Wiegand. 1986. DNA restric-
tion analysis of adenovirus prototype 1 to 41. Arch. Virol. 91:277–290.
3. Akiyama, H., T. Kurosu, C. Sakashita, T. Inoue, S. Mori, K. Ohashi, S.
Tanikawa, H. Sakamaki, Y. Onozawa, Q. Chen, H. Zheng, and T. Kitamura.
2001. Adenovirus is a key pathogen in hemorrhagic cystitis associated with
bone marrow transplantation. Clin. Infect. Dis. 32:1325–1330.
4. Allard, A., B. Albinsson, and G. Wadell. 2001. Rapid typing of human
adenoviruses by a general PCR combined with restriction endonuclease
analysis. J. Clin. Microbiol. 39:498–505.
5. Allard, A., R. Girones, P. Juto, and G. Wadell. 1990. Polymerase chain
reaction for detection of adenoviruses in stool samples. J. Clin. Microbiol.
28:2659–2667.
6. Aoki, K., and Y. Tagawa. 2002. A twenty-one year surveillance of adenoviral
conjunctivitis in Sapporo, Japan. Int. Ophthalmol. Clin. 42:49–54.
7. Arnberg, N., P. Pring-Akerblom, and G. Wadell. 2002. Adenovirus type 37
uses sialic acid as a cellular receptor on Chang C cells. J. Virol. 76:8834–
8841.
8. Bailey, A., and V. Mautner. 1994. Phylogenetic relationships among adeno-
virus serotypes. Virology 205:438–452.
9. Belousova, N., V. Krendelchtchikova, D. T. Curiel, and V. Krasnykh. 2002.
Modulation of adenovirus vector tropism via incorporation of polypeptide
ligands into the fiber protein. J. Virol. 76:8621–8631.
10. Benko¨, M., B. Harrach, and W. C. Russell. 2000. Adenoviridae, p. 227–238.
In M. H. V. Van Regenmortel, C. M. Fauquet, D. H. L. Bishop, E. B.
Carsten, M. K. Estes, S. M. Lemon, J. Maniloff, M. A. Mayo, D. J. McGeoch,
C. R. Pringle, and R. B. Wickner (ed.), Virus taxonomy. Seventh report of
the International Committee on Taxonomy of Viruses. Academic Press, New
York, N.Y.
11. Bhat, A. M., R. G. Meny, E. A. Aranas, and F. Yehia. 1984. Fatal adenoviral
(type 7) respiratory disease in neonates. Clin. Pediatr. (Philadelphia) 23:409–
411.
12. Carrigan, D. R. 1997. Adenovirus infections in immunocompromised pa-
tients. Am. J. Med. 102:71–74.
13. Chuang, Y. Y., C. H. Chiu, K. S. Wong, J. G. Huang, Y. C. Huang, L. Y.
Chang, and T. Y. Lin. 2003. Severe adenovirus infection in children. J.
Microbiol. Immunol. Infect. 36:37–40.
14. Crawford-Miksza, L., and D. P. Schnurr. 1996. Analysis of 15 adenovirus
hexon proteins reveals the location and structure of seven hypervariable
regions containing serotype-specific residues. J. Virol. 70:1836–1844.
15. Crawford-Miksza, L. K., R. N. Nang, and D. P. Schnurr. 1999. Strain vari-
ation in adenovirus serotypes 4 and 7a causing acute respiratory disease.
J. Clin. Microbiol. 37:1107–1112.
16. Dakhama, A., R. G. Hegele, G. Laflamme, E. Israel-Assayag, and Y.
Cormier. 1999. Common respiratory viruses in lower airways of patients with
acute hypersensitivity pneumonitis. Am. J. Respir. Crit. Care Med. 159:
1316–1322.
17. Davison, A. J., M. Benko, and B. Harrach. 2003. Genetic content and
evolution of adenoviruses. J. Gen. Virol. 84:2895–2908.
18. De Jong, J. C., A. G. Wermenbol, M. W. Verweij-Uijterwaal, K. W. Slaterus,
P. Wertheim-Van Dillen, G. J. Van Doornum, S. H. Khoo, and J. C. Hierholzer.
1999. Adenoviruses from human immunodeficiency virus-infected individu-
als, including two strains that represent new candidate serotypes Ad50 and
Ad51 of species B1 and D, respectively. J. Clin. Microbiol. 37:3940–3945.
19. Echavarria, M., M. Forman, J. Ticehurst, J. S. Dumler, and P. Charache.
1998. PCR method for detection of adenovirus in urine of healthy and
human immunodeficiency virus-infected individuals. J. Clin. Microbiol. 36:
3323–3326.
20. Echavarria, M. S., S. C. Ray, R. Ambinder, J. S. Dumler, and P. Charache.
1999. PCR detection of adenovirus in a bone marrow transplant recipient:
hemorrhagic cystitis as a presenting manifestation of disseminated disease.
J. Clin. Microbiol. 37:686–689.
21. Erdman, D. D., W. Xu, S. I. Gerber, G. C. Gray, D. Schnurr, A. E. Kajon, and
L. J. Anderson. 2002. Molecular epidemiology of adenovirus type 7 in the
United States, 1966–2000. Emerg. Infect. Dis. 8:269–277.
22. Felsenstein, J. 2004. PHYLIP (Phylogeny Inference Package), version 3.6.
Department of Genome Sciences, University of Washington, Seattle.
23. Friedrichs, N., A. M. Eis-Hubinger, A. Heim, E. Platen, H. Zhou, and R.
Buettner. 2003. Acute adenoviral infection of a graft by serotype 35 following
renal transplantation. Pathol. Res. Pract. 199:565–570.
24. Gall, J. G., R. G. Crystal, and E. Falck-Pedersen. 1998. Construction and
characterization of hexon-chimeric adenoviruses: specification of adenovirus
serotype. J. Virol. 72:10260–10264.
25. Gruber, W. C., D. J. Russell, and C. Tibbetts. 1993. Fiber gene and genomic
origin of human adenovirus type 4. Virology 196:603–611.
26. Heim, A., C. Ebnet, G. Harste, and P. Pring-Akerblom. 2003. Rapid and
quantitative detection of human adenovirus DNA by real-time PCR. J. Med.
Virol. 70:228–239. (Erratum, 71:320.)
27. Hierholzer, J. C., Y. O. Stone, and J. R. Broderson. 1991. Antigenic rela-
tionships among the 47 human adenoviruses determined in reference horse
antisera. Arch. Virol. 121:179–197.
28. Houng, H. S., S. Liang, C. M. Chen, J. Keith, M. Echavarria, J. L. Sanchez,
S. A. Kolavic, D. W. Vaughn, and L. N. Binn. 2002. Rapid type-specific
diagnosis of adenovirus type 4 infection using a hexon-based quantitative
fluorogenic PCR. Diagn. Microbiol. Infect. Dis. 42:227–236.
29. Kasova, V., M. Bruckova, F. Kotelensky, K. Kotelenska, and R. Ulrichova.
1977. Isolation of adenovirus type 29 from an outbreak of epidemic kerato-
conjunctivitis. Acta Virol. 21:173.
30. Kidd, A. H., M. Jonsson, D. Garwicz, A. E. Kajon, A. G. Wermenbol, M. W.
Verweij, and J. C. De Jong. 1996. Rapid subgenus identification of human
adenovirus isolates by a general PCR. J. Clin. Microbiol. 34:622–627.
31. Kimura, M. 1980. A simple method for estimating evolutionary rates of base
substitutions through comparative studies of nucleotide sequences. J. Mol.
Evol. 16:111–120.
32. Lion, T., R. Baumgartinger, F. Watzinger, S. Matthes-Martin, M. Suda, S.
Preuner, B. Futterknecht, A. Lawitschka, C. Peters, U. Potschger, and H.
Gadner. 2003. Molecular monitoring of adenovirus in peripheral blood after
allogeneic bone marrow transplantation permits early diagnosis of dissemi-
nated disease. Blood 102:1114–1120.
33. Lole, K. S., R. C. Bollinger, R. S. Paranjape, D. Gadkari, S. S. Kulkarni,
N. G. Novak, R. Ingersoll, H. W. Sheppard, and S. C. Ray. 1999. Full-length
human immunodeficiency virus type 1 genomes from subtype C-infected
seroconverters in India, with evidence of intersubtype recombination. J. Vi-
rol. 73:152–160.
34. Madisch, I. 2004. PNCH. Histogram maker of matrix files, version 1.0.
Medizinische Hochschule Hannover, Hannover, Germany.
35. Magnusson, M. K., S. S. Hong, P. Boulanger, and L. Lindholm. 2001.
Genetic retargeting of adenovirus: novel strategy employing “deknobbing”
of the fiber. J. Virol. 75:7280–7289.
36. Matsuse, T., H. Matsui, C. Y. Shu, T. Nagase, T. Wakabayashi, S. Mori, S.
Inoue, Y. Fukuchi, and H. Orimo. 1994. Adenovirus pulmonary infections
identified by PCR and in situ hybridisation in bone marrow transplant re-
cipients. J. Clin. Pathol. 47:973–977.
37. Mellman-Rubin, T. L., R. P. Kowalski, M. Uhrin, and Y. J. Gordon. 1995.
Incidence of adenoviral and chlamydial coinfection in acute follicular con-
junctivitis. Am. J. Ophthalmol. 119:652–654.
38. Mori, K., T. Yoshihara, Y. Nishimura, M. Uchida, K. Katsura, Y. Kawase, I.
Hatano, H. Ishida, T. Chiyonobu, Y. Kasubuchi, A. Morimoto, T. Teramura,
and S. Imashuku. 2003. Acute renal failure due to adenovirus-associated
VOL. 79, 2005 PHYLOGENY OF IMMUNOGENIC ADENOVIRUS DETERMINANTS 15275

obstructive uropathy and necrotizing tubulointerstitial nephritis in a bone
marrow transplant recipient. Bone Marrow Transplant. 31:1173–1176.
39. Norder, H., L. Bjerregaard, L. Magnius, B. Lina, M. Aymard, and J. J.
Chomel. 2003. Sequencing of ‘untypable’ enteroviruses reveals two new
types, EV-77 and EV-78, within human enterovirus type B and substitutions
in the BC loop of the VP1 protein for known types. J. Gen. Virol. 84:827–
836.
40. Oberste, M., D. Schnurr, K. Maher, S. al-Busaidy, and M. Pallansch. 2001.
Molecular identification of new picornaviruses and characterization of a
proposed enterovirus 73 serotype. J. Gen. Virol. 82:409–416.
41. Oberste, M. S., K. Maher, D. R. Kilpatrick, M. R. Flemister, B. A. Brown,
and M. A. Pallansch. 1999. Typing of human enteroviruses by partial se-
quencing of VP1. J. Clin. Microbiol. 37:1288–1293.
42. Pehler-Harrington, K., M. Khanna, C. R. Waters, and K. J. Henrickson.
2004. Rapid detection and identification of human adenovirus species by
adenoplex, a multiplex PCR-enzyme hybridization assay. J. Clin. Microbiol.
42:4072–4076.
43. Pring-Akerblom, P., and T. Adrian. 1994. Type- and group-specific polymer-
ase chain reaction for adenovirus detection. Res. Virol. 145:25–35.
44. Pring-Akerblom, P., F. E. Trijssenaar, and T. Adrian. 1995. Sequence char-
acterization and comparison of human adenovirus subgenus B and E hexons.
Virology 212:232–236.
45. Pring-Akerblom, P., F. E. Trijssenaar, T. Adrian, and H. Hoyer. 1999. Mul-
tiplex polymerase chain reaction for subgenus-specific detection of human
adenoviruses in clinical samples. J. Med. Virol. 58:87–92.
46. Purkayastha, A., S. E. Ditty, J. Su, J. McGraw, T. L. Hadfield, C. Tibbetts,
and D. Seto. 2005. Genomic and bioinformatics analysis of HAdV-4, a
human adenovirus causing acute respiratory disease: implications for gene
therapy and vaccine vector development. J. Virol. 79:2559–2572.
47. Rice, P., I. Longden, and A. Bleasby. 2000. EMBOSS: the European Molec-
ular Biology Open Software Suite. Trends Genet. 16:276–277.
48. Roelvink, P. W., A. Lizonova, J. G. Lee, Y. Li, J. M. Bergelson, R. W. Finberg,
D. E. Brough, I. Kovesdi, and T. J. Wickham. 1998. The coxsackievirus-
adenovirus receptor protein can function as a cellular attachment protein for
adenovirus serotypes from subgroups A, C, D, E, and F. J. Virol. 72:7909–
7915.
49. Rosen, L., J. F. Hovis, and J. A. Bell. 1962. Further observation on typing
adenoviruses and a description of two possible additional serotypes. Proc.
Soc. Exp. Biol. Med. 110:710–713.
50. Saitoh-Inagawa, W., A. Oshima, K. Aoki, N. Itoh, K. Isobe, E. Uchio, S.
Ohno, H. Nakajima, K. Hata, and H. Ishiko. 1996. Rapid diagnosis of
adenoviral conjunctivitis by PCR and restriction fragment length polymor-
phism analysis. J. Clin. Microbiol. 34:2113–2116.
51. Sarantis, H., G. Johnson, M. Brown, M. Petric, and R. Tellier. 2004. Com-
prehensive detection and serotyping of human adenoviruses by PCR and
sequencing. J. Clin. Microbiol. 42:3963–3969.
52. Schnurr, D., and M. E. Dondero. 1993. Two new candidate adenovirus
serotypes. Intervirology 36:79–83.
53. Schu¨tz, E., and N. von Ahsen. 1999. Spreadsheet software for thermody-
namic melting point prediction of oligonucleotide hybridization with and
without mismatches. BioTechniques 27:1218–1224.
54. Seidemann, K., A. Heim, E. D. Pfister, H. Ko¨ditz, A. Beilken, A. Sander, M.
Melter, K. W. Sykora, M. Sasse, and A. Wessel. 2004. Monitoring of ade-
novirus infection in pediatric transplant recipients by quantitative PCR:
report of six cases and review of the literature. Am. J. Transplant. 4:2102–
2108.
55. Shenk, T. 2001. Adenoviridae: the viruses and their replication. Lippincott-
Raven Publishers, New York, N.Y.
56. Shimada, Y., T. Ariga, Y. Tagawa, K. Aoki, S. Ohno, and H. Ishiko. 2004.
Molecular diagnosis of human adenoviruses D and E by a phylogeny-based
classification method using a partial hexon sequence. J. Clin. Microbiol.
42:1577–1584.
57. Stalder, H., J. C. Hierholzer, and M. N. Oxman. 1977. New human adeno-
virus (candidate adenovirus type 35) causing fatal disseminated infection in
a renal transplant recipient. J. Clin. Microbiol. 6:257–265.
58. Strimmer, K., and A. von Haeseler. 1996. Quartet puzzling: a quartet max-
imum-likelihood method for reconstructing tree topologies. Mol. Biol. Evol.
88:22–24.
59. Swenson, P. D., G. Wadell, A. Allard, and J. C. Hierholzer. 2003. Adenovi-
ruses, p. 1404–1417. In P. R. Murray, E. J. Baron, M. A. Pfaller, J. H.
Jorgensen, and R. H. Yolken (ed.), Manual of clinical microbiology, 8th ed.,
vol. 2. ASM Press, Washington, D.C.
60. Tamura, K., and M. Nei. 1993. Estimation of the number of nucleotide
substitutions in the control region of mitochondrial DNA in humans and
chimpanzees. Mol. Biol. Evol. 10:512–526.
61. Tanaka, K., N. Itoh, W. Saitoh-Inagawa, E. Uchio, S. Takeuchi, K. Aoki, E.
Soriano, M. Nishi, R. B. Junior, C. M. Harsi, L. Tsuzuki-Wang, E. L.
Durigon, K. E. Stewien, and S. Ohno. 2000. Genetic characterization of
adenovirus strains isolated from patients with acute conjunctivitis in the city
of Sao Paulo, Brazil. J. Med. Virol. 61:143–149.
62. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W:
improving the sensitivity of progressive multiple sequence alignment through
sequence weighting, position-specific gap penalties and weight matrix choice.
Nucleic Acids Res. 22:4673–4680.
63. Venard, V., A. Carret, D. Corsaro, P. Bordigoni, and A. Le Faou. 2000.
Genotyping of adenoviruses isolated in an outbreak in a bone marrow trans-
plant unit shows that diverse strains are involved. J. Hosp. Infect. 44:71–74.
64. Wadell, G., A. Allard, M. Evander, and Q. Li. 1986. Genetic variability and
evolution of adenoviruses. Chem. Scripta 26B:325–335.
65. Wigand, R. 1987. Pitfalls in the identification of adenoviruses. J. Virol.
Methods 16:161–169.
66. Wigand, R., D. Keller, and I. Werling. 1982. Immunological relationship
among human adenoviruses of subgenus D. Arch. Virol. 72:199–209.
67. Williams, J., T. Grodzicker, P. Sharp, and J. Sambrook. 1975. Adenovirus
recombination: physical mapping of crossover events. Cell 4:113–119.
68. Wu, E., S. A. Trauger, L. Pache, T. M. Mullen, D. J. von Seggern, G. Siuzdak,
and G. R. Nemerow. 2004. Membrane cofactor protein is a receptor for adeno-
viruses associated with epidemic keratoconjunctivitis. J. Virol. 78:3897–3905.
69. Xu, W., M. C. McDonough, and D. D. Erdman. 2000. Species-specific iden-
tification of human adenoviruses by a multiplex PCR assay. J. Clin. Micro-
biol. 38:4114–4120.
15276 MADISCH ET AL. J. VIROL.