Nucleotide sequence and phylogenetic classification of candidate human
papilloma virus type 92
Ola Forslund,a,* Annika Antonsson,aGeoff Higgins,bHoang Ly,bHajo Delius,c
Andreas Hunziker,cand Ethel-Michele de Villiersc
aDepartment of Medical Microbiology, Malmo ¨ University Hospital, Lund University, Malmo ¨, Sweden
bInfectious Diseases Laboratories, Institute of Medical and Veterinary Science, Adelaide, Australia
cDivision for Tumorvirus Characterization, Deutsches Krebsforschungszentrum, Heidelberg, Germany
Received 5 February 2003; returned to author for revision 20 March 2003; accepted 29 April 2003
From a basal cell carcinoma (BCC) the complete genome of candidate human papillomavirus (HPV) type 92 was characterized. Phyloge-
netically, the candidate HPV 92 was relatively distantly related to other cutaneous HPV types within the B1 group. By quantitative real time PCR,
94 viral copies were present per cell in the BCC and another BCC contained 1 viral copy per cell. Lower copy numbers were found in two solar
keratoses (1 copy per 33 cells and 1 copy per 60 cells) and two squamous cell carcinomas (1 copy per 436 cells and 1 copy per 1143 cells). The
high viral load of HPV 92 in two BCCs differs from the low amount of HPV DNA reported from nonmelanoma skin cancers.
© 2003 Elsevier Science (USA). All rights reserved.
Keywords: HPV; NMSC; Quantification; PCR; BCC; SCC; Solar keratosis; Complete genome
Human papillomavirus (HPV) has tropism for keratino-
cytes, and certain genital HPV types are the cause of cer-
vical cancer and their high-grade precursor lesions (zur
Hausen, 1999). HPV types associated with cutaneous le-
sions have been isolated from benign warts and several
studies have reported the presence of HPV DNA in prema-
lignant lesions and nonmelanoma skin cancer (NMSC),
mainly in immunosuppressed patients (Berkhout et al.,
2000, 1995; de Jong-Tieben et al., 1995, 2000; de Villiers et
al., 1997; Hopfl et al., 1997; Shamanin et al., 1996). Many
of the cutaneous HPV types were originally isolated from
lesions occurring in patients with epidermodysplasia verru-
ciformis (EV) (Majewski and Jablonska, 2002). The 15
HPV types originally detected in the EV patients (Howley,
1996) have historically been designated EV types but are
grouped according to phylogenetic classification within group
B1 (hereafter referred to as B1 types), while other related skin
HPV types (HPV types 4, 48, 50, 60, 65) belong to group
B2 (Chan et al., 1995). The HPV types of group B1 have,
however, recently also been demonstrated in the general
population both in malignant lesions (Harwood et al., 2000;
Wieland et al., 2000) and in normal skin (Antonsson et al.,
2000; Astori et al., 1998; Boxman et al., 1997, 2000).
However, no specific HPV type has hitherto been associated
with malignant skin lesions among the general population.
We recently demonstrated a new HPV type in a basal cell
carcinoma (BCC)(Forslund et al., 2003), and the aim was to
characterize its complete genome and to measure the viral
load in a limited number of skin lesions.
The complete genomic sequence
Candidate HPV 92 was originally identified by the am-
plification of a 480-bp fragment (Accession No. AF489730)
* Corresponding author. Department of Medical Microbiology, Malmo ¨
University Hospital, SE-205 02 Malmo ¨, Sweden. Fax: ?46-40-337312.
E-mail address: firstname.lastname@example.org (O. Forslund).
Available online at www.sciencedirect.com
Virology 312 (2003) 255–260www.elsevier.com/locate/yviro
0042-6822/03/$ – see front matter © 2003 Elsevier Science (USA). All rights reserved.
within the L1 open reading frame (ORF). The remaining
part of the genome was amplified by “long” PCR using
primers within this short fragment. The combination of
these two fragments comprises the candidate HPV 92 full-
length genome of 7461 bp having a GC content of 40%
(Accession No. AF531420).
The genome of candidate HPV 92 displayed the early
(E1 to E7) and late (L1 and L2) ORFs (Fig. 1), separated by
an upstream regulatory region (URR) of 390 bp. Analysis
revealed three E2 binding sites [ACC(N)6GGT] in the URR
sequence, with the first nucleotide of the motifs at nucleo-
tides (nt) 7270, 7404, and 7431 and one within the E6 ORF
at nt 114. The function of such an E2 binding site within the
E6 ORF is still unclear but is also found for the B1 type
HPV 22. Four possible binding sites for nuclear factor,
NF-1 (TTGGC), were identified at nt 7323, 7346, 7378, and
7389, one site for activator protein, AP-1 (TGACTAA), at
nt 7250 and a TATA box at nt 52. Polyadenylation signals
(AATAAA) were located at nt 7200 for the late mRNA and
at nt 4146 for the early mRNA.
The putative E6 protein consisted of 138 amino acids and
CxxC(x)29CxxC, separated by 37 amino acids. The zinc
binding domains of E6 are usually separated by 36 amino
acids in other HPV genomes. The E7 protein was 91 amino
acids long and showed one conserved zinc binding domain,
CxxC(x)29CxxC, and a motif (LxCxE) for binding to pRB.
The sizes of putative proteins are shown in Table 1.
The candidate HPV 92 was positioned in the cluster of
HPV types in the B1 group (Fig. 2). The closest related
sequence was FA56 (92% DNA sequence identity), a sub-
type of candidate HPV 92 isolated from the forehead of a
41-year-old Japanese male (Antonsson et al., 2003a). The
entire L1 open reading frame (ORF) of candidate HPV 92
shared the highest similarity with HPV 49 (72%). The
inclusion of new putative HPV types from our recent studies
(Antonsson et al., 2003a, 2003b; Forslund et al., 1999,
2003) led to an expansion of the B1 branch, by an additional
27 types, in comparison to 22 HPV types characterized
previously. Similarly, the branch B2 was expanded by 55
new putative HPV types (FA types) compared to the pre-
vious 5 HPV types (HPV 4, 48, 50, 60, 65).
Quantification of viral DNA
In order to measure the number of viral copies per human
cell, six biopsy samples containing candidate HPV 92 DNA
(Forslund et al., 2003) were analyzed by real time PCR
assay, in parallel with standards of known amounts of
cloned viral DNA and human DNA. In the BCC samples,
from which the candidate HPV 92 was originally cloned, 94
copies per cell was found, and in another BCC 1 copy per
cell was demonstrated. Two solar keratoses each contained
1 copy per 33 cells and 1 copy per 60 cells, whereas two
cases of squamous cell carcinoma harbored only 1 copy per
436 cells and 1 copy per 1143 cells (Table 2). For confir-
mation of the real time PCR assay, a 10-fold dilution series
of the BCC with 94 copies per cell was tested by PCR and
visualized by gel electrophoresis. The dilution series dem-
onstrated an end point at 10?6for HPV 92 (data not shown).
This end point well reflects the viral copy number given by
the real time PCR.
The complete genome of candidate HPV 92 was charac-
terized from two overlapping amplicons. However, several
unsuccessful attempts were performed to clone the genome
without PCR amplification. Hence, we resorted to amplifi-
cation by PCR to enable cloning of the complete genome.
The nucleotide error rate after 40 cycles of PCR using Taq
polymerase has been reported to be approximately 1 per 800
nucleotides (Forslund and Hansson, 1996). The candidate
HPV 92 was therefore cloned from a PCR product gener-
ated with the high-fidelity Expand Taq polymerase resulting
in a 2-fold increased fidelity of DNA synthesis (Gunther et
al., 1998), and its error rate was therefore estimated to be
Fig. 1. Putative open reading frames of candidate HPV type 92 (7461 bp).
ORFs in the strand analogous to mRNA
ORF Nucleotide position
Note. No ATG was found in the nucleotide sequence of the E4 ORF.
Rapid Communication / Virology 312 (2003) 255–260
about 1 per 1600 nucleotides. The patterns of the genes
characteristic of papillomaviruses were not influenced by
the error rate.
The role of HPV infections in the pathogenesis of NMSC
has been debatable due to the low copy numbers reported. It
has even been speculated that the presence of less than one
copy per cell in the lesions may have resulted from con-
tamination with HPV DNA from the surrounding healthy
skin (Antonsson et al., 2000; Astori et al., 1998; Boxman et
al., 1997; Boxman et al., 2000; Forslund et al., 2003). The
viral load of about 94 copies per cell in the BCC analyzed
here is in sharp contrast to the low amount (less than 1 copy
per 10 cells) of viral DNA previously reported in NMSC
(Bens et al., 1998; Meyer et al., 2001). The presence of
normal cells within the biopsy cannot be excluded and could
have contributed to a dilution effect. Therefore, the viral
copy number per infected cell may be even higher than
reported here. Further studies, including the analysis of
microdissected tissues, localizing the HPV DNA to keratin-
ocytes and accurately measuring the HPV load per keratin-
ocyte should enable a more precise quantification.
The phylogenetic analysis revealed that candidate HPV
92 was relatively distantly related to other known HPV
types. This great diversity observed among the HPVs in our
tree may be indicative of other unknown HPV types that
might be positioned between the B1 and B2 clusters. Con-
tinuous improvements in methods used for the detection of
cutaneous HPV types may eventually lead to the identifica-
tion of additional HPV types, which may also be associated
with skin cancer.
Fig. 2. Phylogenetic tree of partial L1 open reading frames (?430 bp) of 111 cutaneous HPV types and candidates. Candidate HPV 92 is positioned at the
left of group B1. The A2 group is composed of the characterized HPV types 3, 10, 28, 29, and 77 but these were not included in the tree. Boldface represents
characterized HPV types.
Rapid Communication / Virology 312 (2003) 255–260
Initial screening of samples demonstrated the presence of
candidate HPV 92 viral DNA in 12% (7/59) of Australian
patients, from BCC, squamous cell carcinoma, solar kera-
tosis, and normal skin (Forslund et al., 2003). A larger
controlled study including healthy matched controls will be
needed to determine whether candidate HPV 92 is a main
player in the development of malignant lesions of the skin
and more specifically in the etiology of basal cell carcino-
Materials and methods
Sample for cloning of candidate HPV 92
A 3-mm biopsy specimen was taken from a BCC on the
left temple of an immunocompetent man, 89 years old,
referred to the Dermatology Clinic at Royal Adelaide Hos-
pital (Sample 1, Table 2, Forslund et al., 2003). The DNA
was extracted and dissolved in 200 ?l of TE buffer, using a
simple phenol-free method as described earlier (Forslund et
Long amplicons and DNA sequencing
HPV DNA was initially identified by PCR amplification
of a 480-bp fragment (Accession No. AF489703) using the
FAP primers (Forslund et al., 1999). Two type-specific
primers were thereafter designed on this sequence, and an
amplicon of about 7 kb was generated by PCR. The 50-?l
PCR solution contained 5 ?l of sample and 0.5 ?M each
primer (FAIMVSF 5?-AGGATGTACACCATGCTTAGG,
FAIMVSR 5?-GATTCCCTGACACCTTAGGC), 0.2 mM
each dNTP (Amersham Pharmacia Biotech, Buckingham-
shire, UK) 0.2% BSA (Fraction V, Sigma-Aldrich, Sydney,
Australia), 3.5 mM MgCl2, 1? HF buffer, and 1.2 U of Exp
enzyme (Expand High Fidelity PCR System, Roche, Castle
Hill, Australia). The PCR was performed in a MJ thermal
cycler (PTC-200 DNA Engine, Watertown, MA), with cal-
culated control settings using 2 min at 94°C, followed by 10
cycles of 15 s at 94°C, 30 s at 53°C followed by 8 min at
68°C, and 25 cycles of 15 s at 94°C, 30 s at 55°C, followed
by 8 min at 68°C (with an extra 5 s/cycle). The amplified
sample was separated by electrophoresis in a 1.5% agarose
gel (LA grade, Promega, Madison, WI) in 0.5? TBE buffer
and a correct sized amplicon was excised and purified (QIA-
quick PCR purification kit, Qiagen, Hilden, Germany). Four
microliters was used for cloning (TOPO TA Cloning Kit,
Invitrogen, Leek, The Netherlands) and the clone with an
insert of the expected size was recultivated and purified
(Qiaprep, Qiagen). The DNA clone with the insert of about
7 kb was submitted to the Human Papillomavirus Reference
Laboratory (Heidelberg, Germany) and the DNA sequence
was obtained via cycle sequencing (ABI Prism BigDye
Terminator Cycle Sequencing, Applied Biosystems, Foster
City, CA) and analyzed with an automated DNA sequencer
(ABI PRISM 377 DNA Sequencer, Foster City, CA). The
final HPV sequence was derived by combination of the
sequences from the two amplification products and was
designated candidate HPV 92. The prefix candidate is ac-
cording to the terminology of cloned HPV genomes from
PCR products (E. M. deVillers, personal communication).
The candidate HPV 92 has been given GenBank Accession
Phylogenetic analysis of 111 cutaneous HPV types and
The phylogenetic analysis was based on alignments of
111 DNA sequences between the primer sites of FA primers
(about 430 nucleotides) within the L1 ORF of 27 cutaneous
HPV types and putative new HPV types [FA1–13 (Forslund
et al., 1999), FA14–43 (Antonsson et al., 2000), FA44–72
(Antonsson et al., 2003a, 2003b), FAIMVS1–15, and can-
dHPV 92 (Forslund et al., 2003)]. Alignments by ClustalX
were edited with Genedoc, Phylip was used for neighbor
joining and maximum likelihood analysis and the phyloge-
Number of viral copies per cell of candidate HPV 92
Site of lesionDiagnosisViral copies/?l
Note. Samples 2 and 3 were collected from the same patient.
aThe coefficient of variation was calculated from quantities in duplicate tests.
bConverted to number of cells by the assumption that 6.6 pg of human DNA is present per diploid cell.
cThe number of viral copies per cell(s) is the ratio between the viral copy number and the number of cells, per microliter of extracted DNA.
Rapid Communication / Virology 312 (2003) 255–260
netic tree was seen with Treeview. The papillomavirus
groups were labeled according to Chang (Chan et al., 1995).
Quantification of viral DNA
Type-specific PCR for HPV 92 was positive in 6 of 64
skin samples (2/25 BCCs, 2/16 solar keratosis, 2/23 squa-
mous cell carcinomas) of Australian skin tumor patients
(Forslund et al., 2003) (Table 2). These six samples were
then analyzed by a real time PCR with type-specific primers
from the L1 ORF, in order to estimate the amount of HPV
DNA in relation to human DNA. For generation of a stan-
dard curve a 10-fold dilution series of plasmid containing
HPV 92 was used, from 100,000 to 10 copies of viral DNA
in the PCR in a background of 100 pg of placental DNA.
Another aliquot of the sample was amplified by PCR
with ?-globin primers for human DNA (PC03/PC04) (de
Roda Husman et al., 1995) and compared to a standard
curve with 250, 25, and 2.5 ng and 250 and 25 pg of
placental DNA (Sigma-Aldrich). The amount of human
DNA was then converted to number of cells by the assump-
tion that 6.6 pg of DNA is present per diploid cell. Standard
curves were generated from mean threshold cycle (CT)
values of each dilution in duplicate, and mean CTvalues of
samples in duplicate were used for quantity calculation. The
threshold level at which CTvalues were calculated was
manually set to 2. A 10?2dilution was used of the BCC
sample with the highest viral load to obtain CTvalues within
the range of the HPV 92 standard curve. For the PCR, 2.5
?l of template was mixed in a 25-?l reaction mixture
containing 0.5 ?M each HPV 92 primer (L1.178F 5?-
GATCCACATGATGTATTAGTGC and L1.430R 5?CA
TCCTTTGTTCCAGTATTGTAC), 200 ?M each dNTP
(Roche, Bromma, Sweden), 1? SYBR Green (Molecular
Probes, Leiden, The Netherlands), 0.2% bovine serum al-
bumin (Fraction V, Sigma-Aldrich), 3.5 mM MgCl2, Amp 1
? buffer II, and 0.625 U AmpliTaq Gold polymerase (Per-
kin–Elmer, Foster City, CA). Using a GeneAmp 5700 Se-
quence Detection System (Perkin–Elmer) the mixture was
heated to 94°C for 10 min and PCR carried out for 45 cycles
[94°C for 1.5 min, 55°C for 1.5 min (50°C for ?-globin),
and 72°C for 1.5 min]. Post-PCR, dissociation curves were
inspected as a quality control that specific amplicons and
not primer–dimers were the majority of amplicons. In ad-
dition, the remaining amplified materials in the 96-well
reaction plate were gel electrophoresed in a 2% gel contain-
ing ethidium bromide (20 ng/?l) and visualized by UV light
to assure that the measured signals from samples were
obtained from amplicons of the correct size.
For confirmation of the finding of 94 copies per cell of
the BCC by real time PCR, a 10-fold dilution series of the
sample was amplified by PCR with type-specific primers
(L1.178F and L1.430R) under the same conditions as above
(without SYBR Green and using another thermocycler;
Mastercycler Eppendorf, Hamburg, Germany). After ampli-
fication, 5-?l aliquots were gel electrophoresed in a 2% gel
and HPV-92-specific amplicons were identified.
Antonsson, A., Hazard, H., Holmgren, V., Simon, M., Kataoka, A., Hossin,
S., Håakangåard, C., Hansson, B. G., 2003a. Prevalence and type
spectrum of human papillomavirus in normal skin from three conti-
nents. J. Gen. Virol., 84(7), 1881–1886.
Antonsson, A., Forslund, O., Ekberg, H., Sterner, G., Hansson, B.G., 2000.
The ubiquity and impressive genomic diversity of human skin papil-
lomaviruses suggest a commensalic nature of these viruses. J. Virol. 74
Antonsson, A., Karanfilovska, S., Lindqvist, P. G., Hansson, B. G., 2003b.
General acquisition of human papillomavirus infections occurs in early
infancy. J. Clin. Microbiol. 753–02, 41(6), 2509–2514.
Astori, G., Lavergne, D., Benton, C., Hockmayr, B., Egawa, K., Garbe, C.,
de Villiers, E.M., 1998. Human papillomaviruses are commonly found
in normal skin of immunocompetent hosts. J. Invest. Dermatol. 110 (5),
Bens, G., Wieland, U., Hofmann, A., Hopfl, R., Pfister, H., 1998. Detection
of new human papillomavirus sequences in skin lesions of a renal
transplant recipient and characterization of one complete genome re-
lated to epidermodysplasia verruciformis-associated types. J. Gen. Vi-
rol. 79 (Pt. 4), 779–787.
Berkhout, R.J., Bouwes Bavinck, J.N., ter Schegget, J., 2000. Persistence
of human papillomavirus DNA in benign and (Pre)malignant skin
lesions from renal transplant recipients. J. Clin. Microbiol. 38 (6),
Berkhout, R.J., Tieben, L.M., Smits, H.L., Bavinck, J.N., Vermeer, B.J., ter
Schegget, J., 1995. Nested PCR approach for detection and typing of
epidermodysplasia verruciformis-associated human papillomavirus
types in cutaneous cancers from renal transplant recipients. J. Clin.
Microbiol. 33 (3), 690–695.
Boxman, I.L., Berkhout, R.J., Mulder, L.H., Wolkers, M.C., Bouwes
Bavinck, J.N., Vermeer, B.J., ter Schegget, J., 1997. Detection of
human papillomavirus DNA in plucked hairs from renal transplant
recipients and healthy volunteers. J. Invest. Dermatol. 108 (5), 712–
Boxman, I.L., Russell, A., Mulder, L.H., Bavinck, J.N., Schegget, J.T.,
Green, A., 2000. Case-control study in a subtropical Australian popu-
lation to assess the relation between non-melanoma skin cancer and
epidermodysplasia verruciformis human papillomavirus DNA in
plucked eyebrow hairs. The Nambour Skin Cancer Prevention Study
Group. Int. J. Cancer 86 (1), 118–121.
Chan, S.Y., Delius, H., Halpern, A.L., Bernard, H.U., 1995. Analysis of
genomic sequences of 95 papillomavirus types: uniting typing, phylog-
eny, and taxonomy. J. Virol. 69 (5), 3074–3083.
de Jong-Tieben, L.M., Berkhout, R.J., Smits, H.L., Bouwes Bavinck, J.N.,
Vermeer, B.J., van der Woude, F.J., ter Schegget, J., 1995. High
frequency of detection of epidermodysplasia verruciformis-associated
human papillomavirus DNA in biopsies from malignant and premalig-
nant skin lesions from renal transplant recipients. J. Invest. Dermatol.
105 (3), 367–371.
de Jong-Tieben, L.M., Berkhout, R.J., ter Schegget, J., Vermeer, B.J., de
Fijter, J.W., Bruijn, J.A., Westendorp, R.G., Bouwes Bavinck, J.N.,
2000. The prevalence of human papillomavirus DNA in benign kera-
totic skin lesions of renal transplant recipients with and without a
history of skin cancer is equally high: a clinical study to assess risk
factors for keratotic skin lesions and skin cancer. Transplantation 69
de Roda Husman, A.M., Snijders, P.J., Stel, H.V., van den Brule, A.J.,
Meijer, C.J., Walboomers, J.M., 1995. Processing of long-stored archi-
val cervical smears for human papillomavirus detection by the poly-
merase chain reaction. Br. J. Cancer 72 (2), 412–417.
Rapid Communication / Virology 312 (2003) 255–260
de Villiers, E.M., Lavergne, D., McLaren, K., Benton, E.C., 1997. Pre- Download full-text
vailing papillomavirus types in non-melanoma carcinomas of the skin
in renal allograft recipients. Int. J. Cancer 73 (3), 356–361.
Forslund, O., Antonsson, A., Nordin, P., Stenquist, B., Hansson, B.G.,
1999. A broad range of human papillomavirus types detected with a
general PCR method suitable for analysis of cutaneous tumours and
normal skin. J. Gen. Virol. 80 (Pt. 9), 2437–2443.
Forslund, O., Hansson, B.G., 1996. Human papillomavirus type 70 genome
cloned from overlapping PCR products: complete nucleotide sequence
and genomic organization. J. Clin. Microbiol. 34, 802–809.
Forslund, O., Ly, H., Reid, C., Higgins, G., 2003. A broad range of human
papillomavirus types are present in skin of Australian patients with
non-melanoma skin cancers and solar keratosis. Br. J. Dermatol., in
Gunther, S., Sommer, G., Von Breunig, F., Iwanska, A., Kalinina, T.,
Sterneck, M., Will, H., 1998. Amplification of full-length hepatitis B
virus genomes from samples from patients with low levels of viremia:
frequency and functional consequences of PCR-introduced mutations.
J. Clin. Microbiol. 36 (2), 531–538.
Harwood, C.A., Surentheran, T., McGregor, J.M., Spink, P.J., Leigh, I.M.,
Breuer, J., Proby, C.M., 2000. Human papillomavirus infection and
non-melanoma skin cancer in immunosuppressed and immunocompe-
tent individuals. J. Med. Virol. 61 (3), 289–297.
Hopfl, R., Bens, G., Wieland, U., Petter, A., Zelger, B., Fritsch, P., Pfister,
H., 1997. Human papillomavirus DNA in non-melanoma skin cancers
of a renal transplant recipient: detection of a new sequence related to
epidermodysplasia verruciformis associated types. J. Invest. Dermatol.
108 (1), 53–56.
Howley, P.M., 1996. Papillomavirinae: The viruses and their replication,
in: Fields, B.N., Knipe, D.M., Chanock, R.M., Monath, T.P., Howley,
P.M., Melnick, J.L., Roizman, B. (Eds.), Field’s Virology, third ed.
Lippincott-Raven, New York, pp. 2045–2109.
Majewski, S., Jablonska, S., 2002. Do epidermodysplasia verruciformis
human papillomaviruses contribute to malignant and benign epidermal
proliferations? Arch. Dermatol. 138 (5), 649–654.
Meyer, T., Arndt, R., Christophers, E., Nindl, I., Stockfleth, E., 2001.
Importance of human papillomaviruses for the development of skin
cancer. Cancer Detect. Prev. 25 (6), 533–547.
Shamanin, V., zur Hausen, H., Lavergne, D., Proby, C.M., Leigh, I.M.,
Neumann, C., Hamm, H., Goos, M., Haustein, U.F., Jung, E.G., et al.,
1996. Human papillomavirus infections in nonmelanoma skin cancers
from renal transplant recipients and nonimmunosuppressed patients.
J. Natl. Cancer Inst. 88 (12), 802–811.
Wieland, U., Ritzkowsky, A., Stoltidis, M., Weissenborn, S., Stark, S.,
Ploner, M., Majewski, S., Jablonska, S., Pfister, H.J., Fuchs, P.G.,
2000. Communication: papillomavirus DNA in basal cell carcinomas
of immunocompetent patients: an accidental association? J. Invest.
Dermatol. 115 (1), 124–128.
zur Hausen, H., 1999. Papillomaviruses in human cancers. Proc. Assoc.
Am. Physicians 111 (6), 581–587.
Rapid Communication / Virology 312 (2003) 255–260