Rescue of a herpes simplex virus type 1 neurovirulence function with a cloned DNA fragment.

Abstract

A herpes simplex virus type 1 (HSV-1) genetic function that is required for viral replication in the murine central nervous system was unambiguously localized. Thus, cosmid clones of either HSV-1 HindIII fragment C (0.64 to 0.87 map units) or fragment B (0.64 to 0.83 plus 0.91 to 1.0 map units) were employed to restore neurovirulence to an intertypic recombinant (RE6) that is specifically deficient in this property. The neurovirulent recombinants were generated in cell culture by cotransfecting the clone fragments and unit-length RE6 DNA and then selected in mouse brains. Either fragment efficiently conferred neurovirulence to RE6, demonstrating that no short region unique sequences are required. Analyses of the genomic structures of the neurovirulent recombinants showed that, in every case, HSV-1 information from 0.71 to 0.83 map units was incorporated into the RE6 genome. Cleavage of HindIII fragment C with EcoRI eliminated its capacity to rescue RE6. Virulence could be restored by the addition of HSV-1 BamHI fragment L (0.71 to 0.74 map units) that spans an EcoRI site at 0.72 map units. The precise location of this HSV-1 neurovirulence function is discussed.

Full text

Page 1

Vol. 55, No. 2
JOURNAL OF VIROLOGY, Aug. 1985, p. 504-508
0022-538X/85/080504-05$02.00/0
Copyright C 1985, American Society for Microbiology
Rescue of a Herpes Simplex Virus Type 1 Neurovirulence Function
with a Cloned DNA Fragment
RICHARD L. THOMPSON,'* GAYATHRI V. DEVI-RAO,' JACK G. STEVENS,2 AND EDWARD K. WAGNER'
Department ofMolecular Biology and Biochemistry, University of California, Irvine, California 92717,1 and Department
ofMicrobiology and Immunology, University of California, Los Angeles, California 900242
Received 10 December 1984/Accepted 2 May 1985
A herpes simplex virus type 1 (HSV-1) genetic function that is required for viral replication in the murine
central nervous system was unambiguously localized. Thus, cosmid clones of either HSV-1 Hindlll fragment
C (0.64 to 0.87 map units) or fragment B (0.64 to 0.83 plus 0.91 to 1.0 map units) were employed to restore
neurovirulence to an intertypic recombinant (RE6) that is specifically deficient in this property. The
neurovirulent recombinants were generated in cell culture by cotransfecting the clone fragments and
unit-length RE6 DNA and then selected in mouse brains. Either fragment efficiently conferred neurovirulence
to RE6, demonstrating that no short region unique sequences are required. Analyses of the genomic structures
of the neurovirulent recombinants showed that, in every case, HSV-1 information from 0.71 to 0.83 map units
was incorporated into the RE6 genome. Cleavage ofHindlll fragment C with EcoRI eliminated its capacity to
rescue RE6. Virulence could be restored by the addition ofHSV-1 BamHI fragment L (0.71 to 0.74 map units)
that spans an EcoRI site at 0.72 map units. The precise location of this HSV-1 neurovirulence function is
discussed.
The identification of viral genes that control pathogenic
properties at the level of the animal and the elucidation of
their function are of fundamental importance to an under-
standing of the basic molecular mechanisms of disease. With
respect to the natural history of herpes simplex virus (HSV),
the nervous system is a critical target organ, and the general
pathologic features of this viral-host interaction have long
been appreciated (4). Thus, after acute infection at the
surface of the body, the virus travels through axons and can
establish latent infections in both the peripheral (11) and
central (7) nervous systems. Viral reactivation produces
secondary movement in axons, and clinically apparent dis-
ease that ranges in severity from relatively benign cutaneous
lesions to acute fatal encephalitis may ensue.
It is of importance to determine the genetic properties of
this virus that predispose it to invade and destroy nervous
tissue. However, the viral genes that control either
neuroinvasiveness or neurovirulence are little understood. It
is known that the viral thymidine kinase gene is required for
maximum demonstration of neurovirulence (3). Increased
neurovirulence has been demonstrated in one avirulent
natural isolate by serial passage of virus in the murine central
nervous system (6). In this case, DNA from the virulent
agent possessed an altered restriction enzyme profile, but
whether this change
is related to the increase in
neurovirulence has not been established. Among wild-type
and laboratory strains, it has been demonstrated that isolates
of HSV, and even subclones of the same isolate, differ in
degree of neurovirulence (2). However, in no instances have
consistent changes in either viral proteins or DNA been
detected.
Our previous studies described an HSV type 1 (HSV-
1)-HSV-2 intertypic recombinant that is completely and
specifically nonneurovirulent after intracranial inoculation of
mice (12). This recombinant (RE6) is at least 10-million-fold
less neurovirulent than either parental strain but replicates
*Corresponding author.
as well or better than the parents in all tested cells and
tissues, with exception of the mouse brain.
To understand the basis of this property, we developed a
technique that allows identification and localization of HSV
genes which control pathogenic phenotypes displayed in
vivo (13). This strategy involves cotransfecting tissue cul-
tures with unit-length DNA from an HSV strain that lacks a
biologically selectable marker with restriction endonuclease-
fragmented DNA of a strain that displays the desired phe-
notype. Phenotypically positive recombinants are then se-
lected in vivo in the tissue of interest. This procedure has the
advantage of eliminating the background of avirulent paren-
tal virus as well as selecting against any recombinants that
replicate poorly at the temperature of the mouse. Utilizing
this approach, our previous results are here confirmed and
extended by demonstrating the rescue of RE6
neurovirulence with cloned HSV-1 DNA fragments.
Previous results demonstrated that an electrophoretically
purified HSV-1 DNA fragment (HindIIl fragment B, 39
kilobases, 0.64 to 0.83 plus 0.91 to 1.0 map units [m.u.])
contains the type
neurovirulence to RE6. To unequivocally confirm the role of
HSV-1 HindIll fragment B in HSV neurovirulence, it was
necessary to utilize a cloned DNA fragment for rescue.
Since the fragment size is large, the cosmid cloning vector
PHC-79 was used. The cloning procedure employed was
essentially as described (5) except for certain modifications
designed to select cosmids that had incorporated the large
Hindlll fragments of strain 17 syn+. When 12 cosmids were
analyzed, two contained HSV-1 HindIIl fragment B (39
kilobases, 0.64 to 0.83 plus 0.91 to 1.0 m.u.), two HindIll
fragment A (41.5 kilobases, 0.25 to 0.52 m.u.), and eight
Hindlll fragment C (35 kilobases, 0.64 to 0.87 m.u.). Despite
a recA- strain ofEscherichia coli being used as the recipient
strain, four of the cosmids contained detectable permuta-
tions when analyzed with six restriction endonucleases.
Two ofHSV-1 Hindlll fragment C cosmids (PHC 17-3 and
PHC 17-11) and one of the HSV-1 Hindlll fragment B
cosmids (PHC 17-1) were then tested for their ability to
1 gene or genes which restore
504

Page 2

NOTES
505
0.7
0.8
0.9
L~~~~~~D
m
A
I |1
K
L
A
Ipl
L
BSQ
K
IN
G
IF
|H
|~~~~~
M
NIIIE
I Q
I
L
E
Q
T
R|
tE *W
E
A
W
K
F
P
V
U
Z
'F'
L
C
Kind III
Eco RI
Bgl
II
Hpa I
Ba
RI
FIG. 1. Reference map of restriction endonuclease sites for HSV-1 (strain 17 syn+) and HSV-2 (strain HG52) in the region 0.6 to 0.95 m.u.
The genome of HSV is shown schematically as a single horizontal line, and the repetitive sequences flanking the long and short regions are
depicted as open boxes. The joint between the long and short regions is indicated by an extended vertical line at about 0.83 m.u. The
restriction enzyme sites for HindIII, EcoRl, and BglII, HpaI, and Ba,nHI are shown as short vertical lines, with HSV-1 sites along the top
and the HSV-2 sites along the bottom. Fragment names (capital letters) are as described in the text.
confer neurovirulence. Initially, cosmid PHC 17-11 (17 syn+
HindIII fragment C) was employed as the donor DNA, and
under the cotransfection and in vivo selection conditions
described previously (13),
neurovirulence upon RE6 with high efficiency. In three
separate neurovirulence rescue experiments (of 10 transfec-
tion plates each), between 70 and 100% of the cultures
contained virus that killed mice. Virus was obtained from the
mouse brains at necropsy, plaque purified, and employed to
generate PFU per 50% lethal dose (PFU/LD50) ratios.
Neurovirulence of the various plaque isolates was ana-
lyzed by the generation of PFU/LD50 ratios after direct
intracranial inoculation as described previously (12). During
the course of these studies, the neurovirulent strain of
HSV-1 (17 syn+) consistently produced ratios of between 5
and 15. In contrast, mice could not be killed with the
avirulent recombinant RE6, even at inoculation titers of 4.1
x
107, the highest stock titer produced. Most of the
neurovirulent recombinants generated with cloned HindIII
fragment C yielded ratios of approximately 103, an example
ofwhich is recombinant RC 14.1, which gave a ratio of 1.1 x
103 in this assay. However, some tested recombinants, such
as RC 18.1 (1.4 x 104) and RC 19.2 (2.2 x 105), gave higher
ratios.
The quantitative differences between PFU/LD50 ratios
established by these recombinants and the wild-type virus
can be explained in two obvious but not exclusive ways.
First, the rescue of RE6 may be a multigenic event and
various amounts of the HindlIl C fragment could have been
incorporated between 0.72 and 0.83
HindIll C fragment employed could contain mutations that
would affect either neurovirulence or general viral replica-
tion. Other possibilities and the basis for variation between
recombinants derived from the same clone will be discussed
below.
this clone conferred
m.u. Second, the
To determine whether differences in the amount ofHSV-1
HindlIl fragment C incorporated into the NV-1A' recombi-
nants could account for the variation in observed lethality,
the genomic structure of three of the plaque-purified
neurovirulent recombinants was analyzed with six restric-
tion endonucleases (BamHI, EcoRI, HindIII, XbaI, HpaI,
and BglII). These restriction enzymes were chosen to con-
veniently differentiate between HSV-1 and HSV-2 DNA in
the region of 0.7 to 0.83 m.u. in which the original genomic
exchange between type 1 and type 2 occurred in RE6. A
reference map for these enzymes in this region is shown in
Fig. 1. RC 14.1, RC 18.1, and RC 19.2 were chosen for
analysis as they reflected the full range of recovered
neurovirulent phenotypes.
The HpaI digests are the easiest to interpret, and an
example of an 0.8% agarose gel of such a digest is shown in
Fig.
neurovirulent recombinants selected in vivo incorporated
HSV-1 DNA between 0.71 and 0.83 m.u. HpaI fragments
corresponding in size to HpaI fragments Q (0.713 to 0.737
m.u.), T (0.737 to 0.759 m.u.), W (0.759 to 0.764 m.u.), R
(0.764 to 0.788 m.u.), and L (0.788 to 0.827 m.u.) are
apparent. Some heterogeneity can be seen in the HpaI
fragment R and L bands, but this region is known to vary
even among plaque isolates ofthe same strain (8). These data
were entirely consistent with those obtained from analysis
with the other five restriction endonucleases employed and
indicate that, in every case, the entire region from 0.71 to
0.83 m.u. was replaced by type 1 information.
These findings were confirmed by Southern blot analysis
(10) in which nick-translated cloned HSV-1 probes labeled
with [32P]dCTP or biotin-conjugated dUTP were used. Rep-
resentatives of such blots are shown in Fig. 3. In Fig. 3A, 17
syn+, RC 14.1, RC 18.1, RC 19.2, and RE6 were cleaved
with HpaI, blotted, and hybridized with a 32P-labeled clone
2.It is apparent from these data that all of the
0.6
VOL. 55, 1985

Page 3

506
NOTES
of 17 syn+ BamHI fragment L (0.71 to 0.745 m.u.). The
radioactive bands in the figure confirm that all three NV-1A+
recombinants incorporated type 1 DNA corresponding to
HpaI fragments Q (0.713 to 0.737 m.u.) and T (0.737 to 0.759
m.u.). The same DNAs were cleaved with BamHI and again
hybridized to the 17 syn+ BamHI fragment L probe (Fig.
3B). Here, it is apparent that all three NV-1A+ recombinants
have lost the fusion fragment RE6 BamHI fragment LIK2
and regained a normal sized 17 syn+ BamHI fragment L.
Thus, in every case, the original site of the crossover at
-0.72 m.u. in RE6 has been replaced. Within the limits of
resolution of these procedures, the HSV-1 neurovirulence
function that rescues RE6 is between 0.71 and 0.83 m.u.
A Zv
X
;f;i_
z
4
I
..
L
_
_
_
_
~~~~L1K 2
BK
HIr
.~~~~-
-.
%D
vA
t
X.
_W*I
4)
I-
C
L.)
cr
zX
t:w~~V'
w
~~
am
am sow-
i
i:
0
T
Hpa I
FIG. 3. Southern blot analysis ofHSV strains 17 syn+, RE6, and
neurovirulent recombinants. (A) DNAs were digested with HpaI,
electrophoresed, and blotted to nitrocellulose filters as described
previously (13). The filters were hybridized to a nick-translated
clone probe of HSV-1 BamHI fragment L (0.71 to 0.74 m.u.). Note
that all the neurovirulent recombinants have acquired HpaI frag-
ments Q (0.713 to 0.737 m.u.) and T (0.737 to 0.759 m.u.). (B) The
DNAs were cleaved with BamHI and again probed with HSV-1
BamHI fragment L. Note that all the recombinants have lost the
fusion fragment LIK2 present in RE6 and regained a normal 17 syn+
BamHI L fragment.
FIG. 2. Electropherogram of HSV HpaI DNA fragments. DNA
from HSV strains 17 syn+, RE6, and neurovirulent recombinants
was digested with HpaI and electrophoresed in a 0.8% agarose gel as
described in the text. To the right of the lane depicting RE6 DNA,
white bars mark HSV-1 bands that
neurovirulent recombinants but not in the parental RE6. These
bands correspond to HSV-1 HpaI fragments L (0.788 to 0.827 m.u.),
Q (0.713 to 0.737 m.u.), T (0.737 to 0.759 m.u.), W (0.759 to 0.764
m.u.), and R (0.764 to 0.788 m.u.). Note that the RE6 fragment that
comigrates with 17 syn+ HpaI fragment J is actually the fusion
fragment RE-6 HpaI Q1E2.
are present in all of the
As mentioned previously, an alternative explanation for
the variation in virulence observed in the recombinants
might be a hidden mutation or permutation in PHC 17-11. To
test this possibility, two additional cosmid clones were
employed in rescue experiments. The first one tested was an
additional cosmid containing HindIII fragment C (PHC
17-3), which contained no detectable permutations after
cleavage with six restriction endonucleases. In a total offour
such experiments of 20 plates each, no successful rescue of
neurovirulence was detected. Two of the experiments were
controlled with transfections employing PHC 17-11 (see
above). In these cultures, 7 of 10 plates and 9 of 10 plates
killed mice. Therefore, despite PHC 17-3 containing no
detectable permutations, it must contain a defect that pre-
vents the restoration of neurovirulence to RE6. One should
note that such a defect need not reside in a gene controlling
HSV neurovirulence per se; a mutation in any essential gene
in this region could result in the failure of recombinants to
replicate.
PHC 17-1, a clone of the 17 syn+ HindIll fragment B, was
tested next. In two transfection experiments, 9 of 10 and 10
of 10 of the tested cultures contained neurovirulent recom-
binants. Two unrelated isolates were plaque purified and
analyzed for genomic structure and neurovirulence. Both
RB 16.1 and RB 17.1 have incorporated HSV-1 sequences
J. VIROL.
.i
ello
Wm

Page 4

NOTES
507
TABLE 1. Rescue of RE6 neurovirulence with subfragments of
HSV-1 HindIII fragment C
PHC 17-11 (Hindlll fragment C) cleaved bya:
PHC 17-11
HindSIV
plus EcoRI
~~plusHSV-1
BamHI-L
ND
ND
3/10
3/10
Expt
HindIII plus
Eol
EcoRI
BamHI-L
ln
alone
HindiII
1
2
3
4
aRabbit skin cells were cotransfected with RE6 DNA and HSV-1 HindlIl
fragment C cleaved from the vector with HindIII, with both HindIII and
EcoRI, with HSV-1 BamHI fragment L alone, or with HindIII-EcoRI-digested
HindIIu fragment C plus HSV-1 BamHI fragment L. Mice were inoculated
with the progeny oftransfection cultures, and when they died from encephali-
tis, virus was isolated from the braln tissue and reinoculated at 1O5 PFU per
mouse. When these mice also died, the transfection culture was scored as
positive for neurovirulent recombinants. Values are the number of transfec-
tion cultures positive over number tested. ND, Not done.
10/10
9/10
10/10
10/10
0/10
0/20
0/20
0/20
ND
0/40
0/20
ND
between 0.71 and 0.83 m.u. (see Fig. 2). One should note
that, in both cases, HpaI fragment R (0.71 to 0.735 m.u?) is
smaller than normal, a change that is not evident in the
cosmid DNA (data not shown). The PFU/LD5Q ratios ofRB
16.1 and RB 17.1 were 1.1 x 104 and 1.4 x 104, respectively.
In an effort to map more closely the defective
neurovirulence function of RE6, HSV-1 HindIII fragment C
was cleaved with both HindIII and EcoRI and used in rescue
experiments. EcoRI cleaves the HindIII C fragment in two
places: at 0.72 and 0.85 m.u. on the viral genome. It should
be noted that all portions of the fragment (HindIII-EcoRI D
through E, 0.64 to 0.72 m.u.; EcoRI E plus K, 0.72 to 0.85
m.u.; and EcoRI-HindIII K through M, 0.85-0.87 m.u.) were
present in the transfection cocktail. This cleavage com-
pletely eliminated the ability to restore neurovirulence to
RE6. Of a total of 60 transfection plates tested, none
contained neurovirulent recombinants. The simplest expla-
nation of these data is that an HSV-1 neurovirulence func-
tion resides at, or close to, one of these EcoRI sites.
The EcoRI site at 0.72 m.u. resides close to the original
point of crossover between HSV-1 and HSV-2 DNA se-
quences in RE6 (13). To determine whether a permutation
introduced at this site in RE6 is involved in its avirulent
properties, HSV-1 BamHI fragment L (0.71 to 0.745 m.u.)
that spans the EcoRI site at 0.72 m.u. was cloned. Since it
was shown above that some clones may contain permuta-
tions that preclude rescue, five independent clones of
BamHI fragment L were derived and pooled for use in these
rescue experiments.
As previously, the viral progeny of the transfection cul-
tures were injected intracranially into mice, and when the
mice died, virus was recovered from brain tissue and
reinjected at 105 PFU per mouse. When these mice also died,
the culture was scored as positive for containing
neurovirulent recombinants.
As before, HSV-1 HindIll fragment C was highly efficient
in restoring neurovirulence to RE6, and greater than 90% of
these cultures contained neurovirulent recombinants (Table
1). When cleaved with EcoRI, this fragment was not able to
rescue RE6 (no neurovirulent recombinants in 60 plates),
However, when HSV-1 BamHI fragment L was added to the
EcoRI-cleaved HindIll C fragment, 30% of the cultures
contained neurovirulent virus. Since BamHI fragment L was
capable ofrestoring the ability of the EcoRI-cleaved HindIlI
fragment to rescue RE6, intact HSV-1 sequences near the
EcoRI site at 0.72 m.u. must have been required. Whether
the BamHI L fragment was incorporated into these
neurovirulent recombinants could not be determined since
recombination near the EcoRI site of the cleaved cosmid
would generate a fragment of nearly the same size as intact
BamHI fragment L, and this fragment would still contain
HSV-2 information.
When the BamHI L fragment was employed alone in
cotransfections, no neurovirulent recombinants were recov-
ered. It should be noted that signs of central nervous system
disease were detected in some of the mice inoculated with
these cultures, and -10% of these cultures were lethal (data
not shown). However, neurovirulent virus could not be
obtained from the brains of these mice at necropsy.
It was previously shown that the intertypic recombinant
RE6 is specifically blocked from replication in mouse brains
(12). Here, through the use of in vivo recombinant selection,
neurovirulence was restored to RE6 with cosmid cloned
HSV-1 DNA sequences comprising HindIII fragments B
(0.64 to 0.83 plus 0.91 to 1.0 m.u.) and C (0.64 to 0.87 m.u.).
Since both fragments rescue RE6, it is demonstrated that no
unique short region sequences are required. These results
unequivocally demonstrate that one or more HSV-1 gene
function(s) residing in HindIII fragment C is associated with
a greater than 100,000-fold increase in neurovirulence. Little
is known about the in vivo functions ofHSV genes encoded
in this region. Genes that control syncytial formation in
tissue culture cells have been mapped to this area (9). By
utilizing this marker, it was found that intratypic recombi-
nants of various syncytial morphology produced character-
istic comeal lesions in rabbit eyes; however, in this instance,
the neurovirulent nature of the syncytial parent was not
transferred to the recombinants (1).
The PFU/LD50 ratios of neurovirulent recombinants gen-
erated with these fragments were higher than those observed
with wild types and varied even when the same clone was
employed in transfection experiments. Two possible reasons
for this result were studied, and there are other possibilities.
First, isolates of the same strain ofHSV have been shown to
vary over several orders of magnitude in neurovirulence,
and the observed variation may reflect the natural variation
in the RE6 population once the defect is repaired. Second,
although HSV-1 information from 0.71 to 0.83 m.u. was
incorporated into the RE6 genome in all cases, it is possible
that a variable and undetectable amount ofHSV-1 sequences
was incorporated into the preexisting type 1 portions of the
RE6 genome. In addition, multiple crossover events, result-
ing in small inserts of HSV-2 sequences, cannot be com-
pletely ruled out. Thus, more than one gene may be needed
to quantitatively restore neurovirulence. Third, although the
replication kinetics of these agents were not systematically
investigated, at least one of the relatively avirulent recom-
binants (RC 19.2) replicates poorly in culture and thus would
fit into an overall category of defectiveness for recombinant
viruses. Several possible reasons for such agents were
previously discussed (13), and they could result from either
permutations introduced at the time of transfectiop or re-
combination or be due to preexisting mutations in the
employed cloned fragments. In this latter case, such muta-
tions must exist since one HindlIl fragment C clone em-
ployed would not rescue RE6 despite its presenting a normal
restriction enzyme fragment profile upon cleavage with six
endonucleases.
The data presented here also suggest that the studied
neurovirulence defect resides near the EcoRI site at 0.72
VOL. 55, 1985

Page 5

508
NOTES
m.u. since cleavage of HindIII fragment C with EcoRI
eliminated the ability of the fragment to restore
neurovirulence to RE6. This can be restored by the addition
of 17 syn+ BamHI fragment L (0.71 to 0.745 m.u.). Since this
EcoRI site lies very close to the original point of crossover
between type 1 and type 2 information in RE6, a permutation
introduced into RE6 at this point during the act of recombi-
nation is a likely cause of its nonneurovirulence.
The neurovirulence of RE6 could not be rescued with
BamHI fragment L alone, and it may be that some other
region of HindIlI fragment C is also involved. However, in
HSV-1, the transcription map of the BamHI L fragment is
rather complex with several overlapping transcripts on op-
posite strands. The same is true of the transcript map of
HSV-2 (R. L. Thompson and E. K. Wagner, work in
progress), and in this instance, incorporation of the BamHI
L fragment would disrupt at least two HSV-2 mRNAs. This
may not be compatible with successful replication of the
virus.
It should be emphasized that HSV neurovirulence and
neurotropism are almost certainly multigenic functions. The
defect studied here is one that specifically precludes repli-
cation in mouse brains. Localization of this defect to a small
region of the RE6 genome will make it possible to analyze
the transcripts encoded by RE6 in this region and to compare
them with those encoded by HSV-1 and HSV-2. Further
experimentation should then lead to an understanding of the
peptide product and in vivo function of this gene.
This research was supported by grants from the Public Health
Service (AI-06426) and the National Multiple Sclerosis Society
(RG-1647-A-1) to J. Stevens and from the Public Health Service
(CA-1186) to E. Wagner. R. Thompson is a Bank of America-
Giannini Foundation Medical Research Fellow.
LITERATURE CITED
1. Centifanto-Fitzgerald, Y. M., T. Yamaguchi, H. E. Kaufman, M.
Togon, and B. Roizman. 1982. Ocular disease pattern induced by
herpes simplex virus is genetically determined by a specific
region of the viral DNA. J. Exp. Med. 155:475-489.
2. Dix, R. D., R. R. McKendall, and J. R. Baringer. 1983. Com-
parative neurovirulence of herpes simplex virus type 1 strains
after peripheral or intracerebral inoculation of BALB/c mice.
Infect. Immun. 40:103-112.
3. Field, H. J., and P. Wildy. 1978. The pathogenicity ofthymidine
kinase-deficient mutants of herpes simplex virus in mice. J.
Hyg. 81:267-277.
4. Goodpasture, E. W. 1929. Herpetic infection, with especial
reference to involvement of the nervous system. Medicine
7:223-245.
5. Ish-Horowicz, D., and J. F. Burke. 1981. Rapid and efficient
cosmid cloning. Nucleic Acids Res. 9:2989-2992.
6. Kaerner, H. C., C. H. Schroder, A. Ott-Hartnan, G. Kumel, and
H. Kirchner. 1983. Genetic variability of herpes simplex virus:
development of a pathogenic variant during passaging of a
nonpathogenic herpes simplex virus type 1 virus strain in mouse
brain. J. Virol. 46:83-93.
7. Knotts, F. B., M. L. Cook, and J. G. Stevens. 1974. Latent herpes
simplex virus in the central nervous system ofmice and rabbits.
J. Exp. Med. 138:740-744.
8. Roizman, B. 1980. Structural and functional organization of
herpes simplex virus genomes, p. 19-51. In F. Rapp (ed.),
Oncogenic herpesviruses, vol. 1. CRC Press, Inc., Boca Raton,
Fla.
9. Ruyechan, W. T., L. S. Morse, D. M. Knipe, and B. Roizman.
1979. Molecular genetics ofherpes simplex virus. II. Mapping of
the major viral glycoproteins and of the genetic loci specifying
the social behavior of infected cells. J. Virol. 29:677-697.
10. Southern, E. M. 1975. Detection of specific sequences among
DNA fragments separated by gel electrophoresis. J. Mol. Biol.
98:503-517.
11. Stevens, J. G., and M. L. Cook. 1971. Latent herpes simplex
virus in spinal ganglia of mice. Science 173:843-847.
12. Thompson, R. L., and J. G. Stevens. 1983. Biological character-
ization of a herpes simplex virus intertypic recombinant which
is completely and specifically non-neurovirulent. Virology
131:171-179.
13. Thompson, R. L., E. K. Wagner, and J. G. Stevens. 1983.
Physical location of a herpes simplex virus type-1 gene func-
tion(s) specifically associated with a 10-million-fold increase in
HSV neurovirulence. Virology 131:180-192.
J. VIROL.