Recombinant vesicular stomatitis virus targeted to Her2/neu combined with anti-CTLA4 antibody eliminates implanted mammary tumors.

Vesicular stomatitis virus (VSV) is being developed for cancer therapy. We created a recombinant replicating VSV (rrVSV) that preferentially infected Her2/neu expressing breast cancer cells. We now used this rrVSV to treat macroscopic peritoneal tumor implants of a mouse mammary tumor cell line stably transfected to express Her2/neu. rrVSV therapy alone prolonged survival but did not cure any animals. rrVSV therapy combined with antibody to TGFb or antibody to IL-10 receptor (IL-10R) each produced cure in one of six animals. Strikingly, rrVSV therapy combined with anti-CTLA4 monoclonal antibody (MAb) produced cure in four of five animals. Anti-CTLA4 MAb was only effective when administered within one day of rrVSV therapy. Cure required CD4 T-cells early (<7 days) and late (>7 days) after rrVSV therapy whereas CD8 T-cells were required only late (>7 days) after rrVSV therapy. Surviving animals were resistant to re-challenge with D2F2/E2 suggesting a memory immune response. Histopathologic analysis demonstrated a dense inflammatory infiltrate of tumor nodules within days of therapy and foamy histiocytes replacing the tumor nodules 2 weeks following therapy. These studies demonstrate that targeted rrVSV combined with anti-CTLA4 MAb can eliminate established macroscopic tumor implants by eliciting an anti-tumor CD4 and CD8 T-cell immunologic response.

Treatment of meningeal breast cancer xenografts in the rat using an anti-p185/HER2 antibody.

The metastatic spread of breast cancer to the leptomeninges (LM) is a painful, debilitating, and usually lethal condition. Current therapies are generally ineffective or extremely toxic. The current study evaluated monoclonal antibody therapy in an animal model of LM human breast cancer. Monoclonal antibody 4D5, which recognizes the extracellular domain of the HER2/neu receptor, was administered into the cerebrospinal fluid of athymic rats implanted with human breast cancer cell lines. Continuous intraventricular administration of 4D5 inhibited growth of SKBR3 cells that overexpress HER2/neu but not of MCF7 cells, which do not. Inhibition was dose-dependent, with higher doses of 4D5 producing an improved response. i.p. administration of cisplatin in addition to 4D5 did not improve results. Continuous administration of 4D5 into the lumbar, as opposed to the ventricular intrathecal space, was not therapeutically effective. Treatment with 4D5 did not result in outgrowth of cells lacking expression of the HER2/neu receptor. These results suggest that 4D5, administered regionally, may palliate LM metastases from HER2/neu-overexpressing breast carcinoma.

Comparison of in vitro antibody-targeted cytotoxicity using mouse, rat and human effectors.

Antibodies can direct tumor cell lysis by activating complement-mediated and cell-mediated cytoxicities (antibody-dependent cell-mediated cytotoxicity, ADCC). Clinical translation of these effects into successful cancer therapy has been slow. Choosing an appropriate animal model to test new therapeutic strategies is difficult because of species differences in immunological effector functions. In previous work, we found that an unmodified anti-ganglioside mouse IgG3 monoclonal antibody (mAb), 3F8, could successfully treat clinical tumors in humans and experimental tumors in rats but not experimental tumors in mice. We explored the reasons for this species difference by performing in vitro antibody-dependent cytotoxicity assays comparing the potency of polymorphonuclear neutrophils (PMN), natural killer (NK) cells and complement from the three species: mouse, rat and human. 3F8-dependent complement-mediated cytotoxicity produced more than 70% specific release when human and rat sera were used and only 20% with mouse serum. PMN-mediated ADCC was 35%-70% with human effectors, 25%-60% with rat and undetectable with mouse. Human eosinophils did not contribute to this ADCC. Cytotoxicity utilizing interleukin-2-activated NK cells was antibody-independent in all three species but the specific release was 60%-70% with human and rat NK cells and 10% with mouse NK cells. These data suggest that, for mouse IgG3, the rat may provide a more relevant rodent model than the mouse for testing the in vivo antitumor effects of monoclonal antibodies.

Treatment of neoplastic meningeal xenografts by intraventricular administration of an antiganglioside monoclonal antibody, 3F8.

Leptomeningeal (LM) neoplastic metastases are painful, debilitating and inevitably lethal. Intrathecal (IT) anti-tumor antibodies may have therapeutic potential. We evaluated 3F8, an anti-G(D2) murine IgG(3) monoclonal antibody (MAb) in the treatment of human melanoma (SKMEL-1) and neuroblastoma (NMB7) xenografts in athymic rats. Both tumors were lysed efficiently in vitro by 3F8 in the presence of rat neutrophils or rat complement. Antibody-dependent cellular cytotoxicity (ADCC) was not augmented by recombinant human GM-CSF (rhGM-CSF), rhG-CSF, recombinant rat MIP-2 (rrMIP-2) or lipopolysaccharide (LPS). In vivo, continuous intraventricular administration of 3F8 and LPS prevented tumor engraftment, retarded tumor growth and eradicated 3-day-old established xenografts whereas 3F8 alone, LPS alone or F(ab)'(2) plus LPS had no or only marginal effects. Tumor establishment in brain was completely prevented in 36% of animals implanted with SKMEL-1 and 65% of animals implanted with NMB7. Twenty percent of established xenografts around the brain were eradicated but all animals had persistent tumor in the lumbosacral meninges despite treatment. Continuous intraventricular infusion of LPS produced a variable polymorphonuclear (PMN) pleocytosis that was dose-dependent. Continuous intraventricular infusion of 3F8 produced immunohistochemically detectable attachment to 86% of persistent brain deposits of tumor but <1% of spinal lumbosacral deposits. We conclude that regional therapy with anti-G(D2) MAb could target neutrophils to inhibit LM tumor growth. However, optimal activation and mobilization of neutrophils into the cerebrospinal fluid (CSF) and improved penetration of MAb to tumor sites remain critical variables.

Intrathecal administration of an anti-ganglioside antibody results in specific accumulation within meningeal neoplastic xenografts in nude rats.

Intrathecal (i.t.) administration of monoclonal antibodies (MAbs) represents a new therapeutic approach for the treatment of leptomeningeal (LM) cancer, which is presently rapidly fatal. In this study, we quantitated the accumulation of an intrathecally administered anti-ganglioside GD2 MAb (3F8) within leptomeningeal neoplastic xenografts of GD2 positive melanoma and neuroblastoma in nude rats by measuring concentrations of radiolabeled and unmodified MAbs and by immunohistochemistry. Intrathecal administration of 125I-3F8 resulted in area under the tissue concentration versus time curve (AUC) values in SK-MEL-1 melanoma xenografts (53.1 microCi*h/g) that were 14-fold greater than in corresponding blood (3.9 microCi*h/g), whereas i.t. administration of a control nonspecific MAb resulted in AUC values in tumors (7.1 microCi*h/g) that were less than those in blood (9.5 microCi*h/g). Administration of acetazolamide and furosemide, which slow the clearance of IgG MAb from rat cerebrospinal fluid resulted in a fivefold increase in AUC of 125I-3F8 in melanoma (262.9 microCi*h/g). The highest concentration of 125I-MAb in tumor after i.t. administration was seen at the first sampling time of 2 h, and this fell to 50% of maximum values at 8-16 h. Pharmacokinetic analysis of unmodified MAb demonstrated retention of MAb within the LM space of animals with tumor. The concentration of MAb 3F8 appearing in serum after i.t. administration was 10-fold lower in animals with melanoma xenografts than in those without tumor implants. Radiation dose estimates after intraventricular administration of radiolabeled MAb indicated delivery to tumor of 1,870 rad/mCi of 125I-3F8 but only 40 rad/mCi of 125I-labeled control MAb. These results indicate that anti-ganglioside MAbs and other MAbs directed to tumor-associated antigens are excellent candidates for i.t. treatment of appropriate leptomeningeal cancers in humans.

Pharmacokinetics of IgG and IgM anti-ganglioside antibodies in rats and monkeys after intrathecal administration.

Intrathecal (i.t.) administration of monoclonal antibodies (mAbs) represents a new therapeutic approach for the treatment of leptomeningeal cancer, which is rapidly fatal. This study describes the pharmacokinetics of intrathecally administered mAbs in rats and monkeys to optimize their use for regional antineoplastic therapy. We hypothesized that mAbs, which are high-molecular-weight, polar compounds, would be eliminated from the cerebrospinal fluid (CSF) at the same rate as bulk flow of CSF. We found that an IgM mAb was cleared from rat CSF at the rate of CSF bulk flow (0.0041 ml/min), but an IgG mAb was cleared at a faster rate (0.011 ml/min). We attempted to reduce the CSF clearance of an IgG mAb by administration of acetazolamide and furosemide, which inhibit the rate of CSF production and CSF bulk flow. We demonstrated that the administration of acetazolamide and furosemide reduced the clearance of IgG mAb from rat CSF by 58%. These results establish that bulk flow of CSF determines a minimum rate of elimination from the CSF for IgM mAbs and that additional mechanisms operate to clear IgG mAbs from the CSF. Inhibition of CSF production by acetazolamide and furosemide increased the area under the CSF concentration vs. time curve of IgG mAbs in the CSF. The increased area under the CSF concentration vs. time curve is likely to improve the therapeutic index of these agents for i.t. therapy.

Leigh-type neuropathology in Pearson syndrome associated with impaired ATP production and a novel mtDNA deletion.

Pearson syndrome is a systemic disorder of oxidative phosphorylation in infants, predominantly affecting the bone marrow and exocrine pancreas and associated with single deletions in mitochondrial DNA (mtDNA). CNS involvement may occur in patients who survive the infantile hematopoietic disorder. We describe a Pearson syndrome patient who developed neurologic manifestations associated with the pathologic features of Leigh syndrome. Biochemical studies in muscle and skin fibroblasts showed partial deficiencies of complexes I and IV of the respiratory chain. Adenosine triphosphate production in mitochondria isolated from skin fibroblasts was reduced to 25% of controls. We detected a novel 3.6 Kb mtDNA deletion in skin fibroblasts from the proband but not in his mother's white blood cells. Leigh syndrome seems to be the common neuropathologic expression of any disorder causing severe impairment of oxidative energy production in the CNS.

Clinical examination findings in neonates with the absence of electrocerebral activity: an acute or chronic encephalopathic state?

Although the presence of an isoelectric electroencephalogram (EEG) in an older patient may reflect brain death caused by an acute brain injury, this electrographic abnormality may appear in more diverse clinical situations in the neonate with encephalopathy. During a 6-year period, 20 neonates were identified with a severe encephalopathy on neurologic examination who had at least one isoelectric EEG during their treatment in a neonatal intensive care unit. Seventy-four EEG recordings were obtained including 36 isoelectric EEG records. Partially preserved clinical brain function was present in 15 (75%) of 20 infants at the time an isoelectric EEG was obtained. The initial EEG was isoelectric in 16 of 20 infants. Although electrographic activity reemerged in nine of these infants, significant clinical improvement was seen in only two patients. Thirteen of 20 neonates also had electrographic or other evidence of clinical seizures. Of the five survivors (25%), three had severe neurologic sequelae. The remaining two had either transient or persistent neurologic deficits. An isoelectric EEG may be obtained in the neonate with partially preserved brain function and, therefore, may not be a reliable confirmatory test of neonatal brain death. In addition, serial EEGs not only can help assess the severity of a neonatal encephalopathy but also may correlate with chronic and acute neurologic insults.

Effects of neutropenia on edema, histology, and cerebral blood flow after traumatic brain injury in rats.

Neutrophils accumulate during the acute inflammatory response to brain injury, but their role in the injury process remains controversial. We tested the hypothesis that neutrophils contribute to cerebral edema, tissue injury, and disturbed cerebral blood flow (CBF) (hyperemia or ischemia) during the first 24 h after traumatic brain injury. Wistar rats (n = 51) were injected with either vinblastine sulfate to induce neutropenia or the saline vehicle. Five days later, under halothane anesthesia, right hemispheric trauma was produced by weight drop (10 g x 5 cm) onto exposed dura. At 24 h after trauma, brain water (wet-dry weight), traumatic infarct size (percent of hemispheric section infarcted), or local CBF (lCBF, 14C-iodoantipyrine autoradiography) was assessed. Vinblastine treatment produced profound neutropenia on the day of trauma (absolute neutrophil count 0.024 +/- 0.008 x 10(9)/L vs 1.471 +/- 0.322 x 10(9)/L, p < 0.05 in neutropenic vs saline, respectively, mean +/- SEM). Neutropenia did not reduce the development of brain edema in the injured hemisphere (brain water 82.38 +/- 0.29% vs 82.73 +/- 0.37% in neutropenic and saline, respectively, mean +/- SEM) or traumatic infarct size (34.5 +/- 3.3% vs 33.2 +/- 2.1% in neutropenic vs saline respectively). In contrast, neutropenic rats exhibited 52%, 41%, and 57% reductions in lCBF in the frontal cortex, parietal cortex, and amygdala, respectively, of the injured hemisphere 24 h after trauma (all p < 0.05 vs nonneutropenic controls). These data suggest that neutrophils and the acute inflammatory process contribute to the level of CBF observed 24 h after trauma, but effects on edema or early posttraumatic infarct size could not be demonstrated.

Early cerebrovascular response to head injury in immature and mature rats.

Clinical studies suggest that children respond to head injury with more pronounced cerebral edema and hyperemia than do adults. We hypothesized that these age-related differences could be demonstrated in an animal model. Anesthetized and ventilated mature (2-3 months) and immature (3.5-4.5 weeks) male Wistar rats were traumatized by weight drop onto the exposed right parietal cortex. Trauma severity was adjusted to keep the ratio of force to brain weight constant. This resulted in an energy delivered to the brain of about 9 x 10(3) ergs.mm-2.g-1 brain in both age groups. Percent right hemispheric brain water (%RBW) was measured at 2, 24, 48, and 168 h posttrauma. Infarct area, intracranial pressure (ICP), and 14C-iodoantipyrine autoradiographic local cerebral blood flow (ICBF) were measured at 2 h or 24 h posttrauma. In mature rats, %RBW was unchanged at 2 h, but increased at 24 and 48 h (both p < 0.05). In immature rats, %RBW increased at 2 h and remained elevated at 24 and 48 h (all p < 0.05). Traumatic infarct area as a percent of hemispheric area at 24 h did not differ between age groups. In mature rats, at 2 h posttrauma ICBF was reduced (p < 0.05) in 16 of 17 regions but in only 4 of 17 regions in immature rats. ICBF as a percent of age-matched control values showed a greater reduction in mature vs immature rats in 9 of 16 regions (p < 0.05). ICP increased at 24 h posttrauma in both age groups. In immature rats posttrauma, brain water increased earlier and cerebral hypoperfusion was less marked than in mature rats.

Timing of brain insults in severe neonatal encephalopathies with isoelectric EEG.

In a neonatal intensive care unit of a large obstetric hospital, 20 neonates (7 preterm, 9 term, 4 postterm) with at least one isoelectric recording were treated over a 6-year period. Seventy-four EEGs were obtained in this cohort, including 36 isoelectric recordings. Seven infants in this group had evidence of a predominant antepartum component of a pathologic process based on placental, postmortem examination findings, or clinical history. Of the 16 placentas available for review, chronic lesions were observed in 13 of 16 specimens, including villitis, infarction, dysmaturity, and thrombosis. Seven of 9 patients with postmortem neuropathologic examinations had evidence of chronic lesions, principally neuronal necrosis, infarction, and microcalcifications. An additional 10 infants had evidence of an antepartum contribution to a pathologic process that continued into either the intrapartum or neonatal periods, based on maternal and/or neonatal medical factors. Clinical findings supportive of antepartum insults included intrauterine growth retardation, antepartum hemorrhage, abnormal antepartum fetal heart rate patterns, and maternal medical complications. Three patients had either intrapartum- or neonatal-onset of injury. Clinical signs of severe encephalopathy, however, were present in the immediate postnatal period in most patients (18 of 20; 90%). Assessment of clinical and pathologic information on neonates with isoelectric EEGs may estimate the timing of brain injury to the antepartum period, as opposed to, or in addition to, the labor and delivery periods. A neonate who suffered brain injury before parturition may be neurologically depressed after birth with absence of electrocerebral activity on EEG.

X inactivation and dystrophin studies in a t(X;12) female: evidence for biochemical normalization in Duchenne muscular dystrophy carriers.

A 4-year-old girl was identified with high creatine kinase (CK) values, and mild muscle weakness in a limb-girdle distribution. Results of dystrophin analysis of the muscle biopsy were consistent with a manifesting heterozygote for Duchenne muscular dystrophy. In peripheral lymphocytes she had a t(X;12) (p21.2;q24.33). Late DNA replication studies demonstrated inactivation of the normal X chromosome in 99.4% of cells. Dystrophin immunofluorescence showed 64% dystrophin-negative muscle fibers. Dystrophin content of muscle by immunoblot was approximately 5% of normal. The discordance between the percent of normal X inactivation and percent of dystrophin-negative cells may be explained by compensatory protection of dystrophin by rare nuclei with the normal X active in multi-nucleated muscle fibers with shared cytoplasm.

Cerebral infarcts and seizures in the neonate.

A retrospective analysis was performed on 54 infants who suffered perinatal hypoxic-ischemic insults and came to autopsy, in an attempt to assess the association of cerebral infarcts with seizures. Fifty infants had several types of cerebrovascular lesions, including intraventricular hemorrhage (32 cases), periventricular leukomalacia (24), ischemic neuronal necrosis (18), pontosubicular necrosis (12), cerebral infarct (9), and cerebellar hemorrhage (7). Of these infants, nine had electroencephalographic seizures. Among a variety of cerebrovascular lesions, cerebral infarcts represented the single lesion most highly correlated with seizures. The incidence of seizures in infants with cerebral infarcts (44%) was significantly higher than with other types of vascular lesions.

Neonatal electroencephalography and neuropathology.

One-hundred-eight EEGs from 47 newborn infants were compared with the postmortem neuropathological findings. The degree of EEG background abnormality had good correlation with the severity of the brain lesion; the more severe the EEG background abnormality, the more extensive and intensive the morphological change. Widespread encephalomalacia was demonstrated in six infants who manifested isoelectric tracings. In particular, cerebral cortex, corpus striatum, thalamus, midbrain, and pons were affected in all patients with this abnormal EEG pattern. Burst-suppression patterns, which were seen in seven infants, also correlated with multifocal severe brain damage, but there was no common structure that was consistently affected for all patients with this pattern. Positive rolandic sharp-wave transients (PRS) appeared highly specific for white matter lesions. All eight infants with PRS had white matter lesions. However, the sensitivity of PRS for white matter lesions was not high (32%), and the white matter lesions of PRS-positive patients were not necessarily composed of periventricular leukomalacia. The sensitivity of EEG asymmetry was also low (40%) for the focality of morphological change, although the specificity was relatively high (85%). The origin of seizure discharges, on the other hand, had poor correlation with the site of the brain lesion.

EEG diagnoses of neonatal seizures: clinical correlations and outcome.

Electroencephalographic seizures were evaluated in 112 neonates. The first portion of the study involved 80 neonates with clinically identified abnormal movements, 8 of whom (10%) had electroencephalographic evidence of seizures coincident with this activity. Patients with abnormal movements (90%) had no concurrent electrical seizures. In the second part of the study, 40 infants who had electrical seizures were investigated. Eight of these infants had been identified during the first part of the study. Two-thirds of the patients (25) were premature. Sixteen patients (40%) died; 90% had brain lesions documented by computed tomography and/or postmortem study. Cerebral infarction and intraventricular hemorrhage were the most common lesions. One-third of the survivors (8 of 24 patients) were normal at a mean age of 3 years, while two-thirds had significant neurologic and developmental abnormalities. Neonatal seizures often are subtle, not associated with observable clinical expression, and associated with adverse development. Electroencephalographic confirmation is important in the evaluation of neonatal seizures.

Estimation of gestational age by electrographic, clinical, and anatomic criteria.

Current methods for estimating gestational age utilizing clinical parameters can be inaccurate in the very premature neonate. Neuropathologic studies have been used to verify these clinical estimates of gestational age. Estimates of maturity using neonatal electroencephalography have documented the evolution of regional and hemispheric electrical patterns at different gestational ages; however, electroencephalographic studies have not been compared with both anatomic brain studies and clinical examination criteria. Therefore, we studied 25 neonates with at least one electroencephalographic study prior to death and subsequent neuropathologic examination. Twenty-three estimates (92%) of gestational age in weeks, which were based on specific electrical patterns, agreed with sulcal-gyral development (e.g., the delineation of inferior frontal, superior temporal, calcarine sulci, and cytoarchitecture in various brain regions). Clinical (Ballard score) criteria and anatomic estimations of gestational age agreed in 18 of 24 patients (74%). Electroencephalographic estimates were significantly better than clinical estimates (p less than .05) for sick neonates who were less than or equal to 30 weeks gestational age.

Congenital alopecia, seizures, and psychomotor retardation in three siblings.

Three siblings, devoid of hair at birth, had an unusual autosomal recessive disorder characterized by universal congenital alopecia, microcephaly, seizures, psychomotor retardation, and severe growth failure. Metabolic and chromosome studies were normal. Skin biopsies disclosed immature hair follicles, some of which were filled with keratotic material but had no hair shafts. Neuropathologic features included cerebral cortical hypoplasia, neuronal depletion, and microcalcifications. The familial occurrence of universal congenital alopecia conjoined with nonprogressive central nervous system abnormalities in this and other kindreds defines a nosologic group of neurocutaneous disorders in which congenital alopecia is the solitary cutaneous manifestation.

Intracerebral hemorrhage in the full-term neonatal infant.

Supratentorial intracerebral hemorrhage was diagnosed in 18 full-term neonates, including six with primarily intraparenchymal hemorrhage and 12 with primarily intraventricular hemorrhage. Precipitating or associated factors were hypoxic-ischemic injury in five patients, polycythemia in two, and cranial birth trauma in two. Nine other infants had no identifiable medical risk events. The pathogenesis of intraparenchymal hemorrhage was probably related to hemorrhagic infarction, but the pathogenesis of intraventricular hemorrhage was often unknown. All 17 survivors returned for neurologic and developmental examinations between 1 and 7 years of age. Follow-up assessments were normal in nine children and abnormal in eight. Two had perceptual difficulties, three had moderate-to-severe cognitive deficiencies (two of the three had hemiplegia), and three had severe mental and neurologic handicaps. Eight of nine children with known or suspected hypoxic-ischemic or traumatic insults suffered moderate-to-severe disabilities whereas eight of nine children with no known precipitating cause for their hemorrhage developed normally.

Acute and chronic effects of intrathecal morphine in monkeys.

In a double-blind study, seven Macaca fascicularis monkeys receiving intrathecal (i.t.) morphine in saline, 0.07 mg kg-1, were compared with a control group of four monkeys receiving either lumbar puncture alone (n = 1) or i.t. saline (n = 3). Neither morphine nor saline solutions contained preservatives. Arterial blood gas tensions, respiration, arterial pressure, e.c.g., state of consciousness and motor function were recorded for 24 h. The control group was sacrificed 42 days later and the study group was sacrificed at 6 (n = 2) or 42 days (n = 5) after injection. The central nervous system, meninges, nerve roots and dorsal root ganglia were examined macroscopically and microscopically. Respiratory depression did not occur in either the control or the study groups. There were moderate but statistically significant decreases in systolic and diastolic arterial pressures following i.t. morphine. In both groups, the pathological findings were localized to the cauda equina region and characterized by mononuclear cell infiltration. In neither group was there evidence of demyelination, arachnoiditis or necrosis. Focal endoneurial fibrosis was found in only one animal in the control group following multiple lumbar punctures associated with paraesthesia. The features appeared to correlate with the physical trauma associated with lumbar puncture rather than with the injectate.

Neurovirulence mutant of vesicular stomatitis virus with an altered target cell tropism in vivo.

Intracerebral infection of weanling Swiss mice with a temperature-sensitive (ts) mutant of vesicular stomatitis virus (VSV), ts pi364, resulted in a unique neuropathological syndrome not previously described with other VSV mutants. Mice infected with wild-type VSV died from an acute encephalitis characterized by neuronal necrosis and efficient virus replication in both brain and spinal cord. In contrast, with VSV ts pi364, the most prominent histopathological feature was destruction of the ependyma of the lateral ventricles. Virus antigen was also limited to the leptomeninges and the lateral ventricles. Infected mice survived and developed hydrocephalus. Replication of ts pi364 in the brain was 10- to 100- fold less than that of wild-type VSV, and appearance of virus in the spinal cord was delayed. VSV ts pi364 was isolated from mouse cells persistently infected with VSV. Another VSV ts pi mutant, isolated from the same persistent infection, behaved in vivo like wild-type VSV, even though both mutants were very similar in plaque size, reversion frequency, cut-off temperature, and synthesis of virus-specific proteins at semipermissive temperature. These results strongly suggest that VSV ts pi364 has a second, non-ts mutation which results in a restricted target cell range in vivo; wild-type VSV can infect both neurons and ependymal cells, whereas ts pi364 does not replicate in neurons.