Frequent Infection of Cortical Neurons by JC Virus in patients with Progressive Multifocal Leukoencephalopathy

Abstract

The human polyomavirus JC (JCV) infects glial cells and causes progressive multifocal leukoencephalopathy (PML), a demyelinating disease of the brain, in immunosuppressed individuals. The extent of JCV infection of neurons is unclear. We determined the prevalence and pattern of JCV infection in grey matter (GM) by immunostaining in archival brain samples of 49 PML patients and 109 control subjects. Among PML patients, 96% had demyelinating lesions in white matter and at the grey-white junction (GWJ); 57% had them in the GM. Most JCV-infected cells in GWJ and GM were glia but JCV infected neurons in PML lesions at the GWJ of 54% and GM of 50% patients, and in GM outside of areas of demyelination in 11% of patients. JCV regulatory T antigen (Ag) was expressed more frequently in cortical neurons than the VP1 capsid protein. None of the control subjects without PML had any cells expressing JCV proteins. Thus, the cerebral cortex often harbors demyelinating lesions of PML and JCV infection of cortical neurons is frequent in PML patients. The predominance of T Ag over VP1 expression suggests a restrictive infection in neurons. These results indicate that JCV infection of cerebral cortical neurons is a previously under-appreciated component of PML pathogenesis.

INTRODUCTION

JC virus (JCV) is a ubiquitous human polyomavirus that infects 50% to 86% of healthy adults without causing any disease (1–3). The virus remains quiescent in the kidney and lymphoid organs and may also remain latent in the brain (4). In immunosuppressed individuals, including those with HIV-AIDS, hematological malignancies, and transplant recipients, and those with autoimmune diseases treated with immunomodulatory medications, JCV may reactivate and causes a productive and lytic infection of oligodendrocytes and astrocytes, leading to an often the often fatal demyelinating disease of the CNS progressive multifocal leukoencephalopathy (PML) (5–8).

PML is characterized by multifocal areas of demyelination containing JCV-infected oligodendrocytes, as well as reactive gliosis with enlarged, bizarre astrocytes, some of which are infected by JCV. PML lesions are localized preferentially in the subcortical white matter (WM) of the cerebrum, but may also be found in the central WM, corpus callosum, and the cerebellar peduncles and hemispheres. However, PML lesions are also found on MRI within grey matter (GM) structures including cerebral cortex, basal ganglia, thalamus and brainstem (9). We and others have appreciated that demyelinating lesions of PML can also be located within the cerebral cortex (10–13), but the nature of JCV-infected cells within GM lesions has not been thoroughly investigated.

JCV has a circular DNA genome that includes early genes coding for the regulatory small t and large T antigens (Ag). T Ag is instrumental in the initiation of viral replication and transcription of the late genes, which include the capsid proteins VP1, VP2 and VP3, and the agnoprotein. The viral capsid is made up of 72 pentamers of VP1, each associated with either VP2 or VP3. Assembly of mature viral particles, which do not contain the regulatory t and T Ag, occurs in the nuclei of infected cells. Thus, detection of T Ag by immunohistochemistry (IHC) in the absence of VP1 suggests a restricted or early infection. Conversely, the presence of the VP1 protein indicates that JCV has undergone a full replicative cycle and has formed mature viral particles, which can then be detected by electron microscopy (14).

Although JCV was long thought to only infect glial cells in the brain, we have described 2 conditions caused by infection of neurons. JCV granule cell neuronopathy is caused by a JCV variant harboring a small deletion in the VP1 capsid protein, with specific tropism for cerebellar granule cell neurons. This infection results in cerebellar atrophy and associated neurologic dysfunction (15–19). JCV encephalopathy is caused by a productive infection of cortical pyramidal neurons and was described in an HIV-negative patient with lung cancer who presented with lesions restricted to the hemispheric GM (20). To determine the prevalence and pattern of JCV infection of the cerebral cortex and the phenotype of infected cells, we studied JCV expression of T Ag and VP1 protein in cerebral samples from a large population of PML patients, including HIV-positive and HIV-negative subjects. Infection of cortical neurons was frequent in HIV-positive/PML patients and the pattern of JCV protein expression in these cells was different from that in glial cells within PML cerebral WM lesions.

MATERIALS AND METHODS

Brain Samples

Formalin-fixed, paraffin-embedded archival postmortem brain tissue samples from 49 patients with histologically confirmed PML were studied. These included 36 (73%) HIV- positive and 13 (27%) HIV-negative patients. None of the HIV-positive PML patients had immune reconstitution inflammatory syndrome. Thirty-one PML patients died within 1 year from PML onset; 7 patients died more than 1 year (range: 1–8 years) after PML onset. The clinical duration was not available for 11 patients. These materials were collected from the Departments of Pathology of the Beth Israel Deaconess Medical Center (Boston, MA), the Miriam Hospital (Providence RI), and from the National Neuro AIDS Tissue Consortium (21) from 1973 to 2010. We studied a total of 159 blocks from postmortem brain samples that had JCV-positive cells (of any type) and/or PML lesions in the subcortical areas of the cerebrum (77% of blocks) or in WM adjacent to GM areas of the brainstem, basal nuclei or the hippocampus (23% of blocks). Cerebellar samples were not included because we had previously reported infection of cerebellar granule cell neurons (19).

The control group included 187 paraffin blocks of postmortem brain samples from 109 subjects without PML (24 HIV-positiveand 85 HIV-negative). Within the HIV-negative group there were 19 patients without CNS abnormalities and 66 patients with either multiple sclerosis (n = 29), amyotrophic lateral sclerosis (n = 12), brain tumors (n = 5), stroke (n = 5), Alzheimer disease (n = 4), demyelination or WM degeneration (n = 3), dementia (n = 2) and one each of Epstein bar virus, myopathy, tauopathy, seizures, focal cerebellar sclerosis and trisomy 21. Ten samples from a patient with JCV encephalopathy (20) were used as a reference for the cortical neurons immunostaining experiments.

IHC and Immunofluorescence Staining

The antibodies used are listed in the Table. Single and double IHC and immunofluorescence (IF) staining assays were performed as previously described (19, 20). In multiple stainings, primary antibodies from different species (mouse monoclonals, rabbit, chicken and sheep polyclonals) were used with appropriately matched (species and isotype) secondary antibodies. For IHC, these antibodies were conjugated either to alkaline phosphatase or horseradish peroxidase or to Alexa Fluor 350, 488 & 568 (Invitrogen, Carlsbad, CA), according to the manufacturer’s instructions. Most single and multiple IHC were performed with the Mouse IgG and/or Rabbit IgG Vectastain Elite ABC Kits (Vector Laboratories, Burlingame, CA), according to the manufacturer’s instructions. Other kits and chromogens were also used (19, 20). Negative controls for IHC and IF included omission of the primary antibodies and the use of isotype-matched controls and sections from individuals known to be free of JCV-infected cells.

Table Antibodies

Antigen; product name	Host	Isotype	Target	Source
PAB597	m	NA	JCV VP1	Walter Atwood, Brown University, Providence, RI
SV40 VP1	r	IgG	JCV VP1	Lee Biomolecular Res., San Diego, CA
SV40 T Ag (v-300); sc-20800	r	IgG	JCV T Ag	Santa Cruz Biotechnology, Santa Cruz, CA
MAP-2; HM-2; M4403	m	IgG1	Neurons	Sigma, St-Louis, MO
MAP-2; ab5622	r	IgG	Neurons	Chemicon, Temecula, CA
MAP-2; ab5392	ch	IgY	Neurons	Abcam, Cambridge, MA
MAP-2; ab92434	ch	IgY	Neurons	Abcam
MAP-2; LS-B290	ch	IgY	Neurons	Lifespan Bioscience, Seattle, WA
NeuN; MAB377; A60	m	IgG1	Neurons	Chemicon
Neuron-specific βIII Tubulin; 2G10; ab78078	m	IgG1	Neurons	Abcam
160 kD Neurofilament Medium antibody; ab39371	ch	IgY	Neurons	Abcam
GFAP; M0761; 6F2	m	IgG1	Astrocytes	Dako, Carpinteria, CA
GFAP; Z0334	r	IgG	Astrocytes	Dako
GFAP; ab90601	sh	IgG	Astrocytes	Abcam
GFAP; ab4674	ch	IgY	Astrocytes	Abcam
CNPase; C5922; 11-5B	m	IgG1	Myelin/Oligo	Sigma,
CNP; HPA023266	r	IgG	Myelin/Oligo	Sigma
MBP; SMI-94	m	IgG1	Myelin/Oligo	Covance Research, Denver, PA
MOSP; MAB328; CE-1	m	IgM	Myelin/Oligo	Chemicon
Myelin PLP; ab9311; plpc-1	m	IgG2a	Myelin/Oligo	Chemicon
Myelin PLP; ab28486	r	IgG	Myelin/Oligo	Abcam

Abbreviations: m: mouse; ch; chicken; r: rabbit; sh: sheep; oligo: oligodendrocytes; MAP-2, microtubule-associated protein-2; GFAP, glial fibrillary acidic protein; CNPase, CNP, 2′,3′-Cyclic-nucleotide 3′-phosphodiesterase; MBP, myelin basic protein; MOSP, myelin/oligodendrocyte-specific protein; PLP, proteolipid protein.

Characterization of PML Lesions

Double IHC for 2′, 3′-cyclic-nucleotide 3′-phosphodiesterase (CNPase) and JCV VP1 (PAB597) were performed to estimate the extent and pattern of the demyelination in the WM, grey-white junction (GWJ) and GM of the whole section, and to correlate them with the location of JCV-infected cells. When CNPase antigenicity was lost due to destruction of the CNPase epitope after long-term formalin preservation, sections were stained with hematoxylin and Luxol fast blue.

Phenotype of the JCV-Infected Cells and Neurons and Cell Counting

The presence of JCV-infected cells of any type (glial cells and neurons) was determined by single staining (IF and/or IHC) with rabbit anti-T Ag and mouse anti-VP1 (PAB597) antibodies. The presence of JCV-infected neurons was investigated with double staining (IF and/or IHC) for rabbit anti-T Ag or mouse anti-VP1 (PAB597) and various neuronal markers (Table).

The frequencies of JCV-infected cells (any type) or neurons were estimated according to a semi-quantitative scale, (0 = none; 1 = 1–10 cells; 2 = 11 cells-1% cells; 3 = 2–5% cells; 4 = 6–25% cells; 5 = 26–100% cells), similar to a previous study (19). The presence of glial cells (astrocytes and oligodendrocytes) was also investigated in double stainings for JCV proteins and the various glial cell markers (Table).

Statistical Analysis

A principal component analysis (PCA) (22) was performed on all the measurements (cell frequencies, demyelination and patient clinical information) to reveal the internal structure of these data and determine if there were particular associations. The exact (permutation) Wilcoxon sign rank test was used to compare paired values and the exact (permutation) Wilcoxon rank sum test was used to compare 2 unmatched groups (e.g. results in HIV-positiveand HIV-negative patients) (23). Fisher exact test (24) was used to compare the proportion of patients with PML lesions or JCV-infected cells in various locations.

RESULTS

Topographic Localization of JCV-Infected Cells and PML Lesions in Subcortical and Cortical Areas of the Cerebrum

We first studied the topography of JCV-infected cells and PML lesions in subcortical and cortical areas of the cerebrum. We defined the topographic location of demyelinating lesions in the WM, GWJ and GM and confirmed the presence of JCV-infected cells within the lesions by IHC against JCV proteins and Luxol fast blue counterstaining (Fig. 1A). We then performed triple immunostaining experiments against oligodendrocyte (brown), astrocytes (blue) and JCV-infected cells (red), on adjacent sections (Fig. 1B). Triple immunostaining illustrated the extent of demyelinating lesions in subcortical and cortical areas and demonstrated that oligodendrocytes are abundant within the deep layers of the cortex whereas astrocytes are present in all the layers of the GM (Fig. 1B–D). Both astrocytes and oligodendrocytes were infected by JCV outside of the WM (Fig. 1C, E). Some patients had widespread lesions extending throughout the subcortical and cortical areas that contained many infected oligodendrocytes (Fig. 1D, F). Some demyelinating lesions were circumscribed within the cerebral GM without evidence of demyelination in the nearby WM (Fig. 2A) and caused extensive destruction of the cortex (Fig. 2B). No JCV-infected cells were found in the WM, subcortical or cortical areas in the controls.

Figure 1.

Demyelinating lesions and JCV-infected cells in the cerebral white matter (WM), grey white junction (GWJ) and grey matter (GM) of 3 PML patients. (A) Immunohistochemistry (IHC) for JCV VP1 (PAB597, dark blue) displayed in a 36,000,000 pixels picture of cerebrum of an HIV-positive PML patient shows that infected cells are present in the WM (inset 1), GWJ (inset 2), and GM (inset 3). H&E and Luxol fast blue staining. (B) Triple IHC for JCV-infected cells (SV40 VP1 rabbit antiserum, red), 2′, 3′-cyclic nucleotide 3′-phosphodiesterase (CNPase) for oligodendrocytes and myelin (C5922, brown) and glial fibrillary acidic protein (GFAP) for astrocytes (Z0334, blue) performed on an adjacent section reveals that demyelinated areas extend from the subcortical to the cortical areas. (C-F) The same triple immunostaining (JCV VP1 red, oligodendrocytes and myelin brown, astrocytes blue) performed on 2 other patients shows lesions scattered in subcortical WM and GWJ (C). The boxed area, in the GWJ at the lower right of this panel is magnified in (E) and reveals that infected cells include astrocytes (arrows) and oligodendrocytes (arrowhead) (D). There is extensive demyelination in the WM, GWJ and GM. The boxed area located in GM at the upper right of the panel is magnified in (F), shows infected oligodendrocytes (arrowhead). Scale bars: 1 mm.

Figure 2.

Demyelinating lesions of progressive multifocal leukoencephalopathy (PML) may be circumscribed within the grey matter (GM). (A) A single round lesion is present in GM of an HIV-negative PML patient (arrows) and contains JCV-infected cells. Double immunohistochemistry (IHC) for JCV T antigen (Ag) (v-300, dark blue) and MAP-2 (M-4403, brown). The region corresponding to the panel is highlighted at lower magnification by the square box in the inset, demonstrating the absence of demyelination in the nearby white matter (WM) (H&E/Luxol fast blue). (B) An extensive coalescent area of tissue destruction restricted to the GM (arrow) is located at a distance from an affected area in GWJ. Double IHC for JCV T Ag (v-300, red) and MAP-2 (M-4403, brown). The boxed area in the inset shows JCV-infected cells (arrows). Scale bars: 1 mm.

Neurons within GM Lesions of PML Can be Infected by JCV

Immunostaining for JCV proteins and neuronal markers demonstrated that some cortical neurons characterized by expression of the microtubule-associated protein 2 (MAP-2) (Fig. 3A, C, D) or NeuN (Fig. 3B) expressed JCV T Ag in their nuclei (Fig. 3A, B, D) or JCV VP1 (Fig. 3C). These cells were interspersed among uninfected neurons and in some cases JCV-infected glial cells (Fig. 3B, C). JCV-infected neurons were found in both HIV-positive (Fig. 3B, D) and HIV-negative (Fig. 3A, C) PML patients.

Figure 3.

Expression of JCV T antigen (Ag) and VP1 capsid protein in cortical neurons of PML patients. (A) Double immunofluorescence (IF) with an anti-T-Ag antibody (v-300, Alexa 568, red) and the neuronal marker anti-MAP-2 (M4403, Alexa 488, green) in an HIV-negative PML patient with malignant thymoma shows one cell in the gray-white junction (GWJ) expressing T Ag (box). Higher magnification views of cells in this box shown in the lower inset, and viewed separately with the green (middle inset) and red channel (upper inset) indicates that this neuron, recognizable by its cytoplasmic staining with MAP-2, expresses T Ag in its nucleus (MAP-2-positive/T Ag-positive ^cell, arrow). An adjacent uninfected neuron is marked by an arrowhead. (B) T Ag-expressing neurons in an HIV-positive PML patient using double IF with anti-T Ag (v-300, Alexa 568, red) and neuronal marker anti-NeuN (MAB377, Alexa 488, green). One NeuN-positive/T Ag-positive neuron in the box shown in the lower inset, and viewed separately with the green (middle inset) and red channel (upper inset) indicates that this neuron express T Ag in its nucleus (NeuN-positive/T Ag-positive cell, arrow). An adjacent uninfected neuron is marked by an arrowhead and an infected glial cell by an asterisk. (C) An HIV-negative PML patient with chronic lymphocytic leukemia has productively JCV-infected cells in grey matter (GM) shown by double IF with anti JCV VP1 (PAB597, Alexa 488, green) and anti-MAP-2 (ab5622, Alexa 568, red); one of them is a neuron expressing VP1 protein (VP1-positive/MAP-2-positive, arrow) seen in the lower inset, viewed separately with the green (upper inset) and red channel (middle inset). (D) A large neuron stained by double immunohistochemistry with anti-MAP-2 antibody (ab5392, VIP, purple) expresses JCV T Ag in its nucleus (v-300, brown) in an HIV-positive PML patient (arrow). An adjacent uninfected neuron is marked by an arrowhead. Scale bars: AC, 100 μm; D, 25 μm.

Isolated Cortical Neurons May Sustain JCV Infection

While most JCV infected glial cells and neurons were located within demyelinating lesions of PML, isolated neurons expressing JCV proteins were also observed in otherwise healthy-appearing cortical areas (Fig. 4A); that example is from an HIV-positive patient who had PML lesions in the cerebellar WM but no evidence of demyelination in the cerebrum. Numerous cells at the mid and lower cortical level expressed JCV T Ag but no VP1 (Fig. 4B). These cells had a distinct neuronal morphology, which was confirmed by double immunostaining for JCV and neuronal markers (Fig. 4C, D). Thus cortical neurons may sustain latent JCV infection even in the absence of demyelinating PML lesions in the nearby GM.

Figure 4.

An HIV-positive patient with PML lesions in the cerebellar WM has a restrictive JCV infection in the cerebral cortex with isolated neurons expressing T antigen (Ag) but not VP1. (A) Low-power view of the cortical and subcortical areas of the cerebrum showing no demyelination (H&E/Luxol fast blue). The boxed area in the grey matter (GM) is magnified in (B) in which immunohistochemistry (IHC) shows numerous cells with morphological characteristics of neurons expressing JCV T Ag (arrows, v-300, brown). Two of these cells in the boxed area of this panel are seen at higher power in the inset. (C) Double immunofluorescence staining against JCV T Ag (v-300, Alexa Fluor 568, red) and the neuronal marker MAP-2 (ab5392, Alexa Fluor 488, green) confirms that these cells are neurons. Three MAP-2-positive/T Ag-positive neurons in the box are shown in the lower inset (arrows), and viewed separately with the green (middle inset) and red channel. An adjacent uninfected neuron is marked by an arrowhead. (D) Double IHC with JCV anti-JCV T Ag (v-300, blue) and neuronal marker Neuron-specific beta III Tubulin (ab78078, brown) shows a large cortical neuron infected by JCV. Scale bars: A, 5 mm; B, 250 μm; D, 25 μm.

JCV Infection Is Mainly Productive in WM and Restrictive in GM

We then sought to determine the relative distribution of JCV T Ag and VP1 expression in the subcortical and cortical areas. In an HIV-positive PML patient with subcortical lesions in WM extending to the GWJ and G, double immunostaining showed that JCV VP1 (indicative of productive, lytic infection) was predominantly expressed in the WM, whereas JCV T Ag (consistent with a restrictive infection) was mainly found in the GM and GWJ (Fig. 5). These results suggest that the cerebral cortex may be a location of a restrictive infection by JCV.

Figure 5.

Relative expression of JCV T antigen (Ag) and VP1 protein in subcortical and cortical progressive multifocal leukoencephalopathy (PML) lesions. (A) Double immunofluorescence (IF) of low-power composite picture including 6,024,000 pixels/picture of an HIV-positive PML patient with anti-VP1 ab (PAB597, Alexa Fluor 488, green) and anti-T Ag, (v-300, Alexa Fluor 568, red). Boxes 1 to 3 showing representative areas of the white matter (WM), grey-white junction (GWJ) and grey matter (GM) areas are magnified. (B) Green channel show that infected cells expressing VP1 protein are mainly present in the PML lesion in the WM and to a lesser extent at the GWJ; they are rare in GM. (C) Conversely, the red channel indicates that cells expressing T Ag predominate in the GM and GWJ. Scale bar: 100 μm.

Hippocampal Neurons and Other Extracortical Neurons May Sustain JCV Infection

JCV-infected neurons were also found near PML lesions of the brainstem and hippocampus. For example, samples from 1 HIV-positive PML patient had widespread lesions of PML in the subcortical WM of the mesial temporal lobe that extended to the GWJ and GM (Fig. 6A). The boxed area of this panel is magnified in Figure 6B and shows the neuronal structure of the subiculum, as well as the CA1 and CA2 areas. JCV T Ag, but only rare VP1 expression, was observed in the hippocampal pyramidal neurons, while expression of VP1 in the glial cells of the nearby GWJ and WM was common (Fig 6C). These results are confirmed by double immunostaining for JCV and the neuronal marker MAP-2 shown in Figure 6D–F. These findings suggest a mostly restrictive infection of hippocampal neurons by JCV without significant production of mature viral particles.

Figure 6.

Hippocampal neurons of an HIV-positive/progressive multifocal leukoencephalopathy (PML) patient are infected with JC virus. (A–C) Low-power composite made of 154,000,000 pixels/picture of the mesial temporal lobe shows widespread demyelination in white matter (WM) and the grey-white junction (GWJ) (H&E/Luxol fast blue) The boxed area in the hippocampus magnified in (B) shows numerous pyramidal neurons that express JCV T antigen (Ag) (v-300, dark brown, arrows) and a few in the granule cell layer of the dentate gyrus (arrowhead). The boxed area from this panel is magnified in (C), showing T Ag staining in the nuclei of some neurons (brown, arrows). (D–F) Double immunofluorescence staining for JCV T Ag (v-300, Alexa Fluor 568, red) and the neuronal marker MAP-2 (ab5392, Alexa Fluor 488, green), shows a JCV-infected pyramidal neuron (arrows) between 2 uninfected neurons (arrowheads), which are shown in the green (E) or red (F) channels. Scale bars: A, 1 mm; C–F, 50 μm.

Frequency of Patients with PML lesions and JCV-Infected Cells and Neurons in Subcortical and Cortical Areas of the Cerebrum

Ninety-six percent of patients who had lesions in the subcortical areas had demyelination involving both WM and GWJ. Although much less common than lesions in the WM and GWJ, 57% of patients had PML lesions in the GM (p < 0.0001) (Fig. 7A). We then calculated the frequency of PML patients who had any T Ag- and VP1-expressing cells, including both glial cells and neurons, in the subcortical and cortical areas. Similar numbers of patients had at least some cells expressing either JCV T Ag or VP1 in all 3 geographic areas; the frequencies for WM and GWJ were 86% and 91%, respectively, and, as expected, were close to the frequencies of PML lesions in the same locations (Fig. 7B). Of note, the frequencies of patients harboring any cells in GM expressing T Ag (80%, p = 0.03) or to a lesser extent VP1 (72%, p = 0.19) were higher than the frequencies of patients with PML lesions in GM (57%, Fig. 7A). These results indicate that a significant proportion of PML patients have JCV-infected cells that are isolated in GM and that are not located within demyelinating lesions. Indeed, of the 20 (43%) PML patients without PML lesions in the GM, one quarter (5 patients) had JCV-infected cells expressing T Ag and/or VP1.

Figure 7.

Frequency of patients harboring progressive multifocal leukoencephalopathy (PML) lesions and JCV-infected cells in subcortical and cortical areas of the cerebrum. (A) PML lesions. (B) Cells (any type) expressing JCV T antigen (Ag) or VP1 protein. (C) Neurons expressing JCV T Ag or VP1 protein. WM: white matter; GWJ: grey white junction; GM: grey matter; n: number of cases that could be evaluated in each location. P values for differences between groups are indicated.

Furthermore, we determined the number of PML patients who had any neurons (identified by expression of neuronal marker) expressing JCV T Ag or VP1 in GWJ and GM. As shown in Figure 7C there was a trend for a smaller number of patients with VP1- (37%) vs. T Ag- (54%) expressing neurons in GWJ (p = 0.09); this difference was significant in GM (18% vs. 50%, p = 0.004). Fewer patients tended to have neurons expressing VP1 in GM compared to GWJ (18 vs. 37%, p = 0.06). Altogether, these results indicate that patients harboring restrictive infection of neurons by JCV are more frequent than those with productive infection, especially when neurons are located within GM.

Frequency of JCV-Infected Cells and Neurons in Subcortical and Cortical Areas of the Cerebrum

To evaluate the frequency of JCV-infected cells and neurons in our patients’ samples, we used a semiquantitative scale to estimate the number of infected cells (glial cells and neurons) in WM, GWJ and GM, and the number of infected neurons in GWJ and GM (Fig. 8). Because the subcortical WM does not contain neurons, these cells were not evaluated in this location.

Figure 8.

Semiquantitative evaluation of JCV-infected cells and neurons in subcortical and cortical areas of the cerebrum in PML patients. (A) Cells (any type) and neurons expressing JCV T antigen (Ag). (B) Cells (any type) and neurons expressing JCV VP1protein. WM: white matter; GWJ: grey white junction; GM: grey matter. 0 = none; 1 = 1–10 cells; 2 = 11 cells-1% cells; 3 = 2–5% cells; 4 = 5–25% cells. P values for differences between groups are indicated.

As expected, WM lesions of PML contained the highest number of JCV-infected cells. These cells expressed VP1 protein more frequently than they expressed T Ag (Wilcoxon signed rank test p = 0.003). T Ag-expressing neurons were more numerous than VP1-expressing neurons both in GWJ (p < 0.0001) and GM (p = 0.0002), and there were more T Ag- or VP1-expressing neurons in GWJ than in GM (p = 0.005 and p = 0.01, respectively). There were also more T Ag-and VP1-expressing cells of any type in the GWJ than in the GM (p < 0.0001). The number of cells expressing T Ag and/or VP1 was much higher (p < 0.0001) than the number of neurons in both GWJ and GM. These results indicate that the majority of JCV-infected cells are glial cells, even in the GM. Together, these results suggest that the number of cells infected by JCV decreases when from the subcortical to cortical areas; cells that are productively infected by JCV are more frequent in WM, whereas there are more numerous restrictively infected cells in GWJ and GM.

There were no significant differences between HIV-positive and HIV-negative PML cases, slow (>1 year) and rapid progressions (<1 year) for all of these parameters. A principal component analysis confirmed that the HIV status, the rapidity of PML progression, and the age of the patients were independent of the histological parameters, i.e. cells and neuron infection frequencies and demyelination (not shown).

DISCUSSION

The present results provide a new perspective on JCV biology in the CNS and considerably expand previous case reports of cortical neuron involvement with JCV infection (14, 20, 25, 26). Furthermore, they indicate that the presence of PML lesions in GM and infection of cortical neurons is an important aspect of JCV pathogenesis.

It is widely accepted that PML (as the name implies) is restricted to the WM. However, the cerebral GM contains numerous myelinated fibers leading to WM tracts (27) and the oligodendrocytes responsible for the myelination of these fibers, as well as nearby astrocytes, can be infected by JCV. Of note, subcortical U-fibers are myelinated fibers at the GWJ that enter and leave the GM and connect areas of cortex. U-fibers begin myelination early in gestation and often are not completely myelinated until the third or fourth decade of life (28). They are frequently affected in PML; indeed, we found that 96% of PML patients have lesions at GWJ and 57% in GM.

The location of PML lesions at the GWJ of such a high proportion of patients is intriguing in that it is similar to the common location of brain metastases. In the latter, clumps of tumor cells circulating in the bloodstream become trapped in these areas where end-arterioles decrease in diameter as they lead into capillaries. It is therefore possible that a hemodynamic phenomenon favors extravasation of leukocytes, perhaps B cells carrying JCV, through the blood-brain barrier at this location (29).

Although the concept that a leukoencephalopathy (by definition a WM disease) also involves the GM may seem counter-intuitive, the cortical location of the lesions correlates well with neurological manifestations. Indeed, PML patients often present with cortical blindness, aphasia, or seizures, which originate from GM. In these patients, deafferentation of the cortex caused by lesions in the subcortical WM, the cortical GM, or both, may be responsible for the neurological symptoms.

We found that up to 54% of patients had JCV-infected neurons in GWJ or GM and that these cells could be located either within or outside of demyelinating lesions. This was confirmed by the morphological aspect of infected cells, as well as double immunostaining with JCV proteins and 4 different neuronal markers. This finding opens a novel field of investigation and several new questions: Are cortical neurons susceptible only when they are in contact with JCV-infected glial cells? Conversely, could neurons be the primary target of JCV infection in the brain, with transmission to glial cells and demyelination occurring only as a secondary event? The presence of neurons expressing JCV proteins in the GM of patients without GM PML lesions suggests that these cells might be a primary site of infection in the CNS. However, most of these neurons express JCV T Ag only and not VP1, consistent with a restrictive infection caused by an abortive replication cycle that fails to produce mature viral particles. Therefore, it is likely that within demyelinated areas of the cortex, JC virions are generated first within glial cells and transmitted subsequently to neurons.

Another question raised by the results is the timing of neuronal infection. We and others have found JCV DNA, but not proteins, in the brains of HIV-negative control subjects without PML (30, 31); therefore it is possible that the virus penetrates in the CNS early on, perhaps at the time of primary infection, and remains latent until immunosuppression or decreased immunosurveillance allow for its reactivation. Regardless of the time of entry of JCV in the brain, it is possible that even a restrictive and focal infection of neurons may be associated with neurologic dysfunction in PML.

Indeed, we observed in one patient that hippocampal neurons could also be infected by JCV. This finding may be significant because the hippocampus, a GM structure located in the mesial temporal lobes, plays a crucial role in memory and may be the site of epileptogenesis (32). Of note, PML patients frequently present with memory dysfunction, and 18% of them develop seizures (33). Therefore, JCV infection of hippocampal neurons in PML patients deserves further study. Because mesial temporal sclerosis is a common cause of intractable epilepsy, the role of JCV in the hippocampus may need to be revisited as well in immunocompetent individuals. We have found that sequence variation located at the C terminus of the VP1 protein is associated with tropism of JCV in cerebellar granule cell neurons (16). Others have described point mutations in the VP1 leading to single amino acid changes in PML isolates, compared to JCV found in urine of the same individuals (34, 35). Investigations are now in progress in our laboratory to determine whether JCV mutants are preferentially infecting cortical and hippocampal neurons.