Memory loss and aberrant neurogenesis in mice exposed to patient anti-N-methyl-D-aspartate receptor antibodies

Abstract

Objective:

Anti-N-methyl-D-aspartate receptor (anti-NMDAR) encephalitis results in chronic epilepsy and permanent cognitive impairment. One of the possible causes of cognitive impairment in anti-NMDAR could be aberrant neurogenesis, an established contributor to memory loss in idiopathic drug-resistant epilepsy. We developed a mouse model of anti-NMDAR encephalitis and showed that mice exposed to patient anti-NMDAR antibodies for 2 weeks developed seizures and memory loss. In the present study, we assessed the delayed effects of patient-derived antibodies on cognitive phenotype and examined the corresponding changes in hippocampal neurogenesis.

Methods:

Monoclonal anti-NMDAR antibodies or control antibodies were continuously infused into the lateral ventricle of male C56BL/6J mice (8–12 weeks) via osmotic minipumps for 2 weeks. The motor and anxiety phenotypes were assessed using the open field paradigm, and hippocampal memory and learning were assessed using the object location, Y maze, and Barnes maze paradigms during weeks 1 and 3–4 of antibody washout. The numbers of newly matured granule neurons (Prox-1+) and immature progenitor cells (DCX+) as well as their spatial distribution within the hippocampus were assessed at these time points. Bromodeoxyuridine (BrdU: 50 mg/kg, i.p., daily) was injected on days 2–12 of the infusion, and proliferating cell immunoreactivity was compared in antibody-treated mice and control mice at week 3 of the washout.

Results:

Mice infused with anti-NMDAR antibodies demonstrated spatial memory impairment during week 1 of antibody washout (p = 0.02, t-test; n = 9–11). Histological analysis of hippocampal sections from these mice revealed an increased ectopic displacement of Prox-1+ cells in the dentate hilus compared to the control-antibody-treated mice (p = 0.01; t-test). Mice exposed to anti-NMDAR antibodies also had an impairment of spatial memory and learning during weeks 3–4 of antibody washout (object location: p = 0.009; t-test; Y maze: p = 0.006, t-test; Barnes maze: p = 0.008, ANOVA; n = 8–10). These mice showed increased ratios of the low proliferating (bright) to fast proliferating (faint) BrdU+ cell counts and decreased number of DCX+ cells in the hippocampal dentate gyrus (p = 0.006 and p = 0.04, respectively; t-tests) suggesting ectopic migration and delayed cell proliferation.

Significance:

These findings suggest that memory and learning impairments induced by patient anti-NMDAR antibodies are sustained upon removal of antibodies and are accompanied by aberrant hippocampal neurogenesis. Interventions directed at the manipulation of neuronal plasticity in patients with encephalitis and cognitive loss may be protective and therapeutically relevant.

1. Introduction

Anti-NMDA receptor (NMDAR) encephalitis, the most common form of antibody-positive noninfectious brain inflammation, is characterized by the presence of neuronal antibodies against glutamate receptor subunits in the serum and cerebrospinal fluid (CSF) (Dalmau et al., 2008). Symptoms manifest as cognitive decline, psychosis, and new-onset seizures (Dalmau, 2016; Dalmau et al., 2007). Seizures and cognitive dysfunction often persist in the chronic phase of encephalitis (Byun et al., 2016; Finke et al., 2012), and the mechanisms underlying the pathophysiology of these symptoms are not well understood. Most patients who recovered from anti-NMDAR encephalitis do not have visible structural or signal changes on conventional brain imaging (Dalmau et al., 2011; Iizuka and Sakai, 2008) supporting the premise that despite recurrent seizures, the gross destruction of neuronal brain tissue does not commonly occur in anti-NMDAR encephalitis. This is also supported by findings from histopathological examination of biopsy and postmortem brain specimens of patients with anti-NMDAR encephalitis and animal models of encephalitis, which have failed to identify neuronal loss in the hippocampus (Bien et al., 2012; Taraschenko et al., 2019). Despite the lack of obvious persistent anatomical changes in the chronic stage of the disease, the impairment of the processing speed, episodic memory, and executive dysfunction leading to permanent disability remain in these patients many months after the initial encephalitis presentation (McKeon et al., 2018; Nicolle and Moses, 2018).

We previously demonstrated that innate immune activation is one of the key pathophysiological mechanisms of cognitive failure in anti-NMDAR encephalitis (Taraschenko et al., 2021b, 2024). Specifically, the attenuation of the interleukin-1 receptor (IL-1R)-mediated signaling promotes recovery of memory and attenuation of seizures in mice treated with anti-NMDAR antibodies (Taraschenko et al., 2021b) while removal of the myeloid differentiation primary response 88 (MyD88) in the toll-like receptor signaling pathway allows protective seizure and memory phenotype in MyD88 knockout mice (Taraschenko et al., 2024). Another potential mechanism of cognitive failure in acquired epilepsies was linked to the disruption of hippocampal neurogenesis (Ammothumkandy et al., 2022; Coras et al., 2010; Fu et al., 2019). Adult neurogenesis in humans and mice takes place in the subgranular zone (SGZ) of the dentate gyrus of the hippocampus where the step-wise maturation of granule cells takes place over 4–6 weeks (Duan et al., 2008; Hsieh, 2012). The expression of the appropriate cellular markers allows the distinction of the doublecortin-positive (DCX+) neuroblasts and Prospero-related homeobox 1-positive (Prox-1+) newborn granule neurons at approximately 1–4 and 1–6 weeks, respectively, from the start of the development of the multipotent neural stem cells (Duan et al., 2008). Newborn granule cells migrate out from SGZ to other hippocampal layers before they become incorporated into the neuronal microcircuits that participate in supporting long-term memory and other cognitive functions (Kempermann et al., 2004). Induced or spontaneous epileptic seizures in rodents lead to aberrant hippocampal neurogenesis, including increased proliferation of neural progenitors and enhanced ectopic migration of granule cells in the hippocampus (Fu et al., 2019; Ledergerber et al., 2006; Overstreet-Wadiche et al., 2006).

Patients with chronic drug-resistant epilepsy demonstrated decreased neurogenesis in the mesial temporal lobe and the longer duration of epilepsy was associated with a more significant decline in the generation of new neurons (Ammothumkandy et al., 2022). Further, altered neurogenesis was associated with accelerated cognitive decline in an Alzheimer’s disease (AD) model with spontaneous seizures; whereas, administration of conventional anti-seizure medications caused cessation of seizures and restoration of neurogenesis and memory function in these mice (Fu et al., 2019). Finally, a study in patients with surgically treated severe temporal lobe epilepsy showed that encoding new memories was related to the regenerative capacity of patients’ hippocampus and that low proliferation or differentiation abilities of stem cells correlated with worse memory function in these patients (Coras et al., 2010). Collectively, these findings provide evidence that seizures can disrupt neurogenesis, which along with other factors can perpetuate memory loss in epilepsy; however, the role of hippocampal neurogenesis in memory loss in autoimmune encephalitis has not been assessed.

In the present study, we use a previously characterized mouse model of antibody-induced seizures in anti-NMDAR encephalitis (Taraschenko et al., 2021a, 2019) to assess changes in hippocampal neurogenesis following the brief and prolonged washouts after the sustained infusion of anti-NMDAR antibodies and establish the relationship between these changes and cognitive performance in mice.

2. Methods

2.1. Animals

C56BL/6J male mice (8–12 weeks, Charles River Laboratories (Roanoke, IL) were housed in groups of five initially and individually after surgery. They were maintained on a 12-hour light cycle (light on/off at 7 a.m. / 7 p.m.) with ad libitum access to a standard irradiated rodent chow diet (Teklad Diet 7912, Envigo, Indianapolis, IN) and water. All experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center (UNMC, Approval ID 18–116-08-FC). The principles outlined in the ARRIVE principles were followed (Kilkenny et al., 2014).

2.2. Cranial surgery and implantation of subcutaneous osmotic pumps

Mice were placed into the stereotaxic apparatus and anesthetized with fluothane. The unilateral injector guide cannula (Protech International Inc., Boerne, TX) was lowered into the lateral ventricle (AP = −0.6; ML= 2.5, DV = −3.2, angle 36 °) and secured with glue as previously described (Taraschenko et al., 2021a, 2019). After a 7-day recovery period, mice were implanted with subcutaneous osmotic minipumps (Alzet, Cupertino, CA; 1002) filled with experimental monoclonal anti-NMDAR antibodies or inactive control antibody solution (Fig. 1A–B). The solution contained the combination of three monoclonal antibodies derived from an anti-NMDAR encephalitis patient with seizures (1D1, 5F5, and 2G6; 0.2 μg/μl) or control 6A IgG1λ human monoclonal antibody, which does not bind GluN, dissolved in phosphate-buffered saline (PBS). The antibodies were developed and tested as previously described (Sharma et al., 2018). The antibody solution was infused for 14 days at a flow rate of 0.25 μl/h. Upon completion of the experiments, the contents of the pumps were examined for residual solution. The residuals were less than 0.05%.

Figure 1.

Schematic detailing of the two experimental designs. The infusion of anti-NMDAR antibodies and assessment of mouse behavior and histopathological changes during week 1 of antibody washout (A). The infusion of anti-NMDAR antibodies or control antibodies, administration of bromodeoxyuridine (BrdU), and assessment of mouse behavior and histopathological changes during weeks 3–4 of antibody washout (B). OF, open field; OL, object location; DCX, doublecortin; Prox-1, Prospero-related homeobox-1.

2.3. Bromodeoxyuridine injections

To establish how the treatment with antibodies and seizures affects proliferation of the newly developed cells, the thymidine analog bromodeoxyuridine (BrdU, Millipore Sigma, St. Louis, MO) was administered daily (50 mg/kg i.p.) to mice treated with anti-NMDAR antibodies or control antibodies on days 2–12 of the infusion (Fig. 1B). This dose and frequency of BrdU administration has been previously used in other rodent models where it was proven successful for studying neurogenesis in the adult rodent DG (Kuhn et al., 1996; Sorrells et al., 2018; Taupin, 2007); the timing was chosen from our previous findings of seizures during that time (Taraschenko et al., 2021b).

2.4. Behavioral tests

Two groups of mice (one treated with anti-NMDAR antibodies and one treated with control antibodies) for 2 weeks underwent the assessment of motor and anxiety functions using the open field paradigms starting one day following the completion of the antibody infusion as previously described (Cho et al., 2015; Prieur and Jadavji, 2019; Taraschenko, 2019); the studies were completed over 3 days (Fig. 1A). Briefly, mice were habituated in the testing room for 30 min before being placed into the custom-made acrylic arena (41 cm length, 41 cm width, 35 cm height) and were allowed to move freely during a 20-min trial during which a total distance traveled and time spent in the center of the arena (25% of the total area) were recorded. Next, memory function was assessed using the object location test and Y maze with spontaneous alterations according to the modified previously described paradigms (Cho et al., 2015; Prieur and Jadavji, 2019) (Fig. 1A). For the object location test, mice were habituated in the arena and on the next day were allowed to explore for 5 min two identical objects placed in the corners of the test arena (acquisition trial) after which, following a 2h retention period, one object was displaced to a new location (NL) to be positioned diagonally from the object in the familiar location (FL); time spent to explore each location was recorded (test trial) to derive the NL index (time NL/time (NL+FL). Mice that spent less than 3 sec in total exploration time with both objects were excluded.

In the Y-maze comprised of three arms positioned 120 degrees apart (Stoelting, Wood Dale, IL) mice were placed at the end of one arm and allowed to move freely through the maze during an 8-min session. The entry was defined as the head and all four paws completely advancing into the arm (Maurice et al., 1994). Alternation was defined as successive entries into the three arms, on overlapping triplet sets. The alternation index was calculated as a ratio of the number of alternations to the number of choices (Maurice et al., 1994). Animal behavior was tracked using video camera recording (Any Maze, Stoelting).

To assess how the potential changes in neurogenesis relate to behavior, the duration of antibody washout was extended to 4 weeks. Thus, upon completion of the 2-week-long washout, two groups of mice, one infused with anti-NMDAR antibodies and one control antibodies, were tested in the open field and object location paradigms, as well as in Barnes maze and Y-maze (weeks 3 and 4 of the washout, Fig. 1B). In a modification of the previously described Barnes maze paradigm (Attar et al., 2013), the apparatus was comprised of a circular platform with 20 holes of 4.45 cm in diameter positioned 2.54 cm from the edge of the 92 cm wide elevated PVC platform (Maze Engineers, Skokie, IL). Following a 3-min habituation trial in which the mice were placed in the center of the platform and allowed to explore the apparatus for 3 min without intervention, mice were subjected to a similar procedure in which they were trained to locate an escape chamber located beneath one of the peripheral holes located on the apparatus (the target hole). Both groups were trained during seven, 3-min-long training trials scheduled over a period of three days. The latency to reach the target hole and the proportion of mice that did not reach the target hole (failure rate) were recorded for each training session to assess learning. In the memory probe trial, the procedure was repeated after the escape chamber was removed. In this probe, the animal’s movement within the maze was recorded for the full 3-min period regardless of the primary latency (i.e., time to identify the target hole the first time). Mice that did not reach the target hole within 3 min were excluded. The primary latency, total time spent in the target quadrant, and time spent in this quadrant after the target hole was identified were recorded. Animal behavior was tracked using an automated tracking equipment (Any Maze, Stoelting). The analysis of all behavioral tests was performed by an experimenter blinded to the treatment status of the mice.

2.5. Histological analysis

Following the completion of behavioral tests, mice were deeply anesthetized with fluothane and transcranially perfused with cold PBS. The brains were dissected out, and the hemispheres were separated by a longitudinal incision. One hemisphere was immersed in 4% paraformaldehyde (PFA) solution for 1 day and transferred to 30% sucrose solution for 2 days after which it was paraffin embedded. Six 5 μm-thick coronal sections through the hippocampus were obtained every 150 μm between the anterior-posterior (AP) bregma −1.4 and −2.3(Paxinos and Franklin, 2019); one section at each AP level was used for immunohistochemistry analysis of each cellular marker (Fig. 2A–B). For fluorescent staining, the brains were frozen in 2-methylbutane and coronally sectioned on a freezing microtome at 30 μm along the anteroposterior dimension of the hippocampus, and one section was used for each cellular marker (6 per mouse). The free-floating sections were stored in PBS.

Figure 2.

Histological evaluation of hippocampus. Schematic representation of the sampling approach and coordinates from bregma (A) and representative sections of the hippocampus at these coordinates (B) after 1 week of antibody washout; hematoxylin-eosin, scale bar: 50 μm. Schematic representation of the hippocampal dentate gyrus subregions assessed during week 4 of antibody washout (C). GZ, granular zone; SGZ, subgranular zone; ML, molecular layer; DH, dentate hilus; scale bar: 30 μm.

Formalin-fixed paraffin-embedded tissue sections were deparaffinized with xylene (LabChem Inc., Zelienople, PA) and rehydrated in descending concentrations of ethanol per standard procedures (Taraschenko et al., 2019). Endogenous peroxidase activity was suppressed with 3% H₂O₂ in 100% MeOH for 20 min at room temperature. The sections were incubated in a citrate buffer mix (0.01M, pH 6.0) at 95° C for 40 min, washed with PBS, and blocked with 2.5% horse serum (Vector Laboratories, Newark, CA) for 30 min. Primary antibodies to identify DCX and Prox-1- positive cells were applied overnight at 4° C (rabbit anti-DCX, 1:1000, Cell Signaling Technology; Danvers, MA, 4604S and rabbit anti-Prox-1, 1:500, Abcam; Waltham, MA, Ab199359, respectively). On the following day, the sections were incubated for 1 h with secondary antibodies (horse anti-rabbit, Vector Laboratories, Newark, CA, MP-7401) and washed in PBS. The NovaRed substrate (Vector Laboratories, Newark, CA) was applied for 5 min and washed with PBS.

Immunohistochemistry for fluorescent markers of BrdU and DCX or Prox-1 was performed on free-floating tissue sections as previously described (Varma et al., 2019). Briefly, sections were permeabilized with 0.4% Triton in TBS (tris-buffered saline) for 30 min and washed with TBS before they were incubated in 2N hydrochloric acid (HCl) and washed with 0.1M Na2B4O7 (borate buffer) and TBS. Primary antibodies (rat anti-BrdU, 1:500; Abcam, Ab6326 and rabbit anti-Prox1, 1:500; Millipore, Burlington, MA, Ab5475) were added and left overnight at 4 °C. On the following day, sections were rinsed repeatedly with TBS and secondary antibodies (donkey anti-rat-cy3, 1:300; Jackson ImmunoResearch, West Grove, PA, 712–165-153 or anti-rabbit IgG 1:300, AlexaFluor 488 donkey; Invitrogen, Waltham, MA, A21206) were added for 3 h and then rinsed with TBS (Varma et al., 2019). DAPI reagent (Sigma, St. Louis, MO) was added, and the sections were rinsed and mounted on microscopic glass slides.

2.6. Visualization and quantification of the progenitor cells

Immunoreactivity for DCX⁺ and Prox-1⁺ in the sections from the paraffin-embedded brains was quantified using a light microscope equipped with a camera (Leitz Diaplan, Grand Rapids, MI) with a 10x objective. The fluorescent immunoreactivity for BrdU⁺ cells and those co-labeled with DCX or Prox-1 in the free-floating sections was assessed using a confocal microscope (Nikon A1R, Nikon Instruments Inc., Melville, NY) with a 10x objective and a stack of images was collected with 1024 × 1024 resolution, at 1 μm steps. Quantification of cells in the dentate gyrus of the hippocampus was performed on maximum intensity projected images (10 μm) using ImageJ software. A rolling ball background subtraction was applied to all images. The cell counts were based on the intensity autothresholding of pixels (triangle function) and size (analyze particles) in ImageJ. The watershed function was applied to all images to separate overlapping cells.

The minimal and maximal fluorescent intensity for BrdU+ cells was recorded for each image, and the cells with intensities above the mean were regarded as “bright” and those below the mean as “faint”. Lightly and heavily labeled nuclei in the BrdU-immunoreactive preparations likely represented subsequent postmitotic generations with a brighter appearance corresponding to less proliferative activity (del Rio and Soriano, 1989). The granular zone (GZ) and SGZ of the dentate gyrus were the focus of our studies as they are the most proliferative regions in the dentate gyrus (Mandyam et al., 2007); but cells were also quantified in molecular layer (ML) and dentate hilus (DH) to assess the proliferative capacity of the dentate gyrus subregions and to contrast them with the more neurogenic GZ and SGZ (Fig. 2C). The cells were regarded to belong to SGZ if they were touching the granular layer or were positioned within the two cellular layers within the DH (Malberg et al., 2000). The cells were considered ectopic if they were positioned more than two cell diameters from the SGZ border into the DH (Varma et al., 2019). The cells were counted in each section by an experimenter blinded to the study code. The counts for each cellular marker were added for all hippocampal sections in two hemispheres using the approach previously applied by others (Varma et al., 2019).

2.7. Statistical analysis

Continuous data including behavioral scores and cell counts were compared between the groups using a two-tailed Student’s test. Longitudinal measurements (i.e., distribution of cells along the axis of the hippocampus or latencies to reach the target hole) were compared using repeated measure ANOVA. The proportions of mice that failed to reach the target hole during each trial were compared using the chi-squared tests (GraphPad Prism, v.9, Boston, MA).

3. Results

3.1. Mice develop selective memory deficits during the first week following the antibody infusion

Mice tested 1–3 days after the infusion of antibodies showed no changes in their motor or anxiety phenotype. Specifically, in the open field test, mice treated with anti-NMDAR (n = 11) and control antibodies (n = 10) traveled 63.5 ± 4.0 and 61 ± 5.4 m, respectively (p = 0.72) during a 20-minute trial. The anti-NMDAR antibody- and control antibody-treated mice spent equal proportions of time in the center of the arena (13.7 ± 2.3 and 12.2 ± 2.6 %, respectively, p = 0.65, Fig. 3A–B). These findings suggest that the motor and anxiety phenotypes were not altered immediately following the withdrawal of antibodies.

Figure 3.

Behavioral phenotype of mice treated with anti-NMDAR and control antibodies during week 1 of antibody washout. Following the habituation in the testing room environment, mice were allowed to explore the test arena; the total distance traveled and time spent in the center of the arena were recorded. Mice treated with anti-NMDAR antibodies did not differ in their motor or anxiety phenotypes in an open field test (A-B). In the object location paradigm, the acquisition trial during which mice were presented with two identical objects was followed by a retention period and a test trial during which one object was displaced to a new location (NL) diagonally from the object in the familiar location (FL). The time spent in the NL and FL were recorded and the NL index was calculated as the ratio of NL and the total time spent in NL and FL. In a Y maze paradigm, mice were allowed to explore three open arms of the Y maze; the alternation index was calculated as a ratio of the number of alternations to the number of choices. Mice treated with anti-NMDAR developed impaired spatial discrimination in the object location paradigm (C) while their memory was intact in a Y-maze paradigm (D). *, p < 0.05, t-test. The data represent mean ± SEM; n = 9–11 per group; NL, new location.

In the object location test, one mouse in the anti-NMDAR-treated group and two mice in the control group were excluded because they failed to satisfy a 3-second cutoff for total exploration time. The mean NL indexes were 43.2 ± 5.0% and 60.1 ± 3.9% in anti-NMDAR antibody- and control antibody-treated groups, respectively (Fig. 3C, Table 1). Mice treated with anti-NMDAR antibodies demonstrated decreased ability to discriminate between the NL and FL in contrast to the mice treated with control antibodies, which preferred the NL over the FL; this indicates the impairment of spatial memory and object discrimination (Murai et al., 2007) (p = 0.02, t- test, n = 9 per group; Fig. 3C). However, in the Y maze paradigm, mice treated with anti-NMDAR antibodies (n = 11) and control antibodies (n = 10) showed similar exploratory behavior and spontaneous alternation indexes (67 ± 3.7% and 66.4 ± 3.4%, respectively; p = 0.9; Fig. 3D, Table 1) indicating that short-term working memory was not affected by antibody treatment.

Table 1.

Summary of the memory test outcomes during the antibody washout in mice treated with anti-NMDAR and control antibodies.

Assessments (weeks from IgG removal), units	Anti-NMDAR IgG	Control IgG	p-value
Memory test indexes, mean ± SEM
Object location (1), %	43.2 ± 5.0	60.1 ± 3.9	p = 0.02
Y maze (1), %	67 ± 3.7	66.4 ± 3.4	p = 0.9
Object location (3), %	43.2. ± 5.7	63.1 ± 3.9	p = 0.009
Barnes maze, failure rate: trials 3 and 4 (4), %	0 and 30	100 and 70	p = 0.03
Barnes maze, primary latency (4), sec	55.1 ± 21.4	82.7 ± 18.7	p = 0.35
Barnes maze, total time in the target quadrant (4), sec	86.2 ± 9.8	73.3 ± 8.0	p = 0.32
Barnes maze, time in the target quadrant after discovery of the hole (4), sec	69.3 ± 14.6	43.6 ± 8.5	p = 0.13
Y maze (4), %	55.0 ± 3.1	71.0 ± 2.3	p = 0.0006

3.2. Mice infused with anti-NMDAR encephalitis demonstrated an enhanced ectopic displacement of granule cells in the dentate hilus

The number of ectopic Prox-1+ cells in the hippocampal hilus was more than three times higher in mice treated with anti-NMDAR antibodies compared to the control mice (p = 0.01; t-test; Fig. 4A–D). The mean total counts per hippocampus (± SEM) were 62.8 ± 13.0 and 20.3 ± 2.4, respectively (n = 8 and n = 7, respectively, Fig. 4A). Analysis of the longitudinal distribution of the cell counts along the AP axis of the hippocampus revealed that cell numbers were reduced equally at all AP bregma levels in both treatment groups (Group effect: p = 0.02, ANOVA; Fig. 4B). The number of DCX+ cells was similar in anti-NMDAR antibody and control antibody-treated groups (532 ± 68.6 and 547 ± 68.7, respectively; p = 0.88; t-test; Fig. 5 A–C).

Figure 4.

Expression of the Prospero-related homeobox-1-positive (Prox-1+) granule cells in the hippocampal dentate gyrus after 1 week of antibody washout. Mice treated with anti-NMDAR antibodies developed an increased ectopic displacement of Prox-1+ granule cells in dentate hilus (DH) after 1 week of the antibody washout (A). The expression of the ectopic Prox-1+ cells in the DH of mice treated with anti-NMDAR antibodies was consistently higher than that in mice treated with control antibodies at all assessed bregma levels (B). The data represent mean ± SEM. **, p < 0.001, t-test; n = 7–8 mice per group. Representative microscopic images showing the Prox-1 immunoreactivity of the nootropic mature granule neuron in the hippocampal granule zone (GZ, arrow, and insert) and ectopic neurons in DH (circle) of mice treated with control antibodies (C) and anti-NMDAR antibodies (D). Scale bar: 50 μm (insert: 30 μm).

Figure 5.

Expression of the doublecortin-positive (DCX+) cells in the hippocampal dentate gyrus after 1 week of antibody washout. Mice exposed to anti-NMDAR antibodies did not show changes in the expression of DCX+ cells in the hippocampal granule zone (GZ) upon removal of anti-NMDAR antibodies or control antibodies for 1 week (A). Longitudinal changes in the expression of DCX+ cells in the hippocampal GZ were similar in both groups at all assessed bregma levels (B). The data represent mean ± SEM; n = 7–8 mice per group. Representative microscopic images showing the DCX immunoreactivity in the hippocampal GZ of mice treated with antibodies (C and insert). Scale bar: 50 μm (insert: 15 μm).

3.3. Mice exposed to anti-NMDAR antibodies demonstrated sustained memory impairment upon removal of antibodies

To test the hypothesis that cognitive impairment in autoimmune encephalitis is accompanied by aberrant neurogenesis, we administered BrdU for 10 days during the infusion of antibodies (Fig. 1B) corresponding to the time when an abundance of seizures was expected based on our previous studies (Taraschenko et al., 2021a). The motor function and anxiety phenotype of mice treated with anti-NMDAR antibodies (n = 10) and control mice (n = 10) assessed during week 3 of washout in an open field paradigm were similar (Fig. 6A–B). Specifically, mice traveled 48.4 ± 3.2 and 45.5 ± 2.6 m and spent 13.7 ± 3 and 12.3 ± 1.7 % of time in the center of the arena, respectively (p = 0.49 and p = 0.69, respectively; t-tests). In the object location test, mice treated with anti-NMDAR antibodies had significantly lower mean NL index compared to mice treated with control antibodies; (43.2. ± 5.7% and 63.1 ± 3.9%, respectively; p = 0.009; t-test; Fig. 6C; Table 1); one mouse in anti-NMDAR antibody-treated group was excluded because it spent less than 3 sec exploring both objects.

Figure 6.

Behavioral phenotype of mice treated with anti-NMDAR and control antibodies during weeks 3 and 4 of antibody washout. Following the infusion of anti-NMDAR and control antibodies, mice were retained in their home cages for an additional two weeks before their motor and anxiety phenotypes were assessed in an open field paradigm; the total distance traveled, and time spent in the center of the arena were recorded. Mice treated with anti-NMDAR antibodies and control antibodies displayed similar motor or anxiety phenotypes during week 3 of antibody washout (A-B). In the object location paradigm, the acquisition trial during which mice were presented with two identical objects was followed by a retention period and a test trial during which one object was displaced to a new location (NL) diagonally from the object in the familiar location (FL). The time spent in the NL and FL were recorded and the NL index was calculated as the ratio of NL and the total time spent in NL and FL. Mice exposed to anti-NMDAR antibodies demonstrated persistent loss of spatial discrimination in an object location test (C). One week later, in the Barnes maze apparatus mice were trained to locate an escape chamber situated beneath one of the peripheral holes (the target hole) over the course of three days (seven sessions). The latency to reach the target hole in the Barnes maze and the proportion of mice that did not reach the target hole (failure rate) were recorded for each training session to assess learning. During the acquisition trial, mice treated with anti-NMDAR antibodies displayed higher latencies to enter the escape chamber and higher failure rates compared to the mice treated with control antibodies demonstrating impaired learning (D-E). In a Y maze paradigm, mice were allowed to explore three open arms of the Y maze; the alternation index was calculated as a ratio of the number of alternations to the number of choices. Mice treated with anti-NMDAR antibodies demonstrated spatial memory impairment in a Y-maze paradigm (F). *, p < 0.05 (chi-squared tests); **, p < 0.01; ***, p < 0.001 (t-tests or ANOVA). The data represent mean ± SEM or proportions of animals; n = 9–10 per group. NL, new location.

In the Barnes maze, during the acquisition training, the latency to enter the escape chamber was higher in anti-NMDAR-treated mice than in control antibody-treated mice (Group effect: p = 0.008, two-way ANOVA, Fig. 6D). The failure rate was also higher in the former group during trials 3 and 4, suggesting an impaired learning in anti-NMDAR antibody-treated mice (p = 0.03; chi-squared tests; Fig. 6E, Table 1). During the memory probe trial, two mice in the anti-NMDAR antibody-treated group were excluded from the final analysis due to overall failure to meet the training criterion after the training procedure (n = 8 and n = 10 in anti-NMDAR antibody-treated and control antibody-treated groups, respectively). The significant differences were not detected when comparing the primary latency for the location of the target hole between the anti-NMDAR antibody-treated and control mice (55.1 ± 21.4 and 82.7 ± 18.7 sec, respectively; p = 0.35; t-test Supplemental Fig. 1A, Table 1). Similarly, there were no significant differences in the total time spent in the target quadrant or time spent in the target quadrant after the target hole was identified (p = 0.32 and p = 0.13, respectively; t-tests; Supplemental Figs. 1B–C). Specifically, the total times spent in the target quadrant in anti-NMDAR antibody and control mice were 86.2 ± 9.8 and 73.3 ± 8.0 sec, respectively and those spent in the target quadrant after the discovery of the target hole were 69.3 ± 14.6 and 43.6 ± 8.5 sec, respectively. Furthermore, anti-NMDAR antibody-treated mice had significantly lower spontaneous alternation index in Y maze compared to the control mice (55.0 ± 3.1% and 71.0 ± 2.3%, respectively; p = 0.0006, t-test, Fig. 6F) further supporting the presence of impairment of working memory weeks after the antibody removal. Collectively, these findings, in conjunction with findings from Y maze assessments, demonstrate short-term and working memory impairments in the NMDAR antibody-treated mice. However, we did not observe similar differences in latencies between groups in the Barnes maze probe trial.

3.4. Mice with memory loss induced by anti-NMDAR antibodies showed disruption of hippocampal neurogenesis

In both experimental and control groups of mice, BrdU+ positive cells were found in GZ and SGZ as well as in ML and DH (n = 6 mice per group). The overall quantities of BrdU+ cells in these regions were similar in anti-NMDAR antibody-treated and control mice (GZ and SGZ: p = 0.71; ML: p = 0. 45; DH: p = 0. 41, respectively; t-tests; Fig. 7A). However, the ratio of bright to faint BrdU+ cell counts in the GZ and SGZ that reflects the dynamics of the precursor cell division (Mandyam et al., 2007) was significantly higher in anti-NMDAR antibody-treated mice compared to the control mice suggesting a shift toward the predominance of low proliferating SGZ precursors in the former group (p = 0.006, t-test; Fig. 7B). There were no differences in the ratios of bright and faint BrdU+ cells in the two other assessed regions, such as ML (p = 0.20) and DH (p = 0.07; t-tests; Supplemental Fig. 2A).

Figure 7.

Assessment of neurogenesis after 4 weeks of antibody washout. Mice treated with anti-NMDAR antibodies and control antibodies displayed similar patterns of total BrdU immunoreactivities in the granular zone (GZ) and subgranular zone (SGZ) as well as a molecular layer (ML) and dentate hilus (DH) (A). Mice treated with anti-NMDAR antibodies demonstrated a higher bright-to-faint BrdU+ cell ratio (B) and lower densities of doublecortin-positive (DCX+) cells in the hippocampal GZ and SGZ compared to the mice treated with control antibodies after 4 weeks of antibody washout (C). The data represent mean ± SEM. *, p < 0.05, ***, p < 0.001, t-tests; n = 6 mice per group. Representative microscopic images showing the DCX and BrdU immunoreactivities in the hippocampus of mice treated with control antibodies (top panel) and anti-NMDAR antibodies (lower panel) (D). Scale bar: 117 μm.

The numbers of ectopic Prox-1+ cells in DH were similar in mice treated with anti-NMDAR antibodies and control mice (p = 0.87, t-test; Supplemental Fig. 2B). The respective Prox-1+ cell counts in these groups were 16.3 ± 4.2 and 17.3 ± 4.2. The total numbers of ectopic and normotropic Prox-1+ cells colocalized with BrdU+ cells of any intensity in all hippocampal regions were low, 4.7 ± 1.5 and 4.8 ± 4.7 in anti-NMDAR and control antibody-treated mice, respectively; there was no difference between the two groups (p = 0. 5, t-test; not shown).

The total number of DCX+ cells in the hippocampal dentate gyrus of mice treated with anti-NMDAR was marginally lower compared to those treated with control antibodies (401 ± 27.7 and 561.8 ± 63.7.3, respectively, p = 0.04, t-test, Figs. 7C–D). The counts of DCX+ cells colocalized with BrdU+ in the dentate gyrus were 2.8 ± 1.4 and 1.2 ± 0.5 in the anti-NMDAR antibody-treated and control mice, respectively (data not shown). There was no significant difference in number of the BrdU+/DCX+ cells between the two groups (p = 0.29, t-test, data not shown). Collectively, these findings suggest that antibody treatment resulted in decreased proliferation of the BrdU+ cells and decreased density of DCX+ early neuronal progenitor cells in the GZ and SGZ of dentate gyrus that were detected at 4 weeks after removal of antibodies. Furthermore, the ectopic displacement of the granule Prox+ cells that were detected after one week of antibody washout was no longer present at this time point.

4. Discussion

In the present study, using the mouse model of anti-NMDAR encephalitis with seizures previously developed in our laboratory (Taraschenko, 2019), we assessed the effects of passive transfer of monoclonal anti-NMDAR antibodies on memory phenotype during the short-term and prolonged antibody washouts and established the corresponding patterns of neurogenesis in the hippocampus. We showed that mice exposed to anti-NMDAR antibodies for 2 weeks developed impairments of spatial memory that were apparent both one and 3–4 weeks following the removal of antibody infusion. The antibody-treated mice also demonstrated ectopic displacement of Prox-1+ granule cells in the DH shortly after the removal of antibodies without changes in the densities of early neuronal progenitor DCX+ cells. Further, mice exposed to anti-NMDAR antibodies had an increased ratio of bright to faint BrdU+ cells in the hippocampus and a decreased number of DCX+ neuroblasts in the GZ and SGZ of dentate gyrus 4 weeks after the antibody removal that is indicative of a decrease in neural progenitor proliferation and/or cell cycle arrest. Collectively, these findings suggest that prolonged exposure to anti-NMDAR antibodies results in sustained memory loss, reduced proliferative activity of progenitor cells, and decreased formation of new neurons in the hippocampus.

Memory loss and executive dysfunction after clinical remission are the cardinal features of the chronic phase of autoimmune encephalitis (Heine et al., 2021). The incomplete recovery of cognitive function in anti-NMDAR encephalitis resulting in the inability to attain gainful employment is particularly devastating for young patients who represent the largest segment of this disease population (Gable et al., 2012). In our previous studies, we showed that mice treated with anti-NMDAR antibodies derived from affected patients demonstrate seizures and memory deficits in a novel object recognition paradigm during week 1 of antibody removal (Taraschenko et al., 2021b); the latter was consistent with reports by others (Planaguma et al., 2015). We reproduced these findings in the present study using another behavioral paradigm, an object location test, and showed that spatial memory deficits supported by the hippocampus persisted beyond week 1 for an additional 3 weeks following the removal of antibodies. Interestingly, the protracted memory deficits at weeks 3–4 were noted in both object location and Y maze paradigms and were also accompanied by impaired learning in the Barnes maze paradigm. The persistence of memory loss was in contrast to one study that showed restoration of memory in mice by day 25 following the cessation of patient anti-NMDAR antibody infusion (Planaguma et al., 2015). This could be due to the use of the different test batteries in our study that were specific to uncover spatial memory deficits compared to the novel object recognition test, which is designed to test a non-spatial memory domain (Costa et al., 2021; Planaguma et al., 2016, 2015). Alternatively, the persistent impairment of learning and memory functions upon removal of antibodies in our study could be attributed to the differential repertoire and dissociation profile of monoclonal anti-NMDAR antibodies and those present in CSF used by other authors (Planaguma et al., 2016, 2015; Sharma et al., 2018). Despite the consistently impaired performance on short-term and working memory tasks such as the Y maze and Barnes maze training trials, we did not detect a difference in performance in the Barnes maze memory probe trial in mice exposed to anti-NMDAR antibodies. This could be explained by the wide variability observed in the performance of the group during the test probe. This pattern was similar to that reported in other mouse epilepsy and encephalitis models that demonstrated impaired learning but no significant differences in long-term memory (Linnoila et al., 2023; Van Den Herrewegen et al., 2019). However, there is evidence that NMDAR antibody-treated mice may demonstrate more significant long-term memory deficits following an increased period between reimmunization and Barnes maze testing (Linnoila et al., 2023).

Neurons born in adult hippocampal SGZ migrate to the GZ where they receive functional inputs. The process of neurogenesis, including the proliferation of the neural stem cells or progenitors, their differentiation, maturation, and integration into the existing networks are continuously supported by several physiological and pathological contributions (Zhao et al., 2008). Previous studies in rodents revealed that epileptic seizures lead to aberrant neurogenesis, including increased proliferation of neural progenitors and enhanced ectopic migration of granule cells in the hippocampus (Fu et al., 2019; Ledergerber et al., 2006). For example, an estimated 1% of the total granule cell population migrates ectopically into the DH instead of the GZ of the hippocampus as a result of pilocarpine-induced status epilepticus in mice (Walter et al., 2007; Zhan et al., 2010). These cells exhibit a high rate of spontaneous activity that may enhance seizure propagation and receive more excitatory inputs from mossy fibers than cells in the GZ (Zhan et al., 2010), which in turn could contribute to memory loss (Ammothumkandy et al., 2022; Fu et al., 2019). However, recurrent seizures in AD and epilepsy may also cause a biphasic effect on neurogenesis with an initial acute increase and subsequent reduction of cell division and proliferation likely to the exhaustion of the finite pool of neural stem cells (Fu et al., 2019). Collectively, these findings provide evidence that seizures can disrupt neurogenesis leading to time-dependent decreased cell proliferation and ectopic cell displacement in the hippocampus, which along with other factors, can support persistent memory loss in autoimmune encephalitis.

In the present study, we demonstrated such an enhanced presence of Prox-1+ cells in mice with memory loss as early as 3 weeks from the initiation of antibody infusion (i.e., 1 week from the removal of antibodies). Since it takes about 6 weeks for the newly born stem cell to fully migrate out of SZG (Duan et al., 2008), exposure to anti-NMDAR antibodies and seizures occurring during that time likely facilitated this process in our study . For example, seizure-induced depletion of the migration factors (e.g., reelin signaling) described in other epilepsy models (Haas and Frotscher, 2010) could have resulted in unopposed aberrant migration of progenitor cells in anti-NMDAR-antibody-treated mice in our study. Consistent with this premise, antibody-mediated blockage of reelin function in naïve mice caused the granule cell dispersion, while exogenous reelin administration after seizure prevented the dispersion (Haas and Frotscher, 2010). The ectopic hilar localization of new granule cells without a change in density of DCX+ cells in our study could support this possibility. Alternatively, hippocampal inflammation maintained during the exposure to antibodies in our mice (Taraschenko et al., 2021b, 2024) may support aberrant neurogenesis and help shorten the typical timeline of stem cell development (Goshen et al., 2008; Zhu et al., 2020). The observed changes in neurogenesis have correlated with memory function. Given that memory loss has been demonstrated early upon antibody removal, this indicates that granule cells are not a single determinant of the cognitive phenotype in anti-NMDAR encephalitis. Interestingly, other studies found a place recognition defect as early as 8 days after the targeted fractionated brain irradiation that blocked the formation of new neurons in mice (Madsen et al., 2003).

Ectopic hilar neurons can survive for months after the initial injury (McCloskey et al., 2006); however, we did not see an increase in ectopic Prox-1+ cells at 6 weeks from the start of the infusion (i.e., 4 weeks after the removal of antibodies) while we still detected the memory deficits. It is unclear why the enhanced ectopic expression of the Prox-1+ cells noted early was not apparent at the later stages of antibody washout; however, it was not likely due to the animal aging given the consistent numbers of the ectopic Prox-1+ cells in the control mice at both time points. Although we did not note any neuronal death in the hippocampus at week 1 of the antibody washout (Taraschenko et al., 2019), ectopic granule cells in DH may have died later. Another intriguing possibility could be that ectopically migrated neurons, noted during week 1 of the washout, may transition to glial cells weeks later and stop expressing the neuronal markers tested in our study. For example, the progenitor cells from human patients with temporal lobe epilepsy favor astroglia cell fate rather than a neuronal fate with longer disease duration and have co-expressed DCX+ along with glial markers in DH (Ammothumkandy et al., 2022). Overall, while the antibodies in our study show a strong phenotype in neurogenesis at week 4 of antibody removal, they still appear to modulate neuronal function at week 1 as behavioral impairments also exist at the earlier point. However, the loss of memory function and findings of the disruption in neurogenesis remain correlative. Further studies with genetic or pharmacological ablation of neurogenesis are needed to establish a causative link between the cellular changes and behavioral phenotype in mice in our model.

In addition to addressing the time course of cell division in the SGZ of mice treated with anti-NMDAR antibodies, we quantified precursors over a similar time course in dentate gyrus subregions that have been reported to contribute to hippocampal plasticity (Mandyam et al., 2007). With labeling studies using BrdU administered during the critical time for seizures, we saw that cells born during this time were proliferating more slowly than cells in the control mice. This was reflected in the increased ratio of bright to faint BrdU+ cells in mice treated with anti-NMDAR antibodies that indicates slower postmitotic evolution (del Rio and Soriano, 1989). Interestingly, these changes were specific to the GZ and SGZ regions where the proliferation and differentiation are expected to take place (Hsieh, 2012). This is consistent with the reduced number of DCX+ neuroblasts in GZ and SGZ in anti-NMDAR-antibody-treated mice at week 4 of antibody washout. However, the impact of the antibodies on DCX+ progenitor survival could also contribute to the number change; and we did not design the BrdU labeling studies to assess the cell survival. The lack of changes in DCX+ cell counts at week 1 and their decrease at week 4 after the antibody removal suggests that the effect of the antibodies on neurogenesis is lagging over the time frame. Lack of significant change in the total number of BrdU+ cells of all intensities may be associated with slow division or quiescence, which protects cells from apoptosis or damage (Coller et al., 2006; Min and Spencer, 2019).

The limitations of this study include the lack of concomitant seizure measurements and the degree of immune activation in the hippocampus. This precludes drawing the conclusion regarding the direct correlation of seizure burden with aberrant neurogenesis and does not allow us to assess the contribution of hippocampal inflammation to the observed changes in neurogenesis. While our previous studies demonstrated higher CCL2 activation in mice with a higher number of seizures (Taraschenko et al., 2021a), we did not see a correlation of seizure burden with early memory impairment which further supports the premise of multifactorial etiology of memory loss in autoimmune encephalitis similar to that proposed for epilepsy (Lenck-Santini and Scott, 2015). The other limitation is our restricted ability to capture the newly born cells during the 4 weeks after the antibody infusion and BrdU administration were stopped. In our preliminary studies, mice continue to have spontaneous seizures for at least 7–10 days upon the removal of antibodies. Therefore, it is possible that new neurons that were born after the BrdU labeling could contribute to the impaired memory phenotype. Further, the numbers of the mature and proliferating cells reported herein reflect the quantities in the samples assessed with our approach and represent a fraction of those present in the entire mouse hippocampus. Despite that, we believe that the reported relationships in our sample reflect the findings in the entire hippocampus of the whole brain. Finally, this study reflects on the postmortem evaluation of neurogenesis that limits the translation importance of such assessments; the study highlights the need for novel noninvasive tools to monitor neurogenesis in vivo in affected patients.

5. Conclusions

In summary, we demonstrated the presence of persistent memory deficits and aberrant neurogenesis in mice with anti-NMDAR encephalitis. These findings will provide the groundwork for future studies in patients with cognitive loss after autoimmune encephalitis to develop new therapeutic interventions. Environmental enrichment and exercise, advances in invasive neurostimulation, and therapeutic implantation of stem cells are emerging tools to manipulate neuronal plasticity in epilepsy, which could be brought to the field of autoimmune epilepsy to combat the major cognitive disability in encephalitis.

Supplementary Material

Supplemental Figure 1 A-C

Supplemental Figure 1. Assessment of the memory function in the Barnes maze paradigm in mice treated with anti-NMDAR and control antibodies during week 4 of antibody washout. Mice were trained to locate an escape chamber situated beneath one of the peripheral holes located on the apparatus (the target hole) over the course of three days (seven sessions). In the memory probe trial, the procedure was repeated after the escape chamber was removed. During the administration of a probe trial, mice treated with anti-NMDAR antibodies and control antibodies demonstrated comparable latencies for the location of target hole (A), total time spent in the target quadrant (B), and time spent in the target quadrant after the discovery of the target hole (C). The data represent mean ± SEM; n = 8–10 mice per group.

Supplemental Figure 2 A-B

Supplemental Figure 2. Dynamics of maturation of the BrdU+ cells in the molecular layer (ML) and dentate hilus (DH) after 4 weeks of antibody washout revealed similar patterns in anti-NMDAR antibody-treated and control mice (A). The numbers of ectopic Prox-1+ cells in DH were similar in mice treated with anti-NMDAR antibodies and control mice after 4 weeks of antibody washout (B). The data represent mean ± SEM; n = 6 mice per group.

Highlights.

• Mice exposed to patient anti-NMDA receptor antibodies develop sustained memory deficits that persist upon removal of antibodies.

• The memory impairment in mice was accompanied by aberrant hippocampal neurogenesis, including the loss of progenitor cells and ectopic displacement of granule cells

• Aberrant neurogenesis may contribute to the persistent cognitive failure in anti-NMDA receptor encephalitis after the resolution of an acute phase.