A mouse model of paralytic myelitis caused by enterovirus D68

Abstract

In 2014, the United States experienced an epidemic of acute flaccid myelitis (AFM) cases in children coincident with a nationwide outbreak of enterovirus D68 (EV-D68) respiratory disease. Up to half of the 2014 AFM patients had EV-D68 RNA detected by RT-PCR in their respiratory secretions, although EV-D68 was only detected in cerebrospinal fluid (CSF) from one 2014 AFM patient. Given previously described molecular and epidemiologic associations between EV-D68 and AFM, we sought to develop an animal model by screening seven EV-D68 strains for the ability to induce neurological disease in neonatal mice. We found that four EV-D68 strains from the 2014 outbreak (out of five tested) produced a paralytic disease in mice resembling human AFM. The remaining 2014 strain, as well as 1962 prototype EV-D68 strains Fermon and Rhyne, did not produce, or rarely produced, paralysis in mice. In-depth examination of the paralysis caused by a representative 2014 strain, MO/14-18947, revealed infectious virus, virion particles, and viral genome in the spinal cords of paralyzed mice. Paralysis was elicited in mice following intramuscular, intracerebral, intraperitoneal, and intranasal infection, in descending frequency, and was associated with infection and loss of motor neurons in the anterior horns of spinal cord segments corresponding to paralyzed limbs. Virus isolated from spinal cords of infected mice transmitted disease when injected into naïve mice, fulfilling Koch's postulates in this model. Finally, we found that EV-D68 immune sera, but not normal mouse sera, protected mice from development of paralysis and death when administered prior to viral challenge. These studies establish an experimental model to study EV-D68-induced myelitis and to better understand disease pathogenesis and develop potential therapies.

Fig 1. EV-D68 causes limb paralysis in neonatal mice.

(A) Phylogenetic tree of EV-D68 based on VP1 sequence showing each strain tested in this study. (B) Bar graph showing the percent of mice with paralysis following intracerebral injection with each EV-D68 strain or control media. (C) Examples of limb paralysis (arrows) in 3 different mice following intracerebral injection of MO/14-18947. Left–day post-infection 3 (dpi 3), center–dpi 12, right–dpi 28. (D) Day of paralysis onset in mice after intracerebral injection with MO/14-18947. Mice that died (n = 9 circles with “X”) generally had earlier onset of paralysis than mice that lived (n = 18 circle). (E) Average number of functional (non-paralyzed) limbs in mice following intracerebral injection with MO/14-18947 in mice that lived (solid line, squares) and mice that died (dotted line, circles). Error bars represent the standard error of the mean (SEM). (F) Serum anti-EV-D68 neutralizing antibody titer in mice (reciprocal serum dilution) following intracerebral injection with MO/14-18947 at dpi 12 and dpi 28 (n = 10), Fermon (n = 8–10), Rhyne (n = 7–10), CA/14-4231 (n = 8–10) or mock (n = 8). The line in each group on the graph represents the median titer. Lower limit of assay detection (LLD) of 1:10 and upper limit of assay detection (ULD) of 1:10,240 are indicated by the dotted lines.

Fig 2. Infectious virus and EV-D68 RNA can be detected in the spinal cords of infected mice.

(A) Graph shows the mean TCID₅₀ titer in spinal cords from mice infected intracerabrally with MO/14-18947. Between 5 and 11 mice from up to 2 litters were collected for each time point. At dpi 0 virus was undetectable in the spinal cords of infected mice, but mean viral titer rose steadily at dpi 2, although the mice in this group had not yet developed signs of paralysis. Paralyzed mice examined on dpi 4 showed further increase in mean spinal cord viral titer. Mean spinal cord titer plateaued from dpi 4 through dpi, 8 and then declined, becoming undetectable by dpi 12 (B) Mean genome copy number in whole spinal cords from the same mice in (A) quantified by two-step RT-PCR targeting EV-D68 VP1 paralleled the titers seen in (A), although viral genome remained detectable on dpi 12 despite absence of infectious virus. This indicates that viral genome can remain detectable in tissue longer than viable infectious particles. Note that estimated genome copies were used as relative comparisons between each time point and are likely an underestimate of absolute viral RNA due to inefficiency of RNA extraction from animal tissue and amplification by two-step RT-PCR. Error bars represent SEM.

Fig 3. EV-D68 infects anterior horn motor neurons and results in motor neuron death.

(A) A cervical spinal cord section at 100X original magnification from a mouse injected intracerebrally with EV-D68 MO/14-18947 that developed right forelimb paralysis on day 4 post-injection. Loss of motor neurons (green, labeled with choline acetyltransferase / ChAT) is observed in the right (“R side”) anterior horn, corresponding to the affected side. In contrast, motor neurons of the left anterior horn corresponding to the unaffected side are relatively intact. (B) A consecutive cervical spinal cord section from the same dpi 4 mouse at 100X original magnification reveals the presence of EV-D68 VP2 capsid protein within the few remaining right anterior horn neurons. The box represents the area imaged at 200X in (C). (C) 200X and (D) 600X images of the right anterior horn stained for EV-D68 VP2 (green). The box represents the area imaged at 600X in (D). (E) 200X and (F) 600X images from a left anterior horn in an intracerebral-injected mouse at day 3 post-injection before the onset of paralysis showing EV-D68 antigen in an intact cluster of motor neurons. The box represents the area imaged at 600X in (F). For all images, neurons (magenta) are labeled with NeuN, a general neuron marker, and nuclei (blue) are labeled with Hoechst 33342. Scale bars for 100X original magnification are 400 μm, 200X are 200 μm, 600X are 50 μm.

Fig 4. Large numbers of particles morphologically consistent with enterovirus are seen in dying anterior horn cells of the spinal cord following intracerebral injection of EV-D68.

TEM images were taken from the cervical spinal cord anterior horn of a day 4 post-injection MO/14-18947 mouse with new onset forelimb paralysis. (A) A dying cell, consistent in position and morphology with a motor neuron, showing nuclear fragmentation, cytoplasmic blebbing, and regions dense with particles morphologically consistent with enteroviruses (arrowheads). The box represents the area imaged in (B). (B) A higher magnification image shows a cluster of particles morphologically consistent with enteroviruses. Scale bars are 2 μm and 500 nm for low (11,000X) and high magnification (98,000X), respectively.

Fig 5. EV-D68 causes paralysis after intranasal and intramuscular infection.

(A) An image of a day 3 post-infection mouse with new onset left hindlimb paralysis following intramuscular injection of EV-D68 strain MO/14-18947. (B) Lumbar spinal cord sections (100X original magnification) from an intramuscular-injected mouse at day 3 post-infection with left hindlimb paralysis showing loss of motor neurons and viral antigen (arrows) in the left anterior horn. (C) An image of a mouse on day 12 post-infection that developed right forelimb paralysis following intranasal infection with EV-D68 strain MO/14-18947. (D) Cervical spinal cord sections (100X original magnification) from a day 8 post-infection intranasal-infected mouse with right forelimb paralysis showing viral antigen in the anterior horn (white arrow). The scale bars are 400 μm. Abbreviation: dpi, days post-infection.

Fig 6. Fulfillment of Koch’s postulates establishes a causal role for EV-D68 in paralytic disease.

(A) Steps taken to fulfill Koch’s postulates in the EV-D68 mouse model of AFM. Mice were infected with MO/14-18947 by intracerebral injection and monitored daily until the onset of paralysis (Step 1). The whole spinal cord of a paralyzed mouse was removed and mechanically lysed (Step 2). The spinal cord lysate was then inoculated into a flask of RD cells (Step 3). After the appearance of cytopathic effect, the RD cell media with passaged spinal cord lysate was collected and syringe filtered. The filtered media with passaged spinal cord lysate was then injected intracerebrally into naïve mice (Step 4). 38% (n = 9 out of 24) of these mice developed signs of paralysis. Spinal cords from several paralyzed mice (n = 3 out of 3) in Step 4 were removed and examined by TCID₅₀ and metagenomic sequencing for the presence of EV-D68. (B) TCID₅₀ analysis revealed infectious virus confirmed to be MO/14-18947 by metagenomic deep sequencing identification for each step in the fulfillment of Koch’s postulates.