The chief outcomes from this study have been the demonstration for the first time that: (i) following MOG-induced spinal cord demyelination in the DA rat, CRT, CHOP and p-EIF2α were present at significantly increased levels within spinal cord lesions; (ii) significantly increased amounts of CRT, CHOP, XBP1 and p-eIF2α were detectable in the region of the central canal of diseased animals; (iii) levels of circulating CRT were significantly lower in EAE rat sera, when compared to control samples; (iv) CRT in human blood sera was present at significantly higher levels in individuals with MS, than in healthy controls.
Using intradermal injection of rmMOG, demyelination was induced in the spinal cord of adult female DA rats. Control animals were injected with IFA or saline only. Nine out of 10 animals developed a phenotype 13–21 days post MOG injection, which was retained in 5 individuals on the day of tissue harvesting. Inflammatory myelin loss was identified by staining of myelin and infiltrating microglia and the presence of myelin fragments within lesions was confirmed using ORO staining. After isolation of total RNA from whole spinal cord lysates, real-time PCR analysis of ER stress-related transcripts did not detect significant differences between control and EAE samples. A semi-quantitative protocol was then applied to immunohistochemically-stained tissue from the 5 rats in which inflammatory pathology was confirmed. On a scale of 1 to 5, the degree of staining was scored within the lesion, lesion edge, NAWM, GM and the region of the central canal and compared to that generated using IFA and saline controls. A total of 240 experimental and control tissue sections were examined, enabling us to profile the expression of BiP, CRT, CHOP, XBP1 and p-eIF2α.
At significantly increased scores, ranging from 1.6 to 5 in all EAE-associated regions examined, CRT was detected in astrocytes, microglia and GM neurons. When present, oligodendrocyte expression of CRT was minimal (not shown). IFA and saline control samples achieved CRT scores of 1 or less. An intriguing finding was the localisation of CRT to the rim of ORO-positive myelin fragments (confirmed in a human MS tissue sample, see additional file 2) and the punctate or ‘patchy’ nature of CRT staining seen when tissue was dual-labelled with CRT and GFAP or IBA1 (Figure 4i and j).
In contrast to CRT, we were unable to detect significant changes in the proportion of BiP-positive cells, when control and EAE samples were analysed. However, with scores of 2.2 to 3.9 AU, compared to staining in saline and IFA sections, which yielded scores of 1 or less, dual-labelling confirmed CHOP expression in astrocytes and microglia. Positive staining for XBP1, which scored 2.1 to 3.5, could be seen in a range of cell types within WM lesions following chromogenic single labelling, but the level of staining differed significantly only from the pattern seen in saline controls.
The final marker of ER stress assessed was p-eIF2α. Following tissue staining, it yielded scores of 1.9 to 3.1 AU. When morphological criteria were applied to the pattern of chromogenic staining seen, p-eIF2α staining appeared to be present in a range of cell types. Dual labelling confirmed p-eIF2α in microglia. Expression of p-eIF2α within lesions, at the lesion edge, in GM and the CC was also significantly higher than levels recorded for equivalent areas in saline and IFA controls.
Results obtained when NAWM was examined hinted that CRT, CHOP and p-eIF2α could be involved in lesion development, as these markers were present at significantly higher levels when compared to control counterparts.
Detection of significantly higher levels of CHOP and phospho-eIF2α within the region of the central canal in diseased animals hints that possible triggers of ER stress may be present in the cerebrospinal fluid and an investigation of candidate molecules is warranted.
We then expanded our investigations to include analysis of CRT secretion into the blood of EAE rats and in samples drawn from a cohort of 14 individuals with MS and 11 healthy controls. Data showed significantly reduced levels of CRT in rat sera when compared to IFA or saline control animals. The reverse was seen when human samples were analysed, in that MS patient samples contained significantly higher levels of secreted CRT. When results generated using samples taken from Natalizumab-treated patients only were analysed separately, mean CRT values were not found to differ significantly from mean values in healthy individuals. In contrast, the amounts of CRT detected in serum from individuals not undergoing treatment with Natalizumab, remained significantly higher than controls.
There are no previous reports of ER stress in rat EAE models. However, in broad terms, the data we have generated in rats confirms and adds to findings published using SJL and C57Bl/6 mice, in that BiP, CHOP, XBP1 and p-eIF2α were reported as significantly altered in these models [15–19] The question of what individual ER stress proteins are doing in the context of demyelination has yet to be answered. Deslauriers et al.  showed that EAE occured to the same degree in CHOP-/- mice as in normal controls, suggesting that CHOP is not required to induce inflammatory demyelination in mice. When present, as in our rat model, CHOP may be involved in perpetuating pro-apoptotic signals leading to loss of CHOP-positive cells within lesions. Alternatively, Popko and colleagues have proposed that CHOP is not pro-apoptotic, but is uniquely required to achieve complete remyelination following inflammatory demyelination .
The second major transcription factor associated with ER stress is XBP1. We found this highly specific marker of ER stress to be significantly altered in all regions of EAE lesions and in NAWM, when compared to saline-injected controls, but levels of XBP1 in EAE lesions did not differ significantly from those detected in control IFA animals. This is at odds with reported significant transcriptional upregulation of spliced XBP1 in samples isolated from EAE C57/BL mice . This may be due to the fact that the antibody used to screen our tissue does not discriminate between proteins encoded by spliced or unspliced variants, or may reflect the differences between the disease profile in our animal model and theirs. It is most likely that translational arrest is occuring within cells which stained positively for p-eIF2α, as phosphorylation of this molecule is known to interfere with protein translation, as part of a well characterised self-protective mechanism aimed at reducing the accumulation of misfolded proteins within the ER. It would be useful to determine the duration of this molecular event and how timing of its appearance or disappearance within lesions relates to tissue repair or ultimate cell and tissue destruction.
It is likely that, when functioning within the ER, proteins may have a role which differs to that found outside the ER. For example, while in the ER lumen, CRT participates in the CRT-calnexin cycle, ensuring correct folding of glycosylated proteins and trafficking of mis-folded proteins to the ER-associated degredation system. By an as yet undefined means which may occur under conditions of ER stress and may involve removal of the KDEL ER retention sequence , CRT could be displayed on the cell surface. This CRT cell surface expression may require interaction with C1q complement and CD91 . On the other hand, given that EAE is a T cell-mediated disease, it is also possible that CRT detected in the tissue samples derived from our EAE rats, was released from granules produced by invading cytotoxic T lymphocytes [22, 23]. Whether originating from brain cells, or released from invading T cells, CRT bound to C1q and CD91, may aid recognition and clearance of apoptotic cells by macrophages/microglia . Furthermore, some investigators have demonstrated a direct interaction between phosphatidylserine (PS) and CRT which was maintained when apoptosis caused PS to be ‘flipped’ to the outer leaflet of the plasma membrane [24, 25]. It may be that CRT’s appearance at the cell surface precedes PS exposure, as reported for tumour cells . Tarr et al also described ‘punctate clusters’ containing PS and CRT in apoptotic Jurkat T cells, reminiscent of the puctate staining seen in confocal imaging of CRT-stained EAE lesions (Figure 4j). Intriguingly, there is also a puncate quality to some of the CRT staining in the ORO-CRT images reported here. Again, it is possible that PS abnormally displayed on the outer surface of myelin fragments could be bound by CRT as part of a myelin debris clearance process.
Once CRT is secreted into the circulation, it may function in modulating innate and adaptive immune responses. In this regard, a large body of literature is accumulating on the role of CRT in the pathogenesis of rheumatoid arthritis (RA), another autoimmune disorder, Tarr et al  detected significantly higher levels of extracellular CRT in the synovial fluid and plasma of rheumatoid arthritis patients. At values between 232 and 623 ng/ml, the quantities of CRT found in the serum of MS patients examined in our study, were 20 to 30 times higher than those reported by Tarr et al, leading to the speculation that circulating CRT could also be playing a role in the pathogenesis of MS. Recent descriptions that various proteins including Hsp70, BiP and αB-crytstallin  have “chaperokine” functions suggest that CRT could also be involved in determining the balance between peripheral pro- and anti-inflammatory T cell subsets, although this remains to be definitively established. Our finding that levels of circulating CRT in the serum of EAE animals were significantly lower than levels detected in control IFA and saline animals is challenging to explain. Further time-course experiments need to be carried out in rat and human cohorts, to determine whether or not circulating CRT is, as we propose, a new robust surrogate biomarker of demyelinating disease and a possible ‘chaperokine’ in MS.