Loss of cerebroside is most profound in corpus callosum and varies in other brain regions
In the corpus callosum, loss of the myelin-specific lipid, cerebroside, was 65% after 6 weeks of cuprizone exposure; after 12 weeks of exposure, the deficit was 78%(Fig. 1a, p < 0.05, mnova) This deficit was almost double that reported for whole brain using the same exposure model (Jurevics et al. 2001). The cerebroside deficit present in corpus callosum by 6 weeks of cuprizone exposure was only partially compensated following removal of the metabolic insult and allowing for 6 weeks of recovery. This also contrasts to whole brain data, where the lesser cerebroside deficit on a whole brain basis was largely eliminated by 6 weeks into the recovery phase. This discrepancy is probably due to the greater precision of analysis when only the severely affected region is analyzed. The differential response to cuprizone exposure for different brain regions is clear when results for cerebellum (maximum of 33% demyelination, Fig. 1d) and brainstem (maximum of 13% demyelination, Fig. 1g) are examined. As for corpus callosum, if a recovery phase is included, there is a hint of partial restoration of cerebroside levels.
Figure 1. Effects of cuprizone exposure on cerebroside concentration, cerebroside synthesis rates, and ceramide galactosyltransferase (CGT) steady-state mRNA levels in corpus callosum, cerebellum, and brainstem of mice. Animals at 8 weeks of age were exposed to cuprizone in their diet for up to 12 weeks. In some cases, cuprizone was removed after 6 weeks of exposure and animals allowed to recover for up to 6 additional weeks. Data for cerebroside concentration and synthesis rates are means ± SEM (n = 3); data for mRNA levels are single determinations and are plotted relative to the mean of the three controls.
Download figure to PowerPoint
Changes in rate of cerebroside synthesis precede changes in accumulation
For corpus callosum, the rate of synthesis of cerebroside is qualitatively as expected (Fig. 1b). The alteration in rate of synthesis of cerebroside is not, however, directly proportional to subsequent changes in cerebroside level. Even in the control situation, although the concentration of cerebroside is constant from 8 to 20 weeks postnatal (Fig. 1a), there is a decrease in rate of synthesis of cerebroside, presumably reflecting a developmental slowing of cerebroside turnover. If animals are exposed to cuprizone, there is an immediate and significant down-regulation of cerebroside synthesis (Fig. 1b, p < 0.03, mnova); as expected, this precedes the physical loss of cerebroside. Yet the down-regulation may not be as marked as expected in consideration of the profound loss of cerebroside noted in Fig. 1(a). Consider that over the 12-week period, the control animals synthesize a total of 10.9 nmoles cerebroside/mg wet weight of corpus callosum (proportional to the area under the closed circle data line of Fig. 1b– after correction to weekly values). The equivalent calculation after cuprizone treatment (area under the closed diamond line) is 7.7 nmoles cerebroside/mg wet weight of corpus callosum. Thus, the decrease in cerebroside synthesized over 12 weeks is 30%; in contrast, the actual decrement in cerebroside level is 78% (noted earlier with respect to Fig. 1a). This suggests there is not just a lack of renewal of cerebroside undergoing normal metabolic turnover, but also pathophysiological events that result in increased catabolism or removal of cerebroside. As expected, during recovery, the total amount of cerebroside synthesized (5.1 nmol/mg wet weight) was more than enough to account for the amount of cerebroside accumulation (3.8 nmol/mg wet weight).
The data relating to cerebellum is less striking; a 20% decrease in total synthesis of cerebroside and a 32% decrease in concentration of cerebroside. Interestingly, the mild 12% cerebroside deficit in brainstem is less than the 30% deficit in cerebroside newly synthesized over the 12-week period (see Discussion).
Changes in levels of mRNA coding for CGT also reflect demyelination/remyelination
There is a profound down-regulation of mRNA for CGT (ceramide galactosyltransferase, the rate-limiting enzyme for cerebroside biosynthesis) in corpus callosum (Fig. 1c). Even in the presence of continued challenge with cuprizone, there is complete return to normal levels of mRNA. If cuprizone exposure is continued, there is another fall in mRNA levels followed by partial recovery. This pattern is as expected (Matsushima and Morell 2001); the available evidence suggests that upon cuprizone challenge, oligodendroglia go through apoptosis, myelin is removed, and precursor cells infiltrate and differentiate into myelinating glia – a cycle which can be repeated (Mason et al. 2001). Again, the pattern in corpus callosum is an exaggerated version of that noted previously in whole brain (Jurevics et al. 2001). Data for CGT mRNA from the other brain regions (Figs 1f and i) are less dramatic, but compatible with the data for corpus callosum.
Alterations in cholesterol levels are as predicted for loss of myelin
During cuprizone-induced demyelination, cholesterol levels in corpus callosum were also significantly reduced (Fig. 2a, p < 0.02, mnova). The reductions in cholesterol levels, if calculated in terms of nanomole deficit, are roughly as expected to match the cerebroside deficits. For example, after 6 weeks of exposure to cuprizone in the diet, the deficiency in cerebroside relative to initiation of exposure is 12.5 nmol/mg wet weight, and the deficit in cholesterol is 22 nmol/mg wet weight. The molecular ratio of cholesterol to galactolipid lost is 1.8, only slightly off the expected value of 2.0 characteristic of myelin (Norton 1977). However, in contrast with the myelin and oligodendroglial-specific distribution of cerebroside, cholesterol is distributed in surface membranes of all cells. Thus the cholesterol deficit in terms of percentage decrement of the total is less dramatic than for cerebroside. Similarly, cuprizone exposure induced decrements in the rate of synthesis of cholesterol (Fig. 2b) and of steady state levels of the mRNA coding for the rate-limiting enzyme in cholesterol synthesis (HMG-CoA reductase, Fig. 2c) that are proportionately less than analogous measures for cerebroside. As expected, for cerebellum and brainstem, the decrements in levels of cholesterol (Figs 2d and g), rates of cholesterol synthesis (Figs 2e and h), and HMG-CoA reductase mRNA (Figs 2f and i) are milder than those for corpus callosum.
Figure 2. Effects of cuprizone exposure on cholesterol concentration, cholesterol synthesis rates, and HMG-CoA reductase (HMG-CoA Red) steady-state mRNA levels in corpus callosum, cerebellum, and brainstem of mice. See Fig. 1 legend for additional details. Data for cholesterol concentration and synthesis rates are means ± SEM (n = 3); data for mRNA levels are single determinations, and are plotted relative to the mean of the three controls.
Download figure to PowerPoint
GeneChip analysis of RNA from corpus callosum
The corpus callosum, as dissected from brains of cuprizone-exposed animals, represents a preparation in which the sequence and time-course of events is quite reproducible. As we have shown above, due to the severity of the demyelination, specific biochemical alterations related to demyelination and remyelination are easy to discern above the general metabolic background. This should be true also for gene expression. Thus, we used Affymetrix GeneChip technology to analyze mRNA expression in corpus callosum at various stages of demyelination and remyelination (all GeneChip data are for 6 weeks of cuprizone exposure followed by 6 weeks of recovery). We did not consider as significantly altered any gene whose peak expression was less than one-third of that of the ‘average’ gene on the chip, although we realize this arbitrary discriminator may exclude genes whose up-regulation is of physiological significance.
Preexisting information regarding the cuprizone exposure model system allows for testing for specific expectations. GeneChip data for CGT (GenBank accession # U48896) is similar to that obtained by northern analysis (compare Fig. 4, closed circles, to that in Fig. 1c). Other data in Fig. 4 indicate the expected lesser alterations for HMG-CoA reductase mRNA (M62766). Two other profiles for mRNA species involved in more general aspects of lipid metabolism are included; those for fatty acid synthase (X13135) and stearoyl desaturase-1 (M21285) also lack the cuprizone-induced profound down/up-regulation of the more myelin-specific CGT mRNA.
Figure 4. GeneChip analysis of effects of cuprizone exposure on steady-state mRNA levels of selected enzymes involved in lipid synthesis in corpus callosum of mice. Data are from animals exposed to cuprizone for up to 6 weeks and then (for later time points) allowed to recover. Relative mRNA levels were determined from Affymetrix GeneChip data using GeneSpring analysis software. Average difference values for each time point were normalized to the overall average value for all genes on the chip; these values were then plotted relative to the corresponding control value for that gene obtained from three control samples.
Download figure to PowerPoint
The patterns of some genes coding for membrane components present in myelin or in oligodendroglial cells are shown in Fig. 5(a). These include the well characterized proteolipid protein (PLP, AI85038) and myelin-associated glycoprotein (MAG; M31811; for references, see Morell and Quarles 1999). Also included are results for myelin-oligodendrocyte glycoprotein (MOG, L20942; for review see Johns and Bernard 1999), myelin-associated oligodendrocyte basic protein (MOBP, U81317, Yamamoto et al. 1994), myelin and lymphocyte protein (MAL, Y07812, Schaeren-Wiemers et al. 1995), and oligodendrocyte transmembrane protein (OTP, U19582, Szuchet et al. 2001). We do not report results for a gene listed by Affymetrix as myelin basic protein (M11533), since they were not readily interpretable. We do not have access to the sequences on the GeneChip and suspect this may be some representation of isoforms of MBP or of sequences from the Golli gene (this larger and earlier-expressed gene includes the MBP sequence; for references, see Campagnoni and Skoff 2001).
Figure 5. GeneChip analysis of effects of cuprizone (exposure for 6 weeks, followed by recovery) on steady-state mRNA levels of myelin or myelin-related proteins (a) and of other genes showing temporal patterns of mRNA expression similar to MOG (b). Data were analyzed and plotted as described in Fig. 4.
Download figure to PowerPoint
We used GeneSpring data analysis software to search for genes that are expressed in the same pattern as MOG, which showed the most dramatic down-/up-regulation among the myelin/oligodendroglial-specific genes. Some of these genes are plotted Fig. 5(b): synectin (AF104358), presumed to be involved in syndecan-4-dependent interactions, with a possible role in assembly of the syndecan-4 signaling complex (Gao et al. 2000); BCL7C (Y11905), related to a gene deleted in Williams' syndrome, possibly resulting from some impaired developmental process (Jadayel et al. 1998); G-protein coupled receptor 26 (GPC R-26; U13370), distantly related to some known receptors and found in brain (Lee et al. 2000); heterogeneous nuclear ribonucleoprotein U (hnRNP U; AF073992), interacts with class III POU factors (Malik et al. 1997) that are well expressed in the oligodendrocyte lineage (Schreiber et al. 1997); and NK6 transcription factor related (NK6 TFR; L08074, also called Gtx), this gene is expressed during myelination and re-expressed during remyelination after experimentally induced demyelination (Sim et al. 2000); kinesin family 5C, kinesin heavy chain (Kinesin HC; X61435); several expressed sequence tags (ESTs) also showed similar temporal patterns of expression − one of these (AA797709) is shown in the figure.
The temporal expression pattern of genes associated with myelin is consistent with demyelination followed by remyelination; other genes with similar temporal expression patterns may be linked to these processes as well. We caution that these microarray data are preliminary – comparison of one chip at each age point to the mean of three control chips. The validity of the interpretations rests upon the normalization procedures built into the analysis of data from each chip and, importantly, the demonstration of expected temporal specificity as related to chemical analysis and histology. With respect to these results and those presented below, we argue only for the utility of this model system and that the observations are worth follow-up with more quantitative techniques such as northern blots or real-time PCR.
Another reference point for temporal analysis of microarray data is the time-course of accumulation of microglia/macrophages in corpus callosum. Lysozyme is a good marker for activated macrophages (Cross et al. 1988; Venezie et al. 1995). Up-regulation of lysozyme mRNA in cuprizone-exposed mice correlates with immunohistochemical evidence for recruitment of microglia/macrophages and consequent demyelination (Morell et al. 1998). Expression of lysozyme M (M21050) and other genes with similar temporal patterns are shown in Fig. 6. Note that these genes go from a normally very low level to a significant level of expression. Thus a plot of mRNA levels ‘relative to control values’ might give misleading results since the control level of base expression is very low but variable. Therefore, we have included in Fig. 6 the magnitude of the peak response relative to the average expression level of all the genes on the chip. Among the representative genes plotted, some are directly involved as macrophage components or are factors involved in their differentiation: macrophage metalloelastase (MME; M82831); histocompatibilty 2 antigen 1-E (HC-2A; X00958); leukocyte common antigen (L-CA, M23158); hematopoietic lineage switch 2 protein (HLS2, AF009513); involved in the erythroid to myeloid lineage switch.
Figure 6. GeneChip analysis of effects of cuprizone (exposure for 6 weeks, followed by recovery) on steady-state mRNA levels of the macrophage marker, lysozyme M, and of other genes showing temporal patterns of mRNA expression similar to lysozyme M. Data were analyzed and plotted as described in Fig. 4.
Download figure to PowerPoint
Also induced were markers less specific but clearly with functions relevant to recruitment and activation of phagocytic cells. Examples include phospholipase D1 (PLase D1; U87868), a widely distributed activity involved in membrane trafficking (see Jones et al. 1999 for review). Other molecules involved in signaling possibly related to such recruitment and activity include TrkB (M3385), known to play a role in neural development (e.g. Frost 2001) and possibly in glial cell biology (Althaus and Richter-Landsberg 2000), and R-ras (M21019), a monomeric GTPase known to activate αMβ2 integrin and thus promote phagocytosis (Self et al. 2001). The above are representative of genes whose increased expression during microglial/macrophage recruitment might be expected. Many genes were identified with similar patterns of expression in response to cuprizone exposure, and these might be relevant for future targeted study. Again, these include a number of ESTs; AI019327 is plotted in Fig. 6 as an example.
Another group of genes of interest (Fig. 7) includes those with a temporal pattern similar to insulin-like growth factor 1 (IGF-1, X04480). Appearance of IGF-1 in this model is prominent in the corpus callosum only after apoptosis of oligodendroglial cells (Mason et al. 2000). A literature suggests IGF-1 acts to promote early differentiation of oligodendroglial precursors (e.g. Ye et al. 1995; see Mason et al. 2000, 2001 for other references). Other potential signaling molecules that may promote oligodendroglial precursor recruitment and differentiation could include those with similar temporal patterns of expression (Fig. 7). Some of these genes could also be involved in recruitment or stimulation of microglia/macrophage. Expression of mRNA species with the temporal pattern characteristic of IGF-1 include: JE protein (M19681); monocyte chemoattractant protein-1 (MCP-1), a CC chemokine involved in recruitment of monocytes; MIP-1 gamma (U49513), mouse macrophage inflammatory protein-1 gamma, a CC chemokine family member; LIX precursor (U2727), a lipopolysaccharide response gene possibly involved in recruitment of inflammatory cells; SOCS-3 (U88328), suppressor of cytokine signaling-3 protein; MIP (M58004), a macrophage inflammatory protein related to MIP-1; CCR5 (AF022990), CC chemokine receptor-5, a coreceptor for HIV-1; T-cell receptor (L2G25B), similar to macrophage inflammatory protein-1 alpha precursor; and angiogenin (U22516), a blood vessel-inducing protein.
Figure 7. GeneChip analysis of effects of cuprizone (exposure for 6 weeks, followed by recovery) on steady-state mRNA levels of insulin-like growth factor-1 (IGF-1) and on other genes showing temporal patterns of mRNA expression similar to IGF-1. Data were analyzed and plotted as described in Fig. 4.
Download figure to PowerPoint
Some of the examples above have been examined for function, such as MIP-1α, or are currently being examined (McMahon et al. 2001).