Suppressor of cytokine signaling 3 limits protectionof leukemia inhibitory factor receptor signalingagainst central demyelinationBen Emery*, Holly S. Cate†, Mark Marriott†, Tobias Merson†, Michele D. Binder†, Cameron Snell†, Pik Ying Soo†,Simon Murray†, Ben Croker‡, Jian-Guo Zhang‡, Warren S. Alexander‡, Helen Cooper§, Helmut Butzkueven†,and Trevor J. Kilpatrick†¶�

*Department of Neurobiology, Stanford University, Stanford, CA 94305; †Howard Florey Institute and ¶Centre for Neuroscience, University of Melbourne,Victoria 3010, Australia; ‡The Walter and Eliza Hall Institute, Parkville, Victoria 3150, Australia; and §Queensland Brain Institute, University of Queensland,Brisbane, Queensland 4072, Australia

Communicated by Donald Metcalf, The Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia, March 30, 2006 (received for reviewMarch 3, 2006)

Enhancement of oligodendrocyte survival through activation ofleukemia inhibitory factor receptor (LIFR) signaling is a candidatetherapeutic strategy for demyelinating disease. However, in othercell types, LIFR signaling is under tight negative regulation by theintracellular protein suppressor of cytokine signaling 3 (SOCS3).We, therefore, postulated that deletion of the SOCS3 gene inoligodendrocytes would promote the beneficial effects of LIFRsignaling in limiting demyelination. By studying wild-type andLIF-knockout mice, we established that SOCS3 expression by oli-godendrocytes was induced by the demyelinative insult, that thisinduction depended on LIF, and that endogenously produced LIFwas likely to be a key determinant of the CNS response tooligodendrocyte loss. Compared with wild-type controls, oligo-dendrocyte-specific SOCS3 conditional-knockout mice displayedenhanced c-fos activation and exogenous LIF-induced phosphory-lation of signal transducer and activator of transcription 3. More-over, these SOCS3-deficient mice were protected against cupri-zone-induced oligodendrocyte loss relative to wild-type animals.These results indicate that modulation of SOCS3 expression couldfacilitate the endogenous response to CNS injury.

oligodendrocyte � SOCS3 � cuprizone � signal transduction �conditional knockout

Demyelinating diseases of the CNS often involve oligoden-drocyte death (1). The in vitro viability of oligodendrocytes

is potentiated by several cytokines, including leukemia inhibitoryfactor (LIF), ciliary neurotrophic factor (CNTF), insulin-likegrowth factor I (IGF-I), and neurotrophin 3 (NT-3) (2, 3). Thesefindings suggest therapeutic approaches, and recent work indi-cates that signaling through the LIF receptor (LIFR) complex byLIF or CNTF ameliorates demyelination in experimental auto-immune encephalomyelitis (4, 5). This effect can be induced byexogenous LIF but also probably represents an innate responsebecause expression of LIF and CNTF is induced during inflam-matory demyelination and CNS injury (4, 6). Both LIF andCNTF signal through the LIFR, which comprises LIFR� andglycoprotein 130 (gp130); CNTF also requires the glycosylphos-phatidylinositol-anchored nonsignaling CNTF receptor to effectits binding (7).

Signaling through LIFR involves activation of the signal trans-ducers and activators of transcription (STATs) and is usuallyterminated rapidly by a negative feedback loop involving STAT-induced expression of the intracellular protein suppressor of cyto-kine signaling 3 (SOCS3), a member of a molecular family com-prising SOCS1–7 and cytokine-inducible SH2-containing protein(CIS) (8). In several systems, SOCS proteins function as negativeregulators, induced in response to cytokine stimulation as well asinhibiting further signaling by the same cytokine receptors. TheSOCS1 and SOCS3 proteins are highly related, both being impli-

cated in negative regulation of Janus kinase (JAK)–STAT signaling(9), the predominant pathway activated by LIFR. Expression of theSOCS3 gene is up-regulated by the STATs (10), and the SOCS3protein binds to phosphotyrosine-759 on gp130 (11).

The above results suggest that SOCS proteins could limit thebeneficial effects that LIFR signaling exerts to limit central demy-elination. However, SOCS3-knockouts are embryonically lethalbecause of placental deficits (12, 13). Here, we utilized a condi-tional-knockout approach in oligodendrocytes, and we establishedthat SOCS3 modulates cuprizone-induced demyelination.

ResultsOligodendrocytes Express SOCS3 in Response to LIFR Signaling in Vitroand During Demyelination. To address whether SOCS3 has a role inmodulating LIFR�gp130 signaling within oligodendrocytes, weutilized Northern blot analysis to assess expression of SOCS3 withinprimary oligodendrocytes isolated from rat optic nerves. TheSOCS3 mRNA was almost undetectable in unstimulated cells, butit was induced after stimulation by LIF. This induction was rapid,occurring within 30 min of stimulation (Fig. 1a). In addition, theSOCS3 protein was also present at very low levels in the CG-4 cellline derived from oligodendrocytes (14), but it was induced by LIFwithin 1 h (Fig. 1b). The mRNA for SOCS1 was not induced by LIFstimulation (data not shown), suggesting that SOCS3 is likely to bethe more relevant protein in modulating LIFR signaling withinoligodendrocytes.

Increased SOCS3 expression has been reported within the con-text of demyelination induced by the copper-binding chemicalcuprizone (15), but the cells responsible for this expression have notbeen identified. We, therefore, assessed the expression profile ofSOCS3 by using immunofluorescence in mice treated with cupri-zone for 10 days. This early time point was chosen because it hasbeen reported that there is a strong down-regulation of myelingenes but relatively little oligodendrocyte loss at this time point(15). In control animals, expression of SOCS3 was restricted toneurons, as described in refs. 16 and 17; this neuronal expressionwas not obviously altered in response to cuprizone. However, thewhite matter of animals exposed to cuprizone exhibited a greaternumber of SOCS3-positive cells than unchallenged controls (Fig.1c). The majority of these cells were oligodendrocytes, as shown bytheir expression of the oligodendrocyte-specific marker CC1. The

Conflict of interest statement: No conflicts declared.

Abbreviations: CNTF, ciliary neurotrophic factor; IGF-I, insulin-like growth factor I; LCM,laser capture microdissection; LFB, luxol fast blue; LIF, leukemia inhibitory factor; LIFR,leukemia inhibitory factor receptor; MBP, myelin basic protein; MSA, murine serum albu-min; NT-3, neurotrophin 3; PAS, periodic acid/Schiff; SOCS, suppressor of cytokine signaling;STAT, signal transducer and activator of transcription.

�To whom correspondence should be addressed. E-mail: t.kilpatrick@hfi.unimelb.edu.au.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0602574103 PNAS � May 16, 2006 � vol. 103 � no. 20 � 7859–7864

NEU

ROSC

IEN

CE

increased oligodendroglial SOCS3 expression extended to white-matter tracts outside the corpus callosum such as the optic nervesand stria medullaris of the thalamus, indicating that cuprizoneadministration causes early reactive changes, even in regions that donot frankly demyelinate.

We next explored whether either LIF or CNTF was responsiblefor the cuprizone-mediated induction of SOCS3. RT-PCR wasperformed on samples obtained from the medial corpus callosumby laser capture microdissection (LCM). After 10 days of cuprizone,the expression of the SOCS3, LIF, and CNTF genes was inducedwithin the corpus callosum of wild-type mice. Two CNTF�/��LIF�/� mice subjected to the same experimental paradigm dis-played absent CNTF and reduced LIF gene expression, as ex-pected. Expression of the SOCS3 gene was also reduced in theCNTF�/��LIF�/� mutant mice relative to wild-type mice, confirm-ing that its induction is downstream from CNTF and�or LIF.Furthermore, there was almost no induction of the SOCS3 gene in

two CNTF�/��LIF�/� mice challenged with cuprizone (as shownfor one of the mice in Fig. 1d), strongly suggesting that LIF is thepredominant inducer of SOCS3 in this system.

LIF Is an Endogenous Survival Factor for Oligodendrocytes. We nextdetermined whether LIF was also the principal cytokine responsiblefor limiting oligodendroglial loss in the context of cuprizone-induced injury by analyzing wild-type and LIF-knockout mice.Although it has been reported that female LIF�/� mice display mildreductions of MBP expression within restricted regions of the brain(18), assessment of the density of CC1-positive oligodendrocyteswithin the corpus callosi of three unchallenged LIF�/� mice andthree wild-type controls revealed no statistically significant differ-ence between the two genotypes (P � 0.05; Fig. 2 and Table 1), nordid the density of oligodendrocytes within the corpus callosumappear to be obviously affected by gender among the knockouts(data not shown). In contrast, after 3 weeks of cuprizone, a timepoint at which wild-type mice usually display only moderate demy-elination, a cohort of four LIF�/� mice displayed an 84.8% reduc-tion in oligodendrocytes (compared with 36.2% loss in four wild-types mice, P � 0.01), a finding corroborated in a second,independent, cohort of mice (data not shown).

Generation of Oligodendrocyte SOCS3 Conditional Knockouts. Ournext goal was to assess whether the SOCS3 expression induced inoligodendrocytes limited the benefits of endogenous cytokinesignaling in the context of central demyelination. To address thisquestion, SOCS3 was deleted from myelinating oligodendrocytes bycrossing mice in which the coding region of SOCS3 is loxP-flanked[SOCS3FL/FL (19)] with transgenic mice expressing Cre recombi-nase from a 1.96-kb fragment of the MBP promoter (20). CremRNA was highly expressed within the spinal cord and hindbrainbut not detected in other organs, indicating that the expression ofthe transgene mimicked that of the endogenous promoter (Fig. 3a).Southern blot analysis of genomic DNA isolated from the CNS and

Fig. 1. SOCS3 is induced by LIFR signaling by oligodendrocytes in vitro andin vivo in the context of cuprizone-mediated demyelination. (a) Northern blotanalysis of SOCS3 expression in primary rat oligodendrocytes treated with LIF.(b) Western blot analysis of SOCS3 expression in CG-4-derived oligodendro-cytes treated with LIF. (c) Immunostaining for CC1 and SOCS3 in the whitematter of an unchallenged mouse and a mouse given 0.2% cuprizone for 10days. Merged images include DAPI nuclear stain (blue). (Scale bar, 25 �m.) (d)RT-PCR analysis of SOCS3, myelin basic protein (MBP), LIF, and CNTF in samplesisolated from the corpus callosum by laser capture microdissection (LCM). Micewere either unchallenged or given cuprizone for 10 days, and they were wildtype or mutant for the LIF and CNTF genes as indicated. Template loadinguniformity between samples was assessed by amplification of GAPDH.

Fig. 2. Deletion of LIF increases the severity of cuprizone-mediated oligo-dendrocyte loss. Representative frozen sections within the medial corpuscallosum of LIF�/� mice and wild-type controls were stained for the oligoden-drocyte marker CC1. Mice were either unchallenged or given cuprizone for 3weeks. Arrows indicate examples of CC1-positive cells. (Scale bar, 25 �m.)

Table 1. LIF deficiency accentuates cuprizone-mediated loss ofCC1-positive oligodendrocytes

Genotype Unchallenged 3 wk cuprizone

LIF��� 2,102 � 210 1,342 � 225LIF��� 2,057 � 285 312 � 167**

Numbers represent cells per mm2, � SD. **, P � 0.01 vs. WT mice.

7860 � www.pnas.org�cgi�doi�10.1073�pnas.0602574103 Emery et al.

other organs indicated that Cre-mediated recombination of theloxP-flanked SOCS3 allele was restricted to the CNS (cortex andhindbrain, Fig. 3b). To establish the spatial profile of recombination

of SOCS3 within the brain, LCM was utilized to isolate and analyzesamples from the corpus callosum and cerebral cortex of aSOCS3FL/FL and a SOCS3�MBP/�MBP mouse. RT-PCR analysis ofsamples from the SOCS3�MBP/�MBP mouse revealed stronger ex-pression of MBP and Cre mRNA within the corpus callosumrelative to cortex (Fig. 3c). The recombined (deleted) allele ofSOCS3 was also readily detectable by PCR from corpus callosalsamples, whereas only the unrecombined (floxed) allele was de-tectable from cortical samples and samples obtained from aSOCS3FL/FL animal (Fig. 3d), indicating that the MBP–Cre trans-gene effects the deletion of SOCS3 within the brain in a predom-inantly white matter-specific distribution. In contrast to wild-typemice (Fig. 1c), no colocalization was seen between SOCS3 and CC1in a SOCS3�MBP/�MBP mouse treated with cuprizone (data notshown).

Molecular Consequences of Oligodendrocyte-Specific Deletion ofSOCS3. To determine how the deletion of SOCS3 influenced keysignaling events, cuprizone was administered to SOCS3WT/WT andSOCS3�MBP/�MBP mice for 10 days. The responsiveness ofSOCS3WT/WT and SOCS3�MBP/�MBP mice to LIF was first assessedat the level of STAT3 activation after a single injection of LIF.Wild-type mice displayed a modest phosphorylation of STAT3within the hindbrain 30 min after receiving LIF, which was signif-icantly reduced by 3 h after LIF stimulation. In contrast, theSOCS3�MBP/�MBP mice displayed a LIF-induced phosphorylation ofSTAT3 that was heightened relative to wild-type mice (Fig. 4a, onemouse per sample, blot representative of two independent exper-iments). Next, we compared the induction of a subset of LIF-responsive target genes in the corpus callosum of SOCS3WT/WT andSOCS3�MBP/�MBP mice that were given cuprizone for 10 days andtreated with LIF or carrier protein [murine serum albumin (MSA)]only. Analysis with RT-PCR revealed that wild-type mice displayeda strong induction of SOCS3 mRNA 1 h after exogenous LIFadministration, whereas, as expected, no such induction was seen inthe SOCS3�MBP/�MBP mice (Fig. 4b, one mouse per sample, gelrepresentative of two independent experiments). By utilizing RT-PCR (Fig. 4b) and real-time PCR, we also established that theimmediate-early gene c-fos was up-regulated in all mice treated withcuprizone for 10 days, presumably in response to locally producedcytokines. This induction was observed to be 3-fold greater in acohort of four SOCS3�MBP/�MBP mice relative to four wild-typemice (14.2 � 3.1-fold vs. 4.6 � 1.6-fold; P � 0.05), but, unlike theabove-described pattern of potentiation of phosphorylation ofSTAT3, the expression of c-fos was not significantly altered by theadministration of exogenous LIF (14.2 � 3.1-fold without LIF vs.8.1 � 1.6-fold with LIF; P � 0.15) in SOCS3�MBP/�MBP micechallenged with cuprizone. Immunofluorescence analysis of c-Fosexpression within the corpus callosum confirmed the real-timePCR expression pattern, and it also confirmed that the collosalc-Fos expression in SOCS3�MBP/�MBP mice was found predomi-

Fig. 3. Generation and confirmation of conditional knockout of SOCS3within oligodendrocytes. (a) Northern blot analysis of Cre expression in MBP–Cre transgenic mice. (b) BamHI digestion and Southern blot analysis of DNAfrom the CNS and immune organs of a SOCS3FL/FL and a SOCS3�MBP/�MBP mousedistinguish between the loxP-flanked (SOCS3FL, 9 kb) and excised allele(SOCS3DEL, 4.9 kb). The image shown is representative of three mice examinedfor each genotype. (c) RT-PCR analysis of genomic DNA isolated from the LCMsamples [cortex (Cx) or white matter (WM)] of a SOCS3�MBP/�MBP mouse,showing white-matter predominance of Cre and MBP gene expression. (d) PCRanalysis of genomic DNA isolated from the LCM samples (cortex or whitematter) of a SOCS3FL/FL and a SOCS3�MBP/�MBP mouse, distinguishing betweenthe SOCS3 loxP-flanked (FL) and excised (DEL) alleles.

Fig. 4. Oligodendrocyte SOCS3 conditional knockouts display heightened sensitivity to LIFR signaling in the context of cuprizone-mediated demyelination. (a)LIF-induced tyrosine phosphorylation of STAT3 in hindbrains from SOCS3WT/WT and SOCS3�MBP/�MBP mice after 10 days of receiving cuprizone. (b) RT-PCR analysis ofSOCS3 and c-Fos in LCM-isolated sections from the corpus callosum of SOCS3WT/WT and SOCS3�MBP/�MBP mice. Mice were either unchallenged or given cuprizone for 10days, followed by an injection of either LIF or mouse serum albumin (MSA) 1 h before being killed. Template loading was assessed by amplification of GAPDH.

Emery et al. PNAS � May 16, 2006 � vol. 103 � no. 20 � 7861

NEU

ROSC

IEN

CE

nantly in cells arranged in linear arrays, typical of oligodendrocytes(data not shown).

Deletion of SOCS3 Inhibits Oligodendrocyte Loss in the Context ofDemyelination. We next assessed the effects of SOCS3 on oligo-dendrocyte numbers in cuprizone-mediated demyelination. Weexposed four SOCS3WT/WT and four SOCS3�MBP/�MBP mice tocuprizone for 4 weeks, a time at which wild-type mice displaywidespread loss of oligodendrocytes within the medial corpuscallosum (21, 22). After 4 weeks of receiving cuprizone, thewild-type mice displayed almost complete loss of CC1-positiveoligodendrocytes (Fig. 5a and Table 2; P � 0.001). Although the

SOCS3�MBP/�MBP mice displayed some oligodendrocyte loss withinthe corpus callosum in response to cuprizone (P � 0.05), this losswas attenuated by 49% compared with wild-type mice (P � 0.05).Diminished oligodendrocyte loss in SOCS3�MBP/�MBP mice was alsoconfirmed in a second, independent, cohort of mice.

To assess whether this reduction in oligodendrocyte loss led to apreservation of myelin, an independent cohort of mice was assessedfor luxol fast blue (LFB) staining of myelin after 4 weeks ofcuprizone administration (Fig. 5b). Although eight wild-type micechallenged in this way displayed severe reductions in myelin levelswithin the corpus callosum relative to unchallenged controls, thisreduction was attenuated by 18% in eight SOCS3�MBP/�MBP micewith the IMAGE analysis program (National Institutes of Health),although not reaching statistically significant levels (relative myelinlevels compared with unchallenged controls: wild-type 40.7% �7.1%, SOCS3�MBP/�MBP 47.5% � 9.8%; P � 0.05).

Deletion of SOCS3 Promotes the Capacity of Exogenous LIF to Ame-liorate Myelin Loss. Groups of five wild-type and fiveSOCS3�MBP/�MBP mice were treated daily with LIF or MSAbetween days 7 and 28 in the context of a 28-day course ofcuprizone to ascertain whether exogenous LIF could furtherreduce oligodendrocyte loss. Neither group treated with LIFdisplayed any significant reduction in the loss of oligodendro-cytes relative to controls of the corresponding genotype thatwere treated with carrier protein (MSA) only (P � 0.32 and0.76, respectively); however, LIF-treated SOCS3�MBP/�MBP

mice still displayed significantly higher densities of oligoden-drocytes than their LIF-treated wild-type counterparts (P �0.05) (Fig. 5a and Table 2). On the other hand, when myeli-nation was assessed in an independent group of 10 LIF and 8vehicle-treated SOCS3�MBP/�MBP mice by analysis of LFBstaining density with the IMAGE program, there was a signif-icant rescue of myelin loss in the LIF-treated animals (relativemyelin levels compared with unchallenged SOCS3�MBP/�MBP

mice: vehicle alone 47.5% � 9.8%, LIF treatment 90.1% �5.9%; P � 0.05). No such statistically significant benefit of LIFwas seen in wild-type animals (Fig. 5b).

DiscussionOur data establish that activation of the LIFR complex aftercuprizone-induced injury limits oligodendrocyte loss and in-duces the inhibitor SOCS3 in these cells. Deletion of SOCS3further reduces oligodendrocyte loss and provides a substratethat enables exogenous LIF to protect against demyelination.

Although a number of cytokine systems influence the remyeli-nation phase after cuprizone-mediated demyelination, includingTNF-�, IGF-I, FGF-2, and IL-1� (22–25), less is known about thecytokines that modulate the initial loss of oligodendrocytes. Al-though IGF-I reduces the loss of oligodendrocytes when transgeni-cally overexpressed, mice lacking the IGF-I receptor within oligo-dendrocytes fail to show any acceleration in cuprizone-mediatedloss of oligodendrocytes, displaying deficiencies confined to thesubsequent remyelination phase (21, 25). These findings, in com-bination with our observations that the loss of oligodendrocytes isexacerbated in LIF�/� mice, suggest that LIFR signaling is animportant endogenous mechanism for limiting oligodendroglialinjury and loss in vivo. Moreover, given that both LIF and CNTFwere induced within the corpus callosum in response to cuprizone,it is likely that cytokines signaling through the LIFR complex willshow a high degree of redundancy.

A number of observations suggest that SOCS3 is a negativeregulator of LIF signaling in cuprizone-induced demyelination. Notonly is the expression of both LIF and SOCS3 induced by cuprizone,but also the induction of SOCS3 is almost completely abrogated inLIF�/� mice. Deletion of the LIF and SOCS3 genes also results inopposing phenotypes; LIF�/� mice display accelerated loss,whereas SOCS3�MBP/�MBP mice display enhanced oligodendrocyte

Fig. 5. Oligodendrocyte SOCS3 conditional knockouts are protected againstcuprizone-mediated demyelination. (a) Representative frozen sections withinthe medial corpus callosum of SOCS3WT/WT and SOCS3�MBP/�MBP mice stainedfor the oligodendrocyte marker CC1. Mice were either unchallenged or givencuprizone for 4 weeks together with daily treatment with either LIF or MSAfrom days 7 to 28. Arrows indicate examples of CC1-positive cells, which aremore numerous in SOCS3�MBP/�MBP mice compared with wild-type mice after4 weeks of cuprizone, whether they were treated with LIF or vehicle (MSA).(Scale bar, 25 �m.) (b) Representative LFB–PAS-stained paraffin sections froman unchallenged mouse and cuprizone-challenged SOCS3WT/WT andSOCS3�MBP/�MBP mice treated daily with either LIF or MSA from days 7 to 28.The medial corpus callosi are indicated by arrows, demonstrating the mostovert demyelination, as determined by loss of LFB staining (blue) and uncov-ering of PAS staining (red) in wild-type mice challenged with cuprizone, withpartial protection in SOCS3�MBP/�MBP mice receiving vehicle and almost com-plete protection in SOCS3�MBP/�MBP mice receiving LIF. (Scale bar, 200 �m.)

Table 2. SOCS3 deficiency protects CC1-positiveoligodendrocytes independent of exogenous LIF

Genotype Unchallenged

4 wk cuprizone

MSA LIF

WT�WT 2,305 � 140 66 � 32 178 � 218�MBP��MBP 1,976 � 133 1,000 � 534* 882 � 572*

Numbers represent cells per mm2, � SD. *, P � 0.05 vs. WT mice.

7862 � www.pnas.org�cgi�doi�10.1073�pnas.0602574103 Emery et al.

survival with cuprizone. Furthermore, the phenotype displayed bythe SOCS3�MBP/�MBP mice coincided with increased expression ofthe LIF-responsive immediate-early gene c-fos and potentiatedactivation of STAT3. Interestingly, c-fos was selectively enhanced inthose cuprizone-challenged SOCS3�MBP/�MBP mice that were notexposed to exogenous LIF, whereas STAT3 phosphorylation wasenhanced by the administration of LIF; whether this apparentdissociation accounts for composite concentration-dependent ef-fects that LIF might exert on oligodendrocyte biology awaits furtherinvestigation. It is also possible that the beneficial effects of SOCS3deficiency could have been mediated by enhanced signaling of othercytokines. However, the known cytokine responsiveness of oligo-dendrocytes is fairly limited, being restricted to the LIF family,IGF-I, NT-3, and neuregulin. Neither NT-3 nor neuregulin isreported to interact with SOCS3, and, as indicated above, thebeneficial effects of IGF-I in this context appear restricted to theremyelination phase.

Perhaps unexpectedly, administration of exogenous LIF to wild-type mice failed to reduce cuprizone-mediated demyelination sig-nificantly, as measured by either oligodendrocyte density or LFBstaining. Given that this exogenous administration resulted in afurther increase in SOCS3 but not c-fos mRNA levels within thecorpus callosum (Fig. 4), it seems likely that the LIF-inducedSOCS3 was rapidly and effectively inhibiting any further signalingthrough the LIFR complex. Other investigators have also reportedthat attempts to enhance LIF signaling within the heart by over-expressing the gp130 receptor subunit resulted in heightenedSOCS3 induction in response to LIF but without other LIF-responsive genes being similarly increased (26). These findingsestablish the difficulty of potentiating LIFR signaling past endog-enous levels in the presence of SOCS3.

Provision of exogenous LIF to SOCS3�MBP/�MBP mice did notaffect oligodendrocyte numbers, which implies that some oligoden-drocyte loss is driven by non-LIF-dependent mechanisms but thatthis residual population can still be induced to limit demyelinationby exogenous LIF in the absence of the negative regulatory activityof SOCS3. In combination, these results suggest that the modula-tion of SOCS3 and enhancement of LIFR signaling provide excitingpotential therapeutic strategies for the treatment of central demy-elinating disease.

Materials and MethodsAnimal Resources. The SOCS3 loxP-flanked strain (line 350,C57BL�6 background) was generated as described in ref. 19. TheMBP–Cre lines were generated in conjunction with F. Koentgen ofOzgene (Bently, Western Australia, Australia) from a plasmidkindly provided by Alex Gow (Wayne State University School ofMedicine, Detroit) by microinjection into the male pronucleus offertilized C57BL�6 oocytes. Homozygous loxP-flanked SOCS3mice positive for the MBP–Cre transgene (SOCS3�MBP/�MBP) wereobtained by crossing MBP–Cre mice onto the 350 line for twogenerations. CNTF�/��LIF�/� and CNTF�/��LIF�/� mice on theC57BL�6 background were obtained by crossing CNTF�/��LIF�/�

mice that were a kind gift from M. Sendtner (Institute for ClinicalNeurobiology, University of Wurzburg). LIF-knockout mice on amixed DBA2 and C57BL�6 background were a kind gift fromPhillipe Brulet (Institut Pasteur, Paris); oligodendrocyte countsfrom these animals were compared with those from wild-typelittermates. Oligodendrocyte purifications were performed onSprague–Dawley rat pups (postnatal day 7) from Monash Univer-sity (Melbourne, Australia). Experiments were conducted in ac-cordance with guidelines of the Howard Florey Animal EthicsCommittee.

Induction of Demyelination. Cuprizone-mediated demyelination wasinduced by feeding mice at 6–10 weeks of age a diet of powderedfeed (Barastoc; Ridley AgriProducts, Pakenham, Victoria) contain-ing 0.2% wt�wt cuprizone (Sigma) for up to 4 weeks.

LIF Treatment. Mice were injected s.c. with LIF daily (Amrad,Melbourne, Australia, 25 �g�kg of body weight) made up in 100 �lof 0.1% MSA (Sigma) or an equivalent dose of 0.1% MSA alone.For assessment of LIF responsiveness at day 10 of cuprizonetreatment, mice received i.v. injections of 2 �g of LIF in 100 �l of0.1% MSA at specified times before being killed.

Cell Culture. Oligodendrocyte precursors were purified from opticnerves of 7-day-old rats and were expanded in SATO medium with10 ng�ml platelet-derived growth factor and NT-3 (PeproTech,Rocky Hill, NJ) as described in ref. 27. Cells were induced todifferentiate by the addition of thyroid hormone for 48 h andtreated with 100 ng�ml LIF for 30 and 60 min. The bipotential glialcell line CG-4 (kind gift from O. Bogler, Center for MolecularGenetics, University of California, San Diego) was cultured inSATO medium with 30% B104 conditioned medium as reported inref. 14 and induced to differentiate for 48 h by withdrawal ofconditioned medium before cytokine treatment.

Histology. For fluorescence microscopy, 10-�m cryostat sectionswere probed with primary antibodies at 1:200 followed by detectionwith appropriate fluorochrome-conjugated secondary antibodies(Molecular Probes, 1:500). Primary antibodies used were APC-7�CC1 (Oncogene Science) and rabbit polyclonal anti-SOCS3 (pro-duced by J.-G.Z.). Paraffin sections (10 �m) were used for assess-ment of demyelination with LFB–periodic acid/Schiff (PAS)reagent.

Oligodendrocyte Density Analysis. Three 10-�m coronal cryostatsections of the brain between bregma �0.58 mm and �0.9 mmat least 50 �m apart were taken from each animal. Sections werestained with DAPI and the CC1 antibody and photographed ata magnification of �200 at the midline of the corpus callosum(corresponding to the medial 425 �m of each section). Thenumber of CC1-immunopositive cells with DAPI-positive nucleiwithin the corpus callosum was counted for each image, and thearea of the corpus callosum was determined in square millime-ters with the IMAGE 1.63 program (National Institutes of Health).All counts and analyses were performed blind to the genotypeand treatment of the animal.

Analysis of LFB Staining. LFB–PAS-stained paraffin sections wereselected for each animal at as close to bregma �0.75 mm as theseries of sections would allow. Sections were photographed with a�10 objective, and the images were imported into PHOTOSHOP(Adobe Systems, San Jose, CA). Within PHOTOSHOP, the bluechannel of each image was selected and exported as a gray scaleimage into IMAGE 1.63, which was used to take a density reading ofthe LFB staining within the central 870 �m of the corpus callosum.Density measurements for each mouse were normalized againstunchallenged corpus callosum optical density values by using thefollowing formula: transformed value � (density reading�unchallenged density average) � 100. All analysis was performedblind to the genotype and treatment of the animal.

Western Blots. Hindbrains were lysed in radioimmunoprecipitationassay buffer with protease inhibitors (28) and 100 �g of protein runon 10% Tris–glycine gels (Invitrogen), and they were transferred topoly(vinylidene difluoride) membrane (Pall). Membranes wereassayed for STAT3 or phospho-STAT3 as described in ref. 28.

Northern Blot Analysis. RNA samples were generated from cul-tured oligodendrocytes or mouse organs with Miniprep andMidiprep kits, respectively (Qiagen, Valencia, CA) as per themanufacturer’s instructions. Northern blotting was performedaccording to standard methodology with probes generated fromthe 3 UTR region of SOCS3 (plasmid kindly supplied by LynnHartley, The Walter and Eliza Hall Institute).

Emery et al. PNAS � May 16, 2006 � vol. 103 � no. 20 � 7863

NEU

ROSC

IEN

CE

LCM. Frozen sections (16 �m) of unfixed brains were collected onmembrane-coated slides (PALM, Zeiss) and immersed in ace-tone at �20°C for 5 min. The slides were air-dried at roomtemperature for 10 min, stained in 1% cresyl violet�95% vol/volethanol for 1 min, and rinsed through three changes of 100%ethanol for 2 min each. The slides were air-dried, and laser-dissected tissue (typically 5 � 105 �m2 taken from 5–10 sections)was isolated from either the medial corpus callosum or thelateral cerebral cortex with the PALM MicroBeam MicroLaserSystem and PALM ROBOSOFTWARE (Millennium Science, Sur-rey Hills, Victoria, Australia). Tissue was collected into the capof a 200-�l tube containing 5 �l of either DNA isolation buffer[1 mM EDTA�20 mM Tris, pH 8.0�0.5% Igepal CA-630 (Ste-pan, Northfield, IL)�200 �g/ml proteinase K] or PicoPure RNAIsolation kit extraction buffer (Arcturus, Mountain View, CA)for RNA isolation.

PCR. Laser-capture DNA samples were amplified by using prim-ers that distinguish among the wild-type, loxP-f lanked, anddeleted SOCS3 alleles. Primer sequences were as follows: for-ward, 5-TCTTGTGTCTCTCCCCATCC-3; coding reverse, 5-TGACGCTCAACGTGAAGAAG-3; and 3 UTR reverse, 5-ACGTCTGTGATGCTTTGCTG-3.

RT-PCR. RNA from LCM samples was isolated by using the Pico-Pure RNA Isolation kit as per the manufacturer’s instructions.RNA (10 ng) was reverse-transcribed, and genes of interest wereamplified for 30 cycles by using standard procedures. Primers usedwere as follows: GAPDH forward, 5-ACCACAGTCCATC-CCATCAC-3, and reverse, 5-TCCACCACCCTGTTGCT-GTA-3 (29); MBP forward, 5-ATGGCATCACAGAAGAGA-3, and reverse, 5-CATGGGAGATCCAGAGCG-3 (GenBankaccession no. L00398); SOCS3 forward, 5-ACCAGCGCCACT-TCTTCACG-3, and reverse, 5-GTGGAGCATCATACT-

GATCC-3 (30); c-fos forward, 5-GGCTCTCCTGTCAACA-CACA-3, and reverse, 5-CCGCTTGGAGTGTATCTGTC-3(31); LIF forward, 5-CGCCTAACATGACAGACTTCCCAT-3,and reverse, 5-AGGCCCCTCATGACGTCTATAGTA-3 (32);CNTF forward, 5-GGGGATGGCTTTCGCAGA-3, and re-verse, 5-GGTACGGTAAGCCTGGAGGT-3 (National Centerfor Biotechnology Information accession no. NM�053007).

Quantitative PCR for c-fos. Brains from day-10 cuprizone-treatedmice were collected in RNAlater (Ambion, Austin, TX) andstored for 2 weeks at 4°C. The corpus callosum was microdis-sected from the brain of each animal, and RNA was isolated withthe PicoPure RNA Isolation kit according to the manufacturer’sinstructions. RNA (8.5 ng) was reverse-transcribed, and quan-titative PCR for 18 S and c-fos RNAs was performed on aSequence Detector 7700 (Applied Biosystems) using SYBRgreen chemistry (Applied Biosystems) according to the manu-facturer’s instructions. PCRs were analyzed with the SEQUENCEDETECTION SYSTEM software (Applied Biosystems), with resultsshown as the relative expression from cohorts of wild-type andSOCS3�MBP/�MBP mice by using the equivalent untreated controlas a baseline and accounting for 18S RNA levels. Primers wereas follows: 18 S forward, 5-CGGCTACCACATCCAAGGAA-3, and reverse, 5-GCTGGAATTACCGCGGCT-3; c-fos for-ward, 5-AGGAGGCCTTCACCCTGC-3, and reverse, 5-TGACTGGCTCCAAGGATGG-3.

Statistics. Data are expressed as the means � SD. Comparisonswere evaluated with an unpaired Student’s t test assumingunequal variances. *, P � 0.05; **, P � 0.01.

This work was supported by the National Health and Medical ResearchCouncil of Australia and by the National Multiple Sclerosis Society of theUnited States of America. Recombinant LIF was a generous gift fromthe Australian Medical Research and Development Corp.

1. Lucchinetti, C., Bruck, W., Parisi, J., Scheithauer, B., Rodriguez, M. &Lassmann, H. (2000) Ann. Neurol. 47, 707–717.

2. Kahn, M. A. & De Vellis, J. (1994) Glia 12, 87–98.3. Barres, B. A., Schmid, R., Sendnter, M. & Raff, M. C. (1993) Development

(Cambridge, U.K.) 118, 283–295.4. Butzkueven, H., Zhang, J. G., Soilu-Hanninen, M., Hochrein, H., Chionh, F.,

Shipham, K. A., Emery, B., Turnley, A. M., Petratos, S., Ernst, M., et al. (2002)Nat. Med. 8, 613–619.

5. Linker, R. A., Maurer, M., Gaupp, S., Martini, R., Holtmann, B., Giess, R.,Rieckmann, P., Lassmann, H., Toyka, K. V., Sendtner, M. & Gold, R. (2002)Nat. Med. 8, 620–624.

6. Sriram, K., Benkovic, S. A., Hebert, M. A., Miller, D. B. & O’Callaghan, J. P.(2004) J. Biol. Chem. 279, 19936–19947.

7. Stahl, N., Davis, S., Wong, V., Taga, T., Kishimoto, T., Ip, N. Y. & Yanco-poulos, G. D. (1993) J. Biol. Chem. 268, 7628–7631.

8. Krebs, D. L. & Hilton, D. J. (2001) Stem Cells 19, 378–387.9. Starr, R., Willson, T. A., Viney, E. M., Murray, L. J., Rayner, J. R., Jenkins,

B. J., Gonda, T. J., Alexander, W. S., Metcalf, D., Nicola, N. A. & Hilton, D. J.(1997) Nature 387, 917–921.

10. Auernhammer, C. J., Bousquet, C. & Melmed, S. (1999) Proc. Natl. Acad. Sci.USA 96, 6964–6969.

11. Nicholson, S. E., De Souza, D., Fabri, L. J., Corbin, J., Willson, T. A., Zhang,J. G., Silva, A., Asimakis, M., Farley, A., Nash, A. D., et al. (2000) Proc. Natl.Acad. Sci. USA 97, 6493–6498.

12. Roberts, A. W., Robb, L., Rakar, S., Hartley, L., Cluse, L., Nicola, N. A.,Metcalf, D., Hilton, D. J. & Alexander, W. S. (2001) Proc. Natl. Acad. Sci. USA98, 9324–9329.

13. Takahashi, Y., Carpino, N., Cross, J. C., Torres, M., Parganas, E. & Ihle, J. N.(2003) EMBO J. 22, 372–384.

14. Louis, J. C., Magal, E., Muir, D., Manthorpe, M. & Varon, S. (1992) J. Neurosci.Res. 31, 193–204.

15. Jurevics, H., Largent, C., Hostettler, J., Sammond, D. W., Matsushima, G. K.,Kleindienst, A., Toews, A. D. & Morell, P. (2002) J. Neurochem. 82, 126–136.

16. Mori, H., Hanada, R., Hanada, T., Aki, D., Mashima, R., Nishinakamura, H.,Torisu, T., Chien, K. R., Yasukawa, H. & Yoshimura, A. (2004) Nat. Med. 10,739–743.

17. Polizzotto, M. N., Bartlett, P. F. & Turnley, A. M. (2000) J. Comp. Neurol. 423,348–358.

18. Bugga, L., Gadient, R. A., Kwan, K., Stewart, C. L. & Patterson, P. H. (1998)J. Neurobiol. 36, 509–524.

19. Croker, B. A., Krebs, D. L., Zhang, J. G., Wormald, S., Willson, T. A., Stanley,E. G., Robb, L., Greenhalgh, C. J., Forster, I., Clausen, B. E., et al. (2003) Nat.Immunol. 4, 540–545.

20. Gow, A., Friedrich, V. L., Jr., & Lazzarini, R. A. (1992) J. Cell Biol. 119,605–616.

21. Mason, J. L., Ye, P., Suzuki, K., D’Ercole, A. J. & Matsushima, G. K. (2000)J. Neurosci. 20, 5703–5708.

22. Arnett, H. A., Mason, J., Marino, M., Suzuki, K., Matsushima, G. K. & Ting,J. P. (2001) Nat. Neurosci. 4, 1116–1122.

23. Gao, X., Gillig, T. A., Ye, P., D’Ercole, A. J., Matsushima, G. K. & Popko, B.(2000) Mol. Cell. Neurosci. 16, 338–349.

24. Mason, J. L., Suzuki, K., Chaplin, D. D. & Matsushima, G. K. (2001)J. Neurosci. 21, 7046–7052.

25. Mason, J. L., Xuan, S., Dragatsis, I., Efstratiadis, A. & Goldman, J. E. (2003)J. Neurosci. 23, 7710–7718.

26. Tone, E., Kunisada, K., Kumanogoh, A., Negoro, S., Funamoto, M., Osugi, T.,Kishimoto, T. & Yamauchi-Takihara, K. (2000) Cytokine 12, 1512–1518.

27. Barres, B. A., Hart, I. K., Coles, H. S., Burne, J. F., Voyvodic, J. T., Richardson,W. D. & Raff, M. C. (1992) Cell 70, 31–46.

28. Brysha, M., Zhang, J. G., Bertolino, P., Corbin, J. E., Alexander, W. S.,Nicola, N. A., Hilton, D. J. & Starr, R. (2001) J. Biol. Chem. 276, 22086–22089.

29. Tokumoto, Y. M., Durand, B. & Raff, M. C. (1999) Dev. Biol. 213,327–339.

30. Bjorbaek, C., Elmquist, J. K., El-Haschimi, K., Kelly, J., Ahima, R. S., Hileman,S. & Flier, J. S. (1999) Endocrinology 140, 2035–2043.

31. Duval, D., Reinhardt, B., Kedinger, C. & Boeuf, H. (2000) FASEB J. 14,1577–1584.

32. Chesnokova, V., Auernhammer, C. J. & Melmed, S. (1998) Endocrinology 139,2209–2216.

7864 � www.pnas.org�cgi�doi�10.1073�pnas.0602574103 Emery et al.