1142 VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 nature neurOSCIenCe

a r t I C l e S

In the CNS, immune functions are coordinated by specialized tissue macrophages called microglia1 that reside in the parenchyma and perivascular spaces, which monitor the tissue for injury or pathologi­cal changes2,3. Alterations to CNS homeostasis leads to microgliosis, a process that is characterized by change in microglia morphology, cell surface antigen expression, proliferation and modulation of inter­actions with other immune cells4.

Several reports have proposed that bone marrow–derived progeni­tors contribute to both microgliosis and microglia homeostasis, on the basis of experiments in which animals were lethally irradiated and their bone marrow was replaced with genetically labeled cells5–8. Our own recent work suggests that these reports must be interpreted with caution, as intravenous delivery of bone marrow obtained by mechan­ical collection leads to altered dynamics of cell migration across the blood brain barrier (BBB)9. Using parabiosis, a surgical procedure that creates peripheral blood chimeras between two animals without the forced mobilization of bone marrow progenitors, we recently found that microglia are maintained through in situ self­renewal rather than recruitment of circulating precursors to the CNS9. Furthermore, we found that expansion of resident microglia, in the absence of blood­borne cell recruitment, is responsible for the microgliosis that ensues in response to denervation and neurodegeneration9. A recent study confirmed that bone marrow–derived cells do not enter the healthy CNS in the adult animal and found that microglia and blood­derived monocytes have distinct embryonic origins, with microglia seeding the CNS early during embryonic development in the absence of a contribution from adult hematopoietic stem cells10. The active exclu­sion of blood­derived monocytes from the CNS during normal homeo­stasis, as well as in some neurodegenerative diseases, suggests that their role is fundamentally different from that of microglia and that,

when mechanisms excluding blood­derived monocytes from the CNS parenchyma fail, monocyte recruitment into the CNS likely heralds a new phase of disease pathology.

EAE is a well­studied mouse model of the human disease multiple sclerosis and is characterized by extensive infiltration of the CNS by inflammatory cells11. Initiation of EAE involves the activation of myelin­specific TH1 or TH17 cells, which in turn trigger the expan­sion of resident microglia and the recruitment of blood­borne myelo­monocytic cells12,13. However, it is yet to be determined whether either of these events lead to further demyelination or, conversely, to neuroprotection by limiting T cell–induced damage14. In particular, the fundamental mechanisms leading to the distinct stages of relaps­ing and remitting disease, and the associated physical impairment, remain highly controversial.

Specific functional roles have been proposed for resident micro­glia and blood­derived monocytes or macrophages15,16. Experiments using bone marrow chimeras suggest that the activation of resident microglia may represent one of the initial steps in EAE pathogenesis, preceding and possibly triggering the infiltration of blood­derived cells17,18. Consistent with the idea that monocytes or macrophages are involved in the exacerbation of EAE, depletion of these cells using sil­ica dust or liposomes containing dichloromethylene diphosphonate19 induces a marked suppression of the clinical signs of EAE20,21. Similarly, blockade or genetic deletion of CCR2, a chemokine receptor that is known to be involved in monocyte and T cell trafficking, pre­vents severe disease and can lead to faster remission22–24. In addition, increases in circulating inflammatory monocytes have been shown to correlate with relapses in EAE mice25.

A key limitation of these transplantation experiments is that they lead to the artificial, long­term replacement of a substantial fraction of

1University of British Columbia, Biomedical Research Centre, Vancouver, British Columbia, Canada. 2Present address: Department of Biomedical Physiology and Kinesiology, Simon Fraser University, Burnaby, British Columbia, Canada, and Division of Neurology, Department of Medicine, Neuromuscular Disease Unit, Vancouver Hospital and Health Sciences Center, Vancouver, British Columbia, Canada. Correspondence should be addressed to F.M.V.R. (fabio@brc.ubc.ca).

Received 15 April; accepted 3 June; published online 31 July 2011; doi:10.1038/nn.2887

Infiltrating monocytes trigger EAE progression, but do not contribute to the resident microglia poolBahareh Ajami1, Jami L Bennett1, Charles Krieger1,2, Kelly M McNagny1 & Fabio M V Rossi1

In multiple sclerosis and the experimental autoimmune encephalitis (EAE) mouse model, two pools of morphologically indistinguishable phagocytic cells, microglia and inflammatory macrophages, accrue from proliferating resident precursors and recruitment of blood-borne progenitors, respectively. Whether these cell types are functionally equivalent is hotly debated, but is challenging to address experimentally. Using a combination of parabiosis and myeloablation to replace circulating progenitors without affecting CNS-resident microglia, we found a strong correlation between monocyte infiltration and progression to the paralytic stage of EAE. Inhibition of chemokine receptor–dependent recruitment of monocytes to the CNS blocked EAE progression, suggesting that these infiltrating cells are essential for pathogenesis. Finally, we found that, although microglia can enter the cell cycle and return to quiescence following remission, recruited monocytes vanish, and therefore do not ultimately contribute to the resident microglial pool. In conclusion, we identified two distinct subsets of myelomonocytic cells with distinct roles in neuroinflammation and disease progression.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature neurOSCIenCe VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 1143

a r t I C l e S

CNS microglia with donor bone marrow–derived cells26. The presence of these donor­derived cells, which can expand locally in response to disease, as well as the absence of a specific marker for distinguishing microglia from blood­borne monocytes, has made it challenging to precisely determine the kinetics and relative importance of the entry of circulating monocytes and expansion of endogenous microglia.

It has been reported that lethal irradiation substantially alters the integrity of the BBB27,28. A recent study in which the heads of recipi­ent mice were shielded during myeloablation supports the notion that irradiation is necessary for the engraftment of microglia precursors in a bone marrow transplant setting29. However, our previous work indicates that, although lethal irradiation is required for donor cells to reach the CNS, it is not sufficient by itself. Instead, collection of bone marrow cells via flushing or crushing donor bone and subsequent introduction into the blood circulation is required in conjunction with irradiation for engraftment of infiltrating cells to occur9.

We took advantage of this finding to establish a new experimental approach based on a combination of parabiosis and irradiation, which avoids the use of mechanically collected bone marrow, allowing for the efficient replacement of bone marrow and peripheral blood cells in the complete absence of CNS repopulation. We combined this approach with transgenic labels to follow the entry of blood­borne monocytes into the CNS and their subsequent fate at specific stages of EAE pro­gression and following remission. Our results reveal a dynamic inter­play between blood­derived myelomonocytic cells and microglia and strongly support a causal link between monocyte invasion and disease progression. Indeed, we found that specifically blocking the entry of circulating inflammatory monocytes into the CNS prevented progres­sion to severe disease. In addition, we found that the invasion of the CNS by blood­borne myelomonocytic cells is a transient event and that these cells do not provide a long­term contribution to the resident microglia pool. Our data identify the invasion of circulating mono­cytes into the CNS parenchyma as a major outcome­altering event in

EAE progression and suggest that therapeutic strategies specifically aimed at inhibiting the migration of monocytes, rather than that of leukocytes in general, would be efficacious.

RESULTSMonocyte infiltration correlates with EAE progressionTo evaluate the kinetics of inflammatory cell infiltration and endog­enous microglia expansion in EAE, we needed to develop a method for tracking circulating cells without affecting the CNS environment or resident cells. First, we induced peripheral blood chimerism by surgically joining two syngeneic mice, one of which ubiquitously expressed GFP30. These parabiotic mice established a rich anastomotic circulation, which quickly led to efficient peripheral chimerism9. We verified the presence of GFP­positive cells in the blood of the non­GFP partner 2 weeks after parabiosis, and EAE was induced in both part­ners by immunization with myelin peptides and adjuvant. Mice were monitored daily for development of clinical signs of EAE (n = 15). We collected the spinal cords of these mice 2 weeks after induction.

Consistent with their role in the pathogenesis of EAE, partner­derived GFP and CD4 double­positive T­lymphocytes were readily identified in the spinal cords of all of the mice (Supplementary Fig. 1). This confirmed that the extent of circulatory exchange in parabiotic pairs was sufficient for cells from one partner to reach the CNS of the other. A caveat of this model is that parabiosis places animals under increased levels of stress, a situation that is well known to dampen EAE progression31,32. Although EAE developed in all of the 15 parabiotic pairs that we examined, we found that, in ten of those pairs, EAE was initiated, but did not progress beyond clini­cal score 2 (hind limb weakness and loss of tail tone). Notably, in the mice in which disease did not progress to paralysis, we found no evidence of partner­derived macrophage (Iba­1 and GFP double positive) infiltration of the CNS, despite the activation of resident microglia (Supplementary Fig. 1). Thus, although disease initiation

GFP Iba-1 PECAM

Parabiosis (P)week 0

100

Per

iphe

ral b

lood

chi

mer

ism

(% G

FP

pos

itive

) 80

60

40

20

0

Num

ber

of c

ells

60

40

20

00 2 4

I

E

72.3

100 101 102

GFP103 104

C and S

Week post-parabiosis6 8 10 12

Irradiation (I)week 2

Chimerism (C)week 6

Separation (S)week 6

EAE induction (E)week 8

a

b c

d

Figure 1 Irradiation and separation of parabiotic mice leads to peripheral blood chimerism in the absence of donor cell entry into the CNS. (a) Schematic representation of the experimental strategy. Parabionts were generated by joining GFP-positive and GFP-negative C57BL/6 mice. The GFP-positive partner was shielded with lead 2 weeks after surgery and the GFP-negative C57BL/6 partner was subjected to a lethal dose of irradiation. Average blood chimerism at this time point before irradiation was 55.7%. During the following 4 weeks, GFP-positive cells migrate from the shielded partner to the irradiated one and reconstitute its bone marrow. We surgically separated the parabionts 4 weeks post-irradiation. EAE was induced in the chimeric mice 2 weeks after irradiation. Average blood chimerism at this time point was 79.6%. (b) Percentage of GFP-positive cells in the blood of GFP-negative partners. The data presented correspond to the steps in a. For each data point, n = 6. Error bars represent s.d. (c) Representative FACS plot showing the frequency of GFP-positive cells in the blood of a GFP-negative partner 4 weeks after lethal irradiation. (d) Representative image of the spinal cord from healthy irradiated-separated parabionts not subjected to EAE induction (n = 10 mice). Spinal cord sections were immunostained for the mature macrophage and microglia marker Iba-1 (red) and the endothelial marker PECAM (blue). GFP, which identifies partner-derived cells, is shown in green. The images shown are maximum intensity projections of stacks of confocal optical sections. Note that no infiltration of GFP-positive cell was detected in healthy spinal cord. Scale bar represents 500 µm.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

1144 VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 nature neurOSCIenCe

a r t I C l e S

leading to mild functional impairment can occur independently of blood­derived monocytes, a clear correlation exists between the pres­ence of inflammatory infiltrates and EAE progression.

To further explore the relationship between myelomonocytic infil­tration and disease progression, we modified our model so that mice would develop more severe disease. Parabiotic pairs of healthy GFP­negative and GFP­positive mice were initially established (Fig. 1a). We placed the GFP­positive partner into a lead container 2 weeks after the parabiosis surgery while leaving the GFP­negative partner unshielded. The pair was then irradiated at the same dosage used for myeloablation (Fig. 1a). An average of 79% of the cells in the periph­eral blood of GFP­negative myeloablated partners were GFP­positive by 4 weeks after irradiation (Fig. 1b,c). Consistent with our previous results and in contrast with data obtained using injected bone marrow cells, spontaneously migrating partner­derived cells did not colonize the CNS of irradiated animals, and the resulting chimeric mice had hematopoietic cells present throughout the body, but not CNS micro­glia, that were GFP positive (Fig. 1d)9.

We surgically separated the partners 4 weeks after irradiation and then allowed them to recover for 2 weeks (Fig. 1a). Separated mice retaining over 60% GFP­positive cells in their peripheral blood were identified and selected for EAE induction (Fig. 1a). Thereafter, blood chimerism was consistently found to be stable, regardless of disease induction (Fig. 1b). Following EAE induction, mice were monitored daily for development of clinical signs and a group of animals was killed at each of the four clinical scores, defined as follows: (1) loss of tail tone, (2) hind­limb weakness with gait abnormality, spasticity and/or ataxia, (3) severe hind­limb weakness and partial paralysis, and (4) complete hind­limb paralysis. This parabiosis­irradiation­separation strategy allowed susceptibility to EAE, with 23 of the 24 mice that were not killed for analysis at disease scores of 1 or 2 developing severe functional impairment.

The presence of infiltrating blood cells in the CNS parenchyma at each clinical score of EAE was measured by assessing the fre­quency of partner­derived GFP­positive cells in spinal cord lesions. Blood­derived cells were already evident in the CNS parenchyma of animals with a disease score of 1 and they increased in number as the disease progressed to a score of 4 (Fig. 2a–f, Supplementary Fig. 2 and Supplementary Table 1). The increase in infiltrating cell number correlated substantially with disease severity, rather than with the time lapsed from disease induction, suggesting that recruit­ment of blood­borne cells is linked to the tissue damage associated with EAE pathology, either as a cause or consequence (Fig. 2b and Supplementary Fig. 3).

Evaluation of lineage­specific markers revealed that most donor­derived cells from mice with a disease score of 1 were T­lymphocytes (Supplementary Fig. 4) and none were positive for myelomonocytic

markers (Fig. 3 and Supplementary Fig. 5a). Recruitment of inflammatory monocytes (CD11b positive, Iba­1 negative) was first observed when animals reached a disease score of 2 (Fig. 3a,g–j and Supplementary Fig. 5b) and cell numbers increased with disease progression to a score of 3 (Fig. 3k–n and Supplementary Fig. 5c). Notably, donor­derived CD11b­positive cells were initially found in proximity to the meninges, suggesting that the entry of monocytes into the CNS is not the result of a generalized disruption of the BBB (Fig. 2d and Supplementary Fig. 5b).

Between clinical scores of 3 and 4, most partner­derived GFP­positive cells expressed Iba­1 (Fig. 3b), leading to a substantial reduction in the number of CD11b­positive, Iba­1–negative cells found in the parenchyma and suggesting that infiltrating monocytes progressively differentiate into mature macrophages (Fig. 3a,b and Supplementary Table 1). The reduction in the number of CD11b­positive, Iba­1­negative parenchymal cells and the limited increase in GFP­positive cells between mice with disease scores of 3 and 4 (Fig. 3a and Supplementary Table 1) strongly suggest that, at this stage, the entry of new circulating monocytes into the CNS is greatly reduced compared to mice with disease scores of 2 and 3, despite the increase in functional impairment. In addition, the number of infil­trating macrophages correlated well with disease severity (R2 = 0.89), but poorly with the time interval between disease induction and tissue collection (R2 = 0.37; Supplementary Fig. 6). Thus, a strong correla­tion exists between the extent of CNS infiltration by myelomonocytic cells and EAE disease progression, consistent with that observed in the conjoined parabionts. Furthermore, most monocytes enter the CNS before the development of severe disease.

Microglia and monocytes accumulate with different kineticsMicroglia activation is believed to be important in the regulation of CNS inflammation33. To establish the temporal relationship between

01 2

EAE clinical score3 4

40P < 0.001

P < 0.001

P < 0.01

80

Num

ber

ofG

FP

-pos

itive

cel

ls 120

01 2EAE clinical score

R2 = 0.8646

3 4

40

80

Num

ber

ofG

FP

-pos

itive

cel

ls 120a

Score of 1 Score of 2

Score of 3 Score of 4

GFP Iba-1 PECAM GFP Iba-1 PECAM

GFP Iba-1 PECAM GFP Iba-1 PECAM

c d

e f

bFigure 2 Monocytic infiltration correlates with progression to paralytic stages of EAE. (a) Total number of GFP-positive cells at different disease scores. To collect this data, we counted two diagonally opposite quadrants in each of three sections from the spinal cords of six animals per disease stage. (b) There was a strong correlation between total number of GFP-positive cells and the progression of EAE disease. Data points represent individual animals (n = 6). (c–f) Distribution of GFP-positive blood-derived cells (green) and Iba-1–positive (red) microglia in the spinal cord at four clinical scores of EAE. Images are collages of maximum intensity confocal stacks projections showing a view of the entire spinal cord. Staining with endothelial marker PECAM (blue) illustrates the spatial relationship between blood-derived inflammatory cells (GFP positive) and blood vessels. Outlined areas are provided in higher magnification as well as separated channels in Supplementary Figure 2. Scale bars represent 500 µm.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature neurOSCIenCe VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 1145

a r t I C l e S

T­cell entry, microglia activation and infiltra­tion of circulating monocytes, we measured the incorporation of bromodeoxyuridine (BrdU) in Iba­1–positive cells of the spinal cord during EAE progression. No BrdU incorporation was observed in the absence of EAE induction. BrdU­positive microglia (Iba­1–positive, GFP­negative cells) were already evident at a disease score of 1, a time in which no monocytic infiltrate is present in the CNS, confirming that activation of resident microglia is an early event in EAE pathogenesis (Fig. 4, Supplementary Fig. 7a and Supplementary Table 1). However, BrdU incorporation in microglia in the absence of infiltrating (GFP positive) monocytes was also observed in three animals that were immunized, but did not develop disease signs (Supplementary Fig. 8), suggesting that, although microglia activation is required for EAE initiation17,18, it may not be sufficient to trigger severe disease.

A significant increase in the frequency of proliferating microglia (P = 4.41 × 10−20), as well as in the absolute number of Iba­1–positive resident cells (P = 2.2 × 10−19), was observed between mice with dis­ease scores of 2 and 3 (Fig. 4a,b and Supplementary Table 1). As the majority of infiltrating monocytes had not differentiated at these time points (Fig. 3a,b), the vast majority of Iba­1–positive cells in spinal cord were derived from resident microglia, whereas CD11b­positive, Iba­1–negative monocytes were derived exclusively from circulating pools.

In contrast, between disease scores of 3 and 4, the fraction of BrdU­positive microglia remained constant (Fig. 4a and Supplementary Table 1) and the resident microglia number declined significantly in mice with a disease score of 4 (P = 1.2 × 10−11), concomitant with the generation of Iba­1–positive macrophages from circulating precur­sors (Fig. 4b). To further investigate the causes of the decline in cell

number, we measured the rate of apoptosis in Iba­1–positive, GFP­negative cells by TUNEL staining. We observed a substantial increase in TUNEL­positive microglia at a disease score of 4, indicating that these cells were lost through apoptosis (Supplementary Fig. 9).

Blocking inflammatory infiltration prevents EAE progressionOur data establish a clear correlation between monocytic infiltra­tion and disease progression. To clarify whether the two are causally linked and whether the lack of disease progression might be a result of a lack of monocytic infiltration, we used the parabiosis­irradia­tion­separation strategy to replace the peripheral blood and bone marrow of Ccr2−/− mice with GFP­positive, Ccr2+/+ cells. The CCL2 (MCP­1) receptor, CCR2, is expressed at high levels on inflammatory monocytes and is required for the entry of these cells, but not T cells, into the CNS. It has been shown that CCR2 is required for the devel­opment of severe EAE23; whether this effect is a result of infiltrating cells or of other cell types that express this receptor is still unclear.

We found that Ccr2−/− mice in which the majority (86%) of circulat­ing cells were replaced with GFP­positive, Ccr2+/+ donor cells were fully susceptible to EAE and readily progressed to severe paralytic disease. This effect correlated with the restoration of blood­derived

01 2

EAE clinical score

Sco

re o

f 1S

core

of 2

Sco

re o

f 3S

core

of 4

GFP

GFP

GFP

GFP

CD11b

CD11b

CD11b

CD11b

Iba-1

Iba-1

Iba-1

Iba-1 GFP CD11b Iba-1

GFP CD11b Iba-1

GFP CD11b Iba-1

GFP CD11b Iba-1

3 4

40

P < 0.001

P < 0.001P < 0.001

P < 0.001

P < 0.001

P < 0.01

Fre

quen

cy o

fbl

ood-

deriv

edm

onoc

ytes

(%

) 80

01 2

EAE clinical score3 4

40

Fre

quen

cy o

fbl

ood-

deriv

edm

acro

phag

es (

%)

80

a

c e fd

g i jh

k m nl

o q rp

bFigure 3 Blood-derived infiltrating cells retain monocyte characteristics during progression to paralytic disease. (a) Graph showing the frequency of infiltrating cells (GFP positive) that displayed a monocytic phenotype (CD11b positive but Iba-1 negative) at each clinical disease score. For each data point, n = 6. (b) Graph showing the frequency of infiltrating cells (GFP positive) that expressed markers of mature macrophages (Iba-1 positive). For each data point, n = 6. (c–r) Representative high-magnification images of spinal cord sections from mice with each disease score, stained for monocytic and macrophage markers. Lower magnification images showing the distribution of the cells across the entire section are available in Supplementary Figure 5. Iba-1 is shown in blue, GFP is shown in green and CD11b is shown in red. CD11b and GFP double-positive cells with round morphology, representing blood-derived myelomonocytic cells, increased as disease progressed to paralytic stages (g–r). At a clinical score of 4, however, the majority of blood-derived cells had changed morphology and acquired Iba-1 expression (o–r). Few monocytes were detected at this stage, suggesting that cell influx is reduced relative to earlier stages of disease. Arrowheads point to representative monocytes. Scale bars represent 50 µm.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

1146 VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 nature neurOSCIenCe

a r t I C l e S

myelomonocytic infiltration into the CNS. Conversely, we found that Ccr2+/+ mice in which the peripheral blood and bone mar­row were replaced with cells from Ccr2−/− mice expressing GFP ubiquitously had only mild EAE and never progressed beyond a disease score of 1 (Fig. 5). As expected, only rare GFP­positive cells were detected in the CNS and none of these cells were positive for myelomonocytic markers (Fig. 5d–g). Thus, interfering with the entry of myelomonocytic cells into the CNS successfully blocked EAE progression, clearly supporting a role for these cells in causing functional deficits.

Blood monocytes do not contribute to CNS-resident microglia Our previous work suggests that, unlike most peripheral macrophages, which are replenished by blood­derived precursors, CNS­resident microglia are capable of self­renewal in situ. However, here we found that, in EAE, circulating monocytes efficiently enter the CNS and give rise to mature macrophages that are phenotypically indistinguishable from resident microglia. Do these macrophages also acquire the ability to self­renew in the CNS, and thereby make long­term contributions to the microglia pool? To determine whether colonization of the CNS by blood­derived myelomonocytic cells is transient or sustained, we analyzed the spinal cords of three irradiated­separated parabiotic mice that were allowed to recover from severe EAE. In these mice, we found a marked reduction in parenchymal GFP­positive cells compared with the numbers observed at the peak of disease (score of 4). In addition, none of the remaining GFP­positive cells were Iba­1 positive (Fig. 6), suggesting that the presence of blood­derived CNS myelomonocytic cells in the CNS parenchyma is a transient event.

Following substantial expansion in the transition to paralytic dis­ease, a large fraction of resident microglia undergo apoptosis, leading to a return of cell numbers to baseline. To determine whether all activated microglia are ablated after remission, we labeled proliferating microglia in three irradiated­separated parabiotic mice using BrdU

during the week preceding maximum disability. BrdU­labeled rami­fied microglia were clearly evident 3 months later, following complete remission (Fig. 6), suggesting that, unlike the progeny of infiltrating monocytes, expanded resident microglia can persist for a long time in the CNS.

Long-term contribution to microglia requires HSCsTransplantation of irradiated animals with whole bone marrow is known to lead to the long­term persistence of donor­derived micro­glia in the CNS26,34. In contrast, our data indicate that monocytes normally found in the circulation are unable to engraft the CNS of healthy mice, and that even when specific pathologies trigger their infiltration, their presence is transient. A possible explanation for these conflicting results is that the bone marrow cells mediating long­term CNS engraftment after irradiation are immature hematopoietic precursors that do not efficiently enter the circulation under normal circumstances. Immature cells in bone marrow belong to one of two functional groups: multipotent progenitors endowed with varying ability to self renew, generally contained in the lineage­negative, c­Kit–positive, Sca­1–positive (KLS) subset, and committed progeni­tors, which can generate all of the mature circulating cells belonging to specific hematopoietic lineages, but are unable to extensively self

Sco

re o

f 1S

core

of 2

Sco

re o

f 3S

core

of 4

GFP GFP

GFP

Iba-1

Iba-1

BrdU

BrdU GFPGFP

GFP Iba-1 BrdUGFP

GFP Iba-1 BrdUGFP

Iba-1 BrdU

GFP Iba-1 BrdU

GFP Iba-1 BrdU

GFP Iba-1 BrdU

c e fd

g i jh

k m nl

o q rp

01

EAE clinical score2 3 4 1

EAE clinical score2 3 4

40

Fre

quen

cy o

f Brd

U-p

ositi

vem

icro

glia

(%

) 80

0

80

Num

ber

of c

ells 160P < 0.001

P < 0.01

P < 0.001P < 0.001

P < 0.001

P < 0.001

P < 0.001

MicrogliaMacrophages

a bFigure 4 Kinetics of microglia expansion and blood-derived monocyte infiltration in EAE. (a) Percentage of proliferating microglia (GFP negative, BrdU positive, Iba-1 positive) during EAE progression. For each data point, n = 6. (b) The absolute number of resident microglia (GFP-negative, Iba-1–positive cells) and blood-derived macrophages (GFP-positive, Iba-1– positive cells) in the spinal cord parenchyma at different stages of disease. Microglia increased in number as disease progressed to a score of 3 and declined between scores of 3 and 4, whereas the number of blood-derived macrophages continued to increase. For each data point, n = 6. (c–r) Representative high-magnification images of spinal cord sections from each stage of disease, stained for BrdU and the macrophage marker Iba-1. Lower magnification images showing the distribution of the cells across the entire section are available in Supplementary Figure 7. The presence of BrdU-positive, Iba-1–positive cells at a clinical score of 1 indicates that microglia activation occurs before inflammatory cell entry. GFP is shown in green, BrdU in blue and Iba-1 in red. Scale bars represent 50 µm.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature neurOSCIenCe VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 1147

a r t I C l e S

renew. To test which of these subsets can permanently contribute to microglia in irradiated recipients, we generated a set of irradiated­transplanted chimeric animals in which C57BL/6 mice were recon­stituted with bone marrow–derived myelomonocytic progenitors from Cx3cr1­GFP/+ heterozygous donors. In this strain, the coding sequence for one allele of Cx3cr1, which encodes the receptor for CX3CL1 (fractalkine), is replaced with GFP, resulting in marker expression in all of the myelomonocytic cells, including bone marrow progenitors, circulating monocytes, tissue resident macrophages and microglia35,36. The CX3CR1­positive fraction of bone marrow comprises a non–self­renewing progenitor population that can tran­siently repopulate the peripheral blood of recipients. Cells positive for the transgenic marker were sorted from the bone marrow of Cx3cr1­GFP/+ animals based on GFP expression and transplanted into lethally irradiated recipients. As these cells cannot generate erythroid cells or platelets, the survival of the myeloablated mice was ensured by co­transplanting a radioprotective dose of GFP­negative, syngeneic KLS cells. As expected, fluorescence­activated cell sorting (FACS) analysis confirmed that GFP­positive cells persisted in the peripheral blood of recipients until 14 weeks post­transplant, after which they became undetectable (Fig. 7a).

We subjected the recipients to EAE induction 2 weeks after transplan­tation, a time at which the circulating cells of most mice consisted of more than 10% GFP­positive cells. The contribution of blood­derived GFP­positive monocytes to the CNS of EAE mice was evaluated at a disease score of 4, approximately 2 weeks post­induction. At this time,

abundant GFP­positive, Iba­1–positive donor cells were observed in the spinal cord (Fig. 7b). A second evaluation after 3 months found few donor­derived myelomonocytic cells (less than one cell per section analyzed), which was independent of whether mice had completely recovered from EAE or had developed permanent functional impair­ment (Fig. 7c,d). In stark contrast, Iba­1–positive, donor­derived microglia were readily detected 3 months after complete recovery from EAE in the spinal cord sections of a parallel group of irradi­ated recipients that received GFP­positive KLS cells (Fig. 7e,f). Thus, although mature inflammatory macrophages can infiltrate the CNS and trigger EAE progression, their presence in the CNS is transient. Only uncommitted stem or progenitor cells are capable of generating long­lived microglia in irradiated recipients.

DISCUSSIONAlthough adoptive transfer studies have established that T cells are both necessary and sufficient for induction of EAE37,38, the cellular mecha­nisms that govern disease progression are still being debated. Microglia and blood­derived myelomonocytes have both been implicated in the development of EAE because of their ability to present antigens, secrete pro­inflammatory cytokines39 and participate in demyelination by phagocytosis of degraded myelin40. However, it is unclear whether these two cell types have distinct roles in EAE pathogenesis.

Here we present an irradiation­parabiosis–based experimental model that can distinguish between blood­derived monocytes and CNS­resident microglia in the absence of bone marrow transplan­tation. Using this model, we have discovered important functional differences between resident and infiltrating cells. During early disease, T­cell infiltration and microglia activation are evident, even in mice that do not develop functional impairment. In contrast, the

Figure 5 Blocking monocyte infiltration prevents EAE progression. The bone marrow and peripheral blood of Ccr2+/+ mice were replaced with that of GFP-expressing Ccr2−/− donors (circles) using the parabiosis-irradiation-separation strategy. Conversely, the peripheral blood and bone marrow of Ccr2−/− animals was replaced with GFP-expressing Ccr2+/+ cells (squares) using a similar experimental approach. EAE was induced in both groups. (a,b) Representative FACS plots showing the frequency of GFP-positive cells in the blood of GFP-negative partners at the time of EAE induction. (c) Average clinical score for each group of animals (Ccr2−/− to wild type, n = 10; wild type to Ccr2−/−, n = 9). Ccr2+/+ animals repopulated with Ccr2−/− cells did not progress beyond a disease score of 1, whereas blood repopulation with GFP-expressing Ccr2+/+ cells restored the susceptibility of Ccr2−/− mice to severe paralytic disease. *P < 0.01. (d,e) Abundant GFP-positive cells stained for the myelomonocytic markers CD11b (d) and Iba-1 (e) were found in spinal cord sections of Ccr2−/− mice reconstituted with GFP-positive bone marrow. (f,g) None of the donor cells found in spinal cord sections of C57BL/6 mice reconstituted with GFP-expressing Ccr2−/− cells expressed myelomonocytic markers. Iba-1 and CD11b are shown in red and GFP is shown in green. Scale bar represents 50 µm.

Iba-1

GFP Iba-1 BrdU

BrdU BrdUGFP GFP Iba-1GFP

a

b c d e

Figure 6 Blood-borne inflammatory cell infiltration is transient. (a) Collages of maximum-intensity projections of stacks of confocal optical sections from the spinal cord of chimeric animals generated by parabiosis-irradiation-separation that recovered from EAE. Notice that the rare partner-derived cells (GFP positive) still present did not stain with Iba-1. At this stage, however, some of the endogenous microglia were still BrdU positive (blue, arrowheads), indicating that they returned to quiescence during or shortly after BrdU treatment. (b–e) Higher magnification images of the white box in a. Arrowheads indicate BrdU-retaining microglia (GFP negative, Iba-1 positive, BrdU positive). Iba-1 is shown in red, GFP in green and BrdU in blue. Scale bars represent 500 µm (a) and 50 µm (b–e).

86.7

0100 101 102

GFP

100

1

2

EA

E c

linic

al s

core 3

4

12 14

**

*

**

16Days post EAE induction

18 20 22 24 26 28 30

103 104

50

100

150

200

Num

ber

of c

ells

GFP-expressing

Ccr2+/+ cells to Ccr–/– mice

GFP-expressing

Ccr2+/+ cells to Ccr–/– mice

GFP-expressing

Ccr–/– cells to Ccr2+/+ mice

GFP CD11b

GFP Iba-1

GFP CD11b

GFP Iba-1

a

c

d f

e g

92.5

0100 101 102

GFP103 104

100

200

300

Num

ber

of c

ells

GFP-expressing

Ccr–/– cells to Ccr2+/+ miceb

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

1148 VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 nature neurOSCIenCe

a r t I C l e S

appearance of infiltrating monocytes correlates with substantial dis­ability, and impairing the CCR2­dependent recruitment of these cells prevents progression from very mild to severe disease. Thus, in EAE, the infiltration of monocytes, a cell type that is normally excluded from the CNS in both healthy animals and a variety of diseases, may represent a pathogenic overreaction by the innate immune system. A similarly deleterious role of monocytes has been recently reported in acute viral meningitis41,42. In this study, however, in situ prolif­eration of microglia was minimal, whereas it was substantial during EAE, suggesting that there may be disease­specific differences in the activation of resident inflammatory cells.

The numbers of resident microglia dropped markedly between mice with disease scores of 3 and 4, a time at which the number of infiltrating myeloid cells increases and coincides with functional impairment. Thus, our data suggest that monocytes and microglia respond differently to environmental stimuli regulating their survival. In addition, the differences between the two cell types were also evi­dent in their ability to colonize the CNS. Although monocyte­derived mature macrophages are phenotypically indistinguishable from resi­dent microglia, they are specifically and completely removed from the CNS. In contrast, the fact that we detected BrdU­retaining cells 3 months after disease remission indicates that microglia can enter the cell cycle, proliferate and then return to quiescence.

If circulating monocytes are unable to generate long­lived micro­glia, which cell type is responsible for their production in irradiated­transplanted animals? We found that the ability to generate microglia was restricted to uncommitted stem or progenitor cells in bone marrow. Notably, other studies have found that hematopoietic stem or progenitor cells can generate myelomonocytic cells in peripheral tissues in response to Toll­like receptor signaling43,44.

Thus, two distinct types of myelomonocytic engraftment occur in the CNS. The first is triggered by neuroinflammatory patholo­gies and transiently recruits inflammatory monocytes to affected areas. The second, observed in irradiated­transplanted subjects, is dependent on hematopoietic stem or progenitor cells and can lead to a permanent contribution to the microglial pool. This second type of engraftment may be exploited to deliver therapeutic gene products across the BBB.

Taken together, these observations lead to a three­step model of EAE progression. First, CD4­positive T cells and endogenous microglia are responsible for disease initiation. Second, the mechanisms

excluding blood­borne monocytes from the CNS break down. Third, infiltrating blood­derived monocytes trigger, directly or indirectly, EAE progression to the severe, paralytic form of the disease. Once the disease reaches a score of 4, the decline of microglia and disappearance of infiltrating myelomonocytic cells leads to remission and, depending on the extent of axonal and neuronal loss, possibly to recovery45.

Blocking the migration of immune cells across the BBB has long been regarded as a likely therapeutic approach for treating CNS­directed autoimmune diseases46. Understanding the pathological cascade of EAE will help us to tailor more specific therapies for mul­tiple sclerosis. For instance, blocking the homing of T lymphocytes to the CNS using an antibody specific for α4 integrin suppresses EAE and reduces relapse rates in humans with multiple sclerosis by 66%47,48. Unfortunately, in a subset of individuals, this treatment leads to the reactivation of viral infections and progressive multi­focal leukoencephalopathy49,50. Our findings provide a rationale for a therapeutic strategy that specifically targets myelomonocytic cell entry, which might have potentially fewer side effects than existing therapies.

METHODSMethods and any associated references are available in the online version of the paper at http://www.nature.com/natureneuroscience/.

Note: Supplementary information is available on the Nature Neuroscience website.

AcknowledgmenTSWe thank B. Chua, T. Godbey, K. Ranta, M. Cowan, L. Rollins and the Biomedical Research Center animal unit personnel for advice and help on animal welfare. We thank A. Johnson, C.K. Chang, J. Kang and D. Mahdaviani for their technical

2 3 4 5 6 7Weeks post myelomonocytic

progenitors transplant

Per

iphe

ral b

lood

chi

mer

ism

(% G

FP

-pos

itive

cel

ls)

02468

1012141618

9 11 12 13 14

Permanent impairment Fully recovered

GFP Iba-1 PECAM

GFP Iba-1 PECAM

GFP

GFP Iba-1 PECAM

GFP Iba-1

Peak of EAEa b

c d

e f

Figure 7 Uncommitted stem or progenitor cells, but not myelomonocytic-committed hematopoietic progenitors, contribute to resident microglia in irradiated-transplanted recipients. (a) Percentage of GFP-positive cells in the peripheral blood of irradiated mice (n = 6) transplanted with the GFP-positive fraction of bone marrow from Cx3cr1-GFP/+ donors, representing myelomonocytic-committed progenitors. As expected, the frequency of donor-derived myelomonocytic cells in the blood declined rapidly during the first month following the transplant. Error bars represent s.d. (b) Spinal cord sections from mice transplanted with Cx3cr1-GFP/+ cells and killed at an EAE clinical score of 4 (approximately 2 weeks after EAE induction) revealed substantial myelomonocytic infiltration. The absence of donor-derived cells in spinal cord sections from mice killed 3 months after the peak of disease suggests that blood-borne cells in the CNS are transient and do not contribute long-term to the microglia pool, independent of whether the outcome of EAE was a permanent functional impairment (c) or full recovery (d). (e,f) Spinal cord sections from mice transplanted with KLS stem or progenitor cells from a donor ubiquitously expressing GFP and killed 3 months after the peak of disease, when the mice had completely recovered from the disease, contained numerous donor-derived ramified microglia (GFP positive, Iba-1 positive). Iba-1 is shown in red, GFP in green and PECAM in blue. Scale bars represent 500 µm (b–d) and 50 µm (e,f).

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature neurOSCIenCe VOLUME 14 | NUMBER 9 | SEPTEMBER 2011 1149

a r t I C l e S

help. This work was supported by Canadian Institute for Health Research (CIHR, MOP 81382) grants and a grant from the Multiple Sclerosis Society of Canada to F.M.V.R., a Neuromuscular Research Partnership grant from the CIHR, a grant from the Amyotrophic Lateral Sclerosis Society of Canada and Muscular Dystrophy Canada to C.K. and F.M.V.R. (JNM­69682), a Collaborative Health Research grant from the CIHR and the Natural Science and Engineering Research Council of Canada to C.K. and F.M.V.R. (CHRP 299119), and a research grant from the Multiple Sclerosis Society of Canada to K.M.M. K.M.M. is a Michael Smith Foundation for Health Research Senior Scholar. B.A is supported by a Michael Smith Foundation Senior Graduate Studentship and a CIHR–Amyotrophic Lateral Sclerosis Doctoral Research Award. J.L.B. is supported by a Multiple Sclerosis Society of Canada Postdoctoral Research Fellowship. This research was undertaken, in part, thanks to funding from the Canadian Research Chairs program to F.M.V.R.

AUTHoR conTRIBUTIonSB.A. designed and conducted all of the experiments, interpreted the data and wrote the manuscript. J.L.B. conducted the EAE induction and participated in the writing of the manuscript. C.K. and K.M.M. participated in the writing of the manuscript. F.M.V.R. designed and interpreted experiments and wrote the manuscript.

comPeTIng FInAncIAl InTeReSTSThe authors declare no competing financial interests.

Published online at http://www.nature.com/natureneuroscience/. Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Kreutzberg, G.W. Microglia: a sensor for pathological events in the CNS. Trends Neurosci. 19, 312–318 (1996).

2. Nimmerjahn, A., Kirchhoff, F. & Helmchen, F. Resting microglial cells are highly dynamic surveillants of brain parenchyma in vivo. Science 308, 1314–1318 (2005).

3. Raivich, G. Like cops on the beat: the active role of resting microglia. Trends Neurosci. 28, 571–573 (2005).

4. Streit, W.J., Walter, S.A. & Pennell, N.A. Reactive microgliosis. Prog. Neurobiol. 57, 563–581 (1999).

5. Lawson, L.J., Perry, V.H. & Gordon, S. Turnover of resident microglia in the normal adult mouse brain. Neuroscience 48, 405–415 (1992).

6. Hickey, W.F. & Kimura, H. Perivascular microglial cells of the CNS are bone marrow–derived and present antigen in vivo. Science 239, 290–292 (1988).

7. Priller, J. et al. Targeting gene-modified hematopoietic cells to the central nervous system: use of green fluorescent protein uncovers microglial engraftment. Nat. Med. 7, 1356–1361 (2001).

8. Hess, D.C. et al. Hematopoietic origin of microglial and perivascular cells in brain. Exp. Neurol. 186, 134–144 (2004).

9. Ajami, B., Bennett, J.L., Krieger, C., Tetzlaff, W. & Rossi, F.M. Local self-renewal can sustain CNS microglia maintenance and function throughout adult life. Nat. Neurosci. 10, 1538–1543 (2007).

10. Ginhoux, F. et al. Fate mapping analysis reveals that adult microglia derive from primitive macrophages. Science 330, 841–845 (2010).

11. Hickey, W.F. Migration of hematogenous cells through the blood-brain barrier and the initiation of CNS inflammation. Brain Pathol. 1, 97–105 (1991).

12. Hickey, W.F., Hsu, B.L. & Kimura, H. T-lymphocyte entry into the central nervous system. J. Neurosci. Res. 28, 254–260 (1991).

13. Swanborg, R.H. Experimental autoimmune encephalomyelitis in rodents as a model for human demyelinating disease. Clin. Immunol. Immunopathol. 77, 4–13 (1995).

14. Steinman, L., Martin, R., Bernard, C., Conlon, P. & Oksenberg, J.R. Multiple sclerosis: deeper understanding of its pathogenesis reveals new targets for therapy. Annu. Rev. Neurosci. 25, 491–505 (2002).

15. Santambrogio, L. et al. Developmental plasticity of CNS microglia. Proc. Natl. Acad. Sci. USA 98, 6295–6300 (2001).

16. Ulvestad, E. et al. Human microglial cells have phenotypic and functional characteristics in common with both macrophages and dendritic antigen-presenting cells. J. Leukoc. Biol. 56, 732–740 (1994).

17. Ponomarev, E.D., Shriver, L.P., Maresz, K. & Dittel, B.N. Microglial cell activation and proliferation precedes the onset of CNS autoimmunity. J. Neurosci. Res. 81, 374–389 (2005).

18. Heppner, F.L. et al. Experimental autoimmune encephalomyelitis repressed by microglial paralysis. Nat. Med. 11, 146–152 (2005).

19. Bauer, J. et al. The role of macrophages, perivascular cells, and microglial cells in the pathogenesis of experimental autoimmune encephalomyelitis. Glia 15, 437–446 (1995).

20. Huitinga, I., van Rooijen, N., de Groot, C.J., Uitdehaag, B.M. & Dijkstra, C.D. Suppression of experimental allergic encephalomyelitis in Lewis rats after elimination of macrophages. J. Exp. Med. 172, 1025–1033 (1990).

21. Brosnan, C.F., Bornstein, M.B. & Bloom, B.R. The effects of macrophage depletion on the clinical and pathologic expression of experimental allergic encephalomyelitis. J. Immunol. 126, 614–620 (1981).

22. Izikson, L., Klein, R.S., Charo, I.F., Weiner, H.L. & Luster, A.D. Resistance to experimental autoimmune encephalomyelitis in mice lacking the CC chemokine receptor (CCR)2. J. Exp. Med. 192, 1075–1080 (2000).

23. Fife, B.T., Huffnagle, G.B., Kuziel, W.A. & Karpus, W.J. CC chemokine receptor 2 is critical for induction of experimental autoimmune encephalomyelitis. J. Exp. Med. 192, 899–905 (2000).

24. Mildner, A. et al. CCR2+ Ly-6Chi monocytes are crucial for the effector phase of autoimmunity in the central nervous system. Brain 132, 2487–2500 (2009).

25. King, I.L., Dickendesher, T.L. & Segal, B.M. Circulating Ly-6C+ myeloid precursors migrate to the CNS and play a pathogenic role during autoimmune demyelinating disease. Blood 113, 3190–3197 (2009).

26. Kennedy, D.W. & Abkowitz, J.L. Kinetics of central nervous system microglial and macrophage engraftment: analysis using a transgenic bone marrow transplantation model. Blood 90, 986–993 (1997).

27. Li, Y.Q., Chen, P., Jain, V., Reilly, R.M. & Wong, C.S. Early radiation-induced endothelial cell loss and blood-spinal cord barrier breakdown in the rat spinal cord. Radiat. Res. 161, 143–152 (2004).

28. Diserbo, M. et al. Blood-brain barrier permeability after gamma whole-body irradiation: an in vivo microdialysis study. Can. J. Physiol. Pharmacol. 80, 670–678 (2002).

29. Mildner, A. et al. Microglia in the adult brain arise from Ly-6Chi CCR2+ monocytes only under defined host conditions. Nat. Neurosci. 10, 1544–1553 (2007).

30. Wright, D.E. et al. Cyclophosphamide/granulocyte colony-stimulating factor causes selective mobilization of bone marrow hematopoietic stem cells into the blood after M phase of the cell cycle. Blood 97, 2278–2285 (2001).

31. Mason, D. Genetic variation in the stress response: susceptibility to experimental allergic encephalomyelitis and implications for human inflammatory disease. Immunol. Today 12, 57–60 (1991).

32. Bukilica, M. et al. Stress-induced suppression of experimental allergic encephalomyelitis in the rat. Int. J. Neurosci. 59, 167–175 (1991).

33. Streit, W.J. Microglia and the response to brain injury. Ernst Schering Res. Found. Workshop 39, 11–24 (2002).

34. Krall, W.J., Challita, P.M., Perlmutter, L.S., Skelton, D.C. & Kohn, D.B. Cells expressing human glucocerebrosidase from a retroviral vector repopulate macrophages and central nervous system microglia after murine bone marrow transplantation. Blood 83, 2737–2748 (1994).

35. Jung, S. et al. Analysis of fractalkine receptor CX(3)CR1 function by targeted deletion and green fluorescent protein reporter gene insertion. Mol. Cell. Biol. 20, 4106–4114 (2000).

36. Saederup, N. et al. Selective chemokine receptor usage by central nervous system myeloid cells in CCR2–red fluorescent protein knock-in mice. PLoS ONE 5, e13693 (2010).

37. Carson, M.J., Thrash, J.C. & Walter, B. The cellular response in neuroinflammation: the role of leukocytes, microglia and astrocytes in neuronal death and survival. Clin. Neurosci. Res. 6, 237–245 (2006).

38. Engelhardt, B. Molecular mechanisms involved in T cell migration across the blood-brain barrier. J. Neural Transm. 113, 477–485 (2006).

39. Benveniste, E.N. Role of macrophages/microglia in multiple sclerosis and experimental allergic encephalomyelitis. J. Mol. Med. 75, 165–173 (1997).

40. Bauer, J., Sminia, T., Wouterlood, F.G. & Dijkstra, C.D. Phagocytic activity of macrophages and microglial cells during the course of acute and chronic relapsing experimental autoimmune encephalomyelitis. J. Neurosci. Res. 38, 365–375 (1994).

41. Kim, J.V., Kang, S.S., Dustin, M.L. & McGavern, D.B. Myelomonocytic cell recruitment causes fatal CNS vascular injury during acute viral meningitis. Nature 457, 191–195 (2009).

42. Getts, D.R. et al. Ly-6C+ ‘inflammatory monocytes’ are microglial precursors recruited in a pathogenic manner in West Nile virus encephalitis. J. Exp. Med. 205, 2319–2337 (2008).

43. Massberg, S. et al. Immunosurveillance by hematopoietic progenitor cells trafficking through blood, lymph, and peripheral tissues. Cell 131, 994–1008 (2007).

44. Nagai, Y. et al. Toll-like receptors on hematopoietic progenitor cells stimulate innate immune system replenishment. Immunity 24, 801–812 (2006).

45. Wujek, J.R. et al. Axon loss in the spinal cord determines permanent neurological disability in an animal model of multiple sclerosis. J. Neuropathol. Exp. Neurol. 61, 23–32 (2002).

46. Luster, A.D., Alon, R. & von Andrian, U.H. Immune cell migration in inflammation: present and future therapeutic targets. Nat. Immunol. 6, 1182–1190 (2005).

47. Yednock, T.A. et al. Prevention of experimental autoimmune encephalomyelitis by antibodies against alpha 4 beta 1 integrin. Nature 356, 63–66 (1992).

48. Miller, D.H. et al. A controlled trial of natalizumab for relapsing multiple sclerosis. N. Engl. J. Med. 348, 15–23 (2003).

49. Steinman, L. Blocking adhesion molecules as therapy for multiple sclerosis: natalizumab. Nat. Rev. Drug Discov. 4, 510–518 (2005).

50. Langer-Gould, A., Atlas, S.W., Green, A.J., Bollen, A.W. & Pelletier, D. Progressive multifocal leukoencephalopathy in a patient treated with natalizumab. N. Engl. J. Med. 353, 375–381 (2005).

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

nature neurOSCIenCe doi:10.1038/nn.2887

ONLINE METHODSmice. C57BL/6 mice expressing GFP ubiquitously under the control of a CMV–β­actin hybrid promoter (C57BL/6; GFP/CD45.2) were a kind gift from I. Weissman (Stanford University). C57BL/6, Cx3cr1-GFP and Ccr2−/− mice were purchased from Jackson Laboratories. In the case of Cx3cr1-GFP/+ mice, heterozygotes obtained by crossing homozygotes with C57BL/6 mice were used as donors in transplantation experiments. Ccr2−/− C57BL/6 mice were bred in house to the ubiquitous GFP strain. All mice were maintained in a pathogen­free facility, and all experiments were performed in accordance with the policies of the University of British Columbia’s Animal Care Committee.

generation of chimeric mice by parabiosis-irradiation. We selected 4–6­week­old female mice that were at least 17 g in weight for parabiosis. Age­ and weight­matched mice were housed together for at least 1 week before surgery. Pairs of parabiotic mice were surgically generated as previously described9. The GFP­positive partner was shielded with lead 2 weeks after surgery and the pair was subjected to a lethal dose of irradiation (1,100 rads) using a gamma cell irradiator with a cobalt­60 source. The parabiotic pairs were surgically separated 4 weeks after irradiation.

generation of chimeric mice by irradiation and transplant. We used 6–10­week­old Cx3cr1-GFP/+ female mice as bone marrow donors. Donor mice were killed with CO2 and their femurs and tibias were removed. Marrow cavi­ties were flushed with FACS buffer (phosphate­buffered saline (PBS) supple­mented with 2% fetal bovine serum (vol/vol) and 2 mM EDTA) using a 25­gauge needle attached to a syringe. Erythrocytes were lysed by resuspending samples in NH4Cl solution and incubating at 20–24 °C for 5 min.

Nucleated cells were suspended in FACS buffer and CX3CR1­expressing cells were sorted on the basis of GFP expression. Sorts were performed on a FACSVantage SE (Becton Dickenson). Cells were sorted in PBS.

Bone marrow cells were isolated and stained on ice for 30 min with an anti­body cocktail containing antibodies to CD3, Mac1, Gr1, Ter119, B220 antibodies directly conjugated to PE­Cy7 (lineage cocktail, BD Biosciences), c­Kit–FITC or c­Kit–PE (clone 2B8, BD Pharmingen), and Sca­1–APC (clone: D7, eBiosciences). Following staining, cells were sorted using a FACSVantage or a FACSAria (Becton Dickinson). Cells were sorted in PBS.

Immediately before transplantation, 4–6­week­old C57BL/6 mice received a lethal dose of irradiation (1,100 rads) as described above. Recipient mice received 1.5 × 106 sorted GFP­positive bone marrow cells from Cx3cr1-GFP/+ mice and/or 500 radioprotective KLS cells via the tail vein.

Peripheral blood analysis. Peripheral blood was obtained from the tail vein. Red blood cells were lysed as described and the white blood cells were resus­pended in FACS buffer. Analysis was accomplished using a FACS Calibur (Becton Dickinson), and flow cytometry data were analyzed using FlowJo (Treestar) analysis software (Supplementary Fig. 10).

Induction of autoimmune experimental encephalitis. Mice were immunized by subcutaneous injection of the myelin peptide MOG35­55 (200 µg) emulsified in complete Freund’s adjuvant containing 4 mg ml−1 Mycobacterium tuberculosis (Difco) 3 weeks after parabiosis or 1 week after separation. In addition, 200 ng of pertussis toxin (Biological Laboratories) was administered intravenously on the day of immunization and 2 d after.

Tissue collection and preparation. Animals were injected with a lethal dose of avertin and monitored. Upon the loss of nociceptive reflexes, animals were perfused transcardially with 20 ml of PBS/EDTA followed by 20 ml of 4% para­formaldehyde (wt/vol) in 0.1 M PBS at 20–24 °C. The spinal cord was removed and the tissue was post­fixed for 24 h in 4% paraformaldehyde at 4 °C and then cryoprotected in a 24% sucrose solution (wt/vol) in PBS for 24 h. Spinal cord tissues were embedded in Optimal Cutting Temperature compound (Tissue­Tek) and frozen at −80 °C.

Immunofluorescence of tissue sections. After tissue processing, 20­µm­thick cryosections were cut from the lumbar region of spinal cords (L1–L5), resulting into approximately 500 sections per lumbar region for each mouse. Sections were examined using a Zeiss Axioplan 2 microscope equipped for epifluorescence (Carl Zeiss) and the sections with most infiltrated cells (GFP­positive cells) selected for immunofluorescence. Sections were allowed to thaw at 20–24 °C, rehydrated in PBS for 2 h and incubated with blocking buffer (25% normal goat serum (vol/vol), 3% BSA (wt/vol) and 0.3% Triton X­100 (vol/vol)) for another hour. FcγR was blocked by incubating the sections with 5–10 µg ml−1 purified antibody to CD16/32 on ice for 10 min. Primary antibody staining was performed overnight. For primary antibodies, we used antibody to Iba­1 as microglia/macrophage marker (Wako), antibody to mouse CD31 as an endothelial marker (PECAM­1, BD Pharmingen), antibody to CD4 as a T­cell marker (clone L3T4, eBiosciences), and antibody to BrdU (clone B44, BD). For secondary antibodies, we used Alexa568­ or Alexa647­conjugated goat antibodies to rat (Molecular Probes), and Alexa568­conjugated goat antibody to mouse IgG1 (Molecular Probes). All sections were analyzed by confocal microscopy using a Nikon C1 Laser Scanning Confocal Microscope. Unless otherwise indicated, all images presented are maxi­mum intensity projections of z stacks of individual optical sections.

Quantification and statistics. Cells were manually counted in two diagonally opposite quadrants in the three sections with the most infiltrated cells (GFP­ positive cells) for each animal (Supplementary Fig. 11). All counts were blinded. All of the statistical tests were performed using Microsoft Excel X software.

The mean number of cells per quadrant counted in each group was calculated. Unpaired t tests (two tailed) were used for all statistical analyses. Linear regression analysis was performed using Microsoft Excel X software. When the data is pre­sented as averages in figures, the individual data points from which the averages were calculated can be found in Supplementary Table 1, and the exact P values between each dataset can be found in the legend to the same table.

© 2

011

Nat

ure

Am

eric

a, In

c. A

ll ri

gh

ts r

eser

ved

.©

201

1 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.