The study “puts a new player on the field that we haven't been thinking about before,” said Steve Greenberg, a CAA specialist at Massachusetts General Hospital in Boston. It puts perivascular macrophages “on board as both a potential determinant of who will get CAA or AD, and also a potential target for treatment.”

As immune scavengers, macrophages scour every nook and cranny in the body for unwanted debris—be it damaged cells, DNA fragments, or amyloid deposits. Such housekeeping, though, often requires release of proinflammatory chemicals and other toxins, making it hard to determine whether these Aβ-gobbling cells harm or help in neurodegenerative disease (see ARF related news story). Numerous AD studies have documented the clearance of parenchymal plaques by microglia, the macrophages of the brain, in response to immunization with Aβ peptide. McLaurin and postdoctoral fellow Cheryl Hawkes are among those who have witnessed plaque removal in vaccinated mice.

While poring over confocal slides of immunostained brain tissue from these animals, the Canadian scientists came across cells that looked like perivascular macrophages staining positive for thioflavin S (a dye that preferentially labels the fibrillar form of Aβ found in plaques). “It suggested to us that perivascular macrophages might be involved in the clearance of CAA,” McLaurin said. “That's how this whole project started.”

To put their hunch to the test, first author Hawkes looked at whether depletion of perivascular macrophages would influence CAA severity in an AD mouse model. She did the studies in TgCRND8 mice, which overexpress thrice-mutated human amyloid precursor protein and develop parenchymal and vascular Aβ deposition already at 12 weeks. To target perivascular macrophages, Hawkes used a method whereby phagocytic cells get tricked into killing themselves by engulfing a liposome-coated toxin (clodronate). At four months of age, the mice received a single brain injection of either clodronate- or vehicle-encapsulated liposomes, and were sacrificed for analysis one month later.

The clodronate treatment put a dent in the perivascular macrophage population, halving CD206 protein levels in brain extracts and reducing CD163 reactivity in immunostained brain tissue. (CD206 and CD163 are macrophage-specific markers expressed by perivascular macrophages but not parenchymal microglia.) More importantly, the macrophage depletion seemed to exacerbate CAA in TgCRND8 mice: compared with vehicle-treated animals, the clodronate group had about five times more cortical tissue (0.28 versus 0.05 percent) covered by thioS-positive blood vessels. Subsequent immunohistochemistry and ELISA studies revealed that higher levels of vascular Aβ42—but not of the less toxic species Aβ40, or of parenchymal or plasma Aβ42—accounted for the increased CAA severity.

Having shown that CAA worsens with perivascular macrophage depletion, the researchers took the opposite tack of stimulating turnover of these cells in five-month-old TgCRND8 mice and seeing whether this could reduce vascular amyloid load. Once again, they turned to an unusual method to target perivascular macrophages, using chitin (yes, the stuff in insect cell walls) to boost turnover rate. (Normally, about a tenth of this macrophage population gets replaced each month from bone marrow.) Other studies had used chitin to stimulate peripheral macrophages, but McLaurin and Hawkes had to show it did the same to the perivascular population.

The scientists administered two sequential chitin injections into the brain—the first along with a green dye, the second two weeks later with a red dye—and determined the ratio of singly and doubly labeled cells. If turnover rate increases, more macrophages would appear as singly labeled red cells because the ones that took up green dye earlier wouldn’t hang around as long. Using this readout, the procedure seemed to work in their hands: the chitin-treated mice had more red macrophages (48 percent red cells) than did vehicle-treated mice (33 percent red cells). Cortical samples from chitin-treated mice had a nearly threefold reduction in vascular amyloid load (0.10 percent of cortex covered in ThioS-positive blood vessels, versus 0.28 percent in vehicle-treated cortex). As in the clodronate experiment, it was Aβ42 and not Aβ40 that seemed to mediate the CAA effects. In chitin-treated mice, total human Aβ42 levels were considerably lower in blood vessels but, curiously, higher in plasma, compared to vehicle-treated animals. By staining the treated cortical samples with antibodies to markers for activated astrocytes (glial fibrillary acidic protein) or microglial cells (Iba-1), and seeing no colocalization with the ThioS signal, the researchers could attribute the CAA relief to the perivascular macrophages and not to those other cell populations.

At this point, though, scientists remain cautious about the clinical relevance of the new findings. The authors themselves note that whereas human CAA consists predominantly of Aβ40, vascular amyloidosis is skewed heavily toward Aβ42 in TgCRND8 mice, which express about five times as much Aβ42 as Aβ40 at five to six months of age. McLaurin sees the CAA relief mediated by perivascular macrophages as “a general Aβ phenomenon” that appears Aβ42-specific in the TgCRND8 strain simply because these mice make so much more Aβ42. Donna Wilcock of Duke University in Durham, North Carolina, thinks the findings would carry more weight if obtained in a mouse strain that more closely models human CAA. “It would be really nice to do the same study in the Tg2576 mouse, where we know that the majority of the CAA is Aβ40 just like it is in AD patients,” she said. That would take it “one step closer to giving information on actual clinical AD.”

Nevertheless, the current findings suggest that perivascular macrophages do play a part in ridding blood vessels of harmful Aβ. The challenge is coming up with a straightforward means to stimulate these cells selectively. “We haven't yet been able to figure out a mechanism where you could treat someone peripherally and affect only this one population of cells,” McLaurin said. In the meantime, her group is studying whether stimulation of peripheral macrophages affects normal blood vessel function. “Are you going to affect blood circulation in the brain? Does the vessel still constrict naturally? That's where we're going right now,” she told ARF.

Berislav Zlokovic of University of Rochester, New York, said he believes the key to targeting perivascular macrophages lies in identifying the molecular mechanisms that regulate recruitment of their precursors from the periphery. Adding to the neuroinflammation puzzle, a report published last month by Zlokovic’s group suggests that CAA may be touched off by a pair of transcription co-factors that regulate smooth muscle cell activity (Bell et al., 2008 and see ARF related news story). A study in this month’s issue of Acta Neuropathology (Weller et al., 2009) proposes that age-related vessel stiffening slows perivascular draining of Aβ, which then deposits as CAA and parenchymal plaques (see also Carare-Nnadi/Weller comment).

Collectively, these studies highlight the need for therapies to keep blood vessels from getting clogged with amyloid. Whether this is accomplished by harnessing perivascular macrophages or other Aβ clearance mechanisms (see, e.g., ARF related conference story and ARF conference story), the time could be ripe for human testing of such strategies once they become available, suggested Greenberg. Based on recent imaging studies he and colleagues have done in sporadic (Johnson et al., 2007) and familial (Greenberg et al., 2008) CAA patients, “there may be a signature of vascular amyloid” in live brain scans, he said. “It may be possible to noninvasively follow the progression of CAA.”—Esther Landhuis.

Reference:
Hawkes CA and McLaurin J. Selective targeting of perivascular macrophages for clearance of beta-amyloid in cerebral amyloid angiopathy. 2009. PNAS Early Edition. Abstract