To the Editor:

We are writing in response to the paper by Yemisci et al.1. The authors propose that ischemia-reperfusion causes widespread, long-lasting dynamic constriction of pericytes, leading to microvascular failure that contributes to brain injury and behavioral deficits1. After careful review of the paper, we have several concerns.

The authors claim pericytes contract in response to ischemia, but they do not present data showing dynamic contractility of pericytes1. Their conclusions are based largely on ex vivo histological observations that cannot be equated to observations of functioning microvessels in vivo2,3,4 and may be prone to artifacts induced by decapitation, tissue fixation and processing. The authors present data from contralateral 'control' hemispheres, but the ischemic hemisphere may be more prone to artifactual disturbances (for example, ischemia-induced edema may impair perfusion unrelated to pericyte contractions)1. Figure 6a–f shows in vivo imaging data of capillaries with trapped erythrocytes, but the suggestion that these capillaries are “constricted” on the basis of comparison of the red blood cell diameter and the fluorescent dye column diameter is not visually convincing. Moreover, Figure 6g, showing trapped erythrocytes in a capillary and an adjacent smooth muscle actin (SMA)-positive pericyte, comes from ex vivo imaging; the colocalization of trapped erythrocytes next to a pericyte does not demonstrate a dynamic pericyte constriction or prove causality. Figure 1d–g shows ex vivo retinal pericytes contracting in response to peroxynitrate (an hypothesized mediator of ischemic injury), similar to work done by others5, but retinal pericytes' response to ischemia may differ from the cortex, given the vast differences in number, capillary coverage and morphology of pericytes in the retina versus cortex6,7. Given the fundamental claim (that ischemia leads to long-lasting pericyte constriction and microvascular failure in cortex), the evidence should include in vivo demonstration of ischemia-induced microvascular failure and red blood cell stasis caused by constricting pericytes in cortex.

Data in Figure 3 show SMA immunolabeling in the frontoparietal cortex, and the labeled cells are identified as pericytes, but we think the identity of these cells is unclear. Previous publications using SMA or NG-2 immunolabeling8,9,10, and other figures in this paper (for example, Fig. 1), show pericytes as spindle-shaped cells whose long axes are parallel to the long axes of microvessels. The cells labeled with SMA in Figure 3c,d,i,j appear circumferentially arrayed around the capillaries, and the density of pericyte-surrounded capillaries appears to be much greater than would be expected. The SMA immunolabeling may have cross-reacted with other cell types, leading the authors to mistakenly infer that all of the immunolabeled profiles represent pericytes surrounding microvessels1. This concern is bolstered by the capillary diameters reported; the average nonischemic capillary lumen diameter reported by the authors is 10 μm1, whereas others have reported (and we have observed) that the average brain capillary lumen diameter is 4.5–6 μm3,4,11, with 10 μm at the extreme end of the spectrum. The perivascular localization of these cells ought to have been directly confirmed with double immunolabeling for CD31 or other endothelial markers; this would also allow for better measurement of capillary diameter that might agree with other published data. Alternatively, the authors might have confirmed the identity of the SMA-labeled cells with other markers for pericytes (for example, NG-2, as demonstrated in Fig. 2)1.

The authors do not address the possibility that microvascular failure after ischemia-reperfusion may be due to changes induced in upstream arteriolar constriction or occlusion1. Recent reports have highlighted the important role of feeding arterioles in microcirculatory function2,4, but the authors do not report on the effects of ischemia-reperfusion on arterioles upstream of the capillaries they examine, nor do they comment on whether peroxynitrate signaling may cause long-lasting arteriolar constriction in spite of reperfusion1.

In summary, the authors present very tantalizing hypotheses suggesting pericytes contribute to ischemia-reperfusion induced injury in the brain1, and we applaud their efforts to draw attention to the role of pericytes in microvascular function. However, we feel that with the expanding utility of in vivo imaging, there could have been more compelling evidence provided by better in vivo experiments, and our concerns about the immunolabeling data make us hesitant to accept the data at face value. Fortunately, none of these issues are especially difficult to address, and we look forward to continued work from Yemisci et al.