Pericytes Preserve Capillary Integrity to Prevent Kidney Hypoxia

Proposed in 1998 as a mechanism of kidney disease progression, “the chronic hypoxia hypothesis”¹ continues to evoke support as well as controversy.² Support comes mainly from observations of vasoconstriction, decreased renal blood flow, and peritubular capillary rarefaction in human and experimental CKD, including CKD after AKI.³ Compelling considerations based on expected or demonstrated diminution of oxygen delivery suggest that tubules in areas with capillary rarefaction become hypoxic. Assessing oxygen tension (PO₂) in such tubules is difficult. Oxygen probes inserted into kidneys puncture capillaries and damage the parenchyma; therefore, such measurements at best reflect averages of PO₂ in blood, interstitial fluid, and tubule cells in surrounding tissue. Noninvasive techniques also are fraught with methodologic imponderables, do not have resolution, and measure PO₂ over large areas. Regardless, both approaches show that PO₂ in the outer medulla, areas most injured by ischemia, is normally much lower than in the cortex. How these measurements translate to intracellular PO₂ within tubule cells is unknown, but rigorous theoretical analyses of data from mitochondrial, isolated cell, and tissue systems suggest that oxygen gradients from capillaries across interstitium to parenchymal cells are steep, so much so that PO₂ in the vicinity of mitochondria normally drops to very low levels, despite which respiration proceeds because of tight oxygen binding to cytochrome c oxidase.^4,5 PO₂ measured by probes or extrapolated to the intracellular milieu varies directly with oxygen delivery and inversely with consumption. Decreased delivery is offset by diminished consumption if glomerular filtration falls during disease, accounting for unaltered PO₂ by microprobe measurement.⁶ Such measurements give spurious information regarding hypoxia in restricted tubule microenvironments, where PO₂ may indeed become limiting and determine pathologic outcomes. If the interstitium in such microenvironments is expanded acutely by edema or chronically by connective tissue, decreased oxygen delivery—caused by persistent vasoconstriction and capillary damage or rarefaction—can damage tubules, despite apparently unchanged average PO₂ measured in the general area. Translated across steeper than normal oxygen gradients caused by expanded interstitium, PO₂ within tubule cells in the most affected areas could fall to levels insufficient to optimally maintain vital cellular functions.

Low PO₂ has been detected within tubule cells of restricted kidney microenvironments in AKI and CKD models, visualized as adduct formation of intracellular thiols with hypoxia probe pimonidazole.^6–9 Two hours after ischemia-reperfusion injury (IRI), pimonidazole adducts were detected by immunohistochemistry in deep cortical and medullary tubule cells surrounded by expanded interstitium, although PO₂ measured with oxygen probes was unchanged.⁶ Chronically after IRI as tubulointerstitial fibrosis developed, pimonidazole adducts were detected in tubule clusters in outer medulla.^8,9 Pimonidazole adducts form at PO₂<10 mmHg (approximately 1.3% O₂), concentrations at which HIF1α expression is expected to be elevated. HIF1α is robustly expressed during tubule regeneration after AKI.¹⁰ Curiously, in diverse CKD models, including CKD after AKI, HIF1α is expressed in structurally preserved tubules but is not expressed in tubules associated with fibrosis (C. Rosenberger, personal communication). Because fibrotic foci also exhibit the most capillary rarefaction and are hypoxic shown by tubule pimonidazole adduct formation, lack of HIF1α expression in these tubules begs explanation. HIF1α expression starts to increase at relatively high oxygen levels (approximately 8%) in cultured cells exposed to decreasing PO₂, rising exponentially to highest levels at approximately 0.5% O₂, but decreases steeply thereafter at lower concentrations.¹¹ Together, these observations suggest that HIF1α, a protective response to hypoxic injury,¹⁰ is expressed during repair by moderately hypoxic tubules that will recover or have recovered normal structure but not by severely hypoxic tubules undergoing atrophy. In such microenvironments destined to become fibrotic, extensive capillary disintegration and steeper oxygen gradients resulting from interstitial expansion may cause tubule oxygen delivery to fall below critical thresholds. PO₂ in these tubule cells may decline to levels below 0.5%. At these concentrations, HIF1 responses may not occur, but other oxygen-dependent systems can become compromised, preventing recovery. Tubule cells do not necessarily die but assume an “atrophic” pathologically signaling phenotype.³ Abnormal paracrine activity by such chronically hypoxic tubules may disturb mutually beneficial pericyte-capillary interactions in the interstitium to cause their separation from each other: a disruptive event that leads to capillary disintegration on one hand and pericyte-myofibroblast transformation on the other hand.¹² This premise is consistent with the operation of a vicious cycle of feedback interactions between damaged tubules and interstitial events that reinforces and makes worse the developing tubulointerstitial pathology.³ In the vicious cycle paradigm, tubule-directed disruption of normal pericyte-endothelial interactions leads to capillary rarefaction and interstitial expansion; the resulting chronic hypoxia maintains epithelial damage and prevents tubule recovery. Each arm of the vicious cycle is made worse by feedback. Eventually, capillary rarefaction increases, interstitial expansion by fibroblasts/connective tissue becomes irreversible, and tubule hypoxia is made progressively more severe.³

Experiments reported in this issue of the Journal of the American Society of Nephrology by Kramann et al.¹³ place kidney interstitial pericytes in context of peritubular capillary disintegration and hypoxia that follows kidney injury. First, they found that pericytes and pericyte-derived myofibroblasts carrying a Gli1 transgenic label increase in number and move away from capillaries during 5 days after IRI. This was associated with capillary rarefaction. Second, Kramann et al.¹³ provide incontrovertible evidence for the requirement of Gli1+ pericytes in maintaining peritubular capillary integrity. They used diphtheria toxin to ablate Gli1+ pericytes expressing diphtheria toxin receptors in a transgenic model. Acute loss of pericytes led to capillary disintegration by 10 days accompanied by tissue hypoxia, proximal tubule damage revealed as Kim1 protein in tubules with pimonidazole adducts, and expression of inflammatory cytokines. These early alterations were followed over the long term of 56 days by peritubular capillary rarefaction. Interestingly, not commented on or quantitated by Kramann et al.,¹³ the electron micrographs that they provide to illustrate capillary disintegration after pericyte ablation (figure 2D in ref. 13) show significant expansion of the interstitium between affected endothelium and adjacent proximal tubules. The authors did not examine pericyte-ablated kidneys by electron microscopy over the long term. However, our morphologic studies suggest that marked expansion of interstitial space between capillaries and tubules is a consistent feature of CKD after AKI (M.A. Venkatachalam and J.M. Weinberg, unpublished data) that the authors might have been able to observe in their 56-day kidneys also. This is important to note; in addition to the expected reduction of oxygen delivery caused by capillary rarefaction and narrowing in such kidneys, steeper oxygen gradients across the wider endothelium-tubule space are certain to make oxygen availability within tubule cells even more precarious.

Pericytes normally make intimate contacts with endothelial cells. They also exhibit complex bidirectional signaling required to maintain pericyte quiescence and endothelial integrity.¹⁴ Experiments reported earlier by Lin et al.¹² suggested that perturbation of these spatial and signaling relationships by pericyte migration away from capillaries results in transformation of pericytes to the proliferative and profibrotic myofibroblast phenotype, loss of endothelial integrity, and associated capillary damage early after kidney injury. These adverse profibrotic events were thought to result from signaling by VEGF and PDGF-B receptors in endothelial cells and pericytes, activated vicariously or more intensely by inappropriately produced and secreted growth factors or cytokines derived from tubules, endothelium, pericytes, and monocytes.¹² Pericyte identity in these innovative studies was inferred only by expression of a Col1a1-GFP transgene, and evidence implicating pericyte activation and migration in the accompanying loss of endothelial integrity was circumstantial. Moreover, the experimental design did not permit examination of the role of pericyte-capillary interactions in the causation of long-term capillary loss. By ablating pericytes in otherwise normal mice, Kramann et al.¹³ now provide clear-cut evidence for pericyte requirement in maintaining capillary integrity acutely and chronically and proof that pathology at the level of the pericyte leads to long-term capillary loss. Moreover, the authors have previously authenticated the pericyte identity of Gli1+ interstitial cells.¹⁵ Although restricted to 10-day analyses after pericyte ablation, the results also show that capillary damage induced by pericyte loss causes tissue hypoxia, tubule damage and also, inflammation.

In sum, these studies highlight and substantiate the actions of one arm of the proposed tubule-interstitium-tubule vicious cycle of interactions that prevents full recovery and gives rise to tubulointerstitial fibrosis after AKI. Although such a vicious cycle could be triggered by a primary interstitial event, such as endothelial injury and inflammation, considerations on the basis of the primacy of tubule damage in AKI, experiments showing that selective tubule damage is sufficient to cause tubulointerstitial fibrosis, and evidence for tubule-interstitium crosstalk all indicate that tubule factors are the major trigger for disturbing the physiologic pericyte-endothelium interactions that maintain capillary integrity and pericyte quiescence. Finally, we note that Gli1+ cells are only a subpopulation of pericytes in the kidney interstitium. Unfortunately, as noted by the authors, the lineage marker FoxD1 that marks all pericytes is also present in mesangial cells, vascular smooth muscle cells, and even some tubule cells. Ablating all pericytes selectively could result in rapidly fatal outcomes, precluding detailed examination of their role. Thus, Kramann et al.¹³ were prescient in selecting Gli1+ cells as their target. This strategy to ablate a pericyte subpopulation produced mild injury but regardless, yielded vital information.

Disclosures

None.