Introduction

In the 19th century, a French scientist, Charles-Marie Benjamin Rouget, reported the presence of a population of contractile cells in small blood vessels, referred to as Rouget cells [1]. Later, in the 20th century, a German scientist, Karl Wilhelm Zimmermann, renamed these cells “pericytes” due to their distinct anatomical position around the vasculature [2]. The word pericyte comes from “peri” meaning “around” and “cyte” from Latin, which has origins from the Greek “kytos” (cell), properly illustrating a cell encircling a blood vessel [3]. Until the end of the 20th century, pericytes were identified still based mainly on their anatomical location and morphology. Pericytes have long processes surrounding blood vessel walls and are widely dispersed in all tissues [4]. They encircle endothelial cells, and communicate with them along the length of the blood vessels by paracrine signaling and physical contact [5]. In the brain, the ratio of endothelial cells to pericytes is ~3:1 [6, 7], implying an immense importance of cerebral pericytes.

Formerly, the accurate distinction of pericytes from other perivascular cells was impossible, as light and electron microscopy were the only technologies able to visualize these cells, limiting the information acquired. This resulted in the illusory notion that pericytes are merely inert supporting cells, limited exclusively to the physiological function of vascular stability. Already in the 21th century, the combination of fluorescent and confocal microscopy with genetic tools, such as fate lineage tracing, enabled the discovery of novel and unexpected roles for pericytes in health and disease [8]. Recently, quickly expanding insights into the pathophysiological functions of pericytes have attracted the attention of many researchers.

Pericytes participate in blood vessel development, maturation, and permeability, as well as contributing to their normal architecture [9, 10]. They regulate blood flow [11, 12], and affect coagulation [13]. Pericytes also collaborate with astrocytes, neurons, and endothelial cells, forming the neurovascular unit [12, 14, 15], to regulate maintenance of the functional integrity of the blood brain barrier [16,17,18,19,20,21]. This may occur via pericyte-derived molecules, such as platelet-derived growth factor subunit B (PDGFB)/PDGF receptor-beta (PDGFRβ) signaling, which is indispensable for the formation and maturation of this barrier [22]. In addition, pericytes perform several immune functions [23], regulate lymphocyte activation in the retina [24, 25], attract innate leukocytes to exit through sprouting blood vessels in the skin [26], and contribute to the clearance of toxic cellular byproducts, having direct phagocytic activity in the brain [27]. Interestingly, following white matter demyelination, pericytes promote the differentiation of oligodendrocyte progenitors involved in central nervous system regeneration via a2-chain of laminin [28]. Pericytes may also behave as stem cells in several tissues [29], generating other cell populations, as well as regulating the behavior of other stem cells, as hematopoietic stem cells in their niches [30,31,32,33,34]. Note that pericytes from distinct peripheral tissues may have various properties, and may differ from those in brain. Increasing evidence also shows that brain pericytes alter their traits following stimuli and develop stemness, demonstrating their plasticity [35,36,37,38,39].

Pericytes exhibit structural plasticity during embryonic cerebral development, participating in vascular remodeling [40]. Understanding pericyte behavior in the adult brain is a central question in neuroscience, as these cells may play central roles in the pathogenesis of neurodegenerative disorders. Nevertheless, whether pericytes participate in vascular remodeling in the adult brain remains unknown. Now, in a recent article in Cell Reports, Berthiaume and colleagues investigated the behavior of pericytes in the adult mouse brain [41]. The authors revealed pericyte plasticity in the adult brain by using elegant state-of-the-art techniques, including two-photon microscopy in combination with sophisticated Cre/loxP in vivo tracing technologies. Berthiaume and colleagues imaged, at high-resolution over several weeks, cerebral pericytes in NG2-CreER/TdTomato, Myh11-CreER/TdTomato, and PDGFRβ-Cre/YFP mice. These experiments unveiled that pericytes comprise a quasi-continuous, non-overlapping network along the entire length of blood vessels. Interestingly, the pericyte prolongations were not stable in length, extending or retracting during the period of analysis. Then, the authors explored the effect of pericyte death on its neighboring pericytes. After pericyte ablation, using targeted two-photon irradiation, Berthiaume and colleagues showed that adjacent pericytes extend their processes into the uncovered area, covering the exposed blood vessel [41]. Strikingly, neighboring pericytes are able to reverse the vascular dilatation that occurs after pericyte depletion [41] (Fig. 1). Thus, this longitudinal imaging study demonstrated pericyte plasticity in the adult brain.

Fig. 1
figure 1

Cerebral pericyte plasticity in response to neighbor ablation. Pericytes are present around blood vessels in the brain. The study of Berthiaume and colleagues now suggests a novel role for pericytes in vascular remodeling in the adult brain [41]. After pericyte ablation, using targeted two-photon irradiation, adjacent pericytes extend their processes to cover the exposed endothelial bed, and reverse the vascular dilatation that occurs after pericyte depletion. Future studies will reveal in detail the cellular and molecular mechanisms involved in this process in the brain microenvironment.

Here, we consider these findings, and evaluate recent advances in our knowledge of pericyte biology in the brain.

Perspectives and Future Directions

Pericyte Heterogeneity in the Brain

Pericytes are heterogeneous regarding their distribution, phenotype, marker expression, origin, and function [42]. In the past century, pericytes were distinguished into three types based on their mural location and morphology: pre-capillary, mid-capillary, and post-capillary [2, 43]. Berthiaume and colleagues exclusively studied pericytes surrounding mid-capillary regions [41]. Thus, it remains unknown whether pre-capillary and post-capillary pericytes present different behavior. This should be taken into consideration in future work; discovering specific markers for pre-capillary and post-capillary pericytes will help to address this question. Separately analyzing the behavior of pericytes from different locations in the blood vessels may reveal their functional heterogeneity. Pericyte heterogeneity is also based on their molecular marker expression profiles. Capillary pericytes express desmin but are commonly negative for α smooth muscle actin (αSMA), while venular pericytes express both desmin and αSMA proteins [44]. Moreover, Kir6.1 is highly expressed in a subset of brain pericytes, but is undetectable in others [45]. In addition, arteriolar pericytes that do not express the leptin receptor (LEPR) have been described to be distinct from sinusoidal pericytes in the bone marrow that express LEPR [31, 46]. Also, both nerve/glial antigen 2-positive (NG2+) and -negative (NG2−) pericytes have been described in the skin [26]. Furthermore, pericytes positive and negative for glutamate aspartate transporter and the cytoskeletal protein Nestin have been described in the spinal cord [47, 48].

Cerebral pericytes also differ in their embryonic origins [42]. While pericytes in coleomic organs are mesoderm-derived [7], most cephalic pericytes are of neuroectodermal origin [49], and recent studies have shown that a subpopulation of pericytes in the embryonic brain may derive from hematopoietic progenitors [42, 50]. All these descriptive characteristics in which pericytes differ are also important regarding their functions, as pericytes in distinct locations [30], with different marker expression profiles [51], and from varying origins [42] differ in their functions. For instance, after brain injury, while one subset participates in scar tissue formation [52], another is capable of generating new blood vessels [53]. Thus, subsets of pericytes can contribute to distinct pathological conditions in varying ways. Importantly, similar analysis as done by Berthiaume et al. (2018), should be performed on cerebral pericytes from other blood vessels and in distinct brain regions, as their behavior may differ.

Pericyte Markers

Several molecular markers help to identify pericytes, such as PDGFRβ, NG2, proteoglycan (CSPG4), myosin heavy chain 11 (Myh11), aminopeptidase N (CD13), αSMA, regulator of G protein signaling 5, desmin, vimentin, ATP-binding cassette, subfamily C (CFTR/MRP), member 9 (SUR2), alkaline phosphatase, CD146, CD133, endosialin, potassium inwardly-rectifying channel, subfamily J, member 8 (Kir6.1), Tbx18, vitronectin, and interferon-induced transmembrane protein 1 (Ifitm1), among others (Table 1) [29, 48, 54,55,56,57,58,59]. The difficulty now is generating markers, antibodies, and mice to take advantage of this information. Unfortunately, there is no single molecular marker yet that can be used to unequivocally and exclusively label the whole population of pericytes. Berthiaume and colleagues imaged labeled pericytes in NG2-CreER/TdTomato, Myh11-CreER/TdTomato, and PDGFRβ-Cre/YFP mice. Although NG2, Myh11, and PDGFRβ proteins can be expressed in pericytes, none of them is specific to pericytes, or labels all pericytes, as oligodendrocyte progenitors also express NG2 [60]. Under special conditions, microglia may also express NG2 [61]. Interestingly, pericytes that do not express NG2 proteoglycan also exist [26]. In addition, Myh11 labels vascular smooth muscle cells, and is expressed only in a subgroup of pericytes [30]. PDGFRβ expression also is not restricted to pericytes. Several stromal cells such as vascular smooth muscle cells [62, 63] and fibroblasts [64] express this cell-surface tyrosine kinase receptor [7]. Note that Berthiaume and colleagues used PDGFRβ-Cre/YFP mice, in which all the cells derived from PDGFRβ-expressing cells are also labeled with fluorescence [41]. Since PDGFRβ is broadly expressed throughout the embryo in embryonic stages in several cell types, PDGFRβ-Cre/YFP mice are not the best mouse model for analyzing pericyte behavior, as several other cell populations may be labeled at the same time [29, 65]. Therefore, as PDGFRβ expression is more restricted in adult animals, the use of PDGFRβ-CreER/YFP mice instead would be more appropriate for the study of PDGFRβ-expressing pericytes in the brain [66]. Importantly, not all cells in perivascular locations are necessarily pericytes [67]. Besides pericytes, other cells surrounding blood vessels have been described, including fibroblasts [68], macrophages [69, 70], microglia [71], adventitial cells [72], and vascular smooth muscle cells [73]. Altogether, this raises the possibility that some of the observations by Berthiaume et al. (2017) are from a different, non-pericytic, cell type. Currently, the state-of-the-art identification of pericytes in tissue preparations relies on a combination of anatomical localization (covering endothelial cells and underlying the basal lamina), morphology, and the co-expression of at least two pericytic molecular markers. The discovery of a single molecular marker specific to all cerebral pericytes will facilitate the study of the behavior of these cells in the brain.

Table 1 Pericyte markers.

Pericytes as Stem Cells

In the last decade, the potential of pericytes to contribute to tissue regeneration/homeostasis as tissue-resident progenitors has been established by numerous studies [74,75,76]. Pericytes not only participate in the formation of new blood vessels [53], but their ability to differentiate into the neural lineage has also been demonstrated [48, 77,78,79,80,81,82,83]. Therefore, pericytes are expected to be able to proliferate and multiply when activated. Surprisingly, Berthiaume and colleagues show that pericytes are not activated to proliferate in response to the death of one adjacent pericyte; instead, they extend prolongations onto the uncovered endothelium [41]. It remains unknown whether pericytes continue to behave similarly when more adjacent pericytes are ablated. Is the elimination of adjacent pericytes unable to activate pericytic multiplication? Does this depend on the number of pericytes that die? Also, a question arises regarding the plasticity of pericytes in vivo. Are cerebral pericytes able to form other pericytes in the adult brain? This could be tested by time-lapse high-resolution imaging analysis of lineage tracing in pericyte-specific mouse models.

Pericyte Communication with Other Tissue Components in Their Microenvrionment

Which signaling molecules are needed to activate the extension of pericyte processes into the adjacent endothelial bed? And which signaling molecules are important for these pericytes to reverse the vasodilation that occurs after pericyte death [84]? Although Berthiaume and colleagues have revealed how pericytes respond to the deletion of a neighboring pericyte, they did not explore the molecular and cellular mechanisms involved in this process. A recent study in the spinal cord has shown that expression of the enzyme aromatic L-amino acid decarboxylase is important for pericyte-induced vasoconstriction after spinal cord injury [85, 86]. Is this enzyme also important in the cerebral pericyte after ablation of its neighbor? In addition to studies of genetic mouse models, transcriptomic and single-cell analysis of pericytes after ablation of their neighbors will help us to understand the molecular mechanisms involved in those processes in the brain microenvironment.

Moreover, it is still not understood whether any other cells are involved in this behavior. Do endothelial cells cross-talk to communicate that they have become uncovered? In addition, as discussed above, there are several other perivascular cells in the cerebral vasculature. How other perivascular cells participate in this process remains to be elucidated. Interestingly, after irradiation of single pericytes [41], what happens to their remnants? Does the cellular debris cause damage to neighboring cells and inflammation? Or are the cellular remnants important for adjacent pericytes to extend their prolongations in this area? Moreover, as macrophages usually engulf cellular debris, do they communicate with neighboring pericytes, activating them to expand their processes? The relationships between brain pericytes and microglia/macrophages have been addressed in recent studies. Interestingly, during development, macrophages may generate pericytes in the brain [50, 87, 88], while after stroke, pericytes may form macrophages [35, 36, 89].

It well known that pericytes produce several signaling molecules as well as responding to various signals and communicating with other cells, such as endothelial cells [90,91,92]. Interestingly, some evidence suggests direct communication with astrocytes as well. Astrocyte-derived glutamate may regulate gene transcription in pericytes [93]. Also, astrocytes may influence pericyte behavior by altering PDGFβ signaling in which pericytes play a key role [94, 95]. In contrast, very little is known about cross-talk within the population of pericytes. Future studies will need to explore how pericytes communicate with their peers.

Pericytes in Disease

Berthiaume and colleagues imaged pericytes in the healthy adult mouse brain [41]. Are the normal functions of the pericytes changed when one cell occupies the space of two? Interestingly, the authors followed pericyte behavior for several weeks up to almost 2 months [41]. It remains unknown whether this pericyte behavior is a temporary solution before the lost pericytes are regenerated. Is this process reversed after a longer period of time? Importantly, as the interruption of pericyte contact with endothelial cells may lead to endothelial hyperplasia [10], the brain vasculature should be followed for a longer time. Also, it remains unknown whether pericytes behave similarly during different life stages, such as embryonic development, the postnatal period, and aging. Furthermore, it remains to be studied how pericytes respond to the ablation of their neighbors in various brain diseases, in which it is well accepted that their dysfunction plays pivotal roles (Table 2), such as Alzheimer’s disease [96,97,98,99], amyotrophic lateral sclerosis [100, 101], diabetic retinopathy [102], cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy [103], epilepsy [104], human immunodeficiency virus-related dementia [105], brain tumors [106], primary familial brain calcification [107], and diabetes-related microangiopathy [108]. As pericyte degeneration also causes breakdown of the blood-brain barrier, leading to the entrance of blood-derived toxic substances into the central nervous system [16], future studies should explore how neighboring pericytes act to protect the brain against such toxicity. Recently, Arango-Lievano and colleagues have used the two-photon microscopy approach to track changes in perivascular cells during pathophysiological progression in the epileptic brain [109].

Table 2 Pericyte modifications occurring in central nervous system diseases.

Modern Technologies to Study Pericyte Biology

Methods to eliminate pericytes from the tissue microenvironment, enabling analysis of the functioning of a tissue without pericytes, may lead to advances in our understanding of the role of pericytes in specific organs. Multiple pharmacological drugs to induce apoptotic cell death are accessible. Nevertheless, most lack spatiotemporal and cell-type specificity [110,111,112,113]. Modern state-of-the-art experimental approaches for specific cell ablation in vivo have been created, including genetically-encoded death receptors [113,114,115,116], two-photon thermal ablation [113, 117,118,119,120], chromophore-assisted light inactivation [113, 121,122,123,124,125,126,127], and more recently two-photon chemical apoptotic targeted ablation [113]. Unfortunately, every tool has limitations, for instance, prolonged illumination requirements, non-specific tissue damage from the spilling of cellular debris, induction of local inflammation, and the need to efficiently and accurately deliver the dyes or genetic materials for targeted cell killing. Ideally, these limitations can be overcome by combining different methodologies to answer the same questions. Thus, in the future, comparing distinct methods for precise ablation of pericytes in the brain should be used to achieve a complete understanding of pericyte behavior in the central nervous system.

Conclusion

The study by Berthiaume and colleagues reveals a novel and important behavior of cerebral pericytes in response to ablation of adjacent pericytes. However, our understanding of cross-talk between different cell types present in the brain vascular microenvironment remains limited, and the complexity of these interactions in distinct physiological and pathological conditions should be elucidated in future studies. The huge challenge that we face now is how to translate animal research to humans.