Abstract

   The pericyte coverage of microvessels is altered in metabolic diseases,
   but the mechanisms regulating pericyte–endothelial cell communication
   remain unclear. This study investigated the formation and function of
   pericyte tunneling nanotubes (TNTs) and their impact on endothelial
   cell metabolism. TNTs were analyzed in vitro in retinas and co-cultures
   of pericytes and endothelial cells. Using mass spectrometry, the
   influence of pericytes on endothelial cell metabolism was examined.
   TNTs were present in the murine retina, and although diabetes was
   associated with a decrease in pericyte coverage, TNTs were longer. In
   vitro, pericytes formed TNTs in the presence of PDGF, extending toward
   endothelial cells and facilitating mitochondrial transport from
   pericytes to endothelial cells. In experiments with
   mitochondria-depleted endothelial cells displaying defective TCA cycle
   metabolism, pericytes restored the mitochondrial network and
   metabolism. 19,20-Dihydroxydocosapentaenoic acid (19,20-DHDP), known to
   disrupt pericyte–endothelial cell junctions, prevented TNT formation
   and metabolic rescue in mitochondria-depleted endothelial cells.
   19,20-DHDP also caused significant changes in the protein composition
   of pericyte-endothelial cell junctions and involved pathways related to
   phosphatidylinositol 3-kinase, PDGF receptor, and RhoA signaling.
   Pericyte TNTs contact endothelial cells and support mitochondrial
   transfer, influencing metabolism. This protective mechanism is
   disrupted by 19,20-DHDP, a fatty acid mediator linked to diabetic
   retinopathy.

   Keywords: polyunsaturated fatty acid, dihydroxydocosapentaenoic,
   diabetes, pericyte-to-endothelial cell communication

1. Introduction

   The microcirculation plays a unique role in tissue homeostasis and has
   the capacity to adjust the supply of blood within tissues to meet the
   demands of specific cell populations for oxygen and nutrients. Exchange
   takes place across the membranes of endothelial cells that make up the
   capillaries, and although these cells can release vasoactive factors to
   promote dilation, i.e., nitric oxide or endothelium-derived
   hyperpolarizing factors, an intimate relation with pericytes in
   precapillary arterioles and postcapillary collecting venules is
   required for optimal function and the maintenance of barrier function.
   The term pericyte is now known to reflect a number of mural cell
   subsets and has been attributed roles in promoting and maintaining
   endothelial cell quiescence in newly generated vessels as well as
   contraction to regulate micro-environmental blood flow (for a review,
   see Wu et al. [[36]1]).

   The role of endothelial dysfunction in the development of
   cardiovascular disease has been well documented [[37]2,[38]3], but the
   contribution of alterations in pericyte function to cardiovascular and
   pulmonary disease development is underappreciated. Indeed, it now seems
   that the pericyte loss or migration away from the vasculature, which
   was previously described in retinas from animals and humans with
   diabetes [[39]4], also occurs during the development of the
   microcirculatory dysfunction that accompanies heart failure
   [[40]5,[41]6]. While the classical contact between pericytes and
   endothelial cells seems to be via their shared basal lamina, long, fine
   structures that enable the transport of material over many tens of
   microns have been reported, which connect pericytes with other
   pericytes or with endothelial cells in an adjacent vessel. These
   structures or tunneling nanotubes (TNTs) were initially described in
   2004 as “nanotubular highways” [[42]7,[43]8] and are now known to
   support the intercellular trafficking of many types of cellular
   compounds ranging from Ca^2+ and electrical signals [[44]9], micro-RNAs
   [[45]10], and proteins [[46]11,[47]12] to preformed MHC complexes
   [[48]13]. They can even transport endosomal vesicles [[49]14],
   organelles such as lysosomes, and mitochondria in a process referred to
   as “organellar diakinesis” [[50]15,[51]16]. Depending on the
   conditions, the transfer of mitochondria through TNTs has been proposed
   to rescue impaired mitochondrial function in recipient cells and to
   propagate degenerative states linked to mutated mitochondrial DNA
   [[52]17,[53]18]. We previously linked the pericyte depletion and the
   loss of retinal barrier function in diabetic retina with the generation
   of a diol of docosahexaenoic acid (DHA), i.e.,
   19,20-dihydroxydocosapentaenoic acid (19,20-DHDP) [[54]19]. Therefore,
   the aim of this study was to identify pericyte TNTs in vivo and in
   vitro to study their generation, the impact of pericyte TNTs on
   endothelial cell metabolism, and the role of 19,20-DHDP in these
   processes.

2. Materials and Methods

2.1. Animals

   Ins2Akita (C57BL/6-Ins2Akita/J) mice carrying a mutation in the insulin
   2 gene [[55]20] were obtained from Jackson Laboratory. The animals were
   housed in conditions that conform to the Guide for the Care and Use of
   Laboratory Animals published by the U.S. National Institutes of Health
   (NIH publication no. 85–23). Mice were sacrificed using 4% isoflurane
   and cervical dislocation for the isolation of the retina. Only
   10-month-old male mice were used in the current study.

2.2. Cell Culture

   Pericytes: Human brain vascular pericytes were purchased from ScienCell
   Research Laboratories (Carlsbad, CA, USA) and cultured in DMEM/F-12
   Media (11320033, Gibco, Waltham, MA, USA) supplemented with 5% FCS
   (Gibco, Waltham, MA, USA), insulin (Sigma-Aldrich, Taufkirchen,
   Germany), and human-FGF. For this study, six independent batches of
   pericytes were used up to passage seven.

   Endothelial cells: Human umbilical vein endothelial cells were isolated
   and cultured, as described previously [[56]21], in ECGM2 Media
   (C-22111, Promocell, Heidelberg, Germany) supplemented with 5% FCS
   (Gibco, Waltham, MA, USA) and the commercially available supplement mix
   (C-39216, Promocell, Heidelberg, Germany). The cultivated cells were
   used up to passage four.

2.3. Endothelial Cell Pericyte Co-Culture

   2D Endothelial cell pericyte co-culture: Endothelial cells were seeded
   on dishes coated with 1% gelatin in serum and growth factor-depleted
   medium containing fatty acid BSA (0.1%). Once the endothelial cells
   adhered, pericytes were added at a ratio of 1:3 and the mixture was
   co-cultured for 6 or 16 h in the presence of PDGF-BB (0.3–3 ng/mL,
   Hz-1308, Proteintech, Planegg-Martinsried, Germany). Experiments were
   performed in the absence and presence of 19,20-DHDP (10 µmol/L, Cayman,
   Ann Arbor, MI, USA) and in the presence of an sEH inhibitor (t-AUCB, 10
   µmol/L). In some experiments, endothelial cells were isolated from the
   co-cultures at the end of the incubation period using CD144-coated
   magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany). Pericytes
   were collected from the flowthrough and were enriched by
   centrifugation. Thus, isolated cells were directly processed for
   immunophenotyping, and for the rest of the experiments, the cells were,
   after separation, frozen in liquid nitrogen and further processed as
   described.

   3D Filter-based co-culture: Endothelial cells were seeded at a high
   density (600,000 cells per well) on the underside of gelatin-coated
   six-well Transwell filter inserts (0.4 µm Sarstedt, Nümbrech, Germany).
   After 6 h, the inserts were inverted and pericytes (600,000 per well)
   were seeded on the opposite side of the filter. The co-culture was
   maintained for 5 days to ensure that endothelial cells and pericytes
   were confluent and were able to form endothelial cell–pericyte
   junctions in the pores of the filter, as described [[57]19]. The
   co-cultures were then treated with either the solvent (0.1% DMSO) or
   19,20-DHPD (10 µmol/L) in the presence of the sEH inhibitor (t-AUCB, 10
   µmol/L) for 12 h. The cells were removed by scraping both sides of the
   filter, and proteins in the filter pores were isolated by incubating
   the filter with RIPA-lysis buffer (Tris/HCL 50 mmol/L pH 7.5, NaCl 150
   mmol/L, NaPi 10 mmol/L, NaF 20 mmol/L, 1% sodium deoxycholate, 1%
   Triton, and 0.1% SDS) and recovered by centrifugation (15,000 rpm, 15
   min, 4 °C).

2.4. TNT Formation

   To study TNT formation, pericytes (10,000) were seeded on a 1%
   gelatine-coated µ-slide 8 well (ibidi, Gräfelfing, Germany), and when
   ~40% confluent, the medium was replaced with serum and growth
   factor-depleted medium containing fatty acid free bovine serum albumin
   (BSA; 0.1%) and either the solvent (0.01% DMSO) or PDGF-BB (0.3–3
   ng/mL) for 6 h. The experiments were performed in the absence and
   presence of 19,20-DHDP (10 µmol/L) and stopped by the addition of 4%
   paraformaldehyde (PFA). TNTs were visualized using a confocal
   microscope (Carl Zeiss LSM 780, Jena, Germany) by rendering the average
   phalloidin-546 signal and quantifying the length in Fiji ImageJ
   (version 1.53t, NIH).

2.5. Detection of Mitochondria and Mitochondrial Exchange

   Mitochondria were visualized after incubating cells with MitoTracker
   Red (1:1000, 22426, Thermo Fisher, Waltham, MA, USA) or MitoTracker
   Green (1:1000, M7514, Thermo Fisher) for 2 h. Thereafter, the cells
   were washed extensively with phosphate buffered saline (PBS) and
   cultured as described above. The MitoTracker signals were visualized in
   cells treated with either the solvent (0.1% DMSO), PDGF-BB (0.3–3
   ng/mL) or 19,20-DHDP (10 µmol/L) for 6 h, and imaging was performed
   using a confocal laser scanning microscope (LSM 780, Zeiss, Jena,
   Germany). In some experiments, the MitoTracker Red signal was assessed
   by flow cytometry (BD LSRFortessa X-20, BD, Heidelberg, Germany) in
   CD31+ endothelial cells isolated from 2D endothelial cell-pericyte
   co-cultures. The data were analyzed using FlowJo Vx software (version
   18.8.1, TreeStar, Ashland, OR, USA).

2.6. Mitochondria Depletion

   The selective deletion of mitochondrial DNA (mtDNA) was induced as
   described [[58]22,[59]23] by culturing freshly isolated endothelial
   cells in endothelial cell growth medium (ECGM)-2 (PromoCell,
   Heidelberg, Germany) containing ethidium bromide (25 µmol/L; Roth,
   Karlsruhe, Germany) and uridine (50 µg/mL; Sigma-Aldrich, Taufkirchen,
   Germany) for 14 days. Mitochondria-depleted cells were selected using
   carbonyl cyanide m-chlorophenyl hydrazone (CCCP, 1 ng/mL; C2920,
   Sigma-Aldrich, Taufkirchen, Germany).

2.7. Immunohistochemistry

   Isolated retinas and cells were fixed in paraformaldehyde (PFA, 4%) for
   2 h at room temperature. The samples were then blocked and
   permeabilized in PBS containing 0.5% Triton X-100 and 5% donkey serum
   (4 h, room temperature). After permeabilization, the samples were
   incubated with primary antibodies overnight (4 °C), and after washing,
   they were incubated with appropriate secondary antibodies (1:200 in
   PBS) together with DAPI (10 ng/mL, ThermoFisher, Waltham, MA, USA) for
   one hour. The samples were then washed and mounted in Fluoromount-G
   (ThermoFisher, Waltham, MA, USA).

   The following primary antibodies were used: NG-2 (1:200, AB5320,
   Millipore, Darmstadt, Germany), CD31 (1:100, 557355, BD Biosciences,
   Heidelberg, Germany), connexin 43 (1:100, 3512, Cell Signaling, Leiden,
   Netherlands), VDAC (1:100, sc390996, Santa Cruz, Heidelberg, Germany),
   Tom20 (1:100, sc17764, Santa Cruz, Heidelberg, Germany). The Alexa
   Fluor-coupled secondary antibodies were from Thermo-Fisher (1:200,
   Waltham, MA, USA) or fluorophore conjugated phalloidin-546 (1:1000,
   A22283, Thermo Fisher, Waltham, MA, USA).

2.8. Immunoblotting

   For immunoblotting, the cells were lysed in RIPA lysis buffer (Tris/HCL
   50 mmol/L, pH 7.5, NaCl 150 mmol/L, NaPi 10 mmol/L, NaF 20 mmol/L, 1%
   sodium deoxycholate, 1% Triton X-100, and 0.1% SDS containing protease
   and phosphatase inhibitors), and detergent-soluble proteins were
   suspended in SDS-PAGE sample buffer. The samples were separated by
   SDS-PAGE and subjected to Western blotting, as described [[60]24].
   Membranes were blocked in Rotiblock (Roth, Karlsruhe, Germany) for 2 h
   and incubated overnight at 4 °C with primary antibodies. Following
   extensive washing, horseradish peroxidase-conjugated secondary
   antibodies were added for 2 h, and protein bands were visualized by
   using Lumi-Light plus (Roche, Mannheim, Germany). Luminescence was
   captured using an image acquisition system (Fusion FX7; Vilber-Lourmat,
   Torcy, France).

   The following primary antibodies were used: anti-phospho Tyr751 PDGFR
   (1:1000, Cat. # 3161), anti-PDGFR (1:1000, Cat. # 3169), anti-phospho
   Ser473 Akt (1:1000, Cat. # 4058), anti-Akt (1:1000, Cat. # 9272), and
   anti-phospho Thr202/Tyr204 Erk1/2 (1:1000, Cat. # 9106), which were all
   from Cell Signaling (Leiden, The Netherlands). The antibody against
   Erk1/2 (1:1000, Cat. # ABS44) was from Sigma-Aldrich (Taufkirchen,
   Germany), the antibodies recognizing VDAC (1:1000, Cat. # sc390996) and
   Tom20 (1:1000, Cat. # sc17764) were from Santa Cruz (Heidelberg,
   Germany), and the anti β-actin (1:3000, Cat. # MAK6019) was from
   Linaris (Eching, Germany). The peroxidase conjugated secondary
   antibodies used at a dilution of 1:20,000 in Tris-buffered-saline-Tween
   were goat anti-rabbit (Cat #. 401393) and goat anti-mouse (Cat. #
   401253), both from Merck (Darmstadt, Germany).

2.9. Measurement of Mitochondrial Oxidation by MitoSOX

   Mitochondrial ROS production was assayed as described [[61]25], with
   some modifications. Mitochondrial depleted endothelial cells were
   cultured with and without pericytes in the presence of PDGF-BB (1
   ng/mL) or PDGF-BB and 19,20-DHDP (10 µmol/L) for 12 h. Thereafter, the
   cells were incubated with MitoSOX (1 µmol/L; ThermoFischer) for 30 min
   at 37 °C. The cells were recovered by enzymatic digestion and incubated
   with antibodies directed against CD31 (1:100, Cat #. 561653,
   BD-Biosciences, Heidelberg, Germany). The MitoSOX and MitoTracker Red
   signals in CD31+ endothelial cells were measured by flow cytometry (BD
   LSRFortessa X-20, BD, Heidelberg, Germany), and the data were analyzed
   using FlowJo Vx software (version 18.8.1, TreeStar, Ashland, OR, USA).

2.10. Proteomics

   The mass spectrometry proteomics data including a detailed method
   description and data analysis have been deposited to the
   ProteomeXchange Consortium via the PRIDE [[62]26] partner repository,
   with the dataset identifier PXD049131.

   MaxQuant Search data were further analyzed using Perseus 1.6.1.3
   [[63]27]. Reverse identifications and common contaminants were removed.
   Proteins were filtered to be identified at least three times (n = 3) in
   one experimental group. Missing values were replaced by the lowest
   value in the matrix. Most affected proteins were determined by
   Student’s t test and permutation-based FDR (q value).

2.11. Metabolomics

   Sample extraction: Trifluoroethanol:H[2]O (1:1) was added to each
   sample. The samples were shortly vortexed and incubated on ice for 10
   min. Next, methanol:ethanol (1:1) was added and the samples were
   incubated for another 10 min. Ice-cold water, containing internal
   standards—13C-fumarate-C2, 13C-citrate-C2 and 13C-succinate-C1,4 (all
   0.02 mmol/L)—was added. After 10 min, the samples were sonicated on ice
   for 30 s and centrifuged (14,000 rpm, 10 min, 4 °C). The supernatant
   was recovered and transferred to two separate tubes before being
   evaporated under a stream of N2 gas. The remaining aqueous fraction was
   shock-frozen in liquid nitrogen and subsequently freeze-dried (Alpha
   3–4 LSCbasic system, Martin Christ, Osterode am Harz, Germany).

   The dried samples were reconstituted in H[2]O:acetonitrile (AcN; 1:3)
   for the analysis of pyruvate, 2-HG, succinate, malate, and fumarate
   (method 1—HILIC separation) or in 100% H[2]O containing 0.2% formic
   acid for the analysis of the remaining metabolites of the TCA cycle and
   the amines asparagine, glutamine, and glutamate (method 2—C18
   separation). The samples were centrifuged for 10 min at 14,000 rpm and
   4 °C and the supernatant was transferred to MS vials and subjected to
   mass spectrometric analysis.

   Quantification and analysis: Method 1 (HILIC separation): The dried
   samples were reconstituted in H2O: AcN (1:3) and then centrifuged
   (14,000 rpm, 4 °C, 10 min), and the supernatant was transferred to mass
   spectrometry (MS) vials. Thereafter, the samples (4 µL) were injected
   via an Infinity II Bio liquid chromatography system into a 6495C triple
   quadrupole mass spectrometer (both Agilent Technologies, Waldbronn,
   Germany). The metabolites were separated on an Acquity BEH Amide HILIC
   column (1.7 µm, 2.1 × 100 mm, Waters) using the following mobile phase
   binary solvent system and gradient at a flow rate of 0.4 mL/min. Mobile
   phase A consisted of 100% water with 10 mmol/L ammonium acetate and 5
   µmol/L medronic acid (pH 9.3). Mobile phase B consisted of 100%
   acetonitrile with 10 mmol/L ammonium acetate. The following 23 min
   gradient program was used: starting with 85% B, 0–1 min 85% B, 1–8 min
   75% B, 8–12 min 60% B, 12–15 min 10% B, 15–18 min 10% B, 18–19 min 85%
   B, and 19–23 min 85% B. The column compartment was set to 20 °C.
   Metabolites were detected with authentic standards and/or via their
   accurate mass, fragmentation pattern, and retention time in polarity
   switching ionization dynamic MRM AJS-ESI mode and quantified (where
   appropriate) via a calibration curve. The gas temperature of the mass
   spectrometer was set to 200 °C, the gas flow was 14 L/min, the
   nebulizer was set to 40 psi, and the sheath gas flow was 12 L/min, 375
   °C. The capillary voltages were set at 4000/2500 V, with a nozzle
   voltage of 500/0 V. The voltages of the High-Pressure RF and
   Low-Pressure RF were set to 150/90 and 60/60 V, respectively. The
   metabolite peaks were annotated with Skyline-daily (version
   22.2.1.278). Quantification was achieved by comparing sample values
   with the slope of a standard curve of the appropriate standard, and the
   data are presented as the concentration per µg protein.

   Method 2 (C18 separation): The dried samples were reconstituted in
   H[2]O containing 0.2% formic acid and centrifuged (14,000 rpm, 4 °C, 10
   min), and the supernatant was transferred to MS vials. Thereafter, the
   samples (8 µL) were injected as described above. The metabolites were
   separated on an Acquity HSS T3 column (1.8 µm, 2.1 × 150 mm, Waters,
   Eschborn, Germany) using the following mobile phase binary solvent
   system and gradient at a flow rate of 0.35 mL/min: mobile phase A
   consisted of H[2]O containing 0.2% formic acid, and mobile phase B
   consisted of acetonitrile containing 0.2% formic acid. The following 11
   min gradient program was used: starting with 1% B, 0–6 min 1% B, 6–7
   min 80% B, 7–8 min 80% B, and 8–11 min 1% B. The column compartment was
   set to 30 °C. Metabolites were detected as mentioned above. The gas
   temperature was 240 °C, the gas flow 19 L/min, the nebulizer was set to
   50 psi, and the sheath gas flow was set to 11 L/min (400 °C). The
   capillary voltages were set at 1000/1000 V, with a nozzle voltage of
   500/500 V. The voltages of the High-Pressure RF and Low-Pressure RF
   were set to 100/100 and 70/70 V, respectively. Metabolite peaks were
   annotated with Skyline-daily (version 22.2.1.278). Quantification was
   achieved by comparing the sample values with the slope of a standard
   curve of the appropriate standard, and the data are presented as the
   concentration per µg protein.

2.12. Statistical Analyses

   The data are expressed as the mean ± SEM. Statistical evaluation was
   performed using Student’s t test for unpaired data. One-way ANOVA
   (Bonferroni) and two-way ANOVA (followed by Tukey’s multiple
   comparisons test) were used, where appropriate. The statistics for
   metabolomics were performed with Metaboanalyst 5.0. The statistical
   tests are described in the figure legend for each experiment. Values of
   p < 0.05 were considered statistically significant.

3. Results

3.1. Tunneling Nanotubes and Migrating Pericytes in the Retina

   The pericyte coverage of vessels in the second capillary layer of
   retinas from 10-month-old wild-type mice was high (approximately 50%),
   and some cells were observed to extend long processes (7–30 µm) that
   spanned capillaries ([64]Figure 1a). Such processes have been
   classified as TNTs [[65]7,[66]8,[67]28], and although, previous reports
   have suggested that 30% of retinal pericytes extrude them [[68]29]; the
   incidence we observed was much lower. In retinas from diabetic
   Ins2Akita mice, pericyte coverage was decreased, i.e., approximately
   20% ([69]Figure 1b), which is consistent with the reported pericyte
   migration and loss in this model [[70]4,[71]19]. Despite this, more
   TNTs were apparent in retinas from diabetic mice than in non-diabetic
   mice ([72]Figure 1c), and they were significantly longer with the size
   distribution skewed to higher values ([73]Figure 1d). Three distinct
   pericyte morphologies were distinguishable in retinas from diabetic
   mice: (1) pericytes that were associated closely with endothelial cells
   and did not form inter-capillary bridges, (2) pericytes whose nuclei
   were aligned with capillaries and extruded long processes (7–70 µm),
   and (3) a subpopulation with nuclei that were clearly no longer aligned
   with endothelial cells and, thus, defined as migrating pericytes
   ([74]Figure 1e).

Figure 1.

   [75]Figure 1
   [76]Open in a new tab

   Pericyte tunneling nanotubes (TNTs) in retinas from non-diabetic and
   diabetic mice. Retinas were obtained from diabetic 10-month-old
   Ins2^Akita mice and their non-diabetic littermates. (a) Pericytes (NG2)
   and capillaries (CD31) in the second capillary layer of retinas from
   non-diabetic and diabetic Ins2^Akita littermates; upper panels bar = 50
   µm, lower panels bar = 20 µm. (b) Pericyte coverage of capillaries in
   the second capillary layer; n = five animals per group, with three to
   four images evaluated per retina. (c) TNTs per field of view (fov); n =
   five animals per group, with three to four images evaluated per retina
   (Student’s t test). (d) The average length and the length distribution
   of TNTs from non-diabetic and Ins2^Akita mice; n = five animals per
   group, with three to four images evaluated per retina (Student’s t
   test), ** p < 0.01, **** p < 0.0001. (e) Morphology of pericytes (NG2+)
   detected in retinas from diabetic mice: (1) non-activated, endothelial
   cell-associated pericytes, (2) endothelial cell-associated pericytes
   that make TNTs, and (3) migrating pericytes.

3.2. Tunneling Nanotubes Transport Mitochondria between Pericytes and
Endothelial Cells

   Pericytes in culture spontaneously formed TNTs/TNT-like structures that
   were rich in F-actin ([77]Figure 2a). The number and length of these
   structures increased in the presence of different concentrations of
   PDGF BB ([78]Figure 2b). Consistent with a previous report [[79]30],
   connexin 43 containing gap junctional plaques were observed at the
   contact points between pericyte TNTs and endothelial cells, as well as
   within TNTs ([80]Figure 2c).

Figure 2.

   [81]Figure 2
   [82]Open in a new tab

   PDGF-BB-dependent in vitro pericyte tunneling. (a) Tunneling nanotubes
   (TNTs) induced by PDGF-BB (0.3–3 ng/mL, 6 h); bar = 50 µm. Comparable
   results were obtained using five additional cell batches. (b) Average
   TNT length per field of view (fov) and percentage of TNT’s longer than
   100 µm; n = six independent cell batches with two evaluated images per
   batch (one-way ANOVA and Tukey’s multiple comparisons test); *** p <
   0.001, **** p < 0.0001. (c) Connexin 43 (Cx43) in pericytes (PC),
   pericyte-derived TNTs, and an endothelial cell (EC). Comparable images
   were obtained using an additional three cell batches; bar = 20 µm.

   TNTs support the intercellular trafficking of many cellular compounds
   and organelles; such as mitochondria
   [[83]15,[84]16,[85]31,[86]32,[87]33]. As pericytes are usually in close
   contact with endothelial cells, we determined whether or not
   mitochondria could be exchanged between these two cell types. To this
   end, endothelial cells containing MitoTracker Red and pericytes
   containing MitoTracker Green were seeded in separate inserts on one
   slide and allowed to form confluent monolayers ([88]Figure 3a).
   Thereafter, the inserts were removed, and the cells were allowed to
   migrate towards each other in the presence of PDGF BB. The first
   contact detected between the two cell types was in the form of
   MitoTracker Red-containing pericyte TNTs extending towards endothelial
   cells ([89]Figure 3b). The fact that mitochondria were contained in the
   growing TNTs was confirmed by the staining of the mitochondrial protein
   and the voltage-dependent anion-selective channel (VDAC) ([90]Figure
   3c). When contact was established between pericytes and endothelial
   cells, mitochondrial transfer took place, as evidenced by the
   appearance of endothelial cells containing both MitoTracker Green and
   MitoTracker Red ([91]Figure 3d). This unidirectional transport, i.e.,
   from pericytes to endothelial cells, was also observed when the
   mitochondrial dyes were reversed ([92]Figure S1). We did not detect any
   evidence of mitochondrial transfer from endothelial cells to pericytes,
   an observation that fits with reports indicating that organelle
   transport is frequently unidirectional, i.e., from the cell that
   initiated the formation of TNT to the receptor cell [[93]8,[94]34].

Figure 3.

   [95]Figure 3
   [96]Open in a new tab

   Transfer of mitochondria from pericytes to endothelial cells in vitro.
   (a) Scheme showing the migration chamber used to study the early
   contact of MitoTracker Red-labelled pericytes (PC) with MitoTracker
   Green-loaded endothelial cells (EC) for up to 6 h. (b) Labelled
   mitochondria in pericyte TNTs during first contact with endothelial
   cells in the presence of PDGF-BB (1 ng/mL) for six hours. Comparable
   results were made using five independent cell batches; bar = 20 µm. (c)
   Presence of the VDAC in pericyte TNTs. Comparable observations were
   made using four additional cell batches; bar = 20 µm. (d) Transfer of
   pericyte-derived mitochondria (red) to endothelial cells (green) in
   cells cocultured for 6 h in the presence of PDGF-BB (1 ng/mL); bar = 50
   µm in overview, 10 µm in magnification. Comparable observations were
   made using six additional cell batches.

3.3. Impact of Pericyte-Derived Mitochondria on EtBr-Treated Endothelial
Cells

   To study the consequences of mitochondrial transfer between the two
   cell types in more detail, experiments were performed using human
   endothelial cells cultured in the presence of ethidium bromide and
   uridine (EtBr) for 14 days to prevent mitochondrial DNA (mtDNA)
   replication and thus deplete cells of mitochondria [[97]23]. The
   approach used disrupted the mitochondrial network but did not result in
   the complete loss of the mitochondrial protein Translocase of outer
   mitochondrial membrane 20 (Tom20) ([98]Figure 4a) or the MitoTracker
   Red signal ([99]Figure 4b). However, the expression of Tom20 and VDAC
   was markedly reduced ([100]Figure 4c) and accompanied by an increased
   generation of mitochondrial reactive oxygen species (ROS) ([101]Figure
   4d). While the latter effect seemed counterintuitive, the EtBr-induced
   decrease in mitochondrial DNA in fibroblasts was also associated with
   increased ROS generation [[102]35], which could be attributed to
   residual mitochondria being dysfunctional.

Figure 4.

   [103]Figure 4
   [104]Open in a new tab

   Consequences of EtBr-induced endothelial cell mitophagy on
   mitochondrial network integrity and metabolism. (a) Mitochondrial
   network in endothelial cells cultured in the presence of EtBr for 14
   days, as visualized using Tom20. The images were taken 6 h after EtBr
   washout; bar = 20 µm. Comparable results were obtained in five
   additional cell batches. (b) FACs-based quantification of mitrotracker
   red (MitoRed) in CD31+ endothelial cells cultured in the absence and
   presence of EtBr; n = four independent cell batches (Student’s t test).
   (c) Expression of VDAC1 and Tom20 in endothelial cells cultured in the
   presence of solvent or EtBr for 14 days; n = seven independent cell
   batches (Student’s t test). (d) FACs-based quantification of
   mitochondrial reactive oxygen species (MitoSox) in CD31+ endothelial
   cells cultured in the absence and presence of EtBr; n = four
   independent cell batches (Student’s t test). *** p < 0.001, **** p <
   0.0001.

   Next, EtBr was removed and endothelial cells were co-cultured with
   pericytes for 6 h. This resulted in mitochondria-filled TNTs emanating
   from pericytes towards endothelial cells to enable the reformation of a
   mitochondrial network that was labelled with MitoTracker Green and
   pericyte-derived MitoTracker Red ([105]Figure 5a). Consistent with
   observations made using the marker dyes, pericyte-derived mitochondria
   restored the protein levels of Tom20 and Translocase of inner
   mitochondrial membrane 17 (Tim17) in endothelial cells isolated from
   the co-culture using CD144-coated magnetic beads ([106]Figure 5b). The
   expression of VDAC in endothelial cells was partially recovered during
   the 6 h of EtBr washout, but the co-culture with pericytes markedly
   increased its levels, clearly indicating the impact of mitochondrial
   transfer from pericytes.

Figure 5.

   [107]Figure 5
   [108]Open in a new tab

   Transfer of mitochondria between pericytes and endothelial cells. (a)
   Mitochondrial transfer between pericytes (red) and EtBr-treated
   endothelial cells (green). The phase contrast images show
   mitochondria-filled TNTs emanating from pericytes towards endothelial
   cells. Comparable results were obtained in three additional experiments
   using independent cell batches; bar = 20 µm. (b) Mitochondrial proteins
   in endothelial cells treated with the solvent or EtBr or treated with
   EtBr prior to washout and culture with pericytes for 6 h (+PC); n =
   four independent cell batches (one-way ANOVA and Tukey’s multiple
   comparisons test). (c) TNT formation by PDGF-treated pericytes cultured
   in the presence of the solvent (Sol, 0.1% DMSO), 19,20-EDP, or
   19,20-DHDP (10 µmol/L, 6 h). Experiments were performed in the presence
   of an sEH inhibitor; n = four independent cell batches with two images
   evaluated per cell batch (one-way ANOVA and Tukey’s multiple
   comparisons test). (d) Mitochondrial transfer from pericytes (red) to
   endothelial cells (green) in the absence and presence of 19,20-DHDP (10
   µmol/L). (e,f) Transfer of mitochondria from Mitotracker Red
   (MitoRed)-labelled pericytes to CD31+ endothelial cells (e), and
   mitochondrial ROS production detected using Mitosox in CD31+ cells (f);
   n = four to eight independent cell batches (one-way ANOVA and Tukey’s
   multiple comparisons test). * p < 0.05, ** p < 0.01, *** p < 0.001,
   **** p < 0.0001.

   We previously reported that an increase in the expression of the sEH in
   the retina and the increased generation of the DHA diol,
   19,20-dihydroxydocosapentaenoic acid (19,20-DHDP), were responsible for
   the dissolution of pericyte endothelial cell junctions and the
   development of non-proliferative diabetic retinopathy [[109]19].
   Therefore, we determined the effect of 19,20-DHDP on the formation of
   pericyte TNTs in vitro. While TNTs rapidly developed in PDGF-treated
   pericytes, the TNTs that formed in the presence of 19,20-DHDP were
   clearly shorter and formed fewer contacts with other pericytes
   ([110]Figure 5c). Treating pericytes with the sEH substrate 19,20-EDP
   (in the presence of an sEH inhibitor to prevent any metabolism to
   19,20-DHDP) had a slight effect on the TNT length but no effect on the
   number of TNTs over 100 µm. Moreover, while pericytes rescued
   mitochondria numbers and free radical production in CD31+ endothelial
   cells, the effects were blunted when pericytes were treated with
   19,20-DHDP ([111]Figure 5d–f).

3.4. 19,20-DHDP Antagonizes the Rescue of Endothelial Cell Metabolism by
Pericytes

   There was a clear impact of endothelial cell mitochondrial depletion on
   the initial part of the TCA cycle, i.e., markedly reduced intracellular
   levels of citrate and isocitrate ([112]Figure 6). Cells depleted of
   mtDNA with EtBr are known to maintain some mitochondrial function
   [[113]36], and we observed that levels of α-ketoglutarate (α-KG),
   succinate, and fumarate were elevated in EtBr-treated cells, indicating
   that residual endothelial cell mitochondria were able to partly refuel
   the TCA cycle using glutamine as an energy source. Indeed, the
   glutamine levels were unexpectedly high, which likely reflects an
   increase in its uptake. α-KG was also metabolized to 2-hydroxyglutarate
   (2HG), the levels of which were also markedly elevated in EtBr-treated
   endothelial cells. To determine the impact of pericyte-derived
   mitochondria on endothelial cell metabolism, EtBr-treated cells were
   washed and incubated with PDGF-treated pericytes overnight. Thereafter,
   endothelial cells were separated by CD144-coated magnetic beads and
   subjected to metabolomic analyses via LC-MS/MS. The co-culture with
   pericytes restored the TCA cycle metabolite levels to those detected in
   mitochondria-intact (solvent-treated) endothelial cells. Pericytes also
   decreased the endothelial cell levels of glutamine but had a minor
   impact on α-KG and 2HG, which can be accounted for by the
   re-establishment of the TCA cycle. Consistent with our observations on
   TNT formation, the pericyte-associated rescue of the endothelial cell
   TCA cycle was largely prevented in cells exposed to 19,20-DHDP.

Figure 6.

   [114]Figure 6
   [115]Open in a new tab

   Rescue of endothelial cell metabolism by pericyte-derived mitochondria.
   Mitochondrial intact (Sol) or mitochondria-depleted (EtBr) endothelial
   cells were cocultured with pericytes (+PC) in the absence and presence
   of 19,20-DHDP for 36 h. Intermediates of the TCA cycle, asparagine, and
   glutamine metabolism were measured by LC-MS/MS; n = six independent
   cell batches (one-way ANOVA and Tukey’s multiple comparisons test). * p
   < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. ACO: aconitase,
   ASNS: asparagine synthetase, CS: citrate synthase, FH: fumarate
   hydratase, GLS: glutaminase, GLUD: glutamate dehydrogenase, IDH:
   isocitrate dehydrogenase, LDH: lactate dehydrogenase, PDH: pyruvate
   dehydrogenase, SDH: succinate dehydrogenase, SUCLA: succinate-CoA
   ligase.

3.5. Effect of 19,20-DHDP on Pericyte-Endothelial Cell Junction Proteins

   To gain insight into the impact of 19,20-DHDP on intracellular
   communication, endothelial cells and pericytes were placed on opposite
   sides of a filter insert and cultured to confluence. Thereafter, the
   filters were treated with the solvent or 19,20-DHDP before the
   endothelial cells and pericytes were carefully removed and the proteins
   present in the filter pores were analyzed by mass spectrometry. The
   endothelial cell–pericyte junctions contained a large number of
   proteins ([116]Figure 7a,b). Overall, 1003 proteins were detected in
   junctions in the presence of the solvent, and a discrete population of
   those were depleted from junctions treated with 19,20-DHDP and included
   transport proteins such as the cationic amino acid transporter, SLC7A5,
   and the alanine-serine-cysteine transporter, SLC1A5. Pathway analysis
   of the junctional proteins altered by 19,20-DHDP identified
   phosphatidylinositol 3-kinase, PDGF receptor β, and RhoA-related
   signaling events as being altered by the DHA diol ([117]Figure 7c). As
   PDGF is essential for endothelial cell–pericyte communication and
   coverage [[118]37,[119]38] and stimulated TNT formation, we determined
   whether or not 19,20-DHDP was able to interfere directly with PDGF
   signaling in pericytes. This seemed a logical hypothesis given that
   this lipid mediator can alter the cholesterol content and rearrange
   protein complexes in lipid rafts that may well affect receptor
   signaling [[120]39]. However, 19,20-DHDP did not directly affect the
   ability of PDGF to elicit the phosphorylation of its receptor (PDGFR)
   or the subsequent phosphorylation of either AKT or ERK1/2 ([121]Figure
   7d). Thus, although the fatty acid diol prevented PDGF-induced TNT
   formation and altered the expression of proteins involved in PDGF
   signaling, it had no direct impact on the early activation of the PDGF
   receptor.

Figure 7.

   [122]Figure 7
   [123]Open in a new tab

   Effect of 19,20-DHDP on TNT formation and proteins in endothelial
   cell–pericyte junctions. (a) Heatmap showing the 50
   pericyte–endothelial cell junctions most affected by 19,20-DHDP (10 µM,
   16 h). Mitochondrial proteins are highlighted in red, PDGFR-β signaling
   in blue, and trafficking-related proteins in green; n = three
   independent cell batches. (b) Volcano plot showing endothelial
   cell–pericyte junctional proteins in co-cultures of endothelial cells
   and pericytes, comparing 19,20-DHDP with non-treated fractions
   (19,20-DHDP/Solvent). (c) Pathway enrichment analysis (NCI-Nature) of
   proteins significantly decreased in the junctional space after exposure
   to 19,20-DHDP. (d) Representative Western blot showing the effect of
   PDGFBB (1 ng/mL, for up to 240 min) on the phosphorylation of the PDGF
   receptor, Akt, and Erk1/2 in pericytes in the absence and presence of
   19,20-DHDP; n = four independent cell batches.

4. Discussion

   The results of the present investigation highlight that a metabolic
   communication exists between pericytes and endothelial cells, which
   involves the transfer of functional mitochondria through TNTs.
   Moreover, a DHA diol, i.e., 19,20-DHDP, generated by sEH, which
   promotes the dissolution of endothelial cell–pericyte junctions to
   initiate pericyte migration in vivo [[124]19], stunted the formation of
   TNT-like structures and prevented the metabolic rescue of
   mitochondrial-depleted endothelial cells by pericytes.

   Pericytes are key regulators of neurovascular coupling and have been
   proposed to contract to regulate blood flow (for reviews, see
   references [[125]40,[126]41]). It has also been suggested that