Niche-specific macrophage loss promotes skin capillary aging

Summary All mammalian organs depend upon resident macrophage populations to coordinate repair processes and facilitate tissue-specific functions1–3. Recent work has established that functionally distinct macrophage populations reside in discrete tissue niches and are replenished through some combination of local proliferation and monocyte recruitment4,5. Moreover, decline in macrophage abundance and function in tissues has been shown to contribute to many age-associated pathologies, such as atherosclerosis, cancer, and neurodegeneration6–8. Despite these advances, the cellular mechanisms that coordinate macrophage organization and replenishment within an aging tissue niche remain largely unknown. Here we show that capillary-associated macrophages (CAMs) are selectively lost over time, which contributes to impaired vascular repair and tissue perfusion in older mice. To investigate resident macrophage behavior in vivo, we have employed intravital two-photon microscopy to non-invasively image in live mice the skin capillary plexus, a spatially well-defined model of niche aging that undergoes rarefication and functional decline with age. We find that CAMs are lost with age at a rate that outpaces that of capillary loss, leading to the progressive accumulation of capillary niches without an associated macrophage in both mice and humans. Phagocytic activity of CAMs was locally required to repair obstructed capillary blood flow, leaving macrophage-less niches selectively vulnerable to both homeostatic and injury-induced loss in blood flow. Our work demonstrates that homeostatic renewal of resident macrophages is not as finely tuned as has been previously suggested9–11. Specifically, we found that neighboring macrophages do not proliferate or reorganize sufficiently to maintain an optimal population across the skin capillary niche in the absence of additional cues from acute tissue damage or increased abundance of growth factors, such as colony stimulating factor 1 (CSF1). Such limitations in homeostatic renewal and organization of various niche-resident cell types are potentially early contributors to tissue aging, which may provide novel opportunities for future therapeutic interventions.


Summary
All mammalian organs depend upon resident macrophage populations to coordinate repair processes and facilitate tissue-specific functions [1][2][3] .Recent work has established that functionally distinct macrophage populations reside in discrete tissue niches and are replenished through some combination of local proliferation and monocyte recruitment 4,5 .Moreover, decline in macrophage abundance and function in tissues has been shown to contribute to many age-associated pathologies, such as atherosclerosis, cancer, and neurodegeneration [6][7][8] .Despite these advances, the cellular mechanisms that coordinate macrophage organization and replenishment within an aging tissue niche remain largely unknown.Here we show that capillaryassociated macrophages (CAMs) are selectively lost over time, which contributes to impaired vascular repair and tissue perfusion in older mice.To investigate resident macrophage behavior in vivo, we have employed intravital two-photon microscopy to non-invasively image in live mice the skin capillary plexus, a spatially well-defined model of niche aging that undergoes rarefication and functional decline with age.We find that CAMs are lost with age at a rate that outpaces that of capillary loss, leading to the progressive accumulation of capillary niches without an associated macrophage in both mice and humans.Phagocytic activity of CAMs was locally required to repair obstructed capillary blood flow, leaving macrophage-less niches selectively vulnerable to both homeostatic and injury-induced loss in blood flow.Our work demonstrates that homeostatic renewal of resident macrophages is not as finely tuned as has been previously suggested [9][10][11] .Specifically, we found that neighboring macrophages do not proliferate or reorganize sufficiently to maintain an optimal population across the skin capillary niche in the absence of additional cues from acute tissue damage or increased abundance of growth factors, such as colony stimulating factor 1 (CSF1).Such limitations in homeostatic renewal and organization of various niche-resident cell types are potentially early contributors to tissue aging, which may provide novel opportunities for future therapeutic interventions.

Main Text
Tissue homeostasis is dependent on multiple macrophage populations that reside in distinct sub-tissue compartments or niches, such as epithelium, blood vessels or nerves, and are thought to support specialized tissue functions 4,12,13 .Recent work has suggested that functional decline and rarefication of vascular niches may contribute to various age-associated tissue pathologies (including sarcopenia, chonic wounds, and Alzheimer's disease) [14][15][16] .It is not yet known how tissue-resident macrophages resist or potentiate such niche specific aging processes 6,17 .

Capillary-associated macrophage decline with age correlates with impaired blood flow.
To model mammalian tissue aging, we adapted an intravital microscopy technique to visualize skin resident macrophage populations non-invasively in live mice throughout the lifetime of the organism 18,19 (Figure 1a, Supplemental Videos 1 and 2).
Unexpectedly, longitudinal imaging of skin macrophages (marked by Csf1r-EGFP) revealed niche-specific decline in macrophage populations.Subdividing the skin into three anatomical layers, epidermis, upper (papillary) and lower (reticulated) dermis, we observed that macrophages of the upper dermis were lost with age at a greater rate than macrophages from both epidermis and lower dermis (Figure 1b-c).Further characterization of this population with additional myeloid fluorescent reporters revealed that most macrophages of the upper dermis express the chemokine receptor CX3CR1, but not CCR2 (Extended Data Fig. 1a,b).A major component of the upper dermal niche is the superficial capillary plexus, which supplies nutrient exchange for the overlying epidermis.To visualize this structure, we utilized third harmonic generation from our imaging to track red blood cell (RBC) flow 20,21 , which was consistent with conventional labeling methods such as rhodamine dextran injection (Extended Data Fig. 1c-e and Supplemental Video 3).With this in vivo marker of blood flow, we found that the macrophages were closely associated with blood capillaries of the superficial plexus, suggesting that they may provide support for this capillary niche (Extended Data Fig. 1fh and Supplemental Video 4).To investigate if these macrophages play a role in capillary function, we assessed if blood flow, via RBC flow (Extended Data Fig. 2), was altered in the presence of capillary-associated macrophages (CAMs).Performing timelapse recordings of fluorescently labeled CAMs in mice with a cre-dependent dual reporter system (Cx3cr1-CreERT2; R26-mTmG), we found that capillaries lacking an associated macrophage had a higher rate of obstructed blood flow (Figure 1d,e and Supplemental Video 5).Longitudinal imaging across multiple ages showed that the fraction of capillaries with CAMs significantly decreased with age (Figure 1f).This decrease in CAMs and coverage outpaced the loss of capillaries (Extended Data Fig. 3a,b), which was previously shown to be an early hallmark of aging in multiple tissues, including central nervous system, lung, kidney, and skin 14,[22][23][24][25][26] .To assess if this phenomenon also occurs in humans, we obtained both young and old human patient skin samples.Consistent with our observations in mice, human capillary-associated macrophages also displayed a decline with age.Moreover, CAM decline also outpaced capillary loss with age, suggesting a similar loss in macrophage coverage of the capillary niche (Extended Data Fig. 3c-f).To assess any functional role of CAMs in maintaining capillary blood flow, we performed chemical and genetic ablation of CAMs, via clodronate liposomes and Cx3cr1-DTR depletion, respectively, and observed an acute loss in capillary flow (Figure 1 g,h and Extended Data Fig. 4).Collectively, this work highlights an evolutionarily conserved loss in skin capillary-associated macrophages with age, which correlates with impaired homeostatic capillary perfusion.

Local CAM recruitment and phagocytosis is required to restore blood flow.
We next aimed to assess the cellular mechanism(s) by which CAMs support capillary blood flow, using a laser-induced blood clotting model to precisely target and stop blood flow in individual capillary segments (Figure 2a, Extended Data Fig. 5a,b and Supplemental Video 6).To assess any macrophage involvement, we tracked the daily displacement of surrounding CAMs to laser-induced clots.These data showed that CAM recruitment to sites of capillary damage as well as RBC engulfment are locally restricted to approximately 80-100µm, and largely occur within the first two days after injury (Figure 2b,c).Under homeostatic conditions, neighboring CAMs were similarly involved locally, taking up RBC debris from naturally occurring capillary clots (Figure 2d).
Given these findings, we next tested if capillary blood flow was preferentially repaired when at least one resident CAM was in close proximity (<75µm) with the injured capillary segment.The results showed a significant improvement in capillary repair for vessels with an associated macrophage (Extended Data Fig. 5c-e), suggesting that the loss of CAMs could lead to compromised capillary blood flow with age.To rule out an indirect association of capillary repair and CAM proximity, we performed laser-induced ablation of CAMs adjacent to capillary segments immediately prior to inducing capillary clots.Indeed, in regions with CAM ablation within 75µm of the capillary clot, repair was significantly impaired and blood flow was not properly reestablished (Figure 2e,f).
Our serial imaging revealed that CAMs at capillary repair sites often contained RBC debris shortly after clot formation.To understand if CAM uptake and clearance of this cellular debris is functionally important for capillary repair, we acutely impaired CAM phagocytosis through inducible cre-dependent knockout of Rac1, a critical component of the phagocytic machinery 27 , one week prior to laser-induced clot formation.While there was no significant change in CAM density (Extended Data Fig. 6a,b), we found significant impairment in capillary repair in Cx3cr1 CreER ;Rac1 fl/fl mice compared to Cx3cr1 CreER ;Rac1 fl/+ littermate controls (Figure 2g,h), suggesting that Rac1-dependent phagocytic clearance is critical for proper capillary repair and tissue reperfusion.Together, our results support a model in which local CAM recruitment and phagocytic clearance of capillary debris is critical to maintain capillary function.Therefore, as CAM density declines with age, so does capillary perfusion of the tissue (Figure 2i).

CAMs do not replenish lost neighbors, resulting in population loss with age
Maintenance of tissue resident macrophage populations in the skin is thought to be mediated through a combination of local proliferation and systemic replacement by blood monocytes 13,28,29 .To understand how CAM density declines with age, we performed single-macrophage lineage tracing, by inducing sparse cre-recombination to label and tracking of individual macrophages over weekly revisits (Extended Data Fig. 7a-c).In doing so, we observed discrete behaviors, including macrophage loss and proliferation (Figure 3a).Analysis of weekly rate of CAM loss and division revealed a significant skew toward more CAM loss than division in the first 4 months of age (Figure 3b).Consistent with these data, there was a significant decline in the fraction of fatemapped CAMs over a 20-week time course (Figure 3c).These results demonstrate that the CAMs are insufficiently replenished by local proliferation, which contributes to their progressive decline in this tissue niche.
The ability of resident macrophages to locally self-renew has largely been studied through methods of near-total macrophage depletion, which have limited nichespecificity and often generate tissue-wide inflammation [9][10][11] .To directly interrogate the steps of macrophage self-renewal over time, we tracked both the replacement of individual macrophages after loss and the distribution of sister macrophages after division.First, to track local macrophage replacement after CAM loss, we performed laser-induced ablation of all CAMs within a defined 500µm 2 region.Serial revisits up to two weeks after ablation revealed minimal repopulation from adjacent capillary regions that retained intact CAM populations (Extended Data Fig. 8a and Figure 3d).We found a similar lack in repopulation following partial CAM depletion in Cx3cr1 DTR mice following low dose diphtheria toxin administration (Figure 3e).To avoid any nonphysiological effects from these cell depletion models, we developed a dual fluorescent macrophage reporter mouse, Cx3cr1-CreERT2; Rosa26-dsRed; Csf1r-EGFP, allowing for cre-dependent recombination to differentially label a small fraction of macrophages and track their homeostatic replacement.Examination on serial weekly revisits indicated that, as we observed in the depletion models, most capillary niches did not recruit a new macrophage for as long as two weeks following CAM loss (Figure 3f).In contrast, emptied capillary niches were readily replenished with new CAMs when laser-induced macrophage loss was accompanied by laser-induced capillary damage (Figure 3g).
Such replenishment was partially CCR2-dependent, suggestive that monocytes may participate in repopulating the capillary niche after injury (Extended Data Fig. 8b-d and Figure 3g).Taken together, these results suggest that CAM loss is not a sufficient trigger to promote neighboring macrophages into the emptied niche.
Second, to precisely track sister macrophage migration following cell division, we studied mice with a dual fluorescent nuclear reporter, Cx3cr1-CreERT2; R26-nTnG, where we can use low or high doses of tamoxifen to either label individual or all CAMs, respectively (Extended Data Fig. 9a).Compared to the average distance between all neighboring macrophages, sister CAMs remained significantly closer to each other even 2 weeks after division (Extended Data Fig. 9b,c).These findings further support the notion that CAM division and loss are not spatiotemporally coupled, which we predict would progressively lead to disorganized patterning and the accumulation of both empty and crowded capillary niches.We tested this prediction by looking at the distribution of neighboring CAMs in both young (2-month old) and old (10-month old) mice.In young mice, the majority of macrophages were within 50µm of each other.In contrast, old mice had a biphasic distribution of macrophage patterning, with most CAMs either within 25µm or further than 75µm apart (Extended Data Fig. 9d).These results highlight two distinct cellular features that contribute to reduced CAM coverage with age: 1) insufficient macrophage repopulation following CAM loss, and 2) insufficient distribution of these cells along the capillary niche, which may promote progressive erosion of the vascular bed (Extended Data Fig. 9e).

Local CAM replenishment in old mice is sufficient to rejuvenate capillary repair and tissue reperfusion
While CAM proliferation was insufficient for maintenance of the population during homeostasis, we asked if extrinsic cues such as broader tissue damage could enhance division rates.We employed large laser-induced wounds in both the epidermal and upper dermal niches (Extended Data Fig. 10a) and found that local CAM proliferation was significantly increased one week following tissue damage in both layers (Extended Data Fig. 10b-e).Given these results we hypothesized that damage-induced CAM proliferation in the aged capillary niche would provide lasting increases in macrophage density.Indeed, broad epidermal damage resulted in a lasting increase in CAMs in damaged regions compared to neighboring control regions (Extended Data Fig. 10f,h).
Our results suggest that the capillary-associated macrophage population in old mice can be stably expanded following environmental changes, such as injury.
Therefore, we next assessed if increased CAM density, without previous injury, would be sufficient to improve future capillary repair and reperfusion.To this end, we utilized a fusion protein of the canonical macrophage growth factor, CSF1, with the Fc region of porcine IgG, as it has been reported to robustly increase macrophage density in multiple tissues, including skin [30][31][32] .We performed daily intradermal injections of either CSF1-Fc or PBS in the left or right hind paws, respectively, of the same mice (Figure 4a).There was a significant increase in CAMs in the CSF1-treated paws compared to contralateral PBS controls, which had no significant change from pretreatment (Figure 4b,c).
Strikingly, we found that CSF1 treatment was sufficient to improve homeostatic capillary blood flow in old mice, as compared to PBS controls which had significantly more obstructed capillary segments (Figure 4d,e).Utilizing the same aged mice, we next tested if this increase in CAM density would be sufficient to improve capillary repair rates.Following laser-induced clotting, there was a significant improvement in capillary repair and reperfusion in CSF1-treated mice compared to PBS controls (Figure 4 f,g).
Consistent with this result, there was also significant improvement in capillary repair and tissue reperfusion in old mice following damage-induced CAM expansion (Extended Data Fig. 10i).

Discussion
Macrophage renewal has largely been studied in non-physiological settings, such as through in vitro cell culture or severe depletion models that often are accompanied by acute inflammation 9,10,[33][34][35] .Our work clearly demonstrates that homeostatic renewal of resident macrophages is not as finely tuned as has been previously suggested.Specifically, we found that neighboring macrophages do not proliferate or reorganize sufficiently to maintain an optimal population across the skin capillary niche unless they receive additional cues from acute tissue damage or increased growth factor abundance.Interestingly, we confirmed previous findings that show epidermal Langerhans cell density also declines with age, which has been associated with impaired epidermal function [36][37][38][39] .This raises the possibility that age-associated loss in macrophage density is a more general phenomenon in populations that rely on local self-renewal.
In addition to self-renewal, we also found that CAM recruitment to repair tissue damage was spatially restricted.To our knowledge, the long-term size and stability of resident macrophage territories or niches in vivo has not been reported.Our work provides strong evidence that injury-induced macrophage recruitment is restricted to approximately 100µm.In young mice, CAM density is high enough to provide substantial niche/territory overlap between neighbors.However, with declining CAM density with age, we show that a significant fraction of the skin capillary network is no longer within a CAM's territory range.It will be important to understand how this property of CAMs is influenced by other aspects of regional heterogeneity, such as innate immune imprinting 13 , to shape local immune responses in tissues.
Collectively, this work demonstrates that loss in CAMs: 1) begins within the first few months of life, 2) is progressive throughout life, and 3) is functionally detrimental to vascular function, which has been shown to be a primary driver of age-associated tissue impairments 15,40 .Furthermore, this work provides a novel platform to investigate ageassociated deviations in tissue homeostasis at the single-cell level in a living mammal.

Mice
Mice were bred and maintained in the Alexandria Center for the Life Sciences animal facility of the New York University School of Medicine, in specific pathogen-free conditions.Albino B6 (B6(Cg)-Tyr c-2J /J, Jax 000058), Csf1r EGFP (B6.Cg-Tg(Csf1r-EGFP)1Hume/J, Jax 018549), Ccr2 RFP (B6.129(Cg)-Ccr2tm2.1Ifc/J,Jax 017586), R26 mTmG (B6.129(Cg)-Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J,Jax 007676), R26 nTnG (B6N.129S6-Gt(ROSA)26Sortm1(CAG-tdTomato*,-EGFP*)Ees/J,Jax 023537), LysM Cre (B6.129P2-Lyz2tm1(cre)Ifo/J,Jax 004781), Rac1 f/f (Rac1tm1Djk/J, Jax 005550) mice were purchased from Jackson Laboratories.R26 dsRed mice were described previously (Luche et al., 2007) and were obtained from the laboratory of Dr. Gordon Fishell.Cx3cr1 CreER , Cx3cr1 GFP , and Cx3cr1 DTR were generated in our laboratory and have been described (3 ref).Cre-induction for the lineage tracing or total CAM labeling experiments was induced with a single intraperitoneal injection of Tamoxifen (Sigma-Aldrich; T5648) (100µg or 4mg in corn oil, respectively) in 1 month old mice.Rac1 f/f recombination was induced with two intraperitoneal injections of Tamoxifen (2mg in corn oil) 48hrs apart in 1 month old mice.All imaging and experimental manipulation are performed on non-hairy mouse plantar (hind paw) skin.Preparation of skin for intravital imaging are performed as described below.Briefly, mice are anesthetized with intraperitoneal injection of ketamine/xylazine (15 mg/ml and 1 mg/ml, respectively in PBS).After imaging, mice are returned to their housing facility.For subsequent revisits, the same mice are processed again with injectable anesthesia.The plantar epidermal regions are briefly cleaned with PBS pH 7.2, mounted on a custom-made stage and a glass coverslip are placed directly against the skin.Anesthesia is maintained throughout the course of the experiment with vaporized isoflurane delivered by a nose cone.Mice from experimental and control groups were randomly selected for live imaging experiments.No blinding was done.All lineage tracing and ablation experiments were repeated in at least three different mice.All animal procedures were performed in accordance with protocols approved by the Institutional Animal Care and Usage Committee of New York University School of Medicine.

In vivo imaging and laser ablation
Image stacks were acquired with an Olympus multiphoton FVMPE-RS system equipped with both InSight X3 and Mai Tai Deepsee (Spectra-Physics) tunable Ti:Sapphire lasers, using Fluoview software.For collection of serial optical sections, a laser beam (940nm for GFP/tdTomato/dsRed/RFP/Rhodamine/Second Harmonic Generation and 1300nm for Third Harmonic Generation, respectively) was focused through a water immersion lens (N.A. 1.05; Olympus) and scanned with a field of view of 0.5mm 2 , at 600 Hz.Z-stacks were acquired in 1-2μm steps for a ~50-100μm range, covering the epidermis and dermis.Capillary blood flow was visualized in some experiments through intravenous injection with 18 mg/kg of dextran-rhodamine 70 kD (Sigma-Aldrich; R9379).Cell tracking analysis was performed by re-visiting the same area of the dermis in separate imaging experiments through using inherent landmarks of the skin to navigate back to the original region, including the distinct organization of the superficial vasculature networks.Cells that were unambiguously separated (by at least 250µm) from another were sampled to ensure the identity of individual lineages.For time-lapse recordings, serial optical sections were obtained between 5-10 minute intervals, depending on the experimental setup.Laser-induced cell ablation, capillary clot, or tissue damage was carried out with the same optics as used for acquisition.An 940nm laser beam was used to scan the target area (1-500μm 2 ) and ablation was achieved using 50-70% laser power for ~1sec.Ablation parameters were adjusted according to the depth of the target (10-50µm).Mice from experimental and control groups were randomly selected for live imaging experiments.No blinding was done.All lineage tracing and ablation experiments were repeated in at least three different mice.

Drug treatments
To induce macrophage depletion, mice received intradermal injections of either Clodronate-liposomes or PBS-liposomes (stock concentration 5mg/ml; Liposoma; CP-005-005) (5µl per paw) every 3 days.Depending on experimental details, Cx3cr1 DTR mice received either intraperitoneal (IP) injection of diphtheria toxin (Sigma-Aldrich; D0564) every other day at 25ng/g body weight in PBS or a single low dose IP injection at 10ng/g body weight in PBS.To induce macrophage expansion, mice receive daily intradermal injections of CSF1-FC (Bio-Rad; PPP031) or PBS in contralateral hind paws (5µl per paw) for 4 days.

Human skin samples
Written informed consent was obtained for postmortem examination from next of kin for 10 patients.Clinical information and laboratory data were obtained from the electronic medical record.Skin samples were obtained from the anterolateral chest and fixed in 10% formalin for at least 24h prior to processing.Slides were stained with hematoxylin and eosin, CD68 (Clone 514H12) and ERG (Clone EPR3864).Macrophages and capillaries were identified using a combination of morphology, CD68 and ERG staining.
Counting was performed on at least eight high-power fields (40x) within 100um of the epidermis.

Image Analysis
Raw image stacks were imported into Fiji (NIH, USA) or Imaris software (Bitplane/Perkin Elmer) for further analysis.Provided images and supplementary videos are typically presented as a maximal projection of 4-8µm optical sections.For visualizing individual labeled cells expressing the dsRed or tdTomato Cre reporters, the brightness and contrast were adjusted accordingly for the green (GFP) and red (dsRed/tdTomato) channels and composite serial image sequences were assembled as previously described.Images were obtained as large tiled image stacks at roughly the same positions and then manually aligned over the experimental time course in Imaris (Bitplane/Perkin Elmer) by using data from all channels.

Statistical Analysis
Data are expressed as mean ± SD.An unpaired Student's t-test was used to analyze data sets with two groups.One-way ANOVA was used to analyze data sets with three or more groups.p < 0.05 to p < 0.0001 indicated a significant difference.Statistical calculations were performed using the Prism software package (GraphPad, USA).

Figure 2 -
Figure 2 -Local CAM recruitment and phagocytosis are required to restore

Figure 4 -
Figure 4 -Local CAM replenishment in old mice is sufficient to rejuvenate

Legends Extended Data Figure 1 .
Scheme of CAM replacement after capillary niche damage in Csf1r-EGFP; Ccr2 RFP/+ or