Introduction

In adaptive immune responses, antigen presentation from peripheral tissues via dendritic cells triggers specific lymphocyte activation with the generation of target-specific T cell responses and ultimately the generation of immune memory(1). Importantly, these events must take place within the context of an inflammatory, or damage-related, challenge at the site of antigen encounter(2). There is also ample evidence of other peripheral tissue-derived molecules providing additional specificity to the immune response especially in terms of assigning selective tissue-homing properties to the antigen-specific T cells(3, 4). This is therefore a complex response which requires the simultaneous, and coordinated, interaction of multiple cell types and molecules. In mammals this is achieved by integrating all these processes within lymph nodes that lie downstream of all tissues and which are connected to the tissues via the lymphatic vasculature(5).

In addition to requiring precise coordination of cells and molecules, the immune response is highly energy-intensive and requires rapid access to local molecular energy sources including free fatty acids(6). Indeed, research indicates that there is a trade-off between the high metabolic costs of immune activation and the maintenance of body temperature(7). In this regard, it is notable that all mammalian lymph nodes are embedded in perinodal adipose tissue(8) which contains, not just adipocytes, but also a range of immune cells with macrophages representing the largest resident population. Considerable research indicates a key role for the perinodal adipose tissue as a source of nutrients, released through lipolysis, to feed the immune response(9). In addition, there is clear evidence of an effect of immune stimulation on perinodal adipose tissue content and size(10). What is not understood is how this process linking adipocytes in perinodal adipose tissue to immune response is regulated.

Here we demonstrate that induction of cutaneous inflammation is sufficient, in the absence of antigen presentation, to remodel lymph nodes and perinodal fat deposits. Specifically cutaneous inflammation induced either by wounding, or administration of the TLR9 agonist imiquimod, is associated with rapid atrophy of the fat pad embedding the inguinal lymph node and a marked increase in size of the lymph node. Both effects are strongly correlated with the degree of peripheral inflammation and are dependent on the CCR2-independent accumulation of a specific macrophage population in the fat pad. We therefore identify inflammation as an antigen-independent trigger for perinodal fat pad, and lymph node, remodelling in the immune response. These data therefore shed new light on the communication between peripheral tissues and lymph nodes in the initiation of an immune response and suggest that the mechanisms of action of pro-inflammatory vaccine adjuvants may not be restricted to the peripheral injection site.

Results

CCR2-/- male mice display reduced inguinal fat pad, and increased inguinal lymph node size

Gross examination of resting male CCR2-/- mice revealed a striking phenotype in the inguinal fat pad, characterised by a marked reduction in the size of the fat pad and a concomitant increase in the size of the embedded inguinal lymph node (Figure 1a). Quantitative analyses confirmed these observations with a significant reduction in inguinal fat pad size and weight (Figure 1bi and ii) and an increase in inguinal lymph node volume and cellularity (Figure 1ci and ii). Overall, a strong negative correlation between lymph node weight and fat pad weight was observed in CCR2-/- males (Figure 1d). Importantly this phenotype was only observed in male mice (Figure 1a) and progressed with age (Figures 1bii and 1cii). Total body weight was unaffected in male (and female) mice (Figure 1e), suggesting that the phenotype observed in male mice is not systemic. Thus, CCR2-/- male mice display reduced inguinal fat pad weight and increased inguinal lymph node size which are correlatively linked.

Figure 1

The observed phenotype is associated with skin wounding

The fact that the above differences were observed in male, but not female, mice suggested a possible contribution of fighting to the phenotype. Therefore, we examined the contribution of co-housing to the phenotype in CCR2-/- male mice. To this end, CCR2-/- males (8 weeks old) were either segregated, or co-housed in cages of 5, and culled 3 weeks later. As expected, co-housed CCR2-/- male mice showed evidence of more severe, fighting-related, skin wounding than their segregated counterparts. Of note, some co-housed CCR2-/- male mice had to be culled due to excessive wounding (Figure 2a). This increased wounding was clearly associated with smaller inguinal fat pads and larger inguinal lymph nodes in comparison to segregated counterparts (Figures 2b-d). As noted above, we observed a strong negative correlation between lymph node weight and fat pad size in CCR2-/- males (Figure 1d). When separating co-housed and segregated CCR2-/- males, we obtained 2 different functions: segregated CCR2-/- males showed no significant correlation between lymph node weight and fat pad size (Figure 2ei), which was similar to what was observed in WT males (Figure 2eii) and CCR2-/- females (Figure 2eiii). In contrast, co-housed CCR2-/- males showed a very strong correlation between lymph node and fat pad weight (Figure 2eiv), indicating that this correlation only holds for co-housed CCR2-/- mice (Figure 2f).

Figure 2.

Whilst the above phenotype was not immediately apparent in WT males, subsequent analysis comparing isolated, and co-housed, WT males revealed a small, but significant, increase in lymph node size, and a reduction in fat pad size, in co-housed WT male mice (Figure 2g). These differences were the consequence of a few males in the co-housed population with markedly enlarged lymph nodes and atrophied fat pads (red dots in Figures 2g and h). Interestingly, the mice with the most extensive alterations in lymph node and fat pad weight, also showed evidence of fighting-associated skin wounds (Figure 2hii) thus further suggesting a link between the observed phenotypes and cutaneous inflammation. Thus, these data link fighting-associated skin inflammation to inguinal fat pad and lymph node alterations.

CCR2-/- mice display extensive wounding-related skin inflammation

Next, we examined why this phenotype was particularly prominent in co-housed CCR2-/- male mice. Counterintuitively, co-housed CCR2-/- males had significantly more CD45+ cells in the skin (Figure 3a) than segregated, or co-housed, WT mice with a significant increase in neutrophils, eosinophils and T-cells (Figures 3b and 3c). Despite the anticipated reduction in infiltrating monocytes in the co-housed CCR2-/- mouse skins(11, 12) (Figure 3di), macrophage numbers were similar in WT and co-housed CCR2-/- mice (Figure 3dii). Further staining revealed the presence of both CD206+ and CD206-macrophages in WT skin, but only CD206+ macrophages in co-housed CCR2-/- mouse skin (Figure 3diii). Importantly, the number of CD45+ cells in the skin displayed a strong negative correlation with fat pad weight (Figure 3e), indicating that the extent of cutaneous inflammation is linked to inguinal lymph node and fat pad size. Thus CCR2-/- mice display exaggerated skin inflammation, the extent of which correlates with the reduction in inguinal fat pad weight.

Figure 3:

Inguinal fat pads and lymph nodes in CCR2-/- mice display altered cellularity

Next, we examined the leukocyte content of the remodelled fat pads and lymph nodes. Whilst, as expected, fat pads from co-housed male CCR2-/- mice were severely depleted in inflammatory monocytes (Figure 4Aii), they displayed a general increase in macrophage numbers (Figure 4Ai and ii) compared to WT mice. Further flow cytometric analysis revealed an increased number of pro-inflammatory CD11c+ CD206+ macrophages in CCR2-/-, compared to WT, male mice and a significant reduction in CD11c-CD206+ anti-inflammatory macrophages (Figure 4bi, ii and iii). Again, these phenotypes progressed with age and were not seen in female CCR2-/- mice. All other tested fat pad depots in CCR2-/- males and females showed a significant reduction in the number of infiltrating inflammatory monocytes with a concomitant decrease in the number of adipose tissue macrophages (ATMs) in visceral and brown fat but not in axillary, fat depots (Supplementary Figure 1A). Further characterisation of the inguinal ATM in CCR2-/- male mice revealed increased expression of CD11b and of M1-type pro-inflammatory markers (Supplementary Figure 1B and 1C). In addition to macrophages, flow cytometric analysis revealed a significant increase in both neutrophils and eosinophils in the inguinal fat pads of male CCR2-/- mice which, again, increased with age (Figures 4ci-iii) and was not seen in female mice. Importantly, the increased numbers of macrophages, neutrophils and eosinophils was significantly ameliorated when the CCR2-/- male mice were isolated from each other again confirming the role for fighting in this phenotype (Supplementary Figure 2A). Analysis of visceral fat showed no differences between co-housed and segregated CCR2-/- male mice (Supplemental Figure 2B), with normal numbers of leukocytes (Supplemental Figure 2C), but a reduction in macrophages compared to WT (Supplemental Figure 2D).

Figure 4:

In the inguinal lymph nodes of CCR2-/- male mice, we observed a 20-fold increase (8-10 weeks) and up to 150-fold increase (16 weeks) in numbers of subcapsular macrophages (Figure 4d). This increased lymph node cellularity was not restricted to macrophages and whilst no differences were observed at 8-10 weeks of age, inguinal lymph nodes of CCR2-/- males displayed increased numbers of B-cells and CD4 and CD8 T-cells by 16 weeks of age (Supplementary Figure 1D). Again, female mice did not show this phenotype. Mandibular and mesenteric lymph nodes showed no increase in subcapsular macrophage numbers between WT and CCR2-/- males (Supplementary Figure 1E), suggesting that the increased lymph node size is restricted to the inguinal lymph nodes. Thus, both the atrophied fat pad, and enlarged lymph node, are characterised by increased levels of resident leucocytes.

Reduced inguinal fat pad weight and increased inguinal lymph node weight is independent of iCCRs

The data above indicate emergence of an ATM population in the fat pads of male CCR2-/- mice despite severe depletion of inflammatory monocytes. We therefore next investigated roles for other iCCRs in ATM recruitment by analysing the leukocyte content of the inguinal fat pad of iCCR-/-(11) (compound deletion of CCR1, CCR2, CCR3, and CCR5) male mice. Importantly, these mice phenocopy CCR2-/- male mice in that they display equivalent reciprocal changes in fat pad and lymph node weights (Figure 5ai and ii). Flow cytometry analysis indicated that, like CCR2-/- mice, the iCCR-/- male fat pads displayed a generally increased inflammatory cellular content with a particularly significant increase in neutrophils (Figure 5bi and ii). In addition, an increase in CD11c+ CD206+ ATMs was also seen (Figure 5c) which increased as a proportion of total ATMs as mice aged (Figure 5d). Importantly, numbers of eosinophils in the iCCR-/- fat pads were not increased likely as a direct result of the deletion of the eosinophil receptor CCR3 (Figure 5a). Thus, these data show that other iCCRs do not contribute to the fat pad macrophage accumulation seen in CCR2-/- mice and discount eosinophils as a cell type contributing to the phenotype.

Figure 5:

The Fat Pad Phenotype can be induced in WT females in a model of skin inflammation

Given the association of the phenotype with inflammation, we next tried to recapitulate it by inducing inflammation in WT female mice using the Imiquimod (IMQ) model of skin inflammation(13). Following IMQ treatment, WT females showed a dramatic reduction in inguinal fat pad weight and an increase in lymph node weight compared to females treated with control cream (Figure 6ai-ii). This phenotype was also apparent in IMQ-treated CCR2-/- female mice. In keeping with the fact that IMQ applied to the dorsal skin enters the circulation and thus become systemic(14), we observed a similar effect in axillary fat pad and lymph nodes (Supplementary Figure 3A). Inguinal lymph node and fat pad weight also correlated with each other, but again the function could be divided in two: vehicle treated females appeared to have no correlation between lymph node and fat pad size, whilst IMQ-treated females showed a strong negative correlation between inguinal lymph node size and fat pad weight (Figure 6b). Analysis of the cellular content of the inguinal fat pad revealed an increase in CD45+ cells in IMQ-treated WT and CCR2-/- females (Figure 6ci) associated with a very strong (∼5 fold) increase in macrophage numbers (Figure 6cii). Unlike fighting CCR2-/- males, we observed no increase in neutrophils, eosinophils, or lymphocyte numbers in the fat pad (Supplementary Figure 3B). Similar to CCR2-/-fighting males, fat pad macrophages in WT females treated with IMQ showed a significant increase in the levels of CD11b and CD11c (Supplementary Figure 3Ci, 3Cii).

Figure 6:

As expected, WT females treated with IMQ showed a strong increase in number of CD45+ cells in the skin (Figure 6e), and most of this increase was attributed to macrophages, however eosinophils, monocytes, DCs and lymphocytes were also significantly increased (Supplementary Figure 3D). Similar to co-housed CCR2-/- males, we observed a strong correlation between the number of CD45+ cells in the skin and the inguinal fat pad weight (Figure 6e), again suggesting that the extent of skin inflammation is correlated with local lymph node and fat pad size.

Overall, these data demonstrate that the phenotype observed in CCR2-/- male mice can be recapitulated in inflamed WT female mice and highlight macrophages as a potential driver cellular population. Our data further demonstrate, again, that the phenotype and the macrophage accumulation are independent of CCR2.

The fat pad phenotype is at least partially dependent on macrophages

Given the common emergence of a distinct macrophage population in the CCR2-/- male inguinal fat pads, as well as in WT female mice treated with IMQ, we next assessed the transcriptome, and the potential role, of this population in inguinal fat pad, and lymph node, remodelling. To determine the transcriptional profile of the ATMs that emerge in inflamed fat pads we used ATMs from resting, or IMQ-treated, females as this inflammatory model allows for precisely timed kinetic analysis. Principal component analysis revealed that the ATM in inflamed fat pads were transcriptionally distinct from those present in resting fat pads with approximately 1100 upregulated and 700 downregulated transcripts in inflamed ATMs (Figure 7a). Bulk comparison of the control and Aldara-treated ATMs revealed that both display a core macrophage signature but that, whilst those from control female mice resembled standard inflammatory macrophages, those from Aldara-treated mice represented a distinct macrophage population which displayed upregulation of a range of genes involved in lipid metabolism, lipid homeostasis and cholesterol efflux (Figure 7b). Further analysis of the transcripts preferentially expressed in the ATM from the IMQ-treated mice revealed considerable overlap with transcripts preferentially expressed in a previously described monocyte derived population of fat pad macrophages which enter fat pads in response to either high-fat diet or calorific restriction(15) (Figure 7d). Overall, these transcriptomic data indicate that the inflammation-associated fat pad (Figure 7c) is characterised by the presence of an atypical macrophage population which displays enhanced expression of lipid-metabolising genes.

Figure 7.

To determine the overall contribution of macrophages to the observed phenotype, we repeated the IMQ model in female mice in the presence or absence of a pharmacological blocker of CSF1-R(16, 17). Assessment of the extent of IMQ-induced cutaneous inflammation using a modified PASI score(13) indicated, surprisingly, that CSF1-R blockade had no effect on the overall severity of the skin inflammation following IMQ treatment (Figure 7e). In contrast, whilst the blocker had no effect on the inguinal fat pads or lymph node in vehicle treated animals, it ameliorated both the reduction in fat pad, and increase in lymph node, weight in the IMQ treated mice (Figure 7f-g).

Overall, these data reveal the presence of an atypical macrophage population in the remodelled fat pad and demonstrate a role for macrophages in the inguinal fat pad and lymph node remodelling downstream of inflamed skin.

The fat pad phenotype is associated with lipolysis and free fatty acid availability

A number of genes with key roles in fatty acid metabolism were up regulated in the inflamed fat pad ATMs (Figure 8a). This is indicative of active lipolysis in the fat pads and, given the importance of fatty acids as energy sources for the immune system(8, 10, 18, 19), suggested that the fat pad may act as a source of energetics for the embedded lymph node. We therefore next examined free fatty acid levels in the fat pads and the associated lymph nodes. As shown in the exemplary data in Figure 8b, the availability of a number of free fatty acid species demonstrated significant increases in both the fat pad, and lymph nodes, over a time-course covering days 1, 3 and 5 post-IMQ treatment. Notably the levels of the individual fatty acids generally increased in the fat pads prior to the increase in the lymph nodes. In addition, at least for the C20 fatty acids, levels in the fat pads far exceed those in the lymph nodes. Together these observations suggest that the fatty acids have originated from the fat pads and not the lymph nodes and raise the possibility of fatty acid donation from the lipolysis-active fat pads to the lymph nodes.

Figure 8:

Discussion

It has been realised for some time that the high energy requirements of an immune response are met, in significant part, by lipolysis of perinodal fat deposits(18, 19). However, the trigger for this has remained elusive. Here we demonstrate a clear link between the extent of peripheral cutaneous inflammation and remodelling of draining lymph nodes and perinodal fat pads. Specifically, we see a rapid reduction in the size of the fat pads with a reciprocal increase in the size of the embedded lymph node. This phenotype was seen downstream of fighting wounds, in CCR2-/- male mice, and in WT female mice treated with the inflammatory TLR7 agonist IMQ and was associated with the accumulation of a specific macrophage population in the atrophied fat pads. Notably, this phenotype was observed in both the fighting CCR2-/- males, as well as IMQ treated CCR2-/- females, demonstrating that the phenotype is independent of CCR2 and peripheral inflamed site and the draining fat pad and lymph node.

Interestingly, we observed a more robust inflammatory response to fighting-associated wounding in co-housed CCR2-/- male mice compared to isolated CCR2-/- male mice or WT male mice. This is a counterintuitive observation given the non-redundant role for CCR2, and its ligands, in monocyte egress from the bone marrow and recruitment to inflamed sites(11, 12, 20). Notably, similar results have been reported from other studies on inflammation in CCR2-/- mice(21–24) and potentially reflect the inability to recruit a specific CCR2-dependent anti-inflammatory macrophage population which would otherwise restrict the extent of the inflammatory response(24). It is also important to note that, despite the anticipated monocytopenia that is seen in CCR2-/- mice, there is no difference in the total number of macrophages in the inflamed skin of co-housed and WT male mice suggesting that in the CCR2-/- mice the macrophages are derived from CCR2-independent monocyte recruitment or in situ macrophage proliferation.

In terms of the altered macrophage population in the atrophied perinodal fat pad, transcript omics analysis reveals these to be similar to a population of ATM described by Weinstock and colleagues as being markedly increased upon calorific restriction(15). There are many other reports of macrophage recruitment to perinodal fat pads following inflammatory stress or calorific restriction(25, 26). The ATMs identified in the current study are characterised by relatively increased expression of genes involved in lipolysis and analysis of free fatty acids indicate active lipolysis in the fat pad in response to inflammation. Furthermore, the use of a well-characterised pharmacological blocker of CSF1R(17) clearly indicates a requirement for macrophages in the remodelling of the perinodal fat pad and lymph nodes. Overall, this points to the recruitment (or in situ proliferation) of a CCR2-independent macrophage population with enhanced lipolytic activity. Interestingly, there have been reports of a direct conduit system between perinodal fat deposits and the embedded lymph nodes(27) suggesting a clear route for transfer of fatty acids from the fat pad to the lymph node.

In summary, our study demonstrates a striking link between peripheral inflammation and perinodal fat pad atrophy and lymph node expansion. This suggests that inflammation is a key determinant of local energy provision from perinodal fat pads to lymph nodes and that this can be seen independently of antigenic challenge. Our data are also important in that they suggest that pro inflammatory vaccine adjuvants may mediate aspects of their function(28, 29), remote from the injection site, by priming downstream fat pads and embedded lymph nodes for the development of subsequent immune responses.

Materials and Methods

Mice and co-housing

Wild-type (WT), CCR2-/- and iCCR-/- mice are on a C57BL/6 background, and were maintained as previously described(11, 20), in a specific-pathogen free facility at the University of Glasgow under the auspices of a U.K Home office Licence. All animal procedures were approved by the local ethical review committee. Animals used were both male and female, ‘young’ mice were classified as 8-10 weeks of age, adult mice were 14-16 weeks of age, while ‘old’ mice were classified as 20+ weeks of age. To assess the impact of fighting and the contribution of skin injuries to the inflammatory phenotype observed, 8-10 week old male litters were separated and co-housed with individuals from other litters of the same age in groups of 4-5 for 6 weeks.

Adipose Tissue, Skin and Lymph Node digestion

Mice were sacrificed and tissues extracted for flow cytometry and RNA sequencing. Excised adipose tissue (inguinal fat pads, infrascapular brown fat, visceral white fat and axillary fat pads) were coarsely cut and digested for 1 hour on a Thermoshaker at 37°C at 850rpm in 1ml of Liberase in Hanks buffered saline (with Ca and Mg, ThermoFisher Scientific), at a final concentration of 0.44 Wunch units/sample. Samples were vortexed at the 30-minute mark, placed on ice after 1 hour, and filtered through a 100µm mesh strainer in a 50ml falcon, and the filter was washed with 10ml of PBS. Samples were spun down at 300g for 5 minutes, and the pellet was resuspended and passed through a 70µm mesh strainer. Skin samples were processed in a similar way however digestion was performed at 1000rpm, with vortexing every 15 minutes to encourage dissociation. Lymph nodes were collected and mechanically digested through a 70µm mesh strainer with the aid of a 2ml syringe plunger. The mesh was washed with 10ml PBS, and the sample was spun down at 300g for 5 minutes. The resulting pellet was resuspended in 10ml of PBS and re-filtered through a new 70um strainer to remove remaining fat and connective tissue.

Isolation of Adipose tissue macrophages

Cell counts from digested and filtered adipose tissue samples were obtained via a Luna Cell Count Reader, and the appropriate amount of Anti-F480 MicroBeads Ultrapure (Miltenyi Biotech) were added to each sample to magnetically isolated adipose tissue macrophages according to the manufacturer’s protocol. Briefly, 10 µl of Anti-F480 MicroBeads Ultrapure were added for every 10⁷ cells, and incubation was performed for 15 minutes on ice in the dark. Cells were washed with 10ml of FACS Buffer and centrifuged at 300g for 10 minutes. The pellet was resuspended in 500µl, and the suspension was placed on an MS MACS column on an OctoMacs separator (Miltenyi Biotech). The flow-through containing unlabelled cells was discarded, while the eluted fraction containing magnetically labelled cells was collected and enriched over a second MS columns to achieve higher purities. At the end of the second magnetic isolation, a small portion of each sample was collected and stained with CD64, Ly6G and SiglecF fluorescent antibodies to test macrophage content and amounts contaminating cells via flow cytometry. RNA extraction was performed on samples with >90% CD64+ cells.

Flow cytometry

Starting from single-cell suspensions, cells were first stained with a fixable viability dye (Zombie Aqua; BioLegend), and Fc receptors were blocked using FcR Blocking-Reagent (Miltenyi Biotec). Surface marker staining was performed with the appropriate Ab mix diluted in FACS Buffer (PBS/2 mM/EDTA/0.5% FBS) on ice for 20 min. Cells were washed, fixed with Fixation Buffer (420801; BioLegend), and resuspended in FACS buffer before analysis. Data were acquired using either the Fortessa (BD Biosciences) flow cytometer and analyzed using the FlowJo 10 software. Voltages and compensation were determined using UltraComp eBeads (01-2222-42; Bioscience) as single controls. Positive staining was determined using Fluorescence-Minus-One controls. The gating strategy and antibody list for inguinal fat pad leukocytes are outlined in Supplementary figure 4.

AZD7507 Treatment

To achieve macrophage depletion, mice received 100l of AZD7507(17) (MedChemExpress, HY-117244), an orally active CSF-1R inhibitor, at a concentration of 100mg/kg (25mg/mL). 250mg of AZD7507 were resuspended in 10ml of MilliQ water with 0.5% HPMC and 0.1% tween 80. 100µl of AZD7507 suspension was added to 1ml syringes, and administered to mice via oral gavage, twice a day, for 8 days. Mouse weights were monitored daily, and after 8 days, the back of the mice was shaved to allow for Aldara application.

Aldara/Imiquimod Model

WT and CCR2-/- female mice were each treated with a ∼62.5 mg dose of Aldara™ (Meda AB) cream containing 5% Imiquimod or an equivalent volume of control cream (Boots Aqueous Cream/10% Vaseline Lanette Cream) by application to shaved dorsal skin near the base of the tail (∼3mm² area)(13). Cream was applied for 3 consecutive days (day 1, day 2 and day 3) and the mice culled 24h after the final application. Mice were monitored daily for weight loss. Following euthanasia blood was collected immediately using a syringe flushed with 0.5 M EDTA. Following blood collection, mice were transcardially perfused with 20ml of ice-cold PBS. When tissue was required for immunostaining, mice were subsequently perfused with 4% PFA.

RNA Extraction and processing

After F480-magnetic positive selection, the eluted cell suspensions from 6 inguinal fat pads were pooled together to obtain enough ATMs (around 30,000 cells) for low-input RNA sequencing. Cells were spun down at 1000g for 10 minutes, lysed, and spun again to remove bound magnetic beads before proceeding with RNA extraction. RNA was extracted using the RNeasy Microkit with on-column DNAse I digestion (Qiagen), following manufacturer’s instructions, obtaining samples of around 100-150ng of total RNA in water.

RNA samples were then sent to a commercial sequencing provider (Novogene), for RNA QC, mRNA library preparation (poly-A enrichment) and Illumina Sequencing PE150.and sequencing. Briefly, raw reads of fastq format were processed through in-house scripts, removing adapter, ploy-N and low-quality reads from raw data. Index of the reference genome was built and paired-end clean 1 reads were aligned using Hisat2 v2.0.5. FeatureCounts v1.5.0-p3 was used to count the reads for each gene. Differential expression analysis of two conditions/groups was performed using the DESeq2Rpackage (1.20.0). The resulting p-values were adjusted suing the Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with an adjusted P-value <= 0.05 found by DESeq2 were assigned as differentially expressed

Comparison of scRNA cell type expression with Aldara

The Weinstock et al dataset(15) was downloaded from GEO (GSE141036) as a Seurat object. The data was analysed and converted into pseudo bulk. In detail, Seurat (v5.0.3)(ref - 2) was used to load the dataset. Cells were filtered for nFeature_RNA > 200 & percent_mito < 5. The data was SCT transformed, and PCA, find clusters, and generate UMAP performed using PC 1-30. Cluster markers were identified using FindAllMarkers with the settings only.pos = TRUE, min.pct = 0.25, logfc.threshold = 0.25. All other settings were left to default. As expected the resulting UMAP and clusters were highly similar to the manuscript, and so cell types were matched. Next a per cell type pseudo-bulk count table was generated from the raw counts, using a loop in R. This was normalised into an expression table using DESeq2 (v3.18)(ref - 3), under default settings. To compare to the Aldara dataset the mean value for each gene in control and Aldara groups was calculated. Both the Aldara and pseudo-bulk data were filtered to include only the marker genes from the overlap analysis (see above). Finally, the pseudo-bulk expression values for each cell type were compared to the equivalent values in the control or Aldara group using a Spearman correlation.

Free fatty acid analysis

A methanol extraction of foetal bovine serum was used as a quality control sample and used for instrument conditioning prior to sample injection. Methanol extracts from samples were centrifuged at 13000 rpm for 5 min at 4°C, to separate protein precipitate from the methanol supernatant. For each sample, an aliquot of supernatant was diluted 1:1 (v/v) with MilliQ water (Merck Millipore) spiked with C17:0 fatty acid at 0.8 µg/ml and transferred into an autosampler vial. Sample vials were kept at 8°C in an Acquity Premier FTN autosampler (Waters corporation). An aliquot of 10µL of each sample was injected onto CORTECS UPLC T3 1.6 µm, 3 x 100mm column (Waters corporation) held at 50°C within an Acquity Premier column manager (Waters corporation). Free fatty acids were separated using Acquity Premier BSM UPLC pump (Waters Corporation), using 0.2 mM Ammonium formate with 0.01% formic acid for solvent A, and Isopropanol:Acetonitrile (1:1 v/v) 0.2 mM ammonium formate with 0.01% formic acid for solvent B. Elution gradient was as follows: 30% B for 0.1 min, 30-65% in 0.9 min, 65-75% B in 2.5 min, 75-98% B for 1.5 min, 98% held for 1min before re-equilibration at 30% for 2 minutes. Fatty acids were detected in negative ion mode using a Xevo TQ Absolute (Waters Corporation) triple quadrupole mass spectrometer. Ionisation conditions were heated electrospray source with capillary voltage 1.9kV, desolvation temperature 500°C, desolvation gas 1000L/Hr, cone gas 150L/Hr and source temperature of 120°C. Fatty acid ions were detected using MRM mode with parent to parent transitions (see attached MRM table), with a +/- 0.5 min window around the expected elution time, with a cone voltage of 30V and collision energy of 15eV. 41 transitions were acquired from C6:0 to C24:0, including common polyunsaturated fatty acids. The following fatty acids were matched by mass and retention time to authentic standards, C6:0, C8:0, C10:0, C12:0, C14:0, C16:0, C18:0, C20:0, C22:0, C24:0, alpha-linolenic acid (18:3n3), gamma-Linolenic acid (18:3n6) and Arachidonic acid (C20:4n6). All other reported fatty acids were determined based on formula mass and estimated retention time. Peak picking of the acquired data was performed using TargetLynx (Waters Corporation), before exporting to a text file. The results text file was then read using custom scripts in R (version 4.41). Signals for background ions were averaged and then subtracted from analyte signals. Any negative values were then removed from the analysis. Chains shorter than C12:0 were determined to be too variable, along with the internal standard C17, which was spiked into the sample, and therefore not reported. The resulting data table was then saved as an Excel spreadsheet.

Statistical analysis

All statistical analyses were performed using Graph Pad Prism.

Data availability

RNA sequence data were deposited in GEO (accession number GSE317411). All data generated or analyzed during this study are included in the manuscript and supporting files; source data files have been provided for all figures.

Acknowledgements

The authors acknowledge the help of the MVLS Core Cellular Analysis Facility. This work was funded by the Wellcome Trust (217093/Z/19/Z) and the Medical Research Council (MR/V010972/1).

Additional files

Supplementary Information

Additional information

Funding

Wellcome

https://doi.org/10.35802/217093

• Gerard J Graham

• Robin Bartolini

UKRI | Medical Research Council (MRC) (MR/V010972/1)

• Gerard J Graham

• Gillian J Wilson

• Robin Bartolini