Metabololipidomic and proteomic profiling reveals aberrant macrophage activation and interrelated immunomodulatory mediator release during aging

Abstract Macrophages adapt distinct pro‐inflammatory (M1‐like) and pro‐resolving (M2‐like) phenotypes with specific tasks in the immune response and tissue homeostasis. Altered macrophage responses with age are causative for unresolved inflammation, so‐called inflammaging, and lead to higher infection susceptibility with unfavorable progression. Here, we reveal molecular determinants of age‐related changes in phenotypic functions of murine peritoneal macrophages (PM) by employing comprehensive mass spectrometry‐based proteomics (4746 protein groups) and metabololipidomics (>40 lipid mediators). Divergent expression of various macrophage‐specific marker proteins and signaling pathways indicates aberrant PM phenotypes in old mice which detrimentally impact their capabilities to release immunomodulatory chemokines and cytokines. We show that aging strikingly compromises the polarization process of macrophages to adapt either pro‐inflammatory or pro‐resolving phenotypes, thereby yielding aberrant and afunctional macrophage subtypes that cannot be readily assigned to either a typical M1 or M2 phenotype. In particular, the phenotypic adaptation of the bacteria‐challenged metabololipidome in macrophages related to inflammation is severely limited by age, which persists across ex vivo polarization towards M1 and M2a macrophages. Our results establish distinct age‐associated PM phenotypes outside of the simplified M1 and M2 dichotomy and challenge the dogma of increased pro‐inflammatory macrophage pre‐activation due to aging by revealing maladaptive functions throughout all phases of inflammation, including resolution.


| INTRODUC TI ON
Although the life expectancy for humans markedly increased in the last decades and is projected to rise further, this positive trend is, however, accompanied by a higher susceptibility to inflammatory and infectious diseases in elderly individuals (Franceschi et al., 2018).
Immunosenescence is widely regarded as the root cause for the ageassociated impairment in the efficiency of the immune response (Aiello et al., 2019). One major hallmark is inflammaging, which is classified as a systemic, chronic, and low-grade inflammation impairing tissue homeostasis and repair during aging (Franceschi et al., 2018). In this context, macrophages play a major role as regulatory innate immune cells which, by releasing immunomodulatory mediators, are involved in initiation, perpetuation, and resolution of inflammatory processes (Okabe & Medzhitov, 2016). Macrophages Furthermore, they are an important linkage to the adaptive immune system through the presentation of antigens, which are recognized by T cells (Gaudino & Kumar, 2019).
Due to their extensive plasticity, macrophages are prone to ageassociated changes within the inflammatory microenvironment, termed macrophaging (Prattichizzo et al., 2016). Several aging-driven defects in macrophage polarization were reported, but conclusive age-related phenotypes remain elusive. Thus, observations include decreased polarization markers in peritoneal macrophages (PM) (Linehan et al., 2014), altered phagocytic activity of Kupffer cells and alveolar macrophages (Hilmer et al., 2007;Linehan et al., 2014;Wong et al., 2017), and a generally increased pro-inflammatory phenotype of bone-marrow derived macrophages (BMDM) and Kupffer cells of old origin (Becker et al., 2018;Gibon et al., 2016). However, in-depth characterization of the phenotypic changes of macrophages with age and evaluation of their capabilities to adapt to the dynamic inflammatory microenvironment are missing.
Here, we provide substantial insights into age-associated aberrations of the proteome and metabololipidome of murine tissue-resident PM and their distinct phenotypes with concomitant alterations in the release of immunomodulatory mediators. We employed comprehensive metabololipidomic and proteomic profiling of PM from adult and old mice to reveal age-related changes of phenotypic markers, signaling and activation pathways, and release of cytokines/chemokines and LM. Our results indicate a deprivation of inflammatory pathways in old PM, which persists across ex vivo polarization and impairs the temporal release of immunomodulatory mediators during bacterial infection and the clearance of bacterial debris.

| Aging increases resident PM numbers and impacts macrophage activation markers
To study how aging affects the composition and phenotype of macrophages within the murine peritoneal cavity, we isolated resident peritoneal cells from adult (4-6 months) and old (>24 months) C57BL/6JRj mice by peritoneal lavage. Cells from both age cohorts were similar in terms of viability, diameter, and average circularity ( Figure 1a). The cell number significantly increased from averagely 4.48 to 13.15 × 10 6 cells as consequence of aging ( Figure 1a). To confirm that after initial seeding PM are the most abundant cell type, we assessed CD11b and F4/80 as surface markers (Ghosn et al., 2010) by flow cytometry. Regardless of the age, most cells (83.2% to 89.1%) were positive for both markers, with slightly higher portion of CD11b + /F4/80 − cells from old mice ( Figure S1a). Notably, the mean fluorescence intensity (MFI) for both CD11b and F4/80 was strongly reduced as consequence of aging (Figure 1b,c), confirming different origin of macrophages (Bain et al., 2016). To study if aging affects the functionality of PM, we assessed their phagocytic capacity for engulfing fluorescent-labelled E. coli particles and found no difference between both age cohorts ( Figure S1b). Interestingly, we found that 390713860; Carl Zeiss Foundation (IMPULS); Federal Government of Germany and the State of Thuringia K E Y W O R D S aging, eicosanoids, inflammation, lipidomics, macrophage activation, mediators of inflammation, peritoneal macrophages, proteomics treatment of PM from adult mice for 6 h with the supernatant of LPSchallenged (18 h) old (but not adult) PM led to a significant reduction in phagocytosis ( Figure S1c). This indicates that the secretome of adult and old PM has divergent impact on macrophage functions.
We then investigated how aging affects activation markers and interrelated inflammatory signaling pathways in PM by employing mass spectrometry (MS)-based proteomic and metabololipidomic profiling. Comprehensive libraries for both the proteome (DDA F I G U R E 1 Aging establishes a distinct macrophage phenotype in the peritoneal cavity. (a) Number, viability, diameter, and circularity of isolated cells from the peritoneal cavity of adult (4-6 months) and old mice (>24 months) were assessed with a Vi-CELL XR system (n = 25). Expression of surface markers (b) CD11b and (c) F4/80 on PM from adult (turquoise) and old mice (grey) was measured by flow cytometry (n = 9-10). Results are shown in representative histograms for both surface markers with mean fluorescence intensity (MFI) of all replicates. (d) Principal component analysis (PCA) of the proteome of PM from adult and old mice measured by DIA mass spectrometry (n = 5). (e) Volcano plot displays proteins significantly increased or decreased by aging (Table S1; n = 5). Not affected proteins are shown in grey. Dashed lines indicate a cutoff for significance of p < 0.05 and absolute fold changes (log 2 ) > 0.58, respectively. Significantly regulated proteins of interest were labelled with an emphasis on those involved in macrophage function during inflammation. (f) Ingenuity pathway analysis of significantly regulated protein clusters in PM from old mice in comparison with adult (n = 5). Displayed pathways are among the Top 100 most significantly regulated pathways and were selected based on the relevance for aging and inflammation. Bonferroni-Holm corrected p-values are implicated by color and z-scores by bar size. Statistics: Data are shown as (a-c): median (min to max) or (e, f) median and p-values were calculated by 'a-c' unpaired two-tailed Student's t-test with or without Welch's correction (Table S7), (e) Spectronaut™ (Table S1) or (f) QIAGEN Ingenuity Pathway Analysis. ***p ≤ 0.001, ****p ≤ 0.0001, ns, not significant.   Figure S1d). Principal component analysis (PCA) of the overall proteome revealed striking differences between age groups as indicated by clearly defined, separated clusters for adult and old PM (Figure 1d). Among the most strikingly regulated proteins by aging, we found several macrophage activation markers and key regulatory enzymes in inflammation ( Figure 1e). Thus, reduced amounts of intercellular adhesion molecule 1 (CD54) that mediates homing and trafficking of inflammatory cells to distant tissues, or chitinase-like protein 3 (YM1) with chemotactic activity for eosinophils, in old PM ( Figure 1e) implicate decreased activation states.
Interestingly, the levels of the LM-biosynthetic proteins COX-1, prostacyclin synthase (PTGIS), and 5-LOX-activating protein (FLAP) were strongly reduced (>twofold) in old PM (Figure 1e), suggesting a close association between aging and the formation of inflammatory LM produced by these enzymes. In contrast, various signaling proteins, for example, mTOR, IGF-2, and NFKBIE were elevated due to aging ( Figure 1e). The proteome data also confirmed the reduced levels of CD11b and F4/80 in old PM (assessed by flow cytometry, see above) alongside decreased CD14 expression ( Figure S1e).
To provide a concise overview of signaling pathways affected in PM by aging, we exploited ingenuity pathway analysis (IPA; Qiagen) to cluster the proteomic data. Among the top-100 affected pathways were several of those that are pivotal for macrophage activation and inflammatory signaling (Figure 1f). Proteins connected to signaling of eicosanoids, EIF2, ERK/MAPK, mTOR, Toll-like receptor, TREM1 and micropinocytosis as well as the nuclear factor erythroid 2-related factor 2 (NRF-2)-mediated oxidative response were markedly downregulated in old PM (Figure 1f), while proteins related to the role of Fcγ receptor (phagocytosis), NFAT (immune response), necroptosis signaling, and oxidative phosphorylation were elevated ( Figure 1f). Furthermore, senescence signaling was upregulated while the sirtuin cascade was impaired, both of which are associated with aging (Imperatore et al., 2017;Prattichizzo et al., 2016). Together, these results imply a distinct macrophage phenotype for old PM, potentially connected to altered immune responses against pathogens and inflammatory signaling.

| Downregulation of COX-1 and FLAP during aging compromises PG and LT formation
Production of either LT and PG or SPM upon macrophage challenge with bacteria distinguishes their pro-inflammatory or pro-resolving characteristics, respectively (Werz et al., 2018). To study how aging impacts the response of PM to bacterial challenge, we infected adult and old PM with pathogenic E. coli (MOI of 50) and performed comprehensive metabololipidomic profiling. PCA indicated similar LM signatures for both PM cohorts with only minor deviations ( Figure 2a). Yet, closer inspection of single bioactive LM revealed marked differences as consequence of aging. Most strikingly, PGE 2 formation was strongly reduced from 3325 to 247 pg/mL, but also the levels of pro-inflammatory LTB 4 (2024-1083 pg/mL) and trans-LTB 4 (3278-1819 pg/mL) were impaired, in old PM ( Figure 2b). Interestingly, besides these pro-inflammatory LM also SPM like maresin-1 (MaR1) from 72.7 to 27.3 pg/mL and protectin DX (PDX) from 20.2 pg/mL to non-detectable (<3 pg/mL) amounts were lowered ( Figure 2b).
Next, we analyzed whether aging impacts the overall composition of bacteria-induced LM signature profiles by clustering all detectable LM according to their biosynthetic pathways. Aging impaired formation of COX-and 5-LOX-derived products from 18% to 6% and 40.3% to 28.5% of total LM, respectively ( Figure 2d). The aging-induced drop of SPM from 1.7 to 1.0% is mainly due to markedly reduced protec-

| Aging alters the macrophage phenotype of naive tissue-resident macrophages
To study how aging affects the polarization state of naive PM resident in the peritoneal cavity, we first assessed surface markers that are generally used to define pro-inflammatory M1 (CD54, CD86) or proresolving M2a (CD200R, CD206) phenotypes in mice using flow cytometry (Schulz et al., 2019). PM from old mice displayed higher levels of CD54 (ICAM-1), while CD86 (B7-2) was overall weakly expressed in PM with lower levels in cells from old mice ( Figure 3a). Interestingly, CD200R was reduced due to aging, while CD206 was strikingly increased ( Figure 3a). These findings essentially agree with the proteomic analysis, although differences of CD54 in old versus adult PM did not correlate ( Figure S1f (Table S6; n = 8). Metabolites were grouped as follows:  (Table S6; n = 8). (h) Ratio of pro-inflammatory PGE 2 , LTB 4 , t-LTB 4, and epi t-LTB 4 to all SPM (see above) released by adult and old PM after E. coli infection (Table S6; n = 8). Statistics: Data are shown as (b, d) mean or (c) median (min to max) or (e-h) mean ± SEM and p-values were calculated by unpaired two-tailed Student's t-test with or without Welch's correction (Table S7). **p ≤ 0.01, ***p ≤ 0.001, ns, not significant.  To examine whether observed age-related changes in polarization and the metabololipidome of PM apply also to macrophages from other immunologic niches, we performed side-by-side analysis of PM with splenic macrophages (SM) and BMDM from the same mice. PCA of the metabololipidome after E. coli infection revealed distinct phenotypes for both SM and BMDM as consequence of aging

F I G U R E 3 Investigation of polarization markers in naive peritoneal macrophages.
(a) Expression of M1 (CD54, CD86) and M2a surface markers (CD200R, CD206) in naive PM from adult and old mice was determined by flow cytometry. Results are shown as representative histograms with MFI of all replicates (n = 4-5). Abundance of (b) M1 and (c) M2a proteomic markers in PM from adult and old mice was determined by DIA mass spectrometry. Each dot represents the median of a different marker (the complete list is given in Table S2) and median of all markers is indicated by vertical bars, whereas dotted lines show median in M1-and M2a-PM (48 h) from adult mice (n = 5). (d) Heatmap showing age-related changes in the expression of M1 and M2a proteomic markers in PM from old mice in comparison with adult (n = 5). Fold change is implicated by color scale. Statistics: Data are shown as (a) median (min to max) or (b-d) as median and p-values were calculated by unpaired two-tailed Student's t-test with or without Welch's correction (Table S7). *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001.     (Table S6). SM mainly released PG, TXB 2 , LT and moderate levels of RvD5 and RvE1   ( Figure S3d), while BMDM produced low amounts of LM, mainly TXB 2 ( Figure S3c). Furthermore, we found significant downregulation of F4/80 and CD54 and a downward trend for CD86 and CD200R in all three macrophage types from different niches ( Figure S3e). Together, our data imply distinct macrophage polarization states of adult and old macrophages, with neither a classical M1-nor M2a-like phenotype, but clearly differing between the age cohorts.

| Impairment of functional polarization of PM as consequence of aging
To elucidate whether aging affects the polarization process of PM, In contrast, in old M1-PM, the release of these cytokines was delayed.
Interestingly, secretion of anti-inflammatory IL-1ra and IL-10 was more rapidly induced from old PM during M1 polarization when compared to adult M1-PM (Figure 5a,c). Note that the IL-10 release from old M1-PMs significantly dropped at 48 h, well below adult levels ( Figure 5c).
Finally, the release of anti-inflammatory TGFβ from M2a-PM remained low after 24 h of polarization as consequence of aging ( Figure 5c).
Together, while chemokine release from M1-PM is broadly increased by aging, the opposite holds true for pro-inflammatory cytokines, and anti-inflammatory ones tend to peak at odd timepoints, potentially compromising the crosstalk with other immune cells.

| Aging shifts the metabololipidome of M2a-PM towards pro-inflammatory LM
Since the M1-or M2-like phenotype is tightly connected to the capacity of forming either pro-inflammatory PG and LTs or pro-resolving F I G U R E 5 Aging leads to aberrant cytokine and chemokine release from polarized macrophages. The cytokine and chemokine secretome of adult and old PM after stimulation with LPS (100 ng/mL) and IFNγ (20 ng/mL was determined by means of a cytokine array ( Figure S7, Table S4). (a) Heatmap showing global time-related changes in the cytokine and chemokine secretome of old M1-PM in comparison with adult. Fold change is implicated by color scale. (b) The temporal release is shown for selected and prominent pro-inflammatory cytokines and chemokines secreted by M1-PM of both age cohorts over the course of 48 h. (c) Concentration of pro-inflammatory IL-1ra, IL-6, IL-10, IL-12, and TNFα in adult and old M1-PM, and TGFβ in adult and old M2a-PM was measured by ELISA (n = 6; except 24 h [IL-10]: n = 4-6). Statistics: Data are shown as (a, b) mean or (c) mean ± SEM and p-values were calculated by unpaired two-tailed Student's t-test with or without Welch's correction (Table S7). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, ns, not significant.    Figure S5e). Figure S6 summarizes the temporal release of both pro-inflammatory and pro-resolving mediators from M1-and M2a-PM and respective shifts as consequence of aging. Similar age-associated patterns are evident among LM derived from either COX ( Figure S6a) or LOX ( Figure S6b), with the exception of the COX product TXB 2 that resembles the pattern of LOX products ( Figure S6a). In conclusion, aging markedly impacts the adaptation of the metabololipidome during PM polarization, particularly in M2a-PM, which extends beyond a simple delay in SPM formation but rather implies a fundamental deficiency in LM biosynthesis.

| DISCUSS ION
Reduced immune response efficiencies of macrophages as con- and unresolved low-grade inflammation in elderly individuals (Doyle et al., 2018;Schädel et al., 2021).
We found markedly increased numbers of PM within in the peritoneal cavity of old mice, which displayed reduced expression of CD11b and F4/80, the most reliable discriminants for PM against other cell types within the niche (Ghosn et al., 2010). PM can be subdivided into small and large subsets, based on differences in size, expression of F4/80, and their ontogeny (Cassado Ados et al., 2015;Ghosn et al., 2010). Previous studies reported a shift in the ratio of small and large PM subtypes as consequence of inflammation or after trauma (Ghosn et al., 2010;Okabe & Medzhitov, 2014), due to monocytes being recruited from the circulation into the peritoneal F I G U R E 6 Impaired M2a macrophage polarization as consequence of aging causes a depletion of DHA-derived SPM levels after E. coli infection. Polarized PM (4-48 h) were infected with pathogenic E. coli for 90 min at a MOI of 50 and subsequent LM levels were determined by UPL-CM-SMS (Table S6; (Table S7). *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001. niche to replace senescent or apoptotic resident macrophages (Louwe et al., 2021). Functionally, small and large PM subsets display comparable phagocytic activity, but their response to TLR4 stimulation by LPS differed significantly (Ghosn et al., 2010). This could explain the increased PM numbers in old mice and may implicate that during aging a substitution of large PM with small PM occurs through monocyte recruitment. However, no significant differences in the cell size and the phagocytic activity of PM of both age cohorts were evident, contradicting findings with mice by others (Linehan et al., 2014). Previous studies selectively addressed large PM but neglected the substantial subpopulation of small PM of old mice that contribute to the inflammatory response within the peritoneal niche (Cassado Ados et al., 2015;Ghosn et al., 2010).
Notably, upon subsequent ex vivo culture for 24 h, the phagocytic capacity of adult, but not of old PM, was elevated regardless of the polarization state but vanished at 48 h. Interestingly, phagocytosis by adult PM was suppressed by the secreted mediators from LPSchallenged old PM but not vice versa, implying that the secretome of old resident PM may detrimentally impact the phagocytic function of macrophages.
Our MS-based proteomic profiling revealed a distinct ageassociated PM activation state, characterized by impaired inflammatory signaling pathways including mTOR, ERK/MAPK, NRF2, TREM1 and a rise in apoptosis signaling. Furthermore, canonical pathways analysis showed reduced sirtuin signaling in old PM, implying a more quiescent state and impaired self-renewal (Imperatore et al., 2017). Also, the protein levels of COX-1, cPGES, and FLAP, which are key enzymes in the biosynthesis of proinflammatory PG and LT (Radmark et al., 2015;Smith et al., 2011), were decreased in old PM, along with considerably lower PGE 2 and LTB 4 formation upon infection with pathogenic E. coli. Note that in old PM also formation of SPM (i.e., PD, RvD5, and MaR1) was impaired, suggesting potential deficits in immune response and tissue repair (Markworth et al., 2021). However, there was no dominance of pro-inflammatory PG and LT over SPM, which has been suggested as key feature of inflammaging (Arnardottir et al., 2014;Minhas et al., 2021). Side-by-side analysis of PM with SM and BMDM showed that age-related suppression of PG and TX after infection with pathogenic E. coli is a common feature among macrophages from different niches.
Previous studies suggest a rather pro-inflammatory M1-like phenotype for tissue-resident macrophages as consequence of aging (Becker et al., 2018;Van Beek et al., 2019). Our data from proteomic and metabololipidomic profiling revealed a distinct age-related phenotype for PM and, at least in terms of LM profile, also for SM and BMDM, which cannot be readily assigned to either an inflammatory M1-like or pro-resolving M2a-like phenotype. There is a general consensus that the classical categorization into the extreme phenotypes M1 and M2 is difficult to apply to macrophage populations in vivo, where a mixture of different activation states is common (Murray et al., 2014). In our study, several proteins and signaling pathways characteristic for M1 (e.g., CD86, HO-1, iNOS, and SOD-2) and for M2a (CD200R, TLR8, YM1, DMP-4) were downregulated in old PM. Yet, other markers for M1 (CD54, GSTA3, IGF-2, mTOR) or M2a (CD206, IRF-8, SLC1A5, PFKFB4) were elevated as consequence of aging. Therefore, our results underline the necessity to expand the classification of macrophages beyond the simple M1 and M2 dichotomy and to consider broader sets of phenotypic protein markers and immunomodulatory mediators, especially LM, to adequately attribute inflammatory or pro-resolving characteristics (Dalli & Serhan, 2012;Ley, 2017;Murray et al., 2014;Werz et al., 2018).
A phenotypic switch from M1-to M2a-PM is crucial for resolving inflammation, tissue repair, and regeneration to eventually return to Tailored temporal release of different chemokines, cytokines, and LM is crucial for mounting the host immune response (Rossol et al., 2011;Serhan & Savill, 2005). Our secretome analysis revealed elevated release of chemoattractants and activators of cells in innate and adaptive immunity from old M1-PM, especially at later timepoints (i.e., 48 h). These data concur with the age-related increased recruitment and persistence of immune cells within various niches like the adipose tissue, liver, lung, peritoneum, and spleen (Almanzar et al., 2020;Mogilenko et al., 2021). Along these lines, aging caused delayed and impaired release of the pro-inflammatory cytokines IL-6, IL-12, and TNFα but accelerated and elevated secretion of anti-inflammatory IL-1ra and IL-10 during M1 polarization. Such aberrations may lead to miscommunication between PM and other immune cells at the site of infection (Davies et al., 2013).
Transition from inflammation to resolution is accompanied by a LM class switch from pro-inflammatory PG and LT to pro-resolving SPM, which is directly connected to the polarization state of macrophages at sites of inflammation (Serhan, 2014;Serhan et al., 2015;Werz et al., 2018). Aging affected LM signature profiles in M1-PM mainly by lowering the formation of the major metabolite PGE 2 across the polarization process, which can be explained by de- aging delayed resolution in a model of self-resolving peritonitis (Arnardottir et al., 2014) with mechanistic insights into the role that PM play.
Conclusively, aging has a tremendous impact on the activation and functionality of PM resulting in aberrant macrophage phenotypes outside of the simplified M1 and M2 dichotomy. Our findings fit with the general observation that elderly individuals are prone to develop unresolved, chronic inflammation and are at higher risk to die from infections, obviously apparent in the COVID-19 pandemic (Fulop et al., 2021). Yet, our data challenge the dogma of increased pro-inflammatory macrophage pre-activation due to aging and instead reveal maladaptive properties in the plasticity and immunological competence of macrophages.

| Phagocytosis assay
After initial seeding, PM were aspirated, supplied with fresh macrophage medium (RPMI 1640, 10% heat-inactivated FCS, 2 mM L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin) and incubated with 5 μL of green-fluorescent E. coli particles (abcam, ab235900) per well (37°C, 2 h, 5% CO 2 ). After incubation, cells were carefully aspirated and washed repeatedly, and finally, 500 μL of assay buffer was added. Fluorescence was measured with a Novostar system (BMG Labtech), and relative phagocytosis activity was calculated through a standard curve of different concentrations of green fluorescent E. coli particles in assay buffer.

| Proteome analysis by UPLC-MS-MS
For spectral library generation, peptides were separated using a

| Sample preparation and metabololipidomic profiling by UPLC-tandem mass spectrometry
To the supernatants from co-incubations of PM and E. coli, deuterium-labelled LM standards were added, and samples were   Table S7. Statistical significance was assumed for comparisons with p < 0.05 and is indicated as: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, ns, not significant. For proteome data, proteins with a q-value below 0.05 and absolute log2-fold change above 0.58 were considered as significantly affected, unless otherwise stated. Displayed q-values were calculated by Spectronaut™ and indicated as *q < 0.05, **q < 0.01, ***q < 0.001, ns, not significant, unless otherwise stated.

ACK N OWLED G EM ENT
Open Access funding enabled and organized by Projekt DEAL.

FU N D I N G I N FO R M ATI O N
The authors gratefully acknowledge support from the mouse facilities of the FLI and BIZ as well as the Proteomics Core Facility at FLI.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare no competing interests.