Towards understanding the cell surface phenotype, metabolic properties and immune functions of resident macrophages of the peritoneal cavity and splenic red pulp using high resolution quantitative proteomics [version 1; peer review: 1 approved, 1 approved with reservations]

Background: Resident macrophages (Mφs) are distributed throughout the body and are important for maintaining tissue homeostasis and for defence against infections. Tissue Mφs are highly adapted to their microenvironment and thought to mediate tissuespecific functions involving metabolism and immune defence that are not fully elucidated. Methods: We have used high resolution quantitative proteomics to gain insights into the functions of two types of resident tissue Mφs: peritoneal cavity Mφs and splenic red pulp Mφs.  The cellular expression levels of many proteins were validated by flow cytometry and were consistently in agreement with the proteomics data. Results: Peritoneal and splenic red pulp macrophages displayed major differences in cell surface phenotype reflecting their adaptation to different tissue microenvironments and tissue-specific functions. Peritoneal Mφs were shown to be enriched in a number of key enzymes and metabolic pathways normally associated with the liver, such as metabolism of fructose, detoxification, nitrogen homeostasis and the urea cycle. Supporting these observations, we show that peritoneal Mφs are able to utilise glutamine and glutamate which are rich in peritoneum for urea generation.  In comparison, splenic red pulp Mφs were enriched in proteins important for adaptive immunity such as antigen presenting MHC molecules, in addition to proteins required for erythrocyte homeostasis and iron turnover. We also show that these tissue Mφs may utilise carbon and nitrogen substrates for different metabolic fates to support distinct tissue-specific roles. Conclusions: This study provides new insights into the functions of tissue Mφs in immunity and homeostasis.  The comprehensive Open Peer Review


Introduction
Resident macrophages (Mϕs) are distributed in tissues throughout the body and contribute to both homeostasis and defence against infectious disease. Resident Mϕs from different tissues are highly heterogeneous in properties and functions and this is reflected in considerable diversity in their transcriptomes 1 . Tissue Mϕs arise from both embryonic and adult sources including yolk sac, foetal liver and bone marrow and recent studies have shown that local signalling within tissues is sufficient to promote the acquisition of specialized Mϕ functions irrespective of their tissue of origin 2 . Therefore, tissue Mϕs have distinct phenotypic and metabolic characteristics that define tissue-specific functions and unique roles, even at steady-state. The detailed molecular characterisation of tissue Mϕs is important in order to fully understand the various roles they play in a given tissue. Important insights into Mϕ functions have been gained by system-wide analyses of transcriptional regulation and epigenetics 1,3,4 . However, much less is known about protein dynamics. The proteome can differ significantly from the transcriptome due to a variety of regulatory processes such as posttranscriptional processing and degradation of mRNA and protein modifications and degradation, all of which are important in controlling steady-state protein abundance. Recent technological advances in mass spectrometry (MS) permit the systematic quantification of thousands of proteins from cellular lysates 5 .
Proteomics data, when combined with complementary approaches, such as bioinformatics and pathway data analysis tools, provides the contextual information required to decode large data sets into meaningful biologically adaptive processes such as cellular functions and metabolic pathways.
Tissue Mϕs can be divided into two categories: those that are 'fixed' within the tissue by being firmly embedded in neighbouring stroma and those that are 'free' to migrate within tissues. To investigate tissue Mϕ proteomes of each category, we selected splenic red pulp Mϕs (SRPM) as an example of fixed Mϕs and peritoneal cavity Mϕs (PM) as an example of free Mϕs 6 . Both of these Mϕ populations are embryo-derived and have some well-known functions. SRPM are important for clearance of aged red blood cells, and iron homeostasis and recycling from senescent and damaged erythrocytes 2,7-9 . In contrast, PM play a role in B1 cell class switching and through their ability to rapidly migrate into visceral organs, they have been shown to play role in tissue repair and liver regeneration 10-15 .
To characterise and investigate other possible tissue-specific functions for SRPM and PM at steady-state, we used MS approaches to obtain accurate quantification of protein copy number per cell that was validated by flow cytometry. Comparisons of the proteomes with the corresponding transcriptomes were made using the ImmGen 16 gene expression database, which provided high-throughput validation for the translation of mRNA to protein for many of the differentially expressed genes and also revealed numerous examples of discordance between mRNA and protein levels. We have combined proteomics with complementary approaches, decoded large data sets and revealed novel functions relating to metabolism and host defence, in addition to previously characterised functions.
Animals C57BL/6J wildtype mice were used throughout. All 60 mice used in the study were maintained in the Biological Resource Unit at the University of Dundee under specific-pathogen-free conditions following procedures that were approved by the University of Dundee Ethical Committee and under the authorization of the UK Home Office Animals (Scientific Procedures) Act 1986. Mice were housed in same sex groups, in individually ventilated cages and were given standard diet RM3 (SDS Special Diet Services). Housing conditions were: 12-h light, 12-h dark cycle, 21°C temperature and relative humidity of 45-60%. Mice had free access to autoclaved drinking water and autoclaved food ad libitum. Mice were humanely killed using a Schedule 1 method of gradual exposure to carbon dioxide and death was confirmed by cervical dislocation.
SRPM isolation and purification SRPM were purified as described previously 17 . In brief, single cell suspensions of splenocytes were obtained by digesting the tissue with 400 U/ml collagenase D and 50 μg/ml DNAse 1 at 37°C for 20 minutes, erythrocytes were lysed and passed through a 100 μm cell strainer following suspension in MACS buffer. The single cell suspensions of splenocytes were then passed over LS separation columns (Miltenyi) without further treatment. Before being removed from the magnet, the columns were washed three times with 2 ml MACS buffer (Miltenyi). After removal of the magnet, SRPM were eluted from the separation column with 5 ml MACS buffer and subjected to fluorescence activated cell sorting.

PM Isolation and purification
The peritoneal cavity was flushed with 4-5 ml PBS + 1% FBS + 2 mM EDTA. After centrifugation, cells were resuspended in RPMI medium supplemented with 4 mM L-glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin and 10% heat inactivated FBS and subjected to fluorescence activated cell sorting.

Flow cytometry
Single-cell suspensions were Fc-receptor-blocked for 30 min at 4°C with rat anti-mouse CD16/CD32 antibody in PBS with 1% FBS. Blocked cells were subsequently incubated with fluorophore-conjugated surface-staining primary antibodies for 60 min at 4°C, prior to washing in PBS containing 1% FBS and 2 mM EDTA. Following surface staining, cells were washed and analysed using a Canto II flow cytometer and FlowJo software version 10.0. Raw data for flow cytometry analysis are available, see Underlying data 18 .
Fluorescence activated cell sorting Following isolation from the spleen and peritoneal cavity as described above, cells were stained for the Mϕ lineage marker, F4/80 and viability dye DAPI, under sterile conditions. Briefly, cells were pelleted and incubated sequentially with 1:100 Fc-receptor block for 10 min at 4°C, followed by Mϕ marker F4/80 antibody (1:200) and Siglec-F PE antibody (1:200) for 30 min at 4°C. Cells were washed and sorted into F4/80 APChigh-expressing single positive populations on a BD InfluxTM cell sorter into RPMI media containing 20% FBS. Sorted cells were pelleted and processed for proteomics.

Urea assay
Magnetically sorted PM were seeded at 0.25 × 10 6 cells per well (96-well) in HBSS with 0.5 μM retinoic acid 20 ng/ml M-CSF and incubated for 2 h. The medium was removed and replaced with HBSS with 0.1% BSA with or without 2 mM glutamine and glutamate (+ 0.5 μM retinoic acid, + 20 ng/ml M-CSF), with or without 25 mM glucose. For urea measurements, the supernatant was transferred to fresh 96-well plate after overnight incubation and urea levels were measured using a QuantiChromTM Urea Assay Kit. Raw data for urea quantification are available, see Underlying data 18 .
Filter-aided sample preparation (FASP) of samples A total of n=4 biological replicates for PM and n=3 biological replicates for SRPM were processed for proteomic analysis. Cell pellets were solubilised in 100 mM Tris HCl containing 4% SDS, and 100 mM dithiothreitol in the presence of protease and phosphatase inhibitors. Samples were processed using the FASP protocol 19 with some modifications. After removal of SDS with 8 M urea and alkylation with iodoacetic acid, Vivacon 500, 10k MWCO HY filters (Sartorius Stedim Biotech, VN01H22) were washed three times with 100 mM Tris-HCl pH 8 then another four times with 100 mM triethyl ammonium bicarbonate (TEAB). Proteins on the filters were then digested twice at 30°C with trypsin (Pierce) (2 x 2 μg), first overnight and then for another 6 h in a final volume of 200 μl, 100 mM TEAB. The resulting tryptic peptides were desalted using C18 solid phase extraction cartridges (Empore, Agilent technologies) and quantified using the Pierce Quantitative Colorimetric Peptide Assay (Thermo Scientific). Raw data for peptide quantification are available, see Underlying data 18 .
High pH reverse phase fractionation Peptides (8 μg for each sample) were dissolved in 200 μL ammonium formate (10 mM, pH 9.5) and fractionated using RP-High pH HPLC. A C18 Column from Waters (XBridge peptide BEH, 130Å, 3.5 μm 2.1 X 150 mm, Ireland) with a guard column (XBridge, C18, 3.5 μm, 2.1X10mm, Waters) are used on a Ultimate 3000 HPLC (Thermo-Scientific). Buffers A and B used for fractionation consist, respectively, of 10 mM ammonium formate in milliQ water and 10 mM ammonium formate with 90% acetonitrile. Both buffers were adjusted to pH 9.5 with ammonia. Fractions were collected using a WPS-3000FC auto-sampler (Thermo-Scientific) at 1 min intervals. Column and guard column were equilibrated with 2% buffer B for 20 min at a constant flow rate of 0.2 ml/min. Samples (180 μl) were loaded onto the column at 0.2 ml/min, and separation gradient start 1 min after sample loading onto the column. Peptides were eluted from the column with a gradient of 2% to 5% buffer B over 6 min and then from 5% to 60% buffer B over 53 min. The column was washed for 16 min in 100% buffer B and equilibrated in 2% buffer B for 20 min as mentioned above. Fraction collection began 1 min after injection and stopped after 80 min (total of 80 fractions, 200 μl each). The total number of fractions concatenated was set to 10 and the content of the fractions was dried and suspended in 50 μl 1% formic acid prior to analysis with LC-MS.

LC-MS analysis
Analysis of peptides was performed on a Velos-Pro (Thermo Scientific) mass spectrometer coupled with a Dionex Ultimate 3000 RS (Thermo Scientific). LC buffers were the following: buffer A (2% acetonitrile and 0.1% formic acid in Milli-Q water (v/v)) and buffer B (80% acetonitrile and 0.08% formic acid in Milli-Q water (v/v). Aliquots of 15 μL of each sample were loaded at 5 μL/min onto a trap column (100 μm × 2 cm, PepMap nanoViper C18 column, 5 μm, 100 Å, Thermo Scientific) equilibrated in 98% buffer A. The trap column was washed for 6 min at the same flow rate and then the trap column was switched in-line with a Thermo Scientific, resolving C18 column (75 μm × 50 cm, PepMap RSLC C18 column, 2 μm, 100 Å). The peptides were eluted from the column at a constant flow rate of 300 nl/min with a linear gradient from 5% buffer B to 35% buffer B in 124 min, and then to 98% buffer B by 126 min. The column was then washed with 98% buffer B for 20 min and re-equilibrated in 98% buffer A for 17 min. Fractions (10) from each sample were analysed using LTQ-Orbitap Velos Pro in data dependent mode. A scan cycle comprised MS1 scan (m/z range from 335-1800) in the Velos orbitrap followed by 15 sequential dependent MS2 scans (the threshold value was set at 5000 and the minimum injection time was set at 200 ms) in LTQ with collision induced dissociation (CID). The resolution of the Orbitrap Velos was set at to 60,000. To ensure mass accuracy, the mass spectrometer was calibrated on the first day that the runs were started.

Proteomics quantification and bioinformatics analysis
A total of n=4 biological replicates for PM and n=3 biological replicates for splenic red pulp Mϕs were subjected to proteomic analysis. The raw mass spectrometric data were processed, searched and quantified using MaxQuant 1.5.5.1 20 and Andromeda as the search engine software 21 . Enzyme specificity was set to that of trypsin/P, allowing for cleavage of N-terminal to proline residues and between aspartic acid and proline residues. Other parameters used were: variable modifications-methionine oxidation, protein N-acetylation; fixed modifications, cysteine carbamidomethylation; database: Uniprot -mouse; minimum peptide length, 7; maximum missed cleavages, 2; PSM & protein false discovery rate, 1%. Default MaxQuant settings included a mass tolerance of 7 ppm for precursor ions, and a tolerance of 0.5 Da for fragment ions. A reverse database was used to apply a maximum false positive rate of 1% for both peptide and protein identification. This cut-off was applied to individual spectra and whole proteins in the MaxQuant output. The match between runs feature was activated with default settings. For bioinformatic and statistical analysis, Perseus software (version 1.5.1.6) was used, all 'Contaminant', 'Reverse' & 'Only identified by site' proteins were removed from the data and the proteomic ruler 5 was used to calculate copy number per cell. Proteins above 2-fold change [log2(2)=1], proteins with nominal p-value less than 0.05 [-log10(0.05)=1.301] were considered as differentially expressed proteins. All bioinformatics, statistical and pathway analyses were performed with the Perseus software 20-22 and Excel was used for basic data analysis and plots. The proteome and the ImmGen mRNA data were merged using R studio based on GeneID. Proteomics data are available via ProteomeXchange with identifier PXD019362.

Statistical analysis
Statistical significance was determined using the two-tailed Student's t-test. All experiments were performed at least twice. P values of <0.05 were considered significant. GraphPad prism version 6 or Perseus version 1.5.1.6 were used.

Results and Discussion
Label-free quantitative proteomics experiments were performed to identify differentially regulated proteins in SRPM and PM that might give insights into their tissue-specific functions under steady-state conditions. Following tissue isolation and cell sorting by flow cytometry, Mϕ populations were around 95% pure ( Figure 1A). Database searches enabled the identification of more than 7000 protein groups in these cells (Extended data, Table 1) 18 . Label-free quantitative intensity of the proteins obtained using MaxQuant software followed a normal distribution for both the SRPM and PM populations 23 ( Figure 1C). The intensities were transformed into absolute quantification of copy number per cell, using the proteomic ruler methodology 5 . Copy numbers for proteins from the 3 (SRPM) or 4 (PM) biological replicates showed strong Pearson correlation coefficients. Variability within replicates for a single population was from 0.968 + 0.0005 for SRPM to 0.985 + 0.002 for PM, indicating robustness and reproducibility of our MS-based peptide quantification methods ( Figure 1D). Principal component analysis (PCA) of all proteins expressed by the two sorted populations revealed a relatively greater distance between the SRPM and PM ( Figure 1E). Volcano plots showing the distribution of fold-change versus p-values were generated by plotting the t-test p-value against the PM:SRPM ratio of copy number per cell on a logarithmic scale.
We observed a large number of protein IDs that are differentially regulated between the two populations of Mϕs. However, for many proteins, the fold increase obtained using copy number per cell correlated with the increase in cell size/ volume determined by forward scatter properties in flow cytometry ( Figure 1B, F, upper panel). It has previously been noted that PM are larger in size and volume than the SRPM 24 . To avoid the bias that enriched proteins in the PM are due to this increased cell size, in addition to evaluating differentially regulated proteins using absolute copy number, we also evaluated them following normalisation, centring the median across samples ( Figure 1F, lower panel) 25 . We used a fold-change of ≥2 and p values of ≤0.05 to define differential regulation across biological replicates. Several proteins were differentially expressed by at least 2-fold when SRPM were compared to PM. Overall, these comparisons indicate minimum variability within replicates for a single population and a noticeable diversity between PM and SRPM populations.
Correlation analysis of mRNA and protein abundance Integrating multiple levels of information is necessary and valuable to fully elucidate complex biological systems 26-28 . In line with this, we integrated our proteomic data sets with the corresponding mRNA data sets from the ImmGen database (http://rstats.immgen.org/DataPage, GSE15907, Extended data, Table 2) 18 . This allowed us to gain global understanding of mRNA and protein expression relationships and to identify differentially regulated proteins that are not evident at the transcriptomic level. Further, this integration of heterogeneous data sets, taken from different laboratories and across experimental platforms, has provided high-throughput data validation and effective use of the data to gain insights into complex biological functions of tissue Mϕ populations.
Out of 7533 unique proteins from our proteome data set, we identified 6903 corresponding transcripts from the Imm-Gen database and used these for correlation analyses (Extended data, Table 3) 18 . The results show moderate positive correlations between mRNA and protein abundance for both Mϕ populations (Figure 2A   Scatter plots of fold changes obtained using mRNA data vs fold changes obtained using proteome data. The coloured dots indicate differentially expressed genes that were statistically significant in the mRNA data set (E), protein data set (F), or both (G). The proteomics data were derived from 4 (PM) or 3 (SRPM) biological replicates. The mRNA data were derived from 3 biological replicates for each Mϕ population.
correlations were observed when PM mRNA was compared with SRPM mRNA or when the PM proteome was correlated with the SRPM proteome ( Figure 2, D) 29 . This can be explained in part by fundamental limitations and differences in transcript and protein behaviour, with mRNA half-life ranging from 0.1 to ~40 h versus protein half-life 0.5 to >500 h 30,31 . On the one hand, when a target protein reaches saturating levels, gene silencing or transcript degradation pathways may predominate, leading to an inverse or moderate correlation. On the other hand, when a target protein is in demand in the cell, protein synthesis and mRNA expression may display a strong correlation.
Next, we asked whether transcripts that showed tissuespecific enrichment, based on ImmGen microarray analysis, were similarly enriched at the protein level. The mRNA data set has 1537 unique transcripts that are differentially and significantly regulated ( Figure 2E), compared to 3697 unique proteins in the proteomic dataset ( Figure 2F). Correlation analysis showed that 930 distinct transcripts (that have tissue-specific enrichment) were positively correlated with their corresponding proteins from the proteome data set ( Figure 2G and Extended data, Table 4) 18 . However, we noticed that more than 35% of the differentially regulated transcripts were not correlated. This may be due to fundamental biological differences in transcript and protein behaviour or regulatory programs as discussed above, as well as variability associated with the global measurement of transcripts and proteins, such as differences in sensitivity, dynamic range, with the proteome being more dynamic (Figure 2A, B: X axis, proteome; Y axis, mRNA). In conclusion, our correlation of mRNA and protein abundance provides cross-platform/laboratory comparisons and high throughput validation for the translation of mRNA to protein for many of the differentially expressed genes. This correlation analysis also identified numerous examples of mismatches between mRNA and protein expression. The proteomes have thus provided many additional novel insights concerning the biology of these Mϕ populations not observed at the mRNA level.
Characterization and validation of common and tissue specific membrane proteins: novel insights not evident from the transcriptome Cell surface proteins play a critical role in sensing and responding to the tissue microenvironment and are commonly expressed differentially amongst leukocyte subsets. Proteomic data showed that the cell adhesion molecules (CAMs) VCAM1, F11R (JAM1), CADM1 and PECAM1 were expressed exclusively in SRPM, whereas ICAM2, NINJ1, BCAM, CEACAM1 and ALCAM were expressed specifically by PM, with the first two being the major adhesion receptors ( Figure 3A). Adhesion receptor ICAM1 was expressed at similar levels in both the Mϕ populations.
Integrins are a large family of heterodimeric adhesion receptors that contain an αand a βsubunit. The β2 integrins exhibited striking differential expression, with CD11b (ITGAM α-chain; ITGB2 β-chain) being dominant on PM, but CD11d (ITGAD α-chain; ITGB2 β-chain) being exclusively expressed by SRPM ( Figure 3B). PM expressed higher levels of ITGA6 (α6-chain) and ITGA4 (α4-chain) which pair with β1 and either β1 or β7, respectively. ITGAV (αV-chain), which can pair with several β chains, was expressed at similar levels on PM and SRPM. The quantitative proteomics further showed that the integrin αand β-chains were in 1:1 stoichiometry in each case (interconnecting arrows, Figure 3B).
Many of the adhesion molecules identified by proteomics were validated by flow cytometry, showing an excellent correlation between proteomics data and transcriptome data ( Figure 3C). The observed tissue specific expression of CAMs and integrins suggest that many of these adhesion molecules have roles in tissue specific interactions 32 .
Next we assessed Mϕ expression of Toll-like receptors (TLRs), which play a key role in the innate immune system. The proteomic analysis showed enrichment of TLR2 and TLR13 in PM and TLR9 in SRPM. Other TLRs were expressed at similar levels in both tissue Mϕs ( Figure 3A and Figure 4A, B). The proteomics data were in disagreement with the transcriptome data for TLR2 expression ( Figure 4A), but its preferential expression in PM shown by proteomics was confirmed by flow cytometry ( Figure 4B). Low expression of TLR4 in both Mϕ populations seen in the proteomics data was also confirmed by flow cytometry (Figure 4A, B).
Similarly, proteomic analysis showed clear enrichment of several Mϕ growth factor receptors (M-CSF-R (CSF1R) and GM-CSF-R (CSF2RA, B) in PM, whereas the transcript levels were similar for both PM and SRPM ( Figure 4A). Expression levels of CSF1R, determined by flow cytometry, were in good agreement with the proteomic data ( Figure 4A, B).
Similar observations were made with a number of other cell surface receptors ( Figure 4C, D). While the proteomics and transcriptome data were consistent for expression levels of TIMD4, EMR1, CD163, CD206 and SIGLEC12, no correlation was seen for CD200R1, CD244, CD80, CD44, SIRP-A, CD86 and MS4A1. As above, the flow cytometry analysis was consistently in agreement with the proteomics data ( Figure 4D).   Another key group of membrane proteins are the Slc and ABC families of transporters. These are essential for the import and export of many small molecules that are vital for regulation of metabolic processes and immune responses. Both PM and SRPM express a large number of these transporters in common, as shown in Figure 8. However, some are differentially expressed and enriched above log 2 3-fold. For PM, these include the ABC family transporter ABCA9 and Slc family transporters SLC2A6, SLC9A3R2, SLC25A15, SLC15A2, SLC16A3, SLC31A1, SLC25A1, SLC25A13 and SLC25A15. For SRPM, these include the ABC family transporter ABCG3 and Slc family Figure 5. Characterization of CD markers and membrane receptors that are common and tissue specific. Graphs show the estimated protein copy numbers per cell using the proteomic ruler approach for proteome data set (upper panels), log fold protein increase in PM versus SRPM (middle panels) and mean fluorescence of mRNA data set obtained from ImmGen (lower panels).
These findings indicate that, while the membrane proteins of PM and SRPM overlap considerably, a large set of membrane proteins is differentially expressed by these Mϕs which are likely to contribute to their tissue-specific functions, as discussed in the next section.
Proteins important in known tissue-specific functions of peritoneal cavity and splenic red pulp macrophages Although PM and SRPM have similar embryonic origins 33 , they undergo tissue-specific differentiation programmes and mediate distinct functions within each microenvironment. This is clearly reflected in the proteomic data, as discussed below.
For PM, the transcription factor GATA6 was shown to be key regulator that defines a tissue-specific gene expression programme 34 . We observed enrichment of GATA6 as well as the reported GATA6-dependent PM-specific proteins such as SERPINB2, SELP (CD62P), THBS1, NT5E (CD73), TGFB2 and LTBP1 in our PM proteome ( Figure 9A).
Vitamin A is required for induction of GATA6-dependent andindependent PM-specific gene expression programmes 34 . For vitamin A (retinol) to be transcriptionally active, it needs to be oxidized to the biologically active metabolite, retinoic acid (RA) 35-37 . Our proteomic analysis showed that RALDH2, the rate-limiting enzyme in RA synthesis 35,38 , is highly expressed in PM. This suggests that PM may be capable of converting vitamin A to RA and that the omentum may not be required for this step as proposed by Okabe (Okabe 2014). Further, RA exerts its biological effects, primarily, through nuclear receptors RARs/RXRs and the non-nuclear receptor M6P/IGF2R 39-43 . In our proteome, we observed enhanced expression of RA receptors such as RXRB, RXRA, RARG and IGF2R in PM compared to SRPM ( Figure 9A). Figure 6. Characterization of membrane proteins that are common and tissue specific. Graphs show the estimated protein copy numbers per cell using the proteomic ruler approach for proteome data set (upper panels), log fold protein increase in PM versus SRPM (middle panels) and mean fluorescence of mRNA data set obtained from ImmGen (lower panels).
Furthermore, in the PM proteome we observed enrichment of proteins such as CXCL13, TGFB2 and CD44 that have been implicated in PM function ( Figure 9A). CXCL13, is essential for B-1 cell (major peritoneal cavity B cell population) migration to the peritoneal cavity 44 . TGF-β generated by PM has been implicated in B-1 cell class switching and the generation of IgA 45 . In addition to enrichment of TGF-β, we also observed enhanced expression of TGF-β-associated proteins such as TGFBRAP1, TGFBI in the PM proteome ( Figure 9A). CD44 has been shown to be important for the movement of PM to the liver in a sterile liver injury model to facilitate tissue repair by phagocytosing damaged tissue 14 .
For SRPM, SPI-C is a key transcription factor required for their development following its induction by Mϕ exposure to heme 46 ( Figure 9B). Major functions of SRPM are in capture and degradation of aged erythrocytes, repair of damaged or infected erythrocytes and iron homeostasis. Our proteomics data show enrichment of TAM receptors AXL, MERTK and STAB2 on SRPM ( Figure 9B), which are likely to be important for recognition of aged or damaged erythrocytes via exposed phosphatidylserine, leading to erythro-phagocytosis 47 . We observed enrichment of heme oxygenase 1 (HOM-1) and ferroportin (SLC40A1) in SRPM ( Figure 9B) which are required for heme metabolism and the release of free iron 7 . We also observed enrichment of CD163, which is a high affinity scavenger receptor for hemoglobin-haptoglobin complexes 48 ( Figure 9B).
Splenic hematopoietic stem and progenitor cells (HSPCs) localize in the red pulp near the sinusoids in parafollicular areas 49,50 . VCAM1 is important for retaining HSPCs and in splenic myelopoiesis and was also enriched in the proteome of SRPM ( Figure 9A).
Insights into metabolic functions of peritoneal cavity macrophages An important conceptual advance in the field of Mϕ biology, emerging from recent studies, is that Mϕ polarisation is critically supported by metabolic shifts 51 . Therefore, in our proteomes we asked if any key metabolic enzymes are differentially regulated which could give insights into tissue-specific functions ( Figure 10). PM were enriched with enzymes related to glycolyticcitrate-lipogenic switch and generation of lipid secondary messengers. We observed enhanced expression of glucose transporter SLC2A6 (GLUT6/9) and key glycolytic enzymes ( Figure 10A, B). These include HK1 (hexokinase-1), which traps glucose in the cell by converting it to glucose-6-phosphate; PFKL (ATP-dependent 6-phosphofructokinase liver type), the rate-limiting enzyme mediating the third step in glycolysis; KHK (ketohexokinase; fructokinase), the key enzyme for dietary fructose catabolism that bypasses the major glycolytic checkpoint (PFK) and converges into the glycolytic pathway. The enrichment of key glycolytic enzymes in PM suggests that these cells exhibit high rates of glycolysis ( Figure 10A, B).
Studies by others showed increased extracellular acidification rates (ECAR) in PM, which are indicators of glycolytic activity 52,53 . The enhanced expression of LDH (lactate dehydrogenase) and SLC16A3/MCT4 ( Figure 8A), the lactate transporter 54 , further supports the observed increase in ECAR and channelling of glycolysis towards lactate production in PM. Lactate at physiologic concentrations has been shown to play a role in visceral organ homeostasis, anti-inflammatory responses, wound healing and also acts as fuel donor through generation of pyruvate (via LDH) 55-60 .
Furthermore, we noted that PM express high levels of ACLY (ATP citrate lyase), the enzyme responsible for generating cytosolic acetyl-CoA from citrate of the TCA cycle ( Figure 10A, B). This suggests that pyruvate generated through glycolysis may be directed to alternate pathways other than TCA/OXPHOS, where citrate production is an end-product of glucose utilisation 61,62 . Supporting this hypothesis, previous studies with PM noted a marked difference in rates of CO 2 production from the 1 st and 3 rd carbon of pyruvate. It was noted that almost 60% of the [1-14 C] pyruvate that was utilized appeared as 14 CO 2 , whereas for the [3-14 C] pyruvate only about 15% of that utilized appeared as 14 CO 2 . This suggests a fate for acetyl-CoA produced from pyruvate other than OXPHOS 52,53,63 . Although these studies on CO 2 production were performed using thioglycollate-elicited PM, the ImmGen database showed high levels of ACLY (ATP citrate lyase) transcripts in both resident and thioglycollate-elicited PM.
ACLY is a key enzyme of de novo FA synthesis, as it generates cytosolic acetyl-CoA, an essential biosynthetic precursor for FA synthesis. In addition, the enhanced ASPA (aspartoacylase) levels in PM ( Figure 10A, B) and availability of its substrate N-acetylaspartate in peritoneal fluid 52,64 could provide an additional source of acetyl donor for FA synthesis (Francis JS, 2016). Consistent with this idea, we observed enrichment of the rate-limiting enzyme of FA synthesis, ACACA (Acetyl-CoA Carboxylase Alpha), which converts cytosolic acetyl-CoA to malonyl CoA ( Figure 10A and B). This pathway is important for M2 Mϕ development, as an inhibitor of acetyl-CoA carboxylase (ACACA) was shown to inhibit IL-4 induced M2 activation 65 . We also observed enrichment of ELOVL5 (fatty acid elongase 5), which catalyzes the rate-limiting reaction of the long-chain fatty acid (FA) elongation cycle and is vital in generation of mono/poly unsaturated FAs. Thus, the PM proteomics data show enrichment of key enzymes related to the glycolytic-citrate-lipogenic pathway. This is critical in many cells, including tumour cells, to fulfil bioenergetic and synthetic requirements 66 . In addition, we observed increased expression of FA-binding proteins FABP3, FABP4 and FABP7 which act as cytoplasmic lipid chaperones and are important in mediating metabolic balance and coordinating FA/lipid trafficking and lipid-mediated signalling in cells ( Figure 10A and B).  FAs are essential for many aspects of cellular function. They not only act as a storage form of fuel, they are also important components of the cell membrane. When partially cleaved from the cell membrane, they form second messengers or local hormones such as arachidonic acid (AA) metabolites. These can mediate signalling in an autocrine or paracrine manner, thereby regulating inflammatory and immune responses and other vital cellular functions. Given that we observed enrichment in the PM proteome of key enzymes of FA synthesis (in particular Elovl5 that is required for the synthesis of monounsaturated and polyunsaturated FAs) as well as enrichment of many enzymes of the AA metabolic pathway (ALOX15, LTC4S, ALOX5, PTGR1, PTGES), this strongly suggests a role of FA synthesis in lipid secondary mediator generation ( Figure 10A, B). Consistent with this idea, previous studies have observed that liver tissues of Elovl5-deficient mice had reduced levels of AA, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which are precursors of the AA pathway mediators and substrates for ALOX15, LTC4S, ALOX5, PTGES 67,68 . Amongst these, ALOX15 (15-lipo-oxygenase) is the most abundant and is well documented for its role in resolution of inflammation via the generation of lipid mediators such as resolvins, protectins and maresins 14,69-71 . Given the fact that PM can be recruited to visceral organs and contribute to tissue repair 10-15 , ALOX15 expression by PM may be important for resolution of inflammation, clearance of apoptotic cells and maintenance of tissue homeostasis, similar to findings made with Ly6C lo , CX3CR1 hi Mϕs of liver and adipose tissue 69,71,72 .
PM were enriched with enzymes related to lipoprotein uptake and cholesterol metabolism. Besides de novo synthesis, FAs can be acquired from extracellular sources via receptor-mediated uptake. The PM proteome exhibited increased levels of LDL/VLDL endocytic receptors, including CD36, MSR1, STAB1, LDLR, LSR. PM also expressed the lipoprotein lipase, LPL, which catalyses hydrolysis of triglycerides from LDL and VLDL ( Figure 10A and C). Given that PM are relatively large and mobile cells, they may require contributions from both exogenous triacylglycerols and de novo synthesized FAs to meet increasing bioenergetics and demand for biosynthetic and lipid secondary metabolites, similar to that of M2 Mϕs 65 .
We also observed enrichment of a number of proteins involved in regulation of cholesterol levels ( Figure 10A Interestingly, we observed that PM have elevated levels of enzymes related to nitrogen metabolism, detoxification, and immunoprotective function. Enzymes involved in catabolism of nitrogen-containing metabolites (amino acids, nucleotides and monoamines) and removal of toxic ammonia via the urea cycle were enriched in PM. These include ARG1 (arginase-1), GLS (glutaminase), HAL (histidase), MAOA (monoamine oxidase A), PADI4 (peptidylarginine deiminase type 4), GDA (guanine deaminase), AMPD2 (AMP deaminase 2), XDH (xanthine dehydrogenase) and UOX (urate oxidase) ( Figure 10A, D). Among these, arginase-1 is the most studied enzyme in Mϕs. In PM it has been shown to play a role in the generation of urea 78 and in addition, we find that PM are able to utilise glutamine and glutamate which are rich in the peritoneum for urea generation ( Figure 10E). This further supports the observation that fuel derived from glutaminolysis is exploited by PM to maintain the respiratory burst as suggested recently 52 . Arginase-1 is a key signature of alternatively activated Mϕs and is involved in a variety of immunomodulatory functions. By limiting the supply of L-arginine for NO synthase and by inhibiting inducible NO synthase activity via the generation of urea, arginase-1 mediates an anti-inflammatory function 78-80 ( Figure 10D). Arginase-1 can also act extrinsically to suppress Th2 cytokine-driven inflammation via nutrient deprivation of L-arginine, which is required to sustain T cell responses 78 . Arginase-1 is also involved in the regulation of wound healing via synthesis of ornithine-proline-collagen 79,80 .
Further, the proteomics data showed that XDH was strongly elevated in PM ( Figure 10A, D). XDH, in addition to its role in purine catabolism (conversion of hypoxanthine/xanthine to uric acid), has been shown to play a significant role as a xenobiotic-detoxifier and also as a drug-metabolizing enzyme due to its relaxed specificity for substrates 81 . PM also expressed high levels of UGT1A7C (UDP-glucuronosyltransferase 1-7C) that catalyses glucuronidation, a major pathway of xenobiotic biotransformation in most mammalian species, that detoxifies many divergent chemical classes, rendering them harmless, more water-soluble, and hence excretable ( Figure 10A). The proteomics therefore shows that PM are enriched with many proteins that play a significant role in nitrogen homeostasis and in detoxification of various endogenous and exogenous compounds.

Insights into immune functions of peritoneal cavity macrophages
Besides playing an important role in homeostasis, proteomics confirmed that PM are also well-equipped to perform important functions in immune defence. They expressed high levels of complement receptors (C5AR1, C3AR1, CD93; C1QR, CD209B; SIGN-R1) 82 and complement factors (C1R, CR1L, C1QB, C1QC, C1S, C4B, CFD, CFB, CFH) which provide a critical first-line defence against infection ( Figure 11A). We also observed enrichment of various coagulation factors such as F5, F13A1, F10, F7, FN1, KNG1, PROS1, PLAUR, PLAU ( Figure 11A) 83-85 . In particular, coagulation factor XIII (F13), which is highly enriched in PM, has been shown to take part in the early innate immune defence and significantly influences wound healing 85 . It limits bacterial dissemination and facilitates bacterial killing through their entrapment within fibrin networks 84 . In addition, a number of chemokines and immunomodulatory factors, as shown in Figure 11A, were enriched in PM that could play an important role in PM-mediated defence functions.

Insights into immune functions of splenic red pulp macrophages
For SRPM, in addition to enrichment of proteins involved in regulation of erythrocyte homeostasis, we observed enhanced expression of many proteins related to immune defence. Here we discuss some of the proteins that potentially play important functions in splenic tissue.
SRPM were enriched with proteins related to lipid and peptide antigen presentation. Antigen presentation via major histocompatibility complex class II molecules is highly critical in adaptive immunity. The processing and presentation of foreign proteins into small peptide fragments bound to class II molecules is a dynamic process consisting of multiple steps and chaperones. Several of the molecules involved in this process were enriched in SRPM compared to PM. We found enrichment of H2-A, H2-E (MHC-II) the class II antigen presenting molecules; H2-G (invariant chain (li/(CLIP)), the chaperone important in stabilising the peptide-binding groove of MHC-II/H2-DM that facilitates the exchange of the class II-associated invariantchain peptide (CLIP) with peptides derived from exogenous antigens; CSTB and CSTL cathepsins, antigen-processing enzymes; the NFYB transcription factor that regulates MHC expression ( Figure 11B). Flow cytometry analysis confirmed higher expression of MHC class II (H-2A and H-2E subclasses) molecules in SRPM compared to PM ( Figure 11B).
We also observed enhanced expression of the non-classical class 1b histocompatibility antigen H2-M3 ( Figure 11B). H2-M3 specifically bind to N-formylated peptides and is important in presenting Listeria monocytogenes and Salmonella typhimurium antigens to CTL 86,87 .
Apart from factors involved in peptide antigen presentation, we also found preferential expression of lipid antigen presenting molecule, CD1D and PLA2G15 (lysosomal phospholipase A2) which processes lipid antigens for presentation by CD1d 88,89 ( Figure 11B).
Thus, we observed enhanced expression of both lipid and foreign peptide presenting MHC-II molecules and their regulators in SRPM. In comparison, classical MHC class I molecules which present mainly endogenous peptides recognised by CD8+ (cytotoxic) T cells, were present at similar levels in the proteomes of SRPM and PM. In conclusion, the proteomics data suggest that SRPM may play a previously unrecognised role in regulation of adaptive immune responses, in addition to their homeostatic role in heme recycling and erythrocyte homeostasis.
SRPM were enriched in a number of defence proteins such as CTS (cathepsins), NGP (neutrophilic granule protein), MPO (myeloperoxidase), EPX (eosinophil peroxidase), EAR (eosinophil cationic protein), GZMA (granzyme A), ELANE (neutrophil elastase) and LTF (lactotransferrin) ( Figure 12A). Many of these lytic mediators are associated with granulocytes but are also reported to be expressed by certain tissue Mϕs for pathogen defence 90-92 . Since SRPM have been shown to mediate efferocytosis of granulocytes in order to regulate steadystate granulopoiesis and HSPC trafficking 93-96 , it is possible they acquired these factors through phagocytosis of apoptotic granulocytes 97,98 . This would also be consistent with low or undetectable mRNA encoding some of these proteins (Figure 12). However, other granulocyte markers like Gr-1(Ly6G/Ly6C) (neutrophils) and Siglec-F (eosinophils) were not detected (Extended data, Table 1) 18 . Finally, a number of immunomodulatory factors shown in Figure 12A, such as TGF-B1, IL1B, IL18, NOS1, IGF2, PPBP, CMC4 and serpin proteases, were enriched in SRPM. These could be important in host defence and clearance of foreign particles.
Insights into metabolic functions of splenic red pulp macrophages SRPM were enriched with proteins related to glycan and lipid degradation enzymes. A major function of SRPM is the uptake of aged erythrocytes. This provides a significant phagocytic burden that has to be rapidly catabolised. We observed enhanced expression of endo-lysosomal glycan degradation enzymes such as ARSB (arylsulfatase B), IDS (iduronate 2-sulfatase), MANBA (beta-mannosidase), NPL (sialate lyase) ( Figure 12B). Npl is a sialic acid degradation enzyme that prevents sialic acids from being recycled and returning to the cell surface. Given that sialic acid is an abundant sugar on erythrocytes, Npl and other glycan degradation enzymes are likely to be important for efficient catabolism of glycans on efferocytosed erythrocytes. We also noted enriched levels of endo-lysosomal lipases/exonucleases ( Figure 12B) known to be involved in the clearance of efferocytosed particles 88,89,99-101 . These include PLBD1 (phospholipase B-like 1), ABHD6 (monoacylglycerol lipase), PLD3 (phospholipase D3) with both lipase and 5' exonucleases activities and PLA2G15 (lysosomal phospholipase A2) which plays a role in the processing and presentation of lipid antigens to CD1d. We also observed enrichment of LAP3 (cytosol aminopeptidase), ACP5 (acid phosphatase 5, lysosomal), a tartrate-resistant acid phosphatase, which catalyses the formation of ROS and participates in degradation of endocytosed bone matrix and in bacterial killing 102 . Key enzymes of beta oxidation were also enriched in SRPM such as ACSF2 (Acyl-CoA synthetase which is required to activate FA to enter beta oxidation) and ACADSB (mitochondrial acyl-CoA dehydrogenase) that initiates the first step of FA oxidation. The enhanced expression of this degradation machinery for glycans, proteins and lipids is likely to be important in the effective clearance of efferocytosed cells and foreign material by SRPM.
We identified several enzymes of lipid metabolism expressed preferentially in SRPM. These include phospholipases PLBD1, PLD3, HPGDS (hematopoietic PGD synthase) and HPGD (hydroxyprostaglandin dehydrogenase) ( Figure 12B). HPGDS is the key enzyme for production of the D and J series of prostanoids in immune cells. PGD2 plays pro-or anti-inflammatory roles, where the outcome depends on the inflammatory milieu [103][104][105][106][107] . Though HPGDS and HPGD were found to be highly expressed in spleen, their tissue-specific functional significance is not well understood [108][109][110][111] . Also of note is the enrichment of GGT (gamma-glutamyltransferase), a membrane-bound enzyme involved in the metabolism of glutathione (gamma-glutamylcysteinyl-glycine; GSH) which can convert leukotriene C4 to leukotriene D4.
Other metabolic pathway enzymes that were enriched in SRPM with a potential role in immunomodulatory function are shown in Figure 12B with their absolute copy numbers, foldchange and mRNA levels. ADA (adenosine deaminase) was first identified in human spleen and is the predominant enzyme that converts the toxic DNA degradation product, deoxyadenosine to the non-toxic deoxyinosine. Elevated extracellular ADA activity has been observed in Mϕ-rich tissues, such as liver and spleen and shown to be important for acute and protracted inflammatory responses 112,113 . TGM1, TGM2 (transglutaminase) is a multi-functional enzyme that catalyses formation of isopeptide bonds between lysine and glutamine and produces protein-protein cross-links. In the context of immune defence, transglutaminases are required for migration of Mϕs towards apoptotic cells and for phagocytosis of apoptotic bodies. They also prevent the release of the pro-inflammatory cell content by crosslinking proteins of apoptotic cells and play both antiinflammatory and inflammatory roles 114-116 . CA1 and CA2 (carbonic anhydrase 1, 2) are important enzymes involved in acid production. In osteoclasts, CA2 has been shown to be critically involved in demineralization of ectopic calcification, but with respect to SRPM, its function is unknown. Given that erythrocytes also express CA1 and CA2 and that we detected haemoglobin subunits in our proteome ( Figure 12B and Extended data, Table 1) 18 , it is possible that SRPM obtained some erythrocyte proteins via phagocytosis of aged erythrocytes 117-119 . This would also be consistent with low or undetectable mRNA encoding these proteins ( Figure 12). However, retention of erythrocyte proteins appears to be selective, as other erythrocyte markers like glycophorins, erythrocyte membrane skeleton band protein band EPB4.2 and hemoglobin subunit-D 120 were low or undetectable (Extended data, Table 1) 18 .
We observed enrichment in SRPM of non-canonical glycolytic pathway enzymes ADPGK (ADP-dependent glucokinase), GPD1 (glycerol-3-phosphate dehydrogenase), SLC2A3 (Glut3glucose transporter) and SLC37A2 (G6PT) ( Figure 12B). ADPGK catalyses noncanonical phosphorylation of glucose to glucose 6-phosphate using ADP as a phosphate donor and is highly expressed in human hematopoietic cells 121-123 . Studies with T-cells have shown that ADPGK activity resulted in a deviation of glycolysis towards the glycerol-3-phosphate dehydrogenase (GPD) shuttle, leading to increased mitochondrial ROS production. Given that we observed enrichment of, ADPGK, GPD1 and SLC2A3/GLUT3, a similar mechanism may be operational in SRPM 124 and contribute to their host defence functions. Furthermore, G6PT functions in glucose recycling between the ER and the cytoplasm and its deficiency leads to defective myeloid phenotypes, displaying impaired glucose homeostasis, neutropenia and neutrophil dysfunction 125-128 . Thus, we observed a number of proteins enriched in the SRPM proteome that could play roles in acute inflammation, pathogen defence and clearance ( Figure 12B).
Creatine synthesis requires three amino acids, methionine, glycine, and arginine, and two enzymes, GATM (glycine amidinotransferase) and GAMT (guanidinoacetate N-methyltransferase). GATM is the rate limiting step in creatine synthesis, that catalyses the transfer of an amidino group from L-arginine to glycine to generate guanidinoacetate (GAA) and GAMT methylates guanidinoacetate (the immediate precursor of creatine) to produce creatine.
We observed elevated levels of GATM in the SRPM proteome and GAMT in the PM proteome (green arrowhead in Figure 10A, Figure 12B and 12C). Previous studies have shown the requirement of two organs for completion of creatine synthesis 129-131 . In the rat, high levels of GATM are found in the kidney, whereas GAMT is present in the liver. Rat hepatocytes were found to readily convert GAA to creatine; the same hepatocytes were unable to produce creatine from methionine, arginine, and glycine. 15 N from 15 NH 4 Cl is readily incorporated into urea but not into creatine, suggesting the liver mainly utilises nitrogen in arginine for urea generation via the urea cycle and not for creatine synthesis 130,131 . Similar to hepatocytes, we found that PM have high levels of GAMT and not the first ratelimiting enzyme, GATM (last column, green arrowhead in Figure 10A, Figure 12B and 12C). Given that PM express high levels of arginine, they may channel the nitrogen in arginine for urea generation via the urea cycle and not for creatine synthesis, similar to the observations made with hepatocytes ( Figure 12C). We hypothesise that, similar to the observations with kidney and liver, SRPM and PM may be coordinated for completion of creatine synthesis.
In general terms, SRPM appear to be functionally similar to a population of Mϕ recently described in human liver termed 'Non-inflammatory Macs -cluster 10' in single cell RNA-Seq data 132 and Kupffer cells of the mouse. All of these cells express high levels of VCAM1, SLC40A1, HMOX1, CD163, AXL, MERTK and play roles in the capture and breakdown of aged erythrocytes passing through the sinusoids 132-134 . In comparison, PM appear to share some of the functions of liver hepatocytes, with both cells expressing high levels of enzymes ARG1, HAL, GAMT, PFK-1, HK1 and SAA3 which play roles in metabolism of nitrogen-containing compounds, carbohydrates and lipids 78,132 .
In conclusion, the present study provides a comprehensive view of the cellular profiling of Mϕs in different immune microenvironments and highlights the possible contributions of SRPM and PM populations in homeostasis and host defence that are adapted to each tissue environment.

Conclusions
This study has made effective use of a vast proteomic data set to obtain a better overall picture and understanding of the steady-state roles of splenic red pulp and peritoneal cavity Mϕs, defined mainly in terms of cell surface phenotype, metabolic pathways and immune functions (Figure 13). Given that several previous studies have demonstrated a low correlation between mRNA and protein expression, our proteomics analysis provides global profiling of these tissue Mϕs at a protein level, providing novel information on absolute copy number per cell. In addition, the integration of our proteomics data sets with corresponding ImmGen transcriptomics data sets provides information on the relationships between the levels of transcripts and the proteins that they encode. This has not been examined previously in a comprehensive manner for tissue Mϕs. By merging these heterogeneous data sets we provide high-throughput data validation for the translation of mRNA to protein for many of the differentially expressed genes. We also show for the genes where mRNA and protein expression do not correlate, the cellular expression levels (as validated by flow cytometry) were in excellent agreement with the proteomics data. Thus, using quantitative proteomics, we have identified and validated several novel cellular differences between splenic and peritoneal Mϕs which were not addressed previously through transcriptomics studies.
We believe our characterization of cell surface markers, metabolic pathways and immune functions for PM and SRPM will be a useful resource in the Mϕ field. In particular, peritoneal Mϕs are easy to isolate and have therefore been used extensively as a model to study tissue Mϕ cell biology and immune functions. The availability of a comprehensive database of proteins expressed by these Mϕs and their relative abundance is important as it can help interpretations of complex results and provide guidance as to the suitability of these Mϕ for addressing specific scientific questions. Dissecting the cellular components and attempting to address why specific components are active at the cellular level using expression data and contextual information helps to understand their possible contribution to tissue specific functions. For example, many of the key enzymes that play roles in the catabolism of dietary fructose, carbohydrate and proteins and the removal of toxic ammonia (from excess amino acids and nitrogen catabolism) via the urea cycle, are enriched in peritoneal Mϕs. As predicted by these observations, we demonstrated that peritoneal Mϕs were able to effectively utilise glutamine and generate urea.
Thus, our approach, which builds upon connecting the mRNA expression data and protein networks, along with concepts proposed from previous publications of target genes, may pave the way to understanding the biological implications of the massive data available at both the mRNA and protein levels. In turn, this can give rise to novel hypotheses that can be tested experimentally.
Mass spectrometry proteomics data are deposited with the Pro-teomeXchange Consortium via the PRIDE 135 partner repository. This project contains the following underlying data: • Data containing raw values for peptide quantification (XLSX) following FASP for use in MS.
• Data containing raw values for the urea assay (XLSX). This project contains the following extended data: • Table 1. Proteomics data set. The table contains protein identifications and quantifications for peritoneal cavity Mϕ (PM) and splenic red pulp Mϕ (SM) obtained using proteomic analysis. The column headers describe the data in each column. The table also reports the copy number per cell for each experimental replicate obtained using the proteomic ruler quantifications and the relative fold change calculated using Perseus software as indicated in Methods.
• Table 2. mRNA data set. The table contains geneexpression profiles for peritoneal and splenic Mϕs obtained from the ImmGen dataset. The table contains mRNA profile data, their mean fluorescence intensities and the relative fold changes wherein affymetrix microarrays were used as the data collection platform. The column headers provide descriptions of the data contained within the table.
• Table 3. Proteome and mRNA data sets combined. The table contains protein quantification from the proteome dataset together with the corresponding transcripts from the ImmGen database which were used for the correlation analysis.
• Table 4. Tissue-specific enriched proteins where proteome and mRNA data were positively correlated.
The table contains protein identifications where the correlation analysis showed 930 distinct transcripts with tissue-specific enrichment that were positively correlated with their corresponding proteins from the proteome data set.
Data hosted with Figshare are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).
Example: Figure 3B, please explain in the legends the meaning of the dotted coloured lines, it is present in the text but figures should be self-explanatory. Same for figure  Flow cytometry experiments reported in Fig 3C and 4B and D. How many replicas were performed, can you run statistical analysis on these data?
Agency for Science, Technology and Research (A*STAR), Singapore Immunology Network (SIgN), Singapore, 138648, Singapore This report by Nagala & Crocker aims at comparing the proteome of murine splenic red pulp and peritoneal macrophages. Briefly, authors have sorted and profiled these two macrophage populations by using high-resolution quantitative proteomics to gain insights into the functions of these two subsets. The report constitutes a solid but simple resource for the community, confirming what is already published. One could argue about the relevance of the study as an extensive catalog of differentially expressed proteins and hypotheses concerning their biological functions. The flow cytometry validation is undoubtedly important here but the biological relevance is questionable as validation is lacking for the biological relevance of the findings. Also, readers could ask about the choice of the authors to have targeted these two populations.
Basically, ones could compare one by one virtually all macrophage populations i.emicroglia vs. Kupffer cells or alveolar macrophages vs. heart macrophages and obtain a similar study. The underlying question is to decide if this kind of catalog deserves a publication in itself or if it should be reported as a publicly available database.
Few more comments below: RBC are lysed for splenic cell preparation but not for peritoneal ones. Even if these are standard procedures, it might artificially induce the aged/damaged erythrocyte recycling program. Do authors have check this with non-lysed samples? ○ Page 8: "Other TLRs were expressed at similar levels in both tissue Mφs ( Figure  3A and Figure 4A, B)". No TLR are displayed on Figure 3A.
○ Figure 3C is not so relevant as data are already reported in Figures 3A and B

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Yes

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? Yes