Main

Essentially all tissues are populated by self-renewing long-lived tissue-resident macrophages (RTMs) that clear damaged cells and debris. In the brain, RTMs prune neuronal synapses, in the lung they remove surfactant, and in the liver they eliminate senescent red blood cells (RBCs). The core responsibility of RTMs is to maintain tissue health and homeostasis through these homeostatic functions1. RTMs colonize tissues during embryogenesis, with distinct regulatory nodes controlling acquisition and maintenance of a given tissue’s macrophage program. For instance, in the lung, GM-CSFR signalling is essential for alveolar macrophage (lung RTM) development, whereas transcription factor GATA-6 controls peritoneal cavity RTM differentiation2. However, we have limited understanding of regulatory factors that are important for RTMs irrespective of their tissue of residency and how these programs influence the differentiation of monocytes into RTMs across tissues3,4.

Polyamines are ubiquitous metabolites whose biosynthesis is augmented in metabolically active cells. Polyamines have critical roles in many cell functions, including in translation, during which spermidine is the substrate for a two-step reaction mediated first by deoxyhypusine synthase (DHPS) and then by deoxyhypusine hydroxylase to post-translationally ‘hypusinate’ eIF5A, a process in which a conserved lysine is converted into the amino acid hypusine5,7. Historically, eIF5A has been considered to be the only protein that contains hypusine, with the only functions of DHPS and deoxyhypusine hydroxylase being to hypusinate eIF5A8. Hypusinated eIF5A enhances the translation efficiency of certain mRNA transcripts that lead to ribosome stalling9, including those with polyproline motifs10,11, but it has been unclear exactly which transcripts are within this set and how this process is affected in diverse biological contexts in mammalian cells. Previous work by our group and others has shown that polyamine metabolism directs T cell lineage choices and pathogenic potential in inflammation12,13, and that hypusinated eIF5A boosts respiration, in part by enhancing the translation of certain tricarboxylic acid cycle enzymes, and as such contributes to macrophage alternative activation14. Other studies have investigated the role of DHPS deficiency in myeloid cell inflammation, observing that in obese mice, DHPS deletion suppresses inflammatory macrophage accumulation in adipose tissue and improves glucose tolerance15, and that myeloid cell DHPS expression is required to clear gastrointestinal pathogens by controlling translation of antimicrobial factors16. Here we investigated the role of DHPS in macrophages in steady state using mouse models of myeloid or macrophage-specific gene deletion and found a striking deficiency in mature RTMs across tissues in the absence of DHPS.

Dhps-ΔM mice have a global RTM defect

To examine the polyamine–hypusine pathway (Fig. 1a) in macrophages, we bred mice with loxP-flanked exons 2–7 of Dhps17 with mice expressing LysM–Cre to generate mice with DHPS deleted in myeloid cells, including monocytes and macrophages (Dhps-ΔM mice). We crossed these mice to Rosa26eYFP mice to generate Dhps+/+ LysM–Cre Rosa26eYFP (control) and Dhps-ΔM Rosa26eYFP mice, in which YFP reports Cre expression and thus DHPS deletion when loxP is present. DHPS was deleted in macrophages from Dhps-ΔM mice, leading to decreased eIF5A hypusine expression (Fig. 1b and Supplementary Figs. 1 and 2). We gated on YFP+ cells and assessed macrophages across tissues. Whereas macrophage number (defined by F4/80 and CD11b, CD11c or CD64, depending on tissue) differed to some extent between 8–10-week old (control) and Dhps-ΔM Rosa26eYFP mice, macrophages were present across all tissues in both genotypes when measured by flow cytometry (Fig. 1c–f and Extended Data Figs. 1–3) and by imaging (defined by F4/80 or IBA1) (Supplementary Figs. 3–6). However, Dhps-ΔM Rosa26eYFP mice exhibited substantial defects in RTMs in the peritoneum (TIM-4+), lung (Siglec-F+), liver (TIM-4+), heart (TIM-4+), brain (CX3CR1+), spleen (TIM-4+) and kidney (CD11c+CD11blow)18,19,20,21,22 compared with controls (Fig. 1c–f and Extended Data Figs. 1–3), findings supported by imaging of TIM-4+ RTMs in the liver and spleen (Extended Data Fig. 4).

Fig. 1: Dhps-ΔM mice have a defect in RTMs.

a, Polyamine biosynthesis and eIF5a hypusination pathways. b, Immunoblot of indicated proteins on FACS-sorted F4/80hiCD11b+ peritoneal macrophages from Dhps-WT and Dhps-ΔM mice. c–e, Representative flow cytometry plots of macrophage subpopulations in peritoneal cavity (c), lungs (d) and liver (e) in Dhps+/+ Rosa26eYFP (control) or Dhps-ΔM Rosa26eYFP reporter mice. CD45+YFP+ cells were gated on singlets and live cells. f, Absolute numbers of macrophages and RTMs across tissues from Dhps+/+ Rosa26eYFP (control) or Dhps-ΔM Rosa26eYFP reporter mice. Images reflect peritoneal cavity, lung, liver, heart, brain, spleen and kidney. Representative plots and graphs summarize results of at least two independent experiments. Data are mean ± s.d., representative of n = 4 biological replicates. Statistical analyses were performed using two-tailed t-tests; P values are shown. DOHH, deoxyhypusine hydroxylase. Illustrations in b–f were created using BioRender (https://biorender.com).

Source Data

Dhps −/− monocytes do not form mature RTMs

We focused on RTMs in peritoneum, lung and liver, distinct niches in which LysM–Cre is strongly expressed (Supplementary Fig. 7). RTMs declined over time in Dhps-ΔM mice (Fig. 2a). In early adulthood, many RTM reservoirs are sustained through self-renewal, receiving little contribution from monocytes, the RTM precursor cell. However, unlike in control mice, we observed persistent monocytes in the peritoneal cavity of Dhps-ΔM mice that we reasoned could reflect sensing of a lack of mature TIM-4+ RTMs (Fig. 2b). We performed parabiosis (Extended Data Fig. 5a) to evaluate the monocyte contribution to RTMs. RTM pools in lung, peritoneal cavity and liver of the wild-type (WT):WT mice contained only host cells, with no contribution from the congenically marked WT parabiont. However, among the WT:Dhps-ΔM mice, RTM niches in the Dhps-ΔM mice comprised cells from the WT parabiont (Fig. 2c and Extended Data Fig. 5b,c). These data indicate that the tissues of Dhps-ΔM mice, despite being populated with F4/80+ macrophages, received continual monocytic influx, which we propose was driven by a dearth of fully developed RTMs. Furthermore, experiments with bone marrow chimeras revealed that monocytes from CD45.2+ Dhps-ΔM bone marrow precursors failed to repopulate RTM pools in irradiated WT recipient mice when competing with CD45.1+ Dhps-WT monocytes (Fig. 2d and Extended Data Fig. 5d,e). These data suggest that the RTM defect in Dhps-ΔM mice results from poor macrophage survival in the tissue, which drives constant monocytic influx.

Fig. 2: DHPS is essential for monocyte-to-RTM maturation and macrophage survival.

a,b, Flow cytometry of RTMs (a) and Ly6C+ monocytes (b); gated: Lin−(CD3, CD19, NK1-1)F4/80−. c, CD45.1 versus CD45.2 8 weeks postparabiosis, lung RTMs (F4/80+CD64+CD11b−Siglec-F+) and chimerism (%), WT:Dhps-ΔM. d, Bone marrow (1:1) CD45.1 plus CD45.2 Dhps-WT or CD45.1 WT plus CD45.2 Dhps-ΔM cells were infused into irradiated CD45.1 WT recipients; chimerism 12 weeks later in lung RTMs is shown. e–g, CD11c+CD64+ macrophages (e), CD11blowSiglec-F+ macrophages (f) and Ly6C+ monocytes (g) in BAL after intratracheal CL; control (CTL) mice (e and f) received PBS; BAL was collected at 7 weeks. h,i, Ki-67 MFI (h) and percentage of active caspase-3 (i) in peritoneal (F4/80+TIM-4+) and alveolar (F4/80+CD11c+) macrophages. j–l, Frequencies (j,k) and numbers per milligram tissue (l) of YFP+ kidney RTMs (KRM; F4/80+CD64+CD11c+CD11blow). m, Active caspase-3 in YFP+ KRM. n,o, YFP+ cardiac RTMs (F4/80+TIM-4+) (n) and YFP+ microglia (CD11b+CD45int) (o) post-tamoxifen i.p. in Dhps+/+ (control) and Dhps**flx/flx CX3CR1–ERT2cre-Rosa26eYFP mice. Representative plots with graphs summarize results of two or more experiments, except in b and e–g, in which they represent one experiment. Data are presented as the mean ± s.d. In a, n = 4 (2, 9, 13 weeks) and 6 (4 weeks), Dhps-WT; and n = 8 (2 weeks) and 4 (4, 9, 13 weeks), Dhps-ΔM for peritoneum, lung. For liver, n = 4 (2, 9, 13 weeks) and 6 (4 weeks), Dhps-WT; and n = 7 (2 weeks), 5 (4 weeks), and 4 (9, 13 weeks), Dhps-ΔM. In b, n = 3 (4 weeks), 4 (8, 12 weeks) and 3 (34 weeks), Dhps-WT; and n = 2 (4 weeks), 4 (8 weeks) and 2 (12, 34 weeks), Dhps-ΔM. In c, d, f and k, n = 4, 5, 3, and 3, respectively. In h, n = 8, Dhps-WT; n = 5, Dhps-ΔM, peritoneal; and n = 7, Dhps-WT and Dhps-ΔM alveolar macrophages. In i, n = 5, Dhps-WT; n = 4, Dhps-ΔM, peritoneal; n = 8, Dhps-WT; and n = 7, Dhps-ΔM alveolar. In l, n = 5 (control) and n = 6 (Dhps**flx/flx). In m, n = 4 (2, 4 weeks) and 3 (10 weeks), control; and n = 3 (2) and 5 (4, 10 weeks), Dhps**flx/flx. In n and o, n = 5 (control), n = 6, Dhps**flx/flx. n indicates biological replicates. Statistics: two-tailed t-tests, two-way analyses of variance; P values are shown. NS, not significant; Tamox, tamoxifen. Illustrations in a–e, g**–j, n and o were created using BioRender (https://biorender.com).

Source Data

To capture what happened to RTM niches after acute macrophage depletion, we injected clodronate liposomes (CL) into the trachea of mice to deplete macrophages in the bronchoalveolar space. In control mice, CL led to rapid macrophage depletion followed by monocyte recruitment and differentiation into SIGLEC-F+CD11blow alveolar macrophages over time (Fig. 2e,f). In Dhps-ΔM mice, however, monocytes entered and differentiated into a persistent population of CD64+CD11c+ macrophages but failed to re-establish the local SIGLEC-F+CD11blow lung RTM pool (Fig. 2e,f). Also evident was a persistent influx of CD11b+Ly6c+ monocytes (Fig. 2g). In control mice, CL injected intraperitoneally (i.p.) rapidly depleted peritoneal macrophages23, followed by monocyte recruitment and differentiation into macrophages (Extended Data Fig. 5f,g). In Dhps-ΔM mice, however, monocytes persisted in the tissue after CL but failed to re-establish the local RTM population (Extended Data Fig. 5f,g). Overall, these data show a collapse in the tissue-residency potential of several DHPS-deficient RTM populations, resulting in persistent but ultimately futile monocytic infiltration to restore the RTM reservoir.

Dhps−/− macrophages have survival defects

The RTM defect in Dhps-ΔM mice manifests as continual turnover of RTM reservoirs, compelled by low macrophage survival driving continual monocytic influx to tissues. We assessed Ki-67 and active caspase-3 expression, proliferation and impending cell death indicators, respectively, in F4/80+TIM-4+ peritoneal cavity and F4/80+ CD11c+ lung macrophages in control and Dhps-ΔM mice. DHPS-deficient macrophages expressed less Ki-67 and more active caspase-3 (Fig. 2h,i). To probe proliferation and survival, we administered IL-4 complexes (IL-4c), which drive accumulation of peritoneal RTMs through self-renewal without recruitment from blood monocytes24, plus EdU, into the peritoneal cavity of Dhps-WT and Dhps-ΔM mice. Control but not DHPS-deficient macrophages accumulated and proliferated (Extended Data Fig. 5h,i). Adoptive transfer of equal numbers of CellTrace Violet (CTV)-labelled peritoneal macrophages from Dhps-WT and Dhps-ΔM mice into congenic recipients showed that DHPS-deficient cells proliferated less than controls, and, although there was some CTV dilution, macrophages did not accumulate (Extended Data Fig. 5j,k), indicating increased cell death. Thus, DHPS-deficient macrophages have a decreased capacity for proliferation coupled with increased death, leading to defective tissue persistence. Notably, there was no difference in peritoneal or liver RTMs, or monocyte expression, between genotypes of colony stimulating factor 1 receptor (CSF1R), the ligand for which is a critical growth and survival factor for monocyte–macrophage development21 (Supplementary Fig. 8).

Mature RTM persistence requires DHPS

Dhps-WT and Dhps**flx mice crossed to CX3CR1–ERT2cre-Rosa26eYFP mice generated offspring in which DHPS could be inducibly deleted in cells expressing CX3CR1. We administered tamoxifen to *Dhps-*WT CX3CR1–ERT2cre-Rosa26eYFP and Dhps**flx/flx CX3CR1–ERT2cre-Rosa26eYFP mice to delete DHPS in mature kidney RTMs25 in adult mice. YFP expression confirmed Cre-mediated gene deletion after tamoxifen. Kidney RTMs were lost by day 45 post-tamoxifen in Dhps**flx/flx CX3CR1–ERT2cre-Rosa26eYFP mice (Fig. 2j–l), a phenotype that was confirmed by imaging (Extended Data Fig. 6) and was correlated with increased active caspase-3 expression in any surviving YFP+ DHPS-deficient RTMs 10 weeks post-tamoxifen (Fig. 2m). We also observed a trend of increased active caspase-3 expression in YFP+ DHPS-deficient RTMs at 5 weeks post-tamoxifen by imaging (Supplementary Fig. 9). Microglia (brain RTMs), and heart RTMs to some extent, also expressed CX3CR1 (ref. 26). Numbers and frequency of YFP+ heart RTMs declined at 8 weeks, (Fig. 2n), whereas microglia numbers were diminished at 11 weeks (Fig. 2o and Supplementary Fig. 10) post-tamoxifen. Thus, mature RTMs rely on DHPS for persistence.

Blocked RTM differentiation

We next performed single-cell RNA sequencing (scRNA-seq) of peritoneal exudate cells from Dhps-WT and Dhps-ΔM mice. Data were subsetted to include only macrophages, on the basis of a range of expression of key macrophage markers (Adgre1, Csf1r, H2-Ab1, Cd68, Lyz2, Itgam, Mertk). Initial analysis revealed differential clustering between genotypes (Fig. 3a), including diminished frequency of canonical, mature RTMs in Dhps-ΔM mice as defined by Timd4 expression (cluster 3) (Fig. 3a–c). Cells from Dhps-ΔM mice also had distinct populations that were largely not observed in controls (clusters 1 and 2) (Fig. 3a,b), of which cluster 1 was associated with Ccr2 expression, a marker of recently infiltrated monocyte-derived macrophages27 (Fig. 3c). We propose that cluster 1 represents monocyte-derived macrophage infiltration triggered by a scarcity of mature RTMs in Dhps-ΔM mice. We then assessed the top differentially expressed genes (DEGs) among all clusters (Supplementary Fig. 11) and found that cluster 1 showed increased expression of Itgb5 (ref. 28), Cx3cr1 (ref. 29) and Ly6c2 (ref. 30), genes associated with protective macrophage phenotypes and cell interactions within tissues. Cluster 2, a population with high expression of Adgre1 (F4/80) that was significantly expanded in Dhps-ΔM mice, had no Ccr2 and Timd4 expression and probably represents an immature monocyte-derived macrophage population unable to develop to mature RTMs (cluster 3). Again, when assessing the top DEGs (Supplementary Fig. 11) we identified Cxcl2 (ref. 31) and Folr2 (ref. 32) expression in cluster 2, and Timd4 along with Tgfb2 (ref. 33), Wnt2 (ref. 34) and Nt5e35 in cluster 3, demonstrating broad differences in phenotypes of peritoneal macrophages that may be correlated with more immature macrophages (cluster 2) versus mature RTMs (cluster 3). Cluster 4 might represent a population that is transitioning from immature macrophages (cluster 2) to RTMs (cluster 3), as evidenced by its Timd4, Marco and Tgfb2 expression (Fig. 3a and Supplementary Fig. 11). Notably, expression of Slc7a2, encoding SLC7A2, which transports ornithine and polyamines into cells, was augmented in clusters 2, 3 and 4, perhaps indicating a role of polyamine uptake in macrophage differentiation (Supplementary Fig. 11). Wnt2, Tgfb2 and Cd63 were among the most significantly downregulated genes in DHPS-deficient cells in clusters 2 and 4 and 3, respectively (Supplementary Fig. 12), perhaps indicating that expression of these genes, which have roles in cell adhesion, trafficking and differentiation33,36,37, is critical for RTM development. Pathway analysis of DEGs between control and DHPS-deficient macrophages indicated that cluster 2 exhibited decreases in chromatin remodelling, translation and proliferation (Fig. 3d and Supplementary Table 1), along with increased expression of inflammatory genes (Fig. 3e and Supplementary Table 2), findings that were to some extent echoed in DHPS-deficient cells in clusters 1 and 3 (Supplementary Fig. 13).

Fig. 3: Single-cell transcriptional analysis indicates a block in monocyte-to-RTM maturation in the absence of DHPS.

a, scRNA-seq clustering analysis of monocytes and macrophages from the peritoneal cavity of Dhps-WT and Dhps-ΔM mice. The proportions of each cluster within conditions are represented as percentages. b, Overlapping clustering distribution between Dhps-WT and Dhps-ΔM mice. c, Ccr2, Adgre1 (F4/80) and Timd4 expression in Dhps-WT and Dhps-ΔM mice. d,e, DAVID pathway enrichment analysis for downregulated (d) and upregulated (e) genes in cluster 2 (Adgre1 (F4/80+; Timd4−) Dhps-ΔM versus Dhps-WT. f, Proteomics analysis: volcano plot of differentially expressed proteins in F4/80+CD11b+ sorted peritoneal macrophages from Dhps-WT and Dhps-ΔM mice. g,h, Selected downregulated (g) and upregulated (h) pathways from DAVID pathway enrichment analysis. scRNA-seq and proteomics data represent one experiment with three biological replicates per condition.

Source Data

To confirm our findings in another tissue, we performed scRNA-seq on CD45+YFP+ (LysM–Cre+) cells sorted from digested lungs of Dhps+/+ LysM–Cre Rosa26eYFP (control) and Dhps-ΔM Rosa26eYFP mice (Supplementary Fig. 14). We focused only on macrophages, selecting cells on the basis of a range of markers relevant to this tissue (Adgre1, Csf1r, H2-Ab1, Cd68, Lyz2, Itgam, Mertk, Itgax, Siglecf). Our analysis revealed significant differences between DHPS-deficient and control macrophages (Extended Data Fig. 7 and Supplementary Fig. 15), with loss of cluster 0 and enrichment of other clusters (including cluster 1) in DHPS-deficient macrophages. When examining genes expressed by alveolar macrophages (Adgre1, Itgax and Siglecf), we found that cluster 0 was exclusively Siglecf positive and therefore represented mature alveolar macrophages, and this cluster was specifically absent from DHPS-deficient macrophages (Extended Data Fig. 7). These results mirror our flow cytometry data, which showed that Dhps-ΔM Rosa26eYFP mice did not lack F4/80+ (Adgre1) CD11c+ (Itgax) macrophages but rather lacked mature Siglec-F+ RTMs (Fig. 1d,f).

On the basis of our scRNA-seq from peritoneal macrophages (Fig. 3a–e), we speculated that the enriched intermediate clusters in DHPS-deficient macrophages (for instance, cluster 2), which were not Timd4+, could represent macrophages blocked in RTM differentiation. In the lung scRNA-seq data (Extended Data Fig. 7), we identified two clusters, 1 and 10, that were almost exclusively found in Dhps-ΔM Rosa26eYFP mice. We questioned whether these clusters represented immature lung macrophages blocked in their differentiation towards mature RTMs, analogous to our observations in the peritoneum. To test this, we identified the top upregulated DEGs of cluster 2 (Fig. 3a,b), the main intermediate cluster, from our peritoneal macrophage dataset and searched for the same signature in the lung macrophage scRNA-seq dataset (Extended Data Fig. 8). This signature mapped to clusters 1 and 10 in the lung, suggesting that the cells we proposed to be blocked in differentiation in one tissue could be observed in a second tissue. This indicated that these cells represent a transitional state from immature macrophages to RTMs that is independent of tissue. When DHPS was absent, whether in the peritoneal cavity or the lung, these cells failed to acquire the RTM signature imposed by the tissue and remained as immature macrophages. These results support a requirement for DHPS for macrophages to differentiate into mature RTMs across tissues.

Defects in cell adhesion and signalling

We next assessed global protein expression in F4/80+ peritoneal macrophages from Dhps-WT and Dhps-ΔM mice (Fig. 3f). Pathway enrichment analysis revealed significant decreases in metabolism, cell adhesion and integrin-mediated signalling pathways, along with increases in immune activation and inflammation (Fig. 3g,h and Supplementary Tables 3 and 4). Many studies have outlined the importance of metabolic remodelling in immune cells38, including RTMs39. Likewise, cell adhesion and signalling are critical facets of RTM biology. A lack of robust expression of components in any of these pathways could lead to a block in RTM differentiation. As hypusinated eIF5A is important for translation, we investigated active translation in WT and DHPS-deficient macrophages by sequencing ribosome-engaged transcripts.

We crossed Dhps+/+ LysM–Cre mice and Dhps-ΔM (Dhps**flx/flx LysM–Cre) mice with RiboTag mice to create a model in which ribosomes become HA-tagged upon LysM–Cre expression, enabling us to efficiently immunoprecipitate ribosomes from macrophages ex vivo and sequence ribosome-associated transcripts (Supplementary Fig. 16a). We used peritoneal macrophages as they are more abundant and their isolation does not require tissue digestion. Before sequencing, we confirmed that of the HA-tagged cells in the peritoneal cavity, 98% of Dhps ‘Ribo’-WT cells) and 93% of Dhps ‘Ribo’-ΔM cells were F4/80+CD11b+ macrophages (Supplementary Fig. 16b). We then sequenced transcripts in all peritoneal exudate cells (input: approximately 60% F4/80+CD11b+HA-TAG+ cells in Dhps ‘Ribo’-WT mice and approximately 50% F4/80+CD11b+HA-TAG+ cells in Dhps ‘Ribo’-ΔM mice; Supplementary Fig. 16b), and all transcripts bound to ribosomes (immunoprecipitation of HA-tagged ribosomes) for each genotype. Although this approach meant that the input included some non-macrophage cells, the pulldown was exclusively HA-tagged LysM-expressing cells and reduced manipulation of the samples.

Hypusinated eIF5A promotes translation of sequences that stall the ribosome, and ribosome stalling can lead to transcript degradation10,11. Therefore, we analysed all transcripts that were stably expressed in the ‘inputs’ for both genotypes and then filtered for transcripts that were differentially enriched on ribosomes between control and DHPS-deficient macrophages (Supplementary Fig. 16c). We aimed to find genes that were significantly decreased on ribosomes in DHPS-deficient macrophages compared with controls but not owing to extraordinary differences in total transcript abundance. This analysis identified 13 genes that were significantly reduced on ribosomes in DHPS-deficient macrophages (Fig. 4a), most of which were involved in cell adhesion or interaction, signalling and apoptosis; these genes included Icos40, Cd28 (ref. 41), Axin2 (ref. 42), Tnik43,44, Amigo2 (ref. 45), Fam83g46, Il1rl1 (refs. 47,48,49), Rab44 (ref. 50) and Oasl1 (ref. 51). These data suggest that the proteins encoded by these genes might be involved in RTM differentiation, and that their expression is decreased in DHPS-deficient macrophages.

Fig. 4: Cell adhesion and signalling pathways critical for macrophage tissue-residency are DHPS-dependent.

a, Heat map of stable transcripts reduced on ribosomes in peritoneal macrophages (pMacs) (*Dhps-*ΔM). b–d, pMacs analysed for IL33R (ST2) (b); cultured with IL-33 (20 ng ml−1) for 72 h or IL-4 (20 ng ml−1) for 24 h and analysed (c); or analysed for IL-4R (d). e, Bulk RNA-seq scheme. f, Volcano plot of shared DEGs. g, DAVID pathway analysis of downregulated genes in Dhps-deficient macrophages. h, L1CAM and E-cadherin, flow cytometry; pMacs from Dhps+/+ Rosa26eYFP and Dhps-ΔM Rosa26eYFP mice. i, Percentage cell recovery by time after EDTA; pMacs were cultured for 24 h. j,k, Confocal images of cultured pMacs with cell structure by brightfield microscopy (j) (arrows indicate rounded cells) and lungs stained for macrophages and stromal cells (F4/80, CD64, and PDGFRa, respectively) from Dhps-WT and Dhps-ΔM mice (k). Nuclei (DAPI); box indicates region of interest. l, Area and sphericity of F4/80+CD64+ macrophages from k. Each data point is the average of five random fields quantified per biological replicate. m, Lungs from Dhps-WT and Dhps-ΔM mice stained for macrophages (F4/80, brown), vasculature (CD31, purple), epithelial cells (EpCAM, teal) and nuclei (haematoxylin, indigo). Frequency of F4/80+ cells not positioned on tissue (non-overlapping) was quantified from five random fields in three biological replicates per genotype. n, Confocal images of YFP+ kidney macrophages from Dhps+/+ (control) and Dhps**flx/flx CX3CR1–ERT2cre-Rosa26eYFP mice 5 weeks post-tamoxifen. Cell volume and sphericity were quantified by Imaris (10.0.1, Bitplane); five random fields in five biological replicates per genotype. All data represent two or more experiments, except in a and e–g, in which data represent one experiment with three biological replicates. Data are mean ± s.d. For b and c, n = 3, Dhps-WT; and n = 4, Dhps-ΔM. For d, n = 5, Dhps-WT; and n = 4, Dhps-ΔM. For h, i, l and m, n = 3. For n, n = 5. n represents biological replicates. Statistical analyses; two-tailed t-tests and two-way analyses of variance; P values are shown. Scale bars, 50 μm (k, top), 20 μm (k, bottom); 20 μm (n). Illustrations in a, b, e and h–n were created using BioRender (https://biorender.com).

Source Data

Il1rl1 codes for the IL-33 receptor (ST2) and is involved in differentiation of bone-marrow-derived monocytes and macrophages to RTMs47,52 and in the self-renewal and tissue repair and maintenance functions of RTMs48,53. Consistent with our ribosome sequencing data, ST2 was decreased on peritoneal macrophages from Dhps-ΔM mice (Fig. 4b), and, functionally, the ability of these macrophages to alternatively activate in response to IL-33 in vitro was impaired47,48 (Fig. 4c). By contrast, DHPS-deficient macrophages expressed IL-4R and partially responded to IL-4 in vitro by inducing alternative activation markers, although not to control cell levels (Fig. 4c,d), consistent with their defective response to IL-4c in vivo (Extended Data Fig. 5h–k) and our previous findings that hypusinated eIF5A directs macrophage alternative activation14. These data suggest that ST2 is among a subset of ribosome-engaged transcripts that depend on hypusinated eIF5A for efficient translation and that its reduced protein expression might contribute to the monocyte-to-RTM transition defect in the peritoneal cavity in Dhps-ΔM mice. However, although decreased ST2 might contribute to this defect in DHPS-deficient peritoneal macrophages, it may not be important for the RTM transition in other tissues. Also, multiple genes, rather than any one gene, probably contribute to RTM differentiation.

As Dhps-ΔM mice manifested a defect in all RTMs, we sought to understand which pathways might be critical for the RTM transition across tissues. We performed bulk RNA-seq on F4/80+ macrophages (Supplementary Fig. 17a–c) isolated from the lung, liver and peritoneal cavity of Dhps-WT and Dhps-ΔM mice (Fig. 4e,f). Pathway analysis of DEGs that were common among all DHPS-deficient macrophages from each tissue identified genes enriched in pathways of cell adhesion, signalling and migration as the most significantly downregulated (Fig. 4g and Supplementary Tables 5 and 6), with increased expression of those enriched in inflammatory response and apoptotic pathways (Supplementary Fig. 17d and Supplementary Tables 7 and 8). Notably, our sequencing of ribosome-engaged transcripts identified several genes involved in cell adhesion (Fig. 4a). For example, Tnik augments expression of target genes in the Wnt/β-catenin pathway, which controls cell adhesion43,44; as such, β-catenin-deficient macrophages lack cell adherence capacity54. The genes encoding cell adhesion molecules L1CAM and E-cadherin (l1cam and Cdh1, respectively) were downregulated in the analysis of DHPS-deficient macrophages isolated from three distinct tissues (Supplementary Table 6); this was consistent with the finding of decreased protein expression (Fig. 4h) and with earlier work showing that E-cadherin expression is polyamine-dependent in alternatively activated macrophages[55](https://www.nature.com/articles/s41586-025-09972-2#ref-CR55 “Van den Bossche, J. et al. Alternatively activated macrophages engage in homotypic and heterotypic interactions through IL-4 and polyamine-induced E-cadherin/cat