Sepsis induces long-term reprogramming of human HSPCs and drives myeloid dysregulation in sepsis survivors

word count: 231 35 Number of figures: 4 in the main text, 3 in Appendix 1 - Supplementary Methods and Figures 36 Number of tables: 3 in Appendix 2 - Supplementary Tables 37

number: 52 38 Supplementary files count: 2 39 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 2

40 Sepsis is a life-threatening condition char acterised by an overwhelming immune response 41 and high fatality. While most research has focused on its acute phase, many sepsis 42 survivors remain immunologically weakened leaving them susceptible to serious 43 complications from even mild infections. The mechanisms underlying this prolonged immune 44 dysregulation remain unclear, limiting effect ive interventions. Here, we analysed whether 45 sepsis induced long-term “training” in hem atopoietic stem and progenitor cells (HSPCs), 46 imprinting changes that persist in their my eloid progeny. Peripheral blood analysis of 8 47 sepsis survivors, 12 patients with septic shock, and 10 healthy donors revealed a significant 48 expansion of CD38+ progenitors in survivors, with increases in megakaryocyte-erythroid and 49 granulocyte-monocyte progenitors, and reduced mature neutrophil counts. This shift 50 suggests impaired granulopoiesis, favouring im mature, immunosuppressive granulocytes. 51 Differentiated macrophages from survivors’ HSPCs exhibited impaired metabolic pathways 52 after lipopolysaccharide stimulation, with dow nregulation of tricarboxylic acid cycle and 53 glycolysis genes, indicating altered immune metabolism. Pathway analysis revealed 54 enhanced type-I interferon (IFN) and JAK-STAT signalling in survivors’ macrophages, 55 reflective of potentially tolerance-prone reprogramming. Finally, exposing healthy donor 56 HSPCs to IFN β during macrophage differentiation reduced HSPC proliferation, increased 57 apoptosis, and induced a metabolic shift towards gl ycolysis over mitochondrial respiration. 58 Together, these findings suggest that sepsis induces lasting reprogramming in HSPCs 59 leading to myeloid progeny with altered i mmune memory that might drive immune 60 dysregulation in survivors. These data open av enues to explore potential targets to better 61 manage long-term immune alterations in sepsis survivors. 62 63 KEY POINTS 64 ● Sepsis induces long-term alterations in HSPCs, leading to the expansion of immature 65 progenitors and metabolic dysregulation of their progeny. 66 ● Type-I IFN signalling reprograms macrophage differentiation, affecting their 67 metabolic function and reducing cell proliferation. 68 69 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 3

70 Sepsis is a life-threatening condition charac terised by a dysregulated host response to 71 infection, leading to organ dysfunction and often high mortality. 1–3 In 2017, 48.9 million 72 people were affected by this syndrome, with 19.7% succumbing to it. 4 Survivors face long-73 term consequences, including increased susceptibility to secondary infections, persistent 74 immune-cell alterations, low-grade inflammation, and release of damage-associated 75 molecular pattern molecules. 5 Mechanistically, monocytes from sepsis survivors undergo 76 inflammatory reprogramming, which drives chronic inflammation and persistent immune 77 activation. This process is sustained by elevated cytokines and nucleotide oligomerization 78 domain-like receptor protein 3 (NLRP3) components that can persist for years after sepsis.6 79 80 Hematopoietic stem cells (HSCs), which are produced in the bone marrow (BM), give rise to 81 all blood-cell lineages. 7 In normal aging, HSCs frequency in the BM increases, with a 82 tendency towards a myeloid bias. 8 Interestingly, sepsis can similarly promote myelopoiesis 83 and trigger emergency haematopoiesis, which eventually leads to accelerated 84 immunosenescence.5 Prolonged myelopoiesis, evidenced by an expansion of myeloid-85 derived suppressor cells (MDSCs) and subsequent long-term changes in myeloid functions, 86 is observed in sepsis survivor s months after the acute phase. 9 Work conducted in murine 87 model of sepsis showed that after acute seps is, HSCs are less responsive to granulocyte-88 colony stimulating factor and thus fail to in duce granulopoiesis, suggesting HSCs exhaustion 89 after severe infection.10 However, the mechanisms by which these changes impact long-term 90 immune function in sepsis survivors remain unclear. 91 Trained immunity – or innate immune memory – is a phenomenon whereby cells of the 92 innate immune system, such as monocytes and macrophages, undergo epigenetic 93 reprogramming after exposure to certain pathogens, enhancing their re sponse to future 94 infection.11 In a process referred to as ‘central trained immunity’, HSCs can also acquire 95 “memory”, generating myeloid progeny that retain this trained state. 12,13 For example, 96 Bacillus Calmette-Guérin (BCG) vaccination instructs murine HSCs to produce trained 97 myeloid progeny, which subs equently protects against M. tuberculosis infection. 14 In the 98 context of sepsis, hematopoietic stem and progenitor cells (HSPCs) and BM-derived 99 macrophages in surviving mice un dergo epigenetic reprogramming, 11 resulting in impaired 100 cytokine production.15 Interestingly, metabolic changes, such as a shift toward glycolysis, are 101 highly linked to the induction of these epigenetic modifications. 11 For example, secondary 102 lipopolysaccharide (LPS) challenges in mice treated with β -glucan (a trained immunity 103 inducer in myeloid cells) leads to increases in myelopoiesis, glycolysis, and cholesterol 104 biosynthesis.16 105 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 4 106 Although the impaired function of myeloid cells in sepsis survivors has been already 107 reported, the signalling pathways and metabo lic mechanisms underlying these long-term 108 alterations remain unexplored. As such, we lack the tools to identify or modulate immune 109 dysfunction in affected patients. Moreover, the long-term consequences of HSPC training 110 during the acute phase of sepsis on immune resi lience in survivors are unclear. To address 111 these knowledge gaps, we compared the profil e of circulating HSPCs and peripheral blood 112 subpopulations in patients with acute septic s hock, long-term survivors (Surv; average 13.5 113 months post-sepsis), and age-matched healthy donors (HD). We then assessed the 114 transcriptional and functional profiles of HSC-derived macrophages from Surv and HD to 115 pinpoint potential contributors to macrophage functional and metabolic impairments. 116 117

118 Study participants 119 Twelve adult patients admitted to the intensive care unit (ICU) at St. Anne's University 120 Hospital in Brno (Czech Republic) with early se ptic shock were prospectively enrolled into 121 the study cohort. Two patients died before reac hing the second collection time point, while 122 the first time point collection for another patient failed. Patients with chronic 123 immunosuppression, ongoing active oncological disease, or who had received antibiotic 124 therapy for more than 2 days were excluded. Addi tionally, eight sepsis survivors (Surv, 8 to 125 26 months - 13.5 months on average after the initial ICU admission) were retrospectively 126 enrolled in the “sepsis survivor cohort”. Fi nally, 10 healthy age- and comorbidity- matched 127 individuals were recruited into the “age-matc hed healthy donors” (HD) cohort at the First 128 Department of Orthopaedic Surgery, St. Anne’s University Hospital in Brno. Patients with an 129 acute infection within the last 28 days, ongoi ng active oncological disease, or chronic 130 immunosuppression were not included. Cohort details are summarised in Supplementary 131 Tables 1 and 2. For in vitro studies on the effect of IFN β on HSPCs differentiation, buffy 132 coats from adult blood donors were obtained from the Department of Transfusion & Tissue 133 Medicine of Brno University Hospital. Cord blood was obtained from women after childbirth 134 at the Institute for the Care of Mother and Child in Prague. Written informed consent was 135 obtained from all enrolled patients. All procedures were approved by the institutional Ethical 136 Committee of St. Anne's University Hospital Brno (4G/2018; 10G/2021), Ethical Committee 137 of the Faculty of Medicine of Masaryk University (18/2023), and Ethical Committee of the 138 Institute for the Care of Mother and Child (31/03/2014). All procedures complied with the 139 Helsinki Declaration of 1975, as revised in 2013. 140 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 5 Flow cytometry 141 Flow cytometry analyses followed the guidelines by Cossarizza et al. 17 A total of 200 µL and 142 500 µL of heparinized blood were used to label mature immune cells and circulating HSPCs, 143 respectively. Whole blood was lysed in 1x RBC lysis buffer for 10 minutes at room 144 temperature (RT). Where indicated, dead cells were labelled with Live/Dead fixable dyes 145 (Thermo Fisher Scientific) at a concentration of 1:800 in PBS. Cells were labelled in FACS 146 buffer for 30 minutes on ice with the antibodies listed in Supplementary Table 3. Where 147 indicated, propidium iodide was added immediately before sample acquisition to discriminate 148 dead cells. Then, 10 µl of Precision Count Beads (BioLegend) were added to each sample 149 to obtain absolute counts of HSPCs and other immune subsets. All samples were acquired 150 on a Sony SA3800 spectral analyser (Sony Biotechnologies). 151 HSPC-derived macrophage (HSDM) differentiation 152 HSDM differentiation was performed as described, with minor modifications. 18 CD34+ cells 153 were isolated directly from PBMCs or fr om enriched HSPCs (RosetteSep Hematopoietic 154 Progenitor Enrichment Cocktail Kit, Stemcell Technologies) via immunomagnetic isolation 155 (Miltenyi Biotec). Extended experimental details are reported in the supplementary methods. 156 HSDM stimulation and RNA-sequencing 157 Mature HSDMs (1 x 10 5) from HD and Surv were seeded in 96-well plates in Media C and 158 stimulated with 100 ng/mL LPS-EB (Invivogen) for 3 hours at 37ºC. Total RNA was extracted 159 using the RNeasy Plus Micro Kit (Qiagen), ac cording to manufacturer’s recommendations. 160 RNA quality was assessed with Bioanalyzer2100 RNA Nano 6000 chips (Agilent 161 Technologies), and samples with an RNA Integrity Number (RIN) > 8 were used for 162 sequencing. An Illumina sequencing library was prepared using the NEBNext Ultra II 163 Directional RNA Library Prep Kit (New England Biolabs) following the manufacturer's 164 protocols. Total RNA was used for poly-A enrichment, then fragmented, and reverse 165 transcribed into cDNA. After universal adapter ligation, samples were barcoded using NEB 166 dual indexing primers and pooled equimolarly after quantitation with PicoGreen. The sample 167 pool was sequenced using a Nextseq 550 sequencer (Illumina) with a 75-cycle high-output 168 cartridge. 169 RNA-seq analysis 170 Raw reads were quality checked, pre-processed, and mapped to the reference genome 171 (Ensembl GRCh38) with gene annotation (Ensembl v94). Mapped reads were counted and 172 summarised by gene. After removing genes with < 10 counts, differentially expressed genes 173 (DEGs) were calculated using DESeq2. 19 Gene ontology (GO) and Gene Set Enrichment 174 Analysis (GSEA) were performed using the clusterProfiler package. 20 GSEA was performed 175 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 6 after adaptive shrinkage of the log2 fold-change (LFC) values. 21 To infer signalling pathway 176 activity, the DecoupleR package 22 was used, fitting a Multivariate Linear Model using as 177 input the Wald statistic results from DES eq2 and the top 500 responsive genes ranked by p-178 value in the PROGENy collection 23. Transcription factor activity was inferred by fitting a 179 Univariate Linear Model (available with the DecoupleR package) using as input the Wald 180 statistic results calculated by DESeq2 and the human regulons in the DoRothEA gene 181 regulatory network with confidence levels “curated/high (A)”, “likely (B)”, and “medium (C)”.24 182 Statistical analyses 183 Statistical analyses were performed with R v4.0.2 . Specific statistical tests are reported in 184 each figure legend. The Shapiro-Wilk test and vi sual inspection of QQ-plots were used to 185 determine the normality of distributions, guiding the selection of parametric or non-186 parametric tests. 187 188 Sample collection and preparation, extended HSPC-derived macrophage (HSDM) 189 differentiation, Immunofluorescence staining, Cell cycle profiling, Apoptosis assay, and 190 Metabolic profiling of HSDMs are descr ibed in Supplementary methods and figures 191 (Appendix 1). 192 193

194 Circulating HSPCs are expanded in sepsis survivors 195 We first aimed to investigate the effect of sepsis on HSPCs and their progeny, in order to 196 understand whether the long-term immunosuppres sion observed in sepsis survivors could 197 be due to the reprogramming of HSPCs during sepsis. To do so, we enrolled 12 patients 198 with septic shock at two time points (within 24 hours (T1) or 3-5 days (T2) from ICU 199 admission), eight Surv (average 13.5 months since ICU discharge; Supp. Table 1), and 10 200 HD (Supp. Table 2). We began by analysing HSPCs in the peripheral blood of all 201 participants by flow cytometry (see Supp. Figure 1 for gating strategy) and found differences 202 in their absolute numbers across groups (Figure 1A, Supp. Figure 2A). 203 204 When comparing HD with Surv, the latter showed a significant expansion of CD38+ 205 progenitors (Figure 1A, p=0.034). Among these committed progenitors, megakaryocyte-206 erythroid progenitors (MEPs) were significantly increased in survivors, while common 207 myeloid progenitors (CMPs) and granulocyt e-monocyte progenitors (GMPs) showed a 208 similar trend (Figure 1A, p=0.068 and p=0.101, respectively). When patients with septic 209 shock were included in the comparison, we found a significant increase in the absolute GMP 210 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 7 counts in survivors compared to septic shock patients at both time points (p=0.0276 for TP1 211 and p=0.0288 for TP2; Supp. Figure 2A). MEP counts showed a similar trend when 212 compared to TP1 (p=0.0528). Taken together, these results suggest that septic shock 213 affects the differentiation of HSCs into committed progenitors long after infection resolution, 214 favouring the accumulation of MEPs and GMPs. 215 216 To determine whether the changes observed in the HSPC compartment were reflected in the 217 terminal differentiation of blood cells, we analysed the peripheral blood of Surv. Compared to 218 HD, Surv showed a decrease in absolute count of mature neutrophils (p=0.057, Figure 1B). 219 Consistent with the expansion of polymorphonuclear (PMN) cells and early MDSCs in the 220 peripheral blood of Surv, 9 these findings suggest that sepsis induces long-term 221 reprogramming of the hematopoietic compartment, promoting a skew towards the 222 granulocytic lineage and the development of immature and immunosuppressive 223 granulocytes. The release of MEP from BM in peripheral blood has been described in a 224 mouse model of sepsis as a result of higher concentration of SCF in peripheral blood 225 compared to BM. 26 Here we report expansion of MEP in Surv long after septic shock, 226 suggesting that a similar mechanism might be still taking place long after recovering from 227 sepsis. 228 Macrophages derived from sepsis survivor HSPCs show metabolic impairments 229 Having shown expanded CD38+ progenitors, increasing trends in CMP and GMP and 230 decrease in absolute counts in neutrophils in Surv compared to HD, we hypothesized that 231 HSPCs in Surv could give rise to a myeloid pr ogeny with altered functionality. To test this, 232 we adapted a protocol18 to differentiate macrophages in vitro from circulating CD34+ HSPCs 233 (HSPC-derived macrophages, HSDMs). We first used flow cytometry to characterise mature 234 HSDMs obtained from adult circulating HSPCs isol ated from adult blood donors. These cells 235 expressed CD68, CD11b, CD33, CD14, CD16, CD206, CD86, HLA-DR, CCR5, TLR2 and 236 TLR4 (Supp. Figure 2B and C), indicative of successfully differentiated macrophages. 237 Moreover, these cells demonstrated the ability to phagocytose and sequester 238 Staphylococcus aureus within their lysosomes, as shown by the uptake of pHrodo S. aureus 239 particles (Supp. Figure 2D), confirming that their core immune functionality is retained. 240 241 Next, we assessed whether the transcriptional pr ofile of HSDMs differentiated from Surv and 242 HD differed as a result of sepsis. Here, HSDMs from Surv and HD were stimulated with LPS 243 and subjected to bulk RNA-seq. We found that the stimulated HSDMs from Surv exhibited 244 upregulation of genes involved in the “respons e to lipopolysaccharide” and “regulation of 245 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 8 innate immune response” pathways when compared to non-treated cells from the same 246 individuals (Figure 2A and B). 247 248 When comparing LPS-stimulated HSDMs from Surv and HD, we found that the former 249 downregulated many genes involved in key metabolic pathways, including the TCA cycle, 250 glycolysis/gluconeogenesis, and pyruvate metabolism, as indicated by the negative gene-set 251 enrichment scores for these pathways (Figure 2C and D). These findings align with other 252 studies indicating that trained immunity relie s on the metabolic reprogramming of myeloid 253 cells, and with studies showing a decrease in all major metabolic pathways as a hallmark of 254 tolerant monocytes.11,27 255 256 As initial exposure to bacterial molecules (LPS in particular) can render myeloid cells more 257 tolerant to subsequent restimulation, 27,28 we explored whether a similar mechanism might 258 affect HSDMs derived from Surv. Using a pub licly available dataset of genes that are 259 “tolerizable” and “non-tolerizable” to TLR4-dependent LPS exposure in human 260 macrophages,28 we performed gene set enrichm ent analysis (G SEA) and found no 261 enrichment in either category (adjusted p va lue=0.484 and 0.604, respectively; Figure 2E). 262 This finding suggests that the mechanisms i nducing a tolerant phenotype in macrophages 263 after LPS exposure differs from the one acting on HSPCs during sepsis. It is, thus, possible 264 that the mechanisms governing endotox in tolerance and sepsis-mediated 265 immunosuppression rely on different processes, the latter likely involving a metabolic and 266 epigenetic rewiring of HSPCs. 267 Type-I IFN signalling characterises HSDMs in sepsis survivors 268 To explore signalling pathways differences t hat might underlie our observations regarding 269 the metabolic and transcriptional reprogrammi ng of HSDMs from Surv, we inferred the 270 activity of signalling pathways and transcription factor (TF) regulons across conditions using 271 our RNA-seq data. In HSDMs from Surv, LPS stimulation led to induction of the NF- κ B 272 pathway (Figure 3A). This was accompanied by the activation of TFs involved in the TLR4-273 mediated response to LPS, including NFKB1, RELA and RELB (Figure 3B), which are 274 central to NF- κ B signalling. 29,30 Additionally, we observed JAK-STAT pathway activation in 275 response to LPS stimulation. While NF- κ B is rapidly activated through TLR4 upon LPS 276 exposure, the JAK-STAT pathway can be activated in an autocrine manner following the 277 induction of IFNβ expression, which is also triggered by TLR4 activation via LPS.29,30 278 279 When directly comparing LPS-stimulated HSDMs derived from Surv with those from HD, we 280 found distinct signalling patterns. Specifically, JAK-STAT pathway activity was higher in 281 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 9 HSDMs from Surv, whereas the phosphatidylinositol 3-kinase (PI3K) pathway was more 282 activated in HSDMs from HD (Figure 3C). Acco rdingly, we noticed the preferential activation 283 of STAT1 and IRF9 in HSDMs from Surv, compared to those from HD (Figure 3D). These 284 TFs are activated downstream of IFNAR1 and IFNAR2 receptors in response to type-I 285 interferons (IFN) such as IFNα or IFNβ , and regulate the expression of IFN-stimulated genes 286 (ISGs). Taken together, our results suggest a hyperactivation of type-I IFN signalling in 287 macrophages differentiated from Surv HSPCs, which could reflect a mechanism to 288 overcome long-term sepsis-related immunosuppression and tolerance.31 289 IFNβ affects the development of macrophages from HSPCs 290 Given that HSDMs from Surv showed increased activation of the JAK-STAT pathway and 291 elevated STAT1 and IRF9 TFs activity, we hypothesised that type-I IFNs could be 292 responsible for the dysregulation observed in the HSPC compartment of Surv. To determine 293 the role of type-I IFNs during myeloid differentiation, we differentiated HSDMs from 294 circulating HSPCs isolated from adult healthy blood donors in the presence or absence (not 295 treated, NT) of IFN β , which is a primary mediator of ty pe-I IFN responses that influences 296 immune cell differentiation and inflammatory signalling. 297 We observed a reduction in total cell numbers after 7 and 14 days of differentiating IFN β -298 stimulated HSPCs, compared to untreated cells (Figure 4A and B). This reduction was 299 similarly observed when using cord-blood HSPCs (Supp. Figure 3A). Nevertheless, IFN β -300 treated cells from adult blood donors showed an increased proportion of CD14+ cells at 14 301 days of differentiation (Supp. Figure 3B), suggesting that IFN β stimulation might support 302 monocytic differentiation over other lineages in vitro. 303 304 To assess whether the reduced cell numbers in IFN β -stimulated HSPCs from adult healthy 305 blood donors were due to decreased proliferation, we analysed a set of genes involved in 306 cell proliferation after 7 days of differentiation. Only two genes ( B2M and ANAPC2) were 307 significantly upregulated in IFN β -stimulated cells compared to untreated cells, indicating that 308 cell proliferation rates likely remained unchanged (Figure 4C). We then tested whether 309 increased cell death during differentiation could account for the lower cell counts. Annexin-V 310 staining revealed a significantly higher frequency of early apoptotic cells (Annexin-V+, PI-) in 311 IFNβ -treated cells (Figure 4D). Interestingly, early and late apoptotic cells showed reduced 312 expression of CD38 compared to live cells (Figure 4E and F), suggesting that these cells 313 might represent less committed progenitors . We speculate that the increase in 314 CD34+CD38+ HSPCs and GMP in Surv compared to HD and the concomitant decrease in 315 mature neutrophils in peripheral blood in Surv could be caused by the IFN β -induced 316 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 10 apoptosis of CD38+ progenitors (Figure 1, Supp. Figure 2A). This would also be supported 317 by the higher activity in the p53 pathway in HSDMs from Surv compared to HD (Figure 3C). 318 319 Finally, we examined whether IFN β stimulation during HSDM differentiation induced any 320 metabolic reprogramming. IFN β -stimulated HSDMs showed a significantly lower oxygen 321 consumption rate-to-extracellular acidification rate (OCR-to-ECAR) ratio at basal levels 322 (Figure 4G, H). This finding suggests that, in the resting state, IFN β -stimulated HSDMs 323 depend less on the TCA cycle and more on glycolysis, compared to untreated cells, which is 324 consistent with a metabolic shift following classical macrophage activation. 32 Nevertheless, 325 in Surv the transcriptional regulation of the glycolytic pathway in LPS-stimulated HSDMs is 326 also suppressed (Figure 2D), likely due to the immunosuppression induced by the cytokine 327 storm and their consequential switch toward s alternatively-activated macrophages, as 328 evidenced elsewhere, 1 despite the activated hypoxia pathway (Figure 3C) that in general 329 favours glycolysis in pro-inflammatory macrophages.33 330 331

332 Sepsis research has traditionally focused on immune-cell responses during the acute phase 333 due to its high fatality rate in this period. However, the persistent health complications 334 observed in many sepsis survivors months after their initial recovery, highlights a significant 335 gap in our understanding of the long-term impacts of sepsis on the immune system. This 336 study aimed to explore these long-term effects specifically on HSPCs, investigating whether 337 sepsis induces “training” in these cells during septic shock that carries over to the myeloid 338 progeny and alters their metabolism and signalling responses. 339 340 Previous studies, including our own,9,34 have identified alterations in immune cell phenotypes 341 in sepsis patients during the acute phase of septic shock, 35–37 as well as notable shifts in 342 MDSCs and PMN-MDSCs that are detectable 6-26 months after recovery. 9 Nevertheless, a 343 thorough analysis of the immunophenotypic changes that are apparent in sepsis survivors 344 has not yet been performed, thus limiting our understanding of the adverse effects and 345 recurrence of sepsis in these patients. We now provide evidence of a likely biologically 346 relevant decrease in absolute neutrophil numbers in the peripheral blood of long-term sepsis 347 survivors (Surv). Furthermore, we observed an expansion of GMPs in Surv compared to 348 patients in the acute phase of septic shock, indicative of defective granulopoiesis in this 349 group. Together with an increase of immature PMN-MDSCs, these data suggest that 350 impaired granulopoiesis gives rise to suppresso r cells rather than mature neutrophils in 351 sepsis survivors. 352 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 11 353 We also observed a long-term shift in the hematopoietic compartment of Surv, characterised 354 by an increased number of committed CD38+ progenitors and MEPs compared to HD. 355 Although thrombocytopenia commonly occurs in septic patients, 38 thromboembolism events 356 are also common. 39 Accordingly, in murine models of severe septic shock it has been 357 reported that MEP counts increase 26 and that platelets can exacerbate inflammation 40. The 358 MEP expansion observed in Surv might explain th e increased risks of myocardial infarction 359 and stroke observed in these patients. 41 The increase of CD38+ progenitors suggests that 360 HSCs might undergo “training” during septic shock, which influences their differentiation and 361 the characteristics of their myeloid progeny. I ndeed, these findings resonate with an earlier 362 finding that BCG vaccination trains monocyte and macrophage progeny. 14 We confirmed 363 whether there exists a memory of sepsis-relat ed training in HSPC progeny by differentiating 364 macrophages (HSDMs) in vitro from Surv HSPCs and re-challenging them with LPS. Bulk 365 RNA sequencing revealed downregulation of gene s involved in glycolysis/gluconeogenesis, 366 the TCA cycle, and pyruvate metabolism in these LPS-challenged Surv HSDMs compared to 367 HD. These findings align with previously described defects in glycolysis and oxidative 368 phosphorylation in monocytes from septic patients. 27 Despite published data suggesting 369 normalisation of these metabolic determinants within days of recovery (> 7 days after 370 sepsis),27 we show that such defects persist in HSDMs differentiated from Surv. These 371 discrepancies might be caused by the different time frame after which the measurements 372 were taken in these studies - quite early after the recovery vs months after the resolution of 373 sepsis. We speculate, therefore, that sepsis primes HSPCs in a way that persists over time, 374 potentially driving a myeloid bias in immune cell production. 375 376 LPS-induced tolerant macrophages can protect against septic-shock-induced cell death and 377 improve survival in mice. 42 Reprogramming into hyporesponsive, immunosuppressive 378 phenotypes depend on the p21-mediated DNA binding of NF- κ B p50 homodimers, which 379 inhibit mRNA transcription and reduce IFN β production in mice.43 Here, we demonstrate that 380 reprogramming during septic shock differentially affects the response to LPS of HSDMs 381 derived from Surv compared to HD HSDMs. Pathway analysis provided further insights into 382 the altered immune responses in Surv as a resu lt of this reprogramming: LPS stimulation 383 induced NF- κ B and JAK-STAT pathway activation in HSDMs from Surv, the JAK-STAT 384 response notably stronger than observed in HD. This effect was coupled with elevated type-I 385 IFN activity, including STAT1 and IRF9 activation that is indicative of sustained type-I IFN 386 signalling. JAK-STAT activation after LPS treatment and IFNAR1 via autocrine IFN β 387 production are consistent with findings from murine BM-derived macrophages. 44 The role of 388 Ifnar1 in immune “training” has also been described in murine alveolar macrophages, 45 as 389 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 12 well as in a mouse model of autoimmune systemic lupus erythematosus, where a type-I IFN 390 signature promoted myelopoiesis. 46 Additionally, type-I IFN signalling impairs macrophage 391 anti-Mycobacterium tuberculosis immunity in mice. 47 While IFNAR1 downregulation 392 contributes to HSCs maintenance, 48 heightened IFN β signalling through JAK-STAT in Surv 393 contrasts with the hyporesponsive phenoty pe seen in murine macrophages, where IFN β 394 production is suppressed.43 Together, these findings suggest t hat immune training in Surv is 395 characterized by a long-lasting reprogrammi ng of myeloid cells characterised by an 396 immunosuppressive metabolic state and heightened type-I IFN responses upon LPS 397 challenge. These results point towards a potential model where “primed” HSPCs differentiate 398 into myeloid cells more prone to mounting antiv iral over antibacterial responses, possibly 399 underlying the susceptibility to bacterial infect ions of sepsis survivors and the potential 400 sepsis recurrence.49 401 402 Compared to HSDMs derived from Surv, we found that HSDMs from HD predominantly 403 activate the PI3K pathway upon LPS treatment. PI3K signalling is associated with 404 modulating inflammation and enhanced cell surviv al, and so might constitute a protective 405 mechanism to counterbalance excessive pro-in flammatory immune activation and cellular 406 apoptosis.50 Interestingly, a dependency on PI3K has been shown in IFN β -driven regulation 407 of glucose metabolism. 51 Our findings suggest that IFN β exposure during HSPC 408 differentiation from adult blood donors leads to increased early apoptosis (after 7 days of 409 differentiation in vitro ) and a bias towards CD14+ cell development (after 14 days of 410 differentiation in vitro). Finally, we found that HSDMs differentiated in the presence of IFN β 411 depend on glycolysis. This finding is consist ent with our RNAseq data, which revealed 412 reduced TCA cycle activity in HSDMs derived from Surv but contradicts the reduced 413 glycolysis observed in HSDMs from Surv. This is likely due to the different activation state of 414 the cells and the length of the stimulation. A study evidenced that a 5-day “chronic” exposure 415 of human monocyte-derived macrophages to IFN β suppresses both basal oxygen 416 consumption rate and glycolysis.52 417 418 Despite the limited number of samples analyzed and the simplified model of type-I IFN 419 activity in sepsis modelled by prolonged IFNβ stimulation of HSPCs from adult blood donors, 420 our study provides evidence of long-term r eprogramming in Surv compared to HD. It is 421 important to note that the cytokine storm and signalling dynamics during septic shock in vivo 422 are greatly more complex than the in vitro system tested in our study. Moreover, different 423 infectious agents and sepsis severity might be driving different responses, and thus 424 reprogramming, in HSPCs. This could not be tested in our study due to the limited size of 425 our cohort. Finally, our results highlight the need to evaluate the metabolic and 426 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 13 transcriptional status of terminally differentia ted immune cells during and after sepsis. These 427 analyses could support the development of ther apeutic targets to overcome the recurrence 428 and adverse effects of septic shock. 429 Taken together, we show that sepsis induces a long-term skew in HSPC differentiation and 430 myeloid cell functionality in Surv. Compared to HD, Surv show increased numbers of 431 committed progenitors and decreased neutrophil counts. Importantly, the immune training of 432 HSPCs during the acute phase of sepsis pred isposes them to differential signalling upon 433 subsequent LPS re-challenge in vitro. The evidence of metabolically impaired macrophages 434 should be elaborated further and potentially used as a model to improve immune cell 435 responses in sepsis survivors who suffer fr om opportunistic infections and face potential 436 sepsis recurrence. Potential therapeutic strategi es targeting immune cell-specific metabolic 437 manipulations using nutrition supplements, or affecting the cytokine availability, such as 438 IFNβ , could help re-establish the homeostatic state of hematopoietic progenitors and 439 immune cells after the septic shock episode. Overall, our results open potential therapeutic 440 opportunities to manage long-term immune-cell reprogramming in septic shock survivors. 441 442

443 The research was supported by the Ministry of Health of the Czech Republic, grant nr. 444 NV21J-05-00056), all rights reserved and DRO (Institute of Hematology and Blood 445 Transfusion – UHKT, 00023736). The research was also supported by project nr. 446 LX22NPO5107 (MEYS): Financed by European Union – Next Generation EU and the 447 European Union's Horizon Europe research and innovation programme under grant 448 agreement No. 101137484. 449 We would like to thank the technical support team of the Center for Translational Medicine 450 for technical support. Core Facility Genomics of CEITEC Masaryk University is gratefully 451 acknowledged for the obtaining of the scientific data presented in this paper. We would also 452 like to thank Dr. Jessica Tamanini from Insight Editing London for critical review of the 453 manuscript. 454 455 AUTHORSHIP 456 Contribution: MDZ and JF designed the study; MH, VS, MV, and JH supervised the cohort 457 recruitment; VT, AM, JS, TT, KrB, and MH recruited the study participants; MDZ, PL, MHK, 458 KaB, VB, IA, NV, and SU performed the experiments and analyzed the data; MDZ, KaB, 459 MHK, and JF secured funding; MDZ, PL, KaB, and JF wrote and reviewed the manuscript. 460 461 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 14 Conflicts-of-interest disclosure: MDZ is an employee and owns stock in Ensocell 462 Therapeutics. Other authors declare no conflict of interest. 463 464 APPENDIXES 465 Appendix 1: Supplementary Methods and Figures. 466 Appendix 2: Supplementary Tables. 467 468 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 15

469 1. Cao M, Wang G, Xie J. Immune dysregulat ion in sepsis: experiences, lessons and 470 perspectives. Cell Death Discov. 2023;9(1):465. doi:10.1038/s41420-023-01766-7 471 2. Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus 472 Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801-810. 473 doi:10.1001/jama.2016.0287 474 3. Jarczak D, Kluge S, Nierhaus A. Sepsis—Pathophysiology and Therapeutic 475 Concepts. Front Med (Lausanne) . 2021;8. 476 https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2021.628302 477 4. Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis 478 incidence and mortality, 1990–2017: analysis for the Global Burden of 479 Disease Study. The Lancet . 2020;395(10219):200-211. doi:10.1016/S0140-480 6736(19)32989-7 481 5. Venet F, Monneret G. Advances in the understanding and treatment of sepsis-482 induced immunosuppression. Nat Rev Nephrol . 2018;14(2):121-137. 483 doi:10.1038/nrneph.2017.165 484 6. Gritte RB, Souza-Siqueira T, Borges da Silva E, et al. Evidence for Monocyte 485 Reprogramming in a Long-Term Postsepsis Study. Crit Care Explor . 2022;4(8). 486 https://journals.lww.com/ccejournal/fulltext/2022/08000/evidence_for_monocyte_repro487 gramming_in_a_long_term.3.aspx 488 7. Swann JW, Olson OC, Passegué E. Made to order: emergency myelopoiesis and 489 demand-adapted innate immune cell production. Nat Rev Immunol. 2024;24(8):596-490 613. doi:10.1038/s41577-024-00998-7 491 8. Pang WW, Price EA, Sahoo D, et al. Human bone marrow hematopoietic stem cells 492 are increased in frequency and myeloid-biased with age. Proceedings of the National 493 Academy of Sciences. 2011;108(50):20012-20017. doi:10.1073/pnas.1116110108 494 9. De Zuani M, Hortová-Kohoutková M, Andrej č inová I, et al. Human myeloid-derived 495 suppressor cell expansion during sepsis is revealed by unsupervised clustering of flow 496 cytometric data. Eur J Immunol. 2021;51(7):1785-1791. 497 doi:https://doi.org/10.1002/eji.202049141 498 10. Biswas N, Bahr A, Howard J, Bonin JL, Grazda R, MacNamara KC. Survivors of 499 polymicrobial sepsis are refractory to G-CSF-induced emergency myelopoiesis and 500 hematopoietic stem and progenitor cell mobilization. Stem Cell Reports . 501 2024;19(5):639-653. doi:10.1016/j.stemcr.2024.03.007 502 11. Ochando J, Mulder WJM, Madsen JC, Netea MG, Duivenvoorden R. Trained 503 immunity — basic concepts and contributions to immunopathology. Nat Rev Nephrol . 504 2023;19(1):23-37. doi:10.1038/s41581-022-00633-5 505 12. Netea MG, Domínguez-Andrés J, Barreiro LB, et al. Defining trained immunity and its 506 role in health and disease. Nat Rev Immunol . 2020;20(6):375-388. 507 doi:10.1038/s41577-020-0285-6 508 13. De Zuani M, Fri č J. Train the Trainer: Hematopoietic Stem Cell Control of Trained 509 Immunity. Front Immunol . 2022;13. 510 https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2022.827250 511 14. Kaufmann E, Sanz J, Dunn JL, et al. BCG Educates Hematopoietic Stem Cells to 512 Generate Protective Innate Immunity against Tuberculosis. Cell . 2018;172(1):176-513 190.e19. doi:10.1016/j.cell.2017.12.031 514 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 16 15. Davis FM, Schaller MA, Dendekker A, et al. Sepsis Induces Prolonged Epigenetic 515 Modifications in Bone Marrow and Peri pheral Macrophages Impairing Inflammation 516 and Wound Healing. Arterioscler Thromb Vasc Biol . 2019;39(11):2353-2366. 517 doi:10.1161/ATVBAHA.119.312754 518 16. Mitroulis I, Ruppova K, Wang B, et al. Modulation of Myelopoiesis Progenitors Is an 519 Integral Component of Trained Immunity. Cell. 2018;172(1):147-161.e12. 520 doi:10.1016/j.cell.2017.11.034 521 17. Cossarizza A, Chang HD, Radbruch A, et al. Guidelines for the use of flow cytometry 522 and cell sorting in immunological studies (second edition). Eur J Immunol . 523 2019;49(10):1457-1973. doi:https://doi.org/10.1002/eji.201970107 524 18. Vogel G, Cuénod A, Mouchet R, et al . Functional characterization and phenotypic 525 monitoring of human hematopoietic stem cell expansion and differentiation of 526 monocytes and macrophages by whole-cell mass spectrometry. Stem Cell Res . 527 2018;26:47-54. doi:https://doi.org/10.1016/j.scr.2017.11.013 528 19. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for 529 RNA-seq data with DESeq2. Genome Biol . 2014;15(12):550. doi:10.1186/s13059-530 014-0550-8 531 20. Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R Package for Comparing 532 Biological Themes Among Gene Clusters. OMICS. 2012;16(5):284-287. 533 doi:10.1089/omi.2011.0118 534 21. Stephens M. False discovery rates: a new deal. Biostatistics. 2017;18(2):275-294. 535 doi:10.1093/biostatistics/kxw041 536 22. Badia-i-Mompel P, Vélez Santiago J, Braunger J, et al. decoupleR: ensemble of 537 computational methods to infer biological activities from omics data. Bioinformatics 538 Advances. 2022;2(1):vbac016. doi:10.1093/bioadv/vbac016 539 23. Schubert M, Klinger B, Klünemann M, et al. Perturbation-response genes reveal 540 signaling footprints in cancer gene expression. Nat Commun . 2018;9(1):20. 541 doi:10.1038/s41467-017-02391-6 542 24. Garcia-Alonso L, Holland CH, Ibrahim MM, Turei D, Saez-Rodriguez J. Benchmark 543 and integration of resources for the estimation of human transcription factor activities. 544 Genome Res . 2019;29(8):1363-1375. 545 http://genome.cshlp.org/content/29/8/1363.abstract 546 25. Stirling DR, Swain-Bowden MJ, Lucas AM, Carpenter AE, Cimini BA, Goodman A. 547 CellProfiler 4: improvements in speed, utility and usability. BMC Bioinformatics . 548 2021;22(1):433. doi:10.1186/s12859-021-04344-9 549 26. Valet C, Magnen M, Qiu L, et al. Sepsis promotes splenic production of a protective 550 platelet pool with high CD40 ligand expression. J Clin Invest . 2 022;132(7). 551 doi:10.1172/JCI153920 552 27. Cheng SC, Scicluna BP, Arts RJW, et al. Broad defects in the energy metabolism of 553 leukocytes underlie immunoparalysis in sepsis. Nat Immunol . 2016;17(4):406-413. 554 doi:10.1038/ni.3398 555 28. Foster SL, Hargreaves DC, Medzhitov R. Gene-specific control of inflammation by 556 TLR-induced chromatin modifications. Nature. 2007;447(7147):972-978. 557 doi:10.1038/nature05836 558 29. Paracha RZ, Ahmad J, Ali A, et al. Formal Modelling of Toll like Receptor 4 and 559 JAK/STAT Signalling Pathways: Insight into the Roles of SOCS-1, Interferon- β and 560 Proinflammatory Cytokines in Sepsis. PLoS One. 2014;9(9):e108466-. 561 https://doi.org/10.1371/journal.pone.0108466 562 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 17 30. Duan T, Du Y, Xing C, Wang HY, Wang RF. Toll-Like Receptor Signaling and Its Role 563 in Cell-Mediated Immunity. Front Immunol . 2022;13. 564 https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2022.812774 565 31. Rackov G, Shokri R, De Mon MÁ, Martínez-A. C, Balomenos D. The Role of IFN- β 566 during the Course of Sepsis Progression and Its Therapeutic Potential. Front 567 Immunol. 2017;8. 568 https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2017.00493 569 32. Chen S, Saeed AFUH, Liu Q, et al. Macrophages in immunoregulation and 570 therapeutics. Signal Transduct Target Ther . 2023;8(1):207. doi:10.1038/s41392-023-571 01452-1 572 33. Li C, Wang Y, Li Y, et al. HIF1 α -dependent glycolysis promotes macrophage 573 functional activities in protecting against bacterial and fungal infection. Sci Rep . 574 2018;8(1):3603. doi:10.1038/s41598-018-22039-9 575 34. Hortová-Kohoutková M, Lázni č ková P, Bendí č ková K, et al. Differences in monocyte 576 subsets are associated with short-term su rvival in patients with septic shock. J Cell 577 Mol Med. 2020;24(21):12504-12512. doi:https://doi.org/10.1111/jcmm.15791 578 35. Inoue S, Suzuki-Utsunomiya K, Okada Y, et al. Reduction of Immunocompetent T 579 Cells Followed by Prolonged Lymphopenia in Severe Sepsis in the Elderly*. Crit Care 580 Med. 2013;41(3). 581 https://journals.lww.com/ccmjournal/fulltext/2013/03000/reduction_of_immunocompet582 ent_t_cells_followed_by.13.aspx 583 36. Jensen IJ, Li X, McGonagill PW, et al. Sepsis leads to lasting changes in phenotype 584 and function of memory CD8 T cells. Belz GT, Rath S, Roquilly A, eds. Elife. 585 2021;10:e70989. doi:10.7554/eLife.70989 586 37. Boomer JS, Shuherk-Shaffer J, Hotchkiss RS, Green JM. A prospective analysis of 587 lymphocyte phenotype and function over the course of acute sepsis. Crit Care . 588 2012;16(3):R112. doi:10.1186/cc11404 589 38. Venkata C, Kashyap R, Farmer JC, Afessa B. Thrombocytopenia in adult patients with 590 sepsis: incidence, risk factors, and its association with clinical outcome. J Intensive 591 Care. 2013;1(1):9. doi:10.1186/2052-0492-1-9 592 39. Kaplan D, Casper TC, Elliott CG, et al. VTE Incidence and Risk Factors in Patients 593 With Severe Sepsis and Septic Shock. Chest. 2015;148(5):1224-1230. 594 doi:https://doi.org/10.1378/chest.15-0287 595 40. Su M, Chen C, Li S, et al. Gasdermi n D-dependent platelet pyroptosis exacerbates 596 NET formation and inflammation in severe sepsis. Nature Cardiovascular Research . 597 2022;1(8):732-747. doi:10.1038/s44161-022-00108-7 598 41. Dalager-Pedersen M, Søgaard M, Schønheyder HC, Nielsen H, Thomsen RW. Risk 599 for Myocardial Infarction and Stroke After Community-Acquired Bacteremia. 600 Circulation. 2014;129(13):1387-1396. doi:10.1161/CIRCULATIONAHA.113.006699 601 42. Melo ES, Barbeiro DF, Gorjão R, et al. Gene expression reprogramming protects 602 macrophage from septic-induced cell death. Mol Immunol . 2010;47(16):2587-2593. 603 doi:https://doi.org/10.1016/j.molimm.2010.06.011 604 43. Rackov G, Hernández-Jiménez E, Shokri R, et al. p21 mediates macrophage 605 reprogramming through r egulation of p50-p50 NF- κ B and IFN- β . J Clin Invest . 606 2016;126(8):3089-3103. doi:10.1172/JCI83404 607 44. Sheikh F, Dickensheets H, Gamero AM, Vogel SN, Donnelly RP. An essential role for 608 IFN-β in the induction of IFN-stimulated gene expression by LPS in macrophages. J 609 Leukoc Biol. 2014;96(4):591-600. doi:https://doi.org/10.1189/jlb.2A0414-191R 610 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 18 45. Zahalka S, Starkl P, Watzenboeck ML, et al. Trained immunity of alveolar 611 macrophages requires metabolic rewiring and type 1 interferon signaling. Mucosal 612 Immunol. 2022;15(5):896-907. doi:10.1038/s41385-022-00528-5 613 46. Grigoriou M, Banos A, Filia A, et al. Transcriptome reprogramming and myeloid 614 skewing in haematopoietic stem and progenitor cells in systemic lupus 615 erythematosus. Ann Rheum Dis . 2020;79(2):242. doi:10.1136/annrheumdis-2019-616 215782 617 47. Mayer-Barber KD, Andrade BB, Oland SD, et al. Host-directed therapy of tuberculosis 618 based on interleukin-1 and type I interferon crosstalk. Nature. 2014;511(7507):99-103. 619 doi:10.1038/nature13489 620 48. Gui J, Zhao B, Lyu K, Tong W, Fuchs SY. Downregulation of the IFNAR1 chain of 621 type 1 interferon receptor contributes to the maintenance of the haematopoietic stem 622 cells. Cancer Biol Ther. 2017;18(7):534-543. doi:10.1080/15384047.2017.1345395 623 49. DeMerle KM, Royer SC, Mikkelsen ME, Prescott HC. Readmissions for Recurrent 624 Sepsis: New or Relapsed Infection?*. Crit Care Med . 2017;45(10). 625 https://journals.lww.com/ccmjournal/fulltext/2017/10000/readmissions_for_recurrent_s626 epsis__new_or_relapsed.14.aspx 627 50. Williams LM, Ricchetti G, Sarma U, Smallie T, Foxwell BMJ. Interleukin-10 628 suppression of myeloid cell activation — a continuing puzzle. Immunology. 629 2004;113(3):281-292. doi:https://doi.org/10.1111/j.1365-2567.2004.01988.x 630 51. Burke JD, Platanias LC, Fish EN. Beta Interferon Regulation of Glucose Metabolism 631 Is PI3K/Akt Dependent and Important for Antivi ral Activity against Coxsackievirus B3. 632 J Virol. 2014;88(6):3485-3495. doi:10.1128/jvi.02649-13 633 52. Leisching G, Yennemadi A, Gogan K, Keane J. Interferon α and β induce differential 634 transcriptional and functional metabolic phenotypes in human macrophages and blunt 635 glycolysis in response to antigenic stimuli. Eur J Immunol . 2024;54(9):2451032. 636 doi:https://doi.org/10.1002/eji.202451032 637 638 639 640 641 642 643 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 19 FIGURE LEGENDS 644 Figure 1. Flow cytometry phenotyping of HSPCs and mature immune cells in 645 peripheral blood of Surv and HD. 646 A. Box plots representing the absolute number of circulating HSPC subsets in age-matched 647 healthy donors (HD) and long-term sepsis survivors (Surv). 648 B. Box plots representing the absolute number of circulating immune cells in HD and Surv. 649 The differences between the two groups were tested with a Wilcoxon rank-sum test. * = p-650 value ≤ 0.05. 651 652 Figure 2. Transcriptional profiling of HSDMs from Surv and HD. 653 A. Volcano plot comparing LPS-treated HSDMs and untreated HSDMs from sepsis survivors 654 (Surv). Coloured dots indicate significant upregul ation in LPS-treated (purple, lfc > 1.5 and 655 adjusted p-value ≤ 0.05) or untreated (orange, lfc < 1.5 and adjusted p-value ≤ 0.05) cells. 656 B. Top 15 upregulated “biological process” pathways in Surv HSDMs stimulated with LPS 657 compared to untreated cells from the same pati ents. The colour of each dot represents the 658 Benjamini-Hochberg adjusted p-value, while the size of the dot represents the number of 659 genes enriched in each pathway. 660 C. Volcano plot comparing HSDMs derived from Surv compared to HD after LPS stimulation. 661 Coloured dots indicate significant upregulation in Surv (purple, lfc > 1.5 and adjusted p-value 662 ≤ 0.05) or HD (green, lfc < 1.5 and adjusted p-value ≤ 0.05). 663 D. Top 15 enriched terms from GSEA on the KEG G database, using differentially expressed 664 genes between LPS-stimulated HSDMs derived from Surv and HD. Dot size indicates the 665 adjusted p-value, and dot colour indicates t he normalised enrichment score (NES). Green 666 dots indicate terms significantly enriched in HSDMs from HD (i. e. NES < 0), while purple 667 dots indicate terms significantly enriched in HSDMs derived from Surv. 668 E. GSEA plot illustrating the enrichment of “toler izable” (orange line) and “non-tolerizable” 669 (green line) genes28 in LPS-stimulated HSDMs derived from Surv compared to HD. The 670 bottom portion of the plot shows the ranked po sitions of genes by differential expression, 671 with upregulated genes concentrated on the left (positive rank) and downregulated genes on 672 the right (negative rank). 673 674 Figure 3. Signalling pathways and transcription factor activity in HSDMs from Surv 675 and HD. 676 A. Enriched signalling pathways in Surv HSDMs stimulated with LPS compared to untreated 677 cells from the same donors. Dot size indicates the adjusted p-value, while dot colour 678 indicates the enrichment score (purple, ES > 0 - enriched in LPS-treated HSDMs; orange, 679 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 20 ES < 0 - enriched in untreated HSDMs). A black border around each dot indicates 680 statistically significant enrichments (adjusted p-value ≤ 0.05). 681 B. Transcription factor activity in Surv HSDMs stimulated with LPS compared to untreated 682 cells from the same donors. Each dot repr esents a transcription factor, with dot size 683 indicating the adjusted p-value, and dot colour indicating the activity score. Positive scores 684 (purple) indicate increased transcription factor activity in LPS-treated HSDMs, while negative 685 scores (orange) indicate increased transcription factor activity in untreated cells. A black 686 border around each dot indicates statistically significant enrichments (adjusted p-value ≤ 687 0.05). 688 C. Enriched signalling pathways in LPS-stimulat ed HSDMs derived from Surv versus HD. 689 Dot size indicates the adjusted p-value, while dot colour indicates the enrichment score 690 (purple, ES > 0 - enriched in Surv HSDMs; green, ES < 0 - enriched in HD HSDMs). A black 691 border around each dot indicates statistically significant enrichments (adjusted p-value ≤ 692 0.05). 693 D. Transcription factor activity in LPS-stimulated HSDMs derived from Surv versus HD. Each 694 dot represents a transcription fa ctor, with dot size indicating the adjusted p-value and dot 695 colour indicating the activity score. Positive scores (purple) indica te increased transcription 696 factor activity in Surv HSDMs, while negat ive scores (orange) indicate increased 697 transcription factor activity in healthy dono rs’ HSDMs. A black border around each dot 698 indicates statistically significant enrichments (adjusted p-value ≤ 0.05). 699 700 Figure 4. The effect of IFN β on differentiation and function of HSDMs from adult blood 701 donors. 702 A. Comparison of cell counts after 7 days of differentiation of HSPCs from adult blood 703 donors in the presence (red) or absence of IFN β (blue). Left: total cell count, tested with a 704 pairwise t-test. Right: fold-change compared to not treated (NT) cells from each donor, 705 tested with a pairwise t-test. * = p-value ≤ 0.05, ** = p-value ≤ 0.01. 706 B. Comparison of cell counts after 14 days of differentiation of HSPCs from adult blood 707 donors in the presence (red) or absence of IFN β (blue). Left: total cell count, tested with a 708 pairwise t-test. Right: fold-change compared to non-treated cells from each donor, tested 709 with a pairwise t-test. * = p-value ≤ 0.05, ** = p-value ≤ 0.01. 710 C. Volcano plot comparing IFN β -treated (red) and NT cells (blue). Dotted lines indicate 711 statistical significance thresholds (|FC| ≥ 1.5, p-value ≤ 0.05). 712 D. Percentage of early apoptotic, late apoptotic, live, and necrotic cells in NT (blue) and 713 IFNβ -treated cells (red) after 7 days of different iation. Differences between the two groups 714 were tested with a pairwise t-test. * = p-value ≤ 0.05. 715 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint 21 E. Top: geometric mean fluorescence intensity (GMFI) of CD38 in live, early-apoptotic and 716 late-apoptotic cells in NT sa mples. Differences between all groups was tested with a Tukey 717 post-hoc test. * = p-value ≤ 0.05. Bottom: representative histogram showing CD38 levels in 718 early-apoptotic and late-apoptotic cells in NT samples. 719 F. Top: GMFI of CD38 in live, early-apoptotic and late-apoptotic cells in IFN β -treated 720 samples. Differences between all groups was tested with a Tukey post-hoc test. * = p-value 721 ≤ 0.05. Bottom: representative histogram showing CD38 levels in early-apoptotic and late-722 apoptotic cells in IFNβ -treated samples. 723 G. Dot plot comparing the OCR-to-ECAR ratio in resting (left) and stressed (right) cells 724 differentiated in the presence (red) or absence (NT, blue) of IFN β . Differences between the 725 two groups were tested with a pairwise t-test. * = p-value ≤ 0.05. 726 H. Graph depicting the oxygen consumption rate (OCR) in HSDM differentiated in the 727 presence (red) or absence (NT, blue) of IFN β at basal level, and after addition of oligomycin, 728 FCCP, and a combination of rotenone and antimycin A (Rot/AA). 729 730 DATA AND CODE AVAILABILITY 731 The code used to analyse the bulk RNA-seq data and generate the figures is available at 732 https://github.com/Deusu/HSDM-sepsis. The RNA-seq data generated in this study is 733 available at Zenodo (10.5281/zenodo.14295543). 734 735 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 17, 2024. ; https://doi.org/10.1101/2024.12.14.628447doi: bioRxiv preprint