The human spleen is the main secondary lymphoid organ involved in the control of systemic infection¹, particularly through its critical role in the removal of encapsulated bacteria from the circulation². Loss of the spleen by splenectomy, and other forms of functional asplenia, expose patients to the risk of overwhelming sepsis, most commonly caused by capsulated bacteria^{3, 4, 5}. This has led to the inclusion of asplenia as a key indication for vaccination in many countries⁶. Despite this the pneumococcus remains the leading cause of post-splenectomy sepsis⁷. While the processes of bacterial clearance by the human spleen are critically understudied, the mechanisms by which the spleen removes Plasmodium falciparum-infected red blood cells are well documented due to exploitation of the translational ex vivo human spleen prefusion models⁸.

In most human tissues, tissue resident macrophages express the mannose receptor (MRC1/CD206)⁹. However, in the human spleen neither the CD68⁺CD163⁺ red pulp macrophages (RPMs) nor the CD68⁺CD169⁺ perifollicular capillary sheath macrophages (PCSAMs) express the mannose receptor⁹. Instead, the CD206 receptor is expressed by the sinusoidal lining littoral cells, which make up the structure of the human spleen red pulp^{9, 11}. Splenic sinuses are the walled irregular structures draining the open circulation of the human splenic red pulp and are lined by sinus endothelia alternatively named sinusoidal lining cells (CD206⁺, LYVE-1⁺, CD141⁺)^{11, 12, 13}. The rationale behind this variation in CD206 expression in the human spleen is still unknown. The CD206 mannose receptor is a key molecule involved in the binding to bacteria, viruses and parasite-infected cells playing a key role in the interface between the human host and microbes. Binding of the mannose receptor to bacterial surface sugars, including capsular polysaccharides of Streptococcus pneumoniae and Klebsiella pneumoniae lipopolysaccharide (LPS)¹⁰, is one of the main mechanisms by which macrophages recognize these pathogens; and endocytosis of the receptor is associated to the uptake and the subsequent killing of the bacteria inside the macrophage compartments¹⁴. In addition, CD206 has been shown to mediate uptake of dengue virus and HBV into host cells^{15, 16} and recent work on the malaria parasite P. falciparum showed that infected red blood cells are efficiently captured via their surface polysaccharides by the CD206⁺ sinusoids, suggesting a crucial role of sinusoidal cells in the reduction of pathogen load¹⁷.

Investigation of host-pathogen interactions in the spleen, using murine infection models, non-human primate infections and ex vivo porcine spleen perfusion, revealed that systemic S. pneumoniae infections were predominantly cleared by splenic red pulp macrophages^{18, 19, 20, 21}. While the spleen was the primary organ responsible for the clearance of bacteria, the rare occurrence of splenic replication in permissive macrophages facilitated re-seeding of bacteria to the bloodstream and initiation of systemic infection^{18, 19}. The translational ex vivo human spleen perfusion model confirmed detection of bacterial clusters in human splenic macrophages²¹. However, the mannose receptor expression on sinusoidal endothelial cells and not macrophages raised the question about pathogen clearance mechanisms. Using the unique translational platform of human organ ex vivo perfusion alongside primary splenic cell co-cultures of macrophages and sinusoids, we tested the hypothesis that CD206-mediated capture by sinusoidal endothelial cells underlies the human spleen’s distinctive ability to filter, retain, and clear encapsulated bacteria. This host-species-specific adaptation sheds new light on splenic innate immune surveillance and has important implications for vaccine efficiency and the development of host-targeted therapies.

To determine the impact of CD206 expression on sinusoidal cells on the clearance of encapsulated bacteria in the human spleen, we applied a translational ex vivo human spleen perfusion model²¹ using organs sourced through the TIMID trial (REC 18/EM/0057; ClinicalTrials.gov NCT04620824) and primary cells through the MOSIE trial (CE: 668/2023/Sper/AOUBo). Microscopy mapped the three red pulp markers, CD163, CD169, and CD206, to three distinct cell types: CD68⁺CD163⁺ red pulp macrophages (RPMs), CD68⁺CD169⁺ perifollicular capillary sheath–associated macrophages (PCSAMs), and sinusoid lining littoral CD206⁺LYVE-1⁺CD31⁻ cells^{9, 22} (Fig. 1A–1B; Fig. S1A1-S1B). High-content scanning fluorescent microscopy revealed that RPMs and sinusoidal cells dominate the red pulp¹¹ (Fig. 1A), while CD169⁺ macrophages clustered in sheaths around perifollicular capillaries, often forming ring-like structures adjacent to white pulp follicles²² (Fig. 1B). CD206⁺ sinusoids, CD163⁺ RPMs, and CD169⁺ macrophages occupied on average 36.5% (SD 4.13), 27.7% (SD 6.1), and 1.3% (SD 1.6) of the total splenic section area respectively, with ranges of 30.3–42.0%, 19.9–35.8%, and 0.07–4.6% (Fig. 1C). Marker abundance varied between spleens, up to 1.2-fold for CD206⁺, 2.7-fold for CD163⁺, and intriguingly 12.1-fold for CD169⁺ macrophages (Fig. 1D–1F), the latter reflecting differences in follicle density and microanatomy (Fig. S1C1–S1C3). Importantly, serial biopsies from the same spleen at different infection stages showed no variation in CD163 or CD206 levels (Fig. S1D1–S1D2), confirming the stability of these populations during perfusion and infection.

Taken together, these observations reveal a compartmentalized immune architecture in human spleen, with CD206⁺ sinusoidal cells forming an extensive, blood-facing capture network; CD163⁺ macrophages distributed throughout the red pulp for removal of pathogens, senescent cells and debris; and rare CD169⁺ macrophages positioned around perifollicular regions, where they may regulate antigen entry into the white pulp²³. This spatial arrangement underscores a division of labour in pathogen surveillance, placing CD206⁺ cells in an ideal position to intercept microbes during their first passage through the splenic open circulation.

Ex vivo organ perfusion.

Using the TIMID trial, human spleens were sourced for ex vivo normothermic organ perfusion to assess the splenic antibacterial clearance capacity. After cannulation and anticoagulant perfusion, spleens were transported on ice, connected to a normothermic circuit, and perfused with polymerised haemoglobin as an oxygen carrier. Systemic infection was simulated by introducing S. pneumoniae directly into the perfusion liquid mimicking hematogenous invasive human infections (Tab. S1), and serial biopsies and perfusate samples were taken over time to monitor the infection dynamics²¹. Across experiments, the spleen demonstrated a robust filtration capacity, removing over 90% of the inoculum within 60 minutes (Fig. 2A1-2B2, Fig. 2D1-2E2). Clearance kinetics were consistent whether challenged with individual strains or serotype mixtures (Tab. S1, Tab. S2). With limited organ availability, mixed infections guaranteed a more controlled experimental set-up allowing for both simultaneous testing of virulent and avirulent serotypes while minimising any potential impact of type-specific immunity of organ donors (Tab. S3). At a high-dose (1x10⁸ cumulative CFU) of five equally counted serotypes (2, 4, 5, 6B, 19F), clearance was equally efficient across strains, regardless of invasive potential (Fig. 2A1–2B2)²⁴. Similarly, the association of the bacteria with CD163⁺ RPMs and the less abundant CD169⁺ macrophages, based on image analysis, did not vary among serotypes (Fig. S2A1-S2A2). Neither the rapidity nor the extent of bacterial clearance in the human spleen was affected by the donor-derived level of antibacterial antibodies in the perfusion fluid (Tab. S3). At a lower dose (1x10⁷ cumulative CFU), all serotypes were cleared from both perfusate and tissue within 30 minutes (Fig. 2D1–2E2). These kinetics underscore the spleen’s extraordinary capacity for rapid, non-discriminatory removal of encapsulated bacteria, even without high antibody titres. This observation aligns with the clinical reality of overwhelming post-splenectomy infection (OPSI), where the absence of splenic tissue dramatically impairs early pathogen removal²⁵. The uniformity of clearance across serotypes highlights a fundamental property of splenic immunity: its filtering capacity operates independently of serotype-specific immune history, a fact critical to understanding infection risk in asplenic patients²⁵.

The mannose receptor (CD206) is a C-type lectin that binds high-mannose structures found on many pathogens, including the capsules of S. pneumoniae and LPS of K. pneumoniae^{10, 26, 27}. In most tissues, CD206 is expressed by macrophages and dendritic cells and mediates bacterial binding and uptake^{10, 28}. In the human spleen, however, macrophages lack CD206 entirely (Fig. 1A–1B, Fig. S2B-S2C3), despite the spleen being the primary site for pneumococcal clearance. Instead, CD206 is confined to sinusoidal lining cells, a distribution that raises an important mechanistic question: does splenic clearance rely on CD206-mediated capture by these sinusoidal cells rather than by macrophages?

To address this, we built on in vitro evidence that free mannose can block CD206–pathogen binding¹⁰. In the ex vivo perfusion model, spleens were pre-treated with 5 mM mannose alongside the 10–20 mM glucose required for host and bacterial metabolism. When challenged with a high pneumococcal dose (1x10⁸ CFU), mannose-treated spleens failed to clear the bacteria, instead, counts instead accumulated progressively in both perfusate and tissue (Fig. 2C1-2C2). At a lower challenge dose (1x10⁷ CFU), usually cleared within minutes, killing was completely abolished; bacteria instead proliferated rapidly (Fig. 2F1-2F2). These results identify the carbohydrate-binding activity of CD206 on sinusoidal lining cells as an essential initial step in pathogen clearance. Since the presence of glucose in the perfusion liquid and the cell culture medium blocks completely the pneumococcal mannose metabolism through carbon catabolite repression²⁹, the observed escape from host-mediated clearance cannot be ascribed to an effect of mannose on the bacteria. Rather, it reflects direct interference with receptor-mediated capture.

To complement the quantification of viable bacteria, we used high-content scanning fluorescence microscopy to examine the association of pneumococci with splenic macrophages and sinusoidal cells in tissue sections obtained during ex vivo perfusion. This analysis reinforced the clearance findings: across all time points, 16–21% of tissue-associated bacteria colocalized with CD206 (Fig. 3A1), a proportion that remained stable despite a 90% reduction in total bacterial load during the first hour (Fig. 2). Correspondingly, the area of sinusoidal tissue occupied by bacteria showed no decline (Fig. 3A2). Three-dimensional confocal reconstructions revealed discrete clusters of CD206 in direct contact with bacteria (Fig. 3B1-3B2). In mannose-treated spleens, bacterial association with CD206 fell sharply at early time points (Fig. 3A1), coinciding with higher viable counts and broader sinusoidal distribution later (Fig. 3A2).

To determine the fate of these captured bacteria, we performed wheat germ agglutinin (WGA) staining, which confirmed that CD206⁺ sinusoidal cells do not internalise bacteria (Fig. 3C1-3C3). Instead, captured bacteria remained extracellular, suggesting that these cells act as stationary traps. Additional quantification showed that 20% of pneumococci colocalised with CD163⁺ RPMs at 30 min, 2 h, and 5 h post-infection (Fig. 3D1), and that mannose treatment significantly reduced this association (Fig. 3D1). Although CD169⁺ macrophages were less abundant, their bacterial association was also diminished by mannose (Fig. 3E1-3E2).

To provide in vitro evidence for the phenotypes observed during ex vivo spleen perfusion, primary adherent cell cultures were established from human spleen homogenates. Primary spleen cultures contained 57% CD206⁺ sinusoidal cells and 35% CD163⁺ macrophages (Fig. 4A1), often linked by extensions (Fig. 4A2). Only CD14⁺ cells comprised only 1.4%, indicating rare monocytes, possibly overlapping with the CD206⁺ population³⁰.Low-MOI infections showed that 2.1% of bacteria attached to CD206⁺ cells (Fig. 4B1) and 0.6% to CD163⁺ macrophages (Fig. 4B2). Mannose or anti-CD206 antibodies reduced bacterial binding to CD206⁺ cells (Fig. 4B1) but not to CD163⁺ macrophages (Fig. 4B2). The expression of CD206 on primary sinusoidal cells, and not macrophages, was confirmed by CD206⁺ co-expression with the sinusoidal marker LYVE-1 (Fig. 4C). Three-dimensional reconstructions confirmed that pneumococci remained on the surface of CD206⁺ cells (Fig. 4D1–4D2), while bacteria appeared intracellular only in macrophages (Fig. 4E1–4E2). At MOI 10, bacterial counts dropped sharply within 30 min for both TIGR4 and D39 serotypes (Fig. 4F1, 4G1), but this bactericidal activity was lost when CD206 was blocked by mannose or antibodies (Fig. 4F1, 4G1). In contrast, when cultures were challenged with a non-encapsulated derivative of the TIGR4 strain, over 99% of bacteria were eliminated within 30 minutes (Fig. 4F2), and this killing was unaffected by mannose supplementation (Fig. 4F2). The same behaviour was observed for a rough non-encapsulated D39 strain (Fig. 4G2). These results strongly suggest that CD206-mediated bacterial attachment to sinusoidal cells is essential for enabling subsequent macrophage-mediated killing of encapsulated, sugar-coated pathogens. The control experiments with the non-encapsulated strains reinforce the conclusion that the CD206-dependent interaction between sinusoidal cells and macrophages is not required for the clearance of un-encapsulated bacteria. In addition to pneumococcal capsules, CD206 had been shown to bind also K. pneumoniae LPS ¹⁰. Repeating the bactericidal assay with capsulated K2 capsule K. pneumoniae and K1 capsule Escherichia coli, confirmed CD206-dependence for any bactericidal activity of macrophages (Fig. 4H1-4H2). These in vitro results fully confirm the observations during organ perfusion, indicating that CD206-mediated bacterial attachment to sinusoidal cells is essential for enabling subsequent macrophage-mediated killing of encapsulated pathogens.

These observations support a sequential “handoff” model: CD206⁺ sinusoidal cells capture encapsulated bacteria without internalisation, then present them to macrophages for phagocytosis and killing. Blocking CD206 prevents this initial capture and reduces downstream bacterial association with macrophages. Importantly, this arrangement mirrors receptor-mediated antigen transfer in other lymphoid organs³¹ but it is specialised for bloodstream pathogen clearance.

Unlike the traditional view that attributes CD206-mediated clearance to macrophages^{9, 27}, our data show that in the human spleen, these cells are non-phagocytic stromal elements. Functionally, they behave like biological flypaper, retaining pathogens at the sinusoidal lining surface until macrophages, particularly CD163⁺ and CD169⁺ subsets, engage them (Fig. S2B-S2C3). The mannose-inhibitable spectrum of this capture extends beyond pneumococcus to multiple encapsulated species, echoing mechanical retention of poorly deformable cells in splenic sinuses^{32, 33} but based on biochemical glycan recognition.

The therapeutic implications are twofold: enhancing CD206-mediated capture could strengthen host defences, while systemic mannose administration, as in some urinary tract infection (UTIs) treatments³⁴, could inadvertently impair splenic clearance. Our findings also clarify that in the spleen, CD206 is not a macrophage marker but a functional receptor on sinusoidal cells. This corrects earlier assumptions that CD206-mediated phagocytosis in spleen was macrophage-driven^{9, 35} repositioning these cells as dedicated antigen-capturing and presenting cells. By capturing encapsulated bacteria via mannose recognition, these sinusoidal cells transform a microbial evasion strategy, the anti-phagocytic capsule, into a vulnerability. The concept parallels earlier work showing that erythrocytes with exposed mannoses, including P. falciparum–infected cells, are preferentially retained and cleared¹⁷. Our results extend this to bacteria, demonstrating that the spleen uses mannose recognition to detect and trap both infected cells and encapsulated microbes. In summary, the human spleen is not just a passive filter but an active immune organ combining physical, biochemical, and receptor-mediated mechanisms to eliminate bloodstream threats. Redefining CD206⁺ sinusoidal cells as non-phagocytic, but indispensable for microbial capture, offers a new perspective on splenic immunity and opens the door to targeted interventions that reinforce this early line of defence.

Mechanism of pathogen clearance by both CD163⁺ and of CD169⁺ macrophages.

The two dominant splenic macrophage subsets, CD163⁺ RPMs and the less abundant CD169⁺ PCSAMs, operate in distinct but complementary niches²². Quantitative colocalization revealed that 15–30% of bacteria were associated with CD163⁺ macrophages, while 4–22% associated with CD169⁺ macrophages (Fig. 5A1-5A2). Notably, these proportions were consistent between early (30 min) and late (5 h) time points, despite > 90% clearance of bacteria from the perfusate within the first hour. In addition, microscopy analyses of tissue sections with high or low bacterial loads revealed a consistent ratio of bacteria-to-macrophage association across both CD163⁺ and CD169⁺ populations (Fig. 5B1-5B2), suggesting that even at high challenge doses, macrophage capacity for bacterial clearance was not saturated. The ratios remained consistent across different bacterial loads and serotypes (2, 4, 19F) (Fig S2A1-S2A2), indicating stable, load-independent pathogen engagement by these macrophage subsets. When fluorescent beads were perfused instead of live bacteria (Fig. S3A-S3C), macrophage uptake was slower, with association rates increasing progressively over time (Fig. 5A1–5A2). This slower uptake highlights the dynamic responsiveness of splenic macrophages to live microbes versus inert particles, reflecting an ability to accelerate engagement when genuine pathogens are detected. While these data do not allow quantification of the relative bactericidal contributions of each macrophage subset, they indicate that both CD163⁺ and CD169⁺ macrophages consistently retained similar bacterial loads over time, implying a comparable role in pneumococcal capture and clearance.

To investigate the mechanisms of bacterial killing by human spleen tissue-resident macrophages, spleen sections, at different time points of infection (from 0 to 6 h) were stained for the phagosome maturation marker LAMP1. LAMP1 staining, marking lysosomal compartments, revealed a progressive increase in infected spleens, but not in uninfected controls (Fig. 5C). This signal was localised mainly within macrophages (Fig. 5D1–5D2), consistent with active lysosomal engagement. Interestingly, minimal LAMP1–pneumococcus colocalization was observed, indicating that at early time points, bacteria-containing phagosomes may not yet have matured into fully degradative compartments (data not shown). Overall, the time-dependent increase in LAMP1 levels supports the conclusion that human spleen macrophages remain engaged in active phagocytosis during our ex vivo perfusion model and exhibit dynamic features of the phagocytic process.

Apoptosis has long been recognized as a key mechanism by which human alveolar macrophages eliminate pneumococci³⁶. To assess whether this mechanism is also active in human splenic macrophages, tissue sections were stained for cleaved caspase-3, the activated form of the caspase responsible for executing apoptotic proteolysis^{37, 38}. The infection triggered an early apoptotic response. Within 60 minutes, cleaved caspase-3 levels rose rapidly, over 10-fold in both CD163⁺ RPMs and in CD169⁺ macrophages (Fig. 5E1–5E2), with the latter showing 10x higher per-cell signal (Fig. S2D1–S2D2). In bead-perfused controls, this apoptotic rise was slower and less pronounced, suggesting that live bacterial infection accelerates cell death programmes. M30 staining, indicative of caspase-3–cleaved cytokeratin-18^{39, 40}, increased steadily over 5–6 hours in both subsets (Fig. 5F1–5F2), with the M30-positive macrophage area expanding accordingly. Mannose-treated and untreated spleens showed similar trends and were pooled in the graphs (Fig. S2E1–S2E2).

Considering that a human spleen contains approximately 2x10¹⁰ macrophages⁴¹, we found that at the 1-hour mark, 0.65% of CD163⁺ macrophages were positive for cleaved caspase-3 (Fig. 5E1–5E2). This corresponds to roughly 7x10⁷ to 3x10⁸ apoptotic cells per spleen; numbers that closely match the bacterial challenge dose of 1x10⁸ CFU. This correspondence suggests a direct link between the quantity of phagocytosed bacteria and the induction of macrophage apoptosis.

The lysosomal activation observed after bacterial uptake, marked by increased LAMP1 expression, confirms that splenic macrophages engage in canonical phagolysosomal killing⁴². However, the concurrent detection of early cleaved caspase-3 within just 60 minutes, followed by M30 expression at 5–6 hours, indicates that bacterial clearance in the spleen is closely coupled to programmed cell death. This differs from the delayed apoptosis reported in other macrophage populations, such as alveolar macrophages, where cell death typically occurs in the resolution phase of infection^{36, 43}. The spleen’s context may explain this difference. As a blood-filtering organ, it is uniquely exposed to high concentrations of circulating pathogens. In such an environment, rapid apoptosis following phagocytosis could serve as both a terminal killing step and a containment measure, ensuring that intracellular pathogens cannot persist within surviving macrophages. This swift apoptotic turnover may also act as a chemotactic signal to recruit additional phagocytes, maintaining clearance efficiency even, as individual macrophages are lost. The timing of this response is likely critical for neutralising pathogens capable of manipulating macrophage survival. Neisseria meningitidis, for example, can delay apoptosis via nitric oxide detoxification⁴⁴, prolonging its intracellular survival. The early initiation of apoptosis in splenic macrophages could therefore represent an evolved countermeasure against such strategies, an immunological “self-destruct” that denies bacteria a long-term niche.

Our findings support a revised model of splenic antibacterial defence in which CD206⁺ sinusoidal cells and macrophages act in coordinated sequence. CD206⁺ lining cells first immobilise encapsulated bacteria via mannose-dependent glycan recognition without internalisation, preventing their return to circulation and enabling targeted clearance. CD163⁺ and CD169⁺ macrophages then receive these retained bacteria, phagocytose them, initiate lysosomal degradation, and undergo rapid apoptosis, ensuring efficient destruction while preventing intracellular persistence and systemic re-seeding.

This architecture bears resemblance to antigen-handling systems in lymph nodes, where lymphatic endothelial cells and subcapsular sinus macrophages coordinate pathogen retention and transfer³¹. However, the splenic system is tuned for speed and broad-spectrum efficiency, particularly against encapsulated bacteria that evade complement-mediated lysis and antibody opsonisation. In effect, the capsule, a key virulence factor, becomes a vulnerability when confronted with CD206-mediated capture. The efficiency of this system underscores why loss of splenic function, whether through splenectomy, infarction, or disease-related atrophy ^{3, 4, 5}, carries such a high risk of overwhelming infection (OPSI), particularly from encapsulated organisms²⁵. These patients lack both the macrophage-rich red pulp and the specialised sinusoidal cell network, severely compromising their capacity to trap and neutralise bloodstream bacteria. Functional hyposplenism, as seen in sickle cell disease or celiac disease, often involves reduced sinusoidal cell density, further increasing vulnerability to severe bacterial infections^{45, 46}.

Spleen ex vivo perfusion, used for malaria⁸,¹⁷ and proposed for bacterial studies²¹, has limitations, some linked to reperfusion injury (e.g., high cytokines and soluble proteins) (Fig. S4)⁴⁷, whose impact we did not assess. Recognising the central role of CD206-mediated capture suggests new interventions, such as mannosylated nanoparticles to enhance vaccine delivery⁴⁸ or strategies to preserve/replace macrophages to boost killing capacity. Conversely, systemic free mannose, used to prevent UPEC adhesion³⁴, could transiently inhibit splenic CD206, impairing clearance of encapsulated pathogens. This risk may be important in patients with complicated UTI, concurrent bacteraemia, or reduced splenic reserve, warranting clinical caution.