Neutrophil Extracellular Traps Contain Calprotectin, a Cytosolic Protein Complex Involved in Host Defense against Candida albicans

Constantin F. Urban, David Ermert, Monika Schmid, Ulrike Abu-Abed, Christian Goosmann, Wolfgang Nacken, Volker Brinkmann, Peter R. Jungblut, Arturo Zychlinsky•View original

OverviewBalancedalloy voice

Picture the scene: a neutrophil, a first responder of your immune system, bursts open, not in defeat, but in strategy. It spills out a sticky web of DNA studded with proteins. That web is a Neutrophil Extracellular Trap, or NET. Think of it as a booby-trapped fishing net: it snares microbes and holds them in a toxic embrace. For bacteria, that story is familiar. For fungi like Candida albicans, which can morph from a tiny yeast into a long filament that slips away from engulfment, the rules were murkier. Were NETs just chromatin shrapnel with a smear of histones, or something more deliberate? Urban, Ermert, Schmid, Abu-Abed, Goosmann and colleagues set out to map that trap. They induced NETs from healthy human neutrophils with a phorbol ester, washed the webs to shake off anything loosely stuck, and then did something clever: they digested the DNA backbone with DNase to release whatever was truly bound and ran that through a high-resolution proteomic pipeline—nano-liquid chromatography coupled to Matrix-Assisted Laser Desorption and Ionization, or MALDI, mass spectrometry. They only counted a protein as NET-associated if it appeared in material from at least two different donors. That filter mattered. It turned a mess of possibilities into a shortlist. The shortlist was surprisingly crisp. They detected twenty-four NET-associated proteins. Nine had been described on NETs before; fifteen were new to the NET club. And that roster looked the same from donor to donor. That’s the first pivot in our story: NETs are selective assemblies, not random protein pollution. You can hear it in the balance of what they found: five nuclear proteins, eight cationic granule proteins, and eleven cytoplasmic proteins that were neutral to acidic—telling us that charge alone isn’t the gatekeeper for binding to these webs. Here’s the second pivot. Yes, histones were abundant. In fact, histones H2A, H2B, H3, and H4 together made up about seventy percent of the NET protein mass. But they didn’t look like their nuclear selves. Urban and colleagues saw histones on NETs running a bit lighter—by roughly two to five kilodaltons—hinting at a modified state. And the makeup shifted: H3 and H4 were relatively less abundant than H2A and H2B compared to what you’d find in a nucleus. When they imaged the NETs at high magnification, they looked like smooth strands the size of stacked nucleosomes—on the order of five by ten nanometers—interspersed with flecks that likely carry granule proteins. Picture a bare chromatin cable with protein-studded knots. Before we jump to function, there’s an important control. NET preparations are notoriously hard to keep clean, so the team tracked potential contaminants like glyceraldehyde three-phosphate dehydrogenase, or GAPDH, and lactate dehydrogenase, or LDH. Those enzymes washed away and showed up in supernatants, not in the DNase-released NET fraction. Even after nine careful washes, a little calprotectin clung on, which is exactly the point: some things are bound, not just smeared on. They even quantified the bulk balance of the trap: protein to DNA came in around one point sixty-seven to one, a simple ratio that gives you a feel for how protein-dense the web is. Now, the star. Among those twenty-four proteins, one jumped out as both abundant and unexpected: calprotectin, the S100A8 and S100A9 dimer. It’s a cytosolic protein that lacks a secretion signal, better known for nutritional immunity, the way host cells starve microbes of essential metals by chelation. You wouldn’t necessarily predict it would decorate a DNA web. And yet, as Urban and colleagues showed, that’s exactly what happens during NETosis. They watched the timing unfold under the microscope. Early on, when neutrophils start to activate, classic granule proteins like myeloperoxidase and lactotransferrin spill into the extracellular space. Calprotectin waits. It surges later, peaking when the nucleus decondenses, the nuclear envelope breaks apart, and the chromatin leaves the cell—three to four hours after activation. It naturally rides out on the NETs and, crucially, it sticks there. Immunofluorescence lit up calprotectin right along the DNA and myeloperoxidase in those webs. How much of it ends up on the traps? When they split the system into three pools—the washed supernatant, the DNase-released NET fraction, and the leftover cell remnants—they found roughly a third of the detectable S100A9 was bound to NETs, with a comparable fraction floating free. In absolute terms, enzyme-linked assays showed about 170 nanograms per milliliter in the NET fraction and about 200 nanograms per milliliter as unbound calprotectin, which works out to roughly twenty-three percent NET-associated and twenty-eight percent unbound under those conditions. Put differently, once a neutrophil commits to NETosis, around half of its calprotectin becomes extracellular, and a big chunk of that is physically anchored to the DNA scaffold. That anchor matters because NET-bound calprotectin can touch the fungus; in experiments with NET fragments, the calprotectin on the trap grabbed onto Candida’s surface, whereas soluble calprotectin didn’t show the same stick. Okay, but does it kill? This is where the story gets sharp. In vitro, NETs reliably knocked down Candida albicans by about two orders of magnitude—a hundred-fold reduction in colony counts. When the team digested the DNA backbone with nuclease, the antifungal effect vanished. The trap has to be intact for the punch to land. Then they played with metals. Calprotectin is a metal chelator; it ties up zinc and manganese to starve microbes. When they added extra zinc or manganese to the NET and Candida co-culture, the antifungal effect collapsed. That’s a smoking gun for mechanism. They went a step farther and physically pulled calprotectin out of the NET protein mix. Antibodies against S100A8 and S100A9 scrubbed calprotectin away while leaving other proteins like lactotransferrin intact. After that immunodepletion, the NET fraction lost its growth-inhibiting power. Control depletions did nothing. That’s pretty conclusive: in these conditions, calprotectin is the principal antifungal effector in NETs. They also looked at genetics. NETs from mice lacking S100A9, which effectively deletes the S100A8 and S100A9 dimer, were far less antifungal than those from wild-type mice, even though both genotypes formed NETs just fine. Same trap, missing payload. It’s not just that calprotectin is there; the NET is presenting it in a way that matters. Hyphal forms of Candida, those long filaments that are notoriously tough to engulf, are actually potent inducers of NETs, and under the assay conditions, the traps still suppress hyphal growth. That’s a nice inversion of the usual problem. What the neutrophil can’t swallow, it can net, and then starve. Before we step into living tissue, one more housekeeping detail that makes this proteome believable. Some proteins reported by others as NET-associated didn’t show up by mass spectrometry here—histone H1, bactericidal permeability-increasing protein, the cathelicidin CAP-18, and pentraxin-3. It doesn’t mean they’re never there. Immunoblotting found a 25-kilodalton cleavage product of BPI, and CAP-18 and PTX-3 may be present at very low levels, loosely attached, or lost in preparation. This is the tradeoff in any proteomics experiment: sensitivity and stringency. Urban and colleagues erred on the side of only what we can call confidently across donors, and the result is a clean parts list that holds up across people. Now, take that mechanistic story into a mouse. The team wanted to know whether NET-bound calprotectin matters in vivo. So they used three Candida models—subcutaneous abscess, lung infection by intranasal inoculation, and bloodstream infection by tail vein—and compared wild-type mice to S100A9 knockouts. Remember, those knockouts lack functional calprotectin. In the skin, the difference wasn’t subtle. After a subcutaneous dose of five times ten to the seventh Candida, abscesses in calprotectin-deficient mice grew to about two hundred square millimeters, roughly double the area in wild-type animals, over days two to six. By day four, sixty percent of the knockout lesions ulcerated, with the necrosis punching up through the epidermis, while wild-type lesions stayed walled off under the skin. In nearly a third of the knockout mice, the infection even spilled out of the original site into surrounding tissue. By day eight, sizes converged, which tells you calprotectin is particularly important in the early containment phase. Importantly, both groups recruited neutrophils and made NETs in the lesions to a similar degree. The missing piece was the payload, not the delivery. In the lungs, the pattern held. After an intranasal hit with the same inoculum, knockout mice succumbed more readily than their wild-type littermates, with susceptibility differing at a p-value of 0.0001. When the team measured fungal burden, the knockouts carried significantly higher loads in their lungs; that comparison came in at a p-value of 0.048. And when Candida went into the bloodstream—half a million cells by tail vein injection—calprotectin-deficient mice were again more vulnerable, with survival curves separating at a p-value of 0.0159. Three tissues, the same theme: calprotectin helps hold the line. If you slice into those infected tissues and ask, "Are NETs really there, carrying calprotectin to the fungus?" the answer is yes. In the subcutaneous abscesses, histology showed web-like extracellular DNA surrounding fungal pockets, with myeloperoxidase and histones lighting up along the strands. Calprotectin colocalized right on that scaffold. In the lungs, similar webs stretched across bronchiolar spaces, again marked by myeloperoxidase and histones, again positive for S100A9. Under the electron microscope, you can literally see strands of the matrix draped over Candida, like sticky tinsel over a branch. It mirrors the in vitro picture: a DNA backbone festooned with proteins, pressing calprotectin into contact with the fungal surface. There’s a satisfying coherence between the bench and the animal. NET-associated calprotectin binds the fungus better than soluble calprotectin; in vitro, adding zinc or manganese or removing calprotectin erases the antifungal effect; in vivo, take calprotectin away and lesions spread, fungal loads climb, and survival drops. Meanwhile, across donors, the NET proteome stays stable. The pieces click. Two caveats are worth keeping in your mental margin. First, the in vitro NETs were induced with a potent phorbol ester, not the physiological cocktails neutrophils encounter in tissues. That’s a common compromise in NET studies, but it does leave open questions about how cues in real infections shape the exact proteome. Second, in living animals, calprotectin does more than just ride on NETs. It can act in the cytosol and the extracellular space independently of the DNA scaffold. Urban and colleagues are explicit about this: they don’t yet have a tool to isolate the exclusive contribution of NET-bound calprotectin from all the other places calprotectin could be working. So the in vivo data are strong on causation for calprotectin and strong on the presence of NETs, but they can’t atomize the fraction of the effect due specifically to calprotectin on NETs. Stepping back, what did this work change? It reframed NETs as organized devices. Predominantly modified nucleosomes, carrying a reproducible suite of proteins, with a star payload—calprotectin—that exploits the DNA scaffold to concentrate and localize antifungal chemistry. Not a random blast of gunk. A selective assembly. And against Candida albicans, especially its hyphal form, that assembly buys the host real time and space: a hundred-fold hit to growth in vitro, smaller and better-contained lesions in vivo, and a clear visual of the trap pressed against the pathogen. You can imagine the implications. If calprotectin is the main antifungal payload on NETs in these settings, assays that measure NET-bound calprotectin might tell you something about the state of antifungal defense in a lesion. Therapeutically, tuning metal availability is risky—pathogens and hosts both need zinc and manganese—but the principle is crisp: nutritional immunity isn’t just a diffuse cloud; it can be wired to a scaffold. For now, the takeaway is simpler. As Urban and colleagues showed, the immune system’s throw a net strategy isn’t a desperate last stand. It’s a targeted delivery system, and calprotectin is the hook on the line.

Think of it as a booby-trapped fishing net: it snares microbes and holds them in a toxic embrace. For bacteria, that story is familiar. For fungi like Candida albicans, which can morph from a tiny yeast into a long filament that slips away from engulfment, the rules were murkier.

Were NETs just chromatin shrapnel with a smear of histones, or something more deliberate?

Urban, Ermert, Schmid, Abu-Abed, Goosmann and colleagues set out to map that trap. They induced NETs from healthy human neutrophils with a phorbol ester, washed the webs to shake off anything loosely stuck, and then did something clever: they digested the DNA backbone with DNase to release whatever was truly bound and ran that through a high-resolution proteomic pipeline—nano-liquid chromatography coupled to Matrix-Assisted Laser Desorption and Ionization, or MALDI, mass spectrometry. They only counted a protein as NET-associated if it appeared in material from at least two different donors. That filter mattered. It turned a mess of possibilities into a shortlist.

The shortlist was surprisingly crisp. They detected twenty-four NET-associated proteins. Nine had been described on NETs before; fifteen were new to the NET club.

And that roster looked the same from donor to donor. That’s the first pivot in our story: NETs are selective assemblies, not random protein pollution. You can hear it in the balance of what they found: five nuclear proteins, eight cationic granule proteins, and eleven cytoplasmic proteins that were neutral to acidic—telling us that charge alone isn’t the gatekeeper for binding to these webs.

Here’s the second pivot. Yes, histones were abundant. In fact, histones H2A, H2B, H3, and H4 together made up about seventy percent of the NET protein mass.

But they didn’t look like their nuclear selves. Urban and colleagues saw histones on NETs running a bit lighter—by roughly two to five kilodaltons—hinting at a modified state. And the makeup shifted: H3 and H4 were relatively less abundant than H2A and H2B compared to what you’d find in a nucleus.

When they imaged the NETs at high magnification, they looked like smooth strands the size of stacked nucleosomes—on the order of five by ten nanometers—interspersed with flecks that likely carry granule proteins. Picture a bare chromatin cable with protein-studded knots.

Before we jump to function, there’s an important control. NET preparations are notoriously hard to keep clean, so the team tracked potential contaminants like glyceraldehyde three-phosphate dehydrogenase, or GAPDH, and lactate dehydrogenase, or LDH. Those enzymes washed away and showed up in supernatants, not in the DNase-released NET fraction.

Even after nine careful washes, a little calprotectin clung on, which is exactly the point: some things are bound, not just smeared on. They even quantified the bulk balance of the trap: protein to DNA came in around one point sixty-seven to one, a simple ratio that gives you a feel for how protein-dense the web is.

Now, the star. Among those twenty-four proteins, one jumped out as both abundant and unexpected: calprotectin, the S100A8 and S100A9 dimer. It’s a cytosolic protein that lacks a secretion signal, better known for nutritional immunity, the way host cells starve microbes of essential metals by chelation.

You wouldn’t necessarily predict it would decorate a DNA web. And yet, as Urban and colleagues showed, that’s exactly what happens during NETosis.

They watched the timing unfold under the microscope. Early on, when neutrophils start to activate, classic granule proteins like myeloperoxidase and lactotransferrin spill into the extracellular space. Calprotectin waits.

It surges later, peaking when the nucleus decondenses, the nuclear envelope breaks apart, and the chromatin leaves the cell—three to four hours after activation. It naturally rides out on the NETs and, crucially, it sticks there. Immunofluorescence lit up calprotectin right along the DNA and myeloperoxidase in those webs.

How much of it ends up on the traps? When they split the system into three pools—the washed supernatant, the DNase-released NET fraction, and the leftover cell remnants—they found roughly a third of the detectable S100A9 was bound to NETs, with a comparable fraction floating free. In absolute terms, enzyme-linked assays showed about 170 nanograms per milliliter in the NET fraction and about 200 nanograms per milliliter as unbound calprotectin, which works out to roughly twenty-three percent NET-associated and twenty-eight percent unbound under those conditions.

Put differently, once a neutrophil commits to NETosis, around half of its calprotectin becomes extracellular, and a big chunk of that is physically anchored to the DNA scaffold. That anchor matters because NET-bound calprotectin can touch the fungus; in experiments with NET fragments, the calprotectin on the trap grabbed onto Candida’s surface, whereas soluble calprotectin didn’t show the same stick.

Okay, but does it kill? This is where the story gets sharp. In vitro, NETs reliably knocked down Candida albicans by about two orders of magnitude—a hundred-fold reduction in colony counts.

When the team digested the DNA backbone with nuclease, the antifungal effect vanished. The trap has to be intact for the punch to land. Then they played with metals.

Calprotectin is a metal chelator; it ties up zinc and manganese to starve microbes. When they added extra zinc or manganese to the NET and Candida co-culture, the antifungal effect collapsed. That’s a smoking gun for mechanism.

They went a step farther and physically pulled calprotectin out of the NET protein mix. Antibodies against S100A8 and S100A9 scrubbed calprotectin away while leaving other proteins like lactotransferrin intact. After that immunodepletion, the NET fraction lost its growth-inhibiting power.

Control depletions did nothing. That’s pretty conclusive: in these conditions, calprotectin is the principal antifungal effector in NETs. They also looked at genetics.

NETs from mice lacking S100A9, which effectively deletes the S100A8 and S100A9 dimer, were far less antifungal than those from wild-type mice, even though both genotypes formed NETs just fine. Same trap, missing payload.

It’s not just that calprotectin is there; the NET is presenting it in a way that matters. Hyphal forms of Candida, those long filaments that are notoriously tough to engulf, are actually potent inducers of NETs, and under the assay conditions, the traps still suppress hyphal growth. That’s a nice inversion of the usual problem. What the neutrophil can’t swallow, it can net, and then starve.

Before we step into living tissue, one more housekeeping detail that makes this proteome believable. Some proteins reported by others as NET-associated didn’t show up by mass spectrometry here—histone H1, bactericidal permeability-increasing protein, the cathelicidin CAP-18, and pentraxin-3. It doesn’t mean they’re never there.

Immunoblotting found a 25-kilodalton cleavage product of BPI, and CAP-18 and PTX-3 may be present at very low levels, loosely attached, or lost in preparation. This is the tradeoff in any proteomics experiment: sensitivity and stringency. Urban and colleagues erred on the side of only what we can call confidently across donors, and the result is a clean parts list that holds up across people.

Now, take that mechanistic story into a mouse. The team wanted to know whether NET-bound calprotectin matters in vivo. So they used three Candida models—subcutaneous abscess, lung infection by intranasal inoculation, and bloodstream infection by tail vein—and compared wild-type mice to S100A9 knockouts. Remember, those knockouts lack functional calprotectin.

In the skin, the difference wasn’t subtle. After a subcutaneous dose of five times ten to the seventh Candida, abscesses in calprotectin-deficient mice grew to about two hundred square millimeters, roughly double the area in wild-type animals, over days two to six. By day four, sixty percent of the knockout lesions ulcerated, with the necrosis punching up through the epidermis, while wild-type lesions stayed walled off under the skin.

In nearly a third of the knockout mice, the infection even spilled out of the original site into surrounding tissue. By day eight, sizes converged, which tells you calprotectin is particularly important in the early containment phase. Importantly, both groups recruited neutrophils and made NETs in the lesions to a similar degree. The missing piece was the payload, not the delivery.

In the lungs, the pattern held. After an intranasal hit with the same inoculum, knockout mice succumbed more readily than their wild-type littermates, with susceptibility differing at a p-value of 0.0001. When the team measured fungal burden, the knockouts carried significantly higher loads in their lungs; that comparison came in at a p-value of 0.048.

And when Candida went into the bloodstream—half a million cells by tail vein injection—calprotectin-deficient mice were again more vulnerable, with survival curves separating at a p-value of 0.0159. Three tissues, the same theme: calprotectin helps hold the line.

If you slice into those infected tissues and ask, "Are NETs really there, carrying calprotectin to the fungus?" the answer is yes. In the subcutaneous abscesses, histology showed web-like extracellular DNA surrounding fungal pockets, with myeloperoxidase and histones lighting up along the strands. Calprotectin colocalized right on that scaffold.

In the lungs, similar webs stretched across bronchiolar spaces, again marked by myeloperoxidase and histones, again positive for S100A9. Under the electron microscope, you can literally see strands of the matrix draped over Candida, like sticky tinsel over a branch. It mirrors the in vitro picture: a DNA backbone festooned with proteins, pressing calprotectin into contact with the fungal surface.

There’s a satisfying coherence between the bench and the animal. NET-associated calprotectin binds the fungus better than soluble calprotectin; in vitro, adding zinc or manganese or removing calprotectin erases the antifungal effect; in vivo, take calprotectin away and lesions spread, fungal loads climb, and survival drops. Meanwhile, across donors, the NET proteome stays stable. The pieces click.

Two caveats are worth keeping in your mental margin. First, the in vitro NETs were induced with a potent phorbol ester, not the physiological cocktails neutrophils encounter in tissues. That’s a common compromise in NET studies, but it does leave open questions about how cues in real infections shape the exact proteome.

Second, in living animals, calprotectin does more than just ride on NETs. It can act in the cytosol and the extracellular space independently of the DNA scaffold. Urban and colleagues are explicit about this: they don’t yet have a tool to isolate the exclusive contribution of NET-bound calprotectin from all the other places calprotectin could be working.

So the in vivo data are strong on causation for calprotectin and strong on the presence of NETs, but they can’t atomize the fraction of the effect due specifically to calprotectin on NETs.

Stepping back, what did this work change? It reframed NETs as organized devices. Predominantly modified nucleosomes, carrying a reproducible suite of proteins, with a star payload—calprotectin—that exploits the DNA scaffold to concentrate and localize antifungal chemistry.

Not a random blast of gunk. A selective assembly. And against Candida albicans, especially its hyphal form, that assembly buys the host real time and space: a hundred-fold hit to growth in vitro, smaller and better-contained lesions in vivo, and a clear visual of the trap pressed against the pathogen.

You can imagine the implications. If calprotectin is the main antifungal payload on NETs in these settings, assays that measure NET-bound calprotectin might tell you something about the state of antifungal defense in a lesion. Therapeutically, tuning metal availability is risky—pathogens and hosts both need zinc and manganese—but the principle is crisp: nutritional immunity isn’t just a diffuse cloud; it can be wired to a scaffold.

For now, the takeaway is simpler. As Urban and colleagues showed, the immune system’s throw a net strategy isn’t a desperate last stand. It’s a targeted delivery system, and calprotectin is the hook on the line.

Related lectures