Infectious-disease specialist Mihai Netea has a small round scar on his upper left arm: the telltale blemish of a Bacillus Calmette-Guérin (BCG) vaccination against tuberculosis that he received as a baby in Romania. The country is a high-risk area for Mycobacterium tuberculosis infections, and the vaccine would have protected Netea against the disease for more than a decade. It might even have protected him, albeit for a shorter time, against a range of other infections. And now Netea, who runs a lab at Radboud University in the Netherlands, and other immunologists are beginning to understand why.

A few years ago, Netea was investigating how the immune response to M. tuberculosis changes following BCG vaccination. One day his group got a bizarre result in a control experiment: blood samples from Radboud students who’d received BCG vaccinations in preparation for...

Netea was stumped. Vaccines are designed to generate pathogen-specific immunity by presenting adaptive immune cells, B and T lymphocytes, with a harmless version of a particular offender—in this case, an attenuated live bovine tuberculosis bacillus. So how could BCG vaccination cause the same heightened response to an unrelated pathogen—indeed, one that wasn’t even in the same domain of life?

After a little more digging, Netea got his answer. In 2012, his team showed that innate immune cells called monocytes, which differentiate into macrophages and recognize, attack, and digest pathogens, were responsible for this off-target protection. These nonspecific, first-responder inflammatory cells—thought to react similarly to all infections, old or new—were somehow being primed by the vaccination to respond more vigorously to secondary exposure, even to an unrelated microbe.1

The discovery adds to a growing body of evidence that, like B and T lymphocytes, certain innate immune cells are influenced by the host’s history of pathogen exposure. A few years before Netea’s discovery, another type of innate responder called a natural killer (NK) cell had been found to have pathogen- or molecule-specific memory, much like the vertebrate adaptive immune system.2,3

“We’re blurring the distinction between innate and adaptive immunity,” says immunologist Andrew Lichtman of Harvard Medical School. “You can’t now say dogmatically, ‘There is no memory in innate cells.’”

Challenging textbook immunology

The idea that innate cells might have an ability to remember pathogens is practically heresy for some researchers, says immunologist Lewis Lanier of the University of California, San Francisco, who discovered a form of NK cell memory. “People believe that only B and T cells could be smart and have memory,” he says. “It is almost like religion.”

We’re blurring the distinction between innate and adaptive immunity.—Andrew Lichtman,
Harvard Medical School

As they mature, B and T lymphocytes rearrange the genes for the surface molecules that detect invaders, allowing the cells to generate a practically limitless supply of pathogen-recognition proteins (antibodies in B cells; T-cell receptors in T cells). Those cells whose antibodies or receptors best fit a given pathogen will proliferate to fight the invader upon initial infection and differentiate into memory cells that stand by for reinfection. Such cells can remain in the body for many years, even a lifetime, ready to expand again should the same pathogen return. Innate cells, on the other hand, were simply believed to attack any threatening pathogen, in order to keep the host alive until the adaptive B and T cells could launch their targeted counterstrikes.

It was this classical view of immunity that so flummoxed Netea when he saw the immune response to C. albicans in the students’ blood samples. B and T cells shouldn’t be able to respond more vigorously to a novel pathogen, and innate immune cells shouldn’t be affected by vaccination at all. But he wasn’t the first one to notice the phenomenon. “To my surprise—maybe it shouldn’t have been a surprise—there were some very old studies” that documented improved off-target immunity following vaccination.

In the 1940s and ’50s, controlled trials in the U.S. and U.K. suggested that the BCG vaccine offered nonspecific protection against a suite of infections, reducing deaths in children from diseases other than tuberculosis by 25 percent.4 Experiments in mice over the next 40 years showed that BCG vaccination prompted protection against C. albicans, Staphylococcus aureus, Listeria monocytogenes, and the parasitic trematode Schistosoma mansoni, as well as the immunization’s intended target, M. tuberculosis. Similarly, injection of attenuated C. albicans not only protected against lethal C. albicans infection, but also against infection with S. aureus bacteria.5

Inspired by these old studies, Netea tested BCG vaccination or injection of nonlethal C. albicans in mice that lacked B cells and T cells. Sure enough, either injection could ramp up protection against not only its intended target but also an unrelated pathogen. On the other hand, mice that were defective at producing monocytes—and their derivative macrophages—did not display increased immune protection, suggesting these cells were at the root of this immune priming.1,6

A few years earlier, roughly 5,000 miles away in California, Lanier’s group was discovering that natural killer (NK) cells also have an unexpected ability to learn from past infection. It started with a simple question from postdoc Joseph Sun: What happens to NK cells after they clear a virus? “It was right after we had found the receptor for cytomegalovirus (CMV),” which is carried by some NK cells, allowing them to recognize and attack the infectious agents, recalls Lanier. The question made Lanier stop and rethink the field’s assumption that NK cells live for only a couple of weeks—“that they came out of the bone marrow, matured, did their thing, died, and were replaced by a fresh set of NK cells two weeks later,” says Lanier. As the researchers looked a bit closer, they realized that the two-week estimate was based on old experiments that had calculated merely the average NK-cell turnover rate. “Nobody had ever tracked NK cells that had encountered a pathogen,” explains Lanier.

To find out what happened, he and Sun transferred mature NK cells that expressed the CMV receptor into mice that lacked the receptor, allowing the researchers to easily track the transferred cells. Then they infected the mice with CMV. “We did it open-ended,” says Lanier. “We thought maybe in two weeks [the CMV-receptor–equipped NK cells] will all be gone.” But they weren’t. “The first surprise was that the cells actually expanded [in number] pretty enormously, like a T cell would,” says Lanier. “But what really shocked me was that [Sun] came back into my office two months later and said, ‘Those cells we put in that responded to the virus—they’re still there!’”

Most NK cells do, as historically believed, live for just a couple of weeks, being constantly replenished as they monitor the body for invaders. But Lanier’s work showed that, when challenged by infection, some of these cells can differentiate into long-lived “memory” NK cells, reminiscent of memory cells of the adaptive immune system. Sure enough, when the team infected the mice with CMV a second time, “we showed they worked even better,” he says: more mice survived.2

Meanwhile, up in Boston, Ulrich von Andrian’s Harvard team was characterizing a form of NK-cell memory that mediated contact hypersensitivity—a heightened immune response upon repeat exposure to chemicals, metals, or other small molecules associated with host proteins (haptens). Simply put: an allergy. Just like long-term immunity to pathogens, contact hypersensitivity was thought to require the adaptive arm of the immune system.3 But the group showed that it could occur in mice in the absence of B and T lymphocytes. The researchers went on to demonstrate that NK cells from a mouse repeatedly exposed to a particular chemical conferred their sensitivity to a recipient mouse that had never been exposed.

In 2011, Netea came up with a name for this ability of both monocytes and NK cells to remember an infection (or hapten) and react more vigorously to subsequent exposure: trained immunity.7 Lanier, however, prefers the term “innate memory.”

“I’ve stuck with the word ‘memory’ despite getting heat from the people who . . . think it can only apply to B and T cells,” he says. “I do it just to stir them up!”

Monocytes on guard

MONOCYTE TRAINING: Monocytes and their derivative macrophages recognize a variety of different pathogens via receptors that bind molecules common to a range of microbes—such as peptidoglycans present on most bacteria and β-glucans found on many fungi and bacteria. Exposure to certain pathogens, such as Mycobacterium tuberculosis, prompts the cells to adopt some macrophage-like characteristics and produce inflammatory cytokines such as TNFα and IL6. This so-called monocyte training results in a heightened immune response upon reexposure to the pathogen.
See full infographic: JPG | PDF
While it’s becoming clear that innate immune cells do more than just blindly attack all invaders, what’s unclear is how infections alter the function of these cells, and how those changes result in a more powerful immune response.

Monocytes and their derivative macrophages recognize a variety of different pathogens via receptors that bind molecules common to a range of microbes—such as peptidoglycans present on most bacteria and β-glucans found on a variety of fungi and bacteria. To figure out how such monocyte receptor activation leads to an invigorated immune response, Netea trained monocytes in culture, exposing them to M. tuberculosis, C. albicans, or β-glucan. Such exposure prompted the cells to adopt some macrophage-like characteristics, expressing certain macrophage surface markers and producing inflammatory cytokines. (See illustration.) Subsequent exposure to M. tuberculosis or C. albicans boosted cytokine production in the cells. Comparing trained with untrained cells, the researchers found alterations in gene expression and epigenetic marks across the genome.8,9

Training is essentially “a signal that makes an epigenetic change that alters the way the cell responds [to an infection] a second time,” says Eicke Latz, director of the Institute of Innate Immunity at the University of Bonn in Germany. Because epigenetic alterations to chromatin can persist through cell division, Latz explains, it acts as a sort of molecular memory in the cells. “It is a different type of memory to what you have with the adaptive immune system, but it is certainly a memory that can be quite long-lasting,” he says.

Indeed, the blood samples from Netea’s student volunteers exhibited the same response to C. albicans at least three months after BCG vaccination—much longer, in fact, than monocytes are expected to live. But the effect is not nearly as long-lived as the adaptive immune system’s memory of the M. tuberculosis pathogen, which lasts 10 to 15 years.1 “We have indications that there is still some effect [of monocyte training] present after one year,” Netea says of the C. albicans response induced by BCG, “but by then it is waning.”

The transcriptional and epigenetic changes in trained monocytes revealed two cellular pathways essential for the process: glycolysis, the cellular metabolic pathway that yields energy from glucose, and cyclic adenosine monophosphate (cAMP) signaling, which is involved in transmitting a range of extracellular signals to targets within cells. Netea found that inhibiting either of these pathways in cultured monocytes impaired the cells’ production of training-associated inflammatory cytokines. Furthermore, inhibiting these pathways in mice reduced the ability of nonlethal C. albicans or β-glucan injection to confer protection against a secondary infection.8,9

Netea also recently uncovered evidence that autophagy—the degradation of cellular components to produce energy or to clear away damaged proteins, organelles, or intracellular pathogens—plays a role in BCG-induced training.10 By promoting degradation, Netea suggests, autophagy might help process the microbial proteins that stimulate training. Increased glycolysis, on the other hand, might increase energy production, preparing the cells to fight pathogens. How these diverse pathways are related to each other, or to cAMP signaling, remains unclear, however.

It is a different type of memory to what you have with the adaptive immune system, but it is certainly a memory that can be quite long-lasting.—Eicke Latz, Institute of Innate Immunity,
University of Bonn


While the training of monocytes—by β-glucan, C. albicans, or M. tuberculosis exposure—has been well characterized in vitro, it’s unknown which cells serve as repositories for the trained response in the body, says Frederic Geissmann, who studies inflammation at King’s College London. “Monocytes have a very short half-life in the blood—less than a day in mice, and in humans it is not much longer,” he says. “So it is difficult to imagine that the memory would be kept in these cells.”

More likely, monocyte memory would have to be stored in either the bone-marrow progenitors, which continually replenish the circulating monocytes, or in the longer-lived macrophages residing in the body’s tissues. Training monocytes in culture certainly activates the cells to become macrophage-like, but if tissue macrophages in the body were the sole cells responsible for memory, it wouldn’t explain how the students’ blood displayed a heightened immune response three months after vaccination—because blood does not contain tissue macrophages. For this reason, Latz puts his money on bone-marrow stem cells as, at least, the primary storage units of monocyte memory. These cells would also have plenty of exposure to an organism’s pathogens, he added. “The bone marrow is not a secluded organ; it is like the blood. So it is very possible that the bone marrow cells can be trained directly by infection.”

But for now, “how it works in vivo is still mysterious,” Geissmann says.

NK cells remember

NK CELL MEMORY: Natural killer (NK) cells also have an unexpected ability to learn from past infection. In contrast to monocytes, NK-cell memory appears to be specific to a particular pathogen or molecule. NK cells carry a handful of receptors that are specific for certain pathogens, such as cytomegalovirus (CMV). Upon invasion, the NK cells possessing a pathogen-specific receptor proliferate and differentiate into “memory” cells that show increased production of the immunostimulatory cytokine IFN-γ and greater cell surface expression of the receptor molecule that binds the pathogen.
See full infographic: JPG | PDF
NK-cell memory also holds many unanswered questions. In contrast to monocyte training, the NK-cell memory described by Lanier appears to be specific to a particular pathogen or molecule. NK cells fight diverse pathogens by recognizing when host cells are infected and destroying them. However, in addition to being a general defense agent, it’s thought that, over millions of years of evolution, NK cells have acquired a handful of specific receptors, says Lanier. The CMV receptor his group identified is one example. Other NK cells carry a receptor for H2-D—an alloantigen, or protein present in only some members of an animal species and recognized as foreign by members who lack it (such as in blood type compatibility). Lanier’s group recently reported that NK-cell memory can be induced by H2-D, just as it can by CMV.11 It is these specific receptors that are thought to drive NK-cell memory, he says. “We’ve tried with other pathogens, like influenza, and the NK cells get stirred up, but they don’t undergo this clonal expansion and the generation of long-lived memory cells,” he says.

In von Andrian’s lab, however, no specific hapten receptors have been identified. Hapten-induced NK memory cells are instead dependent on another receptor—the cytokine receptor CXCR6—which has been found to be essential for the function and survival of these memory cells. Interestingly, CXCR6 is not required for CMV-induced memory cells.12

It’s not known if NK-cell memory, like monocyte memory, is associated with epigenetic alterations, but it does appear to be associated with altered gene regulation. CMV-induced memory NK cells, for example, exhibit increased expression of the CMV receptor—perhaps giving them a lower activation threshold. They also produce more of the immunostimulatory cytokine interferon-γ, both constitutively and when reexposed to the virus.

Just how long-lived NK memory cells are remains to be seen. They appear to last at least as long as, if not longer than, trained monocytes, says Lanier. “We have been able to restimulate memory NK cells at least six months after they were generated, and in a couple of mice we found them a year later.” But “there is no evidence . . . that [NK] cells are going to protect you 5 years or 10 years later,” says Lichtman. On the other hand, he says, the memory B and T cells in his body that encountered measles, mumps, and rubella viruses when he was a kid are still functioning. “I still have protective antibodies, 50 years later,” he says. “Nothing like that has ever been described for innate immunity, whether it is macrophages or NK cells.”

Innate-memory medicines

Despite growing epidemiological data suggesting that BCG has additional benefits,4 people have still thought of the vaccination as being disease-specific, says Christine Stabell Benn, a professor of global health at the State Serum Institute and at the University of Southern Denmark. But the new roles for innate cells in remembering past pathogen encounters could change all that. “What I think is so wonderful about the new papers coming out on trained innate immunity,” says Benn, “is that it is really providing that kind of biological plausibility to the epidemiological findings.”

Acknowledging the nonspecific, protective benefits of BCG vaccination, suggests Benn, would strengthen the case for ensuring this injection is given at birth in developing countries, where the risks of various infections are high but vaccination is sometimes delayed for logistical or other reasons. A better understanding of innate memory may also inform new therapeutics, says Lanier. In the case of CMV infection, for example—which can be life-threatening to immunocompromised individuals and some newborns—targeting therapies to NK memory cells may be crucial for saving lives.

Similarly, boosting monocyte memory may be useful in treating immunoparalysis, which occurs in some sepsis patients when their monocytes and macrophages become tolerant of the microbial infection and allow it to rage out of control. Netea suggests boosting glycolysis to essentially force these cells to remember the microbes and fight. Conversely, too much monocyte memory can be problematic in, for example, some chronic inflammatory processes, such as atherosclerosis. In such instances, suppressing monocyte memory might be desirable, he suggests.

The clinical implications of innate immune memory are “very exciting,” says Benn. “The picture we had of the immune system is really changing dramatically, and that offers new ways to think about how it can be modulated to prevent or treat disease.” 

Ruth Williams, a freelance science writer living in Norwalk, Connecticut, is a staff correspondent for The Scientist.


  1. J. Kleinnijenhuis et al., “Bacille Calmette-Guerin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes,” PNAS, 109:17537-42, 2012.
  2. J.G. O’Leary et al., “T cell- and B cell-independent adaptive immunity mediated by natural killer cells,” Nat Immunol, 7:507-16, 2006.
  3. J.C. Sun et al., “Adaptive immune features of natural killer cells,” Nature, 457:557-61, 2009.
  4. F. Shann, “The non-specific effects of vaccines,” Arch Dis Child, 95:662-67, 2010.
  5. O. Levy, M.G. Netea, “Innate immune memory: Implications for development of pediatric immunomodulatory agents and adjuvanted vaccines,” Pediatr Res, 75:184-88, 2014.
  6. J. Quintin et al., “Candida albicans infection affords protection against reinfection via functional reprogramming of monocytes,” Cell Host Microbe, 12:223-32, 2012.
  7. M.G. Netea et al., “Trained immunity: A memory for innate host defense,” Cell Host Microbe, 9:355-61, 2011.
  8. S.C. Cheng et al., “mTOR- and HIF-1α-mediated aerobic glycolysis as metabolic basis for trained immunity,” Science, 345:1250684, 2014.
  9. S. Saeed et al., “Epigenetic programming of monocyte-to-macrophage differentiation and trained innate immunity,” Science, 345:1251086, 2014.
  10. K. Buffen et al., “Autophagy controls BCG-induced trained immunity and the response to intravesical BCG therapy for bladder cancer,” PLOS Pathog, 10:e1004485, 2014.
  11. T. Nabekura and L.L. Lanier, “Antigen-specific expansion and differentiation of natural killer cells by alloantigen stimulation,” J Exp Med, 211: 2455–2465, 2014.
  12. G. Min-Oo, “NK cells: Walking three paths down memory lane,” Trends Immunol, 34: 251-58, 2013.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!