Search

Text and Interview by Hema Bashyam As a student, Anthony De Tomaso was fascinated by many aspects of biology, from molecular biology to marine biology and development to immunology. Now as head of the same lab that he joined as a postdoctoral fellow, De Tomaso has marshaled his diverse scientific

On page [565][1], Haxhinasto et al. pinpoint the signaling roadblock that stops CD4+ T cells from turning into regulatory T (T reg) cells. Some T reg cells develop directly from thymic progenitor cells, but others seem to be converts that were once CD4+ T cells. These potentially inflammatory CD4+

For more than 70 years, alum has been used as a trusty aide to improve antibody responses to vaccines. But how this aluminum-containing compound boosts the response to vaccines has been a mystery. On page 869, Kool et al. expose its mysterious mechanism; they find that alum causes cells to produce a stimulator of dendritic cells (DCs). Figure 1 Mice injected with antigens mixed with alum produce uric acid that then attracts monocytes. DCs were once the favored hypothetical link between alum and B cells, as they activate CD4+ T cells, which can then enhance B cell activation and antibody production. But DCs were later dismissed when it was found that they were not stimulated by alum in vitro. These in vitro results are now shown to be a red herring by Kool et al., who find that alum activates DCs in vivo by provoking the secretion of uric acid—a molecule that is triggered by tissue and cell trauma. The injection of alum, the group found, induced an influx of neutrophils and inflammatory cytokines and chemokines—a combination that was previously seen in response to the injection of uric acid into mice. In mice injected with antigens mixed with alum, uric acid levels increased within hours. The uric acid might be released by the cells' lining the body's cavities that turn necrotic after contacting the alum. The absence of these uric acid sources in in vitro assays might have led to the previous misleading results. In response to the uric acid, inflammatory monocytes flocked to the injection site, took up the antigens, and broke them down into T cell–stimulating epitopes. The monocytes then migrated to lymph nodes, where they matured into DCs and activated CD4+ T cells. Without alum, the antigens were not taken up at the injection site. Still, they eventually reached lymph nodes via the flowing lymph. The resident node DCs, however, did not process the alum-free antigens efficiently or express T cell costimulating receptors. The resulting subdued immunity was similar to that seen in mice that were depleted of inflammatory monocytes or those injected with enzymes that degrade uric acid.

On page 1917, Nishikii et al. produce a plethora of platelets by picking perfect stem cell progenitors and preventing shearing of the platelets' activating receptors. Figure 1 MMP inhibitors help maximize the numbers and function of cultured platelets (green) expressing integrin receptors (red). Platelets for therapeutic use are currently filtered from donated blood, which increases the risk of infections and other side effects in patients who need frequent transfusions. Scientists have been trying to generate platelets from embryonic stem cell (ESC) lines instead, but their efforts have so far been stymied by two problems. First, ESCs cultured with stromal cells produce a tiny platelet population that is quickly drowned out by other cell lineages. Nishikii et al. resolved this issue by using ESCs that had already differentiated into platelet-committed hematopoietic stem cells, which express the clot-promoting αIIbβ3 receptor. The second and more worrying problem is that the ESC-derived platelets don't aggregate properly. This defect was previously seen in vivo in long-lived platelets whose matrix-binding receptors had been sheared off by matrix metalloproteinases (MMPs). The group found that this also happened in vitro, and platelets cultured with MMP inhibitors formed clots in vitro and enhanced tissue repair in injured mice. This approach awaits testing in humans.

A study by Didierlaurent et al. (page 323) explains why fighting the flu makes us vulnerable to bacterial infections in the following months. A clampdown of innate immune receptors in the lung might leave the door open to opportunistic bacteria. The team previously showed that mice that had just recovered from the flu had a weak inflammatory response that made them more susceptible to other pathogens over the next six months. The group now finds that post-flu inflammation in these mice is tempered by a dampened innate immune system. The innate system usually responds to ligands that activate Toll-like receptors (TLRs). But in mice recovering from the flu, these ligands did not activate TLRs on lung macrophages. The cells therefore did not produce the chemokines necessary to attract neutrophils—the main instigators of lung inflammation. Although the blunted TLR response might protect the lungs against inflammation-induced damage, it made them vulnerable to bacterial infection. After beating influenza, the mice were killed by pneumonia-causing bacteria that are normally held at bay in the airways. Influenza and other respiratory viruses are not known to encode TLR-suppressing proteins. This suppression may thus represent the lung's attempts to protect itself from damaging levels of inflammation. The authors suspect that the lungs might suppress TLR expression in response to any insult that causes a massive inflammatory response, including pathogens, injury, or environmental allergens. Their theory is bolstered by the fact that another respiratory pathogen, respiratory syncytial virus, also suppressed macrophage TLR signaling.

Neutrophils need just enough copies of a gene for an antibody-binding receptor, say Willcocks et al. (page 1573). Whereas too few copies can contribute to lupus, too many can predispose to vasculitis. Figure 1 Neutrophils with fewer copies of FCGR3B bind poorly to IgG complexes. Cells can have as many as four copies of this gene, FCGR3B, which encodes the FcγRIIIb receptor, but some lupus patients have fewer than two. Willcocks et al. now find that this scarcity leads to fewer receptors on neutrophils, which thus fail to bind and destroy inflammation-causing antibody clusters. The paucity of receptors did not impair other neutrophil functions such as the production of oxygen radicals. Some healthy individuals also had fewer copies of the gene and thus similarly defective neutrophils. But these individuals do not develop lupus, most likely because they lack other genetic defects required to cause disease. The compensatory influence of other genes might also explain why low copy number of FCGR3B is not a risk factor for lupus among all populations. Having more FcγRIIIb receptors, however, doesn't guarantee good health. The group found that a high copy number of FCGR3B was associated with vasculitis—a disease in which neutrophils release damaging oxygen radicals in response to cross-linking of FcγRIIIb receptors on blood vessel walls.

On page 1201, Huang et al. find that an antigen-presenting protein suffers from an identity crisis: it looks like MHC class I but behaves like MHC class II. This molecule, MR1, is recognized by a subset of gut-roving T cells. When MR1 carries the right antigen, these T cells suppress gut inflammation, but the identity of that antigen is a mystery. Because MR1 is structurally similar to MHC class I molecules, which usually pick up proteasome-processed peptides in the ER, MR1 was thought to do the same. This notion is now disproved by Huang et al., who show that MR1-expressing cells treated with proteasome inhibitors or lacking ER peptide–loading chaperones can still activate these T cells. Instead, they found, MR1 picked up its antigens in endosomes, where MHC class II molecules get theirs. MR1 bound to a chaperone called the invariant chain, which ferries MHC class II molecules from the ER into endosomes. There, it bound another chaperone, called DM, which loads them up with peptides. MR1-expressing cells that lacked the invariant chain were unable to activate gut T cells, whereas high levels of this protein increased their activation. The immune-suppressing ability of MR1-activated T cells suggests that they might prevent the gut from attacking its helpful resident flora. The authors therefore speculate that the MR1 ligand might be derived from one or more of these residents.

A factor that promotes blood vessel growth keeps two types of brain cells alive, say Maharaj et al. (page 491). Their findings might explain why anticancer drugs that block this factor cause neurological side effects. Figure 1 Mice that express high levels of VEGF-blocking molecules develop brain lesions (arrows). The drug target in question—vascular endothelial growth factor (VEGF)—binds to receptors on blood vessel endothelial cells. A localized increase in VEGF, which is produced by underlying epithelial cells and other neighboring cells, promotes vessel growth in developing organs and injured tissues. Lower levels of VEGF might also be necessary for the survival of endothelial cells, which become apoptotic when cultured in the absence of VEGF. VEGF-blocking drugs that limit tumor growth by inhibiting the tumors' blood supply can cause seizures and brain swelling in some cancer patients. To find out whether these symptoms stem from the destruction of brain blood vessels, Maharaj et al. examined the brains of mice that had been engineered to express high levels of proteins that block VEGF and TGFβ—a cytokine that stimulates VEGF production. Besides damage to brain endothelia, the mice developed brain lesions due to a breakdown of the barrier that prevents cerebral spinal fluid from seeping into the brain. The specialized type of epithelial cells that make up this barrier expressed VEGF receptors and became apoptotic when VEGF was systemically blocked. Why brain complications only occur in a subset of cancer patients on anti-VEGF therapy is unknown.

Try as it might to pull itself forward, an immune cell will go nowhere unless a myosin also pushes it from behind, report Morin et al. (page 195).

[Graphic][1] A citrulline-containing IL-8 isoform fails to recruit neutrophils. On page [2085][2], Proost et al. find that altering a single arginine turns an inflammatory chemokine docile. The chemokine IL-8, which activates and recruits neutrophils, is secreted as a mixture of

On page 2019, Deruaz et al. find that ticks might prefer to take on immune cells one at a time rather than brave the whole lot at once. Figure 1 Ticks secrete several different evasins, each of which blocks a different chemokine. Some worms and viruses fend off immune cells by neutralizing the chemokines that attract them. These species make one chemokine-blocking protein that covers all the bases. But ticks don't use this one-for-all strategy, the authors now find. Instead, they made at least three single-chemokine blockers, which probably prevent host immune cells from swarming into the bite site and clogging up the ticks' food pipeline. These anti-chemokines, called evasins, each had a unique shape that might contribute to their selectivity. Ticks might have evolved this selectivity to counter the host's step-wise immune response to the Lyme disease parasite they carry. In parasite-infected hosts, neutrophils usually arrive first to the bite site, followed by eosinophils and monocytes, and finally other immune cells. Tick saliva has different anti-chemokine activity at different times during feeding, suggesting that having several evasins might somehow better counter this staggered cell influx. Viruses, on the other hand, do not have the luxury of combating one cell type at a time and might therefore depend on a single multi-purpose protein.

On page 1077, Machado et al. reveal a new way in which aspirin reins in inflammation—it triggers the destruction of proinflammatory signaling proteins. Figure 1 The destruction (top) of TRAF2 (green) via SOCS2 (red) is blocked in mice treated with proteasome inhibitors (bottom). Aspirin's power was initially attributed to its inhibition of proinflammatory lipids called prostaglandins—a discovery that won the Nobel Prize in 1982. Later, aspirin was also shown to beef up levels of lipids called lipoxins, which help resolve inflammation by blocking NF-κB activation and the recruitment of inflammatory cells. Lipoxins were recently found to activate SOCS2, a protein that blocks signals from growth hormone receptors by targeting downstream signaling proteins to the proteasome. To determine whether SOCS2 also blocks inflammatory signals, Machado et al. fished for its binding partners among molecules that transmit innate immune receptor signals. They now find that SOCS2 binds TRAF2 and TRAF6—adaptor proteins that are required for cytokine production by activated dendritic cells (DCs)—and seems to target them to the proteasome. Treating mice with aspirin decreased DC levels of cytokines and TRAF2 and TRAF6—effects that were mitigated by proteasome inhibitors. The same effects were not found in DCs from SOCS2-deficient mice. Mice treated with a proteasome inhibitor after parasitic infection beat the bug but died from inflammatory damage. The deadly inflammation probably resulted from prolonged signaling via TRAF2 and TRAF6, as inflammation in SOCS2-deficient mice was severe even without the inhibitor. Infected mice that were left untreated resolved the inflammation and survived the infection. Because proteasome inhibitors also block the degradation of IκB—the molecule that holds NF-κB in check—they are being developed as alternatives to aspirin, which can cause stomach problems and other side effects. But the current findings suggest that this strategy might be counterproductive.

[Graphic][1] Overproliferating T cells infiltrate the liver and trigger lupus-like autoimmunity in older Egr-2–deficient mice. Self-tolerance is well known to be enforced in part by an external force of regulatory T cells. Now, Zhu et al. (page [2295][2]) show that T cells also

Refilling an empty niche with proliferating CD8+ T cells can cause autoimmune disease, say Tajima et al. (page 1019). Figure 1 Autoimmune colitis (top) caused by homeostatically proliferating CD8+ T cells is resolved in mice injected with anti–IL-6 antibody (bottom). Infusion of naive T cells into an environment where there are few native T cells—such as in patients undergoing cancer therapy—spurs two kinds of proliferation. T cells activated by non-self-peptide/MHC complexes proliferate rapidly, whereas those activated by self-peptide/MHC in the presence of cytokines such as IL-7 and IL-15 proliferate slowly. CD4+ T cells that proliferate by these methods have previously been shown to drive the inflammation seen in several autoimmune diseases. Tajima et al. now find a new variation in homeostatic proliferation of CD8+ T cells that also induces autoimmunity. In gut lymph nodes of T cell–deficient mice, injected CD8+ T cells proliferated rapidly in response to IL-6. This inflammatory cytokine was found at high levels in these lymph nodes, perhaps due to their proximity to the bacterium-laden gut. The mice developed a thicker gut epithelium and lost weight—signs of an autoimmune disease called colitis. T cell–injected mice treated with an IL-6–blocking antibody or with bacterium-depleting drugs, however, remained healthy. The proliferating T cells secreted several inflammatory cytokines, including IL-17. The IL-17 alone seemed to be responsible for the colitis, as mice injected with T cells that were unable to produce IL-17 remained healthy. If these IL-17–producing CD8+ T cells are found in patients suffering from diseases such as colitis, the anti–IL-6 antibody might be a good therapeutic option.

On page 1583, Varga-Szabo et al. show that platelets missing a calcium channel opener build puny, feeble clots that may actually save lives. Figure 1 Vessel-blocking clots (top) are avoided when platelets lack STIM1 (bottom). Platelets require calcium to initiate clotting in response to extracellular cues. When calcium is needed, most cells first deplete their ER stores. This depletion is sensed by an ER protein called STIM1, which sits near enough to the plasma membrane to open up calcium channels on the cell surface. But since platelets have only stunted ER-like tubules, they were thought to rely instead on an activation-induced lipid, which opens surface calcium channels independently of internal stores. Platelets also have plenty of STIM1, however, which Varga-Szabo et al. now show is needed for slow-but-steady clot formation. Activated STIM1-deficient platelets, the authors found, did not accumulate calcium from external sources. Internal calcium stores in the mutant platelets were sufficient for rapid clot formation in response to short-lived ligands such as thrombin. But platelets needed STIM1-induced calcium influx to clot in response to collagen on injured vessels, which triggers a slower signaling pathway. In the absence of STIM1, clots were small and disintegrated easily, particularly under conditions that mimicked blood flow. The defective platelets continually broke away from the clots, suggesting that STIM1 helps make platelets stickier. The results suggest that a quick calcium burst from ER-like tubules might be sufficient to build small clots to patch up a minor tear. The STIM1-driven calcium spike, on the other hand, might be necessary to form larger clots under high shear, as occurs in small or fat-clogged vessels. Mice lacking STIM1 were protected from injury-induced vessel blockage and strokes, as their feeble clots allowed some blood to flow into injured brain tissues. As the clotting problem didn't prolong bleeding, a small, transient clot might be all that's needed to prevent hemorrhage, with the added bonus of avoiding a complete halt in blood flow.

[Graphic][1] Large tumors (circles) appear more than a year after mice turn on an oncogene, long after T cells are disabled. Cancer cells outwit the immune system long before they debut as visible tumors, say Willimsky et al. (page [1687][2]). Despite a wealth of studies

The smallpox virus has been eradicated thanks to widespread vaccinations with the vaccinia virus. But because live vaccinia virus is used, some vaccine recipients, particularly those who are immune compromised, have experienced fatal infections. Xu et al. (page 981) now offer up an alternative that might side-step the pitfalls of the old vaccine. Killed vaccinia virus does not induce antibodies against proteins that trigger smallpox, forcing researchers to look for alternative strategies. Xu et al. considered designing vaccines that target virulence proteins that mute the immune response. They reasoned that a vaccine that did not include the entire virus would be safer, while antibodies against virulence proteins, known as immune response modifiers (IRMs), would prevent disease even if they did not neutralize the virus. The group now identifies an IRM that is the major target of protective antibodies in a mouse model of smallpox. This IRM prevents the antiviral cytokine interferon-α from activating its receptor on immune cells. Deletion of the IRM from the mousepox virus, the group found, caused a 107-fold decrease in its virulence and prevented lethality. Mice that were injected with the IRM alone were protected against a later challenge with the wild-type virus. The IRM, the type I interferon binding protein, is well-conserved among poxviruses that infect mice and men, so the hope is that the recombinant IRM protein might be an effective poxvirus vaccine for humans as well.

Two T helper (Th) cell subsets trigger the same multiple sclerosis (MS)-like symptoms but varying disease pathologies, say Kroenke et al. (page [1535][1]). Inflammatory Th1 cells and Th17 cells can both drive experimental autoimmune encephalitis (EAE)—the mouse equivalent of MS. The authors found

[Graphic][1] Mice that tend to have vigorous inflammatory responses (right) compensate by expressing the antiinflammatory gene IRAK1BP1. COURTESY OF P. BRODEUR Mice that tend to overreact to infection appear to have a failsafe mechanism that kicks in before inflammation gets out of

HIV and its simian counterpart, SIV, replicate rapidly after infection but often take many years to subdue the immune system and cause AIDS. Okoye et al. (page 2171) now find that these slow burner viruses only cause disease after they deplete the precursor cells that give rise to pathogen-fighting effector cells. Figure 1 Loss of TCM cells during chronic SIV infection (bottom) depletes pathogen-fighting TEM cells and triggers AIDS. The effector memory T (TEM) cells that battle these viruses at infection sites develop from a pool of CD4+ central memory (TCM) cells that are stored in secondary lymphoid organs. Although the virus infects both types of cells simultaneously, TEM cells are the first casualties. This early depletion of TEM cells is not completely devastating; the TCM population quickly cranks out more TEM cells. But this defensive strategy is obviously not foolproof: infected individuals eventually develop AIDS. Okoye et al. now identify the glitch in the strategy by tracking disease progression in SIV-infected rhesus macaques. Newly generated TEM cells were short lived, they found, as persistent activation by the virus induced their death. And unlike in earlier stages of infection, TCM cells no longer came to the rescue, as they were also crippled by viral infection. The virus thus tamed the immune system by stimulating one population to death and destroying its back-up. Based on these results, stabilizing the TCM pool may be a more effective way to prevent the onset of AIDS than controlling viral load—the goal of current vaccination strategies. Rising viral loads might not be the trigger for AIDS, as viral loads remained constant while CD4+ T cell levels declined during the later stages of disease. Recent studies from other groups show that survival of patients with AIDS correlates with more circulating TCM cells, supporting the idea that protecting the TCM niche might keep AIDS at bay. The team is now trying to identify the factors that bolster TCM cell levels and how HIV/SIV dismantles them.

Young T cells are fast learners, according to McCaughtry et al. (page 2513). They finish part of their schooling in the thymus in just four days. Upon entering the thymus, young T cells are first tested for correctly rearranged T cell receptors (TCRs). Those that pass the exam are selected for further training in the thymic medulla. There, T cells that react to self-antigens are killed off. The rest graduate into functional, proliferating adults. This training program was previously estimated to last for two to three weeks. This notion was based on methods that did not differentiate between young T cells, mature T cells that reentered the thymus, and other thymic TCR-bearing immune cell subsets. McCaughtry and colleagues now reanalyze the kinetics of T cell training with this distinction in mind. Maturation stages were identified by specific markers and by labeling proliferating cells with BrdU. The results revealed that T cells took just four days to pass through the medulla and enter the circulation. The team also found that T cells were educated on a first-come, first-served basis. As they passed the first selection test, the cells shut off expression of a fluorescent protein, resulting in a gradual decrease in brightness. Only dim cells expressed receptors that mark exiting cells, suggesting that the emigration follows a first-in, first-out assembly-line pattern. The authors now want to understand how the plumbing of the thymus helps T cells so rapidly sample the different self-antigens expressed by each epithelial cell.

Viruses that lie low inside cells after infection come under attack by the immune system when they reawaken and resume multiplying. Hislop et al. (page 1863) now reveal the mechanism used by the Epstein-Barr virus (EBV) to dodge host immunity during this reactivation phase. EBV initially infects and replicates within oral epithelial cells but later quietly hides out in B cells. During this latent phase, EBV-infected B cells avoid the immune system by expressing very little viral antigen. To ensure viral spread and survival, however, EBV must reenter the replicative, or lytic, phase and invade new epithelial tissues. Previous studies suggested that this herpesvirus also avoids the attention of the immune system during its reawakening. During the lytic phase, EBV-infected B cells dial down the activity of their transporters associated with antigen processing (TAPs)—transmembrane channel proteins that shuttle antigenic peptides into the ER, where they find their HLA partners. Infected B cells thus display few viral antigens at the cell surface. Known herpesviruses genes encoding TAP-inhibiting proteins were not found in the EBV genome. The authors therefore compared herpesvirus genomes to find a lytic phase TAP inhibitor gene in EBV and its closest relatives. Cloning and expression of candidate genes uncovered BNLF2a, which encodes a protein that blocked both the peptide-binding and ATP-binding sites on TAPs and thereby prevented it from translocating peptides into the ER. This mechanism differs from those used by all other herpesvirus TAP inhibitors. The team is currently investigating how the relatively small BNLF2a protein blocks access to two distant sites on TAP.

Arthritis sufferers might get some therapeutic relief from bacterial nucleic acids that usually goad on inflammation, if results from Wu et al. (page 1911) are any indication. Some of these short CpG repeats (CpGs) are now shown to prevent arthritis progression in a mouse model. Figure 1 The invasion (left) of neutrophils (red) into arthritic joints is prevented (right) by a CpG. CpGs are well-known proinflammatory molecules that activate dendritic cells (DCs) via toll-like receptor (TLR)-9. The DCs then stimulate T cells and B cells. CpGs are therefore used in the clinic to enhance protection against infections and tumor development. But immune enhancement can be harmful if the response is directed against the host's own antigens. CpGs and other TLR ligands, for example, further disease progression in many mouse arthritis models. In some mouse models of allergy and asthma, however, CpGs exert a protective role. In these mice, CpG-activated DCs somehow jumpstart the proliferation of protective T cell subsets and stop B cells from secreting allergy-causing antibodies. Wu et al. wondered whether CpGs had a similar therapeutic effect on arthritis. The team tested several known CpGs in mice that had developed arthritis in response to injected serum antibodies. Two structurally similar CpGs halted disease progression. Unlike disease-promoting CpGs, these helpful CpGs stimulated DCs to activate natural killer (NK) cells instead of T and B cells. The NK cells then produced interferon (IFN)-γ. This normally proinflammatory cytokine blocked neutrophil trafficking into the diseased joints, thereby lessening inflammation. The authors attribute this disease-dampening effect to the fact that the T/B cell-mediated initiation phase had passed by the time the CpGs were delivered. The NK-mediated therapeutic effects may translate to human arthritis treatments, as T and B cells have usually already done their dirty work by the time patients arrive in the clinic.

The immune system can only prevent cancer if it recognizes and destroys precancerous cells. Antigenic targets of such an early immune response, however, have been elusive. Spisek et al. (page 831) now find that an embryonic stem cell marker is a bull's eye for immune cells in patients who have precancerous lesions but never develop multiple myeloma. Figure 1 Patients who have anti-SOX2 T cells (solid line) are less likely to develop multiple myeloma than those lacking these T cells (dashed line). Multiple myeloma is an incurable cancer that occurs throughout the bone marrow. It is caused by the uncontrolled proliferation of antibody-producing B cells, which replace normal immune cells as well as red blood cells and platelets. In some patients, this malignant cancer is preceded by a benign condition known as multiple gammopathy of undetermined significance (MGUS). During this stage, the abnormal B cells do not form tumors, and the patients remain healthy. Although MGUS occurs in ∼3% of the population over 50 years of age, only a third of those with MGUS will eventually develop myeloma. This selective disease protection, the team now shows, correlates with the presence of immune responses against the embryonic transcription factor SOX2. SOX2 is normally switched off after embryonic stem cells differentiate, but, in both MGUS and myeloma patients, it is reactivated in bone marrow cells that are potential tumor progenitors. MGUS patients had anti-SOX2 antibodies and T cells, but the myeloma patients did not. In a two-year observational study, patients with early tumors who had anti-SOX2 T cells remained cancer-free. Those that lacked these T cells, however, developed multiple myeloma. These findings suggest that an immune response against the SOX2-expressing bone marrow cells may prevent malignancy. The group is now trying to determine whether myeloma patients never mount an immune response to SOX2 or simply lose it over time. The group's in vitro evidence suggests that anti-SOX2 T cells prevent tumor formation. But myeloma patients would have to be vaccinated against SOX2 to confirm that these T cells have antitumor activity in vivo.

On page 1037, Aspord et al. find that immune cells are swayed to the dark side by cancerous forces. Instead of priming a killing response, dendritic cells (DCs) that infiltrate breast tumors actually help the tumor grow. Figure 1 Breast tumor size increases when DCs and T cells team up (red). The group previously found that human breast tumors often include a large number of mature dendritic cells that form clusters with CD4+ T cells. This association suggested that immunity was at work in the tumor, but it also hinted at an oddity: mature DCs are normally found in lymph nodes, not the affected tissue. To figure out what these DCs were doing in the tumors, Aspord and colleagues studied humanized mice that were grafted with human breast cancer cell lines. In these mice, the DCs recruited CD4+ T cells to the tumors but did not prime an antitumor immune response. Instead, the DCs instigated the T cells to secrete IL-13 and thereby promote tumor growth. IL-13 helped tumors by activating a signaling pathway that has been implicated in metastasis. But why the DCs initiated this tumor-friendly response is still unclear. The findings might help explain why DC-based vaccine therapies sometimes fail in clinical trials. The team now plans to test IL-13 antagonists in breast cancer patients in combination with candidate DC vaccines. The hope is that the antagonists will prevent the vaccine-induced antitumor immunity from being subverted into a cancer-promoting response.

A DNA-mutating enzyme that fine-tunes B cell antibody specificity can become a dangerous liability if its activity is mistimed, misplaced, or left unchecked. Crouch et al. (page 1145) have now designed a mouse model to track the enzyme's activity in B cells during development and immunity. The model can now be used to determine how the enzyme's destructive force is controlled. Figure 1 Permanently marking B cells that have AID activity with YFP (green) shows that AID is turned on upon B cell activation. The enzyme in question is AID (activation-induced cytidine deaminase), which converts cytidines to uracils. AID activity within activated B cells allows them to better recognize a pathogen by creating mutations in the variable regions of their immunoglobulin (Ig) genes. AID-assisted recombination within Ig constant regions also allows B cells to generate different types of antibodies. Mismatch repair mechanisms then splice out the mutations, reseal the new ends, and thereby create new sequence variants at the Ig loci. But this AID clean-up machinery is itself error-prone and increases the likelihood of further mutations that make the animal vulnerable to diseases such as cancer. To monitor the expression of AID in vivo, the team designed transgenic mice that express a fluorescent protein upon AID activation that then permanently marks all AID-expressing cells and their progeny. AID, the team found, was not expressed by developing B cells. It was first switched on in activated B cells that were beginning to proliferate in germinal centers (GC)—the hub of the B cell immune response. The switch seems to be controlled by a short DNA sequence immediately downstream of the AID gene. This sequence had more histone acetylation (which marks transcriptionally active regions) in activated B cells. Deleting the sequence almost completely abolished AID expression. AID expression was down-regulated as B cells differentiated into either memory cells or antibody-producing plasma cells and began to leave the GC. The exact mechanism that shuts off AID in exiting B cells remains to be worked out. Using this model to resolve this mechanism and understand how AID regulatory mechanisms get derailed in B cell cancers is the team's next step.

Vol. 204, No. 13, December 17, 2007. Pages [3054–3055][1]. Please note that the story inaccurately attributed the observation of dendritic cells in the lung and liver to Poulin et al. This discovery was in fact made by Bursch et al. The html and pdf versions have been corrected. [1]: /lookup/

A mycobacterium that causes skin ulcers protects itself from the immune system by suppressing dendritic cell (DC) activity, according Coutanceau et al. (page 1395). In the initial stages of Mycobacterium ulcerans infection, the skin puckers up into nodules that brim with bacteria and inflammatory cells. But this primary immune response gradually loses steam. As disease progresses, the inflammatory cells disappear and the nodules transform into festering ulcers. The bug seems to enforce immune suppression even outside the lesions as patients with ulcers have poor systemic cellular responses. Coutanceau et al. found several structural similarities between mycolactone—a lipid toxin produced by the bug—and known immune-suppressive drugs, and postulated that the toxin might function as an immune modulator. They now find that mycolactone hinders the DCs that drive T cell activation. The toxin prevented skin-derived DCs from migrating to the draining lymph nodes and blood-derived DCs from acquiring their usual antigen-presenting and chemokine-secreting functions. But why the toxin initially allows the primary immune response to occur and how it hobbles DCs after ulcers form are still unclear. Other mycobacteria, such as the bug that causes tuberculosis, secrete related biologically active lipids. Some of them target macrophages and inhibit the production of inflammatory cytokines, not chemokines. The team is now investigating the structural and molecular basis of this difference in toxin function.

[Graphic][1] A COX-2 inhibitor (right) increases the expression of tissue factor (red) in heart vessels (arrows). On page [2053][2], Ghosh et al. reveal how antiinflammatory COX-2 inhibitors cause dangerous cardiac side effects. COX-2, they find, is needed to shut off an important

In the late 1980s, Tim Mosmann and colleagues isolated functionally distinct T helper (Th)-1 and Th2 clones, and provided evidence that these two subsets were mutually inhibitory. Knowledge of the inhibition led to the discovery that Th2 cells make IL-10 to suppress Th1 cells.

T cells can kill tumors more effectively if a roadblock in their receptor signaling is removed, report Loeser et al. on page 879. Figure 1 Mice that lack cbl-b spontaneously reject UV-induced tumors. Most tumors are poor targets for the immune system as they frequently lack costimulatory molecules and present antigens that are only weakly immunogenic. The few tumor-specific T cells that do get generated are tolerized and fail to see the tumor as a dangerous enemy. Strategies to induce strong antitumor T cell responses—including cytokine treatment to coax tumors into expressing costimulatory proteins or vaccination with host dendritic cells loaded with tumor antigens—have not always worked. Loeser and colleagues decided instead to circumvent tolerogenic mechanisms within the antitumor T cells themselves. Their target was the ubiquitin ligase cbl-b, which inhibits TCR signals by targeting downstream signaling proteins for proteasomal degradation. The group had previously shown that T cells lacking cbl-b are efficiently activated by weak antigens even in the absence of costimulation, suggesting that cbl-b functioned as an internal tolerance enforcer. They now find that these T cells are efficient tumor killers as well. Mice lacking cbl-b were able to spontaneously reject tumors that were either injected into the animals or induced by UV exposure. Cbl-b–deficient CD8+ T cells destroyed tumors when transferred into wild-type animals. These cells developed into memory cells and offered long-lasting protection against tumor recurrence. One pitfall of removing a tolerogenic T cell protein such as cbl-b is the threat of autoimmunity. The cbl-b–deficient mice do show signs of autoimmune damage. However, these side effects are not lethal, and the mice have a normal lifespan. The team now plans to test whether antitumor T cell activity in mice can be mediated by RNAi knock-down of cbl-b, as this strategy may be practical in humans.

The machinery that repairs spontaneous DNA mutations fails when the mutations are purposely induced by an enzyme in activated B cells. Wu and Stavnezer (page 1677) now find that the enzyme wins because it induces too much damage for repair proteins to keep up with. Figure 1 DNA Polβ-deficient B cells switch more frequently when AID targets are sparse (IgG2a). The spontaneous mutation of cytosine bases to uracils is corrected by a process that removes the wrong bases, cuts the DNA at the empty spots, and reinserts the correct bases. This normally efficient process fails in B cells that are switching from producing IgM to making other classes of antibodies. In these cells, cytosines are converted to uracils by the enzyme activation-induced deaminase (AID). The mutated DNA is cut normally, but the ends then recombine to produce new types of antibodies. Recombination thus occurs at the expense of repair. The repair failure is blamed on DNA polymerase β (Polβ), whose job is to add back the correct nucleotides. Scientists have proposed that B cells undergoing class switching have too little Polβ or that the damaged DNA sites are not accessible to the enzyme. But Wu and Stavnezer now show that Polβ is probably as productive in B cells as it is in other cell types. Polymerase levels were normal, and the enzyme found its way to damaged sites. Activated B cells lacking the polymerase had even more mutations—and more recombination events—than usual. The team needed a new explanation. They hypothesized that recombination happens because AID creates too many damaged sites for Polβ to repair. The absence of Polβ caused a noticeable increase in recombination only in antibody genes with relatively few AID target sequences. Recombination events were abundant, however, in genes with lots of AID targets whether or not Polβ was present. Thus Polβ repairs AID-induced lesions, but can only fix so many. Activated B cells need their repair machinery to protect themselves from unwanted mutations during antibody gene rearrangement. The authors speculate that, to switch antibody types, B cells thus have had to dilute the efficiency of the repair process by gaining more AID targets in antibody genes.

Early resistance to pathogens requires a swift response from NK cells. In 1989, Giorgio Trinchieri identified an NK growth factor and activator, later called interleukin-12 (IL-12). This discovery helped reveal the regulatory link between innate and adaptive immunity.

[Graphic][1] Human dendritic cells that lack IRAK-4 (gray bars) do not produce inflammatory cytokines in response to most TLR signals. Humans outgrow their need for a TLR-activated kinase, according to a new study by Ku et al. (page [2407][2]). The kinase helps protect young

[Graphic][1] Highly translated EBNAΔGA induces a stronger T cell response. An effective alarm leads to an effective response. On page [525][2], Tellam et al. show that increased translation of a viral protein results in a more effective display of epitopes to T cells and thus a

Anti-inflammatory lipid mediators help put out flames at an injury site. On page [245][1], Svensson and colleagues now report that these lipids switch off pain signals in the spinal cord as well. Pain occurs when neurons in the periphery get stimulated by tissue damage or inflammatory signals such

[Graphic][1] Activated CD4+ T cells expressing high levels of IL-7Rα (orange) infiltrate transplants. Transplanted organs function well in their new home only if they are left alone by the host's immune system. Now, Codarri et al. (page [1533][2]) have potentially found a way to

Sexually transmitted HIV begins its battle with the immune system in the mucosal tissues of the host gut. Sun et al. (page 705) now recapitulate these events for the first time in a small animal model by replacing a mouse immune system with human lymphocytes.

[Graphic][1] In stressed cells, expanded ER membranes are cannibalized by ER-derived autophagosomes. WALTER/PLOS When under stress, the ER of yeast cells expands greatly in size but also eats its own membrane stacks to form autophagosome-like structures, say Sebastian Bernales

Short lipid chains tweak the responses of natural killer T (NKT) cells by weakening the binding between the T cell receptor (TCR) and the lipid-presenting protein, report McCarthy et al. on page 1131. Figure 1 Lipid antigens have to be long enough for cytotoxic granules (green) to move from the rear of NKT cells to the immunological synapse (left to right). Lipid antigens are presented to NKT cells by the surface glycoprotein CD1d. These fatty antigens are composed of a hydrophilic head that is exposed to the TCR on the NKT cells, and two hydrophobic tails. Each tail—an acyl chain and a fatty acid chain—is tucked away inside separate grooves within the CD1d molecule. Previous studies showed that shorter lipid chains destabilize the binding of the lipid antigen to CD1d. Whereas some of these short ligands simply reduce the ability of CD1d to activate an NKT cell, others switch the NKT response from interferonγ to interleukin-4 production. McCarthy et al. now find that shortening either the acyl or the fatty acid chain destabilizes the CD1d-lipid complex. The shorter fatty acid chain, however, also reduces the binding affinity between the CD1d-lipid complex and the TCR. The NKT cells are therefore unable to induce clustering of CD1d-lipid complexes and fail to form a productive synapse with the antigen presenting cell. The authors propose that the stumpy fatty acid chain doesn't completely fill its CD1d groove, which consequently collapses. This new conformation might lead to a less stable interaction between the complex and the TCR.

[Graphic][1] Tumors (red) are attacked by surrounding mDCs (blue) and infiltrating pDCs (green). Cytotoxic T cells that are trained by dendritic cells (DCs) to recognize tumor antigens are thought to be the main artillery in an antitumor response. But Stary et al. now find that the

When T cells come across a tumor, they kill off the outer layers and then move in, say Boissonnas et al. on page 345. The authors use intravital microscopy to show that this infiltration and obliteration of the tumor is antigen dependent. Figure 1 Solid tumors are destroyed (top) only when tumor cells (green) express antigen. Intravital microscopy has previously revealed the killing strategy of T cells in lymphoid tissue. The T cells use a search-and-destroy tactic, moving rapidly through the tissue while constantly scanning for targets. When an enemy is identified, they stop, kill the intruder, and then resume the hunt. The T cells' method of killing inside a solid nonlymphoid tumor environment, however, is less certain. “We knew that the immune system can reject tumors,” says senior author Sebastian Amigorena, “but we didn't really understand how T cells function inside the tumor.” Amigorena and his colleagues now show that the initial contact between anti-tumor T cells and their targets is at the tumor periphery. Once the peripheral cells are dead, the T cells attack the next layer and the next, thus gradually diminishing the tumor mass. This method of tumor destruction by T cells is dependent on antigen expression by the tumor cells. In tumors that do not express a cognate antigen, T cells are initially detected in the periphery but then drift away when they fail to find an appropriate target. The team now plans to use their imaging set-up to investigate whether anti-tumor T cells are doing the killing on their own or, as seen in lymphoid tissue, are recruiting other immune cells. Knowing what cells work against a tumor might lead to anti-tumor strategies that help these cells penetrate to a tumor's depths.

The only drug that can save patients from severe sepsis carries a risk of internal bleeding. Kerschen et al. (page 2439) have now engineered a variant version that offers the same benefits as the original but minimizes the risk. Figure 1 Mice are protected from sepsis-induced death by an APC variant that lacks its anticlotting activity (circle). Sepsis is a system-wide inflammation that occurs when an infecting microbe enters the bloodstream. The inflammation is meant to fight the pathogen, but it also triggers widespread clotting. Clotting proteins, in turn, reinforce inflammation by activating oxidants and proteases, ultimately causing organ failure. Because the exact cause of death in sepsis patients is unknown, scientists don't know what problem to target when designing an effective therapy. Antiinflammatory agents such as steroids and cytokine antagonists have failed as therapies. So too have anticlotting agents. The only treatment that has had any degree of success is activated protein C (APC). APC is both a clot-busting enzyme and a signaling molecule that blocks apoptosis, inflammatory cytokine production, and immune cell recruitment. APC is therefore thought to work against sepsis by breaking the inflammation/clotting cycle. But patients dosed with APC can experience internal bleeding thanks to the drug's anticlotting activity. Kerschen and colleagues now test whether APC might still be effective but less dangerous if its anticlotting activity is reduced. They engineered an APC variant that was 90% less effective at stopping clotting. This variant was just as good as normal APC at saving septic mice, suggesting that the protective effect of APC did not stem from its anticlotting function but from its signaling ability. Exactly why APC's antiinflammatory signaling activity succeeds in treating sepsis where other antiinflammatory drugs have failed is unclear, but this might be due to its broader effects on both the inflammatory and apoptotic pathways.

[Graphic][1] Vaccine antigens coupled to an antibody (green) that binds to carbohydrates on M cells (right) induce a stronger immune response than when coupled to lectins (red). Pathogens that invade the mucosa are better repelled if the vaccines against them are delivered to the

HIV infection tends to develop into AIDS more quickly in infants than in adults. A study by Hartigan-O'Conner et al. (page 2679) now suggests that more potent infant regulatory T (T reg) cells are to blame. Figure 1 Infants (black lines) have more powerful T reg cells that suppress SIV-fighting CD4+ T cells. T reg cells cool down virus-induced immune responses by suppressing activated T cells. It has been suggested that too much suppression allows HIV to get the upper hand. Others, however, have suggested that activated T cell suppression by T reg cells prevents inflammatory T cell cytokines from damaging tissues and thus enhancing disease. Hartigan-O'Conner and colleagues now show that healthy infant monkeys have more—and more potent—T reg cells than do adult monkeys. When monkeys were infected with SIV—the primate version of HIV—most of the infants rapidly developed AIDS. Their T reg cells suppressed anti-SIV T cell functions, whereas the adult T reg cells did not. Humans also have more T reg cells during infancy. The authors suspect that, like primate T reg cells, the potency of human T reg cells might also decrease with age. Determining why T reg cell numbers decrease with age and why aging T reg cells lose their steam is the next step.

[Graphic][1] By day 11 of a Leiden pregnancy, a fetus with paternal mutations in anticoagulant genes (left) is more decayed than its sibling (right). A pregnancy disorder that leads to an early loss of the fetus is linked to mutations that increase the mother's tendency to form blood