MICROBIOLOGYPreparing the shotChristof R. HauckDirect injection of proteins into host cells is one of the tricks bacteria use during infection. It seems that, to achieve this, the stomach pathogen Helicobacter pylori first grabs the cell by its surface receptors.The bacterium Helicobacter pylori successfully colonizes the stomach of about every third person. Infection with this ubiquitous micro-organism can cause acute and chronic gastri-tis, as well as stomach ulcers1. Moreover, up to 90% of cases of stomach cancer are associated with H. pylori infection. The bacterium’s main weapon is an elaborate apparatus on its sur-face called the type-IV secretion system, which acts as a nano-syringe (Fig. 1a). Using this apparatus, the bacterium delivers a cancer-associated protein, CagA, directly into its host cells. But whether the bacterium anchors the secretion system to the surface of host cells before injection, and if so, how, has remained unclear. On page 862 of this issue, Kwok et al.2 report that transfer of CagA is made possible by another H. pylori protein, CagL, which binds to integrin receptors on gastric epithelial cells. So far, CagA is the only H. pylori protein known to be injected into the host cell. In the bacterial chromosome, the cagA gene is part of a stretch of DNA called cagPAI, which also encodes the structural components of the type-IV secretion machinery3. Bacterial strains harbouring cagPAI are considered to be more virulent than other strains4.Previous work5–7 had shown that, once CagA is delivered into the host cell, kinase enzymes of the Src family add a phosphate group to it. The presence of phosphorylated CagA results in several changes that might promote H. pylori virulence and an unfavourable outcome for infection with this bacterium4,8. These changes include the assembly of signalling complexes, reduced cell–cell adhesion and induction of cell migration. Examining the localization of phosphory-lated CagA in isolated gastric epithelial cells, Kwok et al.2 found that it occurs almost exclu-sively at focal adhesion sites — discrete regions of the cell where integrin receptors ‘glue’ cells to the supporting extracellular matrix. The authors speculated that CagA might not move through the cytoplasm of the infected cells to these sites, but instead be injected directly at these places. Support for this idea came from experiments demonstrating that CagA is not transferred into host cells if H. pylori cannot to which strong interactions become weak at very short distances. In this ‘perturbative’ regime, we understand (at least in principle) how to work with QCD. But for the strong cou-pling that occurs over larger distances, one has to resort to computer-simulation techniques, known as lattice QCD. These techniques have been rather successful (for instance, in explain-ing the spectrum of hadron masses), but rig-orous results remain hard to come by: despite years of effort, we still cannot explain, for example, why there are no free, single quarks in nature. Such unresolved puzzles are coming into renewed focus with the scheduled start of experiments at the Large Hadron Collider at CERN in Geneva next year.The new approach that revives the link to string theory first suggested itself in 1998, when Juan Martín Maldacena conjectured12 a link between a close relative of QCD and a ‘superstring’ living in a ten-dimensional curved space-time. Although the theory in question, known as supersymmetric N = 4 gauge theory, is sufficiently different from QCD to be of no direct interest to experiment, the link raised the prospect of a general connection to some form of compactified string theory. This equivalence is now commonly referred to as the AdS/CFT (Anti-de-Sitter/conformal field theory) corre-spondence. If true, it would mean that string theory was originally not so far off the mark after all — its ingredients just need to be inter-preted in the correct way.The Maldacena conjecture raised a lot of interest, but seemed for a long time to be quantitatively unverifiable. This was because it takes the form of a duality in which the strongly coupled string theory corresponds to weakly coupled QCD-like theory, and vice versa. But to verify the duality, one would need to find a quantity to compare in a regime of intermediate coupling strength, and calculate it starting from both sides. No such quantity was obvious.Help came from an entirely unexpected direction. Following a prescient observa-tion13, the spectrum of the N = 4 theory has been found1,2 to be equivalently described by a quantum-mechanical spin chain of a type discovered by Hans Bethe in 1931 when modelling certain metallic systems. There are not many quantum-mechanical systems that can be solved analytically — the hydro-gen atom is the most prominent example — but Bethe’s ansatz immediately applied in a much wider context, and constructed a bridge between condensed-matter physics and string theory (in this context, see the recent News & Views article by Jan Zaanen14 on the nascent connection to high-temperature superconductivity). Indeed, even though the mathematical description of the duality on the string-theory side is completely differ-ent from that on the condensed-matter side, a very similar, exactly solvable structure has been identified here as well3–5.Puzzling out the details of the exact solution is currently an active field of research. But in one instance, that idea had already been put to such a hard test that a complete solution now seems within reach. The context is a special observable entity, the ‘cusp anomalous dimen-sion’, which was argued6,7 to be ideally suited as a device to test whether string and gauge theory really connect. Some of its structure at strong coupling was also worked out. Just recently, Beisert, Eden and Staudacher8 have extracted the analogue of this observable on the field-theory side, and have been able to write down an equation valid at any strength of the coupling. Since then, work has established that their ‘BES equation’ does indeed seem, for the first time, to offer a means of reformulating theories such as QCD as string theories.Much still needs to be learned from this one exactly solvable case. There is justifiable hope that this solution will teach us how to go back to the physically relevant case of QCD and finally arrive at the long-sought dual description by a string theory. It may even take us closer to realizing the