Language: Specifying the Site of Modality-Independent Meaning

Current Biology Vol 16 No 21 R930

9. Brown, P.N., Mathews, M.A., Joss, L.A., Hill, C.P., and Blair, D.F. (2005). Crystal structure of the flagellar rotor protein FliN from Thermotoga maritima. J. Bacteriol. 187, 2890–2902. 10. Shaikh, T.R., Thomas, D.R., Chen, J.Z., Samatey, F.A., Matsunami, H., Imada, K., Namba, K., and Derosier, D.J. (2005). A partial atomic structure for the flagellar hook of Salmonella typhimurium. Proc. Natl. Acad. Sci. USA 102, 1023–1028, Epub 2005 Jan 1018. 11. Thomas, D.R., Francis, N.R., Xu, C., and DeRosier, D.J. (2006). The threedimensional structure of the flagellar rotor

from a clockwise-locked mutant of Salmonella enterica serovar typhimurium. J. Bacteriol., in press. 12. Yonekura, K., Maki, S., Morgan, D.G., DeRosier, D.J., Vonderviszt, F., Imada, K., and Namba, K. (2000). The bacterial flagellar cap as the rotary promoter of flagellin self-assembly. Science 290, 2148–2152. 13. Khan, S., Khan, I.H., and Reese, T.S. (1991). New structural features of the flagellar base in Salmonella typhimurium revealed by rapid-freeze electron microscopy. J. Bacteriol. 173, 2888– 2896.

Language: Specifying the Site of Modality-Independent Meaning Language processing can be triggered by auditory, visual or somatosensory input. A recent study has provided new insight into a fundamental issue raised by this observation: how is knowledge of language implemented in the human brain such that speakers can use any type of sensory-motor input–output system for comprehension and production? David Poeppel Here is a bad idea: there are separate, modality-specific language systems — at least three — and language processing is completely dependent on the properties of the sensory modality. Baroque explanations would be required to account for why heard and read language appears to be identical at the level of meaning. A better and simpler idea is this: there exists a ‘core’ linguistic computational system that specifies what a speaker knows about his or her mental lexicon, phonology, syntax and semantics [1]. This system represents knowledge in a way that permits its translation into different sensory-motor interfaces. After all, the message ‘‘Your paper is rejected’’ is interpreted the same way — each equally painful — whether experienced by sound, sight or touch. But is there evidence that the different modalities converge onto neuronal populations that mediate meaning independent of modality? There must be some type of convergence using shared neural codes. A new imaging paper from Richard Wise’s group [2] provides stimulating new insights into this question.

After about 150 years of systematic research on the neural basis of language, we know virtually nothing about the neural coding at the basis of linguistic experience. We do, however, have a growing body of data about the neurophysiological foundations of language [3], based principally on the results of EEG and MEG studies as well as on important clinical studies. There is also a rich literature on the functional neuroanatomy of language, deriving both from deficit–lesion studies in patients and, more recently, from functional brain imaging [4]. Recent functional/anatomical models of language processing reflect an emerging consensus, although the emphasis must be on emerging. Most textbooks still provide a cartoon left hemisphere highlighting an inferior frontal region, Broca’s area, and a posterior temporal region, Wernicke’s area, that are suggested to form the basis for language processing. But it is now indisputable that there are many other cortical and subcortical areas implicated in speech and language processing and that the right hemisphere plays a crucial role as well [5–7].

14. Khan, S., Dapice, M., and Reese, T.S. (1988). Effects of mot gene expression on the structure of the flagellar motor. J. Mol Biol. 202, 575–584.

Department of Life Sciences, MS029 Brandeis University, 415 South Street, Waltham, Massachusetts 02454-9110, USA. E-mail: [email protected]

DOI: 10.1016/j.cub.2006.09.053

In new imaging work using positron emission tomography (PET) and working with ecologically natural language stimuli, Spitsyna et al. [2] make a provocative contribution to the question of how ‘verbal meaning’ converges neuroanatomically and functionally in the brain. They report a network of four left-lateralized areas that are argued to mediate meaning independent of modality. In their experiment, participants were presented with two kinds of ecologically natural linguistic stimulus. In one condition, subjects heard a one-minute duration segment of connected speech (per experimental block); in another, they were shown a paragraph of text (per block). To control for modality-specific input processing, subjects also heard ‘rotated speech’ [8] or were shown false-font visual stimuli. The prediction tested was, roughly: if there is convergence onto cortical areas responsible for the processing of ‘verbal meaning’, experimental conditions driving different input modalities should still activate the same areas responsible for the supramodal extraction of meaning. The underlying controversy is the following. According to one view, informed by lesion and imaging data, the processing of (lexical-level) meaning is primarily mediated in posterior aspects of the superior temporal lobe and the middle temporal gyrus [7], reminiscent of the classical findings by Wernicke from 1874. But, on the other hand, clinical research on semantic dementia has implicated the anterior and inferior temporal lobe in the processing of meaning. Who is

Dispatch R931

right — anybody, everybody or nobody? Spitsyna et al. [2] identified four left-lateralized areas associated with processing verbal meaning, regardless of whether the stimulation was visual or auditory. The four major activation foci were the anterior superior temporal sulcus, the posterior temporo-occipito-parietal junction, the lateral temporal pole, as well as the anterior fusiform gyrus. Furthermore, the response profile suggested two groupings, with superior temporal sulcus and temporal pole showing similar responsivity, and the anterior fusiform gyrus and temporo-occipito-parietal junction constituting the second group. This is the first time such a network has been identified. In particular, finding a fusiform gyrus–temporo-occipito-parietal junction axis is surprising. The results seem to suggest that both the ‘anteriorists’ and the ‘posteriorists’ of the temporal lobe are, in some respects, correct. Additionally, Spitsyna et al. [2] challenge the hypothesis that aspects of syntactic computation are executed in the superior anterior temporal lobe. Two issues merit brief comment. First, what exactly is the notion of ‘verbal meaning’? ‘Meaning’ is not monolithic. From the perspective of language research, there are different aspects of meaning that must be distinguished, because extensive evidence shows that these different facets have distinct organizing principles, processing requirements, and neuronal infrastructure. For example, conceptual semantics refers to knowledge one has about the various attributes of a concept, independent of the linguistic realization — for example are typically four-legged and bark. This may be the aspect of meaning compromised in semantic dementia. Lexical semantics, on the other hand, refers to formal linguistic properties of single words that have precise processing consequences — for example, ‘bite’ is an ‘eventive verb’ and differs from ‘admire’, a ‘stative verb’, and verb types differ in their processing requirements. Lexical

meaning at this level is typically associated with posterior middle temporal cortex. Compositional semantics, closely connected to syntactic structure, concerns how meaning is constructed in sentential contexts, for example, allowing one to distinguish ‘dog bites man’ from ‘man bites dog’. Because of the tight link to syntax, areas sensitive to structural information are likely to be critical. The fact that there are different types of meaning makes unsurprising the observation that the ‘neural basis of meaning’ has been associated with many different activation profiles. For instance, recent imaging data from other labs suggest that left inferior frontal gyrus anterior to Broca’s area plays a critical role in verbal meaning [9]; and the potential role of parietal cortex has been highlighted as well [10]. To complicate things further, electrophysiological studies show that right superior and middle temporal lobe structures are robustly implicated [11]. On balance, therefore, the data across methods and studies are not yet converging on a single model of the calculation of meaning in the brain. A second point to consider is the use of ecologically valid materials in an imaging study. Unquestionably, it is a major goal of research in cognitive neuroscience of language to understand the system under ecologically natural conditions. Years of research and dozens of studies on the distinction between ‘ba’ and ‘pa’, or the activation of multiple meanings of the word ‘bank’ have possibly reduced enthusiasm for this area of research, despite the important insights that it has provided. Indeed, much recent visual neurophysiological research is turning to the analysis of real visual scenes. But there is, as usual, a price to pay. For instance, presenting connected speech or written text elicits concurrent semantic processing at all levels — in addition to all the computations on which the extraction of meaning depends, including lexical access, syntactic structure building, and so on. Indeed, an older PET study by

Mazoyer et al. [12] also used ecological stimulation and concluded that anterior superior temporal lobe may be essential for elementary syntactic structure building, a hypothesis congruent with recent DTI tractography data [13]. The new data of Spitsyna et al. [2] do not really challenge this hypothesis. Because of the neuroanatomical and data-analytical sophistication of this new study [2], one must take very seriously the possibility that there exists a network of left-lateralized anterior and posterior areas that underlies the computation of meaning. But it is unresolved what underlying operations are executed in each part of the network. An interpretation at that level requires more integration with psycholinguistic and computational research that attempts to fractionate the processes at the appropriate granularity to assign them to neuronal circuitry [14]. We do, however, now have a set of candidate areas which can be investigated with an eye towards specifying the computational primitives that allow verbal meaning to be constructed. References 1. Chomsky, N. (1986). Knowledge of Language: Its Nature, Origin, and Use (New York: Praeger). 2. Spitsyna, G., Warren, J.E., Scott, S.K., Turkheimer, F.E., and Wise, R.J.S. (2006). Converging language streams in the human temporal lobe. J. Neurosci. 26, 7328–7336. 3. M.D. Rugg and M.G. Coles, eds. (1995). Electrophysiology of Mind (New York: Oxford University Press). 4. G. Hickok and D. Poeppel, eds. (2004). Towards a new functional anatomy of language. Cognition 92, Special Issue. 5. Friederici, A.D. (2002). Towards a neural basis of auditory sentence processing. Trends Cog. Sci. 6, 78–84. 6. Demonet, J.-F., Thierry, G., and Cardebat, D. (2005). Renewal of the neurophysiology of language: Functional neuroimaging. Physiol. Rev. 85, 49–95. 7. Hickok, G., and Poeppel, D. (2004). Dorsal and ventral streams: a framework for understanding aspects of the functional anatomy of language. Cognition 92, 67–99. 8. Scott, S.K., Blank, C.C., Rosen, S., and Wise, R.J.S. (2000). Identification of a pathway for intelligible speech in the left temporal lobe. Brain 123, 2400–2406. 9. Thompson-Schill, S., D’Esposito, M., Aguirre, G.K., and Farah, M.J. (1997). Role of left inferior prefrontal cortex in retrieval of semantic knowledge: A reevaluation. Proc. Natl. Acad. Sci. USA 94, 14792– 14797. 10. Price, C.J. (2000). The anatomy of language: contributions from functional neuroimaging. J. Anat. 197, 335–359.

Current Biology Vol 16 No 21 R932

11. Federmeier, K., and Kutas, M. (1999). Right words and left words: electrophysiological evidence for hemispheric differences in meaning processing. Cog. Brain Res. 8, 373–392. 12. Mazoyer, B., Dehaene, S., Tzourio, N., Frak, V., Murayama, N., Cohen, L., Levrier, O., Salamon, G., Syrota, A., and Mehler, J. (1993). The cortical representation of speech. J. Cog. Neurosci. 4, 467–479.

13. Friederici, A.D., Bahlmann, J., Heim, S., Schubotz, R.I., and Anwander, A. (2006). The brain differentiates human and nonhuman grammars: Functional localization and structural connectivity. Proc. Natl. Acad. Sci. USA 103, 2458–2463. 14. Poeppel, D., and Embick, D. (2005). The Relation between linguistics and neuroscience. In Twenty-First Century Psycholinguistics: Four Cornerstones, A. Cutler, ed. (Mahwah NJ: Erlbaum).

DNA-Damage Control: Claspin Destruction Turns off the Checkpoint The ATR–Claspin–Chk1 pathway is critical for turning on the cellular response to DNA damage and replication stress. Five recent reports uncover new mechanisms controlling the recovery phase of the checkpoint response, and introduce crucial roles for Claspin, Rad17 phosphorylation and the ubiquitin proteasome pathway in Chk1 signaling. Benjamin E. Gewurz and J. Wade Harper Proper duplication of chromosomes is critical to cellular function and organismal development. Errors in this process can result in altered cellular pathways, aneuploidy, and disease. Organisms have consequently evolved elaborate control mechanisms to ensure the faithful transmission of their genetic material [1]. Cells respond to DNA damage and replication blocks by activating the DNA-damage response network, a system which arrests the cell cycle and facilitates DNA repair. The machinery that senses damage and activates repair systems has been well studied. In contrast, far less is known about how the damage signal is terminated upon completion of DNA repair. In a series of elegant papers published recently in Current Biology, Molecular Cell and Cell [2–6], several groups have uncovered important regulatory mechanisms that govern activation and turnover of Claspin — a critical mediator of the DNA-damage signaling system. Controlling Claspin abundance through the ubiquitin proteasome pathway appears to be at the heart of checkpoint recovery.

Linking Rad17 to the ATR– Claspin–Chk1 Signaling Complex The DNA-damage response pathway is composed of sensors, mediators, signal transducers and effectors [1]. However, because sensors form complexes with mediators and transducers, the lines between these different functionalities are somewhat blurred. The ATR–ATRIP protein kinase complex is a critical component of the cellular response to DNA damage and replication blocks [7]. ATR–ATRIP associates with single-stranded DNA at sites of damage with the help of replication protein A (RPA) and thus functions as a sensor of damage [8]. In concert, the Rad17 protein loads the DNA clamp Rad9–Rad1–Hus1 onto chromatin [9]. Rad17 binds chromatin independently of ATR, but in response to damage is phosphorylated by ATR at two sites (serine 635 and serine 645). These events culminate in the activation of Chk1 by ATR–ATRIP. Although it was clear that activation of Chk1 requires ATR, Rad17 and the Chk1 associated protein Claspin, the precise role of Rad17 phosphorylation and the relationship between Rad17 and Claspin was largely unknown [10]. To address these questions, Wang et al. [6] replaced cellular

Department of Linguistics and Department of Biology, University of Maryland College Park, 1401 Marie Mount Hall, College Park, Maryland 20742, USA. E-mail: [email protected]

DOI: 10.1016/j.cub.2006.09.047

Rad17 with a mutant form of Rad17, Rad17AA, which is defective in phosphorylation by ATR–ATRIP. They found that, although Rad17 phosphorylation is not required for chromatin binding or survival in culture, cells expressing Rad17AA displayed a shortened S-phase and increased rates of spontaneous chromosome breakage [6]. Rad17 phosphorylation was also found to be necessary for cell survival in the presence of hydroxyurea, a ribonucleotide reductase inhibitor and replication checkpoint activator, but not for survival after ultraviolet exposure. Thus, distinct damage networks differentially rely on Rad17 phosphorylation by ATR–ATRIP. Chk1 is also required for cell survival in the presence of hydroxyurea, raising the question of whether Rad17 phosphorylation promotes Chk1 activation. Wang et al. [6] showed that, in response to hydroxyurea, Chk1 activation was reduced and prematurely terminated after hydroxyurea removal in Rad17AA-expressing cells. Interestingly, Claspin forms a complex with Rad17 in a manner that depends on Rad17 phosphorylation. Moreover, phosphorylation of Claspin was also blocked in Rad17AA-expressing cells. Taken together, these data are consistent with a model in which ATR–ATRIP phosphorylation of Rad17 promotes ATR phosphorylation of Claspin, possibly via a Claspin–Rad17 complex. This presumably promotes Chk1 recruitment to Claspin, and ultimately Chk1 activation by ATR. This model is consistent with the observation that Rad17 phosphorylation is required to maintain Chk1 activation during the early periods