Network Protocol

If you think understanding a computer's inner workings is complicated, consider how much harder it is to decipher the functioning of the brain. Networks of neurons engage in sophisticated signaling; any hope of interpreting their activity patterns requires the ability to look at a given neuron in the context of its local circuitry. Calcium indicator dyes are reagents of choice for monitoring neuronal activity; however, as the number of neurons increases, the technical challenges become formidable. Direct injection of a calcium indicator into a neuron is an informative approach that quickly becomes impractical for neuronal ensembles. On the other hand, a membrane-permeant dye can easily label thousands of neurons, but background staining limits resolution. Electroporation of membrane-impermeable calcium indicator dyes might represent an effective means to load local neuronal circuits for subsequent imaging; after all, dextran-conjugated calcium dyes have been successfully electroporated into neurons in in vitro preparations of neonatal spinal cord. Sure enough, Nagayama et al., have successfully achieved neuronal ensemble labeling in vivo. Their setup includes a dye-filled glass pipet that is inserted into the brain of a craniotomized mouse. Current flow (from the pipet to the grounded tail) results in cell-membrane electroporation, labeling cells in the vicinity of the pipet. This process does not measurably alter local circuit function and allows measurement of calcium responses to stimulation of the olfactory bulb via delivery of odorants and monitoring of barrel cortex signaling in response to whisker deflection. Data for successful recording of neurons in the fourth and fifth lobules of the cerebellum were also shown, as well as evidence demonstrating electroporation of salt-form calcium indicator dyes. The latter development opens the possibility of multicolor calcium imaging. Overall, the prospects for this methodology are extremely encouraging. The relative paucity of cells that are labeled makes picking out individual axons and dendrites practical, which should bring additional data regarding the interplay of functional connectivity and spatiotemporal activity. The technology may even permit simultaneous DNA transfection of neuronal ensembles, allowing collection of detailed functional data.

- Nagayama et al. 2007. In vivo simultaneous tracing and Ca²⁺ imaging of local neuronal circuits. Neuron 53:789-803.

On Target

Until efficient, targeted gene insertion becomes reliable, human gene therapy will likely remain an approach in which hype far outweighs results. Although retroviral vectors are undeniably effective in integrating stably into chromosomal DNA, directing this process to specific genomic locations is likely to prove problematic. The complications arise because of the nature of the viral integrase, which interacts with chromatin-associated proteins, thereby steering integration toward transcriptionally active genes. These interactions can subvert DNA targeting mediated by DNA binding domains that have been added to integrases to create hybrid proteins. One alternative involves steering clear of retroviral integrases entirely and turning to transposons. The Sleeping Beauty (SB) transposon integrates readily into mammalian chromosomes, though with little sequence selectivity beyond a favored AT-rich palindrome. Although the challenge of redirecting SB from one of its estimated 200 million potential integration sites to a targeted location is daunting, the SB transposase does not appear to interact with any cellular factors that might overrule attempts to engineer the enzyme. Thus, Yant et al., select the SB transposase for an effort to achieve sequence-specific integration into human cells. The authors follow the standard strategy of fusing a DNA binding domain to the enzyme responsible for integration (in this case, the SB transposase). As a DNA binding domain, the authors selected the synthetic zinc-finger protein E2C, which binds an 18-bp DNA sequence on chromosome 17. Targeted integration was tested in HeLa cells cotransfected with donor and acceptor plasmids and a plasmid expressing the hybrid transposase. Although overall transposition events were reduced when the hybrid protein was deployed, the specificity of the integration was enhanced by roughly an order of magnitude. However, no targeted integration was observed when testing for insertion into the native chromosomal sequence. Although this effort is still far away from practical gene therapy applications, the results nonetheless show the first bona fide targeted DNA transposition in human cells.

- Yant et al. 2007. Site-directed transposon integration in human cells. Nucleic Acids Research [Epub ahead of print, March 7, 2007].

Under Pressure

Like their colleagues in genomics, proteomics researchers find themselves under mounting pressure to obtain more (and better-quality) data in a shorter period of time. One increasingly popular strategy involves forgoing multidimensional separations prior to mass spectrometry and instead using single-dimension liquid chromatography performed under ultrahigh pressure. Such experiments use a relatively slow flow rate in a column with a narrow inner diameter, a combination that requires long-duration regeneration times and that is unmanageably sluggish when large injection volumes must be used. The latter problem has inspired the addition of preseparation solid phase extraction columns, which brings more efficient sample loading and convenient desalting. The solution to the regeneration time issue may be an apparatus with a decidedly unstreamlined name: DO-SPE/cRPLC/MS/MS (dual online solid phase extraction/capillary reverse-phase liquid chromatography/tandem mass spectrometry). As implied by its moniker, the system includes two SPE columns that connect to two capillary separation columns. In this case, each SPE column has an inner diameter of 250 µm and feeds into an RPLC column with an inner diameter of 75 µm. As expected, the larger diameter of the SPE column permits faster injection times and efficient desalting. The real advantage of DO-SPE/cRPLC for tandem MS is the improved throughput. Normally, a cRPLC column with these dimensions would need an equilibration time of over 2 hours. The dual system cuts that wait down by a factor of five. Importantly, intercolumn reproducibility over time was strong, as only minor differences in elution time and peak intensity were observed. Since these features were apparent in tests of a complex sample (yeast lysate), the dual system should be applicable to real-world biological samples with only minimal optimization.

- Min et al. 2007. Ultrahigh-pressure dual online solid phase extraction/capillary reverse-phase liquid chromatography/tandem mass spectrometry (DO-SPE/cRPLC/MS/MS): a versatile separation platform for high-throughput and highly sensitive proteomic analyses. Electrophoresis [Epub ahead of print, February 16, 2007].

Sense from Missense

Women who choose to assess their relative risk for breast and ovarian cancers by genetic testing of the BRCA1 gene seek definitive information. In the case of mutations that cause a prematurely truncated protein, the deleterious effects are quite well understood. However, information about whether a given missense mutation is clinically important lags sorely behind. The difficulty is not only that there are so many missense mutations that have been described, but also that the prevalence of these variants is so low that determining risk via epidemiology is essentially impossible. Obviously, biochemical characterization could yield detailed data. However, characterizing just one variant might exhaust 3 weeks of personnel time; with hundreds of mutations to investigate, the time and expense required quickly become overwhelming. In response to this impediment, Karchin et al., describe computational methods that enable prediction of which variants are most likely to be of functional significance. First, the authors collected 16 features that are associated with protein structure (e.g., solvent accessibilities of the wild-type and variant residues), physiochemical properties (e.g., polarity), and evolutionary conservation. These features were fed into supervised learning algorithms, which were also given a training set comprising data relating to 618 well-characterized missense variants in p53. The authors then challenged the algorithms with a validation set of 36 BRCA1 missense mutations that they characterized biochemically (the data appear in an article in press at Cancer Research). Although a number of computational procedures were tested, the three top performers were all supervised learning algorithms; together, these agreed with the functional data 94% of the time. As a next step, the algorithms were thrown 54 uncharacterized variants of BRCA1. Based on the results returned, the authors could often infer structural explanations for the predicted effect (or lack thereof). The strength of this method is the ability to apply rules gleaned from one protein (p53) to another (BRCA1). Although this approach is restricted to proteins that have accurate structural information, the authors intend to demonstrate its generalizability to BRCA2 and to explore applicability to other genes linked to familial cancer syndromes.

- Karchin et al. 2007. Functional impact of missense variants in BRCA1 predicted by supervised learning. PLoS Computational Biology [Epub ahead of print, February 16, 2007].