Adeno-Associated Virus Replication Induces a DNA Damage Response Coordinated by DNA-Dependent Protein Kinase

The parvovirus adeno-associated virus (AAV) contains a small single-stranded DNA genome with inverted terminal repeats that form hairpin structures. In order to propagate, AAV relies on the cellular replication machinery together with functions supplied by coinfecting helper viruses such as adenovirus (Ad). Here, we examined the host cell response to AAV replication in the context of Ad or Ad helper proteins. We show that AAV and Ad coinfection activates a DNA damage response (DDR) that is distinct from that seen during Ad or AAV infection alone. The DDR was also triggered when AAV replicated in the presence of minimal Ad helper proteins. We detected autophosphorylation of the kinases ataxia telangiectasia mutated (ATM) and DNA-dependent protein kinase catalytic subunit (DNA-PKcs) and signaling to downstream targets SMC1, Chk1, Chk2, H2AX, and XRCC4 and multiple sites on RPA32. The Mre11 complex was not required for activation of the DDR to AAV infection. Additionally, we found that DNA-PKcs was the primary mediator of damage signaling in response to AAV replication. Immunofluorescence revealed that some activated damage proteins were found in a pan-nuclear pattern (phosphorylated ATM, SMC1, and H2AX), while others such as DNA-PK components (DNA-PKcs, Ku70, and Ku86) and RPA32 accumulated at AAV replication centers. Although expression of the large viral Rep proteins contributed to some damage signaling, we observed that the full response required replication of the AAV genome. Our results demonstrate that AAV replication in the presence of Ad helper functions elicits a unique damage response controlled by DNA-PK.Replication of viral genomes produces a large amount of extrachromosomal DNA that may be recognized by the cellular DNA damage machinery. This is often accompanied by activation of DNA damage response (DDR) signaling pathways and recruitment of cellular repair proteins to sites of viral replication. Viruses therefore provide good model systems to study the recognition and response to DNA damage (reviewed in reference 48). The Mre11/Rad50/Nbs1 (MRN) complex functions as a sensor of chromosomal DNA double-strand breaks (DSBs) and is involved in activation of damage signaling (reviewed in reference 41). The MRN complex also localizes to DNA DSBs and is found at viral replication compartments during infection with a number of DNA viruses (6, 40, 47, 70, 75, 77, 87, 93). The phosphatidylinositol 3-kinase-like kinases (PIKKs) ataxia telangiectasia mutated (ATM), ATM and Rad3-related kinase (ATR), and the catalytic subunit of the DNA-dependent protein kinase (DNA-PKcs) are involved in the signal transduction cascades activated by DNA damage (reviewed in references 43, 51, and 71). These kinases respond to distinct types of damage and regulate DSB repair during different phases of the cell cycle (5), either through nonhomologous end-joining (NHEJ) or homologous recombination pathways (reviewed in references 63, 81, and 86). The DNA-PK holoenzyme is composed of DNA-PKcs and two regulatory subunits, the Ku70 and Ku86 heterodimer. DNA-PK functions with XRCC4/DNA ligase IV to repair breaks during NHEJ, and works with Artemis to process DNA hairpin structures during VDJ recombination and during a subset of DNA DSB events (46, 50, 86). While the kinase activity of DNA-PKcs leads to phosphorylation of a large number of substrates in vitro as well as autophosphorylation of specific residues (reviewed in references 16 and 85), it is currently unclear how DNA-PKcs contributes to signaling in cells upon different types of damage.The adeno-associated virus (AAV) genome consists of a molecule of single-stranded DNA with inverted terminal repeats (ITRs) at both ends that form double-hairpin structures due to their palindromic sequences (reviewed in reference 52). The ITRs are important for replication and packaging of the viral genome and for integration into the host genome. Four viral Rep proteins (Rep78, Rep68, Rep52, and Rep40) are also required for replication and packaging of the AAV genome into virions assembled from the Cap proteins. Although the Rep and Cap genes are replaced in recombinant AAV vectors (rAAV) that retain only the ITRs flanking the gene of interest, these vectors can be replicated by providing Rep in trans (reviewed in reference 7). Productive AAV infection requires helper functions supplied by adenovirus (Ad) or other viruses such as herpes simplex virus (HSV) (reviewed in reference 27), together with components of the host cell DNA replication machinery (54, 55, 58). In the presence of helper viruses or minimal helper proteins from Ad or HSV, AAV replicates in the nucleus at centers where the viral DNA and Rep proteins accumulate (35, 76, 84, 89). Cellular and viral proteins involved in AAV replication, including replication protein A (RPA), Ad DNA-binding protein (DBP), and HSV ICP8, localize with Rep proteins at these viral centers (29, 33, 76).A number of published reports suggest associations between AAV and the cellular DNA damage machinery. For example, transduction by rAAV vectors is increased by genotoxic agents and DNA damaging treatments (1, 62, 91) although the mechanisms involved remain unclear. Additionally, the ATM kinase negatively regulates rAAV transduction (64, 92), and we have shown that the MRN complex poses a barrier to both rAAV transduction and wild-type AAV replication (11, 67). UV-inactivated AAV particles also appear to activate a DDR involving ATM and ATR kinases that perturbs cell cycle progression (39, 60, 88). It has been suggested that this response is provoked by the AAV ITRs (60) and that UV-treated particles mimic stalled replication forks in infected cells (39). In addition to AAV genome components, the viral Rep proteins have been observed to exhibit cytotoxicity and induce S-phase arrest (3, 65).The role of cellular repair proteins in AAV genome processing has also been explored by examining the molecular fate of rAAV vectors, which are converted into circular and concatemeric forms that persist episomally (18, 19, 66). Proteins shown to regulate circularization in cell culture include ATM and the MRN complex (14, 64), while in vivo experiments using mouse models have implicated ATM and DNA-PK in this process (14, 20, 72). Additionally, DNA-PKcs and Artemis have recently been shown to cleave the ITR hairpins of rAAV vectors in vivo in a tissue-dependent manner (36). Despite these studies, it is not clear how damage response factors function together and how they impact AAV transduction and replication in human cells.In this study we examined the cellular response to AAV replication in the context of Ad infection or helper proteins. We show that coinfection with AAV and Ad activates a DDR that is distinct from that seen during infection with Ad alone. The ATM and DNA-PKcs damage kinases are activated and signal to downstream substrates, but the response does not require the MRN complex and is primarily mediated by DNA-PKcs. Although expression of the large Rep proteins induced some DDR events, full signaling appeared to require AAV replication and was accompanied by accumulation of DNA-PK at viral replication compartments. Our results demonstrate that AAV replication induces a unique DNA damage signal transduction response and provides a model system for studying DNA-PK.     

Adeno-Associated Virus Small Rep Proteins Are Modified with at Least Two Types of Polyubiquitination

Adeno-associated virus (AAV) type 2 and 5 proteins Rep52 and Rep40 were polyubiquitinated during AAV-adenovirus type 5 (Ad5) coinfection and during transient transfection in either the presence or absence of Ad5 E4orf6 and E1b-55k. Polyubiquitination of small Rep proteins via lysine 48 (K48) linkages, normally associated with targeting of proteins for proteasomal degradation, was detected only in the presence of E4orf6. The small Rep proteins were ubiquitinated via lysine 63 (K63) following transfection in either the presence or absence of E4orf6 or following coinfection with Ad5. E4orf6/E1b-55k-dependent K48-specific polyubiquitination of small Rep proteins could be inhibited using small interfering RNA (siRNA) to cullin 5.Together, adenovirus type 5 (Ad5) early gene products E1a, E1b-55k, E2a, E4orf6, and virus-associated (VA) RNA can support efficient replication of adeno-associated virus (AAV) (4, 31). E4orf6 and E1b-55k are known to interact with cellular cullin 5 (cul5), elongins B and C, and the ring box protein Rbx1 to form an E3 ubiquitin ligase complex that specifically targets a small population of cellular proteins for degradation by the proteasome (1, 7, 21, 22, 24, 27). This property has been implicated in a number of functions presumed to be required for both Ad and AAV replication (3, 8-10, 17, 23, 24, 34, 35).Previously, only p53, Mre11, DNA ligase IV, and integrin α3 had been shown to be substrates of the Ad5 E3 ubiquitin ligase complex (1, 7, 21, 22, 24, 27); however, we have recently shown (16, 17) that the small Rep proteins and capsid proteins of AAV5 are also degraded in the presence of Ad E4orf6 and E1b-55k in a proteasome-dependent manner. These proteins were restored to levels required during infection by the action of VA RNA (17). The targeting for degradation of AAV5 protein by the E4orf6/E1b-55k E3 ubiquitin ligase complex required functional BC-box motifs in E4orf6 and could be inhibited by depletion of the scaffolding protein cullin 5 using directed small interfering RNA (siRNA) (16). In addition, the degradation of AAV5 protein was partially prevented by overexpression of pUBR7, a plasmid that generates a dominant-negative ubiquitin (16). The role this targeted degradation plays in the life cycle of AAV has not yet been clarified; however, E4orf6 mutants that cannot function in this regard do not support AAV replication as well as wild-type E4orf6 (R. Nayak and D. J. Pintel, unpublished data). Degradation of Mre11 by the Ad5 E3 ligase has also been implicated in allowing efficient Ad5 and AAV replication (24). Ubiquitination of AAV Rep proteins during viral infection, however, has not previously been reported.     

In vitro and in vivo gene therapy vector evolution via multispecies interbreeding and retargeting of adeno-associated viruses

Adeno-associated virus (AAV) serotypes differ broadly in transduction efficacies and tissue tropisms and thus hold enormous potential as vectors for human gene therapy. In reality, however, their use in patients is restricted by prevalent anti-AAV immunity or by their inadequate performance in specific targets, exemplified by the AAV type 2 (AAV-2) prototype in the liver. Here, we attempted to merge desirable qualities of multiple natural AAV isolates by an adapted DNA family shuffling technology to create a complex library of hybrid capsids from eight different wild-type viruses. Selection on primary or transformed human hepatocytes yielded pools of hybrids from five of the starting serotypes: 2, 4, 5, 8, and 9. More stringent selection with pooled human antisera (intravenous immunoglobulin [IVIG]) then led to the selection of a single type 2/type 8/type 9 chimera, AAV-DJ, distinguished from its closest natural relative (AAV-2) by 60 capsid amino acids. Recombinant AAV-DJ vectors outperformed eight standard AAV serotypes in culture and greatly surpassed AAV-2 in livers of naïve and IVIG-immunized mice. A heparin binding domain in AAV-DJ was found to limit biodistribution to the liver (and a few other tissues) and to affect vector dose response and antibody neutralization. Moreover, we report the first successful in vivo biopanning of AAV capsids by using a new AAV-DJ-derived viral peptide display library. Two peptides enriched after serial passaging in mouse lungs mediated the retargeting of AAV-DJ vectors to distinct alveolar cells. Our study validates DNA family shuffling and viral peptide display as two powerful and compatible approaches to the molecular evolution of novel AAV vectors for human gene therapy applications.A large number of inherited or acquired diseases remain promising targets for human gene therapy. One vector that has shown outstanding potential thus far in numerous preclinical and clinical evaluations is based on nonpathogenic adeno-associated virus (AAV). A unique asset among various properties that make AAV especially attractive over its competitors, such as adenoviral or lentiviral vectors, is the availability of a vast number of natural isolates which differ significantly in their properties (24). We and others have shown previously that the function of an AAV vector particle is determined mainly by the capsid protein and that viral Rep proteins and genomic packaging elements are largely interchangeable (24, 27, 85). Paradoxically, the ever-increasing repertoire of naturally occurring and synthetically generated AAV capsid sequences (>300 to date) is currently creating a dilemma for the rational selection of the optimal serotype for a given application. The importance of finding the ideal capsid for efficient and safe gene transfer has been exemplified in many preclinical studies, as well as in a clinical trial using the AAV type 2 (AAV-2) prototype in human liver tissue (36, 47). In one previous study, the treatment of patients with severe hemophilia B with recombinant AAV-2 expressing human factor IX (hFIX) resulted in mildly elevated, yet therapeutic, levels of this blood coagulation factor. However, expression was short lived, and the hFIX decline was accompanied by a transient asymptomatic increase of liver transaminases, due to a T-cell immune response against the AAV-2 capsid (47). Also, preexisting neutralizing anti-AAV-2 antibodies (frequent in humans) in these individuals likely inhibited the linear vector dose response previously observed in animals.We and others have suggested previously that the use of novel AAV serotypes, in particular, nonhuman isolates, will help to overcome some of these problems (19, 24, 63). Important examples are AAV-8 and AAV-9, which can transduce mouse liver far better than AAV-2, albeit the difference in dogs or primates is less clear (17, 52, 54, 75). The potential for the complete transduction of liver tissue and perhaps other tissues makes these two non-AAV-2 serotypes also particularly interesting for therapeutic RNA interference (RNAi) (28). We recently demonstrated the feasibility of efficiently and persistently suppressing hepatitis B virus with RNAi from a double-stranded AAV-8 vector (28). On the other hand, a potential drawback of AAV-8 and AAV-9 is their lack of specific tissue tropism (34, 52). The resulting frequent vector dissemination into all organs, including the brain, even from low peripheral doses in mice or monkeys (52, 54) is a particular concern for RNAi therapies in which control over vector biodistribution and the limitation of off-target effects will be imperative for the success of the approach (28).In order to overcome the constraints of wild-type AAV serotypes, numerous groups have recently begun to develop novel strategies to engineer “designer” AAVs tailored for the therapeutic transduction of clinically relevant organs (reviewed in detail in references 9, 12, 35, 41, 51, and 85). Briefly, the variety of strategies can be grouped into indirect or chemical approaches and direct physical modification strategies. In the indirect approaches, specific molecules (e.g., bispecific antibodies [6] or avidin-coupled ligands [4]) are allowed to react with the viral surface (biotinylated in the case of avidin [4]), as well as a cellular receptor, forming a conjugate ideally able to retarget the capsid to a refractory cell type. Yet, numerous pharmacological problems, such as concerns about in vivo complex stability and difficulties in upscaling complex manufacturing, continue to prevent the broad adaptation of these approaches. Alternative, more powerful strategies rely on the direct physical modification of the AAV capsid protein and gene. Early examples include the generation of mosaic AAV capsids via the mixing of helper plasmids carrying capsid genes from distinct serotypes, such as AAV-1 and AAV-2 (30) and pairwise combinations of AAV-1 through AAV-5 (62). Similar mosaics were generated previously via a marker rescue approach, yielding AAV-2/AAV-3 recombinants with unique properties (8). A related strategy is the rational creation of chimeric virions via domain swapping among multiple parental serotypes, involving either entire capsid loops or parts thereof or individual residues. Notable examples include AAV-1/AAV-2 chimeras with improved tropism in muscle tissue (31), with one of these chimeras presently being studied in a phase I clinical trial for the treatment of Duchenne muscular dystrophy (85). Most recently, our own group described a battery of unique chimeras comprising elements from serotypes 2 and 8, which were exploited to identify capsid subdomains responsible for efficient AAV transduction in murine liver tissue in vivo (64).A special type of chimeric capsids are those containing foreign proteins or peptides inserted into various positions of the virion shell. The methods and strategies used are widely diverse, and again, we refer to comprehensive reviews (12, 35, 41). Noteworthy here are approaches to fuse targeting ligands to the N termini of AAV capsid proteins (ideally, VP2 [45, 83]), or more powerful, to insert short peptides (up to 14 amino acids [21], typically 7) into exposed regions of the assembled virion. This strategy is referred to as viral display, in analogy to phage display, and has already been used extensively to retarget AAV-2 virions to a multitude of refractory or hard-to-infect cell types, such as vascular endothelial, smooth muscle, and pancreatic islet cells (43, 55, 77, 81, 82) and various tumor lines (22, 58, 65, 66). It has particularly benefited from comprehensive mutational analyses by various groups (e.g., references 21, 33, 56, and 83) that have resulted in the identification of prominent locations within the AAV-2 capsid tolerating peptide insertion. Most notable is the heparin binding domain (HBD), consisting of a total of four arginine (R) residues and one lysine residue, with R585 and R588 representing the most crucial components (37, 56). Numerous groups have now consistently shown that the insertion of 7-mer peptides into this region not only is frequently well tolerated and efficiently mediates virus retargeting, but also provides the extra benefit that the endogenous AAV-2 tropism can be abolished, thus enhancing target specificity (e.g., reference 21).In addition to identifying sites for vector engineering, some of the mutational AAV studies directly yielded novel capsid variants with potential benefits for clinical use. A remarkable case was a recent study by Lochrie et al. (42) in which a set of 127 AAV-2 variants with point or insertion mutations were generated and screened for multiple properties. Several capsids were isolated which differed from the wild-type AAV-2 capsid in having better in vitro transduction efficiencies (albeit being equally efficient in vivo) or, clinically most relevant, higher-level resistance to individual or pooled human antisera. Nonetheless, the limitations of the approach also became clear, most notably, the extreme effort required to generate and manually screen a large number of mutants, which in fact prevented the interesting analyses of all possible combinations of beneficial point mutations in further capsids.Indeed, the factors of time and labor are the main reasons why an increasing number of groups have recently begun to develop novel means for AAV vector evolution that no longer rely on the rational modification of the AAV-2 capsid. Instead, the new combinatorial methodologies allow for the far more efficient creation and selection of interesting candidates in a library-based high-throughput format. Thus far, two different strategies have been reported, both principally expanding on previously developed techniques. One is the use of viral display libraries, in which random 7-mer peptides are inserted into the AAV-2 HBD (at amino acid 587 or 588), yielding between 4 × 10⁶ and 1.7 × 10⁸ capsids potentially exposing new ligands on their surfaces (50, 58, 76). Subsequent iterative selection on diverse cell types refractory to the wild type, e.g., coronary artery endothelial cells, cardiomyoblasts, and carcinoma, leukemia, and megakaryocytic cell lines, led to enrichment with peptide mutants with increased target specificities and efficacies (48, 50, 58, 76). The second library type, independently described by two groups in 2006, relies on error-prone PCR amplification of the AAV-2 capsid gene (46, 59). Similar to the methods in earlier mutational studies, this approach resulted in the identification of AAV-2 point mutations (usually up to two per capsid) which yielded mutants that differed from the wild type in having mildly enhanced efficacies in vitro and/or improved transduction efficiencies in the presence of neutralizing anti-AAV-2 antibodies either generated in rabbits or preexisting in individual human sera.Here, for the first time, we introduce the technology of DNA family shuffling into the realm of AAV vector evolution. The basic concept of this technology is the in vitro recombination of related parental genes with >50% homology, which are first fragmented and then reassembled based on partial homology, resulting in libraries of chimeric genes. Iterative amplification under pressure can then yield hybrids not only combining parental assets, but also ideally exhibiting novel and synergistic properties (70, 71). DNA family shuffling has been used extensively in recent years to evolve and improve all types of proteins, from markers and enzymes to vaccines (e.g., references 10, 13-15, and 39). Importantly, a set of reports also suggested its power to enhance viral gene therapy vectors by creating retro- or lentiviruses with improved stability or efficacy compared to that of the parental wild types (57, 61, 69). Here, we describe the novel use of DNA family shuffling for the highly efficient molecular interbreeding of eight multispecies AAVs to create chimeric capsids and, moreover, document its compatibility and synergism with existing AAV vector evolution technology.     

Replacement of the Replication Factors of Porcine Circovirus (PCV) Type 2 with Those of PCV Type 1 Greatly Enhances Viral Replication In Vitro

Porcine circovirus type 1 (PCV1), originally isolated as a contaminant of PK-15 cells, is nonpathogenic, whereas porcine circovirus type 2 (PCV2) causes an economically important disease in pigs. To determine the factors affecting virus replication, we constructed chimeric viruses by swapping open reading frame 1 (ORF1) (rep) or the origin of replication (Ori) between PCV1 and PCV2 and compared the replication efficiencies of the chimeric viruses in PK-15 cells. The results showed that the replication factors of PCV1 and PCV2 are fully exchangeable and, most importantly, that both the Ori and rep of PCV1 enhance the virus replication efficiencies of the chimeric viruses with the PCV2 backbone.Porcine circovirus (PCV) is a single-stranded DNA virus in the family Circoviridae (34). Type 1 PCV (PCV1) was discovered in 1974 as a contaminant of porcine kidney cell line PK-15 and is nonpathogenic in pigs (31-33). Type 2 PCV (PCV2) was discovered in piglets with postweaning multisystemic wasting syndrome (PMWS) in the mid-1990s and causes porcine circovirus-associated disease (PCVAD) (1, 9, 10, 25). PCV1 and PCV2 have similar genomic organizations, with two major ambisense open reading frames (ORFs) (16). ORF1 (rep) encodes two viral replication-associated proteins, Rep and Rep′, by differential splicing (4, 6, 21, 22). The Rep and Rep′ proteins bind to specific sequences within the origin of replication (Ori) located in the intergenic region, and both are responsible for viral replication (5, 7, 8, 21, 23, 28, 29). ORF2 (cap) encodes the immunogenic capsid protein (Cap) (26). PCV1 and PCV2 share approximately 80%, 82%, and 62% nucleotide sequence identity in the Ori, rep, and cap, respectively (19).In vitro studies using a reporter gene-based assay system showed that the replication factors of PCV1 and PCV2 are functionally interchangeable (2-6, 22), although this finding has not yet been validated in a live infectious-virus system. We have previously shown that chimeras of PCV in which cap has been exchanged between PCV1 and PCV2 are infectious both in vitro and in vivo (15), and an inactivated vaccine based on the PCV1-PCV2 cap (PCV1-cap2) chimera is used in the vaccination program against PCVAD (13, 15, 18, 27).PCV1 replicates more efficiently than PCV2 in PK-15 cells (14, 15); thus, we hypothesized that the Ori or rep is directly responsible for the differences in replication efficiencies. The objectives of this study were to demonstrate that the Ori and rep are interchangeable between PCV1 and PCV2 in a live-virus system and to determine the effects of swapped heterologous replication factors on virus replication efficiency in vitro.     

A New Borrelia Species Defined by Multilocus Sequence Analysis of Housekeeping Genes

Analysis of Lyme borreliosis (LB) spirochetes, using a novel multilocus sequence analysis scheme, revealed that OspA serotype 4 strains (a rodent-associated ecotype) of Borrelia garinii were sufficiently genetically distinct from bird-associated B. garinii strains to deserve species status. We suggest that OspA serotype 4 strains be raised to species status and named Borrelia bavariensis sp. nov. The rooted phylogenetic trees provide novel insights into the evolutionary history of LB spirochetes.Multilocus sequence typing (MLST) and multilocus sequence analysis (MLSA) have been shown to be powerful and pragmatic molecular methods for typing large numbers of microbial strains for population genetics studies, delineation of species, and assignment of strains to defined bacterial species (4, 13, 27, 40, 44). To date, MLST/MLSA schemes have been applied only to a few vector-borne microbial populations (1, 6, 30, 37, 40, 41, 47).Lyme borreliosis (LB) spirochetes comprise a diverse group of zoonotic bacteria which are transmitted among vertebrate hosts by ixodid (hard) ticks. The most common agents of human LB are Borrelia burgdorferi (sensu stricto), Borrelia afzelii, Borrelia garinii, Borrelia lusitaniae, and Borrelia spielmanii (7, 8, 12, 35). To date, 15 species have been named within the group of LB spirochetes (6, 31, 32, 37, 38, 41). While several of these LB species have been delineated using whole DNA-DNA hybridization (3, 20, 33), most ecological or epidemiological studies have been using single loci (5, 9-11, 29, 34, 36, 38, 42, 51, 53). Although some of these loci have been convenient for species assignment of strains or to address particular epidemiological questions, they may be unsuitable to resolve evolutionary relationships among LB species, because it is not possible to define any outgroup. For example, both the 5S-23S intergenic spacer (5S-23S IGS) and the gene encoding the outer surface protein A (ospA) are present only in LB spirochete genomes (36, 43). The advantage of using appropriate housekeeping genes of LB group spirochetes is that phylogenetic trees can be rooted with sequences of relapsing fever spirochetes. This renders the data amenable to detailed evolutionary studies of LB spirochetes.LB group spirochetes differ remarkably in their patterns and levels of host association, which are likely to affect their population structures (22, 24, 46, 48). Of the three main Eurasian Borrelia species, B. afzelii is adapted to rodents, whereas B. valaisiana and most strains of B. garinii are maintained by birds (12, 15, 16, 23, 26, 45). However, B. garinii OspA serotype 4 strains in Europe have been shown to be transmitted by rodents (17, 18) and, therefore, constitute a distinct ecotype within B. garinii. These strains have also been associated with high pathogenicity in humans, and their finer-scale geographical distribution seems highly focal (10, 34, 52, 53).In this study, we analyzed the intra- and interspecific phylogenetic relationships of B. burgdorferi, B. afzelii, B. garinii, B. valaisiana, B. lusitaniae, B. bissettii, and B. spielmanii by means of a novel MLSA scheme based on chromosomal housekeeping genes (30, 48).     

Alignment of the c-Type Cytochrome OmcS along Pili of Geobacter sulfurreducens

Immunogold localization revealed that OmcS, a cytochrome that is required for Fe(III) oxide reduction by Geobacter sulfurreducens, was localized along the pili. The apparent spacing between OmcS molecules suggests that OmcS facilitates electron transfer from pili to Fe(III) oxides rather than promoting electron conduction along the length of the pili.There are multiple competing/complementary models for extracellular electron transfer in Fe(III)- and electrode-reducing microorganisms (8, 18, 20, 44). Which mechanisms prevail in different microorganisms or environmental conditions may greatly influence which microorganisms compete most successfully in sedimentary environments or on the surfaces of electrodes and can impact practical decisions on the best strategies to promote Fe(III) reduction for bioremediation applications (18, 19) or to enhance the power output of microbial fuel cells (18, 21).The three most commonly considered mechanisms for electron transfer to extracellular electron acceptors are (i) direct contact between redox-active proteins on the outer surfaces of the cells and the electron acceptor, (ii) electron transfer via soluble electron shuttling molecules, and (iii) the conduction of electrons along pili or other filamentous structures. Evidence for the first mechanism includes the necessity for direct cell-Fe(III) oxide contact in Geobacter species (34) and the finding that intensively studied Fe(III)- and electrode-reducing microorganisms, such as Geobacter sulfurreducens and Shewanella oneidensis MR-1, display redox-active proteins on their outer cell surfaces that could have access to extracellular electron acceptors (1, 2, 12, 15, 27, 28, 31-33). Deletion of the genes for these proteins often inhibits Fe(III) reduction (1, 4, 7, 15, 17, 28, 40) and electron transfer to electrodes (5, 7, 11, 33). In some instances, these proteins have been purified and shown to have the capacity to reduce Fe(III) and other potential electron acceptors in vitro (10, 13, 29, 38, 42, 43, 48, 49).Evidence for the second mechanism includes the ability of some microorganisms to reduce Fe(III) that they cannot directly contact, which can be associated with the accumulation of soluble substances that can promote electron shuttling (17, 22, 26, 35, 36, 47). In microbial fuel cell studies, an abundance of planktonic cells and/or the loss of current-producing capacity when the medium is replaced is consistent with the presence of an electron shuttle (3, 14, 26). Furthermore, a soluble electron shuttle is the most likely explanation for the electrochemical signatures of some microorganisms growing on an electrode surface (26, 46).Evidence for the third mechanism is more circumstantial (19). Filaments that have conductive properties have been identified in Shewanella (7) and Geobacter (41) species. To date, conductance has been measured only across the diameter of the filaments, not along the length. The evidence that the conductive filaments were involved in extracellular electron transfer in Shewanella was the finding that deletion of the genes for the c-type cytochromes OmcA and MtrC, which are necessary for extracellular electron transfer, resulted in nonconductive filaments, suggesting that the cytochromes were associated with the filaments (7). However, subsequent studies specifically designed to localize these cytochromes revealed that, although the cytochromes were extracellular, they were attached to the cells or in the exopolymeric matrix and not aligned along the pili (24, 25, 30, 40, 43). Subsequent reviews of electron transfer to Fe(III) in Shewanella oneidensis (44, 45) appear to have dropped the nanowire concept and focused on the first and second mechanisms.Geobacter sulfurreducens has a number of c-type cytochromes (15, 28) and multicopper proteins (12, 27) that have been demonstrated or proposed to be on the outer cell surface and are essential for extracellular electron transfer. Immunolocalization and proteolysis studies demonstrated that the cytochrome OmcB, which is essential for optimal Fe(III) reduction (15) and highly expressed during growth on electrodes (33), is embedded in the outer membrane (39), whereas the multicopper protein OmpB, which is also required for Fe(III) oxide reduction (27), is exposed on the outer cell surface (39).OmcS is one of the most abundant cytochromes that can readily be sheared from the outer surfaces of G. sulfurreducens cells (28). It is essential for the reduction of Fe(III) oxide (28) and for electron transfer to electrodes under some conditions (11). Therefore, the localization of this important protein was further investigated.     

p85 Associates with Unphosphorylated PTEN and the PTEN-Associated Complex

The lipid phosphatase PTEN functions as a tumor suppressor by dephosphorylating the D3 position of phosphoinositide-3,4,5-trisphosphate, thereby negatively regulating the phosphoinositide 3-kinase (PI3K)/AKT signaling pathway. In mammalian cells, PTEN exists either as a monomer or as a part of a >600-kDa complex (the PTEN-associated complex [PAC]). Previous studies suggest that the antagonism of PI3K/AKT signaling by PTEN may be mediated by a nonphosphorylated form of the protein resident within the multiprotein complex. Here we show that PTEN associates with p85, the regulatory subunit of PI3K. Using newly generated antibodies, we demonstrate that this PTEN-p85 association involves the unphosphorylated form of PTEN engaged within the PAC and also includes the p110β isoform of PI3K. The PTEN-p85 association is enhanced by trastuzumab treatment and linked to a decline in AKT phosphorylation in some ERBB2-amplified breast cancer cell lines. Together, these results suggest that integration of p85 into the PAC may provide a novel means of downregulating the PI3K/AKT pathway.The phosphoinositide 3-kinase (PI3K)/AKT signaling pathway regulates glucose/nutrient homeostasis and cell survival and plays a central role in both normal metabolism and cancer. The PTEN tumor suppressor gene (29, 30, 54) negatively regulates the PI3K/AKT pathway by dephosphorylating the D3 hydroxyl subunit of phosphoinositide-3,4,5-trisphosphate, a key membrane phosphatidylinositol generated by PI3K (34). PTEN undergoes genetic or epigenetic inactivation in many malignancies, including glioblastoma, melanoma, and endometrial, prostate, and breast cancers, among others (6, 13, 22, 23, 47, 49-51, 55, 68). Similarly, germ line mutations of PTEN are associated with the development of hamartomatous neoplasias such as Cowden disease and Bannayan-Zonana syndrome (17, 21, 41).The tumor suppressor function of PTEN undergoes dynamic regulation involving both C-terminal phosphorylation and protein-protein interactions. Phosphorylation of serine and threonine residues at the PTEN C-terminal tail, mediated by kinases such as CK2 and glycogen synthase kinase 3β, alters its conformational structure and association with PDZ domain-containing proteins and attenuates PTEN enzymatic activity (1, 11, 20, 32, 45, 61-63, 66, 67, 71). Conversely, PTEN function is promoted in large part through its stabilization in unphosphorylated form by incorporation into a high-molecular-weight protein complex (the PTEN-associated complex [PAC]) (66). We first demonstrated the existence of the PAC through gel filtration studies of rat liver extracts, which identified PTEN within a high-molecular-mass peak (>600 kDa), as well as a low-molecular-mass peak (40 to 100 kDa) in which PTEN is monomeric and phosphorylated (66). Subsequently, several PDZ domain-containing proteins were shown to interact with PTEN, including MAGI-1b, MAGI-2, MAGI-3, ghDLG, hMAST205, MSP58/MCRS1, NHERF1, and NHERF2, which mediate indirect binding with platelet-derived growth factor (PDGF) receptor β (25, 36, 42, 57, 66). More recently, LKB1, a serine/threonine kinase tumor suppressor (7), was also found to interact with and phosphorylate PTEN in vitro (36). In aggregate, these data suggest that PTEN functional output is controlled by a complex interplay of protein interactions and regulation of C-terminal phosphorylation.Beyond these interactions, there is also evidence to support additional regulatory mechanisms by which the tumor suppressor function of PTEN is mediated. The herpesvirus-associated ubiquitin-specific protease was shown to interact directly with PTEN and promote its nuclear entry (53). Both ubiquitination and relocalization into the nucleus constitute important PTEN regulatory mechanisms (53, 64). In many tumors, PTEN nuclear exclusion has been associated with poor cancer prognosis and more aggressive cancer development (15, 44, 56). Moreover, successful treatment of acute promyelocytic leukemia was shown to be associated with an increase in monoubiquitinylation and relocation of PTEN into the nucleus (53).Like PTEN, the p85 regulatory subunit of PI3K serves as a prominent modulator of PI3K/AKT signaling. p85, which exists in three isoforms (α, β, and γ), targets the catalytic (110-kDa) PI3K subunit to the membrane, which brings it into proximity with membrane-associated phosphatidylinositol lipids. In the steady state, p85 forms a tight association with the catalytic PI3K subunit, usually p110α or p110β in nonhematopoietic cells, with p110δ predominating in leukocytes (19). Consistent with this notion, p85 and p110 exist in equimolar ratios in a wide variety of mammalian cell lines and tissues (19), although some studies have suggested a role for free p85 in cell signaling (33, 65).Several recent lines of evidence have begun to support a possible regulatory relationship between PTEN and p85 (reviewed in references 3 and 53). For example, liver-specific deletion of PIK3R1, which encodes the p85α regulatory subunit, reduces both the activation of PI3K and PTEN enzymatic activity in this context. As a result, p85α-deficient hepatic cells express elevated levels of phosphoinositide trisphosphate and exhibit prolonged AKT activation (60). In addition, both PTEN and p85 are regulated by small GTPase proteins such as RhoA, but PTEN coimmunoprecipitates with the RhoA effector Rock only in the presence of PI3K (18, 31, 37). Although only correlative in nature, these findings may suggest a possible role for PTEN in p85 regulation or vice versa, in addition to its known function as a direct antagonist of the PI3K/AKT pathway (3, 9, 52, 57, 60).In the present study, we demonstrate an endogenous association between p85 and PTEN. Using newly generated antibodies that selectively recognize the PTEN C-terminal tail in its unphosphorylated form, we demonstrate that this PTEN-p85 association preferentially involves the unphosphorylated form of PTEN. The specificity of this interaction was confirmed using multiple antibodies and through studies of both human cancer cells and murine embryonic fibroblasts (MEFs) deficient for specific p85 subunits. This association, which also engages p110β, is enhanced by trastuzumab treatment and correlates with diminished AKT phosphorylation. These results support a functional role for the PTEN-p85 association that may have important biological and therapeutic implications for PI3K/AKT pathway regulation.     

A New Genetic Vaccine Platform Based on an Adeno-Associated Virus Isolated from a Rhesus Macaque

We created a hybrid adeno-associated virus (AAV) from two related rhesus macaque isolates, called AAVrh32.33, and evaluated it as a vaccine carrier for human immunodeficiency virus type 1 (HIV-1) and type A influenza virus antigens. The goal was to overcome the limitations of vaccines based on other AAVs, which generate dysfunctional T-cell responses and are inhibited by antibodies found in human sera. Injection of a Gag-expressing AAVrh32.33 vector into mice resulted in a high-quality CD8⁺ T-cell response. The resulting Gag-specific T cells express multiple cytokines at high levels, including interleukin-2, with many having memory phenotypes; a subsequent boost with an adenovirus vector yielded a brisk expansion of Gag-specific T cells. A priming dose of AAVrh32.33 led to high levels of Gag antibodies, which exceed levels found after injection of adenovirus vectors. Importantly, passive transfer of pooled human immunoglobulin into mice does not interfere with the efficacy of AAVrh32.33 expressing nucleoproteins from influenza virus, as measured by protection to a lethal dose of influenza virus, which is consistent with the very low seroprevalence to this virus in humans. Studies of macaques with vectors expressing gp140 from HIV-1 (i.e., with AAVrh32.33 as the prime and simian adenovirus type 24 as the boost) demonstrated results similar to those for mice with high-level and high-quality CD8⁺ T-cell responses to gp140 and high-titered neutralizing antibodies to homologous HIV-1. The biology of this novel AAV hybrid suggests that it should be a preferred genetic vaccine carrier, capable of generating robust T- and B-cell responses.The initial interest in vectors based on adeno-associated viruses (AAV) was for applications in gene therapy. Most of the initial work was with vectors derived from AAV serotype 2 (AAV2), which is one of the six initial isolates. In the first in vivo studies, several groups showed stable expression of the transgene Escherichia coli β-galactosidase following intramuscular (i.m.) injection of AAV2-LacZ without immune responses to the transgene (23, 44). The apparent tolerance of the host to AAV-encoded antigens to a variety of transgene products has been demonstrated in mice and some large animals (1, 35, 39). Several mechanisms have been proposed to explain the lack of T-cell responses following in vivo gene transfer with AAV, including ignorance (inadequate presentation of antigen), anergy, and suppression (1, 5, 18, 37).As applications of AAV vectors for in vivo gene transfer expanded, it became clear that the apparent immune privilege of AAV transgene products was not absolute. A number of examples emerged in which the host mounted vibrant T-cell responses to AAV-encoded transgene products. Several key parameters appeared to influence immunogenicity of the transgene. For example, Sarukhan et al. suggest that the subcellular localization of the protein influences the magnitude of the ensuing T-cell response after AAV gene transfer (37). The dose and route of administration of the AAV vector also contribute significantly to B- and T-cell responses to the transgene (3, 13). Wang et al. showed that inflammation at the site of AAV administration promotes antigen-specific immune responses to the transgene (47). A consistent observation has been that B-cell responses to AAV-encoded transgenes are much more intense and more consistently generated than CD8⁺ T-cell responses (8, 46, 51). A number of investigators have begun to explore AAV vectors as genetic vaccines against a variety of infectious and noninfectious diseases, based on the notion that it can be developed to stimulate transgene immune responses (14, 22, 26, 28, 48-50).The discovery of an expanded family of AAV capsids from human and nonhuman primates has provided an opportunity to evaluate the effects of capsid structure on vector performance. Most of this work has focused on the use of novel AAV serotypes for achieving higher levels of transgene expression for applications in gene therapy (7, 12, 36). Xin et al. recently evaluated, in mice, vectors as vaccines for human immunodeficiency virus type 1 (HIV-1) based on the original AAV isolates AAV1 to AAV6 and two novel AAVs we recently discovered, AAV7 and AAV8 (48). They showed significant capsid-dependent effects on T- and B-cell responses to HIV-1 gp160. We recently confirmed these observations and more thoroughly evaluated the quality of the CD8⁺ T-cell responses (26). AAV vectors of multiple serotypes encoding HIV-1 Gag were injected i.m. into mice, which all showed some level of CD8⁺ T-cell responses based on tetramer staining and peptide-induced gamma interferon (IFN-γ) expression. However, the quality of AAV-induced, Gag-specific T cells was substantially lower than that obtained with adenoviral vectors, based on several criteria. A majority of the tetramer-positive (Tet⁺) T cells were nonresponsive to antigen, and those that did respond to antigen produced low levels of IFN-γ and no interleukin-2 (IL-2). Very few memory T cells were generated, and animals primed with AAV vectors were not responsive to a boost with an adenoviral vector. However, all AAV serotypes studied did generate very high levels of antibodies to the Gag transgene product.A final issue to consider in the use of AAV as a genetic vaccine for HIV-1 is the presence of neutralizing antibodies (NAbs) to the vector due to prior AAV infections. We recently conducted an extensive screening of human populations from several continents and found high prevalence and high titers of NAbs to AAV1 and AAV2 and moderate levels of NAbs to AAV7 and -8 (4). In vivo gene transfer experiments indicate that AAV NAbs will likely impinge on vector efficacy (9, 33, 38).This study describes the creation of a novel AAV from rhesus macaque isolates, called AAVrh32.33, and its characterization as a genetic vaccine for HIV-1. AAVrh32.33 has properties unlike those of any others we have studied. We showed that vectors based on this novel capsid elicit strong CD8⁺ T-cell responses to reporter transgene products that are dependent on CD4⁺ T-cell help and dependent on signaling through CD40L and CD28 (L. E. Mays and J. M. Wilson, submitted for publication). Important to the use of this vector in the clinic is a very low incidence of NAbs to it in human populations. This study describes the development of vectors based on AAVrh32.33 as genetic vaccines.     

A Bacillus anthracis S-Layer Homology Protein That Binds Heme and Mediates Heme Delivery to IsdC

The sequestration of iron by mammalian hosts represents a significant obstacle to the establishment of a bacterial infection. In response, pathogenic bacteria have evolved mechanisms to acquire iron from host heme. Bacillus anthracis, the causative agent of anthrax, utilizes secreted hemophores to scavenge heme from host hemoglobin, thereby facilitating iron acquisition from extracellular heme pools and delivery to iron-regulated surface determinant (Isd) proteins covalently attached to the cell wall. However, several Gram-positive pathogens, including B. anthracis, contain genes that encode near iron transporter (NEAT) proteins that are genomically distant from the genetically linked Isd locus. NEAT domains are protein modules that partake in several functions related to heme transport, including binding heme and hemoglobin. This finding raises interesting questions concerning the relative role of these NEAT proteins, relative to hemophores and the Isd system, in iron uptake. Here, we present evidence that a B. anthracis S-layer homology (SLH) protein harboring a NEAT domain binds and directionally transfers heme to the Isd system via the cell wall protein IsdC. This finding suggests that the Isd system can receive heme from multiple inputs and may reflect an adaptation of B. anthracis to changing iron reservoirs during an infection. Understanding the mechanism of heme uptake in pathogenic bacteria is important for the development of novel therapeutics to prevent and treat bacterial infections.Pathogenic bacteria need to acquire iron to survive in mammalian hosts (12). However, the host sequesters most iron in the porphyrin heme, and heme itself is often bound to proteins such as hemoglobin (14, 28, 85). Circulating hemoglobin can serve as a source of heme-iron for replicating bacteria in infected hosts, but the precise mechanisms of heme extraction, transport, and assimilation remain unclear (25, 46, 79, 86). An understanding of how bacterial pathogens import heme will lead to the development of new anti-infectives that inhibit heme uptake, thereby preventing or treating infections caused by these bacteria (47, 68).The mechanisms of transport of biological molecules into a bacterial cell are influenced by the compositional, structural, and topological makeup of the cell envelope. Gram-negative bacteria utilize specific proteins to transport heme through the outer membrane, periplasm, and inner membrane (83, 84). Instead of an outer membrane and periplasm, Gram-positive bacteria contain a thick cell wall (59, 60). Proteins covalently anchored to the cell wall provide a functional link between extracellular heme reservoirs and intracellular iron utilization pathways (46). In addition, several Gram-positive and Gram-negative bacterial genera also contain an outermost structure termed the S (surface)-layer (75). The S-layer is a crystalline array of protein that surrounds the bacterial cell and may serve a multitude of functions, including maintenance of cell architecture and protection from host immune components (6, 7, 18, 19, 56). In bacterial pathogens that manifest an S-layer, the “force field” function of this structure raises questions concerning how small molecules such as heme can be successfully passed from the extracellular milieu to cell wall proteins for delivery into the cell cytoplasm.Bacillus anthracis is a Gram-positive, spore-forming bacterium that is the etiological agent of anthrax disease (30, 33). The life cycle of B. anthracis begins after a phagocytosed spore germinates into a vegetative cell inside a mammalian host (2, 40, 69, 78). Virulence determinants produced by the vegetative cells facilitate bacterial growth, dissemination to major organ systems, and eventually host death (76-78). The release of aerosolized spores into areas with large concentrations of people is a serious public health concern (30).Heme acquisition in B. anthracis is mediated by the action of IsdX1 and IsdX2, two extracellular hemophores that extract heme from host hemoglobin and deliver the iron-porphyrin to cell wall-localized IsdC (21, 45). Both IsdX1 and IsdX2 harbor near iron transporter domains (NEATs), a conserved protein module found in Gram-positive bacteria that mediates heme uptake from hemoglobin and contributes to bacterial pathogenesis upon infection (3, 8, 21, 31, 44, 46, 49, 50, 67, 81, 86). Hypothesizing that B. anthracis may contain additional mechanisms for heme transport, we provide evidence that B. anthracis S-layer protein K (BslK), an S-layer homology (SLH) and NEAT protein (32, 43), is surface localized and binds and transfers heme to IsdC in a rapid, contact-dependent manner. These results suggest that the Isd system is not a self-contained conduit for heme trafficking and imply that there is functional cross talk between differentially localized NEAT proteins to promote heme uptake during infection.     

Cytotoxic-T-Lymphocyte-Mediated Elimination of Target Cells Transduced with Engineered Adeno-Associated Virus Type 2 Vector In Vivo

A recent clinical trial in patients with hemophilia B has suggested that adeno-associated virus (AAV) capsid-specific cytotoxic T lymphocytes (CTLs) eliminated AAV-transduced hepatocytes and resulted in therapeutic failure. AAV capsids elicit a CTL response in animal models; however, these capsid-specific CTLs fail to kill AAV-transduced target cells in mice. To better model the human clinical trial data in mice, we introduced an immunodominant epitope derived from ovalbumin (OVA; SIINFEKL) into the AAV capsid and tested CTL-mediated killing of AAV2-transduced target tissues in vivo. Initially, in vitro experiments demonstrated both classical class I and cross-presentation of the OVA antigen, following endogenous expression or AAV2-OVA vector transduction, respectively. Furthermore, an OVA-specific CTL response was elicited after muscular or systemic injection of the AAV2-OVA vector. Finally, CTL reactivity was enhanced in mice with established SIINFEKL-specific immunity after AAV2-OVA/α1 anti-trypsin (AAT) administration. Most importantly, these OVA-specific CTLs decreased AAT expression in mice treated with AAV2-OVA/AAT vector that followed a time course mimicking uncoating kinetics of AAV2 transduction in OVA-immunized mice. These results demonstrate that AAV capsid-derived antigens elicit CD8⁺ CTL reactivity, and these CTLs eliminated AAV-transduced target cells in mice. Notably, this model system can be exploited to study the kinetics of capsid presentation from different serotypes of AAV and permit the design of novel strategies to block CTL-mediated killing of AAV-transduced cells.Adeno-associated virus (AAV) is a single-stranded DNA parvovirus. Its replication relies on coinfection of a helper virus such as adenovirus or herpesvirus. In the absence of a helper virus, AAV establishes latency to integrate into the AAVS1 site of host chromosome 19 (11). The genome of AAV is ∼4.7 kb and contains two open reading frames encoding replication proteins and structural capsid proteins (21). The capsid proteins (VP) are composed of VP1, VP2, and VP3. The VP3 protein is the major structural component and constitutes nearly 80% of the virion shell with an overall ratio of 1:1:8 for VP1, VP2, VP3, respectively. While VP2 is thought to be nonessential for AAV transduction (30), the VP1 subunit contains a phospholipase A2 domain required for infectivity (9). Recombinant AAV (rAAV) vectors require only the 145-bp terminal repeats of the AAV genome in cis and all other viral factors supplied in trans for production (18). rAAV vectors have rapidly gained popularity in gene therapy applications and have proven effective in preclinical studies/clinical trials for a number of diseases (20, 31, 33).AAV vectors mount a potent humoral immune response against capsid in animals and human. However, AAV vectors only contain the therapeutic gene flanked by two 145-bp AAV terminal repeats devoid of any AAV genes(23). In addition, AAV initiates long-term stable therapeutic gene expression in animal models (3-5, 17, 31). Based on these observations AAV has been thought to be relatively nonimmunogenic regarding the induction of cytotoxic T lymphocytes (CTLs) specific for capsid proteins. In spite of all of these observations, the recent clinical trial for hemophilia B (F9) gene therapy has otherwise suggested that AAV2 capsid initiates cell-mediated immunity that eliminates the AAV2 encoding F9 (AAV2/F9) vector transduced liver cells (15). Against this backdrop, numerous attempts to replicate aforementioned observations in animal models have been made. Preliminary results from these studies support direct presentation and cross-presentation of the AAV2 capsid in animal models (6, 12, 13, 22, 29). However, capsid-specific CTLs did not eliminate AAV2-transduced target cells in mice (12, 13, 29), inconsistent with observations made in a clinical trial for hemophilia B with AAV2/F9 gene therapy. A potential explanation for this discrepancy is the weak immunogenicity of the AAV2 capsid in mice. Accordingly, we hypothesized that incorporation of a peptide epitope into the AAV2 capsid would increase immunogenicity of the rAAV and therefore could be exploited to mimic events ongoing in humans and study approaches to block capsid-specific CTL reactivity in mice.We chose to introduce the MHC-H2K^b-restricted SIINFEKL peptide derived from ovalbumin (OVA) into AAV2 capsid. Integration of the OVA epitope into AAV capsids elicited a specific CTL response. Most importantly, after administration of genetically engineered AAV2 vectors into OVA peptide-immunized mice, OVA-specific CTL reactivity was further enhanced, thereby limiting transgene expression in vivo. The modified vector described herein is a potentially valuable tool for future studies focused on developing strategies to evade capsid-specific CTL-mediated elimination of AAV-transduced target cells in animal models.     

Cultivation of Fastidious Bacteria by Viability Staining and Micromanipulation in a Soil Substrate Membrane System

Soil substrate membrane systems allow for microcultivation of fastidious soil bacteria as mixed microbial communities. We isolated established microcolonies from these membranes by using fluorescence viability staining and micromanipulation. This approach facilitated the recovery of diverse, novel isolates, including the recalcitrant bacterium Leifsonia xyli, a plant pathogen that has never been isolated outside the host.The majority of bacterial species have never been recovered in the laboratory (1, 14, 19, 24). In the last decade, novel cultivation approaches have successfully been used to recover “unculturables” from a diverse range of divisions (23, 25, 29). Most strategies have targeted marine environments (4, 23, 25, 32), but soil offers the potential for the investigation of vast numbers of undescribed species (20, 29). Rapid advances have been made toward culturing soil bacteria by reformulating and diluting traditional media, extending incubation times, and using alternative gelling agents (8, 21, 29).The soil substrate membrane system (SSMS) is a diffusion chamber approach that uses extracts from the soil of interest as the growth substrate, thereby mimicking the environment under investigation (12). The SSMS enriches for slow-growing oligophiles, a proportion of which are subsequently capable of growing on complex media (23, 25, 27, 30, 32). However, the SSMS results in mixed microbial communities, with the consequent difficulty in isolation of individual microcolonies for further characterization (10).Micromanipulation has been widely used for the isolation of specific cell morphotypes for downstream applications in molecular diagnostics or proteomics (5, 15). This simple technology offers the opportunity to select established microcolonies of a specific morphotype from the SSMS when combined with fluorescence visualization (3, 11). Here, we have combined the SSMS, fluorescence viability staining, and advanced micromanipulation for targeted isolation of viable, microcolony-forming soil bacteria.