Excerpted from CEGS Grant Proposal Jun 2002

Aim 2: Mitra & Church   (overall PI & Director: Wu)


C/D2.   In situ measurements of polymer modifications.


            We propose to extend in situ RNA/DNA analyses to measure changes at the level of DNA (mutation (2a), recombination(2b), and methylation (2c)), RNA (alternative ends (2d) and splicing (2e)), and protein epitope combinations (2f).  For each we will use polony concepts and aim for subcellular and single-residue (modified nucleotide or amino-acid) resolution.  Another unifying theme is the assessment of for "cis-ness".  For example, which alleles are in cis in DNA haplotypes, which exons are in cis in a spliced RNA molecule, which protein modifications are in cis in a protein molecule? Here the use of single molecules is crucial for interpretation, as a population average could be very misleading.  Analogously which of the above molecular features occur in the same cell (in contrast to a cell population average)?


The key problem in genomics is that sequence comparisons (and SNPs) alone do not predict functioning of physiological systems accurately. Functional genomics should do better, but is weak and costly due to the enormous multiplicity of cell-types, gene products, and interactions and cross-reactions among network components.  This is a very general problem since nearly all organisms form complex cellular patterns from microbial communities to developing plants and animals. How can we cost-effectively monitor the comprehensive set of cell types? Functional genomics of humans is further complicated by difficulty in accessing key cell types (e.g. brain tissue).  Single nucleotide polymorphisms (SNPs) do not necessarily predict diseases with acceptable precision for individual patients even if the statistics are convincing in populations.   Instead of  "genetic linkage/association" one would favor "functional association".


A desirable long-term scenario for clinical diagnostics illustrates key research goals: Obtain cells from a patient (Aim 3).  Use automated microfluidic differentiation system (mDS, Aim 4) to create and display a variety of cell-types in a compact format. Stain with multiple oligonucleotides and/or antibodies, multiplexing probings and fluorophores to obtain allele-specific and cell-type-specific phenotypes.  Use this data to computationally prioritize significant deviations of the patient relative to normal variations in the population. As can be seen from Aims 3 & 5, common dosage effects have significant phenotypic consequences. Linear 1.5 fold increases or 2-fold decreases must be discriminated from feedback pushing closer to 1-fold, or amplification beyond 1.5 or 2.  Non-causative SNPs can be used to quantitate allele-specific effects (relevant to Aim 1).  This allows for detection of small expression effects in heterozygotes (a 10-fold allele-specific decrease corresponds to only a 1.8-fold decrease in total RNA for that gene).


Preliminary results  Single-molecule cis-typing: We are developing new technologies which exploit intrinsic advantages of single nucleic acid molecules for DNA haplotyping and RNA exon-typing.  These use the polymerase-colony, a.k.a. "polony", hybridization and allele-specific fluorescent-base addition described in Aim 2d.  This can be generalized to include multi-base extension (a.k.a. "in situ sequencing").  An example of such sequencing is given in the figure below.   Similarly, multiple rounds of hybridization have been used to determine the exon composition of single RNA molecules  (and/or single cells) in Aim 2e, below.


We also have developed a variety of image/data processing and network modeling algorithms (see Aim 5 preliminary results).  The problem of computational image alignment and cost will be greatly reduced by increasing the multiplicity of probes per embryo mount. Image processing demands are reduced in various ways e.g. by fluorescent color ratio assays and internal controls for alignment.


2a. DNA haplotyping will be developed to determine which combinations of DNA polymorphisms in cis are associated with inherited allele-specific expression seen in Aim 1.  For example, in collaboration with Brad Hyman (MGH Neurology) we are assessing if the reason for ApoE3/E4 heterozygous variation in Alzheimer's phenotype is related to level of expression of the two protein forms.  Since there is significant tissue specificity, a surrogate tissue approach is unlikely to work and even tissue-specific and allele specific protein abundance correlations with disease could be the result of a protein difference rather than caused by a transcriptional difference. Hence the need to haplotype over the distance between the protein change and the promoter/enhancer polymorphisms.


2b. Recombination and combinatorial tags: Three types of recombination are of potential interest.  One is VDJ recombination which determines a large fraction of the epigenetic state of B and T cells as well as the specificity of the antibodies in Aim 2f.  A second is class switch recombination (CSR) which could act as a lineage reporter (in addition to methylation below).  The third system is homologous recombination (meiotic, mitotic or transfection) as mentioned in Aim 2a, 2f and 5.  By way of preliminary results, our lab invented the use of libraries of short quasi-random recombinants as unique tags [1].  These recombinants are important as a means for uniquely tagging progenitor cells for developmental lineage studies  We were involved in the first such lineage study which was developed for retinal and cerebral cortex [2].  Another example of recombinant libraries with quasi-random short tags would be the antibody libraries in Aim 2f.  Here the sequence of the randomer not only is a tag but also a web-shareable definition of the “reagent” analogous to the SequenceTagged Sites (STS) and Expressed Sequence Tags found useful in the Genome Project (HGP).


The switch of the immunoglobulin isotype from IgM to IgG, IgE or IgA is mediated by class switch recombination (CSR).  Activation Induced Cytidine Deaminase (AID) is indispensable in both CSR and somatic hypermutation.  Ectopic expression of AID induces CSR in an artificial switch construct in fibroblasts at a level comparable to that in stimulated B cells. The features of junction points were similar to those observed in physiological CSR: no consensus sequences around the break points, no long homology between two S sequences at the junctions, and frequent mutations in the proximity of the junctions and dependent on transcription of the target S region, as in endogenous CSR in B cells [3].

Thus the CSR-based recombinations seem to result in unique sequences.  We propose that combining expressed switch regions with a transgenic AID gene under control of a chemically-inducible (e.g. anhydrotetracycline) promoter would allow us to pulse the tagging of all cells in a developing mouse. Using an adjacent constant primer for polony sequencing of a sufficient number of variable bases after recombination would be sufficient to constitute a likely unique tag (by virtue of it including both a recombination junction and mutations). 


2c. DNA methylation and lineage can be assessed by bisulfite deamination of cytidines and sequencing.  Methylation allows (partial) lineage reconstruction of the history of cell-divisions, migrations, and epigenetic changes for each cell as embodied in recent algorithms [4,5]. The primers required for bisulfite analyses have additional constraints beyond those that apply to genomics more generally [6,7]. These are embodied in a public primer resource [8].


We will use in situ polonies amplified on embryos or cells as in Aim 1 or mDS in Aim 4.  This will allow retention of the in situ geometry and lower costs as in aim 1c, 2b,d-f. Yet another advantage of caging cells and nucleic nucleic acids used in polony amplification is reduction of loss during extensive washing [9] allowing "amplified sequences of a single copy gene from as little as 50 pg of bisulphite treated chromosomal DNA (~10 individual cells)." We expect to do even better because our reaction volume is a million times smaller, our PCR products ten times shorter, and we might go to milder treatment than 4 h at 50C pH5 due to the immobilization effect, e.g. recent results claim that "0C for 1 h resulted in 98% conversion" to dU [10]. 


2d.  In situ RNA sequencing and RNA 3' polyA end choice. We will combine the in situ methods used in Aim 1a,b with allele specific methods of 1c.

This will allow us to measure two DNA or RNA alleles close in space (as in transvection) as well as quantitative differences in expression of the two alleles as can occur in regulatory mutations and/or graded versions of epigenetic phenomena.  Aim 1c describes a way to do this on the diluted contents of single cells.  Aim 2d attempts to do the same in situ in the case where a polymorphism lies near a polyA end.  It also allows discovery and quantitation of  RNA 3' end choices.  This is somewhat analogous to Long-SAGE/MPSS [11,12] but with the huge advantage of being in situ capable.


The signal-to-noise for typical FISH is not currently compatible with Single Nucleotide Polymorphism Extension  (SNuPE) assays [13].  We have developed methods for in situ amplification of the number of template molecules (centered on one initial molecule) to enough copies to get robust allele determination [14].  We call this general class of amplification products polonies (short for polymerase colonies, see also Aim 1c).  Although we and others have used rolling circle amplification [15]  or in situ PCR [16-18], these do not cleanly lead to a single primed template as needed for single-base or multi-base extension assays. Although single-molecule extensions can be visualized [19], the added robustness from dual primer constraints in polonies is significant. Instead, our polony method begins with embedding the target RNA or DNA molecules in polyacrylamide such that one PCR primer is covalently part of the polyacrylamide gel via an acrydite moiety [14], while the other primer  is free to diffuse.  This allows removal of one strand for subsequent hybridization and/or polymerase extension.  In this method, polony size is reduced by increasing acrylamide concentration, template length, decreasing number of cycles.  The lower limit so far has been 10 micron diameter polonies.  An alternative would be to immobilize both primers.  Three small barriers prevented this originally, but probably no longer.  First, the small amounts of free primers must be as low as possible.  We now accomplish this by pre-electrophoresis. Second, the presence of the second strand would be a potent (intramolecular) competitor.  To solve the latter problem, we will test photocleavable (PC) linkage of one primer to the matrix by incorporating acrydite (Apogent) and PC-spacer phosphor-amidites (Ambergen) [20,21] on the 5' end of the oligo in the penultimate step of oligonucleotide syntheses. We have already demonstrated the use of PC linkages between dNTPs and their fluorophore (using similar PC nitrobenzyl groups from Ambergen).  The third problem is the fact that the rate of spread of the polony with double immobilization is limited by template length. We will systematically test lower acrylamide and/or bis-acrylamide concentrations to allow the effective radius of to be larger.


Automated multiplex  in situ sequencing and microscopy.   As an extension of the in situ methods (in 1a, 1b, 2b and 2c), we will push the limits of multiplexing , resolution and data integration.  This will mainly be done in the context of sequencing 3' ends of mRNAs in situ, but we will also see to what extent the same concepts can be applied to methyl-CGs (Aim 2c), RNA allele typing (Aim 1c), and exon typing (Aim 2e-f).  In addition to the limits of resolution due to the polonies technology itself, there are limits set by optics and signal-to-noise and scan speed. Very fast scanning is available for microarrays at 3 to 5 micron pixel size.  However, specialized resolutions in the sub-100 nm range have also been demonstrated [22,23]. Multiphoton  fluorescence excitation provides another dimension [24-26]. Some of this will require developing quantitative microscopy with full slide scan and autofocusing very analogous to the array scanners, but more challenging due to resolution required. Computational integration of adjacent sections is also important.  We anticipate this type of alignment to be easier to automate than our "one- or two- color" alignments because of the rich RNA profile available for each voxel in the images. "Optical projection tomography" has been used as a tool for 3D microscopy and gene expression studies [27].    


2e. RNA exon-typing is a key example of how the cell exploits combinatorial opportunities.  The fraction of genes showing alternative RNA splicing has risen from 5% to 50% since 1994 [28]. Many of these alternatives have huge effects on the physiology and pharmacology of the resulting proteins.  Aim 2a.will allow us to accurately determine alternative spliceform distributions without full-length cDNA clone sequencing. This will be done by amplifying potentially variable exons and then sequentially or simultaneously probing for each putative exon.  Exons can also be discovered and quantitated in situ by polony sequencing from a known exon into whatever is 3' or 5' of it (see Aim 2d).  Aim 2a will also feed into the Aim 5a goal of aligning in situ images exploiting rich profile data. 















Figure 2e.  Alternative splicing in Tau. By exon-specific hybridization to polonies amplified from single RNA molecules using exon 1 & 11 primers.  Mutations in the human tau gene cause frontotemporal dementia and Parkinsonism associated with chromosome 17 (FTDP-17). One of the major disease mechanisms in FTDP-17 is the increased inclusion of tau exon 10 during pre-mRNA splicing [29].


2f.  Protein multiplexing in situ.   Clearly "affinity proteomics" is a huge gap in the current application of genomics to biology.  Antibodies are crucial for in situ quantitation, antibody-arrays,  immunoselection/precipitation (IP), etc. The importance is that an enormous fraction of the cell-type specification and disease is specified by differences at the post-translational level.  How can we even begin to get the reagents since they are not currently considered as "designable" as reagents for RNAs?   We will develop methods for obtaining protein epitopes simultaneously with cognate oligonucleotide-tagged-antibodies.  We will strive to make them suitable for highly-multiplexed, single-molecule polony-tag readouts and discovery of novel epitope combinations. We will make versions which are "humanized" to be compatible with human immune systems, or easily reversible as needed in Aim 4d. Finally they will be shareable electronically as described in aim 2b.


Preliminary results.  We have three relevant activities.  First, we have been among the first to establish large-scale proteomics [30] as well as algorithm development and crosslinking experiments [31].  We believe that we have the first proteome of an organism "completed" in the sense of finding peptides for every reasonable open reading frame. Second, we work with the Harvard Institute of Proteomics and FlexGene Consortium [32] with a goal of producing at least one full-length expression cassette for each annotated human gene.  The cassettes are designed to be easily/automatically moved into dozens of vivo or in vitro expression system.  We have adapt lambda-red [33], yeast homologous recombination [34] and Creator in vitro recombination [35] for this purpose. In collaboration with Rick Boyce (MGH Neurology)  we are exploring a particularly safe and efficient VSV-Baculoviral expression system for these cassettes in embryonic stem cells and neuronal derivatives [36] .  


Third, we (Jingdong Tian, Xiaohua Huang in collaboration with Steve Blacklow and Tony Forster at HMS, and Gloria Culver at Iowa State Univ) have coupled DNA, RNA and protein synthesis in such a way that each process lacks normal termination; T7RNA polymerase stops without release near an artificial block (a slide surface) and ribosomes pileup behind. Making protein reading frames continue around the circular DNA template without stop codons long polyproteins are made and remain tethered to the slide via the RNA and DNA. We have done this with a EGFP polyprotein and shown that it folds and functions. Polyproteins may be useful for increasing the binding to other polymeric or surface structures. The ability to display full-length proteins is a potential advantage over the peptide display championed by Phylos [37,38]. On the other hand, the latter method using a puromycin deriv-ative to form a covalent link between the mRNA and encoded protein might allow us to do more strin-gent washes. Tony Forster has shown that modified amino-acids (e.g. biotinyl-Lys-tRNA) can be efficiently incorporated into these translation products in release-factor-free vitro translation systems. [39,40].


Aim 2f. Experimental.  Combinatorial chemistry, amplification, structure design and arrays are well-established and useful for nucleic acids but other classes of ligands are more useful for protein affinity measures.  Hybrid molecules capture the advantages of each.  The idea is that by attaching nucleic acid tags to reagents which bind proteins one can amplify, probe and/or sequence the tags in situ to quantitate a multiplicity of proteins in an array or tissue.  The attachment can be to affinity agents (e.g. antibodies, small ligands, lectins), which would bind to cell arrays or tissue sections or protein arrays and after washing be detected by nucleic acid methods (RCA, PCR, FISH).  Simple defined-length linkers can be used to turn moderate affinity agents into an assembly, which has higher avidity and/or specificity.  These assemblies can combinatorially explored (created and screened) by annealing libraries of oligonucleotide-conjugated ligands (e.g. active site agents, peptide loops) followed by polony analysis of the tag composition (and possibly order). 


One set of methods that we pursue accommodates natural cDNAs without individual manipulation of the stop-codon region of the cDNA.  In addition to the demonstrations above using (i) polyprotein and (ii) purified translation lacking release factors, additional approaches include (iii) inhibitors of translational elongation or release, (iv) dominant negative inhibitory mutant release factors, (v) antibodies to release factors for inhibition or immuno-absorption subtraction, (vi) suppressor tRNAs added to the vitro reaction to allow extension the last 20 amino acids out of the ribosomal exit tunnel, and (vii) finally, the nascent chain could be ensnared by a binding motif on one of the ribosomal proteins, e.g. a biotinylation motif peptide 14-mer BirA substrate[41]  fused to the ribosomal-protein L22 N-terminus, which is free and near the exit channel of the nascent peptide chain [42].


Peptide pair selection 1.  Such expression pairs would be useful for stimulating and measuring cell surface receptors (as in Aim 4),  as well as proteomics generally.  We will gene-fuse   Next a library of single-chain antibodies randomized in the appropriate variable residues regions will be made from a prototype vector using partially degenerate oligonucleotides.  "Camelized human single domain antibodies" [43].  5'-GCCCCAGATATC AAA (MNN)9GCA(MNN)10  TGC TGC ACA GTA ATA-3' (19 randomized codons in bold) will be our initial target, as they are preferred as sterically small (14 kdal) antibody probes and extraordinarily compatible with enzyme active sites [44]. Human framework antibodies have the advantage of compatibility with human immune systems for potential downstream diagnostic and therapeutics applications. Efficient tumor targeting has been observed using single domain VHs with dissociation constants of 2 nM  to 65nM [45]. The camel Ab domains have also proven useful in transmigrating across human blood-brain barrier endothelium [46].


Libraries (non-camel scFv) made entirely in vitro without animal or bacterial cells with an initial complexity of about 2e9 have been screened using mutagenic PCR ribosomal display to obtain antibodies "with monomeric dissociation constants as low as 82 pM" [47].  N-terminal gene-fusions of the antibody with a monomeric avidin [48] as the bridge will allow the antibodies to bind and be released from the ribosome.  The ribosome display library will finally have a library of protein coding fragments.  These fragments could be random exons, designed codon peptides (see below), or membrane-bound polysomal cDNAs from embryonic stem cells and differentiated derivatives.   Various points within the Ig constant scaffold can be replaced with very specific protease sites, e.g. factor Xa site, IEGR, cleavable by taipan snake venom [49].  This allows release of the ribosome display from the selection-solid-phase and also helps produce the reversible visualization antibodies needed in Aim 4.


Figure 2f. (right) Only the ribosome-display vectors in which the released peptide binds to the antibody will bind to the Ni-column.  A positive round with a P-ser-tRNA (see modified peptides section below) is shown, followed by protease release and a counter-selection round with a ser-tRNA replacement and then a second counter-selection round with P-ser but replacing other aminoacids that to create a peptide allele, splice-form, or gene-family relative.  These three rounds could be done on a large library of peptide-antibody bicistrons  simultaneously.  To obtain high affinity antibodies, previously successful whole-gene mutagenesis [47] could be enhanced by focusing on the desired variable regions by use of specifically primed error-prone polymerases (we have been studying these in collaboration with Tom Ellenberger's lab) followed by accurate PCR or RCA for the whole construct [50]


Peptide pair selection 2.  The above selection requires that the first peptide of a pair in a bicistronic mRNA must bind to its mate.  A complementary concept, is that two peptides (or antibody) bind to each other or to a third surface and the proximity is "recorded" as a hybrid nucleic acid. There is some precedent for this is methods based on ligation of aptamers [51] or rolling circles [52].  The idea is that is two affinity tags need to be near enough to one another to prime polymerase or allow a ligation to occur.   This is the protein equivalent of Aim 2d (DNA proximity assay).   There are also conceptual similarities to protein FRET [53], however our proposed method should have higher multiplex capability and robustness.  This method has the advantage over single tags that it is resistant to non-specific binding because it requires two tags to get a signal. The affinity between the two nucleic acid tags is tuned so that they are weak enough stay apart in solution but strong enough to bind together when kept in close proximity in the presence of polymerase or ligase , e.g. 10-mers [51].  One way to accomplish this that ties in well with the overall thrust of Aim 2f is using polyribosomes (from rolling circles optionally) lacking release factors (as above) having free mRNA 3’ ends with tags which co-prime RNA-dependent RNA polymerase followed by RT-PCR.  Another is to set-up recombination between the two DNA circles.  This could result not only in a signal of the cis-ness but also an immediately useful construct.


Epitope- or allele- specific mRNAs for antibody selection:  We are collaborating under separate funding with a group at Univ. Houston (Linxiao Gao) and one at Boston Univ. (Rostem Irani) on micromirror oligonucleotide array synthesizers [54], with Agilent on ink jet arrays and Affymetrix on custom photomasks.  All are typically synthesized such that the 3' ends are blocked (being the point of attachment to the linkers from the glass surface).  Nevertheless chemistry can be chosen such that the 3' ends can be freed before use. We also achieved this easily using restriction enzymes [55].  E.g. we will start with pairs of touching 120 microns square regions of 75-mers on Agilent arrays (10^7 molecules) with 15 bp of 3' end overlap, which can create 11,500 ds-135mers with 17 bp at each end, sufficient for vaccinia in vitro recombination [56] into  rolling-circle (in vitro replication) vectors.  These vectors will contain nicking sites such as N.BbvCIB and Lox sites (for Cre recombinase) to convert from tandem linear repeats to nicked circles as well as in vitro transcription and translation elements. A second recombinase system will be included to allow removal of the peptide to generate a simple antibody (or antibody+RNA); alternatively pairs of peptides or antibodies can be recombined between constructs. This scale to millions of oligomers.  Error elimination proven for PCR [50] will be applied to reannealed ds-synthetic 135-mers. Quality assurance will also include duplicates and polony FISSEQ. 


Modified peptides: Alternative genetic codes [39] are more straightforward to program with pure in vitro translation systems making short designed peptides than for large genomes and more natural sets of translation factors because one has complete control over codon usage and tRNAs present.  The new tRNAs can be charged chemically [57] or by altered aa-tRNA-synthetases [58].  Examples of potentially useful tRNAs would be phospho-Ser/Thr/Tyr, methyl-Lys, methyl-Arg, and acetyl-Lys or enzymatically resistant versions of these. For example, in the histone "code" [59], 11 of the N-terminal 18 amino acids of H3  are modified in many combinations of the above aminoacids (see section B. c.).  For in situ analyses, peptides displaying systematic modification combinations or antibodies directed at them should be "programmable" using the above in vitro system.  In a challenging project and complementary source of selection, we will bind the antibody library  to the differentiated stem cell arrays from Aim 4 and release with the specific protease as above. We will try polyprotein antibody libraries, a no-termination-codon library, and/or the no-ribosome (puromycin) approaches.  Phosphothioate rNTPs will be used if RNAse resistance is desirable. The intention is to broaden the scope to include post-synthetic modifications such as glycosylation. 


Specificity.  Reagents which are specific to one combination of post-synthetic modifications, or one isoform or allele type are a holy grail with multiple uses in analysis and modulation of activity.  For each of the above antibody selection methods, selection and screens for antibody specificity is important. One option for would be to turn the selected library into a set of polonies and screen for those which bind only their own peptide as assayed by FISSEQ. A second option is to run the selection twice with different genetic codes (above).  A third is using antibody combinations (as described above).  A fourth option is only available in conjunction with the synthetic 75-mer programming is choosing peptides most likely to be unique in the first place.    It should be noted that when using non-standard genetic codes both the peptide and the antibody will use the altered code unless design or selection prevents that.


Proteomic cis-allele or cis-modification measures:  Measures in a heterozygote are more reliable if done in an allele-specifically than in the usual pooled measure because the signal is undiluted (for example, a 2-fold change in the internally controlled ratio is going to be better than the externally compared ratio of 1.5 fold in comparing the rare heterozygote with the common homozygote.  Doing this at the protein level is more accurate than at the DNA-SNP level since many currently unpredictable steps lie between the causative SNPs and the final protein levels. In the embodiment where one tag is for an active-site-directed agent and another tag is directed against a common protein surface polymorph-ism, one can assess allele specific differences in protein function. This is analogous to Aim 2b, which assesses allele-specific differences in RNA levels. The extension of HEP to protein function is a step forward in accuracy of assessment of the function of new SNPs by leveraging old SNPs into functional assays.



C/D4.  Voldman, Badarinarayana, & Church.


a)       Demonstrated multiple sequential affinity labeling in flow


            To automatically and viably assess the phenotype of the differentiating cells in our mDS, we will use affinity agents to sequentially probe for various cell-surface markers.  We have demonstrated a first step towards this goal by sequential labeling ES cells with two antibodies in a flow system, photobleaching in between the two (Figure 4‑4).  Furthermore, we used indirect immunofluorescence with the same fluorophore, demonstrating that our ability to multiplex in time obviates the need to multiplex colors and thus dramatically extending the number of targets that we can probe.  We will adapt this scheme to the mDS, allowing us to assay surface marker expression in living cells (Aim 4c).




References for Specific Aim 2

2-1           Church, G.M., and Kieffer-Higgins, S. 1988. Multiplex DNA sequencing. Science 240: 185-188.


2-2           Walsh, C., Ryder, L., Cepko, C., Church, G.M., and Tabin, C. 1992. The dispersion of neuronal clones across the cerebral cortex. Science 258: 317-320.


2-3           Okazaki, I-M, Kinoshita, K, Muramatsu, M, Yoshikawa, K and Honjo, T.  2002. The AID enzyme induces class switch recombination in fibroblasts. Nature 416: 340 - 345.


2-4           Ro, S and Rannala, B. 2001. Methylation patterns and mathematical models reveal dynamics of stem cell turnover in the human colon. Proc Natl Acad Sci USA 9819: 10519-10521.


2-5           Yatabe Y, Tavare S, and Shibata D. 2001. Investigating stem cells in human colon by using methylation patterns. Proc Natl Acad Sci USA 98:10839-44.


2-6           Selinger, D., Cheung, K., Mei, R., Johanson, E.M., Richmond, C., Blattner, F.R., Lockhart, D., and Church, G.M. 2000. RNA expression analysis using a 30-base pair resolution Escherichia coli genome array. Nature Biotechnology 18: 1262-7.


2-7           Wright, M and Church,G.M. 2002. An Open-source Oligonucleotide Microarray Probe Standard for Human and Mouse. Submitted to Nature Biotechnology.


2-8           Li, L.-C.  2002.  MethPrimer - Design Primers for Methylation PCRs  http://itsa.ucsf.edu/~urolab/methprimer


2-9           Olek A, Oswald J, and Walter J.  1996. A modified and improved method for bisulphite based cytosine methylation analysis.  Nucleic Acids Res. 24: 5064-6.


2-10         Grunau C, Clark SJ, and Rosenthal A. 2001. Bisulfite genomic sequencing: systematic investigation of critical experimental parameters. Nucleic Acids Res  29: E65-5.


2-11         Saha S, Sparks AB, Rago C, Akmaev V, Wang CJ, Vogelstein B, Kinzler KW, and Velculescu VE.  2002. Using the transcriptome to annotate the genome. Nat Biotechnol. 20:508-12.


2-12         Brenner S, et al.  2000. Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays. Nat Biotechnol.18: 630-4.


2-13         Singer-Sam J. 1994. Quantitation of specific transcripts by RT-PCR SNuPE assay. PCR Methods Appl. 3:S48-50.


2-14         Mitra, R. and Church, G.M. 1999. In situ localized amplification and contact replication of many individual DNA molecules.  Nucleic Acids Res. 27: 1-6.


2-15         Lizardi, P.M., Huang, X., Zhu, Z., Bray-Ward, P., Thomas, D.C., and Ward, D.C. 1998. Mutation detection and single-molecule counting using isothermal rolling-circle amplification. Nature Genet. 19: 225-232.


2-16         Harrer T, Schwinger E, and Mennicke K. 2001. A new technique for cyclic in situ amplification and a case report about amplification of a single copy gene sequence in human metaphase chromosomes through PCR-PRINS. Hum Mutat. 17: 131-40.


2-17         Alzahrani AJ, Vallely PJ, and McMahon RF. 2002. Development of a novel nested in situ PCR-ISH method for detection of hepatitis C virus RNA in liver tissue. J Virol Methods 99:53-61.


2-18         Sallstrom JF, Zehbe I, Alemi M, and Wilander E. 1993. Pitfalls of in situ polymerase chain reaction (PCR) using direct incorporation of labeled nucleotides. Anticancer Res. 13:1153.


2-19         Hu, X, Aston C, and Schwartz DC.  1999. Optical mapping of DNA polymerase I action and products. Biochem Biophys Res Commun. 254:466-73.


2-20         Olejnik J, Krzymanska-Olejnik E, and Rothschild KJ.  1998. Photocleavable aminotag phosphoramidites for 5'-termini DNA/RNA labeling. Nucleic Acids Res. 26:3572-6.


2-21         Olejnik, J., Krzymanska-Olejnik, E., and Rothschild, K. J. 1996. Photocleavable biotin phosphoramidite for 5'-end-labeling, affinity purification and phosphorylation of synthetic oligonucleotides. Nucleic Acids Res. 24: 361-6.


2-22         Egner A, Jakobs S, and Hell SW.  2002. Fast 100-nm resolution three-dimensional microscope reveals structural plasticity of mitochondria in live yeast. Proc Natl Acad Sci U S A 99:3370-5.


2-23         Lacoste, TD, Michalet, X, Pinaud, F, Chemla, DS, Alivisatos, AP, and S. Weiss.  2000. Ultrahigh-resolution multicolor colocalization  of single fluorescent probes. Proc Natl Acad Sci USA 97: 9461-9466.


2-24         Drummond D.R., Carter N., Cross R.A. 2002. Multiphoton versus confocal high resolution z-sectioning of enhanced green fluorescent microtubules: increased multiphoton photobleaching within the focal plane can be compensated using a Pockels cell and dual widefield detectors. J Microsc. 206:161-9.


2-25 Maiti, S., Shear, J.B., Williams, R.M., Zipfel, W., and Webb, W.W. 1997. Measuring Serotonin Distribution in Live Cells with Three-Photon Excitation.  Science 275: 530-532.


2-26         Xu C, Zipfel W, Shear J.B., Williams R.M., and Webb W.W. 1996. Multiphoton fluorescence excitation: new spectral windows for biological nonlinear microscopy. Proc Natl Acad Sci USA  93:10763-8.


2-27         Sharpe J., Ahlgren U., Perry P., Hill B., Ross A., Hecksher-Sorensen J., Baldock R., and Davidson D. 2002. Optical projection tomography as a tool for 3D microscopy and gene expression studies. Science  296: 541-5.


2-28         Modrek B., and Lee C. 2002. A genomic view of alternative splicing. Nat Genet 30:13-9.


2-29         Kalbfuss B., Mabon S.A., and Misteli T. 2001. Correction of alternative splicing of tau in frontotemporal dementia and parkinsonism linked to chromosome 17. J Biol Chem  276: 42986-93.


2-30         Link, A.J., Robison, K. and Church, G.M. 1997. Comparing the predicted and observed properties of proteins encoded in the genome of Escherichia coli. Electrophoresis 18 (8):1259-1313


2-31 Chen, T., Jaffe, J.D.. and Church, G.M. 2001. Algorithms for Identifying Protein Cross-links via Tandem Mass Spectrometry. Recomb. J Comput Biol. 8:571-583.


2-32         Brizuela L, Braun P, and LaBaer J. 2001. FLEXGene repository: from sequenced genomes to gene repositories for high-throughput functional biology and proteomics. Mol Biochem Parasitol.  118:155-65


2-33         Yu D., Ellis H.M., Lee E.C., Jenkins N.A., Copeland N.G., and Court D.L. 2000. An efficient recombination system for chromosome engineering in Escherichia coli. Proc Natl Acad Sci USA 97:5978-83.


2-34         Raymond C.K., Sims E.H., and Olson M.V. 2002. Linker-mediated recombinational subcloning of large DNA fragments using yeast. Genome Res. 12:190-7.


2-35         Lin, Y.  and Farmer, A. 2002.  BD Bioscience Application note. http://www.clontech.com/products/families/creator/index.shtml


2-36         Barsoum J., Brown R., McKee M., and Boyce F.M. 1997. Efficient transduction of mammalian cells by a recombinant baculovirus having the vesicular stomatitis virus G glycoprotein. Hum Gene Ther. 8: 2011-8.


2-37         Weng S., Gu K., Hammond P.W., Lohse P., Rise C., Wagner R.W., Wright M.C., and Kuimelis R.G.  2002. Generating addressable protein microarrays with PROfusion covalent mRNA-protein fusion technology. Proteomics 2:48-57.


2-38         Roberts R.W. and Szostak J.W.  1997. RNA-peptide fusions for the in vitro selection of peptides and proteins. Proc Natl Acad Sci USA 94:12297-302.


2-39         Forster A.C., Weissbach H., and Blacklow S.C. 2001. A simplified reconstitution of mRNA-directed peptide synthesis: activity of the epsilon enhancer and an unnatural amino acid. Anal Biochem. 297: 60-70.


2-40         Shimizu Y., Inoue A., Tomari Y., Suzuki T., Yokogawa T., Nishikawa K., and Ueda T. 2001. Cell-free translation reconstituted with purified components. Nat Biotechnol 19:751-5.


2-41         Beckett, D., Kovaleva, E., and Schatz, P.J. 1999. A minimal peptide substrate in biotin holoenzyme synthetase-catalyzed biotinylation. Protein Science 8: 921-929.


2-42         Yusupov M.M., Yusupova G.Z., Baucom A., Lieberman K., Earnest T.N., Cate J.H., and Noller H.F. 2001. Crystal structure of the ribosome at 5.5 A resolution. Science 292: 883-96.


2-43         Tanha J., Xu P., Chen .Z, Ni F., Kaplan H., Narang S.A., MacKenzie C.R.  2001. Optimal design features of camelized human single-domain antibody libraries. J Biol Chem. 276: 24774-80.


2-44         Conrath K.E., Lauwereys M., Galleni M., Matagne A., Frere J.M., Kinne J., Wyns L., and Muyldermans S.  2001. Beta-lactamase inhibitors derived from single-domain antibody fragments elicited in the camelidae. Antimicrob Agents Chemother. 45: 2807-12.


2-45         Cortez-Retamozo V., Lauwereys M., Hassanzadeh Gh. G., Gobert M., Conrath K., Muyldermans S., De Baetselier P., and Revets H. 2002. Efficient tumor targeting by single-domain antibody fragments of camels. Int J Cancer 98: 456-62.


2-46         Muruganandam A., Tanha J., Narang S., and Stanimirovic D. 2002. Selection of phage-displayed llama single-domain antibodies that transmigrate across human blood-brain barrier endothelium. FASEB J 16: 240-2.


2-47         Hanes J., Schaffitzel C., Knappik A., and Pluckthun A.  2000. Picomolar affinity antibodies from a fully synthetic naive library selected and evolved by ribosome display. Nat Biotechnol. 18: 1287-92.


2-48         Shin S.U., Wu D., Ramanathan R. Pardridge W.M., and Morrison S.L.  1997. Functional and pharmacokinetic properties of antibody-avidin fusion proteins. J Immunol 158: 4797-804.


2-49         Humm A., Fritsche E., Mann K., Gohl M., and Huber R. 1997. Recombinant expression and isolation of human L-arginine:glycine amidinotransferase and identification of its active-site cysteine residue. Biochem J. 322: 771-6.


2-50         Smith J., and Modrich P. 1997. Removal of polymerase-produced mutant sequences from PCR products. Proc Natl Acad Sci USA 94: 6847-50.


2-51         Fredriksson S., Gullberg M., Jarvius J., Olsson C., Pietras K., Gustafsdottir S.M., Ostman A., and Landegren U.  2002. Protein detection using proximity-dependent DNA ligation assays. Nat Biotechnol. 20: 473-7.





2-52         Schweitzer B., Wiltshire S., Lambert J., O'Malley S., Kukanskis K., Zhu Z., Kingsmore S.F., Lizardi P.M., and Ward D.C. 2000.Inaugural article: immunoassays with rolling circle DNA amplification: a versatile platform for ultrasensitive antigen detection. Proc Natl Acad Sci USA 97: 10113-9.


2-53         Truong K., and Ikura M. 2001. The use of FRET imaging microscopy to detect protein-protein interactions and protein conformational changes in vivo. Curr Opin Struct Biol 11: 573-8.


2-54         Singh-Gasson S., Green R.D., Yue Y., Nelson C., Blattner F., Sussman M.R., and Cerrina F. 1999.  Maskless fabrication of light-directed oligonucleotide microarrays using a digital micromirror array. Nat Biotechnol.17: 974-8.


2-55         Bulyk, M.L., Gentalen, E. Lockhart, D.J., and Church, G.M. 1999. Quantifying DNA-protein interactions by double-stranded DNA arrays. Nature Biotechnology 17: 573-7.


2-56         Willer D.O., Yao X.D., Mann M.J., and Evans D.H. 2000. In vitro concatemer formation catalyzed by vaccinia virus DNA polymerase. Virology. 278: 562-9.


2-57         Mendel D., Cornish V.W., and Schultz P.G. 1995. Site-directed mutagenesis with an expanded genetic code. Annu Rev Biophys Biomol Struct  24:435-62 


2-58         Wang L., Brock A., Herberich B., and Schultz P.G. 2001. Expanding the genetic code of Escherichia coli. Science 292: 498-500.


2-59         Jenuwein T., and Allis C.D. 2001. Translating the histone code. Science 293:1074-80.