Patent Application
20160138013
The genetic material in a single cell can be amplified by DNA polymerase into many clonal copies through whole genome amplification and characterized by shotgun sequencing. Single-cell genome sequencing has been successfully demonstrated on microbial and mammalian cells1-6, and applied to the characterization of microbial genomic diversity of the ocean7, somatic mutations in cancers8,9, and meiotic recombination and mutation in sperm3, 10.
Embodiments herein relate generally to whole-genome amplification. Some embodiments herein related generally to unbiased amplification of a genome.
According to some aspects, a method of producing a substantially unbiased amplification library of a genome of a single cell is provided. The method can comprise amplifying the genome of the single cell in a nanoliter-scale reaction environment configured for substantially unbiased amplification of the genome, and constructing a library comprising a plurality of amplicons of the substantially unbiased amplification of the genome. In some embodiments, amplifying the genome of the single cell comprises multiple strand displacement amplification (MDA) comprising contacting the reaction environment with (a) strand-displacement polymerase, and (b) a plurality of random multimers of DNA, thereby producing a substantially unbiased amplification of the genome of the single cell. In some embodiments, a ratio of amount of nucleic acid of the genome to volume of the nanoliter-scale reaction environment is at least about 0.03 Mega-basepairs per nanoliter. In some embodiments, a ratio of amount of nucleic acid of the genome to volume of the nanoliter-scale reaction environment is at least about 200 Mega-basepairs per nanoliter. In some embodiments, the nanoliter-scale reaction environment is configured for amplification of at least about 90% of the genome at greater than 1× coverage. In some embodiments, the nanoliter-scale reaction environment comprises a volume of no more than about 20 nL. In some embodiments, the nanoliter-scale reaction environment comprises a volume of no more than about 12 nL. In some embodiments, the method further comprises amplifying a plurality of genomes of single cells in a plurality of nanoliter-scale reaction environments on a single substrate, wherein at least 95% of the reaction environments do not comprise any genomes other than a genome of a single cell. In some embodiments, at least 99% of the reaction environments do not comprise any genomes other than a genome of a single cell. In some embodiments, the substrate is configured for a single pipetting action to distribute the genomes of single cells among the reaction environments. In some embodiments, the method further comprises selecting a desired number of reaction environments; and amplifying the plurality of genomes of single cells in only the desired number of reaction environments. In some embodiments, the method further comprises identifying a reaction environment in which a desired level of amplification has been achieved, wherein the library is constructed from the reaction environment in which a desired level of amplification has been achieved. In some embodiments, the method further comprises constructing a plurality of libraries from the plurality of reaction environments, in which the number of the plurality of libraries is the same or different as the number of the plurality of reaction environments. In some embodiments, amplifying the genome of the single cell in the nanoliter-scale reaction environment comprises amplification in the presence of an amplification-detection moiety. In some embodiments, the amplification-detection moiety comprises a cyanine dye. In some embodiments, the amplification-detection moiety comprises SYBR™ green dye. In some embodiments, signal from the amplification-detection moiety identifies a reaction environment in which a desired level of amplification has been achieved. In some embodiments, the reaction environment does not comprise any cells other than the single cell. In some embodiments, the reaction environment does not comprise any genomes other than the genome of the single cell. In some embodiments, the random multimers are selected from the group consisting of: pentamers, hexamers, heptamers, octamers, nonamers and decamers. In some embodiments, the random multimers are hexamers. In some embodiments, substantially all of the plurality of amplicons are unbranched. In some embodiments, the method further comprises removing at least some of the plurality of amplicons from the reaction environment prior to constructing the library. In some embodiments, removing at least some of the plurality of amplicons comprises micromanipulation. In some embodiments, the plurality of amplicons comprises no more than about 100 picograms to about 10 nanograms of DNA. In some embodiments, the library comprises a transposase-based library. In some embodiments, the library comprises a Tn5 transposase-based library. In some embodiments, the library comprises a random fragmentation and ligation library. In some embodiments, the single cell is one of a human cell or a microbial cell. In some embodiments, the single cell comprises a cell of a bacterium that is unculturable, or substantially unculturable. In some embodiments, the MDA comprises real time MDA. In some embodiments, the method is performed in parallel on two or more genomes of two or more single cells, thereby producing two or more unbiased amplification libraries in parallel. In some embodiments, the method further comprises at least one of: de novo assembly of unculturable bacteria in the human gut, de novo assembly of unculturable bacteria in heterogeneous environments such as sea water, copy number variation calling on single neurons, copy number variation calling on single cancerous cells or circulating tumor cells, or human haplotyping. In some embodiments, the strand-displacement polymerase comprises a high-fidelity polymerase. In some embodiments, the strand-displacement polymerase comprises phi29 polymerase.
According to some aspects, a method of producing a substantially unbiased amplification of a genome by multiple strand displacement amplification (MDA). The method can comprise providing the genome in a nanoliter-scale reaction environment, and contacting the nanoliter-scale reaction environment with (a) strand-displacement polymerase, and (b) a plurality of random multimers of DNA, thereby producing a substantially unbiased amplification of the genome. In some embodiments, the method further comprises constructing a library comprising a plurality of amplicons of the substantially unbiased amplification of the genome. In some embodiments, the nanoliter-scale reaction environment is configured for amplification of at least 90% of the genome at greater than 1× coverage. In some embodiments, a ratio of amount of nucleic acid of the genome to volume of the nanolioter-scale reaction environment is at least about 0.3 Mega-basepairs per nanoliter. In some embodiments, a ratio of amount of nucleic acid of the genome to volume of the reaction environment is at least about 200 Mega-basepairs per nanoliter. In some embodiments, the random multimers are selected from the group consisting of: pentamers, hexamers, heptamers, octamers, nonamers, and decamers. In some embodiments, the random multimers comprise hexamers. In some embodiments, substantially all of the plurality of amplicons are unbranched. In some embodiments, the nanoliter-scale reaction environment comprises a nanoliter-scale reaction environment that facilitates substantially unbiased amplification of the single cells. In some embodiments, the nanoliter-scale reaction environment comprises a volume of no more than about 20 nL. In some embodiments, the nanoliter-scale reaction environment comprises a volume of no more than about 12 nL. In some embodiments, there is at least a 99% probability that the reaction environment comprises no more than one genome. In some embodiments, the method further comprises at least one of: de novo assembly of a genome of an unculturable bacterium of the human gut, de novo assembly of an unculturable bacterium of a heterogeneous environment, copy number variation calling on a single neuron, copy number variation calling on a single cancerous cell or circulating tumor cell, or human haplotyping. In some embodiments, the strand-displacement polymerase comprises a high-fidelity polymerase. In some embodiments, the strand-displacement polymerase comprises phi29 polymerase.
According to some aspects, a substrate for substantially unbiased amplification a genome at least one single cell is provided. The substrate can comprise a plurality of loading areas, in which each loading area is configured to receive a liquid sample. Each loading area can comprise a plurality of nanoliter-scale reaction environments that facilitates substantially unbiased amplification of a single cell. In some embodiments, the plurality of nanoliter-scale reaction environments is configured for performing a desired number of amplification reactions in parallel, in which each amplification reaction is conducted in a different nanoliter-scale reaction environment. In some embodiments, the plurality of nanoliter-scale reaction environments is configured for performing a desired number of amplification reactions in parallel without further modification of the substrate. In some embodiments, the plurality of nanoliter-scale reaction environments are not in fluid communication with any microfluidic channels or nanofluidic channels. In some embodiments, each nanoliter-scale reaction environment has a volume of no more than about 12 nL. In some embodiments, each nanoliter-scale reaction environment has a volume of no more than about 20 nL. In some embodiments, each loading area is configured for loading a solution comprising diluted cells into the plurality of nanoliter-scale reaction environments via a single pipetting action. In some embodiments, each reaction environment comprises a plurality of random multimers and strand-displacement polymerase. In some embodiments, the plurality of multimers comprises hexamers. In some embodiments, the substrate comprises at least three loading areas. In some embodiments, each loading area comprises at least ten nanoliter-scale reaction environments. In some embodiments, each loading area comprises at least one hundred nanoliter-scale reaction environments. In some embodiments, the substrate further comprises a detector configured to detect an amplification-detection moiety in each of the reaction environments. In some embodiments, the substrate further comprises a nanopipettor configured to withdraw amplified nucleic acid from a single reaction environment. In some embodiments, the nanoliter-scale reaction environments are configured so that at least 99% of the reaction environments comprise a genome of no more than one cell following a loading of solution comprising single cells or fractions thereof in the loading area. In some embodiments, substantially each reaction environment comprises a genome of no more than one cell, and wherein substantially each reaction environment that comprises a genome further comprises a plurality of amplicons of the genome. In some embodiments, the plurality of amplicons comprises substantially unbiased coverage of the genome. In some embodiments, the plurality of amplicons comprises no more than about 100 picograms to about 10 nanograms of DNA. In some embodiments, the strand-displacement polymerase comprises a high-fidelity polymerase. In some embodiments, the strand-displacement polymerase comprises phi29 polymerase.
Amplification of sub-nanogram quantities of nucleic acids, for example the genome of a single cell, can be useful for a number of applications. According to some embodiments herein, methods and manufactures for substantially unbiased amplification of nucleic acids are provided. In some embodiments, a small quantity of nucleic acid, for example the genomic material of a single cell, is amplified in a nanoliter-scale volume. The nanoliter-scale volume can provide for amplification in a high concentration of reactants. The amplification can comprise multiple strand displacement amplification (MDA). In some embodiments, the amplification is performed in a single reaction space, such as a well, thus minimizing moving parts. In some embodiments, the amplification method can be readily scaled by simply increasing or decreasing a number of nanoliter-scale amplifications that are performed in parallel. In some embodiments, a sequencing library is prepared from the amplified nucleic acid. In some embodiments, the library comprises a random fragmentation and ligation library.
Genome sequencing of single cells can have a variety of applications including, but not limited to characterizing difficult-to-culture microorganisms and identifying somatic mutations in single cells from mammalian tissues. A major hurdle of this process can be bias in amplifying and making multiple copies of the genetic material from a single cell, a procedure known as polymerase cloning. Some embodiments herein provide a microwell displacement amplification system (MIDAS), a massively parallel polymerase cloning method in which single cells are randomly distributed into hundreds to thousands of microwells in nanoliter-scale volumes and simultaneously amplified for shotgun sequencing. In some embodiments, MIDAS dramatically reduces amplification bias by implementing polymerase cloning in nanoliter-scale reactions, allowing the de novo assembly of near-complete microbial genomes from single E. coli cells. In some embodiments, MIDAS allows detection of single-copy number changes in primary human adult neurons at 1-2 Mb resolution. MIDAS can facilitate the characterization of genomic diversity in many heterogeneous cell populations. It is further contemplated that as amplification reactions according to some embodiments herein are performed in a single reaction environment, these reactions can be performed with minimal moving parts (for example, only a pippettor to add or remove solution from a reaction environment). Accordingly, amplification reactions according to some embodiments herein can be performed with a high degree of reliability, while minimizing the need for additional components such as moving parts, and chasses and operating software for such moving parts. In some embodiments, amplification is performed in a single reaction environment. In some embodiments, the amplification is performed without the activity of fluidic channels or other fluidic system other than one or more pipettors for adding and/or removing solution from the reaction environment. In some embodiments, the amplification is performed in a reaction environment that is not in fluid communication with a network of fluidic channels, and is not configured for being in fluid communication with a network of fluidic channels.
Some embodiments allow for whole genome amplification of many single cells in parallel in an unbiased manner. Hundreds (or more) of cells can be amplified simultaneously in nanoliter volumes. Some embodiments include a low input sequencing library construction technique such that DNA directly from the whole genome amplification can be sequenced. The unbiased nature of amplification can allow for a myriad of downstream applications, including de novo assembly of unculturable bacteria and copy number variation calling of single mammalian cells.
According to some embodiments herein, methods of nucleic acid amplification are readily scalable. Depending on the desired number of amplification reactions to be performed, a number of nanoliter-scale reaction environments (for example wells) can be selected. Templates (e.g. single cells, or single cell genomes) can be diluted to a volume such that there is approximately no more than one template per reaction environment, and distributed among the desired number of reaction environments. In some embodiments, at least one substrate comprising a plurality of nanoliter-scale reaction environments is provided. If the desired number of reactions is less than the number of reaction environments on the substrate, only some of the reaction environments can be used.
If the desired number of reactions is greater than the number of reaction environments on the substrate, two or more substrates can be used, for example 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 30, 40, 50, 60, 70, 80, 90, or 100 substrates, including ranges between any two of the listed values. It is contemplated herein that scalability offers flexibility to an operator. Additionally, as amplification reactions according to some embodiments herein can be performed with minimal moving parts, the number of amplification reactions can be readily scaled without any substantial customization or redesign of the substrate architecture (such as operating software, mechanical components, fluidic systems, and the like). Accordingly, in some embodiments, a large number of amplification reactions can be performed in parallel. In some embodiments, at least 2 amplification reactions are performed in parallel, for example at least 2, 3, 4, 5, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000, 7500, 8000, 8500, 9000, 9500, or 10000 amplifications, including ranges between any two of the listed values.
Traditional whole genome amplification techniques for single cells can amplify genomes extremely biasedly. Small regions of the genome can be amplified greatly, whereas most of the genome can be amplified very little. Therefore, a large amount of sequencing effort can be required to resolve any of the genome. Downstream applications, such as de novo assembly or copy number variation calling, thus can be extremely difficult and inaccurate.
In some embodiments, whole genomes of single cells are amplified unbiasedly. In some embodiments, whole genomes of single cells are amplified substantially unbiasedly. As used herein “substantially unbiased” and pluralizations, conjugations, variations, and the like of this root term refers to amplification of a genome wherein, when the amplified genome is divided into at least 100 genomic bins that were previously determined such that each would contain a similar number of reads after mapping (see, e.g. 30), the log10 fold-amplification of at least 80% of the bins is within ±20% of the mean (i.e. for at least 80% percent of the genomic bins, the log10 of the fold amplification, is no more than 20% more, and no less than 20% less than the mean number of copies genome-wide). In some embodiments, the log10 fold-amplification of at least 80% of the bins is within ±20% of the mean, for example at least about 80%, 85%, 90%, 95%, 99%, or 99.9%. When whole genome amplification is substantially unbiased or unbiased, most of the genome can be amplified to a similar degree. Therefore, relatively little sequencing effort can be necessary for downstream analysis. De novo assembly can be accomplished and copy number variations can be called with a much greater accuracy.
As used herein, “nanoliter-scale” refers to a volume, for example in a reaction environment, of at least about one nanoliter and no more than about 50 nanoliters, more preferably about 5 nanoliters to about 30 nanoliters, more preferably about 10 nanoliters to about 25 nanoliters, for example about 12 nanoliters or about 20 nanoliters.
In some embodiments, cells are diluted and spread evenly across a loading area on a substrate, in which the loading area contains hundreds of nanoliter-scale reaction environments such that at least 99% of the reaction environments contain no more than 1 cell per well. In some embodiments, the substrate comprises a PDMS slide. After lysis and denaturing, the DNA can be amplified using multiple displacement amplification (MDA). The MDA reactants can be provided in buffer comprising polymerase, dNTP's, random oligonucleotides, and an amplification-detection moiety such as SYBR™ green dye. The MDA can be performed in a temperature controlled environment and in optical communication with a detector for amplification-detection moiety, such as a microscope. Without being limited by any theory, the small volume and consequent high concentration of template can allow for an unbiased amplification of the whole genome. Staining with an amplification-detection moiety, for example SYBR™ green, during amplification allows for positive amplifications to be observed due to an increase in detectable signal over time. Positive amplifications are then automatically or manually removed using a micromanipulator and deposited into tubes. Some embodiments include a low input sequencing library construction method capable of using sub nanogram inputs of DNA. The complex MDA amplicon can then be denatured and simple linear DNA created. The linear DNA can be used to construct a sequencing library. In some embodiments, transposons with Illumina sequencing adaptors (Nextera) then fragment the DNA while adding sequencing adapters. Accordingly, a sequencing library can be prepared. It is contemplated that nucleic acid amplified substantially unbiasedly in accordance with embodiments herein can be used for a number of downstream applications, including any of a number of genome sequencing techniques known to the skilled artisan.
A variety of techniques for amplifying nucleic acid are known to the skilled artisan. Exemplary techniques for amplifying nucleic acid include, but are not limited to: polymerase chain reaction (PCR), strand displacement amplification (SDA), for example multiple displacement amplification (MDA), loop-mediated isothermal amplification (LAMP), ligase chain reaction (LCR), immuno-amplification, and a variety of transcription-based amplification procedures, including transcription-mediated amplification (TMA), nucleic acid sequence based amplification (NASBA), self-sustained sequence replication (3SR), and rolling circle amplification. See, e.g., Mullis, “Process for Amplifying, Detecting, and/or Cloning Nucleic Acid Sequences,” U.S. Pat. No. 4,683,195; Walker, “Strand Displacement Amplification,” U.S. Pat. No. 5,455,166; Dean et al, “Multiple displacement amplification,” U.S. Pat. No. 6,977,148; Notomi et al., “Process for Synthesizing Nucleic Acid.” U.S. Pat. No. 6,410,278; Landegren et al. U.S. Pat. No. 4,988,617 “Method of detecting a nucleotide change in nucleic acids”; Birkenmeyer, “Amplification of Target Nucleic Acids Using Gap Filling Ligase Chain Reaction,” U.S. Pat. No. 5,427,930; Cashman, “Blocked-Polymerase Polynucleotide Immunoassay Method and Kit,” U.S. Pat. No. 5,849,478; Kacian et al., “Nucleic Acid Sequence Amplification Methods,” U.S. Pat. No. 5,399,491; Malek et al., “Enhanced Nucleic Acid Amplification Process,” U.S. Pat. No. 5,130,238; Lizardi et al., BioTechnology, 6:1197 (1988); Lizardi et al., U.S. Pat. No. 5,854,033 “Rolling circle replication reporter systems,” each of which is hereby incorporated by reference in its entirety. Preferably, MDA can be used in accordance with some embodiments herein. MDA can comprise annealing random oligonucleotide primers to a template nucleic acid and extending the oligonucleotide primers forward to the annealing site of the most immediate downstream oligonucleotide primer so as to form branched amplified nucleic acid. MDA can be performed at a constant temperature, and compared to conventional PCR can produce relatively large products with a relatively low error rate. A variety of MDA reagents can be used in accordance with embodiments herein. In some embodiments, MDA is performed with a strand-displacement polymerase In some embodiments, the strand displacement polymerase comprises a high-fidelity DNA polymerase. for example Φ29 DNA polymerase.
The fold amount of amplification that occurs according to some embodiments herein can depend on the amount of template, and the total mass of reactants. According to some embodiments herein, amplification is performed until saturation (e.g. until additional cycles of amplification are no longer in a logarithmic phase, so that the additional cycles produce few to no additional amplicons). Without being limited by any theory, it is contemplated that the total amount of amplification is proportional to the total mass of the reaction, and inversely proportional to the size of the template being amplification. Accordingly, by way of example, given the same reaction mass and amplification until saturation in accordance with some embodiments herein, a 1 Mb genome would be amplified approximately 10-fold more than a 10 Mb genome.
Without being limited by any theory, it is contemplated herein that a high concentration of amplification reactants and template in accordance with some embodiments herein can facilitate substantially unbiased amplification of all or substantially all of the template, for example genomic material. So as to provide a high concentration of reactants, including, but not limited to, template, the ratio of template to reaction volume can be relatively high in some embodiments herein. Accordingly, in some embodiments, the nanoliter-scale reaction environments are configured for a high ratio of genomic material to reaction volume. In some embodiments, the nanoliter-scale reaction environments are configured for at least about 0.02 megabases of genomic material per nanoliter of reaction volume, for example at least about 0.02, 0.03, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 0.6, 0.7, 0.8, 0.9, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1500, 2000, 2500, 3000, 4500, or 5000 megabases of genomic material per nanoliter, including ranges between any two of the listed values. In some embodiments, the nanoliter-scale reaction environments are configured for at least about 0.03 megabases of genomic material per nanoliter of reaction. In some embodiments, the nanoliter-scale reaction environments are configured for at least about 0.3 megabases of genomic material per nanoliter of reaction. In some embodiments, the nanoliter-scale reaction environments are configured for at least about 100 megabases of genomic material per nanoliter of reaction. In some embodiments, the nanoliter-scale reaction environments are configured for at least about 200 megabases of genomic material per nanoliter of reaction. It is further contemplated herein that the nanoliter-scale reaction environments can be configured so that substantially each nanoliter-scale reaction environment comprises only one genome (or cell comprising a genome) when a liquid comprising diluted whole cells or fractions thereof is applied to a substrate as described herein. Accordingly, in some embodiments, each nanoliter-scale reaction environment is configured so that at least about 95% of the nanoliter-scale reaction environments comprises only one cell after administration of the solution comprising cells or fragments thereof, for example at least about 95%, 96%, 97%, 98%, 99%, 99.5%, 99.6%, 99.7%, 99.8%, 99.9%, or 99.99%.
While substantially unbiased amplification in accordance with some embodiments herein can be useful for many applications, one useful application includes genome sequencing. It is contemplated that the substantially unbiased amplification in accordance with some embodiments herein yields amplification of all or substantially all of the template genome at a coverage level that is useful for sequencing. In some embodiments, the nanoliter-scale reaction environments are configured for amplifying at least about 90% of the entire genome therein with >1× coverage, for example at least about 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, 99.5%, or 99.9%, including ranges between any two of the listed values.
In some embodiments, unbranched amplicons are produced for use in library construction. As used herein, “substantially all amplicons are unbranched” and the like refers to at least about 70% of the amplicons (for example, about 70%, 75%, 80%, 85%, 90%, 95%, 97%, 98%, 99%, or 99.9%) do not have a branch characteristic of multiple strand displacement, but rather, are unbranched double-stranded DNA molecules. Without being limited by any theory, it is noted that MDA products are typically highly branched. In some embodiments, unbranched amplicons can be produced from MDA products by contacting the MDA products with DNA polymerase I.
A variety of sequencing techniques are known to the skilled artisan, and can be used in accordance with embodiments herein. The selection of a sequencing technique can depend on a variety of factors, for example the size and characteristics of a genome being amplified. As a number of embodiments herein comprise or are compatible with massively parallel amplification and sequencing, sequencing techniques compatible with rapid, large-scale “next-generation” sequencing can be useful in accordance with some embodiments herein. Exemplary sequencing techniques include Illumina™ (Solexa) sequencing (Illumina), Ion Torrent™ sequencing (Life Technologies), SOLiD™ sequencing (Life Technologies), and the like.
In some embodiments, an amplification-detection moiety is used to monitor the progress of amplification. As used herein, “amplification-detection moiety” refers broadly to any of number of detectable moieties that produce a detectable type or intensity of signal in the presence of amplification product, for example double-stranded nucleic acid, but do not produce the signal (or produce only low-level or background signal) in the absence of amplification product. A first class of amplification-detection moieties includes dyes that bind specifically to double-stranded DNA, for example intercalating agents. These dyes have a relatively low fluorescence when unbound, and a relatively high fluorescence upon binding to double-stranded nucleic acids. As such, dyes that selectively detect double-stranded can be used to monitor the accumulation of double strained nucleic acids during an amplification reaction. Examples of dyes that selectively detect double-stranded DNA include, but are not limited to SYBR™ Green I dye (Molecular Probes), SYBR™ Green II dye (Molecular Probes), SYBR™ Gold dye (Molecular Probes), Picogreen, dye (Molecular Probes), Hoechst 33258 (Hoechst AG), and cyanine dimer families of dyes such as the YOYO family of dyes (e.g. YOYO-1 and YOYO-3), the TOTO family of dyes (e.g. TOTO-1 and TOTO-3), and the like. Other types of amplification-detection moieties employ derivatives of sequence-specific nucleic acid probes. For example, oligonucleotide probes labeled with one or more dyes, such that upon hybridization to a template nucleic acid, a detectable change in fluorescence is generated. Exemplary amplification-detection moieties in this class include, but are not limited to Taqman™ probes, molecular beacons, and the like. While non-specific dyes may be desirable for some applications, sequence-specific probes can provide more accurate measurements of amplification. One configuration of sequence-specific probe can include one end of the probe tethered to a fluorophore, and the other end of the probe tethered to a quencher. When the probe is unhybridized, it can maintain a stem-loop configuration, in which the fluorophore is quenched by the quencher, thus preventing the fluorophore from fluorescing. When the probe is hybridized to a template nucleic sequence, it is linearized, distancing the fluorophore from the quencher, and thus permitting the fluorophore to fluoresce. Another configuration of sequence-specific probe can include a first probe tethered to a first fluorophore of a FRET pair, and a second probe tethered to a second fluorophore of a FRET pair. The first probe and second probe can be configured to hybridize to sequences of an amplicon that are within sufficient proximity to permit energy transfer by FRET when the first probe and second probe are hybridized to the same amplicon.
In some embodiments, an amplification-detection moiety is used to quantify the double-stranded DNA in each reaction environment. Accordingly, in some embodiments, the products of reaction environments in which a desired amount of amplification has occurred can be selected for downstream applications such as construction of sequencing libraries. Thus, methods according to some embodiments herein can minimize the use of reagents and other resources by only constructing sequencing libraries for single-cell genomes that were actually amplified, and for reducing a need for preparing redundant libraries as a “back-up” against reaction environments that did not amplify.
In some embodiments, the sequence-specific probe comprises an oligonucleotide that is complementary to a sequence to be amplified, and is conjugated to a fluorophore. In some embodiments, the probe is conjugated to two or more fluorophores. Examples of fluorophores include: xanthene dyes, e.g., fluorescein and rhodamine dyes, such as fluorescein isothiocyanate (FITC), 2-[ethylamino)-3-(ethylimino)-2-7-dimethyl-3H-xanthen-9-yl]benzoic acid ethyl ester monohydrochloride (R6G)(emits a response radiation in the wavelength that ranges from about 500 to 560 nm), 1,1,3,3,3′,3′-Hexamethylindodicarbocyanine iodide (HIDC) (emits a response radiation in the wavelength that ranged from about 600 to 660 nm), 6-carboxyfluorescein (commonly known by the abbreviations FAM and F), 6-carboxy-2′,4′,7′,4,7-hexachlorofluorescein (HEX), 6-carboxy-4′,5′-dichloro-2′,7′-dimethoxyfluorescein (JOE or J), N,N,N′,N′-tetramethyl-6-carboxyrhodamine (TAMRA or T), 6-carboxy-X-rhodamine (ROX or R), 5-carboxyrhodamine-6G (R6G5 or G5), 6-carboxyrhodamine-6G (R6G6 or G6), and rhodamine 110; cyanine dyes, e.g. Cy3, Cy5 and Cy7 dyes; coumarins, e.g., umbelliferone; benzimide dyes, e.g. Hoechst 33258; phenanthridine dyes, e.g. Texas Red; ethidium dyes; acridine dyes; carbazole dyes; phenoxazine dyes; porphyrin dyes; polymethine dyes, e.g. cyanine dyes such as Cy3 (emits a response radiation in the wavelength that ranges from about 540 to 580 nm), Cy5 (emits a response radiation in the wavelength that ranges from about 640 to 680 nm), etc; BODIPY dyes and quinoline dyes. Specific fluorophores of interest include: Pyrene, Coumarin, Diethylaminocoumarin, FAM, Fluorescein Chlorotriazinyl, Fluorescein, R110, Eosin, JOE, R6G, HIDC, Tetramethylrhodamine, TAMRA, Lissamine, ROX, Napthofluorescein, Texas Red, Napthofluorescein, Cy3, and Cy5, and the like.
In some embodiments, the sequence-specific probe is conjugated to a quencher. A quencher can absorb electromagnetic radiation and dissipate it as heat, thus remaining dark. Example quenchers include Dabcyl, NFQ's, such as BHQ-1 or BHQ-2 (Biosearch), IOWA BLACK FQ (IDT), and IOWA BLACK RQ (IDT). In some embodiments, the quencher is selected to pair with a fluorophore so as to absorb electromagnetic radiation emitted by the fluorophore. Flourophore/quencher pairs useful in the compositions and methods disclosed herein are well-known in the art, and can be found, e.g., described in S. Marras, “Selection of Fluorophore and Quencher Pairs for Fluorescent Nucleic Acid Hybridization Probes” available at the world wide web site molecular-beacons.org/download/marras,mmb06%28335%293.pdf.
In some embodiments, a fluorophore is attached to a first end of the sequence-specific probe, and a quencher is attached to a second end of the probe. Attachment can include covalent bonding, and can optionally include at least one linker molecule positioned between the probe and the fluorophore or quencher. In some embodiments, a fluorophore is attached to a 5′ end of a probe, and a quencher is attached to a 3′ end of a probe. In some embodiments, a fluorophore is attached to a 3′ end of a probe, and a quencher is attached to a 5′ end of a probe. Examples of probes that can be used in quantitative nucleic acid amplification include molecular beacons, SCORPIONS™ probes (Sigma) and TAQMAN™ probes (Life Technologies).
Substrates comprising a plurality of nanoliter-scale reaction environments can be used in accordance with some embodiments herein.
In some embodiments, the substrate comprises several loading areas, and a plurality of nanoliter-scale reaction environments in fluid communication with each loading area. In some embodiments, applying to a loading area a solution having the total volume of the nanoliter-scale reaction environments for that loading area, and single genomes (for example single cells, or isolated genomes of single cells) at a dilution of about 0.1 genome per reaction environment can result in 99% of the reaction environments in that loading area comprising no more than a single genome (or single cell comprising that genome). For example, if each loading area of the substrate comprises 255 microwell reaction environments, each having a diameter of about 400 μm and a depth of about 100 μm (for a volume of about 12 nl), applying 3 μl of a solution comprising 0.1 cells per microwell (e.g. 26 cells), about 99.5% of the microwells will comprise no more than one cell. It is noted that this number was confirmed via SEM microscopy (see
An exemplary substrate 10 in accordance with some embodiments herein is schematically illustrated in
It is recognized that each loading area can be loaded with a separate sample, so that multiple samples can be amplified on the same substrate in parallel (one sample in each loading area). Accordingly, in some embodiments, the number of samples being amplified in parallel can readily be scaled up or down. For example, if the number of samples is less than or equal to the total number of loading areas on the substrate, the appropriate number of loading areas can be selected for parallel reactions. If the number of samples is greater than the total number of loading areas on the substrate, two or more substrates can be used to accommodate the total number of samples.
In some embodiments, the substrate 100 comprises 16 loading areas 12, and each loading area 12 comprises 255 nanoliter-scale loading environments 14. Each nanoliter-scale reaction environment 14 can have a diameter of about 400 μm and a depth of about 100 μm, for a volume of about 12 nl. The substrate can comprise PDMS. Each loading area can have a height of about 7 mm and a width of about 7 mm. The loading areas can be arranged in a pattern on the substrate.
In some embodiments, the substrate further includes a detector for amplification-detection moieties. The detector need not be attached to the substrate. For example, the substrate can be positioned in optical communication with a fluorescent microscope, and optionally a camera. In accordance with some embodiments herein, an amplification-detection moiety can be present in the nanoliter-scale reaction environments, and can indicate when a desired amount of amplification of nucleic acid has occurred in a particular nanoliter-scale reaction environment. Accordingly, in some embodiments, the detector is configured to detect nanoliter-scale reaction environments in which a desired amount of amplification has occurred. In some embodiments, a manual user can select one or more nanoliter-scale reaction environments for downstream applications such as library construction based on the signal detected by the detector. In some embodiments, one or more nanoliter-scale reaction environments are automatically selected for downstream applications such as for library construction based on the amount of signal detected by the detector.
In some embodiments, the substrate further comprises a pipettor for withdrawing amplified nucleic acid from a selected nanoliter-scale reaction environment. The pipettor can be configured to withdraw nanoliter-scale volumes or less from the selected well. In some embodiments, the pipettor comprises a pipette having a diameter less than the diameter of the nanoliter-scale reaction environment. In some embodiments, the pipette has a diameter of no more than about 50 μm, for example about 50 μm, 45, 40, 35, 30, 25, 20, 15, 10, or 5 μm, including ranges between any two of the listed values. In some embodiments, the pipette has a diameter of about 30 μm. In some embodiments, the pipette is a glass pipette. The pipette can be sterile. In some embodiments, the pipettor is under the mechanical control of a manual micromanipulator so that a user can manually select a nanoliter-scale reaction environment of interest for withdrawing liquid, for example amplified nucleic acid. In some embodiments, the pipettor is under the mechanical control of an automatic micromanipulator in data communication with a detector as described herein, so that the pipettor can automatically withdraw liquid from a nanoliter-scale reaction environment exhibiting a desired level of amplification.
In some embodiments, the genome of microbial and/or human cells is sequenced. Some embodiments include assembly of genomes of single bacterial cells with very little sequencing effort. Some embodiments include calling copy number variations on single human neurons down to a 1-2 megabase resolution.
Methods and manufactures in accordance with some embodiments herein can be useful for one or more of: De novo assembly of unculturable bacteria in the human gut; De novo assembly of unculturable bacteria in heterogeneous environments such as sea water; Copy number variation calling on single neurons; Copy number variation calling on single cancerous cells or circulating tumor cells; and haplotyping, for example Human haplotyping.
In some embodiments, the genome of a single cell is amplified. In some embodiments the cell is a human cell. In some embodiments, the cell is a microbial cell. In some embodiments, the cell is a bacterial cell. In some embodiments, the cell is from a substantially unculturable strain. As used herein, “substantially unculturable” and variations thereof refer to a strain that, when cultured under normal laboratory conditions, fewer than 20% of replicates of that strain will reach a logarithmic growth phase, for example fewer than 20%, 15%, 10%, 5%, 2%, 1%, or 0.1%.
For previous techniques, a major technical challenge was the highly uneven amplification of the one or two copies of each chromosome in a single cell. This high amplification bias leads to difficulties in assembling microbial genomes de novo and inaccurate identification of copy number variants (CNV) or heterozygous single nucleotide changes in single mammalian cells. Recent developments of bias-tolerant algorithms11, 12 have greatly mitigated the effects of uneven read depth on de novo genome assembly and CNV calling, yet an unusually high sequencing depth is still required, making this approach impractical for organisms with large genome sizes.
Several strategies have been previously developed to reduce amplification bias, including reduced reaction volume13, 14 and supplementing amplification reactions with single-strand binding proteins or Threhalose15, 16. Post-amplification normalization by digesting highly abundant sequences with a duplex-specific nuclease has also been utilized to markedly reduce bias17. Despite these efforts, amplification bias still remains the primary technical challenge in single-cell genome sequencing. Using cells that contain multiple copies of the genome or multiple clonal cells has been the only viable solution to achieve near complete genome coverage with MDA18, 19. Without being bound by any particular theory, we reasoned that amplification is always bias-prone, and that by limiting the amplification magnitude to “just-enough” for sequencing in accordance with some embodiments herein, we could potentially reduce the bias. In addition, we supposed that reducing the reaction volume by ˜1000 fold to nanoliter level and thus dramatically increasing the effective concentration of the template genome might reduce contamination and improve primer annealing and hence amplification uniformity13, 14. To these ends, we developed the microwell displacement amplification system (MIDAS) in accordance with some embodiments herein, a microwell-based platform that allows for highly parallel polymerase cloning of single cells in thousands of nanoliter reactors of 12 nL in volume, the smallest volume that has been implemented to date to the best knowledge of Applicants. Coupled with a low-input library construction method, we achieved highly uniform coverage in the genomes of both microbial and mammalian cells. We demonstrated substantial improvement both in de novo genome assembly from single microbial cells and in the ability to detect small somatic copy number variants in individual human adult neurons with minimal sequencing effort.
Due to the extreme bias that can be caused by whole genome amplification from a single DNA molecule, genomic analysis of single cells has traditionally been a challenging task. Traditionally, a large amount of sequencing resources can be required to produce a draft quality genome assembly or determine a low-resolution copy number variation profile due to amplification bias and coverage dropout. MIDAS in accordance with some embodiments herein addresses this issue through the use of nanoliter scale volumes to generate nanogram level amplicons and the use of a low-input transposon-based library construction method. Compared to the traditional single-cell library construction and sequencing protocol, MIDAS in accordance with some embodiments herein provides a more uniform, higher-coverage, and lower cost way to analyze single cells from a heterogeneous population.
MIDAS was applied to single E. coli cells and resolved nearly the entire genome with relatively low sequencing depth. Additionally, using de novo assembly on MIDAS libraries, over 90 percent of the genome was assembled. Thus, in some embodiments, MIDAS is applied to an uncultivated organism to provide a draft quality assembly with more genes covered and less sequencing resource expenditure. Currently, a majority of unculturable bacteria are analyzed metagenomically as part of a mixed population rather than individually. Although metagenomics allows for the discovery of novel genes, individual sequences cannot be resolved. The biased nature of traditional MDA-based methods when applied to single cells has proved single cell microbial analysis challenging in terms of de novo genome assembly. Despite recent success in analyzing partially assembled single cell genomes7, the full potential of single cell genomics remains to be fully explored. As such, in some embodiments the use of MIDAS on heterogeneous environmental samples, novel single-cell organisms and genes can be easily discovered and characterized in a low-cost and high-throughput manner, allowing a much higher-resolution and more complete analysis of single bacterial cells.
In some embodiments, MIDAS is applied to the analysis of copy number variation in single human neuronal nuclei. With a low amount of sequencing effort, MIDAS was able to systematically call single copy number changes of 2 million base pairs or larger in size. It has been shown recently that, in human adult brains, post-mitotic neurons in different brain regions exhibited various levels of DNA content variation (DCV)29. The exact genomic regions that associate with DCV have been difficult to map to single neurons because of the amplification bias with existing MDA-based methods. CNVs in single tumor cells have been successfully characterized with a PCR-based whole genome amplification method8. However, tumor cells tend to be highly aneuploid and exhibit copy number changes of larger magnitude, which are more easily detected. The applicability of this strategy to other primary cell types with more subtle CNV events remains unclear. We have demonstrated that MIDAS greatly reduces the variability of single cell analysis to a level such that a small single-copy change is detectible, allowing characterization of much more subtle copy number variation. MIDAS can be used to simultaneously probe into the individual genomes of many cells from patients with neurological diseases, and thus will allow identification of a range of structural genomic variants and eventually allow accurate determination of the influence of somatic CNVs on brain disorders in a high-throughput manner.
In some embodiments, MIDAS compares very favorably to traditional MDA-based methods. Recently, another single cell sequencing method that dramatically reduces amplification bias and increases genomic coverage was reported. Known as MALBAC, this method incorporates a novel enzymatic strategy to amplify single DNA molecules initially through quasi-linear amplification to a limited magnitude prior to exponential amplification and library construction32. MALBAC was implemented in microliter reactions in conventional reaction tubes. In contrast, MIDAS represents an orthogonal strategy by adapting MDA to a microwell platform. It will therefore be more easily able to analyze a larger number of single cells in parallel in a single experiment. While both MIDAS and MALBAC show relatively unbiased amplification across the genome (
MIDAS can provide researchers with a powerful tool for many other applications, including high-coverage end-to-end haplotyping of mammalian genomes or probing de novo CNV events at the single-cell level during the induction of pluripotency or stem cell differentiation33. MIDAS can allow for efficient high-throughput sequencing of a variety of organisms at a relatively low price. This new technology should help propel single cell genomics, enhance our ability to identify diversity in multicellular organisms, and lead to the discovery of thousands of new organisms in various environments.
With reference to Examples 1-5, the following methods were used. The skilled artisan will appreciate that the following methods can readily be used or adapted or modified in accordance with some embodiments herein:
Microwell arrays were fabricated from polydimethylsiloxane (PDMS). Each array was 7 mm×7 mm, with 2 rows of 8 arrays per slide and 156 microwells per array. The individual microwells were 400 μm in diameter and 100 um deep (˜12 nL volume), and were arranged in honeycomb patterns in order to minimize space in between the wells. To fabricate the arrays, first, an SU-8 mold was created using soft lithography at the Nano3 facility at UC San Diego. Next, a 10:1 ratio of polymer to curing agent mixture of PDMS was poured over the mold. Finally, the PDMS was degassed and cured for 3 hours at 65 C.
E. coli K12 MG1655 was cultured overnight, collected in log-phase, and washed 3× in PBS. After quantification, the solution was diluted to 10 cells/μL. Human neuronal nuclei were isolated as previously described29, 34 and fixed in ice-cold 70% ethanol. Nuclei were labeled with a monoclonal mouse antibody against NeuN (1:100 dilution) (Chemicon, Temecula, Calif.) and an AlexaFluor 488 goat anti-mouse IgG secondary antibody (1:500 dilution) (Life Technologies, San Diego, Calif.). Nuclei were counterstained with propidium iodide (50 ug/ml) (Sigma, St. Louis, Mo.) in PBS solution containing 50 ug/ml RNase A (Sigma) and chick erythrocyte nuclei (Biosure, Grass Valley, Calif.). Nuclei in the G1/G0 cell cycle peak, determined by propidium iodide fluorescence, were electronically gated on a Becton Dickinson FACS-Aria II (BD Biosciences, San Jose, Calif.) and selectively collected based on NeuN+ immunoreactivity.
All reagents not containing DNA or enzymes were first exposed to ultraviolet light for 10 minutes prior to use. The PDMS slides were treated with oxygen plasma to make them hydrophilic and ensure random cell seeding. The slides were then treated with 1% bovine serum albumin (BSA) (EMD Chemicals, Billerica, Mass.) in phosphate buffered saline (PBS) (Gibco, Grand Island, N.Y.) for 30 minutes and washed 3× with PBS to prevent DNA from sticking to the PDMS. The slides were completely dried in a vacuum prior to cell seeding. Cells were diluted to a concentration of 10 cells/μL, and 3 μL of cell dilution was added to each array (30 cells total per array).
Initially, to verify that cell seeding adhered to the Poisson distribution, cells were stained with 1×SYBR green and viewed under a fluorescent microscope. Proper cell distribution was further confirmed with SEM imaging. For SEM imaging, chromium was sputtered onto the seeded cells for 6 seconds to increase conductivity. Note that the imaging of cell seeding was only used to confirm the theoretical Poisson distribution and not performed during actual amplification and sequencing experiments due to the potential introduction of contamination.
After seeding, cells were left to settle into the wells for 10 minutes. The seeded cells were then lysed either with 300 U ReadyLyse lysozyme at 100 U/μL (Epicentre, Madison, Wis.) and incubation at room temperature for 10 minutes, or with 5 1 minute freeze/thaw cycles using a dry ice brick and room temperature in a laminar flow hood. After lysis, 4.5 μL of alkaline lysis (ALS) buffer (400 mM KOH, 100 mM DTT, 10 mM EDTA) was added to each array and incubated on ice for 10 minutes. Then, 4.5 μL of neutralizing (NS) buffer (666 mM Tris-HCl, 250 mM HCL) was added to each array. 11.2 μL of MDA master mix (1× buffer, 0.2×SYBR green I, 1 mM dNTP's, 50 μM thiolated random hexamer primer, 8U phi29 polymerase, Epicentre, Madison, Wis.) was added and the arrays were then covered with mineral oil. The slides were then transferred to the microscope stage enclosed in a custom temperature and humidity controlled incubator set to 30 C. Images were taken at 30-minute intervals for 10 hours using a 488 nm filter.
Images were analyzed with a custom Matlab script to subtract background fluorescence. Because SYBR Green was added to the MDA master mix, fluorescence under a 488 nm filter was expected to increase over time for positive amplifications. If a digital profile of fluorescent wells with increasing fluorescence over time was observed (approximately 10-20 wells per array), the array was kept. If no wells fluoresced, amplification failed and further experiments were stopped. Alternatively, if a majority of the wells fluoresced, the array was considered to be contaminated and subsequent analysis was similarly stopped. If 2 abutting wells fluoresced, neither was extracted due to the higher likelihood of more than one cell in each well existing (as in this case, seeding was potentially non-uniform).
1 mm outer diameter glass pipettes (Sutter, Novato, Calif.) were pulled to ˜30 um diameters, bent to a 45 degree angle under heat, coated with SigmaCote (Sigma, St. Louis, Mo.), and washed 3 times with dH20.
Wells with positive amplification were identified using the custom Matlab script described above. A digital micromanipulation system (Sutter, Novato, Calif.) was used for amplicon extraction. The glass pipette was loaded into the micromanipulator and moved over the well of interest. The microscope filter was switched to bright field and the pipette was lowered into the well. Negative pressure was slowly applied, and the well contents were visualized proceeding into the pipette. The filter was then switched back to 488 nm to ensure the well was no longer fluorescent. Amplicons were deposited in 1 μL dH20.
For quantification of microwell amplification, 0.5 μL of amplicon was amplified a second time using MDA in a 20 μL PCR tube reaction (1× buffer, 0.2×SYBR green I, 1 mM dNTP's, 50 mM thiolated random hexamer primer, 8U phi29 polymerase). After purification using Ampure XP beads (Beckman Coulter, Brea, Calif.), the 2nd round amplicon was quantified using a Nanodrop spectrophotometer. The 2nd round amplicon was then diluted to 1 ng, 100 pg, 10 pg, 1 pg, and 100 fg to create an amplicon ladder. Subsequently, the remaining 0.5 μL of the 1st round amplicon was amplified using MDA along with the amplicon ladder in a quantitative PCR machine. The samples were allowed to amplify to completion, and the time required for each to reach 0.5× of the maximum fluorescence was extracted. The original amplicon concentration could then be interpolated.
1.5 μL of ALS buffer was added to the extracted amplicons to denature the DNA followed by a 3-minute incubation at room temperature. 1.5 μL of NS buffer was added on ice to neutralize the solution. 10 U of DNA Polymerase I (Invitrogen, Carlsbad, Calif.) was added to the denatured amplicons along with 250 nanograms of unmodified random hexamer primer, 1 mM dNTPs, 1× Ampligase buffer (Epicentre, Madison, Wis.), and 1×NEB buffer 2 (NEB, Cambridge, Mass.). The solution was incubated at 37 C for 1 hour, allowing second strand synthesis. 1 U of Ampligase was added to seal nicks and the reaction was incubated first at 37 C for 10 minutes and then at 65 C for 10 minutes. The reaction was cleaned using standard ethanol precipitation and eluted in 4 μL water.
Nextera transposase enzymes (Epicentre, Madison, Wis.) were diluted 100 fold in 1×TE buffer and glycerol. 10 μL transposase reactions were then conducted on the eluted amplicons after addition of 1 μL of the diluted enzymes and 1× tagment DNA buffer. The reactions were incubated for 5 minutes at 55 C for mammalian cells and 1 minute at 55 C for bacterial cells. 0.05 U of protease (Qiagen, Hilden, Germany) was added to each sample to inactivate the transposase enzymes; the protease reactions were incubated at 50 C for 10 minutes followed by 65 C for 20 minutes. 5 U Exo minus Klenow (Epicentre, Madison, Wis.) and 1 mM dNTP's were added and incubated at 37 C for 15 minutes followed by 65 C for 20 minutes. Two stage quantitative PCR using 1×KAPA Robust 2G master mix (Kapa Biosystems, Woburn, Mass.), 10 μM Adapter 1, 10 μM barcoded Adapter 2 in the first stage, and 1×KAPA Robust 2G master mix, 10 μM Illumina primer 1, 10 μM Illumina primer 2, and 0.4×SYBR Green I in the second stage was performed and the reaction was stopped before amplification curves reached their plateaus. The reactions were then cleaned up using Ampure XP beads in a 1:1 ratio. A 6% PAGE gel verified successful tagmentation reactions.
Bacterial libraries were size selected into the 300-600 bp range and sequenced in an Illumina Genome Analyzer IIx, Illumina HiSeq, or Illumina MiSeq using 100 bp paired end reads. E. coli data was both mapped to the reference genome and de novo assembled. For the mapping analysis, libraries were mapped as single end reads to the reference E. coli K12 MG1655 genome using default Bowtie parameters. Contamination was analyzed, and clonal reads were removed using SAMtools' rmdup function. For the de novo assembly, paired end reads with a combined length less than 200 bp were first joined and treated as single end reads. All remaining paired end reads and newly generated single end reads were then quality trimmed. De novo assembly was performed using SPAdes11 v. 2.4.0. Corrected reads were assembled with kmer values of 21, 33, and 55. The assembled scaffolds were mapped to the NCBI nt database with BLAST, and the organism distribution was visualized using MEGAN35. Obvious contaminants (e.g., human) were removed from the assembly and the assembly was analyzed using QUAST36. The remaining contigs were annotated using RAST37 and KAAS38.
To implement “just-enough” amplification and thus limit the effects of the exponential amplification bias from MDA in a highly parallel manner, we designed and fabricated microwell arrays of a size comparable to standard microscope slides. The format of the microwell arrays, including well size, pattern, and spacing, was optimized to achieve efficient cell loading, optimal amplification yield, and convenient DNA extraction. Each slide contained 16 arrays each containing 156 microwells of 400 μm in diameter, allowing for parallel amplification of 16 separate heterogeneous cell populations (
After seeding of cell populations into each microwell array, we performed limited Multiple Displacement Amplification (MDA) on the seeded single cells at a reaction volume of ˜12 nL in a temperature and humidity controlled chamber (
To construct Illumina sequencing libraries from the nanogram-scale DNA amplicons, we used a modified method based on the Nextera Tn5 transposase library construction kit. Previous studies have shown that Nextera transposase-based libraries can be prepared using as little as 10 picograms of genomic DNA21. However, the standard Nextera protocol was unable to generate high-complexity libraries from MDA amplicons, resulting in poor genomic coverage (data not shown). To address this issue, we used random hexamers and DNA Polymerase I to first convert the hyperbranched amplicons into unbranched double-stranded DNA molecules, which allowed effective library construction using the Nextera™ in vitro transposition method (
Thus, a sequencing library was constructed using products of substantially unbiased amplification in accordance with some embodiments herein.
As a proof of concept, we utilized MIDAS on three single MG1655 E. coli cells and analyzed between approximately 2-8 million paired-end Illumina sequencing reads of 100 bp in length for each, which is equivalent to a genomic coverage of between 87× and 364×. We first mapped the reads to the reference E. coli genome and were able to recover between 94% and 99% of the genome at >1× coverage. We then performed de novo genome assembly using SPAdes22. We were able to assemble between 88% and 94% of the E. coli genome (
As a control, we also amplified and sequenced one E. coli cell using the conventional in-tube MDA method, and controlled the reaction time to limit the amplification yield to the nanogram level. A fraction of the control amplicon was further amplified in a second reaction to the microgram level. The two control amplicons were converted into sequencing libraries using the traditional shearing and ligation method. We found that limiting amplification yield resulted in a reduction of amplification bias even for in-tube amplification. However, MIDAS had a markedly reduced level of amplification bias when compared with either control reaction (
Given the highly uniform genome coverage achieved by microwell based polymerase cloning, we next applied MIDAS to the characterization of copy number variation in single mammalian cells. The higher cognitive function of the human brain is supported by a complex network of neurons and glia. It has been long thought that all cells in a human brain share the same genome. Without being bound by any particular theory, recent evidence suggests that individual neurons could have non-identical genomes due to aneuploidy23-26, active retrotransposons27, 28 and other DNA content variation29. However, the presence of somatic genetic variation in individual neurons has never been conclusively demonstrated at the single genome scale. To demonstrate the viability of MIDAS as a platform for investigating copy number variation in single primary human neurons, we prepared nuclei from one post-mortem brain sample from a healthy female donor and a second post-mortem brain sample from a female individual with Down Syndrome. We purified cortical neuronal nuclei by flow sorting based on neuron-specific NeuN antibody staining. Five sequencing libraries (two disease-free, three Down Syndrome) were generated from individual nuclei using MIDAS, and generated sequencing data was analyzed using an SNS method based on circular binary segmentation30. We similarly observed a dramatic reduction of amplification bias in the MIDAS libraries when compared to the conventional in-tube MDA-based method (
We next sought to characterize the sensitivity of detecting single copy-number changes. While it was not possible, even with aggressive binning into large genomic regions, to distinguish true copy number differences from random amplification bias in the conventional single-cell MDA library, the uniform genome coverage in the MIDAS libraries allowed clear detection of Trisomy 21 in each of the Down Syndrome nuclei (
Thus, substantially unbiased amplification in accordance with some embodiments herein can sensitively detect changes in copy number of portions of a genome.
Mammalian single-cell libraries were sequenced in an Illumina Genome Analyzer IIx or Illumina HiSeq using 36 bp single end reads. The CNV algorithm previously published by Cold Spring Harbor Laboratories8 was used to call copy number variation on each single neuron, with modifications to successfully analyze non-cancer cells. Briefly, for each sample, reads were mapped to the genome using Bowtie. Clonal reads resulting from Polymerase Chain Reaction artifacts were removed using samtools, and the remaining unique reads were then assigned into 49,891 genomic bins that were previously determined such that each would contain a similar number of reads after mapping30. Each bin's read count was then expressed as a value relative to the average number of reads per bin in the sample, and then normalized by GC content of each bin using a weighted sum of least squares algorithm (LOWESS). Circular binary segmentation was then used to divide each chromosome's bins into adjacent segments with similar means. Unlike the previously published algorithm, in which a histogram of bin counts was then plotted and the second peak chosen as representing a copy number of two, it was assumed, due to samples not being cancerous and thus being unlikely to contain significant amounts of aneuploidy, that the mean bin count in each sample would correspond to a copy number of two. Each segment's normalized bin count was thus multiplied by two and rounded to the nearest integer to call copy number. MIDAS data clearly showed a CNV call designating Trisomy 21 in all Down Syndrome single cells, while the traditional MDA-based method was not able to call Trisomy 21.
In order to test the ability of the CNV algorithm described above to call small CNVs, artificial CNVs were computationally constructed. Prior to circular binary segmentation, in each Down Syndrome sample, one hundred random genomic regions across chromosomes 1-22 were chosen, each consisting of either 17 or 34 bins of approximately 60 kb in size. Each region was replaced with an equivalently sized region from chromosome 21 or chromosome 4 (Supplementary Table 2). The above algorithm was then run on each “spiked-in” sample, and the number of new CNV calls in each sample that matched each spike-in was tallied. For the chromosome 21 spike-ins, MIDAS was able to accurately call 98% of spiked-in CNVs at the 2 Mb level and 68% of spiked-in CNVs at the 1 Mb level, while the traditional MDA-based method was not able to call any spiked-in CNVs. As expected, spike-ins of chromosome 4 did not result in any additional CNV calls.
Thus, small CNV's can be called in accordance with some embodiments herein.
1
Healthy
1
16,949,551
17,257,431
5
307,881
120,000
1
Healthy
1
147,802,093
149,049,044
3
1,246,952
120,000
1
Healthy
2
133,000,723
133,135,043
4
134,321
120,000
1
Healthy
4
190,664,845
191,154,276
4
489,432
240,000
1
Healthy
6
32,526,395
32,645,736
1
119,342
120,000
1
Healthy
8
39,308,029
39,363,306
1
55,278
60,000
1
Healthy
10
47,008,316
47,538,599
4
530,284
180,000
1
Healthy
15
34,761,777
34,873,738
1
111,962
60,000
1
Healthy
16
71,141,287
71,246,392
7
105,106
60,000
1
Healthy
17
21,257,685
21,374,155
3
116,471
120,000
1
Healthy
20
29,449,066
29,811,435
4
362,370
120,000
2
Healthy
1
16,949,551
17,257,431
4
307,881
120,000
2
Healthy
1
147,802,093
149,049,044
4
1,246,952
120,000
2
Healthy
6
32,526,395
32,699,933
1
173,539
180,000
2
Healthy
8
39,308,029
39,363,306
1
55,278
60,000
2
Healthy
10
47,008,316
47,538,599
4
530,284
180,000
2
Healthy
15
34,761,777
34,873,738
1
111,962
60,000
2
Healthy
16
71,141,287
71,246,392
9
105,106
60,000
2
Healthy
17
21,257,685
21,374,155
3
116,471
120,000
3
Down Syndrome
4
190,664,845
191,154,276
4
489,432
240,000
3
Down Syndrome
8
39,308,029
39,363,306
0
55,278
60,000
3
Down Syndrome
10
38,869,769
42,858,972
6
3,989,204
240,000
3
Down Syndrome
16
71,141,287
71,246,392
11
105,106
60,000
3
Down Syndrome
20
29,449,066
29,868,184
6
419,119
180,000
3
Down Syndrome
21
14,432,540
48,129,895
3
33,697,356
36,180,000
4
Down Syndrome
2
133,000,723
133,065,776
39
65,054
60,000
4
Down Syndrome
8
39,308,029
39,415,337
0
107,309
120,000
4
Down Syndrome
10
38,869,769
42,858,972
8
3,989,204
240,000
4
Down Syndrome
20
29,449,066
29,927,709
6
478,644
240,000
4
Down Syndrome
21
14,432,540
48,129,895
3
33,697,356
36,180,000
5
Down Syndrome
10
38,869,769
42,858,972
9
3,989,204
240,000
5
Down Syndrome
20
29,449,066
29,868,184
9
419,119
180,000
5
Down Syndrome
21
14,432,540
48,129,895
3
33,697,356
36,180,000
The disclosures of all references cited herein are incorporated herein by reference in their entireties.
In this application, the use of the singular can include the plural unless specifically stated otherwise or unless, as will be understood by one of skill in the art in light of the present disclosure, the singular is the only functional embodiment. Thus, for example, “a” can mean more than one, and “one embodiment” can mean that the description applies to multiple embodiments.
The foregoing description and Examples detail certain embodiments. It will be appreciated, however, that no matter how detailed the foregoing may appear in text, the invention may be practiced in many ways and the invention should be construed in accordance with the appended claims and any equivalents thereof.
The present application claims the benefit of U.S. Provisional App. No. 61/829,193, filed May 30, 2013, which is hereby incorporated by reference in its entirety.
This research was supported by National Institutes of Health Grant R01HG004876. The government may have certain rights in the invention.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/US14/39830 | 5/28/2014 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 61829193 | May 2013 | US |
