Anal. Chem. 2008, 80, 7639–7642
High-Efficiency Microarray Printer Using Fused-Silica Capillary Tube Printing Pins Steve M. Clark,†,‡ Gregory E. Hamilton,† Robert A. Nordmeyer,§ Donald Uber,§ Earl W. Cornell,§ Nils Brown,† Richard Segraves,† Randy Davis,† Donna G. Albertson,† and Daniel Pinkel*,† University of California San Francisco, Box 0808, San Francisco, California 94143, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720, and GlaxoSmithKline, UW 2108, 709 Swedeland Road, P.O. Box 1539, King of Prussia, Pennsylvania 19406 We describe a contact printing approach for microarrays that uses fused-silica capillary tubes with tapered tips for printing pins and a pressure/vacuum system to control pin loading, printing, and cleaning. The printing process is insensitive to variable environmental factors such as humidity, and the small diameter of the pins allows routine printing from 1536 well source plates. Pin load capacity, 0.2 µL in the current system, is adjustable by controlling pin length. More than 2000 spots can be printed per 0.2-µL pin load (12 000 spots/cm2 are readily achievable. Solutions with a wide range of viscosities and chemical properties can be printed. The system can print tens of thousands of different solutions at high speed, due to the ability to use large numbers of pins simultaneously, and can produce a large number of replicate arrays since all of the solution picked up by the pins is available for deposition. Many approaches have been developed for producing microarrays. Some use printing pins that contact the array substrate, while others use noncontacting methods such as ink jets and electrosrpray. The deposited material may contain either preformed molecules or building blocks that are sequentially deposited and linked to form the final molecule. Each of these production approaches has advantages and disadvantages, making some more suitable for particular applications than others. These technologies have been comprehensively reviewed by Barbulovic-Nad and colleagues.1 Using these approaches, arrays that contain from several to millions of elements can be produced, with elements spacing ranging from nanometers to millimeters. Laboratory production of arrays is typically accomplished using contact printing methods that employ pins of various sorts to transfer solutions from microtiter source plates to the array substrate. These arrays are typically printed with element densities of 5000->10 000/cm,2 allowing tens of thousands to be printed on a microscope slide. The pins are typically solid and can deposit * To whom correspondence should be addressed. E-mail:
[email protected]. † University of California San Francisco. ‡ GlaxoSmithKline. § Lawrence Berkeley National Laboratory. (1) Barbulovic-Nad, I.; Lucente, M.; Sun, Y.; Zhang, M.; Wheeler, A. R.; Bussmann, M. Crit. Rev. Biotechnol. 2006, 26, 237–259. (2) George, R. A.; Woolley, J. P.; Spellman, P. T. Genome Res. 2001, 11, 1780– 1783. 10.1021/ac8010395 CCC: $40.75 2008 American Chemical Society Published on Web 09/03/2008
only one or a few array elements each time they are loaded, or they contain a capillary slot that is filled on dipping into the source plate. The slotted (split) pins can print several hundred elements per load, but much of the solution may be unavailable for deposition due to evaporation from the open edges of the capillary slot. Occasionally cylindrical capillaries have been used,2 which prevent this evaporation (www.labnext.com). Solid and split pins have been made from a variety of materials including metals (www.arrayit.com, www.genetix.com, www.majerprecision.com, www.pointtech.com), ceramic (www.labnext.com), quartz,3 and silicon (www.parallel-synthesis.com). The most widely used laboratory systems employ passive loading of the pins by capillary action, and most pin designs use the forces generated by impact with the substrate to assist with the deposition of the printing solution on the substrate. These forces may be sufficient to deform malleable materials such as metals, so that printing tip life may be limited. In spite of their drawbacks, the passive printing pins are the most widely used in research laboratories because they do not require incorporating fluid control capability in the printer. Printing with these types of pins has recently been reviewed by George.4 The printing approach described in this note differs from others by using slender, cylindrical fused-silica printing pins and employing pressure and vacuum to assist with control of the printing and reloading. This system can reliably load small amounts of solution and use it to completion, print rapidly since pins can be mounted at high density, and make arrays from solutions with a wide range of viscosities and chemical properties. Array printing is reliable and easy to manage in the laboratory but requires the inclusion of appropriate hardware and software for the fluid control in the printer design. A brief overview of the system and its performance is presented below. Design details of the print head, operational details of the print cycle, and an overview of the mechanical design of two generations of printers are presented in the Supporting Information. EXPERIMENTAL SECTION The printing pins and print head are shown in Figure 1a. The pins are made from 360-µm-outside diameter/75-µm-inside diameter capillary electrophoresis tubing, which has been heated and pulled to reduce the inside diameter to ∼30-40 µm at the tip. (3) Tehan, E. C.; Higbee, D. J.; Wood, T. D.; Bright, F. V. Anal. Chem. 2007, 79, 5429–5434. (4) George, R. A. Methods Enzymol. 2006, 410, 121–135.
Analytical Chemistry, Vol. 80, No. 19, October 1, 2008
7639
Figure 1. Capillary print head design and control system. (a) Details of print head and printing pins. Each printing pin is cemented into a piston. The piston is inserted into a cylinder, and the lateral motion of the pin is further constrained by guide plates. Vacuum holds the pistons in the “down” position but allows the pin and cylinder to move upward relative to the print head when the pins contact the array substrate. The upper left image shows several pistons, with one intentionally raised several millimeters to simulate the (exaggerated) motion after the pin tip contacts the array substrate. During actual printing the motion is ∼200 µm. Detailed images of the beveled print tips are shown in the lower left images. A 64-pin print head for 1536-well plates is shown at the upper right. The tips of the printing pins are shown reflected in the chromium-coated microscope slides on which most printing is performed. (b) Pressure and vacuum system for controlling the print head. The upper end of each printing pin is connected to a port on a manifold using flexible tubing. The right end of the manifold is always kept at equal or higher pressure than the left side so that air and liquid flow is always toward the waste bottle. The manifold is mounted on the stage with the print head. Details of the operation of the system are in the Supporting Information.
The outside of the tip is then beveled by grinding to a diameter of ∼70-90 µm. This design produces a pin with substantial wall thickness just above the tip so that it is very robust. The pins do not chip or deform. Currently, pins ∼40 mm in length are employed, which have a fluid capacity of ∼0.2 µL. The pins are cemented into small pistons and inserted into the print head that contains a vacuum chamber. Low pressure in this chamber holds the pins in their “rest” position between contacts with the printing substrate. A picture of a 64-pin print head for 1536-well plates (2.25mm pin spacing) is shown in the upper right corner of the figure. Pin loading and cleaning and array printing are controlled by the pressure/vacuum system illustrated in Figure 1b. The upper end of each pin is connected to a manifold by flexible tubing. One side of the manifold is connected to the vacuum side of the system, which includes a waste bottle, and the other side is connected to 7640
Analytical Chemistry, Vol. 80, No. 19, October 1, 2008
atmospheric vents and pressure sources. The print head and manifold are mounted together, and this assembly is then mounted on the printer. Precision locating holes and dowels allow reproducible mounting of the print heads so that no realignment is required when a head is changed. Pins are loaded by dipping them into the microtiter source plate and applying a brief vacuum pulse to ensure that loading starts. They complete filling by capillary action. During printing, a small positive pressure is applied to the manifold to keep the liquid at the pin tip so that the tip is always wet. When the pin contacts the substrate the printing solutions wets it, and a very reproducible drop is pulled from the pin as it is raised. Pressure is not used to specifically expel the liquid. The pins are cleaned in sequential sonication baths, the first containing a small amount of glass cleaner, and the second distilled water. Various sequences
Figure 2. Image of four of 64 subarrays printed with the 64-pin print head. Subarrays fill a 2.25-mm square and contain 256 spots on 130µm centers. DNA from ligation-mediated PCR representations of human genomic BAC clones was isolated, dissolved in 20% DMSO/ 80% distilled water at a concentration of 2 µg/µL,7 loaded into a 1536well microtiter plate, and printed on chromium-coated microscope slides using the 64-pin print head shown in Figure 1. After drying in air and UV cross-linking, the slide was mounted in a 90% glycerol buffer containing the DNA stain DAPI under a coverslip. The image was obtained by a custom-built imaging system using ∼ 360-nm excitation from a mercury-xenon arc lamp.5 This ∼4.5-mm square is a portion of the full image, which was obtained with a single exposure and covers the entire 18-mm square print area. Resolution is 10 µm/pixel.
of pressure and vacuum are employed to ensure that the pins finish clean and dry. An inexpensive hair dryer is used for final drying. Each pin behaves independently, so that plugged pins or pins dipping into empty wells in the microtiter plate do not affect the operation of their neighbors. Details of the operation of this system are contained in the Supporting Information. RESULTS AND DISCUSSION Printing Accuracy. Figure 2 shows 4 subarrays from an 18mm square array printed with a 64-pin print head. Each subarray contains 256 spots printed in a continuous sequence from a single load of the pins. All spots printed after the loading are shown; e.g., no sacrificial prespotting was employed to remove an initial transient in spot size. The spots have 130-µm center-to-center spacing and have been imaged using a custom-built CCD system that acquires the entire 18-mm square area in a single exposure.5 Measurements of the positions of the centers of mass of each of the spots indicate a standard deviation of 1 µm in placement precision within a subarray. The positions of the subarrays relative to each other are more variable, primarily due to the precision (5) Hamilton, G.; Brown, N.; Oseroff, V.; Huey, B.; Segraves, R.; Sudar, D.; Kumler, J.; Albertson, D.; Pinkel, D. Nucleic Acids Res. 2006, 34, e58.
with which the tips of the printing pins are located relative to the axis of the pin. Maximum variation is ∼40 µm, or ∼10% of the shaft diameter of the pins. This variation does not interfere with automated analysis of array images6 but could be reduced by more precise fabrication of the pin tips if desired. Each tip produces spots of highly uniform diameter as long as the surface properties of the substrate are uniform. In these images, which are typical of the printing performance, the coefficient of variation of the spot diameter is ∼2.5%. For these four subarrays, the mean diameter of all spots is 100 µm, with a coefficient of variation of 5%. The uniformity in spot size among subarrays is dependent on the uniformity of the outside diameters of the pin tips. We typically employ a pin if it produces a spot diameter between 70 and 100 µm. Given our printing solutions and imaging, we do not find that the variation in spot diameter among pins affects the measurements. Printing Capability. The printer shown in Figure S1a of the Supporting Information has been in routine operation in an academic core facility since early 2002. Most of the printing has involved DNA microarrays, with DNA fragments ranging in size from oligonucleotides to 20 kb. Among the standard arrays that are produced are ∼2500 locus genome scanning arrays7 for the human and mouse and a ∼32 000 locus human genome tiling array, all made from bacterial artificial chromosome clones. The elements of the smaller arrays are printed in triplicate so that each array contains ∼7500 spots, while single spots are printed for the larger array. The smaller arrays are printed 2 per slide on each of 238 slides, so that ∼1500 spots are printed by each pin each time it is loaded. These arrays are printed in a 13.5-mm square from 1536-well plates using 36 printing pins (476 arrays in 7 h), while the genome tiling array is printed from 1536-well plates using a 64-pin head (238 arrays in 20 h). The later array is printed as two 18-mm square blocks (total area 18 mm × 36 mm). Printing times are substantially shorter and slide capacity is somewhat larger for the printer shown in Figure S1b (Supporting Information). Routine printing is performed with center-to-center spacing of the spots of 120-130 µm and spot diameters of 70-100 µm. Spots under 50 µm in diameter, allowing correspondingly closer spacing, can be obtained by increasing the hydrophobicity of the surface. The printing system is very frugal and reliable. More than 10 000 of the 2500 locus arrays are produced from ∼12 µL per locus of printing solution, and the printing reliability is >99%. Details of the mechanical designs of the printers are provided in the Supporting Information. The efficient use of printing solutions makes this printing system particularly well suited for applications where many copies of an array are produced in a single print run. For example, for high-throughput screening purposes, this system could print small arrays on the bottoms of wells in microtiter plates of essentially any standard format. Since the pins are thin over their entire length, printing can be accomplished over most of the area of the bottom of the well using one or several printing pins. With (6) Jain, A. N.; Tokuyasu, T. A.; Snijders, A. M.; Segraves, R.; Albertson, D. G.; Pinkel, D. Genome Res. 2002, 12, 325–332. (7) Snijders, A. M.; Nowak, N.; Segraves, R.; Blackwood, S.; Brown, N.; Conroy, J.; Hamilton, G.; Hindle, A. K.; Huey, B.; Kimura, K.; Law, S.; Myambo, K.; Palmer, J.; Ylstra, B.; Yue, J. P.; Gray, J. W.; Jain, A. N.; Pinkel, D.; Albertson, D. G. Nat. Genet. 2001, 29, 263–264.
Analytical Chemistry, Vol. 80, No. 19, October 1, 2008
7641
the printers shown in Figure S1 (Supporting Information), ∼32 microtiter plates containing ∼3000 wells (for 96-well plates) could be mounted on the platens in the printing area. Printing one spot in each well would be within the 0.2-µL capacity of the capillary pins of the dimensions described in this paper (depending on the degree or wetting of the printing surface by the solutions). The printing rate would be between ∼3 and ∼6 spots/s, depending on which of the two printers were used, or ∼7-15 min per print cycle. This print time would not be problematic since the solutions in the capillaries are not subject to evaporation. The forces at the printing tips are sufficiently low to allow printing on soft surfaces such as nitrocellulose. Soft porous surfaces typically draw more solution from the pin during contact, reducing the number of spots that can be made for a given volume of solution. Therefore, this system may be ideal for making modest numbers of spots on
7642
Analytical Chemistry, Vol. 80, No. 19, October 1, 2008
absorbent substrates. Even if a print run does not use the full pin capacity, the pickup volume of the pins is smaller than most pins of other designs, so that smaller amounts of printing solutions will be wasted. ACKNOWLEDGMENT Work supported by the National Cancer Institute grant CA83040, and Vysis Inc. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review May 21, 2008. Accepted July 1, 2008. AC8010395
