Nucleic Acids Research, 2003, Vol. 31, No. 16 e87 ~
`D01: 10. 1 093/nar/gng0'86
`
`impact of surface chemistry and blocking strategies
`on DNA microarrays
`Scott Taylor‘, Stephanie Smith’, Brad Windlez and Anthony Guiseppi-Elie‘?-*
`
`.
`
`‘Center for Bioelectronics, Blosensors and Biochips (C38), 2Department of Medicinal-Chemistry and 3Department
`of Chemical Engineering, Virginia Commonwealth University, PO Box 843038, 601 West Main Street, Richmond,
`VA 23284-3038, USA
`‘
`
`Received March 24, 2003; Revised May 15. 2003; Accepted June 2. 2003
`
`ABSTRACT
`
`The surfaces and immobilization chemistries of
`
`'
`
`DNA microarrays are the foundation for high quality
`gene expression data. Four surface modification
`chemistries, poly-L-lysine-(PLL), 3-glycidoxypropyl-
`trimethoxysilane
`(GPS),
`DAB-AM-poly(propyl-
`eminime. hexadecaamine) dendrimer (DAB) and-3-
`aminopropyltrimethoxysilane (APS), were evaluated)
`using cDNA and oligonucleotide, sub~arrays.v Two
`RCA—cleaned
`and
`. un-silanized glass
`surfaces,
`immersed in Tris-EDTA buffer were also studied.
`
`DNA on amine-modified ‘surfaces was fixed by UV
`(90 rnJ/cm”), while DNA on‘GPS-modified surfaces
`was immobilized by covalent coupling. Arrays were
`blocked with either succinic anhydride (SA), bovine
`serum albumin (BSA) or left unblocked prior to
`hybridization. with labeled PCR product. Quality fac-
`tors evaluated were surface affinity for cDNA versus"
`oligonucleotides, spot and background intensity,
`spotting concentration and blocking chemistry.
`Contact angle measurements and atomic force
`microscopy were preformed to characterize surface
`wettability and morphology. The GPS surface exhib-
`_ited the lowest.background intensity regardless of
`blocking method. Blocking the arrays did not affect
`raw spot intensity, but affected background inten-
`sity on ‘amine surfaces, BSA blocking being the
`lowest. Oligonucieotides and cDNA on unblocked
`GPS-modified slides gave the best signal (spot-to-
`background intensity ratio). -Under the conditions
`evaluated, the unblocked GPS surface along with
`amine covalent coupling was the most appropriate
`for both cDNA and oligonucleotide microarrays.
`
`INTRODUCTION
`
`genetic complexity with potentially many more applications
`(1-4). Although production capabilities and use of micro-
`arrays are becoming increasingly well established, significant
`differences exist with regard to fabrication techniques and end
`userprotocols. Such differences make it difficult to compare
`results across platforms and present data management chal-
`lenges for the integration of databases. Fabrication parameters
`that may vary include: surface chemistry of slides (5-9), type
`and length of printed DNA (2,9) and immobilization or fixing
`strategies for the spotted DNA. Various end user protocols
`include: pre-hybridization surface blocking (3), mRNA label-
`ing protocols, hybridization protocols, post-hybridization
`wash stringency and data analysis techniques (4,10,l 1). An
`‘additional area of great concern is
`the implementation
`(placement and type) of appropriate controls aimed at quality
`assurance and quality control. The absence of approaches that
`are based on ‘best practices’ for design, fabrication, and end
`use of microarrays makes comparative data analysis between
`groups problematic. Although some work has been recently
`published that addresses several of these issues, (2-7,9—l3) '
`there is still little consensus about which design features and
`end user protocols are optimum for highest quality, microarray,
`data. In a recent attempt to develop microarray standards, the
`authors of the IVIIAME (minimum information about a
`microarray experiment) protocol have introduced guidelines
`for establishing standards conceming the information require-
`ments for a more effective comparative analysis of microarray
`data between groups ( 10).‘ The emphasis on these guidelines is
`however on documentation and not on engineering guidance.
`This paper aims at providing engineering guidance in the
`fabrication of cDNA and oligonucleotide microarrays.
`The glass surfaces of DNA microarrays have been modified
`in various ways to immobilize DNA (oligonucleotides and/or
`cDNA) (5-9). Common surface modifications for printing and
`affixing DNA onto glass slides are: poly-L-lysine (PLL) (14),
`3—aminopropyltrimethoxysilane (APS) (3,5,9), 3-_glycidoxy-
`propyltrimethoxysilane (GPS)
`(7,9) and aldehyde or car-
`boxylic acid (5). DNA has also been directly printed onto '
`unmodified glass (9). Amine-terminated cDNA and amine-
`terrninated oligonucleotides may be covalently coupled to
`' The DNA microarray enables researchers to survey the entire
`epoxide, isothiocyanate and aldehyde activated glass surfaces
`transcriptome of virtually any cell population. This capability
`(7). Non-terminated DNA has also been spotted onto amine-
`produces unprecedented quantities of raw data and enables the
`functionalized surfaces such as PLL, APS and surfaces that
`investigation of gene expression, functional gcnomics and
`
`
`‘To whom correspondence should be addressed at Center for Bioclectronics. Bioscnsors and Biochips (C3B), Virginia Commonwealth University, PO Box
`843038, 601 West Main Street, Richmond, VA 23284-3038, USA. Tel: +1 804 827 7016; Fax: +1 804 827 7029; Email: guiseppi@vcu.cdu
`
`‘ Page 1 of 19
`
`BD EXHIBIT 1017
`
`

`
`
`
`
`
`e87 Nucleic Acids Research, 2003, Vol. 31, No. 16
`
`PAGE 2 OF 19
`
`'
`
`were functionalized and derivatized with polyamidoamine
`dendrimer (PAMAM) (6).
`'
`‘
`One possible advantage of GPS, APS and PAMAM over
`PLL is that the former are covalently immobilizedyto the
`silicon bearing hydroxide functional groups on the surface of
`glass, while PLL is immobilized by adsorption, the result of
`acid—base
`interactions
`and hydrogen bonding with the
`amphoteric glass surface (15). Moreover, it has been reported
`that aminosilanes and PAMAM surfaces offer- a more
`consistent surface than PLL, with lower background and
`higher overall fluorescent signal intensities,(6). Given that
`there are ~5.0 silanol groups/nmz on a fully hydroxylated
`silica surface that is supplemented by a few layers of surface
`bound water, and given that the APS ‘molecule could pack to a
`limit of ~5 molecules/nmz (perfect hydrocarbon chain pack-
`ing, e.g. c-axis of polyethelene crystals packs at ~52-5.4),
`then it is likely that a well-packed APS layer would typically
`present in the range 3.5-4.0 amine groups/nmz (l6,l7), while
`PAMAM derivatized surfaces present ~66 amines/nmz (18). In
`addition, PLL surfaces generally require an induction period
`of ~2 weeks before they .can produce consistent microarray
`results (3). PLL, APS and PAMAM all present amine
`functional groups suitable for interaction with DNA via
`hydrogen bonding and, potentially, via electrostatic inter-
`actions (9) under the appropriate pH conditions. DNA is
`commonly ‘cross-linked’ on these surfaces by exposure to UV
`light, however
`this process is poorly understood but
`is
`believed to involve the creation of radicals’ that induce inter-
`chain cross-linking. GPS, in contrast, allows amine-terminated
`DNA to be covalently immobilized to the surface (19) via an
`amine-initiated nucleophilic ring opening reaction that leads
`to covalent bond formation between the GPS and the amine-
`terminated DNA.
`
`Blocking reactions are typically employed to prevent
`labeled reverse transcription product from adsorbing ‘to the
`surface of the printed microarrayduring the hybridization
`reaction. Blocking methods provide the added advantage of
`washing away unbound DNA from the surface that would
`.otherwise compete with the labeled species (3). Two of the
`most _common blocking methods to address non—specific
`adsorption on amine-modified microarrays involve blocking
`with succinic anhydride (SA) (3,141) or bovine serum albumin
`(BSA) (3). Both are intended to block the unreacted functional
`groups of the printed microarray with chemistries that have
`low affinity for DNA.
`-
`In this paper, "we report an evaluation of spotting concen-
`tration, surface chemistries and blocking strategies for their
`combined role in the performance of oligonucleotide and
`CDNA microarrays. Our goal was to establish optimum
`protocols
`for manufacturing,
`spotting, hybridization and
`scanning of mic-roarrays. cDNA.and oligonucleotide micro-
`arrays were therefore spotted on six different surfaces. These
`surfaces evaluated were: APS, GPS, DAB-AM—l6-poly(pro-
`pyleminime hexadecaamine) (DAB), and PLL. DAB is a
`generation 3 dendrimer that was linked tothe glass surface via
`covalent coupling following surface modification with GPS.
`In addition, two unmodified blank slides: (i) RCA-cleaned, but
`not surface modified (RCA); and (ii) cleaned and immersed in
`Tris—EDTA buffer (TEB) were also evaluated. Microarrays
`were blocked with either SA (SA—blocked), BSA (BSA-_
`blocked) or left unblocked. These surfaces represent a broad
`
`range of available surface chemistries. The GPS presents the
`reactive glycidoxy functional group to which amine-termin-
`ated oligonucleotides and cDNA, derived from amine-termin-
`ated primers, could be covalently affixed. The APS, PLL and
`DAB. surfaces present varying densities of amine functional-
`ities for hydrogen-bonding interactions with DNA. The RCA-
`cleaned glass slides served as a reference surface while the
`TEB immersion deliberately introduced surface contamin-I
`ation to otherwise cleaned glass slide surfaces. The non-
`blocked surface served as the control for blocking. These _
`surfaces and blocking strategies were evaluated by fabricating
`tnicroarrays of CDNA and 30mer oligonuclotides prepared
`using the human GAPDH gene sequence. The oligonucleo-
`tides and cDNA were spotted at five different concentrations
`and hybridized to Alexaflour 553-labeled GAPDH PCR
`product. Wettability of the surfaces was determined by contact
`angle measurements with hexadecane and ultrapure water.
`Surface morphology was characterized by atomic force
`microscopy (AFM).
`-
`
`MATERIALS AND METHODS
`Cleaning, preparation and surface modification of
`mieroarray slides
`
`In a class 1000 clean room, 50 VWR brand glass microscope
`slides (VWR 48300-025) were solvent cleaned by immersion
`for 1 min in boiling acetone followed by 1 min in boiling
`isopropanol. The slides were then washed in ultrapure H20 (18
`MOhm) for 1 min and dried with filtered nitrogen. Next, the
`slides were UV/ozonated for l5 min on one side using a
`Boekel UV Clean Model 135500 followed by ultrasonication
`in a Branson 1510 ultrasonicator in isopropanol for 5 min. The
`slides were then washed in diH2O and dried using filtered
`nitrogen. Finally, the slides were activated by immersion in a
`(5:l:])
`solution of diH2O:hydrogen peroxidezammonium
`hydroxide (RCA) at 60°C for l min, followed by dil-{Z0
`wash, placed in glass slide caniers and dried in a convection
`oven for 30 min at 80°C. After this step, RCA-cleaned slides
`were stored for subsequent spotting.
`The cleaned slides were then partitioned into six groups.
`One group of nine slides was modified by immersion in a
`solution of 'y-APS 0.1% v/v in anhydrous toluene for 30 min at
`40°C, washed three times in anhydrous toluene, placed in a
`glass staining dish and cured in a convection oven for 20 min
`at l 10°C. The slides were then stored until needed for printing.
`Twenty-four slides were chemically modified by immersion in
`a solution of GPS 0.1% v/v in anhydrous toluene for 30 min at
`40°C, washed three times in anhydrous toluene, placed in a
`glass staining dish and cured in a convection oven for 20 min
`at l 10°C. Nine of these slides were stored for printing, and the
`remaining slides were subsequently modified by immersion in
`a solution of DAB l.O% v/v in absolute ethanol overnight at
`room temperature. After the overnight incubation, the slides
`were washed three times in ethanol, placed in a glass staining‘
`dish and cured in a convection oven for 20 min at 110°C. The
`nine remaining slides were immersed in TEB (1.0 M Tris,
`0.1 M EDTA) for 30 min at room temperature, washed in
`dil-I20, dried in a convection oven and stored. Nine slides were
`modified with PLL. The slides were immersed in a solution of
`70 ml phosphate-buffered saline, 70 ml of 0.1% PLL and
`
`Page 2 ofl9
`
`

`
`
`
`
`
`PAGE 3 or 19
`
`Nucleic Acids Research, 2003, V01. 31, N0. 16 e87
`
`560 ml of dil-I20, then incubated with gentle shaking for 1 h at
`room temperature. The slides were then washed five times in
`dil-I20, dried with filtered nitrogen and placed in a 55°C
`vacuum oven for 10 min. All slides were stored in a plastic
`microscope box wrapped in aluminum foil then placed in a
`desiccator cabinet until needed for spotting. The PLL-
`modified slides were stored for 1 week prior to microarray
`spotting.
`'
`~
`
`Contact angle and AFM measurements
`
`Contact angles of de-ionized;water (yg = y,_P + 7;)‘ = 53 + 20 =
`73 mN m“) and anhydrous hexadecane (3/L = 7;)’ = 26 mN m“)
`were measured at
`the cleaned or chemically modified
`microscope glass
`slides using an NRL Contact Angle
`Goniometer
`(Ramé-Hart
`Inc., Mountain Lakes, NJ).
`Octadecyltncholorsilane (OTS) was used as
`a, reference
`surface and was prepared following solvent cleaning ‘by
`immersion in 0.l% v/v OTS in anhydrous toluene at 40°C for
`30 min. The slides were then rinsed three times with toluene
`and dried at 110°C for 20 min. In a contact angle measure-
`ment, a droplet (~15. pl) of probe‘ solvent was placed on the
`cleaned or modified glass slide from a fixed height, and the
`contact angle was directly measured through the focusing lens
`of the goniometer. AFM was performed using a Digital
`Instruments Dimension 3100 Atomic Force Microscope. Scan
`rates were set between 5 and 8 Hz depending on the image
`quality, and the scan size was changed from 1 ‘to 10 ttm upon
`engagement of the cantilever. The instrument was operated in
`tapping mode to obtain the‘ micrographs. The resulting height
`images were processed using Nanoscope HT software. Images
`were flattened to. remove scan lines, and the height scale was
`set
`to 75 nm. Feedback controls such as-
`integral gain,
`proportional gain and amplitude set point were modulated in
`real
`time’ as the image was being generated. Integral and
`proportional gain were always set between 2 and 0.5.
`
`Preparation of GAPDH CDNA for arraying
`
`The glyceraldehyde-3~phosph_ate dehydrogenase (GAPDH)
`gene fragment obtainedfrom PCR was a source of CDNA for
`arraying onto the slides prepared in the previous step. Amine-
`modified PCR primers:
`forward: 5' amine-C6—ccacccatgg-
`caaattccatggcaccgtca and reverse: 5'
`amine‘-C6-ggtttttcta—
`gacggcaggtcaggtccacc, were diluted to a working con-
`centration of 0.001 ltg/ttl and 10 ul was then mixed with
`0.5- it] (5000 U/til) of New England Biolabs (NEB) Taq
`polymerase (M02678), 0.1 pl (200 mM) dNTPs (Invitrogen
`10216-012, 014, 016, 018), 5 ul of 10X NEB PCR buffer,
`0.5 ul of GAPDH template and 34 ul of diH2O per 50 ttl
`reaction for a total of 50' reactions. The reaction was initiated
`at 95°C for 30 s and cycled 29 times under the following
`conditions: melt at 95°C for 30 s, anneal at 50°C for 30 s and
`
`extend at 72°C for l min- using an MJ Research PTC-200
`‘thennal cycler. After PCR,
`the reaction products were
`combined and distributed into three 1.7 til centrifuge tubes.
`To each tube was added 750 pl of 100% ice-"cold isopropanol
`a_nd'the tubes were centrifuged at 14 000 rpm. for 30 min in
`an Eppendorf Model 5804R centrifuge to pelletize the PCR
`product. The pellet was washed in 75% ethanol and re-pelleted
`by centrifugation at
`14 000 r.p.m.
`for 30 min. After
`centrifugationthe pellet was re~suspended in 20 ttl diH2O
`per tube and the contents of each tube were combined. The
`concentration of GAPDH in solution was quantified by UV_
`spectroscopy with a Perkin Elmer Lambda 40 spectrometer.
`The GAPDH cDNA was diluted to the concentrations of 2.0,
`1.0, 0.5, 0.2, 0.02 and 0.002 ttg/ttl. An. equal volume of 2X
`spotting buffer (3 M Betaine, 6X SSC) was added to each of
`the dilutions to make the 1X spotting solution. The solutions
`were then distributed into separate 96 well V bottom micotiter
`plates using a Packard Biochip MultiProbeI_I- Liquid Handling
`robot. The plates were stored at -20°C until needed for
`spotting.
`
`Preparation of oligonucleotides for arraying
`
`‘Oligonucleotide primers were designed using the GAPDH
`sequence (accession no. NM__002046) and synthesized by
`Integrated DNA Technologies. Table 1 lists the oligonucle-
`otides, their 5’ modification and their position in the GAPDH
`sequence. The forward,
`interior and random primers were
`diluted to the 2X concentrations: 2.0, 1.0, 0.5, 0.2, 0.02 and
`0.002 pg/ttl in diH2O and mixed with an equal volume of 2X
`spotting buffer (3 M betaine, 6X SSC). The forward, interior
`and random primers were arrayed on each type of chemically
`modified glass slide as well as onto the two groups of
`unmodified slides (RCA-cleaned and buffer immersed).
`
`Probe immobilization
`
`a Cartesian
`using
`performed
`Array fabrication was
`Technologies PixSys SSOOSQ Pin Array Robot-. and Liquid
`Dispensing System. Forward,
`interior and the
`random
`oligonucleotide sequences were spotted in three sub—arrays
`on slides that were modified with GPS, APS, DAB, PLL and
`the unmodified slides (RCA-cleaned and buffer immersed).
`PCR amplified GAPDH cDNA was also spotted on these
`slides in three additional but separate sub-arrays. The DNA
`arrayed on these surfaces was spotted in graded concentrations
`using the betaine spotting solution. The final DNA microarray
`layoutis shown in Figure 1. After spotting, the APS, DAB,
`PLL, RCA and buffer immersed arrays were cross-linked with
`90 ml/cmz in an Ultra-Violet Products CL-l000 UV cross-
`linker and baked at 80°C for
`l.5 h. The GPS arrays were
`incubated at 42°C in 50% humidity for 8 h, rinsed with 0.2%
`SDS solution for 2 min by vigorous shaking, washed three
`
`Table 1. Oligonucleotidc sequence information
`
`
`
`
`Oligo name Sequence Position Modification
`
`
`
`
`
`Forward
`Reverse
`Interior
`Unlabeled competitor
`Randomcr
`
`228-258
`802——8l l
`502—53l
`Complement of interior
`None
`
`Amine
`Amine
`Amine
`None
`Amine
`
`ccacccatgg caaattccat ggcaccgtca
`ggtttttcta gacggcaggt caggtccacc
`cagcctcaag atcatcagca atgcctcctg
`caggaggcat tgctgatgat cttgaggctg
`acctggacct gaatccgcca tatagcctac
`
`Page 3 of 19
`
`

`
`
`
`
`
`e87 ‘Nucleic Acids Research, 2003, Vol. 31, No. 16
`
`PAGE 4 OF-19 _
`
`
`§§§3{°m°‘°3°“5
`/' a’ o oofio c;>:o o oo_._
`IO 0 o__o.o .0 0.09
`5o0ng‘/pt
`A
`_
`K.’
`
`_o .0 0:0. 0 .o--o- o-ovo
`’250ng/N"
`LDNA Subgrrds
`_'
`
`a o-o-.0-o 0.0 o.o__o _
`10'”-mu '
`'
`'
`_‘
`
`o o o_o_.o-o',o-o:o-.'o
`,,,,£,),p,
`
`° 0'0'.0f’0.0.0 .0 .0.’-oi
`0.lng/pl
`
`
`
`
`
`Q/gligo Subgrids
`’
`'1'
`'
`Probe concentrations Microarray Layout
`
`
`
`"‘-
`
`Random‘,
`Interior
`1
`Eorjward
`
`
`
`Ohgo
`Chg‘)
` ooooooooq
`
`oooooeeca
`
`
`oooooqooo
`
`
`
`‘\
`oo.o—-ooooo,-o‘
`oooooooo‘
`
`O o.-O‘ 0 Of! (0.0
`
`
`
`Ing/pt
`o.1,gg/“1
`
`Figure 1. Microarray layout.
`
`times in diHzO, incubated in .diH?,»,O at 50°C for 20min then
`‘dried with filtered nitrogen. All arrays .were then stored in foil-.
`wrapped slide-boxes in a desiccator cabinet overnight prior ‘to
`hybridization‘.
`
`Labeling of GAPDH'target
`
`The forward and reverse oligonucleotide primers were used to
`. amplify a 600 bp region of the GAPDH gene for fluorophore
`labeling. The previously described‘ PCR protocol was used
`except that aminoallyl dUTP (Molecular Probes A—2l664) was
`included in the reaction mixture at a ratio of 3:1 dUTP:TTP for
`a final concentration of 200 mM in each 80 til reaction for a
`total of 60 reactions. The resulting PCR product was labeled
`using the AR-BS7“ DNA labeling kit from Molecular Probes
`(A—2l665) according to the supplied protocol.
`Pre-hybridization. blocking
`
`Twelve slides were immersed in pre-hybridization buffer.
`containing 5X SSC, 0.1% SDS and 1.0% BSA, incubated at
`42°C for 45 min, washed 5 X in diH2O then dried using filtered -
`nitrogen. Another 12 slides were immersed in SA pre-
`hybridization solu'tion containing l5 ml sodium borate and
`6 g SA in 350 ml 1-methyl-2-pyrrolidinone. The solution
`containing the slides was incubated on an orbital shaker for
`20 min, quenched in boiling dil-I20’, washed five times in 95%
`ethanol and dried using filtered nitrogen. Twelve slides were
`left unblocked. The remaining slides in the GPS and RCA
`groups wereprocessed separately according to the same
`protocol.
`
`Hybridization and imaging
`
`Each group of slides was hybridized using a GenTac
`Hybridization Station (Genomic Solutions).
`100 pl of
`hybridization buffer [4X SSC, 1>< Denhardt’s reagent, 5.0%
`SDS, 10% dextran sulfate, 40% formamide solution (50% v/v
`dil~l2O)] containing 40 ng labeled GAPDH CDNA and, for
`some experiments, 24 ng unlabeled competitor, was added to
`each rnicroarray hybridization solution. The hybridization was
`allowed to proceed for 16 h at 42°C. After hybridization, the
`arrays were sequentially washed with medium stringency
`buffer (2>< SSC, 0.1% SDS) (Genomic Solutions 16004001),
`
`high stringency buffer» (0.lX SSC, 0.05% SDS) (Genomic
`Solutions 16004501), post wash buffer (0.1X SSC) (Genomic
`Solutions 16003501) and dil-I20. The arrays were then dried
`with filtered nitrogen. Each microarray was scanned at 5 pm
`resolution using a Perkin Elmer SeanA1ray 5000 microarray
`scanner using the 488 nm filter.
`
`RESULTS
`
`Surface chemistry and blocking strategy
`
`Four chemically modified and two unmodified glass surfaces
`were studied for their characteristics relating to:
`(i)
`iin-.
`mobilization of CDNA and» oligonucleotides,
`(ii) resulting
`slide background intensity after hybridization‘,
`(iii) signal
`intensity (spot intensity/slide background intensity) following
`hybridization and (iv)
`spotting uniformity. The surface
`chemistries evaluated were y-APS, GPS, DAB (linked to the
`glass surface via GPS), PLL, a cleaned glass surface that had
`-been immersed in TEB and a RCA-cleaned surface. These
`surfaces were selected because they are commonly used or
`otherwise cost effective/easy to implement in the microarray
`fabrication laboratory. While there are several alternative
`attachment chemistries (5,7), we limited this study to the most
`widely used and well-documented examples. Most cDNA
`microarray fabrication has been reported using PLL surfaces
`(2,3,l4,l5). However, Hegde at al. (3) and Liu et a1. (20) have
`used APS surfaces for their CDNA microarray work and APS-
`modified glass surfaces are commercially available from
`Coming [CMT—GAPS slides (catalog no. 40004, Coming)]
`and Telechem [Super Amine slides (catalog no. SMM)] (web
`addresses for microarray substrates: Corning: http://www.
`coming.com/LifeSciences/pdf/gaps_ii_coated~slides_10_0l_
`ss_,cmt,gaps_0O2.pdf
`and Telechemz
`http://arrayit.com/
`Products/Substrates/substrates.html).
`_
`In an effort to identify a better microarray surface, one
`group has examined the amine presenting compound,
`PAMAM (6), and found it to have superior background and
`oligonucleotide capturing characteristics. We chose a closely
`related compound to that used by Benters et al.
`(6)
`for
`comparison with the common amine surfaces. As a means of
`
`Page4iofl9
`
`

`
`
`
`
`
`Nucleic Acids Research, 2003, Val. 31, N0. 16 e87
`
`(')‘ua‘
`'<'>'NB"
`HN‘iC”2)g5U93l-O-F-O-Sugar-o~;;.o.sqga,.oH
`0
`o
`L
`I
`ase
`ass
`Base
`‘f-"3,
`<lTNa.'
`HN'(CH2)§5“93l'O-$3-O-Sugar-O-ii-o.5ugag.o”
`«
`0
`I
`o
`Base
`
`Base
`
`. Base
`
`PAGESOF 19
`
`I A
`
`en».
`9H2
`H(’:-on Hc‘:-0H
`(EH2.
`(EH2
`9
`9
`?“2
`‘.’”2
`‘.’”2
`?”2
`9“:
`?“2
`-O-‘Sl5I-—O—-?i-O-
`0
`O
`zz
`
`\2?s/ as
`as
`V \>N/
`Non )3
`(cH,)3
`K,./
`I
`(<I:H2)2 .,
`R
`
`I
`(CH )3
`
`I
`(5,425
`
`(ca )3
`
`(CH2)3
`
`.
`
`.
`
`Figure 2. Schematic illustration of the various surface chcmisuics studied and the idealized interaction of DNA with functional groups on a glass surface.
`(A) GPS covalently bound to an aminc—terminated oligonuclcotidc[ (B) PLL hydrogen bonding with an oligonucleotidc. (C) One-half of a DAB dendrimer
`hydrogen bonding with an oligonuclcotide. (D) APS hydrogenbonding with an oligonucleotidc.
`'
`
`covalent coupling, it has been reported that epoxy-silane
`(GPS) has been used for
`immobilizing amine«temiinated
`oligonucleotides and CDNA (5,2l). Figure 2 is a schematic
`illustration of the various surfaces studied.
`The pre-hybridization blocking strategies studied were: no
`blocking, the adsorption of BSA and the reaction of SA. The
`ability of each of these three blocking strategies to" reduce
`post~hybridization background intensity was investigated for
`each of the six surfaces. SA is commonly used as a blocking
`reagent in CDNA microarrays prepared on amine-functiona-
`lized surfaces (3,i3). The anhydride readily reacts with the
`available amines forming the amide and thereby eliminating
`the amine from the surface with the intent of avoiding non-
`" specific adsorption of DNA. Such an approach should be
`effective for both oligonucleotide and CDNA microarrays.
`A blocking solution containing BSA has been reported
`
`by Hegde at al. (3) to result in lower background intensities
`when compared with SA. BSA is a neutral globular protein
`that readily adsorbs to surfaces and is commonly used in
`ELISAS.
`'
`
`There are two microarray platfonns in wide usage: cDNA
`and oligonucleotide arrays. The oligonucleotide arrays vary in
`oligonucleotide length but are generally 2S—~70mers while
`printed cDNA typically ranges from 70 to 600 bp. Both types
`were evaluated in this study. The oligonucleotides selected
`were 30mers of the GAPDH gene and the CDNA was an
`-600 bp PCR product amplified from GAPDH using amine-
`. terminated primers. Both types of DNA" were spotted over a
`broad range of concentration (0.001—0.5 ttg/til).
`We measured spot quality as a function of spot and
`background intensities. All intensities were measured under
`the same conditions of laser power and PMT gain. Images
`
`Page 5 of 19
`
`('3-Nair
`j ?‘ Na’
`B
`H2N'(CH2)§5“9al'0‘F'0'5“931'?O-E‘-O-Sugar-OH
`O -
`I
`0
`l3a.se
`
`lI
`
`Base
`I
`
`Base
`
`7
`
`NH2
`.
`'
`I
`,9
`""2
`H(CH2)
`l
`4
`.
`I'”2
`gtgtuzid
`\c/N\(l:H
`I’
`NH
`I 2
`H(cI'I2)4
`II
`?\c/ \c”
`I‘“2
`0
`’
`\C/N\éH , g(<;H2>4
`u
`n((l:H2I4
`0
`\c/ \cH
`II
`\c/ ‘CH
`0
`II
`\
`0
`
`°
`
`O'Na‘
`?_ N34‘
`i
`-D
`H2N'(CH2)5'S"93"0'F’0‘5“93T'O-IE’-O-Sugar-OH
`
`0 _
`
`o
`
`'
`
`base
`
`base
`
`Base
`
`’.°“2
`’I’“2
`‘Ella
`‘EH2
`‘.’“2
`‘.’“2
`_?_“2.
`‘E“2
`-O-§l—Q-§l-O-
`C
`
`O
`
`('3' Na’
`‘I5- Na‘
`,
`_
`C
`”2”‘(CH2)6'5“Qaf-O-}i“-O-Sugar-o-I:>—o_suga,-.0“
`o
`I
`ase
`Base
`3353
`'.
`t
`r
`\
`I
`'\
`'f“2
`Ill";
`t'IH2
`If“:
`C
`(H)(CI-I)
`(CH)‘(CH)
`
`'
`,'
`,'
`NH2
`
`NH2
`I
`
`.
`
`;
`
`NH2
`I
`
`NH2
`I
`
`)
`
`

`
`e87 Nucleic Acids Research, 2003, Vol. 31, N0. 15 .
`
`PAGE 6 OF 19
`
`Ala‘
`
`'
`la RcA'cD'NA
`
`U_nbloc_ked__A_\{e_rage Baekgmqnd _lntensi_t_ie§
`~‘l
`1"
`
`
` I;
`
`
`RCACDNA_.
`
`
`
`opsOligo
`
`Surfaeejchemlstry
`
`Unblocked A\(erage»Back'g'rou_nd. Intensities‘
`ii! .GP.S'Ollgo
`-2:1 Rc_A.cDNA.'
`_ reaougo
`I;;GPS-.cDNA
`- AF’.S:0llgo'
`I;.TEB‘{§DNA
`13 PLL'_.Ol_igo
`an APS:~'r_;D_NA
`‘I‘.DAB.0lIgo
`EVPLL CDNA
`I DAB_.'cDNA
`
`
`
`
`
`‘
`
`
`
`
`
`Concentratlon
`
`were subsequently scanned at the same resolution (5 microns).
`Our findings are presented according to the blocking strategy
`employed.
`
`Background intensities
`
`the average background intensities
`Figure 3A—C shows
`following hybridization to target
`for all
`the CDNA and
`oligonucleotide sectors of all six chemically modified sur-
`faces. Background intensities were measured for each of the
`
`spotting concentrations '(0.001—O.5 pg/pl)‘
`six different
`employed and averaged over the many replicates for that
`concentration. We chose this approach to allow us to discem_
`the influence of spotting concentration, and hence spot
`intensity, on the intensity of
`the background signal as
`perceived by the QuantArray software. All intensities were
`measured using the same QuantArray parameters and were
`plotted on the same scale to allow ready visual comparison of
`the data. Figure 3A shows the background intensity of.
`
`Page6of‘19
`
`
`
`
`
`51 TE8 Ollgo
`- I TEB cD
`
`I APS Ollgof
`5! APS cDNA
`:1 PLL _OIigo
`EIPLL cDNA
`
`in DA8 ongo
`
`IDAB com,
`Concentratlon
`
`
`
`

`
`
`
`
`
`PAGE 7 on 19
`
`Nucleic Acids Research, 2003, Vol.31, No.16 687 i
`
`Ba
`
`’
`
`BSAvAverage B'ackgroun‘d intensities. ‘
`
`:~‘l‘¢f_
`
`_ lTREA CDNA
`I mops Oligo
`IGPS cDNA
`
`
`
`I TEB.Oligo
`El Tee cDNA
`in APS Oligo
`:2 APS VCDNA
`PLL coNA"
`D PLL ongo.
`2: DAB Oll_g_o
`- DAB com
`
`Concentration
`
`Intensity
`
`4
`3“;
`
`(vdc
`’
`..~.
`A’
`'°<_cp<§v§Z-ad§-‘§
`2..
`§9E60o0%°°%
`C30‘-’m°c/J°..:§"nm
`06/>mu:{3o.&’.~:og<
`'z§%$'*,r-<<°'_—J2
`0
`ac
`*
`o.
`
`Surface Chemistry
`
` '
`
`1..
`“_
`
`‘
`
`“Q”:
`
`‘ab
`
`lzntensltyj,
`
`tensities
`‘:2
`
`,-
`
`’
`1;}. RCA CDNA
`’ i G_Psjcor_u_A
`"ta-T_EB,‘c'DNA.
`.£1'AP3..CDNA‘
`ta PLL ongo-
`
`'l..GPS-’O|igoi
`_| Tea-:ongo_'.
`' n;APS5jQllgo§
`:3 PLL eDNA-
`I-DAB'O|igo:
`
`Concentration
`
`
`the
`unblocked slides, while Figure 3B and C shows
`background intensities of the BSA- and SA~blocked slides,
`respectively.
`V
`It can be seen in Figure 3A (unblocked) that the amine-
`bearing surfaces gave the highest background intensities
`(~4000 counts) when compared with the unmodified surfaces,
`RCA and TEB, and the epoxide-bearing surface.‘ Figure 3B
`(BSA~blocked) shows very similar behavior to the unblocked
`slides. That
`is,
`the amine~bearing surfaces gave higher
`
`background intensities when compared with the unmodified
`surfaces, RCA and T138, and the epoxide-bearing surface.
`However, in this case the background intensities are between
`1000 and 2000 counts, half as much as the unblocked slides.
`BSA therefore reduces the background intensity by ~50%
`compared with unblocked slides. It is noteworthy that this
`reduction in background intensity is most significant for the
`amine-bearing surfaces and does not significantly affect the
`background intensities of the unmodified surfaces, RCA and
`
`Page 7 of 19
`
`

`
`
`‘
`
`anca cDNA
`IGPS cDNA
`
`I-TEB. com
`13 APS-cDNA
`B1 PLLCDNA
`"I
`
`_ SA Average B
`IIGRS Oligo
`' E) TEB Oligo
`APS Oligo
`:22 pm. ongo
`I DAB Ollgo
`
`',.-*S-;’
`..
`
`E}
`
`
`
`Concentration
`
`Figure 3. (Previous two pages and above) Average background intensities following hybridization of all cDNA and oligonucieotide sectors at different
`spotting concentrations (0.001-1.0 ttgltzl) for all six surfaces studied. These are
`grouped by the blocking method employed; (A) unblocked, (B) BSA blocked
`and (C) SA blocked. (a) 3D bar charts of average background intensities as a function of spotting concentration and surface chemistry. (b) 2D bar charts
`showing the standard error.
`
`It can be seen in
`the epoxide-bearing surface.
`TEB, or
`Figure 3C (SA-blocked)
`that
`the amine-bearing surfaces
`likewise gave higher background intensities compared with
`the unmodified surfaces, RCA and TEB, and the epoxide-
`bearing surface. However, in the case of SA blocking, these
`background values were considerably higher than those found
`for
`the amine-bearing surfaces on unblocked and BSA-
`
`blocked slides. Here, background intensities ranged from
`3000 to 24000 counts. There is also clear variation in the
`behavior of oligonucleotide and cDNA spots when blocked
`with SA. Oligonucleotide sectors were less prone to high
`background intensity counts while cDNA sectors gave high
`counts. Close observation of the scanned images revealed
`sizable comet tails on the cDNA spots. These observations
`
`Page 8 of 19
`
`e87 Nucleic Acids Research, 2003, Vol.31, No. 16
`
`‘
`
`PAGE 8 OF 19
`
`ckground.lntensltIes
`' SA-Average.-"Ba
`'-r.\
` I 1!Sl‘r3'v:.:.%‘
`
`
`Ca
`
`E1-RCA con7x7'7
`IGPS Ogligo
`I GPS.cDNA
`I TEB Oligo
`B TEB cDNA
`
`I APSiOilgo
`E APS cDNA
`0 PLL cDNA
`
`a PLL Oligo
`Ir oAB.0ugo
`.I DAB cDNA
`...-,....-...,._.......
`
`Concentration
`
`Intensity
`
`RCAcDNA.Q cpsQligo
`
`
`
`TEBOllgo"
`
`APSOligo
`
`APScDNA
`
`PLLcDNA
`
`PLLOllgo
`
`DABcDNA
`
`Surface.IChern_i§u'y
`
`

`
`
`
`
`
`PAGE 9 OF 19
`
`Nucleic Acids. Research, 2003, Vol. 31, No.16 e87
`
`have been previously reported in microarray experiments
`using SA blocking
`(1920, Oregon State Microarray
`Laboratory: http://www.cgrb.orst.edu/CSL/custorn.pdf). SA
`appears to have a deleterious effect on UV-cross-linked
`cDNA— spots, inducing comet tail formation, co_mpromising the
`integrity of DNA spots.
`
`Spot intensities
`
`Oligonucleotide and CDNA sectors were spotted at concen-
`trations of 0.001, 0.01, 0.1, 0.25, 0.5 and 1.0 llg/til. Figure 4A
`and B shows the-resulting raw spot intensities obtained over
`these six concentrations and under the three blocking condi-
`tions studied. For CDNA sectors the plots display a fairly sharp
`rise to plateau between 0.25 and 1.0 ttg/ttl resulting in higher
`spot
`intensity'values
`than oligonucleotide sectors. The
`oligonucleotide plots did not exhibit a’ plateau, rather they
`displayed a constant gradual rise and a smaller andmore even
`slope. Although both types of DNA were spotted at the same
`concentration,
`the-raw spot
`intensities of oligonucleotide
`sectors were generally‘ lower than those of cDNA sectors
`(Fig. 4B versus A) over all surfaces studied. cDNA sectors
`displayed an —-2-8-fold higher raw intensity than oligonucle-
`otide sectors for any given concentration.
`The difference between cDNA and oligonucleotide spot
`intensities was especially apparent among the amine surfaces
`where spot intensities differed 8-fold. Oligonucleotides and
`cDNA exhibited close clustering at each. concent