1997 Oxford University Press

Nucleic Acids Research, 1997, Vol. 25, No. 22

44474454

Oligonucleotide dendrimers: synthesis and use as polylabelled DNA probes

M. S. Shchepinov*, I. A. Udalova1, A. J. Bridgman and E. M. Southern
Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK and 1Department of Paediatrics, Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, Oxford OX3 9DU, UK
Received September 4, 1997; Revised and Accepted October 1, 1997

ABSTRACT Oligonucleotide dendrimers were synthesized using a novel phosphoramidite synthon, tris-2,2,2-[3-(4,4-dimethoxytrityloxy) propyloxymethyl]ethyl-N,N-diisopropylaminocyanethoxy phosphoramidite. Label, incorporated using [-32P]ATP and polynucleotide kinase, was increased in proportion to the number of 5-ends. There was a similar increase in signal when these multiply labelled oligonucleotides were used as probes to oligonucleotide arrays. A dendrimeric oligonucleotide was used successfully as a primer in the PCR. The strand bearing the dendrimer was resistant to degradation by T7 Gene 6 exonuclease making it easy to convert the double-stranded product of the PCR to a multiply-labelled, single-stranded probe. INTRODUCTION Since their first synthesis (1), polyfunctional starburst polymers (cascades, silvanes, arboranes, dendrimers) have attracted considerable attention as a new branch of polymer science (25). Two basic strategies for the synthesis of these structures were proposed: divergent, with the structure grown up from the centre to the periphery (1); and convergent, growth of molecule from the periphery to the centre (6,7). One of the most important parameters governing a dendrimeric structure and its generation is the number of branches generated at each step; this determines the number of repetitive steps necessary to build up the desired molecule and the density of groups at the periphery. The main properties of the molecule are determined by the functional end groups or moieties on its outer shell. Many applications proposed for dendrimers exploit the high density and the large number of these groups. Dendrimers with a positively charged outer surface of globules interact strongly with nucleic acids, a property which was used recently for the transport of nucleic acids through the membranes of living cells (8,9). Dendrimers with internal cavities are capable of non-covalent interactions (topological trapping) with guest molecules to give guesthost systems. The inner space of these cavities can be spacious enough to accommodate relatively big molecules such as the dye Bengal Rose (10).

Atoms and groups with different valence which have been used as centres of branching include carbon (11), a three-substituted phenyl ring (12), ribo- (7) or branched nucleosides (13,14), nitrogen (15), phosphorus (16), etc. Of these, the quaternary carbon atom, which retards chemical reactivity due to neopentyl substitution pattern, leads to the quickest onset of dense packing in the reiterative process. The pentaerythrityl synthon containing this atom was used in carbohydrate chemistry to build up tri-antennary cluster polysaccharides (1719) as well as in a combinatorial strategy for the preparation of functionally diverse scaffolds (20). Branched (dendrimeric) oligonucleotides can be used to amplify radioactive or fluorescent signals in hybridisation tests (21). Such amplification may be particularly important in in situ hybridisation and in the emerging techniques which exploit oligonucleotide arrays (2224), where the signal is limited by the surface density of the oligonucleotides or the target molecule. Phosphitamide reagents have been described which double the amount of reactive 5-hydroxyl groups after each condensation step thus giving 2n reactive OH-groups after n condensations (2530); they have been used in combination with biotin, fluorescein, pyrene and other phosphitamide synthons for multiple 5-labelling of oligonucleotides. But all these compounds share a difficulty: too dense a concentration of reporter groups leads to self-quenching of fluorescence. Here we report the synthesis of a novel phosphoramidite, synthon 7, based on a pentaerythritol structure, which trebles the number of branches at each coupling. We also describe a method for reducing the density of packing of the dendrimer branches and a number of applications for the oligonucleotide dendrimers. Oligonucleotide dendrimers containing bunches of nine oligonucleotides were labelled with [-32P]ATP and polynucleotide kinase. We found that the efficiency of labelling strongly depends on the distance of the 5-hydroxyl group from the centre of branching. When used as probes to oligonucleotide arrays, the multiply labelled structures showed much higher sensitivity than mono-labelled counterparts. These structures were also shown to be compatible with the PCR conditions. The PCR products had reduced mobility in gel electrophoresis and were readily converted to single strands by T7 Gene 6 exonuclease.

*To whom correspondence should be addressed. Tel: +44 1865 275226; Fax: +44 1865 275283; Email: misha@bioch.ox.ac.uk

4448 Nucleic Acids Research, 1997, Vol. 25, No. 22 MATERIALS AND METHODS Oligonucleotides and oligonucleotide dendrimers used for hybridization were made in an Applied Biosystems 392 DNA/ RNA synthesizer using standard phosphoramidite chemistry (31). Dendrimers were synthesised on 1000 LCAA-CPG (Cruachem). 3-DMTr-5-phosphoramidites of base-protected nucleosides and aminated polypropylene were a gift from Beckman Instruments. Oligonucleotides and oligonucleotide dendrimers were labelled according to standard methods using radioisotopes purchased from Amersham International. Polynucleotide kinase was from NEB. UV-spectra were measured on a Spectronic 2000 spectrophotometer, Milton Roy Co., USA. MALDI-TOF massspectra were recorded on a VoyagerTM Elite BiospectrometryTM Research Station, PerSeptive Biosystems. 1H-NMR spectra were recorded on a Varian Gemini 200 200 MHz spectrometer. 31P-NMR spectra were recorded on a Brucker AC-500 500 MHz spectrometer. HPLC system was the Waters (Milford, MA, USA). PhosphorImages were obtained using a Molecular Dynamics PhosphorImager Model 400A. All chemicals were purchased from Aldrich Chemical Company. Phosphitylating reagent and oligonucleotide purification columns were from Sigma Chemical Company. Silica gel for column chromatography and solvents were purchased from BDH/Merck. Tris-2,2,2-[(cyanoethoxy)methyl]ethanol (2) Acrylonitrile (100 ml, 1.55 mol) was added to a stirred mixture of pentaerythrite (1, 68 g, 0.5 mol) and sodium hydroxide (2.5 g, 0.075 mol) in 75 ml of water at RT. The mixture was then stirred at 50_C overnight. The solution obtained was neutralized with 5% HCl and washed three times with 200 ml of ethyl acetate. The combined organic phase was evaporated in vacuo to give 120 g of colourless syrup which was purified by flash chromatography (chloroform) to give 94 g (63.7%) of 2 as a colourless syrup. Rf: 0.68 (methanol:methylene chloride, 1:9, detection of TLC in I2 vapour). (For tetrakis-product: Rf = 0.9). Calculated for C14H21N3O4: 295. Mass-spectrum, MALDI-TOF: 296.217 (MI + H+), 318.475 (MI + Na+), 334.474 (MI + K+). IR: 2240 cm1 (C5N). Calculated for the tetrakis-product C17H24N4O4: 348. Mass-spectrum, MALDI-TOF: 349.452 (MI + H+) Tris-2,2,2-{[(methoxycarbonyl)ethoxy]methyl}ethanol (3) A solution of 2 (15 g, 0.05 mol) in HCl-saturated methanol (200 ml) was refluxed for 2.5 h. To this mixture 10 ml of H2O and 5 ml conc. HCl were added and the mixture was stirred for 15 min. After addition of 400 ml H2O, the mixture was extracted with ethyl acetate (3 150 ml), combined organic fractions were washed with NaHCO3 sat. (4 200 ml), brine, dried over Na2SO4 and evaporated in vacuo to give 15 g (75%) of 3 as a colourless oil. Rf: 0.62 (methanol:chloroform, 1:30, detection of TLC in I2 vapour). IR: disappearance of the band at 2240 cm1 (C5N). Calculated for C17H30O10: 394. Mass-spectrum, MALDI-TOF: 394.972 (MI + H+), 416.934 (MI + Na+), 432.892 (MI + K+). Tert-butyldimethylsilyl-[tris-2,2,2-(5-oxy-2-oxa-pentyl) ethanol] (4) To the alcohol 3 (13.5 g, 34 mmol, coevaporated with dry pyridine) in 150 ml of anhydrous pyridine was added TBDMSCl (6 g, 40 mmol) and the mixture was stirred overnight at room temperature. Mass-spectrum (MALDI-TOF) of the reaction mixture showed complete conversion of initial alcohol into silylated product. Rf: 0.83 in methanol:chloroform, 1:30 and 0.91 in methanol:chloroform, 1:9, detection of TLC in I2 vapour. Calculated for C23H44O10Si: 509; found: 508.882 (MI), 530.864 (MI + Na+), 546.819 (MI + K+). The reaction mixture was evaporated to dryness, residue dissolved in 200 ml of ethyl acetate, washed with NaHCO3 sat. (4 200 ml), brine, dried over Na2SO4 and evaporated in vacuo to give a silylated triester as a colourless oil (17.3 g, 98%). 17 g (33 mmol) of this triester was dried in vacuo over P2O5, dissolved in 20 ml of anhydrous THF and added over 40 min to the stirred ice-cold suspension of LiAlH4 in 350 ml of anhydrous THF. The mixture was then stirred for another 45 min on ice bath and 1.5 h at room temperature. 45 ml of 15% solution of NaOH in water was added over 20 min. The residue was decanted and washed with THF (3 30 ml). Combined THF fractions were diluted with water (150 ml) and carefully neutralized with diluted HCl. The mixture was evaporated and purified by flash chromatography (methanol:chloroform, 1:30 to 1:6) to give 9.5 g (67%) of colorless oil. Rf: 0.49 (methanol:chloroform, 1:9, detection of TLC in I2 vapour). Calculated for C20H44O7Si: 425. Mass-spectrum, MALDI-TOF: 426.005 (MI + H+), 448.131 (MI + Na+). Tert-butyldimethylsilyl-{tris-2,2,2-[3-(4,4-dimethoxytrityloxy)propyloxymethyl]ethanol} (5) Triol 4 (2 g, 4.7 mmol) was dried by coevaporation with anhydrous pyridine and dissolved in 60 ml of it. DMTrCl (6 g, 17.75 mmol) was added, with stirring, over 2 h, after which time 750 mg of DMTrCl, a catalytic amount of DMAP and 1 ml of EDIP were added to the stirred mixture. In 5 h, 1 ml of water was added to the reaction mixture. It was stirred for another 10 min and evaporated to an oil, dissolved in 200 ml of ethyl acetate, washed with sat. NaHCO3 (4 200 ml), brine, dried over Na2SO4 and evaporated in vacuo. The resulting oil was purified by flash chromatography (hexane:dichloromethane, 1:3 to dichloromethane) to give 3.5 g (58%) of white solid. Rf: 0.75 (chloroform with traces of Et3N), 0.86 (hexane:ethylacetate, 3:1, traces of Et3N). Calculated for C83H98O13Si: 1331.6. Mass-spectrum, MALDI-TOF: 1355.62 (MI + Na+). 1H-NMR (CDCl3, , md): 7.56.7 (m, 39H, arom.), 3.73 (s, 18H, OCH3), 3.39 (t, 6H, DMTrOCH2), 3.23 (s, 6H, (OCH2)3C), 3.08 (t, 6H, CH2O), 1.79 (quin., 6H, CH2CH2CH2), 1.58 (s, 2H, CH2OTBDMS), 0.88 (s, 9H, tBu), 0.05 (s, 6H, CH3). Tris-2,2,2-[3-(4,4-dimethoxytrityloxy)propyloxymethyl] ethanol (6) Silylated compound 5 (2.25 g, 1.7 mmol) was dissolved in 40 ml of standard 1.0 M TBAF solution in THF and stirred for 6 h. Ethylacetate (150 ml) was added and the mixture was washed with sat. NaHCO3 (4 200 ml), brine, dried over Na2SO4, evaporated and the resulting oil was purified by flash chromatography (hexane:ethylacetate, 3:1 to 1:1, traces of Et3N) to give 1.8g (88%) of white solid. Rf: 0.4 (hexane:ethylacetate, 3:1, traces of Et3N). Calculated for C77H84O13: 1216.5. Mass-spectrum, MALDI-TOF: 1241.64 (MI + Na+). 1H-NMR (CDCl3, , md): 7.56.75 (m, 39H, arom.), 3.77 (s, 18H, OCH3), 3.45 (t, 6H, DMTrOCH2), 3.3 (s, 6H, (OCH2)3C), 3.1 (t, 6H, CH2O), 2.78 (br. s., 1H, OH), 1.8 (quin., 6H, CH2CH2CH2), 1.65 (br. s, 2H, CH2OH).

4449 Nucleic Acids Research, 1997, Vol. 25, No. Nucleic Acids Research, 1994, Vol. 22, No. 122 Tris-2,2,2-[3-(4,4-dimethoxytrityloxy)propyloxymethyl] ethyl-N,N-diisopropylaminocyan ethoxyphosphoramidite (7) The alcohol 6 was phosphitylated using chloro[diisopropylamino]--cyanoethoxyphosphine as previously described (32) and purified by flash chromatography (hexane:ethylacetate, 4:1, traces of Et3N) to give the title compound in 75% yield. Rf: 0.55 (hexane:ethylacetate, 3:1, traces of Et3N). Calculated for C86H101N2O14P: 1414.9. Mass-spectrum, MALDI-TOF: 1413.22 (MI - H+). 31P-NMR (CH3CN:CD3CN, 1:1, int. standard: 80% H3PO4, , m.d.): 151.312. O1-(4,4-dimethoxytrityl)-O16-(N,N-diisopropylaminocyanethoxyphosphinyl)-pentaethylene glycol (8) Synthesised according to protocol in ref. 33. Synthesis of oligonucleotides 11, 12, 14 and oligonucleotide dendrimers 9, 10 and 13 Oligonucleotide dendrimers 9, 10 and 13 were synthesised using synthons 7 and 8 (Scheme 2). To uniform the melting temperatures of oligonucleotides with different base composition when hybridising them to the same array, we used polypyrimidine sequences close in T-C contents for dendrimer oligonucleotides and a control, and polypurine sequences on the array, all being 15mers. To obtain compound 9, 15mer oligonucleotide 5-TCT TCT TCT TCT TTT was synthesised on 0.2 mol scale using 1000 CPG-support to the 5-end of which the spacer molecule based on structure 8 was attached. Starting from this step, we did not use the capping procedure. While still immobilised to CPG, this structure was detritylated and subjected to two condensations with trebler 7 (increased wait step), thus increasing the amount of 5-OH-groups for each molecule up to nine. Then the synthesis of pentathymidylates was carried out atop of these hydroxyl groups on 1 mol scale. The overall yield of the structure 9 was 80%, as determined by DMTr+-assay. Compound 10 was synthesised in the same way, but additional spacers 8 were introduced between OH-groups of treblers and pentathymidylates. The control 15mer single-stranded oligonucleotide was 5-TTT CTC TTT CTC TTC (11). To synthesise oligonucleotide dendrimer 13, oligonucleotide 5-GGT TTC TCT CTG ACT GCA TCT TGT CC (12, ref. 35) was synthesised on 0.4 mol scale using 1000 CPG-support. Further synthesis using one half of that CPG was carried out as described for 10. Oligonucleotides 12 and 13 were used in PCR together with 5-TCA TGG GGA GAA CCT GCA GAG AA (14, ref. 35). After ammonolysis, oligonucleotides and oligonucleotide dendrimers were evaporated to dryness, dissolved in 0.2 ml of 2 M LiClO4 and precipitated by the addition of 1.3 ml of cold acetone. After centrifugation the pellets were washed with acetone and used without further purification. Oligonucleotide labelling with T4 kinase Control oligonucleotide 11 (20 pmol) and 2 pmol of 9 (or 10) were labelled using 10 Ci of [-32P]ATP (10 Ci/l, 3000 pmol/ml) and 10 U T4 polynucleotide kinase at 37_C for 30 min. The reaction was stopped and products purified by spinning through TE-10 or TE-30 oligonucleotide purification columns. Array synthesis and hybridisation Arrays of oligonucleotides complementary to 9 and 10 and to single-labelled 15mer oligonucleotide control 11 synthesised across the gradient of linker 8 were fabricated as previously described (24,34) using aminated polypropylene as a solid support. Hybridisation reactions were carried out overnight at 30_C in TMA buffer (3.5 M TMA, 50 mM Tris-Cl pH 8, 0.2 mM EDTA, 0.04 mg/ml SDS). After washing under the same conditions, the arrays were exposed to a storage phosphor screen (Fuji STIII) which was then scanned in a PhosphorImager. PCR conditions The PCR was carried out using primers 12 (or 13) and 14 designed for the analysis of the microsatellite located in the first intron of the human LT- gene (35), in a 100 l volume with 100 ng of genomic DNA template, 25 M oligonucleotide primer 12 (or 13) and 14, 250 M each dNTP, 1 PCR buffer and 5 U Taq polymerase (Perkin Elmer) under the following conditions: 94_C for 30 s, 60_C for 30 s, and 72_C for 45 s for 30 cycles. DNA products were analysed on 2% agarose minigel in 1 TAE buffer. Enzymatic digestion The PCR product was treated with 2 U/l of T7 Gene 6 exonuclease (USB-Amersham) in 1 PCR buffer with adjusted concentration of Mg2+ up to 20 mM. After incubation at room temperature for 1 h the exonuclease was inactivated at 80_C for 15 min. One half of the reaction mixture was used in the subsequent polymerisation reaction with 2 M of primer 14, 250 M each dNTP and 5 U Taq polymerase using one modified PCR cycle: 94_C for 60 s, 60_C for 2 min, and 72_C for 10 min. DNA products were analysed on a 2% agarose minigel in 1 TAE buffer. RESULTS AND DISCUSSION Synthesis and characterisation of oligonucleotide dendrimers The key building block 7 was synthesised starting from pentaerythrite 1 (Scheme 1). To make sure there is enough space to accommodate three bulky DMTr-groups, three oxymethyl chains of 1 were lengthened with cyanoethyl moieties to give tri-cyanoethoxy-substituted compound 2 in a way similar to that previously described (11). The alcohol 2 was easily separated from tetra-substituted pentaerythrite, the main byproduct, by column chromatography. The Pinner transformation of 2 followed by hydrolysis of three imino-ester groups gave triester 3 in 75% yield. TBDMS-protection of the hydroxyl group followed by reduction in LiAlH4/THF led to triol 4 in 67% yield. This intermediate was tritylated with a slight excess of DMTrCl using DMAP and EDIP as catalysts. Tri-DMTritylated 5 was readily separated from mono- and di-DMTritylated derivatives by column chromatography due to its high chromatographic mobility compared to mono-, di-DMTritylated species and DMTrCl. Desilylation in TBAF/THF took 6 h to give tri-DMTritylated alcohol 6 in 88% yield, phosphitylation of which by usual procedures led to the final trebler phosphoramidite 7. The condensation efficiency of 7 was tested using a standard protocol for phosphoramidite oligonucleotide synthesis with an extended coupling time of 2 min. We did not come across the 4449

4450 Nucleic Acids Research, 1997, Vol. 25, No. 22

Scheme 1. Synthesis of the trebler phosphoramidite reagent and structure of pentaethyleneglycol spacer phosphoramidite 8. (a) 3.5 equiv. acrylonitrile, aq. NaOH, 50_C, 15 h, 63%; (b) MeOH/HCl, reflux 2.5 h, 75%; (c) TBDMSCl, pyridine, RT, 15 h, 98%; (d) LiAlH4/THF, 2.5 h, 67%; (e) 4 equiv. DMTrCl, pyridine, EDIP, DMAP, 7 h, 58%; (f) TBAF/THF, 6 h, RT, 88%; (g) CNEtO(N(iPr)2)PCl, EDIP, CH2Cl2, 2 h, 75%.

Figure 1. Mass-spectrum (MALDI-TOF) of 3-TpCH2C(CH2OCH2CH2CH2pT)3. Calculated for C54H81O35N8P4: 1526.151. Found: 1526.94 (MI + H+), 1548.94 (MI + Na+). The instrument was used in the positive ion reflector mode with a delay time of 80 ns, 108 averaged scans.

problem of longer acidic deprotection of the DMTr-group (13), probably because the synthesis scale used (1 mol) provided 80 s acidic exposure which was enough to cleave all the DMTr groups. The DMTr+ cation released after each condensation step (36) indicated that reagent 7 gives stable trebling of the amount of DMTr groups, up to three condensations when using 500 CPG supports, with decrease of the yield of condensation in subsequent steps. For the 1000 DMTrT-LCAA-CPG-(long chain alkylamino controlled pore glass) support, coupling yields of 7 were high (>95%). Four couplings gave an 80-fold increase in the number of terminal hydroxyl groups. The addition of a spacer oligonucleotide 510mer between the surface of the

1000 CPG support and the starting point of trebling led to a further increase in the efficiency of condensation; with this it was possible to achieve high yields of phosphotriester bond formation for the first five condensations to produce an 240-fold increase in the number of OH-groups. The structure of the compound obtained after condensing the trebler to the 5-OH group of the DMTrT-LCAA-CPG support and subsequent condensation of T-phosphoramidite to three hydroxyl groups of this dimer was confirmed by MALDI-TOF (Fig. 1). No trace of a product containing two thymidine residues instead of three, as would result from incomplete condensation, was detected. The chromatographic properties of this compound and pentathymidylate

4451 Nucleic Acids Research, 1997, Vol. 25, No. Nucleic Acids Research, 1994, Vol. 22, No. 122 4451

Figure 2. Chromatographic profiles (vertical axis: A254) of (A) 3-TpCH2C(CH2OCH2CH2CH2pTDMTr)3 and 5-DMTrT5 and (B) 3-TpCH2C(CH2OCH2CH2CH2pT)3 and T5, all being crude mixtures after ammonolysis. In both cases, the bottom line shows a control pentamer. Separation conditions: Rainin Dynamax RP 300 , C8 column (4.6 250 mm); 476% acetonitrile concentration gradient in 0.1 M triethylammonium acetate over 30 min. The compound containing three DMTr groups has a much longer retention time compared with DMTr-T5 (A) whereas detritylated species possess much more similar characteristics (B).

Oligonucleotide dendrimers as high sensitivity DNA probes The new synthon was used to create oligonucleotides comprising two parts: a probe moiety, to hybridise to the complementary target sequence; and a label moiety with multiple short runs of thymidylates which could be tagged with 32P-phosphates. Two different labelling moieties, 9 and 10 were synthesised (Scheme 2) with different lengths of branch between the probe moiety and the oligothymidylate chain. The longer branch was necessary to achieve full labelling by polynucleotide kinase (PNK). With this, the incorporation was eight times that of the conventional oligonucleotide, against 2-fold for the dendrimer 9 with the short spacer. The lower labelling of 9 could be explained by tight clustering of the 5-ends imposed by the centre of branching, which probably hinders interaction with the enzyme. For further experiments we used only structures 10 and 13, which is related to 10, with a different R (Scheme 2). We investigated the hybridisation properties of 10 to solid support-bound oligonucleotides synthesised on an aminated polypropylene support as previously described (34). First, an array of spacer 8 (from 0 to 4 units) was synthesised on the polypropylene using a hexagonal mask. Subsequent oligonucleotide synthesis was carried out orthogonal to the linker gradient using a narrow mask. For each target, two complementary oligonucleotides were synthesised: one tethered through its 3-end and the other through its 5-end.

Figure 3. Electrophoretic mobilities of compounds 911. Lane 1, 9 (total number of phosphate groups: 64); lane 2, 10 (total number of phosphate groups: 73); lane 3, control 15mer oligonucleotide (total amount of phosphate groups: 14); lane 4, 100 bp DNA ladder (starting from 100mer). 2% agarose gel, electrophoresis in 1 TAE buffer.

control are shown in Figure 2. Electrophoretic mobilities of 911 are shown in Figure 3.

4452 Nucleic Acids Research, 1997, Vol. 25, No. 22

Scheme 2. Oligonucleotide dendrimers of second generation 9, 10 and 13 used for hybridisation assays. 9 and 10, R = 15mer oligonucleotide 5-TCT TCT TCT TCT TTT with 3-free OH, connected to the dendrimers through 16-atom linker 8. 13, R = 5- GGT TTC TCT CTG ACT GCA TCT TGT CC (35) with free 3-OH, connected to the dendrimers through 16-atom linker 8. All dendrimers built up from nine pentathymidylates with kinated 5-ends.

For the hybridisation reaction equimolar amounts of 32Plabelled control oligonucleotide 11 and dendrimer 10 were applied to the array (Fig. 4). Quantitative analysis showed striking

differences in signal intensities from the areas complementary to the control and to the dendrimer. The two areas containing immobilised oligonucleotides complementary to the control were

4453 Nucleic Acids Research, 1997, Vol. 25, No. Nucleic Acids Research, 1994, Vol. 22, No. 122 4453

Figure 4. (A) PhosphorImage of hybridisation of an equimolar mixture of labelled 10 and 11 to the array manufactured as previously described (24,34). The intensity of the signal in the top lane (complementary to 10, favourable orientation) is 7.5 times higher than that for the controls complementary to 11 (two middle lanes). (B) Schematic representation of favourable and unfavourable orientations of oligonucleotide dendrimer to complementary array.

Figure 5. Use of 12 (13) and 14 as PCR primers. Products of PCR amplification with: lane 1, primers 13 (DMTr ON, i.e., nine DMTr groups at 5-ends of pentathymidylates are still present) and 14; lane 2, primers 13 and 14; lane 3, primers 12 and 14; lane 4, 100 bp DNA ladder (starting from 100mer). 2% agarose gel, electrophoresis in 1 TAE buffer.

of roughly equal intensity, whereas the areas complementary to the dendrimer 10 differed in the intensities both between each other and the controls. The area containing 3-immobilised oligonucleotide gave a signal about twice as high as the controls, whereas the patch containing 5-immobilised oligonucleotide gave a signal about 7.5 times higher. This difference between the 5- and 3-attached oligonucleotides could be explained by favourable and unfavourable orientations of the oligonucleotides: in the case of 5-immobilised oligonucleotide, the dendric part of 10 can be accommodated, as it sticks out of the layer of immobilised oligonucleotides, whereas for the 3-immobilised oligonucleotide the dendric part of 10 will be directed towards the surface where it will meet steric interference. The 7.5-fold augmentation of signal for the array bearing 5-immobilised oligonucleotide complementary to 10 is in good agreement with the increase in the incorporation of label, which was 8-fold that of the control. This result indicates that the large dendrimeric structure does not significally affect the hybridisation yield, even with oligonucleotides tethered to a surface, provided the orientation does not bring the dendrimer in proximity with the surface. Oligonucleotide dendrimers as PCR primers A range of applications would be opened up if dendrimeric oligonucleotides could be incorporated into products of PCR. There were reasons to believe that the presence of a large branched structure might interfere with duplex formation between the oligonucleotide and its target or with the activity of the polymerase. The following experiment shows that dendrimeric oligonucleotides can be incorporated into products of the PCR, and that the branched structure has a considerable effect on the electrophoretic mobility of the product. Oligonucleotides 12 (or 13) and 14 were used as primers to produce a specific fragment corresponding to the intron region of human lymphotoxin gene which contains a microsatellite repeat. Two PCR amplifications were carried out in parallel using genomic DNA template. Non-dendrimeric primers 12 and 14 were used in a control reaction. In another reaction, primer 13

(dendrimeric analog of 12, see Materials and Methods and Scheme 2) was used together with 14. The control gave a band of expected length of 170 bp with genomic DNA from different sources (Fig. 5). With the dendrimeric primer 13, two PCR products were observed on a gel, one being the same length of 170 bp, and another with the mobility of 200 bp duplex. The latter fragment size is in agreement with the number of additional nucleotide residues (45 nucleotides plus four trebling units, 7, and 10 linking units, 8) attached to the 5-end of 13. The fragment of 170 bp could be explained by the presence of non-completed oligonucleotide fragments in 13. These fragments are capped during the oligonucleotide synthesis to avoid synthesis of truncated species. We did not purify 13 after the synthesis, so after ammonolysis the shorter oligonucleotide fragments (like 12) were deblocked and acted as usual PCR primers competing with 13 and giving PCR products similar to that for 12. Conversion of PCR products to single strands The products of the PCR are double-stranded and therefore make poor hybridisation probes without further treatment. Many of the methods for making probes, such as asymmetric PCR, are cumbersome. In our experience, the only consistently reliable methods are transcription, which requires the incorporation of a promoter in one of the primer oligonucleotides; or degradation of one of the strands by an exonuclease which requires that one of the primers is made resistant to the enzyme. This can be achieved by incorporating phosphorothioate residues in one of the primer oligonucleotides (37). The structure of the dendrimer is such that it leaves a bunch of oligonucleotides as a 5-overhang on the double-stranded PCR product. Such a structure is naturally resistant to a number of exonucleases including T7 Gene 6 exonuclease. The product of PCR made as described above was treated with T7 Gene 6 exonuclease. Gel electrophoresis (Fig. 6) showed a faint band, as expected for a single-stranded product. That it was the correct strand was made clear by converting it to a double-strand using only 14 as primer. The product had the same mobility on gel electrophoresis as the initial PCR product. Significantly, the by-product of the initial PCR did not reappear,

4454 Nucleic Acids Research, 1997, Vol. 25, No. 22 Incorporating dendrimers into products of the PCR offers the opportunity to prepare probes with higher labelling capacity. The presence of the branched end also introduces properties that are useful in the preparation of single-stranded probes, which is achieved simply by adding an exonuclease to the PCR product.

REFERENCES
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Buhleier,E., Wehner,W. and Vgtle,F. (1978) Synthesis 155157. Newcome,G.R. (1996) J. Heterocycl. Chem. 33, 14451460. Dagani,R. (1996) Chem. Engng. News 74, 3038. Newkome,G.R., Moorefield,C.N. and Vgtle,F. (1996) Dendritic Molecules: Concepts, Syntheses, Perspectives. VCH, New York. http.//dendrimers.cas.usf.edu/links.html#PIONEERS Miller,T.M. and Neenon,T.X. (1990) Chem. Mat. 2, 346349. Hudson,R.H.E. and Damha,M.J. (1993) J. Am. Chem. Soc. 115, 21192124. Boussif,O., Lezoualch,F., Zanta,M.A., Mergny,M.D., Scherman,D., Demeneix,B. and Behr,J.-P. (1995) Proc. Natl. Acad. Sci. USA 92, 72977301. Kukowska-Latallo,J.F., Bielinska,A.U., Johnson,J., Spindler,R., Tomalia,D.A. and Baker,J.R. (1996) Proc. Natl. Acad. Sci. USA 93, 48974902. Jansen,J.F.G.A., de Brabander-van den Berg,E.M.M. and Meijer,E.W. (1994) Science 266, 12261229. Newkome,G.R. and Lin,X. (1991) Macromolecules 24, 14431444. Amabilino,D., Ashton,P., Balzani,V., Brown,C., Credi,A., Frecht,J.M, Leon,J.W., Raymo,F., Spencer,N., Stoddart,J. and Venturi,M. (1996) J. Am. Chem. Soc. 118, 1201212020. Horn,T. and Urdea,M.S. (1989) Nucleic Acids Res. 17, 69596967. Azhayeva,E., Azhayev,A., Guzaev,A., Horinen,J. and Lnnberg,H. (1995) Nucleic Acids Res. 23, 11701176. de Brabander-van den Berg,E.M.M. and Meijer,E.W. (1993) Angew. Chem. Int. Ed. Engl. 32, 13081310. Lartigue,M.-L., Slany,M., Caminade,A.-M. and Majoral,J.-P. (1996) Chem Eur. J. 2, 14171426. Backinowsky,L.V., Kitov,P.I. and Kochetkov,N.K. (1993) Russ J. Bioorg.Chem. 19, 644654. Lehmann,J. and Weitzel,U.P. (1996) Carbohydr. Res. 294, 6594. Hanessian,S., Qui,D., Prabhanjan,H., Reddy,G.V. and Lou,B. (1996) Can. J. Chem. 74, 17381747. Hanessian,S., Prabhanjan,H., Qui,D. and Nambiar,S. (1996) Can. J. Chem. 74, 17311737. Urdea,M.S., Running,J.A., Horn,T., Clyne,J., Ku,L. and Warner,B.D. (1987) Gene 61, 253264. Maskos,U. and Southern,E.M. (1992) Nucleic Acids Res. 20, 16791684. Southern,E.M. and Maskos,U. (1994) J. Biotechnol. 35, 217227. Southern,E.M., Case-Green,S.C., Elder,J.K., Johnson,M., Mir,K.U., Wang,L. and Williams,J.C. (1994) Nucleic Acids Res. 22, 13681373. Kwiatkowski,M., Sund,C. and Hurskainen,P. (1988) Pat. PCT WO 90/00622. Horn,T., Warner,B.D., Running,J.A., Downing,K., Clyne,J. and Urdea,M.S. (1989) Nucleosides Nucleotides 8, 875877. Bazin,H., Roget,A. and Teoule,R. (1991) Nucleosides Nucleotides 10, 363366. Gryaznov,S., Loginova,L., Potapov,V.K., Stetsenko,D. and Veselovskaya,S. (1991) Nucleic Acids Symp. Ser. 24, 207208. Korshun,V.A, Boreskov,Y.G, Pestov,N. and Berlin,Yu.A. (1991) Nucleic Acids Symp.Ser. 25, 269. Korshun,V.A, Pestov,N.B., Nozhevnikova,E.V., Prokopenko,I.A., Gontarev,S.V. and Berlin,Yu.A. (1996) Synth. Comm. 26, 25312548. Caruthers,M.H. (1991) Acc. Chem. Res. 24, 278284. Seela,F. and Kaiser,K. (1987) Nucleic Acids Res. 15, 31133129. Durand,M., Chevrie,K., Chassignol,M., Thuong,N.T. and Maurizot,J.C. (1990) Nucleic Acids Res. 18, 63536359. Shchepinov,M.S., Case-Green,S.C. and Southern,E.M. (1997) Nucleic Acids Res. 25, 11551161. Nedospasov,S.A., Udalova,I.A., Kuprash,D.V. and Turetskaya,R.L. (1991) J. Immunol. 147, 10531059. Gait,M.J. (1984) Oligonucleotide Synthesis. A Practical Approach. IRL Press, Oxford, UK. Nikiforov,T.T., Rendle,R.B., Kotewics,M.L. and Rogers,Y.-H. (1994) PCR Methods Appl. 3, 285291.

Figure 6. Enzymatic digestion of PCR product. Lane 1, products of PCR amplification with primers 13 and 14; lane 2, treatment of PCR products from lane 1 with T7 Gene 6 exonuclease; lane 3, resynthesis of a second strand using 14 as a primer and products of exonuclease treatment (lane 2) as a template; lane 4, 100 bp DNA ladder (starting from 100mer). 2% agarose gel, electrophoresis in 1 TAE buffer.

as would be expected if it were not capped by the dendrimer, but were the result of priming from truncated failure sequences. Mobility shift by dendrimeric tags It can be useful to modify the electophoretic mobility of a PCR product by attaching a polymeric tag to one of the primers. The resulting mobility shift can be used to spread a set of multiplexed PCR products so that they do not interfere with each other. The addition of a dendrimeric bunch of short oligonucleotides to a PCR primer has a profound effect on the mobility of the PCR product (Fig. 3): addition of nine pentathymidines decreased the mobility in agarose gels by an amount equivalent to adding 50 bp; this effect is even larger in polyacrylamide gels because of the smaller pore size (7). The number of couplings required to make the branched tag is small, especially when compared with the number of couplings required to make an unbranched polymer with the same effect on mobility. A total of seven steps was used to make the dendrimeric tag: 50 steps would be needed to add 50 bases. CONCLUSIONS We have shown that conventional methods of oligonucleotide synthesis can be adapted to the synthesis of dendrimeric oligonucleotides using a branched phosphoramidite with multiple protected primary hydroxyl groups. The branching monomer used in these studies permits the synthesis of a dendrimeric head on top of a conventional monomeric oligonucleotide. The monomeric and the dendrimeric sequences can have different lengths and different orientation: we have shown how this can be exploited to design molecules with desirable features. The novel monomer has three growth points leading to rapid expansion of the branches: we demonstrate the parallel synthesis of nine oligonucleotides on a branched structure produced after only two rounds of expansion. Such structures offer a number of new applications. For example, besides the obvious means of amplifying signal demonstrated in this work, it is possible to introduce different labels on the ends of different branches; these could be fluorophores with different emission spectra, providing potential for a wide palette of colours; or the different ends could bear the donor and acceptor of an energy transfer pair. It is also possible, by using a variety of blocking groups, to synthesise a different sequence on each branch.

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37