The support does two jobs at once in SPOT synthesis, and confusing them causes most of the misreadings. A sheet of cellulose is the solid phase for several hundred independent Fmoc syntheses, and it is also the surface on which those peptides are later probed. Nothing separates the two roles. The peptide you assemble is the peptide you assay, on the same square centimetre of paper, without ever being weighed, purified or moved. Almost everything the method can and cannot tell you follows from that.
Ordinary chemistry, unusual compartment
SPOT synthesis follows standard fluorenylmethoxycarbonyl chemistry on conventional cellulose sheets and can use more than 600 different building blocks [1]. The procedure runs in three phases — preparation of the cellulose membrane, stepwise coupling of the amino acids, and cleavage of the side-chain protecting groups — and where soluble material is needed, peptides can be cleaved from the membrane afterwards [1]. There is no resin, no reaction vessel, no filtration. The reaction compartment is the wetted area of the droplet of activated amino acid delivered to a position on the sheet: that wetted area is the spot, and its size is set by the volume dispensed and by how readily the membrane takes up liquid.
Throughput follows from that geometry rather than from any chemical limit. A cellulose sheet of 19 by 29 centimetres carries around 900 large addressable spots [2]. That sheet is 551 square centimetres, so the working density is under two spots per square centimetre — arithmetic, not a measurement. A run of 400 nine-residue peptides takes about six days, and the cost per peptide is less than 1% of the same peptide made conventionally on resin [1].
Decide against the printed high-density formats before you commit. SPOT was characterised by its originator as not as impressively miniaturised as photolithographic array synthesis, while fulfilling similar demands with the advantages of a reliable and easy experimental procedure, inexpensive equipment needs and highly flexible array and library formatting; it permits rapid and highly parallel synthesis of large numbers of peptides and peptide pools including a large variety of unnatural building blocks, and adapts easily to binding, enzymatic and cellular assays [3]. The density gap is real. Combinatorial synthesis on microelectronic chips, in which solid amino acid microparticles are charged by friction, transferred to defined pixel electrodes and coupled on melting under standard Fmoc chemistry, has reached peptide array densities of up to 40,000 spots per square centimetre [4]. Those authors group cellulose spot membranes with split-pool bead libraries and name the number of synthesisable peptide sequences per area among the specific limitations of both, alongside false-positive results [4]. Scan geometry itself — peptide length, offset, and what a given offset guarantees to present intact — is the same arithmetic on any support, and is treated in the companion guide on printed peptide microarrays. Choose SPOT for chemical freedom and cheap iteration; choose a printed or in-situ high-density format when raw sequence count is the constraint.
What the peptide is actually attached to
Cellulose presents hydroxyls, and the first synthetic decision is what you convert them into. Aminofunctionalising the membrane through a polyethylene glycol spacer gives supports stable to trifluoroacetic acid and to strong aqueous base for days, which extends the scope of synthesis considerably: more efficient side-chain cleavage protocols can be applied, and these yielded peptides of improved purity [5]. Membranes prepared this way were, for the first time, accessible at a loading as high as 5 micromoles per square centimetre, and the membranes reported in that study were compatible with antibody binding as well as with enzymatic phosphorylation assays [5].
Maximum accessible loading is not the density you should run. For binding assays a peptide density of 10 nanomoles per square centimetre has been recommended as a reasonable starting point for further optimisation, because reducing peptide density on the cellulose avoids the ring spot effect — less binding observed in the spot centre than at the rim [6]. Those two figures describe different quantities: what a support can carry, and what you want presented — five hundred times lower. If spots are reading as annuli rather than discs, density is the first parameter to lower.
Cleavable, or not
The linker decides whether the membrane is an assay surface or a synthesis plate, and the fork is chemical. Two rigid bi-aromatic linkers built from cellulose treated with 2,4-dichloro-6-methoxy-1,3,5-triazine illustrate it [7]. Reaction with m-phenylenediamine gives a non-cleavable linker that attaches peptides through amide, ether and amine bonds resistant to harsh reaction conditions; reaction with 3-Fmoc-aminobenzoic acid followed by thermal isomerisation of the intermediate ester gives a linker very stable during peptide synthesis, from which the peptide is cleavable — together with a fragment of the linker — in 1 molar lithium hydroxide [7]. Note what the second case releases: peptide plus linker remnant, not the bare peptide, and any downstream mass measurement has to account for it.
Where the free peptide must carry an authentic C-terminus, the cleavage chemistry has to be acid-labile rather than alkaline. Using hydroxymethylphenoxyacetic acid for 19 amino acids and 4-(4-hydroxymethyl-3-methoxyphenoxy)butyric acid for proline as acid-labile linkers reduced side-chain reactions normally occurring during conventional alkaline peptide cleavage from cellulose membranes, and the improved cleavage and washing steps gave peptides with authentic C-termini in a purity range of 60–95%; the technology is described as meeting most requirements for highly parallel synthesis of soluble peptides within the upper microgram range [8]. Alkaline release also has a cheap form: peptides can be cleaved from the cellulose support by ammonia gas and afterwards spotted onto glass microchips, or synthesised on modified cellulose discs and converted into CelluSpots microarrays [2].
Tethering fixes orientation, and orientation can silence an entire class of interaction. Most solid-support-bound peptide libraries lack a free C terminus precisely because the peptide is fixed to the support through it — a problem when the recognition event reads C-terminal residues, as PDZ domains commonly do [9]. A method for generating membrane-bound peptides with authentic C termini was validated by screening a human library of 6,223 C termini against the syntrophin PDZ domain, and the same library produced new peptide ligands for the ERBIN PDZ domain [9]. If the motif you are chasing sits at a terminus, a conventionally tethered array is the wrong instrument rather than a failed experiment.
Why a spot cannot be quantified absolutely
Nothing in the workflow produces a mass per spot. There is no weighing, no per-spot ultraviolet reading, no aliquot to inject. What you set is a surface density, fixed by how the membrane was derivatised and how much activated amino acid was delivered; what you read is a probe signal. Between those two the method is explicitly semiquantitative — SPOT signals provide an approximation of the dissociation constants of the several thousand interactions analysed simultaneously in array format [10] — and the technique is described as able to distinguish binding affinities semi-quantitatively within one array [11]. That last qualifier is the operative one: within one array.
How far the rank can be trusted has been measured directly. Signal intensity of ligate binding to cellulose-bound peptides and the affinity of the corresponding soluble peptides show good correlation, illustrated across three different ligate binding assays, and statistical analysis of binding assay reproducibility with more than 15,000 peptides gave a mean standard signal deviation of 0.18 [6]. Using one monoclonal antibody as a model system and a set of peptides of differing affinity for it, SPOT was shown to assign spot signal intensities accurately to three different binding affinity classes, with the dissociation constants of the intermediate region falling between pKdis 5 and pKdis 7 — that is, between 10 micromolar and 0.1 micromolar — and altering the experimental parameters caused a directed change of that region [12]. Accuracy improved when regional trends in the membrane surface were taken into account, and the applied concentrations of the binding partners were found not to correspond directly to the effective concentrations in the assay [12].
That last point is what forbids absolute quantification. You know what you pipetted onto the membrane; you do not know the concentration the peptides experienced. Three affinity classes, for one antibody, on one membrane, under one set of conditions, is what an intensity map licenses. A dissociation constant is not.
Incomplete coupling and the population at a spot
Every spot holds a mixture, and the mixture degrades with length. Coupling efficiencies quantified for three coupling methods across four model decapeptides placed diisopropylcarbodiimide with HOBt ahead of methods using pentafluorophenyl esters of Fmoc amino acids, at a mean coupling yield per cycle of 87% to 91% [13]. Carry that forward. A decapeptide needs nine couplings after the first residue is anchored, so a uniform 90% per cycle leaves roughly 39% of chains full length, and a uniform 87% leaves about 29%. Extend to a 15-mer, fourteen couplings, and the same uniform assumption gives roughly 23% and 14%. The arithmetic is crude — real yields vary with residue and with neighbour, and capping changes what the failures become — but its direction is not in question.
The consequence is asymmetric, and that asymmetry is what gets misread. A hit remains a hit: all four decapeptides prepared by all three coupling protocols were reactive with their cognate monoclonal antibody, and using the best coupling method, peptides of increasing length from 10 to 29 residues bearing a model epitope at the N terminus were immunoreactive up to 29 residues [13]. Enough correct chains survive to give signal. The fragile result is the negative, because a spot whose full-length fraction has fallen to a quarter is a weaker presentation of your designed sequence than the layout implies, and a blank position cannot distinguish that from an absent determinant.
Two levers improve the purity of what a spot carries, though neither raises the per-cycle coupling yield itself. Deprotection is one: optimisation showed that a 30-minute pretreatment of the cellulose with 90% trifluoroacetic acid, followed by the standard deprotection protocol, resulted in higher purity of the synthesised products [6]. Support chemistry is the other, since acid- and base-stable membranes admit the more efficient side-chain cleavage protocols that gave improved purity [5]. Neither delivers a purity figure for an individual spot. The purity range quoted earlier comes from cleaved material analysed away from the membrane — 60–95%, a range across a method rather than a value for a position [8]. Treat per-spot purity as unknown unless you cleave that spot and measure it.
The library designs that do the work
Three library types account for most antibody work on SPOT arrays: protein-sequence-derived scans of overlapping peptides, used to locate epitopes within the protein sequence; truncation libraries, used to identify the minimal peptide length required for antibody binding; and complete substitutional analyses, used to identify the key residues important for antibody binding [14].
A worked example fixes what each yields. The 9E10 antibody epitope EQKLISEEDL, widely used as a protein fusion tag, was characterised by substitutional analysis and length variation using peptide spot synthesis on cellulose: the key amino acids of the interaction are the core residues LISE, and the shortest peptide with a strong binding signal is KLISEEDL [15]. Ten residues in, four residues out. That is the resolution a substitutional scan buys, and it is per-residue rather than per-region.
Substitution need not be exhaustive to be informative. Overlapping peptides covering the whole sequence of the major coat protein g8p of filamentous bacteriophage fd were synthesised by spot synthesis on cellulose membranes; each of three tested monoclonal antibodies reacted with a core epitope covering ten amino acid residues at or near the amino terminus of g8p [16]. The contribution of individual residues was then analysed with a set of peptides containing individual amino acids exchanged by glycine, identifying Glu2, Asp4, Asp5, Pro6, Lys8, Phe11 and Asp12 as accessible residues, and the positions of the essential residues within the epitope are in accordance with a helical conformation of the amino-terminal region of g8p [16]. That study also carries its own control: extension of the amino terminus by various sequences did not inhibit binding, indicating that a terminal amino group is not essential for the interaction [16].
The cost of the exhaustive version is arithmetic worth doing before you commit membrane. A complete substitutional analysis of a 10-mer, placing each of the 20 common amino acids at each of the 10 positions, is 200 spots, of which 10 restore the parent residue, leaving 190 changed sequences — per hit peptide. A single-substitution scan to one residue, glycine or alanine, is 10 spots. The gap between those two numbers is why partial scans stay common.
The format is not confined to antibodies. Peptides matching a relaxed consensus derived from phage display were synthesised at high density on a cellulose membrane and probed directly with SH3 domains; eight domains were grouped into five classes with partially overlapping specificity, and within classes displayed high promiscuity, binding a large number of common targets with comparable affinity [10].
Controls that separate binding from surface and reagent chemistry
Run the detection chain alone. False-positive results can occur both when a peptide array is challenged with analyte and when captured analyte is detected with label conjugates, and the cross-reactivity of peptides with the detection agents was for a long time little known; the potential of 5-(and 6)-carboxytetramethylrhodamine, of fluorescein isothiocyanate in the form of the peptide-bound fluorescein-substituted thiourea derivative, and of biotin/streptavidin-peroxidase to cross-react with individual amino acids in a peptide sequence has been described at the amino acid level [11]. A membrane carrying the same peptides, probed with secondary reagent and buffer only and read on the same settings, is the cheapest experiment on the bench.
Read spot shape, not only integrated intensity. The ring spot effect is a density artefact with a documented fix — lower the peptide density [6] — and an annular spot is reporting on your synthesis rather than on your ligand.
Distribute replicates across the sheet. Because assignment accuracy improved once regional trends in the membrane surface were accounted for [12], a block of replicates in one corner measures that corner as much as it measures the sequence. Interleave them, and interleave the controls with them.
Include an orientation control wherever terminal chemistry might matter. A peptide whose recognition depends on a free C-terminus is invisible on a conventionally tethered array, and the comparison that reveals it is the same sequence presented with an authentic C terminus [9].
Confirm hits away from cellulose. In the SH3 domain work, all the targets of the Abp1 SH3 domain identified by the array approach were shown to bind the native protein by coimmunoprecipitation [10]. Cleaving the peptide and re-testing it in solution is the other route, and the method supports it directly [1].
What a SPOT experiment settles
It settles that, under one set of conditions and on one membrane, certain positions in a layout you designed retained a probe more than others. It supports a rank, and with a model system an assignment to broad affinity classes [12]. It does not establish a dissociation constant, a mass or a purity, and it does not license comparison of raw intensities across membranes. The blank position remains the weakest output: it may mean no determinant is present, or that the full-length fraction was too low to present one, or that the determinant needs a free C terminus the tether removed [9], or that peptide density suppressed binding at the spot centre [6].
Report accordingly. Nothing in the following list is a finding; it is the minimum that makes a SPOT experiment legible to someone else. Membrane type, functionalisation, and spacer identity and length. The linker, whether it is cleavable, by what reagent, and what leaves with the peptide. Nominal peptide density and how it was set. Coupling reagent and whether couplings were doubled. The deprotection protocol, including any acid pretreatment. Blocking buffer, wash stringency, probe identity and concentration, and the detection chain. Replicate layout across the sheet, and whether a detection-reagent-only membrane was run on the same settings. And whether hits were re-synthesised and re-tested away from cellulose — a hit reproduced only on the membrane that produced it has been repeated, not validated.