Multiple labelled nanoparticles for bio detection

Remote nanoparticle detection is required for the development of in situ biological probes.

Here we describe the labelling of silver nanoparticles to produce multiply coded particles which can be detected by surface enhanced resonance Raman scattering (SERRS).

There is a potential for thousands of codes to be written and read without the need for spatial resolution of components of the code.

The use of these systems in bioanlaysis and in situ detection is discussed.


The use of nanoparticles for in situ detection of specific reactions has significant potential to probe living cells and tissues.

For example, the detection in a cell of DNA or protein/receptor interactions will provide information of specific functionality.

Fluorescence is a possible detection technology since it is both sensitive and selective and is already widely used for biological analysis.

However, one form of Raman spectroscopy, surface enhanced resonance Raman scattering (SERRS), has superior properties for labelling of nanoparticles.

Both techniques have similar sensitivities1 but with SERRS many more codes can be written onto the surface of the nanoparticles than is the case with fluorescence.

In addition, since both fluorophores and non fluorophores are suitable a more robust, extensive and simpler labelling chemistry can be used.2,3

Raman scattering has many advantages in biological systems in that it provides a vibrational spectrum from aqueous solutions using visible wavelength lasers and detectors.

It involves the irradiation of a sample with a laser and the measurement of components of light scattered from the target system, which are displaced in frequency by one vibrational unit.

The pattern of signals is often diagnostic or is characteristic of a particular molecule or group, enabling in situ detection, particularly if the analyte signal is enhanced over the background.

Water provides very poor Raman scattering and consequently is nearly transparent making the technique useful in bioanalysis.

However, there are two big disadvantages, namely that Raman scattering is weak with only 1 in about 108 scattered photons being Raman scattered and fluorescence from the sample matrix can often swamp the Raman signal.

Enhancement of Raman scattering can be obtained in a number of ways.

Among the most effective is SERRS in which the analyte molecule is adsorbed onto a suitable roughened surface such as that of silver or gold nanoparticles and an excitation frequency close to the plasmon frequency of the particles is used.4–7

The plasmon frequency is often shifted by aggregation of the particles.

The enhancement can be increased further by using an analyte which is a dye for which one of the absorption peaks lies in a frequency region reasonably close to the excitation frequency8,9 (within 50–100 nm).

SERRS enhancement of the Raman signal of up to 1014 has been claimed and single molecule detection has been demonstrated.10,11

In addition to the sensitivity, one of the major improvements of SERRS over existing technologies is that the Raman signals obtained are sharp and can be interpreted easily in terms of the resonance spectrum of the analyte.

The narrow signals can be more easily resolved than those of fluorescence and the signal pattern is molecularly specific.

Thus, individual dyes can be clearly discriminated in mixtures and this opens up the potential for the coding of nanoparticles with multiple labels in a manner which enables them to be identified in situ in a flowing stream or in a biological matrix.

In this paper, the ability to label silver nanoparticles with a number of dyes and with an oligonucleotide suitable for DNA hybridization and hence DNA labeling is demonstrated.

For a dye to be an effective label for reproducible SERRS it must bind strongly to the surface even in the biological environment.

A number of dyes using different types of metal complexing are used and investigated to see if they can simultaneously bind to the nanoparticles or if dyes with one type of complexing group dominate.

The possible future application of these functionalised nanoparticles is investigated by attaching a dye labeled oligonucleotide to create a SERRS active analyte and examining the stability in buffer.

These experiments are carried out in suspensions of the colloidal nanoparticles and it is difficult to assess whether all single particles are multiple labeled in the same way.

We do, however achieve this on a single micro bead which can be optically detected.


All spectra were acquired using a Renishaw 2000 Raman Microprobe with a triple stage Peltier cooled CCD detector.

The excitation was provided by a Spectra-Physics Model 2020 argon-ion laser with a wavelength of 514.5 nm and 3 mW of power at the source.

Samples were analysed in a plastic microtitre plate using a ×50 objective.

The acquisition time for all spectra was 10 s and the grating was centered at 1350 cm−1.

The concentration study was carried out on a Renishaw inVia system with a CCD spectrometer.

The excitation was provided by a Spectra-Physics argon-ion laser with a wavelength of 514.5 nm and 10 mW of power at the source.

The citrate reduced silver nanoparticle suspension used in this work was prepared by a modified Lee and Meisel method.12,13

An aggregating agent is commonly added to the colloid to induce the formation of small clusters, whose surface plasmon matches the frequency of the excitation source.

The aggregating agent used was either a 1 M sodium chloride solution or a 0.01 M spermine solution.

Spermine was used for all experiments involving the labelled oligonucleotide since as well as being an effective aggregating agent for silver colloid,14 it is a known charge neutralisation agent for the negatively charge phosphate backbone of DNA,15 aiding adsorption of the sample to the metal surface.

For each sample 250 μL of nanoparticle suspension, 250 μL of distilled water, 30 μL of each analyte and 10 μL of aggregating agent were mixed together and the SERRS was immediately acquired.

The oligonucleotide examined contained a basic priming sequence of 5′ GTG CTG CAG GTG TAA ACT TGT ACC AG 3′.

The visible chromophore used was the fluorophore 2,5,2′,4′,5′,7′-hexachloro-6-carboxyfluorescein (HEX), which was attached at the 5′ terminus.

HEX is negatively charged and therefore repels the negatively charged metal surface.

Surface attachment was achieved by the incorporation of positively charged modified nucleobases at the 5′-terminus next to the HEX label.16

Rhodamine dye was purchased from Aldrich and all other dyes used were synthesised in our Raman group.

Results and discussion

SERRS can be used for labelling both fluorophores and non fluorophores.

Fluorescence quenching from a dye efficiently adsorbed on the surface is very effective so that the dye is clearly and easily recognised.

Thus, rhodamine 6G, a standard fluorophore, shows excellent Raman scattering with a very high degree of sensitivity.

This dye has been widely used in the development of SERRS spectroscopy.

However, when we attempted to use it in biological media, fluorescence which obscured the Raman scattering was obtained from the silver particle suspensions.

It would appear that rhodamine does not adhere strongly to surfaces and can be easily displaced by biological media such as buffer.

To overcome this problem we have used dyes which have a specific attachment chemistry to the surface.

We have recently published a number of papers on some of these and related dyes showing that they are effective in displacing ligands from the surface and in giving excellent SERRS.

However we also use a new dye here which is discussed later.

To test the sensitivity of the assay, a concentration study was carried out using 3,5-dimethoxy-4-(6′-azobenzotriazolyl)-phenylamine, to discover the minimum amount that could be detected by SERRS.

Fig. 1 shows a plot of SERRS intensity against concentration for this dye.

The acquisition time used was 0.5 s for all measurements.

The intensity of the peak at 1367 cm−1, after the background had been subtracted, was measured for each spectrum.

For each concentration, five consecutive measurements were made and the average number of counts in this peak determined.

The dye was detectable down to a concentration of the order of 10−10 M, calculated as the final concentration that would have been present in solution before adsorption of the dye to the metal surface.

The concentration versus intensity plot is almost linear, apart from the highest concentration.

By calculation of the surface area present and approximating the area each dye molecule occupies, the approximate concentration at which monolayer coverage occurs can be worked out as about 10−6 M. Thus the highest concentration shown is above monolayer coverage.

Since SERRS intensity drops off rapidly with distance from the surface, the levelling of the graph is to be expected.

Previous papers have commented on the fact that there is a change in slope approximately at one tenth of surface coverage, i.e. at about 10−7 M. This dye can affect the charge on the particles and thus above one tenth of monolayer coverage both the state of aggregation and also packing effects can play a role.

However compared to the effect of complete coverage of the surface, Fig. 1 shows that the effect of this is minor.

At 10−10 M, there are about 1000 molecules in the beam with the objective used.

This is not a detection limit, since a longer accumulation time or more laser power would greatly improve the sensitivity.

However, the focus here is not on single molecule detection but on multiple labelling so this was not pursued.

We have already shown single molecule detection elsewhere.16

The experiment was carried out using 514 nm excitation and sodium chloride to aggregate the colloid.

The sodium chloride treatment induces the formation of small clusters which remain in suspension, and whose plasmon is shifted from the frequency of the single particle (406 nm) to higher wavelengths.

Since the clusters formed have a variety of shapes and sizes, and therefore different plasmon frequencies, not all clusters are resonant with 514 nm excitation.

Consequently only a few percent at most of the dye is actually surface and resonance enhanced.

Although an aggregating agent has been employed, we have shown that the dye used in this concentration study actually induces aggregation itself.

To investigate single particle enhancement, similar experiments were carried out using no aggregating agent and a non aggregating dye in which the amine group is replaced by a hydroxyl group.

This ensures that on adsorption of the ligand, the negative charge on the particle is maintained and hence no aggregation occurs.

This was checked using a Malvern particle size analyser.

The single particle enhancement is not as large as with aggregates.

When an excitation frequency of 406 nm, corresponding to the frequency of the unaggregated plasmon, was used, the enhancement was approximately one sixth of that obtained with an aggregating agent at 514 nm.

Thus, eventually in biological use two different approaches are possible.

Single particles can be used directly with some drop in sensitivity, or a specific way can be found to make selected clusters of one particular size and plasmon resonance.

To indicate the advantage of SERRS in selective labelling, a set of three dyes was selected from those available in the laboratory.

The first dye was rhodamine 6G, a standard fluorophore, which attaches to the silver by electrostatic attraction through the positive charge on the dye and the negative charge on the citrate reduced silver.

The second dye was 3,5-dimethoxy-4-(6′-azobenzotriazolyl)-phenylamine, which was used in the sensitivity study.

This dye uses the benzotriazole group to complex strongly to the silver.

We have shown previously that benzotriazole can displace a citrate layer from the surface to form a thermodynamically more stable layer.

We have already shown that in practice, rhodamine will be unsuitable.

Correctly attached to the silver the fluorescence is completely quenched but displacement is easy and the fluorescence from even a small amount of the dye prevents effective SERRS measurement.

Therefore we have begun to explore other mechanisms of binding than the use of benzotriazole to ensure that a wide range of labels with good surface attachment can be synthesised.

The third dye was 4 pyridine azo.

Pyridine was the first molecule to exhibit surface-enhanced Raman scattering through adsorption at a metal surface.4

However since the new dye has the nitrogen at either end of the molecule it complexes to form a linear chain with silver(i).

It is believed that it will self assemble on the silver surface and thus become immobilised.

Fig. 2 shows the spectra obtained with these three dyes.

In this figure, each dye can be identified individually in the dye mixture by the presence of unique marker bands.

By varying the concentration of each dye to alter its relative intensity in the spectrum from the dye mixture, it is possible to write thousands of codes onto the silver colloids without the need for any form of separation.

Thus, with the use of the correct dye set, a system of nanoparticle detection can be constructed in which single nanoparticles can be identified in situ without any extensive preparation or separation.

One way of using the system in biodetection would be to attach the particles to an oligonucleotide containing a chromophore.

We have shown previously that a dye labelled oligonucleotide can be used with standard molecular biology procedures to detect by SERRS the defect in cystic fibrosis genes.17

In essence, the labelled nanoparticle takes the place of a fluorophore in the published assay.

DNA was attached to the nanoparticle surface using a linker and a specific dye molecule, in this case, HEX.

A further three dyes were attached to the metal surface.

Dye 1,1-[4-(8-hydroxy-quinolin-5-ylazo)-phenyl]-ethanone and dye 3, N,N′-bis-quinolin-8-ol-5-ylmethylene-hydrazine, are novel dyes which attach to the metal surface through the 8-hydroxyquinoline complexing group.

Dye 2,1-[4-(7-amino-3H-benzotriazol-4-ylazo)-phenyl-]ethanone complexes through the benzotriazole group.

Fig. 3 shows the spectra acquired from each of the individual dyes, including that with the oligonucleotide attached.

This figure also shows the spectrum recorded from the mixture of all dyes.

The acquisition time for all spectra was 10 s and all have been normalised to have the same maximum intensity in the highest peak in each spectra.

In Fig. 3 dashed lines have been added linking features in the spectrum from the mixture with the spectrum from the pure sample from which they originate.

It is possible to pick out the presence of all the dyes, including the one with the DNA strip attached, by eye, in the spectrum from the dye mixture.

It is clear that the peaks labelled A, B and C are due to the presence of dye 1.

Peak D has contributions from both dye 3 and HEX.

The relative intensity of this peak is higher in the spectrum from the mixture than in either of the individual spectra from HEX or dye 1 showing that the presence of both samples have contributed to this feature.

A shoulder, E*, is visible on peak E. It is likely that E and E* originate from contributions from dye 1 and dye 3 which have peaks at similar positions.

Feature F can be attributed to dye 1 and the peak G, and its shoulder G* look similar to the feature in HEX at that wavenumber.

However, dye 3 also has a peak at the position of peak G. The relative intensity of G* compared to G has decreased in the spectrum from the mixture compared to that of pure HEX, indicating that the contribution from dye 3 has increased the intensity of peak G in the mixture.

In the pure HEX spectrum peak G is less intense than the peak at higher wavenumber.

The difference in intensities between these peaks is not so large in the spectrum from the mixture, providing further evidence for a contribution from dye 3.

Peak H is due to the presence of dye 3 and I can be attributed to dye 2.

Dye 1 also has a peak at the position of peak I. However, examination of the spectrum from dye 1 shows that this peak is less intense than the group of peaks responsible for features A–C in that spectrum.

That group of peaks has a low relative intensity in the spectrum from the mixture, and the peak at 1590 cm−1 will have an even smaller contribution to the spectrum from the mixture.

Therefore peak I will be dominated by the contribution from dye 2 and the effect of dye 1 can be taken to be minor.

Finally, peak J can be assigned to HEX.

Table 1 summarises the marker bands for each dye in the spectrum from the mixture.

The volumes and concentrations of the dyes used in the mixture are also listed in Table 1.

A surprising feature of this experiment which is not fully understood is that the chromophore on the DNA is at least one hundred times more sensitive than that of the other dyes.

In part this is likely to be due to the closeness of the excitation frequency to the HEX chromophore.

Thus, dye labelling, at least in suspension of silver particles containing DNA are practical.

The technique is very sensitive and very selective allowing coding and therefore identification of any one particular suspension.

It is likely that when detecting particles by this method in a biological setting that the oligonucleotide will be in a buffer solution.

To investigate the effect of buffer on the SERRS signal, a suspension containing colloid, water, dyes 1 and 2, HEX and spermine was prepared as previously described, and the SERRS signal was recorded.

A second suspension was then prepared in which the water was replaced by the same volume of saline sodium dihydrogen phosphate buffer, with a pH of 7.

Fig. 4 shows the SERRS signals from the two suspensions.

In the spectrum from the sample with water a peak identifying each dye has been marked.

It is obvious that when water is replaced by buffer a strong SERRS signal can still be obtained.

All dyes are still distinguishable, although their relative intensities in the spectrum have changed.

The buffer contains sodium chloride which is an aggregating agent.

The change in relative intensities may be due to the individual response of each dye to the change in aggregation conditions.

For example, sodium chloride reacts with silver to form a silver chloride layer on the surface of the particles.

This may alter the nature of the surface bonding causing a change in ligand orientation or a change in the relative binding strengths of each ligand to the surface.

However, the dye ratio in the suspension could easily be adjusted so that the signals from all of the dyes were of similar intensities.

To ensure that multiplexed labelling occurred on each individual particle, particles were immobilised on a glass slide by drying out the colloid.

With the optical microscope it is difficult to ensure that signals arrive from a single nanoparticle.

Therefore, somewhat larger particles based on silica polymers were produced by depositing silver on the silica particle surface.

A mixture of two dyes was detected from a single particle.

In another experiment, two dyes and a dye labelled oligonucleotide combination was detected from one μm areas of a cluster of microparticles present on the glass surface.

Further work is in progress to fully characterise the individual nanoparticles.

Particles of this nature have very considerable potential for biochemical analysis and gold particles could be similarly developed.

The advantage of silver is that the range of dyes which remain effective in biological media is much more extensive.

However, there is more difficulty on using silver in viable biological systems such as cell suspensions because silver can react positively with the cell and is well known as a bacteriocide.

Thus, the choice of particle system will depend on the nature of the bioassay to be developed.

Recently, SERRS detection of the cystic fibrosis gene was demonstrated at a level which competes with fluorescence.17

The selectivity of SERRS was used to eliminate a separation step in the assessment of the genetic defect.

However, the detection of one genetic defect provides a detection for only 60% of cystic fibrosis cases.

To achieve over 90% detection, eleven different defects require to be detected.

The new multiple labelled particles provide sufficient coding to make this feasible, for example using the new SERRS/microfluidics methods.18

This type of assay works well with silver and the added sensitivity and more complete development of the silver dye chemistry would make silver the ideal choice in this type of in vitro system.

The difficulty with using silver in in vivo systems such as suspensions of viable cells has meant that the coding systems have not been tested in this type of system.

Here, gold would have a significant advantage but it may be that efficient and complete surface coverage by tightly bound ligands may be sufficient for silver to be used.

If it can be used, the advantages in sensitivity would make it the element of choice for this type of analysis as well.

In either event, SERRS has large advantages in that it provides an unparalleled degree of multiplexing and uses a simpler chemistry than fluorescence.