The Performance of the OLIS RSM 1000F in Three of its Fluorescence Modes

The first Olis RSM 1000F was built in response to the challenge: "If you can build an absorbance spectrophotometer that scans 1,000 times per second, can't you build a fluorimeter which does, too?" Thus, everything in the instrument was chosen to facilitate the acquisition of 1000 emission scans per second. As such, the user can expect an RSM fluorimeter to provide the greatest advantage when operating in this regime: the observation of samples which change rapidly. For this type of measurement, one cannot compare the performance of the RSM to other instruments, because there are no competing products on the market.

However, Olis is also well known for developing truly modular instruments, which can be used for a range of applications. We have, over the years, added additional modes of operation to the Olis RSM 1000F, including absolutely traditional emission scanning with superb sensitivity, fast and submicrosecond fixed wavelength work.

1. Rapid-scanning: 1,000 emission spectral scans per second, best for 'easy' readings and for kinetically changing samples
   
2. Conventional scanning: wavelength-by-wavelength scanning, achieved by (a) signal averaging during millisecond scanning or (b) signal averaging during microsecond reading accompanied by changing wavelengths (scanning).
3. Fixed wavelength: microsecond read rates for monitoring very fast kinetic processes using photomultiplier tube.

Quoting "sensitivity" is not trivial; one must consider:

• concentration of fluorophore
• quantum efficiency of fluorophore
• extinction coefficient of fluorophore
• excitation and emission wavelengths
• bandwidths of excitation and emission
• grating blaze wavelength/groove density
• age and intensity of excitation source
• presence or absence of polarizers, and

Total Collection Time and How That Time Is Used

This document focuses on the final, and perhaps most critical factor: How long do we have to make the measurement and how are we using that time? Knowing this, one can extrapolate with the additional factors, the answer to "What is the sensitivity of the OLIS RSM 1000F?"

I. How sensitive is the RSM when used as a rapid scanning fluorimeter?

A. Easy: a One Second Scan

In one second, the RSM can collect the emission spectrum of ~35 nM fluorescein with an RMS S/N ratio of ~120 (see Figure 1). How does this compare to a conventional instrument in terms of sensitivity? On this time scale, the RSM must be regarded as infinitely more sensitive, since the traditional fluorimeter will produce no result. That this single scan is actually the fitted result of 1,000 is incidental if your sample is not changing; it is fantastic, if you are also hoping to capture kinetic information.

Figure 1: Emission of 36 nM fluorescein in tap water. Excitation bandpass — 10 nm; emission bandpass = 5 nm; 400 1/mm grating. Scan rate of 1,000 scans per second, collected and averaged to single scan. If sample had been changing, one millisecond emission scans would have been the kinetic 'points' available for finding rate constant. When sample is not changing, rate constant is zero.

B. Faster: Using One Millisecond Scans for a 250 ms Protein Folding Study, as during a stopped-flow experiment

Consider the following data which monitor the folding of T4 lysozyme. The instrument acquires full spectral scans of the tryptophan emission at a time resolution of 1 millisecond (see Figures 2-4). The entire experiment takes ~250 ms.

Figure 2: Scans of tryptophan fluorescence of folding T4 lysozyme taken over 273 ms. (scan rate = 1000 s ) Click on image to view at a better resolution.

Figure 3: Eigenvector analysis of lysozyme folding data. The first two eigenvectors show two changes occurring. Eigenvectors for 2 species suggest an AB mechanism. A single rate fit produces the 'answer' - from A ( max = 345) changing to form B ( max = 330).

Figure 4: Global fit results for lysozyme folding data. This analysis procedure not only rates (~18.5 s), but determines the spectra of the folded species as well. Note also the high signal to noise ratio of the results as compared to the raw data in Figure 2.

It is now appropriate to compare the sensitivity of the RSM fluorimeter to conventional stopped-flow fluorimeters which are often used to observe kinetics of this type. (Comparison to conventional scanning fluorimeters remains entirely pointless.) Conventional stopped-flow fluorimeters acquire single wavelength or total emission data; meanwhile, the OLIS RSM fluorimeter outfitted with a stopped-flow mixing unit is acquiring hundreds of wavelengths of spectral information for each shot.

Looking at any particular one millisecond spectral scan that the RSM collects and attempting to draw a comparison between its signal to noise ratio and that of a fixed wavelength kinetic trace acquired over hundreds of milliseconds is illogical. Looking at a single millisecond emission scan is like looking at a single fixed wavelength data point: it is raw data isolated from other raw data collected during a single measurement. Just as one looks at fixed wavelength points in a fixed wavelength trace as a body to be fitted with an appropriate equation that will return the rate constants, so should one look at millisecond spectral scans as an ensemble to be fitted with an equation that will return the rate constants.

Except — and this is the huge except! — we ALSO have spectral information. We can reconstruct the starting spectrum, any intermediate spectra, and the final spectrum(a) and know what was undergoing thekinetic change.

Any one of the 250 one millisecond spectral scans acquired by the RSM during the above protein folding event is quite noisy. But after Global Analysis of the entire ensemble of scans, one has spectra with good signal-to-noise ratios of both the folded and unfolded species, which differ by a wavelength shift as well as amplitude change.

These definitive results — knowledge of the rate constants combined with knowledge of the spectral shapes — could be obtained from a fixed-wavelength stopped-flow instrument only at the expense of many shots and, necessarily, far more sample and more opportunities for systematic error.

It is this combination of the rapid spectral scanning and Global Analysis which accounts for the sensitivity of the RSM during fast kinetic measurements.

II & III: Using the RSM in Static or Very Slow measurements

There are two options: Utilizing the millisecond spectral scan rate or utilizing the 'conventional' scan rate.

A. We will consider the rapid-scanning option first.

As we have seen, the RSM produces 1,000 spectral scans per second. For a scan lasting 250 seconds, a conventional fluorimeter might make one spectral scan. The RSM, in its rapid scanning mode, meanwhile collects and averages 250,000 scans.

This being said, it remains to establish a means of comparing the performance of the RSM in rapid scanning mode to a conventional fluorimeter in the task of acquiring static fluorescence emission spectra. It has been suggested that comparison of the rms S/N of the Raman scattering peak of water provides an objective means of measuring the relative sensitivity of conventional slow-scanning spectrofluorimeters. The validity of this test requires the conditions be specified thoroughly: excitation bandwidth = 5 nm; emission bandwidth = 5 nm; excitation wavelength = 350 nm; integration time = 1 second. It is possible to configure an RSM fluorimeter to approximate these specifications. Since the RSM is completely digital and has no time constant settings, we instruct the instrument to collect for 1 second for each wavelength point in the spectrum. Under these conditions, the RSM will measure the Raman band of water with an rms S/N of ~250. (see Figure 5).

 

Figure 5: Raman scattering peak of distilled water obtained from Global Analysis of 1300 scans collected over 250 second. Signal-to-noise ratio (rms) = 263.

B. Next, conventional scanning.

Below is a fluorescence emission spectrum of ovalene in a PMMA matrix (Starna standard #2). The concentration of fluorophore is 200 nM. Spectrum acquired in conventional scanning mode with analog PMT detector. Instrument settings are as follows: excitation wavelength 342 nm, excitation bandpass = 8 nm, emission bandpass = 3.8 nm. Time per data point is 1 second; time for scan is 160 seconds. High volts on the detector is 576 volts.

Those familiar with high-end research spectrofluorimeters will recognize that for this sort of measurement, the RSM in rapid scanning mode is less sensitive. Two things should be noted, however. First, this performance is still respectable and will be adequate for all but the very weakest signals. Second, this performance demonstrates the remarkable versatility of the RSM: in rapid-scanning mode, the RSM can provide results in many time domains, whereas conventional scanning fluorimeters are useful for slow measurements only.

Moving to Conventional Operation

To this point, we have considered data acquired in a rapid-scanning mode. However, the RSM is built around a monochromator which is much like any other. Furthermore, the specialized photodetector systems used in the RSM can be replaced with other types of detectors. The point of this is that, if desired, the RSM can be operated in the same fashion as more conventional instruments.

In detail, replacement of the Scandisk [the cartridge in which the rapid-scanning slit disk is housed] with a fixed position intermediate slit and the replacement of fast response PMTs with slow response and high-gain photomultiplier tubes or a photon counting detector will convert the RSM system to a conventional scanning fluorimeter with all the advantages and disadvantages such an instrument entails.

Similarly, a user performing stopped-flow fluorescence emission experiments may wish to operate in total-emission mode (as in conventional instruments) rather than rapid scanning. Using an RSM system in this way involves nothing more than repositioning a photomultiplier tube (detector) with an appropriate optical filter in place.

It should be remembered that the option of rapid scanning should be chosen whenever experimental conditions permit; taking advantage of the extra information in multiwavelength data combined with the analytical power of global fitting will always provide better results than if this information is unavailable or ignored.