In ultraviolet microbeam irradiations, characteristics of the monochromator and lamp affect the spectral composition of the ultraviolet light and probably the biological results

A recent article by Stonington et al. (1989) presented results from ultraviolet (u.v.) microbeam irradiations of spindles that are different from our earlier results (Sillers and Forer, 1983; Wilson and Forer, 1988; Hughes et al. 1988; review by Forer, 1988). There are two main differences: (1) we found that the reduction of spindle fibre birefringence occurred with two statistically different peaks of sensitivity, one at 260 run and the other at 280 run. Stonington et al. (1989), on the other hand, found only one broad peak, from 260 nm to 280 nm (Fig. 1). (2) We found that the u.v. sensitivities for blocking chromosome movement were different from those for reducing birefringence (Fig. 2); by altering the wavelength and dose we were able to alter birefringence with no effect on chromosome movement, or conversely (Sillers and Forer, 1983; Wilson and Forer, 1988). Stonington et al. (1989), on the other hand, said that they ‘could not detect these subtle differences’.

These conclusions of Stonington et al. (1989) are counter-indicative to my interpretation of how spindle fibre microtubules function (Forer, 1988), which is based in large part on the observation that destruction of spindle fibre microtubules in the irradiated area does not necessarily block chromosome movement. They are also counter to the rationale for experiments in which we try to irradiate with a wavelength that depolymerizes microtubules but has no effect on movement and hence has the minimum effect on other spindle fibre components (e.g. see Wilson and Forer, 1989). Since the conclusions of Stonington et al. (1989) suggest that these two basic assumptions are not valid, it is important to test whether there are technical differences between their experiments and ours that might account for the different conclusions.

With respect to technical differences, the new u.v. microbeam apparatus described by Stonington et al. (1989) is almost exactly the same as the one we described a few years previously (Wilson and Forer, 1987; Hughes et al. 1988), so the apparatus itself is not likely to contribute to differences in results, except for two main differences: (1) we monitored the intensity of every irradiation, whereas they monitored the intensities only ‘on several occasions’, and assumed that the lamp output was constant in the interim. (2) The sources of ultraviolet light were different, as were the monochromator half-band-widths. These two sets of differences could give rise to the different results, as follows.

Because there is considerable variation in lamp output during any one day, often by as much as 20%, and even more variation from day to day, we monitor the energy flux of every irradiation (Sillers and Forer, 1983; Wilson and Forer, 1987; Wilson and Forer, 1988; Hughes et al. 1988). Stonington et al. (1989) did not do this, which could give rise to inaccuracies in estimates of doses.

The use of different u.v. sources and of monochromators with different half-band-widths could also lead to different results. Stonington et al. (1989) used a mercury-arc lamp for their irradiations, whereas we have used mercury-arc lamps, xenon–mercury-arc lamps and xenon-arc lamps. The spectral distribution (the plot of intensity versus wavelength) for ultraviolet light emitted from a xenon–arc lamp is a smooth curve, a ‘continuum’; those from higher-pressure mercury-arc lamps and xenon–mercury-arc lamps have peaks (originally from lines in lower-pressure lamps) at characteristic wavelengths that are superposed on a ‘continuum’ (e.g. see discussion by Sillers and Forer, 1983, and data presented by Noel, 1940, 1941; Baum and Dunkelman, 1950; Elenbaas, 1965; Koller, 1965; Knowles and Burgess, 1984; Coohill, 1985). How would the different spectral distributions be transmitted through monochromators with different half-band-widths?

The entrance and exit slits of monochromators limit the angles of the light entering and leaving those monochromators and thereby define the ‘half-band-width’ (HBW), a measure of the range of wavelengths present in the light rays leaving the monochromator. In all the experiments under discussion the monochromators had equal-sized entrance and exit slits, which results in the transmission curve illustrated in Fig. 3 (e.g. see discussions by West, 1960; Harrison et al. 1948; Knowles and Burgess, 1984; Johns and Rauth, 1965). If, for example, light incident on the monochromator contains all wavelengths in equal intensities, then the peak intensity of the light leaving the monochromator is at the wavelength to which the monochromator is set, there is 50% of the peak intensity at that setting±one-half of the half-band-width, and there is zero intensity at that setting±the half-band-width. If, on the other hand, the light incident on the monochromator has ‘peaks’ (‘lines’), as would be the case using mercury-arc and xenon–mercury-arc lamps, then the situation is different. Consider, for example, a lamp that emits a “line’ at 270 nm, with the peak intensity of the line four times that of the ‘continuum’ at 275 nm, as in Fig. 4 A, and consider how that light would be transmitted through a monochromator of half-band-width 10 nm that is set to 275 nm. If we consider the transmission of the monochromator at 275 nm to be 100%, the monochromator would transmit 100% of the light at wavelength 275 nm, and 50% of the light at 270 nm, but, since there is four times more energy at 270 nm than at 275 nm, the 270 nm light leaving the monochromator will be twice the intensity of the 275 nm light (Fig. 4B), despite the fact that the monochromator was set to 275 nm. Since the lines of the 100 W mercury-arc lamp are often closer together than the 10 nm half-band-width of the monochromator used by Stonington et al. (1989), the monochromator setting may not be an accurate measure of the actual wavelengths in the irradiating beam.

This analysis suggests, then, that technical differences between our work and that of Stonington et al. (1989) could give rise to the different results. To test this ‘theoretical’ analysis, the output of a 100 W mercury-arc lamp was passed through a monochromator of half-band-width 10 nm, the conditions used by Stonington et al. (1989), and, then, the spectral distribution of the light leaving the monochromator was determined by passing the output from the monochromator through a second monochromator with very narrow half-band-width.

To determine the intensity (watts) in the various light beams, a u.v.-100 photocell (United Detector Technology, CA) was used together with a Keithley picoammeter, as described previously (Wilson and Forer, 1987). The sensitivity of the photocell to different wavelengths (i.e. amps generated by the photocell per watt incident on the photocell, as a function of wavelength) was calibrated as described previously (Wilson and Forer, 1987).

Two monochromators were used, either separately or in series. One was an Oriel 7270 (Oriel Corp., Stamford, CT) and the other a PTR ‘MiniChrom-1’ (Optometries, Ayer, MA.). The exit and entrance slits were always the same size on any given monochromator. The half-band-width that is obtained depends on the sizes of the slits relative to the spectral dispersion of the monochromator (nm wavelength per mm slit width); the spectral dispersion values were supplied with the monochromators.

The intensities of light emerging from each monochromator were measured at 1 nm or 2 nm intervals of wavelength settings. The spectral distributions of intensities in the light emitted from the lamps (i.e. incident on the monochromator) were calcuated from the transmissions of each monochromator: I placed the photocell after the monochromator, measured amps, converted amps to watts, and then divided this value by the transmission of the monochromator (at that wavelength) to obtain watts emitted by the lamp at that wavelength.

The transmissions of each monochromator were measured as follows. Each monochromator was set to a half-band-width of 3.6 nm, each was set at the same wavelength, and they were positioned so that the output from one monochromator entered the second. The photocell was used to monitor the intensity first after the first monochromator and then after the second monochromator: the ratio of the two readings is the transmission of the second monochromator. I made this measurement for wavelengths between 258 nm and 320 nm, and used the average of three readings (at three different times) as the transmission at that wavelength for that monochromator. For the readings after the first monochromator, the photocell was placed at the position where the slit of the second monochromator would subsequently be placed, and in front of the photocell I placed a slit that was exactly the same size as the entrance slit of the second monochromator. For the readings after the second monochromator, the photocell (with no covering) was placed right next to the exit slit of the second monochromator: the photocell was larger than the exit slit and intercepted the entire output from the monochromator. In the range of wavelengths from 260 nm to 320 nm the PTR monochromator had transmissions in the range of 7% to 11%, whereas the Oriel monochromator had transmissions in the range of 13% to 16%.

The spectral composition of the light leaving a monochromator was measured by placing in series with that monochromator a second monochromator that had narrower half-band-width than the first (1.2nm compared to ⩾ 3.6nm), and placing the photocell after the second monochromator. I recorded amps incident on the photocell at different wavelength settings on the second monochromator, for each wavelength converted amps to watts, divided each watt value by the transmission of the second monochromator at that wavelength, and then divided each resultant value by the total intensity (all wavelengths added together) to determine for each wavelength the fraction of the total output of the first monochromator that that wavelength comprised. For example, in one experiment the first monochromator was of half-band-width 10 nm and set at 275 nm; the second monochromator was of halfband-width 1.2 nm and readings after the second monochromator were taken at 1 nm intervals from 263 nm to 290 nm. The intensities (watts) at each wavelength were summed and then the individual values were divided by the total.

Mercury-arc and other lamps were from UltraViolet Products (UVP), San Gabriel, California.

The spectral distribution of intensities in the u.v. emissions from a 100 W mercury-arc lamp was measured for two different half-band-widths, 1.3 nm and 3.6 nm (Fig. 5). Increasing the half-band-width of the monochromator ‘smooths’ the peaks due to an increased wavelength spread in the output of the monochromator. The spectral distribution from the xenon-mercury-arc lamp (Fig. 6) also has peaks, but the peaks are often of different wavelengths or relative intensities compared with those from the mercury-arc lamps. The spectral distribution from the xenon-arc lamp is relatively smooth in the u.v. (Fig. 6), as also described by the suppliers and others (e.g. see Baum and Dunkelman, 1950; Calvert and Pitts, 1966; Sillers and Forer, 1983; Knowles and Burgess, 1984).

I calculated the spectral distribution expected in the output from a monochromator when the light arises from a mercury-arc lamp and the monochromator has a halfband-width of 10 nm, the experimental conditions used by Stonington et al. (1989). Assuming that the monochromator was set to 260 nm, the expected output within the range 260nm±10nm was calculated by multiplying the outputs from the lamp at each wavelength (the data of Fig. 5, HBW=1.3nm) by the relative percentage transmissions of the monochromator at each wavelength, assuming that the monochromators were perfect and had transmissions as a function of wavelength as given in Fig. 3. Thus, the output at 260 nm was multiplied by 1, the outputs at 261 nm and 259 nm were multiplied by 0.9, those at 262 nm and 258 nm were multiplied by 0.8, etc.; the values, displayed as a graph illustrating the expected relative intensities of each wavelength in the output (Fig. 7), are quite different from the nominal monochromator setting: the predominant wavelengths were at 263 nm to 266 nm and each wavelength from 262 nm to 268 nm was present with more than twice the intensity of the light at 260 nm, the wavelength the monochromator was set to.

I measured the spectral distributions of light arising from a mercury-arc lamp after it passed through a monochromator with half-band-width 10 nm, for comparison with the calculated values. It is important to note that the measurements are limited by the half-band-width of the second monochromator: a half-band-width of 1.2 nm was used, which is as small as is practical, but even with this the measurements are distorted by peaks and valleys in the spectral distribution. One could imagine, for example, that if the actual output from the first monochromator is moderately high at say 269 nm and zero at 270 nm, the measured spectral distribution would have a considerable reading at 270 nm, because of the finite half-band-width of the second monochromator. Thus, the calculations, if accurate, would give a better assessment of the spectral distribution of light in the output than would the actual measurements. Given that the half-band-width of the second monochromator imposes these limitations on the measurements, the measured spectral distribution for light leaving the first monochromator (Fig. 8) is close to the calculated distribution (Fig. 7), from which I conclude that the calculations are indeed correct, and that calculations such as these can be relied on to give accurate spectral distributions for light leaving a monochromator.

I both calculated and measured the spectral distributions of u.v. emitted by a xenon-mercury-arc lamp after passing through a monochromator set to 260 nm and in which half-band-widths were varied: the measurements agree with the calculations in distribution and shape (Figs 9 and 10), confirming, again, that the calculations can be relied upon to give accurate spectral distributions. The curves also confirm intuitive expectations in showing that the narrower the half-band-width of the first monochromator the more closely the output of that monochromator represents the wavelength to which the monochromator is set.

Because of their relevance to issues raised in the Discussion, I calculated the spectral distributions after light from various lamps had passed through monochromators with various settings (Figs 11-15); these will be considered in the Discussion.

The data show that for any given setting on the monochromator the output from the monochromator depends on the source of light and the half-band-width of the monochromator. Different lamps and half-band-widths were used in different experiments (Table 1); could these have caused the biological results reported by Stonington et al. (1989) to be different from those of Sillers and Forer (1981, 1983) and Hughes et al. (1988)? Let us first consider the comparison between effects on spindle fibre birefringence and effects on chromosome movement.

When spindle fibres in crane-fly spermatocytes were irradiated, birefringence could be reduced while at the same time chromosome movement remained normal (Forer, 1966; Sillers and Forer, 1983). On the other hand, when spindle fibres in PtK cells were irradiated, in no case was birefringence reduced and movement normal (Stonington et al. 1989). There are two important technical differences between these two sets of experiments.

One difference is in the spectral outputs from the monochromators. There was only one wavelength with which crane-fly spindle fibre birefringence could be reduced with no effect on chromosome movement (Sillers and Forer, 1983), namely 260 nm (Fig. 2). The outputs calculated for the two different conditions used in irradiating with 260 run light are illustrated in Fig. 11: the output used by Sillers and Forer (1983) has a peak at 260 nm, whereas that used by Stonington et al. (1989) peaks at 264 nm to 265 nm, not 260 nm, and has less than 5% of its output at 260 nm. The question then becomes: would one expect to see reduced birefringence with no effect on movement after irradiating with the spectral distribution used by Stonington et al. (1989)? There are no data on crane-fly spermatocytes in which effects on birefringence were compared with effects on chromosome movement in irradiations using the spectral distribution used by Stonington et al. (1989), so the question cannot be answered directly. Extrapolating from Fig. 2, however, assuming that the spectral distribution of Stonington et al. (1989) would be equivalent to irradiations with 264-265 nm wavelength light, one would expect movement to be blocked either at a lower dose than required to reduce birefringence or at roughly the same dose. Thus, one would expect only rarely to see birefringence reduced with no effect on velocity under the conditions used by Stonington et al. (1989). A second technical difference is relevant to this same issue, as follows.

In order to reduce birefringence with no effect on chromosome movement the wavelength must alter birefringence at a lower dose than that which blocks movement, and the u.v. must be turned off as soon as birefringence is reduced. In our experience in analysing the videotapes taken during irradiations, an area of reduced birefringence is generally produced at least 2-3 s before we are sure enough of its presence to turn off the u.v. Assuming that we keep the u.v. on for 2–3 s longer than an irradiation time of 20 s, our irradiations are with the minimum dose plus 10–15%; the extra 15% does not block motion because using 260 nm u.v. there is a five-fold difference in sensitivity between reducing birefringence and blocking chromosome movement (Fig. 2). On the other hand, Stonington et al. (1989) irradiated for only 3 s in producing areas of reduced birefringence; in their case, having the u.v. on for 2 s extra corresponds to an increase of 67% above the minimum dose. Since, as discussed above, irradiations with their spectral distribution are expected to reduce birefringence at about the same dose as is needed to block chromosome movement, a 67% increase in dose almost guarantees that both processes would be affected by the same irradiation. This analysis suggests, then, that the difference in monochromator output is the main factor contributing to the apparently different biological result, that chromosome movement is always blocked when birefringence is reduced (Stonington et al. 1989) rather than birefringence being able to be reduced without blocking chromosome motion (Sillers and Forer, 1983).

Let us now consider the other different result, the wavelength sensitivities for reducing spindle fibre birefringence in crane-fly spermatocytes versus PtK cells (Fig. 1). The ‘action spectrum’ (the curve of sensitivity versus wavelength) for reducing spindle fibre birefringence in crane-fly spermatocytes in vivo is the same as that for reducing birefringence and depolymerizing microtubules isolated from marginal bands of newt red blood cells (Hughes et al. 1988), and is close to that for reducing birefringence of spindle fibres in Haemanthus endosperm cells (Forer et al. 1990, and unpublished data). Because of the similar sensitivities in such disparate cells and situations, it would seem most likely that the same action spectrum applies to all microtubules; if so, why are the results presented by Stonington et al. (1989) different?

In looking at the action spectrum for crane-fly spermatocytes (Fig. 1), the points at 270, 280 and 290 nm are relatively close in sensitivity, whereas that at 260 nm is quite different from the other three. Those for PtK cells are all quite close in sensitivity (Fig. 1). There are two important technical differences between the two sets of experiments. One is that different lamp and half-bandwidths were used (Table 1). Irradiations of crane-fly spermatocytes with 260 nm wavelength light were actually with 260 nm, whereas, as discussed previously, those of PtK cells were not (Fig. 11). Further, from the outputs of the irradiations used for determining the action spectrum for crane-fly spermatocytes (Fig. 12) one sees that the irradiations had quite distinct wavelength distributions. For example, irradiations with the monochromator set to 260 nm were of a completely separate set of wavelengths compared with those with the monochromator set to 270 nm. The only ‘overlap’ in the irradiations is from the “tails’ of the spectral distribution when the monochromator was set to 280 nm, and the intensities in the overlapping regions are relatively small compared with the peak intensities at 270 nm and 290 nm. In the irradiations used for determining the action spectrum for PtK cells, on the other hand, there was a great deal of overlap in the different irradiations (Fig. 13). For example, when PtK cells were irradiated with the monochromator set to 270 nm or 280 nm, light of wavelengths 275 nm and 276 nm was present in considerable intensities compared to the peak wavelengths. A consequence of this is that, even if the sensitivities for reducing birefringence were different at the peak wavelengths, the large amounts of other wavelengths in the ranges that overlapped heavily with other monochromator settings would narrow differences in sensitivities between the irradiations with different wavelength settings. A second technical difference could contribute to this homogenization, and that is that in irradiating PtK cells the doses were not measured for each irradiation; as noted in the Introduction, we see variations in dose rate from irradiation to irradiation, and from day to day, so variations in dose rates could have masked differences between different wavelengths. Thus, this analysis suggests that Stonington et al. (1989) did not see a large difference between irradiations with 260 nm and the other wavelengths because their irradiations were not with 260 nm, and that differences between all the wavelengths were greatly narrowed because of overlap in the spectral outputs at the different settings and because of unmeasured variations in dose rates.

It is important to emphasize that these same criticisms apply to some of the data we previously published. With respect to the action spectrum for blocking chromosome movement in crane-fly spermatocytes (Fig. 2), the spectral distributions for the irradiations had considerable overlap (Fig. 14). Though we did detect differences between irradiations with different wavelengths (Sillers and Forer, 1981), we required a large number of cells to do so, and the large amount of overlap in the spectral distributions minimized differences between nominally different wavelengths. Further, because of the overlaps it is not clear which wavelengths in the output are responsible for the biological effects. With respect to the action spectrum for reducing birefringence in crane-fly spermatocytes, those irradiations by Hughes et al. (1988), nominally with 260 nm, in fact were done with a range of wavelengths, with a peak at 263nm to 265nm (Figs 9, 10: HBW=6nm); for that reason they were omitted from Figs 1 and 2. Thus there is no doubt that the accuracy of some of our own data can be improved by irradiating with narrower half-bandwidths and using different sources of u.v.

For the two apparent biological differences discussed above, the ability to reduce birefringence without blocking chromosome movement and the action spectrum for reducing spindle fibre birefringence, I have argued that the apparent biological differences were most likely caused by technical differences. In the general case, the way to decide whether any given difference is for biological or technical reasons is to use the same instrumentation specifications for irradiations of the different biological circumstances, or to scrutinize the technical differences and test to determine if these might account for the apparent biological differences. For the technical differences considered here, one wants to keep the wavelength distributions as narrow as possible, so that the interpretation that there are indeed biological differences is not clouded by technical differences. For example, Spurck et al. (1990) found that after areas of reduced birefringence (ARBs) were formed on spindle fibres in PtK and newt cells the poleward sides of the ARBs were unstable, and rapidly disappeared, unlike those in crane-fly spermatocytes where the poleward side is relatively stable (Wilson and Forer, 1989). This could be a biological difference - that spindle fibres in PtK and newt cells respond differently from those in crane-fly spermatocytes - but there also are important technical differences as well as differences in cell type: the irradiations of Spurck et al. (1990) with nominally 285 nm wavelength u.v. were of the spectral distribution illustrated in Fig. 15, i.e. the irradiations were with a broad range of wavelengths of more-or-less equal intensity. It is quite possible that irradiations with a broad range of wavelengths would give different results from those obtained with irradiations with a narrower range, because the broader range of wavelengths could affect more components than would irradiations with a narrower range. To test if this result is a real biological difference, one needs to rule out technical differences as a cause. Since irradiations of Haemanthus endosperm spindle fibres with narrow ranges of wavelengths also produce ARBs with unstable poleward sides (Forer et al. 1990, and in preparation), this particular difference is most probably a real biological difference, and not due to different wavelength distributions. Thus, in comparing experiments on different cells it is important to be aware of technical as well as biological differences.

The data in this article are also relevant to the issue of how one interprets biological effects of microbeam irradiations. If, for example, one uses heterochromatic irradiation to irradiate until spindle microtubules have gone, one cannot necessarily attribute the biological effects that one observes to the absence of spindle microtubules in the irradiated region, because the irradiations also might affect components in addition to microtubules. Indeed we know from previously published data (Fig. 2) that spindle fibres contain systems with different wavelength sensitivities. One way to minimize the effects on other components is by using narrow halfband-widths (i.e. obtaining as monochromatic a spectral distribution as possible) and irradiating at a wavelength that affects one component but not the other. Irradiating at a peak sensitivity for microtubules would not guarantee that only microtubules and no other components are affected by the irradiation, but at least this would seem to be a good place to start in trying to minimize the effects on other systems that have different wavelength sensitivities. Data in this paper have shown that irradiations with supposedly monochromatic light can actually contain a range of wavelengths; since these could affect several spindle fibre components, it would be difficult to decide which spindle fibre component caused the biological effect in question, or to decide whether the biological events that occur after the depolymerization of microtubules in the irradiated region might in fact result from effects on several components, not only microtubules.

This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada. I am grateful to Barbara Czaban and Julia Swedak for critical readings of several drafts of this article.