In the interpretation of our results we follow the general consensus going back to the original work of Duysens and Sweers (1963) (see Introduction) that oxygenic photosynthetic organisms have two pigment systems, which display different Chl a fluorescence emission spectra, with the ratio of F > 700/F < 710 being larger in PS I compared to PS II. We further assume that under the conditions of our experiments the fluorescence emission at room temperature consists exclusively of contributions from Chl a in PS I, F(I), and Chl a in PS II, F(II). As was shown by Franck et al. (2002) using intact leaves, the relative contribution of F(I) is distinctly higher at emission wavelengths > 700 nm, where it shows a peak around 730 nm, whereas peak emission of F(II) is around 685 nm. For measuring fluorescence signals enriched in F(I) and F(II), respectively, we used the detector filter sets depicted in Fig. 1. While fluorescence above 700 nm (F > 700) was selected by 1 mm of Schott RG9, for selection of fluorescence below 710 nm (F < 710) a Balzers shortpass 710 nm filter was applied. In both cases these filters were protected by 2 mm low-fluorescent RG665 in order to avoid fluorescence of the RG9 and SP710 filters excited by stray pulse-modulated measuring light.
The polyphasic fluorescence rise induced upon the onset of saturating light was measured repetitively with the help of dedicated pre-programmed Script-files (PamWin-3 program) which control the timing between consecutive measurements and the different amplifier gains for the F > 700 and F < 710 recordings. For both signals the same detector was used and, therefore, the optical filters had to be changed manually. F > 700 and F < 710 were measured alternatingly at a constant repetition rate (5 min intervals). Averaging of the separate F > 700 and F < 710 recordings was started when an overall stationary state of the sample was reached and the differences between consecutive recordings of each kind had become negligibly small. Under these conditions the averaged F > 700 and F < 710 signal changes are close to being quasi-simultaneously measured, i.e. reflecting quasi-identical responses of the same sample. In this case, any difference in the F > 700 kinetics compared with the F < 710 kinetics may be considered to indicate differences in the F(I) contribution to the F > 700 and F < 710 responses.
If F(I) would contribute to the dark fluorescence yield (Fo) only, the kinetics of light induced changes of variable fluorescence yield should be equal in F > 700 and F < 710. On first inspection of the original raw data, this indeed seems to be the case. Figure 2 shows the screenshot of original F > 700 (red) and F < 710 (blue) recordings of the polyphasic rise of fluorescence yield induced by strong continuous light, as measured with a highly dilute suspension of Chlorella (200 µg Chl l−1). However, as will be shown below, closer inspection afterall reveals significant reproducible differences which argue in favour of variable fluorescence yield of PS I in vivo, Fv(I).
For a thorough investigation of the question of whether F(I) contributes to the variable fluorescence induced upon illumination, appropriate scaling of the F > 700 responses for comparison with the F < 710 responses is essential. In this context, it is important that the O-I1 transient of the polyphasic fluorescence rise may be considered a specific PS II response which specifically reflects the closure of PS II reaction centers. Under the given conditions of illumination, this so-called “photochemical phase” is completed at 1 ms. The underlying physiological responses are identical, irrespectively of whether they are measured at F > 700 or F < 710. Hence, it makes sense to rescale the F > 700 curve to display an O-I1 amplitude that equals the O-I1 amplitude in the F < 710 curve. After such O-I1 equalization, not only the O-I1 rise, but also any other PS II response should be reflected with equal amplitudes in the F > 700 and F < 710 signal changes. Any increase of F(I) should be reflected by a larger rise of F > 700 compared to F < 710 (for further details on O-I1 equalization, see Material and methods).
In Fig. 3 the O-I1 equalized F > 700 curve is compared with the F < 710 curve. After export to Excel and before equalization, small unavoidable constant blank signals were subtracted, as determined for F > 700 and F < 710 with the cuvette being filled with the BG11 suspension medium. Comparison of the O-I1 equalized F > 700 and F < 710 responses in Fig. 3a,b reveals two small but clear-cut differences. First, the dark fluorescence yield FO > 700 is higher than FO < 710. Second, the amplitude of I2-P > 700 is larger than that of I2-P < 710. While the former is not unexpected in view of previous work (Peterson et al. 2001; Franck et al. 2002), the latter may be considered a new finding.
The differences in variable fluorescence are more clearly apparent in Fig. 4, in which the O-I1 equalized Fv > 700 and Fv < 710 curves are compared, i.e. after subtraction of the respective FO values.
After equalization of the O-I1 amplitudes, the kinetics of F > 700 and F < 710 are close to identical, except for the I2-P part of the polyphasic fluorescence rise. The distinctly higher amplitude of I2-P in F > 700 supports the notion that I2-P reflects variable PS I fluorescence, Fv(I) (Schreiber et al. 1989). The difference between the O-I1 equalized Fv curves, which may be considered to specifically reflect the light-induced changes of F(I), i.e. Fv(I), during the course of the polyphasic fluorescence rise, is depicted in Fig. 5.
The light-induced increase of PS I fluorescence depicted in Fig. 5a, b shows a lag phase up to about 10 ms and a peak around 150 ms, followed by a decline. The latter may be assumed to reflect light activation of electron transport at the PS I acceptor side. The 10 ms lag phase is clearly apparent in panel b with logarithmic time scale. To our knowledge this is the first unequivocal experimental evidence for rapid changes of PS I fluorescence under in vivo conditions. Ikegami (1976) previously reported on changes of PS I fluorescence yield in P700-enriched particles isolated from spinach chloroplasts. In agreement with Ikegami (1976) we believe that the observed PS I fluorescence changes are controlled by the redox states of both primary PS I donor and acceptor, which both act as quenchers in the oxidized forms. Recent simultaneous measurements of dark–light induction kinetics of chlorophyll fluorescence, P700 and ferredoxin (Fd) in intact leaves suggested that the I2-P phase indeed correlates with the reduction of both P700 and Fd (Klughammer and Schreiber 2016).
The properties of the observed Fv(I) are impressively similar to those theoretically derived by Lazar (2013) by in silico simulations based on his new PS I model combined with his older PS II model (Lazar 2003) (see Introduction).
Suppression of I
2-P and Fv(I)
It has been known for some time that the amplitude of the I2-P phase is maximal after thorough dark-adaptation and becomes suppressed at relatively low quantum flux densities of background illumination (Schreiber et al. 1995a; Schansker et al. 2006). Figure 6 shows the result of polyphasic rise measurements of F > 700 and F < 710 (carried out with the same Chlorella suspension as used for the measurements in Figs. 2, 3, 4 and 5) after 2 h continuous illumination at 96 µmol m−2 s−1. In the given stationary state of illumination, the sample is sufficiently stable to allow repetitive alternating measurements of F > 700 and F < 710 with 1 min intervals, so that a high signal/noise ratio can be reached by averaging of alternatingly measured responses for quantitative comparison of the kinetics.
The data in Fig. 6 can be directly compared with the corresponding data in Fig. 4. Notably, in the illuminated state the F > 700 and F < 710 kinetics are practically identical. In both curves, the I2-P phases have disappeared, which means that there is no indication of Fv(I) anymore. The clarity of the outcome of this experiment argues for the reliability of the chosen approach for identification and quantification of Fv(I).
The I2-P phase can also be selectively suppressed by addition of the artificial PSI acceptor methyl viologen (Neubauer and Schreiber 1987; Schansker et al. 2005). Measurements of F > 700 and F < 710 analogous to those of Figs. 2, 3 and 4 in the presence of 1 mM methyl viologen resulted in similar responses as shown for the stationary light state in Fig. 6, i.e. the O-I1 equalized changes of Fv > 700 and Fv < 710 were similar, there was no I2-P phase and, hence, no indication of Fv(I).
It is a common feature of methyl viologen application and stationary illumination that both treatments open the “bottle neck” at the PS I acceptor side, which develops upon dark-inactivation of the reactions downstream of Fd. Analogous to the enhancement of the PSII fluorescence yield by the reduction of QA at the acceptor side of PS II, PS I fluorescence is supposed to be stimulated upon accumulation of reduced Fd (Lazar 2013; Klughammer and Schreiber 2016). In addition, as previously suggested by Ikegami (1976), PS I fluorescence yield may be also controlled by the redox state of P700, which in its oxidized form quenches the excitation energy. Upon a sudden dark–light transition the initial reduction of Fd goes hand in hand with the oxidation of P700, so that initially no increase of Fv(I) is expected. Whether this will occur or not, depends on the relative rates of Fd reoxidation and P700 re-reduction. The latter can be prevented by the PQ analogue dibromothymoquinone (DBMIB), which not only blocks the reduction of P700 by electrons that arrive from PS II, but also the reduction of P700 by cyclic flow (Trebst 2007). Hence, I2-P should be eliminated by DBMIB and with it also any increase of Fv(I). As shown in Fig. 7, this is indeed the case. The O-I1 equalized Fv > 700 and Fv < 710 curves are practically identical. The curves in Fig. 7 were measured with the same sample as those in Figs. 4 and 6, after addition of DBMIB. While it is tempting to compare the P-levels, in order to decide whether the suppression of I2-P is due to a decrease of P or an increase of I2, we prefer not to draw definite conclusions from these data, as DBMIB is known to quench Chl a fluorescence in its oxidized form. Without knowledge on the extent of such quenching under the given conditions, for the time being we can conclude only that the O-I1 equalized F > 700 and F < 710 curves are practically identical in the presence of DBMIB, i.e. that DBMIB has eliminated the Fv(I) apparent in the control sample (Fig. 4).
Deconvolution of the F(I) and F(II) contributions
In view of the evidence presented above, it appears justified to assume that the I2-P transient is due to Fv(I), which contributes more to F > 700 than to F < 710. Based on this assumption, returning to the data of Fig. 3, it is possible to deconvolute the O-I1 equalized F > 700 response into the respective contributions of F(I) and F(II), i.e. Fo(I), Fo(II), Fv(I) and Fv(II). Deconvolution involves the following steps:
-
(1)
Equalization of the O-I1 amplitudes of F > 700 and F < 710
-
(2)
Subtraction of F < 710 from the O-I1 equalized F > 700; the amplitude in the resulting F(I) response (as depicted above in Fig. 5) is smaller than that contained in the F > 700 curve, as it was diminished by subtraction of the F(I) contained in the F < 710 curve.
-
(3)
Rescaling of the F(I) response to give the same I2-P amplitude as in the F > 700 curve. The resulting response constitutes the contribution of F(I) to the O-I1 equalized F > 700 curve.
-
(4)
The complementary contribution of F(II) to the O-I1 equalized F > 700 signal is obtained by subtraction of the F(I) contribution: F(II) = F > 700–F(I)
For further details see Section on "Deconvolution of F(I) and F(II)" under Materials and methods.
The result of deconvolution of the data displayed in Fig. 3 is presented in Fig. 8.
In the given example the deconvolution suggests a 37% contribution of Fo(I) to the Fo > 700 in Chlorella, which is almost identical to the value determined by Franck et al. (2002) at the 722 nm maximum of F(I) emission in barley. Fv(I) contributes 14% to the total Fv > 700 in Chlorella under the given conditions, which agrees with the theoretically derived 8–17% reported by Lazar (2013). The deconvoluted F(II) response is characterized by an Fv/Fm (II) value of 0.75, which is distinctly higher than the apparent Fv/Fm = 0.69 in the original F > 700 response. For comparison, the apparent Fv/Fm value in the original F < 710 response (see Fig. 3a) amounted to 0.70.
Fv(I) in state 2 of Synechococcus leopoliensis
For the above presentation of evidence for Fv(I) in vivo the model system of a dilute suspension of Chlorella was chosen in spite of the fact that the I2-P phase, i.e. the suggested “indicator” of Fv(I), is relatively small in this organism. Decisive advantages of this model system are the absence of light intensity gradients, the stability of the continuously stirred sample over many hours and the excellent reproducibility of the light induced responses. In principle, these advantages also apply for measurements with suspensions of cyanobacteria. However, reliable measurements and interpretation of light induced chlorophyll fluorescence changes in cyanobacteria are more demanding (Campbell et al. 1998; Stirbet et al. 2019). Cyanobacteria display pronounced reversible state 1 < – > state 2 transitions (Mullineaux and Emilyn-Jones (2005). After prolonged dark-acclimation state 2 is formed, characterized by rather low values of apparent Fv/Fo. So far few measurements of rapid dark–light induction kinetics in the sub-s and sub-ms time ranges of cyanobacteria have been reported and to our knowledge no previous attempts were made to compare F > 700 and F < 710 after O-I1 equalization. Due to its outstanding sensitivity and flexibility in terms of excitation and emission wavelengths, the Multi-Color-PAM is ideally suited for such measurements.
Figure 9 shows the result of measurements with Synechococcus leopoliensis in the dark state 2 analogous to the above measurements with Chlorella. Notably, following O-I1 equalization the amplitude of the I2-P phase is clearly higher with F > 700 compared to F < 710, similarly and even more pronounced as with Chlorella, thus impressively confirming the existence of Fv(I) also in cyanobacteria.
The 440 nm pulse-modulated measuring light (ML) used in the experiment of Fig. 9, directly excites Chl a of PS I and PS II, located within the thylakoid membrane. In this way, excitation of phycobiliprotein fluorescence is avoided. Hence, it may be assumed that the Fo values of the O-I1 equalized polyphasic rise curves of F > 700 and F < 710 in Fig. 9a are composed of Fo(I) and Fo(II) only. The amplitude of Fo > 700 is close to twice that of Fo < 710 (factor 1.935). Notably, this is also true for the amplitudes of I2-P > 700 and I2-P < 710 (factor 1.985).
The polyphasic rise kinetics of Synechococcus leopoliensis in the dark state 2 differs considerably from the kinetics measured with Chlorella (see Figs. 2, 3 and 4). In particular, there is hardly any I1-I2 phase. Actually, this is not surprising considering that the PQ pool in cyanobacteria becomes readily reduced in the dark and that the I1-I2 phase normally is paralleled by the light driven reduction of the PQ pool. When briefly before onset of actinic illumination a strong pulse of far-red light is given, the I1 level is lowered and an I1-I2 rise similar to the one in Chlorella is recorded (not shown).
As described above in Fig. 8 for Chlorella, in Fig. 10 the deconvoluted F(I) and F(II) signal changes are presented that are contained in the light-induced polyphasic F > 700 rise kinetics measured with Synechococcus leopoliensis. Notably, the F(I) changes are restricted to the I2-P part of the curve, whereas the by far largest part of the apparent F(II) changes occurs during the O-I1 part of the curve. The overwhelming part of Fo > 700 consists of F(I). Under the given conditions, Fo(I) exceeds Fo(II) by a factor of 23.8. This is a consequence of dark state 2, in which distribution of excitation energy to PS I is favored. Furthermore, cyanobacteria display substantially higher PS I: PS II ratios (see e.g. Stirbet et al. 2019). Wang et al. (1977) reported that only about 15% of total Chl a is associated with PS II in Anacystis nidulans (former name of Synechococcus leopoliensis). In spite of the extremely low Fo(II), the Fv(II) component, which is mostly due to the O-I1 rise, is larger than the Fv(I) component by a factor of 1.4. Consequently, a rather large Fv/Fm(II) of 0.88 results, whereas the Fv/Fm (I) amounts to not more than 0.18. Hence, the notoriously low values of apparent Fv/Fm that have been observed in previous work with cyanobacteria (see e.g. Badger and Schreiber 1993; Campbell et al. 1998; Stirbet et al. 2019) are mostly due to a large contribution of Fo(I), particularly when 440 nm excitation is applied, like in the present study. The most surprising new finding, however, is the observed almost 40% contribution of Fv(I) to overall Fv in Synechococcus in the dark state 2. This finding was confirmed and extended by numerous further measurements with cyanobacteria, presentation of which would go beyond the scope of the present communication.
Apparent Fv(I) from analogous measurements with leaves
Analogous measurements with leaves encounter two major problems, both of which are caused by the much higher optical density compared to that of the highly dilute suspensions of Chlorella and Synechococcus in the above measurements. First, due to the much higher chlorophyll content the PS II emission, which peaks around 685 nm, is strongly reabsorbed and, therefore, F(II) is rather low. In principle, this problem can be overcome by signal averaging. The second problem is more serious, as it unavoidably means heterogeneous origins of the F > 700 and F < 710 responses: As F > 700 is much less reabsorbed than F < 710, in the mean it originates from deeper cell layers than F < 710, where the effective quantum flux density is lower than at the leaf surface, from where most of the measured F < 710 originates. Hence, in leaves F > 700 and F < 710 report on heterogeneous populations of chloroplasts that not only “see” different actinic light intensities during the measurements of the polyphasic rise kinetics, but also have developed under different light conditions and may display different physiological properties. This problem can be minimized by the use of light-green young leaves and by applying strongly absorbed 440 nm pulse-modulated measuring light, most of which is absorbed in the uppermost cell layers.
Figure 11 shows the result of analogous measurements as carried out above for dilute suspensions using a light-green young ivy leaf (Hedera helix). For this purpose a special leaf holder was applied that was developed for fluorescence measurements from leaf surfaces with the Multi-Color-PAM (see Sect. "Materials and methods").
The ivy data in Fig. 11a may be compared with the corresponding data for Chlorella in Fig. 4. Not unexpectedly, the O-I1 rise in F > 700 is somewhat slower than that in F < 710, due to the somewhat lower mean effective quantum flux density. By application of a saturating single turnover flash at 1 ms in both curves the I1 level is determined, so that O-I1 equalization can be carried out (see Section on "Rescaling for comparison of F > 700 and F < 710 data" under Materials and methods). As in the case of Chlorella and Synechococcus leopoliensis, the amplitude of the I2-P phase is distinctly larger in the O-I1 equalized F > 700 curve, thus suggesting that also in leaves the I2-P phase is associated with Fv(I). We note that after O-I1 equalization the I1-I2 phases of the F > 700 and F < 710 responses are practically identical, which argues for the correctness of the rescaling procedure. At the applied high actinic intensity, in contrast to the photochemical O-I1 phase, the rate of the “thermal” I1-I2 phase is independent of light intensity.
In view of the unavoidable problems outlined above for comparative measurements of F > 700 and F < 710 in leaves, the deconvolution of the F(I) and F(II) components in Hedera helix presented in Fig. 11b should be considered tentative. While the F(I) dip in the O-I1 range is a trivial consequence of the lower effective PAR, at the present stage it cannot be excluded that also other kinetic details result from distortions caused by optical or physiological heterogeneities.
We have carried out analogous measurements with a variety of C3 and C4 leaf species, all of which displayed a larger amplitude of I2-P in F > 700 compared to F < 710. However, as the optical densities were generally higher than in the case of the light green ivy leaf (Fig. 11), reliable O-I1 equalization was problematic.