Studies on the primary productivity of phytoplankton have revealed that iron (Fe) limitation is the major factor controlling phytoplankton growth in vast regions of the contemporary oceans (“iron hypothesis”), most notably in the high nutrient low chlorophyll regions (HNLC) (Martin and Fitzwater, 1988; Boyd et al., 2000; de Baar and Boyd, 2000; Dalton, 2002; Gervais et al., 2002; Moore et al., 2007). Fe is an essential micronutrient for phytoplankton because it is a cofactor of proteins directly involved in photosynthesis, respiration, nitrate, nitrite and sulfate reduction, N₂ fixation, chlorophyll synthesis, and a number of other biosynthetic or degradative reactions (Geider et al., 1993; Geider and La Roche, 1994; Morel and Price, 2003). Fe plays a crucial role as component of different photosynthetic electron carriers as cytochrome (Cyt) b₆f and Cyt c₆ and iron–sulfur complexes and as an integral part of both photosystem I (PSI) and photosystem II (PSII; Greene et al., 1991, 1992; McKay et al., 1997; Erdner and Anderson, 1999).

Iron deficiency causes important alterations in thylakoid membrane structure and the basic processes involved in photochemical energy conversion. A general decrease in transcripts associated with photosynthesis has been shown under this stress condition (Allen et al., 2008). Chlorosis is one of the most important symptoms of Fe deficiency (Greene et al., 1991; Davey and Geider, 2001; Moseley et al., 2002; Allen et al., 2008). It is attributed to the inhibition of the chlorophyll (Chl) biosynthetic pathway, which requires the function of iron-containing enzymes (van Leeuwe and Stefels, 1998, 2007). Disconnection between light-harvesting centers as a consequence of ultrastructure changes of the thylakoid membranes under Fe-limited conditions (Meisch et al., 1980; Terry and Abadía, 1986; Hilt et al., 1987) has been described as responsible for a decline of PSII efficiency, electron transport and carbon fixation (Raven, 1990; Behrenfeld and Milligan, 2012; Petrou et al., 2014). Reduced photosynthetic efficiency due to loss of the D1 protein (Greene et al., 1992), reduced concentration of Cyt b₆f and Cyt c₆ and decreasing Cyt f:Chl a, P700:Chl a, and PSI:PSII ratios (Greene et al., 1991; Allen et al., 2008) have been observed in Fe-limited continuous cultures of the diatom P. tricornutum.

Diatoms (Bacillariophyceae) are the most important eukaryotic group of phytoplankton (Guiry, 2012) with a key role in influencing the global climate since they are responsible for up to 20% of the global primary productivity and 40% of the carbon sequestration in the oceans (Treguer et al., 1995; Field et al., 1998; Bowler et al., 2010; Falkowski and Raven, 2013). Fe fertilization experiments in HNLC regions have shown the appearance of blooms dominated by diatoms, suggesting that these algae have adaptations that allow survival in Fe limited waters and a subsequent rapid multiplication when Fe becomes available (Morrissey and Bowler, 2012). Different strategies have been developed by diatoms to minimize their Fe requirements: a decrease of the cellular pigment concentrations at the cost of light capture efficiency (Petrou et al., 2011), biochemical alteration of the photosynthetic Fe demand through decreased expression of the Fe-rich PSI and Cyt b₆f components (Strzepek and Harrison, 2004; Allen et al., 2008) and/or substitute Fe-containing enzymes as ferredoxin by proteins with Fe-free equivalents, as flavodoxin (La Roche et al., 1996; Marchetti et al., 2009). Diatoms have also developed specific regulatory mechanisms to dissipate energy excess under environmental stress conditions (Grouneva et al., 2009; Goss and Jakob, 2010; Lavaud and Goss, 2014).

In this study, we have investigated the photosynthetic response of the pennate marine diatom P. tricornutum to Fe deficiency, using thermoluminescence (TL), Chl fluorescence, and P700 redox state measurements. These techniques are very simple, precise and non-destructive, and provide valuable in vivo measurements of the effects of environmental perturbations on PSII and PSI activity (Ducruet, 2003; Allen et al., 2008; Kalaji et al., 2014b). Chl a fluorescence technique has been extensively used for studying the effects of different environmental stresses on photosynthesis (Maxwell and Johnson, 2000; Kalaji et al., 2011, 2014b; Goltsev et al., 2012).

Phaeodactylum tricornutum is highly tolerant to Fe limitation and can grow in steady-state laboratory cultures at Fe levels 50 times lower than those tolerated by others diatoms (Kustka et al., 2007). Studies on this diatom have shown that it is also unusually resistant to damage by exposure to high light intensities (Olaizola et al., 1994). Moreover, this organism has a xanthophyll-dependent non-photochemical quenching (NPQ) that is induced more rapidly, and can compete with excitation transfer to the PSII reaction center (RC), much more efficiently than does the analogous process in higher plants (Lavaud et al., 2002a). In addition, P. tricornutum may be able to short-circuit its PSII RC by a cyclic electron transfer path when the charge separation cannot be stabilized by normal secondary electron transport (Lavaud et al., 2002b).

Thermoluminescence provides an in vivo measure of the response of PSII activity to environmental stresses (Rahoutei et al., 1990; Briantais et al., 1992; Misra et al., 1997; Walters and Johnson, 1997; Roman and Ducruet, 2000). Photosynthetic luminescence is a process that originates from PSII by recombination of charge pairs separated by a prior irradiation. Luminescence decay phases can be better resolved by TL emission technique, which consists in recording luminescence emission during the warming of a sample after an irradiation given at a relatively low temperature (for a review, see Vass and Inoue, 1992; Inoue, 1996; Ducruet, 2003). Therefore, the properties of the two principal TL emission bands, B band and AG band (Ducruet, 2003), may be used to obtain information on the effect of Fe deficiency on the photochemical activity of PSII. The B band is the result of the recombination of S₂/S₃Q_B^- pairs, Q_B being the secondary quinone acceptor, and S₂/S₃ being the states of the oxygen-evolving complex (OEC) storing two or three positive charges (Rutherford et al., 1984; Vass and Inoue, 1992). The AG band, although originating from PSII, is governed by the electron back-transfer from the stroma to Q_B, which requires (i) a sufficient potential gap between the acceptors side of PSI (NADPH/NADP) and PSII (PQH₂/PQ); (ii) an activated chlororespiratory pathway involving both non-photochemical reduction and oxidation of plastoquinones (PQs) (Bennoun, 1982; Sundblad et al., 1988; Rumeau et al., 2007).

Chlororespiration has been defined as a respiratory electron transport chain in interaction with the photosynthetic electron transfer in thylakoid membranes of chloroplasts. It involves mostly a NAD(P)H-PQ oxidoreductase activity (Ndh activity), the thylakoid PQ pool and a terminal oxidase named PTOX (Peltier and Cournac, 2002). A chlororespiratory reduction of the PQ pool has been found in diatoms and algae (Wilhelm and Duval, 1990; Dijkman and Kroon, 2002) leading to the build-up of a proton gradient without the participation of PSII electron transport (Jakob et al., 1999, 2001). Several studies have also proposed that chlororespiratory components may be involved in protective or adaptive mechanisms of photosynthetic organisms to environmental stress conditions (Rumeau et al., 2007).

In this work, a combination of TL, Chl fluorescence and P700 (PSI primary donor) absorbance measurements have been used to elucidate the effects of Fe deficiency in the photosynthetic electron transport activity of P. tricornutum. The results obtained in this work suggest that Fe deficiency induces the partial blocking of electron transfer from PSII to PSI, and consequently, leads to a more reduced state of the PQ pool. This blocking is likely to be due to an important reduction of the amount of Cyt c₆. Fe deficiency induced also a significant increase of the light sensitivity of PSII. The possible activation under low Fe concentration of alternative secondary electron transfer pathways, as chlororespiration, is discussed.

The effects of Fe deficiency on various physiological and biochemical parameters of P. tricornutum cells have been investigated. Table 1 shows the results obtained for these parameters in cells harvested after 21 days of growing in both Fe-replete and Fe-deficient conditions under our experimental conditions. After 1 week of Fe limitation, growth gradually slowed down (see Supplementary Figure S1). A significant decrease of the growth rates in Fe-deficient P. tricornutum cells was observed after 21 days of cultures in comparison with cells cultured in Fe-replete conditions (Table 1). Fe-deficient P. tricornutum cells showed lower concentration of Chl a (56%). Fe deficiency also induced a significant decrease of about 60 and 80% in the oxygen evolving and respiration activities of the cells, respectively. Overall, the effects observed under Fe limitation are consistent with those described previously (Kudo et al., 2000; Allen et al., 2008).

The effects of Fe deficiency on PSII photochemistry of P. tricornutum cells were investigated using Chl a fluorescence and TL techniques. Measurements of Chl a fluorescence showed clear differences in the photosynthetic activity of PSII between Fe-replete and Fe-deficient cells (Table 1 and Figure 1). The maximum quantum yield of PSII, measured as F_v/F_m, was significantly decreased in Fe-deficient cells. Thus, whereas Fe-replete cultures showed an F_v/F_m value of 0.615, the value of the Fe-deficient cultures decreased to 0.404 (Table 1). Steady-state light curves showed that the effective quantum yield of PSII, Y(II), was lower in Fe-deficient cultures at all irradiance levels tested as compared with Fe-replete cultures (Figure 1A). At the maximum irradiance (1957 μE m^-2 s^-1), values dropped well below 0.1 under both Fe-culture conditions. The relative electron transport rates (rETR) were significantly greater in the Fe-replete cultures than those measured in the Fe-deficient cultures in the complete range of tested irradiances (Figure 1B). The maximal rETR was 50% lower in the Fe-deficient cultures. Taddei et al. (2016) have recently reported the severe decrease of both F_v/F_m (70%) and rETR (64%) parameters induced by Fe limitation in P. tricornutum cells. In Fe-deficient cells rETR was almost completely inhibited at light intensities above 2000 μE m^-2 s^-1 (Figure 1B). However, in Fe-replete cultures the rETR value measured at this light intensity remained at about 50% of the maximum value. rETR started to decrease above a 200 μE m^-2 s^-1 irradiance value in Fe-deficient cells; however, in Fe-replete cells inhibition of electron transfer was observed above irradiance values of 300 μE m^-2 s^-1 (Figure 1B). Thus, a substantially higher sensitivity to light was observed in P. tricornutum cells grown under low Fe concentration.

Before the analysis of the effects of Fe deficiency on TL emissions of P. tricornutum cells, we have characterized some of the most relevant TL bands detected in vivo using healthy cells from cultures of this diatom. Not much is known about the characteristics of TL bands in diatoms because only a few TL studies have been carried out previously (Vavilin et al., 2002; Eisenstadt et al., 2008; Materna et al., 2009). Excitation of dark-adapted P. tricornutum cells at 1°C with a series of saturating single turn-over flashes induced the appearance of very complex TL glow curves, with differences in the temperature of the maximum (t_max) and signal intensity. TL curves induced by 1, 2, and 3 flashes are shown in Figure 2 as examples. The light emission curve obtained after illumination with two flashes was the largest of the series and showed a t_max at about 24°C and a small shoulder around 39°C (Figure 2, 2F). These TL signals could be well simulated by two decomposition components, with different t_max and contributions to the total signal intensity. The decomposition analysis of the emission curve induced by two flashes is shown in Figure 2 (2F, dotted lines) as an example. This first component can be assigned to the well-known TL B band originating from the recombination reactions of S₃Q_B^- and S₂Q_B^- charge pairs in PSII. A t_max value of 23°C and a signal contribution of 63% were obtained for this band (Table 2). We tentatively assigned the second TL component appearing at higher temperatures to the AG band, usually induced by FR illumination in intact photosynthetic materials (Miranda and Ducruet, 1995). A t_max value of 40°C and a signal contribution of 37% were obtained for this AG band (Table 2). A similar AG band has been observed in leaves of pea, Arabidopsis and tobacco and in cells of the green alga Chlamydomonas reinhardtii excited by white light (Miranda and Ducruet, 1995; Ducruet et al., 2005, 2011; Ducruet and Vass, 2009). This TL emission seems to reflect a back-flow of electrons from unknown reductants present in the stroma to the quinone acceptors of PSII, allowing their recombination with S₂ and S₃ states (Sundblad et al., 1988; Miranda and Ducruet, 1995). Whereas recombination of S₂Q_B^- and S₃Q_B^- centers produces a B band, the S₂Q_B and S₃Q_B centers should not lead to luminescence emission, unless an electron is progressively fed back to Q_B, resulting in AG emission. AG band also appears after illumination with continuous white light or flashes in some metabolic conditions: when the use of photosynthetic energy is slowed down due to a lack of CO₂ (Mellvig and Tillberg, 1986), in young pea leaves (Miranda and Ducruet, 1995) and when CAM metabolism is activated in a CAM-inducible species (Krieger et al., 1998). AG emission has been associated to the activation of cyclic/chlororespiratory electron flows in leaves by stress conditions (Ducruet, 2003).

Analysis of TL yields in dark-adapted samples illuminated by a train of short saturating flashes allows the estimation of the ratio between S₀:S₁ and Q_B:Q_B^- in PSII (Ducruet, 2003; Zurita et al., 2005; Roncel et al., 2007). The intensity of the B band exhibited a typical four-oscillation period with maxima after the second and sixth flashes (inset of Figure 2). According to Inoue (1996), this pattern may suggest that in dark-adapted P. tricornutum cells the ratio S₀:S₁ and Q_B^-:Q_B is about 25:75. Thus, after one single flash, the S₁Q_B centers will go to the luminescence-emitting state S₂Q_B^-, generating a B band peaking at about 30–32°C (Figure 2, 1F), generally identified as the B₂ band component. Two flashes induced the largest glow curve because they generate a large amount of PSII centers in both S₂Q_B^- and S₃Q_B^- luminescence states (Rutherford and Inoue, 1984). Besides, the yield from the latter recombination is higher than that from the former, by a factor of 1.7–2.0 (Rutherford et al., 1984). The much higher contribution of the second component of the B band (B₁ band) after two flashes induces the appearance of a TL glow curve significantly shifted to lower temperatures and broadened.

To confirm that the proposed AG band observed in P. tricornutum cell suspensions after white light flash excitation can be identified as a typical AG band (normally induced by FR light), we have also performed TL measurements after continuous illumination of cell samples with 720 nm monochromatic light (Figure 2, FR/2F). After this illumination, samples were also excited with two white light flashes to ensure induction of maximal signals for B and AG bands. FR illumination generated a more prominent AG band, while the B band was reduced. FR light preferentially excites PSI and consequently oxidizes the PQ pool, thus favoring in the dark a back transfer of electrons from stromal reductants to the oxidized Q_B and finally to the S₂ and S₃ states of the manganese cluster (Ducruet et al., 2005). This overall recombination reaction leads to AG emission. The mathematical analysis of the two components found by the simulation software showed t_max values of 21 and 40°C with signal contributions of 43 and 57% for the B and AG bands, respectively (Table 2). These results support that the 40°C band observed in P. tricornutum cell suspensions after excitation with white light flashes (Figure 2, 2F) corresponds to the same recombination reaction which gives rise to the FR-induced AG TL band previously described (Roncel et al., 2007).

Thermoluminescence was used to determine the effects of Fe deficiency on PSII electron transfer activity of P. tricornutum cells (Figure 3). Excitation of Fe-deficient P. tricornutum cells with two flashes at 1°C induced the appearance of a TL glow curve with significant differences in comparison with the curves obtained in Fe-replete cells: a decrease on the total TL signal intensity of about 10% and also a significant decrease of the 40°C component of the signal (AG band; Figure 3, 2F). The decomposition analysis of this emission curve allowed obtaining t_max values of 22 and 41°C and signal contributions of 71 and 29% for the B and AG bands, respectively (Table 2). Thus, although the t_max values were similar for both iron conditions, a significant increase of the signal contribution of the B band was detected (about 8%) in parallel with a similar decrease for the AG band (Figures 2 and 3; Table 2).

Thermoluminescence measurements were also performed after continuous illumination of Fe-deficient P. tricornutum cell samples with 720 nm (FR) monochromatic light (Figure 3, FR/2F). The mathematical analysis of the two components found by the simulation software showed t_max values of 20 and 39°C and signal intensity contributions of 66 and 34% for the B and AG bands, respectively (Table 2). Thus, the signal contributions for B and AG bands remained similar to that observed in white light excitation experiments without previous FR illumination (Figure 3, 2F; Table 2). Interestingly, FR excitation did not increase the amount of AG band in P. tricornutum cells cultivated in Fe deficiency, as observed for Fe-replete cells (Figures 2 and 3, FR/2F).

The effects of high light intensity on the TL emission curves induced in P. tricornutum cells in both Fe-replete and Fe-deficient culture conditions have been also investigated (Figure 4). Cells from 21 days cultures of both Fe conditions were harvested and suspended in the TL cuvette at the same Chl concentration. Cells were then illuminated with white light of 300 μE m^-2 s^-1 intensity during different time periods. This medium light intensity was chosen because when using a high photoinhibitory light intensity (1000 μE m^-2 s^-1) not TL signal was detected for either Fe conditions (data not shown). After these illuminations, TL emission was recorded as described in “Materials and Methods” section. Figure 4 shows the results obtained. The application of a light intensity of 300 μE m^-2 s^-1 during increasing time periods induced the progressive decrease of luminescence emission in Fe-deficient cells (Figure 4, upper). After 6 min of illumination a decrease in emission intensity of TL of about 30% was observed. However, after 10 min of light illumination, the TL signal was almost abolished. The effects of illumination in Fe-replete cells were significantly different. The application of 300 μE m^-2 s^-1 light during increasing time periods did not induce changes in the total intensity of the luminescence emission in Fe-replete cells (Figure 4, lower). However, after 6 or 10 min of illumination a significant increase of the signal contribution of the AG band was detected (from 30% to about 68%) in parallel with a similar decrease for the B band. Thus, these results suggest that Fe deficiency induced an increase in light sensitivity of PSII.

In photosynthetic cells, photoinhibitory conditions (as exposure to high light intensities) increase the probability to generate the very reactive and toxic ¹O₂ species (singlet oxygen) in PSII. The formation of singlet oxygen could initiate the peroxidation of unsaturated lipids in membranes (Vavilin and Ducruet, 1998). The level of lipid peroxidation in photosynthetic membranes can be measured by the HTL technique (see Materials and Methods section) (Roncel et al., 2007). Several luminescence high temperature bands (HTL bands) have been observed without prior illumination at temperatures above 60°C (Ducruet and Vavilin, 1999; Roncel et al., 2007). A broad HTL band centred near 130°C (known as the HTL2 band) is generated because of the thermal radiative decomposition of lipid peroxides that, in turn, leads to the formation of carbonyl groups in a triplet state followed by migration of excitation energy toward Chl (Vavilin and Ducruet, 1998; Ducruet and Vavilin, 1999; Vavilin et al., 2002). The amplitude of this band has been well correlated with the accumulation of malondialdehyde, an indicator of lipid peroxidation in standard chemical tests (Vavilin and Ducruet, 1998; Vavilin et al., 2002).

The HTL technique was applied to detect lipid peroxidation in P. tricornutum cells cultured in both Fe-replete and Fe-deficient conditions. The measurements have been performed using cells from 21 days growth cultures under standard light intensity conditions (20 μE m^-2 s^-1). Figure 5 shows that a broad HTL2 band with maximum between 140 and 150°C was present in Fe-deficient cells (Figure 5). This band was significantly lower in Fe-replete cells (Figure 5). Thus, a very high level of lipid peroxidation was observed in cells grown in Fe-deficient conditions under a low light intensity (20 μE m^-2 s^-1). These results suggest that Fe deficiency may induce the appearance of acceptor-side photoinhibitory processes, and consequently, the generation of singlet oxygen at a very low light intensity, which are usually not photoinhibitory.

The effect of Fe deficiency on PSI activity was also investigated by measuring the P700 redox state changes during illumination (Figure 6). The oxidized form of P700 displays a broad absorbance peak around 800–840 nm. Thus, it is possible to analyze its redox state monitoring changes in the absorbance at 830 nm. In dark-adapted cultures, P700 is found reduced since the acceptor side of P700, i.e., the Calvin-Benson cycle and subsequent reactions, are de-activated. Under actinic light P700 is oxidized and re-reduced by electrons coming from the PQ-pool, and thus by applying saturating pulses its ability to become oxidized and re-reduced can be determined.

Induction-recovery curves were performed in Fe-deficient and Fe-deplete cultures using red actinic light (λ 635 nm). As shown in Figure 6A, the calculated quantum yield of PSI photochemistry, Y(I), was substantially reduced in the Fe-deficient cultures. The loss of PSI activity proved to be caused by a lack of availability of electron donors for PSI, as shown by the higher degree of donor side limitations, Y(ND), in Fe-deficient cells (Figure 6B). In contrast, Fe-replete and Fe-deficient cultures were indistinguishable with respect to acceptor-side limitations, Y(NA) (Figure 6C). Thus, Fe limitation leads to a deficiency in PSI activity affecting specifically the supply of electrons to this photosystem in the light. However, the demand for electrons from PSI appears not to be altered.

Iron deficiency induced an important decrease in the relative content of Cyt c₆ protein in P. tricornutum cells (Figure 7). Changes in the amount of Cyt c₆ were determined by measuring the spectra of soluble cell fractions after 21 days of growing in Fe-replete and Fe-deficient conditions. Figure 7 shows the ascorbate minus ferricyanide absorbance difference spectra in the region of 400–600 nm for both Fe conditions. The estimated amounts of Cyt c₆ obtained from these spectra were 218 and 57 μg per grams of total cell wet weight in Fe-replete or Fe-deficient P. tricornutum cultures, respectively. When normalized to Chl content the values were 0.52 and 0.15 μg of Cyt c₆ per mg of Chl in Fe-replete and Fe-deficient cells, respectively. Thus, under Fe-deficient conditions Cyt c₆ concentration is reduced to less of 30% of the protein present in Fe-replete cells. This significant reduction of the Cyt c₆ concentration as a consequence of the Fe deficiency was confirmed by Western blot analysis (Figure 7, inset).

From this work, it is shown that the culture of the marine diatom P. tricornutum under low Fe concentration led to a significant decline in photosynthetic and respiratory electron transfer processes, as well as to an increased sensitivity to light. To partially compensate for the negative effects of Fe limitation, it seems that a secondary electron transfer pathway, such as chlororespiration, can be activated in P. tricornutum cells.

Photosynthetic electron transport activity is appreciably lowered by Fe-limitation stress (Morales et al., 1991; Kudo et al., 2000; Moseley et al., 2002; Allen et al., 2008; Page et al., 2012; Urzica et al., 2012; Petrou et al., 2014; Kalaji et al., 2014a; Taddei et al., 2016). A significantly decreased photosynthetic electron transport rate (about 40% of the control) was here found in Fe-deficient cells of P. tricornutum (Table 1 and Figure 1B), confirming previous results (Kudo et al., 2000; Allen et al., 2008). The substantial Fe requirement in both photosystems RC (three atoms for PSII; 12 atoms for PSI) and in the photosynthetic electron transport chain (six atoms for Cyt b₆f complex, one atom for Cyt c₆ and two atoms for ferredoxin molecule) (Behrenfeld and Milligan, 2012; Petrou et al., 2014) seems to be the origin of this effect. One of the objectives of this work has been to distinguish between deleterious effects of Fe deficiency on PSII and PSI activities.

Photosystem II activity of P. tricornutum cells was strongly affected by the culture under low Fe concentration. The inhibitory effect on the PSII photochemical activity was reflected by the decrease of the maximum quantum yield of PSII (38%; Table 1 and Figure 1A) and also by the decrease on the intensity of the total TL emission signal (10%; Figures 2 and 3). However, the TL results obtained in this work have shown that the recombination electron transfer reaction between Q_B and the S₂/S₃ states of the manganese cluster is not affected in cells grown with low Fe concentration (Figures 2 and 3). The t_max for the B and AG TL bands were almost identical for the two Fe conditions (Table 2), thus indicating that the energetic of the electron transfer reactions involved are not affected by Fe deficiency. Thus, the low Fe concentration seems not to affect the PSII primary photochemistry in P. tricornutum cells.

The loss of PSII photochemical activity may be attributed to a decrease in the amount of PSII pigments, as Fe is required for their biosynthesis (Morales et al., 1991). An increased disconnection of antennae from the PSII RC in response to Fe starvation has been also proposed (Petrou et al., 2014). Due to this, the transfer of excitons to the PSII RC may be hindered, and thus, the efficiency of PSII reduction, causing a decline in the F_v/F_m (Figure 1A). This finding is consistent with general photosynthetic responses to Fe limitation reported in diatoms (Greene et al., 1991, 1992; Geider et al., 1993; Allen et al., 2008; Lommer et al., 2012; Petrou et al., 2014), green algae (Vassiliev et al., 1995), cyanobacteria (Guikema and Sherman, 1983; Riethman and Sherman, 1988), and higher plants (Morales et al., 1991; Larbi et al., 2006; Timperio et al., 2007; Kalaji et al., 2014a). In P. tricornutum cells grown under Fe deficiency a significantly lower concentration of Chl a was observed (Table 1). The decreased PSII activity may be also attributed to a reduced amount of functional PSII complexes in cells (Msilini et al., 2011) or the presence of light harvesting complexes connected to inactive PSII complexes due to Fe depletion (Wydrzynski, 1982). Recently, Taddei et al. (2016) have reported that the photosynthetic capacity is severely impaired in P. tricornutum when Fe is limited, as demonstrated by the lower F_v/F_m and rETR. These authors have proposed that the decreased maximal rETR was probably caused by a diminished capacity of carbon fixation.

Photosystem I activity has been described to be more sensitive than PSII activity to Fe-limitation (Pushnik and Miller, 1989). The results of this work have showed that the electron donation to PSI is severely inhibited by Fe deficiency. In Fe-deficient cultures of P. tricornutum, a significant lower quantum yield of PSI was detected (Figure 6A). A deficiency of donors or acceptors of PSI may be the reason for this effect. In the first case, the pool of P700, the primary donor of PSI, could not be reduced whereas in the second case the pool of P700 cannot be oxidized. The calculated quantum yield of donor-side limitations Y(ND) (Figure 6B) and acceptor-side limitations Y(NA) (Figure 6C) showed that the significant decreased quantum yield of PSI [Y(I)] obtained in Fe-deficient cultures is due to limitations on the donor side of PSI. In dark-adapted cultures, P700 is found reduced and the acceptor side of P700, i.e., the Calvin–Benson cycle and subsequent reactions, are de-activated. Under actinic light, P700 becomes oxidized and later reduced by electrons coming from the PQ pool. Under Fe-deficient conditions, P700 cannot become reduced possibly due to fewer Fe-containing electron transfer complexes downstream of the PQ pool, such as the Cyt b₆f complex or the Cyt c₆ soluble donor (Figure 7) (Bruce and Malkin, 1991; Greene et al., 1992; Allen et al., 2008). In particular, in P. tricornutum cells grown under low Fe conditions, the Cyt c₆ concentration is reduced to less than 30% of the protein present in Fe-replete cells, thus presumably disfavouring the PQ pool re-oxidation. This decrease in Cyt c₆ is significantly larger than the previously estimated following photochemically induced absorbance changes in whole cells (Allen et al., 2008).

An over-reduction of the PQ pool may induce the appearance of the acceptor side photoinhibition process in PSII, thus generating singlet oxygen species (Murata et al., 2007). This highly reactive form of oxygen can cause peroxidation of the membrane lipids. Analysis of HTL2 bands of TL obtained in P. tricornutum cells (Figure 5) clearly showed a much higher level of lipid peroxidation in Fe-deficient conditions, suggesting a high rate of generation of reactive singlet oxygen. Interestingly, this photoinhibitory process seems to be activated in Fe-deficient cells under a very low light intensity.

The severe decline of the photosynthetic electron transport activity induced by Fe deficiency in P. tricornutum cells led probably to a significant reduction of the level of synthesis of ATP and NADPH in the stroma. There is a TL band emission associated to PSII, the AG band, directly related to the assimilatory potential ([NADPH +ATP]) in the stroma (Heber et al., 1986; Mellvig and Tillberg, 1986; Palmqvist et al., 1986; Krieger et al., 1998, Roman and Ducruet, 2000). This TL AG band is observed in higher plants and algae after FR pre-illumination or, sometimes, after two or three flashes, peaking at about 45°C at a 0.5°C s^-1 warming rate. The AG band corresponds to the fraction of PSII centers in the S₂/₃Q_B non-radiative state immediately after pre-illumination, in which the arrival of an electron transferred from stroma along cyclic/chlororespiratory pathway(s) produces the S_2/3Q_B^- radiative state that emits luminescence (Sundblad et al., 1988). The analysis of the emission curves induced by two flashes has shown the existence of an AG band of similar energetic characteristics (t_max at 40–41°C) in dark-adapted P. tricornutum cells grown under both Fe-replete and Fe-deficient conditions (Figures 2 and 3). However, it has been detected a significant decrease of the contribution of this component to the total TL intensity in Fe-deficient cells, thus suggesting the existence of a lower assimilatory potential ([NADPH +ATP]) in the stroma of these cells.

Far red light preferentially excites PSI and consequently oxidizes the PQ pool, thus favoring in the dark a back transfer of electrons from stromal reductants to the oxidized Q_B and, finally, to the S₂ and S₃ states of the manganese cluster (Ducruet et al., 2005; Roncel and Ortega, 2005). This overall recombination reaction leads also to AG emission. Interestingly, FR illumination did not increase the amount of the AG band in P. tricornutum cells cultured under Fe deficiency, as observed for Fe-replete cells (Figures 2 and 3). This phenomenon can be explained taking into account several possibilities. First, the substantial inhibition of the synthesis of Cyt c₆ (this work) and functional components of the Cyt b₆f complexes (Bruce and Malkin, 1991; Greene et al., 1992) would disfavor the PQ pool re-oxidation and consequently the appearance of PSII centers in the S_2/3Q_B non-radiative state, which is initially required to generate this band. However, the oxidation process of the PQ pool was only partially inhibited under Fe-deficient conditions, as shown by polarography (Table 1) and fluorescence (Figure 6) experiments. Therefore, FR illumination should induce a detectable increase in PSII centers in the S₂/₃Q_B non-radiative state. Other possible explanation may be a dramatic decrease of the assimilatory potential ([NADPH +ATP]) in the stroma of cells grown under low Fe concentration.

The inability to generate the AG band in Fe-deficient cells illuminated with FR light could be related with the activation of cyclic/chlororespiratory electron transport pathway(s) induced by this stress condition. The back electron transfer responsible for the AG band is usually induced by warming above 35°C to activate the cyclic pathway(s). But if these pathways are already activated prior to the TL recording, the AG band emission fuses with the B band, because Q_B becomes reduced efficiently by stroma electrons before warming (Ducruet et al., 2011). Thus, we propose that in Fe-deficient cells of P. tricornutum the contribution of the 40°C AG band to the total TL emission after FR illumination not increases because it is already fused with the lower temperature B band. Fe deficiency could thus induce the activation of the chlororespiratory electron transfer pathway. This process has been proposed to be involved in protective or adaptive mechanisms of photosynthetic organisms to environmental stress conditions (Bennoun, 1982; Morehouse and Mason, 1988; Peltier and Schmidt, 1991; Rumeau et al., 2007).

The marine diatom P. tricornutum is highly tolerant to damage induced by exposure to high light intensities (Olaizola et al., 1994). However, the results obtained in this work have showed that this algae becomes sensitive to low and medium light intensities if is cultivated under low Fe concentration (Figures 2 and 4). A high level of lipid peroxidation has been detected in Fe-deficient cells under the standard culture conditions, i.e., light/dark cycles of 16/8 h and illumination with a low light intensity of 20 μE m^-2 s^-1 (Figure 5). Thus, even under these dim light conditions singlet oxygen is generated, probably due to an acceptor side photoinhibition process in PSII. Besides, the rETR observed in Fe-deficient cells at different light intensities were significantly lower than rETR from Fe-replete cells (Figure 1). The illumination with a light intensity of 300 μE m^-2 s^-1 during different time periods caused a dramatic disappearance in the TL signal amplitude in P. tricornutum cells grown under low Fe concentration (Figure 4). However, this treatment did not affect the intensity of the TL signal in Fe-replete cells, which were also capable of generating an important AG band component (Figure 4). These data provide strong evidence supporting the proposal that light sensitivity of the photosynthetic apparatus is substantially increased in P. tricornutum cells grown under low Fe concentration conditions. The partial blocking of the oxidation of the PQ pool, and the consequent induction of the acceptor side photoinhibition in PSII, might be the reason for this high light sensitivity.

Cyclic electron flow around PSII, presumably via Cyt b559, has been suggested earlier as a photoprotection mechanism that could retard both acceptor and donor side photoinhibition (for review, see Whitmarsh and Pakrasi, 1996). Such a cycle was shown to occur in vivo at high light intensities in the green alga Chlorella pyrenoidosa (Falkowski et al., 1986) and in the diatom P. tricornutum, also accompanied with the activation of chlororespiration (Lavaud et al., 2002b). Fe deficiency could induce the activation of the cyclic electron flow in PSII even at low and medium light intensities in P. tricornutum. However, a low synthesis of one of the proposed components of cyclic electron transfer pathway in PSII, the Cyt b559, would significantly decrease the efficiency of such protection mechanism.

In summary, our results show that decreasing Fe concentration in the culture medium results in a significant decrease of the photochemical efficiency of both PSII and PSI complexes, as well as to an increased sensitivity to light because the activation of the acceptor side PSII photoinhibition process. We propose that the possible induction of chlorespiratory electron transfer pathway under Fe restricted conditions could partially compensate some of the metabolic negative effects of this stress condition: (1) the low levels of ATP generated by the linear photosynthetic electron transfer; and (2) the over-reduction of the PQ pool, and the consequent induction of the acceptor side photoinhibition of PSII.

MR, AL, MH, JN, and JO conceived and designed experiments; MR, MH, JN, AG-R, BN, PB-B performed experiments; MR and JO wrote the manuscript; all the authors contributed to the discussion and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.