Variations in the first steps of photosynthesis for the diatom Cyclotella meneghiniana grown under different light conditions.

In this work we have applied picosecond and steady-state fluorescence measurements to study excitation energy transfer and trapping in intact Cyclotella meneghiniana diatom cells grown at different light intensities. Different excitation and detection wavelengths were used to discriminate between Photosystem I and II (PSI and PSII) kinetics and to study excitation energy transfer from the outer antenna to the core of PSI and PSII. It is found that the light-harvesting fucoxanthin chlorophyll proteins (FCPs) transfer their excitation energy predominantly to PSII. It is also observed that the PSII antenna is slightly richer in red-absorbing fucoxanthin than the FCPs associated with PSI. The average excitation trapping time in PSI is around 75 ps whereas this time is around 450 ps for PSII in cells grown in 20 µmol of photons per m(2) per s. The latter time decreases to 425 ps for 50 µmol of photons and 360 ps for 140 µmol of photons. It is concluded that cells grown under higher photon flux densities have a smaller antenna size than the ones grown in low light. At the same time, the increase of growth light intensity leads to a decrease of the relative amount of PSI. This effect is accompanied by a substantial increase in the amount of chlorophyll a that is not active in excitation energy transfer and most probably attached to inactivated/disassembled PSII units.

Disentangling picosecond events that complicate the quantitative use of the calcium sensor YC3.60.

Yellow Cameleon 3.60 (YC3.60) is a calcium sensor based on Förster resonance energy transfer (FRET). This sensor is composed of a calmodulin domain and a M13 peptide, which are located in between enhanced cyan-fluorescent protein (ECFP) and the Venus variant of enhanced yellow-fluorescent protein (EYFP). Depending on the calcium concentration, the efficiency of FRET from donor ECFP to acceptor EYFP is changing. In this study, we have recorded time-resolved fluorescence spectra of ECFP, EYFP, and YC3.60 in aqueous solution with picosecond time resolution, using different excitation wavelengths. Detailed insight in the FRET kinetics was obtained by using global and target analyses of time- and wavelength-resolved fluorescence of purified YC3.60 in calcium-free and calcium-bound conformations. The results clearly demonstrate that for both conformations, there are two distinct donor populations: a major one giving rise to FRET and a minor one not able to perform FRET. The transfer time for the calcium-bound conformation is 21 ps, whereas it is in the order of 1 ns for the calcium-free conformation. Ratio imaging of acceptor and donor fluorescence intensities of YC3.60 is usually applied to measure Ca(2+) concentrations in living cells. From the obtained results, it is clear that the intensity ratio is strongly influenced by the presence of donor molecules that do not take part in FRET, thereby significantly affecting the quantitative interpretation of the results.

Monitoring photosynthesis in individual cells of Synechocystis sp. PCC 6803 on a picosecond timescale.

Picosecond fluorescence kinetics of wild-type (WT) and mutant cells of Synechocystis sp. PCC 6803, were studied at the ensemble level with a streak-camera and at the cell level using fluorescence-lifetime-imaging microscopy (FLIM). The FLIM measurements are in good agreement with the ensemble measurements, but they (can) unveil variations between and within cells. The BE mutant cells, devoid of photosystem II (PSII) and of the light-harvesting phycobilisomes, allowed the study of photosystem I (PSI) in vivo for the first time, and the observed 6-ps equilibration process and 25-ps trapping process are the same as found previously for isolated PSI. No major differences are detected between different cells. The PAL mutant cells, devoid of phycobilisomes, show four lifetimes: ∼20 ps (PSI and PSII), ∼80 ps, ∼440 ps, and 2.8 ns (all due to PSII), but not all cells are identical and variations in the kinetics are traced back to differences in the PSI/PSII ratio. Finally, FLIM measurements on WT cells reveal that in some cells or parts of cells, phycobilisomes are disconnected from PSI/PSII. It is argued that the FLIM setup used can become instrumental in unraveling photosynthetic regulation mechanisms in the future.

5-fluorotryptophan as dual probe for ground-state heterogeneity and excited-state dynamics in apoflavodoxin.

The apoflavodoxin protein from Azotobacter vinelandii harboring three tryptophan (Trp) residues, was biosynthetically labeled with 5-fluorotryptophan (5-FTrp). 5-FTrp has the advantage that chemical differences in its microenvironment can be sensitively visualized via (19)F NMR. Moreover, it shows simpler fluorescence decay kinetics. The occurrence of FRET was earlier observed via the fluorescence anisotropy decay of WT apoflavodoxin and the anisotropy decay parameters are in excellent agreement with distances between and relative orientations of all Trp residues. The anisotropy decay in 5-FTrp apoflavodoxin demonstrates that the distances and orientations are identical for this protein. This work demonstrates the added value of replacing Trp by 5-FTrp to study structural features of proteins via (19)F NMR and fluorescence spectroscopy.

Probing the structure and dynamics of a DNA hairpin by ultrafast quenching and fluorescence depolarization.

DNA hairpins have been investigated in which individual adenines were replaced by their fluorescent analog 2-aminopurine (2AP). The temperature dependence of the time evolution of polarized emission spectra was monitored with picosecond time resolution. Four isotropic decay components for each oligonucleotide indicated the coexistence of at least four conformations. The fluorescence for three of these was significantly quenched, which is explained by hole transfer from 2AP to guanine(s). An approximately 8-ps component is ascribed to direct hole transfer, the approximately 50-ps and approximately 500-ps components are ascribed to structural reorganization, preceding hole transfer. At room temperature, a fraction remains unquenched on a 10-ns timescale, in contrast to higher temperatures, where the flexibility increases. Besides quenching due to base stacking, a second quenching process was needed to describe the data. Evidence for both intrastrand and interstrand hole transfer was found. The extracted probability for stacking between neighboring bases in double-stranded regions was estimated to be approximately 75% at room temperature and approximately 25% at 80 degrees C, demonstrating structural disorder of the DNA. Fluorescence depolarization revealed both local dynamics of the DNA and overall dynamics of the entire oligonucleotide. Upon raising the temperature, the C-N terminus of the hairpin appears to melt first; the rest of the hairpin denatures above the average melting temperature.

Fluorescence lifetime heterogeneity in aggregates of LHCII revealed by time-resolved microscopy.

Two-photon excitation, time-resolved fluorescence microscopy was used to investigate the fluorescence quenching mechanisms in aggregates of light-harvesting chlorophyll a/b pigment protein complexes of photosystem II from green plants (LHCII). Time-gated microscopy images show the presence of large heterogeneity in fluorescence lifetimes not only for different LHCII aggregates, but also within a single aggregate. Thus, the fluorescence decay traces obtained from macroscopic measurements reflect an average over a large distribution of local fluorescence kinetics. This opens the possibility to resolve spatially different structural/functional units in chloroplasts and other heterogeneous photosynthetic systems in vivo, and gives the opportunity to investigate individually the excited states dynamics of each unit. We show that the lifetime distribution is sensitive to the concentration of quenchers contained in the system. Triplets, which are generated at high pulse repetition rates of excitation (>1 MHz), preferentially quench domains with initially shorter fluorescence lifetimes. This proves our previous prediction from singlet-singlet annihilation investigations (Barzda, V., V. Gulbinas, R. Kananavicius, V. Cervinskas, H. van Amerongen, R. van Grondelle, and L. Valkunas. 2001. Biophys. J. 80:2409-2421) that shorter fluorescence lifetimes originate from larger domains in LHCII aggregates. We found that singlet-singlet annihilation has a strong effect in time-resolved fluorescence microscopy of connective systems and has to be taken into consideration. Despite that, clear differences in fluorescence decays can be detected that can also qualitatively be understood.

Singlet-singlet annihilation kinetics in aggregates and trimers of LHCII.

Singlet-singlet annihilation experiments have been performed on trimeric and aggregated light-harvesting complex II (LHCII) using picosecond spectroscopy to study spatial equilibration times in LHCII preparations, complementing the large amount of data on spectral equilibration available in literature. The annihilation kinetics for trimers can well be described by a statistical approach, and an annihilation rate of (24 ps)(-1) is obtained. In contrast, the annihilation kinetics for aggregates can well be described by a kinetic approach over many hundreds of picoseconds, and it is shown that there is no clear distinction between inter- and intratrimer transfer of excitation energy. With this approach, an annihilation rate of (16 ps)(-1) is obtained after normalization of the annihilation rate per trimer. It is shown that the spatial equilibration in trimeric LHCII between chlorophyll a molecules occurs on a time scale that is an order of magnitude longer than in Photosystem I-core, after correcting for the different number of chlorophyll a molecules in both systems. The slow transfer in LHCII is possibly an important factor in determining excitation trapping in Photosystem II, because it contributes significantly to the overall trapping time.

Generation of fluorescence quenchers from the triplet states of chlorophylls in the major light-harvesting complex II from green plants.

Laser flash-induced changes of the fluorescence yield were studied in aggregates of light-harvesting complex II (LHCII) on a time scale ranging from microseconds to seconds. Carotenoid (Car) and chlorophyll (Chl) triplet states, decaying with lifetimes of several microseconds and hundreds of microseconds, respectively, are responsible for initial light-induced fluorescence quenching via singlet-triplet annihilation. In addition, at times ranging from milliseconds to seconds, a slow decay of the light-induced fluorescence quenching can be observed, indicating the presence of additional quenchers generated by the laser. The generation of the quenchers is found to be sensitive to the presence of oxygen. It is proposed that long-lived fluorescence quenchers can be generated from Chl triplets that are not transferred to Car molecules. The quenchers could be Chl cations or other radicals that are produced directly from Chl triplets or via Chl triplet-sensitized singlet oxygen. Decay of the quenchers takes place on a millisecond to second time scale. The decay is slowed by a few orders of magnitude at 77 K indicating that structural changes or migration-limited processes are involved in the recovery. Fluorescence quenching is not observed for trimers, which is explained by a reduction of the quenching domain size compared to that of aggregates. This type of fluorescence quenching can operate under very high light intensities when Chl triplets start to accumulate in the light-harvesting antenna.

Peridinin chlorophyll a protein: relating structure and steady-state spectroscopy.

Peridinin chlorophyll a protein (PCP) from Amphidinium carterae has been studied using absorbance (OD), linear dichroism (LD), circular dichroism (CD), fluorescence emission, fluorescence anisotropy, fluorescence line narrowing (FLN), and triplet-minus-singlet spectroscopy (T-S) at different temperatures (4-293 K). Monomeric PCP binds eight peridinins and two Chls a. The trimeric structure of PCP, resolved at 2 A [Hofmann et al. (1996) Science 27, 1788-1791], allows modeling of the Chl a-protein and Chl a-Chl a interactions. The FLN spectrum shows that Chl a is not or is very weakly hydrogen-bonded and that the central magnesium of the emitting Chl a is monoligated. Simulation of the temperature dependence of the absorption spectra indicates that the Huang-Rhys factor, characterizing the electron-phonon coupling strength, has a value of approximately 1. The width of the inhomogeneous distribution function is estimated to be 160 cm(-)(1). LD experiments show that the two Chls a in PCP are essentially isoenergetic at room temperature and that a substantial amount of PCP is in a trimeric form. From a comparison of the measured and simulated CD, it is concluded that the interaction energy between the two Chls a within one monomer is very weak, <10 cm(-)(1). In contrast, the Chls a appear to be strongly coupled to the peridinins. The 65 cm(-)(1) band that is visible in the low-frequency region of the FLN spectrum might indicate a Chl a-peridinin vibrational mode. The efficiency of Chl a to peridinin triplet excitation energy transfer is approximately 100%. On the basis of T-S, CD, LD, and OD spectra, a tentative assignment of the peridinin absorption bands has been made.

Förster excitation energy transfer in peridinin-chlorophyll-a-protein.

Time-resolved fluorescence anisotropy spectroscopy has been used to study the chlorophyll a (Chl a) to Chl a excitation energy transfer in the water-soluble peridinin-chlorophyll a-protein (PCP) of the dinoflagellate Amphidinium carterae. Monomeric PCP binds eight peridinins and two Chl a. The trimeric structure of PCP, resolved at 2 A (, Science. 272:1788-1791), allows accurate calculations of energy transfer times by use of the Förster equation. The anisotropy decay time constants of 6.8 +/- 0.8 ps (tau(1)) and 350 +/- 15 ps (tau(2)) are respectively assigned to intra- and intermonomeric excitation equilibration times. Using the ratio tau(1)/tau(2) and the amplitude of the anisotropy, the best fit of the experimental data is achieved when the Q(y) transition dipole moment is rotated by 2-7 degrees with respect to the y axis in the plane of the Chl a molecule. In contrast to the conclusion of, Biochemistry. 23:1564-1571) that the refractive index (n) in the Förster equation should be equal to that of the solvent, n can be estimated to be 1.6 +/- 0.1, which is larger than that of the solvent (water). Based on our observations we predict that the relatively slow intermonomeric energy transfer in vivo is overruled by faster energy transfer from a PCP monomer to, e.g., the light-harvesting a/c complex.

Circularly polarized chlorophyll luminescence reflects the macro-organization of grana in pea chloroplasts.

Circular polarization of luminescence (CPL; Steinberg IZ (1978) Annu Rev Biophys Bioeng 7: 113-137) was applied to study pea chloroplasts in different structural states. The structural changes of chloroplasts were induced by variation of osmotic pressure, concentration of magnesium-ions or photoinhibition. Both large CPL and psi-type circular dichroism (psi, polymerization and salt induced) signals appeared in the presence of granal macrostructure and were sensitive to structural changes of the grana. The relation was studied between the amount of CPL expressed as an emission anisotropy factor g(em) and amplitudes of the red psi-type CD bands. The positive psi-type CD band was not directly correlated with g (em) possibly due to a large contribution of circular intensity differential scattering to the measured CD spectra. However, a linear correlation between the amplitude of the negative psi-type CD band and g (em) was found. The CPL signal of pea chloroplasts was attributed to a psi-type origin, which is observed in macroaggregates with densely packed chromophores with a long-range chiral order, and directly depends on the level of macroorganization. With the use of CPL-based microscopy, the long-range packing of LHC II particles can be studied in individual chloroplasts in future. In addition, the CPL method in general allows the study of the macro-organization of grana in green leaves, where conventional light-transmission methods fail.

Spectroscopic characterization of the spinach Lhcb4 protein (CP29), a minor light-harvesting complex of photosystem II.

A spectroscopic characterization is presented of the minor photosystem II chlorophyll a/b-binding protein CP29 (or the Lhcb4 protein) from spinach, prepared by a modified form of a published protocol [Henrysson, T., Schroder, W. P., Spangfort, M. & Akerlund, H.-E. (1989) Biochim. Biophys. Acta 977, 301-308]. The isolation procedure represents a quicker, cheaper means of isolating this minor antenna protein to an equally high level of purity to that published previously. The pigment-binding protein shows similarities to other related light-harvesting complexes (LHCs), including the bulk complex LHCIIb but more particularly another minor antenna protein CP26 (Lhcb5). It is also, in the main, similar to other preparations of CP29, although some significant differences are discussed. In common with CP26, the protein binds about six chlorophyll a and two chlorophyll b molecules. Two chlorophyll b absorption bands are present at 638 and 650 nm and they are somewhat more pronounced than in a recent report [Giuffra, E., Zucchelli, G., Sandonà, D., Croce, R., Cugini, D., Garlaschi, F.M., Bassi, R. & Jennings, R.C. (1997) Biochem. 36, 12984-12993]. The bands give rise to positive and negative linear dichroism, respectively; both show negative CD bands (cf. bands with similar properties at 637 and 650 nm in CP26). Chlorophyll a absorption is dominated by a large contribution at 674 nm which also shows similarities to the major band in LHCIIb and CP26, while (as for CP26) a reduction in absorption around 670 nm is observed relative to the bulk complex. Principal differences from LHCIIb and CP26, and from other CP29 preparations, occur in the carotenoid region.

Decreasing the chlorophyll a/b ratio in reconstituted LHCII: structural and functional consequences.

Trimeric (bT) and monomeric (bM) light-harvesting complex II (LHCII) with a chlorophyll a/b ratio of 0.03 were reconstituted from the apoprotein overexpressed in Escherichia coli. Chlorophyll/xanthophyll and chlorophyll/protein ratios of bT complexes and 'native' LHCII are rather similar, namely, 0.28 vs 0. 27 and 10.5 +/- 1.5 vs 12, respectively, indicating the replacement of most chlorophyll a molecules with chlorophyll b, leaving one chlorophyll a per trimeric complex. The LD spectrum of the bT complexes strongly suggests that the chlorophyll b molecules adopt orientations similar to those of the chlorophylls a that they replace. The circular dichroism (CD) spectra of bM and bT complexes indicate structural arrangements resembling those of 'native' LHCII. Thermolysin digestion patterns demonstrate that bT complexes are folded and organized like 'native' trimeric LHCII. Surprisingly, in the bT complexes at 77 K, half of the excitations that are created on either chlorophyll b or xanthophyll are transferred to chlorophyll a. No or very limited triplet transfer from chlorophyll b to xanthophyll appears to take place. However, the efficiency of triplet transfer from chlorophyll a to xanthophyll is close to 100%, even higher than in 'native' LHCII at 77 K. It is concluded from the triplet-minus-singlet and CD results that the single chlorophyll a molecule that on the average is present in each bT complex binds preferably next to a xanthophyll molecule at the interface between the monomers.

The flow of excitation energy in LHCII monomers: implications for the structural model of the major plant antenna.

Spectral and kinetic information on energy transfer within the light-harvesting complex II (LHCII) monomer was obtained from this subpicosecond transient absorption study, by using selective excitation (663, 669, 672, 678, and 682 nm) of various Chl a absorption bands and detecting the induced changes over the entire Qy region (650-700 nm). It is shown that transfer from the pigment(s) absorbing around 663 nm to the low energy ones occurs in 5 +/- 1 ps, whereas the 670-nm excitation is delivered to the same "destination" in two phases (0.30 +/- 0.05 ps, and 12 +/- 2 ps), and a fast equilibration (lifetime 0.45 +/- 0.05 ps) takes place within the main absorption band (675-680 nm). From comparison with results from similar time-resolved measurements on trimeric samples, it can be concluded that the intramonomeric energy transfer completely determines the spectral equilibration observed in native LHCII complexes. To correlate the measured lifetimes and their associated spectra with the pigment organization within the available structural model of LHCII (. Nature. 367:614-621), extensive but straightforward theoretical modeling was used. Thus it is demonstrated that the pigment assignment (Chl a or Chl b) given by Kuhlbrandt and co-workers cannot simultaneously describe the dichroic spectra and the transient absorption results for the rather homologous LHCII and CP29 proteins. A more recent assignment for CP29, in which a Chl b molecule ("Chl b5") is identified as a Chl a (Dr. R. Bassi, personal communication), leads to a much better description of both CP29 and LHCII. Furthermore, the orientations of the transition dipole moments, which have not been obtained in the crystal structure, are now assigned for most of the Chl's.

Fluorescence and absorption spectroscopy of the weakly fluorescent chlorophyll a in cytochrome b6f of Synechocystis PCC6803.

A spectroscopic characterization of the chlorophyll a (Chl) molecule in the monomeric cytochrome b6f complex (Cytb6f) isolated from the cyanobacterium Synechocystis PCC6803 is presented. The fluorescence lifetime and quantum yield have been determined, and it is shown that Chl in Cytb6f has an excited-state lifetime that is 20 times smaller than that of Chl in methanol. This shortening of the Chl excited state lifetime is not caused by an increased rate of intersystem crossing. Most probably it is due to quenching by a nearby amino acid. It is suggested that this quenching is a mechanism for preventing the formation of Chl triplets, which can lead to the formation of harmful singlet oxygen. Using site-selected fluorescence spectroscopy, detailed information on vibrational frequencies in both the ground and Qy excited states has been obtained. The vibrational frequencies indicate that the Chl molecule has one axial ligand bound to its central magnesium and accepts a hydrogen bond to its 13(1)-keto carbonyl. The results show that the Chl binds to a well-defined pocket of the protein and experiences several close contacts with nearby amino acids. From the site-selected fluorescence spectra, it is further concluded that the electron-phonon coupling is moderately strong. Simulations of both the site-selected fluorescence spectra and the temperature dependence of absorption and fluorescence spectra are presented. These simulations indicate that the Huang-Rhys factor characterizing the electron-phonon coupling strength is between 0.6 and 0.9. The width of the Gaussian inhomogeneous distribution function is 210 +/- 10 cm-1.

Ultrafast evolution of the excited states in the chlorophyll a/b complex CP29 from green plants studied by energy-selective pump-probe spectroscopy.

The energy transfer process in the minor light-harvesting antenna complex CP29 of green plants was probed in multicolor transient absorption experiments at 77 K using selective subpicosecond excitation pulses at 640 and 650 nm. Energy flow from each of the chlorophyll (Chl) b molecules of the complex could thus be studied separately. The analysis of our data showed that the "blue" Chl b (absorption around 640 nm) transfers excitation to a "red" Chl a with a time constant of 350 +/- 100 fs, while the 'red' Chl b (absorption at 650 nm) transfers on a picosecond time scale (2.2 +/- 0.5 ps) toward a "blue" Chl a. Furthermore, both fast (280 +/- 50 fs) and slow (10-13 ps) equilibration processes among the Chl a molecules were observed, with rates and associated spectra very similar to those of the major antenna complex, LHC-II. Based on the protein sequence homology between CP29 and LHC-II, a basic modelling of the observed kinetics was performed using the LHC-II structure and the Förster theory of energy transfer. Thus, an assignment for the spectral properties and orientation of the two Chl's b, as well as for their closest Chl a neighbors, is put forward, and a comparison is made with the previous assignments and models for LHC-II and CP29.

The influence of aggregation on triplet formation in light-harvesting chlorophyll a/b pigment-protein complex II of green plants.

The influence of aggregation on triplet formation in the light-harvesting pigment-protein complex of photosystem II of green plants (LHCII) has been studied with time-resolved laser flash photolysis. The aggregation state of LHCII has been varied by changing the detergent concentration. The triplet yield increases upon disaggregation and follows the same dependence on the detergent concentration as the fluorescence yield. The rate constant of intersystem crossing is not altered by disaggregation, and variations of the triplet yield appear to be due to aggregation-dependent quenching of singlet excited states. The efficiency of triplet transfer in LHCII aggregates from chlorophyll (Chl) to carotenoid (Car) is 92 +/- 7% at room temperature and 82 +/- 6% at 5 K, and does not change upon disaggregation. The Chl's that do not transfer their triplets to Car's seem to be bound to LHCII and are capable of transfering/accepting their singlet excitations to/from other Chl's. Two spectral contributions of Car triplets are observed: at 525 and 506 nm. Disaggregation of macroaggregates to small aggregates reduces by 10% the relative contribution of Car triplets absorbing at 525 nm. This effect most likely originates from a decreased efficiency of intertrimer Chl-to-Car triplet transfer. At the critical micelle concentration, at which small aggregates are disassembled into trimers, the interactions between Chl and Car are changed. At room temperature, this effect is much more pronounced than at 5 K.

Xanthophylls in light-harvesting complex II of higher plants: light harvesting and triplet quenching.

A spectral and functional assignment of the xanthophylls in monomeric and trimeric light-harvesting complex II of green plants has been obtained using HPLC analysis of the pigment composition, laser-flash induced triplet-minus-singlet, fluorescence excitation, and absorption spectra. It is shown that violaxanthin is not present in monomeric preparations, that it has most likely a red-most absorption maximum at 510 nm in the trimeric complex, and that it is involved in both light-harvesting and Chl-triplet quenching. Two xanthophylls (per monomer) have an absorption maximum at 494 nm. These play a major role in both singlet and triplet transfer. These two are most probably the two xanthophylls resolved in the crystal structure, tentatively assigned to lutein, that are close to several chlorophyll molecules [Kühlbrandt, W., Wang, N. D., & Fujiyoshi, Y. (1994) Nature 367, 614-621]. A last xanthophyll contribution, with an absorption maximum at 486 nm, does not seem to play a significant role in light-harvesting or in Chl-triplet quenching. On the basis of the assumption that the two structurally resolved xanthophylls are lutein, this 486 nm absorbing xanthophyll should be neoxanthin. The measurements demonstrate that violaxanthin is connected to at least one chlorophyll a with an absorption maximum near 670 nm, whereas the xanthophylls absorbing at 494 nm are connected to at least one chlorophyll a with a peak near 675 nm.

Transient electric birefringence study of intermediate filament formation from vimentin and glial fibrillary acidic protein.

Mg2+-induced polymerization of type III intermediate filament proteins vimentin and glial fibrillary acidic protein was studied by transient electric birefringence. In the absence of MgCl2 we found a net permanent dipole moment, approximately 45-nm-long dimers for vimentin, approximately 65-nm-long tetramers, hexamers, and possibly octamers for both proteins, and 100-nm aggregates for glial fibrillary acidic protein. Controlled oligomerization occurred after the addition of MgCl2. Although the solutions contained (small) aggregates of different sizes, more or less discrete steps in polymer formation were observed, and it was possible to discriminate between an increase in width and length. At the first stage of polymerization (in 0.3 mM MgCl2 for vimentin and 0.2 mM MgCl2 for glial fibrillary acidic protein), the permanent dipole moment disappeared without a change in length of the particles. At higher MgCl2 concentrations, structures of approximately 100 nm were formed, which strongly tended to laterally assemble into full-width intermediate filament structures consisting of about 32 monomers. This contrasts with previous models where first full-width (approximately 10-nm) aggregates are formed, which then increase in length. Subsequently, two discrete elongation steps of 35 nm are observed that increase the length to 135 and 170 nm, respectively. Possible structural models are suggested for the polymerization.