Unravelling the Roles of Integral Polypeptides in Excitation Energy Transfer of Photosynthetic RC-LH1 Supercomplexes

Elucidating the photosynthetic processes that occur within the reaction center-light-harvesting 1 (RC-LH1) supercomplexes from purple bacteria is crucial for uncovering the assembly and functional mechanisms of natural photosynthetic systems and underpinning the development of artificial photosynthesis. Here, we examined excitation energy transfer of various RC-LH1 supercomplexes of Rhodobacter sphaeroides using transient absorption spectroscopy, coupled with lifetime density analysis, and studied the roles of the integral transmembrane polypeptides, PufX and PufY, in energy transfer within the RC-LH1 core complex. Our results show that the absence of PufX increases both the LH1 → RC excitation energy transfer lifetime and distribution due to the role of PufX in defining the interaction and orientation of the RC within the LH1 ring. While the absence of PufY leads to the conformational shift of several LH1 subunits toward the RC, it does not result in a marked change in the excitation energy transfer lifetime.


cryo-EM Structure
Figure S2.Separation of photosynthetic membrane complexes from Rba sphaeroides WT and ΔpufXY, using a 10-25 % continuous sucrose gradient.The RC-LH1 core complexes present as exclusively monomers in ΔpufXY.The separation of membrane proteins of the other species were identical to those reported previously. 1 Figure S3.Top.UV/Vis spectrum of dimeric WT RC-LH1 supercomplex.Bottom.TA spectra at selected times, obtained for the dimeric WT RC-LH1 supercomplex at a pump wavelength 877 nm, coinciding with the of peak absorbance of the LH1 BChl(Qy) band.The 862-892 nm spectral region highlighted by the grey box is excluded from subsiquent analysis owing to detection of scattered pump light.Bottom (Inset) TA spectra over the 525-775 nm spectral region magnifided x 12. Figure S4.Top, UV/Vis spectrum of the monomeric ΔpufY RC-LH1 supercomplex.Bottom, TA spectra at selected times, obtained for the monomeric ΔpufY RC-LH1 supercomplex at a pump wavelength 877 nm, coinciding with the of peak absorbance of the LH1 BChl(Qy) band.The 860-890 nm spectral region highlighted by the grey box is excluded from subsiquent analysis owing to detection of scattered pump light.Bottom (Inset), TA spectra over the 525-775 nm spectral region magnifided x 11.

Analysis of the Transient Absorption Data
Several methods are commonly employed to provide kinetic insight into complex datasets generated from time-resolved spectroscopies.3][4][5][6] Briefly, the simplest method is to fit kinetic traces at a single wavelength to a series of exponential decay/growth functions. 7lternatively, global fitting procedures, such as global lifetime analysis (GLA) offer a route to analyse the kinetics of all wavelengths simultaneously.GLA provides a route to visualise the complex timeresolved spectra by decomposing them into a small number (~2-5) of compartment populations.The lifetime of each compartment is fitted as a single exponential decay function, with the pre-exponential factor (xj) allowed to vary with wavelength, Equation 1. 4 Conventionally, GLA results in decayassociated difference spectra (DADS) if the compartments are allowed to decay in parallel.The DADS are presented as plots of pre-exponential factor vs. wavelength, providing a representation of the change to the time-resolved spectra for each lifetime component, τj. 4 ΔA(, ) = ∑   (  , ) In this work we employ lifetime density analysis (LDA) to examine the RC-LH1 kinetics.LDA is based on the principle that the time-resolved data can be represented by a continuous distribution of single exponential functions, Equation 2, where ∅() is the spectral distribution function. 5To make Equation 2readily solvable, the integral needs to be discretized into a quasi-continuous sum of n exponential functions (n = 500 in this work), becoming analogous to Equation 1.Although this discretization indicates that LDA and GLA are similar, and differ only by the value of n, an important distinction is that although both methods are effectively model independent, LDA requires no user input into the expected number of kinetic pathways and their associated lifetimes. 2,5(, LDA results in a three-dimensional density map, x(τ,λ), termed a "lifetime density map", shown in Figure S8 for all six RC-LH1 complexes.Comparison of lifetime density maps is complicated owing to the difficulty in accurately representing the magnitude of the pre-exponential factor with contour/colour maps.Instead, we reduce the three-dimensional lifetime density map to a series of twodimensional plots.Information on the kinetics is obtained through the integration of the modulus of the pre-exponential factor between 750 -950 nm for each lifetime, (which displays the most intense TA spectra features), which we term lifetime density kinetic trace, LDKT.The wavelength dependent average pre-exponential factor of lifetimes associated with each band observed in the LDKT can be calculated, allowing the spectral change associated with each kinetic process to be plotted in two dimensions, which we denote as lifetime averaged difference spectra, LADS.LDKT and LADS are shown in Figure 3

Analysis of kinetic processes with broad distributions
Lifetime distributions obtained from LDA are particular sensitive to noise within the dataset, with even minimal noise levels introducing significant artificial oscillating pre-exponential amplitudes. 8,9egularization is an important procedure within LDA in order to penalize large amplitude coefficients within the fitting process, effectively smoothing the data.Tikhonov regularization, as employed in OPTIMUS, can lead to broadening of kinetic distributions, which is exacerbated by increasing noise levels within the TA dataset. 8To explore whether the broader distributions of lifetimes observed for RC ← LH1 EET of the ΔpufXY and ΔpufX supercomplexes is a result of regularization within the LDA fit, GLA of the TA data was performed.
Initially, a three-compartment parallel scheme was employed, the initial lifetime guesses were taken from the two largest peaks in the LDKT and a third lifetime to deal with species decaying beyond the timeframe probed.The decay associated difference spectra (DADS) resulting from this analysis for t1 and t2 resemble the LADS assigned to EEA and EET, with very similar lifetimes, from the LDA.t3 decays beyond the experimental window, as such no relevance can be interpreted from the fitted lifetime, and contains features assignable to the decay of RC features.The results of these fits to the 900 nm kinetic trace is shown in Figure S17.A good fit is observed for the monomeric and dimeric WT and ΔpufY RC-LH1 supercomplexes, however a poor fit around 50-200 ps is observed for the ΔpufX and ΔpufXY species.
A 4-compartment parallel scheme was then employed, the initial lifetime guesses for the shortest (t1) and longest (t4) lifetime compartments were chosen from the results of the 3-compartment fit, however, those for compartments t2 and t3 were chosen from the half-maximum of the EET lifetime distribution observed in Figure 4 a and b for each RC-LH1 supercomplex.The DADS resulting from this analysis for t2 and t3 are remarkably similar in appearance to both one another, and that obtained in the 3compartment fit.A small improvement of this 4-compartment fit is observed to the data at 900 nm for the monomeric and dimeric WT and ΔpufY RC-LH1 supercomplexes (as may be expected from inclusion of an additional fitting parameter), however, a notable improvement to the fit between 50-200 ps is now observed ΔpufX and ΔpufXY species, Figure S17.Consistent with the increased distribution of EET lifetimes observed in the LDKT, the difference between t2 and t3 is significantly larger for the ΔpufX and ΔpufXY RC-LH1 supercomplexes.Although the resulting decay-associated spectra contain spectral signatures of chromophores of interest, and hence are often assigned to specific photophysical processes, it is important to acknowledge that the fitting results are simply a mathematical model of the data, and they do not necessarily represent a physical kinetic model.These observations (the similarity of the DADS associated with t2 and t3 in the 4-compartment GLA fit) and the considerable improvement in fit upon inclusion of a second decay function representing the EET process for the ΔpufX and ΔpufXY species is consistent with an increased EET distribution in these RC-LH1 supercomplex.However, a direct comparison of EET lifetimes across the range of RC-LH1 supercomplexes studied cannot be easily made from such analysis.Our aim is not to fully describe the complex kinetics observed for these RC-LH1 supercomplexes using GLA, however, we note that dispersive kinetics can be modelled through the use of stretched exponential functions within GLA, 6 and, to the best of our knowledge, there is no publicly available software to perform these fits.

Figure S5 .
Figure S5.Top, UV/Vis spectrum of the dimeric ΔpufY RC-LH1 supercomplex.Bottom, TA spectra at selected times, obtained for the dimeric ΔpufY RC-LH1 supercomplex at a pump wavelength 877 nm, coinciding with the of peak absorbance LH1 BChl(Qy) band.The 862-892 nm spectral region highlighted by the grey box is excluded from subsiquent analysis owing to detection of scattered pump light.Bottom (Inset) TA spectra over the 525-775 nm spectral region magnifided x 12.

Figure S6 .
Figure S6.Top, UV/Vis spectrum of the monomeric ΔpufX RC-LH1 supercomplex.Bottom, TA spectra at selected times, obtained for the monomeric ΔpufX RC-LH1 supercomplex at a pump wavelength 877 nm, coinciding with the of peak absorbance LH1 BChl(Qy) band.The 862-892 nm spectral region highlighted by the grey box is excluded from subsiquent analysis owing to detection of scattered pump light.Bottom (Inset), TA spectra over the 525-775 nm spectral region magnifided x 9.

Figure S7 .
Figure S7.Top.UV/Vis spectrum of the monomeric ΔpufXY RC-LH1 supercomplex.Bottom, TA spectra at selected times, obtained for the monomeric ΔpufXY RC-LH1 supercomplex at a pump wavelength 877 nm, coinciding with the of peak absorbance LH1 BChl(Qy) band.The 862-892 nm spectral region highlighted by the grey box is excluded from subsiquent analysis owing to detection of scattered pump light.Bottom (Inset), TA spectra over the 525-775 nm spectral region magnifided x 12.
Figure S8.Lifetime density map generated from LDA of TA spectra for RC-LH1 complexes.a) WT monomer, b) ΔpufY monomer, c) ΔpufX, d) WT dimer, e) ΔpufY dimer, and f) ΔpufXY.With the colour bar showing the lifetime amplitude.The L-curves for each LDA fit are shown in figureS9.LDA fits performed over 750-950 nm region of TA spectra smoothed by 5 nearest neighbours over the whole spectral window apart from 15 nm on either side of the pump wavelength which is contaminated by pump scatter.The TA spectra are fitted with 500 lifetimes spread on a log scale between 0.03 ps to 20 ns.Lifetimes < 0.9 ps have been cut for clarity as at early timeframes LDMs are dominated by features assignable to remaining coherent artefact components.

Figure S11 .
Figure S11.Combined lifetime density kinetic traces (left panel), UV-VIS (top right) and lifetime averaged difference spectra (right panel 2 to 5) of ΔpufY monomer.The lifetime averaged difference spectra panels show the wavelength dependent average pre-exponential factor of lifetimes within the shaded area of same colour.

Figure S12 .
Figure S12.Combined lifetime density kinetic traces (left panel), UV-VIS (top right) and lifetime averaged difference spectra (right panel 2 to 5) of ΔpufY dimer.The lifetime averaged difference spectra panels show the wavelength dependent average pre-exponential factor of lifetimes within the shaded area of same colour.

Figure S13 .
Figure S13.Combined lifetime density kinetic traces (left panel), UV-VIS (top right) and lifetime averaged difference spectra (right panel 2 to 4) of ΔpufX monomer.The lifetime averaged difference spectra panels show the wavelength dependent average pre-exponential factor of lifetimes within the shaded area of same colour.

Figure S14 .
Figure S14.Combined lifetime density kinetic traces (left panel), UV-VIS (top right) and lifetime averaged difference spectra (right panel 2 to of ΔpufXY monomer.The lifetime averaged difference spectra panels show the wavelength dependent average pre-exponential factor of lifetimes within the shaded area of same colour.

Figure
Figure S15.DADs generated from performing a 3-compartment parallel GLA with the initial guess coming from peaks found in LDKT for RC-LH1 complexes, a) WT monomer, b) ΔpufY monomer, c) ΔpufX, d) WT dimer, e) ΔpufY dimer, and f) ΔpufXY.Lifetimes resulting from the fit are indicated within the legend.

Figure S16 .
Figure S16.DADs generated from performing a 4 compartment parallel GLA with initial guess of t1 coming from peak found in LDA while t2 and t3 our set to be either side of the main LDA peak assigned to EET with t4 set to 10000ps to deal with residual signal of TA data for RC-LH1 complexes, a) WT monomer, b) ΔpufY monomer, c) ΔpufX, d) WT dimer, e) ΔpufY dimer, and f) ΔpufXY.Lifetimes resulting from the fit are indicated within the legend.

Figure S17 .
Figure S17.Fits to the kinetics observed at 900nm of the raw data (black dots), from 3-compartment GLA with single EET lifetime (blue) and 4-compartment GLA with two EET lifetimes (red) [see text] for a) WT monomer, b) ΔpufY monomer, c) ΔpufX, d) WT dimer, e) ΔpufY dimer, and f) ΔpufXY.A magnified view of the 50 -500 ps timescale, along with the associated residuals of the fits, are shown in the insets.

8 .
Figure S18.UV/Vis spectra of the WT monomer, ΔpufY monomer, ΔpufXY monomer and ΔpufX monomer as indicated.The inset shows a zoom of the LH1 BChl(QY) absorption peak.Note that the ΔpufX and ΔpufXY strains were grown microoxically in the dark, given their inability to photosynthesize.The ΔpufX and ΔpufXY RC-LH1 complexes have a broad absorption and maxima at 481, 509 and 545 nm, and thus a visibly different color, because of the conversion of the carotenoid spheroidene to spheroidenone under the microoxic culture conditions.

Figure S19 .
Figure S19.TA spectra of the WT monomer, ΔpufY monomer, ΔpufXY monomer and ΔpufX monomer as indicated at a pump-probe delay of a) 0.25 ps and b) 10 ps.

Figure S23 .
Figure S23.Distances between LH1 BChl a to the closest BChl a of the special pair in the RC of: (a) WT monomer, (b) ΔpufY monomer, and c) ΔpufX monomer.Measurements in bracket indicates difference to the WT monomer

Figure S24 .
Figure S24.Distances between LH1 BChl a to the closest BChl a of the special pair in the RC of: (a) WT dimer type 1, and (b) WT dimer type 2. Measurement in bracket indicates difference to the WT dimer type 1.

Figure S25 .
Figure S25.Distances between LH1 BChl a to the closest BChl a of the special pair in the RC of: (a) ΔpufY dimer type 1 , and (b) ΔpufY dimer type 2. Measurement in bracket indicates difference to the WT dimer type 1.