Reversible dark state transitions in fluorophores represent a limiting factor in fluorescence-based ultrasensitive spectroscopy, are a necessary basis for fluorescence-based super-resolution imaging, but may also offer additional, largely orthogonal fluorescence-based readout parameters. In this work, we analyzed the blinking kinetics of Cyanine5 (Cy5) as a bar-coding feature distinguishing Cy5 from rhodamine fluorophores having largely overlapping emission spectra. First, fluorescence correlation spectroscopy (FCS) solution measurements on mixtures of free fluorophores and fluorophore-labeled small unilamellar vesicles (SUVs) showed that Cy5 could be readily distinguished from the rhodamines by its reversible, largely excitation-driven trans-cis isomerization. This was next confirmed by transient state (TRAST) spectroscopy measurements, determining the fluorophore dark state kinetics in a more robust manner, from how the time-averaged fluorescence intensity varies upon modulation of the applied excitation light. TRAST was then combined with wide-field imaging of live cells, whereby Cy5 and rhodamine fluorophores could be distinguished on a whole cell level as well as in spatially resolved, multiplexed images of the cells. Finally, we established a microfluidic TRAST concept and showed how different mixtures of free Cy5 and rhodamine fluorophores and corresponding fluorophore-labeled SUVs could be distinguished on-the-fly when passing through a microfluidic channel. In contrast to FCS, TRAST does not rely on single-molecule detection conditions or a high time resolution and is thus broadly applicable to different biological samples. Therefore, we expect that the bar-coding concept presented in this work can offer an additional useful strategy for fluorescence-based multiplexing that can be implemented on a broad range of both stationary and moving samples.

Fluorescence methods have developed very strongly in the last decade, allowing single- molecule detection sensitivity, high specificity, high time- and spatial resolution, as well as high readout speeds. Together, this makes fluorescencea very central readout modality for biomolecular and cellular studies. The photophysics of the fluorescence emitters, fluorophores, used is of central importance or the performance. Here, the photostability and brightness ofthe fluorophores, but also their blinking properties set the limits of the performance. Fluorophore blinking, arising from photo-induced, non-fluorescent transient states of fluorophores, can however also be taken advantage of cellular and biomolecular studies. Blinking is a prerequisite for essentially allso-called fluorescence-based super-resolution techniques. Moreover, blinkingis often sensitive to micro-environmental parameters such as pH, oxygen concentrations,redox conditions and viscosity. This follows from the fact that the underlying, non-luminescent, dark transient states typically have 10³ to 10⁶ longer lifetimes than the fluorescent excited states of the fluorophores, thereby giving fluorescent molecules in the dark states more time to interactwith their surrounding in biological environment. It can thus be utilized as an alternative readout parameter to provide useful information on molecules and cells and their environments, beyond what can be monitored by traditional fluorescence methods. This thesis takes as one starting point the transient state (TRAST) spectroscopy technique, designed to monitor such long-lived, dark transient states, including triplet, photo-oxidation, photo-reduction and photo-isomerized states of fluorophores, by measuring how the time-averaged fluorescence signal detected in the sample is changed upon systematically varying the excitation modulation.

The major focus of the present thesis work is to further extent the use of long-lived dark transient states of fluorescence emitters in solution, lipid membranes and live cells. For this different TRAST modalities were adapted and developed, and then demonstrated as useful characterization methods. First, we showed how the relaxed brightness requirements of TRAST made it possible to characterize the photo-physical properties of the high triplet yield carboxy-fluorescein dye and its brominated derivatives (paper I). By widefield TRAST measurements, we demonstrated its capability to sense heavy atom effect of bromine and iodide atoms, and how they affected the triplet and long-lived photo-oxidation states and their transitions rates. Next, we developed and demonstrated a concept based on TRAST method and its ability to distinguish fluorophores with different blinking properties as a way to perform fluorescence-based barcoding and multiplexing. This concept, demonstrated by exploiting the by TRAST well distinguishable photophysical transitions of two fluorescent dyes, which emit in the same spectral range, was demonstrated in paper II. In the same work, we also developed a TRAST modality for microfluidic measurements of molecules and lipid vesicles, on which the bar-coding concept could be demonstrated on-the-fly, as the molecules and vesicles passed through the microfluidic channel. Furthermore, with TRAST implemented in camera-based wide-field microscopy, multicolor barcoded images of cells with high spatial resolution could be further investigated for the first time due to the specific blinking dynamics of these labels. The last two papers of this thesis describe further extensions of the TRAST concept, and the monitoring of fluorescence blinking to live cell studies. In paper III, TRAST in a widefield microscopy setting was employed in combination with FCS to study the folding of dye-labelled RNA strands into G-quadruplex structures in solution and live cells using photo-isomerization kinetics of cyanine dye as a readout parameter. Here, we took advantage of the high sensitivity of cyanine dye photoisomerization, to viscosity and steric constraints,and the resulting blinking of the cyanines, to monitor conformation changes of RNAs in live cells. Finally, in paper IV, we demonstrated how it by TRAST imaging, taking advantage of the photo-induced dark states of a mitochondrial localization fluorophore (n-Nonyl Acridine Orange, NAO), is possible to give this localization probe environmental sensing properties as well.

To sum up, the experimental findings and papers included in this thesis show that fluorescence blinking represent a rich source of information for biomolecular and cellular studies. By the TRAST technique, and the variants further adapted and developed in this work, it is shown that it possible tocapture this rich source of complementary information in a broad range of samples. The work in this thesis suggest that further combination of classical fluorescence readouts and a continued development of different TRAST modalities will open yet new windows and provide insights into molecular interaction studies in biological research.

Fluorescensmetoder har utvecklats mycket starkt under det senaste decenniet, vilket har möjliggjort detektion av enstaka molekyler, med hög specificitet, hög upplösning i tid och rum, samt snabba avläsningshastigheter. Sammantaget gör detta fluorescens till en mycket central avläsningsmodalitet för biomolekylära och cellulära studier. Fotofysiken för de fluorescensmarkörer, fluoroforer, som används är av central betydelse hur fluorescensmetoder presterar. Här sätter fluoroforernas fotostabilitet och ljusstyrka, men också deras blinknings- egenskaper tydliga gränser för prestandan. Fluoroforblinkande, som härrör från fotoinducerade, icke-fluorescerande övergående tillstånd av fluoroforer, kan emellertid också utnyttjas inom cellulära och biomolekylära studier. Blinkande är en förutsättning för i stort sett alla så kallade fluorescensbaserade superupplösningstekniker. Dessutom är fluorophorers blinkningsbeteenden ofta känsliga för mikromiljöparametrar som pH, syrekoncentrationer, redoxförhållanden och viskositet. Detta följer av det faktum att de underliggande, icke-luminescerande, mörka transienta tillstånden vanligen har 10³ till 10⁶ gånger längre livslängder än de fluorescerande exciterade tillstånden hos fluoroforerna. Detta ger fluorescerande molekyler i mörka tillstånd mer tid att interagera med sin omgivning i biologiska miljöer. Blinkningskineitk hos fluoroforer kan således användas som en alternativ avläsningsparameter för att ge användbar information om molekyler, celler och deras miljöer, utöver vad som kan erhållas med traditionella fluorescensmetoder. Denna avhandling utgår från en spektroskopiteknik, s.k. transient state (TRAST) spectroscopy, utformad för att övervaka sådana långlivade, mörka transienta tillstånd, inklusive triplett, fotooxidation, fotoreduktion och fotoisomeriserade tillstånd av fluoroforer. Principen med TRAST bygger på att mäta hur den tidsgenomsnittliga fluorescenssignalen som detekteras i ett prov ändras vid systematisk variation av excitationsmoduleringen.

Huvudfokus för denna avhandling är att utvidga möjligheterna med TRAST att monitorera långlivade mörka transienta tillstånd hos fluoroforer i lösning, lipidmembran och levande celler. För detta anpassade och utvecklade vi olika TRAST-modaliteter och visade sedan deras användbarhet i olika applikationer. Först visade vi hur de relativt låga krav på ljusemission från fluoroforer studerade med TRAST gjorde det möjligt att karakterisera de fotofysikaliska egenskaperna hos karboxy-fluorescein, en fluorofor med hög tripletformering (artikel I). Vi visade även metodens förmåga att känna av hur brom- och jodidatomer påverkade bildningen av långlivade triplett-tillstånd och fotooxidation hos karboxyfluorescein, samt de inblandade övergångshastigheterna. Därefter utvecklade och demonstrerade vi ett koncept baserat på TRAST-metoden och visade dess förmåga att särskilja fluoroforer med olika blinkningsegenskaper, som ett sätt att utföra fluorescensbaserad streckkodning och multiplexering. Detta koncept, demonstrerat genom att utnyttja de av TRAST väl urskiljbara fotofysiska övergångarna hos två olika fluoroforer som emitterar i samma spektralområde, demonstrerades i artikel II. I samma arbete utvecklade vi också en TRAST-modalitet för mikrofluidik-mätningar av molekyler och lipidvesiklar, i vilka streckkodningskonceptet kunde demonstreras när molekylerna och vesiklarna passerade genom mikrofluidkanalen. Dessutom, med TRAST implementerad i kamerabaserad widefield mikroskopi, visade vi att flerfärgade bilder av celler kan erhållas med hög rumslig upplösning, där för första gången blinkningsegenskaperna hos de olika fluorforerna kunde användas för att särskilja dem i cellerna. De två sista artiklarna i denna avhandling beskriver ytterligare utvidgningar av TRAST-konceptet, byggande på monitorering av fluorescensblinkningar i levande celler. I artikel III användes TRAST tillsammans med widefield mikroskopi samt FCS för att studera veckningen av färgämnesmärkta RNA-strängar till s.k. G-quadruplex-strukturer, i lösning samt i levande celler. Här utnyttjade vi fotoisomeriseringskinetiken och resulterande fluorescensblinkningar hos ett cyaninfärgämne som avläsningsparameter, och den höga känsligheten hos denna fotoisomerisering för viskositet och steriska begränsningar, för att övervaka konformationsförändringar av RNA i cellerna. Slutligen, i artikel IV, visade vi hur TRAST-avbildning gör det möjligt att utnyttja fotoinducerade mörka tillstånd hos en mitokondriell lokaliseringsfluorofor (n-Nonyl Acridine Orange, NAO), för att ge denna lokaliseringsprob även omgivnings-avkännande egenskaper.

Sammanfattningsvis visar de experimentella resultaten och artiklarna som ingår i denna avhandling att fluorescensblinkningar representerar en rik källa till information för biomolekylära och cellulära studier. Genom TRAST-tekniken, och de varianter som ytterligare anpassats och utvecklats i detta arbete, visar vi att det är möjligt att utnyttja denna rika källa av kompletterande information i ett brett urval av prover. Arbetet i denna avhandling tyder på att ytterligare kombinationer av klassiska fluorescensavläsningar och en fortsatt utveckling av olika TRAST-modaliteter kommer att kunna öppna nya möjligheter för molekylära interaktionsstudier inom biologisk forskning.

Photo-induced dark transient states of fluorophores can pose a problem in fluorescence spectroscopy. However, their typically long lifetimes also make them highly environment sensitive, suggesting fluorophores with prominent dark-state formation yields to be used as microenvironmental sensors in bio-molecular spectroscopy and imaging. In this work, we analyzed the singlet-triplet transitions of fluorescein and three synthesized carboxy-fluorescein derivatives, with one, two or four bromines linked to the anthracence backbone. Using transient state (TRAST) spectroscopy, we found a prominent internal heavy atom (IHA) enhancement of the intersystem crossing (ISC) rates upon bromination, inferred by density functional theory calculations to take place via a higher triplet state, followed by relaxation to the lowest triplet state. A corresponding external heavy atom (EHA) enhancement was found upon adding potassium iodide (KI). Notably, increased KI concentrations still resulted in lowered triplet state buildup in the brominated fluorophores, due to relatively lower enhancements in ISC, than in the triplet decay. Together with an antioxidative effect on the fluorophores, adding KI thus generated a fluorescence enhancement of the brominated fluorophores. By TRAST measurements, analyzing the average fluorescence intensity of fluorescent molecules subject to a systematically varied excitation modulation, dark state transitions within very high triplet yield (>90%) fluorophores can be directly analyzed under biologically relevant conditions. These measurements, not possible by other techniques such as fluorescence correlation spectroscopy, opens for bio-sensing applications based on high triplet yield fluorophores, and for characterization of high triplet yield photodynamic therapy agents, and how they are influenced by IHA and EHA effects.

Guanine (G)-rich nucleic acids are prone to assemble into four-stranded structures, so-called G-quadruplexes. Abnormal GGGGCC repeat elongations, and in particular their folding states, are associated with amyotrophic lateral sclerosis and frontotemporal dementia. Due to methodological constraints however, most studies of G quadruplex structures are restricted to in vitro conditions. Evidence of how GGGGCC repeats form into G-quadruplexes in vivo is sparse. We devised a readout strategy, exploiting the sensitivity of trans-cis isomerization of cyanine dyes to local viscosity and sterical constraints. Thereby, folding states of cyanine-labeled RNA, and in particular G-quadruplexes, can be identified in a sensitive manner. The isomerization kinetics, monitored via fluorescence blinking generated upon transitions between a fluorescent trans isomer and a non-fluorescent cis isomer, was first characterized for RNA with GGGGCC repeats in aqueous solution using fluorescence correlation spectroscopy and transient state (TRAST) monitoring. With TRAST, monitoring the isomerization kinetics from how the average fluorescence intensity varies with laser excitation modulation characteristics, we could then detect folding states of fluorescently tagged RNA introduced into live cells. 

Deep learning methods are of utmost importance in the field of nanotechnology due to their practical applications provide insights into an optimal design of nanomaterials with multi-characteristics. In the context of present research, we propose fully connected deep neural network (DNN) classifiers which have the capability to classify light transmission from polymer nanocomposite films made up of polystyrene (PS) latex particles and multi-walled carbon nanotubes (MWCNTs). For this purpose, collected spectroscopic data were first divided into three classes based on transmitted light intensity, mass fraction of MWCNT nanofillers, annealing temperature, and particle diameter of PS latexes. Bayesian optimizer has then been implemented for each proposed DNN classifier and the most optimal hyperparameters such as activation functions and hidden layer sizes, which provide the best classification accuracy, were acquired through trial-and-error method. Accuracy, cross-entropy loss, precision, recall and F1-score metrics together with confusion matrix and area under curve computed for receiver operating characteristic (ROC) curves have been extensively employed to assess the performance of proposed DNN classifiers. The highest accuracy, precision, recall and F1-score metrics can be achieved for both training and testing data sets when two hidden layer sizes are set to 30 and 20, respectively and sigmoid functions are used in those of hidden layer units. Computational results have indicated that DNNs can be exploited to classify optical transparency of film samples even with a limited amount of experimental data. 

Mitochondrial membranes and their microenviron-ments directly influence and reflect cellular metabolic states but aredifficult to probe on site in live cells. Here, we demonstrate astrategy, showing how the widely used mitochondrial membranelocalizationfluorophore 10-nonyl acridine orange (NAO) can betransformed into a multifunctional probe of membrane micro-environments by monitoring its blinking kinetics. By transient state(TRAST) studies of NAO in small unilamellar vesicles (SUVs),together with computational simulations, we found that NAOexhibits prominent reversible singlet-triplet state transitions andcan act as a light-induced Lewis acid forming a red-emissivedoublet radical. The resulting blinking kinetics are highlyenvironment-sensitive, specifically reflecting local membrane oxy-gen concentrations, redox conditions, membrane charge,fluidity, and lipid compositions. Here, not only cardiolipin concentrationbut also the cardiolipin acyl chain composition was found to strongly influence the NAO blinking kinetics. The blinking kinetics alsoreflect hydroxyl ion-dependent transitions to and from thefluorophore doublet radical, closely coupled to the proton-transfer eventsin the membranes, local pH, and two- and three-dimensional buffering properties on and above the membranes. Following the SUVstudies, we show by TRAST imaging that thefluorescence blinking properties of NAO can be imaged in live cells in a spatiallyresolved manner. Generally, the demonstrated blinking imaging strategy can transform existingfluorophore markers intomultiparametric sensors reflecting conditions of large biological relevance, which are difficult to retrieve by other means. This opensadditional possibilities for fundamental membrane studies in lipid vesicles and live cells

The present study addresses a comparative performance assessment of multivariate regression (MVR) and well-optimized feed-forward, generalized regression and radial basis function neural network models which aimed to predict transmitted light intensity (I-tr) of carbon nanotube (CNT)-loaded polymer nanocomposite films by employing a large set of spectroscopic data collected from photon transmission measurements. To assess prediction performance of each developed model, universally accepted statistical error indices, regression, residual and Taylor diagram analyses were performed. As a novel performance evaluation criterion, 2D kernel density mapping was applied to predicted and experimental I-tr data to visually map out where the correlations are stronger and which data points can be more precisely estimated using the studied models. Employing MVR analysis, empirical equation of I-tr was acquired as a function of only four input elements due to sparseness and repetitive nature of the remaining input variables. Relative importance of each input variable was calculated separately through implementing Garson's algorithm for the best ANN model and mass fraction of CNT nanofillers was found as the most significant input variable. Using interconnection weights and bias values obtained for feed-forward neural network (FFNN) model, a neural predictive formula was derived to model I-tr. in terms of all input variables. 2D kernel density maps computed for each ANN model have shown that correlations between measured data and ANN predicted values are stronger for a specific I-tr range between 0% and 18%. To measure the stability of the ANN models, as a final analysis, 5-fold cross-validation method was applied to whole measurement data and 5 different iterations were additionally performed on each ANN model for 5 different training and test data splits. Statistical results found from 5-fold cross-validation analysis have reaffirmed that FFNN model exhibited outperformed prediction ability over all other ANN models and all FFNN predicted It,. values agreed well with experimental I-tr data. Taken all computational results together, one can adapt our proposed FFNN model and neural predictive formula to predict I-tr of polymer nanocomposite films, which can be made from different polymers and nanofillers, by considering specific data range as presented in this study with statistical details.