WO2002086472A1 - Molecular rotor derivatives and methods of use - Google Patents

Abstract

The present invention concerns the development and use of fluorescent dyes known as molecular rotors which vary in fluorescence intensity based on viscosity of the environment. The inventor further relates to a class of molecular motors that at modified with a hydrocarbon chain or hydrophilic group to allow for the measurement of membrane or liquid viscosity, respectively.

Description

MOLECULAR ROTOR DERIVATIVES AND METHODS OF USE

BACKGROUND OF THE DSfVENTION Field of the Invention

The present invention generally concerns the development and use of fluorescent dyes known as molecular rotors which vary in fluorescence intensity based on the viscosity of the environment. The invention further relates to a class of fluorescent dyes called molecular rotors that are modified with a hydrocarbon chain or hydrophilic group to allow for the measurement of membrane or liquid viscosity, respectively.

Description of the prior art

Molecular Rotors

Molecular rotors are fluorescent molecules that exhibit viscosity-dependent fluorescence quantum yield, potentially allowing direct measurements of cell membrane viscosity in cultured cells.

A group of related vinyl-aromatic compounds have been shown to undergo intramolecular charge transfers when excited by photon absorption [5]. One member of the group, 9-(dicyanovinyl)-julolidine (Figure 1), has been extensively characterized. For the intramolecular charge-transfer, the julolidine group functions as the electron-donor, and the dicyano-group as electron acceptor [5]. The relaxation of the excited state can either occur through photon emission or non-radiative processes [7]. When k_r and knr represent the rates of radiative and non-radiative relaxation, respectively, the quantum yield D is expressed by Equation 1[7]:

Considering the fact that the reorientation of the rotor - the process which leads to the non-radiative relaxation - is dependent on the free volume of the environment, the quantum yield depends on the temperature T and the viscosity D of the medium according to Equation 2 [7]:

C'and x are constants. In a simplified form, and under the assumption of a constant temperature, this relationship is better known as the Fόrster-Hoffmann-equation, Equation 2 [2]:

If the absorbed light intensity I_ab is known, the emission intensity I_em and the quantum yield are directly and proportionally related through () [7]. Therefore, changes in the viscosity of the environment can directly be determined by measuring the emission intensities 1} and I₂ before and after the viscosity change. With the corresponding viscosities D ι and D₂ , and under the assumption of constant temperature and constant absorption, Equation 4 can be obtained [3]:

In Equation 4, a background intensity I₀ has been allowed for, which is the non- fluorescent amount of light caused by filter bleed-through, scatter, and ambient light. Depending on the measurement setup, the background intensity may be non-neglectable.

For DCVJ, the constant x has been shown to be 0.6 [2]. Experiments that were conducted in media with different viscosities have confirmed the above equation [6]. The quantum yield increases approximately by a factor of 30 when dissolved in 1-propanol and glycerol, respectively [6]. Furthermore, DCVJ has been used to probe the microviscosity changes in phospholipid bilayers, showing a viscosity change between 70 and 120 cP at temperatures from 60 to 10°C [6].

Potential Applications of Molecular Rotors

Plasma Blood Viscometer For Resuscitation Therapy As world-wide blood shortages grow more acute, there is a growing need for more effective means of stabilizing and maintaining physiologic well being in the millions of patients world-wide subjected to acute blood loss as a consequence of trauma or surgical procedures. This need continues to drive forward extensive efforts to develop blood substitutes and new plasma expanders. However, recent experimentation shows that none of the commercially available plasma expanders or blood substitute products are able to reverse changes that occur in the microcirculation due to severe blood losses. This universal limitation in current technology results leads to impaired tissue delivery of oxygen and reduced clearance of metabolic waste products at the cellular level. These irreversible changes are highly correlated with dramatic elevations in morbidity and mortality in animal models and patients.

Conventional theories used to implement fluid replacement and volume expansion advocate the use of low viscosity fluids to maintain adequate flow rate in the microcirculation during conditions of severe volume depletion. However, recent research indicates that below a critical threshold of hemodilution, low viscosity fluids contribute to a shutdown of the microvascular circulation, preventing tissue oxygen delivery and permitting build-up of toxic byproducts of cell metabolism. In contrast, utilization of high viscosity plasma expanders successfully counters the shutdown of these vessels even when used far beyond the limits in which current plasma expanders are used as volume restitution fluids.

Blood viscosity

Blood is a viscous medium composed of cells and liquid plasma. The hematocrit (Hct) of blood is the percent of the blood that is taken up by cells, predominantly red blood cells. A hematocrit of 45% means that 45 percent of the blood volume is cells and the remainder is plasma. Normal hematocrit ranges from approximately 38 for women to 42 for men. As the percentage of red blood cells increases, there is a corresponding rise in the friction that exists between successive layers of blood as it passes through the circulatory system. This friction caused by the viscosity of blood. Normal blood viscosity is a function of hematocrit squared and is about 4 cP (37 DC). There are also various plasma proteins which contribute to the viscosity of blood. Normal plasma viscosity is in the range of 1.15 - 1.35 cP (37 DC; the viscosity of water at this temperature is 0.71 cP). In general, once the viscosity of plasma is measured the viscosity of blood can be calculated if the hematocrit is known. Blood losses diminish the viscosity of both blood and plasma due to autotransfusion, a mechanism by which the organism causes relatively protein free tissue fluid to move into the vascular compartment.

Blood losses and the transfusion trigger

The concept of transfusion trigger has been widely described and accepted as the maximum Aacceptable@ blood loss, below which whole blood transfusions are required. The normal hemoglobin (Hb) content in adults is approximately 14 gm/dl and is important to consider since it is the Hb that carries the oxygen to be delivered to the tissues. When half of the red blood cell population is lost, the Hb level falls to 7 gm/dl (Hct ~22.5). It is at this point that blood transfusions are immediately implemented since it is believed that there is an insufficient number of red blood cells to continue to deliver oxygen to the tissues. This transfusion trigger is universally accepted, as is the explanation for requirement of whole blood replacement.

There is an explicit assumption that must be met for a whole blood transfusion to be useful in acute trauma settings:

The newly transfused red blood cells must be distributed throughout the circulatory system, particularly in the capillary network where red blood cells transfer their oxygen to cells and carry away toxic waste products from cell metabolism

Plasma viscosity and functional capillary density

The assumption that the red blood cells are distributed uniformly through the tissue is almost assuredly incorrect for has demonstrated that in settings of acute blood loss, at or below the transfusion trigger, normal capillary flow is tremendously reduced, preventing red blood cells from reaching their destination. Even if the transfused blood were capable of carrying oxygen, it would be largely ineffective at the tissue capillary level. The work by Tsai et. al, 1998, shows that the mechanism for capillary flow interruption resides in a loss of mechanical and physical forces normally present to maintain capillary distension. The most important of these forces is plasma viscosity and whereas normal or low viscosity plasma expanders fail to maintain capillary flow, high viscosity plasma expanders such as high molecular weight starch and PEG-dextrans have been demonstrated to maintain flow well below the transfusion trigger.

When blood viscosity becomes too low there is a widespread and catastrophic closure of functional capillary beds, expressed as a reduction in the functional capillary density (FCD). In settings of severe blood loss, this reduction in FCD is highly correlated with mortality in animal models (Kerger el al, 1996) and is not corrected by any current low viscosity plasma expanders.

Blood and plasma viscosity plays a critical role in maintaining the microcirculation open and functional via two major mechanisms. First, viscosity acts on the vasculature by stimulating the endothelial cells, which line the blood vessels, to produce crucial chemical signals, which maintain vessel patency and vasodilation. Secondly, the physical forces exerted by normal viscosity plasma result in open capillaries. Under conditions of reduced viscosity, the capillaries collapse on themselves, preventing further flow of plasma or red blood cells, impairing oxygen delivery and waste product exchange. In conditions under which the red blood cell mass is sufficient to deliver oxygen to tissues impaired capillary blood flow results in no removal of metabolic waste products.

Plasma viscosity and microvascular function

There is a specific relationship between plasma viscosity and functional capillary density described in the Tsai paper can be summarized as follows: Reduction of hematocrit by 75% of the normal value with a plasma expander that decrease plasma viscosity below 1.5 cp cause a pathological reduction of functional capillary density and tissue perfusion, while the opposite is true if plasma viscosity can be maintained above 2.0 cp.

Thus a major limitation of current volume replacement or volume expanding therapy is that it does not addresses the reduction in plasma viscosity and the prevention of significant closure of capillary networks. The new theory indicates that it is critical to maintain microvascular function as a primary endpoint of resuscitation therapy. If this is accomplished, oxygen delivery and clearance of metabolic waste products will be maintained even under conditions of extreme lowering of the red blood cell mass. This has wide raging implications for economizing blood transfusions, while improving the efficacy of current available plasma expanders, and blood substitutes under development.

The need for a plasma viscometer

The present understanding of fluid resuscitation indicates that information on plasma and blood viscosity will become an integral part of the decision making process on the form of resuscitation to be administered, and determine the selection of the physical properties of the fluid to be used. Moderate losses of blood (red blood cells) can be safely corrected with simple fluids, and do not require special considerations in terms of blood viscosity, and in fact may be beneficial corrected with low viscosity plasma expanders. Up to a 50% change in red blood cell concentration the cardiovasculature easily adapts to the lowered blood viscosity by increasing cardiac output with the net effect that the distribution of mechanical effects remains nearly normal up to this point.

A critical point is reached when blood losses exceed 50%. When this occurs conventional therapy demands the use of blood indicates to restore volume and oxygen carrying capacity, up to a reposition of hematocrit to about 75% of normal. However the same, or even improved resuscitation, can be obtained by blood volume restitution with a high viscosity solution that restores the viscosity of the fluid in the circulation to a value intermediate between normal blood and the value prevalent in the hemorrhaged condition. This is significant because this procedure can be carried out in the field, does not require blood typing, and becomes life saving when blood is not available.

In attempting to restore the viscosity of the fluid in the circulation it is important to achieve a specific threshold, and also not to exceed normal blood viscosity. The principal unknown in performing this procedure is the volume of remaining circulating fluid, thus the safest approach is that of monitoring the resuscitation operation by direct plasm viscosity measurements.

Blood plasma is an important factor in physiology and disease. Blood plasma viscosity anomalies are associated with diseases such as diabetes, hypertension, infections and infarctions [8-12], A direct cause and effect relationship has not been established between plasma viscosity and the development of these conditions, thus clinical medicine has not to date provided a strong stimulus for the development of practical techniques for characterizing either blood or plasma viscosity.

In transfusion medicine, however, the management and measurement of blood plasma viscosity is becoming a central issue affecting blood replacement with plasma expanders [13] and the development of artificial blood [7] consisting of colloidal solutions of modified hemoglobin.Information about the distribution of blood viscosity in the circulation upon the introduction of plasma expanders or artificial blood is particularly relevant in terms of shear stress mechanotransduction to the endothelium, which regulates the shear stress dependent production of vasoactive mediators [15] such as prostacyclin and NO, and modulates apoptosis [16]. In this context, information on plasma viscosity is a key element in understanding the magnitude of the stimulus to which endothelium is exposed, since shear stress is the product of vessel wall velocity gradient and local plasma/blood viscosity. The measurement of viscosity of macromolecular solutions and particularly blood plasma presently requires mechanical tests. Standard mechanical viscometers include the capillary viscometer, where fluid is sheared by flow past the stationary inner wall of a capillary tube [17,18], the falling-ball viscometer, and the rotational viscometer, where the test liquid is sheared between two surfaces, one fixed and one moving, with the torque being related to viscosity [17]. For the measurement of blood plasma, a specialized capillary viscometer (Harkness Viscometer) has been recommended by the International Committee for Standardization in Haematology [19] which allows to measure sample sizes as low as 0.5 ml within one minute. However, reliable results cannot be obtained unless the effects of air/solution interfaces can be accounted for or controlled [8,20] and the surfaces in contact with the test solutions are meticulously cleaned. These requirements complicate obtaining viscosity data on blood plasma in a clinical setting and limit the rate at which sequential measurements can be made if the same apparatus is used, if sample sizes of 3 - 4 ml are necessary due to the time required for separating red blood cells from plasma. Also, mechanical instruments are unable to assess microviscosity in microscopic environments or samples. Thus a potentially useful parameter for characterizing biological fluid properties is not available due to the lack of a suitable measurement technique/principle. A solution to this problem may be offered through the use of fluorescent molecules that are viscosity-sensitive.

Those molecules, commonly referred to as fluorescent molecular rotors, belong to the group of twisted intramolecular charge-transfer complexes (TICT). Photoexcitation leads to an electron-transfer from the donor group to the acceptor group. Relaxation can either occur through radiation (fluorescence) or intramolecular rotation (thermally induced non-fluorescent relaxation). The preferred relaxation mechanism, intramolecular rotation, is reduced in solvents with low free volume. Therefore, the quantum yield of a molecular rotor increases with decreasing free volume of the solvent [21]. Free volume and viscosity are related [22], which links quantum yield directly to the viscosity of the microenvironment. The relationship between fluorescence quantum yield D and the viscosity D of the solvent has been derived analytically [21,2,6] and experimentally [40,42] and is known as the Fόrster-Hofrmann-Equation [2 supra].

*Cell Membranes

When used for probing viscosity changes in cell membranes, however, DCVJ is associated with some problems. DCVJ is water-insoluble. A staining solution for the cells, though, is aqueous. For experiments on liposomes [6], this difficulty can be overcome by preparing the phospholipid base readily stained before forming the liposomes. In live cells, DCVJ can be bound to serum proteins [3]. Depending on the serum, the binding capacity is limited. Furthermore, protein-bound DCVJ migrates into the cytoplasm, thus adding a high level of constant background signal which reduces the sensitivity of the probe.

Most often, measurement of membrane viscosity is carried out using FRAP or fluorescence anisotropy. The method of FRAP is based on directing a focused laser beam at the cell membrane, which destroys the fluorescent dye in a defined region. Diffusion within the membrane allows dye from the neighboring environment to migrate into the bleached spot. The half-life of the fluorescence recovery is inversely proportional to the diffusion coefficient of the dye[46], which in turn is proportional to the fluidity. There are two major disadvantages associated with FRAP. Firstly, the measurement process is slow, since times between 20 and 60 s are required in order to allow sufficient recovery for the reliable determination of the half-life. This fact strongly limits the temporal resolution of any FRAP measurement and does not permit measurement of dynamic processes. Secondly, FRAP is limited in its resolution, as reduction of the radius of the bleached spot increases the error associated with the measurement. In addition, the bleaching pulse introduces power densities up to 1 MW/cm² to the membrane [46], leading both to the generation of free radicals by photolysis as well as local heating with potential damage to proteins [46]. Especially the second effect may cause artefacts, because the resulting cross-linking of the proteins [46] leads to decreased membrane diffusivity.

Membrane viscosity measurements by fluorescence anisotropy utilize the fact that certain molecules can only be excited by light waves in one polarization plane. The emitted light, again, is polarized. During the excited lifetime, the molecule can rotate, thus shifting the plane of emitted light. The ratio of emitted light parallel versus perpendicular to the plane of excitation is a nonlinear function of the viscosity [42]. While the spatial resolution is essentially determined by the limits of the microscope, rapid photobleaching of the probes both confines the maximum sensitivity and the exposure time of the probe to the excitation light. Furthermore, the polarizing filters in the emission light path generally absorb a significant portion of the emitted light. Moreover, it was shown in a recent study that the deformation of cells caused by shear stress may produce artifacts that are similar to an increase in membrane fluidity [47], which makes anisotropy observations difficult to interpret.

More recently, environment-sensitive probes have been used in membrane viscosity studies. In this category of fluorescent probes belong a group of compounds that have been shown to undergo intramolecular charge transfer when excited by photon absorption [5]. One member of the group, DCVJ (1), has been extensively characterized.

The assessment of membrane viscosity using molecular rotors overcomes the limits posed by FRAP or fluorescence anisotropy measurements. The spatial resolution of the method is potentially as high as it is for fluorescence anisotropy, because it is primarily limited by the optics of the microscope. The molecular rotor also responds to changes of viscosity almost instantaneously (within the nanosecond range) [5]. This allows a very high temporal resolution of the measurements, so that dynamic observations become possible.

The major disadvantage associated with DCVJ when used to probe cell membranes is the affinity for tubulin and proteins in the cytoplasm. The experiments presented in [39] are based on cultured cells, where a significant cytoplasmic staining exists, and a continuous exchange of the dye between the cytoplasm and the membrane takes place. Both the cytoskeleton with its slow response to shear stress [48,49], and the cytoplasm, which slightly increases viscosity with shear stress [50], do not contribute to the signal measured, but provide a background fluorescence which reduces the measurement sensitivity. Furthermore, the relative intensity of the background is unknown and not easily determined, making quantitative measurements of membrane viscosity changes difficult.

It would therefore be desirable to obtain a fluorescent molecule that has the viscosity-dependent properties of DCVJ, but shows a higher affinity to the cell membrane.

SUMMARY OF THE INVENTION The present invention contemplates

These and other aspects and attributes of the present invention will become mcreasingly clear upon reference to the following drawings and accompanying specification.

BRIEF DESCRIPTION OF THE DRAWINGS

BRIEF DESCRJPπON OF THE DRAWINGS

Figure 1: 9-(dicyanovinyl)-julolidine (DCVJ)

Figure 2: Excitation and emission spectra of CCVJ in pure plasma and a plasma:PS solution mixture. This figure shows representative excitation (dashed lines) and emission (solid lines) spectra for CCVJ, taken at 475 nm emission and 430 nm excitation wavelength, respectively. All other probes showed similar spectral behavior. It can clearly be seen that the fluorescence intensity is markedly higher in the high-viscosity sample (40% plasma and 60% PS solution; thick lines) as compared to pure plasma (thin lines).

Figure 3: Intensity values (average ± SD) for 5 independent fluorescence-based viscosity measurements using CCVJ as probe. All average values are significantly different from their neighbors (p<0.0001; p=0.0016 between 3.34 mPas and 425 mPas). There is no overlap of the values between adjoining bars.

Figure 4: Intensity values (average ± SD) or 5 independent fluorescence-based viscosity measurements with smaller viscosity increments than those in Figure 2. Shown are the results of measurements taken with CCVJ (panel a) and CCVJ-ME (panel b). All average values are significantly different (p<0.0001) from their neighbors. There is no overlap of the values between adjoining bars.

Figure 5: Apparent non-Newtonian behavior of blood plasma when tested mechanically. Shown are representative measurements for plasma (left graph) and a mixture of 60% pentastarch solution with 40% plasma (right graph). The measured viscosity decreases with increasing shear stress. At high shear stress values, the change is relatively small, the viscosities where data points below and above 450 s^"1 normally deviate less than 5%. For this reason, viscosity values at 450 s^"1 were used as viscosity reference values.

Figure 6: Calibration curves to obtain the constants and in Equation 2. Intensity is plotted over viscosity in double-logarithmic scale The slope of each line is lower in the high viscosity range, which may in part be attributed to the apparent saturation effect visible in the data point with the highest viscosity.

Figure 7: Chemical structure of 9-(dicyanovinyl)-julolidine (1) and the basic structure of the compounds synthesized (2). One of the nitrile functionalities of DCVJ has been replaced by a carboxylic group, allowing the addition of hydrocarbon chains R. Figure 7A: Further synthesis as in Figure 7 resulting in a double side chain having one or two specificities.

Figure 8: Synthesis of the compounds 2a - 2g. Julolidine 3 was formylated to afford julolidine aldehyde 4. The beta-cyanosters 7 were obtained by esterification of cyanoacetic acid 5 with the corresponding alcohol 6. Condensation of the esters 7 with julolidine aldehyde 4 produced the desired probes 2.

Figure 9: Excitation spectra (dashed lines) and emission spectra (solid lines) for DCVJ and FCVJ in ethylene glycol. The spectra of the other compounds are similar. All probes exhibit a double excitation maximum in the blue range, and a single emission maximum in the green range.

Figure 10: Changes of fluorescence intensity as a function of the increase of shear stress for compounds 2b, 2c, 2d, and 2g (FCVJ), as well as for DCVJ. All compounds except 2c differ significantly (pO.OOOl) from DCVJ.

Figure 11: Measured signal intensity of cells stained with DCVJ and FCVJ (2g) as a response to different step stimuli of shear stress. Both curves were subjected to a noise filter and an exponential correction for photobleaching. FCVJ shows a dramatically higher response. This allows a significantly higher sensitivity, while the noise component is greatly reduced.

Figure 12: Decay of fluorescence intensity of the probes in stained cells due to photobleaching. FCVJ shows a significantly higher decay half-time than DCVJ (p<0.0005). Compound 2c, however, is less stable than DCVJ.

Figure 13: Cuvette geometries:

Computer simulations of a typical perpendicular illumination arrangement (left), common to all commercially available fluorometers, and a front-face illumination arrangement (right). Both drawings show LED excitation (A), fluorescence detection (B) and - as an extension to commercial fluorometers - absorption detection (C). The front-face illumination model features a special, narrow cuvette (D) to minimize the influence of fluid turbidity. Excitation light passes through this cuvette at a 45° angle. Depending on turbidity and dye concentration, attenuated excitation light reaches the absorption sensor and can be used to compensate for fluid absorption characteristics and dye concentration. Some excitation light reflects off the cuvette walls at an angle of 45° and thus does not reach the emission sensor. The large area of the cuvette facing the emission sensor allows it to collect more light, increasing the sensitivity.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Example I

A Novel Approach to Blood Plasma Measurement

Materials and Methods

Chemicals and instrumentation

Human blood plasma was purchased from the San Diego Blood Bank. 9- (dicyanovinyl)-julolidine (DCVJ) and 9-(2-carboxy-2-cyanovinyl)-julolidine (CCVJ) were purchased from Helix Research (Springfield, OR). All other fluorescent compounds were synthesized as described in [19]. Stock solutions of the probes were prepared at a concentration of 20 mM in fluoroscopy-grade dimethylsulfoxide (DMSO, Sigma). Plasma viscosity was changed by adding the clinical-grade high-viscosity plasma expander pentastarch (10% by weight, average molecular weight 260 kD) in physiological saline solution (Pentaspan^®, DuPont Pharma, Wilmington DW). Fluorescence measurements were preformed on a Shimadzu RF-1501 fluorospectrometer (Shimadzu, Kyoto, Japan) using standard methacrylate cuvettes (Fisher Scientific, Pittsburg, PA). Viscosity was measured using a Brookfϊeld DV-JJ+ cone-and-plate viscometer with CP-40 spindle (Brookfield, Middleboro, MA). Viscosity values were obtained at discrete shear rate settings of 75, 90, 150, 225, 450, 750 s^"1 (10, 12, 20, 30, 60, 100 rpm). Viscosities measured at 60 rpm (450 s^"1 ) were used as a reference in order to avoid the apparent non-Newtonian behavior at low shear rates, since this was not present beyond this shear rate. This shear rate allowed to measure the viscosity of all the samples without exceeding the torque limitations of the instrument.

Examination of the suitability of different probes

Blood plasma was kept frozen and thawed at room temperature overnight for use. Coarse precipitates were eliminated by centrifugation at 180 Dg for 10 minutes. In order to ensure homogeneous and precise distribution of the fluorescent probe, pre-stained plasma was prepared by mixing 20 Dl of the probe stock solution in 6 ml plasma under vigorous stirring. For each probe, a series of mixtures of plasma with pentastarch solution (PS solution) was prepared according to Table la. Fluorescence emission spectra were acquired at an excitation wavelength of 440 nm, and the maximum intensity determined irrespective of the wavelength of the maximum. Immediately after the fluorescence measurement, the viscosity of the plasma: PS solution mixture was determined in the cone-and-plate viscometer as described above. All experiments, fluorescence measurement and viscometry, were performed at a constant temperature of 21 DC. The above experiment was repeated for five probes: DCVJ, CCVJ, CCVJ- methylester, CC VJ-ethylester, and CC VJ-butylester. As a result, a 5x5 matrix of 5 probes and 5 viscosities was obtained (Table 2).

Precision assessment of the measurement method

In order to determine the repeatability of the measurement procedure each sample of the matrix was divided into 5 aliquots which were independently mixed with the probe CCVJ (Table la). Fluorescence was measured and viscosity was calculated according to equations 2 & 3, leading to the calculation of standard deviations and coefficients of variability. In order to obtain values at lower viscosity increments, the experiment was repeated with modified mixture ratios as described in Table lb using both CCVJ and CCVJ-ME as probes. Figure 3 shows the measured average values ± SD of the 5 measurements for each mixture of plasma and PS solution. There is no overlap between the measured values, and all average values are significantly different with p < 0.0001 (p = 0.0016 for the last two bars) (Figure4). Coefficients of variability (standard deviation divided by the average within each column) ranged from 0.017 to 0.046.

Results

General behavior of molecular rotors in plasma

All probes dissolved in plasma without forming precipitates, and blood plasma stained with the fluorescent probes exhibited typical spectral behavior with absorption maxima in the blue range (440 - 454 nm, depending on the probe) and emission in the green range (464 - 494 nm). Figure 2 shows typical excitation and emission spectra of CCVJ in blood plasma and a mixture of 40% plasma and 60% PS solution. All other spectra were similar with only minor shifts of the maxima, and all probes exhibited increased emission intensity when dissolved in fluids with igher viscosity as covered in detail in the following sections.

Viscosity-dependent emission intensity for all probes in plasma: PS solution mixtures

Table 2 gives an overview over the measured maximum intensities of the probes DCVJ, CCVJ, CCVJ-methylester, CCVJ-ethylester, and CCVJ-butylester, normalized by the intensity measured in pure blood plasma. Pure blood plasma and mixtures of 80% plasma with 20% PS solution, 60% plasma with 40% PS solution, 40% plasma with 60% PS solution and 20% plasma with 80% PS solution were used. The intensity increase of fluorescence intensity in a high-viscosity mixture containing 80% PS solution over pure plasma is about two-fold (with DCVJ and CCVJ) to 3.5-fold (with CCVJ-esters). Precision assessment of the measurement method

In order to determine the repeatability of the measurement procedure each sample of the matrix was divided into 5 aliquots which were independently mixed with the probe CCVJ (Table la). Fluorescence was measured and viscosity was calculated according to Equations 2 and 3, leading to the calculation of standard deviations and coefficients of variability. In order to obtain values at lower viscosity increments, the experiment was repeated with modified mixture ratios as described in Table lb using both CCVJ and CCVJ-ME as probes. Figure 3 shows the measured average values ± SD of the 5 measurements for each mixture of plasma and PS solution. There is no overlap between the measured values, and all average values are significantly different with p < 0.0001 (p = 0.0016 for the last two bars) (Figure 4). Coefficients of variability (standard deviation divided by the average within each column) ranged from 0.017 to 0.046.

Relationship between viscosity and intensity

Measured fluorescence data was compared to mechanically obtained viscosity. Pure blood plasma as well as mixtures of plasma and PS solution exhibited shear-rate dependent viscosity (Figure 5). Viscosity values for further data analysis were obtained at matched shear rates, 450 s^'1, which is the highest shear rate common to all measurements. Thus, for each sample, one data pair (viscosity, intensity) was obtained. The application of the Fόrster-Hoffmann-Equation (Equation 1) to the data pairs leads to the calibration curve required to mathematically relate intensity and viscosity values.

Data points of intensity over viscosity were plotted in double-logarithmic scale (Figure 6) and the slope determined using a least-squares fit. The following calibration equation to calculate viscosity D from fluorescence intensity/ was derived from Equation 1:

with the constants D and D related to the constants in Equation 1 through Also, in Equation 2 the measured intensity I was used, which is proportionally related to the quantum yield D, but also depends on geometry, concentration and incident light intensity. Empirical curve-fitting yielded the constants D = 1.18 and D = 7.2 xlO^"3 for the experiments with the high viscosity range (Table la). For the experiments with the smaller viscosity increment (Table lb) D = 0.77, D = 11.7^χ 10^"3 were obtained for CCVJ and G = 0.78, D = 7.8^χ 10^"3 for CCVJ-ME. The average deviation between the fluorescence-based results and measurements carried out by mechanical viscometry for the first set of experiments was 0.8% (range: -3.9% to 6.9%). For the second experiments at lower viscosity increments the average deviation using CCVJ was 1.8% (range: -6.5% to 6.4%) when compared to measured values. An average deviation of 1.5% (range: -7.2% to 4.8%) against measured values was found using CCVJ-ME. Correlation coefficients between mechanical and fluorescence-based viscosity values were ri =0.987 for the first experiment (Table la), r₂ =0.996 for the second experiment (Table lb) with CCVJ and r₃ =0.998 for CCVJ-ME.

Discussion

The principal finding of this study is that the viscosity-dependent fluorescence of molecular rotors can be used to measure blood plasma viscosity, thus providing a new method for viscosity measurement in a field where mechanical methods to measure viscosity (capillary viscometer, falling ball viscometer, rotational viscometer) have been used exclusively. All mechanical methods have in common that the fluid is subjected to shear forces, and the resistance of the fluid to these forces (internal friction) is measured. The internal friction of a fluid is proportional to the dynamic *viscosity D and the velocity gradient (shear rate) between layers of different velocities.

Fluorescent molecular rotors, used as viscosity probes, rely on a different mechanism than the application of shear forces. They provide two modes of relaxation, fluorescence and nonradiative intramolecular rotation. The latter is dependent on the free volume of the environment, which in turn is related to its viscosity. Through the free-volume theory of viscosity, photophysical measurements using molecular rotors become independent from mechanical forces. The fluid under observation is not sheared during the measurement, and protein-surface interactions that may lead to artifacts at low shear rates do not affect the result. Further advantages over mechanical measurements include small sample volumes needed to perform fluorescence measurements (microcuvettes typically have a volume of 200-250 Dl), and the high speed of the readout: Using fixed wavelength filters, intensity can be measured within fractions of a second.

Molecular rotors provide a measurement of viscosity for conditions at zero shear rate (zero flow), therefore the viscosity data is applicable solely if the fluid has

Newtonian behavior. Presently it is generally accepted that plasma is a Newtonian fluid as indicated by the studies of Cokelet [20], Harkness [8] and Reinhart et al. [18] although the work of Dintenfass [24] and Sharma et al. [25] indicates that it exhibits shear thinning. Our own measurements show that plasma is Newtonian at shear rates above 250 s^"1, the shear rate found in blood vessels under normal conditions. At lower shear rates our mechanically measured viscosity appears to increase slightly, a behavior that is qualitatively similar for plasma and mixtures of plasma and colloidal plasma expanders, and which is probably due to the added force needed to deform the protein layer at the air-liquid interface [20].

The viscosity measurements derived from molecular rotors obtained with plasma samples at different concentrations and mixtures with colloids to obtain different viscosities are linearly related to the viscosity measured in a conventional mechanical viscometer operated in the range of 450 s^"1. Molecular rotors show the same difference in viscosity between different plasma samples as that shown by a mechanical viscometer used in the Newtonian range of plasma viscosity. Therefore, in principle for the samples used in this study the viscosity at zero shear stress is quantitatively and linearly related to the viscosity in the Newtonian regime for plasma. Consequently measurements with the rotors are representative of conventional measurements, when the rotors are calibrated against a sample that can be measured by both methods, mechanical and fluorescent.

The mathematical relationship between viscosity and quantum yield (thus, under constant excitation conditions, between viscosity and measured emission intensity) has been established experimentally and theoretically. The precision within the experiments of this study was similar to that of mechanical viscosity measurements under routine conditions. Scatter between similar experiments was of the order of 1 - 2%, and precision was never worse than 7.5%. These small deviations between fluorescence and mechanically based viscosity data suggest that the method is suitable for most applications. There are problems, however, that influence precision, mainly temperature, fluid turbidity and dye concentration. Viscosity and the intrinsic relaxation rate of the dye (which is one determinant of the quantum yield) are temperature-dependent functions, therefore temperature control is more important than for mechanical measurements. Ideally, the sample cuvette should be temperature-controlled, which is feasible due to the low amounts of volume used. Fluid turbidity, common in blood plasma samples, strongly affects fluorescence through absorption of excitation and emission light. Optimized cuvette geometries, such as a narrow rectangular cross- section, and offsetting the angle of excitation light, may solve this problem. Dye concentration linearly affects emission intensity, therefore the precision of the measurement depends on the precision of dye delivery. This was the main reason to use pre-stained plasma in this study as aliquots of 1 ml are easier to produce than aliquots of 10 Dl or less. A solution would be to measure absorption simultaneously to emission. Since absorption is not a function of viscosity, it should be dependent only on dye concentration and therefore provide a measurement standard. Another approach would be the simultaneous calibration with samples of known viscosity.

In summary, this study shows that fluorescent molecular rotors allow the measurement of biofluid viscosity through different means than through shearing of the fluid, and that the measurement results are comparable to mechanical measurements in both Example π

Lipophilic Fluorescent probes for the measurement of cell membrane viscosity

Membrane viscosity is a physical property of the cell, which describes the movement of molecules within the phospholipid bilayer. In general, membrane viscosity depends on the chemical composition of the bilayer and is shown to have optimum values for the proper function of various membrane-bound enzymes and receptors. For example, artificial phospholipid bilayers were found to have viscosity values between 70 and 120 cP, depending on the temperature [6], while in human epidermal cells, viscosity values ranging from 30 to 100 cP have been reported [26]. In liver cells, viscosities ranged from 108 to 217 cP [27].

Depending on the membrane viscosity (or membrane fluidity, which is its reciprocal), phospholipids and membrane-bound proteins show different vertical and lateral displacements, as well as lateral and rotational diffusion behavior [28]. Consequently, changes in membrane viscosity have been linked with alterations in physiological properties, such as carrier-mediated transport, activities of membrane- bound enzymes and receptor binding [29]. In addition, variations in membrane viscosity are linked to a variety of diseases, such as atherosclerosis[30], cell malignancy [31], hypercholesterolemia [32] and diabetes [33,34]. Further examples include an increase in membrane viscosity in several blood cells and specific brain cells in patients with

Alzheimer's disease [35,36]. Increased membrane viscosity is also associated with aging [37]. Erythrocytes showed permanently increased membrane viscosity, thus altering hepatic microcirculation [38]. Furthermore, endothelial cells are able to sense fluid shear stress through the cell membrane [39] and effect changes that serve to maintain a specific level of flow in blood vessels [40] .

The importance of membrane viscosity in cellular biology and physiology led to the development of fluorescent-based methods for quantitative measurements. To date, there are three dominant methods to measure membrane viscosity: (a) fluorescence recovery after photobleaching (FRAP), in which measurement of the diffusivity of a fluorophore in the membrane is related to membrane viscosity [41], (b) fluorescence anisotropy, in which the out-of-plane rotation of a fluorophore is related to membrane viscosity [42] and (c) use of environment-sensitive fluorescent probes. In the latter category are included compounds such as 2-dimethylamino-6-lauroylnaphthalene, whose emission wavelength shifts with the viscosity of the medium [43] and 9-(dicyanovinyl)- julolidine (DCVJ, 1) with a viscosity-dependent fluorescence quantum yield.

The intriguing fluorescent properties of DCVJ (Fig. 7) rest upon its ability to lose excited state energy either by radiation or by intramolecular rotation, the ratio of which depends on the free volume of the microenvironment [44]. This ability has defined a new class of molecules, commonly referred to as molecular rotors, that were found to react instantly (within 10 ns) to changes in the environment, thereby allowing real-time measurements with high temporal and spatial resolution. Moreover, in the case of DCVJ, its fluorescence quantum yield was shown to increase by a factor of 30 when the solvent was changed from 1-propanol to glycerol [6]. Due to the above properties, DCVJ has been employed to address a wide variety of problems, including polymerization processes [44], tubulin remodeling [45], and membrane viscosity measurements in liposomes and micelles [6]. More recently, DCVJ was employed to probe changes of membrane viscosity induced by fluid shear stress in cultured cells [3]. Nonetheless, the use of DCVJ in membrane viscosity measurements is often limited by its poor water solubility, which renders the preparation of DCVJ-based staining solutions problematic. For experiments on liposomes [6], this difficulty can be overcome by preparing a DCVJ-stained phospholipid base before forming the liposomes. In cultured cells, DCVJ can be bound to serum proteins [3], but their binding capacity is limited. Furthermore, protein-bound DCVJ migrates readily into the cytoplasm, thus adding a high level of constant background signal during biophysical measurements thereby reducing the sensitivity of the probe. In light of the side effects mentioned above, it was deemed desirable to obtain a fluorescent molecule that has the viscosity-dependent properties of DCVJ, but shows a higher affinity to the cell membrane. For this purpose, we synthesized various compounds (2a-2g), in which one of the nitrile functions of DCVJ was replaced by a carboxylic ester bearing aliphatic hydrocarbon chains of different length. Our results demonstrate clearly that it is possible to alter the chemical structure of DCVJ, as shown in Fig. 7, without affecting its viscosity-dependent fluorescence properties. Specifically, compounds 2b-2g have very similar viscosity-dependent fluorescence properties with DCVJ, but improved membrane localization. Moreover, the farnesyl-containing probe 2g (farnesyl-(2-carboxy-2-cyanovinyl)-julolidine, FCVJ) displayed a dramatically (more than 20-fold) higher sensitivity than DCVJ when probing membrane viscosity changes and was shown to be more stable than 1 against photobleaching.

Results

Synthesis of the molecular rotors

The synthesis of compounds 2a-2g is shown in Fig. 8. Commercially available

(Aldrich) julolidine 3 was formylated with phosphorus oxychloride and dimethylformamide to afford aldehyde 4 in a 95% yield. The desired -cyanoesters 7 were obtained via dicyclohexyl carbodiimide-induced esterification of cyanoacetic acid (5) with the corresponding alcohols in 80-95% yield. Condensation of esters 7 with aldehyde 4 in the presence of triethylamine produced the desired probes 2a-2g in very good yields (90-95%). These compounds were purified by chromatography on silica gel and crystallized twice prior to use.

Physical properties of the synthesized probes

Absorption and emission maxima of the examined compounds were all in the blue and green range, respectively, with emission maxima between 490 and 505 nm and two distinctly different excitation maxima. The first lies at about 405 nm, and the second, major, maximum in the range of 450-490 nm. Fig. 9 shows the excitation and emission spectra of DCVJ and FCVJ. The other compounds exhibited very similar spectra. The excitation and emission maxima are listed for all compounds in Table 3.

a Compounds dissolved in ethylene glycol. b In Eq. 3, x represents the slope of a line fitted into the data points log, over log,. c Slope of regression line into the data points of Fig. 10 under the assumption of a linear relationship. d Compound 2f is insoluble in DMSO, methanol and ethanol.

All compounds exhibited a viscosity-dependent fluorescence emission, while a viscosity-dependent change of the emission wavelength could not be detected. Using media with different viscosities (different mixtures of glycerol and ethylene glycol), it was possible to verify the double logarithmic relationship of quantum yield and viscosity (Fόrster-Hoffmann equation, Eq. 3) which was established for DCVJ [6,39]. Furthermore, all compounds showed similar values of x in Eq. 3. This indicates that the photophysical properties of the molecular rotor had not been affected by the attachment of the hydrocarbon chain.

Only 2a is water-soluble. Therefore, staining solutions were prepared using fetal calf serum (FCS) and bovine serum albumin (BSA) as carriers. All compounds, with the exception of 2f, could be solubilized in aqueous media by binding it to BSA. Using FCS as a carrier, compounds 2e and 2f were insoluble, and compound 2d as well as DCVJ showed a low solubility with precipitates forming within a few hours after the media preparation. The cells grown and stained on the flow chamber glass slides also exhibited the characteristic fluorescence emission spectrum of DCVJ. Under the microscope, the staining of the cells could be observed. In spite of its direct water solubility, 2a was the only probe that did not stain the cell membrane in a significant way. Measurement of membrane viscosity changes in cultured cells

Step-shaped flow profiles of a successively rising magnitude were applied to the cell layer. In all cases, a marked decrease of emission intensity could be observed when flow was turned on, followed by a recovery of the intensity when the flow was turned off. To quantitatively analyze the sensitivity to shear stress, the average intensity of each phase of shear stress was subtracted from the average intensity during low shear stress (0.07 Pa), resulting in data pairs of intensity change over shear stress increase Fig. 11. Linear regressions were computed from these data pairs, and tests for statistical significance (t-test) as well as the deviation from linearity (runs test) were performed. Compounds 2b, 2c, 2d and FCVJ showed a higher amplitude of the intensity when used to probe shear stress in cells than DCVJ (PO.0001 for compounds 2b, 2d and FCVJ, P=0.07 for 2c). The data points Fig. 10 showed no significant deviations from the straight line (runs test, values ranging from _P=0.2 for DCVJ to R=1.0 for 2b). These results, as well as the amplitude of the response, are summarized in Table 3.

FCVJ had the strongest response to shear stress. The amplitude of the fluorescence change in stained cells exposed to shear stress was more than 20 times higher than it could be observed with DCVJ. Fig. 6 shows the measured intensities of the fluorescence of DCVJ in comparison to FCVJ in stained cell membranes as a response to step changes of fluid shear stress of different magnitudes in an experimental set-up similar to the one described in [3]. The signal-to-noise ratio was significantly lower. While a statistically significant detection of shear stress with DCVJ was possible with values higher than 0.6 Pa [3], FCVJ allowed the detection of changes lower than 0.1 Pa (data not shown).

Microscope images indicated that rotors with longer chains exhibited slower migration rates and less intense cytoplasmic staining. In the confocal images, nucleole staining became visible for compounds with shorter chains, including DCVJ and compound 2c. Stained nucleoles were not visible for compounds 2d and FCVJ. This indicates lower intracellular migration rates of these probes.

The stability against photobleaching was also tested in compounds 2c, 2g and DCVJ. In a stained cell monolayer, all probes degraded under the influence of the excitation light in an exponential fashion. FCVJ (2g) showed a significantly higher stability against photobleaching than DCVJ. In a stained cell monolayer, all compounds degraded under the influence of the excitation light. With DCVJ, the half-life was 96 s. The half-life of compound 2c was 70 s, and that of FCVJ was 112 s (Fig. 12), both significantly different from DCVJ (_P<0.0005).

Discussion

Most often, measurement of membrane viscosity is carried out using FRAP or fluorescence anisotropy. The method of FRAP is based on directing a focused laser beam at the cell membrane, which destroys the fluorescent dye in a defined region. Diffusion within the membrane allows dye from the neighboring environment to migrate into the bleached spot. The half-life of the fluorescence recovery is inversely proportional to the diffusion coefficient of the dye [46], which in turn is proportional to the fluidity. There are two major disadvantages associated with FRAP. Firstly, the measurement process is slow, since times between 20 and 60 s are required in order to allow sufficient recovery for the reliable determination of the half-life. This fact strongly limits the temporal resolution of any FRAP measurement and does not permit measurement of dynamic processes. Secondly, FRAP is limited in its resolution, as reduction of the radius of the bleached spot increases the error associated with the measurement. In addition, the bleaching pulse introduces power densities up to 1 MW/cm² to the membrane [46], leading both to the generation of free radicals by photolysis as well as local heating with potential damage to proteins [46]. Especially the second effect may cause artefacts, because the resulting cross-linking of the proteins [46] leads to decreased membrane diffusivity.

Membrane viscosity measurements by fluorescence anisotropy utilize the fact that certain molecules can only be excited by light waves in one polarization plane. The emitted light, again, is polarized. During the excited lifetime, the molecule can rotate, thus shifting the plane of emitted light. The ratio of emitted light parallel versus perpendicular to the plane of excitation is a nonlinear function of the viscosity [42]. While the spatial resolution is essentially determined by the limits of the microscope, rapid photobleaching of the probes both confines the maximum sensitivity and the exposure time of the probe to the excitation light. Furthermore, the polarizing filters in the emission light path generally absorb a significant portion of the emitted light. Moreover, it was shown in a recent study that the deformation of cells caused by shear stress may produce artifacts that are similar to an increase in membrane fluidity [42], which makes anisotropy observations difficult to interpret.

More recently, environment-sensitive probes have been used in membrane viscosity studies. In this category of fluorescent probes belong a group of compounds that have been shown to undergo intramolecular charge transfer when excited by photon absoφtion [5]. One member of the group, DCVJ (1), has been extensively characterized. For the intramolecular charge transfer, the julolidine group functions as the electron donor, and the dicyano-group as electron acceptor [5]. The relaxation of the excited state can either occur through photon emission or non-radiative processes [26].

The assessment of membrane viscosity using molecular rotors overcomes the limits posed by FRAP or fluorescence anisotropy measurements. The spatial resolution of the method is potentially as high as it is for fluorescence anisotropy, because it is primarily limited by the optics of the microscope. The molecular rotor also responds to changes of viscosity almost instantaneously (within the nanosecond range) [5]. This allows a very high temporal resolution of the measurements, so that dynamic observations become possible.

The major disadvantage associated with DCVJ when used to probe cell membranes is the affinity for tubulin and proteins in the cytoplasm. The experiments presented in [39] are based on cultured cells, where a significant cytoplasmic staining exists, and a continuous exchange of the dye between the cytoplasm and the membrane takes place. Both the cytoskeleton with its slow response to shear stress [48,49], and the cytoplasm, which slightly increases viscosity with shear stress [51,52], do not contribute to the signal measured, but provide a background fluorescence which reduces the measurement sensitivity. Furthermore, the relative intensity of the background is unknown and not easily determined, making quantitative measurements of membrane viscosity changes difficult.

To address the above issues encountered with DCVJ (and related rotors), we have synthesized several fluorescent probes, structurally related to DCVJ, with hydrophobic chains of different length attached to it. All probes synthesized showed a similar photophysical behavior to DCVJ with excitation and emission wavelengths in the same range. Also, the dependency of the quantum yield on the viscosity of the environment (slope x in Eq. 3) was similar. This shows that the modifications performed do not inhibit the function of the rotor, i.e. that the molecule can lose its excited state either through intramolecular rotation or through photon emission, depending on the viscosity of the environment.

Tests were performed to determine whether each compound is suitable for membrane viscosity measurements. In the first step, the fluorescent properties of each probe were determined. All compounds were similar in their physical properties. In the second step, solubility in aqueous media was determined, a prerequisite for the staining of attached cells. Both BSA and FCS may be used as a carrier. BSA binding was possible for all compounds except 2f. The probes with shorter chains, 2a through 2d, as well as FCVJ, were also soluble in media prepared with FCS. Due to its carboxylic acid, 2a is directly water-soluble. All compounds capable of binding to serum albumins were then tested for their cell staining capability. All compounds except 2a and 2f were able to stain cells. Finally, stained cells were submitted to shear stress, and the fluorescence intensity change as a function of the shear stress applied was recorded and compared between the compounds. Out of five probes tested (2b through 2e, and 2g), 2c showed a similar response to DCVJ when used to test cells exposed to shear stress, while 2b, 2d and FCVJ showed a higher amplitude of the response.

FCVJ (2g), a rotor in which a farnesyl group was attached to the julolidine part via an ester bond, showed a superior response when probing cell membrane viscosity changes than all other probes, including DCVJ. Farnesyl is a hydrocarbon chain that improves membrane localization [31,32]. Substituting R for farnesyl (Fig 7) resulted in a dye that shows an improved affinity for the cell membrane over DCVJ. This leads to a slower migration into the cytoplasm, a lower background signal and a higher signal-to- noise ratio in the measured intensity time-courses. The confocal images show that FCVJ can be found in the membrane at a higher concentration. With the given -resolution of the confocal microscope, it is difficult to distinguish between the cell membrane on top of the nucleus and the cytoplasm layer between the nucleus and the membrane, but it is reasonable to assume that with an active exchange of the dye between the cytoplasm and the membrane, both layers would be stained.

According to the confocal images, DCVJ and the short-chained compound 2b showed a high affinity for the cytoskeleton, and both dyes can be found in the nucleoles. With compound 2d, there is less tubulin staining and no nucleole staining. This suggests that longer chains allow a stronger binding of the dye to the membrane, and also reduces the migration into the inner compartments of the cell.

Experiments have shown that some of the compounds, particularly FCVJ, show a higher resistance to photobleaching, than DCVJ. This makes FCVJ a more adequate probe for long-term photophysical measurements. It should be noted that the half-life of DCVJ is about 50-100 times higher than.the half-life of l,6-diphenyl-2,3,5-hexatriene, a probe commonly used for viscosity measurements through fluorescence anisotropy under the same conditions.

The presented results show a significant improvement over DCVJ by combining a farnesyl chain with a molecular rotor with respect to membrane localization, stability against photobleaching, and sensitivity of membrane viscosity measurements. Membrane viscosity changes could be detected for fluid shear stress as low as 0.1 Pa, while DCVJ did not lead to statistically significant changes for shear stress values of 0.6 Pa and below [39]. However, cytoplasmic staining still persists even for FCVJ, which likely is a consequence of the cell metabolism. Further experiments are required to determine if the metabolic activity of the cell can be inhibited, and to what extent such an inhibition would affect the cell membrane viscosity.

Significance

Molecular rotors are new and promising tools for cell membrane viscosity measurements. One of the applications is the measurement of cell membrane viscosity in cultured cells or in vivo: membrane viscosity changes are crucial in the understanding of many signaling processes and diseases. Commercially available rotors such as DCVJ, however, are associated with disadvantages such as binding to the cell cytoskeleton and migration into inner compartments of the cell. While DCVJ allows the measurement of cell membrane viscosity changes, a higher sensitivity can be expected from DCVJ- derived molecules that show a higher affinity to the membrane.

We synthesized a group of related compounds that are composed of a DCVJ-like rotor and a hydrocarbon chain. These molecules retain their viscosity sensitivity, but show a higher affinity for the cell membrane with increasing chain length. This is generally associated with a higher amplitude of the fluorescence intensity change that can be observed when changes of membrane viscosity occur. Notably the attachment of a farnesyl chain leads to a more than 20-fold increase in amplitude over DCVJ. This new molecular rotor, referred to as FCVJ, improved the signal-to-noise ratio, thereby allowing the measurement of far smaller viscosity changes than DCVJ. Materials and methods

Synthesis of aldehyde

To a solution of julolidine (3) (0.5 g, 2.88 mmol) and NN-dimethyl formamide (0.27 ml, 3.46 mmol) in dichloromethane (5 ml), was added dropwise phosphorous oxychloride (0.29 ml, 3.17 mmol) and the mixture was stirred at 25°C for 8 h. The reaction was treated with an aqueous solution of sodium hydroxide (2 M) and the mixture was stirred at 0°C for 4 h. The organic layer was extracted with ethyl ether (3x10 ml), collected, dried (MgSO ) and concentrated under reduced pressure. The residue was crystallized twice from dichloromethane/hexane (1:10) to afford aldehyde 4 (0.55 g, 2.74 mmol, 95% yield). 4: Colorless solid; R_f= .37 (silica, 30% ethyl ether in hexanes); 1H ΝMR (400 MHz, CDC1₃) 9.58 (s, 1 H), 7.28 (s, 2H), 3.27-3.29 (m, 4H), 2.74-2.76 (m, 4H), 1.94-1.96 (m, 4H); ¹³C MR (100 MHz, CDC1₃) 89.9, 147.7, 129.3, 123.8, 120.2, 50.0, 27.7, 21.3; HRMS, calcd for C₁₃H₁₅NO (M+H⁺) 202.2754, found 202.2761.

Synthesis of farnesyl ester 7g

To a solution of cyanoacetic acid (5) (0.43 g, 5 mmol) and tr rø,tr /rø-farnesol

(6g) (1.11 g, 5 mmol) in dichloromethane (5 ml) was added DCC (1.03 g, 5 mmol) and the mixture was stirred at 25°C for 10 h. The reaction was then diluted with dichloromethane (20 ml) and the formed DCU was filtered under gravity. The filtrate was dried (MgSO ) and concentrated and the residue was purified by column chromatography (silica, 5-10% ethyl ether in hexane) to afford farnesyl ester 7g (1.39 g, 4.8 mmol, 96% yield). 7g: Colorless oil; R_f=0.63 (silica, 30% ethyl ether in hexanes); 1R (film) ^ 2968.6, 2931.2, 2857.7, 1757.9, 1669.5, 1448.7, 1383.2, 1349.0, 1327.3, 1274.7, 1181.9, 985.5, 932.7; 1HNMR (400 MHz, CDC1₃) 5.33 (t, 1H), 5.05-5.08 (m, 2H), 4.68-4.71 (d, 2H), 3.44 (s, 2H), 1.9-2.18 (m, 8H), 1.71 (s, 3H), 1.66 (s, 3H), 1.58 (s, 6H); ¹³C NMR (100 MHz, CDC1₃) 162.7, 143.9, 135.3, 131.0, 124.0, 123.1, 116.7, 112.9, 63.5, 39.6, 39.4, 26.7, 26.1, 25.7, 24.7, 17.7, 16.5, 16.0; HRMS, calcd for C₁₈H₂₇NO₂ (M+H*) 290.2120, found 290.2141. Synthesis of FCVJ 2g

To a solution of farnesyl cyanoacetic ester (7g) (1.08 g, 3.73 mmol) and aldehyde 4 (0.5 g, 2.48 mmol) in tetrahydrofuran (15 ml) was added triethylamine (0.7 ml, 5 mmol) and the mixture was stirred at 50°C for 10 h. After consumption of the starting material, the reaction mixture was concentrated and the residue purified by column chromatography (silica, 10% ethyl ether in hexane) to afford FCVJ (2g) (1.1 g, 2.36 mmol, 95% yield). Compound 2g was crystallized twice from ether/hexane. FCVJ: Yellow orange solid; Rp0.34 (silica, 30% ethyl ether in hexanes); IR (film) _ma_x 2931.4, 2210.6, 1708.0, 1613.4, 1571.6, 1522.3, 1442.9, 1316.5, 1227.1, 1163.1, 1095.8; 1H NMR (400 MHz, CDC1₃) 7.93 (s, IH), 7.50 (s, 2H), 5.42-5.43 (t, IH), 5.08-5.10 (m, 2H), 4.76-4.78 (d, 2H), 3.30-3.33 (m, 4H), 2.72- 2.76 (m, 4H), 1.90-2.20 (m, 8H), 1.74 (s, 3H), 1.67 (s, 3H), 1.60 (s, 6H); ¹³C NMR (100 MHz, CDC1₃) 164.6, 154.1, 147.3, 142.1, 135.2, 131.4, 131.1, 124.2, 123.5, 120.6, 118.3, 118.1, 117.9, 91.7, 62.6, 50.2, 39.7, 39.6, 27.6, 26.8, 26.3, 25.8, 21.2, 17.8, 16.7, 16.1; HRMS, calcd for C^HωNzOz (M+H⁺) 473.3168, found 473.3189.

Probe 2b: Red solid; R =0.30 (silica, 40% ethyl ether in hexanes); IR (film) _max 2947.9, 2209.6, 1711.8, 1613.1, 1569.6, 1521.0, 1435.3, 1317.2, 1295.8, 1231.2, 1163.0, 1097.5; 1H NMR (400 MHz, CDCI₃) 7.94 (s, IH), 7.50 (s, 2H), 3.86 (s, 3H), 3.31-3.34 (m, 4H), 2.72-2.74 (m, 4H), 1.94-1.96 (m, 4H); ¹³C NMR (100 MHz, CDCI3) 165.1, 154.3, 147.5, 131.6, 120.7, 118.2, 118.1, 90.9, 52.7, 50.2, 27.6, 21.2; HRMS, calcd for C₁₇H₁₈N₂O₂ (M+H⁴) 283.1446, found 283.1461.

Probe 2c: Red solid; R =0.36 (silica, 40% ethyl ether in hexanes); JR. (film) _ma_X 2937.1, 2209.7, 1708.8, 1613.2, 1570.4, 1522.3, 1443.2, 1317.7, 1294.6, 1231.5, 1164.2, 1097.5, 1021.3; 1H NMR (400 MHz, CDC1₃) y.91 (s, 1 H), 7.48 (s, 2H), 4.28-4.31 (m, 2H), 3.29-3.32 (m, 4H), 2.71-2.74 (m, 4H), 1.93-1.95 (m, 4H), 1.33-1.37 (m, 3H); ¹³C NMR (100 MHz, CDC1₃) 164.5, 154.0, 147.4, 131.4, 120.6, 118.2, 117.9, 91.4, 61.6, 50.1, 27.6, 21.1, 14.4; HRMS, calcd for C₁₈H₂₀N₂O₂ (M+H⁴) 297.1603, found 297.1611. Probe 2d: Orange red solid; Rf=0.39 (silica, 40% ethyl ether in hexanes); IR (film) _max 2956.0, 2209.3, 1707.9, 1613.0, 1569.9, 1521.3, 1442.9, 1316.9, 1228.4, 1163.5, 1097.8; 1H NMR (400 MHz, CDC1₃) 7.93 (s, 1 H), 7.50 (s, 2H), 4.23-4.27 (m, 2H), 3.30- 3.33 (m, 4H), 2.72-2.76 (m, 4H), 1.94-1.97 (m, 4H), 1.43-1.72 (m, 4H), 0.94-0.98 (m, 3H); ¹³C NMR (100 MHz, CDCl₃) 184.7, 164.7, 154.1, 131.5, 120.6, 118.3, 65.5, 50.2, 30.8, 27.7, 21.2, 19.2, 13.9; HRMS, calcd for C₂₀H₂₄N₂O₂ (M+H^1") 325.4246, found 325.4250.

Probe 2e: Orange yellow solid; R_f=0.40 (silica, 40% ethyl ether/hexanes); IR (11^) ^_x 2925.9, 2853.1, 2210.5, 1709.5, 1613.0, 1571.6, 1522.6, 1443.3, 1317.2, 1229.4, 1163.0, 1099.9; 1H NMR (400 MHz, CDCI3) 7.93 (s, 1 H), 7.50 (s, 2H), 4.22-4.24 (m, 2H), 3.30-3.33 (m, 4H), 2.72-2.76 (m, 4H), 1.94-1.97 (m, 4H), 1.68-1.75 (m, 2H), 1.25-1.45 (m, 16H), 0.86-0.89 (m, 3H); ¹³C NMR (100 MHz, CDC1₃) 164.6, 154.1, 147.4, 131.5, 120.6, 118.3, 117.9, 91.7, 65.8, 50.2, 31.9, 29.6, 29.4, 29.3, 28.7, 27.6, 25.9, 22.8, 21.2, 14.3 HRMS, calcd for C₂6H₃₆N₂O₂ (M+H⁺) 409.2855, found 409.2861.

Probe 2f : Orange yellow solid; R_f=0.45 (silica, 40% ethyl ether in hexanes); IR

(film) _max 2917.4, 2849.9, 1715.5, 1570.7, 1521.1, 1444.3, 1316.3, 1234.4, 1165.3, 1101.5; 1H NMR (400 MHz, CDC1₃) 7.93 (s, 1 H), 7.50 (s, 2H), 4.22-4.24 (m, 2H), 3.30- 3.33 (m, 4H), 2.72-2.76 (m, 4H), 1.94-1.97 (m, 4H), 1.43-1.72 (m, 6H), 1.24-1.26 (m, 30H), 0.85-0.89 (m, 3H); ¹³C NMR (100 MHz, CDC1₃) 164.7, 154.1, 147.4, 131.5, 131.4, 120.6, 118.3, 117.9, 65.8, 50.2, 32.0, 29.8, 29.7, 29.6, 29.5, 29.4, 28.8, 27.7, 25.9, 22.8, 21.2, 14.3 HRMS, calcd for C₃₆H₅₆N₂O₂ (M+H*) 549.4420, found 549.4439.

For the determination of the relationship between quantum yield and viscosity of the medium, mixtures of ethylene glycol and glycerol were prepared. Different viscosities were achieved through different vol/vol mixture ratios of ethylene glycol.glycerol as follows: 7:3 (49 cP), 5:5 (115 cP), 4:6 (163 cP), 3:7 (245 cP), 2:8 (391 cP) following an experiment described in [4]. 3.5 ml of each of these mixtures was filled into a spectroscopic cuvette under addition of 10 μl of probe stock solution (10 mM probe in methanol). The emission intensity of each of the five samples was acquired using an RF- 1501 fluorophotospectrometer (Shimadzu, Kyoto, Japan) set at

nm and emission -OO n .

For the experiments on cultured cells, procedures similar to the ones described elsewhere [39] were used. Briefly, human umbilical cord endothelial cells were harvested using fresh umbilical cords, where the veins were treated with a 0.02% solution of collagenase in phosphate-buffered saline solution for 30 min at 25°C [1]. The cells were suspended in culture medium (Medium M199 with 2 mM L-glutamine, 50 U/ml penicillin, 50 μg ml streptomycin and 20% FCS) and seeded on tissue culture dishes. These dishes were kept in the incubator at 37°C in CO₂ enriched air (5% CO₂) until confluency was reached. Once confluency was achieved, the cells were lifted with cell dissociation solution and resuspended in Dulbecco's modified Eagle's medium with 20% FCS at a concentration of 2.5^χl0⁶ cells/ml. Aliquots of 350 μl of the suspension were used to seed the cells on glass slides (dimensions 10x40 mm) that were previously treated for 2 h with 0.5 M NaOH. The cells attached and foπned a confluent layer within 24 h.

A staining solution was prepared from 50 μl stock solution of 20 mM of the examined probe in dimethyl sulfoxide (DMSO), which was dispersed in 2 ml FCS under vigorous stirring. 10 ml medium Ml 99 was added. The slides with the cells were covered with the staining solution. Incubation took place over 10 min in the dark at 37°C. After the incubation period, the cells were rinsed with HBSS, and the fluorescence checked under an epifluorescent microscope (Diaphot TMD, Nikon, Garden City, NY, USA) using the G2B filter set. Flow chambers were assembled using a standard methacrylate spectroscopic cuvette (Fisher Scientific, Pittsburgh, PA, USA) and a parallel-plate black Delrin flow chamber with a channel width of 6 mm and a depth of 500 μm. The flow channel was formed between the flow chamber wall and the glass slide with the confluent cell monolayer. In the parallel-plate flow chamber, the following relationship between flow Q and wall shear stress holds: where μ is the viscosity of the medium, h is the channel depth and w is the channel width. On the Shimadzu RF- 1501 fluorophotospectrometer with time-course module, excitation wavelength was set to 455 nm, and emission wavelength to 505 nm. An additional 475 nm LP emission filter (Chroma, Brattleboro, VT, USA) blocked scattered blue light. A microcontroller-driven syringe pump (Pump 22, Harvard

Apparatus, Holliston, MS, USA) attached to a PC provided a controlled flow of the flow medium, HBSS. Throughout the experiments, a minimum flow of 1 ml/ in was maintained in order to provide the cells with fresh medium and avoid hypoxia. This minimal flow caused a shear stress of 0.07 Pa. Shear stress rates of 0.67, 1.33, 2.0, 2.67 and 3.33 Pa (1 Pa=10 dyn/cm²) were applied successively for 30 s with a 30 s period of 0.07 Pa in between. The drop in fluorescence intensity was determined for each step change of shear stress and plotted over the shear stress value. To determine the amplitude of the response, a straight line fit was perfoπned (least-squares method) with the slope of this line representing the amplitude and the deviation from linearity determined with the runs test.

Summarized, the following tests were performed for each compound (not all of the tests have been performed on all of the compounds):

1. Basic properties: acquisition of excitation and emission spectra of the probe dissolved in DMSO at a concentration of 60 μM. 2. Solubility in water, FCS and BSA, as well as cell culture media with 20%

FCS and 3% BSA, respectively.

3. Dependency of the quantum yield on the viscosity of the solvent.

4. Capability of the dye to stain cells.

5. Response of the quantum yield of the compound in cells to fluid shear stress.

For DCVJ and compounds 2b, 2c, 2g, confocal images were acquired to more accurately determine the localization of the dye. Z-Sections of 200 μm thickness were obtained on a BioRad MRC-1024 confocal microscope using the FITC program. For DCVJ as well as compounds 2c and 2g, the resistance against photobleaching was tested by acquiring one image of the stained cells every 15 s using the epifluorescent microscope. In each of three independent experiments, the average intensity of the confluent cell area in the field of view was determined in each frame using Scion Image (Scion, Frederick, NY, USA). A one-phase exponential decay curve was fitted into the data points and the half-life of the compound was calculated.

Example IV Replacing both CN-groups with ester groups

We have experimentally shown that the ester group also functions as electron acceptor group in a similar fashion as the CN group. Replacing both CN-groups with ester groups retains the viscosity dependent quantum yield, while at the same time it allows to attach two recognition elements (Ri and R₂ in Figure 2 (right)). We also confirmed the synthesis and the physical properties of a rotor moiety attached to a phospholipid.

The following properties have been experimentally tested using the diethylester and difarnesylester forms of the compound as well as the rotor attached to the head of phosphatidylethanolamine. We have confiπned the following properties: -Dependency of the quantum yield on the viscosity of the medium (diethylester only)

-Ability to stain the cell membrane

-Response (decreased fluorescence intensity) of stained cell membranes when exposed to fluid shear stress (diethylester only) Namely the probe attached to phosphatidylethanolamine showed an exclusive membrane localization with some staining of the internal membranes. However, there was no response to shear stress, which can be attributed to the fact that the rotor moiety was attached to the head of the phospholipid and thus was not integrated into the membrane. This now covers the more general form of molecular rotors as given in Fig. 2 (right). Table 4 is now completed by vinyljulolidine-diethylester (8) and vinyljulodidine-difarnesylester (9), and CVJ attached to phosphatidylethanolamine, the compounds which were synthesized and examined for this purpose. As indicated in the previous sections, we found that replacement of one nitrile group of DCVJ with a farnesyl chain increases the compatibility with the cell membrane, which reduces cytoplasmic staining and increases the sensitivity more than ten-fold. This finding can be further explored to yield new molecular rotors with enhanced membrane localization properties. /. Molecular rotors bound to stearic acid for the determination of fluidity gradients

Construction of this group probes is based on literature indicating that attachment of fluorescent anisotropy probes, such as DPH, to hydroxyl stearate side chains has yielded a family of fluorophores for gradient viscosity changes. The reported studies have also demonstrated that the stearic acid positions itself in the outer leaflet of the membrane with the carboxylate functionality being at the interface of membrane and extracellular environment.

It is important to state that neither FRAP nor fluorescence anisotropy can be used to deliver these results. By using FRAP the photobleaching pulse affects the entire thickness of the membrane. During recovery it is not possible to distinguish at what depth the migration takes place. On the other hand, the experimental undertaking of fluorescence anisotropy requires the use of a flow cytometer. In the cytometer the cells have to pass through a capillary, which means that they cannot be attached and therefore they cannot be exposed to defined levels of fluid shear stress. Furthermore, fluid shear stress changes the chape of the cell membrane, which has been shown to strongly affect the anisotropy results.

2. Molecular rotors for the assessment of membrane fluidity in different microdomains of the bilayer

It is well established that the outer (exoplasmic) and inner (cytoplasmic) leaflets of cell membranes differ in lipid composition. In most cells, it has been shown that outer leaflet is mainly composed of phosphatidyl choline (PC) and sphingomyelin (SP), while the inner leaflet is mainly composed of phosphatidyl ethanolamine (PE) and phosphaidyl serine (PS). For example, between 80-100% of phosphatidyl serine (PS) can be found in the inner leaflet. Due to its structure, PS provides a high concentration of negative charges in the cytoplasmic part of the membrane. The lipids are also organized in the lateral dimension due to preferential packing of sphingolipids and cholesterol into microdomains. These microdomains, which include detergent-resistant areas, lipid rafts, and caveolae, have been shown to play an essential role to a variety of biological and physiological functions of the cell. For example, caveolae are membrane micropatches enriched in cholesterol, glycosphingolipids, signaling molecules and caveolin (a 22-kDa protein, involved in the structure/formation of caveolae). It has been found that caveolae are a sort of transport vesicles involved in protein (and small molecules) trafficking from the Golgi to the cell membrane and back and are participating in transmembrane signaling.

By attaching a DCVJ-based rotor to a phospholipid backbone we will overcome the adverse influence of non-membrane background staining (as discussed in Section C.2.). Once a neglectable background staining (e.g.cytosolic) is achieved, the determination of viscosity changes will reach the point of highest accuracy. A further advantage of a strict localization will be the ability to perform long-time measurements without significant migration of the dye into the cytoplasm. It is also possible to examine the effect of shear stress in membrane microdomains, such as caveolae and attempt to coπelate mechanical stress to chemical signaling initiated at these microdomains. An additional advantage of these probes is that they are self-micellating and therefore they do not require an additional delivery system in order to enter the membrane. Example V

Molecular Rotors - Special Calibration Techniques

Molecular rotors are unique fluorescent molecules that allow the real-time monitoring of the viscosity of their solvent. This measurement involves the determination of the quantum yield. Since the direct measurement of the quantum yield is a relatively complicated process, it would be desirable to use methods where the measurement of emission intensity is sufficient. Emission intensity depends on the molecule's quantum yield as well as the concentration of the fluorophore, the incident light intensity, and the absorption characteristics of the fluid. In order to relate viscosity to intensity instead of quantum yield, it is therefore necessary to eliminate the above influences. While the incident light can be kept constant by relatively simple means, dye concentration and fluid absoφtion require special techniques, which are the subject of this disclosure.

(l)The use of a reference moiety

Molecular rotors show a viscosity-dependent quantum yield according to the Forster- Hoffmann-Equation

logφ = C + x logη

Here, η is the viscosity of the solvent, φ is the quantum yield, x is a dye dependent constant and C is a constant that reflects the rate of intramolecular rotation. By assuming a proportional relationship between the measured emission light and quantum yield, the equation can be rewritten as

I = C' .η^* with I as the emission intensity. The new constant, C, reflects not only the constant C, but also the incident light intensity and the dye concentration, which at this point are unknown. Thus, emission intensity and quantum yield are linked through

with f denoting a function of the incident light I_ex and dye concentration Cd_ye.

To compare two fluids with different concentrations, the molecular rotor can be coupled to a fluorescent reference moiety, which is not viscosity-sensitive. Since the quantum yield of the reference moiety is independent of the environmental properties, and thus can be determined for any reference molecule, concentration can be directly determined from the ratio of incident to emitted light intensities.

In a practical environment, the measurement process can further be simplified by computing the ratio of the emitted light at the wavelength of the molecular rotor to the emitted light at th wavelength of the reference moiety, requiring two successive measurements at two different wavelengths.

Candidates for the reference moiety are any fluorophores with non-overlapping spectra. Particularly, the emission wavelengths must be different, and each pair of excitation and emission wavelength mist be different to prevent fluorescent resonance energy transfer (FRET).

For measurements of the cell membrane, an ideal fluorescent molecule is Dil-C₁₈ (1,1 - dioctadecyl-3,3,3',3'-tetramethylindodicarbocyanine perchlorate) with an excitation maximum at 540 nm and an emission maximum at 560 nm (DCVJ absorbs at 460 mn and emits at 500 nm).

(2) Use of independent absoφtion measurement

Any fluorescent dye absorbs light at its absoφtion wavelength, part of which is emitted in foπn of fluorescence. This absoφtion is concentration-dependent, but - in the case of molecular rotors - not viscosity dependent, as can be shown in an experiment.

Thus, a simultaneous absoφtion measurement at the excitation wavelength can be used to determine the dye concentration. Similar to the reference moiety, a ratio of the emission intensity to the absoφtion coefficient can be computed, which is independent of dye concentration.

In a practical device, the measurement principles of a spectrometer and a fluorometer will be combined. The design of such a device is shown in Figure 13. As an addition, a narrow cuvette minimizes the influence of fluid turbidity.

In accordance with these and other possible variations and adaptations of the present invention, the scope of the invention should be determined in accordance with the following claims, only, and not solely in accordance with that embodiment within which the invention has been taught.

Claims

CLAIMSWhat is claimed is:

1. A compound molecular rotor probe molecule of the general structure

having a dicyanov yl-aromatic derivative base structure; and at least one side chain wherein Ri is a side chain moiety having a defined specificity and R₂ is a side chain having a defined specificity

2. The compound according to claim 1, wherein Rι is a lipophilic moiety having an affinity for cell membrane.

3. The compound according to claim 1, wherein R is a lipophilic moiety having an identical affinity for cell membrane.

4. The compound according to claim 1, wherein R₂ is a moiety having a different affinity.

5. The compound according to claim 1 , wherein the dicyanovinyl-aromatic molecule is 9-(dicyanovinyl)-julolidine.

6. A method for producing a compound molecular rotor probe of the general formula

comprising the steps of: formylating julolidine to form julolidene aldehyde; esterifying cyanoacetic acid with the corresponding alcohol; and condensing the esters with julolidine aldehyde.

7. A method for measuring blood plasma viscosity comprising adding an effective amount of molecular rotor molecules to the plasma solution; and determining the viscosity of the plasma sample by the intensity of emitted light energy wherein the intensity of the emitted light coπelates with blood plasma viscosity.

8. A method for determining the viscosity and fluidity of cell membrane structures, comprising contacting cells with an effective amount of molecular rotor moleculeshaving a specific affinity for membrane structures of interest; and determimng the viscosity and fluidity of the membrane structures by measuring the intensity and staining patterns of emitted light energy,

wherein the intensity of the emitted light coπelates with membrane viscosity and fluidity. Table 1: Mixture ratios of plasma and pentastarch solution used for the high- viscosity range (Experiment 1) and for the lower viscosity increment (Experiment 2)

Table 2: Changes of the relative mtensity of the tested probes with increasing viscosity of the plasma : pentastarch mixture. All probes show a distinct increase of the emission peak with increased viscosity. At the highest viscosity examined (plasma : pentastarch solution 20:80), intensity is twofold (DCVJ, CCVJ) to 3.5-fold (CCVJ-esters) higher than in pure plasma. The slope is computed from the logarithms of intensity and viscosity and represents the constant x in Equation 1.