The Issue of Tissue: Approaches and Challenges to the Light Control of Drug Activity

Introduction

Light is a powerful chemical reagent:

Where light goes, when it is applied, and the amount of light delivered are all factors that are relatively easy to regulate. Because of this, there has been a significant interest in controlling biological phenomena using light, since so much of biology is linked to the amount of key molecules, as well as the timing and location of their appearance. The areas of study influenced by these factors are numerous and include gene expression, developmental biology and neuroscience. The molecular tools developed for these studies have been wide ranging, and include nucleic acids, proteins and small molecules.^[1] Within this larger category of light controlled biology has been the development of light controlled drug activity.^{[1k, 2]} The motivation behind light controlled drug activity has been similar to that of light controlled biology in general: There are multiple drug classes for which control of the timing, amount and location of activity is critical for optimal treatment of disease.

This review will focus on the literature that deals with light control of actual drugs, that can potentially be used to treat known diseases. This is as opposed to the larger (and important) literature that deals with light control of biological probes, used typically to understand biological phenomenon. We have attempted to identify the broad themes in the literature, and illustrated them with representative examples. In addition we have highlighted the strengths and weaknesses of the approaches. Needless to say, many of the issues that are involved in the light control of drug release also apply to the larger photo-bio community.

The Dimensions that Light Can Control

There are three principal aspects of a process that light can modulate: a) Its spacing, b) its timing and c) the degree to which it happens. This is because once a process is linked to light irradiation, you can control where light goes, when light irradiation is initiated and how much light is applied. Each of these dimensions of control confers different potential advantages.

The control of spacing has the potential to reduce a drug’s toxicity, by limiting its activity to a specific site, for example a tumor or a site of infection. Many chemotherapeutics have dose-limiting toxicities associated with healthy tissue, for example non-cancerous but rapidly dividing cells. Limiting a drug’s activity only to target sites where they are activated by light could allow them to be used for longer periods of time or at higher doses.

The control of the timing and amount of release can be useful for drugs where the required amount varies continuously throughout the day. The majority of drugs likely do not fall into this category, as all that is required for their effective application is that the systemic concentration remain above a critical therapeutic threshold. Normally this can be achieved through more conventional means, such as extended release formulations.^[3] However, there is a subset of applications where timing and amount is critical, such as with hormones and other signaling molecules where the requirements vary continuously throughout the day. Light in this context has the potential to confer a much needed level of timing and degree control.

The Ways In Which Light has Been Used To Control Drug Activity

We can classify light controlled drug activity into four broad categories, which are linked to the mechanism of light control. These are 1) Photocleavage control 2) Photoconformational control 3) Photothermal control and 4) Photodegradation control (Figure 1). With all four of these mechanisms, there are two principle characteristics of the photoactivated group that determine the ultimate performance: The wavelength required of the photoactivation, and the quantum yield for this process. The wavelength strongly influences the depth that light can penetrate through tissue, with UV and short visible wavelengths penetrating a millimeter or less, and infrared having the potential to penetrate on the order of centimeters. This determines the ultimate number of photons that can reach the desired target and photoactivated group, based on the amount of light the body is irradiated with. In addition to wavelength, a critical descriptor of the photoactivated group is the quantum yield, a unitless factor that determines the number of photostimulated events (e.g. photolysis) per photon absorbed by the group, and varies between 0 and 1. The ultimate nature of the photoactivation mechanism also influences the specific photoactivated group that can be used.

Figure 1.

Four main approaches to the light control of drug action.

Photocleavage:

In photocleavage, irradiation is accompanied by the breaking of a covalent bond between the drug and a moiety that is responsible for modulating the activity of the drug. Although not a drug, the earliest example of photocontrol of the activity of a biomolecule is the caging of ATP by Hoffman and coworkers using an ortho-nitro benzyl group that was capable of blocking the activity of ATP until 365nm irradiation released native ATP.^[4] In the subsequent years, many actual drugs have been modified with photocleavable (PC) groups, to demonstrate the possibility of modulating their activity with light.

The two photocleavable groups that have been most investigated for light activated drug control are the ortho-nitro benzylic derivatives^[5] and coumarin derivatives^[6], although a wide range of other PC groups have also been explored. The ortho-nitro benzylic derivatives have absorption maxima in the far UV (~365nm). This is not an ionizing wavelength of UV light, but being shorter than visible light still suffers from lower tissue penetration. For example, Nishimoto and coworkers used an ortho-nitro benzylic PC group to act as a linker between the drug, 5 fluoro uracil, and a cyclic peptide targeting moiety.^[7] Lin and coworkers linked a porphyrin, itself capable of potential photodynamic therapy to 5 fluoro-uracil, via a light cleaved ortho-nitro benzylic group, showing light dependent cell toxicity.^[8] The coumarin system has also been extensively examined, in studies of light activated drug and biological probe release. This has resulted in the synthesis of derivatives with increasingly higher wavelengths of deprotection that now stretch to 500nm and above.^{[6a, 6c, 6e]} For example Feringa and coworkers used two different coumarins with different deprotection wavelengths to inhibit two separate strains of bacteria.^[9] This was accomplished through the caging of two different classes of antibiotics, a penicillin and a fluoroquinolone.

Outside of these two main PC groups, there have been multiple other photocleavable groups explored. For example McCoy and coworkers used the dimethoxy benzoin PC group to mask a range of carboxyl containing NSAID drugs including ketoprofen and ibuprofen.^[10] In addition, multiple examples of BODIPY modified drugs have extended photocleavage well into the visible range. For example Chakrapani and coworkers used an aryloxy-BODIPY derivative to modify a fluoroquinolone antibiotic, and then stimulate its release using 470nm visible light.^[11] In very promising work, Schnermann and coworkers have explored the cyanine chromophore for photocaging, and showed it capable of being activated at an unusually high wavelength (690nm) through a two step process of photooxidation and hydrolysis. This represents the highest wavelength single photoclevage process.^[12]

In much of this work, the focus is chemical, demonstrating that the PC group can be installed, and that light will release the native drug. In addition, model systems have typically been used to demonstrate activity, such as enzyme activity or a specific bioactivity monitored in cell culture studies. What is often missing from the work is a description of how the resultant molecules would be used in an actual in-vivo setting, for example to treat human disease. One potential application is to block activity until irradiation releases active drug in the desired tissue, thus avoiding the toxicity associated with systemic exposure. One challenge is that upon uncaging of the active drug in the target tissue, diffusion away from these tissues is possible, with the potential for toxicity resulting.

Our lab has used the photocleavage approach to control insulin release from a shallow, light accessible dermal depot. We accomplish this by linking photocleavage to a change in solubility of the drug, in our case insulin. These approaches for solubility modulation include linking insulin to a polymer, linking insulin to itself in a “macropolymer”, linking insulin to highly non-polar moieties, and linking insulin to iso-electric point shifting groups.^[13] In all these cases, the result is a species that is insoluble and able to form a depot but upon irradiation will release native soluble insulin. The rationale behind such materials is to use them in shallow, dermal depots that can be stimulated by surface irradiation to release insulin in response to blood glucose information.

Photoconformational Control:

Multiple groups are pursuing the light control of drug action by incorporating photoinduced conformational changes into drugs.^{[1k, 14]} A common motif to introduce this control is the azobenzene linkage, that can switch from the typically thermodynamically more stable trans conformation to the cis conformation upon application of light. The rationale behind this approach is to switch the drug from an inactive state, unable to bind to the target of interest, to an active state after light is applied. The purpose of this, like much of photocleavage activation, is to spatially trigger activity only in the target tissue, for example a site of bacterial infection or cancer metastasis. This can potentially limit toxicity in surrounding tissues, as well as in the environment. Often, with a photoinduced conformational change there is a subsequent reversion to the initial conformational state, typically thermally initiated. This is a potential advantage of this approach over photocleavage, in that activated drug molecules that have been activated in the target tissue and then diffuse away can spontaneously revert to their inactive, and potentially non-toxic forms.

For example, Feringa and coworkers made multiple versions of the known proteasome inhibitor bortezomib, by grafting azo benzene moieties onto it. Several of these showed promise by having preferential inhibition of the target in the photoactivated conformation.^[15] In a rare in-vivo demonstration, the multi-disciplinary team of Trauner, Rutter, Hodson and coworkers used the azo benzene approach to control the activity of a second generation sulfonylurea to regulate blood glucose, in-vivo, through the use of a fiber optic delivered light.^[16] This stimulated the formation of the active conformation of the sulfonylurea, with commensurate blood glucose reduction. The use of a fiber optic to irradiate the pancreas may prove to be impractical in a real world setting. In another compelling example, Peifer and coworkers demonstrated the possibility of a photoconformational control of drug activity in an existing drug, the tyrosine kinase inhibitor axitinib.^[17]

A challenge that exists with the photoconoformational approach is the difficulty of identifying molecules that have strongly different inhibition profiles in their two different conformations. Often, inhibition of the light induced conformer is similar to that of the thermally more stable conformer. Inhibition of the target can also be stronger with the “wrong” i.e. thermally stable conformation. Significant iterations have been pursued recently, with some examples showing significant preference of the light induced conformer for the target.^[18] Other improvements include the increase of the photoisomerization wavelength to increase the tissue penetration ability of the applied light. The overall challenge of tissue light penetration will be discussed in greater detail later in this review.

Photothermal Control:

One of the challenges of light controlled drug release is the difficulty of delivering significant numbers of photons to the target site. With photochemical methods such as photocleavage, one is limited by the inherent photochemistry: high energy processes such as bond breaking typically require high energy photons to accomplish. These in turn are associated with shorter wavelengths which have poor tissue penetration. An alternative to this is to use photons in an indirect way, as a source of localized heat that then stimulates a physical change in a material that then allows drugs to be released. The photons utilized in such applications can be in the infrared range and therefore capable of significant tissue penetration.

Gold nanoparticles are a common motif used to absorb light and convert it to heat. For example Halas and coworkers demonstrated release of anti-cancer compounds such as taxol from gold shell SiO2 nanoparticles, using near IR radiation.^[19] The nanoparticles were modified with albumin which bound drug, until irradiation denatured the protein and released active drug. They were able to demonstrate cellular uptake of the nanocomplexes as well as a light dependent toxicity towards cells in culture. In another example, Xia and coworkers used gold “nanocages” to control the release of doxorubicin using near-IR laser irradiation.^[20] In this example, heat generated by the interaction of light with the gold nanocages triggered a thermally-induced conformational change in polymers adhered to the cages. This then allowed entrapped drug to be released. The effectiveness was demonstrated with both small molecules and proteins (the model lysozyme). Zhou and coworkers used a gold/silver hybrid nanoparticle system, to control the release of a model (curcumin) using near-IR irradiation. Curcumin release was triggered by the temperature responsive swelling/deswelling transition of an outer PEG layer.^[21] A potentially significant challenge with the photothermal approach is the cost and unknown toxicity associated with long term exposure to metal nanoparticles.

Photodegradation Control:

In this mechanism, photocleavable groups are used to link oligomers and in so doing create larger species, such as polymers or micelles, that can entrap drug molecules non-covalently. Upon irradiation the links between oligomers can be photocleaved, changing the structure of the species (e.g. increasing porosity) allowing entrapped drug molecules to be released. Almutairi and coworkers applied the photodegradation approach using a new light sensitive polymer which contains multiple quinone methide self immolative moieties installed along the backbone. Upon irradiation it showed burst release of a model small molecule dye, a proxy for a drug.^[22] Gillies and coworkers used a related approach for the light stimulated release of paclitaxel. They incorporated a light cleavable ortho nitro benzylic crosslinking group into a poly (ester amide). This when formulated with poly(ethylene oxide) and paclitaxel produced a micellar material that responded to 365nm UV light. The light activated crosslinks were cleaved, destabilizing the micelles, and releasing paclitaxel.^[23] Using a similar ortho nitro benzyl based crosslinker, Tong and coworkers created nanocapsules of polyethyleneimine (PEI). Amine groups on PEI were both alkylated and acylated using the crosslinker 4-bromomethyl-3-nitrobenzoic acid. The resulting crosslinked PEI was fashioned into “nanocapsules” that contained a fluorescently labeled dextran as a proxy for a drug payload. Upon photolysis, the crosslink was broken, allowing the contents of the nanocapsule to be released.

A significant advantage of the photodegradation approach is the potential, at least in theory, to have greater than one drug molecule released for each photo-stimulated event. In photocleavage or photoconformational control, each photostimulated event is associated with a single drug being released or activated. This potential for amplification is expected to be dependent on the nature of the drug entrapment. In addition, there is the potential that one material can be applied to any therapeutic, once it has been optimized. A potential disadvantage is the unknown systemic effects of the polymers/oligomer left behind after photodegradation.

The Issue of Tissue Light Penetration

Independent of which mechanism is being used, the promise of light activated drug action hinges on light reaching the target site in an amount sufficient to release a therapeutically relevant amount of drug. This is a major and possibly insurmountable problem in the most general sense. It hinges on many variables: the concentration of drug required for therapeutic action, the pharmacokinetics of the drug (i.e. rate of uptake/clearance and the likelihood of the target concentration being achieved), the differential efficacy found between different photo states (for photo switched as opposed to photoreleased drugs), the location of the site of action, the location of the site of irradiation, the light absorptivity of the drug, the quantum yield for turning that absorbed light into the desired activation, and finally the amount of light reaching the site of irradiation. Most of these factors have the potential to be optimized, by careful design and iteration by the chemist. The last factor however is the most challenging as it is bound by physical and biological limitations: the inherent amount of light that can penetrate tissue to reach the activation site without inducing thermal damage to the intervening tissues.

In the earliest days of this discipline, the issue of tissue light penetration was largely ignored, while the fundamental concepts and approaches for light activated drug release were explored. These early studies focused on the synthesis of modified drug species whose activity could be toggled with light and then followed up with a demonstration of in-vitro release of the drug. The selection of photoswitchable or photocleavable groups was often based on synthetic accessibility, frequently resulting in systems that utilized short wavelengths that can only penetrate tissue to a depth of a few millimeters. The kinetics of drug release were carefully monitored but in in-vitro settings which allowed for convenient control of important factors, such as light intensity. We can see this work as the necessary early steps in the development of the discipline. If these steps fail then there is no need to try the later and more challenging steps (e.g. in-vivo activity). New iterations have attempted to push the field in the “right direction” i.e. through the incorporation of higher and higher wavelength photocleavable/switchable groups, that should allow easier access by light. As the field has matured, and the number of examples of photo released drugs has increased, and as multiple iterations have solved early challenges, it seems fair to examine closely the issue of the penetration of light through tissue to better define the limits of this discipline and to also determine the disease states and applications that would be best suited for a photo activated drug approach.

Recently, Feringa, Szemanski and coworkers made an initial critical analysis in this direction, breaking down potential medical applications of light activated drugs based on the ease of light access to the afflicted tissue.^[14] These range from the skin and eye (i.e. directly accessible via light without significant intervention), to the interior of the body (much larger barrier to light). Sites of intermediate potential include targets that are accessible endoscopically. Our lab’s work on the photoactivated depot approach largely side steps these issues, as it designed around a shallow, skin based depot in which drug is contained, and then released by trans-cutaneous irradiation. The amount of tissue that needs to be traversed by light is inherently on the order of millimeters. What is lost in the PAD approach is spatial targeting. For some applications, for example insulin release, spatial targeting is less important than the timing and amount of drug release, both of which can be controlled by light being directed at the known site of depot injection. But for many applications, the long stated goal of photopharmacology has been to spatially target tissues of the body for drug activation, thus producing active drug only at the target site, and therefore reducing the overall toxic burden to the healthy tissues of the body, or reducing the toxic burden of the drug molecules on the environment. These are significant and reasonable goals. Two major stated applications are cancer and bacterial infection.

The question we now ask is what is the ultimate limit for non-surface applications of light activated drug release? Is it a reasonable goal or does it lie outside of the realm of possibility given the physical constraints of the system? To begin to answer this question we will examine the parameters for a hypothetical target, and predict the total energy needed to produce the desired therapeutic effect. The ultimate constraint is the amount of heat generated in tissue through the process of irradiation. There is a limit to how much heat can be dissipated from tissues, after which irreversible thermal damage will be the result. Our approach for this analysis is to start at the target site and then conceptually work our way back out to the body surface, to ultimately determine the required amount of surface applied photons to achieve the desired therapeutic effect. We will make our assumptions explicit and attempt to make them conservatively so as to define a limiting case.

We consider a spherical target site with a radius of 1 cm, though clearly some targets can be smaller or larger. This is a reasonable value for a “typical” treatment site for a tumor or bacterial infection site. This results in a target volume of 4.2 × 10⁻³ liters. We place this target at a depth of 5cm from the skin’s surface. This is clearly not as deep a target as is possible in the human body, but is a reasonable representative of a “non-surface” target. In this volume we require active drug. A good drug will have a K_d value for its target of 1nM and if we wish for a majority of the target to be bound, a concentration that is minimally ten times this is reasonable or 10nM of active drug in the target volume. This then corresponds to 4.2×10⁻¹¹ moles.

Photons arriving at the target tissue first have to be absorbed by the drug, and then converted into active drug by a photochemical process. The proportion of target-applied photons that are absorbed by the drug is linked to the concentration of the drug, the extinction coefficient of the drug and the pathlength. We have already determined that we need 1×10⁻⁸ M minimum of drug concentration. The pathlength will vary over the cross-section of the spherical target, but we will assume a 2cm distance, the largest diameter of the target. Previously described light activated drug candidates have varying extinction coefficients, so we use a generous value of 10,000 M⁻¹cm⁻¹, understanding that most will be lower than this, and less light will be absorbed. Given these parameters of 1×10⁻⁸ M drug concentration, 2cm pathlength and 10,000 M⁻¹cm⁻¹ extinction, we calculate an absorbance of the drug in the target tissue of 2×10⁻⁴, although with many photo-drugs this absorption could be an order of magnitude less. We are not considering photon losses within the target due to scattering or tissue absorption in the target, though this too will reduce the number of photons able to be absorbed by the drug. We will explicitly deal with the light attenuation due to tissue between the skin and target shortly.

Given the calculated absorption value of the drug in the target tissue, what proportion of photons arriving at the target tissue will be absorbed by the drug? This proportion is (I₀-I)/I₀, where I₀ is the intensity of the light arriving at the target, and I is the intensity of the light departing the target after absorption has taken place. This corresponds to 1-I/I₀. Given that the absorbance A = log (I₀/I), and I/I₀=10^−A then the value of 1- I/I₀ evaluates to 1– 10^−A. This results in a proportion of the light arriving at the target that is absorbed by drug as 4.6×10⁻⁴. The amount of drug converted by light into its active state is related to this amount of light absorbed by the target and the quantum yield of the photochemical process. The quantum yield too is a highly variable number. We have assigned a quantum yield of 0.1 to our hypothetical system which is arguably generous. Given this quantum yield, the proportion of light arriving at the target that is absorbed by the drug and then converted into active drug is 4.6×10⁻⁵.

We will now consider the amount of light that reaches the target in the first place, after having been applied to the skin and traversing 5 cm of tissue. The light attenuation of tissue has been extensively studied, in varying tissues and with varying wavelengths. The existence of a “window” in the infra-red region has been well described. This does not mean however that tissue is transparent in this spectral region, just that absorbance is at a relative minimum. Bashkatov and coworkers have extensively studied light transmittance in a wide range of tissues, including skin, colon, and others.^[24] In reviewing the absorption coefficients in this IR window of 600–1300nm, they indicated that the typical value was 2.5±1.5 cm⁻¹.^[25] This is a value that is consistent with other group’s analyses of light penetration in tissues.^[26] Based on this we used a conservative value of 1 cm⁻¹, although some tissues such as skin appear to be significantly higher, an observation confirmed by multiple authors. This is an absorbance coefficient, so a value of 1 cm⁻¹ corresponds to an attenuation factor of 10 for each cm of tissue traversed. At a depth of 5 cm, this is a proportion of 10⁻⁵ of the skin applied light arriving at the target. This, combined with the factor of 4.6×10⁻⁵ for the proportion of light arriving at the target that is converted to active drug, means that a total of 4.6×10⁻¹⁰ of the skin applied photons get converted into active drug in the target site.

From the previous calculation, we determined that we needed 4.2×10⁻¹¹ moles of active drug in the target tissue. This corresponds to 2.5×10¹³ molecules. Given that 4.6×10⁻¹⁰ of the skin applied photons get converted into active drug, this indicates that we need to apply 2.5×10¹³/4.6×10⁻¹⁰ photons or 5.4×10²² photons to the skin to get the required drug release. A typical infrared photon of 700nm carries 2.84×10⁻¹⁹ J of energy. Thus the total amount of energy applied to the skin is 5.4×10²² × 2.84×10⁻¹⁹ = 1.54×10⁴ J. Only a tiny proportion of this total energy results in drug activation and the majority of this energy will be converted to heat in the tissue, through both absorptive and internal scattering processes. The 1 cm radius, 5 cm high cylinder of tissue between the skin and the target has a volume of 5xπ or 15.7cc. 1.54×10⁴ J is about four times the energy required to boil the equivalent amount of water in the volume between the skin and the target tissue. Needless to say irreversible tissue damage would occur at even a fraction of the amount of energy required to achieve a therapeutic effect. This effect is intensified because a majority (~90%) of the light is absorbed (and hence heat generated) in the outermost 1cm of tissue which will take the majority of the thermal damage.

There are multiple assumptions made in this calculation, but we have attempted to be conservative, to define a limiting case. One parameter that can be varied is the depth, as it is associated with an ~10-fold change in intensity with each cm. Shallower targets, as has been previously suggested, will likely have a better chance at success. There may be ways of mitigating tissue thermal damage by spreading the light “dose” out over time, or through the equivalent of tomography, wherein multiple converging beams reduce the overall thermal burden. But given the apparently very high thermal burden of whole body applications of photopharmacology that this analysis indicates, it seems reasonable to attempt to explore the ways in which this burden can be alleviated or indeed if it can be alleviated in the most general sense.

Conclusions

The manipulation of drug activity using light has significant promise, to allow the spacing, timing and amount of a drug’s action to be controlled using light. This, in turn, may allow major challenges in drug delivery, such as off-target toxicity and the need for continuously variable delivery to be addressed. The discipline has generated a wide range of approaches and intense creativity in the design of photoactivatable drug molecules. The eventual efficacy of these approaches for treating actual disease in an in-vivo setting will ultimately depend on the fundamental physical nature of living tissue, which appears to be a formidable barrier to a general solution. The ultimate application and success of the approach may require even greater innovation to overcome these fundamental barriers, but the promise of the discipline makes such efforts potentially worthwhile.

Figure 2.

Analysis of active drug formation versus surface applied photons.