IEEE COMMUNICATIONS SURVEYS & TUTORIALS 1

A Survey of Biological Building Blocks for Synthetic Molecular Communication Systems Christian A. Soldner,¨ Eileen Socher, Vahid Jamali, Wayan Wicke, Arman Ahmadzadeh, Hans-Georg Breitinger, Andreas Burkovski, Kathrin Castiglione, Robert Schober, and Heinrich Sticht

Abstract—Synthetic molecular communication (MC) is a new experimental evaluation of MCSs. One of the main advantages of communication engineering paradigm which is expected to enable employing proteins for signal emission and detection is that they revolutionary applications such as smart drug delivery and can be modified with tools from synthetic biology and be tailored real-time health monitoring. The design and implementation to a wide range of application needs. We discuss the properties, of synthetic MC systems (MCSs) at nano- and microscale is limitations, and applications of the proposed biological building very challenging. This is particularly true for synthetic MCSs blocks for synthetic MCSs in detail. Furthermore, we outline new employing biological components as transmitters and receivers research directions for the implementation and the theoretical or as interfaces with natural biological MCSs. Nevertheless, design and analysis of the proposed transmitter and receiver since such biological components have been optimized by nature architectures. over billions of years, using them in synthetic MCSs is highly promising. This paper provides a survey of biological components Index Terms—Molecular communications, transmitter and that can potentially serve as the main building blocks, i.e., receiver architecture, signaling particles, synthetic biology, and transmitter, receiver, and signaling particles, for the design test-bed implementation. and implementation of synthetic MCSs. Nature uses a large variety of signaling particles of different sizes and with vastly different properties for communication among biological entities. I.INTRODUCTION Here, we focus on three important classes of signaling particles: The development of nanomachines for medical applications cations (specifically protons and calcium ions), neurotransmit- such as real-time health monitoring and targeted drug delivery ters (specifically acetylcholine, dopamine, and serotonin), and phosphopeptides. These three classes have unique and distinct is a focus area of current nanotechnology research [1]–[3]. features such as their large diffusion coefficients, their specificity, In order to realize the full potential of such applications, and/or their uniqueness of signaling that make them suitable it is necessary that the nanomachines be able to efficiently candidates for signaling particles in synthetic MCSs. For each communicate with each other [4]–[7]. In particular, it is of these candidate signaling particles, we present several specific envisioned that a network of communicating nanomachines transmitter and receiver structures mainly built upon proteins that are capable of performing the distinct physiological func- can help realize the concept of the Internet of Bio-NanoThings tionalities required from the transmitters and receivers of MCSs. which is expected to enable nanomachines to perform complex Moreover, we present options for both microscale implementation tasks [8], [9]. For instance, a group of nanomachines may of MCSs as well as the micro-to-macroscale interfaces needed for detect a metabolic condition and communicate this obser- vation to another nanomachine which is then responsible This work was supported in part by the German Research Foundation under for triggering the release of a drug into the body. Since Projects SCHO 831/7-1 and SCHO 831/9-1, in part by the Friedrich-Alexander conventional communication techniques are not well suited for University Erlangen-Nurnberg¨ under the Emerging Fields Initiative, and in part by the STAEDTLER Foundation. communication at nano- and microscale, especially in liquid Christian A. Soldner¨ and Heinrich Sticht are with the Division of media, molecular communication (MC), where molecules are arXiv:1901.02221v3 [cs.ET] 9 Jul 2020 Bioinformatics, Institute of Biochemistry, Friedrich-Alexander-Universitat¨ used as information carriers, has been proposed as a promising Erlangen-Nurnberg¨ (FAU), Fahrstr. 17, 91054 Erlangen, Germany. (email: {christian.soeldner, heinrich.sticht}@fau.de) bio-inspired mechanism for enabling communication among Eileen Socher is with the Institute of Biochemistry and the Institute of nanomachines [4], [5]. Anatomy, Friedrich-Alexander-Universitat¨ Erlangen-Nurnberg¨ (FAU), 91054 The general structure of a (synthetic) MC system (MCS) Erlangen, Germany. (email: [email protected]) Vahid Jamali, Wayan Wicke, Arman Ahmadzadeh, and Robert Schober is depicted in Fig. 1. In response to a certain input signal, are with the Institute for Digital Communications, Department of Elec- which may be artificial (e.g. a light impulse or an electrical trical, Electronics, and Communication Engineering (EEI), Friedrich- stimulation) or biological (e.g. a nerve signal), the transmitter Alexander-Universitat¨ Erlangen-Nurnberg¨ (FAU), Cauerstr. 7, 91058 Er- 1 langen, Germany. (email: {vahid.jamali, wayan.wicke, arman.ahmadzadeh, releases a pattern of signaling particles , which represents the robert.schober}@fau.de) information to be conveyed. Depending on how sophisticated Hans-Georg Breitinger is with the Department of Biochemistry, Faculty of the transmitter is, it may also apply advanced encoding and Pharmacy and Biotechnology, German University in Cairo (GUC), New Cairo 11835, Egypt. (email: [email protected]) modulation techniques for efficient representation of the data Andreas Burkovski is with the Division of Microbiology, Department of before releasing the corresponding signaling particles into Biology, Friedrich-Alexander-Universitat¨ Erlangen-Nurnberg¨ (FAU), Staudtstr. the channel. The signaling particles propagate through the 5, 91058 Erlangen, Germany. (email: [email protected]) Kathrin Castiglione is with the Institute of Bioprocess Engineering, De- channel, e.g. via free diffusion where the propagation may be partment of Chemical and Bioengineering, Friedrich-Alexander-Universitat¨ Erlangen-Nurnberg¨ (FAU), Paul-Gordanstr. 3, 91052 Erlangen, Germany. 1Throughout this paper we use the terms molecules and particles interchange- (email: [email protected]) ably, although the latter term is broader as not all particles are molecules. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 2

Focus of This Survey Transmitter

Information Source Control/Computing Unit Physical Channel Natural/Synthetic Encoding, Release Stimulus, ... Modulation, ... Mechanism

Signaling Receiver Particles

Information Sink Control/Computing Unit

New Action, Detection, Reception Readout, ... Decoding, ... Mechanism

Fig. 1. General block diagram of an MCS. This paper surveys suitable biological building blocks for implementation of the release and reception mechanisms for several classes of signaling particles. The color code used to represent transmitter, channel, and receiver will be applied throughout the paper. further accelerated by advection [10]. The receiver observes the is the development of communication-theoretical models signaling particles and recovers the data by applying suitable for the release, propagation, and reception of the signaling demodulation and decoding techniques. Thereby, the data may particles that account for the features and constraints of either be read out using an artificial mechanism (e.g. via a the adopted biological building blocks [12]–[17]. Based light emission or an electrical current) or trigger a biological on these models, the basic functionalities of MCSs such as process (e.g. a nerve signal). channel coding [18], [19], modulation [20], [21], detection [21]–[23], decoding [19], [24], synchronization [25], [26], A. Motivation and Scope and estimation [27], [28] can be developed and their Although synthetic MC has received considerable interest performance can be analyzed. from the research community over the past decade, the research • Stage 3 – Control and Computing Modules: The area is still in its infancy. In particular, the design, analysis, implementation of the communication-theoretical concepts and implementation of microscale biological MCSs require developed in Stage 2 depends on where the correspond- inherently a multidisciplinary approach with contributions ing operations are to be performed. For instance, for from different engineering disciplines, including electrical, health monitoring applications where the observations biological, and chemical engineering, and different branches of can be collected and accessed from outside the MC science, including biology, chemistry, physics, and medicine. environment, a personal computer may be responsible Particularly, the field of synthetic biology is expected to play a for part of the processing. For other applications, such as crucial role in the fabrication and implementation of the main targeted drug delivery, sophisticated nano-transmitters and components of future synthetic MCSs, i.e., the transmitter, nano-receivers may have to process the data themselves. receiver, and signaling particles. Various options have been proposed for realizing con- In this paper, we review biological components suitable trol/computing units at nano- and microscale for biological for implementation of MCSs. In order to define the scope transmitters/receivers including molecular circuits (i.e., of this survey paper and to facilitate the classification of the cascaded networks of chemical reactions) [24], [29], [30] different research directions in the field of synthetic MC, we and genetic circuits [31]–[33]. present a roadmap for the development of synthetic biological • Stage 4 – Experimental Verification: The concepts and MCSs from the basic biological building blocks to commercial designs developed in Stages 1-3 have to be verified via applications, see Fig. 2. laboratory experiments [34]–[36]. Stage 4 is challenging due to the fact that controlling an MCS at microscale • Stage 1 – Enabling Basic Biological Hardware: The is difficult. Therefore, in addition to the development of fundamental feature of MCSs is that signaling particles microscale MCSs, it is advantageous to also develop micro- are employed as information carriers [11]. Therefore, in to-macroscale interfaces that enable their observation its most basic form, an MCS consists of a transmitter and test. The role of this stage in the development of that is able to release signaling particles into the channel MCSs is analogous to employing spectrum analyzers, and a receiver which is able to detect the presence of the channel sounders, and other measurement devices to test signaling particles. The primary focus of this survey is and analyze the components of wireless communication the compilation of various biological options for realizing systems [37]. Therefore, this paper does not only survey the release mechanism at the transmitter and the reception options for the implementation of MCSs but also the mechanism at the receiver for several different types of interfaces required for their experimental evaluation. signaling particles. • Stage 5 – Prototyping for Commercial Applications: • Stage 2 – Communication-Theoretical Modeling and Depending on the application, suitable building blocks Design: The next step needed for the design of an MCS IEEE COMMUNICATIONS SURVEYS & TUTORIALS 3

Stage 1 molecules, release mechanisms, and reception mechanisms Basic Biological Hardware needed in Stage 1 as well as the micro-to-macroscale interfaces

Release Signaling Reception needed for their experimental evaluation in Stage 4. Mechanism Particles Mechanism Several survey and tutorial papers focusing on different aspects of MC have been published over the past few years Stage 2 [5]–[7], [10], [50]–[67], see Table I for a brief summary of Communication-Theoretical Modeling and Design these survey and tutorial papers. In particular, the authors of Transmitter Channel Receiver [5]–[7], [56] provide general overviews of the field of MC, its Design Modeling Design future applications, and related challenges. In [7], [50], [51], [55], networking aspects of MCSs are discussed and potential Stage 3 network layer architectures for MCSs are proposed. A general Control and Computing Modules survey on synthetic MCSs, including aspects such as particle transport, communication engineering aspects, testbeds, and Molecular Genetic Circuits Circuits applications, is presented in [52]. The theoretical aspects of MC are surveyed in detail from the perspectives of information theory in [58], [59], physical-layer channel modeling in [10], and transmitter and receiver design in [60]. The design of genetic circuits is surveyed in [62] and the mutual impact of connected biological components is discussed in [63]. Potential medical applications of synthetic MCSs are surveyed in [53], [67]. Drug delivery applications are discussed in Stage 5 [55] and applications of mobile MCSs are presented in [61]. Prototyping for Commercial Applications In addition, the authors of [64] survey research works with a particular focus on MCs in the synaptic cleft and [57] surveys tools from bioinformatics for the analysis of protein- protein interactions. Finally, the authors of [65] propose Fig. 2. This figure illustrates a roadmap for the development of synthetic electrochemical methods to interface with biological MCSs biological MCSs towards commercial applications. The focus of this survey is on Stage 1, where we present different biological options for realizing whereas [66] studies optogenomic interfaces for controlling the release mechanism at the transmitter and the reception mechanism at genes and their interactions in the cell nucleus. These survey the receiver for several different types of signaling particles. The biological papers are either general overviews (e.g., [5]–[7], [52], [56]) or building blocks proposed in this paper may help theoreticians to develop models and communication-theoretical system designs that account for the relevant focus mainly on Stage 2 (e.g., [10], [58]–[61]) and Stage 3 (e.g., biological constraints and features imposed by the proposed building blocks (in [62], [63]) of the development roadmap illustrated in Fig. 2, Stage 2). Moreover, we propose architectures suitable for implementation of and although most of them also consider nanomachines as both microscale MCSs as well as the micro-to-macroscale interfaces required for experimental evaluation of synthetic MCSs (in Stage 4). components of MCSs, the corresponding survey of the related literature is very brief. A comprehensive survey of potential biological building blocks of MCSs (in Stage 1) and their micro-to-macroscale interfaces (in Stage 4) is not available in from Stages 1-3 (that have also been experimentally the literature, yet. verified in Stage 4) are chosen to develop first-order While biological building blocks of MCSs have received prototypes and ensure that these blocks successfully work little attention in the MC literature, in the field of synthetic together. At this level, there are numerous applications for biology, there is a vast body of literature on biological systems MCSs including smart drug delivery, health monitoring, that can potentially be used as components of MCSs. However, and even the realization of the Internet of Bio-NanoThings for researchers not well versed in synthetic biology, it can [8], [9], [38], [39]. be challenging to find the relevant literature and to relate So far, the main focus of the MC literature has been on it to MCS design. Therefore, in this paper, we provide a Stages 2 and 3 assuming often quite abstract and simple models comprehensive survey of biological building blocks that can for the underlying biological building blocks in Stage 1. In potentially be engineered to serve as components of nano- and addition, as a proof of concept, MCSs have been demonstrated microscale synthetic MCSs operating in aqueous environments. at macroscale [12], [40]–[44] and at microscale [34]–[36], Since, unlike what is often assumed in the MC literature, [45]–[49] (i.e., Stage 4). The latter systems employ biological biological systems are very specific, the signaling particles, components as transmitter and/or receiver but were either transmitters, and receivers have to be carefully matched to demonstrated only for single pulse transmission or offer very each other. In particular, the design of the transmitter and low data rates on the order of one symbol per hour. However, receiver in MCSs crucially depends on the adopted signaling for nano- and microscale MC to become practical, continuous particles. Hence, in this survey, we adopt a signaling particle transmission at much higher data rates is needed. Both the centric approach and first present several candidate signaling design and the implementation of such systems require a sound particles for synthetic MCSs. Then, for each of the considered understanding of the biological building blocks that can be used signaling particles, we provide several candidate transmitter to construct them. This paper surveys candidates for signaling and receiver structures. We believe that this survey is useful IEEE COMMUNICATIONS SURVEYS & TUTORIALS 4

TABLE I SUMMARY OF SURVEY AND TUTORIAL PAPERS ON MCSS.MOST OF THE PAPERS LISTED IN THIS TABLE PROVIDE AN OVERVIEW OF MCSS AND CONTAIN CONTENTRELATEDTOALLDEVELOPMENTSTAGESSHOWNIN FIG. 2. FORCLARITY, THEMAINFOCUSOFEACHSURVEY/TUTORIAL PAPER IS REPORTED IN THE FOURTH COLUMN OF THIS TABLE.

Reference Year Type Main Focus More Detailed Notes

Akyildiz et al. [5] 2008 Survey Overview Nanonetworks, architectural aspects, and expected features Dressler et al. [6] 2010 Survey Overview Bio-inspired networking approaches Nakano et al. [7] 2012 Survey Stage 2 Physical and network layers of MCSs Darchini et al. [50] 2013 Survey Stage 2 MCSs via microtubules and physical contact Nakano et al. [51] 2014 Survey Stage 2 Layered architecture of MCSs Farsad et al. [52] 2016 Survey Overview/ Underlying physical principles of MCSs, communication engineering aspects, and Stage 2 simulation tools Felicetti et al. [53] 2016 Survey Overview Medical applications of MCSs, diagnostic and treatment applications, and implementation interfaces Chahibi et al. [54] 2017 Survey Overview Targeted drug delivery, component and modeling approaches Okonkwo et al. [55] 2017 Survey Stages 1, 2, Targeted drug delivery, application concepts, propagation channel modeling, and and 4 system design Wang et al. [56] 2017 Survey Overview Diffusive MCSs, communication theoretical designs, and cooperative relay-based networks Jamali et al. [10] 2019 Tutorial Stage 2 Channel modeling of diffusive MCSs, physical principles, communication theoretical, simulation-based, and data-driven models, and model derivation methodologies Akyildiz et al. [58] 2019 Tutorial Stage 2 MC theory and models for functional blocks of MCSs based on chemical kinetics and statistical mechanics Rose et al. [59] 2019 Tutorial Stage 2 Capacity of point-to-point MCSs Kuscu et al. [60] 2019 Survey Stage 2 Transmitter and receiver architectures of MCSs, modulation, coding, and detection techniques Nakano et al. [61] 2019 Survey Stage 2 Mobile MCSs, modeling approaches, and networking Nguyen et al. [62] 2019 Tutorial Stage 3 Asynchronous genetic circuits McBride et al. [63] 2019 Tutorial Stage 3 Synthetic biomolecular circuits Veletic´ et al. [64] 2019 Survey Stage 2 Synaptic communication engineering and brain–machine interface Kim et al. [65] 2019 Survey Overview Reduction/oxidation (redox) reactions Jornet et al. [66] 2019 Survey Overview Optogenomic interfaces Qadri et al. [67] 2020 Survey Stage 2 Internet of Nano-Things for healthcare applications Soldner¨ et al. (this paper) – Survey Stages 1 and 4 Potential biological building blocks of MCSs, microscale implementations, and macroscale interfaces to both theoreticians and experimentalists. For researchers which encloses a liquid substance [68]. Being much working on the theoretical aspects of MCS design, taking into smaller than a cell, natural vesicles are formed by defor- account the specific properties of the underlying biological mation and subsequent budding from the cell membrane building blocks, which are reviewed here and described in or the membrane of cellular organelles such as the Golgi detail in the provided references from synthetic biology, will apparatus or the endoplasmic reticulum. They are used allow them to develop more realistic communication-theoretical for transport purposes, e.g. in the context of secretion, for models and designs of MCSs. For researchers interested in the storage of certain biomolecules, and as compartments developing MC testbeds and experiments, the survey outlines with particular reaction conditions. Different proteins may the advantages and disadvantages of potential design choices be embedded in their lipid membrane which may facilitate and the provided references contain the detailed information for instance transmembrane transport. Moreover, vesicles needed for implementation. In the following subsections, we with transmembrane proteins can be created artificially first explain some basic biological concepts and components. using biochemistry and molecular biology tools. Such Then, we provide a brief overview of the considered signaling artificially generated lipid vesicles are called liposomes. particles and matching biological components, which can be • Ion channels: An ion channel is a special type of trans- used to construct transmitters and receivers. membrane protein with a pore that becomes permeable for specific types of ions under certain circumstances [69]. Ion B. Some Important Basic Biological Concepts and Components channels only allow passive transport which means that they merely facilitate diffusion along an already existing In the following, to assist readers that do not have a concentration gradient. Ion channels may be subdivided background in biology, we explain some essential basic into different groups based on the conditions that lead to biological concepts and components used throughout this article. channel opening (“gating”). For instance, there are voltage- A summary of additional biological concepts and terminology gated (a certain transmembrane potential is required), appearing in the text is provided in Appendix A. ligand-gated (a certain molecule has to bind from the • Vesicles: A vesicle is a small, round or oval-shaped outside), and mechanosensitive (stretching of/pressure on container whose wall consists of a lipid bilayer membrane IEEE COMMUNICATIONS SURVEYS & TUTORIALS 5

the membrane is required) ion channels [69]. ness, in this survey, we focus on three important types of • Carriers: Another kind of protein involved in the move- signaling particles, namely cations, neurotransmitters (NTs), ment of ions or small molecules across membranes are and phosphopeptides as a representative class of modified carriers [70]. The transported particles are also referred to proteins, see Fig. 3. These classes of signaling particles are as substrate in this context. So called uniporters transport attractive for use in synthetic MCSs as they allow the design only one specific type of substrate. Upon stimulation, they of simple transmitter and receiver structures employing only a undergo a conformational change whereby the particle is small number of protein components. Furthermore, as will be carried through the membrane to be released at the other explained in the following, the considered classes of signaling side. As other secondary carriers (see below) they are particles differ substantially in their behavior and properties, able to accumulate their substrate against a concentration such that they are collectively suitable for a wide class of gradient [70]. However, there are carriers which do not different MCSs. transport one type of substrate alone, but two or more Cations: Cations2 are small positively charged ions that different types of substrates (e.g. particles A and B) simul- have the advantage of fast diffusion [74], [75]. In addition, taneously, either both in the same direction (symporter) or protons, as a specific ion, can jump from one water molecule in opposite directions (antiporter). This principle, which to the next through the formation and concomitant cleavage of is called secondary active transport, allows to transport covalent bonds, the so-called Grotthuss mechanism [76], [77], substrate A against its concentration gradient if there is which makes them appear to move even faster than they would a sufficiently high concentration or charge gradient for already be by classic diffusion due to their small size. Since substrate B that can be used as a source of energy for the ions interact with a plethora of proteins in organisms [78]– transport process. Often, the concentration gradient for [80], there exists a large set of possible biological structures substrate B is actively maintained by use of an ion pump that can be used as components of transmitters and receivers. [70]. Although many types of cations and anions may be appropriate • Ion pumps: Similar to ion channels and carriers, ion for MCSs, the present article focuses on protons due to their pumps are transmembrane proteins which can transport unique speed of diffusion, and on calcium ions, because they ions across a membrane. In contrast to ion channels, ion play an important role as messengers in cellular signaling pumps use an external source of energy such as light [81]. For instance, calcium ions are involved in the process of or adenosine triphosphate (ATP) to facilitate an active apoptosis (programmed cell death of damaged cells) [82] as transport which also works against the concentration well as in the coupling between the electrical excitation and gradient of the respective ion [71]. This type of transport the consecutive contraction of muscle fibers [83]. is also called primary active transport. Neurotransmitters: As an alternative class of particles for • Voltage-clamp method: The voltage-clamp technique is a MCSs, we consider NTs. NTs are bigger than cations and method where a microelectrode is placed inside a vesicle therefore diffuse more slowly. On the other hand, they are or a cell to manipulate or measure the current across more specific and are therefore more likely to avoid unintended the vesicle membrane at a certain voltage [72]. Thereby, signal interference from and to different natural processes e.g. changes in membrane potential can be induced, e.g. in in the human body. The class of NTs comprises several distinct order to open a voltage-gated ion channel. Moreover, the molecules (e.g. acetylcholine, dopamine, and serotonin) that all two-electrode voltage clamp technique can be used to function in a similar fashion in signal transduction. This has measure the transmembrane current that arises when ion the general advantage that similar building blocks can be used channels are opened. The two-electrode voltage clamp for the construction of transmitters and receivers for different technique allows adjustment of the transmembrane poten- NTs. In nature, NTs are used to trigger or suppress neuronal tial and recording of currents through separate electrodes firing (the so-called action potentials) between neurons or from [72], and is mostly used for measurements on oocytes or a neuron to a muscle fiber [84]. Thus, NTs are particularly very large vesicles/cells (> 1-2 mm in diameter) [73]. interesting because they could be used to directly interact with • Reversibility: In this article, we refer to transmitters as neurons [7], e.g. in the context of nerve lesions which need “reversible” if the signaling particles are recycled after their to be bridged, the control of a prosthesis, and the treatment release so that they can be used repeatedly. For example, of neurological diseases. One possible application would be a simple vesicle-based transmitter may eventually get targeted drug delivery. For example, Parkinson’s disease which exhausted over time having released all signaling particles is caused by a selective loss of dopamine producing neurons in that were stored inside at the beginning. In contrast, a a specific region of the brain, the so-called substantia nigra, is “reversible” transmitter is able to regenerate its content, e.g. nowadays treated by a systemic oral administration of levodopa, by pumping the signaling particles back inside. In addition, a precursor of dopamine synthesis [85]. In this context, a reversibility will require that the vesicle exhibits a high direct delivery of dopamine via MC to the region where it is stability and remains intact during multiple regeneration required would greatly enhance the treatment effectiveness of cycles of the transmitter. Parkinson’s disease. Modified proteins: The final class of signaling particles C. Signaling Particles

Nature uses a vast number of different molecules for 2In a similar manner, anions (i.e., small negatively charged ions) can be information exchange between different entities. For concrete- used for signaling in MCSs. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 6 discussed in this article are proteins that become signaling par- that have analyzed the performance and the design of MCSs ticles by post-translational modification (PTM). PTM describes for the signaling particles introduced above. For example, the a covalent enzymatic modification of proteins that occurs after use of calcium/proton ions as signaling particles in MCSs is protein biosynthesis. There are different types of PTMs (e.g. considered in [36], [94]–[100]. Furthermore, several theoretical phosphorylation, methylation, acetylation, and ubiquitination), works investigate the design and analysis of neuronal MCSs which differ in their functional groups, their size, and the way employing NTs as information particles, see e.g. [101]–[106]. they are linked to the protein [86]. Protein phosphorylation, However, these existing theoretical works either considered one of the most frequent types of PTM, is an important abstract models for the transmitter and receiver or focused cellular regulatory mechanism as many proteins (e.g. enzymes, on transmitter and/or receiver structures of natural MCSs. In receptors) are activated/deactivated by phosphorylation and this paper, unlike the existing works, we investigate different dephosphorylation events. More than two-thirds of the 21,000 transmitter and/or receiver structures for each of the considered proteins encoded by the human genome were shown to be signaling particles with the design of synthetic MCSs in mind. phosphorylated [87] indicating that phosphorylation plays a Furthermore, the proposed transmitter and receiver structures role in almost every physiological process. Abnormal pro- are not limited to those existing in nature. tein phosphorylation may lead to severe diseases including Alzheimer’s disease, Parkinson’s disease, and cancer. From a D. Biological Components of Transmitter and Receiver communication-theoretic perspective, the protein moiety is always present in the channel but it is only activated for The signaling particles described in Section I-C are compat- signaling if it carries a phosphate group (orange phosphorous ible with several biomolecules that can be used to construct with red oxygen atoms in Fig. 3) that is transferred by a kinase transmitters and receivers for MCSs. An overview of some [88], a special type of enzyme. We consider phosphorylations relevant classes of such biomolecules is given in Fig. 4. One because of their versatility due to the large number of different relevant protein family are channels that allow ions to pass signaling particles that can be generated by variation of the through a membrane via a channel pore in response to an protein sequence. In addition, phosphorylated proteins may be external stimulus (e.g. voltage, pH, ligands, or mechanical miniaturized to generate smaller signaling particles (henceforth stress) (Figs. 4a, d, e, f). In contrast to such ion channels, termed ‘phosphopeptides’) that have the advantage of faster which only allow diffusion of ions along a concentration diffusion due to their smaller size compared to proteins. gradient, pumps (Fig. 4b) and symporters (Fig. 4g) can move Fig. 3 provides an overview of the properties of the different molecules against a concentration gradient at the expense of signaling particles that we survey for their applicability in energy consumption. G-protein coupled receptors (Fig. 4h) MCSs. From top to bottom, they are ordered according to their detect ligands outside the cell and convert the corresponding molecular weights (MWs). As mentioned before, one common signal to intracellular responses. These biomolecules can be property of all three considered classes of signaling particles integrated into the membrane of cells or artificial vesicles is that their respective transmitters and receivers require only using the tools of synthetic biology [117]–[121]. In addition a few biological components which makes them attractive for to these membrane-bound components, there exist also soluble application in synthetic MCSs. In addition, they are intrinsically proteins with properties relevant for MCSs. For instance, rather homogeneous groups of signaling particles, which kinases (Fig. 4j) can convert peptides into signaling particles facilitates the design of transmitter and receiver structures and enzymes (Fig. 4i) allow the degradation of signaling as well as communication protocols that can be applied to particles. multiple members of these classes. Fig. 5 presents different classifications of the biological Remark 1: We refrained from considering hormones as transmitter and receiver architectures discussed in this paper. signaling particles in this survey, because they are very As mentioned in Section I-A, experimental verification of the divergent as far as their size (molecular weights from about concepts conceived in Stages 1-3 of the proposed roadmap 150 g/mol up to insulin which is a protein with 51 amino for the development of MCSs is in general difficult since acids and a molecular weight of 5800 g/mol), the involved controlling an MCS at microscale is a challenging task. receptors (transmembrane, intracellular, and nuclear), and also Therefore, for each class of signaling particles, we present the effect that they have on their target cell are concerned options for both microscale implementation of the MCSs and [91]. Therefore, it is not possible to develop a general MCS microscale-to-macroscale interfaces that can be used for their architecture valid for the entire group of hormones. We also do experimental evaluation (i.e., in Stage 4). For instance, light- not consider complex biological communication systems such driven ion-pumps can be inserted into the membrane of the as chemotaxis or bacterial quorum sensing, which would require transmitting cell to enable the release of ions (as signaling the challenging implementation of the signaling cascades particles) in response to an external light stimulus which is easy involved [92], [93]. Quorum sensing, for example, depends on to control in laboratory experiments. For practical microscale the conditional activation of gene transcription [93] and would MCSs, other stimuli may be preferred such as other molecules thus only be possible if complete bacterial cells were used as (originating potentially from other synthetic or natural sources) transmitters/receivers. In this paper, we focus on the design which open corresponding ligand-gated channels to release the of simple transmitters and receivers consisting of only a few signaling particles. protein components. Conceptually, transmitter architectures can be also cate- Remark 2: In the MC literature, there are several works gorized based on whether the particle release is induced IEEE COMMUNICATIONS SURVEYS & TUTORIALS 7

Fig. 3. Overview of the considered signaling particles, their molecular weight (MW), diffusion coefficient (D), and chemical structure. Carbon atoms are colored in black, hydrogen atoms in white, oxygen atoms in red, nitrogen atoms in blue, and phosphorous atoms in orange. Diffusion coefficients were taken from [74] (protons, acetylcholine), [75] (calcium ions), and [89] (dopamine, serotonin). The structure images were created using UCSF Chimera [90]. manually (e.g., via a pipette in a laboratory experiment); particles are presented. Section V introduces transmitters that initiated by binding other molecules to ligand-gated channels use phosphorylation as a mechanism for controlling the release on the transmitter surface; or triggered by an electrical, optical, of peptide particles as well as different receivers that can detect or chemical stimulus. Similarly, receiver architectures can be phosphorylated peptides. Several extensions of the proposed classified based on whether the signaling particles bind to transmitter/receiver structures to more complex architectures, receptors on the receiver surface and trigger another secondary capable of performing more elaborate processing tasks, are also signal inside the receiver or the reception process causes provided. In Sections II-V, the respective different synthetic an electrically, optically, or chemically detectable signal. In transmitter and/or receiver architectures are presented in addition to design principles for synthetic transmitters and order of increasing sophistication and complexity required for receivers, we also discuss natural mechanisms in the human integrating the necessary components into the transmitter and body for the release and detection of the considered signaling receiver. Comparatively simple building blocks, which require particles. These mechanisms can be interpreted as natural external devices or the use of detergents, are rather intended for transmitters and receivers or natural sources of interference. in vitro purposes such as testbeds or benchmarking experiments. They could be used for an initial proof of concept whereas more complex cell- or vesicle-based systems may be better E. Organization of the Paper suited for in vivo applications. We note that there is no one- In the following, we outline how the remainder of the to-one mapping between the orders in which the transmitters paper and the individual sections are organized. In particular, and the receivers are presented. In other words, depending on in Sections II and III, we discuss transmitter and receiver the specific application, the proposed transmitter and receiver mechanisms for protons and calcium ions, respectively. In structures can be flexibly combined. Moreover, in Section VI, Section IV, transmitter and receiver structures for NT signaling IEEE COMMUNICATIONS SURVEYS & TUTORIALS 8

Fig. 4. Exemplary protein structures of potential MCSs building blocks. For transmembrane proteins, the position of the membrane is highlighted by two planes. All structures are listed with a unique identifier (PDB code) referring to the protein data bank (www.rcsb.org). (a) Voltage-gated channel: Cav1.1 (PDB code 5GJV [107]). (b) Light-driven pump: Bacteriorhodopsin (PDB code 1M0K [108]). (c) Fluorescent protein: Green fluorescent protein (GFP; PDB code 1GFL [109]). (d) pH-dependent channel: Acid-sensing ion channel (ASIC; PDB code 4FZ0 [110]). (e) Ligand-gated ion channel: Serotonin receptor subtype 5-HT3A (PDB code 4PIR [111]). (f) Mechanosensitive ion channel: Transient receptor potential vanilloid 4 (TRPV4; PDB code 6C8F [112]). (g) NT symporter: Dopamine transporter (DAT; PDB code 4XP1 [113]). (h) G-protein coupled receptor (GPCR): M3 muscarinic receptor (PDB code 4DAJ [114]). (i) NT degrading enzyme: Acetylcholinesterase (PDB code 4BDT [115]). (j) Kinase: Mitogen-activated protein (MAP) kinase 1 (MEK1; PDB code 1S9J [116]). Structure images were created using UCSF Chimera [90]. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 9 Scale Interface Microscale Macroscale Implementation R5, Fig. 7 R5, Fig. 7 R7, Fig. 7 R3, Fig. 9 Example R3/R4, Fig. 7 R1/R2, Fig. 9 R3/R4, II Table R1/R2, II Table R1/R2/R6,Fig.7 R2,Fig.9,Fig.10 Reception Mechanism Optical Natural Electrical Chemical Type Ligand-gated Scale Interface Microscale Macroscale Implementation T4, Fig. 8 T1, Fig. 6 T1, Fig. 8 T2, Fig. 6 T3, Fig. 8 T3, Fig. 8 T6, Fig. 6 T6, Fig. 8 T1, II Table T2, II Table Example T4/T5, Fig. 6 T2/T3, Fig. 6 T4/T5, II Table T2/T3, II Table Fig.10/Table IV Fig.10/Table T2/T4/T5, Fig. 8 Release Mechanism Proposed Biological MC Building Blocks Optical Manual Natural Electrical Chemical Type Ligand-gated peptides Phospho- Protein Modifications Serotonin Dopamine Acetylcholine Neurotransmitter Signaling Particles Protons Ions Calcium Ions

Fig. 5. Classification of the biological building blocks for the MCSs presented in this paper. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 10 we compare the different signaling particles studied in this vesicle [123], which makes them perfect candidates for outward paper, discuss potential medical applications of the considered transportation. MCSs, and present several practical considerations for the T4 (Ion pumps): Alternatively, for the fourth transmitter implementation of these MCSs including relevant biological model, light-driven or ATP-driven pumps are exploited for mechanisms for inter-symbol interference (ISI) mitigation and the proton transmitter system (Fig. 6, T4). While voltage- bottlenecks for the achievable data rates. Furthermore, in gated ion channels conduct cations or anions in a passive Section VII, we outline possible future research directions manner, ion pumps need an external source of energy and including the new research problems that should be tackled function as active transporters, which allows them to build up for the modeling and design of MCSs based on the proposed a proton concentration gradient [124]. As light-driven outward biological building blocks as well as challenges that need to proton pump, for instance, bacteriorhodopsin [125] may be be addressed for the implementations of such MCSs. Finally, used (Fig. 4b), which can be embedded in the membrane of the conclusions of the paper are drawn in Section VIII. proton containing vesicles. In particular, there are several known variants of bacteriorhodopsin which differ in the wavelength II.PROTONSAS SIGNALING PARTICLES required for activation [126]. Recently, an experimental testbed Protons (represented by the symbol H+) are the first type of employing bacteriorhodopsin has been reported in [36]. As an signaling particles we consider for MCSs. The corresponding ATP-driven pump, V-ATPases or P-ATPases [127] could be options for transmitter systems are depicted in Fig. 6, those for used and activated by addition of ATP in the vicinity of the receiver systems are presented in Fig. 7. They are described vesicle. in detail in the following. T5 (Inward ion pumps): The previous two proposed trans- mitter mechanisms (Fig. 6, T3 and T4) enable the controlled release of signaling particle; however, they lack reversibility, A. Transmitters i.e., the recycling of the signaling particles by the transmitter. We consider six different transmitter structures (Fig. 6, By recycling the signaling particles, the transmitter can harvest T1-T6). While the first transmitter (T1) is quite simple (for some of the previously released protons for future releases. laboratory experiment purposes), transmitters T2-T5 are more Moreover, when transmitter and receiver are placed close to complex and consist of a vesicle containing an acidic solution each other, as e.g. in the synaptic cleft, recycling the signaling and employ different mechanisms for the controlled release particles aids in clearing the channel for future releases, and and/or reuptake of protons. Finally, we also briefly consider thereby, reducing ISI. Reversibility can be achieved by adding proton emitters (T6) which may naturally occur in the human a second biomolecule that transports the signaling particle back body. into the vesicle after signal detection. As inward transporter, T1 (Pipette): The simplest transmitter for protons is a pipette light-driven inward proton pumps can be additionally integrated by which an acidic solution is released dropwise into the into the vesicle [124]. These inward pumps are able to transport channel (Fig. 6, T1). This transmitter may be suitable for protons from the surrounding solution into the interior of the controlling the release of protons at macroscale. vesicle (Fig. 6, T5). In case that light-driven outward proton T2 (Degenerating vesicle): As second and more complex pumps are used for signaling particle release (Fig. 6, T4), it transmitter system for protons, we propose the usage of is important to choose outward and inward proton pumps that vesicles that contain an acidic solution (Fig. 6, T2). In the are activated by different wavelengths in order to avoid the simplest case, the release of the vesicle content could be simultaneous operation of both proton pumps. triggered by adding a detergent destroying the membrane Remark 3: In Fig. 6, light-driven pumps are labelled by hν or by means of electroporation [122] where the membrane because the energy of the photon used as a driving force may permeability is increased due to an externally applied electric be calculated as E = hν, where h denotes the Planck constant field. Both approaches are very effective; however, they also and ν is the frequency of the light [128]. In T5, the inward destroy the entire vesicle and thus release the entire content light driven pump is referred to as hν0, indicating the fact that at once. In order to transport only partial quantities of the a different activation frequency/wavelength is needed compared signaling particles from the interior of the vesicle to the to the outward pump. outside, different transporter proteins can be incorporated into T6 (Natural transmitters): Besides these designed syn- the vesicle membrane as will be explained in the following. thetic transmitter systems (T1-T5), there are also some pro- T3 (Ion channels): For the third transmitter model, we propose cesses in the human body that may be interpreted as proton to use ion channels (e.g. voltage-gated proton channels) for a emission. For example, some tissue has a higher proton controlled release of the signaling particles from the vesicle concentration (lower pH value) than other tissue (Fig. 6, T6). (Fig. 6, T3). Voltage-gated ion channels are transmembrane Lower pH values can be observed e.g. in inflamed tissue but proteins (Fig. 4a), which undergo conformational rearrange- also the extracellular pH of tumors can be heterogeneous and ments due to changes in the electrical membrane potential acidic [129], which may be used for detection of cancer cells. near the channel. Such changes of membrane potential can be induced artificially, e.g. by the voltage-clamp technique [72]. The channel opens through these conformational changes B. Receivers and ions can leave the vesicle. Voltage-gated proton channels Protons as signaling particles lead to an increase of the exhibit a high selectivity allowing only protons to leave the proton concentration at the receiver or equivalently a reduction IEEE COMMUNICATIONS SURVEYS & TUTORIALS 11

T1: Pipette with acid (protons ) T4: Vesicle with acid and light-driven proton pump (e.g. bacteriorhodopsin)

light-driven pump

hν

T2: Vesicle with acid, destruction of T5: Vesicle with acid and light-driven vesicle (detergents, electroporation) proton pump; reuptake of protons with an inward light-driven proton pump light-driven pump

hν

hν' ght-driv pump

T3: Vesicle with acid and voltage- T6: Tissue with decreased extracellular gated proton channel pH (e.g. tumor tissue, inflammation) tumor tissue voltage-gated hannel

Fig. 6. Transmitters for protons as signaling particles. The suggested building blocks for different transmitter systems are presented in order of increasing complexity and include designed artificial systems and physiological emitters, which may naturally occur in the human body. of the solution pH. We discuss five possible synthetic receiver different variants were developed, which possess an altered architectures (R1-R5) for protons and two natural processes in pH-sensitive excitation spectrum. For some of these variants, a the human body (R6, R7) that are triggered by changes in the response time of less than 20 ms could be demonstrated [130]. proton concentration. Besides GFP variants with disappearing fluorescence through R1 and R2 (pH sensors): The simplest approaches to detect decreasing pH values, also GFP variants which emit light of a change in the proton concentration of a solution are pH- different color at different pH values were reported [131]. For measuring instruments (Fig. 7, R1) and pH-sensitive dyes (pH example, for excitation at 388 nm, the GFP from the sea cactus indicators) (Fig. 7, R2) as they are routinely used in chemical Cavernularia obesa has a blue fluorescence at pH 5 and below, and biological laboratories. The changes of dye colors can for whereas it shows a green fluorescence at pH 7 and above example be detected by photometry in a second step. [132]. In case of long-term fluorescent measurements, the light- R3 (Fluorescence proteins): In biological experiments, pH- induced bleaching of the fluorophores has to be considered. sensitive fluorescence proteins are also often used as pH Photobleaching results in a decreasing signal that is independent indicators and they are good candidates for use in nanoscale from pH variations. Here, ratiometric pH-sensitive GFP variants, MCSs. One example for such a fluorescent protein is the green such as pHluorin2, have a big advantage since the pH is not fluorescent protein (GFP) (Fig. 7, R3) with its characteristic β- estimated from simple changes in the fluorescence level but barrel structure and the fluorescent chromophore (fluorophore) from the ratio of the fluorescence intensities at two different in the center (Fig. 4c). The fluorescence is dependent on the excitation wavelengths [133]. Thus, photobleaching will not protonation of the fluorophore and in the last years many affect the correctness of the pH measurement, but only the IEEE COMMUNICATIONS SURVEYS & TUTORIALS 12

Fig. 7. Receivers for protons as signaling particles. The suggested building blocks for different receiver systems are presented in order of increasing complexity and include designed artificial systems and physiological receivers, which naturally occur in the human body. Protons are indicated as blue dots. lifetime of the receiver. mophore, which is excited by irradiation with light of a certain R4 (FRET): Another possibility for a receiver system is the wavelength, transfers energy to an acceptor chromophore. The usage of biomolecules, which undergo large conformational shorter the distance between the two chromophores, the more changes upon a decrease in the surrounding pH value. One energy can be transferred from the donor to the acceptor example are peptides consisting of polyglutamate, which chromophore. Hence, FRET is very sensitive to small changes form α-helices in acidic solution while being disordered at in distance between the chromophores. This effect can be read intermediate or high pH (Fig. 7, R4) [134]. This conformational out by monitoring the ratio of the respective light intensities rearrangement of the structure can be visualized and made emitted by the donor and the acceptor at different wavelengths. detectable with a coupled FRET (Forster¨ resonance energy Alternatively, the intensity of the donor can be compared transfer or fluorescence resonance energy transfer) experiment. in the presence and absence of the acceptor [135].3 We The underlying mechanism is an energy transfer between two light-sensitive molecules (chromophores). A donor chro- 3We note that the underlying FRET mechanism as an option for communi- cation at nanoscale has been theoretically investigated in [136]–[140]. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 13 propose the construction of a FRET-based receiver employing blocks similar to those presented in Section II. Due to their a fusion protein with two FRET partners at the termini and a particular role as second messengers4 in the human body, we polyglutamate sequence in between as described in [134]. Upon illustrate in this section how the concepts proposed for protons helix formation in acidic pH, the distance between the two ends may be transferred to calcium ions (represented by the symbol decreases. In biological experiments, a common donor-acceptor Ca2+). Table II summarizes the proposed building blocks and pair is the cyan fluorescent protein (CFP) combined with the underlines the high similarity of the underlying components yellow fluorescent protein (YFP) [131], [141] which are both for both types of signaling particles. color variants of GFP (Fig. 4c). R5 (Proton-gated channels): As a further option for a pH- A. Transmitters dependent receiver system, we propose the usage of proton- gated channels, like acid-sensing ion channels (ASICs, Fig. 4d) Transmitter T1 (pipette) as well as the simple vesicle-based [142], [143] or proton-gated proton channels (e.g. the viral transmitter T2 (degenerating vesicle), which does not contain p7 protein) [144] embedded in a vesicle (Fig. 7, R5). These specific membrane proteins, may be adapted for calcium ions ligand-gated ion channels are permeable for certain types of simply by replacing the acid with a solution containing a cations after activation by high proton concentration at their calcium salt such as CaCl2. Compared to protons, there are some additional degenerating carriers triggered by light or extracellular side [142], [143]. The cations can flow via the 5 channel pore in the vesicle membrane into the interior of the temperature which may be used as an alternative . vesicle and increase the concentration of cations in the vesicle, Biological transmitters T3-T5 can be constructed by replac- which can be detected by measuring the transmembrane current ing the described proton channels and proton pumps in Fig. 6 by methods such as the two-electrode voltage clamp method by calcium-specific proteins with an analogous function. Some [73]. For reversibility, an additional light-driven outward pump suggestions for how this could be realized in detail are given (e.g. KR2 of the marine bacterium Krokinobacter eikastus below. [145] for sodium ions or bacteriorhodopsin for protons) can be T3 (Ion channels): As proposed for protons, the outward embedded into the vesicle membrane. With such light-driven transport of calcium ions may be accomplished with ion pumps, it would be possible to pump the cations from the channels as well. There exist some voltage-gated channels interior of the vesicle back into the surrounding solution. for calcium ions such as Cav1.1 [151] (Fig. 4a). Although a R6 (Proton-triggered protein dissociation): In nature, there stimulation by electricity would be convenient in testbeds and are numerous examples where a protein complex dissociates for applications outside the human body, the medical use of in response to decreasing pH values (Fig. 7, R6). One such MCSs may require a release of calcium ions by triggers which example is described for the periplasmic protein HdeA, which are less invasive and therefore more biocompatible. Fortunately, forms a well-folded homodimer at neutral pH. Under acidic for calcium ions, there are some additional gating mechanisms conditions, HdeA unfolds and exhibits an enhanced tendency to available compared to protons. dissociate into monomers [146], [147]. Another example is the One particular interesting example are ligand-gated channels. human Hsp47 protein, which is a collagen-specific molecular These ion channels become only permeable if a ligand binds chaperone and indispensable for molecular maturation of from the outside causing thereby a conformational change of collagen [148]. Hsp47 transiently binds to procollagen in the the protein. One possible candidate for calcium ion receptors is endoplasmic reticulum (ER, neutral pH) and dissociates from it the 5-HT-3A receptor [152] (Fig. 4e) which opens in response in a pH-dependent manner once this complex is transported to a to serotonin. So, serotonin could be administered in order to compartment with lower pH (e.g. the cis-Golgi or the ER-Golgi release calcium ions from the vesicle. Since serotonin is a NT intermediate compartment) [148]. These examples demonstrate (see Section IV) and thus a natural signaling particle of the human body, it might even be possible to couple the calcium that various physiological processes may be triggered by 6 modulating proton concentrations. release directly to the activation of a serotonergic neuron. This R7 (Proton-modulated “acidic-metabolic” vasodilatation): would allow to use the input from a nerve fiber for activation In the human body, decreasing pH values during hy- of the transmitter. This general principle can also be applied poxia/ischaemia can also lead to a physiological mechanism to other pairs of ligand-gated channels and their physiological known as “acidic-metabolic” vasodilatation [149] (Fig. 7, R7), ligand, of course. This allows for the direct coupling of a which improves blood flow and oxygen supply. Vasodilator biological process and a synthetic MCS. nitric oxide (NO) is generated through a non-enzymatic Another interesting option are mechanosensitive calcium − channels, which are opened in response to mechanical stress reduction of inorganic nitrite (NO2 ) to NO, a reaction that takes place predominantly during acidic/reducing conditions [149]. 4Intracellular chemical substance whose concentration changes upon a Thus, “acidic-metabolic” vasodilatation represents another primary signal, e.g. a ligand binding to a transmembrane receptor [150]. physiological process that may be modulated by proton-based 5One type of such carriers may be so-called light-sensitive caged compounds synthetic MCSs. [160], [161], which are already commercially available. Like in the vesicle, the calcium ions are shielded at the beginning and thus biologically inactive, which is due to a bound photoswitchable molecule. Upon irradiation with light of a certain wavelength, the shielding agent gets cleaved and the calcium ions III.CALCIUM IONSAS SIGNALING PARTICLES are released from their cage [160], [161]. The same general concept, but with heat instead of light as a trigger mechanism, could also be realized by the use Besides protons, other types of ions may be suitable for of thermosensitive microcapsules as described in [162]. synthetic MCSs as well. In many cases, there exist building 6Neuron/synapse which produces serotonin or uses serotonin as an NT. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 14

TABLE II TRANSFERABILITYOFTHEBUILDINGBLOCKSSUGGESTEDFORPROTONS (FIGS. 6 AND 7) TO MCSSUSING CALCIUM IONS AS SIGNALING PARTICLE. COMPARISONOFPOSSIBLETRANSMITTERSANDRECEIVERS.

Component Protons (H+) Calcium ions (Ca2+)

T1 (Pipette) Pipette with acidic solution Pipette with solution containing calcium salt (e.g. CaCl2) T2 (Degenerating carriers) Vesicle with acidic solution Vesicle with calcium ions, alternatively light-sensitive caged calcium or thermosensitive microcapsules T3 (Ion channels) Vesicle with acid and passive transport via voltage-gated Vesicle with calcium ions and passive transport via calcium proton channel [123] channel, different gating mechanisms: • Voltage-gated (e.g. Cav1.1 [151]) • Ligand-gated (e.g. 5-HT-3A + serotonin [152]) • Mechanosensitive (e.g. TRPV4 [153]) T4/T5 (Ion pumps) Vesicle with acid and active transport via proton pump, Vesicle with calcium ions and active transport via calcium different energy sources: pump, different energy sources: • Light-driven (e.g. bacteriorhodopsin [125]) • Light-driven (e.g. engineered pump from [154]) • ATP-driven (e.g. V-ATPases, P-ATPases [127]) • ATP-driven (e.g. Ca2+-ATPase [155])

R1 (Measuring instrument) pH-meter Conductivity measuring instrument R2 (Dye) pH-sensitive dye/pH indicator Calcium-sensitive fluorescent dye (e.g. Fura-2, Indol-1, Fluo-3, Fluo-4, Calcium Green-1 [156], [157]) R3 (Fluorescence proteins) GFP (different variants available) [130], [132] Fusion constructs of GFP variants and calmodulin (e.g. GCaMP [158]) R4 (FRET) α-helix stabilization due to protonation, fluorescence Calcium indicators containing troponin C and a FRET pair proteins at the termini forming a FRET pair [134] (e.g. TN-XL (CFP+YFP) [159])

[163]. One possible approach could be the insertion of the background concentration of other ions in the environment vesicles between the fibers of the extracellular matrix at some requires the detection of rather small changes of the total ionic location in the body. Upon mechanic shear or pressure on strength. the respective tissue, a calcium release would be triggered. R2 (Fluorescent dye): Similar to a pH indicator, a calcium- This principle could be useful in the context of targeted drug sensitive fluorescent dye could be used to detect changes in the delivery, e.g. in order to facilitate the local administration calcium ion concentration. When such a dye is illuminated with of an anesthetic. One example for such a mechanosensitive light of a certain wavelength, fluorescence occurs if calcium calcium channel involved in nociception, i.e., the encoding and ions are present. Examples for such dyes are Fura-2, Indol-1, processing of pain stimuli in the human body, is the transient Fluo-3, Fluo-4, and Calcium Green-1 [156], [157], which all receptor potential vanilloid 4 (TRPV4) [153] which is depicted have a high specificity for calcium ions in common. in Fig. 4f. R3 (Fluorescence proteins): For the third receiver structure, T4/T5 (Ion pumps): Besides passive channels, similar to we propose the use of calcium-sensitive fluorescent proteins. protons, calcium ions can actively be transported using either a They are generally fusion constructs of GFP (Fig. 4c) or one of light-driven pump, that has been artificially created [154], or an its variants, and the calcium-binding protein calmodulin. One ATP-driven Ca2+-ATPase [155]. If two light-driven pumps are example is GCaMP [158] which shows only feeble activity if intended to be used in the same vesicle to allow for transmitter calcium ions are absent, but undergoes a conformational change regeneration (reversibility), the second pump would need to be upon calcium binding leading to a pronounced fluorescence. engineered to work at a different wavelength as has already R4 (FRET): Alternatively, as described for protons, calcium been reported for some bacteriorhodopsin mutants [126]. ions may be detected via receivers relying on the FRET mechanism [159], [164]. There exist calcium ion indicators B. Receivers which are based on a FRET pair of two different fluorescent proteins, connected via the calcium-binding protein troponin Regarding possible receivers for calcium ions, building C. One example, TN-XL [159], consists of CFP and YFP. blocks which are similar to R1-R4 presented for protons can If no calcium ions are present, the protein has an extended be employed (see Table II for a comparison). conformation where the two fluorescent subunits are distant R1 (Conductivity measurement): In response to the release from each other. If CFP is activated by illumination with a of calcium ions, the conductivity of the medium surrounding certain wavelength, only cyan fluorescence occurs. As soon as a the transmitter would increase. Analogous to the usage of a pH- calcium ion binds to troponin C, the conformation of the fusion meter in case of protons, the easiest way to construct a receiver protein changes, such that the two fluorescent building blocks for calcium ions is thus an instrument, which can measure the get into mutual vicinity. Upon illumination and activation of conductivity of the solution in the channel. While this approach CFP, the energy is partly transferred to YFP via FRET so that might be suitable for testbeds and applications outside the yellow fluorescence can be observed as well. body, it is challenging for biological systems because the high IEEE COMMUNICATIONS SURVEYS & TUTORIALS 15

IV. NEUROTRANSMITTERSAS SIGNALING PARTICLES special proteins for the outward transport of the NTs maybe Another important class of signaling particles well suited for inserted into the vesicle membrane. One possible outward the design of synthetic MCSs are NTs such as acetylcholine, transporter are NT sodium symporters (Fig. 8, T4), which dopamine, and serotonin. In the human body, NTs are used are driven by a sodium concentration gradient [168]. The to transmit nerve signals (action potentials) at the chemical structure of sodium symporters for the target NT dopamine, synapses between two neurons (e.g. dopamine, serotonin) or i.e., dopamine transporters, is shown in Fig. 4g. In particular, from a neuron to a muscle fiber (acetylcholine) [84]. In this the underlying mechanism of sodium symporters is as follows: section, we describe some general principles and building When both a sodium ion and an NT bind to the transporter blocks for the design of synthetic transmitters and receivers for simultaneously, they are carried across the vesicle membrane by NTs. The general design strategies for the transmitter systems a conformational change of the transporter. This means that the are depicted in Fig. 8, those for the receiver systems are NT will be pumped outwards as soon as a sodium concentration presented in Fig. 9. These general design principles can be gradient from the inside to the outside is established. Such applied to all three NTs discussed in this survey. Thus, the most a sodium gradient may be realized by a light-driven sodium suitable NT can be selected depending on the requirements pump which transports sodium inside when it is illuminated imposed by the desired application. It is important to note that by light of a certain wavelength. Thus, the combination of an the design of a specific signaling pathway requires the choice NT-sodium-symporter and a light-driven sodium-pump can be of suitable protein components depending on the type of NT used for a finely controllable release of NTs from the vesicle. used. Candidate components for acetylcholine, dopamine, and T5 (Antiporters): An alternative strategy with the same level serotonin are summarized in Table III. of complexity and controllability as the previous transmitter structure (T4) is to use vesicles with NT antiporters instead of symporters [169] (Fig. 8, T5). In contrast to sodium symporters, A. Transmitters in this case, the driving force for transportation of NTs across In the following, we consider six different possible types of the vesicle is a proton gradient. In particular, if a proton transmitters for NTs. The first two transmitters (T1, T2) are binds to the antiporter from the outside of the vesicle and an simple and can be used as macroscale interfaces. Transmitters NT simultaneously from the inside, a conformational change T3-T5 are vesicle based and employ different biological occurs by which the proton is transported inside and the NT mechanisms for NT release. The final transmitter (T6) is the is transported outside. To avoid a constitutive NT release, the axon terminal of a nerve fiber. inside of the vesicle has to be more acidic than the surrounding T1 (Pipette): Analogous to cations (Fig. 6, Table II), the channel when the transmitter is inactive. Coupled with a simplest and least sophisticated transmitter is a pipette by light-driven proton pump such as bacteriorhodopsin, one can which a solution containing the NTs can be released dropwise then remove protons from the vesicle upon illumination with into the channel (Fig. 8, T1). a certain wavelength and thereby induce a proton gradient T2 (Caged compounds): As described for calcium ions, towards the inside which triggers a controlled release of the another interesting concept, which is based on shielding NTs. the NTs until they need to be released, are so called light- T6 (Natural transmitters): Besides the vesicle-based trans- sensitive caged compounds (Fig. 8, T2). In this approach, the mitters (T3-T5), a main advantage of NTs is the possibility to NTs are enclosed and thus inactivated by a photoswitchable directly use a physiological transmitter, i.e., the axon terminal molecule. Upon irradiation with light of a certain wavelength, (presynaptic part) of a nerve fiber in the human body (Fig. 8, the photoswitchable molecule is either cleaved or changes its T6). Upon excitation of a nerve fiber, an electrical signal (action conformation and as a consequence, the NTs are released potential) is generated, moves along the axon terminal, and from their cage. Caged compounds have been developed triggers the release of the NTs, which are stored in vesicles for acetylcholine, dopamine, serotonin, and many other NTs [169]. This natural biological transmitter is inexhaustible [165]–[167]. Some of them are already commercially available. because the vesicles are regenerated by the neuron. Possible Recently, caged serotonin has been suggested to be used for applications of such a direct interface to a neuron [7] are targeted drug delivery in the context of neurodegenerative the bridging of nerve lesions, the release of drugs in certain diseases [167]. Alternatively, the NTs could be shielded using conditions (e.g. an analgesic combined with a neuron involved thermosensitive microcapsules [162] which release their content in pain reception), and the movement of a prosthesis. upon an increase of temperature. T3 (Degenerating vesicles): Vesicles containing a solution of particular NTs may be used as transmitter (Fig. 8, T3). As a B. Receivers simple option, the vesicle can be destroyed to release its content In this subsection, we discuss two synthetic receivers for as described previously. However, this has the disadvantage NTs which employ NT receptors embedded into the membrane that only a one-time release is possible. of a vesicle. Such NT receptors exist in the human body, T4 (Symporters): If a vesicle-based approach is used as e.g. in postsynaptic cells, and there are different types of NT suggested in T3, a transmitter which pumps the NTs in a receptors. Here, we consider ligand-gated sodium channels more controlled manner from the inside of the vesicle into such as the serotonin 5-HT3A receptor (R1, Fig. 4e) and G- the channel may be a better option than simply destroying the protein coupled receptors (R2, Fig. 4h). In addition, we also vesicle and releasing its entire content at once. To this end, consider one natural receiver for NT which can serve as an IEEE COMMUNICATIONS SURVEYS & TUTORIALS 16

Fig. 8. Transmitters for NTs as signaling particles. The suggested building blocks for different transmitter systems are presented in order of increasing complexity and include designed artificial systems and physiological emitters, which naturally occur in the human body. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 17

R1: Vesicle with ligand-gated sodium R3: Postsynaptic terminal channels and measurement of the transmembrane current ligand-gated channel

time

A

+ light-driven pump for ion reversibility

light-driven hν pump

R2: Cell with G-protein coupled receptor (GPCR), detection with FRET-experiment

GPCR

G-protein

FRET assay

Fig. 9. Receivers for NTs as signaling particles. The suggested building blocks for different receiver systems are presented in order of increasing complexity and include designed artificial systems and physiological receivers, which naturally occur in the human body. Neurotransmitters and sodium ions are indicated as orange and green dots, respectively. interface for the control of biological systems (R3). transmissions to pump out the sodium ions so that the receiver R1 (Ligand-gated ion channels): For ligand-gated sodium is replenished. channels, upon binding of the target NT, a pore in the receptor is R2 (GPCRs): Besides ligand-gated ion channels, G-protein opened which allows sodium ions to pass through the membrane coupled receptors (GPCRs) are another class of receptor for [180]. Based on the concentration gradient, sodium ions move NTs [180]. The structure of one example GPCR, namely the from the outside to the inside of the cell, which leads to the M3 muscarinic receptor, is shown in Fig. 4h. When an NT formation of a transmembrane current that can be measured binds to a GPCR, this leads to a conformational change of by methods such as the two-electrode voltage clamp method the receptor. This conformational change is conveyed to the (Fig. 9, R1, upper panel) [73]. For proper functionality of the inside of the cell via an intracellular binding protein such as a vesicle, the concentration of sodium ions inside the vesicle G-protein or arrestin, where it may activate or inhibit a variety has to be lower than the concentration outside. To ensure that of second messenger molecules and thereby have an impact on after detection the required sodium ion gradient is restored to the cell metabolism [182] rendering it suitable for microscale enable future receptions, reversibility has to be integrated in the applications. However, a GPCR could also be used as reception vesicle. This can be accomplished by further integrating light- mechanism for a macroscale interface (Fig. 9, R2). Such a driven sodium-pumps into the membrane of the vesicle (Fig. 9, setup will most probably require an intact cell, which has such R1, lower panel) [181]. Then, the light-driven sodium pumps a receptor in its membrane, because it would be extremely can operate e.g. in the time interval between two consecutive difficult to synthetically reconstruct the corresponding complex IEEE COMMUNICATIONS SURVEYS & TUTORIALS 18

TABLE III COMPARISON OF DIFFERENT NTS AS SIGNALING PARTICLES.EXEMPLARY OPTIONS FOR DIFFERENT COMPONENTS.

Component Acetylcholine Dopamine Serotonin

T3-T5 Vesicle yes yes yes T4 Symporter — dopamine transporter (DAT) [114] serotonin transporter (SERT) [170] (sodium-driven) T5 Antiporter vesicular acetylcholine transporter vesicular monoamine transporter 2 vesicular monoamine transporter 2 (proton-driven) (VAChT) [171] (VMAT2) [172] (VMAT2) [172] T6 Physiological transmitter axon terminal at neuromuscular axon terminal in the central axon terminal in the CNS; junction [173] nervous system (CNS); regulation regulation of mood, emotion, of executive functions, motor memory processing, sleep, control, motivation, arousal, cognition [174] reinforcement, and reward [174]

R1 Ligand-gated channel nicotinic acetylcholine receptor — serotonin receptor subtype 5-HT3 (nAChR) [175] [176] R2 G-protein coupled muscarinic acetylcholine receptors dopamine receptors DRD1-DRD5 serotonin receptor subtypes receptor (GPCR) (mAChR) M1−5 [177] [178] 5-HT1,2,4−7 [179] R3 Physiological receiver trigger contraction of muscle fiber trigger nerve impulses in the CNS trigger nerve impulses in the CNS [173] [174] [174]

signaling cascade in a vesicle. There are several commercially but also plays an important role for the specificity of the available kits which use FRET experiments allowing optical signal transduction process. The proteins discussed as receivers detection of the level of GPCR activation. below generally do not only recognize the phosphoryl group R3 (Natural receivers): NTs can be used to directly interact itself, but also the physico-chemical properties of the peptide with biological systems. For example, if the NTs are released in in its vicinity. This allows the design of various types of proximity of a postsynaptic terminal or a muscle fiber, they may phosphopeptides with different signaling specificity. stimulate a nerve or induce a muscle contraction (Fig. 9, R3). This stimulation occurs by binding of an NT to a receptor, e.g. a A. Transmitter ligand-gated ion channel, on the membrane of the postsynaptic terminal. The resulting ion influx leads to a depolarization In contrast to the systems described in Sections II-IV, in of the cell membrane which propagates then as a new action the case of peptide modifications, the signaling particles do potential along the cell. This principle can be exploited in not need to be stored at the transmitter but can be generated medical applications of MC [7] for bridging of nerve lesions upon a stimulus (e.g. light or external ligand), see Fig. 10. The or targeted intervention into a deregulated neuronal circuitry, stimulation is provided by kinases that transfer a phosphoryl e.g. in the context of neurodegenerative diseases. group from the chemical energy carrier molecule ATP to a peptide, thereby creating a phosphopeptide, the signaling particle. In this process, ATP is hydrolyzed to adenosine V. PHOSPHOPEPTIDESAS SIGNALING PARTICLES diphosphate (ADP), which can be regenerated to ATP by other As outlined in Section I-C, protein modifications represent cellular processes. a widespread principle of signal transduction in nature. Phos- For phosphorylation, several different peptides and corre- phorylation, where a phosphoryl group is added to a peptide, sponding kinases are available. The human proteome contains represents one of the most frequent modifications in cellular at least 518 different protein kinases [183]. Most protein signaling, and will be considered in the following in more detail. kinases phosphorylate either the amino acids serine/threonine To exploit this principle for the design of MCSs, it appears or tyrosine (specific types of amino acids). However, there are advisable to reduce the size of the respective phosphoproteins also dual-specificity protein kinases that can phosphorylate to the vicinity of the phosphorylation sites. These smaller both serine/threonine and tyrosine residues. Within these ‘phosphopeptides’ have the advantage of faster diffusion due groups, kinases additionally differ in their specificity, i.e., to their smaller size compared to intact proteins. the amino acid sequence in the environment of the potential Phosphopeptides are complementary to the particles de- phosphorylation site can determine whether an amino acid scribed in the previous sections as they have very different becomes phosphorylated by a certain kinase or not. Due properties. The most important difference is that peptides do to the large variability of the peptide sequences with the not function in isolation, but require attachment to a chemical corresponding kinases, a large number of different signaling functional group. This step is mediated by a kinase, a specific particles is available. type of enzyme. It is important to note that the peptide unit An important prerequisite for the use of signaling particles represents more than a mere carrier molecule to transport in MCSs is the controllability of particle generation, i.e., in this the phosphoryl group from the transmitter to the receiver case, the ability to switch the kinases on and off. Because kinase IEEE COMMUNICATIONS SURVEYS & TUTORIALS 19

Fig. 10. Phosphopeptides as signaling particles. General composition of the communication system, which uses switchable kinases (activation by light, ligands, or pH change) as transmitters. The receiver consists of a phosphopeptide-binding domain, which allows detection of the binding process via changes in tryptophan fluorescence or in the mass of the complex. activity has profound effects on cellular processes, protein C. More Complex Architectures for MC kinases are generally highly regulated, i.e., there are many mechanisms for switching them on and off. An overview of Compared to cations and NTs, phosphopeptides have a physiological and engineered mechanisms for kinase regulation more sophisticated structure which provides more degrees of is given in Table IV. freedom for system design. In particular, by combining several kinases with receiver domains of corresponding specificity, various communication concepts can be realized. Here, we B. Receiver discuss orthogonal channels, diversity, coding, and jamming, For the detection of phosphorylated peptides, there exists see Fig. 11. a large set of protein domains in nature that may be used as P1 (Orthogonal channels): Orthogonal channels can be receivers in synthetic MCSs. The binding of the phosphopeptide realized by using two kinases, which differ in the type to such domains can for example be detected by a change in of their activation mechanism and the specificity of their tryptophan fluorescence that occurs upon binding (Fig. 10). phosphorylation (Fig. 11a). For example, transmitter 1 (T1) Alternatively, the change in mass upon peptide binding may could be a light-activated kinase and T2 a pH-activated kinase, be detected via surface plasmon resonance [194]. each combined with a specific receiver domain (R1 or R2). In Similar to the versatility on the transmitter side, there exist this system, changes in irradiation and pH can then be detected many different adapter domains that can be used as specific in the same setup based on the signals S1 and S2. This setup receivers. Serine/threonine phosphorylated peptides can be allows the interference free multiplexing of signals. recognized by a large number of different domain types, P2 (Diversity): By selecting suitable signal peptides and re- including 14-3-3, BRCT, FF, WW, and FHA domains [86]. ceiver domains, two signals cannot only be observed separately, Tyrosine phosphorylated peptides can be recognized by SH2 but can also be processed jointly to produce a combined output or PTB domains [195]. The SH2 domain family represents the signal. One setup for such a processing is shown in Fig. 11b. largest class of tyrosine phosphopeptide recognition modules Here, it is sufficient if one of the two stimuli is present to and is found in 111 different human proteins [196]. In trigger the signal at the receiver. The difference to the situation addition to the phosphorylated tyrosine residue (pTyr) itself, shown in Fig. 11a is that instead of two specific recognition these domains also recognize peptide residues adjacent to the domains, a receiver domain (R3) with low specificity is now phosphorylation site. For example, the SH2-domains of the SHP used which can bind the phosphorylation sites x and y of protein preferentially bind to a pTyr-X-X-Leu sequence stretch, both peptides. This can be interpreted as a form of diversity. i.e., they recognize a leucine (Leu), which is three amino acids For example, let’s assume that both peptides convey the same apart from the phosphorylation site (“X” denotes a variable information (e.g. both convey information bit “1”). If we further amino acid). In contrast, CRK SH2-domains recognize a pTyr- assume that diffusion is the main transportation mechanism X-X-Pro sequence, which contains a proline (Pro) instead to bring the phosphorylation sites of the peptides into contact of leucine at the respective sequence position [197]. This with the recognition domain at the receiver, then, due to the recognition of additional residues in the peptide ensures a random nature of the diffusion process, one of the peptides high specificity at the receiver side and underscores that the may not arrive at the receiver. Alternatively, one of the peptides peptide moiety of the signaling particle is more than a mere may not be phosphorylated at all, because the respective kinase carrier, but instead plays an important role for the construction was not activated by a stimulus. However, for the considered of specific transmitter-receiver pairs. setup, it is sufficient if one of the peptides carrying one of the IEEE COMMUNICATIONS SURVEYS & TUTORIALS 20

TABLE IV SUMMARYOFTHETYPEOFSTIMULITHATCANBEUSEDTOCONTROLTHEACTIVITYOFPROTEINKINASES.THETABLEDISTINGUISHESBETWEEN PHYSIOLOGICAL (p) PRINCIPLESOFACTIVATIONANDTHOSETHATWEREACHIEVEDBYMOLECULARENGINEERING (e). IN ADDITION TO THE TARGET KINASEANDTHETYPEOFSTIMULUS, ABRIEFDESCRIPTIONOFTHEUNDERLYINGMOLECULARMECHANISMISGIVEN.

Type of stimulus Origin Target kinase Molecular mechanism

Phosphorylation p Lck Lck contains two regulatory tyrosyl residues (Tyr394, Tyr505). Phosphorylation of these residues controls Lck activity in T cells [184].

Ubiquitination p receptor RTKs can become modified by ubiquitin, which causes their endocytosis from the tyrosine kinases plasma membrane and degradation [185]. (RTKs)

pH change p egg cortex This kinase shows significant changes of activity within the physiologically relevant pH tyrosine kinase range from 6.8 to 7.3 and may therefore be used as a pH sensitive transmitter [186].

Regulatory protein p cyclin The activity of CDKs is modulated by the interaction with specific cyclins that act as dependent regulatory partners [187]. kinases (CDKs)

Allosteric ligand e Fyn, Src, Lyn, Through insertion of a modified FK506 binding domain, these kinases were engineered Yes, PAK1 to allow activation by the allosteric ligand rapamycin [188], [189].

Photoresponsive lig- e Protein kinase C When exposed to light, a photoresponsive small molecule becomes an active inhibitor and (PKC) of PKC. This turning on of enzyme inhibition with light allows to control enzyme function [190].

Light e receptor RTKs were engineered to include light-oxygen-voltage (LOV)-sensing domains, tyrosine kinases resulting in kinases that can be activated by light [191]. (RTKs)

Light e Tropomyosin- Trk was engineered to include the photolyase homology region of cryptochrome 2 (a related kinase blue-light photoreceptor) resulting in a light-controllable kinase [192]. (Trk)

Light e Raf1, MEK1, A photodissociable dimeric protein (Dronpa) was engineered that dissociates in cyan MEK2, CDK5 light and re-associates in violet light. Insertion of Dronpa into protein kinases allowed to create photo-switchable kinases [193]. phosphorylation sites arrives at the receiver, which implies a such that no signal is detected. In a communication context, diversity gain. this may be interpreted as a jamming of the signal. In particular, P3 (Coding): For the architecture shown in Fig. 11c, both if the intended message is encoded via phosphorylation site x, stimuli (e.g. light and pH change) must be present so that a adding the second phosphorylation y jams the received signal. signal can be detected at the receiver. The carrier molecule An example of such a receiver in nature is a 14-3-3 protein that used is a peptide that has two distinct phosphorylation sites specifically recognizes a Cdc25B signal protein phosphorylated for kinases T1 and T2. This requires a receiver with two at the serine 323 position. If a second phosphorylation is added recognition sites for phosphoryl groups, each of which on its at the adjacent serine 321, the interaction with the receiver is own binds the phosphoryl groups too weakly to trigger the disrupted [199]. signal. The simultaneous binding of two phosphoryl groups The principles for switchable interactions described in Fig. 11 results in a significantly stronger binding, which triggers a are widely used in nature. The ELM.switches database [200] detectable signal at the receiver. This may be seen as a form provides an overview of known switchable systems, which of repetition coding as a signal is generated only if both might be exploited for MCS design. As a long-term goal, a phosphoryl groups are observed at the receiver. This principle quantitative understanding of these signaling processes may is used in nature, for example, by the ZAP70 adapter protein, guide the design of signaling particles that interfere with cellular which has two SH2 domains. The simultaneous binding of both signal transduction processes in a desired fashion, e.g. by SH2-domains causes a >100-fold increase in affinity compared counter-balancing impaired signaling originating from disease. to the interaction of a single SH2 domain [198]. P4 (Jamming): For the architecture shown in Fig. 11d, the Remark 4: We emphasize that cations and NTs also allow the transmitter and the signal peptides are similar to those for realization of some of the above complex architectures albeit the coding scheme shown in Fig. 11c. However, the receiver to a lesser degree. For instance, the number of NT-receptor (R4) has different properties compared to R3. If a second pairs that allow realization of orthogonal channels is much phosphorylation is added at position y, this leads to a weakening smaller compared to what can be realized by phosphopeptides. of the binding due to unfavorable interactions with the receiver, In fact, the sophisticated structure of phosphopeptides allows the system designer to engineer multiple types of signaling IEEE COMMUNICATIONS SURVEYS & TUTORIALS 21

a

x P x P T1 S1

T2 y P y S2

b P T1 x P R3 S T2 y P c P T1 x x P S

T2 y P y P

d P P T1 x x

y P y P R4 S

T2

Fig. 11. More complex architectures for MC based on phosphopeptides. (a) Simultaneous orthogonal transmission of two different signals (S1, S2). MCSs realizing (b) diversity, (c) coding, and (d) jamming. Transmitters (T) and receivers (R) of different types are labelled with small subscript numbers. x and y denote two distinct phosphorylation sites either in two different peptides (a, b) or within the same peptide (c, d). molecules whose release and detection can be controlled either implementation. For example, ions exhibit a high particle separately or jointly, depending on the desired application. stability and speed of diffusion, but different ions may generate Such high level of flexibility does not exist for cations and the same signal at the receiver. This may cause considerable NTs. interference in physiological environments where synthetic and natural MCSs employ different ions that interact with the VI.COMPARISON,APPLICATIONS, AND PRACTICAL same receiver. NTs provide a highly unique and specific signal, CONSIDERATIONS but the reversibility of their use still represents a bottleneck in the designs proposed here. In contrast to the other classes, In this section, we first compare the properties, advantages, phosphopeptide-based communication does not require vesicles and disadvantages of the signaling particle classes studied in and is highly versatile7, but the particle stability (susceptibility this paper. Subsequently, we present several medical applica- to proteases and phosphatases) may constitute a limitation. The tions of the proposed biological MCSs and corresponding de- properties of the considered signaling particles are summarized sign options for the transmitter, receiver, and signaling particles. in Table V and will be discussed more in detail in the following Furthermore, we discuss some practical communication-related with respect to their medical applications, options for mitigation considerations of the considered MCSs, namely ISI mitigation of ISI, and the speed of communication. and the speed of communication.

A. Comparison of the Considered Signaling Particles In Sections II-V, we have described three different classes 7Phosphopeptides are versatile signaling particles because they can be used of candidate signaling particles for synthetic MCSs. Each of with large variations in the setup, i.e., different transmitters (different kinases) these classes has certain advantages and disadvantages for and receivers (different recognition domains). IEEE COMMUNICATIONS SURVEYS & TUTORIALS 22

TABLE V PROPERTIES AND ADVANTAGES OF THE DIFFERENT CLASSES OF SIGNALING PARTICLES.

Properties Ions NTs Phosphopeptides

Transmitter control light, voltage, mechanical stress, indirect via ion channel (light, light, ligand ligand voltage) Membrane vesicles required yes yes no Regeneration possible yes, by returning signaling particles difficult: in the presented settings yes, kinase is not destroyed (ATP is (Reversibility) to the vesicles only with pipette or axon terminal regenerated in living cells) Energy source light by using light-driven pumps irreversible chemical (ATP) (if reversible) Receiver output signal optical (fluorescence), ion flow ion flow (conductivity), biological optical (fluorescence), surface (conductivity), ligand release from (GPCR signal) plasmon resonance protein Signal removal external addition of chemicals: base enzymatic degradation: enzymatic dephosphorylation with (OH−), chelation (EDTA) acetylcholinesterase (AChE), phosphatases monoamine oxidases (MAO) Underlying physiological intercellular intercellular intracellular communication type System complexity moderate high moderate Signaling particle stability very high high moderate Speed (diffusion) very fast fast moderate Uniqueness of signal moderate very good good Versatility high moderate very high

B. Potential Medical Applications transduction is interrupted. Depending on the type of nerve, this could result in a paralysis or a loss of sensory perception. In the following, we present some practical examples for One conceivable option to repair this damage is the insertion potential medical applications of the proposed building blocks of a bridging device in order to replace the damaged neuron of synthetic MCSs. The signaling particles presented in this (Fig. 12, lower panel). Such an artificial cell or micelle could article were chosen because changes in their concentration are be fully integrated within physiological signaling. Theoretically, either causative for diseases or because they may be used as it would consist of two MCSs, one between the healthy neuron important diagnostic markers to detect pathological conditions. on the left-hand side and the bridging device (MCS1) and Consequently, a combination of appropriate transmitters and one between the bridging device and the healthy neuron on receivers could allow for the construction of medical nanoma- the right-hand side (MCS2). In MCS1, the healthy neuron chines that would be able to recognize specific regions of the on the left side would serve as a physiological transmitter body or abnormal tissue such as inflammations or tumors due (building block NT, T6). As soon as a nerve signal occurs, to an altered concentration of certain signaling particles. An NTs are released within the synaptic cleft. On the surface of overview of possible medical applications covering a broad the bridging device, ligand-gated Na+ channels (building block range of diseases such as bacterial or viral infection, cancer, NT, R1) could be expressed, which upon binding of the NT inflammation, and neurodegeneration or circulatory diseases is would transport Na+ ions to the inside. This would lead to provided in Table VI. This table lists the respective signaling an enhanced concentration of Na+ inside the vesicle. Thereby, particles and proposes building blocks which could be used to MCS2 on the right-hand side of the bridging device would be design interfering nanomachines. activated. Na+-NT symporters (building block NT, T4) would Which combination of transmitter and receiver is most export NTs into the second synaptic cleft from where they appropriate critically depends on the application of interest, would be detected by the healthy neuron on the right-hand side of course. In the following, we focus on one of the medical (building block NT, R3). Consequently, the interruption within applications listed in Table VI, namely the design of a bridging the transmission of the nerve signal would be repaired. This device for a damaged neuron, cf. Fig. 12, in order to illustrate exemplary MCS may be further optimized: Reversibility could the design principles for choosing suitable biological building be ensured by adding transporters for a reuptake of the NT to blocks for the desired MCS. In the physiologic state (Fig. 12, the left-hand side of the second synaptic cleft. Such a bridging upper panel), a nerve signal is transduced consecutively device is well suited for in vivo application as it is based on a from one neuron to the next via the release of NTs at the biocompatible vesicle or cell and no external equipment such chemical synapse between adjacent neurons. If one of the as light sources or devices for the measurement of membrane neurons within this chain is damaged due to an injury or potentials are needed. a neurodegenerative disease (Fig. 12, middle panel), signal IEEE COMMUNICATIONS SURVEYS & TUTORIALS 23

TABLE VI EXAMPLESFORMEDICAL in vivo APPLICATIONS OF MCSS INCLUDING THE RESPECTIVE MEDICAL CATEGORY AND THE INVOLVED SIGNALING PARTICLES. BUILDINGBLOCKS, WHICHARESUITABLEFORTHEDESIGNOFTHECORRESPONDINGNANOMACHINES, MAYBEREALIZEDUSINGTHETRANSMITTERS ANDRECEIVERSPROPOSEDIN SECTIONS II TO V.

Medical Exemplary disease mechanism and signaling particle Proposed building blocks category

Bacterial Some bacterial proteins interfere with physiological signaling pathways The depleted kinases could be replaced by a synthetic infection based on phosphopeptides or phosphoproteins [201]. An example is transmitter in order to restore the physiological immune SpvC from Salmonella Typhimurium, which can inactivate MAP kinases response (e.g., transmitters in Fig. 10 and Table IV). that are involved in the activation of the immune response [202].

Viral infection Some viral proteins interfere with physiological signaling pathways One possibility to interfere with this pathological signaling based on signal peptides [203]. One example is HIV Vpu [204] which pathway would be to design an artificial receiver that binds contains phosphorylated residues that are detected by the receiver protein and sequesters Vpu in a more affine manner than h-βTrCP h-βTrCP within cells of the human immune system. This interaction (e.g., receiver in Fig. 10). leads to a degradation of the immune receptor CD4 which is required for pathogen recognition [205].

Cancer or Proton concentrations are elevated in inflamed tissue [206] or cancer MCSs with receivers for protons could be used to detect inflammation [207], [129] due to an enhanced/altered metabolism. affected tissue, which would be useful for targeted drug delivery (e.g., R5 in Fig. 7).

Depressive Dysregulations within the dopaminergic system resulting in a lack of Receivers which are activated by dopamine could be used to disorders or the NT dopamine play a role for the development of the hedonic detect dopaminergic synapses. Then, transmitters releasing Parkinson’s deficits of patients suffering from depression [208]. A selective loss dopamine could be used to potentiate the signal (e.g., T4 in disease of dopaminergic neurons within the Substantia nigra is causative for Fig. 8 and R1 in Fig. 9). Parkinson’s disease [85].

Dysregulated Many physiological signaling pathways are based on phosphopeptides Synthetic receivers for mutant c-Myc could be used in order signaling or phosphoproteins [209], [210]. Mutations of the respective phospho- to bind and thus inactivate the signaling pathway that leads to pathways rylation sites which occur in various diseases can thus have deleterious tumor formation (e.g., receiver in Fig. 10). effects [211]. One example are several mutations in the protein c-Myc that lead to the loss of a phosphorylation site. As a result, c-Myc is inefficiently degraded which may lead to tumor formation (Burkitt’s lymphoma) [212].

Hypertension One approach to treat hypertension is the administration of drugs which Transmitters for protons could be used to increase the proton lead to a widening of blood vessels and can thereby reduce the vascular concentration in peripheral blood vessels in order to treat resistance [213]. An increased proton concentration results in “acidic- hypertension (e.g., T3 in Fig. 6). metabolic” vasodilatation [149].

Nerve lesions NTs such as acetylcholine, dopamine, or serotonin are missing because The damaged nerve could be repaired using a bridging device the physiological transmitter, i.e., the releasing neuron is damaged. consisting of two MCSs, cf. Fig. 12 (e.g., T4 in Fig. 8 and R1 Therefore, the transmission of nerve signals is interrupted [214]. in Fig. 9).

Remark 5: We note that for the exemplary application partially synthetic MCS whereby the transmitter is a synthetic scenarios in Table VI, the end-to-end MCS may be in general nanomachine, the particles are the drug molecules, and the partially synthetic and partially natural. For example, in Fig. 12, receiver is the cancerous tissue. the receiver in MCS 1 is a synthetic cell whereas the transmitter is a natural cell (the nerve cell on the left-hand side in Fig. 12). C. ISI Mitigation via Signal Removal On the other hand, for MCS 2, the transmitter is a synthetic cell whereas the receiver is a natural cell (the nerve cell on the right- MCSs are impaired by ISI which is caused by the dispersive hand side in Fig. 12). More advanced applications of MCSs nature of the diffusion process. In particular, when the trans- may necessitate the use of both partially and fully synthetic end- mitter emits consecutive symbols, the signaling particles of to-end MCSs. For instance, in targeted drug delivery for cancer previously transmitted symbols may be observed at the receiver therapy, on the one hand, the drug-carrying nanomachines may in the current symbol interval. Since ISI is a common problem need to communicate with each other in order to collaboratively also in conventional wireline and wireless communication find cancerous tissue [38], [55]. Here, the end-to-end MCS is systems, many approaches for ISI mitigation exist [215]. synthetic, i.e., transmitter and receiver are nanomachines and In conventional communication systems, equalization at the the signaling particles have to be chosen by the system designer receiver is employed to combat ISI. Although, equalization by considering the specific characteristics of the application. has also been proposed as an option for MCSs [22], it may be On the other hand, after the targeted cancerous tissue has been challenging to implement even simple schemes, such as linear localized, the drug-carrying nanomachines that are close to the equalization, at nano-scale. Alternatively, in natural MCSs, ISI tissue have to release their payload drug for cancer therapy is usually mitigated by removing the signaling particles from [2], [39]. This communication system can be interpreted as a the channel. This biological approach also seems to be a viable option for ISI mitigation in synthetic MCSs. In the following, IEEE COMMUNICATIONS SURVEYS & TUTORIALS 24

Physiologic state

nerve signal

Pathologic state/Nerve lesion

nerve signal damaged neuron no signal

Bridging device MCS 1 MCS 2 NT, T6 NT, R1 NT, T4 NT, R3 nerve signal

ligand-gated Na+ channel Na+-NT symporter

Fig. 12. Bridging device for nerve lesions as a potential medical application for MCSs. The upper panel shows the physiological state with three intact neurons. The middle panel illustrates a nerve lesion in which the second neuron is damaged so that the signal is not transmitted. In the lower panel, a potential vesicle- or cell-based bridging device is shown that could be used to repair the damaged nerve. The building blocks used in the MCS are described in Figs. 8 and 9. we will discuss mechanisms for particle removal for each of the current symbol interval are neutralized and cannot the considered signaling particles. cause interference in future symbol intervals. Hence, ISI • Protons: If protons are used as signaling particles, then is mitigated. adding a basic solution capable of reversing the pH change • NTs: For NTs, in nature, signal removal is either achieved caused by the transmitter system is the simplest option by symporters, which pump the NTs into separated for removal of the signaling particles. Like the acidic compartments [168], or by enzymes, which degrade the solution (Fig. 6, T1), this basic solution can be released NTs [84]. The latter principle has the advantage that into the channel e.g. by an electrically controlled pipette it can be easier to implement in synthetic MCSs as after symbol detection. In a physiological environment, fewer biological building blocks are needed. A theoretical where such a pipette is not available, raising the pH value study of NT removal via enzyme has been provided can be achieved for example by decarboxylation reactions for acetylcholine in [220], which can analogically to of amino acids or urea, which consume protons [216]. In a physiological process at neuromuscular junctions be the presence of the enzyme urease, urea is cleaved into eliminated by the enzyme acetylcholinesterase (Fig. 4i). CO2 and NH3. The latter acts as a base and can thus In an analogous manner, dopamine and serotonin can be lead to a neutralization of the acidic liquid in the channel. degraded by monoamine oxidase-A. This cleavage of urea occurs also in nature, e.g. as a • Phosphopeptides: For modified peptides as signaling mechanism to remove acidic carbohydrate fermentation particles, the phosphorylation can be removed by phos- products in human saliva [217]. Moreover, pathogenic phatases to avoid interference with subsequent signals. Helicobacter pylori, which evokes gastritis and gastric One option for regulating such phosphatases are light- cancer in humans, uses urease to buffer the highly acidic controlled inhibitors, which are reversibly activated by ir- gastric liquid [218]. radiation with UV light [221]. Alternatively, phosphatases • Calcium ions: Calcium ions can be chemically shielded with pH-dependent activity were developed [222]. A with a chelating agent such as ethylenediaminetetraacetate modular protein assembly approach was used to design (EDTA) [219] or ethylene glycol-bis(β-aminoethyl ether)- a prototype that can detect tyrosine phosphorylation and N,N,N’ immediately activate phosphatase without requiring further ,N’-tetraacetic acid (EGTA). In particular, these substances external stimuli [223]. can bind to calcium ions and prevent them from fur- ther interaction with their environment. By releasing an D. Speed of Communication EDTA/EGTA solution into the channel as soon as a The time scales at which the different processes needed signal has been detected, the calcium ions released in for signal transmission occur at transmitter and receiver as IEEE COMMUNICATIONS SURVEYS & TUTORIALS 25 well as in the channel are crucial for MCS design as they membrane surface [224]. How many molecules can be ultimately limit the achievable data rate. The quantity, which is exported per time unit will critically depend on how easily the most straightforward to measure, is the speed of diffusion the corresponding membrane protein can be inserted into (Fig. 3). However, in order to estimate the speed of the entire a vesicle membrane and what density can be achieved. communication process, the speed of the proteins involved • Receptors: For GPCRs, ligand binding happens on a as well as the time for building up concentration gradients microsecond time scale [234] and the conformational sufficient for detection need to be taken into account as well. A rearrangements required for receptor activation take place comprehensive review of these processes is beyond the scope on a micro- to millisecond range [235]. However, current of this article. We will therefore only give a broad overview FRET experiments to readout the receptor activation at of the respective time scales; a comprehensive overview of the cellular level would then take multiple tens of seconds time ranges of biological processes with many useful model or even minutes [236]. calculations and the corresponding database are given in [224] • Enzymatic reactions: For some of our proposed building and [74], respectively. One approach that may be exploited blocks, such as the degradation of NTs and the addi- to speed up synthetic MCSs is flow. In particular, flow is tion and removal of peptide modifications, enzymes are also used in nature as a biological process that facilitates the required. The speed at which an enzyme catalyzes its transportation of chemical species across the human body. reaction is dependent on the concentration of the enzyme itself and the substrate (target molecule) the enzyme • Transport in channel: Individual molecular events, such binds to. The general form of an enzymatic reaction is as as an ion flowing through an ion channel or an enzyme follows: k catalyzing a reaction, are typically very fast. For example, 1 kcat E + S ES * E + P (1) one bacteriorhodopsin molecule can pump about 188 k−1 protons per second [74]. However, depending on the density of bacteriorhodopsin in the vesicle, the proton where enzyme E forms with its substrate S an enzyme- k gradient needed to activate the receiver, and the distance substrate complex ES with a certain rate constant 1. between transmitter and receiver, it may take considerably This complex can either dissociate, with the reaction rate k longer to achieve a desired macroscopic effect. Some constant −1, and form E and S, or react, with the reaction k model calculations for such macroscopic processes for rate constant cat, and form an E molecule and a product bacteriorhodopsin have been performed in [125]. In an ex- molecule P [237]. Note that the enzyme itself is not v perimental study, E. coli cells were used to synthetize with consumed in this process. The speed of this reaction blue-absorbing and green-absorbing proteorhodopsins, may be approximated (apart from additional factors such v which are two different types of proton pumps closely as cooperative binding that can affect ) by use of the related to bacteriorhodopsin. A change of 5 − 30 nM in Michaelis-Menten equations [237] as follows proton concentration of the cell suspension was achieved [S] [S] after about 60 s with an approximately linear evolution v = kcat · [E]t · = vmax · (2) KM + [S] KM + [S] over time [225]. • Ion channels: For ion channels, a good estimate for where square brackets denote the concentrations of the the number of ions that can be transported through a respective molecules, [E]t is the total concentration of 7 single channel is on average about 10 per second [74] the enzyme (free+substrate-bound), vmax is the maximum with a range from about 103 s−1 to 108 s−1. Some reaction speed at the given enzyme concentration, and exemplary permeability rates (number of ions that flow KM is the Michaelis constant. Thus, knowing kcat, KM , through the channel in a certain time span) are summarized and the concentrations of enzyme and substrate, the speed in Table VII. This compilation includes a subset of of the reaction can be calculated. In particular, in (2), kcat the channels suggested in this article for which such stands, in simple terms, for the number of reactions an information was available in the literature. Although the enzyme makes per unit time and KM for the substrate channels are rather fast on the level of a single molecule, concentration at which 50 % of the maximum reaction the time needed to detect an effect at the cellular level speed is reached [224]. −2 −1 (for example in Xenopus oocytes) is in the range of a Remark 6: Typical values of kcat vary between 10 s few seconds [226]. However, if artificial vesicles with and 107 s−1 with the median approximately at 14 s−1. ion channels are created, the required time span will be Moreover, the range of the values of KM varies approxi- highly dependent on the density of the ion channels in mately from 10−5 mM up to 103 mM with a median value the membrane. of 0.130 mM [224], [246]. Please note that these values • Transporters: Transporters are in most cases much slower are not only enzyme-specific, but they also depend on the than ion channels with a typical transport rate of ≈ 100 respective substrate and environmental conditions such as molecules s−1 [224]. Nevertheless, if a cell has e.g. pH or ionic strength. For illustration, we provide concrete 10,000 transporters of a certain type in its membrane, values of kcat and KM for some selected exemplary 106 molecules can be transported in total per second. For enzyme-substrate combinations in Table VIII. Specific glucose transporters in certain cell types, it has been values for many more enzymes and substrates are available estimated that they make up about 2 % of the total in the BioNumbers data bank [74]. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 26

TABLE VII PERMEABILITYRATESFORSELECTEDIONCHANNELS.CHANNEL 1 IS AN EXAMPLE FOR A VOLTAGE-GATED PROTON CHANNEL (PROTONS, T3), CHANNEL 2 IS A VOLTAGE-GATED CALCIUM CHANNEL (CALCIUMIONS, T3). CHANNEL 3 CANBEUSEDASARECEIVERFORACETYLCHOLINE (NTS, R1). CHANNELS 4, 5, 6, AND 7 AREADDITIONALEXAMPLESDEMONSTRATINGTHETYPICALRANGEOFIONTRANSPORTRATES.

Ion channel Permeability rate [s−1] References

3 1 Voltage-gated proton channel Hv1 6.0 · 10 [227], [228] 6 2 Voltage-gated calcium channels (e.g. Cav1.1) ≈ 10 [81] 3 Nicotinic acetylcholine receptor 2.5 · 107 [229] 4 Potassium channel KcsA ≈ 108 [230]

7 5 Voltage-gated sodium channel Nav1.4 ≈ 10 [231] 6 Cation channel gramicidin 1.5 · 107 [232] 7 Calcium channel CRAC 1.1 · 104 [233]

TABLE VIII KINETICCONSTANTSFORSELECTEDENZYME-SUBSTRATE COMBINATIONS.ENZYMES 1 AND 2 CANBEUSEDFOR NT DEGRADATION, ENZYME 3 ISAN EXEMPLARY ENZYME FOR PROTEIN PHOSPHORYLATION AND ENZYME 4 FOR PROTEIN DEPHOSPHORYLATION (THEIR INDICATED KINETIC VALUES WERE DETERMINEDFORASAMPLEPEPTIDE.).ENZYMES 5, 6, AND 7 ARE ADDITIONAL EXAMPLES DEMONSTRATING THE ACTIVITY RANGE OF DIFFERENT ENZYMES.

−1 Enzyme Substrate kcat [s ] KM [mM] References 1 Acetylcholinesterase acetylcholine 3.0 · 104 0.1 [238], [239]

Monoamine oxidase-A dopamine 1.83 0.23 2 [240] (MAO-A) serotonin 1.80 0.40 3 Protein kinase N1 peptide 0.7 0.013 [241] 4 Protein phosphatase 2Cα peptide 0.3 0.09 [242]

7 5 Catalase H2O2 4.0 · 10 1.6 [239], [243] 6 Triosephosphate isomerase glyceraldehyde 4.2 · 103 0.5 [239], [244] 3-phosphate 7 Chymotrypsin N-acetyl-Val-OMe 1.7 · 10−1 88 [239], [245]

VII.FUTURE RESEARCH DIRECTIONS in this paper, communication-theoretical models have not been In this section, we outline some future research directions developed by the MC community, yet. In the following, we including open research problems that should be tackled in highlight some of the related open research problems: Stage 2 of the roadmap in Fig. 2 by theoreticians for modeling • Ion pumps vs. ion channels: While ion channels allow and designing the biological building blocks proposed in this passive diffusion of the signaling particles across the paper as well as the challenges that should be addressed by membrane and rely on a concentration gradient, ion pumps experimentalists for implementation of these building blocks enable the transport of the signaling particles across the in Stage 4 of the roadmap. A summary of the proposed future membrane even against a concentration gradient at the research directions are provided in Fig. 13. cost of energy. Communication-theoretical channel models have been developed for ion pumps and ion channels A. Directions and Challenges for Theoretical Research in [125] and [247], respectively. However, these results The design of reliable and efficient synthetic MCSs crucially are preliminary and contain no comparative study of ion depends on the accuracy and simplicity of the corresponding pumps and ion channels. To quantitatively compare these transmitter, channel, and receiver models. Simplicity is desir- options, the impact of the energy consumption of ion able since complicated models typically do not allow for the pumps and the concentration gradient of ion channels derivation of insightful design guidelines which are required have to be modeled and analyzed. This has not been fully for efficient system design while a certain level of accuracy addressed in the existing MC literature, yet. is also needed to ensure the relevance of the results and the • Reuptake: The particle reuptake mechanism enables corresponding design. Although there exists a rich literature in a reduction of ISI and the possibility of “harvesting” biology that analyzes the proposed biological building blocks signaling particles for re-transmission. However, several for MCSs, the corresponding models are too complex to be aspects affect the performance of the reuptake process readily applicable to communication system design. In fact, including the position of the reuptake pumps on the cell for many of the transmitter and receiver architectures proposed surface, their number/density, and the required energy IEEE COMMUNICATIONS SURVEYS & TUTORIALS 27

Future Research Directions

Theoretical Research Experimental Research

Recombinant Membrane Ion Pumps vs. Reuptake Phosphopeptides as Protein Expression Protein Polymersomes Ion Channels Process Signaling Particles Engineering of Proteins Reconstitution

End-to-end Impact of Membrane Optical Artificial Vesicles Channel Model Interfaces Permeability Stimulation vs. Entire Cells

Fig. 13. Summary of potential future research directions.

consumption. Related preliminary results by the MC inside the receiver if the concentration of sodium ions community can be found in [248]–[250], where the authors outside the receiver is low. proposed the transmission of signaling molecules by one • Impact of interfaces: We presented various micro-to- node and harvesting them by inward reuptake pumps by macroscale interfaces to facilitate the experimental eval- the same or another node. uation of the proposed biological building blocks for • Phosphopeptides: As discussed in Section V-C, phospho- MCSs. For the experimental verification in Stage 4 of the peptides are quite flexible which allows the construction proposed development roadmap in Fig. 2, the relative of complex system architectures. In addition, unlike other impact of the microscale MCSs and the interface on types of signaling particles, peptide molecules can be the received signal have to be carefully investigated. For always present in the channel and become phosphorylated instance, the impairments (noises and nonlinearities) that to create the signaling molecules, namely phosphopeptides, diffusion, reactions, and interference from natural cells upon an external stimulus (e.g. light or external ligand). cause to the received signal are different from those that This interesting class of signaling particles has not been a light source, a pH meter, or a current meter cause. investigated in the MC literature, yet. Hence, there exist These two different categories of impairments have to be many open research problems ranging from the modeling separately and quantitatively analyzed. of the synthesis and reception of phosphopeptides to the design of new network architectures addressing the needs B. Implementation Challenges of different applications. • End-to-end channel model: In communication theory, In order to realize the biological building blocks proposed we are often interested in an end-to-end channel model in this article, some technical challenges have to be met, which which relates the received signal to the transmit signal. In are discussed in the following. the context of MCs, the number of signaling molecules • Recombinant expression of proteins: To generate suffi- released by the transmitter is typically considered as the cient amounts of the required proteins (e.g. those shown in transmit signal and the number of signaling molecules Fig. 4), they have to be expressed recombinantly in cells counted at the receiver constitute the received signal. from either prokaryotic (e.g. E. coli) or eukaryotic origin However, the release and reception mechanisms in MCSs (e.g. yeast, insect or mammalian cells). Therefore, a DNA may include multiple stages which should be carefully coding for the respective protein has to be inserted, such taken into account for end-to-end channel modeling. For that the native cellular machinery of protein biosynthesis instance, for the NT receiver R1 in Fig. 9, the NTs can be utilized [251]. The fact that crystal structures for activate ligand-gated sodium channels which allow the most of the proteins shown in Fig. 4 exist indicates that sodium ions to enter the receiving cell and potentially protocols for the recombinant expression and subsequent trigger another action. Therefore, the number of sodium purification of larger amounts of these proteins are already ions entering the receiver should be considered as the available. actual received signal rather than the number of activated • Additional engineering: The vast majority of the proteins sodium channels. Although these two quantities are related, proposed as biological transmitter and receiver compo- the latter accounts for the available concentration of nents were directly adapted from functional biological sodium ions outside the receiver, which may considerably systems and do not require further modification, i.e., they affect the performance. For example, although signaling can be readily obtained by recombinant expression or by particles (i.e., NTs) may activate many ligand-gated using them directly in the context of the native biological sodium channels, a strong signal cannot be generated system. However, protein engineering [252], [253] offers the possibility to modify protein function in order to make IEEE COMMUNICATIONS SURVEYS & TUTORIALS 28

them more suitable for application in MCSs. The simplest substances over the compartment boundaries and, thus, the strategy is the introduction of point mutations, i.e., the background noise in MCS is much lower. The excellent exchange of single amino acids within the protein. This stability and encapsulant retention are also reasons why strategy has for example been applied to engineer a light- polymersomes have attracted significant attention as driven calcium pump [154], bacteriorhodopsin mutants vehicles for (targeted) drug delivery [262]. In addition, the with activity at different wavelengths [126], and GFP chemical versatility of the amphiphilic polymer monomers variants with different pH optima [130]. This strategy from which the polymersomes are formed enables the may also be used to design light-driven calcium pumps membrane properties to be tailored to the intended applica- which operate at different wavelengths (prerequisite for tion. For some polymers, such as the ABA-type triblock co- reversibility, Table II, T5). Such components have not yet polymer poly(2-methyloxazoline)-poly(dimethylsiloxane)- been reported, but can presumably be constructed in a poly(2-methyloxazoline) (PMOXA-PDMS-PMOXA) it similar manner as the already available bacteriorhodopsin has been shown that they do not cause cytotoxic effects variants (Fig. 6, T5) by introducing point mutations. or inflammatory responses in biological systems [263], Besides point mutations, another strategy for protein [264]. PMOXA-PDMS-PMOXA also allows the functional engineering is the fusion to unrelated protein domains of integration of diverse transmembrane proteins, as reviewed synergistic function8, which can be considered as part of a recently [265]. With regard to the communication speed, modular toolbox [253]. For instance, a fusion construct of the number of proteins per vesicle is decisive. In poly- GFP (Fig. 4c) and the calcium-binding protein calmodulin mersomes with a mean diameter of 100 nm, a maximum has led to the development of the calcium sensor GCaMP number of 120 transmembrane proteins per vesicle has which shows a pronounced fluorescence only upon the been inserted so far [266]. This number corresponds to a presence of calcium ions [158]. Moreover, light-inducible coverage of 3% of the available surface area. However, protein kinases have been developed using a similar due to the rapid advances in the field of preparative approach [191]–[193]. Another interesting application purification of membrane proteins and their functional would be a fusion to domains that allow for a fixation reconstitution in artificial membranes, it is expected that at certain carrier materials (e.g. magnetic nanoparticles) this value will increase in the future [267]. A systematic [254]–[256] to facilitate the design of MCSs with a defined study of membrane protein reconstitution in liposomes geometry. and polymersomes revealed that similar amounts of Additional engineering will also be required to implement proteins can be incorporated in both types of vesicle, some of the more complex systems proposed in this survey. with a tendency to slightly higher values in liposomes. For example, membrane proteins have to be forced into Interestingly, the specific activity of reconstituted proton the correct orientation when they are inserted into vesicles pumps was higher in polymer membranes, which points (see below). to the possibility that the same biological effect can be • Membrane protein reconstitution: Some of the pro- achieved with a smaller number of membrane proteins posed systems require the reconstitution of membrane [268]. proteins into vesicles, which is a challenging task. How- • Membrane permeability: The permeability coefficients ever, the artificial construction of such proteoliposomes of lipid membranes for charged molecules are generally has been widely reported [117]–[121], [257], and due low. As a result, pure lipid membranes are often referred to to the fast progress in the field of synthetic biology, as “impermeable” for ions if there is no protein-mediated there is a continuous development of corresponding new transport. However, whether or not a certain membrane methods [257]. As vesicles containing only physiological can be considered (im)permeable also depends on the types of lipids are sometimes not sufficiently stable, so- considered time frame, since any molecule will permeate called polymersomes which have membranes consisting at some point due to passive diffusion if a concentration of non-lipid polymers, or hybrid vesicles containing both gradient across the membrane is applied. This leads to polymers and lipids may be used as an alternative [117]. an uncontrolled mass transport over the compartment • Polymersomes: The significantly higher mechanical sta- boundaries of vesicles, which increases the background bility is not the only advantage that polymersomes have noise in MCS. This aspect is of special importance for over liposomes in the context of MC. The higher stability proton-based transmitters involving lipid vesicles (T2-T5, goes hand in hand with a lower intrinsic permeability, Fig. 6) since the permeability coefficients of protons (10−4 since both aspects correlate with membrane thickness cm s−1) are several orders of magnitude higher than the [258]. While lipid membranes are usually 3 – 5 nm thick corresponding values for other ions, such as calcium (2.5 [259], polymersomes with up to 40 nm thick membranes x 10−11 cm s−1) [269]. As a consequence, the intrinsic have been reported [260]. This leads to permeability membrane permeability may not be negligible for certain coefficients that are several orders of magnitude lower than MCSs. the corresponding values determined for liposomes [261]. • Optical stimulation: Some of the suggested building As a consequence, the uncontrolled passive diffusion of blocks require external optical stimulation. Although it would be preferable to use internal stimuli for in vivo 8Cooperative function of two or more agents which allows for faster applications because they are less invasive, light is still achievement of a common goal. considered as a unique stimulus since it is easy to control IEEE COMMUNICATIONS SURVEYS & TUTORIALS 29

and does not interfere with other physiological stimuli. APPENDIX A For research purposes, it is already nowadays the state GLOSSARY OF BIOLOGICAL CONCEPTSAND TERMS of the art to use optical systems in animals (e.g, living For convenience, in Table IX, we explain the most important rats) in order to stimulate light-dependent proteins [270]. biological concepts and terms appearing in this article. For instance, optogenetics employs light to stimulate genetically engineered neurons, which express light-driven REFERENCES ion channels [271]–[274]. A further miniaturization of the respective equipment could eventually allow for a medical [1] H. Hess, G. D. Bachand, and V. Vogel, “Powering nanodevices with biomolecular motors,” Chemistry – A European Journal, vol. 10, no. 9, use in humans. The example of deep brain stimulation, pp. 2110–2116, Apr. 2004. which is widely applied for therapy of Parkinson’s disease, [2] W. Gao and J. Wang, “Synthetic micro/nanomotors in drug delivery,” dystonia, and obsessive compulsive disorders [275], shows Nanoscale, vol. 6, pp. 10 486–10 494, Jul. 2014. [3] D. Peer, J. M. Karp, S. Hong, O. C. Farokhzad, R. Margalit, and that such interfaces between technical devices and the R. Langer, “Nanocarriers as an emerging platform for cancer therapy,” human body are feasible. Nature Nanotechnology, vol. 2, no. 12, pp. 751–760, Dec. 2007. • Artificial vesicles vs. entire cells: Instead of employing [4] Y. Moritani, S. Hiyama, and T. Suda, “Molecular communication for health care applications,” in Proc. Fourth Annual IEEE International artificial vesicles, most of the presented building blocks Conference on Pervasive Computing and Communications Workshops, could be realized using entire cells in which the respective Mar. 2006, pp. 5–pp. membrane proteins are recombinantly expressed. While [5] I. F. Akyildiz, F. Brunetti, and C. Blazquez,´ “Nanonetworks: A new communication paradigm,” Computer Networks, vol. 52, no. 12, pp. this may be easier from a technical point of view and 2260–2279, 2008. facilitate the implementation of reversibility, it would also [6] F. Dressler and O. B. Akan, “A survey on bio-inspired networking,” require higher concentrations of signaling particles due to Computer Networks, vol. 54, no. 6, pp. 881–900, 2010. [7] T. Nakano, M. J. Moore, F. Wei, A. V. Vasilakos, and J. Shuai, “Molec- the much bigger size of the transmitter/receiver systems. ular communication and networking: Opportunities and challenges,” Moreover, it would most certainly lead to a high level of IEEE Transactions on NanoBioscience, vol. 11, no. 2, pp. 135–148, noise because of the high complexity and the plethora of Jun. 2012. [8] I. F. Akyildiz, M. Pierobon, S. Balasubramaniam, and Y. Koucheryavy, natural transmembrane proteins and signaling cascades “The Internet of Bio-Nano Things,” IEEE Communications Magazine, present in cells which could interfere with the desired vol. 53, no. 3, pp. 32–40, Mar. 2015. signaling processes. Finally, in the context of applications [9] N. A. Abbasi and O. B. Akan, “An information theoretical analysis of human insulin-glucose system toward the Internet of Bio-Nano Things,” within the human body (e.g. targeted drug delivery), the IEEE Transactions on NanoBioscience, vol. 16, no. 8, pp. 783–791, usage of entire cells may be problematic because of the Dec. 2017. high amount of surface proteins which could trigger an [10] V. Jamali, A. Ahmadzadeh, W. Wicke, A. Noel, and R. Schober, “Channel modeling for diffusive molecular communication—A tutorial immune reaction. review,” Proceedings of the IEEE, vol. 107, no. 7, pp. 1256–1301, Jul. 2019. [11] T. Nakano, A. Eckford, , and T. Haraguchi, Molecular Communication. VIII.CONCLUSIONS Cambridge, U.K.: Cambridge Univ. Press, 2013. [12] N. Farsad, N.-R. Kim, A. W. Eckford, and C.-B. Chae, “Channel and In this paper, we provided a comprehensive survey of the noise models for nonlinear molecular communication systems,” IEEE biological components that can serve as building blocks for Journal on Selected Areas in Communications, vol. 32, no. 12, pp. 2392–2401, Dec. 2014. the transmitter and the receiver and as signaling particles [13] H. B. Yilmaz and C. B. Chae, “Arrival modelling for molecular for synthetic MCSs. Adopting a signaling particle centric communication via diffusion,” IEEE Electronics Letters, vol. 50, no. 23, presentation, we argue that cations, NTs, and phosphopeptides pp. 1667–1669, Nov. 2014. [14] A. Ahmadzadeh, H. Arjmandi, A. Burkovski, and R. Schober, “Com- represent promising candidate signaling particles. They can be prehensive reactive receiver modeling for diffusive molecular commu- used to interact with natural physiological processes and as nication systems: Reversible binding, molecule degradation, and finite information carriers in synthetic MCS employing engineered number of receptors,” IEEE Transactions on NanoBioscience, vol. 15, no. 7, pp. 713–727, Oct. 2016. protein systems as transmitters and receivers. The engineered [15] A. O. Bicen, I. F. Akyildiz, S. Balasubramaniam, and Y. Koucheryavy, transmitter/receiver systems considered in this survey mainly “Linear channel modeling and error analysis for intra/inter-cellular Ca2+ rely on physiological protein components of well-characterized molecular communication,” IEEE Transactions on NanoBioscience, vol. 15, no. 5, pp. 488–498, Jul. 2016. functionality. However, the engineering of functional trans- [16] M. Damrath, S. Korte, and P. A. Hoeher, “Equivalent discrete-time mitter/receiver systems based on these isolated components channel modeling for molecular communication with emphasize on an still remains a challenging task, particularly when insertion of absorbing receiver,” IEEE Transactions on NanoBioscience, vol. 16, no. 1, pp. 60–68, Jan. 2017. proteins into vesicle membranes with defined inside-outside [17] V. Jamali, N. Farsad, R. Schober, and A. Goldsmith, “Diffusive geometry is required. For each of the presented synthetic molecular communications with reactive molecules: Channel modeling MCSs architectures, the ensuing advantages and limitations are and signal design,” IEEE Transactions on Molecular, Biological and Multi-Scale Communications, vol. 4, no. 3, pp. 171–188, Sep. 2018. outlined and references to the relevant literature in synthetic [18] P. Shih, C. Lee, P. Yeh, and K. Chen, “Channel codes for reliability biology are provided. This survey will help both theoreticians enhancement in molecular communication,” IEEE Journal on Selected and experimentalists to develop a better understanding of Areas in Communications, vol. 31, no. 12, pp. 857–867, Dec. 2013. [19] V. Jamali, A. Ahmadzadeh, N. Farsad, and R. Schober, “Constant- the options available for the design and implementation of composition codes for maximum likelihood detection without CSI in biological synthetic MCSs and interfaces for natural MCSs. diffusive molecular communications,” IEEE Transactions on Communi- Some of the related open research problems for communication- cations, vol. 66, no. 5, pp. 1981–1995, May 2018. [20] M. C. Gursoy, E. Basar, A. E. Pusane, and T. Tugcu, “Index modulation theoretical modeling/design and technical implementation of for molecular communication via diffusion systems,” IEEE Transactions the proposed MCS building blocks were outlined. on Communications, vol. 67, no. 5, pp. 3337–3350, May 2019. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 30

[21] M. S. Kuran, H. B. Yilmaz, T. Tugcu, and I. F. Akyildiz, “Modulation Selected Areas in Communications, vol. 34, no. 3, pp. 600–614, Mar. techniques for communication via diffusion in nanonetworks,” in Proc. 2016. IEEE International Conference on Communications (ICC), Jun. 2011, [43] H. Unterweger, J. Kirchner, W. Wicket, A. Ahmadzadeh, D. Ahmed, pp. 1–5. V. Jamali, C. Alexiou, G. Fischer, and R. Schober, “Experimental [22] D. Kilinc and O. B. Akan, “Receiver design for molecular communi- molecular communication testbed based on magnetic nanoparticles in cation,” IEEE Journal on Selected Areas in Communications, vol. 31, duct flow,” in Proc. 2018 IEEE 19th International Workshop on Signal no. 12, pp. 705–714, Dec. 2013. Processing Advances in Wireless Communications (SPAWC), Jun. 2018, [23] A. Noel, K. Cheung, and R. Schober, “Optimal receiver design for pp. 1–5. diffusive molecular communication with flow and additive noise,” IEEE [44] N. Schlechtweg, S. Meyer, H. Unterweger, M. Bartunik, D. Ahmed, Transactions on NanoBioscience, vol. 13, no. 3, pp. 350–362, Sep. W. Wicke, V. Jamali, C. Alexiou, G. Fischer, R. Weigel, R. Schober, 2014. and J. Kirchner, “Magnetic steering of superparamagnetic nanoparticles [24] C. T. Chou, “Maximum a-posteriori decoding for diffusion-based in duct flow for molecular communication: A feasibility study,” in EAI molecular communication using analog filters,” IEEE Transactions International Conference on Body Area Networks. Springer, 2019, pp. on Nanotechnology, vol. 14, no. 6, pp. 1054–1067, Nov. 2015. 161–174. [25] L. Lin, C. Yang, M. Ma, and S. Ma, “Diffusion-based clock synchroniza- [45] B. Krishnaswamy, C. M. Austin, J. P. Bardill, D. Russakow, G. L. tion for molecular communication under inverse Gaussian distribution,” Holst, B. K. Hammer, C. R. Forest, and R. Sivakumar, “Time-elapse IEEE Sensors Journal, vol. 15, no. 9, pp. 4866–4874, Sep. 2015. communication: Bacterial communication on a microfluidic chip,” IEEE [26] V. Jamali, A. Ahmadzadeh, and R. Schober, “Symbol synchronization Transactions on Communications, vol. 61, no. 12, pp. 5139–5151, Dec. for diffusion-based molecular communications,” IEEE Transactions on 2013. NanoBioscience, vol. 16, no. 8, pp. 873–887, Dec. 2017. [46] T. Nakano, S. Kobayashi, T. Suda, Y. Okaie, Y. Hiraoka, and [27] V. Jamali, A. Ahmadzadeh, C. Jardin, C. Sticht, and R. Schober, T. Haraguchi, “Externally controllable molecular communication,” IEEE “Channel estimation for diffusive molecular communications,” IEEE Journal on Selected Areas in Communications, vol. 32, no. 12, pp. 2417– Transactions on Communications, vol. 64, no. 10, pp. 4238–4252, Oct. 2431, Dec. 2014. 2016. [47] L. Felicetti, M. Femminella, G. Reali, P. Gresele, M. Malvestiti, and [28] A. Noel, K. C. Cheung, and R. Schober, “Joint channel parameter J. N. Daigle, “Modeling CD40-based molecular communications in estimation via diffusive molecular communication,” IEEE Transactions blood vessels,” IEEE Transactions on NanoBioscience, vol. 13, no. 3, on Molecular, Biological and Multi-Scale Communications, vol. 1, no. 1, pp. 230–243, Sep. 2014. pp. 4–17, Mar. 2015. [48] D. P. Martins, K. Leetanasaksakul, M. T. Barros, A. Thamchaipenet, [29] H. Awan and C. T. Chou, “Molecular communications with molec- W. Donnelly, and S. Balasubramaniam, “Molecular communications ular circuit-based transmitters and receivers,” IEEE Transactions on pulse-based jamming model for bacterial biofilm suppression,” IEEE NanoBioscience, vol. 18, no. 2, pp. 146–155, Apr. 2019. Transactions on NanoBioscience, vol. 17, no. 4, pp. 533–542, Oct. 2018. [30] C. T. Chou, “Designing molecular circuits for approximate maximum [49] T. Nakano, Y. Okaie, S. Kobayashi, T. Koujin, C. Chan, Y. Hsu, a posteriori demodulation of concentration modulated signals,” IEEE T. Obuchi, T. Hara, Y. Hiraoka, and T. Haraguchi, “Performance Transactions on Communications, vol. 67, no. 8, pp. 5458–5473, Aug. evaluation of leader–follower-based mobile molecular communication 2019. networks for target detection applications,” IEEE Transactions on [31] A. Marcone, M. Pierobon, and M. Magarini, “Parity-check coding based Communications, vol. 65, no. 2, pp. 663–676, Feb. 2017. on genetic circuits for engineered molecular communication between [50] K. Darchini and A. S. Alfa, “Molecular communication via microtubules biological cells,” IEEE Transactions on Communications, vol. 66, no. 12, and physical contact in nanonetworks: A survey,” Nano Communication pp. 6221–6236, Dec. 2018. Networks, vol. 4, no. 2, pp. 73–85, 2013. [32] C. Madsen, C. J. Myers, T. Patterson, N. Roehner, J. T. Stevens, and [51] T. Nakano, T. Suda, Y. Okaie, M. J. Moore, and A. V. Vasilakos, C. Winstead, “Design and test of genetic circuits using ibiosimibiosim,” “Molecular communication among biological nanomachines: A layered IEEE Design Test of Computers, vol. 29, no. 3, pp. 32–39, June 2012. architecture and research issues,” IEEE Transactions on NanoBioscience, [33] N. Ramakrishnan, U. S. Bhalla, and J. J. Tyson, “Computing with vol. 13, no. 3, pp. 169–197, Sep. 2014. proteins,” Computer, vol. 42, no. 1, pp. 47–56, Jan. 2009. [52] N. Farsad, H. B. Yilmaz, A. Eckford, C.-B. Chae, and W. Guo, “A [34] T. Nakano, Y.-H. Hsu, W. C. Tang, T. Suda, D. Lin, T. Koujin, comprehensive survey of recent advancements in molecular communi- T. Haraguchi, and Y. Hiraoka, “Microplatform for intercellular com- cation,” IEEE Communications Surveys & Tutorials, vol. 18, no. 3, pp. munication,” in Proc. IEEE Nano/Micro Engineered and Molecular 1887–1919, thirdquarter 2016. Systems, Jan. 2008, pp. 476–479. [35] A. O. Bicen, C. M. Austin, I. F. Akyildiz, and C. R. Forest, “Efficient [53] L. Felicetti, M. Femminella, G. Reali, and P. Lio,` “Applications of sampling of bacterial signal transduction for detection of pulse-amplitude molecular communications to medicine: A survey,” Nano Communica- modulated molecular signals,” IEEE Transactions on Biomedical tion Networks, vol. 7, pp. 27–45, 2016. Circuits and Systems, vol. 9, no. 4, pp. 505–517, Aug. 2015. [54] Y. Chahibi, “Molecular communication for drug delivery systems: A [36] L. Grebenstein, J. Kirchner, R. S. Peixoto, W. Zimmermann, F. Irn- survey,” Nano Communication Networks, vol. 11, pp. 90–102, 2017. storfer, W. Wicke, A. Ahmadzadeh, V. Jamali, G. Fischer, R. Weigel, [55] U. A. K. Chude-Okonkwo, R. Malekian, B. T. Maharaj, and A. V. A. Burkovski, and R. Schober, “Biological optical-to-chemical signal Vasilakos, “Molecular communication and nanonetwork for targeted conversion interface: A small-scale modulator for molecular commu- drug delivery: A survey,” IEEE Communications Surveys & Tutorials, nications,” IEEE Transactions on NanoBioscience, vol. 18, no. 1, pp. vol. 19, no. 4, pp. 3046–3096, Fourthquarter 2017. 31–42, Jan 2019. [56] J. Wang, B. Yin, and M. Peng, “Diffusion based molecular communi- [37] N. B. Carvalho and D. Schreurs, Microwave and Wireless Measurement cation: Principle, key technologies, and challenges,” China Communi- Techniques. Cambridge University Press, 2013. cations, vol. 14, no. 2, pp. 1–18, Feb. 2017. [38] R. Mosayebi, A. Ahmadzadeh, W. Wicke, V. Jamali, R. Schober, and [57] H. Chen, W. Guo, J. Shen, L. Wang, and J. Song, “Structural principles M. Nasiri-Kenari, “Early cancer detection in blood vessels using mobile analysis of host-pathogen protein-protein interactions: A structural nanosensors,” IEEE Transactions on NanoBioscience, vol. 18, no. 2, bioinformatics survey,” IEEE Access, vol. 6, pp. 11 760–11 771, 2018. pp. 103–116, Apr. 2019. [58] I. F. Akyildiz, M. Pierobon, and S. Balasubramaniam, “An information [39] T. N. Cao, A. Ahmadzadeh, V. Jamali, W. Wicke, P. L. Yeoh, J. Evans, theoretic framework to analyze molecular communication systems based and R. Schober, “Diffusive mobile MC with absorbing receivers: on statistical mechanics,” Proceedings of the IEEE, vol. 107, no. 7, pp. Stochastic analysis and applications,” IEEE Transactions on Molecular, 1230–1255, Jul. 2019. Biological and Multi-Scale Communications, vol. 5, no. 2, pp. 84–99, [59] C. Rose, I. S. Mian, and M. Ozmen, “Capacity bounds on point-to-point Nov. 2019. communication using molecules,” Proceedings of the IEEE, vol. 107, [40] N. Farsad, W. Guo, and A. W. Eckford, “Tabletop molecular communi- no. 7, pp. 1342–1355, Jul. 2019. cation: Text messages through chemical signals,” PLOS ONE, vol. 8, [60] M. Kuscu, E. Dinc, B. A. Bilgin, H. Ramezani, and O. B. Akan, no. 12, p. e82935, Dec. 2013. “Transmitter and receiver architectures for molecular communications: [41] N. Farsad, D. Pan, and A. Goldsmith, “A novel experimental platform A survey on physical design with modulation, coding, and detection for in-vessel multi-chemical molecular communications,” in Proc. IEEE techniques,” Proceedings of the IEEE, vol. 107, no. 7, pp. 1302–1341, Global Communications Conference, Dec. 2017, pp. 1–6. Jul. 2019. [42] B.-H. Koo, C. Lee, H. B. Yilmaz, N. Farsad, A. Eckford, and C.-B. [61] T. Nakano, Y. Okaie, S. Kobayashi, T. Hara, Y. Hiraoka, and Chae, “Molecular MIMO: From theory to prototype,” IEEE Journal on T. Haraguchi, “Methods and applications of mobile molecular commu- IEEE COMMUNICATIONS SURVEYS & TUTORIALS 31

nication,” Proceedings of the IEEE, vol. 107, no. 7, pp. 1442–1456, [89] G. Gerhardt and R. N. Adams, “Determination of diffusion coefficients Jul. 2019. by flow injection analysis,” Analytical Chemistry, vol. 54, no. 14, pp. [62] T. Nguyen, T. S. Jones, P. Fontanarrosa, J. V. Mante, Z. Zundel, 2618–2620, 1982. D. Densmore, and C. J. Myers, “Design of asynchronous genetic circuits,” [90] E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Green- Proceedings of the IEEE, vol. 107, no. 7, pp. 1356–1368, Jul. 2019. blatt, E. C. Meng, and T. E. Ferrin, “UCSF Chimera–a visualization [63] C. McBride, R. Shah, and D. Del Vecchio, “The effect of loads in system for exploratory research and analysis.” Journal of Computational molecular communications,” Proceedings of the IEEE, vol. 107, no. 7, Chemistry, vol. 25, no. 13, pp. 1605–1612, Oct. 2004. pp. 1369–1386, Jul. 2019. [91] A. W. Norman and G. Litwack, Hormones. Academic Press, 2014. [64] M. Veletic´ and I. Balasingham, “Synaptic communication engineering [92] G. Wtzany and M. Nowacki, Biocommunication of Ciliates. Springer, for future cognitive brain–machine interfaces,” Proceedings of the IEEE, 2016. vol. 107, no. 7, pp. 1425–1441, Jul. 2019. [93] M. B. Miller and B. L. Bassler, “Quorum sensing in bacteria,” Annual [65] E. Kim, J. Li, M. Kang, D. L. Kelly, S. Chen, A. Napolitano, L. Panzella, Review of Microbiology, vol. 55, no. 1, pp. 165–199, Oct. 2001. X. Shi, K. Yan, S. Wu, J. Shen, W. E. Bentley, and G. F. Payne, “Redox [94] T. Nakano, T. Suda, M. Moore, R. Egashira, A. Enomoto, and K. Arima, is a global biodevice information processing modality,” Proceedings of “Molecular communication for nanomachines using intercellular calcium the IEEE, vol. 107, no. 7, pp. 1402–1424, Jul. 2019. signaling,” in Proc. 5th IEEE Conference on Nanotechnology, 2005., [66] J. M. Jornet, Y. Bae, C. R. Handelmann, B. Decker, A. Balcerak, Jul. 2005, pp. 478–481 vol. 2. A. Sangwan, P. Miao, A. Desai, L. Feng, E. K. Stachowiak, and M. K. [95] M. T. Barros, S. Balasubramaniam, and B. Jennings, “Error control for Stachowiak, “Optogenomic interfaces: Bridging biological networks calcium signaling based molecular communication,” in Proc. Asilomar with the electronic digital world,” Proceedings of the IEEE, vol. 107, Conference on Signals, Systems and Computers, Nov. 2013, pp. 1056– no. 7, pp. 1387–1401, Jul. 2019. 1060. [67] Y. A. Qadri, A. Nauman, Y. B. Zikria, A. V. Vasilakos, and S. W. Kim, “The future of healthcare internet of things: A survey of emerging [96] M. T. Barros, S. Balasubramaniam, B. Jennings, and Y. Koucheryavy, technologies,” IEEE Communications Surveys & Tutorials, pp. 1–1, “Transmission protocols for calcium-signaling-based molecular com- 2020. munications in deformable cellular tissue,” IEEE Transactions on [68] M. Rosoff, Ed., Vesicles. Marcel Dekker, 1996. Nanotechnology, vol. 13, no. 4, pp. 779–788, Jul. 2014. [69] J. Zheng and M. C. Trudeau, Eds., Handbook of Ion Channels. CRC [97] A. O. Bicen, I. F. Akyildiz, S. Balasubramaniam, and Y. Koucheryavy, 2+ Press, 2016. “Linear channel modeling and error analysis for intra/inter-cellular Ca [70] M. G. Wolfersberger, “Uniporters, symporters and antiporters,” Journal molecular communication,” IEEE Transactions on NanoBioscience, of Experimental Biology, vol. 196, pp. 5–6, 1994. vol. 15, no. 5, pp. 488–498, Jul. 2016. [71] D. C. Gadsby, “Ion channels versus ion pumps: The principal difference, [98] M. T. Barros, S. Balasubramaniam, and B. Jennings, “Comparative end- in principle,” Nature Reviews Molecular Cell Biology, vol. 10, no. 5, to-end analysis of calsium-signaling-based molecular communication pp. 344–352, Apr. 2009. in biological tissues,” IEEE Transactions on Communications, vol. 63, [72] H. N. Rubaiy, “A short guide to electrophysiology and ion channels,” no. 12, pp. 5128–5142, Dec. 2015. Journal of Pharmacy and Pharmaceutical Sciences, vol. 20, p. 48, Mar. [99] ——, “Using information metrics and molecular communication to de- 2017. tect cellular tissue deformation,” IEEE Transactions on NanoBioscience, [73] B. Guan, X. Chen, and H. Zhang, “Two-electrode voltage clamp,” Ion vol. 13, no. 3, pp. 278–288, Sep. 2014. Channels. Methods in Molecular Biology (Methods and Protocols), pp. [100] T. Nakano and J. Liu, “Design and analysis of molecular relay 79–89, 2013. channels: An information theoretic approach,” IEEE Transactions on [74] R. Milo, P. Jorgensen, U. Moran, G. Weber, and M. Springer, NanoBioscience, vol. 9, no. 3, pp. 213–221, Sep. 2010. “Bionumbers-the database of key numbers in molecular and cell biology,” [101] B. Maham, “A communication theoretic analysis of synaptic channels Nucleic Acids Research, vol. 38, no. suppl 1, pp. D750–D753, 2009. under axonal noise,” IEEE Communications Letters, vol. 19, no. 11, pp. [75] B. S. Donahue and R. Abercrombie, “Free diffusion coefficient of ionic 1901–1904, Nov. 2015. calcium in cytoplasm,” Cell Calcium, vol. 8, no. 6, pp. 437–448, 1987. [102] M. Veletic, P. A. Floor, Z. Babic, and I. Balasingham, “Peer-to-peer [76] C. De Grotthuss, “Memoire´ sur la decomposition´ de l’eau et des corps communication in neuronal nano-network,” IEEE Transactions on qu’elle tient en dissolution a` l’aide de l’electricit´ e´ galvanique,” 1805. Communications, vol. 64, no. 3, pp. 1153–1166, Mar. 2016. [77] S. Cukierman, “Et tu, Grotthuss! and other unfinished stories,” Biochim- [103] P. He, Y. Mao, Q. Liu, and H. Zhai, “A diffusion-neuron hybrid channel ica et Biophysica Acta (BBA) - Bioenergetics, vol. 1757, no. 8, pp. for molecular communication,” in Proc. International Symposium on 876–885, Aug. 2006. Intelligent Signal Processing and Communication Systems (ISPACS), [78] J. B. Matthew, F. R. Gurd, B. E. Garcia-Moreno, M. A. Flanagan, K. L. Nov. 2017, pp. 855–860. March, and S. J. Shire, “pH-dependent processes in proteins,” Critical [104] D. Malak and O. B. Akan, “Communication theoretical understanding Reviews in Biochemistry, vol. 18, no. 2, pp. 91–197, 1985. of intra-body nervous nanonetworks,” IEEE Communications Magazine, [79] D. Riccardi, “Calcium ions as extracellular, first messengers,” Zeitschrift vol. 52, no. 4, pp. 129–135, Apr. 2014. fur¨ Kardiologie, vol. 89, no. 2, pp. 9–14, 2000. [105] H. Ramezani and O. B. Akan, “A communication theoretical modeling [80] E. M. Brown, P. M. Vassilev, and S. C. Hebert, “Calcium ions as of axonal propagation in hippocampal pyramidal neurons,” IEEE extracellular messengers,” Cell, vol. 83, no. 5, pp. 679–682, 1995. Transactions on NanoBioscience, vol. 16, no. 4, pp. 248–256, Jun. [81] D. E. Clapham, “Calcium signaling,” Cell, vol. 131, no. 6, pp. 1047– 2017. 1058, Dec. 2007. [106] H. Ramezani, T. Khan, and O. B. Akan, “Sum rate of MISO neuro- [82] P. Pinton, C. Giorgi, R. Siviero, E. Zecchini, and R. Rizzuto, “Calcium spike communication channel with constant spiking threshold,” IEEE and apoptosis: ER-mitochondria Ca2+ transfer in the control of Transactions on NanoBioscience, vol. 17, no. 3, pp. 342–351, Jul. 2018. apoptosis,” Oncogene, vol. 27, no. 50, pp. 6407–6418, Oct. 2008. [83] J. C. Calderon,´ P. Bolanos,˜ and C. Caputo, “The excitation–contraction [107] J. Wu, Z. Yan, Z. Li, X. Qian, S. Lu, M. Dong, Q. Zhou, and N. Yan, coupling mechanism in skeletal muscle,” Biophysical Reviews, vol. 6, “Structure of the voltage-gated calcium channel Cav 1.1 at 3.6 A no. 1, pp. 133–160, Jan. 2014. resolution,” Nature, vol. 537, no. 7619, p. 191, 2016. [84] L. Wecker, S. Watts, C. Faingold, G. Dunaway, and L. Crespo, Brody’s [108] B. Schobert, J. Cupp-Vickery, V. Hornak, S. O. Smith, and J. K. Lanyi, Human Pharmacology: Molecular to Clinical, 5th ed. Elsevier, 2010. “Crystallographic structure of the K intermediate of bacteriorhodopsin: [85] R. Dhall and D. L. Kreitzman, “Advances in levodopa therapy for Conservation of free energy after photoisomerization of the retinal,” Parkinson disease,” Neurology, vol. 86, no. 14 Supplement 1, pp. S13– Journal of Molecular Biology, vol. 321, no. 4, pp. 715 – 726, 2002. S24, Apr. 2016. [109] F. Yang, L. G. Moss, and G. N. Phillips Jr, “The molecular structure [86] B. T. Seet, I. Dikic, M.-M. Zhou, and T. Pawson, “Reading protein of green fluorescent protein,” Nature Biotechnology, vol. 14, no. 10, p. modifications with interaction domains,” Nature Reviews Molecular 1246, 1996. Cell Biology, vol. 7, no. 7, p. 473, 2006. [110] I. Baconguis and E. Gouaux, “Structural plasticity and dynamic [87] F. Ardito, M. Giuliani, D. Perrone, G. Troiano, and L. L. Muzio, “The selectivity of acid-sensing ion channel – Spider toxin complexes,” crucial role of protein phosphorylation in cell signaling and its use as Nature, vol. 489, no. 7416, pp. 400–405, Jul. 2012. targeted therapy (review),” International Journal of Molecular Medicine, [111] G. Hassaine, C. Deluz, L. Grasso, R. Wyss, M. B. Tol, R. Hovius, vol. 40, no. 2, pp. 271–280, Jun. 2017. A. Graff, H. Stahlberg, T. Tomizaki, A. Desmyter et al., “X-ray structure [88] C. J. Miller and B. E. Turk, “Homing in: Mechanisms of substrate of the mouse serotonin 5-HT 3 receptor,” Nature, vol. 512, no. 7514, p. targeting by protein kinases,” Trends in Biochemical Sciences, 2018. 276, 2014. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 32

[112] Z. Deng, N. Paknejad, G. Maksaev, M. Sala-Rabanal, C. G. Nichols, [132] K. Ogoh, T. Kinebuchi, M. Murai, T. Takahashi, Y. Ohmiya, and R. K. Hite, and P. Yuan, “Cryo-EM and X-ray structures of TRPV4 H. Suzuki, “Dual-color-emitting green fluorescent protein from the sea reveal insight into ion permeation and gating mechanisms,” Nature cactus cavernularia obesa and its use as a pH indicator for fluorescence Structural and Molecular Biology, vol. 25, no. 3, p. 252, 2018. microscopy,” Luminescence, vol. 28, no. 4, pp. 582–591, Mar. 2013. [113] K. H. Wang, A. Penmatsa, and E. Gouaux, “Neurotransmitter and [133] M. J. Mahon, “Phluorin2: An enhanced, ratiometric, ph-sensitive green psychostimulant recognition by the dopamine transporter,” Nature, vol. fluorescent protein,” Advances in Bioscience and Biotechnology, vol. 2, 521, no. 7552, p. 322, 2015. no. 3, p. 132, 2011. [114] A. C. Kruse, J. Hu, A. C. Pan, D. H. Arlow, D. M. Rosenbaum, [134] J. M. Finke, P. A. Jennings, J. C. Lee, J. N. Onuchic, and J. R. Winkler, E. Rosemond, H. F. Green, T. Liu, P. S. Chae, R. O. Dror, D. E. Shaw, “Equilibrium unfolding of the poly(glutamic acid)20 helix,” Biopolymers, W. I. Weis, J. Wess, and B. K. Kobilka, “Structure and dynamics of vol. 86, no. 3, pp. 193–211, 2007. the M3 muscarinic acetylcholine receptor,” Nature, vol. 482, no. 7386, [135] R. M. Clegg, “Forster resonance energy transfer – FRET what is it, pp. 552–556, Feb. 2012. why do it, and how it’s done,” in FRET and FLIM techniques, T. W. J. [115] F. Nachon, E. Carletti, C. Ronco, M. Trovaslet, Y. Nicolet, L. Jean, and Gadella, Ed. Elsevier, 2009, pp. 1–58. P.-Y. Renard, “Crystal structures of human cholinesterases in complex [136] M. Kuscu and O. B. Akan, “A communication theoretical analysis of with huprine W and tacrine: elements of specificity for anti-Alzheimer’s FRET-based mobile ad hoc molecular nanonetworks,” IEEE Transac- drugs targeting acetyl-and butyryl-cholinesterase,” Biochemical Journal, tions on NanoBioscience, vol. 13, no. 3, pp. 255–266, Sep. 2014. vol. 453, no. 3, pp. 393–399, 2013. [137] ——, “A physical channel model and analysis for nanoscale molecular [116] J. F. Ohren, H. Chen, A. Pavlovsky, C. Whitehead, E. Zhang, P. Kuffa, communications with forster¨ resonance energy transfer (FRET),” IEEE C. Yan, P. McConnell, C. Spessard, C. Banotai, W. T. Mueller, Transactions on Nanotechnology, vol. 11, no. 1, pp. 200–207, Jan. 2012. A. Delaney, C. Omer, J. Sebolt-Leopold, D. T. Dudley, I. K. Leung, [138] ——, “FRET-based nanoscale point-to-point and broadcast communi- C. Flamme, J. Warmus, M. Kaufman, S. Barrett, H. Tecle, and C. A. cations with multi-exciton transmission and channel routing,” IEEE Hasemann, “Structures of human MAP kinase kinase 1 (MEK1) Transactions on NanoBioscience, vol. 13, no. 3, pp. 315–326, Sep. and MEK2 describe novel noncompetitive kinase inhibition,” Nature 2014. Structural and Molecular Biology, vol. 11, no. 12, pp. 1192–1197, Nov. [139] K. Wojcik, K. Solarczyk, and P. Kulakowski, “Measurements on MIMO- 2004. FRET nano-networks based on alexa fluor dyes,” IEEE Transactions [117] P. A. Beales, S. Khan, S. P. Muench, and L. J. Jeuken, “Durable vesicles on Nanotechnology, vol. 14, no. 3, pp. 531–539, May 2015. for reconstitution of membrane proteins in biotechnology,” Biochemical [140] K. Solarczyk, K. Wojcik, and P. Kulakowski, “Nanocommunication via Society Transactions, vol. 45, no. 1, pp. 15–26, Feb. 2017. FRET with dylight dyes using multiple donors and acceptors,” IEEE [118] M. Garni, S. Thamboo, C.-A. Schoenenberger, and C. G. Palivan, “Bio- Transactions on NanoBioscience, vol. 15, no. 3, pp. 275–283, Apr. pores/membrane proteins in synthetic polymer membranes,” Biochimica 2016. et Biophysica Acta (BBA) - Biomembranes, vol. 1859, no. 4, pp. 619– [141] B. Bajar, E. Wang, S. Zhang, M. Lin, and J. Chu, “A guide to fluorescent 638, Apr. 2017. protein FRET pairs,” Sensors, vol. 16, no. 9, p. 1488, Sep. 2016. [119] R. Skrzypek, S. Iqbal, and R. Callaghan, “Methods of reconstitution to [142] S. Grunder¨ and M. Pusch, “Biophysical properties of acid-sensing ion investigate membrane protein function,” Methods, Feb. 2018. channels (ASICs),” Neuropharmacology, vol. 94, pp. 9–18, Jul. 2015. [120] A. Kuhn, M. Haase, and S. Leptihn, “Assisted and unassisted protein [143] I. Hanukoglu, “ASIC and ENaC type sodium channels: Conformational insertion into liposomes,” Biophysical Journal, vol. 113, no. 6, pp. states and the structures of the ion selectivity filters,” The FEBS Journal, 1187–1193, Sep. 2017. vol. 284, no. 4, pp. 525–545, Sep. 2016. [121] I. L. Jorgensen, G. C. Kemmer, and T. G. Pomorski, “Membrane [144] U. Breitinger, N. S. Farag, N. K. Ali, and H.-G. A. Breitinger, “Patch- protein reconstitution into giant unilamellar vesicles: A review on clamp study of hepatitis C p7 channels reveals genotype-specific current techniques,” European Biophysics Journal, vol. 46, no. 2, pp. sensitivity to inhibitors,” Biophysical Journal, vol. 110, no. 11, pp. 103–119, Jul. 2016. 2419–2429, Jun. 2016. [122] J. C. Weaver and Y. Chizmadzhev, “Theory of electroporation: A review,” [145] K. Inoue, H. Ono, R. Abe-Yoshizumi, S. Yoshizawa, H. Ito, K. Kogure, Bioelectrochemistry and Bioenergetics, vol. 41, no. 2, pp. 135–160, Dec. and H. Kandori, “A light-driven sodium ion pump in marine bacteria,” 1996. Nature Communications, vol. 4, p. 1678, Apr. 2013. [123] T. E. DeCoursey, “Voltage and pH sensing by the voltage-gated proton [146] L. Salmon, F. Stull, S. Sayle, C. Cato, S¸. Akgul,¨ L. Foit, L. S. Ahlstrom, channel, hv1,” Journal of The Royal Society Interface, vol. 15, no. 141, E. Z. Eisenmesser, H. M. Al-Hashimi, J. C. Bardwell, and S. Horowitz, p. 20180108, Apr. 2018. “The mechanism of HdeA unfolding and chaperone activation,” Journal [124] K. Inoue, S. Ito, Y. Kato, Y. Nomura, M. Shibata, T. Uchihashi, S. P. of Molecular Biology, vol. 430, no. 1, pp. 33–40, Jan. 2018. Tsunoda, and H. Kandori, “A natural light-driven inward proton pump,” [147] E. Socher and H. Sticht, “Probing the structure of the Escherichia Nature Communications, vol. 7, p. 13415, Nov. 2016. coli periplasmic proteins HdeA and YmgD by molecular dynamics [125] H. Arjmandi, V. Jamali, A. Ahmadzadeh, A. Burkovski, R. Schober, simulations,” The Journal of Physical Chemistry B, vol. 120, no. 46, and M. N. Kenari, “Ion pump based bio-synthetic modulator model pp. 11 845–11 855, Nov. 2016. for diffusive molecular communications,” in Proc. 17th International [148] S. Ito and K. Nagata, “Biology of Hsp47 (serpin H1), a collagen-specific Workshop on Signal Processing Advances in Wireless Communications molecular chaperone,” Seminars in Cell and Developmental Biology, (SPAWC)., Jul. 2016, pp. 1–6. vol. 62, pp. 142–151, Feb. 2017. [126] J. Soppa and D. Oesterhelt, “Bacteriorhodopsin mutants of Halobac- [149] A. Modin, H. Bjorne,¨ M. Herulf, K. Alving, E. Weitzberg, and terium sp. GRB. I. the 5-Bromo-2’-deoxyuridine selection as a method to J. Lundberg, “Nitrite-derived nitric oxide: A possible mediator of ‘acidic- isolate point mutants in halobacteria.” Journal of Biological Chemistry, metabolic’ vasodilation,” Acta Physiologica Scandinavica, vol. 171, vol. 264, no. 22, pp. 13 043–13 048, 1989. no. 1, pp. 9–16, Jan. 2001. [127] R. Anandakrishnan and D. M. Zuckerman, “Biophysical comparison of [150] G. Krauss, Biochemistry of Signal Transduction and Regulation, 4th ed. ATP-driven proton pumping mechanisms suggests a kinetic advantage Wiley-VCH Verlag, 2008. for the rotary process depending on coupling ratio,” PLOS ONE, vol. 12, [151] W. A. Catterall, E. Perez-Reyes, T. P. Snutch, and J. Striessnig, no. 3, p. e0173500, Mar. 2017. “International union of pharmacology. xlviii. nomenclature and structure- [128] P. A. Tipler and G. Mosca, Physik fur¨ Wissenschaftler und Ingenieure, function relationships of voltage-gated calcium channels,” Pharmaco- 7th ed. Springer Spektrum, 2015. logical Reviews, vol. 57, no. 4, pp. 411–425, 2005. [129] R. Van Sluis, Z. M. Bhujwalla, N. Raghunand, P. Ballesteros, J. Alvarez, [152] P. Ronde´ and R. A. Nichols, “High calcium permeability of serotonin S. Cerdan,´ J.-P. Galons, and R. J. Gillies, “In vivo imaging of 5-HT3 receptors on presynaptic nerve terminals from rat striatum,” extracellular pH using 1H MRSI,” Magnetic Resonance in Medicine: Journal of Neurochemistry, vol. 70, no. 3, pp. 1094–1103, 1998. An Official Journal of the International Society for Magnetic Resonance [153] T. C. Ho, N. A. Horn, T. Huynh, L. Kelava, and J. B. Lansman, in Medicine, vol. 41, no. 4, pp. 743–750, 1999. “Evidence TRPV4 contributes to mechanosensitive ion channels in [130] G. Miesenbock,¨ D. A. D. Angelis, and J. E. Rothman, “Visualizing mouse skeletal muscle fibers,” Channels, vol. 6, no. 4, pp. 246–254, secretion and synaptic transmission with pH-sensitive green fluorescent 2012. proteins,” Nature, vol. 394, no. 6689, pp. 192–195, Jul. 1998. [154] I. M. Bennett, H. M. Vanegas Farfano, F. Bogani, A. Primak, P. A. [131] R. N. Day and M. W. Davidson, “The fluorescent protein palette: Tools Liddell, L. Otero, L. Sereno, J. J. Silber, A. L. Moore, T. Moore, for cellular imaging,” Chemical Society Reviews, vol. 38, no. 10, p. and D. Gust, “Active transport of Ca2+ by an artificial photosynthetic 2887, 2009. membrane,” Nature, vol. 420, pp. 398–401, 12 2002. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 33

[155] C. Toyoshima, “How Ca2+-ATPase pumps ions across the sarcoplasmic [178] J.-M. Beaulieu and R. R. Gainetdinov, “The physiology, signaling, reticulum membrane,” Biochimica et Biophysica Acta (BBA) - Molecular and pharmacology of dopamine receptors,” Pharmacological Reviews, Cell Research, vol. 1793, no. 6, pp. 941–946, Jun. 2009. vol. 63, no. 1, pp. 182–217, Feb. 2011. [156] M. D. Bootman, K. Rietdorf, T. Collins, S. Walker, and M. Sanderson, [179] L. C. Berumen, A. Rodr´ıguez, R. Miledi, and G. Garc´ıa-Alcocer, “Ca2+-sensitive fluorescent dyes and intracellular Ca2+ imaging,” Cold “Serotonin receptors in hippocampus,” The Scientific World Journal, vol. Spring Harbor Protocols, vol. 2013, no. 2, pp. pdb–top066 050, 2013. 2012, pp. 1–15, 2012. [157] S.-K. Lee, J.-Y. Lee, M.-Y. Lee, S.-M. Chung, and J.-H. Chung, “Ad- [180] H. Lodish, A. Berk, L. Zipursky, P. Matsudaira, D. Baltimore, and vantages of Calcium Green-1 over other fluorescent dyes in measuring J. Darnell, Molecular Cell Biology. W. H. Freeman, 2000. cytosolic calcium in platelets,” Analytical Biochemistry, vol. 273, no. 2, [181] H. E. Kato, K. Inoue, H. Kandori, and O. Nureki, “The light-driven pp. 186–191, Sep. 1999. sodium ion pump: A new player in rhodopsin research,” Bioessays, [158] J. Nakai, M. Ohkura, and K. Imoto, “A high signal-to-noise Ca2+ probe vol. 38, no. 12, pp. 1274–1282, 2016. composed of a single green fluorescent protein,” Nature Biotechnology, [182] V. V. Gurevich and E. V. Gurevich, “Molecular mechanisms of GPCR vol. 19, no. 2, pp. 137–141, Feb. 2001. signaling: A structural perspective,” International Journal of Molecular [159] M. Mank, D. F. Reiff, N. Heim, M. W. Friedrich, A. Borst, and Sciences, vol. 18, no. 12, p. 2519, 2017. O. Griesbeck, “A FRET-based calcium biosensor with fast signal kinetics [183] G. Manning, “The protein kinase complement of the human genome,” and high fluorescence change,” Biophysical Journal, vol. 90, no. 5, pp. Science, vol. 298, no. 5600, pp. 1912–1934, Dec. 2002. 1790–1796, Mar. 2006. [184] H. Xu and D. R. Littman, “The kinase-dependent function of Lck in T-cell activation requires an intact site for tyrosine autophosphorylation,” [160] G. Ellis-Davies, J. Kaplan, and R. Barsotti, “Laser photolysis of caged Annals of the New York Academy of Sciences, vol. 766, pp. 99–116, calcium: Rates of calcium release by nitrophenyl-EGTA and DM- Sep. 1995. nitrophen,” Biophysical Journal, vol. 70, no. 2, pp. 1006–1016, Feb. [185] W. Critchley, C. Pellet-Many, B. Ringham-Terry, M. Harrison, I. Zachary, 1996. and S. Ponnambalam, “Receptor tyrosine kinase ubiquitination and de- [161] G. C. R. Ellis-Davies, “Caged compounds: Photorelease technology for ubiquitination in signal transduction and receptor trafficking,” Cells, control of cellular chemistry and physiology,” Nature Methods, vol. 4, vol. 7, no. 3, p. 22, Mar. 2018. no. 8, pp. 619–628, Aug. 2007. [186] W. Jiang, P. A. Veno, B. W. Wood, G. Peaucellier, and W. H. Kinsey, [162] O. Nuyken, J. Dauth, and W. Pekruhn, “Synthesis and application of “pH regulation of an egg cortex tyrosine kinase,” Developmental Biology, thermosensitive microcapsules containing azo groups,” Angewandte vol. 146, no. 1, pp. 81–88, Jul. 1991. Makromolekulare Chemie, vol. 187, no. 1, pp. 207–224, Mar. 1991. [187] J. Pines, “Cyclins and cyclin-dependent kinases: A biochemical view,” [163] B. Martinac, “Mechanosensitive ion channels: An evolutionary and Biochemical Journal, vol. 308, no. 3, pp. 697–711, Jun. 1995. scientific tour de force in mechanobiology,” Channels, vol. 6, no. 4, pp. [188] P.-H. Chu, D. Tsygankov, M. E. Berginski, O. Dagliyan, S. M. Gomez, 211–213, 2012. T. C. Elston, A. V. Karginov, and K. M. Hahn, “Engineered kinase [164] L. Lindenburg and M. Merkx, “Engineering genetically encoded FRET activation reveals unique morphodynamic phenotypes and associated sensors,” Sensors, vol. 14, no. 7, pp. 11 691–11 713, Jul. 2014. trafficking for Src family isoforms,” Proceedings of the National [165] T. Milburn, N. Matsubara, A. P. Billington, J. B. Udgaonkar, J. W. Academy of Sciences, vol. 111, no. 34, pp. 12 420–12 425, Aug. 2014. Walker, B. K. Carpenter, W. W. Webb, J. Marque, and W. Denk, [189] O. Dagliyan, A. V. Karginov, S. Yagishita, M. E. Gale, H. Wang, “Synthesis, photochemistry, and biological activity of a caged photolabile C. DerMardirossian, C. M. Wells, N. V. Dokholyan, H. Kasai, and acetylcholine receptor ligand,” Biochemistry, vol. 28, no. 1, pp. 49–55, K. M. Hahn, “Engineering Pak1 allosteric switches,” ACS Synthetic Jan. 1989. Biology, vol. 6, no. 7, pp. 1257–1262, Apr. 2017. [166] R. Araya, V. Andino-Pavlovsky, R. Yuste, and R. Etchenique, “Two- [190] D. Wilson, J. W. Li, and N. R. Branda, “Visible-light-triggered activation photon optical interrogation of individual dendritic spines with caged of a protein kinase inhibitor,” ChemMedChem, vol. 12, no. 4, pp. 284– dopamine,” ACS Chemical Neuroscience, vol. 4, no. 8, pp. 1163–1167, 287, Jan. 2017. May. 2013. [191] M. Grusch, K. Schelch, R. Riedler, E. Reichhart, C. Differ, W. Berger, [167] R. Cabrera, O. Filevich, B. Garc´ıa-Acosta, J. Athilingam, K. J. Bender, A. Ingles-Prieto, and H. Janovjak, “Spatio-temporally precise activation K. E. Poskanzer, and R. Etchenique, “A visible-light-sensitive caged of engineered receptor tyrosine kinases by light,” The EMBO Journal, serotonin,” ACS Chemical Neuroscience, vol. 8, no. 5, pp. 1036–1042, vol. 33, no. 15, pp. 1713–1726, Jul. 2014. May 2017. [192] K.-Y. Chang, D. Woo, H. Jung, S. Lee, S. Kim, J. Won, T. Kyung, [168] L. Shi, M. Quick, Y. Zhao, H. Weinstein, and J. A. Javitch, “The H. Park, N. Kim, H. W. Yang, J.-Y. Park, E. M. Hwang, D. Kim, and mechanism of a neurotransmitter: Sodium symporter—inward release W. D. Heo, “Light-inducible receptor tyrosine kinases that regulate of Na+ and substrate is triggered by substrate in a second binding site,” neurotrophin signalling,” Nature Communications, vol. 5, Jun. 2014. Molecular Cell, vol. 30, no. 6, pp. 667–677, 2008. [193] X. X. Zhou, L. Z. Fan, P. Li, K. Shen, and M. Z. Lin, “Optical control [169] G. J. Siegel, B. W. Agranoff, W. R. Albers, S. K. Fisher, and M. D. of cell signaling by single-chain photoswitchable kinases,” Science, vol. Uhler, Basic Neurochemistry: Molecular, Cellular and Medical Aspects. 355, no. 6327, pp. 836–842, Feb. 2017. Lippincott-Raven, 1999. [194] J. Homola, S. S. Yee, and G. Gauglitz, “Surface plasmon resonance [170] L. J. D. Felice, “A current view of serotonin transporters,” sensors: Review,” Sensors and Actuators B: Chemical, vol. 54, no. 1-2, F1000Research, vol. 5, p. 1884, Jul. 2016. pp. 3–15, Jan. 1999. [195] J. Jin and T. Pawson, “Modular evolution of phosphorylation-based [171] U. Arvidsson, M. Riedl, R. Elde, and B. Meister, “Vesicular acetyl- signalling systems,” Philosophical Transactions of the Royal Society B: choline transporter (VAChT) protein: A novel and unique marker for Biological Sciences, vol. 367, no. 1602, pp. 2540–2555, Aug. 2012. cholinergic neurons in the central and peripheral nervous systems,” [196] B. A. Liu, E. Shah, K. Jablonowski, A. Stergachis, B. Engelmann, Journal of Comparative Neurology, 1998. and P. D. Nash, “The SH2 domain-containing proteins in 21 species [172] K. Wimalasena, “Vesicular monoamine transporters: Structure-function, establish the provenance and scope of phosphotyrosine signaling in pharmacology, and medicinal chemistry,” Medicinal Research Reviews, eukaryotes,” Science Signaling, vol. 4, no. 202, pp. ra83–ra83, Dec. vol. 31, no. 4, pp. 483–519, Feb. 2010. 2011. [173] B. R. MacIntosh, P. F. Gardiner, and A. J. McComas, Skeletal muscle: [197] M. Tinti, L. Kiemer, S. Costa, M. L. Miller, F. Sacco, J. V. Olsen, Form and function. Human Kinetics, 1952. M. Carducci, S. Paoluzi, F. Langone, C. T. Workman, N. Blom, [174] R. Cools, K. Nakamura, and N. D. Daw, “Serotonin and dopamine: K. Machida, C. M. Thompson, M. Schutkowski, S. Brunak, M. Mann, Unifying affective, activational and decision functions,” Neuropsy- B. J. Mayer, L. Castagnoli, and G. Cesareni, “The SH2 domain chopharmacology, vol. 36, no. 1, pp. 98–113, Aug. 2010. interaction landscape,” Cell Reports, vol. 3, no. 4, pp. 1293–1305, [175] E. X. Albuquerque, E. F. R. Pereira, M. Alkondon, and S. W. Apr. 2013. Rogers, “Mammalian nicotinic acetylcholine receptors: From structure [198] J. Y. Bu, A. S. Shaw, and A. C. Chan, “Analysis of the interaction to function,” Physiological Reviews, vol. 89, no. 1, pp. 73–120, Jan. of ZAP-70 and Syk protein-tyrosine kinases with the T-cell antigen 2009. receptor by plasmon resonance.” Proc. of the National Academy of [176] A. Thompson and S. R. Lummis, “5-HT3 receptors,” Current Pharma- Sciences, vol. 92, no. 11, pp. 5106–5110, May 1995. ceutical Design, vol. 12, no. 28, pp. 3615–3630, Oct. 2006. [199] P. Astuti, R. Boutros, B. Ducommun, and B. Gabrielli, “Mitotic [177] T. Haga, “Molecular properties of muscarinic acetylcholine receptors,” phosphorylation of Cdc25B Ser321 disrupts 14-3-3 binding to the Proc. of the Japan Academy, Series B, vol. 89, no. 6, pp. 226–256, high affinity Ser323 site,” Journal of Biological Chemistry, vol. 285, 2013. no. 45, pp. 34 364–34 370, Aug. 2010. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 34

[200] K. V. Roey, H. Dinkel, R. J. Weatheritt, T. J. Gibson, and N. E. Davey, Modulation of the pH-activity profile,” Biomolecular Engineering, “The switches.ELM resource: A compendium of conditional regulatory vol. 23, no. 5, pp. 247–251, Oct. 2006. interaction interfaces,” Science Signaling, vol. 6, no. 269, pp. rs7–rs7, [223] J. Sun, L. Lei, C.-M. Tsai, Y. Wang, Y. Shi, M. Ouyang, S. Lu, J. Seong, Apr. 2013. T.-J. Kim, P. Wang, M. Huang, X. Xu, V. Nizet, S. Chien, and Y. Wang, [201] X.-S. Zhang, N. Tegtmeyer, L. Traube, S. Jindal, G. Perez-Perez, “Engineered proteins with sensing and activating modules for automated H. Sticht, S. Backert, and M. J. Blaser, “A specific A/T polymorphism reprogramming of cellular functions,” Nature Communications, vol. 8, in Western tyrosine phosphorylation B-motifs regulates Helicobacter no. 1, Sep. 2017. pylori CagA epithelial cell interactions,” PLoS Pathogens, vol. 11, no. 2, [224] R. Milo and R. Phillips, Cell biology by the numbers. Garland Science, 2015. 2015. [202] H. Li, H. Xu, Y. Zhou, J. Zhang, C. Long, S. Li, S. Chen, J.-M. Zhou, [225] W.-W. Wang, O. A. Sineshchekov, E. N. Spudich, and J. L. Spudich, and F. Shao, “The phosphothreonine lyase activity of a bacterial type “Spectroscopic and photochemical characterization of a deep ocean III effector family,” Science, vol. 315, no. 5814, p. 1000–1003, Feb. proteorhodopsin,” Journal of Biological Chemistry, vol. 278, no. 36, 2007. pp. 33 985–33 991, Jun. 2003. [203] N. E. Davey, G. Trave,´ and T. J. Gibson, “How viruses hijack cell [226] M. D. Carattino, W. Liu, W. G. Hill, L. M. Satlin, and T. R. Kleyman, regulation,” Trends in Biochemical Sciences, vol. 36, no. 3, pp. 159–169, “Lack of a role of membrane-protein interactions in flow-dependent 2011. activation of ENaC,” American Journal of Physiology-Renal Physiology, [204] F. Margottin, S. P. Bour, H. Durand, L. Selig, S. Benichou, V. Richard, vol. 293, no. 1, pp. F316–F324, Jul. 2007. D. Thomas, K. Strebel, and R. Benarous, “A novel human WD protein, [227] B. Musset and T. DeCoursey, “Biophysical properties of the voltage- h-βTrCP, that interacts with HIV-1 vpu connects CD4 to the ER gated proton channel HV1,” Wiley Interdisciplinary Reviews: Membrane degradation pathway through an f-box motif,” Molecular Cell, vol. 1, Transport and Signaling, vol. 1, no. 5, pp. 605–620, May 2012. no. 4, p. 565–574, Mar. 1998. [228] T. DeCoursey and V. Cherny, “Potential, pH, and arachidonate gate [205] P. Devarajan, M. C. Jones, O. Kugler-Umana, A. M. Vong, J. Xia, and hydrogen ion currents in human neutrophils,” Biophysical Journal, S. L. Swain, “Pathogen recognition by CD4 effectors drives key effector vol. 65, no. 4, pp. 1590–1598, Oct. 1993. and most memory cell generation against respiratory virus,” Frontiers [229] J. M. Berg, J. L. Tymoczko, and L. Stryer, Eds., Biochemistry, 5th in Immunology, vol. 9, p. 596, 2018. edition. W. H. Freeman, 2002. [206] I. M. Cartwright, V. F. Curtis, J. M. Lanis, E. E. Alexeev, N. Welch, [230] J. H. Morais-Cabral, Y. Zhou, and R. MacKinnon, “Energetic optimiza- M. S. Goldberg, R. E. Schaefer, R. Y. Gao, C. Chun, B. Fennimore tion of ion conduction rate by the K+ selectivity filter,” Nature, vol. et al., “Adaptation to inflammatory acidity through neutrophil-derived 414, no. 6859, pp. 37–42, Nov. 2001. adenosine regulation of SLC26A3,” Mucosal Immunology, vol. 13, no. 2, [231] S. Bendahhou, “Keeping hyperactive voltage-gated sodium channels in pp. 230–244, 2020. silent mode,” The Journal of Physiology, vol. 590, no. 11, pp. 2543– [207] I. F. Tannock and D. Rotin, “Acid pH in tumors and its potential 2544, May 2012. for therapeutic exploitation,” Cancer Research, vol. 49, no. 16, pp. [232] W. D. Stein and T. Litman, “Ion channels across cell membranes,” in 4373–4384, 1989. Channels, Carriers, and Pumps. Elsevier, 2015, pp. 81–130. [208] P. Belujon and A. A. Grace, “Dopamine system dysregulation in major [233] A. Zweifach and R. S. Lewis, “Mitogen-regulated Ca2+ current of depressive disorders,” International Journal of Neuropsychopharmacol- T lymphocytes is activated by depletion of intracellular Ca2+ stores.” ogy, vol. 20, no. 12, p. 1036–1046, Jun. 2017. Proceedings of the National Academy of Sciences, vol. 90, no. 13, pp. [209] J. Jin and T. Pawson, “Modular evolution of phosphorylation-based 6295–6299, Jul. 1993. signalling systems,” Philosophical Transactions of the Royal Society B: [234] T. Clark, “G-protein coupled receptors: Answers from simulations,” Biological Sciences, vol. 367, no. 1602, pp. 2540–2555, 2012. Beilstein Journal of Organic Chemistry, vol. 13, pp. 1071–1078, Jun. [210] M. Tinti, L. Kiemer, S. Costa, M. L. Miller, F. Sacco, J. V. Olsen, 2017. M. Carducci, S. Paoluzi, F. Langone, C. T. Workman et al., “The [235] M. Heyden, H. E. Jardon-Valadez, A.-N. Bondar, and D. J. Tobias, SH2 domain interaction landscape,” Cell Reports, vol. 3, no. 4, pp. “GPCR activation on the microsecond timescale in MD simulations,” 1293–1305, 2013. Biophysical Journal, vol. 104, no. 2, p. 115a, Jan. 2013. [211] K. Kadaveru, J. Vyas, and M. R. Schiller, “Viral infection and human disease-insights from minimotifs,” Frontiers in Bioscience: A Journal [236] M. A. Ayoub, F. Landomiel, N. Gallay, G. Jegot,´ A. Poupon, P. Crepieux,´ and Virtual Library, vol. 13, p. 6455, 2008. and E. Reiter, “Assessing gonadotropin receptor function by resonance [212] F. Bahram, N. von der Lehr, C. Cetinkaya, and L.-G. Larsson, “c-Myc energy transfer-based assays,” Frontiers in Endocrinology, vol. 6, Aug. hot spot mutations in lymphomas result in inefficient ubiquitination and 2015. decreased proteasome-mediated turnover.” Blood, vol. 95, p. 2104–2110, [237] J. Koolman and K.-H. Rohm,¨ Taschenatlas der Biochemie. Georg Mar. 2000. Thieme Verlag, 2003. [213] J. A. Levenson, A. C. Simon, and M. E. Safar, “Vasodilatation of [238] T. L. Rosenberry, “Acetylcholinesterase,” Advances in Enzymology and small and large arteries in hypertension,” Journal of Cardiovascular Related Areas of Molecular Biology, vol. 43, pp. 103–218, 1975. Pharmacology, vol. 7, p. S115–S120, 1985. [239] D. L. Purich, Ed., Enzyme kinetics: Catalysis and control: A reference [214] S. Peer and G. Bodner, High-Resolution Sonography of the Peripheral of theory and best-practice methods. Elsevier, 2010. Nervous System. Springer, 2008. [240] R. Ramsay and K. Tipton, “Assessment of enzyme inhibition: A [215] J. G. Proakis and M. Salehi, Digital Communications 5ed. McGraw- review with examples from the development of monoamine oxidase Hill, 2008. and cholinesterase inhibitory drugs,” Molecules, vol. 22, no. 7, p. 1192, [216] J. W. Foster, “Escherichia coli acid resistance: Tales of an amateur Jul. 2017. acidophile,” Nature Reviews Microbiology, vol. 2, no. 11, pp. 898–907, [241] M. D. Falk, W. Liu, B. Bolanos,˜ K. Unsal-Kacmaz, A. Klippel, Nov. 2004. S. Grant, A. Brooun, and S. Timofeevski, “Enzyme kinetics and distinct [217] V. A. Khramov, “The utilization of amino acids and urea by human modulation of the protein kinase N family of kinases by lipid activators oral fluid,” Stomatologiia (Mosk)., vol. 76, no. 6, pp. 13–15, 1997. and small molecule inhibitors,” Bioscience Reports, vol. 34, no. 2, pp. [218] H. T. L. Mobley, “The role of Helicobacter pylori urease in the patho- 93–106, Mar. 2014. genesis of gastritis and peptic ulceration,” Alimentary Pharmacology [242] C. C. Fjeld and J. M. Denu, “Kinetic analysis of human serine/threonine and Therapeutics, vol. 10, no. Suppl 1, pp. 57–64, Apr. 1996. protein phosphatase 2Cα,” Journal of Biological Chemistry, vol. 274, [219] F. Ullmann, Ullmann’s Encyclopedia of Industrial Chemistry. Wiley- no. 29, pp. 20 336–20 343, Jul. 1999. VCH Verlag GmbH and Co. KGaA, Jun. 2000. [243] Y. Ogura, “Catalase activity at high concentration of hydrogen peroxide,” [220] A. Noel, K. Cheung, and R. Schober, “Improving diffusion-based Archives of Biochemistry and Biophysics, vol. 57, no. 2, pp. 288–300, molecular communication with unanchored enzymes,” in International 1955. Conference on Bio-Inspired Models of Network, Information, and [244] S. J. Putman, A. F. W. Coulson, I. R. T. Farley, B. Riddleston, and J. R. Computing Systems. Springer, 2012, pp. 184–198. Knowles, “Specificity and kinetics of triose phosphate isomerase from [221] S. Wagner, A. Schutz,¨ and J. Rademann, “Light-switched inhibitors chicken muscle,” Biochemical Journal, vol. 129, no. 2, pp. 301–310, of protein tyrosine phosphatase PTP1b based on phosphonocarbonyl 1972. phenylalanine as photoactive phosphotyrosine mimetic,” Bioorganic and [245] J. G. Voet and D. Voet, Eds., Biochemistry and molecular biology. Medicinal Chemistry, vol. 23, no. 12, pp. 2839–2847, Jun. 2015. Education. Wiley, 2005. [222] R. D. Makde, K. Dikshit, and V. Kumar, “Protein engineering of class-a [246] A. Bar-Even, E. Noor, Y. Savir, W. Liebermeister, D. Davidi, D. S. non-specific acid phosphatase (PhoN) of Salmonella Typhimurium: Tawfik, and R. Milo, “The moderately efficient enzyme: Evolutionary IEEE COMMUNICATIONS SURVEYS & TUTORIALS 35

and physicochemical trends shaping enzyme parameters,” Biochemistry, [269] M. Rossignol, T. Uso, and P. Thomas, “Relationship between fluidity and vol. 50, no. 21, pp. 4402–4410, May 2011. ionic permeability of bilayers from natural mixtures of phospholipids,” [247] H. Arjmandi, A. Ahmadzadeh, R. Schober, and M. Nasiri Kenari, The Journal of Membrane Biology, vol. 87, no. 3, pp. 269–275, 1985. “Ion channel based bio-synthetic modulator for diffusive molecular [270] K. Abe and H. Yawo, “Optogenetic conditioning of paradigm and communication,” IEEE Transactions on NanoBioscience, vol. 15, no. 5, pattern discrimination in the rat somatosensory system,” PLOS ONE, pp. 418–432, Jul. 2016. vol. 12, no. 12, p. e0189439, Dec. 2017. [248] Y. Deng, W. Guo, A. Noel, A. Nallanathan, and M. Elkashlan, “Enabling [271] S. Balasubramaniam, S. A. Wirdatmadja, M. T. Barros, Y. Koucheryavy, energy efficient molecular communication via molecule energy transfer,” M. Stachowiak, and J. M. Jornet, “Wireless communications for IEEE Communications Letters, vol. 21, no. 2, pp. 254–257, Feb. 2017. optogenetics-based brain stimulation: Present technology and future [249] W. Guo, Y. Deng, H. B. Yilmaz, N. Farsad, M. Elkashlan, A. Eckford, challenges,” IEEE Communications Magazine, vol. 56, no. 7, pp. 218– A. Nallanathan, and C. Chae, “SMIET: Simultaneous molecular 224, Jul. 2018. information and energy transfer,” IEEE Wireless Communications, [272] S. A. Wirdatmadja, M. T. Barros, Y. Koucheryavy, J. M. Jornet, and vol. 25, no. 1, pp. 106–113, Feb. 2018. S. Balasubramaniam, “Wireless optogenetic nanonetworks for brain [250] S. Lotter, A. Ahmadzadeh, and R. Schober, “Channel modeling stimulation: Device model and charging protocols,” IEEE Transactions for synaptic molecular communication with re-uptake and reversible on NanoBioscience, vol. 16, no. 8, pp. 859–872, Dec. 2017. receptor binding,” Submitted to IEEE ICC, 2019. [Online]. Available: [273] N. Li and P. Miao, “Let there be light: A tutorial on optogenetics.” https://arxiv.org/pdf/1912.04025.pdf IEEE Pulse, vol. 5, no. 4, pp. 55–59, Jul. 2014. [251] J. Kubicek, H. Block, B. Maertens, A. Spriestersbach, and J. Labahn, [274] R. Pashaie, P. Anikeeva, J. H. Lee, R. Prakash, O. Yizhar, M. Prigge, “Expression and purification of membrane proteins,” in Methods in D. Chander, T. J. Richner, and J. Williams, “Optogenetic brain interfaces,” Enzymology. Elsevier, 2014, pp. 117–140. IEEE Reviews in Biomedical Engineering, vol. 7, pp. 3–30, 2014. [252] M. W. Lluis, J. I. Godfroy, and H. Yin, “Protein engineering methods [275] C. Hammond, R. Ammari, B. Bioulac, and L. Garcia, “Latest view applied to membrane protein targets,” Protein Engineering Design and on the mechanism of action of deep brain stimulation,” Movement Selection, vol. 26, no. 2, pp. 91–100, Oct. 2012. Disorders, vol. 23, no. 15, p. 2111–2121, Nov. 2008. [253] C.-Y. Lin and J. C. Liu, “Modular protein domains: An engineering [276] P. Turner, A. McLennan, A. Bates, and M. White, BIOS Instant Notes approach toward functional biomaterials,” Current Opinion in Biotech- in Molecular Biology, 3rd ed. Taylor & Francis, 2005. nology, vol. 40, pp. 56–63, Aug. 2016. [277] C. Branden and J. Tooze, Introduction to Protein Structure, 2nd ed. [254] A. H. C. Horn and H. Sticht, “Synthetic protein scaffolds based on Garland Publishing, 1999. peptide motifs and cognate adaptor domains for improving metabolic [278] J. M. Berg, L. Stryer, J. L. Tymoczko, and G. J. Gatto, Biochemistry, productivity,” Frontiers in Bioengineering and Biotechnology, vol. 3, 8th ed. WH Freeman & Co Ltd, 2015. Nov. 2015. [255] A. Banerjee and M. Howarth, “Nanoteamwork: Covalent protein assembly beyond duets towards protein ensembles and orchestras,” Current Opinion in Biotechnology, vol. 51, pp. 16–23, Jun. 2018. [256] T. U. J. Bruun, A.-M. C. Andersson, S. J. Draper, and M. Howarth, “Engineering a rugged nanoscaffold to enhance plug-and-display vacci- nation,” ACS Nano, Jul. 2018. [257] J.-L. Rigaud and D. Levy,´ “Reconstitution of membrane proteins into liposomes,” in Methods in Enzymology. Elsevier, 2003, pp. 65–86. [258] D. E. Discher and A. Eisenberg, “Polymer vesicles,” Science, vol. 297, no. 5583, pp. 967–973, 2002. [259] J.-F. Le Meins, O. Sandre, and S. Lecommandoux, “Recent trends in the tuning of polymersomes’ membrane properties,” The European Physical Journal E, vol. 34, no. 2, pp. 1–17, 2011. [260] Q. Chen, H. Schonherr,¨ and G. J. Vancso, “Mechanical properties of block copolymer vesicle membranes by atomic force microscopy,” Soft Matter, vol. 5, no. 24, pp. 4944–4950, 2009. [261] S. T. Poschenrieder, L. Klermund, B. Langer, and K. Castiglione, “Determination of permeability coefficients of polymersomal membranes for hydrophilic molecules,” Langmuir, vol. 33, no. 24, pp. 6011–6020, 2017. [262] C. G. Palivan, R. Goers, A. Najer, X. Zhang, A. Car, and W. Meier, “Bioinspired polymer vesicles and membranes for biological and medical applications,” Chemical Society Reviews, vol. 45, no. 2, pp. 377–411, 2016. [263] P. Tanner, S. Egli, V. Balasubramanian, O. Onaca, C. G. Palivan, and W. Meier, “Can polymeric vesicles that confine enzymatic reactions act as simplified organelles?” FEBS Letters, vol. 585, no. 11, pp. 1699–1706, 2011. [264] P. Broz,ˇ S. M. Benito, C. Saw, P. Burger, H. Heider, M. Pfisterer, S. Marsch, W. Meier, and P. Hunziker, “Cell targeting by a generic receptor-targeted polymer nanocontainer platform,” Journal of Con- trolled Release, vol. 102, no. 2, pp. 475–488, 2005. [265] J. Gaitzsch, X. Huang, and B. Voit, “Engineering functional polymer capsules toward smart nanoreactors,” Chemical Reviews, vol. 116, no. 3, pp. 1053–1093, 2016. [266] T. S. Schwarzer, L. Klermund, G. Wang, and K. Castiglione, “Mem- brane functionalization of polymersomes: Alleviating mass transport limitations by integrating multiple selective membrane transporters for the diffusion of chemically diverse molecules,” Nanotechnology, vol. 29, no. 44, p. 44LT01, 2018. [267] P. A. Beales, S. Khan, S. P. Muench, and L. J. Jeuken, “Durable vesicles for reconstitution of membrane proteins in biotechnology,” Biochemical Society Transactions, vol. 45, no. 1, pp. 15–26, 2017. [268] R. Goers, J. Thoma, N. Ritzmann, A. Di Silvestro, C. Alter, G. Gunkel- Grabole, D. Fotiadis, D. J. Muller,¨ and W. Meier, “Optimized reconstitu- tion of membrane proteins into synthetic membranes,” Communications Chemistry, vol. 1, no. 1, pp. 1–10, 2018. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 36

TABLE IX EXPLANATIONOFRELEVANTBIOLOGICALANDCHEMICALTERMS.FORMOREIN-DEPTHDISCUSSION, WERECOMMENDREFERENCES [276]–[278].

Term Description From amino acids to proteins Amino acid Organic compound containing an amine (NH2) group, a carboxyl (COOH) group, and a specific side chain. Building block of a protein. Peptide Short chain of amino acids (smaller than a protein) linked by peptide bonds. Protein Large biomolecule, consisting of one or more long chains of amino acids. Crystal structure Three-dimensional model of the structure of a molecule (e.g. a protein) resolved by X-ray crystallography. Residue Single amino acid within a peptide or protein. Proteome Entire set of proteins expressed by a certain organism, tissue or cell. Protein folding and modifications Chaperone Protein that assists the folding process of other proteins or the assembly of macromolecular structures. Molecular maturation Collective term for additional modifications taking place after protein biosynthesis, such as cleavage, posttranslational modifications and oligomerization. Posttranslational modification of Covalent modification of a protein or a peptide after protein biosynthesis. proteins/peptides

Methylation/Acetylation/ Examples for posttranslational modifications: addition of a methyl (CH3) / acetyl (CH3CO) / ubiquitin (a Ubiquitination small regulatory protein) group. Different oligomerization states Monomer Single molecule which can undergo oligomerization and thereby contribute a constitutional unit to a macromolecule. Dimer Chemical structure formed from two subunits. Oligomer Macromolecule formed from a few subunits. Homo-/Hetero- Consisting of identical/different sub-molecules.

Protein domains Protein domain Part of a protein which can evolve, fold, and function independently from the rest of the protein chain. Adapter domain Domain mediating specific interactions with a binding partner. Homologous domain Domain which is similar to another one due to a common evolutionary origin. Enzymes – Biological catalysts Kinase Transfers a phosphate group from ATP to a protein. ATPase Catalyzes cleavage of ATP to ADP and a phosphate group, uses released energy to drive a chemical reaction that would not occur otherwise. Protease Cleaves a protein by hydrolysis of peptide bonds. Phosphatase Removes a phosphate group from a phosphorylated protein. Allosteric regulation Regulation of an enzyme by binding an effector molecule at a site other than the enzyme’s active site. Overview of chemical terms Moiety Chemical group (e.g. methyl moiety, acetyl moiety, peptide moiety) Acidic conditions Environmental conditions characterized by a low pH value, i.e., a high proton concentration. Protonation Covalent addition of a proton to a molecule, formation of the conjugate acid. Hydrolysis Cleavage of a biomolecule accompanied by the consumption of a water molecule. Chelating agent Chemical agent that binds certain metal ions by forming two or more coordinate bonds. May be used to remove or chemically shield the metal ions. Miscellaneous Surface-plasmon resonance Biophysical method to detect biomolecular interactions. Recombinant DNA/protein DNA/protein sequence engineered by laboratory methods of genetic recombination such as molecular cloning. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 37

Christian A. Soldner¨ received the B. Sc. and M. Wayan Wicke (S’17) was born in Nuremberg, Sc. degree in molecular medicine from the Friedrich Germany, in 1991. He received the B.Sc. and M.Sc. Alexander University Erlangen-Nurnberg,¨ Germany, degrees in electrical engineering from the Friedrich- in 2014 and 2016, respectively. During his Ph.D. in Alexander University Erlangen-Nurnberg¨ (FAU), Er- the subject of bioinformatics, he focused his research langen, Germany, in 2014 and 2017, respectively, interest on the analysis of receptor-ligand interactions. where he is currently pursuing the Ph.D. degree. His He studied GPCRs and the macrophage surface research interests include statistical signal processing receptor Mincle by means of molecular dynamics and digital communications with a focus on molecular simulations and developed a metadynamics-based communication. protocol for the determination of ligand binding modes. After his Ph.D. degree, which he received in 2020, he joined Siemens Healthineers and is currently working as an application specialist and sequence developer for magnetic resonance imaging.

Arman Ahmadzadeh (S’14) received the B.Sc. degree in electrical engineering from the Ferdowsi University of Mashhad, Mashhad, Iran, in 2010, and the M.Sc. degree in communications and multimedia engineering from the Friedrich-Alexander-Universitat¨ Eileen Socher received the B.Sc. and M.Sc. degrees (FAU) Erlangen-Nurnberg,¨ Erlangen, Germany, in in Molecular Medicine (with a focus on pathol- 2013, where he is currently working toward the Ph.D. ogy and bioinformatics) from Friedrich-Alexander- degree in electrical engineering at the Institute for University Erlangen-Nurnberg¨ (FAU), Erlangen, Ger- Digital Communications. His current research inter- many, in 2011 and 2012, respectively. In 2017, she ests include physical layer molecular communications. received the doctoral degree in the field of bioin- Mr. Ahmadzadeh has served as a member of the formatics from the Faculty of Sciences, Friedrich- Technical Program Committee of the Communication Theory Symposium Alexander-University Erlangen-Nurnberg¨ (FAU). Cur- for the IEEE International Conference on Communications (ICC) 2017-2020. rently, she is a postdoctoral researcher at the FAU He received several awards including the Best Paper Award from the IEEE Faculty of Medicine in Erlangen, Germany. Her ICC in 2016, IEEE ICC in 2020, and the Student Travel Grants for attending research interests include structural bioinformatics in the Global Communications Conference (GLOBECOM) in 2017. He was general and the molecular modelling of proteins as well as their investigation recognized as an Exemplary Reviewer of IEEECOMMUNICATIONS LETTERS by using molecular dynamics simulations in particular. Within this field, she in 2016. works, for instance, on the computer-based analysis of pH-induced effects on protein structure or protein-protein interactions between viral and human proteins.

Hans-Georg Breitinger received the Ph.D. in Or- ganic Chemistry from Heinrich Heine University (HHU), Dusseldorf,¨ Germany, in 1993. He was a post- doctoral researcher at Cornell University from 1993- Vahid Jamali (S’12, M’20) received the B.Sc. and 1997, and completed the Habilitation (University M.Sc. degrees (honors) in electrical engineering teaching certificate) for Biochemistry at Friedtrich- from the K. N. Toosi University of Technology, Alexander University, Erlangen, in 2003. In 2004 he Tehran, Iran, in 2010 and 2012, respectively, and the joined the German University in Cairo as Full Pro- Ph.D. degree (with distinctions) from the Friedrich- fessor and Head of the Department of Biochemistry. Alexander-University (FAU) of Erlangen-Nurnberg,¨ His research focuses on neuronal ion channels and Erlangen, Germany, in 2019. In 2017, he was a viroporins, the biochemistry of membrane receptors, Visiting Research Scholar with Stanford University, antimicrobial peptides and isolation of new pharmaceuticals from natural CA, USA. He is currently a Postdoctoral Researcher resources. His postdoctoral stay at Cornell University was supported by a with the Institute for Digital Communication, FAU. fellowship of the Swiss Federation of Sciences. His research interests include wireless and molecular communications, Bayesian inference and learning, and multiuser informa- tion theory. Dr. Jamali has served as a member of the Technical Program Committee for several IEEE conferences and he is currently an Associate Editor of the IEEECOMMUNICATIONS LETTERS and IEEEOPEN JOURNALOFTHE COMMUNICATIONS SOCIETY. He received several awards for his publications and research work including the Best Paper Award from the IEEE International Andreas Burkovski received his diploma in Biology Conference on Communications in 2016, the Doctoral Research Grant from in 1989 and his doctoral degree in 1993, both at the the German Academic Exchange Service (DAAD) in 2017, the Goldener University of Osnabruck.¨ After postdoc positions Igel Publication Award from the Telecommunications Laboratory (LNT), at the University of Osnabruck¨ and the Research FAU, in 2018, the Best Ph.D. Thesis Presentation Award from the IEEE Center Julich,¨ he became group leader at the Uni- Wireless Communications and Networking Conference in 2018, the Best Paper versity of Cologne in 1997. In 2005 he received a Award from the ACM International Conference on Nanoscale Computing and professorship in Microbiology at the University of Communication in 2019, and the Postdoctoral Research Fellowship by the Erlangen-Nuremberg. His research interests focus on German Research Foundation (DFG) in 2020. the analysis of host-pathogen-interactions, regulatory networks and synthetic biology. IEEE COMMUNICATIONS SURVEYS & TUTORIALS 38

Kathrin Castiglione received the B.Sc. and M.Sc. degrees in Molecular Biotechnology from the Tech- nical University of Munich (TUM). After her PhD at the Institute of Biochemical Engineering of TUM in 2009, she moved to the Toyama Prefectural University in Japan as a postdoctoral fellow of the Japanese Society of the Promotion of Science. At the end of 2010, she returned to TUM and became head of the biocatalysis group at the Institute of Biochemical Engineering. Since 2012 she has been the leader of an independent junior research group working on artificial reaction compartments based on functionalized polymer vesicles. Since 2018 Kathrin Castiglione has headed the Institute of Bioprocess Engineering of the Friedrich-Alexander University Erlangen-Nurnberg¨ (FAU). Among other things, she works on the design of functionalized polymersomes for molecular communication.

Robert Schober (S’98, M’01, SM’08, F’10) received the Diplom (Univ.) and the Ph.D. degrees in electrical engineering from the Friedrich-Alexander University (FAU) Erlangen-Nurnberg,¨ Erlangen, Germany, in 1997 and 2000, respectively. From 2002 to 2011, he was a Professor and Canada Research Chair at the University of British Columbia (UBC), Vancouver, Canada. Since January 2012, he is an Alexander von Humboldt Professor and the Chair for Digital Communication at FAU. His research interests fall into the broad areas of Communication Theory, Wireless Communications, and Statistical Signal Processing. Robert received several awards for his work including the 2002 Heinz Maier-Leibnitz Award of the German Science Foundation (DFG), the 2004 Innovations Award of the Vodafone Foundation for Research in Mobile Communications, a 2006 UBC Killam Research Prize, a 2007 Wilhelm Friedrich Bessel Research Award of the Alexander von Humboldt Foundation, the 2008 Charles McDowell Award for Excellence in Research from UBC, a 2011 Alexander von Humboldt Professorship, a 2012 NSERC E.W.R. Stacie Fellowship, and a 2017 Wireless Communications Recognition Award by the IEEE Wireless Communications Technical Committee. He is listed as a 2017 Highly Cited Researcher by the Web of Science and a Distinguished Lecturer of the IEEE Communications Society (ComSoc). Robert is a Fellow of the Canadian Academy of Engineering and a Fellow of the Engineering Institute of Canada. From 2012 to 2015, he served as Editor-in-Chief of the IEEETRANSACTIONSON COMMUNICATIONS. Currently, he is the Chair of the Steering Committee of the IEEETRANSACTIONSON MOLECULAR, BIOLOGICALAND MULTI-SCALE COMMUNICATIONS, a Member of the Editorial Board of the Proceedings of the IEEE, a Member at Large of the Board of Governors of ComSoc, and the ComSoc Director of Journals.

Heinrich Sticht was born in 1968. He received the Diploma degree in biochemistry from the University of Bayreuth in 1993, the Ph.D. degree in 1995, and the Dr.habil. degree and Venia legendi from the University of Bayreuth in 1999. From 1996 to 1997, he was a Post-Doctoral Fellow with Oxford Uni- versity, U.K. In 1997, he returned to the University of Bayreuth as a Research Assistant. From 1999 to 2002, he was a Group Leader with the University of Bayreuth and, in 2002, he became a Professor of Bioinformatics with Friedrich-Alexander University. His major research interests include molecular interactions and molecular communication processes, which are studied by a combination of computational tools (e.g., sequence data analysis, molecular docking, and molecular dynamics). He has authored or co-authored about 250 peer-reviewed journal papers, reviews, or book chapters.