Integrated Photosynthetic and Metabolic Constitutional Dynamic Networks ― An “Articial Leaf”

Integration of a photosynthetic network with an assimilation, metabolic network is the fundamental prerequisite to construct an “articial leaf”. Nucleic acid-based constitutional dynamic networks provide the building modules to construct integrated, intercommunicated networks mimicking photosynthesis. Two constitutional dynamic networks composed each of four constituents provide the photosynthetic and metabolic networks. In the photosynthetic network, photoinduced electron transfer from the Zn(II)-protoporphyrin photosensitizer to a bipyridinium electron acceptor is activated, followed by the biocatalytic reduction of NADP + to NADPH, in analogy to photosystem I in native photosynthesis. In the metabolic network, the biocatalyzed-oxidation of lactate to pyruvate proceeds, followed by the metabolic transformation of pyruvate to L-alanine. The guided dynamic feedback-driven intercommunication of the networks is accomplished, leading to the function as an “articial leaf”.

the constituents, e.g., AA′ results in the adaptive dynamic recon guration of the CDN into a new equilibrated network, where the content of AA′ is enriched on the expense of AB′ and BA′ sharing components with AA′. The dynamic separation of AB′ and BA′ leads to the recombination of B and B′ and to the concomitant enrichment of BB′. The base sequence comprising nucleic acids provides a rich "toolbox" to control the stabilization/destabilization of nucleic acids by fuel/anti-fuel strands [28][29][30] , the formation/dissociation of G-quadruplexes (GQ) 31 , and the reversible stabilization/destabilization of duplexes by photoisomerizable intercalator units 32 . A variety of CDNs revealing adaptive and hierarchically adaptive recon guration properties were demonstrated, using fuel/anti-fuel strand 33 , Gquadruplex formation/dissociation 34 and light 35 as input triggers. CDNs of variable complexities, such as intercommunicating 36 and feedback-driven CDNs 34 , and the assembly of [3×3] and three-dimensional CDNs 37 were realized. The applications of CDNs are still scarce, and the use of CDNs to build hydrogels exhibiting switchable stiffness for controlled drug release and self-healing 38 , and the CDN-guided aggregation of nanoparticles for controlled catalysis and switchable optical properties 39 were demonstrated. Recently, enzymes were coupled to CDNs and the switchable operation of biocatalytic cascades and the intercommunication of enzyme networks were highlighted 40 .
In the present study, we introduce CDNs as functional modules to construct an integrated "arti cial leaf".
A CDN-guided photosynthetic network that drives the light-induced electron transfer and the photosynthesis of NADPH is coupled to a metabolic CDN assembly that stimulates the biocatalytic oxidation of lactate to pyruvate and the cascaded metabolic amination of pyruvate to L-alanine. We demonstrate the intercommunication between the photosynthetic process and the metabolic assimilation process and highlight the tight inter-relation between the two networks, where the CDN-driven enhanced photosynthesis of NADPH signals the acceleration of the metabolic network as output, and the CDNdriven enhancement of the metabolic network is translated to "information transfer" dictating the acceleration of the photosynthetic network.

Results
A photosynthetic constitutional dynamic network Fig. 1a introduces the dynamic photocatalytic module mimicking photosynthesis, CDN "X". The network consists of four constituents where the photosensitizer Zn(II)-protoporphyrin IX (Zn-PPIX) intercalates into the G-quadruplex unit, tethered as photosensitizer to component A of constituent AA′, and the N,N'dialkyl-4,4'-bipyridinium (V 2+ ) electron acceptor is covalently linked to component A′ of constituent AA′ (Figs. S1-S2). Irradiation of the photocatalytic module results in the effective quenching of the photosensitizer to yield the redox intermediates Zn-PPIX +· /GQ and MV +· (Fig. 1a, panel I). In the presence of 1,4-nicotinamide adenine dinucleotide phosphate (NADP + ) and ferredoxin-NADP + -reductase (FNR), the formation of reduced cofactor NADPH proceeds, in analogy to the photosystem I. Each of the constituents in CDN includes a loop domain, used to shift the CDN equilibrium through the stabilization of a target loop domain, panel II. In addition, each of the constituents is engineered to include a different Mg 2+ -dependent DNAzyme reporter unit to cleave uorophore-quencher ribonucleobase-modi ed substrates for the quantitative evaluation of the concentrations of constituents, panel III. Figure. 1a and S3 depicts the triggered recon guration of CDN "X" in the presence of auxiliary triggers.
Subjecting CDN "X" to the strand T 1 leads to the stabilization of the T-A . T triplex structure in the loop domain, resulting in the stabilization of AA′ and the recon guration of CDN "X" into CDN X a , where AA′ is up-regulated, constituents AB′ and BA′ are down-regulated and the constituent BB′ is up-regulated (Figs. S4-6). The reverse displacement of the trigger T 1 by the counter trigger T 1 ′ regenerates CDN "X". Similarly, treatment of CDN "X" with the trigger T 2 stabilizes the constituent BA′. BA′ and AB′ are up-regulated, and AA′ and BB′ are down-regulated (Fig. S6). The treatment of CDN X b with the counter trigger T 2 ′ restores CDN "X". The quantitative contents of the constituents in different CDNs are shown in Fig. 1b and Table  S1. The photosensitized electron transfer process proceeding in different CDNs, is stimulated by constituent AA′. The absorption spectra of the photogenerated V +· (Fig. 1c) reveal that the T 1 -up-regulated constituent AA′ in CDN X a lead to the enhanced photoinduced electron transfer, whereas the T 2 -downregulation of AA′ in CDN X b inhibited the photoinduced electron transfer. is observed. It is evident that the built-up of NADPH by CDN X a (Fig. 1f) is enhanced as compared to the NADPH generated by CDN "X" (Fig. 1e), whereas the built-up of NADPH by CDN X b (Fig. 1g) is inhibited as compared to CDN "X". Fig. 1h shows the time-dependent formation of NADPH at time-intervals of illumination of CDNs X, X a , and X b . The e ciency of the photogenerated NADPH is controlled by the e ciency of the primary photoinduced electron transfer process that yields V +· .

A metabolic constitutional dynamic network
The metabolic CDN module is introduced in Fig. 2 and S7. CDN "Y" is composed of the constituents CC′, DC′, CD′ and DD′, where the components of DD′ are modi ed with lactate dehydrogenase (LDH) and nicotinamide adenine dinucleotide (NAD + ), Figs. S2, S8. The metabolic biocatalytic transformation proceeding in CDN "Y" involves the LDH-biocatalyzed reduction of NAD + to NADH by lactic acid, and the concomitant formation of pyruvic acid. The biocatalyzed formation of NADH is coupled to the secondary reduction of methylene blue (MB + ) to colorless MBH, a process that allows the spectroscopy readout of the time-dependent formation of NADH (Fig. 2a, panel I). In addition, the biocatalyzed formation of NADH and pyruvic acid is coupled to the secondary reductive amination of pyruvic acid, in the presence of NH 4 + and alanine dehydrogenase (AlaDH), to form L-alanine as metabolic product.
Subjecting CDN "Y" to trigger T 3 stabilizes constituent DD′, resulting in the recon guration of CDN "Y" to CDN Y a , where DD′ is up-regulated, DC′ and CD′ are down-regulated and CC′ is up-regulated (Figs. S9-11).
The reverse treatment of CDN Y a with T 3 ′ displaces T 3 from DD′ and resulting in the regeneration of CDN "Y". In addition, treatment of CDN "Y" with T 4 stabilizes the constituent CD′, leading to the recon guration of CDN "Y" to CDN Y b , where CD′ and DC′ are up-regulated and the constituents CC′ and DD′ are downregulated. Fig. 2b and Table S2 show the concentrations of the constituents. The CDNs-guided timedependent operation of the biocatalytic cascade corresponding to LDH-catalyzed reduction of NAD + by lactic acid to NADH, and the subsequent reduction of MB + (l = 630 nm) are presented in Fig. 2c and S12-13. The time-dependent depletion of MB + to MBH is enhanced, in the presence of the T 3 -triggered recon gured CDN Y a , and retreated in the presence of the T 4 -recon gured CDN Y b , consistent with the upregulation of the constituent DD′ in CDN Y a and the down-regulation of DD′ in CDN Y b , respectively. In addition, Fig. 2d presents the time-dependent CDNs-driven metabolic cascade, where the LDHbiocatalyzed reduction of NAD + to NADH by lactate is followed by the AlaDH-catalyzed amination of the generated pyruvic acid to yield L-alanine (Fig. S14, Table S3-S5). The rate of formation of L-alanine metabolite is enhanced in the presence of CDN Y a and dampened by CDN Y b , respectively.

Intercommunicated photosynthetic and metabolic dynamic networks
In the next step, efforts to couple photosynthetic module and the metabolic assimilation module were undertaken, similarly to the interlinked processes in plants. The principle to intercommunicate between the two modules is displayed in Fig. 3a. The constituents BB′ and CC′ in CDNs "X" and "Y" were preengineered to include each extra Mg 2+ -dependent DNAzyme units. These units are termed "activators", integrated into the composite in order to intercommunicate between the networks. To intercommunicate between the networks, we added two hairpins, H a or H d into the mixture of two CDNs. The H a is designed to be cleaved by the activator associated with constituent CC′ to yield the fragmented strand H a-1 that In the rst step, the unidirectional intercommunication between the networks using hairpin H a or H d was evaluated. Fig. 3b and S21 show the formation of NADPH by the photosynthetic module upon exposure to the CDN "Y"-synthesized H a-1 at different time-intervals. As the time-interval is prolonged, the photosensitized generation of NADPH by the photosynthetic CDN "X"/FNR is enhanced, consistent with the continuous enrichment of the constituent AA′ by H a-1 . Fig. 3c depicts the metabolic performance of CDN "Y" before and after the strand H a-1 was supplied to CDN "X", by following the reduction of MB + to MBH. As expected, the metabolic module is unaffected upon supplying H a-1 as a trigger to CDN "X". Fig.   3d shows the rate of synthesis of L-alanine by CDN "Y" upon feeding the two CDNs with H d-1 generated at different time-intervals (Fig. S22, Tables S6-S9). As the time interval of the generation of H d-1 is prolonged, the synthesis of L-alanine is enhanced, consistent with the stabilization and time-dependent overexpression of DD′, in the presence of H d-1 . Fig. 3e shows the spectra of NADPH generated by the photosynthetic module (1 h-irradiation) before and after the generation of H d-1 , implying CDN "X" is unaffected upon the transfer of the information strand (H d-1 ) to CDN "Y". Noted that the discussion introduced the positive intercommunication "dialog" between the CDNs. One may envisage, however, the negative intercommunication between the CDNs. For example, subjecting coupled CDNs to hairpin H n (cleaved by the activator of BB′) leads to the generation of fragmented product H n-1 that stabilizes CD′, the down-regulation of DD′ proceeds, resulting in the inhibition of the metabolic module (Figs. S23-26).
Subjecting the mixture of the two CDNs to the two hairpin has, however, a signi cant effect on the intercommunication between the two CDNs (Fig. 4a). The cleavage of the hairpin H a yields the strand H a-1 that provides the information to up-regulate AA′ in the photosynthetic module, thus enhancing the photosynthetic module leading to the time-dependent increase in the photosensitized electron transfer process and the FNR-catalyzed synthesis of NADPH. The up-regulation of AA′ is accompanied by the upregulation of BB′ that leads to the time-dependent enhancement of the cleavage of H d to form H d-1 . The latter product provides the information strand to enhance the metabolic module synthesizing L-alanine.
The stabilization and up-regulation of DD′ is accompanied by the up-regulation of CC′ and, thus, the further enhancement of the cleavage of H a and the enhancement of the photosynthetic module. In the presence of the hairpin H a and H b , a positive feedback mechanism intercommunicating the CDNs is established (Figs. S27-29). The time-dependent increase in the performance of the photosynthetic model is re ected by an information transfer to the metabolic CDN module to enhance its activity and visa versa. The control over the concentrations of the constituents by the two hairpins is re ected in the photosynthetic and metabolic processes occurring in CDNs "X" and "Y". Fig. 4b and S30 show the photosensitized-NADPH generated at time-intervals of the feedback-driven intercommunication of the two networks, indicating the generation of NADPH by the photosynthetic module is enhanced. Fig. 4c depicts the rates of the metabolism (lactate/LDH/AlaDH/MB + cascade) at time-intervals of the intercommunication between the networks. As the feedback process enriches AA′ and DD′, the concentration of NADPH is higher and the biocatalytic cascade is enhanced. The results demonstrate a tight relation between the photosynthetic module and the assimilation, metabolic module. The enhancement of the photosynthetic process channels the information to enhance the metabolic path, and the enhanced metabolic reactions are translated into transfer of information for enhanced photosynthesis.

Conclusions
The study introduces two complementary dynamic networks that mimic the functions of plants-an "arti cial leaf". One network introduces photosynthetic path, where the control over the light-induced electron transfer and the subsequent catalyzed synthesis of NADPH proceeds, in analogy to the transformation driven by photosystem I. A second dynamic network demonstrates a metabolic path mimicking plant assimilation and mitochondrial metabolism by the input-driven oxidation of lactate and its metabolic transformation to L-alanine. The two networks are intercommunicated by demonstrating the guided activation of the metabolic network by the photosynthetic network and the counter control over the photosynthetic network by means of the metabolic network. Finally, the integrated feedback-driven operation of the photosynthetic network and the metabolic network is established by introducing the coupled "leaf-like" operation of the two networks, where "information transfer" between the two networks exists. The photosynthetic network enhances the activity of the metabolic network, and the activity of the metabolic network guides transfer of information for enhancing the photosynthetic network. Beyond highlighting the integration of the photosynthetic and metabolic networks in leaf-mimicking module, the study introduces an important path to guide the synthesis of useful materials (L-alanine) by the metabolic network 41 .
For the triggered transition of CDN X, triggers T 1 , T 1 ′ or T 2 , T 2 ′ are 1.67-fold excess than each component of CDN. After adding triggers into initial CDN, the nal concentration of each component of CDN was 1 µM and the nal concentration of trigger was 1.67 µM. The solution was incubated at 28°C overnight to equilibrate. For the triggered transition of CDN Y, triggers T 3 , T 3 ′ or T 4 , T 4 ′ are 2.5-fold excess than each component of CDN. After adding triggers into initial CDN, the nal concentration of each component of CDN was 1 µM and the nal concentration of trigger was 2.5 µM. The solution was incubated at 28°C overnight to equilibrate. After equilibration, the equilibrated CDN (each component 1 µM) was treated with one substrate (5 µM) (sub 1 for AA′, sub 2 for BB′, sub 3 for BA′, sub 4 for AB′, sub 5 for DC′, sub 6 for CD′, sub 7 for CC′ and sub 8 for DD′). The time-dependent uorescence changes generated by the cleavage of the different substrates by DNAzyme reporter units were measured. By following the rate of formation of the uorophore-labeled fragment and using appropriate calibration curves of the intact constituent (Figs. S4-S5 and S9-S10), the quantitative evaluation of the concentrations of constituents is achieved.    Integrated feedback-driven intercommunicating networks X and Y using the mixture of stimulants Ha and Hd. A) Subjecting the mixture M to the two stimulants Ha and Hd results in the Ha-triggered time-