Photosynthetic hydrogen production by droplet-based microbial micro-reactors under aerobic conditions

The spontaneous self-assembly of multicellular ensembles into living materials with synergistic structure and function remains a considerable challenge in biotechnology and synthetic biology. Here, we exploit the aqueous two-phase separation of dextran-in-PEG emulsion micro-droplets for the capture, spatial organization and immobilization of algal cells or algal/bacterial cell communities to produce discrete multicellular spheroids capable of both aerobic (oxygen producing) and hypoxic (hydrogen producing) photosynthesis in daylight under air. We show that localized oxygen depletion results in hydrogen production from the core of the algal microscale reactor, and demonstrate that enhanced levels of hydrogen evolution can be achieved synergistically by spontaneously enclosing the photosynthetic cells within a shell of bacterial cells undergoing aerobic respiration. Our results highlight a promising droplet-based environmentally benign approach to dispersible photosynthetic microbial micro-reactors comprising segregated cellular micro-niches with dual functionality, and provide a step towards photobiological hydrogen production under aerobic conditions.

Overall this is a detailed and interesting study on a novel fabrication method using two-phase aqueous aqueous polymers to entrap algal cells (and other non-photosynthetic cells) into a relatively uniform (20-160 um) particles. Much of the paper especially the supplemental material addresses the methodological approaches such as shearing rpms, shrinking time, etc., Although these experiments do indicate the method is relatively well-controlled and although this is important, it seems this could have been published separately in a more methodological journal. In fact, I also think the paper would be complete and significant if it ended at Figure 4 and S19, thus omitting the E. coli cell inclusion work. Although quite well written and complete, there were also details that seem critical yet were not shown. For example, how denatured was the BSA from the heat treatment this could have been explored using CD, BN-PAGE, even tryptophan fluorescence. It was not even clear that the BSA needed to be denatured?
It was a bit disappointing to see that even after optimization, that hydrogen production stopped after ~72 hours and then the particles "dissolved" after five days. One would hope the "living" nature of the Chlorella cells would extend this activity longer than the in vitro PSI hydrogen evolution yields shown by others such as the Golbeck lab and in reference 26. These researchers have demonstrated higher yield 5.5 mmol H2/h/mg chlorophyll and a sustained activity for over 85 days? It would be very interesting to see the novel fabrication methods demonstrated here could be applied to include the incorporation of isolated PSI and a catalysis such as the platinum nanospheres or even a Ni/Fe hydrogenase. The integration of enzymatic oxygen scavenging methods (glucose oxidase and catalase) could also make this microsphere particle method compatible with more open environmental applications where anaerobiosis might not be feasible. Overall, it is an interesting and well done study, that may inspire other similar approaches in the field.
Minor points/suggestions: -On Figure 2-it was unclear if the Red fluorescence on Panel 2K was the Chl autofluorescence or the fluorescence from the Nile Red from the denatured BSA?
-It is also surprising that the algal cells seem somewhat clumped together and only on the interior? Is this representative or simply one observations. More images would be preferred or possibly the use of flow cytometry to measure the number and distribution using the Slit-scan feature -With the low cell density (6.9 x 107) there are only a few cells per droplet (I would guess ~20, yet this should be known with statistics using either LSCM or FACS) . However, when you make the droplets using high cell density (3.3 x109) the number of cells goes up to 9000. This is a very large increase 450 times more yet only a 50-fold increase in cell density? This suggests some type of aggregation or algal clustering? Although this has been somewhat systematically studied in the Fig. S3-S9, there are still somewhat inconsistent results. For example, Fig. 3 with Fig. S7. The hyperosmotic shrinking seems to be very different This would be interesting to explore more systematically.
-Assuming diameter of 171 um before hyperosmotic shrinking (~1,964,000 um3) but after the diameter has only reduced to 162 with a volume of 1,764,000 um3. This is only a 11% reduction not the 50% shown in Fig. 3b.
-Figures 3E and F seem redundant to those images shown in Figure 2 K and I.
--Explanation of large variation of size of their droplets? Is there an issue with reproducibility/controllability of the fabrication? --With the photosynthetic Chlorella cells in the interior of these micro-reactors, are there any shading effects that are noted, or heterogeneity in productivity going deeper into these droplets?
--I was hoping to get more of a broad context in the conclusion to get an idea of where to place this study in comparison to previous work on biomass production by whole cells.
-In trying to explain the lack of hydrogen production in the 22 um sized spheroids as compared to the 92 um, the authors claim that this is due to " to the increased number of algal cells in the aerobic surface regions". Since the particle size is smaller and the surface area much lower, I think the authors actually mean an increased percentage of Algal cells are in an aerobic environment and/or there is a drastic loss of cells in the anaerobic environment. The number of cells is certainly overall less (not more as stated) in the 22 um vs the 92 um.
-On page 12, It is also interesting that the hydrogen yield decreased when they go from 92-165 um, this may be due to the shading as suggested yet this might be a key area for further optimization. It might also be interesting to incorporate antennae reduced mutants to see if increasing the anaerobic environment while also increase light penetrance could further boost the hydrogen yield.

Responses to Reviewer #1:
This work describes the use of an aqueous two phases system-based microdroplets to culture algal cells or algal/bacterial cell communities to enable photosynthesis under daylight.
In particular, by co-culturing algal cells and bacterial cells in core-shell geometry, it is shown that hydrogen production can be enhanced. The work demonstrates a clever approach to develop aqueous two-phase system-based microniches with dual functionality to enable photobiological production of hydrogen. Overall, this is a timely work that combines a unique multiphasic system with synergism between two types of biological cells to achieve a potentially useful function of hydrogen production.

Response:
We appreciate the reviewer's positive and valuable comments on the scientific merit of our work.
Specific questions: 1. What is the efficiency of the system? How does it compared to other synthetic systems?
Response: The efficiency of the system is shown in two aspects: the preparation process of the microbial micro-reactors and hydrogen production rate of the system. For the former, during the preparation of the microbial spheroids, almost all the Chlorella or E. coli cells could be encapsulated into the droplets, and the preparation efficiency of our droplet-based microbial micro-reactors is therefore very high. Thus, this is a high throughput method and is suitable for preparing large quantities of microbial spheroids. As for the latter aspect, the hydrogen production rate of the microbial reactor system (Chlorella/E. coli hybrid spheroids) is 0.45 μmol H 2 (mg chlorophyll) -1 h -1 . In comparison with other biology-engineered algal cell-based hydrogen production systems, our micro-reactor system exhibited nearly 1.7-fold enhancement compared to that when Chlamydomonas reinhardtii Cr849 algal cells and Pseudomonas bacteria were co-cultured in sulfur-deprived TAP medium (TAP-S) (Int. J. Hydrogen Energy., 2013, 38, 10779-10787), and at least 1.3-fold enhancement compared to that when pure Chlamydomonas reinhardtii cc124 was cultured in TAP-S medium (Green Chem., 2014, 16, 4716-4727). In addition, our micro-reactor system showed 1.3 and 1.4-fold rate enhancements, respectively, when compared to systems based on silicification-induced algal cell aggregation (0.35 μmol H 2 (mg chlorophyll) -1 h -1 , Angew. Chem. Int. Ed., 2015, 54, 11961-11965), and enzyme-mediated anaerobic encapsulation (0.32 μmol H 2 (mg chlorophyll) -1 h -1 , Angew. Chem. Int. Ed., 2019, 58, 3992-3995).
2. Why do the algal cells segregate to the dextran-rich dispersed phase? Is there specific interactions between dextran and the cell? What happens if PEG-rich phase is the dispersed phase.

Response:
In agreement with our work, previous studies have shown that biomacromolecules and colloidal objects such as yeast cells and human/animal cells are strongly inclined to disperse in the dextran-enriched phase of demixed aqueous two-phase systems (ACS Macro Lett., 2017, 6, 679-683;Mater. Horiz., 2017, 4, 1196-1200. The phase selectivity has been generally attributed to the increased polarity of dextran compared with PEG as well as the branched vs linear/flexible molecular confirmations (Front. Chem., 2019, 7:44).
In response to the Reviewer's question, we undertook a new control experiment in which a dispersed PEG phase was used for the encapsulation of Chlorella cells, by interchanging the composition concentrations of dextran and PEG. In detail, the Chlorella cell dispersion (1.2 × 10 8 cells/mL, 120 μL) was added to 240 μL of the aqueous solution of PEG (120 μL, 160 mg/mL) and denatured BSA particles (120 μL, 3 wt%), and the resulting mixture slowly injected into 1.2 mL of 160 mg/mL dextran solution under stirring (100 rpm) for several minutes. Although PEG-in-dextran droplets could be prepared with BSA particles, the droplets were relatively small and extremely unstable such that they rapidly coalesced (new Figure S8). When Chlorella cells were encapsulated into the PEG phase, the droplets fused with each other immediately and released the Chlorella cells into the dextran phase; in contrast, both the empty and cell-loaded dextran-in-PEG droplets were much more stable ( Figure S4 and Figure S7).
The following text has been added on page 6 of the revised manuscript: "In contrast, no stable cell-loaded emulsion droplets were produced when PEG was used as the loading phase ( Figure S8)." The following new Figure has been added as Figure S8 in the SI.
New Figure S8. Time sequences of optical microscope images of PEG-in-dextran droplets (a) and Chlorella-loaded PEG-in-dextran droplets (b) stabilized by BSA particles (3 wt%). Scale bars, 100 μm. The PEG-in-dextran droplets were relatively small and extremely unstable with respect to coalescence. Attempts to encapsulate the Chlorella cells within the PEG phase resulted in droplet fusion and release of the algal cells into the continuous dextran phase.
3. Could hydrogen production be enhanced by simply adding the bacterial cell to the external PEG-rich phase? Such an experiment must be run as a control to demonstrate that localization of bacteria at the water/water interface.
Response: Many thanks for the valuable suggestion. We performed control experiments in which the bacterial cells were added into the external PEG-rich phase. In the absence of the Chlorella spheroids, no hydrogen production was observed. On addition of the free bacterial cells into the external PEG-rich phase, hydrogen production was observed in the presence of Chlorella spheroids in the preformed dextran-enriched droplets. However, no enhanced hydrogen production was observed confirming that localization of the bacterial cells at the droplet interface by hyperosmotic shrinkage of assembled multicellular hybrid droplets coli cells (blue triangles), and Chlorella/E. coli multicellular spheroids (blue squares); corresponding decrease in oxygen content for the algal/bacterial cell spheroids is also shown (red squares). All samples were in sealed vials and exposed to daylight at an intensity of 100 μE m -2 s -1 . Error bars indicate standard deviations (n=3).
4. Does the PEGylated bacteria adsorb to the interface in the absence of algal cells in the core of the disperse phase?
Response: In the absence of algal cells, PEGylated E. coli did not strongly absorb at the interface of the two aqueous phases. A spatial distribution of Chlorella cells predominantly within the core domain and bacterial cells in the peripheral layer was only obtained when both cell types were mixed in the dextran. As the native Chlorella cells were preferentially concentrated in the dextran phase, crowding effects also facilitated the specific location of the PEG-modified E. coli at the w/w interface. 5. One critical shortcoming of this approach is that the microbial bioreactor becomes ineffective after 5 days. What are the potential strategies to make hydrogen production a continuous process? I think this is an extremely important point since a lot of the synthetic systems can operate continuously.
Response: Many thanks for the comments. In this study, because of the inherent proliferation of Chlorella cells which disrupt the anaerobic microenvironment in the spheroids, the synthesized microbial micro-reactors disassemble and stop hydrogen production after 3 days. In comparison with many reported synthetic or biology-engineered microorganism systems, our studies show that by spatially organizing the living cells we could switch from normal photosynthetic oxygen production to hydrogen production. Therefore, in terms of orchestrating the assembly of different living cells into multicellular ensembles, our approach displays excellent biocompatibility without impairing the viability of the assembled cells; we consider this an important breakthrough.
With regard to the reviewer's suggestion, we undertook further experiments to attempt to prolong hydrogen production by covalently fixing the formed multicellular ensembles to minimize cell proliferation and spheroid disassembly. To do so, aldehyde-functionalized dextran was specially synthesized and used as a cross-linker to further fix the formed multicellular spheroids in the dextran phase. As a consequence, hydrogen production in the optimal bioreactor system (Chlorella/E. coli hybrid spheroids) was successfully prolonged from 72 to 168 h.

Details of the new experiments have been now incorporated into the "Materials and
Methods" section in the SI labelled. The following text has been added on page 2 and page 4/5 of the SI: Page 2, SI: "… dextran (MW 70 kDa, Pharmacia), 4-formylbenzoic acid (Energy Chemical, 98%), N,N-dicyclohexylcarbodiimide (DCC, Energy Chemical), 4-(dimethylamino) pyridine (DMAP, Sigma)" Page 4/5, SI: "Synthesis of aldehyde-functionalized dextran (Dex-CHO). The functionalized dextran was prepared as follows. Mixtures of dextran (2 g, 0.0285 mmol, MW 70 kDa), 4-formylbenzoic acid (0.96 g, 6.4 mmol), and 4-(dimethylamino) pyridine (DMAP; 0.12 g, 0.984 mmol) were dissolved in 40 mL of dimethyl sulfoxide (DMSO), followed by the addition of N,N-dicyclohexylcarbodiimide (DCC; 1.2 g, 5.83 mmol). The system was stirred at room temperature for 18 h and then the impurity removed by filtration. The Dex-CHO product was obtained as a white solid after precipitation in a mixture of ethyl acetate and petroleum ether with a volume ratio of 1 : 9. The white solid was dissolved in the deionized water, and any insoluble impurities was removed by filtration, and then freeze-dried.

Construction of crosslinked robust Chlorella/E. coli multicellular spheroids.
Algal/PEGylated bacteria cell mixtures (OD 600 (E. coli) : OD 750 (Chlorella) = 1 : 80, 120 μL) were added to 360 μL of an aqueous solution of dextran (120 μL, 160 mg/mL), denatured BSA particles (120 μL, 3 wt%) and Dex-CHO (120 μL, 2.4 wt%). The w/w emulsion droplets were slowly transferred into a hyperosmotic PEG solution dissolved in 10 mM PBS (pH = 7.4, 50 % w/w, MW 2000 Da) and left for 20 min to induce simultaneously shrinking of the droplets, compaction of the entrapped Chlorella and E. coli cells as well as the crosslinking reaction between Dex-CHO and the BSA particles. The vials were left unstirred and the sedimented spheroids collected by carefully removing the supernatant followed by washing several times using DI water." The following revised texts have been added to the Abstract and on page 16 of the main manuscript: Abstract: "…Moreover, the duration of hydrogen production is further prolonged by covalent crosslinking of the droplet microenvironment."

Responses to Reviewer #2:
The study exploited utilization of water-in-water dextran-in-PEG emulsion micro-droplets for encapsulating microalga Chlorella pyrenoidosa or assembling Chlorella/E.coli for the photosynthetic production of hydrogen under air. I am not convinced that the study shows enough novelty for publication in Nature Communications. I have the following major concerns, and I recommend rejection of the manuscript in its current form.

Response:
We have addressed each of the Reviewer's comments (please see below). We disagree that the manuscript is not sufficiently novel for NCOMM. Indeed, reviewer 1 writes: "The work demonstrates a clever approach to develop aqueous two-phase system-based microniches with dual functionality to enable photobiological production of hydrogen.
Overall, this is a timely work that combines a unique multiphasic system with synergism between two types of biological cells to achieve a potentially useful function of hydrogen production." In addition, we now show that the duration of hydrogen production can be increased by further modifications of our procedures (please see response 5 above to comments from Reviewer 1).
In brief, the advances in knowledge relate to: (i) Spatial organization of two different populations of living cells: Recently, the emerging topic on living cell self-assembly is attracting much attention. However, the development of new techniques capable of orchestrating the assembly of living cells into multicellular ensembles with synergistic structure and function remains a considerable challenge. Our work exploits the aqueous two-phase separation of dextran-in-PEG emulsion micro-droplets and develops a new methodology for the capture, spatial organization and immobilization of algal cells or algal/bacterial cells, which then allow for the high throughput fabrication of photosynthetic microbial micro-reactors comprising spatially segregated communities of algal or algal/bacterial cells.
(ii) Functionality switching between single cell and organized cell colony: We demonstrate that cooperation between the respiration of E. coli and the physical barrier of the close-packed Chlorella cells allows a functionality switching of the single Chlorella cell colony from normal photosynthetic oxygen production to hydrogen production from the organized Chlorella cell colony. To the best of our knowledge, the demonstration of spatially dependent symbiosis in the constructed micro-niches of segregated communities of algal and non-photosynthetic bacterial cells has not been reported, especially with regard to their synergistic use to enhance hydrogen production.
We apologize for not citing the paper mentioned by the reviewer (Nature Communications, 7:12934, DOI: 10.1038/ncomms12934) in the original manuscript. The suggested paper is now cited in the main text as reference 9 (along with an additional Ref 10).
2. The authors claim that a major contribution of the current study is the spatial organization and immobilization of algal or algal/bacterial cells which creates anaerobic niche in the core of the micro-droplets and induces production of hydrogen. While the basic concept has been demonstrated in a previous study (Ref 24), the current study does not show any groundbreaking progress in innovation. Utilizing w/w immobilization technique improves the stability of the assembly of cells but disassembly occurs over time and completely disassembled after 5 days ( Figure S23).
Response: Although we agree that Tang et al (Angew. Chem. Int. Ed. 2015, 54, 11961-11965) reported hydrogen production in air from algal cells housed within a mineralized core-shell structure, their method employed inorganic silicification rather than the living assembly of multiple cell colonies in dynamic w/w emulsion droplets. In our opinion, the experimental systems are fundamentally different as the novelty of our conceptual approach is based on the spatial organization and functional switching of multicellular microbial communities that can operate synergistically. As summarized on page 18: "In conclusion, we developed a novel strategy for the preparation of populations of discrete microscale microbial reactors with spatial segregated multicellular micro-niches capable of aerobic or anaerobic photosynthesis at room temperature in air." This is not a concept applicable to the immobilization of algal cells in silica.
We have also undertaken addition experiments to prolong the duration of hydrogen production (please see comments to point 5 from Reviewer 1). Whilst we have not yet achieved long-term continuous production, our opinion is that further developments will be possible and we have made some suggestions for further work in the revised Conclusions.
But our principle focus is to demonstrate the proof-of-concept.

The following new text and references have been added to the Conclusions:
"Overall, our methodology provides a proof-of-principle for utilizing aqueous two-phase separated droplets as vectors for controlling algal cell organization and photosynthesis in synthetic micro-spaces. The procedure is facile and capable of high throughputs for modulating algal cell functionality towards hydrogen production without impairing the viability of the living cells. Moreover, it should be possible to combine our methodology with more complex bioengineering approaches involving sulfur deprivation, 24 genetically modified oxygen-tolerant [FeFe]-hydrogenases 25 or cellular surface modifications. 27 Compared with synthetic hydrogen producing systems, 30 the limited rates and yields in the multicellular spheroids remain challenging aspects of future work. In this regard, incorporating chemical-based hydrogen generating machinery 31, 32 or antennae-reduced mutants 33 into the algal cell spheroids could be promising strategies. More generally, our approach provides the possibility for modulating the functionality of other living cells; for example, the droplet-based microbial systems can be readily extended towards ethanol production via the programmed capture of large numbers of yeast cells within the multicellular spheroids ( Figure S32)." 3. Additionally, the hydrogen production stopped after 72 h (Figure 5h), and the authors did not have appropriate discussion of possible reasons.
Response: As this question was also raised by Reviewer 1, we include below a repeat of our response.
We undertook further experiments to attempt to prolong hydrogen production by covalently fixing the formed multicellular ensembles to minimize cell proliferation and spheroid disassembly. To do so, aldehyde-functionalized dextran was specially synthesized and used as a cross-linker to further fix the formed multicellular spheroids in the dextran phase. As a consequence, hydrogen production in the optimal bioreactor system (Chlorella/E. coli hybrid spheroids) was successfully prolonged from 72 to 168 h. ether with a volume ratio of 1 : 9. The white solid was dissolved in the deionized water, and any insoluble impurities was removed by filtration, and then freeze-dried.

Construction of crosslinked robust Chlorella/E. coli multicellular spheroids.
Algal/PEGylated bacteria cell mixtures (OD 600 (E. coli) : OD 750 (Chlorella) = 1 : 80, 120 μL) were added to 360 μL of an aqueous solution of dextran (120 μL, 160 mg/mL), denatured BSA particles (120 μL, 3 wt%) and Dex-CHO (120 μL, 2.4 wt%). The w/w emulsion droplets were slowly transferred into a hyperosmotic PEG solution dissolved in 10 mM PBS (pH = 7.4, 50 % w/w, MW 2000 Da) and left for 20 min to induce simultaneously shrinking of the droplets, compaction of the entrapped Chlorella and E. coli cells as well as the crosslinking reaction between Dex-CHO and the BSA particles. The vials were left unstirred and the sedimented spheroids collected by carefully removing the supernatant followed by washing several times using DI water." The following revised texts have been added to the Abstract and on page 16 of the main manuscript: Abstract: "…Moreover, the duration of hydrogen production is further prolonged by covalent crosslinking of the droplet microenvironment." Page 16: "As the Chlorella and Chlorella/E. coli micro-reactors were stabilized only by non-covalent interactions along with in situ compaction and hydrogelation of the denatured BSA microparticles during hyperosmotic compression, the spheroids disassembled over time as the cell colonies proliferated (Figure S25-S27). As a consequence, hydrogen production was terminated typically within 72 h even though typical levels of cell viability after disassembly at 108 h were 68 and 96% for E. coli and Chlorella cells, respectively ( Figure S28).
To prolong the lifetime of the hybrid bioreactor, we included an aldehyde-functionalized dextran (Dex-CHO) into the w/w dextran-rich emulsion droplets to chemically crosslink the BSA protein hydrogel in situ. The resulting spheroids were physically more robust and displayed extended periods of hydrogen production over 168 h (Figure S29-S31

Response:
We thank the reviewer for their thoughtful question. We agree that the energy used for hydrogen production must continue to come from the light source. Even though the Chlorella cells are tightly packed inside the formed spheroids, the peripheral cells do not shield all of the external light from reaching the core domain. Therefore, Chlorella cells located in the core still receive light illumination but at a relatively lower intensity, which then leads to the decrease of photosynthetic oxygen production. Thus, the anaerobic microenvironment inside the spheroids gradually emerges into enabling hydrogenase activation for hydrogen production.
To aid clarification, we have made the following revision in the main text on page 11: "We attributed the onset of an anaerobic micro-environment within the centre of the spheroids to partial shielding of the buried algal cells to the light source as well as restrictions in the diffusion of atmospheric oxygen into the core regions." 5. In Figure 5j, addition of DCMU inhibits PSII and also the hydrogen evolution, but the authors did not explain why. How much DCMU was added into the system? Response: DCMU is a type of photosynthetic inhibitor (Proc. Natl. Acad. Sci., 2013, 110, 7223-7228). Upon addition of DCMU, the electron transfer from Q A to Q B in PSⅡ is disrupted, thus terminating the hydrogen production of Chlorella spheroids or hybrid Chlorella/E. coli spheroids. We mention the generalities of this process in the main text on page 16, where we write "In both cases, addition of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) inhibited hydrogen production (Figure 5j), confirming that anaerobic photosynthesis in the closely packed algal cells could be readily curtailed by disruption of the PSII electron transfer pathway." As DCMU is often used in studies of photosynthesis, we do not feel that details of the intervention in the electron transfer chain are necessary. But we can add a reference if deemed necessary.  (Green Chem., 2014, 16, 4716-4727;Algal Res., 2017, 28, 161-171). Accordingly, the substrate for the respiration of E. coli in the Chlorella/E. coli hybrid system is most likely associated with the excreta from the Chlorella cells.

We have added the following experimental details in the
We have added the following sentence on page 15 and two new references in the main text: "Moreover, as excreted algal-derived organic matter can be used as a substrate for bacterial respiration in natural and co-cultured algal/bacterial communities, 28, 29 we speculated that nutrient exchange between the co-located cells would lead to an enhancement in photobiological production."

These observations have been included in the revised manuscript and SI.
We have added the following text on page 16: "As a consequence, hydrogen production was terminated typically within 72 h even though typical levels of cell viability after disassembly at 108 h were 68 and 96% for E. coli and Chlorella cells, respectively ( Figure S28 and what the cells at the core do and how they are connected. I feel a lot of detailed experiment-based analysis need to be done to answer these questions.

Response:
The Reviewer raises a number of very interesting questions that go far beyond the scope of the current work, which is principally focused on"..a methodology that offers a proof-of-principle for utilizing aqueous two-phase separated droplets as vectors for controlling algal cell organization and photosynthesis in synthetic micro-spaces and provides a step towards the bottom-up assembly of photobiological micro-reactors with multiple functionalities." (page 3). The mechanistic details being considered by the Reviewer could take years to complete, and will be incorporated into aspects of our future work

Responses to Reviewer #3:
Overall this is a detailed and interesting study on a novel fabrication method using two-phase aqueous polymers to entrap algal cells (and other non-photosynthetic cells) into a relatively uniform (20-160 um) particles. Much of the paper especially the supplemental material addresses the methodological approaches such as shearing rpms, shrinking time, etc., Although these experiments do indicate the method is relatively well-controlled and although this is important, it seems this could have been published separately in a more methodological journal. In fact, I also think the paper would be complete and significant if it ended at Figure 4 and S19, thus omitting the E. coli cell inclusion work. Although quite well written and complete, there were also details that seem critical yet were not shown. For example, how denatured was the BSA from the heat treatment this could have been explored using CD, BN-PAGE, even tryptophan fluorescence. It was not even clear that the BSA needed to be denatured?
Response: Many thanks for the reviewer's positive comments and the questions. As suggested by the reviewer, more details about the denatured BSA are now given in the revised manuscript.
The following text has been added on page 5: "Droplet coalescence was minimized by addition of specifically synthesized microparticles of denatured BSA (Figure S1-S3) to the dextran phase prior to emulsification (Figure S4)." The following text related to the methods characterizing denatured BSA have been added to the SI "Methods" (page 2): "Circular dichroism (CD) spectral measurements were conducted using a Chirascan Plus CD spectrometer (Applied Photophysics Ltd, Leatherhead, UK). Samples were diluted with DI water to concentrations of 0.6 mg/mL in quartz cuvettes of 1 mm pathlength. Fluorescence spectroscopy was performed by a fluorescence spectrophotometer (PerkinElmer, USA, LS 55). Samples were diluted to a protein concentration of 1 mg/mL, and recorded at an excitation wavelength of 295 nm and slit widths of 6.0 nm (Excitation slit) and 3.0 nm (Emission slit)."

The following new SI figures have been added:
New Figure S2. Denaturation of BSA after heating. CD spectra (a) and FTIR spectra (b) of BSA before and after heating at 90 o C for 20 h. The CD results indicate that BSA before heating contains 67.7% α-helix, 2.8% β-sheet and 7.0% β-turn, while BSA after heating contains 34.9% α-helix, 20.3% β-sheet and 10.0% β-turn, indicating heat-induced BSA unfolding. This is in agreement with the FTIR spectra of BSA before and after heat treatment, which show Amide A, Amide I and Amide III peaks with different levels of blue shifts. The blue shift of the Amide It was a bit disappointing to see that even after optimization, that hydrogen production stopped after ~72 hours and then the particles "dissolved" after five days. One would hope the "living" nature of the Chlorella cells would extend this activity longer than the in vitro PSI hydrogen evolution yields shown by others such as the Golbeck lab and in reference 26.
These researchers have demonstrated higher yield 5.5 mmol H 2 /h/mg chlorophyll and a sustained activity for over 85 days? It would be very interesting to see the novel fabrication methods demonstrated here could be applied to include the incorporation of isolated PSI and a catalysis such as the platinum nanospheres or even a Ni/Fe hydrogenase. The integration of enzymatic oxygen scavenging methods (glucose oxidase and catalase) could also make this microsphere particle method compatible with more open environmental applications where anaerobiosis might not be feasible. Overall, it is an interesting and well done study, that may inspire other similar approaches in the field.

Response:
We thank the reviewer's comments, and also really appreciate the highly aspiring suggestions and perspective, which we will definitely consider in future work. As the reviewer mentioned, in comparison with the artificial synthesized hydrogen-producing systems such as by the integration of the platinum nanospheres with isolated PSI or hydrogenase etc, our studies are not currently competitive for long-term hydrogen production. But developing mild and high throughput methods based on the spatial arrangement of living cells could have advantages especially for the modulation of functionality of single cell colonies and mixed cell communities.
We have undertaken additional experiments on extending the duration of hydrogen production from 72 to 168 h by in situ cross-linking of the hybrid spheroids. This new data is included in the revised manuscript (please refer to the response above to the 5 th comment of the first reviewer). 2. It is also surprising that the algal cells seem somewhat clumped together and only on the interior? Is this representative or simply one observation. More images would be preferred or possibly the use of flow cytometry to measure the number and distribution using the Slit-scan feature.
Response: In general, very few aggregates were observed when relatively low numbers of cells were encapsulated (1.2 × 10 8 cells/mL, Figure 2f,g), while clumping occurred at higher loadings (3.3 × 10 9 cells/mL, Figure 2h,i). The images in Figure 2j-m were recorded at very low cell number density (6.9 × 10 7 cells/mL) so that individual cells could be visualized.
3. With the low cell density (6.9 x 10 7 ) there are only a few cells per droplet (I would guess ~20, yet this should be known with statistics using either LSCM or FACS). However, when you make the droplets using high cell density (3.3 x 10 9 ) the number of cells goes up to 9000. This is a very large increase 450 times more yet only a 50-fold increase in cell density? This suggests some type of aggregation or algal clustering?
Response: It was difficult to count accurately the cell numbers per droplet at medium and high loadings using LSCM because of aggregation within the interior. However, the loading efficiency of the algal cells was close to 1.0 (100%) (minimal numbers of cells were observed in the continuous phase); we therefore calculated the statistical average cell number per droplet (n) using the following equation: Where q was the encapsulation efficiency (1.0); c and V 1 were the cell density (cells/mL) and volume (mL) of used algal cells, respectively; V 2 was the volume of dextran phase (mL); r indicated the average radius of Chlorella-entrapped droplets (μm). The calculation gives n values of ca. 8700 and 180 for Chlorella-entrapped droplets prepared with cell densities of 3.3 x 10 9 or 6.9 x 10 7 cells/mL. Thus, the cell density was proportional to cell number per droplet, and appeared independent of preferential cell aggregation or clustering in the micro dextran phase.
We have included the above equation and following text on page 4 of the SI "Methods" section: "The statistical average cell number per droplet (n) was calculated using the following equation: where q was the encapsulation efficiency (ca. 1.0); c and V 1 were the cell density (cells/mL) and volume (mL) of used algal cells, respectively; V 2 was the volume of dextran phase (mL); and r the average radius of the Chlorella-entrapped droplets (μm). The calculation gave n values of ca. 8700 and 180 for Chlorella-entrapped droplets prepared with cell densities of 3.3 x 10 9 or 6.9 x 10 7 cells/mL, respectively." 4. Although this has been somewhat systematically studied in the Fig. S3-S9, there are still somewhat inconsistent results. For example, Fig. 3 with Fig. S7. The hyperosmotic shrinking seems to be very different. This would be interesting to explore more systematically.
Response: Many thanks for reviewer's careful reading and pointing this out. The original Figure S7 was a control experiment related to Figure 3. It was our negligence that there were some mistakes in the caption of old Figure S7. We have now revised the original description "Volume contraction of Chlorella-loaded w/w emulsion droplets after immersion in hyperosmotic PEG solution" to "Volume contraction of w/w emulsion droplets after immersion in hyperosmotic PEG solution" The legend has now been corrected (now Figure   S10).

5.
Assuming diameter of 171 um before hyperosmotic shrinking (~1,964,000 um 3 ) but after the diameter has only reduced to 162 with a volume of 1,764,000 um 3 . This is only a 11% reduction not the 50% shown in Fig. 3b. Response: The mean size of the water-in-water droplets was well controlled (from 180 to 20 μm) by varying the shear force (from 100 to 1000 rpm) such that the procedure showed good reproducibility in this regard. However, the polydispersity was relatively high, which was expected due to the conventional emulsification techniques employed. But our approach allows for a simple, fast and large scale preparation of the droplets. Alternatively, microfluidic fabrication could be used to decrease the polydispersity.

Response: As
8. With the photosynthetic Chlorella cells in the interior of these micro-reactors, are there any shading effects that are noted, or heterogeneity in productivity going deeper into these droplets?
Response: Many thanks for the insightful question. We have not investigated whether there are distinct gradients in hydrogen production due to shading effects.
9. I was hoping to get more of a broad context in the conclusion to get an idea of where to place this study in comparison to previous work on biomass production by whole cells.

Response:
The following text and references have now been added to the Conclusions section of the revised manuscript (page 19). "Overall, our methodology provides a proof-of-principle for utilizing aqueous two-phase separated droplets as vectors for controlling algal cell organization and photosynthesis in synthetic micro-spaces. The procedure is facile and capable of high throughputs for modulating algal cell functionality towards hydrogen production without impairing the viability of the living cells. Moreover, it should be possible to combine our methodology with more complex bioengineering approaches involving sulfur deprivation, 24 genetically modified oxygen-tolerant [FeFe]-hydrogenases 25 or cellular surface modifications. 27 Compared with synthetic hydrogen producing systems, 30 the limited rates and yields in the multicellular spheroids remain challenging aspects of future work. In this regard, incorporating chemical-based hydrogen generating machinery 31, 32 or antennae-reduced mutants 33 into the algal cell spheroids could be promising strategies. More generally, our approach provides the possibility for modulating the functionality of other living cells; for example, the droplet-based microbial systems can be readily extended towards ethanol production via the programmed capture of large numbers of yeast cells within the multicellular spheroids ( Figure S32)." 10. In trying to explain the lack of hydrogen production in the 22 um sized spheroids as compared to the 92 um, the authors claim that this is due to "to the increased number of algal cells in the aerobic surface regions". Since the particle size is smaller and the surface area much lower, I think the authors actually mean an increased percentage of Algal cells are in an aerobic environment and/or there is a drastic loss of cells in the anaerobic environment.
The number of cells is certainly overall less (not more as stated) in the 22 um vs the 92 um.
Response: We agree with the Reviewer and have revised the sentence on page 12 as follows: "… in agreement with the relatively high oxygen production under these conditions due to the increased percentage of algal cells in the aerobic surface regions (Figure 4a)." 11. On page 12, It is also interesting that the hydrogen yield decreased when they go from 92-165 um, this may be due to the shading as suggested yet this might be a key area for further optimization. It might also be interesting to incorporate antennae reduced mutants to see if increasing the anaerobic environment while also increase light penetrance could further boost the hydrogen yield.
Response: Many thanks for these interesting ideas.
We have added these points with references in the Conclusions (see revised text shown in response 9 above).