Toward standardized solid medium cultivations: Online microbial monitoring based on respiration activity

Cultivating microorganisms on solid agar media is a fundamental technique in microbiology and other related disciplines. For the evaluation, most often, a subjective visual examination is performed. Crucial information, such as metabolic activity, is not assessed. Thus, time‐resolved monitoring of the respiration activity in agar cultivations is presented to provide additional insightful data on the metabolism. A modified version of the Respiration Activity MOnitoring System (RAMOS) was used to determine area‐specific oxygen and carbon dioxide transfer rates and the resulting respiratory quotients of agar cultivations. Therewith, information on growth, substrate consumption, and product formation was obtained. The validity of the presented method was tested for different prokaryotic and eukaryotic organisms on agar, such as Escherichia coli BL21, Pseudomonas putida KT2440, Streptomyces coelicolor A3(2), Saccharomyces cerevisiae WT, Pichia pastoris WT, and Trichoderma reesei RUT‐C30. Furthermore, it is showcased that several potential applications, including the determination of colony forming units, antibiotic diffusion tests, quality control for spore production or for pre‐cultures and media optimization, can be quantitatively evaluated by interpretation of the respiration activity.

the same time, acetate is produced in the agar-facing center. The produced acetate is cross-fed to the upper gas-facing cell layers. [5] These studies from the literature are essential to understand the evolution of single colonies. However, they are not designed for mass balancing solid media cultivations as a whole. The fluorophore used by Tschiersch et al. (2012) for the sensor foil is limited to the detection of oxygen and is not broadly applicable for regular respiration quantification in agar cultivations. [4] Carbon dioxide cannot be monitored, which negates an application for, for example, anaerobic cultivations.
For cultivations in liquid media, the gas consumption and production are typically monitored in the off-gas of the respective cultivation vessel. In Erlenmeyer flasks, the monitoring can be conducted, for example, with the Circulation Direct Monitoring and Sampling System (CDMSS), the BCpreFerm system (BlueSens, Herten, Germany), the Transfer-rate Online Measurement (TOM) system (Kuhner Shaker, Herzogenrath, Germany), or the Respiration Activity MOnitoring System (RAMOS). [6][7][8] The RAMOS monitors the changes in the partial pressure of a gas within a defined cultivation vessel's gas phase. The changes in the partial pressures allow for the calculation of mass transfer rates between the gas and the liquid phase for the respective gases.
The RAMOS was continuously adapted to allow for the online monitoring of several gases, such as oxygen, carbon dioxide, and ethylene, as well as total gas production. [8][9][10] The system was extensively utilized for cultivations in liquid media while correlating gas transfer rates with substrate uptake, growth patterns, and product formation. [11][12][13] To the best of our knowledge, the RAMOS technology has never been used for respiration monitoring of cultivations on solid media.
Potential reasons for this could be lower gas consumption and production rate, in contrast to liquid cultivations, as well as diffusion gradients forming in the gas phase, leading to inaccurate measurements. A circulation of the gas phase, as introduced by Takahashi et al. (2017) for the CDMSS, could prevent formation of gradients. [6] Hence, in a proof-ofconcept, this work showcases that a modified version of the RAMOS technology with gas circulation by a microfluidic piezo-membrane pump enables the respiration monitoring of agar cultivations. Furthermore, the system allows for reliable measurements even for cultures forming large films or lawns, at which conventional optical methods are less applicable. To highlight potential application scenarios, four different exemplary use cases are demonstrated in this study ( Figure 1): the respiration monitoring of agar cultivations enables the determination of colony forming units ( Figure 1A), the evaluation of antibiotic disk diffusion tests ( Figure 1B), quality control for spore production or for pre-cultures ( Figure 1C), and an assessment of the metabolic activity ( Figure 1D). In addition, references from literature are given to outline how respiration monitoring can benefit research using solid media.

Microorganisms
The applicability of respiration monitoring for agar cultivations was  Karlsruhe, Germany). [14,15] The mineral medium was supplemented with 1 g L −1 glucose, 2 g L −1 lactose, and 4 g L −1 glycerol as carbon source, and the pH was adjusted to a pH of 7.4 with 5 M NaOH.

Media composition and cultivation procedures
Experiments for S. coelicolor were conducted with soy flour mannitol agar (SFM-agar). The SFM-agar consisted of 20 g L −1 agar, 20 g L −1 soy flour (Schoenenberger, Magstadt, Germany), and 20 g L −1 mannitol (Carl Roth, Karlsruhe, Germany). Yeast extract peptone dextrose agar (YPD-agar) consisted of 20 g L −1 peptone (Tryptic digest from casein, Carl Roth, Karlsruhe, Germany), 20 g L −1 glucose (Carl Roth, Karlsruhe, Germany), 15 g L -1 agar as well as 10 g L −1 yeast extract (Carl Roth, Karlsruhe, Germany) and was used for cultivations with S. cerevisiae and P. pastoris. For T. reesei cultivations, 39 g L −1 of potato dextrose agar (PD-agar) (Carl Roth, Karlsruhe, Germany) powder was used. The pH of the complex media was not adjusted. All media were autoclaved at 121 • C for 20 min. After autoclaving, the agar medium was liquid.
While liquid, 20 mL of the respective agar medium was carefully pipetted into 250 mL Erlenmeyer flasks. Occurring bubbles were removed by manual punctuation with a sterile needle to prevent an uneven agar surface. The agar was cooled at room temperature and, therefore, solidified. For the inoculation, 200 µL of the respective organism stock solution was spread on the solidified agar with a Drigalski spatula.
For the optical density (OD) corresponding to the biomass within a F I G U R E 1 Exemplary fields of application, in which respiration monitoring of agar cultivations may be advantageous. (A) Oxygen uptake increases with the viable biomass and, thereby, the colony forming units in a sample. (B) Oxygen uptake decreases with the antibiotic concentration due to the diffusion of the respective antibiotic into the adjacent agar (red zone). (C) Oxygen uptake patterns can be utilized as a quality indicator to secure reproducible cultivations, for example, for the production of spores on solid medium. (D) The ratio of oxygen uptake and carbon dioxide evolution can indicate formation of certain products (green zone).
sample, the absorbance of the inoculum was measured at a wavelength of 600 nm and is stated in arbitrary units (a.u.). These measurements were either conducted with a Genesys 20 photometer (Thermo Scientific, Darmstadt, Germany) or with an Ultraspec 2100 UV-Visible spectrophotometer (biochrom, Holiston, USA). Cultivations for E. coli were performed at 30, 33.5, and 37 • C. All other organisms were cultivated at 30 • C.
After the cultivations were conducted, water was added to the agar within the Erlenmeyer flasks. The Erlenmeyer flasks were then autoclaved at 121 • C for 20 min to inactivate the biological material. The temperature increase resulted in liquifying of the agar and mixing with the water. Due to the reduced agar concentration, the mixture did not solidify any more after cooling, and the Erlenmeyer flasks were cleaned similarly to liquid cultivations. For the agar diffusion tests, Whatman No 1 filter paper (Whatman plc, Little Chalfont, UK) was stamped to receive filter disks with a diameter of 6 mm. The antibiotic tetracycline (Fluka, Buchs, Switzerland) was dissolved with 10 g L −1 in water, sterile filtered and a tetracycline dilution series was prepared. The required amount of tetracycline was applied by pipetting 10 µL of the respective concentration from the dilution series on top of the filter disk. For comparison, the agar diffusion tests were also perfomed in Petri dishes without respiration monitoring, which were sealed with parafilm (Bemis, Neenah, USA).

Respiratory activity monitoring of agar cultivations
All cultivations were monitored with modified versions of the RAMOS depicted in Figure S1. [7,8]  Oxygen and carbon dioxide partial pressure Δp i Δt changes were fitted linearly during the measurement phase. Thereby, with the available information about the total gas phase volume V G , the agar surface area A, the temperature T, and the gas constant R, the agar area-specific oxygen transfer rate (OTR′) and carbon dioxide transfer rate (CTR′) could be calculated according to Equations (1) and (2). The agar area was assumed to be equal to the inner area of the Erlenmeyer flask with 51.9 cm 2 . In contrast, the respiration of cultivations with liquid medium is calculated, as usual, as the volume-specific oxygen and carbon dioxide transfer rates. [7,8] The respiratory quotient (RQ) was quantified by dividing the CTR′ by the OTR′ according to Equation (3). The measurement technique was described in detail by Anderlei et al. (2001Anderlei et al. ( , 2004). [7,8] The OTR′ of agar cultivations was also monitored in 48-round well . [17] Additionally, oxygen sensor spots (PyroScience, Aachen, Germany) were utilized in combination with the RAMOS. Thereby, the potential heterogeneity in the gas phase and formation of an oxygen partial pressure gradient above the agar cultivations was studied. Sensor spots were glued to the Erlenmeyer flask wall at 2, 4, and 6 cm above the agar surface ( Figure S2). The oxygen partial pressure was measured after a 2-point calibration with a fiber optic multiple analyte meter FireSting PRO (PyroScience, Aachen, Germany). The oxygen partial pressure data was used to calculate the OTR′ according to Equation (1) during the measurement phase of the RAMOS.

Respiration monitoring for cultivations on agar medium allows tracking of the growth of Escherichia coli colonies
Gas uptake and evolution can lead to gradient formation of oxygen and carbon dioxide concentration above the cultures on solid media. These concentration gradients may result in an inaccurate determination of the real gas transfer rates. Hence, a homogenous gas phase is required for a successful respiration monitoring of microbial growth on agar.
In the applied RAMOS device, a convective gas flow is generated by a piezo-membrane pump, which circulates the gas phase of the Erlenmeyer flasks through a loop with oxygen, carbon dioxide, and pressure sensors. To prove the successful homogenization of the gas phase, respiration measurements were conducted at different heights above the LB-agar during the cultivation of E. coli (Figure 2, Figures S2 and S3).
Thereby, it was assessed, if insufficient gas homogenization occurs and, thus, influences the measurement. The determined course of the oxygen partial pressure values over time of the cultivation can be found in Figure S3A. Despite a slight calibration offset, the overall course of all measurements is highly similar. The measurements did not deviate by more than 1.7 hPa from each other in the last hour, when respiratory activity ceased. No influence of the measurement position on the online signal of the oxygen partial pressures could be noticed ( Figure S3A, red, blue, and green lines). The measurement with the electrochemical sensor in the RAMOS was less noisy compared to the optical sensors. Over a 5 min period without respiratory activity, the signal standard deviation was ±0.02 hPa for the electrochemical sensor. In contrast, the deviation for the optical sensors was ±0.11, ±0.10, and ± 0.25 hPa (2, 4, and 6 cm). The spikes occurring in regular intervals are due to the measurement principle of the RAMOS. [7,8] The gas phase in the Erlenmeyer flask represents a closed system during the measurement phase by closing the in-and outlet valves. Therefore, depending on the oxygen uptake of the E. coli cultivation, the oxygen partial pressure decreases more or less rapidly. With this decrease, the OTR′ can be determined ( Figure S3B Figure 2A) results from the noise of the oxygen partial pressure signal of the sensor spots. [18] The resulting OTR′ signals start increasing after approximately 3 h, due to the growth and colony formation of E. coli (Figure 2A). After a rapid exponential increase, a peak value of 36.0 ± 3.1 mmol h −1 m −2 is reached after 6.9 ± 0.4 h. Thereafter, the OTR′ steadily declines and reaches almost 0 mmol h -1 m −2 at 24 h. However, these findings solely indicate that the respiration monitoring is independent of the measurement height. For a further validation, a visual comparison with the biomass formation was conducted.
The observable increase in the OTR′ ( Figure 3A) over time is caused by biomass growth resulting in colony formation. To prove this, E. coli cultivations, conducted on LB-agar in Erlenmeyer flasks, were additionally monitored by recording a video ( Figure 3B). The video can be found in the supplementary material (Video S1), while representative frames during the cultivation are depicted in Figure 3C. Colony formation becomes noticeable once the OTR′ increases exponentially.
The OTR′ reaches a peak at around 14 h, and subsequently, the OTR′ F I G U R E 2 Comparison of the (A) area-specific oxygen transfer rates area-specific oxygen transfer rate (OTR′) of an Escherichia coli BL21 cultivation on a solid lysogeny broth agar (LB-agar) medium monitored with the (B) Respiratory Activity MOnitoring System (RAMOS) and optical oxygen sensor spots. The RAMOS is equipped with a gas circulation pump. Further details are depicted in Figure S1 and described in Section 2. Sensor spots were glued to the inner flask wall at a height of 2 (green triangles), 4 (blue circles), and 6 cm (red squares). Culture conditions: non-shaken 250 mL Erlenmeyer flasks, V L = 20 mL, T = 37 • C, V X,0 = 200 µL with optical density (OD) of 10 −1 .

F I G U R E 3
Cultivation of Escherichia coli BL21 on lysogeny broth agar (LB-agar) in Erlenmeyer flasks. Two parallel agar cultures were monitored with the Respiratory Activity MOnitoring System (RAMOS). (A) Area-specific oxygen transfer rates area-specific oxygen transfer rate (OTR′). (B) A time-lapse video was generated by taking one picture every 10 min of the cultivations with a GoPro HERO4 camera from below. (C) Pictures at 8, 10, 12, 14, 16, and 18 h (colored in the same way as the arrows in (A)) are depicted for the respective parallel measurements (filled and empty square symbols). The contrast of the pictures was increased by 40% for a better visualization of colony formation. The video with original footage can be found in the supplementary files (Video S1). The round, gray and red cap of the RAMOS flasks (black arrows in (C)) can be seen in the background of the pictures. Culture conditions: non-shaken 250 mL Erlenmeyer flasks, V L = 20 mL, T = 33.5 • C, V X,0 = 200 µL with optical density (OD) of 10 −3 , duplicate. steadily decreases. During the decreasing phase, the size of colonies is observed to remain similar. These observations can be explained by the competition for nutrients on the agar, which also depends on the initial inoculum density and the inoculum distribution on the agar. The inoculum density defines how many colonies are formed, whereas the distribution relates to the available space on the agar and, thus, the available nutrients. In contrast to liquid cultivations, in which mixing provides homogenous distribution of nutrients, in agar cultivations, nutrient heterogeneities occur from the beginning. Here, the growth is dependent on the nutrient diffusion in the agar. Neighboring colonies lead to a decreased nutrient availability and therefore, the colony sizes vary depending on the spatial proximity to other colonies. Similar colony size responses to other colonies' spatial proximity are reported in the literature. [19,20] This effect is represented in the respiratory activity and is further illustrated in 3.2.

The inoculum density of Escherichia coli defines the trajectory of the respiratory activity
The viable cell count or correlating colony forming units in a sample are crucial parameters for microbiological methods such as antimicrobial susceptibility testing. [21] Standard procedures to determine the cell count include plating a sample dilution on agar plates and counting the formed colonies. [22] However, not all organisms form defined colonies and a suitable sample dilution is necessary to allow colony counting.  [23] An increase in the colony count leads to decreasing space between the colonies and, therefore, nutrients in the surrounding of the colonies on the agar. [24] The radial growth of each colony depends on available nutrients. Fewer colonies can grow larger in size ( Figure 4A). This prolonged radial growth of colonies is represented by a prolongation of the OTR′ peak. In contrast to these cultivations on agar, the effect of spatial nutrient limitations does not occur in liquid cultivations. In liquid cultivations, up to a scale of several dozen cubic meters, unicellular organisms experience homogenous environmental conditions due to good mixing. Hence, for liquid cultivations of E. coli in LB-medium, the variation of the inoculum density resulted in almost no change of the OTR trajectory ( Figure S4). For a standard determination of colony forming units between 30 and 300 colonies of E. coli are optimal for manual counting on a Petri dish of 8 cm in diameter. [25,26] A broader range for colony forming units is countable with the respiratory method, presented in this work, while additionally providing an objective and quantitative evaluation parameter. For an increased throughput, it was successfully tested, whether the method could also be applied to MTP ( Figure S5). The test indicated an excellent transferability (R 2 = 1.00) even up to 10 −6 a.u. OD of the inoculum. Thereby, labor for plate preparation and manual counting can be reduced. The evaluation of quantitative respiration parameters, such as the OTR′, averts also decisions on whether and how to count neighboring partially overlapping colonies. However, further tests with different organisms should be conducted comparing respiration monitoring with automated counting techniques. It has to be examined, whether the proposed method is beneficial in conditions, where visual evaluation is challenging. More complex morphologies of the tested organisms such as filaments and, particularly, hyphal growth into the agar media, can hardly be observed visually. This is the case for many Streptomyces sp., of which the model representative S. coelicolor was examined in Section 3.4.

Respiration monitoring allows for the evaluation of agar diffusion tests
Agar diffusion tests are most commonly used to examine the susceptibility of bacteria to an antibiotic substance. Despite of the continued development of methods for antibiotic susceptibility testing over the years, the antibiotic disk diffusion method by Bauer and Kirby remains the gold standard. [27,28] Similar to colony counting, automated image analysis systems are utilized for the evaluation of inhibition zones forming around the disk containing the antibiotic. [29] Hence, similarly, a respiratory evaluation can be applied, if an optical analysis is difficult or even impossible. Furthermore, the information on the respiratory activity could indicate the antibiotic's mechanism of action. For example, whether an antibiotic is bacteriostatic or bactericidal.
An exemplary antibiotic disk diffusion test was conducted with E. coli and the antibiotic tetracycline ( Figure 5). Tetracyclines exert a bacteriostatic effect by inhibiting bacterial protein synthesis. This is achieved by blockage of the ribosomal acceptor site for aminoacyl-tRNA. [30] The amount of tested tetracycline was chosen to range up to 30 µg, applied to the respective filter disks. Furthermore, the tests were simultaneously performed in Petri dishes and Erlenmeyer flasks to evaluate, whether there was a difference due to the cultivation vessel and the active aeration in the RAMOS device. The same volume of LB-agar was used in both cultivation vessels.
The visual examination of the bacterial lawn and inhibition zones in the Erlenmeyer flask and Petri dish ( Figure 5A) indicates differences.
The bacterial lawn in the Erlenmeyer flask is homogenous compared to the Petri dish. Therefore, the circular shape of the inhibition zone is clearer presented. This could indicate better growth of E. coli in the Erlenmeyer flask with active aeration by the RAMOS device. However, the improved growth is not necessarily a result of active aeration.
The cross-sectional area in Petri dishes is slightly larger, compared to the Erlenmeyer flasks. Hence, the ratio of the nutrients to the growth area is improved for the cultivations in Erlenmeyer flasks, potentially leading to a homogeneous bacterial lawn. The different cross-sectional areas also affect to a certain extent the diffusion of tetracycline. As the volume of LB-agar was kept constant, the agar height is different, depending on the cross-sectional area. In Petri dishes, the agar height is lower than in the Erlenmeyer flasks, resulting in increased radial diffusion of tetracycline and, therefore, increased inhibition zones. The measured inhibition zones ( Figure 5C) for Petri dishes were slightly increased compared to the zones recorded for Erlenmeyer flasks.
Nonetheless, inhibition zones and the tetracycline amount on a logarithmic scale have a nearly perfect linear correlation (R 2 for Petri dishes = 1.00, R 2 for Erlenmeyer flasks = 0.99) for both cultivation vessels. A linear correlation of the logarithmic antibiotic amount and the size of the inhibition zone is well expected and described in the literature. [31] For the cultivations in Erlenmeyer flasks, the OTR′ was additionally  Figure 5B, red line and filled circles) did not follow the general trend and indicated an even lower peak OTR′ than for the condition with 3 µg tetracycline. The lower peak OTR′ was not an erroneous measurement, but resulted from a moved filter disk during the setup of the experiment. Due to the movement of the disc, tetracycline was further spread on the agar surface, leading to an enlarged inhibition area ( Figure S6). Therefore, the respiration measurement of this cultivation was not considered for the evaluation ( Figure 5C, gray upward triangle).
For comparing the respiratory activity with inhibition zones, the differences in the peak OTR′ of the conditions with tetracycline to the average of the condition without tetracycline (0 µg) are presented ( Figure 5C). The linear fit of this respiration data (R 2 = 0.97) yields a comparable result to the fit of the inhibition zones in Erlenmeyer flasks (R 2 = 0.99). The general deviating trend of the linear fits for inhibition zones and the difference in the peak OTR′ could indicate the bacteriostatic effect of tetracycline on E. coli. However, an antibiotic exhibiting a bactericidal effect needs to be tested for confirmation.

Respiration monitoring can assist in standardization of agar cultivations for pro-and eukaryotic microorganisms
Depending on the metabolism and growth type, different organisms should exhibit a characteristic respiratory pattern in agar cultivations. Hence, it is to be tested, how the respiratory signal changes and whether it could be used to standardize pre-cultivations or spore production protocols on agar. In addition, it is conceivable that contaminants attached to mycelia or hidden under produced spores could change the respiration trajectory, serving as a simple online quality indicator.
For these purposes, it was examined, whether the presented method can be applied for various representative pro-and eukaryotic microorganisms ( Figure 6). As prokaryotic organisms, P. putida and S. coelicolor were chosen. P. putida is of interest, due to its ability to oxidize glucose into gluconate and further into 2-ketogluconate within the periplasm. [32] This leads to an increased oxygen consumption and elevated OTR′ compared to E. coli cultures ( Figure 6A). [33] Thereby, even though for both cultivations LB-agar was used, the cultivations can be clearly distinguished from each other. Additionally, S. coelicolor was cultivated on SFM-agar. It is a gram-positive soil bacterium, which forms filamentous structures and is able to produce spores. [34] For the cultivation, spores were spread on the agar. Therefore, and due to the coelicolor. At first, a vegetative mycelium is formed, which can serve as the substrate for the aerial mycelium upon nutrient depletion. [35] However, also prolonged substrate availability due to a slow hydrolysis and consumption of the soy flour in the medium could cause the differences in the respiratory activity. Nonetheless, the two replicates indicated excellent reproducibility, which underlines the robustness of the presented method. The spore production by S. coelicolor and other spore-formers can be subject to potential contaminations remaining unnoticed below the spore lawn. Contaminations, especially with fast growing organisms prevalent in the laboratory, such as E. coli, are detrimental for the subsequent liquid cultivation process. However, these contaminants can be detected early on by deviating trends in the respiration activity, compared to the axenic reference ( Figure S7). The higher the initial biomass and the growth rate of the contaminant, in contrast to the spore-producer, the more pronounced the deviating trend.
As representative eukaryotic organisms, the yeasts S. cerevisiae and P. pastoris, as well as the fungus T. reesei, were examined ( Figure 6B). Again, the OTR′ over time indicates a unique trajectory for each studied organism. Especially noticeable is the formation of a declining plateau in the OTR′ at 24 h for the P. pastoris and T. reesei cultivations.
In liquid cultivations, a declining plateau shape in the OTR could be attributed to a secondary substrate limitation. [7] It has to be verified, whether this explanation also applies to these cultures on agar. At first glance, the late increase of the OTR′ for S. cerevisiae, which appears at 20 h, is peculiar. However, as S. cerevisiae can utilize carbon sources with the respiro-fermentative metabolism and can form and grow on ethanol, the simultaneous monitoring of the by-product carbon dioxide is necessary to fully understand the late increase in the OTR′. [36]

Additional monitoring of carbon dioxide transfer rates and respiratory quotient reveals substrate consumption and product formation phenomena
The addition of an infrared carbon dioxide sensor to the measurement loop of the RAMOS device allows for determining the CTR′ of the conducted agar cultivations (Figure 7 and Figures S8-S13). Thereby, the RQ can be calculated and, thus, the early respiro-fermentative activity of the Crabtree-positive S. cerevisiae can be quantified. [36] A rise in the CTR′ could be detected after 10 h and, therefore, approximately 10 h before the OTR′ increase (Figure 7). This indicates two phases with F I G U R E 7 Cultivation of Saccharomyces cerevisiae WT on yeast extract peptone dextrose agar (YPD-agar), depicted in Figure 6B, with area-specific oxygen transfer rate (OTR′) and carbon dioxide transfer rate (CTR′), as well as the resulting respiratory quotient (RQ). An RQ  [8,37] However, the two phases of ethanol production and consumption are more clearly distinguishable in liquid cultivations, whereas for agar cultivations both phases overlap. [8] This effect can be attributed to the distribution of cells on the agar and the associated heterogeneity within a colony, resulting in potentially multiple phenotypes. [5,38] Considering the different phenotypes and cross-feeding of products such as acetate or ethanol within a colony, as well as the fact that the respiration monitoring yields a sum signal of the whole cultivation, the observation of these phases in agar cultivations is noteworthy. [39] The respiration monitoring of agar cultivations does not allow for the differentiation of subpopulations and only summarizes the total gas consumption and production. However, depending on the synchronization of the colony growth, the data is useful for validating single colony respiration models.
In contrast to the Crabtree-positive S. cerevisiae, P. pastoris is regarded as a Crabtree-negative yeast. [40] Therefore, ethanol is only produced under oxygen limiting conditions. [41] Oxygen is increasingly limiting in the center of expanding colonies. [42] Accordingly, the CTR′ does not rise before the OTR′ ( Figure S10), in contrast to the S. cerevisiae cultivation (Figure 7). For P. pastoris, the initial growth and ethanol production are indicated by an RQ greater than 1. Afterwards, the RQ drops below 1 due to the sequential ethanol consumption. Interestingly, with the switch to ethanol consumption, no visual changes over time are visible anymore, indicating low biomass formation (Video S2).
The OTR′ and CTR′ remain at a high level for at least 24 h, which is not conceivable from visual evaluation.
The CTR′ and the resulting RQ also underline the observations regarding oxidation reactions in the P. putida cultivations ( Figure S8).
The RQ below 1 between approximately 5-26 h indicates the formation of an oxidized product, which may be metabolites, such as gluconate and 2-ketogluconate. [32] . After about 26 h, the OTR′ drops below the CTR′, suggesting the consumption of oxidized products.
Comparable to the S. cerevisiae cultivation, for P. putida, the observed trajectories fit respiratory and product formation data from liquid cultivations. [33,43] Similar to liquid cultivations, respiration monitoring of cultivations on solid medium could also assist in designing and optimizing solid medium compositions. To showcase this possibility, an E. coli cultivation was performed on a defined medium without complex components ( Figure S12). In contrast to the cultivation of E. coli on LB-agar ( Figure S13), a delayed and slower increase of the OTR′ and CTR′ can be noticed for the cultivation on synthetic Wilms-MOPS-agar. This effect can be attributed to the faster growth on complex components.

3.6
Further potential of respiration monitoring in solid medium cultivations In Section 3.4, S. coelicolor was briefly introduced as a spore-forming bacterium with a complex life cycle. However, in liquid media, specifically S. coelicolor does not form spores. In fact, most Streptomycetes do not form spores in liquid cultivations. [44] Instead, spores develop out of the aerial hyphae, by which they can escape nutrient depleted aqueous environments. [45] Upon depletion of nutrients, as well as due to other signals, the initial vegetative mycelium of Streptomyces undergoes programmed cell death, leading to the formation of aerial hyphae. [46] This is not the case in liquid cultivations. Hence, spore production and the differentiation-associated antibiotic production are frequently observed on solid media. [47] The proposed respiration monitoring could assist in these studies. The respiratory signal should indicate nutrient depletion and programmed cell death. Thereby, influencing factors on sporulation and, ultimately, antibiotic production could be investigated in more detail.
Additionally, the method can be applied to plant cell cultures or adherent mammalian cell cultures. For example, plant cell cultures were already cultivated in liquid media in the RAMOS device, while monitoring the oxygen and the plant cell hormone ethylene. [10] However, plant callus induction is mainly performed on solid media, as some callus cannot be induced in liquid media. [48] Hence, it is conceivable that the presented method is used to track the callus induction and the proliferation on solid media. Thereby, respiration could potentially serve as an online indicator for the induction rate and proliferation efficiency, which could help in selecting optimal culture conditions.

CONCLUSION
The introduced application of the RAMOS for cultivations on solid medium provides valuable information on the respiration activity. It was demonstrated that the respiration activity can indicate growth trajectories, metabolic state, and particular production and consumption phenomena of carbon sources. The OTR′ and CTR′ allow for a quantitative evaluation and comparison with cultivations in liquid media.
Gathering such data is crucial for mass balancing and is especially beneficial, if gaseous substrates and products are to be investigated.
Conclusions can be drawn, validating models of the respiration for single colonies. The respiratory method provides an online and objective process indicator, which can substitute subjective visual evaluations by the naked eye for agar cultivations. Furthermore, the method can assist and be used in combination with automated optical monitoring techniques. It was showcased that respiration monitoring is independent of the growth phenotype and can be applied for various pro-and eukaryotic organisms. In the future, a more detailed investigation of some application areas could be of interest. For example, it could be examined, whether the method can be used to evaluate a positive effector (e.g., disc soaked with an essential nutrient, which is limiting in the agar medium) in an agar diffusion test. Moreover, applying the method for adherent cell cultures could provide essential data for scaling these cultivations.

ACKNOWLEDGMENTS
The authors gratefully acknowledge support and funding from the Deutsche Forschungsgemeinschaft (priority program SPP2170, project no. 427899901). The strain Streptomyces coelicolor A3 (2) DSMZ40783 was kindly provided by the Hans-Knöll-Institute culture collection (Jena, Germany). The authors are thankful to Carl Brehl for his valuable input.
Open access funding enabled and organized by Projekt DEAL.