Oxygen-binding proteins aid oxygen diffusion to enhance fitness of a yeast model of multicellularity

Citation: Wong W, Bravo P, Yunker PJ, Ratcliff WC, Burnetti AJ (2025) Oxygen-binding proteins aid oxygen diffusion to enhance fitness of a yeast model of multicellularity. PLoS Biol 23(1):
e3002975.

https://doi.org/10.1371/journal.pbio.3002975

Academic Editor: Wenying Shou, Fred Hutchinson Cancer Research Center, UNITED STATES OF AMERICA

Received: February 2, 2024; Accepted: December 8, 2024; Published: January 30, 2025

Copyright: © 2025 Wong et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: Numerical data and simulation code can be accessed at the GitHub repository: https://github.com/Ratcliff-Lab/oxygen-binding-proteins_paper. An archive can be found at https://zenodo.org/records/14512540.

Funding: W.C.R. acknowledges support from the NIH (grant no. 5R35GM138030) and NSF (CAREER Grant no. 1845363), A.J.B. acknowledges support from the Human Frontiers Science Program grant (RGY0080/2020), W.W. and W.C.R acknowledges support from the NSF Division of Environmental Biology (grant no. DEB-1845363), and P.J.Y. acknowledges support from the NIH (grant no. 1R35GM138354). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

While molecular clock studies frequently place the origin of the metazoan clade prior to the Ediacaran period 635 million years ago [1], it is generally regarded as the period in which animal life diversified and became macroscopic [2]. Prior to this, most of the Proterozoic Eon (2.5 billion to 542 million years ago), bears relatively little trace of large complex multicellular life, with microbial mats and relatively simple multicellular algae inhabiting the oceans [3–9] alongside poorly understood microfossils that might represent relatives of the Metazoa or early sponges [10,11]. It has long been hypothesized that the dramatic rise in atmospheric oxygen levels marking the end of the Proterozoic, from approximately 1% of modern atmospheric levels to near-modern concentrations, was necessary for or triggered the evolution of animal multicellularity [12–14]. With more abundant oxygen, larger organisms would have been expected to be able to overcome diffusion limitations and achieve higher metabolic rates.

Somewhat paradoxically, recent work has shown that increasing oxygen availability can be a powerful force constraining the evolution of increased multicellular size. Oxygen is a crucial metabolic cofactor, increasing ATP returns from metabolism and allowing otherwise unfermentable carbon sources to be utilized for growth [15,16]. However, the evolution of large multicellular bodies creates a strong barrier to diffusion. This can limit the ability of interior cells to access oxygen, reducing their growth rates. The oxygenation of Earth’s atmosphere thus results in counterintuitive evolutionary dynamics, constraining the evolution of macroscopic multicellular size by generating a novel and powerful trade-off between organismal size and growth rate [17,18] that did not previously exist.

To overcome these anatomical diffusion barriers and support the evolution of large body size, modern animals have evolved specialized oxygen-binding and transport proteins such as tetrameric hemoglobins, dimeric hemerythrins, and multimeric hemocyanins. These proteins are all able to bind and release oxygen cooperatively as oxygen levels change as hemoglobin does [19,20], enabling rapid loading at a respiratory organ and rapid unloading in respiring tissue. These familiar respiratory proteins are components of circulatory systems that thus enable rapid bulk transport of oxygen bound to carriers throughout the body, enabling the evolution of large, active organisms.

However, phylogenetic studies tracing the origins of animal respiratory proteins reveal that freely circulating oxygen carriers with cooperative oxygen binding in blood and hemolymph evolved well after the origin of macroscopic size, and are in fact derived from more ancient monomeric stationary globins and hemerythrins that were not transported in circulatory fluids [21–24]. These more primitive proteins were expressed in body cells and tissues, but still served to facilitate oxygen diffusion from cell to cell and through tissues by increasing the effective diffusion rate of oxygen much as modern tissue globins such as myoglobin do today [25–32]. The ancestral genes encoding these simple intracellular oxygen diffusion facilitators likely originated near or just prior to the most recent common ancestor of the Bilateria [21,23,33–35], a critical clade in animal evolution which is defined by the origin of triploblastic embryos with endoderm, ectoderm, and mesoderm germ layers which give rise to complex tissues more than 2 cells thick [36].

The selective drivers favoring the origin of facilitated diffusion via static oxygen-binding proteins remain unresolved. While it seems clear that they provide a benefit by increasing O₂ diffusion, no prior work has directly examined this benefit as a function of exogenous O₂ concentration and organism size. In this work, we examine these dynamics through synthetic biology and mathematical modeling. We heterologously express the well-studied model oxygen-binding proteins myoglobin from sperm whales (Physeter macrocephalus) [37] and myohemerythrin from peanut worms (Themiste zostericola) [38] in oxygen-limited snowflake yeast (modified Saccharomyces cerevisiae), a model system of diffusion-limited multicellularity capable of rapid in vitro evolution [39–42].

We expected that while enhanced oxygen diffusion would benefit snowflake yeast at all oxygen levels, the benefit would be largest at low oxygen because this environment presents the harshest limitation to the evolution of increased size in the snowflake model system [17]. However, as we examine the fitness consequences of facilitated diffusion, we find an unexpected result: oxygen-binding proteins confer the greatest advantage for large multicellular clusters not when oxygen is rare and diffusion is slow, but when it is abundant. Mathematical modeling provides insight into this counterintuitive result, showing that facilitated diffusion leads to greater increase in oxygen flux through the surface of a cluster when oxygen levels are high compared to when they are low, alleviating anatomical limitations to the use of an abundant resource rather than compensating for low total O₂ availability. By directly examining the fitness consequences of proteins driving facilitated oxygen diffusion to changes in organismal size and the availability of environmental oxygen, our results provide novel context for the evolution of oxygen-binding proteins in the Metazoa.

To investigate whether heterologously expressed oxygen-binding proteins can enhance oxygen diffusion in our snowflake yeast model system, we first needed to quantify oxygen penetration depth into multicellular clusters. Prior work has shown that internal cells in snowflake yeast are strongly diffusion limited in a low-oxygen environment, with only peripheral cells able to respire actively [17].

To examine the effect of oxygen-binding proteins on the depth of oxygen diffusion, we constructed snowflake yeast strains expressing either myoglobin or myohemerythrin integrated at the HO locus (Fig 1A) and introduced the MitoLoc reporter system (preSU9-GFP + preCOX4-mCherry) to visualize aerobic respiration. The MitoLoc system uses a constitutive GFP tag on the outer mitochondrial membrane protein TOM70 to visualize mitochondrial morphology and an mCherry tag on the inner membrane COX4 protein. This protein’s rate of import from the cytoplasm to the inner membrane is dependent on mitochondrial inner membrane potential, allowing colocalization of these 2 tagged proteins to detect membrane potential indicative of high respiration rates [43].

Fig 1. Heterologous expression of oxygen-binding proteins increases O₂ diffusion depth in snowflake yeast.

(A) Experimental strains were constructed by inserting GFP, myohemerythrin, or myoglobin at the HO locus in the yeast chromosome. By cutting plasmids in a region of HO homology, the genes are introduced as a high-expression tandem repeat array. Created in BioRender [45]. (B) Box and whisker plot of oxygen diffusion depth using the mitochondrial MitoLoc reporter in snowflake yeast clusters expressing oxygen-binding proteins or wild-type controls under oxygen limitation. Myohemerythrin and myoglobin expression significantly increased oxygen diffusion depth in the low oxygen environment—21 μm mean diffusion depth for both myoglobin and myohemerythrin vs. 16 μm for the ancestor (p F_2,155 = 12.57 one-way ANOVA, pairwise comparisons via Tukey’s HSD with α = 0.05). No significant difference was observed under supplemental oxygen (p = 0.48, F_2,85 = 0.74 one-way ANOVA with α = 0.05). Points indicate individual clusters. (C) Representative fluorescence micrographs of MitoLoc showing increased depth of oxygen penetration (white colocalization between GFP in cyan and RFP in magenta) in clusters expressing myoglobin (right) compared to a wild-type control (left) under low oxygen. The data underlying this figure can be found in S1 Data and at http://zenodo.org/records/14512540.

https://doi.org/10.1371/journal.pbio.3002975.g001

We imaged snowflake yeast clusters expressing the oxygen-binding proteins or a wild-type control containing MitoLoc after obligately aerobic growth in Yeast Extract Peptone Glycerol media (YEP-Glycerol) under both low oxygen (10 ml cultures shaken without aeration) or supplemental oxygen levels (cultures aerated with room air bubbled through the culture [17,44]). Fractions of clusters exhibiting evidence of aerobic respiration were quantified as described in Bozdag and colleagues [17]. In low oxygen, the majority of the daily culture cycle is spent under 5% saturation (approximately 0.0125 mM). In contrast, under supplemental oxygen (see Methods for details) the majority of the daily culture cycle is spent above 50% saturation (approximately 0.125 mM) and it never drops below 32% (approximately 0.08 mM) (S1 Fig). Strains expressing oxygen-binding proteins showed significantly increased depth of oxygen diffusion under oxygen limitation to wild-type controls (Fig 1B, 21 μm mean diffusion depth for both myoglobin and myohemerythrin, versus 16 μm for the ancestor), indicating the heterologous proteins enhance penetration of oxygen to internal cells (Fig 1B and 1C). No significant difference in O₂ diffusion depth was observed under supplemental oxygen, however.

The fitness effects of expressing oxygen-binding proteins should depend on both the concentration of oxygen in the environment and the size of the multicellular cluster expressing these proteins. Higher environmental oxygen levels should increase the depth that oxygen penetrates, while increasing the size of the cluster should reduce the proportion of cells that are oxygenated. We examined this experimentally, by competing myohemerythrin and myoglobin-expressing strains against their isogenic GFP-tagged ancestor without oxygen-binding protein expression during aerobic growth on YEP-Glycerol. We engineered a small (approximately 10 μm radius) cluster variant as well as unicellular strains to test how size impacts the fitness advantage of oxygen-binding protein expression. Small clusters were generated through deletion of the BUD8 landmark polarity gene, which gives rise to rounder, smaller snowflake cells, and markedly smaller groups (approximately 1.8 and approximately 5.8 times smaller in radii and volume, Figs 2A and S2). Unicells were generated via inclusion of the wild-type ACE2 allele. Normal-sized clusters, small clusters, and unicells were competed under the same 2 oxygen regimes: low (cultures shaken without aeration) and supplemental (cultures aerated with room air bubbled through the liquid) (S1 Fig).

Fig 2. Oxygen-binding proteins are most beneficial for large snowflake yeast in a high oxygen environment.

(A) Small snowflake yeast (ace2Δ bud8Δ), engineered by inducing a mutant that causes a random budding pattern to normal snowflake yeast, were approximately half the radius wild-type (ace2Δ) snowflake yeast. (B) Normal-sized snowflake yeast possessed a fitness advantage when expressing myohemerythrin (one-sample t tests, p = 0.0098, t = 4.62 and p = 0.002, t = 7.08, low and supplemental oxygen, respectively) and myoglobin (one-sample t tests, p = 0.01, t = 4.24 and p = 0.0001, t = 14.06 low and supplemental oxygen, respectively) relative to their isogenic ancestor under both low and supplemental oxygen. This advantage was 2.6-fold (two-sample t test, p = 0.008, t = 3.72, myohemerythrin) and 1.9-fold (two-sample t test, p = 0.016, t = 3.25, myoglobin) higher under supplemental oxygen relative to low oxygen conditions. The benefits of expressing myoglobin and myohemerythrin in small-sized snowflake yeast were comparatively modest. Myohemerythrin and myoglobin provided no detectable fitness benefit under supplemental oxygen (one-sample t tests, p = 0.40, t = 0.94 and p = 0.13, t = 1.91, respectively; Fig 2B) and a marginally significant fitness advantage in low oxygen (p = 0.07, t = 2.44 and p = 0.03, t = 3.41, respectively). In a unicellular (ACE2) background, myohemerythrin expression resulted in comparable fitness levels to small snowflakes with a lower slightly fitness in high oxygen than low (two-sample t test, p = 0.013, t = 3.59), while myoglobin showed a significant fitness defect in high oxygen compared to low oxygen (two-sample t test, p = 0.000026, t = 10.15). Dots represent average relative fitness with bars as one standard deviation, n = 5 independent competitions for each group. The data underlying this figure can be found in S1 Data and at http://zenodo.org/records/14512540.

https://doi.org/10.1371/journal.pbio.3002975.g002

Unicellular yeast bearing myohemerythrin and myoglobin exhibited similar fitness to small bud8Δ snowflake yeast in low oxygen conditions, suggesting that small clusters are indeed small enough to allow oxygen to diffuse deeply into them under experimental conditions. These finesses are similar to those observed for unicells in YEP-Dextrose (see S3 Fig), an environment where fermentation dominates and respiration is of low importance, suggesting that the benefit of expression in low oxygen conditions is minimal. However, while unicellular yeast bearing myohemerythrin showed similar fitness to small clusters in supplemental oxygen conditions, myoglobin-bearing unicells exhibit a significant fitness defect of approximately 1%. We suspect that this is due to a lower maximum respiratory capacity, possibly due to abundant myoglobin proteins binding up heme or iron that would otherwise be bound for the electron transport chain. Nonetheless, this minor cost is more than made up for by the benefits of myoglobin in larger clusters (Fig 2B).

The finding that normal snowflake yeast clusters gained the greatest benefit from oxygen-binding protein expression under supplemental oxygen rather than low oxygen was surprising. We expected the impact of oxygen limitation to be most severe under low oxygen conditions like those that may have suppressed the origins of animals [14,46,47]. Why should proteins that are known for alleviating oxygen limitations be more beneficial when oxygen is plentiful? To better understand this counterintuitive result, we modeled the interplay between oxygen availability, cluster size, and myoglobin expression using reaction-diffusion equations in the Julia modeling environment (see S1 Code).

We simulated spherical yeast clusters with radii ranging from 5 to 70 μm, approximating the range of single cells to large (but not yet macroscopic) snowflake yeast that have undergone significant laboratory evolution for increased size [39–41,48,49]. Oxygen diffuses into the cluster from the surrounding environment using diffusion parameters derived from flocculating yeast [50], while being consumed aerobically. Metabolism was modeled using Monod kinetics with parameters extracted from prior literature on yeast metabolism [51–54]. Myoglobin was added to the simulation at concentrations from 0 to 0.2 mM, a concentration typical of muscle tissue after correcting for cell packing fractions [54,55]. Oxygen was allowed to bind and unbind from myoglobin as it diffused, according to measured properties of the protein [26,56,57]. Each simulation was initiated with a snowflake yeast at a given radius, myoglobin concentration, and external oxygen concentration, and was allowed to run for 100 simulated seconds to reach equilibrium. We collected data on the final profiles of oxygen concentration across the organism, oxygen metabolism, bound and unbound myoglobin concentration, and average metabolic rates.

This computational model allowed us to predict the profile of oxygen penetration and aerobic respiration for different combinations of cluster size, environmental oxygen, and myoglobin expression. The results provide insight into how oxygen carrier proteins can enhance oxygen flux that is dependent on multicellular geometry and oxygen availability. The results of this simulation were broadly consistent with experimental observations. A typical cluster in our model (Fig 3A) exhibits a high oxygen concentration at its periphery, which rapidly falls as depth increases. This results in an outer region undergoing high rates of aerobic respiration but a hypoxic core with low or no aerobic respiration, consistent with experimental observations (Fig 1B). Notably, the addition of simulated myoglobin proteins to the model results in a significant increase in the quantity of aerobic respiration and total oxygen within the cluster—oxygen-bound myoglobin reaches equilibrium penetrating deeper below the surface of a cluster than high levels of dissolved oxygen does, contributing to aerobic respiration of inner cells. Depending on the concentration of myoglobin, this dissolved myoglobin-bound oxygen can represent a large fraction of the total oxygen flux in the system.

Fig 3. Modeling the relationship between size, environmental oxygen, and the fitness effects of myoglobin.

(A) Output of a spherically symmetrical model of the coupled oxygen diffusion, myoglobin diffusion, oxygen/myoglobin binding, and aerobic respiration. Oxygen concentration falls rapidly as distance to the surface of the cluster increases, followed by oxygenated myoglobin concentration, followed by oxygen metabolism. (B) Simulated metabolic rate vs. depth of large (20 μm radius) and small (10 μm radius) clusters with and without 0.1 mM myoglobin expression. Large clusters attain larger metabolic rate increases in high oxygen environments than low; small clusters are metabolically saturated at high oxygen levels and thus attain no benefit while attaining low benefits at low oxygen levels. (C) The magnitude of myoglobin-induced metabolic rate increase, and thus myoglobin-induced selective advantage, of clusters at varied radii and oxygen levels at 0.1 mM of myoglobin expression. Selective advantage is maximized at large radii and intermediate oxygen levels. (D) Selective advantage provided by different degrees of myoglobin expression at 5 μm, 10 μm, and 20 μm radii. At small sizes myoglobin only provides benefits at low oxygen levels with severe diminishing returns to expression, as size increases myoglobin provides larger benefits at intermediate and high oxygen levels with reduced diminishing returns. (E) The fraction of growth-rate decrease caused by increasing in radius from 5 to 20 microns ameliorated by expression of different myoglobin concentrations. As the environmental oxygen level increases, the fraction of size-induced growth retardation that can be reversed by myoglobin expression rapidly increases before leveling off at intermediate values. The data underlying this figure can be found in S1 Data and at http://zenodo.org/records/14512540.

https://doi.org/10.1371/journal.pbio.3002975.g003

Further, the differential effects of oxygen-binding proteins on the growth of clusters of different sizes seen from the fitness assays (Fig 2B) is captured by this model. Experimentally, we observed that small (approximately 10 μm radius) clusters and unicells exhibited a slight fitness benefit from myoglobin expression at low oxygen levels, but no benefit at high oxygen levels. Our model indicates that at high oxygen levels, small clusters are already oxygenated to their core such that myoglobin expression does not appreciably increase the rate of respiration, while at low oxygen levels, the expression of myoglobin allows oxygen to penetrate more deeply and the average rate of aerobic respiration to increase (Fig 3B). Our model was similarly conciliant with observations that large (approximately 20 μm radius) clusters exhibited moderate fitness advantages at low oxygen levels, which rose considerably at higher oxygen levels. Our simulation indicated that large clusters are never oxygenated to their cores (which is consistent with experimental observations, Fig 1B and 1C) and so always get a fitness benefit from increased diffusion rates, with a larger total benefit arising at higher oxygen levels when facilitated diffusion has a larger impact on overall oxygen consumption.

Exploring the size and oxygen parameter space revealed an important general trend. Testing all radii from 5 to 70 μm and all oxygen levels from 0.01 to 0.25 mM (Fig 3C, simulated at a myoglobin concentration of 0.1 mM), it is apparent small clusters never gain a large benefit from expressing oxygen-binding proteins, and the little benefit they do experience is limited to low oxygen levels. As size increases, the maximum benefit obtained from facilitated oxygen diffusion and the oxygen concentration at which that benefit is maximized continually increases. It does, however, plateau at radii large enough that much of the metabolically active cluster is hypoxic. The maximum benefit increases very little above a radius of 40 μm, with the growth rate increase above that size roughly indicating the fold increase in oxygen penetration depth.

Interestingly, we find that for large clusters and high myoglobin concentrations the maximum benefit obtained from myoglobin expression is realized at intermediate oxygen concentrations of approximately 0.05 mM—roughly a fifth of saturation—with smaller but significant benefits remaining at higher oxygen levels. This is unlike the effect in small clusters, in which the benefits of myoglobin expression are maximized at very low oxygen levels and then decrease to zero with increasing oxygen (Fig 3D).

Finally, we examine how much myoglobin can ameliorate the growth costs of multicellularity (that is, the growth of the group relative to that of a single cell). The fraction of growth costs ameliorated by myoglobin expression rises linearly as environmental oxygen increases from zero before leveling out at intermediate oxygen levels (circa 0.05 mM) and remains roughly constant above this level (Fig 3E). The effect is dependent upon myoglobin expression level, with rapidly diminishing returns to higher expression at low oxygen levels and slowly diminishing returns at high levels.

To test the degree of diminishing returns of expression of oxygen-binding proteins, we put myoglobin under the control of the LexA-ER-B42 system [58] in snowflake yeast, with myoglobin expression driven by a synthetic promoter activated by a β-estradiol. This system is expected to show rapidly increasing protein expression from 0 to approximately 50 μm β-estradiol, with slower increasing expression at higher levels to approximately 200 μm [58]. Yeast were induced with 0, 10, 20, 50, and 200 μm β-estradiol and competed against GFP-bearing snowflake control clusters in low and supplemental oxygen conditions as previously described. We found that uninduced, there was no significant difference in fitness observed between low and supplemental oxygen, and that with up to 50 μm β-estradiol induction a slight cost was observed under low oxygen with significantly higher fitness observed under supplemental oxygen (Fig 4). However, further increasing induction intensity resulted in high oxygen fitness defects. We believe that further optimization of the iron import and heme production machinery of yeast, as is being explored in the bioproduction of hemoglobin and leghemoglobin [59–61], is necessary to accurately observe the dose-dependence of oxygen-binding proteins in this system at high levels of expression. Particularly, high levels of globin expression in yeast has been known to deplete heme and saturate iron import capacity at high expression levels, possibly affecting metal homeostasis and total respiratory capacity.

Fig 4. Effects of varied myoglobin expression.

Experimental effect of induction of myoglobin expression at varied levels in low and high oxygen environments in snowflake yeast. Induction of myoglobin in supplemental oxygen to low (0 μm β-estradiol using shows a low level of constitutive expression) and intermediate levels (10–50 μm) shows a fitness advantage relative to low oxygen levels, with increased induction (200 μm) showing a fitness cost, while induction shows a mild fitness deficit at low oxygen (two-sample t tests: 0 μm t = −1.90, p = 0.11, 10 μm t = −2.52, p = 0.037, 20 μm t = −8.19, p = 0.000038, 50 μm t = −3.62, p = 0.0077, 200 μm t = 1.43, p = 0.197). The data underlying this figure can be found in S1 Data and at http://zenodo.org/records/14512540.

https://doi.org/10.1371/journal.pbio.3002975.g004

Oxygen-binding proteins appear to have originated in the common ancestor of bilaterian animals, a clade characterized by complex anatomy and thick tissues. However, the role of oxygen-binding proteins as a key innovation facilitating the early evolution of large size has remained unclear. Much previous work has suggested that the evolution of multicellular size and complexity was limited by the availability of oxygen [12,14,46], and many modern organisms regulate myoglobin levels with increasing expression during hypoxia [62,63]. This limitation would be consistent with work on the de novo evolution of multicellularity in the snowflake yeast model system, in which O₂ limitation strongly impeded the evolution of larger, tougher multicellular organisms [17,64]. Facilitated oxygen diffusion, mediated by O₂-binding proteins, could have lifted this constraint when oxygen was low but present, allowing small multicellular organisms to overcome the constraint of low environmental oxygen.

Our biophysical model illuminates the mechanism underpinning these results. The addition of globins to a system that previously lacked it (Fig 3B) causes internal oxygen gradients to become more shallow at the core of the cluster, indicating increased O₂ flux from the exterior, as bound myoglobin diffuses more deeply and gives up oxygen to deficient tissues. However, globin expression also leads to a steepening of the oxygen and aerobic metabolism gradient at the surface of a cluster, as deoxygenated myoglobin diffuses outwards and takes it up. This leads to increased O₂ diffusion rates down this steepened concentration gradient into the organism from the environment when it is available (see S4 Fig for additional detail).

We therefore suggest an alternative hypothesis for the role and timing of the origins of oxygen-binding proteins in the Metazoa: that they may have evolved not as an adaptation to circumvent low atmospheric oxygen early in animal evolution, but instead became favored to emerge due to rising oxygen levels during the Neoproterozoic, allowing multicellular organisms to better exploit the increased metabolic potential of abundant environmental oxygen. Our hypothesis is consistent with the fact that oxygen-binding globins and hemerythrins appear to have an origin circa the common ancestor of Bilateria [21,23,33–35]. This corresponds to the Ediacaran period, after oxygen levels rose to near today’s values [46,65]. An independent origin of oxygen-binding globins appears to have also occurred in land plants with nitrogen-fixing root nodules, where a similarly highly metabolically active non-photosynthetic tissue must maintain high metabolic rates despite diffusion limitations and other oxygen constraints [32,66].