Manganese co-localizes with calcium and phosphorus in Chlamydomonas acidocalcisomes and is mobilized in manganese-deficient conditions

Exposing cells to excess metal concentrations well beyond the cellular quota is a powerful tool for understanding the molecular mechanisms of metal homeostasis. Such improved understanding may enable bioengineering of organisms with improved nutrition and bioremediation capacity. We report here that Chlamydomonas reinhardtii can accumulate manganese (Mn) in proportion to extracellular supply, up to 30-fold greater than its typical quota and with remarkable tolerance. As visualized by X-ray fluorescence microscopy and nanoscale secondary ion MS (nanoSIMS), Mn largely co-localizes with phosphorus (P) and calcium (Ca), consistent with the Mn-accumulating site being an acidic vacuole, known as the acidocalcisome. Vacuolar Mn stores are accessible reserves that can be mobilized in Mn-deficient conditions to support algal growth. We noted that Mn accumulation depends on cellular polyphosphate (polyP) content, indicated by 1) a consistent failure of C. reinhardtii vtc1 mutant strains, which are deficient in polyphosphate synthesis, to accumulate Mn and 2) a drastic reduction of the Mn storage capacity in P-deficient cells. Rather surprisingly, X-ray absorption spectroscopy, EPR, and electron nuclear double resonance revealed that only little Mn2+ is stably complexed with polyP, indicating that polyP is not the final Mn ligand. We propose that polyPs are a critical component of Mn accumulation in Chlamydomonas by driving Mn relocation from the cytosol to acidocalcisomes. Within these structures, polyP may, in turn, escort vacuolar Mn to a number of storage ligands, including phosphate and phytate, and other, yet unidentified, compounds.

from oxygenic photosynthesis caused their massive precipitation as oxide complexes: their availability dropped by orders of magnitude (2 for Mn, and 4 for Fe), placing tremendous pressure on early unicellular organisms to meet their transition metal quotas for enzymes that had incorporated Fe or Mn cofactors (8) and presumably providing the driving force for the evolution of metal-selective transporters and intracellular metal sequestration mechanisms.
Mn and Fe deficiency are symptomatically similar in plants, as they both decrease the efficiency of photosynthesis and are both accompanied by leaf chlorosis (9), although the underlying causes are different. In the case of Fe, chlorosis results from inhibition of chlorophyll biosynthesis (10), now attributed to the di-iron cyclase in the pathway (11), and also from programmed reorganization of the photosynthetic apparatus for photoprotection and iron sparing (12)(13)(14)(15). Mn deficiency likely destabilizes the PSII complex, perhaps due to the release of the extrinsic lumenal proteins (5). Mn deficiency also increases intracellular redox stress because of decreased SOD activity, which manifests as necrotic regions on leaves (3,16) or replacement of Mn in enzymes with Fe (17).
On the other end of the nutrient assimilation spectrum, excess Fe or Mn can be detrimental for growth, although for different reasons. High cellular Fe levels generate reactive oxygen species via the Fenton reaction, whose toxicity is exacerbated under high photon flux densities (18). For Mn, there is the danger of mismetalation and potential inactivation of the substituted metalloenzyme (17). Two notable exceptions are the budding yeast and the bacterium Deniococcus radiodurans. In both organisms, high concentrations of cellular Mn help protect proteins against oxidative stress damage due to the high superoxide dismutation activity displayed by Mn-phosphate low-molecular weight metabolites (19 -21).
To prevent experiencing either metabolic extreme, all living organisms have devised elaborate systems for the uptake, transport, storage, and remobilization of micronutrients. Mn is typically transported as Mn 2ϩ species, often by members of the same transporter families that transport Fe 2ϩ . These transporters include members of the NRAMP (natural resistanceassociated macrophage protein), MTP (metal tolerance protein), and VIT1/CCC1 (vacuolar iron transporter in Arabidopsis) families (16,(22)(23)(24)(25)(26)(27). Primary Mn 2ϩ uptake takes place at the plasma membrane by NRAMPs and is subsequently routed to the endoplasmic reticulum, Golgi, mitochondria, and chloroplasts (in algae and plants) to fulfill the metalation needs of each compartment. In addition, cells can store Mn 2ϩ in vacuoles for later use under conditions of nutrient limitation or to minimize the toxic effects of overaccumulation when placed in an environment with an overabundance. In this regard, the vacuole acts as a temporary cellular sponge for excess Mn 2ϩ (this work).
Because Mn 2ϩ and Fe 2ϩ can be transported by the same transporters, a greater understanding of Mn metabolism is relevant not only for appreciating Mn biochemistry but also for a systems-level view of Fe homeostasis that includes potential cross-talk and interactions with other metals. For instance, high Mn availability can induce Fe-deficiency symptoms (28,29) by competing for the same transporters (22,(30)(31)(32). The Arabidopsis high-affinity Mn 2ϩ transporter NRAMP1 can also function as a low-affinity Fe 2ϩ transporter under Fe-replete conditions (33). Yet, despite its critical function in photosynthesis, Mn metabolism is underinvestigated. In the unicellular alga Chlamydomonas reinhardtii (hereafter referred to as Chlamydomonas), Mn deficiency strongly induces the expression of NRAMP1 and can also be associated with secondary Fe deficiency (5). The effects of high Mn supply in Chlamydomonas are unknown.
High concentrations of Mn in the soil typically rise after heavy rains and are exacerbated in acid soils. The analysis of various yeast mutants pointed to a role for phosphate in Mn uptake, perhaps as a co-transported counterion on phosphate transporters (21,(34)(35)(36). Polyphosphate chains are also critical for magnesium (Mg 2ϩ ) uptake, suggested to act as a cation filter that attracts Mg 2ϩ to the vacuole (37). It is becoming more appreciated that lysosome-type organelles (of which the vacuole is one), called acidocalcisomes (because of their low pH and high calcium content), are key sites for metal storage (38 -41).
We tested the metal content of Chlamydomonas as a function of Mn 2ϩ -EDTA supply in the medium. Surprisingly, we found that cells accumulate Mn 2ϩ in linear proportion to supply many times over their demands for metalating all Mn sites in proteins (also referred to as quota). Spectroscopic analysis (X-ray absorption near edge structure; XANES) indicates that intracellular Mn is mostly present as Mn 2ϩ species, as expected. Based on various imaging methods, including nanoscale secondary ion MS (nanoSIMS), scanning X-ray fluorescence microscopy (XFM), energy-dispersive X-ray spectroscopy (EDX), and fluorescence imaging, we conclude that Mn 2ϩ is concentrated in the acidocalcisome. Mutants blocked in polyphosphate (polyP) synthesis do not accumulate Mn 2ϩ . However, spectroscopic studies (extended X-ray absorption fine structure (EXAFS) and electron nuclear double resonance (ENDOR)) do not support Mn 2ϩ association with polyP. Therefore, we conclude that polyP functions as an escorting and temporary ligand within acidocalcisomes rather than an end-point sequestration agent as suggested for Mg ions (37).

Chlamydomonas cells have a linear capacity for Mn accumulation
Mn deficiency in Chlamydomonas is associated with slow growth and decreased chlorophyll content due to compromised photosynthetic electron transfer (5). To determine the behavior of Chlamydomonas cells under conditions of Mn excess, we exposed cultures to Mn concentrations ranging from 6 to 1000 M and monitored growth, chlorophyll levels, and photosynthetic parameters. All Mn concentrations were welltolerated by Chlamydomonas and even provided an apparent growth advantage at 50 M and above (Fig. 1A). Higher concentrations of cellular Mn may participate in quenching of reactive oxygen species harmful to cells, as was shown in yeast (20,42). All cultures were healthy, as demonstrated by robust chlorophyll accumulation and F v /F m values, with no indication of cellular stress (Fig. S1). Other metals like Fe, copper (Cu), zinc (Zn), and phosphorus (P) remained constant over the range of Mn concentrations tested here, with the exception of a slight Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis increase in intracellular calcium (Ca) under very high Mn concentrations (Fig. S2).
The observed tolerance to excess external Mn was accompanied by intracellular accumulation of the metal well beyond its normal quota (Fig. 1B). Chlamydomonas cells indeed displayed a linear capacity for Mn uptake from the medium that was not saturated even at 1 mM Mn. This behavior is unique among the metals we have tested and suggests that a distinct control mechanism operates for Mn 2ϩ uptake and storage. Indeed, we have previously observed rapid saturation of Cu uptake that does not extend beyond 2-3 ϫ 10 7 atoms/cell (41,43). The Fe quota is much higher, and its accumulation continues even after the quota is reached, but nevertheless more slowly (13).
We conclude that Chlamydomonas cultures exhibit a linear capacity for Mn uptake to many times its normal quota for photosynthesis and that this intracellular accumulation appears to be beneficial to cell growth.

Mn co-localizes with Ca and P inside acidocalcisomes
In Arabidopsis, the major Mn storage site is the vacuole, from which it can be remobilized by NRAMP family members (16,22,44). We explored intracellular Mn distribution in Chlamydomonas cells grown in Mn-replete and excess conditions by multiple techniques: XFM, transmission EM coupled with energy-dispersive X-ray spectroscopy (TEM-EDX), and nano-SIMS. Each method offers particular advantages, and all three were largely congruent; we present the results from nanoSIMS below. XFM and TEM-EDX are shown in Figs. S3 and S4.
NanoSIMS is a method that maps elements, in surfaces of solid samples based on their mass with high spatial resolution. The method is compatible with visualizing populations of cells, which enables statistical analysis of regions of interest. In conjunction with the use of standards, quantitative information can be extracted from the data. Fixed cells grown in 6 or 1000 M MnEDTA were analyzed with an O Ϫ analysis beam, and secondary ions were collected for carbon ( 12 C ϩ ), phosphorus ( 31 P ϩ ), calcium ( 40 Ca ϩ ), and manganese ( 55 Mn ϩ ). We report elemental abundances as uncorrected ion ratios, as reflected by the specific isotope species detected. As shown in Fig. 2A, P, Ca, and Mn showed clear co-localization in a field of cells grown in excess Mn. XFM observations of fixed whole cells painted a similar picture of intracellular metal distribution (Fig. S3). Sites of P and Ca accumulation coincide with electron-dense particles seen by TEM-EDX (Fig. S4) and are consistent with the defining elemental content of the acidocalcisome, a lysosomerelated organelle. We exploited the sensitivity and dynamic range of nanoSIMS to quantify relative P, Ca, and Mn levels by subdividing each cell into nonoverlapping regions of interest (see Fig. S5) and normalizing ion counts to 12 C ϩ (41). We used the sites of high 40 Ca ϩ / 12 C ϩ and 31 P ϩ / 12 C ϩ concentration as proxy for acidocalcisomes (Fig. 2, B and C). The relative concentration of Ca and P did not change inside acidocalcisomes as a function of Mn supply (Fig. S6). However, Mn accumulation in cells exposed to excess MnEDTA was 100-fold higher than in cells supplied with 6 M MnEDTA and coincided with foci of high Ca (Fig. 2B) and P (Fig. 2C). Because regions of interest were drawn randomly within cellular sections, they did not perfectly overlap with P, Ca, and Mn (see Fig. S5). Nevertheless, 55 Mn ϩ / 12 C ϩ correlated positively with 40 Ca ϩ / 12 C ϩ (R 2 ϭ 0.76) and 31 P ϩ / 12 C ϩ levels (R 2 ϭ 0.67) under excess-Mn conditions. The relative concentrations of all three elements co-varied linearly over a 100-fold range. To further illustrate these relationships, we selected 10 Mn foci and four low-Mn regions and determined associated Ca and P levels (shown as red and blue circles in Fig. 2, B and C). We note that a small Mn fraction may remain in the cytosol of cells grown under Mn excess, where it does not co-localize with Ca or P foci (blue circles). At these locations, Mn levels were still higher than within most Ca foci in cells grown under normal Mn supply. These results, obtained from three independent imaging techniques, therefore collectively demonstrate sequestration of Mn into the P-and Ca-rich acidocalcisomes, over a wide range of Mn concentrations in the growth medium.

A mutant in the VTC complex does not accumulate manganese
Based on the yeast literature on manganese metabolism, and in view of its concentration within acidocalcisomes, we hypothesized that Mn ions were stored inside the organelle complexed with polyP (45,46). The vacuolar transporter chaperone (VTC) complex catalyzes the synthesis of polyP from cytosolic P i and translocates the polymer across the vacuolar membrane in an ATP-dependent manner. The VTC complex is composed of five subunits in yeast, Vtc1-Vtc5 (35,47). The Chlamydomonas

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
genome encodes one orthologue for Vtc1 and Vtc4 each, and a mutant in VTC1, lacking detectable polyP, is available for phenotypic characterization (48). Growth of the vtc1-1 mutant was not adversely affected by excess Mn (Fig. 3A), yet a role for polyP in Mn accumulation became evident when we determined the elemental profiles of vtc1-1 versus VTC1 complemented strains cultured in medium with varied Mn content. The complemented strain accumulated Mn with the same linear capacity as did the WT laboratory strain CC-4532, with an 11-fold increase in Mn content between 6 and 600 M (Figs. 1B and 3B). This capacity was greatly reduced in the vtc1-1 mutant: starting with only half of the Mn content normally seen in VTC1, the mutant managed only a 2-fold rise in its Mn content, even when supplied with 600 M MnEDTA (Fig. 3B). Other elements that differentiated VTC1 and vtc1-1 included P (Fig.  3C) and Ca (Fig. 3D); neither element content changed as a function of Mn supply in either genotype, but P levels in vtc1-1 were one-third of those in VTC1, whereas Ca levels reached at most one-tenth of those in VTC1, suggesting that Ca ions may associate directly with polyP. That vtc1-1 mutants have a lower P quota is expected, as vacuolar polyP makes a substantial contribution to the total cellular P content in WT cells, equivalent to molar concentrations of P inside acidocalcisomes (49). The elemental profile of the Chlamydomonas vtc1-1 mutant is in line with the yeast vtc1 mutant, which has an even more pronounced decrease in internal P content (50).
If the lower Ca content of vtc1-1 were responsible for its inability to accumulate Mn, then growing WT cells with a frac-tion of the Ca normally supplied in our standard growth medium (340 M) should phenocopy the vtc1-1 mutant phenotype. However, we did not discern any drop in Mn accumulation, even when the Ca supply was reduced to only 10% of normal levels, leading to a ϳ60% reduction in cellular Ca (Fig. S7, A and B). Cellular P contents only decreased slightly at the lowest Ca supply levels (Fig. S7C). In contrast, withholding P from the growth medium, which resulted in a 40-fold drop in cellular P levels ( Fig. S8), did interfere with both Mn and Ca accumulation. Indeed, P-limited cells only contained one-quarter of the Mn taken up by P-replete cells when exposed to excess Mn (Fig.  S8A); their Ca content was also very low and reminiscent of Ca levels in vtc1 mutants ( Fig. 3D and Fig. S8C). By similarity with Escherichia coli and yeast, we hypothesize that P limitation is accompanied by a severe drop in polyP that would prevent Mn (and Ca) accumulation, in agreement with our results with the vtc1-1 mutant (Fig. 3). As shown recently for Mg ions (37), polyP therefore also plays a crucial role in Mn and Ca accumulation inside acidocalcisomes.
Aksoy et al. (48) reported that acidocalcisomes were missing in vtc1-1, which would easily explain the loss of Mn accumulation in the mutant. However, their conclusion relied on identifying acidocalcisomes as electron-dense particles in TEM images (48). Such particles inside acidocalcisomes are likely polyP crystals (49,51,52). vtc mutants do not accumulate polyP in yeast (35) or Chlamydomonas (48), and we wondered whether the presented evidence for the lack of acidocalcisomes in the vtc1-1 mutant might be inconclusive. We therefore used Scale bar, 5 m. B and C, correlative quantification of 12 C ϩ -normalized 55 Mn ϩ with 31 P ϩ and 40 Ca ϩ from nanoSIMS imaged cells grown in 1000 M (three replicates) and 6 M MnEDTA (two replicates). The subcellular correlation of ROIs for 55 Mn/ 12 C versus 40 Ca/ 12 C is shown in B, and that for 55 Mn/ 12 C versus 31 P/ 12 C is shown in C. Arrows, regions of high (4 and 3) and low (2) concentration of Mn, Ca, and P in the nanoSIMS image for each element; the same areas of interest are shown as red dots (high) and blue dots (low), respectively, in the correlation plots in B and C.

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
the lysosensor dye DND189, which is used to visualize acidic organelles like lysosomes and acidocalcisomes. We observed many small acidic compartments in VTC1 and vtc1-1 cells grown under both standard conditions and Mn excess, which argues that the vtc1 mutant can generate a significant proton gradient across the acidocalcisome membrane (Fig. 4, A and B). Staining of polyP with DAPI was only positive in VTC1, as described previously (Fig. 4 (C and D); see Aksoy et al. (48)). The inability of vtc1-1 to accumulate Mn thus cannot be attributed trivially to a loss of acidocalcisomes in the mutant and points to another process that is impaired.
To determine whether the acidocalcisome membrane was compromised in the absence of VTC1, we searched for the vacuolar membrane proteins H ϩ -pyrophosphatase and V-ATPase in a proteomics data set performed in another vtc1 allele (vtc1-2, with similar defects in P, Ca, and Mn accumulation as vtc1-1; Fig. S9A) and its associated WT strain CC-4533. We detected comparable spectral counts for the H ϩ -pyrophosphatase and all subunits of the V-ATPase in vtc1-2 and its WT strain (Fig. S9B), which is consistent with the presence of an organelle with a functional and energized boundary membrane (as noted from lysosensor DND189 staining, mentioned above). These results also argue that although the yeast and Chlamydomonas vtc1 mutants are phenotypically close in terms of their elemental profile ( Fig. 3; Yu et al. (50)), they are not biochemically identical. Indeed, the yeast vtc1 mutant is distinct in that it causes a general drop in the levels of several V-ATPase subunits at the vacuolar membrane (47), and the locus is named for that phenotype (vacuolar transporter chaperone). Also not observed in the Chlamydomonas vtc1 mutant was a large change in the levels of the H ϩ -pyrophosphatase or in the predicted homologues to the yeast membrane H ϩ -ATPase Pma1p (Fig. S9C). We did, however, detect fewer peptides for the VTC4 subunit of the VTC complex in the mutant, as might be expected when a multisubunit protein complex is missing a component. Although we cannot rule out more minute changes in the levels or localization of vacuolar membrane proteins in the Chlamydomonas vtc1 mutant, we conclude that the acidocalcisome boundary membrane is largely functional.  (C) and mutant vtc1-1 cells (D) exposed to replete (6 M) or high (600 M) Mn were stained with phosphate dye (DAPI) to observe intracellular polyphosphate accumulation (which appears yellow when present). Chlorophyll autofluorescence is shown in green. Confocal images were collected on a Zeiss 880 microscope using Airyscan in channel mode, and exposures were adjusted as needed to observe intracellular staining. Scale bar, 5 m. Images are representative of one replicate for each condition.

Mobilization of stored Mn when challenged by Mn deficiency
It is well-established that cells will prioritize particular metalloenzymes when faced with a limited metal supply (43,53). Although vtc1 mutants only accumulate one-third of the Mn normally seen in WT cells under standard growth conditions ( Fig. 3 and Fig. S9A), they show no signs of Mn deficiency, and Mn-containing proteins are just as abundant in the vtc1 mutant as they are in the WT strain (Fig. S9C). We therefore asked whether the Mn pool that accumulates in WT cells was biologically accessible. Accordingly, we first grew vtc1-1 and VTC1 strains photoautotrophically in 6 and 600 M MnEDTA and washed cells with EDTA to remove trace metals from the cell surface before transferring them to fresh medium with no added Mn. We chose photoautotrophic conditions to place stronger pressure on the photosynthetic apparatus, the main Mn sink with an essential function in photosynthetic cells, to more easily discern a phenotype (54). Indeed, the growth potential of Chlamydomonas cells was determined by the genotype at the VTC1 locus and by prior exposure to excess Mn. VTC1 cultures grew well if the inoculum originated from a 600 M MnEDTA preculture, but not from a 6 M MnEDTA preculture (Fig. 5A). The elemental profile for Mn across conditions confirmed Mn accumulation in VTC1 cells grown in 600 M MnEDTA at all times. This stored Mn was sufficient to sustain growth for at least 1 week when cells were transferred to growth medium with no added Mn, demonstrating that it was biologically accessible under Mn limitation. The vtc1-1 mutant presented an important control: in the absence of Mn accumulation typical for this mutant, cells could not survive unless provided with some MnEDTA in the growth medium, even when the preculture was exposed to 600 M MnEDTA (Fig.  5A). This control also argues against the trivial explanation of Mn carryover from the pretreatment condition. Actively growing photosynthetic cells require 2-3 ϫ 10 7 atoms of Mn/cell to perform adequately, and this held true in this experiment. Growth-arrested cells were well below this mark, with Mn levels closer to 0.5-0.7 ϫ 10 7 atoms of Mn/cell (Fig. 5B). In summary, stored Mn from acidocalcisomes can be remobilized to sustain growth and photosynthesis when Chlamydomonas is challenged by Mn limitation.

Determination of Mn oxidation and speciation within acidocalcisomes
To validate the hypothesis that polyP was acting as the main Mn ligand by defining the complexation and speciation of cellular Mn, we applied XANES and EXAFS spectroscopies to probe nearest-neighbor and long-range metal-ligand environments. We focused on the Mn K-edge, during which a metal 1s core electron is promoted to a vacant 3d orbital (in the pre-edge region) or to the continuum for all metal collected from cells grown in excess Mn (1000 M MnEDTA). As shown in Fig. 6A, all samples showed a limited pre-edge feature in the XANES, consistent with metal existing in an octahedral ligand symmetry (9,55). The excitation edge for Mn in cells matched that from the MnSO 4 standard in both inflection energy and overall shape, indicating that, as expected, accumulated Mn was predominantly present in the low-oxidation Mn 2ϩ species and not in a high-oxidation state.
Oscillations in an EXAFS spectrum are caused by scattering of the excited electron from the absorbing atom interacting with neighboring scattering atoms, the deconvolution of which provides a means to determine the distance between Mn and the neighboring atom and to estimate the chemical nature of this atom through simulation of a theoretical curve to the empirical data (56). Fourier transforms of EXAFS data provide a ϳ0.5 Å phase-shifted radial distribution function of the ligand environments around the absorbing metal ( Fig. 6B (right, top,  and bottom)), whereas simulations of the EXAFS data can be used to provide direct metrical details of the metal-ligand environment ( Fig. 6B (left, top, and bottom)). The best-fit simulation results suggested a nearest-neighbor ligand environment consisting of 5-6 oxygen and/or nitrogen ligands (based on error bars of the data) at a distance of 2.18 Ϯ 0.02 Å, whereas longrange scattering could be equally explained by 2-4 carbon (C) ligands or a single P ligand (around 4 Å from Mn atoms), with weaker support for P scattering (Table 1 and Fig. 6). A double peak between 2-4k was poorly explained by a fit with either C or P. Hydrated Mn Mn-(H 2 O) 6 can exhibit such oscillations in the 2-4k range, especially at lower pH (9), which would be compatible with the acidic environment of the acidocalcisome. A contribution by an imidazole-containing ligand (e.g. histidine) is another (nonexclusive) possibility, as Mn-histidine complexes only display a single peak in the 2-4k range (9).

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
We next employed EPR, ENDOR, and electron spin echo envelope modulation (ESEEM) spectroscopies to resolve the nature of the Mn ligands in our samples (19,21). EPR spectra of intracellular Mn 2ϩ qualitatively paralleled the elemental analysis (Fig. 1B), showing an over 10-fold greater accumulation of Mn in WT cells grown in the presence of 600 M MnEDTA (Fig.  7A, black line) relative to cells grown with the standard 6 M MnEDTA (red line). vtc1-1 cells exhibited reduced capacity to accumulate Mn (ϳ4-fold) compared with CC-4532 and VTC1 complemented strain (Fig. S10). We therefore used the BODIPY-based fluorescent probe M1, specific for Mn 2ϩ (57), to assess Mn accumulation in vtc1-1 and the complemented strain. Fig. S11 shows that the M1 probe both 1) supported reduced Mn accumulation in the mutant and 2) showed local-ization within a lysosensor-positive organelle. Note that the sensitivity of the M1 probe is insufficient to detect Mn within acidocalcisomes of cells grown under normal Mn nutrition conditions. Mn 2ϩ in both genotypes showed the narrow EPR signal (small zero-field splitting) with a resolved six-line 55 Mn hyperfine pattern centered at g-2 (ϳ12 kG) of Mn 2ϩ coordinated in the high-symmetry, octahedral geometry commonly observed in Mn-metabolite complexes (Fig. S10). From the EXAFS results above, we hypothesized that Mn metabolites could be P-containing compounds (e.g. P i , phytate (Phy), or polyPs, the last providing an elegant explanation for the loss of Mn accumulation in vtc1-1 mutants, defective in polyP biogenesis). We therefore applied ENDOR/ESEEM spectroscopies to distin-

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
guish among these candidate metabolites by comparing spectra derived from cells grown in 6 or 600 M MnEDTA with spectra generated from standard solutions of Mn-P i , Mn-Phy, and Mn-polyP (shown in Fig. 7B). The 1 H signals represent protons from water molecules bound to Mn, their intensity decreasing as the H 2 O ligands of Mn 2ϩ (H 2 O) 6 are replaced by metabolite ligands. For each P i molecule binding to Mn, one water molecule is displaced, whereas in Mn-polyP complexes, virtually all water molecules are displaced. The presence and abundance of P-based ligands can itself be inferred from the presence of their 31 P signals: in short, the Mn 2ϩ acts as an "indicator" of the relative concentrations of competing ligands free within the cellular environment. 31 P ENDOR spectra from Mn 2ϩ in general show a doublet centered around the 31 P Larmor frequency and separated by a hyperfine coupling, A Pi ϳ4; the peaks at lower/higher frequency are labeled as ϩ / Ϫ (Fig. S12). Although in many spectra of P i and polyP complexes, the lower-frequency ( Ϫ ) peak is not well-resolved, the single ϩ peak is sufficient to characterize the complex. Mn-P i and Mn-polyP complexes show similar 31 P couplings, but they exhibit different relative amplitudes of the 31 P and 1 H ENDOR responses, which together allow identification of the contribution from P i /polyP (21). Phytate complexes exhibit two partially overlapping doublets with hyperfine coupling A phy ϳ4, 8 MHz. (Fig. S12); again, the ϩ peaks are sufficient to characterize the complex, and the relative amounts of P i and phytate complexes can be deduced by decomposing an observed cellular 31 P ENDOR spectrum into individual component contributions.
For Mn-replete cells grown in 6 M MnEDTA, by combining 31 P, 1 H absolute ENDOR response (Fig. 7C) and comparing them with standards (Fig. 7B), we discovered that Mn-P i and Mn-Phy complexes accounted for ϳ50 and ϳ21%, respectively, of the total cellular Mn 2ϩ pool, with minimal contribution from polyP ( Fig. 7 (C, D, and F) and Fig. S13): for every 10 cellular Mn 2ϩ ions, 5 will have bound P i , and another 2 will be in a complex with Phy. This result was surprising to us, as we had hypothesized a polyP-based ligand in light of the vtc1 mutant phenotype. ENDOR spectra from cells grown in 600 M MnEDTA were likewise incompatible with a polyP ligand; in these cells, the 31 P peak decreased, whereas the 1 H peak was higher, suggesting that the Mn was not predominantly associated with polyP, as predicted (Fig. 7, C, E, and F). Indeed, the analysis suggested that on average only about 1 in 10 Mn atoms was bound to P i , whereas the fraction associated with Phy remained constant, with roughly 2 of 10 Mn complexed to Phy for every 10 Mn (Fig. 7 (E and F) and Fig. S13).
Cells grown under excess-Mn conditions accumulate Mn to about 10-fold the levels of Mn 2ϩ found in cells grown in Mnreplete conditions (Figs. 1B and 7G). As a result, although the fraction of Mn-P i decreases with increasing Mn accumulation (Fig. 7, D-F), ENDOR-derived speciation shows that the total cellular pool of Mn-P i complexes in fact doubles between Mn supply levels of 6 and 600 M. The fraction of Mn that binds Phy

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
does not change when cells are grown under excess Mn, so the total Mn-Phy cellular pool sees a 10-fold increase from 6 to 600 M Mn. Thus, using Mn 2ϩ as an indicator for its ligands and their concentrations, this means that as the cellular Mn 2ϩ increases, so does the ratio of ligand concentrations, [Phy]/[P i ].
Despite higher absolute contributions from Mn-P i and Mn-Phy, even under excess-Mn conditions, little Mn is bound by polyP. In addition, a careful investigation using ENDOR/ ESEEM revealed no 14 N signals, which indicates that WT cells do not contain a significant population of Mn 2ϩ coordinated by nitrogenous ligands. A high percentage of Mn may therefore be in a complex with ligand(s) not visible by ENDOR spectroscopy (most probably carboxylato metabolites).

Reversible sequestration of Mn by an ENDOR-silent ligand
The ENDOR results described above did not support the initial hypothesis that intracellular Mn 2ϩ is bound to polyP. We wondered whether Mn 2ϩ might, however, be transiently bound to polyP, which could be visualized in a time-course experiment. To this end, we first grew cells in 6 M MnEDTA and then shifted them to fresh medium containing 600 M MnEDTA to induce accumulation. We collected aliquots daily and measured intracellular Mn by elemental analysis and determined the fraction bound to Phy and P i . Mn 2ϩ accumulation only took place when cells reached the end of their exponential growth phase and transitioned to stationary (Fig. 8A). These results align well with the timing of polyP accumulation inside acidocacisomes in Chlamydomonas and other unicellular eukaryotes (58), again suggesting that polyP is an important contributor to Mn accumulation, even if we cannot detect substantial Mn-polyP metabolites. ENDOR spectroscopy of the same samples revealed constant Mn-Phy levels of about 20% and Mn-P i levels around 60% before cells reached stationary phase. However, as cells began to accumulate 15 times the levels of Mn between 36 and 48 h after inoculation, Mn-P i levels concomitantly decreased by 50%, with minimal contribution from Mn-polyP complexes (Fig. 8B), further arguing against a role for polyP as the Mn ligand at any stage during Mn accumulation (Fig. 8B).
In a reciprocal experiment, we transferred cells adapted to Mn excess into fresh medium with no added MnEDTA and collected samples every day. Cellular Mn content decreased drastically within 24 h and stabilized within 2 days, as cells first mobilized their intracellular stores to sustain growth (Fig. 8C). Mn-P i levels mirrored the changes in intracellular Mn: the fraction of Mn-P i complexes increased sharply within 2 days following the step down from 600 M to 0 M MnEDTA, but Mn-polyP complexes did not appear (Fig. 8D).
These results demonstrate that cells 1) accumulate Mn at the end of the exponential growth phase, 2) do not rely on polyP as a final storage form of Mn complexes, and 3) can quickly remobilize accumulated Mn during micronutrient scarcity, consistent with our earlier observation (Fig. 5).

Imidazole becomes a manganese ligand in the absence of polyP
What is the Mn 2ϩ speciation in the vtc1-1 mutant, which lacks polyP? ENDOR spectroscopy identified the Mn-P ligand complexes mainly as Mn-Phy (40%), with minimal contribution from Mn-P i (Fig. 9, A and B). Surprisingly, in contrast to WT (CC-4533 and VTC1 complemented strains), the vtc1-1 mutant had a significant population of Mn-14 N metabolites when grown under Mn-replete conditions. As depicted in Fig. 9C, the Mn-imidazole complex shows a strong three-pulse ESEEM time-wave modulation, which is caused by Mn-14 N electron nuclear hyperfine coupling to a bound imidazole. Under identical assay conditions (namely, samples were set to have equal amounts of Mn during the assay), the vtc1-1 mutant generated time-wave traces similar to those of the reference Mn-imidazole complex but with a signal amplitude only 20% of that from Mn-imidazole.

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
These results therefore indicate that ϳ20% of the cellular Mn 2ϩ ions in vtc1-1 are bound to an imidazole-type ligand in vtc1-1, presumably histidine-based, which could arise from low-molecular-weight metabolites and/or Mn-containing enzymes. In the green lineage, MnSODs and the oxygenevolving complex are the main Mn-bound enzymes, but the EPR spectra show no contribution from either and allow us to rule both out as the source of the imidazole-based ligand in vtc1 (59,60). The two WT strains (CC-4532 and the VTC1 complemented strain) have no discernable Mn-imidazole signal (Fig. 9C). Mn-imidazole metabolites may be present in WT strains, but their signal could be masked by Mn-P metabolites.

Discussion
To prevent mismetalation of enzymes that incorporate iron, zinc, or manganese as cofactors, cells maintain a limited cytosolic pool of these transition metals by the concerted action of regulated uptake and sequestration from the cytosol (17). In the unicellular alga C. reinhardtii and all photosynthetic organisms, mitochondrial superoxidase dismutases and the chloroplast photosystem II complex represent the main Mn sinks in the cell. Although Mn sequestration by the plant vacuole, Golgi, and endoplasmic reticulum have been established, how Chlamydomonas copes with and stores excess Mn is unclear. Our results, comprising observations from complementary approaches, implicate a lysosome-related organelle, the acidocalcisome, as the main sequestration site for cellular Mn. NanoSIMS (Fig. 2), XFM (Fig. S3), and EDX spectroscopy (Fig. S4) (Figs. 3 and 5). The VTC complex localizes to the vacuolar membrane and catalyzes the synthesis of polyP chains of variable length (from tens to hundreds of P i residues) by pulling P i from the cytosol and translocating the elongating polyP chain into the vacuole in an ATP-dependent manner. The polyP inside acidocalcisomes is likely to accumulate in a low-pH environment, surrounded by cations like Ca 2ϩ , Mg 2ϩ , and K ϩ , bringing polyP close to quantitative precipitation (49). Chlamydomonas vtc1 mutants lack electron-dense regions (observed by TEM and TEM-EDX in WT cells; Fig. S4) due to the loss of polyP crystals, likely associated with Ca 2ϩ . Nonetheless, we visualized intact acidocalcisomes in both WT and vtc1 mutant cells with Lysosensor (Fig. 4), and their membrane appears functional based on their ability to maintain internal acidity from the concerted proton concentration gradient generated by the H ϩ -pyrophosphatase and V-ATPase, whose abundances are not changed in the vtc1 mutant ( Fig. 4 and Fig. S7).
Chlamydomonas cells do not accumulate Mn when their cellular P levels are lowered genetically (as in the vtc1-1 mutant; Fig. 3) or nutritionally (Fig. S8). Mutations in the Vtc1 and Vtc4 subunits of the yeast VTC complex prevent accumulation of another divalent metal ion (Mg) inside the vacuole (37), suggesting that polyP is a general and critical component of metal sequestration. The underlying mechanism for Mg 2ϩ uptake does not rely on a dedicated transporter at the plasma membrane, but instead calls upon endocytosis from the growth medium and concentration into the vacuole of a Mg-EDTA

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
complex, with later stripping of the chelator agent and formation of a Mg-polyP complex (37).
Unlike Mg 2ϩ , which has no unpaired electrons and is hence invisible to paramagnetic resonance spectroscopies, the coordination state of Mn 2ϩ can be probed using EPR/ENDOR/ ESEEM spectroscopies (19,21). Hypotheses concerning divalent cation accumulation (37) can therefore be tested with Mn 2ϩ , but not for Mg 2ϩ . EPR spectra from Mn-accumulating cells categorically demonstrate that cellular Mn is not found complexed to EDTA, as the observed EPR signals display only the relatively narrow signal of a high-symmetry metabolite complex with well-defined six-line 55 Mn hyperfine pattern, which is very distinct from the broad and low-symmetry signal of Mn-EDTA, with a negligible 55 Mn hyperfine pattern (62). EPR spectroscopy therefore argues against significant uptake of EDTA from the growth medium by Chlamydomonas cells, even in the presence of 1 mM EDTA.
Our results show that Mn 2ϩ accumulates in the acidocalcisomes and further reveal that Mn 2ϩ is found as mononuclear complexes with inorganic P i and Phy, but not polyP. Further, EPR measurements of Mn 2ϩ in the presence of polyP do not suggest that polyP can form antiferromagnetically coupled polynuclear polyP-Mn 2ϩ complexes (Fig. S9), and the conclusion that virtually all of the Mn is in the Mn 2ϩ state is supported by the parallelism of changes in total Mn and in EPR-detectable Mn 2ϩ . Under replete conditions (6 M MnEDTA), about 70% of total Mn 2ϩ was complexed to either P i or Phy, but the contribution from polyP complexes was minimal (Figs. 7 and 9 and Figs. S11 and S12). Under Mn excess conditions (600 M MnEDTA), Mn-P i levels doubled, whereas Mn-Phy levels increased 10 times; however, together, Mn-P i and Mn-Phy only accounted for one-third of the total Mn accumulated in these cells, with the remaining two-thirds of the pool of Mn 2ϩ likely coordinated to carboxylato metabolites, which are ENDOR-silent. Again, polyP had a minimal contribution to final Mn complexation. Nevertheless, polyP must be present in cells to permit translocation of Mn into acidocalcisomes, as Chlamydomonas vtc1 mutants cannot accumulate Mn, pointing to a transitory role for polyP in Mn homeostasis.
The speciation profile of Mn in vtc1 mutants was very distinct from WT cells: ENDOR features derived from 31 P markedly decreased in the vtc1-1 mutant (Fig. 9, A and B), consistent with a much greater contribution from ENDOR-silent ligands. However, ESEEM spectroscopy shows that ϳ20% of Mn 2ϩ takes on an imidazole-based metabolite as a ligand in the absence of polyP, possibly acting to concentrate this divalent transition metal into acidocalcisomes. A similar population of Mn 2ϩ -imidazole complexes was observed previously in D. radiodurans (19).
We were surprised by how effective Chlamydomonas cells were at accumulating Mn, which is in stark contrast to the rates of uptake for Cu (41) and Fe (13), which becomes restricted once the cell has the amount it needs to maintain metalloprotein quotas. Much of the work on high Mn tolerance comes from yeast: a loss of function in the MnSOD enzyme Sod1p causes elevated oxidative stress, but this can be rescued by high concentrations of exogenous Mn. Once inside the cell, cytosolic Mn forms complexes with P i and can then compensate for the loss of Sod1p due to its high intrinsic superoxide dismutase activity (20). Mn and P metabolism are largely co-dependent: Mn and P are thought to be co-transported by phosphate transporters, which is supported by the lower Mn content of phosphate transporter mutants (36,63,64). However, not all yeast mutants with altered P content agree with this model, unless their polyP pool is also taken into account. For example, pho80 mutants exhibit a constitutive P deficiency phenotype, leading to 1) increased P uptake, 2) high polyP levels, and 3) much higher intracellular Mn content (34,46). The Mn pool in pho80 mutants is presumed to be cytosolic, but we hypothesize a vacuolar localization, where polyP acts as a molecular magnet to sequester Mn away from the cytosol, thereby robbing pho80 cells from Mn-catalyzed detoxification protection. In agreement with our hypothesis, Chlamydomonas vtc1 mutants and yeast vtc mutants fail to accumulate Mn and have much lower polyP levels (35,36,48,50) and may offer an alternative and more comprehensive explanation that ties Mn and P metabolism via polyP.
The crystal structure of the catalytic domain of yeast Vtc4 exposes another factor potentially contributing to the high capacity of cells to accumulate Mn (52). The substrate-binding site is coordinated with Mn 2ϩ via six positively charged and one negatively charged amino acids and a tyrosine residue that are conserved between yeast and Chlamydomonas (Fig. S14). Vtc4p is inactive without a metal cofactor, and Mn 2ϩ cations offered the strongest stimulation of polyP synthesis, followed by Zn 2ϩ , Co 2ϩ , Mg 2ϩ , and Ni 2ϩ (52). The potential for accumulation may become obscured by the associated toxicity of the transition metal, as would be expected for Zn 2ϩ , Co 2ϩ , and Ni 2ϩ . Only in the case of Mn 2ϩ (this study) and Mg 2ϩ (37) will the metals be taken up by cells and stored inside acidocalcisomes, all the more so that Mn 2ϩ and Mg 2ϩ will stimulate polyP synthesis in a feed-forward loop, as long as cytosolic inorganic P is available.
Previous attempts at estimating the Mn quota of healthy Chlamydomonas cells did not distinguish between Mn pools participating in photosynthesis or stored in acidocalcisomes. The pool of Mn sequestered inside the vacuole is fully bioavailable when cells are faced with Mn scarcity conditions (Fig. 5). The Chlamydomonas vtc1 mutant cannot synthesize or store polyP in its acidocalcisomes and therefore lacks one major cellular Mn sink, which grants us access to the photosynthetic Mn quota, which we estimate to be ϳ2 ϫ 10 7 Mn atoms/cell (instead of the previous estimate of 4 -5 ϫ 10 7 Mn atoms per cell; Fig. 3) (65). Likewise, we can estimate the polyP content of Chlamydomonas cells relative to total P content using the vtc1 mutant: about 60% of all cellular P is engaged in polyP and other low molecular weight complexes.
In conclusion, we show here that Chlamydomonas cells can accumulate the transition metal Mn 2ϩ to very high levels, complexed in part with P i and phytate metabolites. PolyP is critical for this process but is not the final ligand for Mn, as little Mn-polyP complex is found in any of the variants studies. Instead, we propose that Chlamydomonas cells utilize polyP to concentrate Mn 2ϩ cations into acidocalcisomes. At least in yeast, high cytosolic levels of polyP are toxic, and polyP cannot be translo-

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
cated to the vacuole from the cytosol (66). The role of polyP in Mn 2ϩ uptake is likely restricted to the acidocalcisomes, where we propose it plays an escorting role to a final and currently unknown ligand(s). Our results may also cast the role of polyP in Mg 2ϩ acquisition in a different light (37). Our results offer the testable alternative that the final ligand for Mg 2ϩ does not in fact contain phosphates, as initially presumed (37). The speciation state of Mn 2ϩ will provide an excellent indicator for the yeast vacuolar environment in cells grown in the presence of high concentrations of EDTA. The ability to identify whether EDTA is an intracellular ligand will shed light on the sequestration of metals into acidocalcisomes/vacuoles in Chlamydomonas and yeast and will refine the role of polyP within this cellular context.

Strains and culture conditions
C. reinhardtii WT strain CC-4532, mutant strain vtc1-1 (CC-5321), and vtc1-1 strain complemented with the VTC1 locus (here referred to as VTC1, CC-5324) were cultured in Tris acetate-phosphate (TAP) medium with revised trace elements (65). Cultures were incubated at 24°C with constant agitation (180 rpm) in an Innova incubator (New Brunswick Scientific, Edison, NJ) in continuous light (ϳ100 mol of photons per m Ϫ2 ⅐s Ϫ1 ) provided by cool white fluorescent bulbs (4100 K) and warm white fluorescent bulbs (3000 K) in a ratio of 2:1. High photon flux density (350 -420 mol of photons per m Ϫ2 ⅐s Ϫ1 ) was provided by a white fluorescent bulb (3800 K).
Where indicated, excess manganese was supplied as EDTAchelated MnCl 2 . Manganese-deficient medium was prepared without Mn supplementation (5). For photoautotrophic growth, acetate was eliminated from TAP medium, the pH was adjusted to 7.4 with HCl, and the cultures were bubbled with air. Phosphorus-free medium (Tris acetate, TA), was prepared by replacing potassium phosphate with 1.2 mM potassium chloride (67).
For all treatments, precultures were grown in TAP medium until mid-log growth phase (3-5 ϫ 10 6 cells/ml) and then used to inoculate test cultures at an initial cell density of 1 ϫ 10 4 cells/ml. Cultures were collected for analysis at a density of 3-4 ϫ 10 6 cells/ml. For phosphorus deficiency, cells grown in TAP medium were collected by centrifugation (3,500 ϫ g, 3 min) and washed twice with TA medium prior to inoculation in TA medium containing the indicated amounts of manganese at cell density of 1 ϫ 10 5 cells/ml (67).
The vtc1-1 mutant and complemented strain (VTC1) were grown in photoautotrophic conditions for 2 rounds (1 week each) with 6 M Mn with air bubbling. For pretreatment, flasks were inoculated at 1 ϫ 10 4 cells/ml with 6 or 600 M Mn and allowed to grow for 1 week with air bubbling. Cells were collected by centrifugation for 5 min at room temperature and washed twice in 5 mM EDTA and once in milliQ water. Test flasks with no added Mn were inoculated with washed cells at a density of 5 ϫ 10 4 cells/ml and grown for 7 days with air bubbling.
For cell size analysis, a Cellometer Auto M10 (Nexcelom Bioscience) was used. Samples from three independent experi-ments were collected at mid-log growth phase, and cell diameter was determined from 1,100 cells/sample.

Chlorophyll content and chlorophyll fluorescence measurements
Chlorophyll was extracted from whole cells with an acetone/ methanol (80:20, v/v) mixture. Total chlorophyll (a and b) content was estimated from the absorbance at 646.6 and 663.6 nm measured on a PerkinElmer Life Sciences LAMBDA 25 UVvisible spectrometer with absorbance at 750 nm as a reference to remove background from cell debris, as described (68). Each measurement was done in two technical replicates collected from three independent experimental replicates.
Minimum chlorophyll fluorescence (F o ) and maximum quantum yield of photosystem II (F v /F m ) were determined in cells after 10-min dark adaptation under a saturating pulse with an IMAGING-PAM MAXI chlorophyll fluorometer and Imag-ingWin software (Heinz Walz, Effeltrich, Germany).

Elemental content measurements
1 ϫ 10 8 cells from all cultures were collected by centrifugation at 3,500 ϫ g for 3 min and washed three times with 1 mM Na 2 -EDTA, pH 8.0, to remove metals associated with the cell surface and twice with Milli-Q water. After removing the remaining water by brief centrifugation, cell pellets were digested with 70% nitric acid at room temperature overnight and at 65°C for 2 h. Digested samples were diluted with Milli-Q water to a final nitric acid concentration of 2% (v/v). To measure metal content of culture medium, aliquots of the medium were treated with nitric acid and diluted with Milli-Q water to reach a final concentration of 2% nitric acid (v/v). Elemental analysis was measured by inductively coupled plasma MS on an Agilent 8800 Triple Quadrupole ICP-MS instrument using three standards for calibration (an environmental calibration standard (Agilent 5183-4688), phosphorus standard (Inorganic Ventures CGP1), and sulfur standard (Inorganic Ventures CGS1)) and two internal standards ( 89 Y and 45 Sc (Inorganic Ventures MSY-100PPM and MSSC-100PPM, respectively)). Elements were determined in MS/MS mode and measured in a collision reaction cell using helium for the measurement of 55 Mn, 63 Cu, and 66 Zn; hydrogen for 56 Fe and 40 Ca; and oxygen for 31 P as gas, as described previously (69). Each sample was measured in four technical replicates, and variation between technical replicates did not exceed 5%.

Nanoscale secondary ion MS
3 ϫ 10 6 cells from WT laboratory strain CC-4532 from three independent cultures were collected by centrifugation (3,500 ϫ g, 5 min) and washed twice with 10 mM sodium phosphate, pH 7.0, before immersion in a fixing solution containing 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M sodium phosphate, pH 7.4, for 2 h at room temperature and then incubated at 4°C.
Cells were washed five times with 0.1 M sodium phosphate buffer, pH 7.4, and post-fixed in a solution of 1% OsO 4 in sodium phosphate buffer, pH 7.4. Samples were further washed four times in Na acetate buffer, pH 5.5, and dehydrated in an ethanol gradient (50, 75, 95, 100, 100, and 100%) for 10 min each. Dehydrated samples were passed through propylene Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis oxide and infiltrated in a 1:1 mixture of Epon 812 and propylene oxide for 2 h and in a 2:1 mixture for 2 h. Finally, the previous mixture was infiltrated in pure Epon 812 overnight and cured in an oven at 60°C for 48 h. Sections of 200-nm thickness (gray interference color) were cut on an ultramicrotome (RMC MTX) using a diamond knife and deposited on 100-mesh carbon-coated molybdenum grids (G100-Mo, Electron Microscopy Sciences).
The nanoSIMS ion image data were processed quantitatively using custom software (L'Image, L. R. Nittler, Carnegie Institution for Science, Washington, D. C.). The ion images were corrected for detector dead time and image shifts between scans and then used to produce ion images. Regions of interest (ROIs) were defined using an automated algorithm to subdivide the analyzed area into hexagons (Fig. S4). Hexagons that were not on cells were deleted. Ion ratios ( 31 P ϩ / 12 C ϩ , 40 Ca ϩ / 12 C ϩ , and 55 Mn ϩ / 12 C ϩ ) for each ROI were calculated by averaging the ratios over replicate scans. These data provide relative quantitative composition, but they were not standardized to provide concentration data.

Confocal microscopy
Chlamydomonas cells were cultured to early stationary phase and collected by centrifugation (3,500 ϫ g, 3 min) and washed twice with 10 mM sodium phosphate, pH 7.5. All fluorescent dyes were diluted in 10 mM sodium phosphate, pH 7.5, to a final concentration of 2 M. Cells were treated with a Lysosensor DND189, Lysosensor DND167, or M1 Mn probe and mounted on glass slides for immediate visualization. Cells treated with DAPI were first fixed for 10 min with 2% glutaraldehyde, followed by permeabilization with 40 M digitonin for 10 min before mounting on glass slides for visualization. Microscopy was performed on a Zeiss LSCM Airyscan 880 ( Fig.  6 (C and D) and Fig. S6) equipped with a ϫ63/1.4 oil immersion objective or a Zeiss Elyra using Lattice SIM with a ϫ100/1.46 oil objective (Fig. 6, A and B). Images were recorded using filter sets or spectral mode as indicated in the figure legends. All aspects of image capture were controlled via Zeiss ZEN Black software, including fluorescent emission signals from probes and/or chlorophyll.

X-ray absorption spectroscopy
Multiple independent samples of 5 ϫ 10 8 cells from cultures treated with excess manganese were collected by centrifugation (3,500 ϫ g, 5 min) and washed twice with 1 mM EDTA and twice with Milli-Q water to remove metals associated with the cell surface and subsequently washed with 10 mM sodium phosphate. After removing the liquid, cells were suspended in 30% glycerol (v/v) (estimated from volume of cell pellet) and loaded with a Hamilton syringe into a Kapton-wrapped lucite XAS sample cell and flash-frozen in liquid nitrogen. Samples were stored in liquid nitrogen until beam exposure. XAS data were collected at the Stanford Synchotron Radiation Lightsource on beamline 9-3, which is equipped with a Si(220) double-crystal monochromator, equipped with a focusing mirror that also provides harmonic rejection. Fluorescence spectra were collected using a 100element germanium solid-state detector from Canberra. During data collection, the Oxford Instruments continuousflow liquid helium cryostat was stabilized at 10 K. Manganese data were collected using a 3-m chromium filter placed between the cryostat and the detector to reduce unassociated scattering. Mn foil spectra were collected simultaneously with the protein data for energy calibration. The first inflection point for Mn was set at 6,543.3 eV. Manganese XAS spectra were recorded using 5-eV steps in the pre-edge regions and 0.25-eV steps in the edge regions and 0.05 Å Ϫ1 increments in the EXAFS region (to k ϭ 13 Å Ϫ1 ), integrating from 1 to 20 s in a k 3 -weighted manner. An average of 6 scans were collected and averaged for each sample. Each scan lasted ϳ45 min. Each spectrum was closely monitored for X-ray-induced radiation damage.
XAS spectra were processed and analyzed using the EXAF-SPAK program suite written for Macintosh OSX. Fluorescence scans corresponding to each channel were examined for anomalies. A Gaussian function was used in the pre-edge region, and a three-region cubic spline was used in the EXAFS region. EXAFS data were converted to k space using E 0 values of 6,560.00 eV. Spectra were simulated using single and multiple scattering amplitude and phase functions generated using the Feff version 8 software package integrated within EXAFSPAK. Single scattering models were calculated for oxygen, nitrogen, phosphorous, and carbon to simulate possible manganese ligand environments. Calibrated scale factors (Sc) and model E 0 values were not allowed to vary during fitting; the Sc for manganese samples was 0.95. Mn data were fit out to a k value of 13.0 Å Ϫ1 . Calibration from Mn (II) and Mn(III) theoretical model compounds were used to determine the fit E 0 and Sc parameters. E 0 values for Mn-O, Mn-C were set at Ϫ10, and Mn-P was set at Ϫ12. EXAFS spectra were simulated using both filtered and unfiltered data; however, simulation results were presented for only fits to raw (unfiltered) data. Simulation protocols and criteria for determining the best fit were as described previously (56).

Robust Chlamydomonas acidocalcisome-mediated Mn homeostasis
Electron paramagnetic and electron nuclear double resonance 2 ϫ 10 8 cells from cultures were collected by centrifugation (3,500 ϫ g, 5 min), and metals associated with the cell surface were removed by washing twice with 1 mM EDTA, pH 8, and twice with Milli-Q water. After a subsequent wash with 50 mM HEPES, pH 7.0, cells were resuspended in 30% glycerol in 50 mM HEPES, pH 7.0, and loaded into an ENDOR sample tube, flash-frozen in liquid nitrogen, and stored at Ϫ80°C until analysis (21). 35-GHz CW EPR spectra were recorded using a laboratory-built 35-GHz EPR spectrometer (74). As described previously (21), the absorption-display EPR spectra (collected from cellular Chlamydomonas and Mn-metabolites by using continuous-wave (CW) "rapid passage" methods at 2 K) are characteristic of an S ϭ 5 ⁄ 2 ion with small zero-field splitting (ZFS), with the principal ZFS parameter, D, much less than the microwave quantum (h). Such spectra show a central 55Mn (I ϭ 5 ⁄ 2) sextet arising from hyperfine interactions, A ϳ91 G, that is associated with transitions between the m s ϭ ϩ 1 ⁄ 2 and Ϫ 1 ⁄ 2 electron-spin substrates. These features "ride on" and are flanked by significantly broader signals from the four "satellite" transitions involving the other electron-spin substrates (m s Ϯ 5 ⁄ 2 7 Ϯ 3 ⁄ 2; Ϯ 3 ⁄ 2 7 Ϯ 1 ⁄ 2). The net absorption spectrum is the sum of the five envelopes of these five transitions among substrates.
Pulsed ENDOR/ESEEM spectra were recorded using a laboratory-built 35-GHz pulsed EPR spectrometer (75). All spectra were recorded at 2 K, which was achieved using an immersion helium cryostat. 31 P, 1 H Davies ENDOR spectra were recorded using the pulse sequence (-T-/2----echo), where T is the time interval for which the radiofrequency pulse is randomly hopped. ENDOR of a paramagnetic metal ion center, such as Mn 2ϩ , provides an NMR spectrum of the nuclei that is hyperfine-coupled to the electron spin (76) and thus can be used to identify and characterize coordinating ligands (77,78). The frozen-solution spectrum of an I ϭ 1 ⁄ 2 nucleus, such as 31 P and 1 H, coupled to Mn 2ϩ comprises a set of doublet features centered at the nuclear Larmor frequency and split by multiples of the electron-nuclear hyperfine coupling (A). The primary doublet is associated with the m s ϭ Ϯ 1 ⁄ 2 electron spin sublevels of Mn 2ϩ and is split by A; weaker satellite doublets associated with the m s ϭ Ϯ 3 ⁄ 2 and Ϯ 5 ⁄ 2 sublevels are split by 3A and 5A. All spectra in this study display 1 H signals that can be assigned to the protons of bound water. In addition, all of the spectra, except for the aqueous solution, show a sharp m s ϭ Ϯ 1 ⁄ 2 31 P doublet from a phosphate moiety bound to the Mn 2ϩ center. The intensities of 31 P and 1 H signals differ significantly among the spectra, and analysis of these intensities provides a means of assessing cellular Mn 2ϩ speciation (21).
Three-pulse ESEEM spectra were recorded using the pulse sequence, /2 Ϫ Ϫ /2 Ϫ T Ϫ /2 Ϫ Ϫ echo, where T is the time varied between second and third microwave pulses, with four-step phase cycling to suppress unwanted Hahn and refocused echoes (79). A 14 N nucleus (I ϭ 1) directly coordinated with 55 Mn creates modulation in the electron spin echo decay, which is dominated by 14 N hyperfine interaction. To quantitate 14 N ESEEM responses from cellular Mn 2ϩ , we chose as a standard the 14 N response from the Mn-imidazole complex, which binds one imidazole and (presumably) five waters.